&EPA
United States
Environmental Protection
Agency
      Fourth External Review Draft
      of Air Quality Criteria for
      Particulate Matter (June, 2003)
      Volume I

-------
                                                     EPA/600/P-99/002aD
                                                            June 2003
                                               Fourth External Review Draft
Air Quality Criteria for Particulate Matter
                      Volume I
       National Center for Environmental Assessment-RTP Office
               Office of Research and Development
              U.S. Environmental Protection Agency
                  Research Triangle Park, NC

-------
 1                                         DISCLAIMER
 2
 3           This document is an external review draft for review purposes only and does not constitute
 4      U.S. Environmental Protection Agency policy.  Mention of trade names or commercial products
 5      does not constitute endorsement or recommendation for use.
 6
 7
 8                                           PREFACE
 9
10           National Ambient Air Quality Standards (NAAQS) are promulgated by the United States
11      Environmental Protection Agency (EPA) to meet requirements set forth in Sections  108 and 109
12      of the U.S. Clean Air Act (CAA). Sections 108 and 109 require the EPA Administrator (1) to
13      list widespread air pollutants that reasonably may be expected to endanger public health or
14      welfare; (2) to issue air quality criteria for them that assess the latest available scientific
15      information on nature and effects of ambient exposure to them; (3) to set "primary" NAAQS to
16      protect human health with adequate margin of safety and to set "secondary" NAAQS to protect
17      against welfare effects (e.g., effects on vegetation, ecosystems, visibility, climate, manmade
18      materials, etc.); and (5) to periodically (every 5 years) review and revise, as appropriate, the
19      criteria and NAAQS for a given listed pollutant or class of pollutants.
20           The original NAAQS for particulate matter (PM), issued in 1971 as "total suspended
21      parti culate" (TSP) standards, were revised in 1987 to focus on protecting against human health
22      effects  associated with exposure to ambient PM less than 10 microns (< 10 jim) that are capable
23      of being deposited in thoracic (tracheobronchial  and alveolar) portions of the lower respiratory
24      tract. Later periodic reevaluation of newly available scientific information, as presented in the
25      last previous version of this "Air Quality Criteria for Parti culate Matter" document published in
26      1996, provided key scientific bases for PM NAAQS decisions published in July 1997. More
27      specifically, the PM10 NAAQS set in  1987 (150 |ig/m3, 24-h; 50 |ig/m3, annual average) were
28      retained in modified  form and new standards (65 |ig/m3, 24-h; 15 |ig/m3, annual average) for
29      particles < 2.5 jim (PM25) were promulgated in July 1997.
30           This Fourth External Review Draft of revised  Air Quality Criteria for Particulate Matter
31      assesses new scientific information that has become available mainly between early  1996

        June 2003                                 I-ii        DRAFT-DO NOT QUOTE OR CITE

-------
 1     through April 2002. The present draft is being released for public comment and review by the
 2     Clean Air Scientific Advisory Committee (CASAC) to obtain comments on the organization and
 3     structure of the document, the issues addressed, the approaches employed in assessing and
 4     interpreting the newly available information on PM exposures and effects, and the key findings
 5     and conclusions arrived at as a consequence of this assessment. Public comments and CASAC
 6     review recommendations will be taken into account in making any appropriate further revisions
 7     to this document for incorporation into a final draft. Evaluations contained in the present
 8     document will be drawn on to provide inputs to associated PM Staff Paper analyses prepared by
 9     EPA's Office of Air Quality Planning and Standards (OAQPS) to pose alternatives for
10     consideration by the EPA Administrator with regard to proposal and, ultimately, promulgation of
11     decisions on potential retention or revision of the current PM NAAQS.
12           Preparation of this document was coordinated by staff of EPA's National Center for
13     Environmental Assessment in Research Triangle Park (NCEA-RTP).  NCEA-RTP scientific
14     staff, together with experts from other EPA/ORD laboratories and academia, contributed to
15     writing of document chapters; and earlier drafts of this document were reviewed by experts from
16     federal and state government agencies, academia, industry, and non-governmental organizations
17     (NGOs) for use by EPA in support of decision making on potential public health and
18     environmental risks of ambient PM.  The document describes the nature, sources, distribution,
19     measurement, and concentrations of PM in outdoor (ambient) and indoor environments.  It also
20     evaluates the latest data on human exposures to ambient PM and consequent health effects in
21     exposed human populations (to support decision making  regarding primary, health-related PM
22     NAAQS).  The document also evaluates ambient PM  environmental effects on vegetation and
23     ecosystems, visibility, and man-made materials, as well as atmospheric PM effects on climate
24     change processes associated with alterations in atmospheric transmission of solar radiation or its
25     reflectance from the Earth's surface or atmosphere (to support decision making on secondary
26     PM NAAQS).
27           The NCEA  of EPA acknowledges the contributions provided by authors, contributors, and
28     reviewers and the diligence of its staff and contractors in the preparation of this document.
29
        June 2003                                 I-iii         DRAFT-DO NOT QUOTE OR CITE

-------
              Air Quality Criteria for Particulate Matter


                             VOLUME I


EXECUTIVE SUMMARY	E-l

1.   INTRODUCTION  	1-1

2.   PHYSICS, CHEMISTRY, AND MEASUREMENT OF PARTICULATE
    MATTER 	2-1
    APPENDIX 2A:  Techniques for Measurement of Semivolatile Organic
                   Compounds 	 2A-1
    APPENDIX 2B:  Analytical Techniques	2B-1

3.   CONCENTRATIONS, SOURCES, AND EMISSIONS OF ATMOSPHERIC
    PARTICULATE MATTER	3-1
    APPENDIX 3 A:  Spatial and Temporal Variability of the Nationwide
                   AIRS PM25 and PM10.25 Data Sets  	 3A-1
    APPENDIX 3B:  Aerosol Composition Data from the Speciation
                   Network	3B-1
    APPENDIX 3C:  Organic Composition of Particulate Matter  	3C-1
    APPENDIX 3D:  Composition of Particulate Matter Source Emissions .... 3D-1
    APPENDIX 3E:  Variability Observed in PM2 5 and PM10_2 5 Concentrations
                   at IMPROVE Sites	3E-1

4.   ENVIRONMENTAL EFFECTS OF AIRBORNE PARTICULATE
    MATTER 	4-1
    APPENDIX 4A:  Colloquial and Latin Names 	 4A-1

3.   CONCENTRATIONS, SOURCES, AND EMISSIONS OF
    ATMOSPHERIC PARTICULATE MATTER	3-1

4.   ENVIRONMENTAL EFFECTS OF AIRBORNE PARTICULATE
    MATTER 	4-1

5.   HUMAN EXPOSURE TO PARTICULATE MATTER AND ITS
    CONSTITUENTS	5-1
June 2003                         I-iv    DRAFT-DO NOT QUOTE OR CITE

-------
              Air Quality Criteria for Particulate Matter
                                (cont'd)
                              VOLUME II
6.   DOSIMETRY OF PARTICULATE MATTER	6-1

7.   TOXICOLOGY OF PARTICULATE MATTER IN HUMANS AND
    LABORATORY ANIMALS  	7-1

8.   EPIDEMIOLOGY OF HUMAN HEALTH EFFECTS ASSOCIATED
    WITH AMBIENT PARTICULATE MATTER	8-1
    APPENDIX 8A:  Short-Term PM Exposure-Mortality Studies:
                   Summary Tables	  8A-1
    APPENDIX 8B:  Particulate Matter-Morbidity Studies: Summary Tables  . . . 8B-1

9.   INTEGRATIVE SYNTHESIS	9-1
    APPENDIX 9A:  Key Quantitative Estimates of Relative Risk for
                   Particulate Matter-Related Health Effects Based on
                   Epidemiologic Studies of U.S. and Canadian Cities
                   Assessed in the 1996 Particulate Matter Air Quality
                   Criteria Document	  9A-1
June 2003                          I-v     DRAFT-DO NOT QUOTE OR CITE

-------
                                Table of Contents

                                                                              Page

List of Tables	  I-xiii
List of Figures  	I-xix
Authors, Contributors, and Reviewers	I-xxxi
U.S. Environmental Protection Agency Project Team for Development of Air Quality
        Criteria for Particulate Matter	I-xxxix
Abbreviations and Acronyms  	  I-xlii

EXECUTIVE SUMMARY	E-l

1.     INTRODUCTION	1-1
      1.1    LEGISLATIVE REQUIREMENTS	1-1
      1.2    HISTORY OF PREVIOUS PM CRITERIA AND NAAQS REVIEWS  	1-2
            1.2.1    The 1997 PM NAAQS Revision  	1-3
            1.2.2    Coordinated Particulate Matter Research Program 	1-6
      1.3    CURRENT PM CRITERIA AND NAAQS REVIEW	1-9
            1.3.1    Key Milestones	1-9
            1.3.2    Methods and Procedures for Document Preparation	1-13
            1.3.3    Approach  	1-15
      1.4    DOCUMENT ORGANIZATION AND CONTENT 	1-15
      REFERENCES 	1-17

2.     PHYSICS, CHEMISTRY, AND MEASUREMENT OF PARTICULATE
      MATTER	2-1
      2.1    PHYSICS AND CHEMISTRY OF PARTICULATE MATTER	2-2
            2.1.1   Basic Concepts 	2-2
            2.1.2   Physical Properties and Processes  	2-3
                    2.1.2.1    Definitions of Particle Diameter	2-3
                    2.1.2.2   Aerosol Size Distributions 	2-5
                    2.1.2.3    Ultrafme Particles	2-20
            2.1.3   Chemistry of Atmospheric Particulate Matter 	2-28
                    2.1.3.1    Chemical Composition and Its Dependence  on
                              Particle Size 	2-29
                    2.1.3.2   Primary and Secondary Particulate Matter	2-29
                    2.1.3.3    Particle-Vapor Partitioning	2-31
                    2.1.3.4   Atmospheric Lifetimes and Removal Processes	2-34
            2.1.4   Comparison of Fine and Coarse Particles	2-36
      2.2    MEASUREMENT OF PARTICULATE MATTER	2-38
            2.2.1   Particle Measurements of Interest	2-38
            2.2.2   Issues in Measurement of Particulate Matter	2-40
                    2.2.2.1    Artifacts Due to Chemical Reactions	2-41
                    2.2.2.2   Treatment of Semivolatile Components of
                              Particulate Matter	2-42
                    2.2.2.3    Upper Cut Point 	2-43

June 2003                                I-vi        DRAFT-DO NOT QUOTE OR CITE

-------
                                 Table of Contents
                                       (cont'd)
                                                                                 Page
                     2.2.2.4    Cut Point for Separation of Fine and Coarse
                               Paniculate Matter	2-44
                     2.2.2.5    Treatment of Pressure, Temperature, and Relative
                               Humidity	2-47
                     2.2.2.6    Time Resolution 	2-48
                     2.2.2.7    Accuracy and Precision   	2-49
             2.2.3    Measurement of Semivolatile Particulate Matter	2-51
                     2.2.3.1    Particle-Bound Water	2-52
                     2.2.3.2    Nitrate and Organic Species	2-57
                     2.2.3.3    Continuous Measurement of Semivolatile and
                               Nonvolatile Mass 	2-64
             2.2.4    U. S. Environmental Protection Agency Monitoring Methods .... 2-68
                     2.2.4.1    The Federal Reference Methods for Measurement
                               of Equilibrated Mass for PM10, PM2 5, and PM10_25	2-68
             2.2.5    Speciation Monitoring 	2-82
             2.2.6    Inorganic Elemental Analyses 	2-85
             2.2.7    Elemental and Organic Carbon in Particulate Matter	2-86
             2.2.8    Ionic Species	2-88
             2.2.9    Continuous Monitoring	2-88
             2.2.10   Measurements of Individual Particles	2-90
             2.2.11   Low Flow Filter Samples for Multiday Collection of
                     Particulate Matter	2-92
      2.3     SUMMARY AND KEY POINTS	2-93
             2.3.1    Atmospheric Physics and Chemistry of Particles	2-93
             2.3.2    Measurement of Atmospheric Particles  	2-96
             2.3.3    Key Points	2-100
      REFERENCES  	2-102

APPENDIX 2 A.  Techniques for Measurement of Semivolatile Organic Compounds ....  2A-1
APPENDIX 2B.  Analytical Techniques 	2B-1
      2B.1 INORGANIC ELEMENTS 	2B-1
           2B.1.1   Energy Dispersive X-Ray Fluorescence	2B-1
           2B.1.2   Synchrotron Induced X-Ray Fluorescence	2B-1
           2B.1.3   Proton (or Particle) Induced X-Ray Emission	2B-2
           2B.1.4   Proton Elastic Scattering Analysis	2B-3
           2B.1.5   Total Reflection X-Ray Fluorescence  	2B-3
           2B.1.6   Instrumental Neutron Activation Analysis	2B-4
           2B.1.7   Atomic Absorption Spectrophotometry	2B-5
           2B. 1.8   Inductively Coupled Plasma with Atomic Emission
                    Spectroscopy  	2B-6
           2B.1.9   Inductively Coupled Plasma with Mass Spectroscopy  	2B-6
           2B.1.10  Scanning Electron Microscopy 	2B-7

June 2003                               I-vii        DRAFT-DO NOT QUOTE OR CITE

-------
                               Table of Contents
                                    (cont'd)
                                                                           Page
     2B.2  ORGANIC AND INORGANIC CARBON	2B-8
     2B.3  CONTINUOUS METHODS  	2B-17
           2B.3.1   Continuous Measurement of Mass	2B-17
           2B.3.2   Continuous Measurement of Organic and/or Elemental Carbon . . . 2B-21
           2B.3.3   Continuous Measurements of Nitrate and Sulfate	2B-24
     2B.4  OTHER CONTINUOUS MEASUREMENTS	2B-26

3.    CONCENTRATIONS, SOURCES, AND EMISSIONS OF ATMOSPHERIC
     PARTICULATE MATTER  	3-1
     3.1   INTRODUCTION  	3-1
     3.2   PATTERNS AND TRENDS IN AMBIENT PM CONCENTRATIONS 	3-2
           3.2.1     Seasonal Variability inPM Concentrations	3-13
           3.2.2     Diurnal (Circadian) Variability inPM Concentrations 	3-30
           3.2.3     Relations Among Particulate Matter in Different Size Fractions  ... 3-32
           3.2.4     Relations Between Mass and Chemical Component
                   Concentrations  	3-34
           3.2.5     Spatial Variability in Particulate Matter and its Components 	3-41
     3.3   SOURCES OF PRIMARY AND SECONDARY PARTICULATE
           MATTER 	3-59
           3.3.1     Chemistry of Secondary PM Formation	3-62
           3.3.2     Source Contributions to Ambient PM Determined by Receptor
                   Models  	3-72
           3.3.3     Background Concentrations of PM in the United States	3-82
           3.3.4     Emissions Estimates for Primary Particulate Matter, and
                   Precursors to Secondary Particulate Matter (SO2, NOX, VOCs,
                   and NH3) in the United States  	3-91
           3.3.5     Uncertainties of Emissions Inventories  	3-97
     3.4   SUMMARY AND KEY CONCLUSIONS  	3-99
     REFERENCES  	3-105

APPENDIX 3 A:  Spatial and Temporal Variability of the Nationwide AIRS PM2 5
               and PM10_25 Data Sets	  3A-l
     REFERENCES	  3A-3
APPENDIX 3B:  Aerosol Composition Data from the Speciation Network	3B-1
     REFERENCES  	3B-3
APPENDIX 3C:  Organic Composition of Particulate Matter	3C-1
     REFERENCES  	3C-14
APPENDIX 3D:  Composition of Particulate Matter Source Emissions	  3D-1
     REFERENCES  	  3D-25
APPENDIX 3E:  Variability Observed in PM2 5 and PM10_2 5 Concentrations at
               IMPROVE Sites	3E-1
     REFERENCES  	3E-5

June 2003                             I-viii       DRAFT-DO NOT QUOTE OR CITE

-------
                                Table of Contents
                                      (cont'd)
                                                                               Page
4.     ENVIRONMENTAL EFFECTS OF AIRBORNE PARTICIPATE MATTER	4-1
      4.1     INTRODUCTION	4-1
      4.2     EFFECTS OF AMBIENT AIRBORNE PM ON VEGETATION AND
             NATURAL ECOSYSTEMS	4-1
             4.2.1    Ecological Attributes  	4-2
             4.2.2    Ecosystem Exposures - Particle Deposition	4-4
                     4.2.2.1    Fine and Coarse Paniculate Matter	4-5
                     4.2.2.2    Diversity of Deposition Modes	4-8
                     4.2.2.3    Magnitude of Deposition  	4-44
             4.2.3    Assessment of Atmospheric PM Deposit!on Effects 	4-55
                     4.2.3.1    Effects on Vegetation and Ecosystems	4-60
                     4.2.3.2    Ecosystem Response to Stresses	4-79
             4.2.4    Urban Ecosystems	4-149
      4.3     AIRBORNE PARTICLE EFFECTS ON VISIBILITY	4-153
             4.3.1    Introduction	4-153
             4.3.2    Factors Affecting Atmospheric Visibility	4-155
                     4.3.2.1    Optical Properties of the Atmosphere and
                              Atmospheric Particles	4-155
                     4.3.2.2    Relative Humidity Effects on Particle Size and
                              Light-Scattering Properties	4-163
             4.3.3    Relationships Between Particles and Visibility  	4-166
             4.3.4    Photographic Modeling of Visibility Impairment  	4-174
             4.3.5    Visibility Monitoring Methods and Networks 	4-175
             4.3.6    Visibility Impairment:  Trends and Current Conditions	4-178
                     4.3.6.1    Trends in Visibility Impairment	4-178
                     4.4.6.2    Current Conditions 	4-187
             4.3.7    Economics of Particulate Matter Visibility Effects 	4-187
      4.4     PARTICULATE MATTER EFFECTS ON MATERIALS	4-191
             4.4.1    Corrosive Effects of Particles and Sulfur Dioxide on
                     Man-Made Surfaces 	4-192
                     4.4.1.1    Metals	4-192
                     4.4.1.2    Painted Finishes  	4-194
                     4.4.1.3    Stone and Concrete	4-197
             4.4.2    Soiling and Discoloration of Man-Made Surfaces	4-202
                     4.4.2.1    Stones and Concrete	4-203
                     4.4.2.2    Household and Industrial Paints	4-204
      4.5     ATMOSPHERIC PARTICULATE MATTER, CLIMATE CHANGE,
             AND EFFECTS ON SOLAR UVB RADIATION TRANSMISSION	4-205
             4.5.1    Atmospheric Particle Interactions with Solar and Terrestrial
                     Radiation Related to Climate Change	4-206
                     4.5.1.1    The Projected Impacts of Global Climate Change .... 4-206
June 2003                                I-ix        DRAFT-DO NOT QUOTE OR CITE

-------
                                 Table of Contents
                                       (cont'd)
                                                                                Page
                     4.5.1.2    Airborne Particle Relationships to Global
                               Warming and Climate Change  	4-209
             4.5.2    Atmospheric Particulate Matter Effects on the Transmission
                     of Solar Ultraviolet Radiation Transmission: Impacts on
                     Human Health and the Environment	4-220
                     4.5.2.1    Potential Effects of Increased Ultraviolet Radiation
                               Transmission	4-220
                     4.5.2.2    Airborne Particle Effects on Atmospheric
                               Transmission of Solar Ultraviolet Radiation 	4-223
      4.6     SUMMARY AND KEY CONCLUSIONS	4-227
             4.6.1    Particulate Matter Effects on Vegetation and Ecosystems	4-227
             4.6.2    Particulate Matter-Related Effects on Visibility	4-235
             4.6.3    Particulate Matter-Related Effects on Materials	4-236
             4.6.4    Effects of Atmospheric Particulate Matter on Global Warming
                     Processes and Transmission of Solar Ultraviolet Radiation	4-237
      REFERENCES  	4-240

APPENDIX 4A:  Colloquial and Latin Names	  4A-1

5.     HUMAN EXPOSURE TO PARTICULATE MATTER AND ITS
      CONSTITUENTS	5-1
      5.1     INTRODUCTION	5-1
             5.1.1    Purpose	5-1
             5.1.2    Particulate Matter Mass and Constituents  	5-2
             5.1.3    Relationship to Past Documents	5-3
             5.1.4    Chapter Structure 	5-4
      5.2     BASIC CONCEPTS OF EXPOSURE	5-5
             5.2.1    The Concept of Exposure	5-5
             5.2.2    Components of Exposure  	5-6
             5.2.3    Quantification of Exposure	5-9
             5.2.4    Methods To Estimate Personal Exposure	5-10
                     5.2.4.1    Direct Measurement Methods  	5-11
                     5.2.4.2    Indirect or Modeling Methods  	5-12
                     5.2.4.3    Methods for Estimating Personal Exposure to Ambient
                               Particulate Matter	5-20
      5.3     SUMMARY OF PARTICULATE MATTER MASS  DATA	5-26
             5.3.1    Types of Particulate Matter Exposure Measurement Studies	5-26
             5.3.2    Available Data  	5-27
                     5.3.2.1    Personal Exposure Data 	5-27
                     5.3.2.2    Microenvironmental Data	5-32
                     5.3.2.3    Traffic-Related Microenvironments	5-40
June 2003                                I-x         DRAFT-DO NOT QUOTE OR CITE

-------
                                 Table of Contents
                                       (cont'd)
                                                                                 Page
                     5.3.2.4    Reanalyses of Previously-Reported Parti culate Matter
                               Exposure Data	5-40
             5.3.3    Factors Influencing and Key Findings on Particulate Matter
                     Exposures  	5-43
                     5.3.3.1    Relationship of Personal/Microenvironmental
                               Particulate Matter with Ambient Particulate Matter .... 5-43
                     5.3.3.2    Factors That Affect Relationships Between Personal
                               Exposure and Ambient PM	5-63
                     5.3.3.3    Effect of Ambient Sources on Exposures to Particulate
                               Matter	5-84
                     5.3.3.4    Correlations of Particulate Matter with Other
                               Pollutants 	5-87
      5.4     SUMMARY OF PARTICULATE MATTER CONSTITUENT DATA  .... 5-91
             5.4.1    Introduction	5-91
             5.4.2    Monitoring Studies That Address Particulate Matter
                     Constituents  	5-91
             5.4.3    Key Findings	5-91
                     5.4.3.1    Correlations of Personal and Indoor Concentrations
                               with Ambient Concentrations of Particulate Matter
                               Constituents 	5-91
             5.4.4    Factors Affecting Correlations  Between Ambient Measurements
                     and Personal or Microenvironmental Measurements of
                     Particulate Matter Constituents	5-102
             5.4.5    Limitations of Available Data	5-103
      5.5     IMPLICATIONS OF USING AMBIENT PARTICULATE MATTER
             CONCENTRATIONS IN TOXICOLOGICAL AND
             EPIDEMIOLOGICAL STUDIES OF PARTICULATE MATTER
             HEALTH EFFECTS  	5-104
             5.5.1    Toxicology  	5-104
             5.5.2    Potential Sources of Error  Resulting from Using
                     Ambient Particulate Matter Concentrations in Epidemiological
                     Analyses 	5-104
                     5.5.2.1    Associations Between Personal Exposures and
                               Ambient Particulate Matter Concentrations	5-107
                     5.5.2.2    Role of Compositional Differences in Exposure
                               Characterization for Epidemiology	5-111
                     5.5.2.3    Role of Spatial  Variability in Exposure
                               Characterization for Epidemiology	5-112
             5.5.3    Analysis of Exposure Measurement Error Issues in Particulate
                     Matter Epidemiology 	5-113
                     5.5.3.1    Time-Series Analyses	5-113
                     5.5.3.2    Studies of Chronic Effects  	5-114

June 2003                                 I-xi         DRAFT-DO NOT QUOTE OR CITE

-------
                              Table of Contents
                                    (cont'd)
                                                                          Page
            5.5.4   Conclusions from Analysis of Exposure Measurement Errors
                   on Particulate Matter Epidemiology	5-116
     5.6    SUMMARY OF OBSERVATIONS AND LIMITATIONS	5-118
     REFERENCES 	5-128
June 2003                             I-xii        DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Tables

Number                                                                          Page

1-1       Key Scientific Uncertainties Related to the Source-To-Response Framework ....  1-8

1-2       Research Topics and Questions Recommended by National Research Council
          to be Addressed by Expanded Particulate Matter Research Program	1-10

1 -3       Schedule for Development of the Current Revised Parti culate Matter Air
          Quality Criteria Document	1-11

2-1       Comparison of Ambient Particles, Fine Particles (Ultrafme Plus
          Accumulation-Mode), and Coarse Particles 	2-37

2-2       Parti culate Matter Components/Parameters of Interest for Health, Ecological,
          or Radiative Effects; for Source Category Apportionment Studies; or for
          Air Quality Modeling Evaluation Studies	2-39

2-3       Summary of Sensitivity Studies of WINS Impactor Performance  	2-73

2-4       PM2.5 Samplers Currently Designated as FRMs for PM2 5 Mass
          Concentrations	2-74

2-5       Measurement Methods for Inorganic Elements	2-86

2-6       Methods for Continuous Measurement of PM Mass, Components, etc	2-89

3-1       Distribution of Ratios of PM2 5 to PM10 and Correlations Between PM2 5
          and PM10, PM2 5 and PM10_2 5, and PM10_2 5 and PM10 Found at Collocated
          Monitoring Sites in Seven Aerosol Characteristic Regions in 1999	3-33

3-2       Concentrations of PM25, PM10_25, and Selected Elements in the PM25 and
          PM10_2 5 Size Ranges with Standard Deviations and Correlations Between
          Elements and PM25 Mass in Philadelphia, PA  	3-35

3-3       Concentrations of PM25, PM10_25  and Selected Elements in the PM25 and
          PM10_2 5 Size Range with Standard Deviations and Correlations Between
          Elements and PM25 and PM10_25 Mass in Phoenix, AZ	3-36

3-4a-d    Measures of the Spatial Variability of PM25 Concentrations Within Selected
          Metropolitan Statistical Areas (MSAs)	3-42

3-5       Measures of the Spatial Variability of PM10_25 Concentrations Within Selected
          Metropolitan Statistical Areas (MSAs)	3-49
June 2003                                I-xiii       DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Tables
                                       (cont'd)

Number                                                                        Page

3-6       Correlation Coefficients for Spatial Variation of PM2 5 Mass and Different
          Sources for Pairs of Sampling Sites in the South Coast Air Basin (1986)  	3-57

3-7       Correlation Coefficients for Spatial Variation of PM2 5 Mass and Different
          Components for Pairs of Sampling Sites in Philadelphia (1994) 	3-57

3-8       Constituents of Atmospheric Particles and their Major Sources	3-60

3-9       Receptor Model Source Contributions to PM2 5	3-76

3-10      Receptor Model Source Contributions to PM10	3-77

3-11      Emissions of Primary PM25 by Various Sources in 1999	3-93

3-12      Emissions of Precursors to Secondary PM25 Formation by Various Sources
          in 1999	3-94

3-13      Nationwide Changes in Ambient Concentrations and Emissions of PM10 and
          Gaseous Precursors to Secondary Particulate Matter from 1992 to 2001	3-96

3A-1      Performance Metrics for PM25 from Collocated Samplers	  3A-31

3B-1      PM2 5 Speciation Samplers by Location:  Sites Selected for PM CD
          Summary 	3B-4

3B-2a     Burlington, VT Summary Data	3B-5

3B-2b     Philadelphia, PA Summary Data  	3B-6

3B-2c     Atlanta, GA Summary Data  	3B-7

3B-2d     Detroit, MI Summary Data  	3B-8

3B-2e     Chicago, IL Summary Data  	3B-9

3B-2f     ST. Louis, MI Summary Data	3B-10

3B-2g     Houston, TX Summary Data	3B-11

3B-2h     Minneapolis, MN Summary Data  	3B-12

3B-2i     Boulder, CO Summary Data 	3B-13

June 2003                               I-xiv        DRAFT-DO NOT QUOTE OR CITE

-------
                                  List of Tables
                                       (cont'd)

Number                                                                        Page

3B-2J     Phoenix, AZ Summary Data 	3B-14

3B-2k     Seattle, WA Summary Data  	3B-15

3B-21     Sacramento, CA Summary Data  	3B-16

3B-2m    Riverside-Rubidoux, CA Summary Data  	3B-17

3B-3      Blank Corrections for Elemental Carbon, Organic Carbon, and Total Carbon
          in the Speciation Network	3B-18

3C-1      Particulate Organic and Elemental Carbon Concentrations Based on Studies
          Published after 1995  	3C-2

3C-2      Particulate Organic Compound Concentrations Based on Studies Published
          after 1990 at Selected Sites 	3C-5

3D-1      Average Abundances of Major Elements in Soil and Crustal Rock	  3D-2

3D-2      Summary of Particle-Phase Organic Compounds Present in Fine Particle
          Road Dust Sample	  3D-5

3D-3      Composition of Fine Particles Released by Various Stationary Sources in
          the Philadelphia Area	  3D-6

3D-4a     Organic and Elemental Carbon Fractions of Diesel and Gasoline Engine
          Particulate Matter Exhaust	  3D-12

3D-4b     Contribution of Organic Carbon to Particulate Matter Carbon Emissions
          in Motor Vehicle Exhaust Collected from Vehicles Operated on Chassis
          Dynamometers	  3D-13

3D-5      Emission Rates for Constituents of Particulate Matter from Gasoline and
          Diesel Vehicles  	  3D-14

3D-6      Summary of Particle-Phase Organic Compounds Emitted from Motor
          Vehicles	  3D-16

3D-7      Mass Emissions, Organic Carbon, and Elemental Carbon Emissions from
          Residential Combustion of Wood 	  3D-19
June 2003                               I-xv        DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Tables
                                       (cont'd)

Number                                                                          Page

3D-8      Summary of Particle-Phase Organic Compounds Emitted from the
          Combustion of Wood in Fireplaces	  3D-20

3D-9      Mean Aerosol Composition at Tropical Site (Sriwijaya University, Sumatra,
          Indonesia) Affected Heavily by Biomass Burning Emissions	  3D-22

3E-1      The Range of Annual Mean PM25, PM10_25, and PM10 Concentrations at
          IMPROVE Particulate Matter Monitoring Sites	3E-3

3E-2      The Range of 24-Hour PM25, PM10_25, and PM10 Average Concentrations at
          the 90th Percentile Level at IMPROVE Particulate Matter Monitoring Sites .... 3E-4

4-1       Essential Ecological Attributes and Reporting Categories 	4-3

4-2       Types and Determinants of Particulate Deposition and Impact to Vegetation  ... 4-10

4-3       Relative Importance of Wet, Dry, Particulate, and Total Deposition to Three
          Forest Sites	4-12

4-4       Reported Mean Deposition Velocities for Sulfate, Chlorine, Nitrate, and
          Ammonium and Ion-Containing Particles	4-37

4-5       Representative Empirical Measurements of Deposition Velocity for
          Particulate Deposition  	4-38

4-6       Reported Mean Deposition Velocities for Potassium, Sodium, Calcium, and
          Magnesium Base-Cation-Containing Particles 	4-38

4-7       Mean Annual Nitrogen Deposition from Fine and Coarse Particles
          Compared to Total Nitrogen Deposition from all Sources to a Variety
          of Forest Ecosystems	4-46

4-8       Mean Annual Sulfate Deposition from Fine and Coarse Particles
          Compared to Total Sulfur Deposition from all Sources to a Variety of
          Forest Ecosystems	4-47

4-9       Mean Annual Base Cation Deposition from Fine and Coarse Particles
          Compared to Total Base Cation Deposition from all Sources to a Variety
          of Forest Ecosystems	4-48

4-10      Mean Particle Size, Deposition Rates, and Derived Deposition Velocities for
          Heavy Metal Deposition to the Upper Canopy of an Upland Oak Forest	4-52

June 2003                               I-xvi       DRAFT-DO NOT QUOTE OR CITE

-------
                                    List of Tables
                                        (cont'd)

Number                                                                           Page

4-11      Total Heavy Metal Deposition to Temperate Latitude Forests	4-52

4-12      Primary Goods and Services Provided by Ecosystems	4-58

4-13      Ecosystem Functions Impacted by Air Pollution Effects on Temperate Forest
          Ecosystems	4-84

4-14      Nitrogen-Saturated Forests in North America, Including Estimated N Inputs
          and Outputs 	4-102

4-15      Visibility Measurement Techniques  	4-172

4-16      Residential Visibility Contingent Valuation Study Results	4-190

4-17      Residential Visibility Valuation Study Results for Los Angeles and
          San Francisco	4-191

4-18      Corrosive Effects of Particulate Matter and Sulfur Dioxide on Metals	4-195

4-19      Corrosive Effects of Parti culate Matter and Sulfur Dioxide on Stone  	4-198

4-20      Examples of Impacts Resulting from Projected Changes in Extreme
          Climate Events	4-208

4-21      Effects of Reactive Nitrogen	4-231

5-1       Types of Parti culate Matter Used in Exposure and Concentration Variables	5-7

5-2       Activity Pattern Studies Included in the Consolidated Human Activity
          Database	5-15

5-3       Personal Exposure Models for Parti culate Matter	5-16

5-4       Summary of Recent PM Personal Exposure Studies	5-28

5-5       Summary of Recent Microenvironmental PM Measurement Studies	5-33

5-6       Papers Reporting Reanalyses of Parti culate Matter Exposure Studies 	5-41

5-7       Personal Monitoring Studies for Particulate Matter: Measured Concentrations
          and Correlation Coefficients  	5-46


June 2003                                I-xvii       DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Tables
                                       (cont'd)

Number                                                                          Page

5-8       Mean Concentration for PM Mass Reported for the Baltimore
          (Williams et al., 2000a,b,c) and the Fresno (Evans et al., 2000) Studies 	5-55

5-9       Daily-average Correlation Coefficients Reported for the Baltimore
          (Williams et al., 2000a,b,c) and the Fresno (Evans et al., 2000) Studies 	5-55

5-10      Regression Analysis Reported for Indoor/Outdoor Relationships for PM25
          in the Baltimore (Williams et al., 2000a,b,c) and Fresno (Evans et al., 2000)
          Studies  	5-56

5-11      Volume Mean Diameter and Maximum PM25 Concentrations of Indoor
          Particle  Sources	5-81

5-12      Correlations Between Personal PM2 5 and Ambient Pollutant Concentrations  . . . 5-88

5-13      Correlations Between Hourly Personal PM25 and Gaseous Pollutants	5-90

5-14      Studies That have Measured Particulate Matter Constituents in Personal
          Exposure Samples  	5-93

5-15      Studies That Have Measured Particulate Matter Constituents in
          Microenvironmental Samples 	5-94

5-16      Summary Statistics for Personal, Indoor, and Outdoor Concentrations of
          Selected Aerosol Components in Two Pennsylvania Communities  	5-96

5-17      Regression Analysis of Indoor Versus Outdoor Concentrations	5-98

5-18      Mixed Model Analysis of Personal Versus Outdoor Concentrations 	5-100

5-19      Regression Analysis of Indoor Versus Outdoor Concentrations	5-100

5-20      Regression Analysis of Indoor Versus Outdoor Concentrations	5-101
June 2003                               I-xviii       DRAFT-DO NOT QUOTE OR CITE

-------
                                    List of Figures

Number                                                                            Page

1-1       A general framework for integrating particulate-matter research	1-7

2-1       Distribution of coarse, accumulation, and nuclei mode particles by three
          characteristics: (a) number, (b) surface area, and (c) volume for the grand
          average continental size distribution  	2-8

2-2       Particle size distributions by number, (a) number concentrations are shown
          on a logarithmic scale to display the wide range by site and size;  (b) number
          concentrations for the average urban distribution are shown on a  linear scale ....  2-9

2-3       Size distribution by volume (a) for the averaged rural and urban-influenced
          rural number distributions shown in Figure 2-2a and a distribution from south
          central New Mexico, and (b) for the averaged urban and freeway-influenced
          urban number  distributions shown in Figure 2-2a	2-10

2-4       Volume size distribution, measured in traffic, showing fine and coarse
          particles and the nuclei and  accumulation modes  of fine particles	2-11

2-5       Submicron number size distribution observed in a boreal forest in Finland
          showing the tri-modal structure of fine particles	2-12

2-6       Specified particle penetration (size-cut curves) through an ideal
          (no-particle-loss) inlet for five different  size-selective sampling criteria	2-15

2-7       An idealized distribution of ambient particulate matter showing fine-mode
          particles and coarse-mode particles and the fractions collected by
          size-selective samplers	2-16

2-8       Comparison of penetration curves for two PM10 beta gauge samplers using
          cyclone inlets	2-18

2-9       Typical engine exhaust size distribution	2-23

2-10      Number size distributions showing measurement of a nucleation  burst made
          in a boreal forest in Finland	2-25

2-11      Examples of the measured one hour average particle number size distributions
          and the log normal fits to the modes of the data 	2-26

2-12(a-f)  Fitted multi-model particle size distribution at different sampling distances from
          Freeway 405 (a) 30 m downwind, (b) 60 m downwind, (c) 90 m downwind,
          (d) 150 m downwind, (e) 300 m downwind, (f) 300 m upwind 	2-27
June 2003                                 I-xix        DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Figures
                                       (cont'd)

Number                                                                           Page

2-12(g)   Ultrafine particle size distribution at different sampling locations near the
          405 Freeway in Los Angeles, CA	2-28

2-13      Particle growth curves showing fully reversible hygroscopic growth of sulfuric
          acid particles, deliquescent growth of ammonium sulfate particles at the
          deliquescent point, reversible hygroscopic growth of ammonium sulfate
          solution droplets at relative humidity relative humidity greater than 80%,
          and hysteresis until the crystallization point is reached  	2-34

2-14      Theoretical predictions and experimental measurements of growth of
          NH4HSO4 particles at relative humidity between 95 and 100%  	2-35

2-15      Schematic showing major nonvolatile and semivolatile components of PM25 . . . 2-42

2-16      Particulate matter concentrations in Spokane, WA, during the August 30,
          1996 dust storm	2-46

2-17      Aerosol water content expressed as a mass percentage, as a function relative
          humidity	2-55

2-18      Amount of ammonium nitrate volatilized from Teflon filters, expressed as a
          percentage of the measured PM25 mass, for the SCAQS and CalTech studies,
          for spring and fall sampling periods  	2-58

2-19      Average concentration of nonvolatile and semivolatile PM components in
          three cities 	2-65

2-20      Comparison of mass measurements with collocated RAMS, PC-BOSS,
          FRM PM2 5 sample, and a conventional TEOM monitor	2-67

2-21      Schematic diagram of the sample collection portion of the PM25 FRM
          sampler	2-70

2-22      Schematic view of the final design of the WINS	2-71

2-23      Evaluation of the final version of the WINS	2-72

2-24      Schematic diagram showing the principle of virtual impaction  	2-80

2-25      Size distribution of particles divided by chemical classification into organic,
          marine, and crustal	2-91
June 2003                                I-xx        DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Figures
                                       (cont'd)

Number                                                                           Page

2B-1      This thermogram, for a sample containing rock dust and diesel exhaust,
          shows three traces that correspond to temperature, filter transmittance,
          and FID detector response	2B-10

2B-2      Examples of thermograms obtained by (a) the IMPROVE protocol, and by
          (b) the NIOSH protocol	2B-12

3-la      1999-2001 county-wide average annual mean PM10 concentrations for
          counties with PM10 monitors	3-7

3-lb      1999-2001 highest county-wide 98th percentile 24-h average PM10
          concentrations for counties with PM10 monitors	3-7

3-2       Nationwide trend in ambient PM10 concentration from 1992 through 2001 	3-8

3-3       Trend in PM10 annual mean concentrations by EPA region, 1992
          through 2001   	3-9

3-4a      1999-2001 county-wide average annual mean PM25 concentrations for
          counties with PM2 5 monitors	3-10

3-4b      1999-2001 highest county-wide 98th percentile 24-h average PM2 5
          concentrations for counties with PM25 monitors	3-10

3-5       Collection of annual  distribution of 24-h average PM2 5 concentrations
          observed in U.S. and Canadian health studies conducted during the 1980s
          and early 1990s  	3-12

3-6a      1999-2000 estimated county-wide average annual  mean PM10_2 5
          concentrations for counties with collocated PM25 and PM10 monitors	3-14

3-6b      1999-2000 estimated county-wide highest 98th percentile 24-h PM10_2 5
          concentrations for counties with collocated PM25 and PM10 monitors	3-15

3-7a,b    Quarterly distribution of 24-h average PM25 concentrations for selected
          monitors in the (a) Philadelphia, PA and (b)  Cleveland, OH  	3-16

3-7c,d    Quarterly distribution of 24-h average PM25 concentrations for selected
          monitors in the (c) Dallas, TX and (d) Los Angeles, CA MS As	3-17

3-8       Seasonal concentrations of PM25 and PM10 measured in the four
          MAACS cities  	3-20

June 2003                                I-xxi        DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Figures
                                        (cont'd)

Number                                                                           Page

3-9a,b    Quarterly distribution of 24-h average PM10_25 concentrations for selected
          sites in the (a) Cleveland, OH; (b) Dallas, TX MS As	3-22

3-9c      Quarterly distribution of 24-h average PM10_25 concentrations for selected
          sites in the Los Angeles, CA MSA  	3-23

3-10      Frequency distribution of 24-h average PM25 concentrations measured at
          the Presbyterian home monitoring site in southwestern Philadelphia from
          1992 to 1995 	3-26

3-11      Concentrations of 24-h average PM25 measured at the EPA site in
          Phoenix, AZ from 1995 to 1997 	3-27

3-12      Frequency distribution of 24-h average PM25 concentrations measured at the
          EPA site in Phoenix, AZ from 1995 to 1997	3-27

3-13      Frequency distribution of 24-h average PM25 measurements obtained
          from all California Air Resources Board dichotomous sampler sites from
          1989 to 1998 	3-28

3-14      Frequency distribution of 24-h average PM10_25 concentrations obtained
          from all California Air Resource Board Dichotomous sampler sites from
          1989 to 1998 	3-28

3-15      Concentrations of 24-h average PM25 measured at the Riverside-Rubidoux
          site from 1989 to 1998	3-29

3-16      Frequency distribution of 24-h average PM25 concentrations measured at the
          Riverside-Rubidoux site from 1989 to 1994	3-29

3-17      Intraday variability of hourly average PM25 concentrations across the
          United States 	3-31

3-18      Occurrence of differences between pairs of sites in three metropolitan
          statistical areas	3-48

3-19      Intersite correlation coefficients forPM25, PM10, andPM10_25  	3-52

3-20      PM2 5 chemical components in downtown Los Angeles and Burbank (1986)
          have similar characteristics  	3-55
June 2003                                I-xxii       DRAFT-DO NOT QUOTE OR CITE

-------
                                 List of Figures
                                     (cont'd)

Number                                                                      Page

3-21     Concentrations of PM25 chemical components in Rubidoux and downtown
         Los Angeles (1986) 	3-56

3-22     Monthly average Saharan dust components of the aerosol sampled in
         Miami, FL, from 1974 to 1996  	3-84

3-23     PM2 5 and PM10 concentrations measured at Chilliwack Airport, located in
         northwestern Washington State, just before and during the Asian desert
         dust episode of April and May 1998 	3-85

3-24     Time series of 24-h average PM10 concentrations observed in the Rio Grande
         Valley during May 1998 	3-87

3-25     PM10  concentrations observed in St. Louis, MO, during May 1998  	3-88

3A-1     Philadelphia, PA-NJ MSA	  3A-4

3A-2     Washington, DC MSA	  3A-5

3 A-3     Norfolk, VA MSA	  3 A-6

3A-4     Columbia, SC MSA	  3A-7

3A-5     Atlanta, GA MSA  	  3A-8

3A-6     Birmingham, AL MSA 	  3A-9

3A-7     Tampa, FL MSA 	  3A-10

3A-8     Cleveland, OH MSA  	  3A-11

3A-9     Pittsburgh, PA MSA  	  3A-12

3A-10    Steubenville, OH-Weirton, WV MSA	  3A-13

3A-11    Detroit, MI MSA	  3A-14

3A-12    Grand Rapids, MI MSA	  3A-15

3A-13    Milwaukee, WI MSA	  3A-16

3A-14    Chicago, IL MSA	  3A-17

June 2003                              I-xxiii       DRAFT-DO NOT QUOTE OR CITE

-------
                                List of Figures
                                   (cont'd)

Number                                                                   Page

3A-15    Gary, IN MSA	 3A-18

3A-16    Louisville, KY MSA 	 3A-19

3A-17    St. Louis, MO MSA	 3A-20

3A-18    Baton Rouge, LA MSA	 3A-21

3A-19    Kansas City, KS-MO MSA	 3A-22

3A-20    Dallas, TX MSA  	 3A-23

3A-21    Boise, ID MSA	 3A-24

3A-22    Salt Lake City, UT MSA	 3A-25

3A-23    Seattle, WA MSA 	 3A-26

3A-24    Portland, ORMSA	 3A-27

3 A-25    Los Angeles-Long Beach, CA MSA	 3 A-28

3 A-26    Riverside-San Bernadino, CA MSA 	 3 A-29

3A-27    San Diego, CA MSA	 3A-30

3A-28    Columbia, SC MSA	 3A-32

3A-29    Tampa, FL MSA  	 3A-33

3 A-30    Cleveland, OH MSA 	 3 A-34

3A-31    Steubenville, OH  MSA 	 3A-35

3A-32    Detroit, MI MSA	 3A-36

3A-33    Milwaukee, WI MSA	 3A-37

3A-34    Chicago, IL MSA	 3A-38

3A-35    Gary, IN MSA	 3A-39


June 2003                            I-xxiv       DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Figures
                                       (cont'd)

Number                                                                          Page

3A-36    Louisville, KY MSA  	 3A-40

3A-37    St. Louis, MI MSA 	 3A-41

3A-38    Baton Rouge, LA MSA	 3A-42

3A-39    Dallas, TX MSA 	 3A-43

3A-40    Salt Lake City, UT MSA  	 3A-44

3A-41    Portland, ORMSA	 3A-45

3 A-42    Los Angeles, CA MSA  	 3 A-46

3A-43    Riverside, CA MSA	 3A-47

3A-44    San Diego, CA MSA	 3A-48

3D-1      Size distribution of particles generated in a laboratory resuspension
          chamber	  3D-2

3D-2      Size distribution of California source emissions, 1986	  3D-4

3E-1      Locations of the rural PM IMPROVE sites selected for the study  	3E-2

3E-2a,b   Interannual variability in 24-h average PM2 5 concentrations observed at
          selected IMPROVE sites: (a) Bridger Wilderness, WY and (b) Yellowstone
          National Park, WY	3E-6

3E-2c,d   Interannual variability in 24-h average PM2 5 concentrations observed at
          selected IMPROVE sites: (c) Dolly Sods/Otter Creek Wilderness, WV and
          (d) Brigantine National Wildlife Refuge, NJ	3E-7

3E-3a,b   Interannual variability in 24-h average PM10_25 concentrations observed at
          selected IMPROVE sites: (a) Bridger Wilderness, WY and (b) Yellowstone
          National Park, WY; 	3E-8

3E-3c,d   Interannual variability in 24-h average PM10_25 concentrations observed at
          selected IMPROVE sites: (c) Dolly Sods/Otter Creek Wilderness, WV
          and (d) Brigantine National Wildlife Refuge, NJ  	3E-9
June 2003                               I-xxv       DRAFT-DO NOT QUOTE OR CITE

-------
                                    List of Figures
                                         (cont'd)

Number                                                                             Page

3E-4a,b   Seasonal variability in 24-h average PM2 5 concentrations observed at
          selected IMPROVE sites:  (a) Bridger Wilderness, WY and (b) Yellowstone
          National Park, WY	3E-10

3E-4c,d   Seasonal variability in 24-h average PM2 5 concentrations observed at selected
          IMPROVE sites: (c) Dolly Sods/Otter Creek Wilderness, WV and
          (d) Brigantine National Wildlife Refuge, NJ	3E-11

3E-5a,b   Seasonal variability in 24-h average PM10_2 5 concentrations observed at
          selected IMPROVE sites:  (a) Bridger Wilderness, WY and (b) Yellowstone
          National Park, WY	3E-12

3E-5c,d   Seasonal variability in 24-h average PM10-2.5 concentrations observed
          at selected IMPROVE sites: (c) Dolly Sods/Otter Creek Wilderness, WV
          and (d) Brigantine National Wildlife Refuge, NJ  	3E-13

4-1       The diversity of fine PM from sites in the western and eastern U.S	4-7

4-2       Relative importance of three modes of deposition of nitrate (A) and sulfate
          (B) at high elevation sites  	4-11

4-3       A simplified resistance catena representing the factors controlling deposition
          of particles to the surface	4-16

4-4       The relationship between deposition velocity of selected paniculate
          materials and the distribution of the material between the coarse- and
          fine-aerosol fractions	4-17

4-5       The relationship between particle diameter and deposition velocity for
          particles  	4-18

4-6       Vertical stratification  of diverse, chemically speciated particles in a mixed
          oak forest	4-33

4-7       The relationship between particle size and concentration below a spruce
          canopy with wind velocity at a height of 16.8 m equaling 5ms"1   	4-34

4-8       Range of percent of total deposition delivered in the dry particulate
          fraction across the sites of the Integrated Forest Study	4-40

4-9       Contribution of parti culate deposition to total deposition of nitrogen, sulfur,
          and base cations	4-40

June 2003                                I-xxvi       DRAFT-DO NOT QUOTE OR CITE

-------
                                    List of Figures
                                         (cont'd)

Number                                                                             Page

4-10      Mean (±SE) percent of total nitrogen, sulfur, or base cation deposition
          contributed by fine plus coarse particles	4-49

4-11      Annual total deposition of heavy metals to Hubbard Brook Experimental
          Forest, NH	4-54

4-12      Sample stressors and the essential ecological attributes they affect	4-56

4-13      Linkages among various ecosystem goods and services (food, water,
          biodiversity, forest products) and other driving forces (climate change) 	4-61

4-14      Effects of environmental stress on forest trees presented on a hierarchial
          scale for the leaf, branch, tree, and stand levels of organization	4-80

4-15      Illustration of the nitrogen cascade showing the movement of the human-
          produced reactive nitrogen as it cycles through the various environmental
          reservoirs in the atmosphere, terrestrial ecosystems, and aquatic ecosystems  . . . 4-97

4-16      Nitrogen cycle  	4-100

4-17      Diagrammatic overview of excess nitrogen in North America	4-104

4-18      Schematic of sources and sinks of hydrogen ions in a forest  	4-118

4-19      Key elements of proposed framework for determining critical loads for
          nitrogen and sulfur in the United States	4-125

4-20      Calcium deposition in > 2-|im particles, < 2-|im  particles, and wet forms
          and leaching in the Integrated Forest Study sites  	4-129

4-21      Magnesium deposition in > 2-|im particles, < 2-|im particles, and wet forms
          and leaching in the Integrated Forest Study sites  	4-130

4-22      Potassium deposition in > 2-|im particles, < 2-|im particles, and wet forms
          and leaching in the Integrated Forest Study sites  	4-131

4-23      Base cation deposition in > 2-|im particles, < 2-|im particles, and wet forms
          and leaching in the Integrated Forest Study sites  	4-132

4-24      Total cation leaching balanced by sulfate and nitrate  estimated from
          particulate deposition and by other sources of sulfate and nitrate and
          by other anions in the Integrated Forest Study sites 	4-133

June 2003                                I-xxvii        DRAFT-DO NOT QUOTE OR CITE

-------
                                    List of Figures
                                        (cont'd)

Number                                                                            Page

4-25      Soil exchangeable Ca+2 pools and net annual export of Ca+2 in the
          Integrated Forest Study sites 	4-134

4-26      Soil exchangeable Mg+2 pools and net annual export of Mg+2 in the
          Integrated Forest Study sites 	4-134

4-27      Soil exchangeable K+2 pools and net annual export of K+2 in the Integrated
          Forest Study sites	4-135

4-28a     Simulated soil solution mineral acid anions and base cations in the red
          spruce site with no change, 50% N and S deposition, and 50% base
          cation deposition  	4-138

4-28b     Simulated soil solution mineral acid anions and base cations in the red
          spruce site with no change, 50% N and S deposition, and 50% base
          cation deposition  	4-139

4-29a     Simulated soil solution Al and soil  base saturation in the red spruce site
          with no change, 50% N and S deposition, and 50% base cation deposition  .... 4-140

4-29b     Simulated soil solution Al and soil  base saturation in the red spruce site
          with no change, 50% N and S deposition, and 50% base cation deposition  .... 4-141

4-3Oa     Simulated soil solution mineral acid anions and base cations in the
          Coweeta site with no change, 50% N and S deposition, and 50% base
          cation deposition  	4-143

4-3Ob     Simulated soil solution mineral acid anions and base cations in the
          Coweeta site with no change, 50% N and S deposition, and 50% base
          cation deposition  	4-144

4-31      Relationship of plant nutrients and trace metals with vegetation  	4-148

4-32      Light reflected from a target toward an observer  	4-157

4-33      Light-scattering efficiency factor for a homogeneous sphere with an index
          of refraction of 1.50 as a function of the size parameter	4-161

4-34      Volume-specific light-scattering efficiency as a function of particle
          diameter	4-162
June 2003                               I-xxviii       DRAFT-DO NOT QUOTE OR CITE

-------
                                    List of Figures
                                        (cont'd)

Number                                                                             Page

4-35      Particle growth curve as a function of relative humidity showing deliquescent
          growth of ammonium sulfate particles at the deliquescent point, reversible
          hygroscopic growth of ammonium sulfate solution droplets at relative
          humidity greater than 80%, and hysteresis until the crystallization point
          is reached	4-164

4-36      Comparison of extinction and visual range  	4-168

4-37      Proportionality of observed daytime haziness to fine particle mass
          concentration in Los Angeles   	4-169

4-38      Relative humidity adjustment factor for ammonium sulfate as a function of
          relative humidity 	4-171

4-39a     Aggregate visibility trends for 10 eastern Class 1 areas 	4-181

4-39b     Aggregate visibility trends for 26 western Class 1 areas	4-181

4-40      (a) Eastern class I area aggregate trends in aerosol light extinction on the
          20% haziest days, including trends by major aerosol component; (b) Western
          class I area aggregate trends in aerosol light extinction on the 20% haziest
          days, including trends by major aerosol component	4-183

4-4la     Light extinction trends in Tucson, Arizona from  1993  to 2002  	4-185

4-4Ib     Light extinction trends in Phoenix, Arizona from 1994 to 2001	4-186

4-42      Estimated global mean radiative forcing exerted by gas and various particle
          phase species for the year 2000, relative to 1750  	4-213

5-1       Hypothetical exposure time profile: Pollutant exposure as a function of time
          showing how the averaged exposure, integrated exposure, and peak exposure
          relate to the instantaneous exposure  	5-9

5-2       Distribution of FINF (a) and cc (b) for daytime as estimated from PTEAM
          study data	5-23

5-3       Comparison of correlation coefficients for longitudinal analyses of personal
          exposure versus ambient concentrations for individual subjects for PM25
          and sulfate  	5-58
June 2003                                I-xxix       DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Figures
                                        (cont'd)

Number                                                                           Page

5-4       Personal exposure versus ambient concentrations for PM25 and sulfate	5-61

5-5       Regression analyses of aspects of daytime personal exposure to PM10
          estimated using data from the PTEAM study 	5-62

5-6       Air exchange rates measured in homes throughout the United States	5-69

5-7       Box plots of hourly air exchange rates stratified by season in Boston, MA,
          during 1998  	5-70

5-8       Geometric mean infiltration factor (indoor/outdoor ratio) for hourly nighttime,
          nonsource data for two seasons	5-71

5-9       Regression of air exchange rate on absolute indoor-outdoor temperature
          difference	5-73

5-10      Comparison of deposition rates from Long et al., 2001a with literature
          values	5-75

5-11      Penetration efficiencies and deposition rates from models of nightly
          average data	5-76

5-12      Mean hourly indoor/outdoor particle concentration ratio from an unoccupied
          residence in Fresno, CA, during spring 1999 	5-82

5-13      Personal versus outdoor SO4"2 in State College, PA  	5-97

5-14      Plots of nonambient exposure to PM10, (a) daily individual daytime values
          from PTEAM data and (b) daily-average values from THEES data	5-110
June 2003                                I-xxx       DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                              EXECUTIVE SUMMARY
Principal Authors

Dr. William E. Wilson—National Center for Environmental Assessment (B243-01),
U. S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Lester D. Grant—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Ms. Beverly Comfort—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. J.H.B. Garner—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Dennis J. Kotchmar—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Joseph P. Pinto—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711
                           CHAPTER 1.  INTRODUCTION
Principal Authors

Dr. Lester D. Grant—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Dennis J. Kotchmar—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711
June 2003                               I-xxxi       DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                      (cont'd)
            CHAPTER 2. PHYSICS, CHEMISTRY, AND MEASUREMENT
                            OF PARTICULA TE MA TTER
Principal Authors

Dr. William Wilson—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Candis S. Claiborn—Washington State University, Laboratory for Atmospheric Research,
Department of Civil and Environmental Engineering, P.O. Box 642910, Pullman, WA 99164

Dr. Brooke L. Hemming—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Contributing Authors

Dr. Tom Cahill, University of California, Davis, One Shields Ave., Davis, CA 95616

Dr. Judith C. Chow, Desert Research Institute, 2215 Raggio Parkway, Reno, NV 89512

Max R. Peterson, Research Triangle Institute, P.O. Box 12194, Research Triangle Park, NC
27709

Dr. James J. Schauer, University of Wisconsin, College of Engineering, Department of Civil and
Environmental Engineering, Madison, WI 53706

Dr. Barbara J. Turpin, The State University of New Jersey, Rutgers, Department of
Environmental Sciences and Rutgers Cooperative Extension, New Brunswick, NJ 08901-8551

Dr. John G. Watson, Desert Research Institute, 2215 Raggio Parkway, Reno, NV  89512

Contributors and Reviewers

Dr. Edward O. Edney—National Exposure Research Laboratory (E205-02)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. R.R. Eldred, Crocker Nuclear Laboratory, University of California, Davis, Davis, CA 95616

Mr. William Ewald—National Center for Environmental  Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
June 2003                               I-xxxii       DRAFT-DO NOT QUOTE OR CITE

-------
                      Authors, Contributors, and Reviewers
                                       (cont'd)
Contributors and Reviewers
(cont'd)

James B. Flanagan—Research Triangle Institute, P.O. Box 12194, Research Triangle Park, NC
27709

Dr. Judith Graham—American Chemistry Council, 1300 Wilson Boulevard, Arlington, VA
22207

Mr. Jim Homolya—Office of Air Quality Planning and Standards (C339-02),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

R.K.M. Jayanty—Research Triangle Institute, P.O. Box 12194, Research Triangle Park, NC
27709

Mr. Scott Mathias—Office of Air Quality Planning and Standards (C539-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Frank McElroy—National Exposure Research Laboratory (MD-46),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Joseph Pinto—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Joann Rice—Office of Air Quality Planning and Standards (C339-02),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27709

Ed E. Rickman—Research Triangle Institute, P.O. Box 12194, Research Triangle Park, NC
27709

Dr. Bret Schichtel—Colorado State University, Cooperative Institute for Research in the
Atmosphere, Foothills Campus, Fort Collins, CO 80523-1375

Dr. Russell Wiener—National Exposure Research Laboratory (D205-03)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
June 2003                              I-xxxiii      DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                      (cont'd)
                 CHAPTER 3. CONCENTRATIONS, SOURCES, AND
              EMISSIONS OF A TMOSPHERIC PARTICVLA TE MA TTER
Principal Author

Dr. Allen S. Lefohn—A.S.L. & Associates, 111 North Last Chance Gulch, Helena, MT 59601

Dr. Joseph P. Pinto—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Contributing Authors

Dr. Barbara J. Turpin—The State University of New Jersey, Rutgers, Department of
Environmental Sciences and Rutgers Cooperative Extension, New Brunswick, NJ 08901-8551

Dr. James J. Schauer—University of Wisconsin, College of Engineering, Department of Civil
and Environmental Engineering, Madison, WI 53706

Contributors and Reviewers

Dr. JoEllen Bandemeyer—Research Triangle Institute, PO Box 12194, Research Triangle Park,
NC 27709

Dr. Lyle Chinkin—Sonoma Technology,  1360 Redwood Way, Suite C, Petaluma, CA 94549

Dr. Candis S. Claiborn—Washington State University, Laboratory for Atmospheric Research,
Department of Civil and Environmental Engineering, P.O. Box 642910, Pullman, WA 99164

Dr. Steven Colome—Integrated Environmental Services, 5319 University Drive, #430, Irvine,
CA 92612

Mr. Tom Coulter—National Exposure Research Laboratory (D234-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Edward O. Edney—National Exposure Research Laboratory (E205-02),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Terence Fitz-Simons—Office of Air Quality Planning and Standards (C3 04-01),
U. S. Environmental Protection Agency, Research Triangle Park, NC 27711
June 2003                              I-xxxiv       DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                      (cont'd)
Contributors and Reviewers
(cont'd)

Dr. Sheldon Friedlander—University of California at Los Angeles, Department of Chemical
Engineering, 5531 Boelter Hall, Los Angeles, CA 90095

Dr. Lynn Hildemann—Civil and Environmental Engineering Department, Stanford University,
Stanford, CA 94305

Mr. Phil Lorang—Office of Air Quality Planning and Standards (D205-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Ms. Karen Magliano—California Air Resources Board, 2020 L Street, Sacramento, CA 95814

Mr. Scott Mathias—Office of Air Quality Planning and Standards (C539-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Ms. Nehzat Mottallebi—California Air Resources Board, 2020 L Street, Sacramento, CA 95814

Mr. Tom Pace—Office of Air Quality Planning and Standards (D205-01), U.S. Environmental
Protection  Agency, Research Triangle Park, NC 27711

Dr. Richard Poirot—VT Air Program, Building 3 South, 103 South Main Street, Waterbury, VT
05671

Mr. Win Setiawan—California Air Resources Board, 2020 L Street, Sacramento, CA 95814

Mr. Dane Westerdahl—California Air Resources Board, 2020 L Street, Sacramento, CA 95814
       CHAPTER 4. ENVIRONMENTAL EFFECTS OF PARTICULA TE MA TTER
Principal Authors

Ms. Beverly Comfort—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. William Ewald—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
June 2003                              I-xxxv      DRAFT-DO NOT QUOTE OR CITE

-------
                      Authors, Contributors, and Reviewers
                                       (cont'd)
Principal Authors
(cont'd)

Dr. J.H.B. Garner—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Lester D. Grant—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. David A. Grantz—University of California/Riverside, Kearney Agricultural Center,
9240 South Riverbend Avenue, Parlier, CA 93648

Dr. Brooke L. Hemming—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Dale W. Johnson—Environmental and Resource Science, 1000 Valley Road, University of
Nevada, Reno, NV 89512

Dr. William Malm—National Park Service, Air Resources Division, CIRA, Colorado State
University,  Fort Collins, CO

Dr. Joseph P. Pinto—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Contributing Authors

Dr. Paul J. Hanson—Environmental Sciences Division, Oak Ridge National Laboratory,
P.O. Box 2008, Bethel Valley Road, Building 1059, Oak Ridge, TN 37831-6422

Dr. William H. Smith—Professor Emeritus, Yale University School of Forestry and
Environmental Studies, 370 Prospect Street, New Haven, CT 06511

Contributors and Reviewers

Dr. Larry T. Cupitt—National Exposure Research Laboratory (D305-01),
U. S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Rich Damberg—Office of Air Quality Planning and Standards (C539-02),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Russell  R. Dickerson—University of Maryland, Department of Meteorology,
Stadium Drive, College Park, MD 20742
June 2003                              I-xxxvi       DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                      (cont'd)
Contributors and Reviewers
(cont'd)

Dr. Anne Grambsch—National Center for Environmental Assessment (860ID),
U. S. Environmental Protection Agency, Washington, DC 20036

Dr. Sagar V. Krupa—University of Minnesota, Department of Plant Pathology,
495 BorlaugHall, 1991 Upper Buford Circle, St. Paul, MN 55108

Dr. Alan J. Krupnick—Quality of the Environment Division, Resources for the Future,
1616 P Street, NW, Washington, DC 20036

Mr. Paul T. Roberts—Sonoma Technology, Inc.,1360 Redwood Way - Suite C,
Petaluma, CA 94954

Mr. John Spence-Cary, NC 27511

Ms. Debra Meyer Wefering—Duckterather Weg 61, Bergisch Gladbach, Germany 54169

Ms. Kay Whitfield—Office of Air Quality Planning and Standards (C243-02),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Richard Zepp—National Exposure Research Laboratory (IOD),
U. S. Environmental Protection Agency, Athens, GA
           CHAPTER 5.  HUMAN EXPOSURE TO PARTICULA TE MA TTER
                             AND ITS CONSTITUENTS
Principal Authors

Dr. Linda S. Sheldon—National Exposure Research Laboratory (E205-02),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Haluk Ozkaynak—National Exposure Research Laboratory (E205-02),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. William E. Wilson—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
June 2003                              I-xxxvii       DRAFT-DO NOT QUOTE OR CITE

-------
                      Authors, Contributors, and Reviewers
                                       (cont'd)
Contributing Authors
Dr. Janet Burke—National Exposure Research Laboratory (E205-02),
U. S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Roy Fortmann—National Exposure Research Laboratory (E205-02),
U. S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. David T. Mage—Institute for Survey Research, Temple University,
Philadelphia, PA 19122-6099

Mr. Thomas McCurdy—National Exposure Research Laboratory (E205-02),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Gary Norris—National Exposure Research Laboratory (E205-03),
U. S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Anne Rea—National Exposure Research Laboratory (E205-04),
U. S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Alan Vette—National Exposure Research Laboratory  (E205-04),
U. S. Environmental Protection Agency, Research Triangle Park, NC 27711

Contributors and Reviewers

Mr. John Langstaff—Office of Air Quality Planning and Standards (C539-01)
U.S. Environmental Protection Agency, Research Triangle park, NC 27711

Dr. Joseph Pinto—National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Harvey Richmond—Office of Air Quality Planning and Standards (C539-01),
U. S. Environmental Protection Agency, Research Triangle Park, NC 27711
June 2003                             I-xxxviii      DRAFT-DO NOT QUOTE OR CITE

-------
             U.S. ENVIRONMENTAL PROTECTION AGENCY
  PROJECT TEAM FOR DEVELOPMENT OF AIR QUALITY CRITERIA
                       FOR PARTICULATE MATTER
Executive Direction

Dr. Lester D. Grant—Director, National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Scientific Staff

Dr. Robert W. Elias—PM Team Leader, Health Scientist, National Center for Environmental
Assessment (B243-01), U.S. Environmental Protection Agency, Research Triangle Park, NC
27711

Dr. William E. Wilson—Air Quality Coordinator, Physical Scientist, National Center for
Environmental Assessment (B243-01), U.S. Environmental Protection Agency, Research
Triangle Park, NC  27711

Ms. Beverly Comfort—Health Scientist, National Center for Environmental Assessment
(B243-01), U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. J.H.B. Garner—Ecological Scientist, National Center for Environmental Assessment
(B243-01), U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Dennis J. Kotchmar—Medical Officer, National Center for Environmental Assessment
(B243-01), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Joseph P. Pinto—Physical Scientist, National Center for Environmental Assessment
(B243-01), U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Technical Support Staff

Ms. Nancy Broom—Information Technology Manager, National Center for Environmental
Assessment (B243-01), U.S. Environmental Protection Agency, Research Triangle Park, NC
27711

Mr. Douglas B. Fennell—Technical Information Specialist, National Center for Environmental
Assessment (B243-01), U.S. Environmental Protection Agency, Research Triangle Park, NC
27711

Ms. Emily R. Lee—Management Analyst, National Center for Environmental Assessment
(B243-01), U.S. Environmental Protection Agency, Research Triangle Park, NC  27711
June 2003                              I-xxxix       DRAFT-DO NOT QUOTE OR CITE

-------
             U.S. ENVIRONMENTAL PROTECTION AGENCY
  PROJECT TEAM FOR DEVELOPMENT OF AIR QUALITY CRITERIA
                       FOR PARTICULATE MATTER
                                      (cont'd)
Technical Support Staff
(cont'd)

Ms. Diane H. Ray—Program Specialist, National Center for Environmental Assessment
(B243-01), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Ms. Donna Wicker—Administrative Officer, National Center for Environmental Assessment
(B243-01), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Richard Wilson—Clerk, National Center for Environmental Assessment (B243-01),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Document Production Staff

Dr. Carol A. Seagle—Technical Editor, Computer Sciences Corporation, 2803 Slater Road, Suite
220, Morrisville, NC  27560

Ms. Jessica Long—Graphic Artist, Computer Sciences Corporation, 2803 Slater Road, Suite
220, Morrisville, NC  27560

Ms. Carolyn T. Perry—Word Processor, Computer Sciences Corporation, 2803 Slater Road,
Suite 220, Morrisville, NC 27560

Ms. Kelly Quifiones—Word Processor, InfoPro, Inc., 8200 Greensboro Drive, Suite 1450,
McLean, VA 22102

Mr. Charles E. Gaul—Graphic Artist, Computer Sciences Corporation, 2803 Slater Road, Suite
220, Morrisville, NC  27560

Mr. John M. Havel—Graphic Artist, Computer Sciences Corporation, 2803 Slater Road, Suite
220, Morrisville, NC  27560

Ms. Laura A. Rhodes—Graphic Artist, Computer Sciences Corporation, 2803 Slater Road, Suite
220, Morrisville, NC  27560

Mr. Keith A. Tarpley—Graphic Artist, Computer Sciences Corporation, 2803 Slater Road, Suite
220, Morrisville, NC  27560
June 2003                                I-xl        DRAFT-DO NOT QUOTE OR CITE

-------
            U.S. ENVIRONMENTAL PROTECTION AGENCY
 PROJECT TEAM FOR DEVELOPMENT OF AIR QUALITY CRITERIA
                      FOR PARTICULATE MATTER
                                   (cont'd)


Technical Reference Staff

Mr. John A. Bennett—Technical Information Specialist, SANAD Support Technologies, Inc.,
11820 Parklawn Drive, Suite 400, Rockville, MD 20852

Ms. Lisa Greenbaum—Records Management Technician, Reference Retrieval and Database
Entry Clerk, InfoPro, Inc., 8200 Greensboro Drive, Suite 1450, McLean, VA 22102

Ms. Sandra L. Hughey—Technical Information Specialist, SANAD Support Technologies, Inc.,
11820 Parklawn Drive, Suite 400, Rockville, MD 20852
June 2003                             I-xli       DRAFT-DO NOT QUOTE OR CITE

-------
                     Abbreviations  and Acronyms
 AAS



 AC



 ACGffl



 ACS



 ADS



 AES



 AIRS



 ANC



 AQCD



 AQI



 ARIES



 ASOS



 ATOFMS



 AWOS



 BaP



 BASE



 BC



 BNF



 BOSS



 BYU



 CAA



 CAAM



 CAMNET



 CARB



 CASAC



 CASTNet
atomic absorption spectrophotometry



air conditioning



American Conference of Governmental Industrial Hygienists



American Cancer Society



annular denuder system



atomic emission spectroscopy



Aerometric Information Retrieval System



acid neutralizing capacity



Air Quality Criteria Document



Air Quality Index



Aerosol Research and Inhalation Epidemiology Study



Automated Surface Observing System



time-of-flight mass spectrometer



Automated Weather Observing System



benzo(a)pyrene



Building Assessment and Survey Evaluation



black carbon



bacterial nitrogen fertilization



Brigham Young University Organic Sampling System



Bringham Young University



Clean Air Act



continuous ambient mass monitor



Coordinated Air Monitoring Network



California Air Resources Board



Clean Air Scientific Advisory Committee



Clean Air Status  and Trends Network
June 2003
                   I-xlii
DRAFT-DO NOT QUOTE OR CITE

-------
 CB
 cc
 CCPM
 CCSEM
 CEN
 CFCs
 CFR
 CHAD
 GIF
 CUT
 CMAQ
 CMB
 CMSA
 CO CD
 COD
 COPD
 CPC
 CRP
 CSIRO
 CSMCS
 CSS
 CTM
 CV
 CVM
 Da
 DAQ
 DCFH
 DMPS
base cations
carbonate carbon
continuous coarse particle monitor
computer-controlled scanning electron microscopy
European Standardization Committee
chlorofluorocarbons
Code of Federal Regulations
Consolidated Human Activity Database
charcoal-impregnated cellulose fiber
Chemical Industry Institute of Technology
Community Multi-Scale Air Quality
chemical mass balance
Consolidated Metropolitan Statistical Area
Air Quality Criteria Document for Carbon Monoxide
coefficient of divergence
chronic obstructive pulmonary disease
condensation particle counter
Coordinated Research Program
Commonwealth Scientific and Industrial Research Organisation
Carbonaceous Species Methods Comparison Study
coastal sage scrub
chemistry-transport model
coeffi ci ent of vari ati on
contigent valuation method
aerodynamic diameter
Department of Air Quality
di chl orofluore scin
differential mobility particle sizer
June 2003
                   I-xliii
DRAFT-DO NOT QUOTE OR CITE

-------
 DMS

 DP
 dv

 BAD

 EC

 ECAO

 EDXRF

 EEA

 ENSO

 EPA

 ESP

 ETS

 EXPOLIS


 FID

 FRM

 GAM

 GC

 GCMs

 GCVTC

 GC/MSD

 GHG

 GSD

 HBEF

 HDS

 HEADS

 HEI

 HI

 hivol
dimethyl sulfide

particle diameter

deciview index

electrical aerosol detector

elemental carbon

Environmental Criteria and Assessment Office

energy dispersive X-ray fluorescence

Essential Ecological Attributes

El Nino-Southern Oscillation

Environmental Protection Agency

electrostatic precipitator

environmental tobacco smoke

Air Pollution Exposure Distribution within Adult Urban Populations in
Europe

flame ionization detection

Federal Reference Method

general additive models

gas chromatography

General Circulation Models

Grand Canyon Visibility Transport Commission

gas chromatography/mass-selective detection

greenhouse gases

geometric standard deviation

Hubbard Brook Experiment Forest

honeycomb denuder/filter pack sampler

Harvard-EPA Annular Denuder Sampler

Health Effects Institute

Harvard Impactors

High volume sampler
June 2003
                   I-xliv
DRAFT-DO NOT QUOTE OR CITE

-------
 HTGC-MS



 HVAC



 1C



 ICP



 IPS



 IMPROVE



 INAA



 IOVPS



 IPCC



 IPM



 IPN



 ISO



 K



 LAI



 LOD



 LWC



 LWCA



 MAA



 MAACS



 MADPro



 MAQSIP



 mCa



 MDL



 MOUDI



 MS



 MSA



 MSA



 mv
high temperature gas chromotography-mass spectrometry



heating, ventilation, or air conditioning



ion chromatography



inductively coupled plasma



Integrated Forest Study



Interagency Monitoring of Protected Visual Environments



instrumental neutron activation analysis



integrated organic vapor/particle sampler



Intergovernmental Panel on Climate Change



inhalable paniculate matter



Inhalable Particulate Network



International Standards Organization



Koschmieder constant



leaf area indices



level of detection



liquid water content



liquid water content analyzer



mineral acid anion



Metropolitan Acid Aerosol Characterization Study



Mountain Acid Deposition Program



Multiscale Air Quality Simulation Platform



membrane-associated calcium



minimum detection level



micro-orifice uniform deposit impactor



mass spectroscopy



methane sulfonic acid



metropolitan statistical area



motor vehicle
June 2003
                    I-xlv
DRAFT-DO NOT QUOTE OR CITE

-------
 NAAQS



 NAMS



 NAPAP



 NARSTO



 NAST



 NCEA



 NDDN



 NERL



 NESCAUM



 NFRAQS



 NGOs



 NHAPS



 NIOSH



 NIR



 NIST



 NOPL



 NOX



 NPP



 Nr



 NRC



 NuCM



 OAQPS



 OAR



 OC



 ORD



 PAH



 PAN



 PAR
National Ambient Air Quality Standards



National Ambient Monitoring Stations



National Acid Precipitation Assessment Program



North American Research Strategy for Tropospheric Ozone



National Assessment Synthesis Team



National Center for Environmental Assessment



National Dry Deposition Network



National Exposure Research Laboratory



Northeast States for Coordinated Air Use Management



North Frontal Range Air Quality Study



non-governmental organizations



National Human Activity Pattern  Survey



National Institute for Occupational Safety and Health



near infared radiation



National Institute of Standards and Technology



naso-oro-pharyngo-laryngeal



nitrogen oxides



net primary production



reactive nitrogen



National Research Council



nutrient cycling model



Office of Air Quality Planning and Standards



Office of Air and Radiation



organic carbon



Office of Research and Development



polynuclear aromatic hydrocarbon



peroxyacetyl nitrate



photosynthetically active radiation
June 2003
                   I-xlvi
DRAFT-DO NOT QUOTE OR CITE

-------
 PEL

 PBP

 PBY

 PC

 PC

 PC-BOSS


 PCA

 PCBs

 PCDD

 PCDF

 PCM

 PEM

 PESA

 PIXE

 PM

 PM10.25

 PM25

 PMF

 POP

 PRB

 PTEAMS

 PTEP

 PUF

 PWC

 RAMS

 RAMS

 RAS

 RCS
planetary boundary layer

primary biological particles

Presbyterian Home

particle concentrator

pyrolitic carbon

Particulate Concentrator-Brigham Young University Organic Sampling
System

principal  component analysis

polychloronated biphenyls

poly chlorinated dibenzo-p-dioxins

polychlorinated dibenzofurans

particle composition monitor

personal exposure monitor

proton elastic scattering analysis

proton induced X-ray emission

particulate matter

coarse particulate matter

fine particulate matter

positive matrix factorization

persistent organic pollutant

policy-relevant background

Particle Total Exposure Assessment Methodology

PM10 Technical Enhancement Program

polyurethane foam

precipitation-weighted concentrations

Regional  Air Monitoring Study

Real-Time Air Monitoring System

roll-around system

Random Component Superposition
June 2003
                   I-xlvii
DRAFT-DO NOT QUOTE OR CITE

-------
 RFC



 RH



 RIVM



 RPM



 RPM



 RRMS



 RTF



 RUE



 SA



 SAB



 SCAQS



 SCOS



 sd



 sec



 SEM



 SES



 SEV



 SIP



 SLAMS



 SMPS



 SMSAs



 SOC



 SoCAB



 SOPM



 SP



 SRM



 STN



 SUVB
residual fuels oils



relative humidity



Directorate-General for Environmental Protection



Regional Particulate Model



respirable particulate matter



relatively remote monitoring sites



Research Triangle Park



radiation use efficiency



Sierra Anderson



Science Advisory Board



Southern California Air Quality Study



Southern California Ozone Study



standard deviation



secondary



scanning electron microscopy



sample equilibration system



Sensor Equivalent Visibility



State Implementation Plans



State and Local Air Monitoring Stations



scanning mobility particle sizer



Standard Metropolitan Statistical Areas



semivolatile organic compounds



South Coast Air Basin



secondary organic particulate matter



Staff Paper



standard reference method



Speciation Trends Network



solar ultraviolet B radiation
June 2003
                   I-xlviii
DRAFT-DO NOT QUOTE OR CITE

-------
 svoc



 SVM



 TAR



 TDMA



 TEO



 TEOM



 THEES



 TNF



 TOFMS



 TOR



 TOT



 TPM



 TRXRF



 TSP



 TVOC



 UAM-V



 UCM



 UNEP



 URG



 USGCRP



 VAPS



 VMD



 VOC



 WMO



 VR



 WINS



 WMO



 WRAC
semivolatile organic compounds



semivolatile material



Third Assessment Report



Tandem Differential Mobility Analyzer



trace element oxides



tapered element oscillating microbalance



Total Human Environmental Exposure Study



tumor necrosis factor



aerosol time-of-flight mass spectroscopy



thermal/optical reflectance



thermal/optical transmission



thoracic particulate matter



total reflection X-ray fluorescence



total suspended particulates



total volatile organic compounds



Urban Airshed Model Version V



unresolved complex mixture



United Nations Environment Programme



University Research Glassware



U.S. Global Change Research Program



Versatile Air Pollution Samplers



volume mean diameter



volatile organic compounds



World Meteorological Organization



visual range



Well Impactor Ninety-Six



World Meteorological Organization



Wide Range Aerosol  Classifier
June 2003
                   I-xlix
DRAFT-DO NOT QUOTE OR CITE

-------
 WTA              willingness to accept



 WTP               willingness to pay



 XAD               polystyrene-divinyl benzene



 XRF               X-ray fluorescence
June 2003                                1-1         DRAFT-DO NOT QUOTE OR CITE

-------
 i                              EXECUTIVE SUMMARY
 2
 3
 4     E.I   INTRODUCTION
 5           The purpose of this document, Air Quality Criteria for Particulate Matter (PM AQCD),
 6     is to present air quality criteria for particulate matter (PM) in accordance with Clean Air Act
 7     (CAA) Sections 108 and 109, which govern establishment, review, and revision of U.S. National
 8     Ambient Air Quality Standards (NAAQS) as follows:

 9      •  Section 108 directs the U.S. Environmental Protection Agency (EPA) Administrator to list
           pollutants that may reasonably be anticipated to endanger public health or welfare and to
           issue air quality criteria for them. The air quality criteria are to reflect the latest scientific
           information useful in indicating the kind and extent of all identifiable effects on public
           health and welfare expected from the  presence of the pollutant in ambient air.

10      •  Section 109 directs the EPA Administrator to set and periodically revise, as appropriate,
           (a) primary NAAQS, which in the judgement of the Administrator, are requisite to protect
           public health, with an adequate margin of safety, and (b) secondary NAAQS which, in the
           judgement of the Administrator, are requisite to protect the public welfare from any known
           or anticipated adverse effects (e.g.,  impacts on vegetation, crops,  ecosystems, visibility,
           climate, man-made materials, etc.).

11      •  Section 109 of the CAA also requires periodic review and, if appropriate, revision of
           existing criteria and standards. Also,  an independent committee of non-EPA experts, the
           Clean Air Scientific Advisory Committee (CASAC), is to provide the EPA Administrator
           advice and recommendations regarding the scientific soundness and appropriateness of
           criteria and NAAQS.

12           To meet these CAA mandates, this document assesses the latest scientific information
13     useful in deriving criteria as scientific bases for decisions on possible revision of current
14     PM NAAQS.  A separate EPA PM Staff Paper will draw upon assessments in this document,
15     together with technical analyses and other information, to identify alternatives for consideration
16     by the EPA Administrator with regard to possible retention or revision of the PM NAAQS.
       June 2003                                  E-l         DRAFT-DO NOT QUOTE OR CITE

-------
 1           The present document is organized into a set of 10 chapters, as follows:
 2      • This Executive Summary summarizes key points from the ensuing chapters.

 3      • Chapter 1 provides a general introduction, including information on legislative requirements
          and history of the PM NAAQS and an overview of approaches used to prepare this document.

 4      • Chapters 2, 3 and 5 provide background information on PM atmospheric science, air quality,
          and human exposure aspects to help place the succeeding discussions of PM health and
          environmental effects into perspective.

 5      • Chapter 4 deals with environmental effects of PM on vegetation and ecosystems, visibility,
          man-made materials, and climate.

 6      • Human health issues related to PM are addressed in Chapter 6 (Dosimetry); Chapter 7
          (Toxicology); and Chapter 8 (Community Epidemiology).

 7      • Chapter 9 provides an integrative synthesis of key points from the preceding chapters.
 9      E.2  AIR QUALITY AND EXPOSURE ASPECTS
10           The document's discussion of air quality and exposure aspects considers chemistry and
11      physics of atmospheric PM; analytical techniques for measuring PM mass, size, and chemical
12      composition; sources of ambient PM in the United States; temporal/spatial variability and trends
13      in ambient U.S. PM levels; and ambient concentration-human exposure relationships.  Overall,
14      the atmospheric science and air quality information provides further evidence substantiating the
15      1996 PM AQCD conclusion that distinctions between fine and coarse mode particles (in terms of
16      emission sources, formation mechanisms, atmospheric transformation, transport distances, air
17      quality patterns, and exposure relationships) warrant fine and coarse PM being viewed as
18      separate subclasses of ambient PM.  Key findings are summarized in the next several sections.
19
20
       June 2003                                 E-2        DRAFT-DO NOT QUOTE OR CITE

-------
1      E.2.1  Chemistry and Physics of Atmospheric Particles
2       •  Airborne PM is not a single pollutant, but rather is a mixture of many subclasses of
          pollutants with each subclass containing many different chemical species.  Particles
          suspended in the atmosphere originate from a variety of sources and possess a range of
          morphological, chemical, physical, and thermodynamic properties. A complete description
          of atmospheric particles would include an accounting of the chemical composition,
          morphology, and size of each particle, and the relative abundance of each particle type as a
          function of particle size.  Recent developments in single particle analysis techniques are
          bringing such a description closer to reality.

3       •  Particle size distributions show that atmospheric particles exist in two classes:  fine particles
          and coarse particles.  Fine and coarse  particles are defined primarily in terms of their
          formation mechanisms and size; they  also differ in sources, chemical composition, and
          removal processes (see Table 2-1).  Subsequent chapters show that fine and coarse particles
          also differ in aspects of concentration, exposure, dosimetry, toxicology, and epidemiology.
          These differences support the setting of separate standards for fine and coarse particles.

4       •  Fine and coarse particles overlap in the size range between 1 and 3 jim aerodynamic
          diameter where ambient PM concentrations are at a minimum. Coarse particles are generally
          larger than this minimum and are generally formed by  mechanical processes. Energy
          considerations limit the break-up of large mineral particles and small particle aggregates
          generally to  a minimum size of about  1 jim in diameter. Coarse particles and coarse-mode
          particles are equivalent terms.

5       •  Fine PM is derived primarily from combustion material that has volatilized and then
6         condensed to form primary PM or from precursor gases reacting in the atmosphere to form
          secondary PM.  New fine particles are formed by the nucleation of gas phase species; they
          grow by coagulation (existing particles combining) or  condensation (gases condensing on
          existing particles). Fine particles are subdivided into accumulation, Aitkin, and nucleation
          modes. In earlier texts, nuclei mode referred to the size range now split into the Aitkin and
          nucleation modes (see Figures 2-4 and 2-5).  Particles in the size range below 0.1 jim
          diameter are called ultrafine or nanoparticles and include the Aitkin and nucleation modes.

       June 2003                                  E-3        DRAFT-DO NOT QUOTE OR CITE

-------
1       •  PM2 5 is an indicator for fine particles. PM10_2 5 is an indicator for thoracic coarse particles
          (coarse particles capable of reaching the thoracic portion of the respiratory system - trachea,
          bronchi, and alveolar regions).  PM2 5 has been selected as the indicator for fine particles to
          include all accumulation-mode particles during high relative humidity, while recognition that
          it also includes some coarse-mode particles between 1 and 2.5 jim. PM2 5 specifies a sample
          collected through a size-selective inlet with a specified penetration curve at 50% cut point at
          2.5 |im aerodynamic diameter.

2       •  Aerosol scientists use three different approaches or conventions in the classification of
          particles by size: (1) modes, based on the observed size distributions and formation
          mechanisms; (2) cut point, usually based on the 50% cut point of the specific sampling
          device, including legally specified, regulatory sizes for air quality standards; and
          (3) dosimetry or occupational health sizes, based on the entrance into various compartments
          of the respiratory system.

3
4      E.2.2  Sources of Airborne Particles in the United States
5       •  Because of the complexity of the composition of ambient PM25 and PM10_25, sources are best
          discussed in terms of individual constituents of both primary and secondary PM25 and
          PM10_25.  Each of these constituents can have anthropogenic and natural sources as shown in
          Table 3-8. Major components of fine particles are sulfates, strong acid, ammonium nitrate,
          organic compounds, trace elements (including  metals), elemental carbon, and water.

6       •  Primary particles are emitted directly from sources.  Secondary particles are formed from
          atmospheric reactions of sulfur dioxide (SO2), nitrogen oxides (NOX), and certain organic
          compounds. NO reacts with ozone  (O3) to form NO2.  SO2 and NO2 react with hydroxy
          radical (OH) during the daytime to form sulfuric and nitric acid. During the nighttime, NO2
          reacts with ozone and forms nitric acid through a sequence of reactions involving the nitrate
          radical (NO3). These acids may react further with ammonia to form ammonium sulfates and
          nitrates.  Some types of higher molecular weight organic compounds react with  OH radicals,
          and olefmic compounds also react with ozone to form oxygenated organic compounds,

       June 2003                                 E-4        DRAFT-DO NOT QUOTE OR CITE

-------
           which nucleate or can condense onto existing particles. SO2 also dissolves in cloud and fog
           droplets, where it may react with dissolved O3, H2O2, or, if catalyzed by certain metals, with
           O2, yielding sulfuric acid or sulfates, that lead to PM when the droplet evaporates.

 7       •  Organic compounds constitute from 10 to 70% of dry PM2 5 mass. Whereas the chemistry of
           particulate nitrate and sulfate formation has been relatively well studied, the chemistry of
           secondary organic particulate matter formation is especially complex. Although additional
           sources of secondary organic PM might still be identified, there appears to be a general
           consensus that biogenic compounds (monoterpenes, sesquiterpenes) and aromatic
           compounds (e.g., toluene and ethylbenzene) are the most significant precursors.
           Atmospheric transformations of the compounds, which are formed in the particle phase
           during the aging of particles, are still not adequately understood.

 8       •  The results of receptor modeling studies throughout the United States indicate that the
           combustion of fossil and biomass fuels is the major source of measured  ambient PM25.
           Fugitive  dust, found mainly in the PM10_2 5 range size, represents the largest source of
           measured ambient PM10 in many locations in the western United States.  The application of
           any of the source apportionment techniques is still limited by the availability of source
           profile data. Whereas the Chemical Mass Balance approach relies directly on source profile
           data, solutions from the Positive Matrix Factorization technique yield profiles for the factors
           that contribute to PM.

 9       •  The use of organic compounds in source apportionment studies could potentially result in the
           attribution of PM to many more source categories  than is possible using only trace elements.
           However, in the relatively few studies of the composition of the organic fraction of ambient
           particles  that have been performed, typically only about 10 to 20 % of organic compounds
           have  been quantified.  The separation of contributions from diesel- and gasoline-fueled
           vehicles using organic marker compounds is still somewhat problematic. Additional efforts
           to develop protocols for extraction and analysis of organic markers are needed.

10
       June 2003                                  E-5         DRAFT-DO NOT QUOTE OR CITE

-------
1     E.2.3  Atmospheric Transport and Fate of Airborne Particles
2       •  Primary and secondary fine particles have long lifetimes in the atmosphere (days to weeks)
          and travel long distances (hundreds to thousands of kilometers).  They tend to be uniformly
          distributed over urban areas and larger regions, especially in the eastern United States.  As a
          result, they are not easily traced back to their individual sources.

3       •  Coarse particles normally have shorter lifetimes (minutes to hours) and generally only travel
          short distances (<10's of km). Therefore, coarse particles tend to be unevenly distributed
          across urban areas and tend to have more localized effects than fine particles. However, dust
          storms occasionally cause long range transport of small coarse-mode particles.

4
5     E.2.4  Airborne Particle Measurement Methods
6       •  Measurements of ambient PM mass and chemical composition are needed to determine
          attainment of standards; to guide progress towards attainment of a standard (including
          determination of source categories and validation of air quality models); and to determine
          health, ecological, and radiative effects. A comprehensive approach requires a combination
          of analytical techniques to assess (1) mass, (2) crustal and trace elements, (3) water-soluble
          ionic species including strong acidity, (4) elemental carbon, and (5) organic compounds.

7       •  There are no calibration standards for suspended particle mass; therefore, the accuracy of
8         particle mass measurements cannot be definitively determined.  The precision of particle
          mass measurements can be determined by comparing results from collocated samplers.
          Intercomparisons, using different techniques and samplers of different designs, coupled with
          mass balance studies (relating the sum of components to the measured mass), provide a
          method for gaining confidence in the reliability of PM measurements.

9       •  Mass concentration measurements with a precision of 10% or better have been  obtained with
          collocated samplers of identical design. Field  studies of EPA PM10  and PM25 reference
          methods and reviews of field  data from collocated PM10 and PM25 samplers show high
          precision (better than ± 10%). The use of more careful techniques, including double

      June 2003                                 E-6        DRAFT-DO NOT QUOTE OR CITE

-------
           weighing of filters, can provide higher precision and may be needed for precise
           determination of PM10_25 by difference.

10      •  Analytical techniques exist for measurement of the mass and chemical composition of PM
           retained on a filter (nonvolatile mass) in terms of elements (except carbon) and certain key
           ions (sulfate, nitrate, hydrogen, and ammonium). Acceptable measurements can be made of
           the total carbon retained on a filter. However, the split into organic carbon and elemental
           carbon depends on the operational details of the analytical methods and varies somewhat
           among methods.  Determination of the various organic compounds in the organic carbon
           fraction remains a challenge.

11      •  Important components of atmospheric PM (particle-bound water, ammonium nitrate, and
           many organic compounds) are termed semivolatile because significant amounts of both the
           gaseous and condensed phases may exist in the atmosphere in equilibrium.  Particle-bound
           water is not considered a pollutant. Most of the particle-bound water is removed by heating
           the particles or by equilibration of the collected particles at a low relative humidity (40%)
           for 24 hours.  However, these processes also cause the loss of other semivolatile
           components.  Semivolatile components also evaporate from the filter during sampling due to
           the pressure drop across the filter or due to a reduction in the atmospheric concentration
           during the sampling time.

12      •  Continuous methods must dry the PM to remove particle-bound water.  If heating is used to
           dry the particles, more of the semivolatile components may be removed than are lost in filter
           sampling.  Collection and retention of ammonium nitrate and semivolatile organic
           compounds represents a major challenge in the effort to move to continuous measurement of
           PM mass.  The use of diffusion dryers, which dehumidify the air stream without heating,
           represents a promising approach. Uncertainty in the efficiency of retention  of ammonium
           nitrate and organic compounds on filters also affects source category attribution and
           epidemiologic studies.
       June 2003                                 E-7        DRAFT-DO NOT QUOTE OR CITE

-------
1       •  Semivolatile organic compounds and semivolatile ammonium compounds (such as NH4NO3)
          may be lost by volatilization during sampling.  Such losses may be very important in
          woodsmoke impacted areas for organic compounds or in agricultural and other areas where
          low sulfate and high ammonia lead to high NH4NO3 concentrations. New techniques are
          now in use for measurement of nitrates and new research techniques are being tested for
          measurement of mass of semivolatile organic compounds in PM and of the total
          (semivolatile plus non-volatile) PM mass. The Federal Reference Methods (FRM) for PM10
          and PM25 give precise (± 10%) measurements of "equilibrated mass."  However, the loss of
          semivolatile PM (ammonium nitrate and organic compounds) and the possible retention of
          some particle-bound water in current PM mass measurements contribute to uncertainty in the
          measurement of the mass of PM as it exists suspended in the atmosphere.

2       •  Techniques are available to separate fine particles from coarse particles and collect the fine
          particles on a filter. No such technique exists for coarse particles. As yet, no consensus
          exists on the best technique for collecting a coarse particle sample for determination of mass
          and composition. Candidates include multistage impaction, virtual impaction, and
          difference (subtracting PM2 5 mass or composition from PM10 mass or composition).
          Advances in the theory and practice of virtual impaction suggest that it should be possible to
          design virtual impactors with much less than the 10% of fine PM collected in the coarse PM
          sample, as is now the case for the dichotomous samplers used in air quality studies and with
          penetration curves as sharp as those used in the current FRM for PM2 5.

3
4     E.2.5  Ambient PM Concentrations in the U.S.
5       •  The recently deployed PM2 5 FRM network has provided data for a large number of sites
          across the United States.  The data are stored in the Aerometric Information Retrieval
          System (AIRS). Data have also been collected at remote sites as part of the IMPROVE and
          NESCAUM networks. Annual mean U.S. PM25 concentrations from 1999 to 2000 range
          from about 5 |ig/m3 to about 30 |ig/m3. In the eastern United States, the data from 1999 to
          2001 indicate that highest quarterly mean concentrations and maximum concentrations most
          often occur during the summer.  In the western United States, highest quarterly mean values

      June 2003                                E-8         DRAFT-DO NOT QUOTE OR CITE

-------
    and maximum values occur mainly during the winter at a number of sites although there
    were exceptions to these general patterns. PM2 5 and PM10 concentrations many urban areas
    have generally declined over the past few decades.  However, they appear to have leveled off
    in the past few years.

    The median PM25 concentration across the United States during 1999, 2000, and 2001, the
    first three years of operation of the PM25 FRM network, was 13 |ig/m3, with a 95th percentile
    value of 18 |ig/m3. The corresponding median PM10_25 concentration was 10 |ig/m3, with a
    95th percentile value of 21 |ig/m3.

    Although PM2 5 concentrations within a given Metropolitan Statistical Area (MSA) can be
    highly correlated between sites, there can still be significant differences in their
    concentrations.  The degree of spatial uniformity in PM2 5 concentrations and the strength of
    site to site correlations in urban areas varies across the country. These factors should be
    considered in using data obtained by the PM2 5 FRM network to approximate community-
    scale human exposures, and caution should be exercised in extrapolating conclusions as to
    spatial uniformity or correlations obtained in one urban area to another. Limited information
    also suggests that the spatial variability in urban source contributions is likely to be larger
    than for regional source contributions to PM2 5 and for PM2 5, itself.

    Data for PM10_2 5 concentrations are not as abundant as they are for PM2 5. The difference
    method used in their derivation is subject to the effects of uncertainties in measuring both
    PM10 and PM2 5.  As a result, estimates of PM10_2 5 concentrations, at times, come out  as
    negative values, based on currently available data (e.g., in the EPA AIRS Database). In
    most cities where significant data is available, PM10_2 5 is spatially less uniform than PM2 5.

    Ambient PM contains both primary  and secondary components. The results of ambient
    monitoring studies and receptor modeling studies indicate that PM2 5 is dominated by
    secondary components in the eastern United States.  General statements about the origin of
    OC in ambient PM2 5 samples cannot yet be made and so the contribution of secondary
    components throughout the rest of the United States is still highly uncertain. Primary
    constituents represent a smaller but still important component of PM2 5. Crustal materials,
June 2003                                 E-9        DRAFT-DO NOT QUOTE OR CITE

-------
           which are primary constituents, constitute the largest measured fraction of PM10_2 5
           throughout the United States. Data for the concentration of bioaerosols in both the PM25 and
           PM10_2 5 size ranges are sparse.

10       •  Recent but limited information about policy-relevant background concentrations have not
           provided sufficient evidence to warrant any changes in estimates of the annual average
           background concentrations given in the 1996 PM AQCD.  These are: 1 to 4 |ig/m3 in the
           West and 2 to 5 |ig/m3 in the East for PM2 5; and approximately 3 |ig/m3 in both the East and
           the West for PM10_2 5, with a range of 0 to 9 |ig/m3 in the West and 0 to 7 |ig/m3 in the East.
           Such concentrations are likely to be highly variable both spatially  and temporally. Further
           information regarding the frequency distribution of 24-hour concentrations based on
           analyses of observations at relatively remote monitoring sites and  on source apportionment
           analyses has become available and can be used for selected sites.

11
12     E.2.6  Human Exposure to PM
13       •  Personal exposure to PM mass or its constituents results when individuals come in contact
           with  particulate pollutant concentrations in locations or microenvironments that they
           frequent during a specific period of time. Various PM exposure metrics can be defined
           according to its source (i.e., ambient,  nonambient) and the microenvironment where
           exposure occurs.

14       •  Most people spend most of their time indoors where they are exposed to indoor-generated
           PM and ambient PM that has infiltrated indoors.

15       •  Indoor-generated and ambient PM differ in sources, sizes, chemical composition, and
           toxicity.

16       •  The ambient PM concentration and the indoor PM concentration can be measured by outdoor
           and indoor monitors.  The total personal exposure can be measured by a personal exposure
           monitor carried by the person. However, the concentrations of indoor-generated PM and
       June 2003                                 E-10        DRAFT-DO NOT QUOTE OR CITE

-------
           ambient PM that has infiltrated indoors and the related values of ambient and nonambient
           PM exposures must be estimated.

17       •  The intercept of a regression of individual, daily values of total personal exposure on daily
           PM concentrations, gives the average nonambient PM exposure; and the slope gives the
           average attenuation factor (the ratio of ambient PM exposure to ambient PM concentration).

18       •  Similarly, the intercept of a regression of individual, daily values of indoor PM concentration
           on daily ambient concentrations, gives the average concentration of indoor-generated PM;
           and the slope gives the average infiltration factor (concentrations of ambient PM that has
           infiltrated indoors/ambient PM concentration).

19       •  The attenuation factor and the infiltration factor depend on the penetration coefficient, the
           fraction of ambient PM that penetrates through the walls, doors, windows, etc.; the
           deposition or removal rate, a measure of how rapidly PM within the indoor
           microenvironment is removed by deposition to surfaces or by filtration in a heating/cooling
           system; and the air exchange rate, a measure of how rapidly indoor air is replaced by outdoor
           air. The attenuation factor also depends on the fraction of time spent outdoors.

20       •  The air exchange rate is an important variable for determining the concentration of ambient
           PM found indoors.  It can be measured by release and measurement of an inert tracer gas
           indoors.  The air exchange rate increases with opening of windows  or doors or operation of
           window or attic fans. It also increases as the indoor/outdoor temperature difference
           increases. For closed homes, i.e., no open windows or doors, the air exchange rate does not
           appear to be a function of wind speed or direction.

21       •  The penetration coefficient and the deposition rate can be estimated from measurements of
           outdoor and indoor concentrations under conditions when there are no indoor sources (night
           time or unoccupied home).  These parameters are functions of particle size. The penetration
           coefficient is high and the deposition rate is low for accumulation mode particles (0.1 to
           1.0 jim). The penetration coefficient is lower and the deposition rate is higher for ultrafme
           particles (< 0.1 jim) and coarse mode particles (> 1.0 jim). The attenuation factor and the
        June 2003                                 E-11        DRAFT-DO NOT QUOTE OR CITE

-------
           infiltration factor are higher for particles in the accumulation mode than for ultrafme or
           coarse particles.

22       • The attenuation factor and the infiltration factor will vary as the air exchange rate does, and,
           therefore, will vary with season and housing characteristics.  These factors will increase with
           increased opening of windows and doors.  For closed homes, these factors will increase with
           an increase of the indoor/outdoor temperature difference; but they do not appear to be
           affected by wind speed or direction.

23       • The regression technique is useful for finding average values of the attenuation factor and the
           nonambient exposure and possibly for estimating the distribution of individual, daily values
           of the nonambient PM exposure.

24       • Individual, daily values of the ambient PM exposure, the nonambient PM exposure, and the
           attenuation factor may be determined from individual, daily values of the total PM personal
           exposure and daily ambient PM concentrations by several techniques:
25         -  Mass balance technique.  Direct measurement of the air exchange rate, measurement of
              the fraction of time spent outdoors by a diary of the subject's activity pattern, and use of
              the equilibrium mass balance equation for the attenuation factor with estimated values of
              the penetration coefficient and the deposition rate.

26         -  Sulfate ratio technique. Individual, daily values of the attenuation factor (for PM25) will
              be given by individual, daily values of personal exposure to sulfate / the daily ambient
              sulfate concentration provided there are no indoor sources of sulfate and sulfate and PM2 5
              have similar particle size distributions.

27         -  Recursive technique.  Indoor-generated emissions, which tend to be episodic, can be
              removed from a continuous record of indoor PM concentration, allowing separation of
              indoor-generated PM and ambient PM that has infiltrated indoors.

28       • In pooled studies (different subjects measured on different days), individual, daily values of
           total PM exposure are usually not well correlated with daily ambient PM concentrations.
        June 2003                                 E-12        DRAFT-DO NOT QUOTE OR CITE

-------
           In longitudinal studies (each subject measured for multiple days), individual, daily values of
           total PM personal exposure and daily ambient PM concentrations are highly correlated for
           some, but not all subjects.

29       •  Only one study has reported estimated individual, daily values of ambient and nonambient
           PM exposure.  Individual, daily values of total PM personal exposure and daily ambient PM
           concentrations were poorly correlated. However, individual, daily values of ambient PM
           exposure and daily ambient PM concentrations were highly correlated.  Individual daily
           values of ambient and nonambient PM exposure were not well correlated. Individual daily
           values of nonambient PM exposure and daily ambient PM concentrations were also not well
           correlated.

30       •  As long as the nonambient PM exposure is not correlated with the ambient PM exposure, it
           will not bias the estimated health effect of PM.  However, the effect per jig/ambient PM
           concentration will be biased low compared to the health effect per jig/ambient PM exposure
           by the attenuation factor. This effect probably explains some of the heterogeneity in PM10
           effects observed in multicity epidemiology studies, as indicated by  a correlation of PM
           effects in different cities with air conditioning use in those cities (i.e., the higher the air
           conditioning use, the lower the health effect estimate per |ig/m3 of ambient PM).

31       •  Exposure relationships also provide  some insight into the issue of confounding. While the
           data base is small, concentrations of gaseous co-pollutants, NO2, O3, and SO2 (and probably
           CO) are likely poorly correlated, and sometimes not significantly correlated, with personal
           exposure to the respective co-pollutant. However, they are frequently significantly
           correlated with both the  ambient PM concentration and the ambient PM exposure. Thus, in a
           regression, where associations are found between gaseous co-pollutants and a health effect, it
           may be because they are a surrogate for PM rather than a confounder. That is, the health
           effect due to PM is transferred to the gaseous pollutant because of the positive correlation
           between the ambient concentration of the gas and the ambient PM exposure.

32
33

        June 2003                                 E-13        DRAFT-DO NOT QUOTE OR CITE

-------
 1     E.3   DOSIMETRY OF PARTICULATE MATTER
 2      • Dosimetry establishes the relationship between PM exposure and the dose of inhaled PM
          delivered to and retained at the target site. Deposition, clearance, translocation, and retention
          comprise the essential elements of dosimetry.

 3      • Dosimetric information is critical to extrapolating effects found in controlled exposure
          studies of laboratory animals to those observed in human exposure studies and for relating
          effects in normal healthy persons to those in potentially susceptible persons.

 4      • Dosimetry separates the respiratory tract into three regions, extrathoracic (ET),
          tracheobronchial (TB), and alveolar (A), based on anatomical features and particle deposition
          and clearance phenomena that occur within each region.

 5      • Particles in the accumulation mode size range (0.1  to 1.0  jim Dp) have the lowest deposition
          fraction in all three regions.

 6      • Coarse and ultrafme particles have higher fractional deposition. For coarse particles,
          fractional deposition peaks between 5 and 10 jim Dp for the TB region and 2.5 and 5  jim Dp
          for the A region.

 7      • For ultrafme particles, fractional deposition peaks between 0.0025 and 0.005 |im Dp for the
          TB region and between 0.01 and 0.05 for the A region.

 8      • A significant fraction of ultrafme and coarse particles, but not particles in the accumulation-
          mode size range, are deposited in the ET region.

 9      • Such transport could provide a mechanism whereby particles could affect cardiovascular
          function as reported in the epidemiologic studies

10      • Fractional  deposition, as a function of particle size, depends on lung size, tidal volume, and
          breathing rate. Exercising subjects receive higher doses of particles per cm2 of lung surface
          than subjects at rest.
       June 2003                                 E-14       DRAFT-DO NOT QUOTE OR CITE

-------
1       •  Airway structure and physiological function vary with age. Such variations may alter the
          deposition patterns for inhaled particles.  Airflow distribution is very uneven in diseased
          lungs, and deposition can be enhanced locally in areas of active ventilation. Total lung
          deposition is generally increased by obstructed airways so that particle deposition can be
          enhanced in people with chronic lung disease. Unfortunately, deposition studies in another
          susceptible population, the elderly, are still lacking.

2       •  Particles depositing on airway surfaces may be completely cleared from the respiratory tract
          or translocated to other sites within this system by regionally specific clearance mechanisms.
          Clearance is either absorptive (dissolution) or nonabsorptive (transport of intact particles).
          Deposited particles may  be dissolved in body fluids, taken up by phagocytic cells, or
          transported by the mucociliary system. Retained particles tend to be small (< 2.5 jim) and
          poorly soluble (e.g., silica, metals). Ultrafine particles can be rapidly cleared from the lungs
          into the systemic circulation where they can be transported to extrapulmonary regions.

3       •  Tracheobronchial clearance has  both a fast and a slow component.  In the fast phase particles
          deposited in the TB region clear out rapidly during the first several hours and continue to
          clear out for 24 hours. A small remaining portion may clear out over several days (slow
          phase). Translocation of poorly soluble PM to the lymph nodes takes a few days and is more
          rapid for smaller (< 2 jim) particles; elimination rates of these retained particles are on the
          order of years. People with COPD have increased particle retention partly because of
          increased initial deposition and impaired mucociliary clearance and use cough to augment
          mucociliary clearance.

4       •  Alveolar clearance takes months to years. Particles may be taken up by  alveolar
          macrophages within 24 hours, but some phagocytosed macrophages translocate into the
          interstitium or lymphatics whereas some  remain on the alveolar surface. Penetration of
          uningested particles into the interstitium increases with increasing particle load and results in
          increased translocation to lymph nodes.

5       •  Acute effects of PM are probably best related to deposited dose; whereas chronic effects may
          be related to cumulative  or retained dose.  Retention of particles is  a function  of deposition
       June 2003                                 E-15        DRAFT-DO NOT QUOTE OR CITE

-------
          site, clearance of particles by macrophages or the mucociliary system, and particle
          characteristics, especially solubility.  Chronic effects may also arise from recurring cycles of
          pulmonary injury and repair.

          Computational models allow calculation of fractional deposition and dose per cm2 of lung
          surface as a function of particle size and respiratory parameters for humans and some animals
          (including the laboratory rat). Such calculations can be used to predict the exposures needed
          to produce comparable doses for animal to human extrapolation.  Computational models have
          been improved in recent years but experimental validation of model predictions is still
          required.
 8     E.4  TOXICOLOGY OF PARTICULATE MATTER IN HUMANS AND
 9           LABORATORY ANIMALS
10     E.4.1  Health Effects of Specific PM Components
11      •  There is relatively little new information on the effects of acid aerosols, and the conclusions
           of the 1996 PM AQCD, i.e., Air Quality Criteria for Parti culate Matter (U.S. Environmental
           Protection Agency, 1996a), are unchanged. It was previously concluded that acid aerosols
           cause little or no change in pulmonary function in healthy subjects, but asthmatics may
           develop small changes in pulmonary function.  Although pulmonary effects of acid aerosols
           have been the subject of extensive research in past decades, the cardiovascular effects of acid
           aerosols have received little attention and should not be ruled out as possible mediators of
           PM health effects.

12      •  Health effects of particle-associated soluble metals have been demonstrated by in vivo and
           in vitro studies using residual oil fly ash (ROFA) or soluble transition metals. Although
           there are some uncertainties about differential effects of one transition metal versus another,
           water soluble metals  leached from ROFA or ambient filter extracts have been shown
           consistently (albeit at high concentrations) to cause cell injury and inflammatory changes
           in vitro and in vivo.
       June 2003                                E-16       DRAFT-DO NOT QUOTE OR CITE

-------
 • Even though it is clear that combustion particles that have a high content of soluble metals
   can cause lung injury and even death in compromised animals and correlate well with
   epidemiologic findings in some cases (e.g., Utah Valley Studies), it has not been fully
   established that the small quantities of metals associated with ambient PM are sufficient to
   cause health effects. Moreover, it cannot be assumed that metals are the only or the primary
   toxic component of ambient PM, rather than there possibly being many different toxic agents
   contributing to ambient PM health effects.

 • There is growing toxicological evidence that diesel exhaust particles exacerbate the allergic
   response to inhaled antigens. The organic fraction of diesel exhaust has been linked to
   eosinophil degranulation and induction of cytokine production, suggesting that the organic
   constituents of diesel PM are the responsible part for the immune effects. It is not known
   whether the adjuvant-like activity of diesel PM is unique or whether other combustion
   particles have similar effects.  It is important to compare the immune effects of other
   source-specific emissions, as well as concentrated ambient PM, to diesel PM to determine
   the extent to which exposure to diesel exhaust may contribute to the incidence and severity
   of allergic rhinitis and asthma.

 • Published research on the acute effects  of particle-associated organic carbon constituents is
   conspicuous by its relative  absence, except for diesel exhaust particles.

 • Studies with ultrafme particles have demonstrated a significantly greater inflammatory
   response than that seen with fine particles of the same chemical composition at similar mass
   doses. In other more limited studies, ultrafines also have generated greater oxidative stress
   in experimental animals.  However, when the particle surface area is used as a dose metric,
   the inflammatory response  to both fine  and ultrafme particles may be basically the same.
   Thus, it may be the higher surface area  of ultrafme particles is the important factors
   contributing to health effects.

 • Concentrated ambient particle  (CAPS)  studies should be among the most relevant in helping
   to understand the characteristics of PM producing toxicity, susceptibility of individuals to
   PM, and the underlying mechanisms. Studies have used collected urban PM for intratracheal
   administration to healthy and compromised animals. Despite the difficulties in extrapolating

June 2003                                 E-17        DRAFT-DO NOT QUOTE OR CITE

-------
           from the bolus delivery used in such studies, they have provided strong evidence that the
           chemical composition of ambient particles can have a major influence on toxicity.

           Ambient particle concentrators, that concentrate particles in situ, provide particles for
           inhalation studies.  The concentration of particles without collection avoids the changes in
           size and composition that occur during collection during collection and resuspension.
           Studies with inhaled CAPs have observed cardiopulmonary changes in rodents and dogs at
           high concentrations of PM between 0.15 and 2.5  jim diameter.  A new generation of ambient
           particle concentrators that allow separation and concentration of coarse-mode, accumulation-
           mode, and ultrafme particle separately will  permit the direct toxicological comparison of
           these various ambient particle sizes.

           Recent studies support the conclusion of the 1996 PM AQCD, which stated that bioaerosols,
           at concentrations present in the ambient environment, cannot account for the reported health
           effects of ambient PM.  However, it is possible that bioaerosols could contribute to the health
           effects of PM.
 9      E.4.2  Mechanisms of Action
10      E.4.2.1  Cardiovascular Effects
11       •  Changes in heart rate, heart rate variability, and conductance associated with ambient PM
           exposure have been reported in animal studies, in several human panel studies, and in a
           reanalysis of data from the MONICA study. Some of these studies included endpoints
           related to respiratory effects but few significant adverse respiratory changes were detected.
           This raises the possibility that ambient PM may have effects on the heart that are
           independent of adverse changes in the lung.
12       •  Inhaled particulate matter affects the heart by uptake of particles into the circulation
           or release of a soluble substances into the circulation.
13       •  Inhaled parti culate matter affects autonomic control of the heart and cardiovascular system.
14
15
        June 2003                                  E-18         DRAFT-DO NOT QUOTE OR CITE

-------
1      E.4.2.2  Respiratory Effects
2       •  Particularly compelling evidence pointing towards ambient PM causing lung injury and
          inflammation derives from the study of ambient PM materials on filter extracts collected
          from community air monitors before, during the temporary closing of a steel mill in Utah
          Valley, and after its reopening.  Intratracheal instillation of filter extract materials in human
          volunteers provoked greater lung inflammatory responses for materials obtained before and
          after the temporary closing versus that collected during the plant closing.  Further, the
          instillation in rats of extract materials from before and after the plant closing resulted in a
          50% increase in air way hyperresponsiveness to acetylcholine compared to 17 or 25%
          increases with saline or extract materials for the period when the plant was closed,
          respectively.  Analysis of the extract materials revealed notably greater quantities of metals
          for when the plant was opened suggesting that such metals (e.g., Cu, Zn, Fe, Pb, As, Mn, Ni)
          may be important contributors to the pulmonary toxicity observed in the controlled exposure
          studies, as well as to health effects shown epidemiologically to vary with PM exposures of
          Utah Valley residents before,  during, and after the steel mill closing.

3       •  Still other toxicological studies point towards lung injury and inflammation being associated
          with exposure of lung tissue to complex combustion-related PM materials, with metals again
          being likely contributors. Rats with SO2-induced bronchitis and monocrotaline-treated rats
          have been reported to have a greater inflammatory response to concentrated ambient PM
          than normal rats.  These studies suggest that exacerbation of respiratory disease by ambient
          PM may be caused in part by lung injury and inflammation.

4       •  Particulate air pollution causes increased susceptibility to respiratory infections. Exposure
          of rats to ROFA and L. monocytogenes, a bacterial pathogen, led rats treated with saline or
          L. monocytogenes in the absence of ROFA. Preexposure of rats to ROFA significantly
          enhanced injury and delayed the pulmonary clearance of a subsequent challenge of L.
          monocytogenes, when compared to saline-treated control rats. Acute exposure to ROFA
          appeared to slow pulmonary clearance of L. monocytogenes and to alter AM function.  Such
          changes could lead to increased susceptibility to lung infection in exposed populations.
       June 2003                                 E-19        DRAFT-DO NOT QUOTE OR CITE

-------
 1       •  Particulate air pollution increases airway reactivity and exacerbates asthma.  Diesel
           paniculate matter (DPM) has been shown to increase production of antigen-specific IgE in
           mice and humans. In vitro studies have suggested that the organic fraction of DPM is
           involved in the increased IgE production. ROFA leachate also has been shown to enhance
           antigen-specific airway reactivity in mice, indicating that soluble metals can also enhance an
           allergic response.
 2
 3     E.4.2.3  Systemic Effects Secondary to Lung Injury
 4       •  Lung injury from inhaled PM causes impairment of oxygenation and increased work of
           breathing that adversely affects the heart.

 5       •  Lung inflammation and cytokine production cause adverse systemic hemodynamic effects.

 6       •  Lung inflammation from inhaled PM causes increased blood coagulability that increases the
           risk of heart attacks and strokes.

 7       •  Interaction of PM with the lung affects hematopoiesis.
 9     E.4.3  Susceptibility
10       •  Older animals or animals with certain types of compromised health, either genetic or
           induced, are more susceptible to instilled or inhaled particles, although the increased animal-
           to-animal variability in these models has created greater uncertainty for the interpretation of
           the findings. Moreover, because PM seems to affect broad categories of disease states,
           ranging from cardiac arrhythmias to pulmonary infection, it can be difficult to know what
           disease models to use in evaluating the biological plausibility of adverse health effects of
           PM.  Nevertheless, particularly interesting new findings point toward ambient PM
           exacerbation of allergic airway hyperresponsiveness and/or antigen-induced immune
           responses. Both metals and diesel particles have been implicated, with an expanding array of
           new studies showing DPM in particular as being effective in exacerbating allergic asthma
           responses.
       June 2003                                 E-20        DRAFT-DO NOT QUOTE OR CITE

-------
 1     E.5  EPIDEMIOLOGY OF HUMAN HEALTH EFFECTS ASSOCIATED
 2           WITH AMBIENT PARTICULATE MATTER
 3           Chapter 8 of this document assesses the extensive PM epidemiologic data base that has
 4     become newly available since the 1996 PM AQCD.  The most important types of additions to
 5     the epidemiologic database beyond that assessed in the 1996 PM AQCD include the following:
 6       •  New multi-city studies on a variety of endpoints which provide more precise estimates of the
           average PM effect sizes than most smaller-scale  individual city studies;

 7       •  More studies of various health endpoints using ambient PM10 and/or closely related mass
           concentration indices (e.g., PM13 and PM7), which substantially lessen the need to rely on
           non-gravimetric indices (e.g., BS or CoH);

 8       •  New studies evaluating relationships of a variety of health endpoints to the ambient PM
           coarse fraction (PM10_2 5), the ambient fine-particle fraction (PM2 5), and even ambient
           ultrafine particles measures (PM0 x and smaller),  using direct mass measurements and/or
           estimated from site-specific calibrations;

 9       •  A few new studies in which the relationship of some health endpoints to ambient particle
           number concentrations were evaluated;

10       •  Many new studies which evaluated the sensitivity of estimated PM effects to the inclusion of
           gaseous co-pollutants in the model;

11       •  Preliminary attempts to evaluate the effects combinations or mixtures of air pollutants
           including PM components, based on factor analysis or source profiles;

12       •  Numerous new studies of cardiovascular endpoints, with particular emphasis on assessment
           of cardiovascular risk factors as well as symptoms;

13       •  Additional new studies on asthma and other respiratory conditions potentially exacerbated
           by PM exposure;

14       •  New analyses of lung cancer associations with long-term exposures to ambient PM;
       June 2003                                E-21        DRAFT-DO NOT QUOTE OR CITE

-------
 1       •  New studies of infants and children as a potentially susceptible population.

 2           In addition, Chapter 8 discusses statistical issues posed by use of General Additive Mode
 3     (GAM) analyses involving default convergence criteria in widely used commercially-available
 4     software employed in time-series analyses of ambient PM-health effects relationship reported in
 5     a few published studies assessed in the 1996 PM AQCD and in numerous more recent studies
 6     assessed in the present PM AQCD. This includes discussion of the following key points:
 7       • Use of GAM analyses with default convergence criteria (GAM default) has variable impacts
           on PM effect size estimates from study to study, depending on many factors (numbers of
           observations, numbers of potential effect modifiers or potential confounders included,
           specific "smoothing" approaches used to control for their effects, numbers of degrees of
           freedom used, etc.). The effect of GAM (default) use tends most often to be some
           (unusually only slight) overestimation of the PM effect size compared to results obtained
           with use of GAM with stringent convergence criteria or other appropriate modeling
           approaches, e.g., general linear models (GLM).

 8       • The results of EP A-encouraged reanaly ses of a number of (> 3 5) of important PM time-
           series analyses comparing PM effect sizes and standard error (confidence interval) estimates
           from GAM (default) analyses versus GAM (stringent) or other appropriate statistical
           approaches (GLM). These reanalyses appear in a Health Effects Institute Special Report
           (HEI, 2003) that includes not only short communications on the GAM reanalyses by the
           original investigators, but also commentaries on the reanalyses and their implications for
           interpreting the PM time-series analyses results by a special peer-review panel convened by
           HEI at EP A's request.

 9
10     E.5.2  Key Epidemiologic Findings
11           The epidemiologic studies discussed in Chapter 8 demonstrate biologically-plausible
12     responses in humans exposed at ambient concentrations. These observational study findings are
13     further enhanced by supportive findings of causal studies from other scientific disciplines
14     (dosimetry, toxicology, etc.), in which other factors could be experimentally controlled, as

       June 2003                                 E-22       DRAFT-DO NOT QUOTE OR CITE

-------
1      discussed in Chapters 6 and 7. The most salient findings derived from the PM epidemiologic
2      studies include the following:
3       • A large and reasonably convincing body of epidemiology evidence confirms earlier
          associations between short- and long-term ambient PM10 exposures (inferred from stationary
          community air monitor measures) and mortality/morbidity effects and suggest that PM10
          (or one or more PM10 components) is a probable contributing cause of adverse human health
          effects.

4       • There appears to be some spatial heterogeneity in city-specific excess risk estimates for the
          relationships between short-term ambient PM10 concentrations and acute health effects. The
          reasons for such variation in effects estimates are not well understood at this time but do not
          negate ambient PM's likely causative contribution to observed PM-mortality and/or
          morbidity associations in many locations. Possible factors contributing to the apparent
          heterogeneity include geographic differences in air pollution mixtures, composition of PM
          components, and personal and sociodemographic factors affecting PM exposure (such as use
          of air conditioners, education, and so on).

5       • A growing body of epidemiologic studies confirm associations between short- and long-term
          ambient PM2 5 exposures (inferred from stationary air monitor measures) and adverse health
          effects and suggest that PM2 5 (or one or more PM2 5 components) is a probable contributing
          cause of observed PM-associated health effects. Some new epidemiologic findings also
          suggest that health effects are associated with mass or  number concentrations of ultrafme
          (nuclei-mode) particles, but not necessarily more so than for other ambient fine PM
          components.

6       • A smaller body of evidence appears to support an association between short-term ambient
          thoracic coarse fraction (PM10_2 5) exposures (inferred from stationary air monitor measures)
          and short-term health effects in epidemiologic studies.  This suggests that PM10_2 5, or some
          constituent component(s) of PM10_25, may be a contributory cause of adverse health effects
          in some locations.  Reasons for differences among findings on coarse-particle health effects
          reported for different cities are still poorly understood,  but several of the locations where
       June 2003                                 E-23        DRAFT-DO NOT QUOTE OR CITE

-------
           significant PM10_2 5 effects have been observed (e.g., Phoenix, Mexico City, Santiago) tend to
           be in drier climates and may have contributions to observed effects due to higher levels of
           organic particles from biogenic processes (endotoxins, molds, etc.) during warm months.
           Other studies suggest that particles of crustal origin are generally unlikely to exert notable
           health effects under most ambient exposure conditions. Also, in some western U.S. cities
           where PM10_2 5 is a large part of PM10, the relationship between hospital admissions and PM10
           may be an indicator of response to coarse thoracic particles from wood burning.

7       •  Long-term PM exposure durations, on the order of months to years, as well as on the order
           of a few days, are statistically associated with serious human health effects (indexed by
           mortality, hospital admissions/medical visits, etc.). More chronic PM exposures, on the
           order of years or decades, appear to be associated with life shortening well beyond that
           accounted for by the simple accumulation of the more acute effects of short-term PM
           exposures (on the order of a few days).  Substantial uncertainties  remain regarding the
           magnitude of and mechanisms underlying chronic health effects of long-term PM exposures
           and the relationship between chronic exposure and acute responses to short-term exposure.

8       •  Recent investigations of the public health implications of such chronic PM exposure-
           mortality effect estimates were also reviewed. Life table calculations by Brunekreef (1997)
           found that relatively small differences in long-term exposure to airborne PM of ambient
           origin can have substantial effects on life expectancy.  For example, a calculation for the
           1969-71 life table for U.S. white males indicated that a chronic exposure increase of
           10  |ig/m3 PM was associated with a reduction of 1.3 years for the entire population's life
           expectancy at age 25. Also, new evidence of associations of PM  exposure with infant
           mortality and/or intrauterine growth retardation and consequent increase risk for many
           serious health conditions associated with low birth weight, if further substantiated, would
           imply that life shortening in the entire population from long-term PM exposure could well
           be significantly larger than that estimated by Brunekreef (1997).

9      •   Considerable coherence exists among effect size estimates for ambient PM health effects.
           For example, results derived from several multi-city studies,  based on pooled analyses of


       June 2003                                 E-24         DRAFT-DO NOT QUOTE OR CITE

-------
           data combined across multiple cities (thought to yield the most precise estimates of mean
           effect size), show the percent excess total (non-accidental) deaths estimated per 50 |ig/m3
           increase in 24-h PM10 to be -1.4% in the 90 largest U.S. cities with the estimate for the
           Northeast being the largest (approximately twice the nationwide estimate); -3.4% in
           10 large U.S. cities; -3.6% in the 8 largest Canadian cities; and -3.0% in western European
           cities.  These combined estimates are consistent with the range of PM10 estimates previously
           reported in the 1996 PM AQCD.  These and excess risk estimates from many other
           individual-city studies, generally falling in the range of ca. 1.5 to 8.0% per 50 |ig/m3 24-h
           PM10 increment, also comport well with numerous new studies confirming increased cause-
           specific cardiovascular- and respiratory-related mortality.  They are also coherent with larger
           effect sizes reported for cardiovascular and respiratory hospital admissions and visits,  as
           would be expected for these morbidity endpoints versus those for PM10-related mortality.

10       • Several independent panel studies (but not all) that evaluated temporal associations between
           PM exposures and measures of heart beat rhythm in elderly subjects provide generally
           consistent indications of decreased heart rate variability (HRV) being associated with
           ambient PM exposure (decreased HRV being an indicator of increased risk for serious
           cardiovascular outcomes, e.g., heart attacks).  Other studies point toward changes in blood
           characteristics (e.g., C-reactive protein levels) related to increased risk of ischemic heart
           disease also being associated with ambient PM exposures. However,  these heart rhythm and
           blood characteristics findings should currently be viewed as providing only limited or
           preliminary support for PM-related cardiovascular effects.

11       • Notable new evidence now exists which substantiates positive associations between ambient
           PM concentrations and increased respiratory-related hospital admissions, emergency
           department, and other medical visits, particularly in relation to PM10 levels. Of much
           interest are new findings tending to implicate not only fine particle  components but also
           coarse thoracic (e.g., PM10_25) particles as likely contributing to exacerbation  of asthma
           conditions.  Also of much interest are  emerging new findings indicative of likely increased
           occurrence of chronic bronchitis in association with (especially chronic) PM exposure. Also
           of particular interest are reanalyses or extensions of earlier prospective cohort studies  of
        June 2003                                 E-25        DRAFT-DO NOT QUOTE OR CITE

-------
           long-term ambient PM exposure effects which demonstrate substantial evidence for
           association of increased lung cancer risk with such PM exposures, especially exposure to
           fine PM or its subcomponents.

12       • One major methodological issue affecting epidemiologic studies of both short-term and
           long-term PM exposure effects is that ambient PM of varying size ranges is typically found
           in association with other air pollutants, including gaseous criteria pollutants (e.g, O3, NO2,
           SO2, CO), air toxics, and/or bioaerosols. Available statistical methods for assessing
           potential confounding arising from these associations may not yet be fully adequate. The
           inclusion of multiple pollutants often produces statistically unstable estimates.  Omission of
           other pollutants may incorrectly attribute their independent effects to PM.  Second-stage
           regression methods may have certain pitfalls that have not yet been fully evaluated.  Much
           progress in sorting out relative contributions of ambient PM components versus other
           co-pollutants is nevertheless being made and, overall, tends to substantiate that observed PM
           effects are at least partly due to ambient PM acting alone or in the presence of other
           covarying gaseous pollutants. However, the statistical association of health effects with PM
           acting alone or with other pollutants should not be taken as an indicator of a lack of effect of
           the other pollutants.

13       • It is possible that differences in observed health effects will be found to depend on site-
           specific differences in chemical and physical composition  characteristics of ambient
           particles and on factors affecting exposure (such as air conditioning) as well as on
           differences in PM mass concentration. For example, epidemiologic Utah Valley studies
           showed that PM10 particles, known to be richer in metals during exposure periods while the
           steel mill was operating, were more highly associated with adverse health effects than was
           PM10 during the PM exposure reduction while the steel mill was closed.  In contrast, PM10 or
           PM2 5 was relatively higher in crustal particles during windblown dust episodes in Spokane
           and in three central Utah sites than at other times, but was  not associated with higher total
           mortality.  These differences require more research that may become more feasible as the
           PM2 5 sampling network produces air quality data related to speciated samples.
        June 2003                                 E-26        DRAFT-DO NOT QUOTE OR CITE

-------
1       • The above reasons suggest it is inadvisable to pool PM epidemiologic studies for different
          locations or time periods, with different population sub-groups, or different health endpoints
          (as is often done in "meta-analyses"), without careful assessment of potential causes and
          consequences of these differences and appropriate caveating of results. Published multi-city
          analyses using common data bases, measurement methods, analytical strategies, and
          extensive independent external review (as carried out in the APHEA and NMMAPS studies)
          are useful. Pooled analyses  of more diverse collections of independent studies of different
          cities, using varying methodology and/or data quality or representativeness, are likely less
          credible and should not, in general, be used without careful assessment of their underlying
          scientific comparability.

2       • It may be possible that different PM size components or particles with different composition
          or sources produce effects by different mechanisms manifested at different lags or that
          different preexisting conditions may lead to different delays between exposure and effect.
          Thus, although maximum effect sizes for PM effects have often been reported for 0-1 day
          lags, evidence is also beginning to suggest that more consideration should be given to lags of
          several days. Also, if it is considered that all health effects occurring at different lag days
          are all real effects, so that the risks for each lag day should be additive, then overall risks
          may exist that are higher than implied by maximum estimates for any particular single or
          two-day lags.  In that case, multi-day averages or distributed lag models may provide more
          accurate estimate of the total impact of PM on the population.

3       • Certain classes of ambient particles may be distinctly less toxic than others and may not
          exert human health effects at typical ambient exposure concentrations or only under special
          circumstances. Coarse thoracic particles of crustal origin, for example, may be relatively
          non-toxic under most circumstances compared to those of combustion origin such as wood
          burning. However, crustal particles may be sufficiently toxic to cause human health effects
          under some conditions; resuspended crustal particles, for example, may carry toxic trace
          elements and other components from previously deposited fine PM, e.g., metals from
          smelters (Phoenix) or steel mills (Steubenville, Utah Valley), PAH's from automobile
          exhaust, or pesticides from administration to agricultural lands.  Likewise, fine particles
       June 2003                                 E-27        DRAFT-DO NOT QUOTE OR CITE

-------
    from different sources have different effect sizes. More research is needed to identify
    conditions under which one or another class of particles may cause little or no adverse health
    effects, as well as conditions under which different classes of particles may cause relatively
    more notable effects.

    Certain epidemiologic evidence from "intervention studies" suggests that reducing ambient
    PM10 concentrations may reduce a variety of health effects on a time scale from a few days
    to a few months, as found in epidemiologic studies of "natural experiments" such as in the
    Utah Valley, and supported by toxicology studies using particles extracts from ambient
    community sampling filters from the Utah Valley. Recent studies in Germany and in the
    Czech Republic also tend to support a hypothesis that reductions in particle-related air
    pollution are associated with reductions in the incidence of adverse health effects.

    Studies that combine the features of cross-sectional and cohort studies provide the best
    evidence for chronic effects of PM exposure. Gauderman et al. (2000; 2002) have found
    significant decreases in lung function growth related to PM10 levels using these techniques.

    Adverse health effects in children reported in a limited number of more recent studies are
    emerging as an important area of more concern than was the case in the 1996 PM AQCD.
    Unfortunately, relatively little is known about the relationship of PM to the most serious
    health endpoints, neonatal and infant mortality, emergency hospital admissions and
    mortality in older children, as well as low birth weight and preterm birth.

    Little is yet known about involvement of PM exposure in the progression from less serious
    childhood conditions, such as asthma and respiratory symptoms,  to more serious disease
    endpoints later in life. This is an important health issue because childhood illness or death
    may cost a very large number of productive life-years.

    Lastly, new epidemiologic studies of ambient PM associations with increased non-hospital
    medical visits (physician visits) and asthma effects suggest likely much larger health effects
    and costs to society due to ambient PM than just those indexed by mortality and/or hospital
    admissions/visits.
June 2003                                 E-28       DRAFT-DO NOT QUOTE OR CITE

-------
1      E.6   ENVIRONMENTAL EFFECTS OF PM
2      E.6.1  Vegetation and Ecosystem Effects
3      •  Deposition of particulate matter from the atmosphere has the potential to alter ecosystem
          structure and function. Human existence on this planet depends on the essential life-support
          services that ecosystem structure and functions provide. Concern has risen in recent years
          regarding the consequences of changing the biological diversity of ecosystems because
          human activities are creating disturbances that are altering the structure (complexity and
          stability) and functioning (producing changes in the processes of energy and water flow and
          nutrient cycling) of ecosystems.

4      •  Human-induced changes in biotic diversity and alterations in the structure and ecosystems
          processes are the two most dramatic ecological trends in the past century. Biodiversity
          encompasses all levels of biological organization, including individuals, populations, species,
          and ecosystems. For this reason, there is a need to understand the effects of PM deposition
          on vegetation and  ecosystems and biodiversity.

5      •  Ecosystem functions maintain clean water, pure air, biodiversity, and impart the following
          benefits: fixation of solar energy, absorption and breakdown of pollutants, cycling of
          nutrients, binding of soil, degradation of organic wastes, maintenance of a balance of
          atmospheric gases, regulation of radiation balance, and climate.

6      •  The deposition of PM onto vegetation and soil, depending on its chemical composition
          (acid/base, trace metal, or nutrients, e.g., nitrates or sulfates), can produce direct or indirect
          responses within an ecosystem.

7      •  Ecosystem response to pollutant deposition is a direct function of the level of sensitivity of
          the ecosystem and its ability to  ameliorate resulting changes. Changes in ecosystem
          structural patterns  and the functioning of ecological processes must be scaled in both time
          and space  and propagated to the more complex levels of community interaction to produce
          observable ecosystem changes.
       June 2003                                 E-29        DRAFT-DO NOT QUOTE OR CITE

-------
   The major effect of atmospheric PM on ecosystems is indirect and occurs through the
   deposition of nitrates and sulfates and the acidifying effects of the associated H+ ion in wet
   and dry deposition in the soil environment.  In the soil, they have the potential to inhibit
   nutrient uptake, alter the ecological processes of energy flow and nutrient cycling, change
   ecosystem structure, and affect ecosystem biodiversity.  Most nitrate is not deposited or
   measured as PM but is a combination of wet and dry deposition.

   The soil environment, one of the most dynamic sites of biological interaction, is inhabited by
   communities of bacteria, fungi, and actinomycetes.  Bacteria, as essential components of the
   nitrogen and sulfur cycles, and fungi in the uptake of mineral nutrients play an important role
   in plant nutrition. Their actions make these elements available for plant uptake. Fungi form
   mycorrhizae, a mutualistic symbiotic relationship with plant roots that is integral to the
   uptake of mineral nutrients.  Changes in the soil environment resulting from deposition of
   nitrates and  sulfates that alter the role of the bacteria in the nutrient cycles and mycorrhizal
   fungi in making minerals available for plant utilization, determine plant and, ultimately,
   ecosystem response.

   Although nitrogen as molecular nitrogen (N2) is the most abundant element in the
   atmosphere, it only becomes available to support plant growth after its conversion into
   reactive forms. In nature, nitrogen may be divided into two groups:  reactive (Nr) and
   nonreactive  (N2). Reactive nitrogen or Nr includes the inorganic reduced forms of nitrogen
   (e.g.,  ammonia [NH3]  and ammonium [NH4+]), inorganic oxidized forms (e.g., nitrogen
   oxides [NOJ, nitric acid [HNO3], nitrous  oxide [N2O], and nitrate [NO3"]), and organic
   compounds (e.g., urea, amine, proteins, and nucleic acids).

   Reactive nitrogen can be widely dispersed and accumulate in the environment when the rates
   of its  formation exceed the rates of removal via denitrification.  As a result of human food
   production, it is now accumulating in the  environment on all spatial scales - local, regional
   and global; and its creation and accumulation is projected to increase as per capita use of
   resources by human populations increases.
June 2003                                 E-30        DRAFT-DO NOT QUOTE OR CITE

-------
1      •  The major changes in the nitrogen cycle can be both beneficial and detrimental to the health
          and welfare of humans and ecosystems. The cascade of environmental effects resulting from
          reactive nitrogen accumulation include the following:  (1) production of tropospheric ozone
          and aerosols that induce human health and ecosystem problems; (2) increases in the
          productivity in forests and grasslands followed by decreases wherever Nr deposition
          increases significantly and exceeds critical thresholds; (3) reactive nitrogen additions
          probably decrease biodiversity in many natural habitats; (4) in association with sulfur is
          responsible for acidification and loss of biodiversity in lakes and streams in many regions of
          the world; (5) eutrophication, hypoxia, loss of biodiversity, and habitat degradation in coastal
          ecosystems.  [Eutrophication is now considered the biggest pollution problem in coastal
          waters.], and (6) contributes to global climate change and stratospheric ozone depletion,
          which can in turn affect ecosystems and human health.

2      •  "Nitrogen saturation" results when reactive nitrogen concentrations exceed the capacity of a
          system to utilize it.  Saturation implies that some resource other than nitrogen is limiting
          biotic function. Water and phosphorus for plants and carbon for microorganisms are most
          likely to be the secondary limiting factors.  The appearance of nitrogen in soil solution is an
          early symptom of excess nitrogen.  In the final stage, disruption of ecosystem structure
          becomes visible.

3      •  Possible ecosystem responses to nitrogen saturation include (1) a permanent increase in foliar
          nitrogen and reduced foliar phosphorus and lignin caused by the lower availability of carbon,
          phosphorus,  and water; (2) reduced productivity in conifer stands because of disruptions of
          physiological function; (3) decreased  root biomass and increased nitrification and nitrate
          leaching; (4) reduced soil fertility, resulting from increased cation leaching and increased
          nitrate and aluminum concentrations in streams, (5) decreased water quality,  and (6)  changes
          in  soil microbial communities from predominantly fungal (mycorrhizal) communities to
          those dominated by bacteria during saturation.
       June 2003                                 E-31        DRAFT-DO NOT QUOTE OR CITE

-------
   Intensive research over nearly a decade indicates that, although the soils of most
   North American forests are nitrogen limited, severe symptoms of nitrogen saturation have
   been observed in the high-elevation, non-aggrading spruce-fir ecosystems in the Appalachian
   Mountains; the eastern hardwood watersheds at the Fernow Experimental Forest near
   Parsons, WV; forests in southern California; the southwestern Sierra Nevada in Central
   California; and the Front Range in northern Colorado. The mixed conifer forest and
   chaparral watershed with high smog exposure in the Los Angeles Air Basin exhibit the
   highest stream water NO3" concentrations in wilderness areas of North America.

   Increases in soil nitrogen play a selective role in ecosystems.  Plant succession patterns and
   biodiversity are affected significantly by chronic nitrogen additions in some North American
   ecosystems. Long-term nitrogen fertilization studies in both New England and Europe
   suggest that some forests receiving chronic inputs of nitrogen may decline in productivity
   and experience greater mortality. Declining coniferous forest stands with slow nitrogen
   cycling may be replaced by deciduous fast-growing forests that rapidly cycle nitrogen.

   In experimental studies nitrogen deposition over a  12-year period Minnesota grasslands
   dominated by native warm-season grasses shifted to low-diversity mixtures dominated by
   cool-season grasses.
 •  Excess nitrogen inputs to unmanaged heathlands in the Netherlands has resulted i
   nitrophilous grass species replacing slower growing heath species. Over the past several
   decades the composition of plants in the forest herb layers has been shifting toward speci
   commonly found in nitrogen-rich areas. It also was observed that the fruiting bodies of
   mycorrhizal fungi had decreased in number, indicating that formation of mycorrhizae w
   species
   s of
dzae were
 •  Soil nitrogen enrichment of the soil significantly alters the composition and richness of the
   arbuscular mycorrhizal fungal community and markedly decreases overall diversity of the
   mycorrhizal community. Decline in the coastal sage scrub in southern California was
   directly linked to the decline of arbuscular mycorrhizal fungal community.

June 2003                                 E-32        DRAFT-DO NOT QUOTE OR CITE

-------
   The effects of excessive deposition of nitrogen, particularly NH3 and NH4+, have led to
   changes in Dutch heathlands via:  (1) acidification of the soil and the loss of cations of K+,
   Ca2+ and Mg2+; and (2) nitrogen enrichment, which results in increased plant growth rates
   and altered competitive relationships. Alteration of any of a number of parameters
   (e.g., increased nitrogen) can alter ecosystem structure and function.

   Notable effects of excess nitrogen deposition have been observed with regard to aquatic
   systems.  Atmospheric nitrogen deposition onto soils in watershed areas feeding into
   estuarine sound complexes (e.g., the Pamlico Sound of North Carolina) appear to contribute
   to excess nitrogen flows in runoff (especially during and after heavy rainfall events such as
   hurricanes) from agricultural practices or other uses (e.g., fertilization of lawns or gardens).
   Massive influxes of such nitrogen into watersheds and sounds can lead to algae blooms,
   which, upon degradation, deplete  oxygen and cause extensive fish kills and damage to
   commercial fish and sea food harvesting.

   Acidic precipitation, a critical environmental stress that affects forest landscapes and aquatic
   ecosystems in North America, Europe, and Asia, is linked to the effects associated with the
   deposition of Nr and sulfates and the associated hydrogen ion. The  resulting soil acidity can
   lead to nutrient deficiencies and to high aluminum-to-nutrient ratios that limit uptake of
   calcium and magnesium and create a nutrient deficiency adversely affect tree growth.
   Calcium is essential in the formation of wood and the maintenance of the primary plant
   tissues necessary for tree growth.

   The critical loads concept is useful for estimating the amounts of pollutants (e.g., reactive
   nitrogen and acidic precipitation)  that sensitive ecosystems can absorb on a sustained basis
   without experiencing measurable  degradation.  The estimation of ecosystem critical loads
   requires an understanding of how  an ecosystem will respond to different loading rates  in the
   long term and can be of special value for ecosystems receiving chronic deposition of Nr and
   sulfur independently and as acidic deposition when in combination.
June 2003                                 E-33        DRAFT-DO NOT QUOTE OR CITE

-------
 •  The biological effect of PM is determined by the amount deposited during the wet and dry
   deposition onto the plant surfaces or soil in the vicinity of the roots. The three major routes
   involved are (1) precipitation scavenging in which particles are deposited in rain and snow;
   (2) occult (fog, cloud water, and mist interception); and (3) dry deposition, a much slower,
   yet more continuous removal to place surfaces.

 •  Few heavy metals (e.g., copper, nickel, zinc) have been documented to have direct
   phytoxicity under field conditions. Ecosystems immediately downwind of major emissions
   sources such as power generating, industrial, or urban complexes can receive locally heavy
   inputs.  Heavy metal  accumulation in the richly organic forest floor where biological activity
   is greatest by affecting litter decomposition present the greatest potential for influencing
   nutrient cycling. The presence of cadmium, copper and nickel have been observed to affect
   the symbiotic activity of fungi, bacteria, and actinomycetes.

 •  Phytochelatins produced by plants as a response to sublethal concentrations of heavy metals,
   are indicators of metal stress and can be used to indicate that heavy metals are involved in
   forest decline. Increasing concentrations of phytochelatins with increasing altitude, and their
   increases across regions that show increased levels of forest injury, have implicated them in
   forest decline.

 •  Secondary organics formed in the atmosphere have been variously subsumed under the
   following terms: toxic substances, pesticides, hazardous air pollutants (HAPS), air toxics,
   semivolatile organic compounds (SOCs), and persistent organic pollutants (POPS).  Such
   substances are controlled under CAA Sect. 112,  Hazardous Air Pollutants (not as criteria
   pollutants controlled by NAAQS under CAA Sections 108 and 109).  Their possible effects
   on humans and ecosystems are discussed in many other government documents  and
   publications.  They are noted in this document because, in the atmosphere, many of the
   chemical compounds are partitioned between gas and particle phases and are deposited as
   particulate matter.  As particles, they become airborne and can be distributed over a wide
   area and affect remote ecosystems.  Some are also of concern to humans because they may
   reach toxic levels in food chains of both animals and humans; whereas others may tend to
   maintain or decrease  some toxicity as they move through the food chain.

June 2003                                E-34        DRAFT-DO NOT QUOTE OR CITE

-------
 1      E.6.2  Visibility
 2           Chapter 4 of this document includes information supplementary to several other significant
 3      reviews of the science of visibility, including the 1991 report of the National Acid Precipitation
 4      Assessment Program, the National Research Council' s Protecting Visibility in National Parks
 5      and Wilderness Areas (1993), and EPA's 1995 Interim Findings on the Status of Visibility
 6      Research.  The following points are made in Chapter 4 and/or in the above referenced
 7      documents.
 8      • The relationships between air quality and visibility are well understood.  Ambient fine
          particles are the major cause of visibility impairment.  Significant scientific evidence exists
          showing that reducing fine particle concentrations will improve visibility.

 9      • The National Research Council defines visibility qualitatively as "the degree to which the
          atmosphere is transparent to visible light." This definition may be expressed quantitatively in
          terms of contrast transmittance. The EPA has defined visibility impairment as a reduction in
          visual range (the farthest distance at which a large black object can be distinquished against
          the horizontal sky is the visual range) and/or atmospheric discoloration.

10      • Light, as it passes through the atmosphere from a scene to an observer, is both scattered and
          absorbed.  The rate of loss of transmitted light intensity with distance is measured by the
          light-extinction coefficient, which may be expressed as the sum of the coefficients for:
          (a) light scattering due to gases; (b) light scattering due to particles; (c) light absorption by
          gases, and; (d) light absorption by particles.  Light scattering by particles is the major
          component of light extinction. Light absorption by gases is almost entirely due to NO2, and is
          typically significant only near NO2 sources.  Light absorption by particles is primarily caused
          by elemental carbon.

11      • Light scattering efficiency depends on particle size, falling off rapidly for particles below
          0.3 or above 1.0 jim in diameter.  Therefore, particles in the accumulation mode (of the fine
          particle mode) are most effective in scattering light and are more important in visibility
          degradation than either nuclei-mode or coarse-mode particles.  Light absorption  is not a
          strong function of particle size. Under exceptional circumstances, such as dust storms, coarse
          particles can dominate scattering.

        June 2003                                  E-3 5         DRAFT-DO NOT QUOTE OR CITE

-------
 • In addition to reducing the intensity of light carrying information about a scene (transmitted
   radiance), particles also scatter light into the observer's view. This extraneous light, called
   air light or path radiance, carries no information about the scene. The competition between
   these two sources of light, expressed as the ratio of transmitted radiance from the scene to
   path radiance, determines the contrast transmittance and the visual quality of the view.

 • Visibility at any location is affected by air quality and non-air quality related effects.  The
   visibility effects of atmospheric constituents are dependant upon not just the mass of
   pollutants, but on the size distribution and refractive index of particles, which are strongly
   influenced by relative humidity.  Non-air quality effects include the angle between the sun
   and the observer's sight path, location of clouds,  and reflectivity of the ground. These effects
   are independent of effects due to changes in atmospheric constituents. Lighting and scene
   effects can be accounted for by defining a range of these effects when estimating visibility
   changes due to air quality influences.

 • The relationship between air pollution and the appearance of a scenic view is well
   understood. Models exist that, given an adequate description of the air quality and non-air
   quality variables,  can produce a simulated photograph that accurately depicts a cloud-free
   scene as it would  appear to a human observer.

 • There are several  potential quantitative indicators of visibility, including: (a) fine particle
   mass and composition (fine particle mass alone provides less of both types of information);
   (b) scattering by dried ambient particles; (c) scattering by particles under ambient conditions;
   (d) extinction (calculated from measurements of scattering plus absorption); (e) light
   extinction measured directly; and (f)  contrast transmittance.

 • A new index, the  deciview (dv), is now being used as a quantitative measure of haziness. It is
   related to the light extinction coefficient, bext, by Haziness (dv) = JO ln(bex/10 Mn).  The
   deciview is more  nearly linearly related to perceived changes in haze level than either visual
   range or light extinction. A change of 1 or 2 dv in uniform haze under many viewing
   conditions will be seen as a small but noticeable change in the appearance of a scene
   regardless of the initial haze condition.


June 2003                                  E-36         DRAFT-DO NOT QUOTE OR CITE

-------
 1      • Visibility in the United States is best in the western, intermountain region.  Visibility
          impairment or haziness is greatest in the eastern United States and southern California.
          Haziness in the eastern United States is caused primarily by atmospheric sulfate. Haziness in
          southern California is primarily caused by nitrate and organic PM. Nitrates contribute about
          40% to the total light extinction in southern California. Nitrates account for 10 to 20% of the
          total extinction in other areas of the United States.

 2      • Organics are the second largest contributors to light extinction in most areas in the United
          States. Organic carbon is the greatest cause of light extinction in the Pacific Northwest,
          Oregon,  Idaho, and Montana, accounting for 40 to 45% of the total extinction. Organic
          carbon contributes between 15 to 20% to the total extinction in most of the western United
          States and 20 to 30% in the remaining  areas of the United States.

 3      • Light absorption by carbon is relatively insignificant but is highest in the Pacific Northwest
          (up to 15%) and in the eastern United States (3%).

 4      • High dust concentrations transported from southern California and the subtropics have
          contributed to regional haze in the Grand Canyon and other Class I areas in the southwestern
          United States.

 5
 6     E.6.3  Particulate Matter-Related Effects on Materials
 7           Atmospheric PM and SO2 exert effects on materials that are related both to aesthetic appeal
 8     and physical damage. Studies have demonstrated particles, primarily consisting of carbonaceous
 9     compounds, cause soiling of commonly used building materials and culturally important items
10     such as statues and works of art. Physical damage from the dry deposition of SO2, particles, and
11     the absorption or adsorption of corrosive agents on deposited particles can also result in the
12     acceleration of the weathering of manmade building and naturally occurring cultural materials.
13      • The natural process of metal corrosion from exposure to environmental elements (wind,
          moisture, sun, temperature fluctuations, etc.) is enhanced by exposure to anthropogenic
          pollutants, in particular SO2.
       June 2003                                 E-37        DRAFT-DO NOT QUOTE OR CITE

-------
          Dry deposition of SO2 enhances the effects of environmental elements on calcereous stones
          (limestone, marble, and carbonated cemented) by converting the calcium carbonate (calcite)
          in the stone to calcium sulphate dihydrate (gypsum). The rate of deterioration is determined
          by the SO2 concentration, the stone's permeability and moisture content, and the deposition
          rate.

          Sulfur dioxide limits the life expectancy of paints by causing discoloration, loss of gloss, and
          loss of thickness of the paint film layer.

          A significant detrimental effect of particulate pollution is the soiling of painted surfaces and
          other building materials. Soiling is a degradation process requiring remediation by cleaning
          or washing, and depending on the soiled surface, repainting. Soiling decreases the
          reflectance of a material and reduces the transmission of light through transparent materials.
          Soiling may reduce the life usefulness of the material soiled.
5      E.6.4  Effects of Atmospheric Particulate Matter on Global Warming
6             Processes and Transmission of Solar Ultraviolet Radiation
7      •  The physical processes (i.e., scattering and absorption) responsible for airborne particle
          effects on transmission of solar visible and ultraviolet radiation are the same as those
          responsible for visibility degradation.  Scattering of solar radiation back to space and
          absorption of solar radiation determine the effects of an aerosol layer on solar radiation.

8      •  Atmospheric particles greatly complicate projections of future trends in global warming
          processes because of emissions of greenhouse gases; consequent increases in global mean
          temperature; resulting changes in regional and local weather patterns; and mainly deleterious
          (but some beneficial) location-specific human health and environmental effects.  The body of
          available evidence, ranging from satellite to in situ measurements of aerosol effects on
          radiation receipts and cloud properties, is strongly indicative of an important role in climate
          for aerosols.  This role, however, is poorly quantified. No significant advances have been
          made in reducing the uncertainties assigned to forcing estimates provided by the IPCC for
          aerosol-related forcing, especially for black carbon-containing aerosol.  The IPCC

       June 2003                                 E-3 8       DRAFT-DO NOT QUOTE OR CITE

-------
           characterizes the scientific understanding of greenhouse gas-related forcing as "high" in
           contrast to that for aerosol, which it describes as "low" to "very low."

 9      •  Quantification of the effect of anthropogenic aerosol on hydrological cycles requires more
           information than is presently available regarding ecosystems responses to reduced solar
           radiation and other changes occurring in the climate system. However, several global scale
           studies indicate that aerosol cooling alone can slow down the hydrological cycle, while
           cooling plus the nucleation of additional cloud droplets can dramatically reduce precipitation
           rates.

10      •  In addition to direct climate effects through the scattering and absorption of solar radiation,
           particles also exert indirect effects on climate by serving as cloud condensation nuclei, thus
           affecting the abundance and vertical distribution of clouds.  The direct and indirect effects of
           particles appear to have significantly offset global warming effects caused by the buildup of
           greenhouse gases on a globally-averaged basis. However, because the lifetime of particles is
           much shorter than that required for complete mixing within the Northern Hemisphere, the
           climate effects of particles generally are felt much less homogeneously than are the effects of
           long-lived greenhouse gases.

11      •  Any effort to model the impacts of local alterations in particle concentrations on projected
           global climate change or consequent local and regional weather patterns  would be subject to
           considerable uncertainty.

12      •  Atmospheric particles also complicate estimation of potential future impacts on human health
           and the environment projected as possible to occur because of increased  transmission of solar
           ultraviolet radiation (UV-B) through the Earth's atmosphere, secondary to stratospheric
           ozone depletion due to anthropogenic emissions of chlorofluorcarbons (CFCs), halons, and
           certain other gases.  The transmission of solar UV-B radiation is affected strongly by
           atmospheric particles.  Measured attenuations of UV-B under hazy conditions range up to
           37% of the incoming solar radiation. Measurements relating variations in PM mass directly
           to UV-B transmission are lacking.
        June 2003                                 E-39        DRAFT-DO NOT QUOTE OR CITE

-------
1      •  Particles also can affect the rates of photochemical reactions occurring in the atmosphere,
          e.g., those involved in catalyzing tropospheric ozone formation. Depending on the amount of
          absorbing substances in the particles, photolysis rates either can be increased or decreased.
          Thus, atmospheric particle effects on UV-B radiation, which vary depending on size and
          composition of particles, can differ substantially over different geographic areas and from
          season to season over the same area.  Any projection of effects of location-specific airborne
          PM alterations on increased atmospheric transmission of solar UV radiation (and associated
          potential human health or environmental effects) due to stratospheric ozone-depletion would,
          therefore, also be subject to considerable uncertainty.

2
3      E.7   KEY CONCLUSIONS
4      •  Epidemiologic studies show consistent positive associations of exposure to ambient PM with
          health effects, including mortality and morbidity. The observed associations of ambient PM
          exposure with health effects must take into account the effects of other environmental or
          demographic factors, depending on whether the effects are acute or chronic. Effect-size
          estimates for PM-attributable excess relative risk for a given health outcome vary somewhat
          depending on specific analytic models used, but nevertheless have shown reasonable
          quantitative consistency across different studies.

5      •  Issues surrounding potential bias in PM risk estimates from time-series studies using
          generalized additive models analyses and default convergence criteria were recently raised.
          Reanalyses of some important  studies were included in an HEI Special Report (2003) that
          generally confirmed that PM effect estimates generally decline somewhat when using more
          stringent convergence criteria,  as compared to the default GAM analyses, with the new
          estimates being well within confidence intervals of the original estimates.  Overall, the
          absolute effect was relatively small, and the basic direction of effect and conclusions
          regarding the significance of the PM effect on mortality and hospital admissions remained
          unchanged in these analyses when the GAM convergence requirement was made more
          stringent. A modeling issue of particular importance highlighted by HEI (2003) is the
       June 2003                                 E-40        DRAFT-DO NOT QUOTE OR CITE

-------
           sensitivity of all models to the degrees of freedom allotted to potentially confounding weather
           variables and time. The HEI panel recommended further exploration of the sensitivity of
           these studies to a wider range of alternative degrees of smoothing and to alternative
           specifications of weather variables in time-series models.

 6       •  Individuals with cardiovascular or pulmonary disease, especially if they are elderly, are more
           likely to suffer severe health effects (mortality or hospitalization) related to PM exposure
           than are healthy young adults. Children and asthmatics are also susceptible to certain PM
           effects, e.g., increased respiratory symptoms and decreased lung  function.  Smokers also
           constitute a population group at increased risk for ambient PM exposure effects.

 7       •  Daily ambient fine particle concentrations measured at a community air-monitoring site are
           useful surrogates for daily personal exposures to ambient fine particles. This is consistent,
           for example, with observed high correlations of personal sulfate exposures with ambient
           sulfate concentrations. The relationship between personal exposure to thoracic coarse
           particles and the ambient concentration of thoracic coarse fraction particles is not as strong,
           making effects due to coarse fraction particles harder to detect in epidemiologic studies.

 8       •  Development of a comprehensive biologically based exposure-dose-response model to aid
           health risk assessment requires further dosimetry data characterizing differences among
           species in percent deposition and regional deposition patterns including differences in
           inhalability, airway geometry, and clearance rates. More information is also needed on
           mechanism(s) of clearance, pathological processes affecting deposition and clearance of
           particles, and factors influencing the response(s) of respiratory tract tissues to particle burden.

 9       •  The percent deposition and regional patterns of deposition depend strongly on particle size.
           Percent deposition is higher in smaller lungs (children; women),  during exercise, and in the
           functioning parts of the lungs in people with compromised lungs (e.g., those with chronic
           obstructive lung disease).

10       •  Estimation of public health impacts of ambient airborne particle exposures in the United
           States would most credibly combine information on exposure-response relationships derived

        June 2003                                 E-41        DRAFT-DO NOT QUOTE OR CITE

-------
          for particular U.S. urban areas, in combination with estimates of exposures to ambient
          particle concentrations for the general population and/or specific susceptible subgroups (e.g.,
          the elderly) those particular areas.  At this time, risk assessment is necessarily limited to use
          of available information from concentration-response relationships relating ambient
          concentrations to health effects in populations.  In view of geographic differences in ambient
          PM mixtures and demographics, broad generalization of some single "best estimate"  of
          relative risk for a given increment in concentration of a given particle indicator (e.g., PM10,
          PM25, etc.) would be subject to much uncertainty.

11       • Toxicological studies of PM using controlled inhalation exposure of humans and laboratory
          animals, intratracheal instillation in humans and animals, and exposure of human and animal
          cells  grown in culture find numerous biological effects which may be related to adverse
          health effects. Newer studies are finding different biological effects for a variety of different
          particle components.  Newer studies also are beginning to identify biological mechanisms
          whereby PM deposited  in the lung can produce adverse effects on the cardiovascular and
          respiratory systems.

12       • Epidemiologic studies indicate increased health risks associated with exposure to PM, alone
          or in  combination with other air pollutants.  PM-related increases in individual health risks
          are small, but likely significant from an overall public health perspective because of the large
          numbers of individuals  in susceptible risk groups that are exposed to ambient PM.

13       • Numerous new studies, including multicity studies, continue to show a consistent association
          of PM10 exposure with mortality and various morbidity  endpoints, thus substantiating the
          relationship of PM exposure with various health effects. However, some new studies using
          PM2 5 as an indicator find higher statistical significance and higher excess risk for PM2 5
          compared to PM10.  Several studies have also observed statistically associations of PM10_2 5
          with  health effects.

14       • Epidemiologic studies, in which factors identified with  source categories or individual
          chemical components of PM have been used as indicators, also show significant associations
        June 2003                                 E-42        DRAFT-DO NOT QUOTE OR CITE

-------
          with nitrates, sulfates, various indicators of elemental carbon, the organic component of PM,
          and some elements.  PM-related source category factors, such as regional sulfate, vehicular
          traffic, fossil fuel combustion, vegetative combustion, and oil burning have also been found
          to be significantly associated with mortality.

15       • Data from multicity  studies, comparisons of effects found in single and multiple regressions,
          exposure relationships, and toxicity suggest that the gaseous co-pollutants (CO, NO3, SO2,O3)
          are not responsible for the relationships observed with PM indicators in community, time-
          series epidemiology. This does not, however, necessarily imply lack of an independent
          association of gaseous co-pollutants with health effects.

16       • Fine and thoracic coarse PM, indexed respectively by PM25 and PM10_25, should be
          considered as separate subclasses of PM. Considerations of emissions sources, atmospheric
          chemistry, physical behavior, exposure relationships, respiratory deposition, toxicologic
          findings, and epidemiologic observations argue for monitoring fine and  thoracic coarse
          particles separately.

17       • Assessment of health risk in epidemiologic studies of ambient air pollutants, including PM,
          has relied largely on studies that focus on changes in health risks that occur in relation to
          normal changes in ambient air pollutant concentrations.  Further evidence of the effects of air
          pollution on health may be deduced from intervention studies, i.e,  studies of changes in
          health effects that occur when air pollution concentrations have been temporarily or
          permanently reduced through regulatory action, industrial shutdown, or other intervening
          factor(s). Only a few epidemiologic intervention studies are available; however, taken
          together, these studies lend confidence that further reduction of ambient air pollution
          exposures in the U.S. would reduce both respiratory and cardiovascular health effects.
          Available studies also give reason to expect that further reductions in both paniculate and
          gaseous  air pollutants would benefit health. Furthermore, experimental  studies of Utah
          Valley filter extracts points to PM-associated metals as a likely cause or promoter of at least
          some of the health disorders associated with ambient PM.
        June 2003                                 E-43        DRAFT-DO NOT QUOTE OR CITE

-------
1       • The overall weight of evidence, based on information concerning PM exposure, dosimetry,
         toxicology, and epidemiology, supports the conclusion that PM, especially fine PM, is the
         primary contributor to a variety of adverse health effects associated with air pollution.
         However, difficult technical issues still remain in further separating the effects of fine and
         coarse particles and in delineating respective contributions of PM acting along or in
         combination with gaseous co-pollutants in increasing risks of health effects anticipated to
         occur in response to exposures to contemporary particle-containing ambient air mixes in the
         United States.
      June 2003                                  E-44        DRAFT-DO NOT QUOTE OR CITE

-------
  1       REFERENCES
  2
  3       Brunekreef, B. (1997) Air pollution and life expectancy: is there a relation? Occup. Environ. Med. 54: 781-784.
  4       Health Effects Institute. (2003) Revised analyses of time-series studies of air pollution and health. Boston, MA:
  5             Health Effects Institute; special report. Available: http://www.healtheffects.org/Pubs/TimeSeries.pdf
  6             [27 June 2003].
  7       Intergovernmental Panel on Climate Change (IPCC). (200la) Climate change 2001: the scientific basis.
  8             Contribution of working group I to the third assessment report of the Intergovernmental Panel on Climate
  9             Change. Cambridge, United Kingdom: Cambridge University Press.
10       Intergovernmental Panel on Climate Change (IPCC). (200 Ib) Climate change 2001: impacts, adaptation, and
11             vulnerability. Contribution of working group II to the third assessment report of the Intergovernmental Panel
12             on Climate Change. Cambridge, United Kingdom: Cambridge University Press.
13       National Acid Precipitation Assessment Program. (1991) National Acid Precipitation Assessment Program 1990
14             integrated assessment report. Washington, DC: National Acid Precipitation Assessment Program.
15       National Research Council. (1993) Protecting visibility in national parks and wilderness areas. Washington, DC:
16             National Academy Press. 3v.
17       United Nations Environment Programme (UNEP). (1998) Environmental effects of ozone depletion: 1998
18             assessment. J. Photochem. Photobiol. B 46: 1-4.
19       United Nations Environment Programme (UNEP). (2000) Environmental effects of ozone depletion: interim
20             summary. Available at: http://www.gcrio.org/ozone/unep2000summary.html  [9 April 2002].
21       U.S. Code. (1991) Clean Air Act, §108, air quality criteria and control techniques, §109, national ambient air quality
22             standards. U. S. C. 42: §§7408-7409.
23       U.S. Environmental Protection Agency. (1995) Interim findings on the status of visibility research. Research
24             Triangle Park, NC: Office of Research and Development; report no. EPA/600/R-95/021.
25       U.S. Environmental Protection Agency. (1996) Air quality criteria for paniculate matter. Research Triangle Park,
26             NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
27       World Meteorological Organization (WMO). (1999)  Scientific assessment of ozone depletion: 1998. Geneva,
28             Switzerland: World Meteorological Organization, Global Ozone and Monitoring Project; report no. 44.
         June 2003                                       E-45          DRAFT-DO NOT QUOTE OR CITE

-------
 i                                  1.  INTRODUCTION
 2
 3
 4           This document is an update of "Air Quality Criteria for Particulate Matter" published by
 5     the U.S. Environmental Protection Agency (EPA) in 1996, and it will serve as the basis for
 6     Congress!onally-mandated periodic review of the National Ambient Air Quality Standards
 7     (NAAQS) for particulate matter (PM). The present document critically assesses the latest
 8     scientific information relative to determining the health and welfare effects associated with the
 9     presence of various concentrations of PM in ambient air.  The document builds  upon the
10     assessment in the previous 1996 EPA Air Quality Criteria Document for Particulate Matter (PM
11     AQCD) by focusing on assessment and integration of information most relevant to PM NAAQS
12     criteria development, based on  pertinent peer-reviewed literature published or accepted for
13     publication mainly through April 2002.  This introductory chapter presents a brief summary of
14     legislative requirements  and history of the PM NAAQS, provides an overview of issues
15     addressed and procedures utilized in the preparation of the present document, and provides
16     orientation to the general organizational structure of the document.
17
18
19     1.1   LEGISLATIVE REQUIREMENTS
20           As indicated in U.S. Code (1991), the U.S. Clean Air Act (CAA), Sections 108 and 109
21     (42 U.S.C. Sections 7408 and 7409) govern the  establishment, review, and revision of National
22     Ambient Air Quality Standards (NAAQS). Section 108(a) directs the EPA Administrator to list
23     pollutants, which, in the Administrator's judgement, cause or contribute to air pollution which
24     may reasonably be anticipated to endanger either public health or welfare  and to issue air quality
25     criteria for them.  The air quality criteria are to reflect the latest scientific information useful in
26     indicating the kind and extent of all identifiable  effects on public health and welfare that may be
27     expected from the presence of the pollutant in ambient air.
28           Section 109 directs the Administrator of EPA to propose and promulgate "primary" and
29     "secondary" NAAQS for pollutants identified under Section 108. Section 109(b)(l) defines a
30     primary standard as a level of air quality, the attainment and maintenance  of which, in the
31     judgement of the Administrator, based on the criteria and allowing for an adequate margin of

       June 2003                                 1 -1        DRAFT-DO NOT QUOTE OR CITE

-------
 1      safety, is requisite to protect the public health. Section 109(b)(2) defines a secondary standard as
 2      one which, in the judgement of the Administrator, based on the criteria, is requisite to protect
 3      public welfare from any known or anticipated adverse effects associated with the presence of
 4      such pollutants.  Welfare effects, defined in Section 302(h), include, but are not limited to,
 5      effects on soils, water, crops, vegetation, man-made materials, animals, wildlife, weather,
 6      visibility and climate, damage to and deterioration of property, and hazards to transportation, as
 7      well as effects on economic values and personal comfort and well-being. Section 109(d)(l)
 8      requires periodic review and, as appropriate, revision of existing criteria and standards.
 9      Section 109(d)(2) requires an independent committee of non-EPA experts, the Clean Air
10      Scientific Advisory Committee (CASAC), to provide advice and recommendations to the EPA
11      Administrator regarding the scientific soundness and appropriateness of criteria and NAAQS for
12      PM and other "criteria air pollutants" (i.e., ozone, nitrogen dioxide, sulfur oxides, carbon
13      monoxide, lead) regulated under CAA Sections 108-109.
14
15
16      1.2   HISTORY OF PREVIOUS PM CRITERIA AND NAAQS REVIEWS
17           "Particulate matter" is the generic term for a broad class of physically and chemically
18      diverse substances that exist in ambient air as discrete particles (liquid droplets or solids) over a
19      wide range of sizes.  These airborne particles originate from a variety of stationary and mobile
20      sources.  Primary particles are emitted directly into ambient air; whereas secondary particles are
21      formed in the atmosphere by transformation of gaseous emissions such as sulfur oxides (SOX),
22      nitrogen oxides (NOX), and volatile organic compounds (VOCs).  The physical and chemical
23      properties of PM vary greatly with time, region, meteorology, and source category, thus
24      complicating assessment of ambient PM health and welfare effects. Particles in ambient air are
25      usually distributed bimodally in two somewhat overlapping size categories:  (1) fine (diameter
26      generally less than 2.5 mm) and (2) coarse (diameter generally greater than 1.0 |im). Particles in
27      these two size fractions tend to differ in terms of formation mechanisms, sources  of origin,
28      composition, and behavior in the atmosphere and human respiratory tract.
29           EPA first promulgated primary and secondary NAAQS for PM on April 30, 1971 (Federal
30      Register, 1971).  These standards measured PM as "total suspended particulate" (TSP), which
31      refers to ambient PM up to a nominal size of 25 to 45 micrometers (|im). The primary standards

        June 2003                                  1 -2         DRAFT-DO NOT QUOTE OR CITE

-------
 1     for PM (measured as TSP) were 260 |ig/m3 (24-h average), not to be exceeded more than once
 2     per year, and 75 |ig/m3 (annual geometric mean).  The secondary standard (measured as TSP)
 3     was 150  |ig/m3 (24-h average), not to be exceeded more than once per year.
 4          EPA completed the next review of PM air quality criteria and standards in July 1987,
 5     revising the 1971 standards to protect against adverse health effects of inhalable airborne
 6     particles which can be deposited in the lower (thoracic) regions of the human respiratory tract,
 7     with "PM10" as the indicator (i.e., those particles collected by a sampler with a specified
 8     penetration curve yielding an upper 50% cut-point of 10-|im aerodynamic diameter (Federal
 9     Register, 1987). EPA established identical primary and secondary PM10 standards for two
10     averaging times: 150 |ig/m3 (24-h average), with no more than one expected exceedance per
11     year and 50 |ig/m3 (expected annual arithmetic mean), averaged over three years.
12
13     1.2.1   The 1997 PM NAAQS Revision
14          The last previous review of the air quality criteria and standards for PM was initiated in
15     April 1994 by EPA announcing its intention to develop revised Air Quality Criteria for
16     Particulate Matter.  Several workshops were held by EPA's Environmental Criteria and
17     Assessment Office in Research Triangle Park, NC (ECAO-RTP) in November 1994 and January
18     1995 to discuss important new health effects information useful in preparing initial PM AQCD
19     draft materials.  Also, plans for review of the PM criteria and standards under a highly
20     accelerated, court-ordered schedule were presented by EPA at a public meeting of the CAS AC in
21     December 1994. A court order entered in American Lung Association v. Browner., CIV-93-643-
22     TUC-ACM (U.S. District Court of Arizona, 1995), as subsequently modified, required
23     publication of EPA's final decision on the review of the PM NAAQS by July 19,  1997.
24          Several external review drafts of the revised PM AQCD were prepared by the RTF
25     Division of EPA's newly created National Center for Environmental Assessment  (i.e., by
26     NCEA-RTP, the successor office to ECAO-RTP), and each were made available for public
27     comment followed by CAS AC review (at public meetings held in August 1995, December 1995,
28     and February 1996). The  CAS AC came to closure on its review of the PM AQCD in early 1996,
29     advising the EPA Administrator in a March 15, 1996 closure letter (Wolff, 1996)  that "although
30     our understanding of the health effects of PM is far from complete, a revised Criteria Document
31     which  incorporates the Panel's latest comments will provide an adequate review of the available

       June 2003                                 1 -3         DRAFT-DO NOT QUOTE OR CITE

-------
 1      scientific data and relevant studies of PM." Revisions made in response to public and CASAC
 2      comments were then incorporated by NCEA-RTP, as appropriate, into the final 1996 PM AQCD
 3      (U.S. Environmental Protection Agency, 1996a). A PM Staff Paper (SP), prepared by the Office
 4      of Air Quality Planning and Standards (OAQPS) within EPA's Office of Air and Radiation
 5      (OAR), drew upon the 1996 PM AQCD and other exposure and risk assessments to pose options
 6      for PM NAAQS decisions. Drafts of the PM SP also underwent public comment and CASAC
 7      review, with consequent revision into the final  1996 PM SP (U.S. Environmental Protection
 8      Agency, 1996b).
 9           The 1996 PM AQCD and the 1997 PM SP (U.S. Environmental Protection Agency,
10      1997a,b) provide detailed information on atmospheric formation, ambient concentrations, and
11      health effects of ambient air PM, as well as quantitative estimates of human health risks
12      associated with exposure to ambient PM.  The principal focus of these documents was on recent
13      epidemiologic evidence reporting associations between ambient concentrations of PM and a
14      range of serious health effects.  Special attention was given to several size-specific classes of
15      particles, including PM10 and the principal fractions of PM10, referred to as the fine (PM25) and
16      coarse (PM10_2 5) fractions.  PM2 5 refers to those particles collected by samplers having
17      penetration curves yielding an upper 50% cut point of 2.5 jim aerodynamic diameter. PM10_25
18      refers to those particles in an aggregate sample having an upper 50% cut point of 10 jim and a
19      lower 50% cut point of 2.5 |im aerodynamic diameter. In other words, the coarse fraction
20      (PM10_25) refers to the inhalable particles that remain if fine (PM25) particles are removed from a
21      sample of PM10 particles. As discussed in the 1996 PM AQCD, fine and coarse fraction particles
22      can be differentiated by their sources and formation processes and by their chemical and
23      physical properties, including behavior in the atmosphere.
24           Taking into account information and assessments presented in the 1996 PM  AQCD and
25      PM SP, advice and recommendations of CASAC, and public comments received on proposed
26      revisions to the PM NAAQS published in December 1996 (Federal Register, 1996), the EPA
27      Administrator promulgated significant revisions to the PM NAAQS in July 1997 (Federal
28      Register, 1997).  In that decision, the PM NAAQS were revised in several respects.  While it was
29      determined that the PM NAAQS should continue to focus on particles less than or equal to
30      10 jim in diameter, it was also determined that the fine and coarse fractions of PM10 should be
31      considered separately. New standards were added,  using PM2 5 as the indicator for fine particles,

        June 2003                                  1 -4        DRAFT-DO NOT QUOTE OR CITE

-------
 1      and PM10 standards were retained for the purpose of regulating coarse-fraction particles. Two
 2      new PM25 standards were set:  an annual standard of 15 |ig/m3, based on the 3-year average of
 3      annual arithmetic mean PM2 5 concentrations from single or multiple community-oriented
 4      monitors; and a 24-hour standard of 65 |ig/m3, based on the 3-year average of the 98th percentile
 5      of 24-hour PM2 5 concentrations at each population-oriented monitor within an area.  To continue
 6      to address coarse-fraction particles, the annual PM10 standard was retained, and the form, but not
 7      the level, of the 24-hour PM10 standard was revised  to be based on the 99th percentile of 24-hour
 8      PM10 concentrations at each monitor in an area.  The secondary standards  were revised by
 9      making them identical in all respects to the primary standards.
10           Following promulgation of the revised PM NAAQS, legal challenges were filed by a large
11      number of parties, addressing a broad range of issues. In May 1998, the U.S. Court of Appeals
12      for the District of Columbia Circuit issued an initial opinion that upheld EPA's decision to
13      establish fine particle standards, holding that such standards were amply justified by the growing
14      body of empirical evidence demonstrating a relationship between fine particle pollution and
15      adverse health effects. American Trucking Associations v. Browner. 175  F. 3d 1027, 1055-56
16      (D.C. Cir. 1999) (rehearing granted in part and denied in part, 195 F. 3d 4 (D.C. Cir. 1999b),
17      affirmed in part and reversed in part, Whitman v. American Trucking Associations. 531 U.S. 457
18      (2001). Further, the court found  "ample support" for EPA's decision to regulate coarse fraction
19      particles, although it vacated the  revisions to the 1987 PM10 standards on the basis  of PM10 being
20      a "poorly matched indicator for coarse particulate pollution" because PM10 includes fine
21      particles. Id. at 1053-55.  As a result of this aspect of the court's ruling, which EPA did not
22      appeal, the 1987 PM10 standards  remain in effect.
23           In addition, the court broadly held that EPA's approach to establishing the level of the
24      standards in its 1997 decisions on both the PM and ozone NAAQS (which were promulgated on
25      the same day and considered together by the court in this aspect of its opinion) effected "an
26      unconstitutional delegation of legislative authority." Id. at 1034-40.  EPA appealed this aspect of
27      the court's ruling to the United States Supreme Court.  In February 2001, the U.S.  Supreme
28      Court unanimously reversed the Court of Appeals' ruling on the constitutional issue, and sent the
29      case back to the Court of Appeals for resolution  of any remaining issues that had not been
30      addressed in that court's earlier rulings. Whitman v. American Trucking Associations. 531 U.S.
31      457, 475-76 (2001). In March 2002, the Court of Appeals rejected all remaining challenges to

        June 2003                                 1 -5        DRAFT-DO NOT QUOTE OR CITE

-------
 1     the standards, finding that the 1997 PM2 5 standards were reasonably supported by the record and
 2     were not "arbitrary or capricious." American Trucking Associations v. EPA. 283 F. 3d 355,
 3     369-72 (D.C. Cir. 2002). Thus, the 1997 PM25 standards are in effect.
 4
 5     1.2.2   Coordinated Particulate Matter Research Program
 6          Shortly after promulgation of the 1997 PM NAAQS decisions, NCEA-RTP published a
 7     PM Health Risk Research Needs Document (U.S. Environmental Protection Agency, 1998a) that
 8     identified research needed to improve scientific information supporting future reviews of the PM
 9     NAAQS.  The document provided a foundation for PM research coordination among Federal
10     agencies and other research organizations and also provided input to later National Research
11     Council (NRC) deliberations on PM research. The Office of Research and Development (ORD)
12     of EPA also moved quickly to broaden its ongoing PM research activities by developing, in
13     partnership with other Federal agencies, a coordinated interagency PM research program. This
14     interagency program has and continues to focus mainly on expanding scientific knowledge of
15     ambient PM exposure and health effects, as well as including development of improved
16     monitoring methods and cost-effective mitigation strategies. The interagency effort also
17     promotes substantially expanded coordination with other research organizations, including the
18     Health Effects Institute (HEI) and other  state-, university-, and industry-sponsored research
19     groups. Beginning in the fall of 1997, public participation was and continues to be encouraged
20     through workshops and review of EPA's PM Research Program documentation.
21          In response to Congressional requirements in EPA's Fiscal Year 1998  Appropriation, the
22     NRC established its Committee on Research Priorities for Airborne Particulate Matter in January
23     1998.  This NRC PM Research Committee's charge is to identify the most important research
24     priorities relevant to setting PM NAAQS, to develop a conceptual plan for PM research, and to
25     monitor EPA's research progress toward improved understanding of the relationship between
26     PM and public health. The  Committee issued its first report in early 1998 (National Research
27     Council, 1998), a second one in  1999 (National Research Council, 1999), and a third one in 2001
28     (National Research Council, 2001). In the above-noted series of reports, the NRC PM Research
29     Committee recommended that expanded PM research efforts be planned and carried out in
30     relation to a general conceptual framework as shown in Figure 1-1. That framework essentially
31     calls for research aimed at:  (a) identifying sources of airborne particles or gaseous precursor

       June 2003                                1 -6         DRAFT-DO NOT QUOTE OR CITE

-------
         Sources of Airborne
         Particulate Matter or
          Gaseous Precursor
             Emissions
Indicator in Ambient
  (Outdoor) Air
    (e.g. Mass
  Concentration)
                 Mechanisms determining emissions,
                 chemical transformation (including
                formation of secondary particles from
                    gaseous precursors), and
                       transport in air
            Human time-activity
            patterns, indoor (or
            microenvironmental)
            sources and sinks of
             parti cul ate matter
Personal
Exposure

fc.
A
W
k
Dose to
Target
Tissues

   Deposition,
clearance, retention
 and disposition of
 parti dilate matter
   presented to
   an individual
 Mechanisms of
damage and repair
        Figure 1-1. A general framework for integrating particulate-matter research.  Note that
                    this figure is not intended to represent a framework for research management.
                    Such a framework would include multiple pathways for the flow of
                    information.
        Source: National Research Council (2001).
 1      emissions and characterization of processes involved in atmospheric transformation, transport,
 2      and fate of ambient PM; (b) delineation of temporal and spatial patterns of air quality indicators
 3      (e.g., PM2 5, PM10_2 5, PM10 mass concentrations) of ambient PM and apportionment of observed
 4      variations in such ambient PM indicators to various emission sources; (c) characterization of
 5      human exposures to ambient PM as one important component of total personal exposure to
 6      particles, as modified by time-activity patterns and varying microenvironmental exposure to
 7      particles of indoor or ambient origin;  (d) characterization of resulting respiratory tract
 8      deposition, clearance, retention, and disposition of inhaled particles, as determinants of dose to
 9      target tissues (e.g., locally in the lungs or via systemic translocation to the heart or other organs);
10      (e) delineation of mechanisms of damage and repair plausibly leading to (f) human health
11      responses, as extrapolated from  or quantified by experimental animal or human exposure
12      (toxicology) studies and/or observational (epidemiology) studies.
13           Research conducted under a PM Research Program structured in relation to the conceptual
14      framework shown in Figure 1-1  would be expected (a) to reduce key  scientific uncertainties
15      regarding interrelationships between PM sources, ambient concentrations, exposures, dose to
16      target tissues,  and resulting health effects and (b) thereby improve the scientific underpinnings
17      for both current and future periodic PM NAAQS reviews.  Table 1-1  highlights some types  of
18      key uncertainties identified by the NRC PM Research Committee in relation to elements of the
        June 2003
                   1-7
 DRAFT-DO NOT QUOTE OR CITE

-------
         TABLE 1-1.  KEY SCIENTIFIC UNCERTAINTIES RELATED TO THE
                        SOURCE-TO-RESPONSE FRAMEWORK

 Source                ^            Concentration (or other indicator)

     • Contribution of various emission sources to ambient and indoor particulate matter concentrations

     • Relative contribution of various sources to the most toxic components of particulate matter

 Concentration (indicator)                    ^        Exposure

     • Relationship between ambient (indoor) particulate matter and the composition of particles to
      which people are exposed

     • Contribution of ambient particulate matter to total personal exposure for:

       - Susceptible subpopulations
       - General population

     • Variation in relationship of ambient particulate matter concentrations to human exposure by place

     • Variation in contribution of ambient particulate matter concentrations to total human exposure
      over time

     • Covariance of particulate matter exposures with exposures to other pollutants

     • Relationships between outdoor ambient and personal exposures for particulate matter and co-
      pollutants

 Exposure              ^        Dose

     • Relationship between inhaled concentration and dose of particulate matter and constituents at the
      tissue level in susceptible subjects

       - Asthma
       - Chronic obstructive pulmonary disease (COPD)
       - Heart disease
       - Age: infants and elderly
       - Others

 Dose                  ^        Response

     • Mechanisms linking morbidity and mortality to particulate matter dose to or via the lungs

       - Inflammation
       - Host defenses
       - Neural mechanisms

 Source: National Research Council (2001).
June 2003                                    1 -8         DRAFT-DO NOT QUOTE OR CITE

-------
 1     source-to-response conceptual framework illustrated in Figure 1-1.  The NRC Committee went
 2     on to delineate a series of 10 research topics that they recommended be addressed in an
 3     expanded PM research program aimed at answering a set of broadly stated questions, as shown
 4     in Table 1-2.
 5          The EPA PM Research Program, structured to address topics shown in Table 1-2 includes,
 6     for example, studies to improve understanding of the formation and composition of fine PM,
 7     improved measurements and estimation of population exposures to ambient PM, the
 8     characteristics or components of PM that are responsible for its health effects, factors increasing
 9     susceptibility to PM effects in some subpopulations, and mechanisms by which these effects are
10     produced. The results from these efforts, and related efforts by other Federal agencies and the
11     general scientific community  during the past several years, have substantially enhanced the
12     scientific and technical bases  for future decisions on the PM NAAQS.
13
14
15     1.3   CURRENT  PM CRITERIA AND NAAQS REVIEW
16     1.3.1   Key Milestones
17          As with other NAAQS reviews, rigorous assessment of relevant scientific information is
18     presented in this updated, revised PM AQCD. As shown in Table 1-3, development of the
19     document has involved substantial external peer review through (a) public workshops involving
20     the general aerosol scientific community, (b) iterative reviews of successive drafts by CASAC,
21     and (c) comments from the public. The final document will reflect input received through these
22     reviews and will serve to evaluate and integrate the latest available scientific information to
23     ensure that the review of the PM standards is based on rigorous evaluation of the available
24     science.
25          An earlier (October 1999) First External Review Draft of this updated document was
26     released in the fall of 1999 for public comment and CASAC review. A Second External Review
27     Draft (March 2001) took into account earlier public comments and the December 1999 CASAC
28     review and was reviewed by CASAC in July 2001. A Third External Review Draft similarly
29     took into account prior public comments and CASAC recommendations and was released in
30     early May 2002 for a  60-day public comment period; CASAC reviewed it at a public meeting in
31     July 2002.

       June 2003                                1 -9        DRAFT-DO NOT QUOTE OR CITE

-------
      TABLE 1-2. RESEARCH TOPICS AND QUESTIONS RECOMMENDED BY
   NATIONAL RESEARCH COUNCIL (NRC) TO BE ADDRESSED BY EXPANDED
                                PM RESEARCH PROGRAM

 RESEARCH TOPIC 1.    OUTDOOR MEASURES VERSUS ACTUAL HUMAN EXPOSURES

     • What are the quantitative relationships between concentrations ofparticulate matter and gaseous co-
      pollutants measured at stationary outdoor air monitoring sites and the contributions of these concentrations
      to actual personal exposures, especially for subpopulations and individuals?

 RESEARCH TOPIC 2.    EXPOSURES OF SUSCEPTIBLE SUBPOPULATIONS TO TOXIC
                         PARTICULATE MATTER COMPONENTS

     • What are the exposures to biologically important constituents and specific characteristics ofparticulate
      matter that cause responses in potentially susceptible subpopulations and the general population?

 RESEARCH TOPIC 3.    CHARACTERIZATION OF EMISSION SOURCES

     • What are the size distribution, chemical composition, and mass-emission rates ofparticulate matter emitted
      from the collection of primary-particle sources in the United States, and what are the emissions of reactive
      gases that lead to secondary particle formation through atmospheric chemical reactions?

 RESEARCH TOPIC 4.    AIR-QUALITY MODEL DEVELOPMENT AND  TESTING

     • What are the linkages between emission sources and ambient concentrations of the biologically important
      components ofparticulate matter?

 RESEARCH TOPIC 5.    ASSESSMENT OF HAZARDOUS PARTICULATE MATTER
                         COMPONENTS

     • What is the role ofphysicochemical characteristics ofparticulate matter in eliciting adverse health effects?

 RESEARCH TOPIC 6.    DOSIMETRY: DEPOSITION AND FATE OF PARTICLES IN THE
                         RESPIRATORY TRACT

     • What are the deposition patterns and fate of particles in the respiratory tract of individuals belonging to
      presumed susceptible subpopulations?

 RESEARCH TOPIC 7.    COMBINED EFFECTS OF PARTICULATE MATTER AND GASEOUS
                         POLLUTANTS

     • How can the effects ofparticulate matter be disentangled from the effects of other pollutants? How can the
      effects of long-term exposure to particulate matter and other pollutants be better understood?

 RESEARCH TOPIC 8.    SUSCEPTIBLE SUBPOPULATIONS

     • What subpopulations are at increased risk of adverse health outcomes from particulate matter?

 RESEARCH TOPIC 9.    MECHANISMS OF INJURY

     • What are the underlying mechanisms (local pulmonary and systemic) that can explain the epidemiologic
      findings of mortality/morbidity associated with exposure to ambient particulate matter?

 RESEARCH TOPIC 10.   ANALYSIS AND MEASUREMENT	

     • To what extent does the choice of statistical methods in the analysis of data from epidemiologic studies
      influence estimates of health risks from exposures to particulate matter? Can existing methods be
      improved? What is the effect of measurement error and misclassification on estimates of the association
	between air pollution and health?	

 Source: National Research Council (2001).


June 2003                                   1-10        DRAFT-DO NOT QUOTE OR CITE

-------
          TABLE 1-3. SCHEDULE FOR DEVELOPMENT OF THE CURRENT REVISED
          PARTICULATE MATTER AIR QUALITY CRITERIA DOCUMENT (PM AQCD)
       Major Milestones
            Target Dates
       PM NAAQS Review Plan to CASAC
       Prepare AQCD Development Plan
       Begin Literature Search
       Federal Register Call for Information/Sources Sought
       CASAC Meeting on AQCD Development Plan
       Prepare Workshop Drafts of Chapters
       Peer Review Workshop
       Prepare External Review Draft AQCD
       First External Review Draft
       Public Comment Period on Draft AQCD
       CASAC Meeting on First Draft AQCD
       Second External Review Draft
       Public Comment Period on Second Draft
       CASAC Meeting on Second Draft
       Third External Review Draft
       Public Comment Period on Third Draft
       CASAC Meeting on Third Draft
       Final PM AQCD	
            October 1997
            November 1997 to January 1998
            February 1998
            April 1998
            May 1998
            May to December 1998
            April 1999
            March to September 1999
            October 1999
            October 1999 to January 2000
            December 1999
            March 2001
            April to July 2001
            July 2001
            April 2002
            May to July 2002
            July 2002
            Oct/Nov 2002
1          Shortly after EPA released the Third External Review Draft in May 2002, the Health
2     Effects Institute (HEI) announced that researchers at Johns Hopkins University had discovered
3     problems with applications of statistical software used in a number of important studies on links
4     between ambient air particulate matter (PM) and death and disease. In response to the surfacing
5     of such statistical issues, which affected numerous PM time-series studies that used General
6     Additive Models (GAM) and were published post-1995, EPA took steps in consultation with
7     CASAC to encourage researchers to reanalyze affected studies and to submit them expeditiously
      June 2003
1-11
DRAFT-DO NOT QUOTE OR CITE

-------
 1     for peer review by a special expert panel convened by HEI. The results of reanalyses for more
 2     than 30 studies of likely importance in informing U.S. standard setting decisions for ambient PM
 3     have now been peer-reviewed and are included, along with commentary by the HEI expert peer-
 4     review panel, in a recently released HEI Special Report (Health Effects Institute, May 2003).
 5     The subject statistical issues and reanalyses results are discussed in more detail in Chapter 8 of
 6     this document. After a 60-day public comment period, starting in June 2003, and CASAC
 7     review of this Fourth Draft, the final version of this PM AQCD is targeted for issuance by
 8     December 31,2003.
 9          After CASAC review of the First External Review Draft of this revised PM AQCD in
10     December 1999, EPA's OAQPS started to prepare the associated PM Staff Paper (SP).
11     A preliminary draft SP was made available to the public and was discussed at a July 2001
12     CASAC meeting; and additional consultations have been held via teleconference with CASAC
13     to obtain feedback on proposed approaches to be used in conducting analyses to be included in
14     the First External Review Draft of the PM SP due to be released in the summer of 2003. The
15     draft PM SP will draw on the updated findings and conclusions from this draft PM AQCD and
16     will also undergo public comment and CASAC review (now being scheduled for November,
17     2003). Ultimately drawing on information in the final version of this newly revised PM AQCD,
18     the PM SP will evaluate policy implications of the key studies and scientific findings contained
19     in the AQCD, present related staff analyses of air quality and human health risk, and identify
20     critical elements that EPA staff believes should be considered in reviewing the PM standards.
21     The PM SP is intended to bridge the gap between the scientific review in the AQCD and the
22     public health and welfare  policy judgements required of the Administrator in reviewing the PM
23     NAAQS.  In doing so, the PM SP will include staff conclusions and recommendations of options
24     for the Administrator's consideration.
25          Based on the final versions of the PM AQCD and the PM SP, and on the advice of
26     CASAC, the Administrator will propose decisions as to whether to retain or revise the current
27     PM NAAQS and provide  opportunities for public and CASAC comments on the proposed
28     decisions. Taking into account comments on the proposed decisions, the Administrator will then
29     make final decisions on the PM NAAQS, which  are now expected to be published around the
30     end of 2005.
31

       June 2003                                 1-12        DRAFT-DO NOT QUOTE OR CITE

-------
 1      1.3.2  Methods and Procedures for Document Preparation
 2           The procedures for developing this revised PM AQCD build on the experience and
 3      methods derived from the most recent previous PM, Ozone, and CO AQCD preparation efforts.
 4      Briefly, the respective responsibilities for production of the present PM AQCD are as follows.
 5      An NCEA-RTP PM Team was formed and is responsible for developing and implementing the
 6      project plan for preparation of the PM AQCD, taking into account inputs from individuals in
 7      other EPA program and policy offices identified as part of the EPA PM Work Group. The
 8      resulting project plan (i.e., the PM Document Development Plan) was discussed with CASAC in
 9      May 1998 and was appropriately revised. An ongoing literature search has continued to be
10      conducted to identify, to the extent possible, all pertinent PM literature published since early
11      1996. Additionally, EPA published in the Federal Register (1) a request for information asking
12      for recently available research information on PM that may not yet be published and
13      (2) a request for individuals with the appropriate type and level of expertise to contribute to the
14      writing of PM AQCD materials to identify themselves (U.S. Environmental Protection Agency,
15      1998b).  The specific authors  of chapters or sections of the proposed document were selected on
16      the basis of their expertise on the subject areas and their familiarity with the relevant literature;
17      these include both EPA and non-EPA scientific experts. The project team defined critical issues
18      and topics to be addressed by  the authors and provided  direction in order to emphasize
19      evaluation of those studies most clearly identified as important for standard setting. It should be
20      noted that materials contributed by non-EPA authors are incorporated and, at times, modified by
21      EPA PM team staff to reflect internal and/or external review comments, e.g., by the public or
22      CASAC, and that EPA is responsible for the ultimate content of the PM AQCD.
23           The main focus of this revised criteria document is the evaluation and interpretation of
24      pertinent atmospheric science information, air quality data, human exposure information, and
25      health and welfare effects information newly published since that assessed in the 1996 PM
26      AQCD and likely to be useful in deriving criteria for U.S. EPA PM NAAQS decisions.  Initial
27      draft versions of AQCD chapters were evaluated via expert peer-review workshop discussions
28      and/or written peer reviews that focused on the selection of pertinent studies included in the
29      chapters, the potential need for additional information to be added to the chapters, and the
30      quality of the summarization and interpretation of the literature.  The authors of the draft
31      chapters then revised them on the basis  of workshop and/or written expert review

        June 2003                                1-13        DRAFT-DO NOT QUOTE OR CITE

-------
 1      recommendations. These and other integrative summary materials were incorporated into the
 2      First External Review Draft of the PM AQCD (October 1999), which underwent public
 3      comment and was the subject of consultation with CAS AC at a December 1999 public meeting.
 4           In order to foster timely presentation and publication of newly emerging PM research
 5      findings, EPA co-sponsored an Air and Waste Management Association International Speciality
 6      Conference, entitled "PM 2000: Particulate Matter and Health," held in January 2000 in
 7      Charleston, SC.  The conference was co-sponsored in cooperation with several other government
 8      agencies and/or private organizations that also fund PM research.  Topics covered included new
 9      research results concerning the latest advances in PM atmospheric sciences (e.g., PM formation,
10      transport, transformation), PM exposure, PM dosimetry and extrapolation modeling, PM
11      toxicology (e.g., mechanisms, laboratory animal models, human clinical responses), and PM
12      epidemiology. The main purpose of the conference was to facilitate having the latest scientific
13      information available in time for incorporation as quickly as possible into the Second External
14      Review Draft of this revised PM AQCD. Hence, arrangements were made for scientists to
15      submit written manuscripts on papers or posters presented at the PM 2000 Conference for
16      expedited peer-review by several major journals, so that decisions on acceptance for publication
17      could be made by mid-2000.  The evaluations and findings set forth in the Second External
18      Review Draft (March 2001) of the revised PM AQCD included consideration of such published
19      PM 2000 papers and  extensive additional information published elsewhere since the previous
20      First External Review Draft;  and the Second Draft was reviewed by CAS AC in July 2001.
21           Further revisions were then incorporated into the Third External Review Draft (April 2002)
22      to reflect both public comment and CAS AC review of the Second Draft, as well as assessment of
23      additional extensive new information published since that addressed in the Second Draft.  This
24      Fourth Draft AQCD incorporates changes made in response to earlier public comments and
25      CASAC reviews; and it includes pertinent peer-reviewed literature published or accepted for
26      publication mainly through April 2002.  The final PM AQCD will include revisions made in
27      response to public comment and CASAC review of this draft document.
28
29
30
        June 2003                                 1-14        DRAFT-DO NOT QUOTE OR CITE

-------
 1      1.3.3  Approach
 2           An initial step in development of this revised PM AQCD was to focus on selection of
 3      pertinent issues to include in the document as the basis for the derivation of PM NAAQS criteria.
 4      Preliminary issues were identified by the NCEA PM Team via inputs from other EPA program
 5      and policy offices, as derived from the 1996 PM AQCD and SP, their CAS AC and public
 6      reviews, from the 1997 standard promulgation process, and from the 1998 PM Research Needs
 7      Document (alluded to in Section 1.2.2).  Still further identification and refinement of issues
 8      resulted from NRC review and reports on PM research priorities (also discussed in Section 1.2.2
 9      above). The CAS AC review of the PM AQCD Development Plan and public comments on
10      earlier draft AQCD materials at various stages of their development also provided important
11      inputs regarding issue identification.
12           In developing draft materials for inclusion in the revised PM AQCD, detailed review of
13      key new research was undertaken as a first step. However, instead of presenting a
14      comprehensive review of all the literature, emphasis in this revised AQCD is placed  on (1) first,
15      the concise summary of key findings derived from previous PM criteria reviews (especially the
16      1996 PM AQCD) and, then, (2) evaluation of the most pertinent new key information, with
17      greater emphasis on more interpretive assessment — an approach reflecting CAS AC
18      recommendations.  To aid in the development of a more concise document than the 1996 PM
19      AQCD, compilation of summary tables of relevant new literature published since completion of
20      that previous document and selective text discussion of such new literature has been  undertaken,
21      with increased emphasis being placed in text discussions on interpretive evaluation and
22      integration of key points derived from the newly summarized research results.
23
24
25      1.4   DOCUMENT ORGANIZATION AND CONTENT
26           The present draft document attempts to critically review and assess relevant scientific
27      literature on PM published, since early 1996, including materials accepted for publication mainly
28      through April 2002 (and thus appearing mostly during 2002).  Limited coverage of some more
29      recent studies is also included as deemed appropriate in light of its special importance. For
30      example, information derived from the recently released HEI Special Report (Health Effects
31      Institute, May 2003), discussed above in Section 1.3.1, is being integrated into this assessment.

        June 2003                                1-15        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Literature discussed in detail in the 1996 EPA PM AQCD (U.S. Environmental Protection
 2      Agency, 1996a) generally is not discussed in depth in this document; rather key findings from
 3      that 1996 review are summarized as appropriate.  Limited treatment is included, however, for
 4      some earlier studies judged to be of particular importance in the review of the PM NAAQS.
 5      Key literature is mainly summarized in tables; and the text mainly attempts to integrate and
 6      discuss overall interpretive points.  An attempt is made to present alternative points of view
 7      where scientific controversy exists.
 8           Emphasis is placed in the document on studies conducted at or near PM pollutant
 9      concentrations found in ambient air. However,  although emphasis has been placed on studies in
10      the range of current ambient levels, studies examining effects of higher concentrations have been
11      included if they contain unique data or documentation of a previously unreported effect or
12      mechanism.
13           The present document, basically organized to assess information related to topics along the
14      same flow of issues presented in the NRC conceptual framework shown in Figure 1-1, includes
15      an Executive Summary and nine chapters presented in two volumes.  Volume I contains the
16      Executive Summary, this general introduction (Chapter 1), and Chapters 2 through 5.  Chapters 2
17      and 3 provide background information on physical and chemical properties of PM and related
18      compounds; sources and emissions; atmospheric transport, transformation, and fate of PM;
19      methods for the collection and measurement of PM; and U.S. ambient air PM concentrations.
20      Chapter 4 assesses PM environmental effects on vegetation and ecosystems, visibility, man-
21      made materials, and climate-related effects (including effects on atmospheric transmission of
22      solar radiation), and includes limited information on economic impacts of some welfare effects.
23      Chapter 5 discusses factors affecting exposure of the general population to ambient PM.
24           The second volume contains Chapters 6 through 9.  Chapter 6 evaluates information
25      concerning dosimetry of inhaled particles in the respiratory tract. Chapter 7 assesses the
26      toxicology of specific types of PM constituents  and potential mechanisms of action, based on
27      both laboratory animal studies and controlled human exposure studies.  Chapter 8 discusses
28      observational, i.e., epidemiologic, studies.  Lastly, Chapter 9 integrates key information on
29      exposure, dosimetry, and critical health risk issues derived from studies reviewed in the prior
30      chapters, as well as highlighting key points regarding important welfare effects associated with
31      ambient PM.

        June 2003                                 1-16        DRAFT-DO NOT QUOTE OR CITE

-------
  1       REFERENCES

  2       Federal Register. (1971) National primary and secondary ambient air quality standards. F. R. (April 30)
  3             36:8186-8201.
  4       Federal Register. (1987) Revisions to the national ambient air quality standards for paniculate matter. F. R. (July 1)
  5             52:24,634-24,669.
  6       Federal Register. (1996) National ambient air quality standards for paniculate matter; proposed rule. F. R.
  7             (December 13) 61: 65,638-65,713.
  8       Federal Register. (1997) National ambient air quality standards for paniculate matter; final rule. F. R. (July 18)
  9             62: 38,652-38,752.
10       Health Effects Institute. (2003) Revised analyses of time-series studies of air pollution and health. Boston, MA:
11             Health Effects Institute; special report.
12       Lioy, P. J. (1990) Assessing total human exposure to contaminants: a multidisciplinary approach. Environ. Sci.
13             Technol. 24: 938-945.
14       National Research Council. (1983) Risk assessment in the federal government: managing the process. Washington,
15             DC: National Academy Press. Available: http://books.nap.edu/books/PODl 15/html/index.html (3 June 2003).
16       National Research Council. (1994) Science and judgment in risk assessment. Washington, DC: National Academy
17             Press. Available: http://www.nap.edu/books/030904894X/html/ (4 June 2003).
18       National Research Council. (1998) Research priorities for airborne paniculate matter. I. Immediate priorities and a
19             long-range research portfolio. Washington, DC: National Academy Press. Available:
20             http://www.nap.edu/catalog/6131.html (4 June 2003).
21       National Research Council. (1999) Research priorities for airborne paniculate matter. II. Evaluating research
22             progress and updating the portfolio. Washington, DC: National Academy Press. Available:
23             http://www.nap.edu/books/0309066387/html/ (4 June 2003).
24       National Research Council. (2001) Research priorities for airborne paniculate matter. III. Early research progress.
25             Washington, DC: National Academy Press. Available: http://www.nap.edu/books/0309073375/html/ (4 June
26             2003).
27       Sexton, K.;  Selevan, S. G.; Wagener, D. K.; Lybarger, J. A. (1992) Estimating human exposures to environmental
28             pollutants: availability and utility of existing databases. Arch. Environ. Health 47: 398-407.
29       U.S. Code. (1991) Clean Air Act, §108, air quality criteria and control techniques, §109, national ambient air quality
30             standards. U.  S. C. 42:  §§7408-7409.
31       U.S. Court of Appeals for the District of Columbia. (1999a) American Trucking Associations, Inc. vs. U.S.
32             Environmental Protection Agency. 175 F3d 1027 (D.C. Cir. 1999).
33       U.S. Court of Appeals for the District of Columbia. (1999b) American Trucking Associations, Inc. vs. U.S.
34             Environmental Protection Agency. 195 F.3d 4 (D.C. Cir.  1999),  affirmed in part, reversed in part, and
35             remanded..
36       U.S. Court of Appeals for the District of Columbia. (2002) American Trucking Associations, Inc. vs. U.S.
37             Environmental Protection Agency. 283 F.3d 355, 378-79 (D.C. Cir. 2002).
38       U.S. District Court of Arizona. (1995) American Lung Association v. Browner. West's Federal Supplement 884
39             F.Supp. 345 (No. CIV 93-643 TUC ACM).
40       U.S. Environmental Protection Agency. (1996a) Air quality criteria for paniculate matter. Research Triangle Park,
41             NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
42       U.S. Environmental Protection Agency. (1996b) Review of the national ambient air quality standards for paniculate
43             matter: policy assessment of scientific and technical information. OAQPS staff paper. Research Triangle
44             Park, NC: Office of Air Quality Planning and Standards;  report no. EPA/452/R-96-013.  Available from:
45             NTIS, Springfield, VA; PB97-115406REB.
46       U.S. Environmental Protection Agency. (1998a) Paniculate matter research needs for human health risk assessment
47             to support future reviews of the national ambient air quality standards for paniculate matter. Research
48             Triangle Park, NC: National Center for Environmental Assessment; report no. EPA/600/R-97/132F.
49       U.S. Environmental Protection Agency. (1998b) Review of national ambient air quality standards for paniculate
50             matter. Commer. Bus. Daily: February 19. Available: http://cbdnet.access.gpo.gov/index.html [1999, November 24].
51       U.S. Supreme Court. (2001) Whitman v. American Trucking Association. 531  U.S. 457 (nos. 99-1257 and 99-1426).
52       SWolff, G. T. (1996) Closure by the Clean Air Scientific Advisory Committee (CASAC) on the staff paper for
53             paniculate matter [letter to Carol M. Browner, Administrator, U.S. EPA]. Washington, DC: U.S.
54             Environmental Protection Agency, Clean Air Scientific Advisory Committee; EPA-SAB-CASAC-LTR-96-
55             008; June 13.
         June 2003                                       1-17         DRAFT-DO NOT QUOTE OR CITE

-------
 i          2.  PHYSICS, CHEMISTRY, AND MEASUREMENT
 2                          OF PARTICULATE  MATTER
 3
 4
 5          Chapter 3 of the 1996 EPA document Air Quality Criteria for Particulate Matter (1996 PM
 6     AQCD; U.S. Environmental Protection Agency, 1996a) contained an extensive review of the
 7     physics and chemistry of airborne parti culate matter (PM). Chapter 2 of this revised version of
 8     the PM AQCD also provides background information on  the physics and chemistry of
 9     atmospheric particles, information useful in aiding the understanding of subsequent chapters.
10     Those  subsequent chapters are basically organized to follow the sequence of key elements
11     comprising the risk assessment framework described in Chapter 1 (Section 1.2.2), beginning
12     with sources and continuing to effects as shown in Figure 1-1.  Thus, this chapter provides
13     new information useful in evaluating PM effects of PM on human health and welfare and in
14     preparing related risk assessments used to support PM standard-setting decisions. Information
15     important for implementation of PM standards, but not essential to the standard setting process,
16     is not the focus in this chapter. The reader is referred to the NARSTO Fine Particle Assessment
17     (NARSTO, 2003) for information relevant to air quality management for PM.
18          Unlike other criteria pollutants (O3, CO, SO2, NO2,  and Pb), PM is not a specific chemical
19     entity but is a mixture of particles from different sources  and of different sizes, compositions,
20     and properties. Emphasis is placed here on discussion of differences between fine and coarse
21     particles and differences between ultrafine particles and accumulation-mode  particles within fine
22     particles.
23          PM is defined quantitatively by the measurement techniques employed. Therefore, it will
24     be useful to discuss our understanding of the relationship between PM suspended in the
25     atmosphere, PM inhaled by people, and PM measured by various sampling and analytical
26     techniques. Chapter 4 of the 1996 PM AQCD (U.S. Environmental Protection Agency, 1996a)
27     contained a review of the state-of-the-art of PM measurement technology.  Since that time,
28     considerable progress has been made in understanding problems in the measurement of PM
29     mass, chemical composition, and physical parameters. There also has been progress in
30     developing new and improved measurement techniques, especially for continuous
31     measurements. Therefore, a more extensive survey on measurement problems and on newly

       June 2003                                2-1        DRAFT-DO NOT QUOTE OR CITE

-------
 1      developed measurement techniques is included in Section 2.2.  For more detail and older
 2      references, the reader is referred to Chapters 3 and 4 of the 1996 PM AQCD (U.S.
 3      Environmental Protection Agency, 1996a).
 4
 5
 6      2.1  PHYSICS AND CHEMISTRY OF PARTICULATE MATTER
 7      2.1.1   Basic Concepts
 8           Atmospheric particles originate from a variety of sources and possess a range of
 9      morphological, chemical, physical, and thermodynamic properties. Examples of atmospheric
10      particles include combustion-generated particles, such as diesel soot or fly ash; photochemically
11      produced particles, such as those found in urban haze; salt particles formed from sea spray; and
12      soil-like particles from resuspended dust. Some particles are liquid; some are solid. Others may
13      contain a solid core surrounded by liquid. Atmospheric particles contain inorganic ions, metallic
14      compounds, elemental carbon, organic compounds, and crustal compounds.  Some atmospheric
15      particles are hygroscopic and contain particle-bound water.  The organic fraction is especially
16      complex, containing hundreds (probably thousands) of organic compounds.  (See Appendix 3C
17      for information on the composition of the organic fraction and the concentration of specific
18      organic compounds.) Primary particles are  emitted directly from sources;  whereas secondary
19      particles are formed from gases through chemical reactions in the atmosphere involving
20      atmospheric oxygen (O2) and water vapor (H2O); reactive species such as ozone (O3); radicals
21      such as the hydroxyl («OH) and nitrate («NO3) radicals; and pollutants such as sulfur dioxide
22      (SO2), nitrogen oxides (NOX), and organic gases from natural and anthropogenic sources.  The
23      particle formation process includes nucleation of particles from low-vapor pressure gases
24      emitted from sources or formed in the atmosphere by chemical reactions, condensation of
25      low-vapor pressure gases on existing particles,  and coagulation of particles.  Thus, any given
26      particle may contain PM from many sources. Because a particle from a given source is likely to
27      be composed of a mixture of chemical components and particles from different sources may
28      coagulate to form a new particle, atmospheric particles may be considered a mixture of mixtures.
29           The composition and behavior of particles are fundamentally linked with those of the
30      surrounding gas.  An aerosol may be defined as a suspension of solid or liquid particles in air.
31      The term aerosol includes both the particles and all vapor or gas phase components of air.

        June 2003                                 2-2        DRAFT-DO NOT QUOTE OR CITE

-------
 1      However, the term aerosol is sometimes used to refer to the suspended particles only. In this
 2      document, "particulate" is used only as an adjective, as in particulate matter.
 3           A complete description of the atmospheric aerosol would include an accounting of the
 4      chemical composition, morphology, and size of each particle and the relative abundance of each
 5      particle type as a function of particle size (Friedlander, 1970).  However, the physical and
 6      chemical characteristics of particles are usually measured separately.  Size distributions by
 7      particle number, used to calculate surface area and volume distributions, often are determined by
 8      physical means, such as electrical mobility, aerodynamic behavior, or light scattering.  Chemical
 9      composition usually is determined by analysis of collected samples although some species can
10      be measured in situ. The mass and average chemical composition of particles, segregated
11      according to aerodynamic diameter by cyclones or impactors, can also be determined.  However,
12      recent developments in single particle analysis techniques by electron microscopy with X-ray
13      analysis of single particles (but not agglomerates) collected on a substrate or by mass
14      spectroscopy of individual suspended particles provide elemental composition of individual
15      particles by particle size and, thus, are bringing the description envisioned by Friedlander closer
16      to reality.
17
18      2.1.2   Physical Properties and Processes
19      2.1.2.1  Definitions of Particle Diameter
20           The diameter of a spherical particle may be determined by optical or electron microscopy,
21      by light scattering and Mie theory, by its electrical mobility, or by its aerodynamic behavior.
22      However, atmospheric particles often are not spherical. Therefore, their diameters are described
23      by an "equivalent" diameter (i.e., the diameter of a  sphere that would have the same physical
24      behavior).  An optical diameter is the diameter of a spherical particle, with the same refractive
25      index as the particle used to calibrate the optical particle sizer, that scatters the same amount of
26      light into the solid angle measured.  Diffusion and gravitational settling are important physical
27      behaviors for particle transport, collection, and removal processes, including deposition in the
28      respiratory tract. Different equivalent diameters are used depending on which process is more
29      important.  For smaller particles diffusion is more important and the Stokes diameter is often
30      used. For larger particles  gravitational setting is more  important and the aerodynamic diameter
31      is often used.

        June 2003                                  2-3         DRAFT-DO NOT QUOTE OR CITE

-------
 1           The Stokes diameter, Dp, describes particle size based on the aerodynamic drag force
 2      imparted on a particle when its velocity differs from that of the surrounding fluid. For a smooth,
 3      spherically shaped particle, Dp exactly equals the physical diameter of the particle. For
 4      irregularly shaped particles, Dp is the diameter of an equivalent sphere that would have the same
 5      aerodynamic resistance.  Electrical mobility analyzers classify particles according to their
 6      electrical mobility. Particles of equal Stokes diameters that carry the same  electric charge will
 7      have the same electrical mobility. Hence, for spherical particles, the electrical mobility diameter
 8      would equal the Stokes diameter. The mobility diameter can be considered the diameter of a
 9      spherical particle that would have the same electrical mobility. The particle mobility can be
10      related to the particle diffusion coefficient and Brownian diffusion velocity through the Stokes-
1 1      Einstein equation. Thus, the Stokes diameter is the appropriate parameter for particle behavior
12      governed by diffusion. The Stokes diameter, Dp, is used in size distributions based on light
13      scattering and mobility analysis. The Stokes diameter is independent of density.
14           The aerodynamic diameter, Da, however, depends on particle density. It is defined as the
15      diameter of a  spherical particle with an equal gravitational settling velocity but a material density
16      of 1 g/cm3.  Cascade impactors separate particles based on their aerodynamic diameter and
17      aerodynamic particle sizers measure the aerodynamic diameter.  Respirable, thoracic, and
18      inhalable sampling and PM25 and PM10 sampling are based on particle aerodynamic diameter.
19      For particles greater than about 0.5 jim, the aerodynamic diameter is generally the quantity of
20      interest. For smaller particles, the Stokes diameter may be more useful. Particles with the same
21      physical size and shape but different densities will have the same Stokes diameter but different
22      aerodynamic diameters.
23           Aerodynamic diameter, Da, is related to the Stokes diameter, Dp, by:
24
                                                      1/2
25
26      where p is the particle density, and Cp and Ca are the Cunningham slip factors evaluated for the
27      particle diameters Dp and Da respectively.  The slip factor is a function of the ratio between
28      particle diameter and mean free path of the suspending gas (0.066 jim for air at one atmosphere

        June 2003                                 2-4         DRAFT-DO NOT QUOTE OR CITE

-------
 1      pressure and 20 °C). C is an empirical factor that accounts for the reduction in the drag force on
 2      particles due to the "slip" of the gas molecules at the particle surface.  It is important for
 3      particles less than 1 |im in diameter, for which the surrounding air cannot be modeled by a
 4      continuous fluid. For large particles (Dp > 5 jim) C = 1; while for smaller particles C > 1.
 5           For particles with diameters greater than the mean free path, A, the aerodynamic diameter
 6      given by equation (2-1) is approximated by:
 7
                                          DP              (Dp » A)                      (2-2)
        This expression, which shows that aerodynamic diameter is directly proportional to the square
        root of the particle density, is often used for particles as small as 0.5 jam.  For particles with
        diameters much smaller than the mean free path, the slip factor must be taken into account.
        In this case the aerodynamic diameter is directly proportional to the particle density,
                                    Da=  (/7)Dp(Dp«/l)                             (2-3)
14
15      Detailed definitions of the various sizes and their relationships are given in standard aerosol
16      textbooks (e.g., Friedlander [2000], Reist [1984, 1993], Seinfeld and Pandis [1998], Hinds
17      [1999], Vincent [1989, 1995], Willeke and Baron [1993], Baron and Willeke [2002], and Fuchs
18      [1964, 1989]).
19
20      2.1.2.2  Aerosol Size Distributions
21           Particle size as indexed by one of the "equivalent" diameters is an important parameter in
22      determining the properties, effects, and fate of atmospheric particles. The atmospheric
23      deposition rates of particles and therefore their residence times in the atmosphere are a strong
24      function of their Stokes and aerodynamic diameters. The diameter also influences deposition
25      patterns of particles within the lung. Because light scattering is strongly dependent on the
26      optical particle size, the amount of light scattering per unit PM mass will be dependent on the
27      size distribution of atmospheric particles. Therefore, the effects of atmospheric particles on

        June 2003                                  2-5         DRAFT-DO NOT QUOTE OR CITE

-------
 1      visibility, radiative balance, and climate will be influenced by the size distribution of the
 2      particles. Studies using cascade impactors or cyclones measure the particle-size distribution
 3      directly in aerodynamic diameter. The diameters of atmospheric particles range from 1 nm to
 4      100 |im, spanning 5 orders of magnitude.  A variety of different instruments, measuring a variety
 5      of equivalent diameters, are required to cover this range.
 6           Older particle counting studies used optical particle counters to cover the range of 0.3 to
 7      30 |im diameter.  Diameters of particles below 0.5 jim were measured as mobility diameters.
 8      The particle diameters used in size distribution graphs from these studies usually are given as
 9      physical or  Stokes diameters rather than aerodynamic diameters. In recent years, aerodynamic
10      particle sizers have been developed that give a direct measurement of the aerodynamic diameter
11      in the range of approximately 0.7 to 10 |im diameter.  These instruments have been used with
12      electrical mobility analyzers that measure the mobility diameter of particles  from 3 nm to
13      approximately 0.5 jim (McMurry, 2000).  Unfortunately, there is no agreed-upon technique for
14      combining the various equivalent diameters. Some workers use various assumptions to combine
15      the various  measurements into one presentation; others report each instrument separately.
16      Therefore, the user of size distribution data should be careful to determine exactly which
17      equivalent diameter is reported.
18
19      Particle Size Distribution Functions
20           The distribution of particles with respect to size is an important physical parameter
21      governing their behavior.  Because atmospheric particles cover several orders  of magnitude in
22      particle size, size distributions often are expressed in terms of the logarithm of the particle
23      diameter on the X-axis and the measured differential concentration on the Y-axis: AN/A(logDp)
24      = the number  of particles per cm3 of air having diameters in the size range from log Dp to log(Dp
25      + ADp).  Because logarithms do not have dimensions, it is  necessary to think of the distribution
26      as a function of log(Dp/Dp0), where the reference diameter  Dp0 = 1 jim is not explicitly stated. If
27      AN/A(logDp) is plotted on a linear scale, the number of particles between Dp and Dp + ADp is
28      proportional to the area under the curve of AN/A(logDp) versus logDp. Similar considerations
29      apply to distributions of surface,  volume, and mass. When approximated by a function,  the
30      distributions are usually given as dN/d(log Dp) rather than AN/A(log Dp).
31

        June 2003                                 2-6        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Atmospheric Aerosol Size Distributions
 2           In 1978, Whitby (1978) published an analysis of over 1000 particle size distributions
 3      measured at various locations in the U.S. Figure 2-1  shows the number, surface area, and
 4      volume distributions for the grand average continental size distribution. Volume, surface area,
 5      and number distributions are plotted on an arithmetic scale such that the volume, surface area, or
 6      number of particles in any specified size range is proportional to the corresponding area under
 7      the curve.  These distributions show that most of the particles are quite small, below 0.1 jam;
 8      whereas most of the particle volume (and therefore most of the mass) is found in particles
 9      > 0.1 jim.  Other averaged atmospheric size distributions are shown in Figure 2-2 and 2-3
10      (Whitby, 1978; Whitby and Sverdrup,  1980). Figure 2-2a and b describe the number of particles
11      as a function of particle diameter for rural, urban-influenced rural, urban, and freeway -
12      influenced urban aerosols. For some of the same data, the particle volume distributions are
13      shown in Figure 2-3a and b. Whitby (1978) observed that the size distributions typically had
14      three peaks which he called "modes."  The entire size distribution could be characterized well by
15      a trimodal model consisting of three additive log-normal distributions. The mode with a peak
16      between 5 and 30 jim diameter, formed by mechanical processes, was named the coarse particle
17      mode; the mode with a peak between 0.15 and 0.5 jim, formed by condensation and coagulation,
18      was called the accumulation mode; and the mode with a peak between 0.015 and 0.04 jam,
19      whose size was influenced by nucleation as well as by condensation and coagulation, was called
20      the transient nuclei or Aiken nuclei range, subsequently shortened to the nuclei mode. The
21      nuclei mode could be seen in the number and surface distribution but only in special situations
22      was it noticeable in the mass or volume distributions. The accumulation and nuclei modes taken
23      together were called fine particles. An idealized size distribution showing modes and formation
24      mechanisms is shown in Figure 2-4.
25           Whitby (1978) concluded
26
27                 The distinction between "fine particles" and "coarse particles" is a fundamental one.
28            There is now an overwhelming amount of evidence that not only are two modes in the mass or
29            volume distribution usually observed, but that these fine and coarse modes are usually
30            chemically quite different.  The physical separation of the fine and coarse modes originates
31            because condensation produces fine particles while mechanical processes produce mostly
32            coarse particles ... the dynamics of fine particle growth ordinarily operate to prevent the fine

        June 2003                                  2-7         DRAFT-DO NOT QUOTE OR CITE

-------
0> O
ft
z t?
   en
   _o
   *-*
        3  O
        CO
               15-
               10 -
                                   Nn = 7.7x10
                                       Na = 1.3x10
                                                        Nc = 4.2
                                                     DGNC = 0,97
        co ^  600
              400-
           *  200 -
                Sn = 74
             DGSn = 0.023
                                                 Sa = 535
       Sc=41
    DGSC = 3.
                     ""I	I	'I	fTI-Hf-
                  0.001
                                                               100
Figure 2-1. Distribution of coarse (c), accumulation (a), and nuclei (n) mode particles by
           three characteristics, a) number (N), b) surface area (S) and c) volume (V) for
           the grand average continental size distribution. DGV = geometric mean
           diameter by volume; DGS = geometric mean diameter by surface area;
           DGN = geometric mean diameter by number; Dp = particle diameter.

Source: Whitby (1978).
June 2003
                               2-8
DRAFT-DO NOT QUOTE OR CITE

-------
           E
           -O
           Q.
           Q
           O)
           jo
           5
           z
           "a
             1,000,000 -

               10,000 -

                  100 -
0.01 •
                0.0001 -
              0.000001 -
                           • Clean Rural
                            Urban Influenced Rural
                      	Average Urban
                      — - — - — Urban + Freeway
                                                          200,000
                    0,01    0,1     1     10    100
                        Particle Diameter, Dp (pm)
                                                               o
                                               0,01        0.1         1
                                                   Particle Diameter, Dp (|jm)
                                10
        Figure 2-2. Particle size distributions by number, (a) number concentrations are shown on
                    a logarithmic scale to display the wide range by site and size; (b) number
                    concentrations for the average urban distribution are shown on a linear scale.
                    For the linear scale, the area under any part of the curve is proportional to
                    particle number in that size range.
        Source: Whitby (1978); Whitby and Sverdrup (1980).
 1            particles from growing larger than about 1 um. Thus, the fine and coarse modes originate
 2            separately, are transformed separately, are removed separately, and are usually chemically
 3            different. . . practically all of the sulfur found in atmospheric aerosol is found in the fine
 4            particle fraction. Thus, the distinction between fine and coarse fractions is of fundamental
 5            importance to any discussion of aerosol physics, chemistry, measurement, or aerosol air quality
 6            standards.
 7
 8           Whitby's (1978) conclusions were based on extensive studies of size distributions in a
 9      number of western and midwestern locations during the  1970s (Whitby et al., 1974; Willeke
10      and Whitby, 1975; Whitby, 1978; Wilson et al., 1977; Whitby and Sverdrup, 1980).
11      No size-distribution  studies of similar scope have been published since then. Newer results
        June 2003
                                   2-9
DRAFT-DO NOT QUOTE OR CITE

-------
          I
         CO
         Q
         TO
70
65-
60-
55-

45-
40-
35-
30-
25-
20-
15-
10-
 5-
               [IT
              o-
              0.01
—•  r^
   0.1
 Particle
                                      Clean Rural
                                      Urban Influenced
                                      Rural
                                      South-Central
                                      New Mexico
    1       10
Diameter, Dp (|jm)
                        ^
                        n
   70
   65-
   60-
   55-
   50-
   45-
t_ 40-
^ 35-
Q
O) 30-
5 25-
•O 20-
   15-
   10-
    5-
                                     100
                                                    Average Urban
                                                    Urban + Freeway
    0
    0.01      0.1       1       10      100
         Particle Diameter, Dp (|jm)
       Figure 2-3.  Size distribution by volume (a) for the averaged rural and urban-influenced
                   rural number distributions shown in Figure 2-2a and a distribution from
                   south central New Mexico, and (b) for the averaged urban and freeway-
                   influenced urban number distributions shown in Figure 2-2a.
       Source: Whitby and Sverdrup (1980); Kim et al. (1993) south central New Mexico.
 1     from particle counting and impactor techniques, including data from Europe (U.S.
 2     Environmental Protection Agency, 1996a) and Australia (Keywood et al., 1999, 2000), show
 3     similar results for the accumulation and coarse modes.  Extensive measurements of particle size
 4     distributions, as part of EPA's Supersites program, are providing much new data for analysis.
 5          Whitby's (1978) conclusions have held up remarkably well.  However, ideas about the
 6     sub-0.1  |im diameter range have changed somewhat as newer instruments provided
 7     measurements extending to smaller sizes and with greater resolution in size and time (McMurry
 8     et al., 2000). Depending on the source, temperature, saturated vapor pressure of the  components,
 9     and the  age of the aerosol, size distributions have been observed with peaks (including multiple
10     peaks) throughout the sub-0.1  jim diameter size range.  Sub-0.1 |im diameter peaks have been
       June 2003
                    2-10
                                                DRAFT-DO NOT QUOTE OR CITE

-------
                                                                Mechanically
                                                                 Generated
                    0.002
                             0.01
                           Nuclei Mode
 0.1            1
Particle Diameter, Dp(|jm)
Accumulation Mode
                                    Fine Particles
      Coarse Mode
    Coarse Particles
                                                                                  100
      Figure 2-4. Volume size distribution, measured in traffic, showing fine and coarse
                  particles and the nuclei and accumulation modes of fine particles. DGV
                  (geometric mean diameter by volume, equivalent to volume median diameter)
                  and og (geometric standard deviation) are shown for each mode.  Also shown
                  are transformation and growth mechanisms (e.g., nucleation, condensation,
                  and coagulation).
      Source: Adapted from Wilson and Suh (1997).
1     observed in rural areas (O'Dowd, 2002) as well as for brief periods (nucleation bursts) in urban
2     areas (Woo et al., 2001a). Based on these and other observations, discussed in detail in
3     Section 2.1.2.3, aerosol scientists now classify particles in the sub-0.1 |im size range as ultrafine
4     particles and divide this size range into a nucleation region (< 10 nm) and an Aitkin (nuclei)
5     region (10-100 nm), as shown in Figure 2-5. Other studies, discussed in detail in the 1996 PM
6     AQCD (U.S. Environmental Protection Agency, 1996a), have shown that in fog  or clouds or at
7     very high relative humidities the accumulation mode may split into a larger size  (more
8     hygroscopic or droplet) submode and a smaller size (less hygroscopic or condensation) submode.
9
      June 2003
      2-11
DRAFT-DO NOT QUOTE OR CITE

-------
                      
-------
 1      several log-normal distributions. Definitions of terms used to describe size distributions in
 2      modal terms are given below.
 3
 4          NucleationMode:  Freshly formed particles with diameters below 10 nm, observed during
 5          active nucleation events.  The lower limit, where particles and large molecules overlap, is
 6          uncertain. Current techniques limit measurements to particles 3 nm or greater.
 7
 8          AitkinMode: Larger particles with diameters between 10 and 100 nm. The Aitken mode
 9          may result from growth of smaller particles or nucleation from higher concentrations of
10          precursors. Nucleation and Aitkin nuclei modes are normally observed in the number
11          distribution.
12
13          Accumulation Mode: Particles with diameters from about 0.1 jim to just  above the
14          minimum in the mass or volume distributions which usually occurs between 1  and 3 jim.
15          Accumulation-mode particles normally do not grow into the coarse mode. Nucleation-
16          mode and Aitkin-mode particles grow by coagulation (two particles combining to form
17          one) or by condensation (low-equilibrium vapor pressure gas  molecules condensing on a
18          particle) and "accumulate" in this size range.
19
20          Coarse Mode or Coarse Particles:  Particles with diameters mostly greater than the
21          minimum in the particle mass or volume distributions, which  generally occurs  between
22           1 and 3 jim. These particles are usually formed by mechanical breakup of larger particles
23          or bulk material.
24
25          Fine Particles: Fine particles include the nucleation, Aitkin,  and accumulation modes, i.e.,
26          particles from the lowest measurable size, currently about 3 nm, to just above the minimum
27          in the mass or volume distribution which generally occurs between 1 and 3 |im.  These
28          particles are generated during combustion or formed from gases.
29
30           Ultrafine Particles: That portion of fine particles with diameters below about  0.1 |im
31          (100 nm), i.e., the Aitkin and nucleation modes.

        June 2003                                 2-13       DRAFT-DO NOT  QUOTE OR CITE

-------
 1           Modes are defined primarily in terms of their formation mechanisms but also differ in
 2      terms of sources, composition, age, and size. Nucleation mode applies to newly formed particles
 3      which have had little chance to grow by condensation or coagulation.  Aitkin mode particles are
 4      also recently formed particles that are still actively undergoing coagulation. However, because
 5      of higher concentrations of precursors or more time for condensation and coagulation, the
 6      particles have grown to larger sizes. Accumulation mode applies to the final stage as particles,
 7      originally formed as nuclei, grow to a point where growth slows  down. These three modes,
 8      which together are called fine particles, are formed primarily by  combustion or chemical
 9      reactions of gases yielding products with low saturated vapor pressures.  Fine particles include
10      metals and elemental and organic carbon (primary PM) and sulfate, nitrate, ammonium ions, and
11      organic compounds (secondary PM).
12           The coarse  mode refers to particles formed by mechanical breakdown of minerals, crustal
13      material,  and organic debris.  The composition includes primary  minerals and organic material.
14      The accumulation mode and the coarse mode overlap in the region between 1 and 3 jim (and
15      occasionally over an even larger range).  In this  region, chemical composition of individual
16      particles can usually, but not always, allow identification of a source or formation mechanism
17      and so permit identification of a particle as belonging to the accumulation or coarse mode.
18
19           Sampler Cut Point.  Another set of definitions of particle size fractions arises from
20      considerations of size-selective sampling. Size-selective sampling refers to the collection of
21      particles below or within a specified aerodynamic size range. Size fractions are usually specified
22      by the 50% cut point size; e.g., PM2 5 refers to particles collected by a sampling device that
23      collects 50% of 2.5 jim particles and rejects 50% of 2.5 jim particles. However, size fractions
24      are defined, not merely by the 50% cut point, but by the entire penetration curve. Examples of
25      penetration curves are given in Figure 2-6. Thus, as shown by Figure 2-6, a PM2 5 sampler, as
26      defined by the Federal  Reference Method, rejects 94% of 3 jim particles, 50% of 2.5 jim
27      particles, and 16% of 2 |im.  Samplers with the same 50% cut point but differently shaped
28      penetration curves would collect different fractions of PM. Size-selective sampling has arisen in
29      an effort to measure particle size fractions with some special significance (e.g., health, visibility,
30      source apportionment,  etc.), to measure mass size distributions, or to collect size-segregated
31      particles for chemical analysis.  Dichotomous samplers split the particles into smaller and larger

        June 2003                                2-14       DRAFT-DO NOT QUOTE OR CITE

-------
                          100
                                                                  • IPM
                                                                  • TPM
                                                                  ORPM
                                                                  V PM
                                           4          10    20        50
                                          Aerodynamic Diameter (|_im)
                          100
       Figure 2-6. Specified particle penetration (size-cut curves) through an ideal (no-particle-
                   loss) inlet for five different size-selective sampling criteria.  Regulatory size
                   cuts are defined in the Code of Federal Regulations; PM25 (200Ic), PM10
                   (2001a). PM2 5 is also defined in the Federal Register (1997).  Size-cut curves
                   for inhalable particulate matter (IPM), thoracic particulate matter (TPM) and
                   respirable particulate matter (RPM) size cuts are computed from definitions
                   given by American Conference of Governmental and Industrial Hygienists
                   (1994).
 1     fractions that may be collected on separate filters.  However, some fine particles (=10%) are
 2     collected with the coarse particle fraction.  Cascade impactors use multiple size cuts to obtain a
 3     distribution of size cuts for mass or chemical composition measurements.  One-filter samplers
 4     with a variety of upper size cuts are also used, e.g., PM25, PM10.
 5          Regulatory size cuts are a specific example of size-selective sampling. In 1987, the
 6     NAAQS for PM were revised to use PM10, rather than total  suspended particulate matter (TSP),
 7     as the indicator for the NAAQS for PM (Federal Register, 1987).  The use of PM10 as an
 8     indicator is an example of size-selective sampling based on a regulatory size cut (Federal
 9     Register, 1987). The selection of PM10 as an indicator was based on health considerations and
10     was intended to focus regulatory concern on those particles small enough to enter the thoracic
11     region of the human respiratory tract.  The PM25 standard set in 1997 is also an example of
       June 2003
2-15
DRAFT-DO NOT QUOTE OR CITE

-------
 1     size-selective sampling based on a regulatory size cut (Federal Register, 1997).  The PM2 5
 2     standard was based primarily on epidemiologic studies using concentrations measured with
 3     PM2 5 samplers as an exposure index. However, the PM2 5 sampler was not designed to collect all
 4     respirable particles; it was designed to collect fine particles because of their different sources and
 5     properties (Whitby et al., 1974). Thus, the need to attain a PM2 5 standard will tend to focus
 6     regulatory concern on control of sources of fine particles.
 7          Prior to 1987, the indicator for the NAAQS for PM was TSP.  TSP is defined by the design
 8     of the High Volume Sampler (hivol) that collects all of the fine particles but only part of the
 9     coarse particles (Figure 2-7).  The upper cut-off size of the hivol depends on the wind speed and
10     direction and may vary from 25 to 40 |im. The Wide Range Aerosol Classifier (WRAC) was
11     designed specifically to collect the entire coarse mode using an impaction system designed by
12     Lundgren to collect particles up to 100 jim in diameter (Lundgren and Burton, 1995).
13
                           0.1
                                       0.5    1     2      5    10
                                            Particle Diameter, Dp(prn)
                                       Total Suspended Particles (TSP) —
                                            PM,
                                       PM.
                                         !2,5
    - PM
                                                          10-2,8
                                                                    20
                                                                          50    10Q
       Figure 2-7. An idealized distribution of ambient particulate matter showing fine-mode
                   particles and coarse-mode particles and the fractions collected by size-selective
                   samplers.  (WRAC is the Wide Range Aerosol Classifier which collects the
                   entire coarse mode [Lundgren and Burton, 1995].)
       Source: Adapted from Wilson and Suh (1997) and Whitby (1978).
       June 2003
2-16
DRAFT-DO NOT QUOTE OR CITE

-------
 1           An idealized distribution with the normally observed division of ambient aerosols into
 2     fine-mode particles and coarse-mode particles and the size fractions collected by the WRAC,
 3     TSP, PM10, PM2 5 and PM10_2 5 samplers is shown in Figure 2-7. PM10 samplers, as defined in
 4     Appendix J to Title 40 Code of Federal Regulations (40 CFR) Part 50 (Code of Federal
 5     Regulations, 200 la; Federal Register, 1987), collect all of the fine-mode particles and part of the
 6     coarse-mode particles. The upper cut point is defined as having a 50% collection efficiency at
 7     10 ± 0.5 |im aerodynamic diameter. The slope of the collection efficiency curve is defined in
 8     amendments to 40 CFR, Part 53, (Code of Federal Regulations, 2001b). An example of a PM10
 9     size-cut curve is shown in Figure 2-6.
10           An example of a PM25 size-cut curve is also shown in Figure 2-6. The PM25 size-cut
11     curve, however, is defined by the design of the Federal Reference Method (FRM) Sampler. The
12     basic design of the FRM sampler is given in the  Federal Register (1997, 1998) and as 40 CFR
13     Part 50, Appendix L (Code of Federal Regulations, 200 Ic). Additional performance
14     specifications are given in 40 CFR Parts 53 and  58 (Code of Federal Regulations, 2001b,d).
15     In order to be used for measurement of PM2 5 to determine compliance with the PM2 5 NAAQS,
16     each specific sampler design and its associated manual of operational procedures must be
17     designated as a reference method under 40 CFR Part 53 in Section 1.2 of Appendix L (Code of
18     Federal Regulations, 200Ic). Thus PM25 FRM samplers may have somewhat different designs
19     (see Table 2-4 in Section 2.2.4.1.2).
20           Papers discussing PM10 or PM2 5 frequently insert an explanation such as "PMX (particles
21     less than x jim diameter)" or "PMX (nominally, particles with aerodynamic diameter < x |im)."
22     While these explanations may seem easier than "PMX, (particles collected with an upper 50% cut
23     point of x jim aerodynamic diameter and a specified penetration curve)," they are not entirely
24     correct and may be misleading since they suggest an upper 100% cut point of x jim. Some
25     countries use PM10 to refer not to samplers with  a 50% cut at 10 |im Da but to samplers with
26     100% rejection of all particles greater than 10 jim Da.  Such samplers miss a fraction of coarse
27     thoracic PM.  An example is shown in Figure 2-8.
28           PM10, as defined by EPA, refers to particles collected by a sampler with an upper 50% cut
29     point of 10 |im Da and a specific, fairly sharp, penetration curve. PM25 is analogously defined.
30     Although there is not yet an FRM,  PM10_2 5 refers either to particles collected by a sampler with
31     an upper 50% cut point of 10 jim Da and a lower 50% cut point of 2.5 jim Da or to the difference

       June 2003                                2-17        DRAFT-DO NOT QUOTE OR CITE

-------
                        0)
                        0
                        CL
50--
40-
30-
20-
10-
                                                    °^
                            (H
                                                               Wedding Cyclonic Inlet
                                                                 O  (J=2 km/h
                                                                 D  U=8 km/h
                                                                 A  \J-24 km/h
                                                                Wedding and Weigand
                                                                     (1993)
                               Kimoto cyclonic inlet
                                                     \
                               	Manufacturer
                                 •  Tsai and Cheng (1996)   xjj
  1
                                                  4     6    8  10
                                            Aerodynamic Diameter (|jm)
           20
                                                -0	1
30
        Figure 2-8.  Comparison of penetration curves for two PM10 beta gauge samplers using
                    cyclone inlets. The Wedding PM10 sampler uses the U.S. EPA definition of
                    PMX as x = 50% cut point. The Kimoto PM10 defines PMX as x = the 100% cut
                    point (or zero penetration).
        Source: Tsai and Cheng (1996).
 1     between the particle concentration measured by a PM10 monitor and a PM2 5 monitor. In all
 2     cases, the fraction of PM collected depends on the entire penetration curve (or curves); i.e., for
 3     PM2 5 some particles > 2.5 jim Da are collected and not all particles < 2.5 jim Da are collected.
 4           In an analysis reported in 1979, EPA scientists endorsed the need to measure fine and
 5     coarse particles separately (Miller et al., 1979). Based on the availability of a dichotomous
 6     sampler with a separation size of 2.5 jim Da, they recommended 2.5 jim Da as the cut point
 7     between fine and coarse particles.  Because of the wide use of this cut point, the PM25 fraction is
 8     frequently referred to as "fine" particles. However, although the PM2 5 sample will usually
 9     contain all of the fine particles, it may collect a small fraction of the coarse particles, especially
10     in dry areas or during dry conditions.  A PM10_2 5 size fraction may be obtained from a
11     dichotomous sampler or by subtracting the mass collected by a PM2 5 sampler from the mass
12     collected by a PM10 sampler.  The resulting PM10_25 mass, or PM10_2 5, is sometimes called
13     "coarse" particles. However, it would be more correct to call PM2 5 an indicator of fine particles
        June 2003
                       2-18
DRAFT-DO NOT QUOTE OR CITE

-------
 1      (because it contains some coarse particles) and PM10_25 an indicator of the thoracic component of
 2      coarse particles (because it excludes some coarse particles below 2.5 jim Da and above 10 jim
 3      Da). It would be appropriate to call PM10 an indicator of thoracic particles. PM10 and thoracic
 4      PM, as shown in Figure 2-6, have the same 50% cut point.  However, the thoracic cut is not as
 5      sharp as the PM10 cut; therefore, thoracic PM contains some particles between 10 and 30 jim
 6      diameter that are excluded from PM10.
 7           Over the years, the terms fine and coarse, as applied to particles, have lost the precise
 8      meaning given in Whitby's (1978) definition.  In any given article, therefore, the meaning of fine
 9      and coarse, unless defined, must be inferred from the author's usage. In this document, fine
10      particles means all particles in the nucleation,  Aitken, and accumulation modes; and coarse
11      particles means all particles in the coarse mode. PM2 5 and fine particles are not equivalent
12      terms.
13
14           Occupational Health or Dosimetric Size Cuts. The occupational health community has
15      defined size fractions in terms of their entrance into various  compartments of the respiratory
16      system. This convention classifies particles into inhalable, thoracic, and respirable particles
17      according to their upper size cuts. Inhalable particles enter the respiratory tract, including the
18      head airways. Thoracic particles travel past the larynx and reach the lung airways and  the
19      gas-exchange regions of the lung. Respirable particles are a subset of thoracic particles that are
20      more likely to reach the gas-exchange region of the lung.  In the past, exact definitions of these
21      terms have varied among organizations.  As of 1993, a unified set of definitions was adopted by
22      the American Conference of Governmental Industrial Hygienists (ACGIH, 1994), the
23      International Standards Organization (ISO), and the European Standardization Committee
24      (CEN). The curves which define inhalable (IPM), thoracic (TPM), and respirable (RPM)
25      particulate matter are shown in Figure 2-6. These curves should not be taken to indicate that
26      particles > 4 jim Da do not reach the gas exchange regions or that particles < 4 jim Da do not
27      deposit in the bronchi.  See Figure 6-13 for a graphical characterization of particle deposition in
28      regions of the respiratory system as a function of particle size.
29
30
        June 2003                                 2-19        DRAFT-DO NOT QUOTE OR CITE

-------
 1      2.1.2.3  Ultrafine Particles
 2           As discussed in Chapter 7 (Toxicology of Particulate Matter in Humans and Laboratory
 3      Animals) and in Chapter 8 (Epidemiology of Human Health Effects Associated with Ambient
 4      Particulate Matter), some scientists argue that ultrafine particles may pose potential health
 5      problems and that some health effects may be associated with particle number or particle surface
 6      area as well as or more closely than with particle mass.  Some additional attention will be given
 7      to ultrafine particles because they contribute the major portion of particle number and a
 8      significant portion of particle surface area.
 9
10      Formation and Growth of Fine Particles
11           Several processes influence the formation and growth of particles. New particles may be
12      formed by nucleation from gas phase material. Particles may grow by condensation as gas phase
13      material condenses on existing particles; and particles also may grow by coagulation as two
14      particles combine to form one.  Gas phase material condenses preferentially on smaller particles,
15      and the rate constant for coagulation of two particles decreases as the particle size increases.
16      Therefore, ultrafine particles grow into the accumulation mode; but accumulation-mode particles
17      do not normally grow into the coarse mode (see Figure 2-4). More information and references
18      on formation and growth of fine particles can be found in the 1996 AQCD PM (U.S.
19      Environmental Protection Agency, 1996a).
20
21      Equilibrium Vapor Pressures
22           An important parameter in particle nucleation and in particle growth by condensation is the
23      saturation ratio,  S, defined as the ratio of the partial pressure of a species, p, to its equilibrium
24      vapor pressure above a flat surface at a specified temperature, p0: S = p/p0. For either
25      condensation or nucleation to occur, the species vapor pressure must exceed its equilibrium
26      vapor pressure.  For particles, the equilibrium vapor pressure is not the same as p0.  Two effects
27      are important: (1) the Kelvin effect, which is an increase in the equilibrium vapor pressure
28      above the surface due to its curvature (very  small particles have higher vapor pressures and will
29      not be stable to evaporation until they attain a critical size) and (2) the solute effect, which is a
30      decrease in the equilibrium vapor pressure of the liquid due to the presence of other compounds
31      in solution.  Organic compounds may also be adsorbed on ultrafine carbonaceous particles.

        June 2003                                 2-20       DRAFT-DO NOT QUOTE OR CITE

-------
 1           For an aqueous solution of a nonvolatile salt, the presence of the salt decreases the
 2      equilibrium vapor pressure of the water over the droplet. This effect is in the opposite direction
 3      of the Kelvin effect, which increases the equilibrium vapor pressure above a droplet because of
 4      its curvature.  The existence of an aqueous solution will also influence the vapor pressure of
 5      water-soluble species.  The vapor pressure behavior of mixtures of several liquids or of liquids
 6      containing several solutes is complex.
 7
 8      New Particle Formation
 9           When the vapor concentration of a species exceeds its equilibrium concentration
10      (expressed as its equilibrium vapor pressure), it is considered condensable.  Condensable species
11      can either condense on the surface of existing particles or can nucleate to form new particles.
12      The relative importance of nucleation versus condensation depends on the rate of formation of
13      the condensable species and on the surface or cross-sectional area of existing particles (McMurry
14      and Friedlander, 1979). In ambient urban environments, the available particle surface area is
15      usually sufficient to rapidly scavenge the newly formed condensable species. Formation of new,
16      ultrafine particles is usually not observable in mass or volume distributions except near sources
17      of condensable species. Wilson et al. (1977) report observations of the Aitkin nuclei mode in
18      traffic.  However, bursts of new particle formation can be observed in urban areas in the number
19      distribution (Woo et al., 2001a; McMurray et al., 2000).  New particle formation also can be
20      observed in cleaner, remote regions.  Bursts of new particle formation in the atmosphere under
21      clean conditions usually occur when  aerosol surface area concentrations are low (Covert et al.,
22      1992).  High concentrations of nuclei mode particles have been observed in regions with low
23      particle mass concentrations indicating that new particle formation is inversely related to the
24      available aerosol surface area (Clarke,  1992).
25
26      Sources of Ultrafine Particles
27           Ultrafine particles are the result of nucleation of gas phase species to form condensed
28      phase species with very low equilibrium vapor pressure.  In the atmosphere there are four major
29      classes of substances that yield particulate matter with equilibrium vapor pressures low enough
30      to form nuclei mode particles:
31

        June 2003                                 2-21        DRAFT-DO NOT QUOTE OR  CITE

-------
1      (1)  Particles containing heavy metals. Nuclei mode particles of metal oxides or other metal
            compounds are generated when metallic impurities in coal or oil are vaporized during
            combustion and the vapor undergoes nucleation. Metallic ultrafine particles also may be
            formed from metals in lubricating oil or fuel additives that are vaporized during
            combustion of gasoline or diesel fuels.  Ultrafine metallic particles were discussed in
            Section 6.9 of the 1996 PM AQCD (U.S. Environmental Protection Agency, 1996a).
2
3      (2)  Elemental carbon or soot (EC).  EC particles are formed primarily by condensation of C2
            molecules generated during the combustion process. Because EC has a very low
            equilibrium vapor pressure, ultrafine EC particles can nucleate even at high temperatures
            (Kittelson, 1998; Morawska et al., 1998).
4
5      (3)  Organic carbon (OC).  Recent smog chamber studies and indoor experiments show that
            atmospheric oxidation of certain organic compounds found in the atmosphere can
            produce highly oxidized organic compounds with an equilibrium vapor pressure
            sufficiently low to result in nucleation (Kamens et al., 1999; Weschler and Shields,
            1999).
6
7      (4)  Sulfates.  Sulfuric acid (H2SO4) molecules are generated in the atmosphere by conversion
            of sulfur dioxide (SO2) to H2SO4. As H2SO4 is formed, it can either nucleate to form new
            ultrafine particles, or it can condense on existing ultrafine or accumulation mode particles
            (Clark and Whitby, 1975; Whitby, 1978).  Nucleation theory allows calculation of
            nucleation rates for both binary nucleation, involving water and sulfuric acid (Easter and
            Peters, 1994)  or ternary nucleation, which requires sulfuric acid, ammonia (NH3), and
            water (Korhonen et al., 1999). Kulmala et al. (2000) compared nucleation rates of binary
            and ternary nucleation and concluded that the ternary rate is 1,000 times the binary rate.
            Results from an aerosol dynamics model with a ternary nucleation scheme indicate that
            nucleation in the troposphere should be ubiquitous and yield a reservoir of
            thermodynamically stable sulfate clusters 1-3 nm in diameter. The growth of these
            clusters to a detectable size (> 3  nm diameter) is limited by the availability of
            condensable vapor. Observations of atmospheric particle formation and growth from a
            continental and a coastal site suggest that a growth process including ternary nucleation is

       June 2003                                2-22       DRAFT-DO NOT QUOTE OR CITE

-------
             responsible for the formation of cloud condensation nuclei. Nucleation processes in the
             atmosphere may also involve organic compounds as well as sulfuric acid, ammonia, and
             water.  However, current formulations of nucleation theory only include the three
             inorganic components. (The possible formation of ultrafine NH4NO3 by reaction of NH3
             and nitric acid (HNO3) vapor has not been investigated.)
 9
10
11
12
13
14
15
16
     Vehicle engine exhaust may include all these substances.  Ultrafine particles are observed
in the emissions from spark, diesel, and jet engines (Kittelson, 1998).  In these cases it seems
likely that elemental carbon, organic compounds, ammonia and sulfuric acid from sulfur in the
fuel, as well as metal additives in the fuel or fuel oil, may contribute to the formation of ultrafine
particles (Tobias et al., 2001).  An example of particles from a diesel engine showing number
and volume distributions is given in Figure 2-9.
                         0.001           0.01
                          Number Weighting ..,„„„„„,„..„„„„„
                                             0.1
                                          Diameter (|im)
                                                                 Mass Weighting
              Figure 2-9. Typical engine exhaust size distribution.
              Source: Kittelson (1998).
        June 2003
                                          2-23
DRAFT-DO NOT QUOTE OR CITE

-------
 1     Recent Measurements of Ultraftne Particles
 2          Instruments, developed during the past decade, permit measurement of size distributions
 3     down to 3 nm diameter particles. Use of these techniques have led to new information on the
 4     formation of new particles by nucleation.  Such measurements have been carried out during
 5     intensive field measurement campaigns, during continuous measurements in urban areas in
 6     several European cities, and in the U.S. as a part of the Supersite program (McMurry et al.,
 7     2000; Woo et al., 2001a). Nucleation has been observed in the free troposphere (Weber et al.,
 8     1999; Clarke, 1992; Schroder and Strom, 1997; Raes et al., 1997); in outflows of evaporating
 9     convective clouds (Clarke et al., 1998; Hegg et al., 1990, 1991; Radke and Hobbs, 1991; Perry
10     and Hobbs, 1994); in the marine boundary layer (Covert et al., 1992; Hoppel et al., 1994; Van
11     Dingenen et al., 1995; Weber et al., 1998; Clarke et al., 1998); downwind of coastal regions
12     during low tide (McGovern et al., 1996; McGovern, 1999); on mountains (Weber et al., 1995,
13     1997; Raes et al., 1997; Wiedensohler et al., 1997); over forests (MaKela et al., 1997; Kulmala
14     et al., 1998; O'Dowd et al., 2002); downwind of certain biogenic emissions (Weber et al., 1998);
15     in urban areas (Birmili and Wiedensohler, 1998; McMurry et al., 2000; Woo et al., 2001a); near
16     freeways (Zhu et al., 2002a,b); in engine exhaust (Kittelson, 1998; Tobias et al., 2001); and in
17     homes (Wallace and Howard-Reed, 2002). Nucleation events in outdoor air almost always occur
18     during daylight, indicating that photochemistry plays a role in producing the gas phase
19     precursors of new particles.
20          The number size distributions observed over a boreal forest in Finland before and during
21     the initial stages of a nucleation event are shown in Figure 2-10. The Aitken and accumulation
22     modes can be seen clearly before the nucleation event. The nucleation mode, with a peak
23     between 3 and 7 nm, appears during the event. Figure 2-11 shows the variety of size
24     distributions that may be observed as nuclei are formed and grow, based on size distributions
25     measured in the Arctic marine boundary layer (Covert et al., 1996). These distributions all show
26     a trimodal distribution within the fine particle size range.  The changes in size distribution due to
27     coagulation (and dilution) immediately downwind of a freeway (Zhu et al., 2002b) are shown in
28     Figure 2-12(a)-(g).  At 30 m downwind the nucleation mode is larger than the Aitken mode, but
29     by 60 m downwind  coagulation has removed particles from the nucleation mode and added
30     particles to the Aitken mode so that the Aitken mode is larger than the nucleation mode.
31

       June 2003                                2-24       DRAFT-DO NOT QUOTE OR CITE

-------
                 104-
               o
               O
                 102.
                                                                            Before nucleation
                                                                            During nucleation
                                          10                     100
                                            Particle Diameter (nm)
                                    1,000
       Figure 2-10.  Number size distributions showing measurement of a nucleation burst made
                     in a boreal forest in Finland.
       Source: O'Dowd et al. (2002).
 1           There is strong evidence that sulfuric acid vapor often participates in nucleation.  However,
 2     condensation of sulfuric acid and its associated water and ammonium ions typically can account
 3     for only 10% to 20% of the observed growth rates for freshly nucleated particles. Therefore,
 4     organic compounds may account for much of the formation and growth of freshly nucleated
 5     particles.  Evidence of nucleation of organic particles comes from smog chamber studies
 6     (Kamens et al., 1999) and from field studies over forests (MaKela et al., 1997; Kulmala et al.,
 7     1998; O'Dowd et al., 2002). Nucleation of organic particles may also occur indoors due to the
 8     reaction of infiltrated ozone with indoor terpenes from air fresheners or cleaning solutions
 9     (Weschler and Shields, 1999). The observation of bursts of nuclei-mode particles in Atlanta
10     (Woo et al., 2001a), perhaps due to unusually high rates of production of condensable species,
11     suggests that high concentrations of ultrafine particles may be a normal occurrence in polluted
12     urban areas.
13
14     Concentration of Ultrafine Particles:  A Balance Between Formation and Removal
15          Nuclei-mode particles may be removed by dry deposition or by growth into the
16     accumulation mode.  This growth takes place as other low vapor pressure material condenses on
17     the particles or as nuclei-mode particles coagulate with themselves or with accumulation mode
       June 2003
2-25
DRAFT-DO NOT QUOTE OR CITE

-------
S"~*.
1=
o
^~
-— -
o
01
o
13
•z.
T3
XbUU
1400 -
1200 -

1000 -
SOO -

600 -

400 -
200 -










.'
1
August 30, 1991, 1000 UTC 1 g





"


I
\- 	
                                                      600-

                                                      500-

                                                      400-

                                                      300-

                                                      200-

                                                      100-
                        August 31, 1991, 0200 UTC
      ro
       O
       Q
       Ol
       O
       Z
       13
              September 5, 1991, 1300 UTC
      700-

      600-

      500-

      400-

      300-

      200-

      100-
Septernber 22, 1991, 1200 UTC
                                    100
                                               1000
                                                                    10
                                                                                100
                                                                                          1000
                              Q  40-
                              cn
                              _o
                              T3  20-
                              z
                              T3
                                    September 24, 1991, 0800 UTC
      Figure 2-11.  Examples of the measured one hour average particle number size
                    distributions and the log normal fits to the modes of the data. Squares are
                    measured data, solid lines are the fitted log normal modes determined by
                    DistFit™. These modes, nucleation between 3 and 20 nm, Aitken between
                    20 and 100 nm, and accumulation above 100 nm can be observed in most
                    examples.

      Source: Covert etal. (1996).
1     particles.  Because the rate of coagulation would vary with the concentration of accumulation-
2     mode particles, it might be expected that the concentration of nuclei-mode particles would
3     increase with a decrease in accumulation-mode mass. On the other hand, the concentration of
4     particles would be  expected to decrease with a decrease in the rate of generation of particles by
5     reduction in emissions of metal and carbon particles or a decrease in the rate of generation of
6     H2SO4 or condensable organic vapor.  The rate of generation of H2SO4 depends on the
      June 2003
2-26
  DRAFT-DO NOT QUOTE OR CITE

-------
      2.0e+5
       0.0
      1.0e+4
      S.Oe+3 -
      6.06+3 -
    2 4.06+3-
      2.06+3 -
                       10             100
                   Particle Diameter (nm)
                                                                               100
                  Particle Diameter (nm)
                                                                                    LJ:
                                                     150m
                                                     Down
                       10             100
                   Particle Diameter (nm)
                      10
                  Particle Diameter (nm)
                                                                               100
                                                1.06+4
                                                S.Oe+3
                                                6.06+3
                                              ° 4.06+3
                                                2.06+3
                                                  0.0
                                                                   Mg=84.7nm
                                                                   0>1.71
                       10             100
                   Particle Diameter (nm)
                      10
                  Particle Diameter (nm)
                                                                               100
Figure 2-12(a-f).  Fitted multi-model particle size distribution at different sampling
                  distances from freeway 405 (a) 30 m downwind, (b) 60 m downwind,
                  (c) 90 m downwind, (d) 150 m downwind, (e) 300 m downwind, (f) 300 m
                  upwind.  Size distributions were normalized to the control CPC's
                  reading.  Note different scales for dN/d log Dp axis. Modal parameters
                  given are: geometric mean diameter, jig; and geometric standard
                  deviation, og.

Source: Zhu et al. (2002b).
June 2003
2-27
DRAFT-DO NOT QUOTE OR CITE

-------
                         1.6e+5
                         1.2e+5-
                         8.0e+4 -
                       Q
                       Oi
                       O
                         4.0e+4 -
                                      10                        100
                                           Particle Diameter (nm)
       Figure 2-12(g). (Combination of a-e with dN/d logDp scale.) Ultrafine particle size
                      distribution at different sampling locations near the 405 freeway in
                      Los Angeles, CA.
       Source: Zhu et al. (2002b).
 1     concentration of SO2 and hydroxyl radical («OH), which is generated primarily by reactions
 2     involving ozone (O3). Thus, reductions in SO2 and O3 would lead to a decrease in the rate of
 3     generation of H2SO4 and condensable organic vapor and to a decrease in the concentration of
 4     nuclei-mode particles.  The balance between formation and removal is uncertain. However,
 5     these processes can be modeled using a general dynamic equation for particle size distribution
 6     (Friedlander, 2000) or by aerosol dynamics modules in newer air quality models (Binkowski and
 7     Shanker, 1995; Binkowski and Ching, 1995).
 8
 9     2.1.3   Chemistry of Atmospheric Participate Matter
10          The major constituents of atmospheric PM are sulfate, nitrate, ammonium, and hydrogen
11     ions; particle-bound water;  elemental carbon; a great variety of organic compounds; and crustal
12     material. Atmospheric PM also contains a large number of elements in various compounds and
13     concentrations.  More information and references on the composition of PM measured in a large
14     number of studies in the United States, may be found in 1996 PM AQCD (U.S. Environmental
       June 2003
2-28
DRAFT-DO NOT QUOTE OR CITE

-------
 1     Protection Agency, 1996a). In this document, the composition and concentrations of PM are
 2     discussed in Chapter 3, Section 3.1, Patterns and Trends in Ambient PM25 Concentrations and
 3     ambient data for concentrations and composition of PM25 are given in Appendices 3 A, 3B,
 4     and 3C.
 5
 6     2.1.3.1   Chemical Composition and Its Dependence on Particle Size
 7           Studies conducted in most parts of the United States indicate that sulfate, ammonium, and
 8     hydrogen ions; elemental carbon, secondary organic compounds and primary organic species
 9     from cooking and combustion; and certain transition metals are found predominantly in the fine
10     particle mode. Crustal materials such as calcium, aluminum, silicon, magnesium, and iron are
11     found predominately in the coarse particles.  Some primary organic materials such as pollen,
12     spores, and plant and animal debris are also found predominantly in the coarse mode. Some
13     components such as potassium and nitrate may be found in both the fine and coarse particle
14     modes but from different sources or mechanisms. Potassium in coarse particles comes from soil.
15     Potassium also is found in fine particles in emissions from burning wood or cooking meat.
16     Nitrate in fine particles comes primarily from the reaction of gas-phase nitric acid with gas-
17     phase ammonia to form particulate ammonium nitrate. Nitrate in coarse particles comes
18     primarily from the reaction of gas-phase nitric acid with preexisting coarse particles.
19
20     2.1.3.2   Primary and Secondary Particulate Matter
21           Particulate material can be primary or secondary. PM is called "primary" if it is in the
22     same chemical form in which it was emitted into the atmosphere. PM is called "secondary"  if it
23     is formed by chemical reactions in the atmosphere.  Primary coarse particles are usually formed
24     by mechanical processes. This includes material emitted in particulate form such as wind-blown
25     dust, sea salt, road dust, and combustion-generated particles such as fly ash and soot. Primary
26     fine particles are emitted from sources either directly as particles or as vapors that rapidly
27     condense to form ultrafme or nuclei-mode particles.  This includes soot from diesel engines,
28     a great variety of organic compounds condensed from incomplete combustion or cooking, and
29     compounds of As, Se, Zn, etc., that condense from vapor formed during combustion or smelting.
30     The concentration of primary particles depends on their emission rate, transport and dispersion,
31     and removal rate from the atmosphere.

       June 2003                               2-29        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Secondary PM is formed by chemical reactions of free, adsorbed, or dissolved gases.  Most
 2      secondary fine PM is formed from condensable vapors generated by chemical reactions of
 3      gas-phase precursors. Secondary formation processes can result in either the formation of new
 4      particles or the addition of particulate material to pre-existing particles. Most of the sulfate and
 5      nitrate and a portion of the organic compounds in atmospheric particles are formed by chemical
 6      reactions in the atmosphere.  Secondary aerosol formation depends on numerous factors
 7      including the concentrations of precursors; the concentrations of other gaseous reactive species
 8      such as ozone, hydroxyl radical,  peroxy radicals,  or hydrogen peroxide; atmospheric conditions
 9      including solar radiation and relative humidity (RH); and the interactions of precursors and
10      pre-existing particles within cloud  or fog droplets or in the liquid film on solid particles. As a
11      result, it is considerably more difficult to relate ambient concentrations of secondary species to
12      sources of precursor emissions than it is to identify the sources of primary particles.
13      A significant effort is currently being directed toward the identification and modeling of organic
14      products of photochemical smog including the conversion of gases to particulate matter. More
15      information of the transformation of precursor gases into secondary PM is given in Chapter 3,
16      Section 3.3.1, Chemistry of Secondary PM Formation.
17           Particle  strong acidity is due  almost entirely to H2SO4 or NH4HSO4. Thus, the acidity of
18      atmospheric particles depends on both the amount of SO2 that is oxidized to SO3 and
19      subsequently forms H2SO4 and the amount of ammonia available to react with the  sulfuric acid.
20      Nitric acid is more volatile than sulfuric acid. Thus, if gas phase SO3 or sulfuric acid or particles
21      containing H2SO4 or NH4HSO4 contact particles containing NH4NO3, nitric acid gas will be
22      released with the remaining ammonia contributing to further neutralization of the acid.  Little
23      NH4NO3 is found in  atmospheres containing significant particle strong acidity. However, as SO2
24      emissions are reduced to the point that there is more than enough ammonia to neutralize the
25      sulfuric acid, NH4NO3 particles will begin to form.  Thus, ammonia emissions and
26      concentrations relative to those of SO2 and H2SO4 are important in determining the strong acidity
27      in the atmosphere and the  concentration of particulate NH4NO3.  Therefore, once SO2 emissions
28      have been reduced to the point that ammonia and sulfate are in balance to form (NH4)2SO4,
29      further reductions in SO2 will not result in an equivalent reduction in airborne PM because one
30      (NH4)2SO4 unit will be replaced by two NH4NO3 units.
31

        June 2003                                 2-30        DRAFT-DO NOT QUOTE OR CITE

-------
 1      2.1.3.3  Particle-Vapor Partitioning
 2           Several atmospheric aerosol species, such as ammonium nitrate and certain organic
 3      compounds, are semivolatile and are found in both gas and particle phases. A variety of
 4      thermodynamic models have been developed to predict the temperature and relative humidity
 5      dependence of the ammonium nitrate equilibria with gaseous nitric acid and ammonia.
 6      However, under some atmospheric conditions,  such as cool,  cold, or very clean air, the relative
 7      concentrations of the gas and solid phases are not accurately predicted by equilibrium
 8      considerations alone, and transport kinetics can be important. The gas-particle distribution of
 9      semivolatile organic compounds depends on the equilibrium vapor pressure of the  compound,
10      total particle surface area, particle composition, atmospheric temperature, and relative humidity.
11      Although it generally is assumed that the gas-particle partitioning of semivolatile organics is in
12      equilibrium in the atmosphere, neither the equilibria nor the kinetics of redistribution are well
13      understood. Diurnal temperature fluctuations cause gas-particle partitioning to be  dynamic on a
14      time scale of a few hours and can cause semivolatile compounds to evaporate during the
15      sampling process.  The pressure drop across the filter can also contribute to the loss of
16      semivolatile compounds.  The dynamic changes in gas-particle partitioning caused by changes in
17      temperature,  pressure, and gas-phase concentration, both in the atmosphere and after collection,
18      cause serious sampling problems that are discussed in Section 2.2.3, Measurement of
19      Semivolatile Particulate Matter.
20
21      Equilibria with Water Vapor
22           As a result of the equilibrium of water vapor with liquid water in hygroscopic particles,
23      many ambient particles contain liquid water (particle-bound water).  Unless removed, this
24      particle-bound water will be measured as a component of the particle mass. Particle-bound
25      water is important in that it influences the size of the particles, and in turn, their light scattering
26      properties and their aerodynamic properties, which are important for deposition to  surfaces, to
27      airways following inhalation, and in sampling instrumentation.  The aqueous solution provides a
28      medium for reactions of dissolved gases including reactions that do not take place  in the gas
29      phase.  The aqueous solutions also may act as a carrier to convey soluble toxic species to the
30      gas-exchange regions of the respiratory system, including  species that would be removed by
31      deposition in the upper airways if the particles had remained in the gas phase (Friedlander and

        June 2003                                2-31        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Yeh, 1998; Kao and Friedlander, 1995; Wilson, 1995).  An extensive review of equilibrium with
 2      water vapor as it pertains to ambient aerosols was given in Chapter 3 of the 1996 PM AQCD
 3      (U.S. Environmental Protection, Agency, 1996a).
 4           The interaction of particles with water vapor may be described briefly  as follows.
 5      As relative humidity increases, particles of crystalline soluble salts, such as (NH4)2SO4,
 6      NH4HSO4, or NH4NO3, undergo a phase transition to become aqueous solution particles.
 7      According to the phase rule, for particles consisting of a single component, this phase transition
 8      is abrupt, taking place at a relative humidity that corresponds to the vapor pressure of water
 9      above the saturated solution (the deliquescence point).  With a further increase in relative
10      humidity, the solution particle adds water (and the concentration of the solute decreases) so that
11      the vapor pressure of the solution is maintained equal to that of the surrounding relative
12      humidity; thus, the solution particle tends to follow the equilibrium growth curve.  As relative
13      humidity decreases,  the solution particle follows the equilibrium curve to the deliquescence
14      point. However, rather than crystallizing at the deliquescence relative humidity, the solute
15      remains dissolved in a supersaturated solution to considerably lower relative humidities.
16      Ultimately the  solution particle abruptly loses its water vapor (efflorescence) and typically
17      returns to the initial  crystalline form.
18           For particles consisting of more than one component, the solid to liquid transition will take
19      place over a range of relative humidities with an abrupt onset at the lowest deliquescence point
20      of the several components and with subsequent growth as crystalline material in the particle
21      dissolves according  to the phase diagram for the particular multicomponent system. Under such
22      circumstances, a single particle may undergo several more or less abrupt phase transitions until
23      the soluble material  is fully dissolved. At decreasing relative humidity, such particles tend to
24      remain in solution to relative humidities well below the several deliquescence points. In the case
25      of the sulfuric  acid-ammonium sulfate-water system, the phase diagram is fairly well
26      understood.  For particles of composition intermediate between NH4HSO4 and (NH4)2SO4, this
27      transition occurs in the range from 40% to below 10%,  indicating that for certain compositions
28      the solution cannot be dried in the atmosphere.  At low relative humidities, particles of this
29      composition would likely be present in the atmosphere  as supersaturated solution droplets (liquid
30      particles) rather than as solid particles. Thus, they would exhibit hygroscopic rather than
31      deliquescent behavior during relative humidity cycles.

        June 2003                                 2-32        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Other pure compounds, such as sulfuric acid, are hygroscopic (i.e., they form aqueous
 2      solutions at any relative humidity and maintain a solution vapor pressure over the entire range of
 3      relative humidity). Soluble organic compounds may also contribute to the hygroscopicity of the
 4      atmospheric aerosol (Saxena et al., 1995; Saxena and Hildeman, 1996), but the equilibria
 5      involving organic compounds and water vapor, and, especially for mixtures of salts, organic
 6      compounds, and water, are not so well understood. These equilibrium processes may cause an
 7      ambient particle to significantly increase its diameter at relative humidities above about 40%
 8      (Figure 2-13).  A particle can grow to five times its dry diameter as the RH approaches 100%
 9      (Figure 2-14).  The Federal Reference Methods, for filter measurements of PM25 and PM10 mass,
10      require equilibration at a specified, low relative humidity after collection (for PM2 5, between
11      30 and 40% RH with control of ±5% RH [Code of Federal Regulations, 2001a]). This
12      equilibration removes much of the particle-bound water and provides  a relatively stable PM
13      mass for gravimetric measurements (see Section 2.2 for details and references).  Otherwise,
14      particle mass would be a function of relative humidity, and the particle mass would be largely
15      particle-bound water at higher relative humidities.  However, some particle-bound water may be
16      retained even after equilibration. Recent studies have shown that significant amounts of particle-
17      bound water are retained in particles collected on impaction surfaces even after equilibration and
18      that the amount of retained particle-bound water increases with relative humidity during
19      collection (Hitzenberger et al., 1997).
20           The retention of particle-bound water is a greater problem for continuous monitors that
21      measure changes in mass collected on a filter over long sampling times.  If particle-bound water
22      is not removed, changes in relative humidity would cause changes in the mass of PM collected
23      over previous hours or days. These changes could be much greater than amount of PM mass
24      added in one hour. Therefore, continuous monitoring techniques generally attempt to remove
25      particle-bound water before measurement either by heating or dehumidification.  However, other
26      semivolatile materials (e.g., ammonium nitrate and organic compounds) that may be partially
27      lost during sampling or equilibration of an unheated filter are certainly lost when the collected
28      sample is heated above ambient temperature. These changes in particle size with relative
29      humidity also mean that particle measurements such as surface area or volume, or composition
30      as a function of size, should be made at the same RH in order for the results are to be
31

        June 2003                                 2-33        DRAFT-DO NOT QUOTE OR CITE

-------
                              2.0-
                            O
                           "ro
                           O
                           -
                           "
                           E
                           CO
                           b
                              1.5-
                              1.0-


                               0-
                               Hygroscopic Growth
                                Curve for (H2SO4) •
                                     30
                                      Hysteresis Loop
                                      for (NH4)2 S04
                                     	M	
                                               50
                                               RH (%)
                                                        i
                                                        70
                                                                 - 7
                                                                 - 6
                                                           - 5  .y
                                                                      CO
                                                                      o:
                                                                 - 4
                                                                 - 3
                                                           - 2
                                                                 - 1
                                                                90
                                                               I
                                                               0
                                                               CD
                                                               E
      Figure 2-13.  Particle growth curves showing fully reversible hygroscopic growth of
                    sulfuric acid (H2SO4) particles, deliquescent growth of ammonium sulfate
                    [(NH4)2 SO4] particles at the deliquescent point (A, about 80% relative
                    humidity [RH]), reversible hygroscopic growth of ammonium sulfate
                    solution droplets at RH greater than 80%, and hysteresis (the droplet
                    remains supersaturated as the RH decreases below 80%) until the
                    crystallization point (B, about 38% RH) is reached (adapted from National
                    Research Council, 1993 and Tang, 1980).

      Source: National Research Council (1993) adapted from Tang (1980).
1

2

3

4

5
comparable. These problems are addressed in more detail in Section 2.2, Measurement of
Particulate Matter.


2.1.3.4  Atmospheric Lifetimes and Removal Processes
     The lifetimes of particles vary with size. Nuclei-mode particles rapidly grow into the
accumulation mode. However, the accumulation mode does not grow into the coarse mode.
Accumulation-mode fine particles are kept suspended by normal air motions and have very low
deposition rates to surfaces.  They can be transported thousands of km and remain in the
atmosphere for a number of days.  Coarse particles can settle rapidly from the atmosphere within
      June 2003
                                         2-34
DRAFT-DO NOT QUOTE OR CITE

-------
                                                                          216
                                        i           i
                                       Theoretical Prediction at 22 c
                                ooooo Experimental Measurements
                                                             150
                      200
                                     NH4 HSO4 Dry Particle Diameter (nm)
       Figure 2-14.  Theoretical predictions and experimental measurements of growth of
                     NH4HSO4 particles at relative humidity between 95 and 100%.
       Source: Lietal. (1992).
 1     hours and normally travel only short distances. However, when mixed high into the atmosphere,
 2     as in dust storms, the smaller-sized coarse-mode particles have longer lives and travel greater
 3     distances. Dry deposition rates are expressed in terms of a deposition velocity that varies with
 4     particle size, reaching a minimum between 0.1 and 1.0 jim aerodynamic diameter (e.g., Lin
 5     et al., 1994). Accumulation-mode particles are removed from the atmosphere primarily by cloud
 6     processes.  Fine particles, especially particles with a hygroscopic component, grow as the
 7     relative humidity increases, serve as cloud condensation nuclei, and grow into cloud droplets.
 8     If the cloud droplets grow large enough to form rain, the particles are removed in the rain.
 9     Falling rain drops impact coarse particles and remove them.  Ultrafine or nuclei-mode particles
10     are small enough to diffuse to the falling drop, be captured, and be removed in rain. Falling rain
11     drops, however, are not nearly as effective in removing accumulation-mode particles as the
12     cloud processes mentioned above.  A more detailed discussion of particle deposition, including
13     acid deposition, especially as it applies to deposition to vegetation, soil, and water surfaces is
14     given in Chapter 4 (Environmental Effects of Airborne Particulate Matter).  Acid deposition and
       June 2003
2-35
DRAFT-DO NOT QUOTE OR CITE

-------
 1      PM are intimately related, first, because particles contribute to the acidification of rain and,
 2      secondly, because the gas-phase species that lead to dry deposition of acidity are also precursors
 3      of particles.  Therefore, reductions in SO2 and NOX emissions will decrease both acidic
 4      deposition and PM concentrations.
 5           Sulfate, nitrate, and some partially oxidized organic compounds are hygroscopic and act as
 6      nuclei for the formation of cloud droplets. These droplets serve as chemical reactors in which
 7      (even slightly) soluble gases can dissolve and react. Thus, SO2 can dissolve in cloud droplets
 8      and be oxidized to sulfuric acid by dissolved ozone or hydrogen peroxide. These reactions take
 9      place only in aqueous solution, not in the gas phase. Sulfur dioxide also may be oxidized by
10      dissolved oxygen.  This process will be faster if metal catalysts such as iron or manganese are
11      present in solution. If the droplets evaporate, larger particles are left behind. If the droplets
12      grow large enough, they will fall as rain; and the particles will be removed from the atmosphere
13      with potential effects on the materials,  plants, or soil on which the rain falls.  (Similar
14      considerations apply to dew.)  Atmospheric particles that nucleate cloud droplets  also may
15      contain other soluble or nonsoluble materials such as metal salts and organic compounds that
16      may add to the toxicity of the rain.  Sulfuric acid, ammonium nitrate, ammonium  sulfates, and
17      organic particles also are deposited on  surfaces by dry deposition. The utilization of ammonium
18      by plants leads to the production of acidity. Therefore, dry deposition  of particles can also
19      contribute to the ecological impacts of acid deposition. These effects are discussed in Chapter 4
20      (Environmental Effects of Airborne Particulate Matter).
21
22      2.1.4   Comparison of Fine and Coarse Particles
23           The physical and chemical properties of fine particles (including ultrafine particles and
24      accumulation-mode particles)  and coarse particles are summarized for  comparison purposes in
25      Table 2-1. These include important differences in sources, formation mechanisms, composition,,
26      atmospheric residence time, removal processes, and travel distances. Ensuing chapters in this
27      document will also show that fine and coarse particles differ in aspects of concentrations,
28      exposure, dosimetry, toxicology, and epidemiology. Collectively, these differences continue to
29      warrant consideration of fine particles as a separate air pollutant class from coarse particles and
30      the setting of separate standards for fine and coarse particles.
31

        June 2003                                  2-36        DRAFT-DO NOT QUOTE OR CITE

-------
                 TABLE 2-1.  COMPARISON OF AMBIENT PARTICLES,
  FINE PARTICLES (Ultrafine Plus Accumulation-Mode) AND COARSE PARTICLES
                                     Fine
                    Ultrafine
                              Accumulation
                                           Coarse
Formation
Processes:

Formed by:
 Composed
 of:
 Solubility:
 Sources:
            Combustion, high-temperature
         processes, and atmospheric reactions
Nucleation
Condensation
Coagulation
Sulfate
Elemental Carbon
Metal compounds
Organic compounds
with very low
saturation vapor
pressure at ambient
temperature
Probably less soluble
than accumulation
mode

Combustion
Atmospheric
transformation of SO2
and some organic
compounds
High temperature
processes
Atmospheric   Minutes to hours
half-life:
Removal
Processes:
Travel
distance:
Grows into
accumulation mode
Diffuses to raindrops


-------
 1      2.2  MEASUREMENT OF PARTICULATE MATTER
 2           The 1996 PM AQCD (U.S. Environmental Protection Agency, 1996a) summarized
 3      sampling and analytical techniques for PM and acid deposition that had appeared in the literature
 4      since the earlier 1982 PM AQCD (U.S. Environmental Protection Agency, 1982). Excellent
 5      reviews have also been published by Chow (1995) and McMurry (2000).  This section discusses
 6      problems in measuring PM; new techniques that attempt to alleviate these problems or measure
 7      problem species; Federal Reference Methods, speciation monitors, analytical methods for
 8      inorganic elements, organic and elemental carbon, and ionic species; and continuous and
 9      multiday monitors.
10
11      2.2.1   Particle Measurements of Interest
12           There are many PM components and parameters that are of interest across the various
13      types of uses to which PM measurement data are applied.  These uses include analyses of
14      compliance with air quality  standards and trends; source category apportionment studies related
15      to the develop of pollution reduction strategies and the validation of air quality models; studies
16      related to health, ecological, and radiative effects; and characterization of current air quality for
17      presentation to the public in the context of EPA's Air Quality Index. PM measurement
18      components and parameters of specific interest for these various purposes are noted below and
19      summarized in Table 2-2.
20           Particle measurements are needed to determine if a location is in compliance with air
21      quality standards, to determine long-term trends in air quality patterns, and for epidemiologic
22      studies. For these purposes, precision of the measurements by a variety of measurement
23      instruments in use is a critical consideration. Therefore, intercomparisons of various samplers
24      under a variety of atmospheric and air quality conditions are important.
25           In order to reduce pollution to attain a standard, pollution control agencies and national
26      research organizations need measurements to identify source categories and to develop and
27      validate air quality models.  For these purposes, PM parameters other than mass, such as
28      chemical composition and size distribution, must also be measured.  Moreover, measurements
29      are needed with shorter time resolution in order to match changes in pollution with diurnal
30      changes in the boundary layer.
31

        June 2003                                 2-3 8        DRAFT-DO NOT QUOTE OR CITE

-------
            TABLE 2-2. PARTICULATE MATTER COMPONENTS/PARAMETERS OF
              INTEREST FOR HEALTH, ECOLOGICAL, OR RADIATIVE EFFECTS;
                    FOR SOURCE CATEGORY APPORTIONMENT STUDIES;
         	OR FOR AIR QUALITY MODELING EVALUATION STUDIES	

         Particle number

         Particle surface area

         Particle size distribution

         PM mass (fine PM mass [PM2 5] and coarse thoracic PM mass [PM10_2 5]) including both
         nonvolatile mass as measured by the current Federal Reference method and total mass
         (including semivolatile components such as ammonium nitrate and semivolatile organic
         compounds, but not particle-bound water)

         Ions (sulfate, nitrate, and ammonium)

         Strong acidity (FT)

         Elemental carbon

         Organic carbon (total, nonvolatile, and semivolatile; functional groups and individual
         species)

         Transition metals (water soluble, bioavailable, oxidant generation)

         Specific toxic elements and organic compounds

         Crustal elements

         Bioaerosols

         Particle refractive index (real and imaginary)

         Particle density

         Particle size change with changes in relative humidity
1          A number of PM measurements are needed for use in epidemiologic and exposure studies

2     and to determine components of PM to guide planning and interpretation of toxicologic studies.

3     For these purposes, size and chemical composition measurements are important, as is

4     measurement across different time intervals. For epidemiologic studies of acute (i.e., short-term)

5     PM exposures, 1-h or continuous measurements can provide important information beyond that

6     provided by 24-h measurements. However, for epidemiologic studies of chronic PM exposures,

7     measurements that permit integration over longer intervals (e.g., a week to a month)  are more

8     relevant.  For dosimetric studies and modeling, information will be needed on the particle size

      June 2003                                2-39       DRAFT-DO NOT QUOTE OR CITE

-------
 1      distribution and on the behavior of particles as the relative humidity and temperature changes
 2      found in the atmosphere are increased to those found in the respiratory system.
 3           For studies of ecological effects and materials damage, measurements of particles and of
 4      the chemical components of particulate matter in rain, fog, and dew are needed to understand the
 5      contributions of PM to soiling of surfaces and damage to materials and to understand the wet and
 6      dry deposition of acidity and toxic substances to surface water, soil, and plants. Some
 7      differentiation into particle size is needed to determine dry deposition.
 8           For studies of visibility impairment and radiative effects, information is needed that relates
 9      to how particles scatter and absorb light, including refractive index, ratio of scattering to
10      absorption, size distribution, and change in particle size with change in  relative humidity.
11
12      2.2.2   Issues in Measurement of Particulate Matter
13           The EPA decision to revise the PM standards by adding daily and yearly standards for
14      PM2 5 has led to a renewed interest in the measurement of atmospheric particles and also to a
15      better understanding of the problems in developing precise and accurate measurements of
16      particles. It is very difficult to measure and characterize particles suspended in the atmosphere;
17      however, numerous improvements in PM monitoring are in use and others are in development.
18      EPA's PM standards are based, in part, on epidemiologic relationships between health effects
19      and PM concentrations as measured with existing monitoring methods.  As understanding of
20      suspended PM has advanced and new monitoring information has become available, EPA has
21      changed the indicator for the PM NAAQS from TSP to PM10, and added PM25. During the
22      current PM NAAQS review, consideration will be given to a standard for coarse thoracic  PM.
23           The U.S. Federal Reference Methods (FRM) for PM25 and PM10 provide relatively precise
24      (±10 %) methods for determining the mass of material remaining on a Teflon filter after
25      equilibration.  However, numerous uncertainties remain as to the relationship between the mass
26      and composition of material remaining on the filter as determined by the FRM measurement
27      procedure and the mass and composition of material that existed in the atmosphere as suspended
28      PM.  As a result, EPA defines accuracy for PM measurements in terms  of agreement of a
29      candidate sampler with a reference sampler.  Therefore, intercomparison of samplers is very
30      important in determining how well various samplers agree and how various design choices
31      influence what is actually measured.

        June 2003                                 2-40       DRAFT-DO NOT QUOTE OR  CITE

-------
 1           There are eight general areas where choices are made in the design and use of an aerosol
 2      sampler.  These include (1) consideration of positive artifacts due to chemical reaction or
 3      adsorption; (2) treatment of semivolatile components; (3) selection of particle size cut
 4      characteristics for the upper cut point; (4) separation of fine and coarse PM; (5) treatment of
 5      pressure, temperature, and relative humidity; (6) time resolution; (7) assessment of the reliability
 6      of the measurement technique; and (8) operation and maintenance procedures needed to
 7      maintain consistent measurements over time. In many cases, choices have been made without
 8      adequate recognition of the consequences. As a result, measurement methods developed by
 9      different organizations may give different results when sampling the same atmosphere even
10      though the techniques appear to be similar.
11
12      2.2.2.1   Artifacts Due to Chemical Reactions
13           When TSP was  collected on glass fiber filters, the reaction of SO2 (and other acid gases)
14      with basic sites on the glass fiber or with basic coarse particles on the filter led to the formation
15      of sulfate (or other nonvolatile sales, e.g., nitrate, chloride). These positive artifacts led to
16      overestimation of mass, sulfate (and probably also of nitrate). The metal impurities in the glass
17      fiber caused a high background that led to low precision in the measurement of trace metals.
18      These problems were largely overcome by changing to quartz fiber or Teflon filters and by
19      separate collection of PM25. However, the possible reaction of acidic gases with basic coarse
20      particles remains a possibility, especially with measurements of PM10 and PM10_25. The reaction
21      of NH3 with acidic particles, either during sampling or during transportation, storage, and
22      equilibration remains a problem in areas such as the eastern U.S. where PM is frequently acidic.
23      Techniques have been developed that overcome this problem by use of a denuder to remove NH3
24      during sampling and to protect the collected PM from NH3 (Suh et al., 1992, 1994; Brauer et al.,
25      1991; Koutrakis et al., 1988a,b).  However, this technique has been applied primarily for
26      measurement of particle strong acidity, not for the measurement of artifact-free ammonium or
27      mass concentrations.  In the measurement of particle strong acidity, basic coarse particles must
28      be separated from acidic fine particles (Koutrakis et al., 1992).
29
30
31

        June 2003                                2-41        DRAFT-DO NOT QUOTE OR CITE

-------
1
2
3
4
5
6
7
2.2.2.2   Treatment of Semivolatile Components of Particulate Matter
     Current filtration-based mass measurements can experience significant evaporative losses,
during and possibly after collection, of a variety of semivolatile components (i.e., species that
exist in the atmosphere in dynamic equilibrium between the condensed phase and gas phase).
Important examples include ammonium nitrate, semivolatile organic compounds, and particle-
bound water.  This problem is illustrated in Figure 2-15.
                          o.
                          d
                          05
                          o
                          w
                          (D
                          E
                                Should be
                                retained
                                         (NH4)XS04
                                          X= 0 to 2
                                        Mineral/Metal
                            0.1                             1.0      2.5
                                         Aerodynamic Diameter (|jm)
                           hi::: Semivolatile components subject to evaporation during or after sampling
      Figure 2-15.  Schematic showing major nonvolatile and semivolatile components of PM25.
                    Semivolatile components are subject to partial to complete loss during
                    equilibration or heating. The optimal technique would be to remove all
                    particle-bound water but no ammonium nitrate or semivolatile organic PM.
1
2
3
4
     Possible approaches that have been used to address the problem of potentially lost
semivolatile components include those that follow, which will be discussed in more detail in
subsequent sections.
      June 2003
                                         2-42
DRAFT-DO NOT QUOTE OR CITE

-------
 1        1.    Collect/measure all components present in the atmosphere in the condensed phase except
              particle-bound water.  (Examples:  Brigham Young absorptive sampler and Harvard
              pressure drop monitor. Both require preconcentration of the accumulation mode and
              reduction of ambient humidity.)
 2
 3        2.    Stabilize PM at a specified temperature high enough to remove all, or almost all, particle-
              bound water.  This results in loss of much of the semivolatile PM.  (Examples: tapered
              element oscillating microbalance (TEOM) operated at 50 °C, beta gauge with heated
              inlet.)
 4
 5        3.    Equilibrate collected material at fixed, near-room temperature and moderate relative
              humidity to reduce particle-bound water.  Accept the loss of an unknown but possibly
              significant fraction of semivolatile PM. (Example:  U.S. Federal Reference Method and
              most filter-weighing techniques.) Equilibration originally was designed to remove
              adsorbed water vapor  from glass fiber filters in order to maintain a stable filter weight.
              The designated RH (35±5%) was a compromise.  If the RH is too low, electrostatic
              charging becomes a problem. The equilibration process does help provide a stable and
              reproducible mass. It  also reduces the particle-bound water. However, it may not
              remove all particle-bound water.
 6
 7     2.2.2.3   Upper Cut Point
 8          The upper cut point of the high volume sampler varied with wind speed and direction.
 9     Newer PM samplers are usually designed to have an upper cut point and its standard deviation
10     that are independent of wind direction and relatively independent of wind speed. Current PM
11     samplers have upper cut points that are stable under normal operating conditions. However,
12     problems may occur under unusual or adverse conditions. Ono et al.  (2000) reported the results
13     of a study in which several PM10 samplers were collocated and operated at various sites at
14     Owens Lake, CA, a location with high concentrations of coarse PM.  Samplers included the
15     Partisol sampler, the TEOM,  a dichotomous sampler, the Wedding high-volume sampler,  and the
16     Graseby high-volume sampler.  They found that the TEOM and Partisol samplers agreed to
17     within 6% on average. The dichotomous sampler and the Graseby and Wedding high-volume

       June 2003                                2-43        DRAFT-DO NOT QUOTE OR CITE

-------
 1      samplers, however, measured significantly lower PM10 concentrations than the TEOM (on
 2      average 10, 25, and 35% lower, respectively).  These lower concentrations were attributed to a
 3      decrease in the cut point at higher wind speeds and to a dirty inlet.
 4           The choice of the cut point characteristics depends upon the application for the sampling
 5      device.  A separation that simulates the removal of particles by the human upper respiratory
 6      system might appear to be a good choice for both health risk and regulatory monitoring (i.e.,
 7      measure what gets into the lungs).  The ACGIH-ISO-CEN penetration curve for thoracic
 8      particles (particles able to pass the larynx and penetrate into the bronchial and alveolar regions of
 9      the lung) has a 50%  cut point at 10 jim aerodynamic diameter (Da).  The U.S. PM10 separation
10      curve is sharper than the thoracic penetration curve but has the advantage of reducing the
11      problem of maintaining the finite collection efficiency specified by the thoracic penetration
12      curve for particles larger than 10 |im Da. (See  Section 2.1.2.2 and Figure 2-6).
13
14      2.2.2.4  Cut Point  for Separation of Fine  and Coarse Particulate Matter
15           As shown in Table 2-1, fine and coarse particles differ not only in size but also in
16      formation mechanisms; sources; and chemical, physical, and biological properties. They also
17      differ in concentration-exposure relationships,  dosimetry (deposition and retention in the
18      respiratory  system),  toxicity, and health  effects as observed by epidemiologic studies.  Thus, it is
19      desirable to measure fine and coarse PM separately in order to properly allocate health effects to
20      either fine or coarse  PM and to correctly determine sources by receptor modeling approaches.
21      For example, sulfates in fine particles are associated with hydrogen or ammonium ions while
22      sulfates in coarse particles are associated with basic metal ions. Transition metals in coarse
23      particles are likely to be associated with soil  and tend to be less soluble (and presumably less
24      bioavailable) than transition metals in fresh combustion particles found in fine particles.
25           The 2.5 jim Da cut point was  chosen in  the early 1970s as the cut point for a new
26      dichotomous sampler (Loo et al., 1976; Jaklevic et al., 1977) for use  in the Regional Air
27      Pollution Study in St. Louis, MO.  At that time aerosol scientists were beginning to realize that
28      there was a minimum between 1 and 3 |im in the distribution of particle size by volume (Whitby
29      et al., 1974). The 2.5  jim cut point was subsequently used as an indicator of fine-mode PM in a
30      number of studies including the Harvard Six-City Studies of the relationships between mortality
31      and PM concentrations (Dockery et al., 1993; Schwartz et al.,  1996). A 2.5 jim cut point was

        June  2003                                 2-44        DRAFT-DO NOT QUOTE OR CITE

-------
 1      also used in the Inhalable Particle Network (Suggs and Burton, 1983) which provided data for
 2      another major epidemiologic study of PM - mortality relationships using an American Cancer
 3      Society cohort (Pope et al., 1995). Therefore, at the time of the last review of the NAAQS for
 4      PM (U.S. Environmental Protection Agency, 1996a,b), there were a number of epidemiologic
 5      studies demonstrating a statistical relationship between PM2 5 concentrations and mortality.
 6           It is now understood that the size range between 1.0 and 2.5 jim, sometimes called the
 7      intermodal region may contain either accumulation-mode or coarse-mode material or both, i.e.,
 8      the two modes may overlap in this region (Kegler et al., 2001). The experimental information on
 9      the composition and source of the intermodal mass was discussed extensively in the 1996 PM
10      AQCD (U.S.  Environmental Protection Agency, 1996a).  Depending on conditions, a significant
11      amount of either accumulation- or coarse-mode material may be found in the intermodal region
12      between 1.0 and 2.5 |im.  The analysis demonstrated the important role of relative humidity in
13      influencing the size of particles in both the accumulation and coarse modes.
14           As the relative humidity increases, hygroscopic accumulation-mode particles will increase
15      in size due to accumulation of particle-bound water. At high relative humidities, some originally
16      submicrometer accumulation-mode PM may be found with a Da above 1 jim.  At a relative
17      humidity of 100%, such as found in fog and clouds, accumulation-mode PM may extend above
18      2.5 |im Da. What is not well understood is whether such particles will shrink to diameters below
19      1 |im as the RH decreases or whether reactions occurring in the wet particles will result in an
20      increase in non-aqueous mass so that even at low RH the diameters would exceed 1 |im. On the
21      other hand, at very low relative humidity, coarse particles may be fragmented into smaller sizes,
22      and small amounts of coarse PM may be found with an Da below 2.5 jim (Lundgren et al., 1984;
23      Lundgren and Burton, 1995). Thus, a PM2 5 sample will contain all of the accumulation-mode
24      PM except during periods of RH near 100 %.  However, under conditions of low RH, it may also
25      contain a small fraction of the coarse PM.  The selection of a cut point of 2.5 jim as a basis for
26      EPA's 1997 NAAQS for fine particles (Federal Register, 1997) and its continued use in many
27      health effects studies reflect the importance placed on more complete inclusion of accumulation-
28      mode particles while recognizing that intrusion of coarse particles can occur under some
29      conditions with this cut point.
30           In addition to the influence of relative humidity, in areas where winds cause high
31      concentrations of windblown soil there is evidence that a significant amount of coarse-mode PM

        June 2003                                 2-45        DRAFT-DO NOT QUOTE OR CITE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
may be found below 2.5 jim. An example, taken from data collected during the August 1996
dust storm in Spokane, WA, is shown in Figure 2-16. Note that the PM10 scale is 10 times that
of the other size fractions. PMj, although high in the morning, goes down as the wind increases
and PM10, PM2 5, and PM^.j go up.  During the peak of the dust storm, around 9 pm, PM2 ^ was
88% of PM25. For the 24-h period, PM25.l was 54% of PM25. However, PMj was not affected
by the intrusion  of coarse particles.  Similar considerations probably apply to short-term
intrusions of dust transported from distant sources such as the Sahara and Gobi deserts (Husar
etal.,  2001).
                                                                                     600
                                                                                     500
                                                                                     400
                                                                                     300
                                                                                     200
                                                                                     100
              12am 2am  4am  6am   Sam  10am 12pm 2pm  4pm  6pm  8pm  10pm
                                         Time, August
       Figure 2-16.  Particulate matter concentrations in Spokane, WA, during the August 30,
                    1996 dust storm.
       Source: Claiborn et al. (2000).
                                                                                          o>
                                                                                          n.
                                                                                          E
                                                                                          CL
 1          A cut point of 1.0 jim could reduce the misclassification of coarse-mode material as fine,
 2     especially in areas with high levels of wind blown soil, but under high RH conditions could
 3     result in some accumulation-mode material being misclassified as coarse.  A reduction in RH,
       June 2003
                                        2-46
DRAFT-DO NOT QUOTE OR CITE

-------
 1      either intentionally or inadvertently, will reduce the mass mean diameter of the accumulation-
 2      mode particles. A sufficient reduction in RH should yield a dry accumulation mode with very
 3      little material above 1.0 jim. Studies of the changes in particle size with changes in relative
 4      humidity suggest that only a small fraction of accumulation-mode particles will be above 1.0 jim
 5      in diameter at RH below 60%, but a substantial fraction will grow above 1.0 jim for RH above
 6      80% (Hitzenberger et al., 1997; McMurry and Stolzenburg, 1989; U.S. Environmental Protection
 7      Agency, 1996a).
 8           Under high relative humidity circumstances, a monitor using a 1.0 jim Da cut point can
 9      achieve better modal separation if the air stream is dehumidified to  some fixed humidity that
10      would remove all or most particle-bound water without evaporating semivolatile components.
11      New techniques which require reduction of RH prior to collection have been developed for
12      measurement of fine particulate matter minus particle-bound water but including semivolatile
13      nitrate and organic compounds. With such techniques, PMX measurements in conjunction with
14      concurrent PM2 5 measurements would be useful for exposure, epidemiologic, and source
15      apportionment studies especially in areas where intrusion of coarse-mode particles into the
16      intermodal range is likely.
17
18      2.2.2.5  Treatment of Pressure, Temperature, and Relative Humidity
19           There are a variety of techniques for defining (or ignoring) the pressure, temperature,  and
20      relative humidity during and after sampling. For example, the sample volume may  be based on
21      the mass or volumetric flow corrected to standard temperature and pressure (273 °K and 1 atm.)
22      (current FRM for PM10), or it may be based on the volumetric flow  at ambient conditions of
23      temperature and pressure (current FRM for  PM2 5).
24           There are also a variety of options for the control of temperature during collection.  The
25      particles may be heated enough to remove much of the particle-bound water (i.e., TEOM at
26      50 °C); the particles may be heated several degrees, just enough to prevent condensation of
27      water in the sampling system; the particles and the sampler may be  maintained near ambient
28      temperature (±5 °C of ambient temperature  is required for EPA FRM samplers); or the particles
29      and sampler may be maintained at constant  temperature inside a heated or air conditioned
30      shelter. There are also options for control of temperature after collection:  (a) no control (room
        June 2003                                 2-47       DRAFT-DO NOT QUOTE OR CITE

-------
 1      temperature) or (b) ship and store at cool temperature (4 °C is the current EPA FRM
 2      requirement).
 3           Consideration must also be given to relative humidity.  Changes in relative humidity cause
 4      changes in particle size of hygroscopic or deliquescent particles.  Changing relative humidity by
 5      adding or removing water vapor affects measurements of particle number, particle surface area,
 6      and particle size distribution and the amount of overlap of accumulation-mode and coarse-mode
 7      particles.  Changing relative humidity by intentional or inadvertent changes in temperature also
 8      affects the amount of loss of ammonium nitrate and semivolatile organic compounds.
 9      Monitoring personnel should be aware of the various options for treatment of pressure,
10      temperature, and relative humidity; make appropriate selections; and document which options
11      are used.
12           Studies of relationships between personal/indoor/outdoor measurements present special
13      problems.  Indoor  environments are typically dryer than outdoors and may be warmer or, if
14      air-conditioned,  cooler.  These differences may change particle size and the amount of
15      volatilization of semivolatile components. Such  changes between indoors and outdoors will
16      complicate the comparison  of indoor to outdoor PM concentrations; the modeling of personal
17      exposure to all particles; and apportionment of personal exposure into particles of ambient
18      origin, particles of indoor origin, and particles originating from personal activity.
19
20      2.2.2.6  Time Resolution
21           The classic 24-hour filter collection technique is being supplemented by a variety of
22      continuous monitors for various PM constituents. This process is being accelerated by the lower
23      operational cost of continuous monitors and the availability of new continuous monitors for
24      mass, number, and certain chemical components, as well as refinements of older methods based
25      on beta attenuation or light  scattering.  Most epidemiologic studies have used 24-hour
26      concentrations as exposure  indicators.  However, one epidemiologic study of chronic effects
27      uses a filter sampler with a  two-week collection period (Gauderman et al., 2000).  Another
28      recent study used 1-2 h concentrations (see Peters et al., 2000). Continuous methods are
29      discussed in Section 2.2.5.
30
31

        June 2003                                 2-48        DRAFT-DO NOT QUOTE OR CITE

-------
 1      2.2.2.7  Accuracy and Precision
 2           Precision is typically determined by comparison of collocated samplers or through
 3      replicate analyses; whereas accuracy is determined through the use of traceable calibration
 4      standards. Unfortunately, no standard reference calibration material or procedure has been
 5      developed for suspended, atmospheric PM. It is possible to determine the accuracy of certain
 6      components of the PM measurement system (e.g., flow control, inlet aspiration, PM25 cut,
 7      weighing, etc.). The absolute accuracy for collecting a test aerosol can also be determined by
 8      isokinetic sampling in a wind tunnel.  However, it is not currently feasible to provide a simulated
 9      atmospheric aerosol with naturally occurring semivolatile components. It is particularly
10      challenging to develop an atmospheric aerosol calibration standard suitable for testing samplers
11      in the field. Therefore, it is not possible at the present time to establish the absolute accuracy of
12      a PM monitoring technique. Intercomparison studies may be used to establish the precision of
13      identical monitors and the extent of agreement between different types of monitors.  Such studies
14      are important for establishing the reliability of PM measurements.  Intercomparison studies have
15      contributed greatly to our understanding of the problems in PM measurement.  Such studies will
16      be discussed as they apply  to specific measurement problems, monitoring instruments, or
17      analytical techniques.
18           Some measurement errors of concern in PM10 sampling, including those that arise  due to
19      uncertainty tolerances in cutpoint, particle bounce and reentrainment, impactor surface
20      overloading, and losses to sampler internal surfaces, were discussed in detail in the 1996 PM
21      AQCD (U.S. Environmental Protection Agency, 1996a).  Other measurement errors of concern
22      in PM2 5 sampling arise because of our inability to assess accuracy in an absolute sense due to a
23      lack of an atmospheric aerosol calibration  standard, the inclusion in PM2 5 of a small amount of
24      coarse particles as discussed in Section 2.2.1.3, problems associated with the definition  of PM2 5
25      as what remains on a filter after collection and equilibration rather than as the mass of particles
26      as they exist in the air.  Still, it is possible to measure PM indicators with high precision.
27      Detailed information on precision and quality assurance may  be found on EPA's Technology
28      Transfer Network website (U.S. Environmental Protection Agency, 2002a). See discussion in
29      Section 2.2.4.
30           Because of the difficulties associated with determining the accuracy of PM measurements,
31      EPA has sought to make FRM measurements equivalent by specifying operating conditions and,

        June 2003                                  2-49        DRAFT-DO NOT QUOTE  OR CITE

-------
 1      in the case of PM2 5 samplers, by specifying details of the sampler design. Thus, both the PM10
 2      as well as the PM2 5 standards are defined with consistency of measurement technique rather than
 3      with the accuracy of the true mass concentration measurement in mind (McMurry, 2000).  It is
 4      acknowledged in the Federal Register (1997) that, "because the size and volatility of the particles
 5      making up ambient particulate matter vary over a wide range and the mass concentration of
 6      particles varies with particle size, it is difficult to define the accuracy of PM2 5 measurements in
 7      an absolute sense...."  Thus, accuracy is defined as the degree of agreement between a field
 8      PM2 5 sampler and a collocated PM2 5 reference method audit sampler (McMurry, 2000). The
 9      Federal Reference Method for PM2 5 is discussed in Section 2.2.3.3. As mentioned earlier,
10      volatilization of organic compounds and ammonium nitrate during sampling or post-sampling
11      handling can lead to significant underestimation of the fine particulate mass concentration in
12      some locations.  Sources of error in the measurement of mass of PM25 suspended in the
13      atmosphere also arise because of adsorption or desorption of semivolatile vapors onto or from
14      collected PM, filter media, or other sampler surfaces; neutralization of acid or basic vapors on
15      either filter media or collected PM; and artifacts associated with particle-bound water.
16          During the past 25 years, there have been advancements in the generation and
17      classification of monodisperse aerosols as well as in the development of electron microscopy and
18      imaging analysis that have contributed to the advancement in aerosol calibration (Chen,  1993).
19      Still, one of the limitations in PM sampling and analysis remains the lack of primary calibration
20      standards for evaluating analytical methods and for intercomparing laboratories. Klouda et al.
21      (1996) examined the possibility  of resuspending the National Institute of Science and
22      Technology (NIST) Standard Reference Material 1649 (Urban Dust) in air for collection on up to
23      320 filters simultaneously using Standard Research International's dust generation and collection
24      system. However, the fine component is not resuspended and the semivolatile component has
25      evaporated. Consequently, this material is not a suitable standard for suspended PM. NIST is
26      continuing work in this area with EPA support.
27          Methods validation was discussed in the  1996 PM AQCD (U.S. Environmental Protection
28      Agency, 1996a), and the usefulness of intercomparisons and "internal redundancy" was
29      emphasized.  For example, a number of internal consistency checks are applied to the IMPROVE
30      network (Malm et al.,  1994). These include mass balances, sulfur measurements by both
31      proton-induced X-ray  emission (PIXE) and ion chromatography (1C), and comparison of organic

        June 2003                                 2-50        DRAFT-DO NOT QUOTE OR CITE

-------
 1     matter by combustion and by proton elastic scattering (PESA) of hydrogen. Mass balances
 2     compare the gravimetrically determined mass with the mass calculated from the sum of the
 3     major chemical components (i.e., crustal elements plus associated oxygen, organic carbon,
 4     elemental carbon, sulfate, nitrate, ammonium, and hydrogen ions). Mass balances are useful
 5     validation techniques; however, they do not check for, or account for, artifacts associated with
 6     the absorption of gases during sampling or the loss of semivolatile material during sampling.
 7     The mass balance check may appear reasonable even if such artifacts are present because only
 8     the material collected on the filter is included in the balance.  Mass balance checks may also
 9     suffer from errors due to some particle-bound water remaining in the PM even after equilibration
10     and from the use of an arbitrary factor to account for the amount of oxygen and hydrogen atoms
11     per carbon atom in the organic carbon fractions (Andrews et al., 2000).
12
13     2.2.3   Measurement of Semivolatile Participate Matter
14           PM suspended in the atmosphere is composed of many chemical species having a wide
15     range of vapor pressures.  Substances with vapor pressures below 10"9 Torr (mmHg) will
16     typically be nonvolatile; substances with vapor pressures above 10"1 Torr will be primarily in the
17     gas phase; substances with vapor pressures between 10"1 and 10"9 Torr may exist in an
18     equilibrium state with some material  in both the gas phase and the condensed (particulate) phase
19     and are known semivolatile material (SVM) (Pankow, 1994a).  SVM, originally in the
20     atmosphere in the particulate phase and collected on a filter, may subsequently be lost from the
21     filter.  SVM may evaporate during sampling due to a reduction in its concentration in the
22     atmosphere being sampled or due to the pressure  drop across the filter. SVM may evaporate
23     after sampling; during intentional equilibration at a low relative humidity;  or during transport,
24     handling, and storage if exposed to an atmosphere in which the vapor pressure of one or more
25     semivolatile components is lower than in the atmosphere sampled. Since water is not a
26     pollutant, it is necessary to remove most of the particle-bound water before weighing (Chow,
27     1995). However, collection and measurement of ammonium nitrate and semivolatile organic
28     compounds in suspended atmospheric PM represents a major analytical challenge (McMurry,
29     2000).
30
31

       June 2003                                2-51        DRAFT-DO NOT QUOTE OR CITE

-------
 1      2.2.3.1  Particle-Bound Water
 2           It is generally desirable to collect and measure ammonium nitrate and semivolatile organic
 3      compounds.  However, for many measurements of suspended particle mass, it is desirable to
 4      remove the particle-bound water before determining the mass. The mass of particle-bound water
 5      is strongly dependent on the relative humidity.  However, the dependence is not linear since
 6      there is significant hysteresis in the water adsorption-desorption pathways (Seinfeld and Pandis,
 7      1998).  Water vapor is not a pollutant and cannot be controlled. Particle-bound water is not
 8      included in the mass of PM subject to regulation and control. Because the mass of particle-
 9      bound water could be equal or greater than that of the other components, a measurement of PM
10      mass including particle-bound water would depend more on relative humidity that pollution.
11      For all these reasons, it is usually desirable to remove most, if not all, particle-bound water
12      before weighing collected PM.  However, in  some situations it may be important to know how
13      much of the suspended particle's mass or volume results from particle-bound water.
14      Figures 2-13 and 2-14 show the change in diameter of sulfate particles as a function of relative
15      humidity.  Figure 2-13 also shows hysteresis  resulting from the difference between
16      deliquescence and crystallization points.
17           Pilinis et al. (1989) calculated the water content of atmospheric particulate matter above
18      and below the deliquescent point.  They predicted that aerosol water content is strongly
19      dependent on composition and concluded from their calculations that liquid water could
20      represent a significant mass fraction of aerosol concentration at relative humidities above 60%.
21      Since then, a few researchers have attempted to measure the water content of atmospheric
22      aerosol. Most techniques have focused on tracking the particle mass as the relative humidity is
23      changed and are still in the development phase. There have been only a few demonstrations
24      using actual ambient aerosol to date.  Of interest, in particular, is the development of the Tandem
25      Differential Mobility Analyzer (TDMA) and  its applications in investigations of the effects of
26      relative humidity on particle growth.
27           Lee et al. (1997) examined the influence of relative humidity on the size of atmospheric
28      aerosol using a TDMA coupled with a scanning mobility particle sizer (SMPS).  They reported
29      that the use of the TDMA/SMPS system allowed for the abrupt size changes of aerosols at the
30      deliquescence point to be observed precisely. They also reported that at relative humidities
        June 2003                                2-52        DRAFT-DO NOT QUOTE OR CITE

-------
 1     between 81 and 89% the water content of ammonium sulfate aerosols (by mass) ranged from
 2     47 to 66%.
 3           Andrews and Larson (1993) investigated the interactions of single aerosol particles coated
 4     with an organic film within a humid environment. Using an electrodynamic balance, they
 5     conducted laboratory experiments in which sodium chloride and carbon black particles were
 6     coated with individual organic surfactants (intended to simulate the surface-active, organic films
 7     that many atmospheric aerosol particles may exhibit) and their water sorption curves were
 8     examined. Their results showed that when ordinarily hydrophobic carbon black particles were
 9     coated with an organic surfactant, they sorbed significant amounts of water (20 to 40% of the dry
10     mass of the particle).
11           Liang and Chan (1997) developed a fast technique using the electrodynamic balance to
12     measure the water activity of atmospheric aerosols. In their technique, the mass of a levitated
13     particle is determined as the particle either evaporates or grows in response to a step change in
14     the relative humidity.  Their technique was demonstrated using laboratory experiments with
15     NaCl, (NH4)2SO4, NaNO3, and (NH4)2SO4/NH4NO3 solutions.  They concluded that one of the
16     advantages of their fast method is the  ability to measure the water activity of aerosols containing
17     volatile solutes such as ammonium chloride and some organics.
18           Mclnnes et al. (1996) measured  aerosol mass concentration, ionic composition, and
19     associated water mass of marine aerosol over the remote Pacific Ocean.  The mass of
20     particle-bound water was determined by taking the difference between the mass obtained at 48%
21     RH and at 19% RH, assuming the aerosol particles were dry at 19% RH. Based on a comparison
22     of the remote Pacific aerosol to aerosol collected at a site at the marine/continental interface of
23     the Washington coast, the amount of water associated with the aerosol was observed to be a
24     function of the ammonium to sulfate ratio. They found that the amount of water associated with
25     the submicrometer aerosol comprised  29% of the total aerosol mass collected at 47% RH and
26     9% of the total mass at 3 5% RH.
27           Ohta et al.  (1998) characterized  the chemical composition of atmospheric fine particles
28     (50% cut point of 2 jim) in Sapporo, Japan, and as part of their measurements, determined the
29     water content using the Karl Fischer method (Meyer and Boyd, 1959). After exposing a Teflon
30     filter, a portion of the filter was equilibrated at 30% RH for 24 h. Then the filter piece was
31     placed in a water evaporator heated at 150 °C, vaporizing the particle-bound water.  The vapor

       June 2003                                 2-53        DRAFT-DO NOT QUOTE OR CITE

-------
 1      evolved was analyzed for water in an aqua-counter where it was titrated coulometrically in Karl
 2      Fischer reagent solution (containing iodine, sulfur, and methanol). The accuracy of the aqua-
 3      counter is ±1 mg. Using this technique, they determined that the water content of the particles
 4      ranged from 0.4 to 3.2% of the total particulate mass (at RH < 30%).  This represents a smaller
 5      portion of water compared to their previous reported values (Ohta and Okita, 1990) that were
 6      determined by calculation at RH of 50%.
 7           Speer et al. (1997) developed an aerosol  liquid water content analyzer (LWCA) in which
 8      aerosol samples are collected on PTFE filters and then placed in a closed chamber in which the
 9      relative humidity is closely controlled. The aerosol mass is monitored using a beta-gauge, first
10      as the relative humidity is increased from low RH to high RH, and then as the RH is decreased
11      again. They demonstrated the LWCA on laboratory-generated aerosol and on an ambient PM2 5
12      sample collected in Research Triangle Park, NC.  The ambient aerosol sample was also analyzed
13      for chemical constituents.  It is interesting to note that, although their laboratory-generated
14      (NH4)2SO4 aerosol demonstrated a sharp deliquescent point, their atmospheric aerosol, which
15      was essentially (NH4)2SO4, did not  show a sharp deliquescent point.
16           Hygroscopic properties of aerosols have been studied from the viewpoint of their ability to
17      act as condensation nuclei.  The hygroscopic properties of fresh and aged carbon and  diesel soot
18      particles were examined by Weingartner et al.  (1997) who found that fresh, submicron-size
19      particles tended to shrink with increasing relative humidity because of a structural change.
20      Lamm el and Novakov (1995) found, in laboratory studies, that the hygroscopicity of soot
21      particles could be increased by chemical modification and that the cloud condensation nucleation
22      characteristics of diesel soot were similar to those of wood smoke aerosol.
23           The results of several of the above studies in which aerosol water content as a function of
24      relative humidity was determined are summarized in Figure 2-17.  In this figure, the results of
25      Lee et al. (1997), Mclnnes et al. (1996), and Ohta et al. (1998) are included.  Relative humidity
26      ranged from 9%, at which the aerosol water content was assumed to be zero (Mclnnes et al.,
27      1996), to 89%,  at which the aerosol water content was determined to be 66% by mass (Lee et al.,
28      1997).  Koutrakis et al. (1989) and Koutrakis and Kelly (1993) also have reported field
29      measurements of the equilibrium size of atmospheric sulfate particles as a function of relative
30      humidity and acidity.
31

        June 2003                                 2-54        DRAFT-DO NOT QUOTE OR CITE

-------
I
T3
                     o
                     m
   90-
   80-

«  7°-
1  60-
Q.
o  50-
0
Q  40-
cn
J  30-
^  20-
I  1CH
                                                                 • Mclnnes et al., 1996
                                                                 • Leeetal., 1997
                                                                 A Ohtaetal., 1998
                               10    20    30    40    50    60
                                              Relative Humidity (%)
                                                                70
                                                                     80
                                                                                100
       Figure 2-17.  Aerosol water content expressed as a mass percentage, as a function of
                     relative humidity.
 1           The effects of relative humidity on particle growth were also examined in several studies.
 2     Fang et al. (1991) investigated the effects of flow-induced relative humidity changes on particle
 3     cut sizes for aqueous sulfuric acid particles in a multi-nozzle micro-orifice uniform deposit
 4     impactor (MOUDI).  Laboratory experiments were conducted in which polydisperse sulfuric
 5     acid aerosols were generated and the RH was adjusted.  The aerosols were analyzed by a
 6     differential mobility analyzer.  Fang et al. (1991) observed that for inlet RH less than 80%, the
 7     cut sizes for the sulfuric acid aerosols were within 5% of that for nonhygroscopic particles
 8     except at the stage for which the cut size was 0.047 jim where the cut size was 10.7% larger than
 9     the nonhygroscopic particle cut size. They concluded that flow-induced RH changes would have
10     only a modest effect on MOUDI cut sizes at RH < 80%.
11           Hitzenberger et al. (1997) collected atmospheric aerosol in the size range of 0.06 to  15 jim
12     in Vienna, Austria, using a nine-stage  cascade impactor and measured the humidity-dependent
13     water uptake when the individual impaction foils were exposed to high RH. They observed
14     particle growth with varying growth patterns. Calculated extinction coefficients and single
15     scattering albedo increased with humidity.
16           Hygroscopic properties, along with mixing characteristics,  of submicrometer particles
17     sampled in Los Angeles, CA, during the summer of 1987 SCAQS study and at the Grand
       June 2003
                             2-55
                                         DRAFT-DO NOT QUOTE OR CITE

-------
 1      Canyon, AZ, during the 1990 Navajo Generating Station Visibility Study were reported by
 2      Zhang et al. (1993). They used a tandem differential mobility analyzer (TDMA; McMurry and
 3      Stolzenburg, 1989) to measure the hygroscopic properties for particles in the 0.05- to 0.5-|im
 4      range. In their experimental technique, monodisperse particles of a known size are selected from
 5      the atmospheric aerosol with the first DMA. Then, the relative humidity of the monodisperse
 6      aerosol is adjusted, and the new particle size distribution is measured with the second DMA.
 7      At both sites, they observed that monodisperse particles could be classified according to "more"
 8      hygroscopic and "less" hygroscopic.  Aerosol behavior observed at the two sites differed
 9      markedly.  Within the experimental uncertainty (±2%) the "less" hygroscopic particles sampled
10      in Los Angeles did not grow when the RH was increased to 90%; whereas at the Grand Canyon,
11      the growth of the "less" hygroscopic particles varied from day to day, but ranged from near 0 to
12      40% when the RH was increased to 90%.  The growth of the "more" hygroscopic particles in
13      Los Angeles was dependent  on particle size (15% at 0.05 jim to 60% at 0.5 jim); whereas at the
14      Grand Canyon, the "more" hygroscopic particles grew by about 50% with the growth not
15      varying significantly with particle size. By comparison of the TDMA data to impactor data,
16      Zhang et al. (1993) surmised that the  more hygroscopic particles contained more sulfates and
17      nitrates while the less  hygroscopic particles contained more carbon and crustal components.
18           Although most of the work to date on the hygroscopic properties of atmospheric aerosols
19      has focused on the inorganic fraction, the determination of the contribution of particle-bound
20      water to atmospheric particulate mass is greatly complicated by the presence of organics.  The
21      effect of RH on adsorption of semivolatile organic compounds is discussed elsewhere in this
22      chapter.  Saxena et al. (1995) observed that particulate organic compounds also can affect the
23      hygroscopic behavior of atmospheric particles.  They idealized the organic component of aerosol
24      as containing a hydrophobic fraction  (high-molecular weight alkanes, alkanoic acids, alkenoic
25      acids, aldehydes,  and ketones)  and a hydrophilic fraction (e.g., lower molecular weight
26      carboxylic acids,  dicarboxylic acids, alcohols, aldehydes, etc.) that would be likely to absorb
27      water. They then analyzed data from a tandem differential mobility analyzer in conjunction with
28      particle composition observations from an urban site (Claremont, CA) and from a nonurban site
29      (Grand Canyon) to test the hypothesis that, by adding particulate organics to an inorganic
30      aerosol, the amount of water absorbed would be affected, and the effect could be positive or
31      negative, depending on the nature of the organics added. They further presumed that the

        June 2003                                 2-56        DRAFT-DO NOT QUOTE OR CITE

-------
 1      particulate organic matter in nonurban areas would be predominantly secondary and thus
 2      hydrophilic, compared to the urban aerosol that was presumed to be derived from primary
 3      emissions and thus hydrophobic in nature.  Their observations were consistent with their
 4      hypothesis, in that at the Grand Canyon, the presence of organics tended to increase the water
 5      uptake by aerosols; whereas at the Los Angeles site, the presence of organics tended to decrease
 6      water uptake.
 7           Peng and Chan (2001) also recently studied the hygroscopic properties of nine water
 8      soluble organic salts of atmospheric interest using an electrodynamic balance operated at 25°C.
 9      Salts studied included sodium formate, sodium acetate, sodium succinate, sodium pyruvate,
10      sodium methanesulfonate, sodium oxalate,  ammonium oxalate, sodium malonate, and sodium
11      maleate. They observed that hygroscopic organic salts have a growth factor of 1.76-2.18 from
12      RH= 10-90%, comparable to that of typical  hygroscopic inorganic salts such as NaCl and
13      (NH4)2SO4.
14           Nonequilibrium issues may be important for the TDMA, as well as for other methods of
15      measuring water content. Although approach to equilibrium when the RH is increased is
16      expected to be rapid for pure salts, it may be much slower for aerosols containing a complex mix
17      of components (Saxena et al., 1995). For example, if an aerosol contains an organic film or
18      coating, that film may impede the transport of water across the particle surface, thus increasing
19      the time required for equilibrium (Saxena et al.,  1995). Insufficient time to achieve equilibrium
20      in the TDMA  could result in underestimation of the water content.
21
22      2.2.3.2   Nitrate and Organic Species
23      Particulate Nitrates
24           It is now well  known that volatilization losses of particulate nitrates occur during sampling
25      on Teflon filters (e.g., Zhang and McMurry [1992]; see also Hering and Cass [1999] and Babich
26      et al., [2000]).  The effect on the accuracy of atmospheric particulate measurements from these
27      volatilization losses is more significant for  PM2 5 than for PM10.  The FRM for PM2 5 will likely
28      suffer loss of nitrates similar to that  experienced with other simple filter collection systems.
29      Sampling artifacts resulting from the loss of particulate nitrates represents a significant problem
30      in areas such as southern California  that experience high amounts of nitrates.  Hering and Cass
31      (1999) reported on errors in PM2 5 mass measurements due to  volatilization of particulate nitrate

        June 2003                                  2-57        DRAFT-DO NOT QUOTE OR CITE

-------
 1     (Figure 2-18). They examined data from two field measurement campaigns conducted in
 2     southern California: (1) the Southern California Air Quality Study (SCAQS) (Lawson, 1990)
 3     and (2) the 1986 CalTech study (Solomon et al., 1992). In both these studies, side-by-side
 4     sampling of PM25 was conducted. One sampler collected particles directly onto a Teflon filter.
 5     The second sampler consisted of a denuder to remove gaseous nitric acid followed by a nylon
 6     filter that absorbs the HNO3 which evaporates from ammonium nitrate. In both studies, the
 7     denuder consisted of MgO-coated glass tubes (Appel et al., 1981). Fine particulate nitrate
 8     collected on the Teflon filter was compared to fine particulate nitrate collected on the denuded
 9     nylon filter. In both studies,  the PM25 mass lost because of volatilization of ammonium nitrate
10     represented a significant fraction of the total PM2 5 mass.  The fraction of mass lost was higher
11     during summer than during fall (17% versus 9% during the SCAQS study and 21% versus 13%
12     during the CalTech study; Figure 2-18). In regard to percentage loss of nitrate, as opposed to
13     percentage loss of mass discussed above, Hering and Cass (1999) found that nitrate remaining on
14     the Teflon filter samples was on average 28% lower than that on the denuded nylon filters.
15
.i  60% -
ro
•5
|  40°,
y
S.
ra
%  20% -
            0%
                   °°
                     oo
                                SCAQS Data Set
                                o Summer Measurements
                                • Fall Measurements
                     50
                           100
                                 150
                                        200
                                              250
                    PM25 Gravimetric Mass (pg/m3)
                                                      |
                                                      'o
                                                      0)
                                                      CL
                                                      
-------
 1           Hering and Cass (1999) also analyzed these data by extending the evaporative model
 2      developed by Zhang and McMurry (1987).  The extended model utilized by Hering and Cass
 3      (1999) takes into account dissociation of collected particulate ammonium nitrate on Teflon
 4      filters into nitric acid and ammonia via three mechanisms: (1) scrubbing of nitric acid and
 5      ammonia in the sampler inlet (John et al., 1988 showed that clean PM10 inlet surfaces serve as an
 6      effective denuder for nitric acid), (2) heating of the filter substrate above ambient temperature by
 7      sampling, and (3) pressure drop across the Teflon filter. For the sampling systems modeled, the
 8      flow-induced pressure drop was measured to be less than 0.02 atm, and the corresponding
 9      change in vapor pressure was 2%, so losses driven by pressure drop were not considered to be
10      significant in this work. Losses from Teflon filters were found to be higher during the summer
11      compared to the winter, higher during the day compared to night, and reasonably consistent with
12      modeled predictions.
13           Finally, during the SCAQS  study, particulate  samples also were collected using a Berner
14      impactor and greased Tedlar substrates in size ranges from 0.05 to 10 jim in aerodynamic
15      diameter. The Berner impactor PM2 5 nitrate values were much closer to those from the denuded
16      nylon filter than those from the Teflon filter with the impactor nitrate being approximately
17      2% lower than the nylon filter nitrate for the fall measurements and approximately 7% lower
18      during the summer measurements. When the impactor collection was compared to the Teflon
19      filter collection for a nonvolatile species (sulfate), the results were in agreement.
20           It should be noted that filters or collection surfaces were removed immediately after
21      sampling and placed into vials containing a basic extraction solution during these
22      intercomparison studies. Therefore, losses that might occur during handling, storage, and
23      equilibration of filters or impaction surfaces were avoided. The loss of nitrate observed from
24      Teflon filters and impaction surfaces in this study, therefore, is a lower limit compared to losses
25      that might occur during the normal processes involved in equilibration and weighing of filters
26      and impaction surfaces. Brook and Dann (1999) observed much higher nitrate losses during a
27      study in which they measured particulate nitrate in Windsor and Hamilton, Ontario, Canada, by
28      three techniques:  (1) a single Teflon filter in a dichotomous sampler, (2) the Teflon filter in an
29      annular denuder system (ADS), and (3) total nitrate including both the Teflon filter and the
30      nylon back-up filter from the ADS. The Teflon filter from the dichotomous sampler averaged
31      only 13% of the total nitrate. The Teflon filter from the ADS averaged 46% of the total nitrate.

        June 2003                                2-59       DRAFT-DO NOT QUOTE OR CITE

-------
 1      The authors concluded that considerable nitrate was lost from the dichotomous sampler filters
 2      during handling, which included weighing and x-ray fluorescence (XRF) measurement in a
 3      vacuum.
 4          Kim et al. (1999) also examined nitrate sampling artifacts by comparing denuded and
 5      undenuded quartz and nylon filters, during the PM10 Technical Enhancement Program (PTEP) in
 6      the South Coast Air Basin of California. They observed negative nitrate artifacts (losses) for
 7      most measurements; however, for a significant number of measurements they observed positive
 8      nitrate artifacts.  Kim et al. (1999) pointed out that random measurement errors make it difficult
 9      to measure true amounts of nitrate loss.
10          Diffusion denuder samplers, developed primarily to measure particle strong acidity
11      (Koutrakis et al., 1988a,b, 1992), also can be used to study nitrate volatilization.  Such
12      techniques were used to measure loss of particulate nitrate from Teflon filters in seven U.S.
13      cities (Babich et al., 2000). Measurements were made with two versions of the Harvard-EPA
14      Annular Denuder System (HEADS).  Nitric acid vapor was removed by a Na2CO3-coated
15      denuder. Parti culate nitrate was the sum of nonvolatile nitrate collected on a Teflon filter and
16      volatized nitrate collected on a Na2CO3-coated denuder downstream of the Teflon filter (full
17      HEADS) or on a Nylon filter downstream of the Teflon filter (Nylon HEADS). It was found that
18      full HEADS (using a Na2CO3 denuder) consistently underestimated the total parti culate nitrate
19      by approximately 20% compared to the nylon HEADS.  Nonvolatilized nitrate was also
20      measured on a Teflon filter from collocated Harvard Impactors (HI). The PM2 5 HI sampler, like
21      the PM2 5 FRM, use impactors with a 50% cut at 2.5 jim.  The HI uses a 37 mm filter and a flow
22      rate of 10 L/min (0.465 L/min/cm2) which the FRM uses a 47 mm filter and a flow rate of 16.7
23      L/min (0.481 L/min/cm2). Therefore, the flow rate and pressure drop across the filters are
24      similar and the loss of nitrate should be similar in both types of samples.  Babich et al. (2000)
25      found significant nitrate losses in Riverside, CA; Philadelphia, PA; and Boston, MA but not in
26      Bakersfield, CA; Chicago, IL; Dallas, TX; or Phoenix, AZ where measurements were made only
27      during winter.
28          Eatough et al.  (1999a) developed a high-volume diffusion denuder system in which
29      diffusion denuder and particle concentrator techniques were combined (see Section 2.2.3.2). The
30      particle concentrator reduces the flow through the denuder so that the denuder can be operated
31      for weeks without a loss of collection efficiency, making the sampler suitable for routine field

        June 2003                                 2-60         DRAFT-DO NOT QUOTE OR CITE

-------
 1      sampling. The system was evaluated for the collection of fine particulate sulfate and nitrate in
 2      Riverside, CA (Eatough et al., 1999b).  Concentrations of PM25 nitrate obtained from the PC-
 3      BOSS agreed with those obtained using the Harvard-EPA Annular Denuder Sampler, HEADS
 4      (Koutrakis et al., 1988b).
 5
 6      Paniculate Organic Compounds
 1           Many semivolatile organic compounds (SVOC) are of interest because of their possible
 8      health effects.  SVOC include products of incomplete combustion such as poly cyclic aromatic
 9      hydrocarbons (PAHs) and polycyclic organic matter, which has been identified as a hazardous
10      air pollutant.  PAHs also have been suggested as alternative parti culate tracers for automobile
11      emissions because the phase-out of organo-lead additives to gasoline means that lead is no
12      longer a good tracer for automobiles (Venkataraman et al., 1994).  PAHs also are emitted during
13      biomass burning, including burning of cereal crop residues and wood fuels (Jenkins et al., 1996;
14      Roberts and Corkill, 1998).
15           A number of studies have used absorbing material following quartz filters to determine
16      phase equilibria of specific organic compounds (Liang et al., 1997; Gundel et al., 1995; Kamens
17      et al., 1995). Much work has also gone into the development of a theory to help understand the
18      phase relationships (Yamasaki et al., 1982; Rounds and Pankow, 1990; Pankow, 1987, 1994a,b;
19      Pankow et al., 1993; Rounds et al., 1993;  Odum et al., 1994). The theory describing phase
20      equilibria of semivolative organic compounds (SVOC) and the gas/particle partitioning of SVOC
21      on inorganic, organic, and ambient smog aerosols continues to be developed (Liang et al., 1997;
22      Jang et al., 1997, 1999; Strommen and Kamens, 1997; Jang and Kamens, 1998,  1999, 2001;
23      Leach et al., 1999; Kamens et al., 1999, 2001).
24           The mass of organic and elemental carbon is usually determined by analysis of PM
25      collected on a  quartz filter. However, quartz fiber filters have a large specific surface area on
26      which adsorption  of gases can occur. Possible artifacts associated with adsorption of organic
27      gases onto quartz  filters have been examined in experiments in which two quartz fiber filters
28      were  deployed in  series. The second quartz filter may indicate gaseous volatile organic
29      compounds (VOC) adsorbed on both filters (positive artifact), SVOC evaporated from particles
30      on the first filter and subsequently adsorbed on the second filter (negative artifact), or a
31      combination of both effects. Unless the individual compounds are identified, the investigator

        June 2003                                 2-61        DRAFT-DO NOT QUOTE OR CITE

-------
 1      does not know what to do with the loading value on the second filter (i.e., to add or subtract from
 2      the first filter loading value).  Moreover, even if the individual compounds were identified on the
 3      back-up filter, the decision concerning adding or subtracting the back-up filter loading would not
 4      be straightforward.
 5           The positive quartz filter artifact has been discussed by Gundel et al. (1995) and Turpin
 6      et al. (2000). It is also possible that some SVOC may desorb from the filter resulting in a
 7      negative artifact (Eatough et al., 1993; Tang et al., 1994; Eatough et al., 1995; Gundel et al.,
 8      1995; Cui et al., 1998;  Pang et al., 2001; Finn et al., 2001).  Semivolatile organic compounds can
 9      similarly be lost from Teflon filters because of volatilization, causing the PM2 5 mass to be
10      significantly underestimated (negative artifact). Like particulate nitrates, the FRM for PM25 will
11      suffer loss of SVOC, similar to the losses experienced with other simple filter collection
12      systems. Most studies that have focused on the positive and negative sampling artifacts
13      associated with SVOC compounds have utilized either diffusion denuder technology (Eatough
14      et al., 1995; Mader et al., 2001) or placed an adsorbent media, such as a back-up quartz filter or a
15      polyurethane foam adsorbent behind the main filter (Wallace and Kites, 1995). Further
16      information on denuder techniques are given in Appendix 2A.
17           Using their multichannel diffusion denuder sampling system (BOSS), Eatough et al. (1995)
18      reported that, for samples collected at the South Coast Air Quality Management District
19      sampling site at Azusa, CA, changes in the phase distribution of SVOC could result  in a loss on
20      average of 35% of the parti culate organic material.  Cui et al. (1998) found that losses of SVOC
21      from particles in the Los  Angeles Basin during the summer were greater during the night
22      (average = 62%) than during the day (average = 42%).
23           The percent SVOC lost from the front filter in a filter-denuder system may be greater than
24      that lost in a filter-only system such as the FRM.  In a filter-denuder system, the gas-phase
25      component of the SVOC is removed.  The absence of the gas-phase causes the gas-particle
26      equilibrium to shift so the SVOC collected on the filter may evaporate more rapidly  in a filter-
27      denuder system than in a filter-only collection system. To determine the fraction of SVOC lost
28      from a Teflon filter in a filter-only system, it is necessary to compare the amount measured by a
29      nondenuder system with that measured by a denuder system. At present, little information is
30      available on the volatilization losses of SVOC.  However, in one study (Pang et al., 2001), the
31      total mass on denuded  and undenuded filters were compared and found to be identical within

        June 2003                                 2-62        DRAFT-DO NOT QUOTE OR CITE

-------
 1      error limits (R2 = 0.816, slope = 0.961 ± 0.027 for total mass compared to R2 = 0.940, slope =
 2      0.986 ± 0.020 for sulfate).  Pang et al. interpreted this result as suggesting that the major cause of
 3      loss of SVOC is the pressure drop across the filter.
 4           Positive artifacts may occur during sample collection because of the adsorption of gases
 5      onto the filter materials (e.g., Gundel et al., 1995).  Using a quartz filter behind a Teflon filter,
 6      Kim et al. (2001a) estimated that on an annual average basis 30% of the PM25 organic carbon
 7      concentration resulted from positive artifacts.  There is a larger positive artifact because of
 8      greater adsorption of organic vapor onto quartz fiber filters than onto Teflon filters (Turpin et al.,
 9      1994; Chow et al., 1994a,b, 1996; Eatough et al., 1996; Finn et al., 2001; Kirchstetter et al.,
10      2001).
11           Kirchstetter et al. (2001) report that adsorptive properties of quartz fiber filters vary with
12      lot number; therefore, front and back-up filters should be taken from the same lot.  The literature
13      suggests that a Teflon filter followed by a  quartz back-up filter appears to provide a better
14      estimate of the adsorption of gases on a quartz fiber front filter than does a quartz filter followed
15      by a quartz backup and that the difference between these two adsorption estimates  can be
16      substantial for short durations (Novakov et al., 1997; Kirchstetter et al., 2001; Turpin et al.,
17      2000).  The typically lower organic carbon loadings on concurrently collected quartz followed
18      by quartz filters relative to Teflon followed by quartz filters are believed to occur because
19      adsorption on the quartz front filter acts to reduce the gas-phase concentration downstream until
20      adsorption equilibrium has been achieved in the vicinity of the front quartz filter surface.
21      Because Teflon filters have little affinity for organic vapors, this equilibrium occurs almost
22      instantaneously for Teflon filters, and the Teflon-quartz back-up filter is exposed to the ambient
23      concentration of organic vapors from the beginning of the sampling period. It might be expected
24      that the quantity of organic vapor adsorbed on quartz filters would depend on the organic
25      composition and would vary by season and location. However, it is also possible that the quartz
26      possesses a limited number of adsorption sites which are rapidly occupied so that the quantity of
27      OC on  the back up filter would be relatively constant and depend on the pretreatment of the
28      quartz.
29
30
31

        June 2003                                 2-63        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Combined Measurement of Semivolatile Nitrate and OC and Nonvolatile Species (OC, EC,
 2      Nitrate, and Sulfate)
 3           Fine particles in urban atmospheres contain substantial quantities of semivolatile material
 4      (e.g., NH4NO3 and semivolatile organic compounds [SVOC]) that are lost from particles during
 5      collection on a filter. Several diffusion denuder samplers have been developed for the
 6      determination of both NO3" and organic semivolatile fine particulate components as well as
 7      nonvolatile nitrate and organic compounds and nonvolatile sulfate (Pang et al., 2001; Eatough
 8      et al., 1993).  The combination of technology used in the BOSS diffusion denuder sampler and
 9      the Harvard particle concentrator has resulted in the Particle Concentrator-Brigham Young
10      University Organic Sampling System (PC-BOSS) for the 24-hr integrated collection of PM25,
11      including NH4NO3 and semivolatile organic material. Modifications of the BOSS sampler allow
12      for the determination of these same species on a time scale from a few hours to weekly (Ding et
13      al., 2002; Eatough et al., 1999a,b; 2001). Episodic studies have been conducted in Riverside,
14      CA and Bakersfield, CA (Obeidi et al., 2002) and Provo, UT (Obeidi and Eatough, 2002). The
15      average concentration of semivolatile and nonvolatile components in the three cities, during the
16      study periods, are shown in Figure 2-19.
17
18      2.2.3.3   Continuous Measurement of Semivolatile and Nonvolatile Mass
19           Techniques for the continuous measurement of PM mass is needed both to provide
20      real-time information on pollution levels (Long et al., 2002) and to reduce the costs involved in
21      visiting sites to change filters and in the equilibration and weighing of filters. Two techniques
22      are currently in use for the continuous measurement of PM mass.  The TEOM is normally
23      operated at 50° C in order to remove particle-bound water.  However, at 50° C most semivolatile
24      material is also evaporated. Therefore, the TEOM, operated at 50° C, may be considered to
25      measure the mass of nonvolatile PM. The beta gauge mass monitor  changes the filter more
26      frequently than the TEOM and is less sensitive to changes in mass caused by changes in relative
27      humidity. It does not control the temperature at the filter.  However, most beta gauge monitors
28      heat the inlet.  This heating causes evaporation of a substantial fraction of the particle-bound
29      water and an unknown fraction of the semivolatile PM.  Thus, the beta gauge may be considered
30      to measure the nonvolatile PM plus a small fraction of the particle-bound water and an unknown
31      fraction of the semivolatile PM.  Three new techniques have been developed to address the issue
32      of lost semivolatile PM mass.

        June 2003                                2-64         DRAFT-DO NOT QUOTE OR CITE

-------
                                              Riverside, CA; September 1997
                                     Semivolatile^
                                    Components^
                                                                      Non-volatile
                                                                      K Components
                                            Ammonium
                                              Nitrate   Elemental  Ammonium
                                               2-6     Carbon   Sulfate
                                                       3.1       1.2
                                                Bakersfield, CA; March 1998
                                        Semivolatile ^
                                      Components ,X
                                          f=^~  Lost   ~^EE
                                            : Semivolatile OC =
                                                8.5
                                    Lost
                                 Ammonium/
                                     Nitrate
                                       0.8
                                                                   *° Non-volatile
                                                                     X, Components
                  Ammonium
             „_„,,  >: Sulfate

                        17
               gf Elemental
                   Carbon
                     1.T
                                               Provo, UT; September 1998
                                      Semivolatile
                                    Components
                                     Lost
                                 Ammonium
                                    Nitrate
                                       0.2 t
                                                                     Non volatile
                                                                       Components
Figure 2-19.   Average concentration of nonvolatile and Semivolatile PM components in
                 three cities (ug/m3).

Source:  Obeidi and Eatough (2002); Obeidi et al. (2002).
June 2003
2-65
DRAFT-DO NOT QUOTE OR CITE

-------
 1     Real-Time Total Ambient Mass Sampler (RAMS)
 2           The RAMS, a monitor based on diffusion denuder and TEOM monitor technology, has
 3     been developed, validated, and field tested for the real-time determination of total fine PM mass,
 4     including semivolatile PM (Eatough et al., 1999a; Obeidi and Eatough, 2002; Obedi et al., 2002;
 5     Pang et al., 2001).  The RAMS measures the total mass of collected particles including
 6     semivolatile species with a TEOM monitor using a "sandwich filter."  The sandwich contains a
 7     Teflon coated particle collection filter followed by a charcoal-impregnated filter (GIF) to collect
 8     any semivolatile species lost from the particles during sampling. Because the instrument
 9     measures total mass collected by the sandwich filter, all gas phase compounds that can be
10     adsorbed by a GIF must be removed from the sampling stream prior to the TEOM monitor.
11     Laboratory and field validation data indicate that the precision of fine PM mass determination is
12     better than 10%. The RAMS uses a Nafion dryer to remove particle-bound water from the
13     suspended particles and a particle concentrator to reduce the amount of gas phase organics that
14     must be removed by the denuder. An example of data from the  RAMS, the TEOM, and the
15     PC-BOSS is shown in Figure 2-20. This figure also shows the PM2 5 mass from the TEOM as
16     being negative for the hours of 16 to 19.  This likely results from the loss of volatile materials
17     from the heated filter.
18
19     Differential TEOM
20           Patashnick et al. (2001) developed a differential TEOM system that is  based on a pair of
21     TEOM sensors, each of which is preceded by its own electrostatic precipitator (ESP) and
22     downstream from a common size selective inlet.  By alternately switching the ESPs on and off
23     and out of phase with each other, the two sensors measure "effective mass" that includes both
24     the nonvolatile component and the volatile component sampled  by the TEOM, less the volatile
25     component that vaporized during the sampling interval. On the  sensor side with the ESP turned
26     on, there is no particle collection on that filter so  that only volatilization of previously collected
27     particles continues.  This allows a correction for the effective mass as measured by the first
28     sensor by subtracting out the volatilization artifact and leaving the nonvolatile and volatile
29     components of the particulate matter.  This system has yet to be well characterized for other
30     biases or interferences such as reactions on the filters, particle collection efficiency of the ESPs,
31     and particle and semivolatile material losses.

       June 2003                                2-66       DRAFT-DO NOT QUOTE OR CITE

-------
                            PC-BOSS (Nonvolatile Material)
                     -20
                             TEOM
                              at35C
—B—  RAMS
      at35C
| PC-BOSS (Lost From Particles)
      	FRM PM25
           24 h average
                                           Riverside, CA
                        13    15    17    19    21    23 0  1
                                              Time of Day
      Figure 2-20.  Comparison of mass measurements with collocated RAMS (real-time data),
                   PC-BOSS (1-h data), FRM PM25 sampler (average of 24-h data), and a
                   conventional TEOM monitor (real-time data).  The semivolatile fine
                   particulate matter is sampled with the RAMS and PC-BOSS, but not with
                   the TEOM monitor or the FRM PM2 5 single filter sampler.  The PC-BOSS
                   provides information on both the nonvolatile component (NV) and the
                   semivolatile organic component (SVOC).
      Source: Eatoughetal. (1999a).
1     Continuous Ambient Mass Monitor (CAMM)
2          Koutrakis and colleagues (Koutrakis et al., 1996; Wang, 1997) have developed CAMM, a
3     technique for the continuous measurement of ambient particulate matter mass concentration
4     based on the measurement of pressure drop increase with particle loading across a membrane
5     filter. Recently, Sioutas et al.  (1999) examined the increase in pressure drop with increasing
6     particle loading on Nuclepore filters. They tested filters with two pore diameters (2 and 5 jim)
7     and filter face velocities ranging from 4 to 52 cm s"1 and examined the effects of relative
8     humidity in the range of 10 to 50%. They found that, for hygroscopic ammonium sulfate
9     particles, the change in pressure drop per unit time and concentration was a strong function of
      June 2003
    2-67
  DRAFT-DO NOT QUOTE OR CITE

-------
 1      relative humidity, decreasing with increasing relative humidity. These results suggest that
 2      particulate concentration measurements, similar to the method of Koutrakis et al. (1996) that
 3      uses the pressure drop method, may be subject to additional uncertainties if used in an
 4      environment where the ambient relative humidity  is quite variable and the relative humidity
 5      where the particles are measured cannot be controlled accurately. The current version of the
 6      CAMM (Wang, 1997) uses a particle concentrator, a Nafion dryer, and frequently  moves the
 7      filter tape to avoid artifacts due to evaporation of semivolatile components from the active
 8      portion of the filter tape which would occur if the  atmospheric concentration of the semivolatile
 9      components decreased.
10           The CAMM was recently operated alongside a gravimetric PM method (the Harvard
11      Impactor, or HI) in seven U.S. cities selected for their distinctly different ambient particulate
12      compositions and densities.  The correlation between the two methods was high, with an overall
13      r2 of 0.90 and average CAMM/HI ratio of 1.07 (Babich et al., 2000).
14
15      2.2.4   U. S. Environmental Protection Agency Monitoring Methods
16      2.2.4.1  The Federal Reference Methods for Measurement of Equilibrated Mass for PM10,
17              PM25, andPM1025
18           In 1997, EPA promulgated new standards for PM25 to address fine-fraction thoracic
19      particles and retained with minor revisions the 1987 PM10 standards to continue to address
20      coarse-fraction thoracic particles (Federal  Register, 1997). In partial response to numerous
21      challenges to these standards, the U.S. Court of Appeals for the District of Columbia Circuit in
22      American Trucking Association v. EPA, 175 F. 3d 1027 (U.S. Court of Appeals, D.C. Cir. 1999)
23      found "ample support" for regulating coarse-fraction  particles but revoked the revised PM10
24      standards (leaving in effect the 1987 PM10 standards) on the basis of PM10 being a  "poorly
25      matched indicator for coarse particulate pollution" because PM10 includes fine particles.
26      Consistent with this specific aspect of the Court's  ruling, which EPA did not appeal, EPA is now
27      considering use of PM10_25 as the indicator for coarse-fraction thoracic particles, in conjunction
28      with PM2 5 standards that address fine-fraction thoracic particles.  Thus, EPA is now developing
29      a Federal Reference Method for the measurement  of PM10_25.
30
31
        June 2003                                 2-68        DRAFT-DO NOT QUOTE OR CITE

-------
 1      2.2.4.1.1 PM10
 2           The FRM specified for measuring PM10 (Code of Federal Regulations, 2001a,b) has been
 3      discussed in previous PM AQCD's and will only be mentioned briefly.  The PM10 FRM defines
 4      performance specifications for samplers in which particles are inertially separated with a
 5      penetration efficiency of 50% at an aerodynamic diameter (Da) of 10 ± 0.5 jim.  The collection
 6      efficiency increases to =100% for smaller particles and drops to ~ 0% for larger particles.
 7      Particles are collected on filters and mass concentrations are determined gravimetrically.
 8      Instrument manufacturers are required to  demonstrate through field tests a measurement
 9      precision for 24-h samples of ± 5 |ig/m3 for PM10 concentrations below 80 |ig/m3 and 7% above
10      this value. A number of samplers have been designated as PM10 reference samplers. The TEOM
11      and several beta gauge samplers with 1-h time resolution have been designated as automated
12      equivalent methods (U.S. Environmental  Protection Agency, 2001).
13
14      2.2.4.1.2 PM25
15           In contrast to the performance-based FRM standard for PM10, the FRM for PM2 5 (Code of
16      Federal Regulations, 200 la) specifies certain details of the sampler design, as well as of sample
17      handling and analysis, whereas other aspects have performance specifications (Noble et al.,
18      2001). The PM25 FRM sampler consists of a PM10 inlet/impactor, a PM25 impactor with an oil-
19      soaked impaction substrate to remove particles larger than 2.5 |im Da, and a 47-mm PTFE filter
20      with a particle collection efficiency greater than 99.7%. The sample duration is 24 h, during
21      which time the sample temperature is not to exceed ambient temperatures by more than 5 °C.
22      A schematic diagram of the PM2 5 FRM sample collection system is shown in Figure 2-21. After
23      collection, samples are equilibrated for 24 h at temperatures in the range of 20 to 23 °C (± 2  °C)
24      and at relative humidities in the range of 30 to 40% (± 5%).  The equilibration tends to reduce
25      particle-bound water and stabilizes the  filter plus sample weight.  Filters are weighed before and
26      after sampling under the same temperature and relative humidity  conditions. For sampling
27      conducted at ambient relative humidity less than 30%, mass measurements at relative humidities
28      down to 20% are permissible (Code of Federal Regulations, 200 la).
29           The PM10 inlet specified for the PM2 5 FRM is modified from a previous low flow-rate
30      PM10 inlet that was acceptable in both EPA-designated reference  and equivalent PM10 methods.
31      The modification corrects a flaw that was reported for the previous sampler, in that under some

        June 2003                                2-69        DRAFT-DO NOT QUOTE OR CITE

-------
                                         Ambient
                                         aerosol
                                         sampling
                                          inlet
                                           fractionator
                                             Downtube
                                              WINS
                                              PM25
                                            fractionator
       Figure 2-21.  Schematic diagram of the sample collection portion of the PM25 FRM
                     sampler.
       Source: Noble et al. (2001).
 1     meteorological conditions, the inlet may allow precipitation to penetrate the inlet. The
 2     modification includes a larger drain hole, a one-piece top plate, and louvers. Tolocka et al.
 3     (200 la) evaluated the performance of this modified inlet in a series of wind tunnel experiments.
 4     The modified inlet was found to provide a size cut comparable to the original inlet, for both
 5     PM25 and PM10 sampling.  Since the modification did not change the characteristics of the size
 6     cut, the modified inlet may be substituted for the original inlet as part of a reference or
 7     equivalent method for PM10 and PM2 5 (Tolocka et al., 2001a).
 8
 9           WINS Impactor. Design and calibration of the EPA PM2 5 Well Impactor Ninety-Six
10     (WINS) is given by Peters et al. (2001a). The WINS impactor was designed to be deployed
11     downstream of the Graseby-Andersen 246B PM10 inlet as part of a sampler operating at a flow
12     rate of 16.7 L/m.  The WINS is pictured in Figure 2-22.  The PM2 5 inlet consists of a single jet,
13     directed toward a round hole, with a jet exit impaction surface comprised of a 37 mm diameter
       June 2003
2-70
DRAFT-DO NOT QUOTE OR CITE

-------
                                                 PM-10 Aerosol
                                                  frotr* I
                                                           Nozzle
                                                          .— Collection cup
                                                            with antispitl ring
                                                          - Impaction surface:
                                                          filter immersed in 1 mL
                                                            Dow Corning 704
                                                           diffusion pump oil
,


PM-2
samp
                                                     filter
              Figure 2-22.  Schematic view of the final design of the WINS.
              Source: Peters etal. (200la).
 1      glass fiber filter immersed in 1 ml of low volatility diffusion pump oil (i.e., the well). Particles
 2      not having enough inertia to be removed by the impactor are captured downstream on the sample
 3      collection filter. This design was selected to minimize impactor overloading that would
 4      otherwise result in particle bounce.  The oil wicks through the particulate deposit on the
 5      impactor to provide a continuously wetted surface for impaction.  The penetration curve
 6      indicated a 50% cutpoint of 2.48 jim Da with a geometric standard deviation of 1.18
 1      (Figure 2-23).
 8           The WINS separator was evaluated for its loading characteristics (Vanderpool  et al., 2001)
 9      by monitoring the performance after repeated operation in an artificially generated, high
10      concentration, coarse-mode aerosol composed of Arizona Test Dust, as well as in the field in
11      Rubidoux, Phoenix, Philadelphia, Research Triangle Park, and Atlanta.  In the wind  tunnel
12      experiments, the WINS performance was found to be a monotonic function of loading. A minus
13      5% bias in the PM2 5 measurement resulted from a coarse particulate loading of approximately
14      16 mg. This negative bias was due to a slight reduction in the separator cutpoint. It  was also
        June 2003
2-71
DRAFT-DO NOT QUOTE OR CITE

-------
                                    100
                                     80
                                    60
                                S
                                CD
                                    40
                                                        WINS Final Version
                                                       	 WINS Best Fit Line
                                                        ^   Aerosizer Detector
                                                        X   Fluorometer Detector
                                    20
                                        1.0        2.0        3.0        4.0
                                             Aerodynamic Diameter (|jm)

              Figure 2-23. Evaluation of the final version of the WINS.
              Source: Peters et al. (200la).
 1     found that the predictable results from the controlled laboratory experiments could not be
 2     extrapolated to field settings and that the WINS performance was more sensitive to the impactor
 3     loading in the field than it was in experiments with the single component aerosol. Significant
 4     particle bounce was not observed in either the laboratory or the field experiments.  Vanderpool
 5     et al. (2001) concluded that their study supports the recommendation that the FRM WINS wells
 6     should be replaced after every 5 days of 24-h operation (U.S. Environmental Protection Agency,
 7     1998).
 8           A detailed sensitivity study of the WINS impactor was conducted (Vanderpool et al., 2001)
 9     in which the effects on the impactor performance of a number of parameters were examined.
10     The results of this study are summarized in Table 2-3.
11           The regulations also allow for Class I, II, and III equivalent methods for PM2 5 (Code of
12     Federal Regulations, 200 Ic).  Class I equivalent methods use samplers with relatively small
13     deviations from the sampler described in the FRM.  Class II equivalent methods include "all
14     other PM25 methods that are based upon 24-h integrated filter samplers that are subjected to
       June 2003
2-72
DRAFT-DO NOT QUOTE OR CITE

-------
             TABLE 2-3. SUMMARY OF SENSITIVITY STUDIES OF WINS IMP ACTOR
                                          PERFORMANCE
Parameter
Manufacturing tolerances
on WINS components
Flow control biases
T and P measurement
Diffusion oil volume
Impactor loading
Ambient P variations
Air Properties
Impactor oil crystallization
Impactor oil viscosity
Amount of variance
Specified tolerances
4%
Allowable ambient
0.75 ml to 3 ml
After 5 24 h events

OC

-20 C
-35 C
Cutpoint variation
0.05 micrometers
0.05 micrometers
+ 0.02 micrometers
No effect
-0.07 micrometers
Negligible
2.40 micrometers
No effect
No effect
Need to change WINS
5 days
PM2 s mass
concentration bias
<1%
Cutpoint shift partially offset
volume bias
+ 0.4%

<1.5%
Negligible
NA
No effect
No effect
more frequently than every
        Source: Vanderpooletal. (2001).
 1     subsequent moisture equilibration and gravimetric mass analysis." Class III equivalent methods
 2     include non-filter-based methods such as beta attenuation, harmonic oscillating elements, or
 3     nephelometry (McMurry, 2000). As of July 2001, 11 PM2 5 samplers (listed in Table 2-4) had
 4     been tested and led to the conclusion that the PM10 sampling systems can be designed such that
 5     concentration measurements are precise to ±  10%. For PM25, cutpoint tolerances are not
 6     expected to affect the mass concentration as much as for PM10, because the 2.5 jim Da cutpoint
 7     generally occurs near a minimum in the mass distribution (e.g., Figure 2-5).
 8          The PM2 5 mass concentration will be affected, on the other hand, by other sampling issues
 9     mentioned but not discussed extensively in the previous 1996 PM AQCD (U.S. Environmental
10     Protection Agency, 1996a).  These issues have been discussed earlier in this chapter and include
11     gas/particle, particle/particle, and particle/substrate interactions for sulfates  and nitrates (e.g.,
12     Appel et al., 1984), volatilization losses of nitrates (Zhang and McMurry, 1992), semivolatile
13     organic compound artifacts (e.g., Eatough et  al., 1993), and relative humidity effects (e.g.,
14     Keeleretal., 1988).
       June 2003
2-73
DRAFT-DO NOT QUOTE OR CITE

-------
             TABLE 2-4. PM25 SAMPLERS CURRENTLY DESIGNATED AS FRMs FOR
                                    ?, MASS CONCENTRATIONS
Sampler
RAAS2.5-100
RAAS2.5-300
RAAS2.5-200
Partisol 2000
Partisol-Plus 2025
Partisol 2000 audit
PQ200
PQ 200A
605 CAPS
MASS 100
MASS 300
Manufacturer
Andersen Instruments
Andersen Instruments
Andersen Instruments
Rupprecht & Patashnick
Rupprecht & Patashnick
Rupprecht & Patashnick
BGI, Inc.
BGI, Inc.
ThermoEnvironmental Instruments
URC
URC
Description
FRM single
FRM sequential
FRM audit
FRM single
FRM sequential
FRM audit
FRM single
FRM audit
FRM single
FRM single
FRM sequential
Federal Register Reference
Vol. 63, p. 31991,6/11/98
Vol. 63, p. 31991,6/11/98
Vol. 64, p. 12167,3/11/99
Vol. 63, p. 18911,4/16/98
Vol. 63, p. 18911,4/16/98
Vol. 64, p. 19153,4/19/99
Vol. 63, p. 18911,4/16/98
Vol. 63, p. 18911,4/16/98
Vol. 63, p. 58036, 10/29/98
Vol. 65, p. 26603, 05/08/00
Vol. 65, p. 26603, 05/08/00
        Source: Peters et al. (200Ib); U.S. Environmental Protection Agency (2001).
 1          Several studies now have been reported in which the FRM was collocated with other PM2 5
 2     samplers in intercomparison studies.  During the Aerosol Research and Inhalation Epidemiology
 3     Study (ARIES), several PM2 5 samplers were collocated at a mixed industrial-residential site near
 4     Atlanta, GA (Van Loy et al., 2000). These samplers included a standard PM25 FRM, a TEOM
 5     with Nafion drier, a particulate composition monitor (PCM; Atmospheric Research and
 6     Analysis, Gary, NC), a medium-volume (113 L/min flow rate) fine particle (PM2 5) and
 7     semivolatile organic compound (i.e., a filter followed by a solid adsorbent) sampler, operated by
 8     the Desert Research Institute, a HEADS sampler, and a dichotomous sampler for  coarse PM.
 9     The PCM sampler has three channels, all of which have PM10 cyclone inlets. The first two
10     channels both have two denuders preceding a 2.5-|im WINS impactor and filter packs. The first
11     denuder is coated with sodium carbonate to remove acid gases, and the second is  coated with
12     citric acid to remove ammonia. The third channel has a carbon coated parallel-plate denuder
13     preceding the WINS impactor. Measurements of 24-h mass from the FRM, PCM, and TEOM
14     samplers, as well as reconstructed PM25 mass (RPM), were compared for a  12-mo period.  The
15     slopes for the TEOM-FRM, PCM-FRM, and RPM-FRM correlations were 1.01, 0.94, and 0.91,
       June 2003
2-74
DRAFT-DO NOT QUOTE OR CITE

-------
 1      respectively; whereas the y-intercepts for each were 0.68, 0.04, and 0.98.  Particulate sulfate
 2      measurements on the FRM Teflon filter, the PCM Teflon filter, and PCM Nylon filter were
 3      nearly identical. Nitrate results from the three filters were much less consistent, with the FRM
 4      collecting substantially less nitrate than that collected on either the denuded nylon filter or a
 5      denuder followed by a Teflon-nylon filter sandwich. Particulate ammonia measurements were
 6      also compared and showed more scatter than the sulfate measurements but less than the nitrate
 7      measurements.
 8           An intercomparison of both PM10 and PM25 mass measurements was conducted during the
 9      1998 Baltimore PM Study (Williams et al., 2000). PM monitors were collocated at a residential
10      indoor, residential outdoor, and ambient monitoring site within Baltimore County, MD. PM
11      samplers included TEOMs, PM2 5 FRMs, cyclone-based inlets manufactured by University
12      Research Glassware (URG), and Versatile Air Pollution Samplers (VAPS).  The VAPS sampler
13      is a dichotomous sampler operating at 33 L/min (one coarse particle channel at 3 L/min, and two
14      fine particle channels at 15 L/min, each). In the configuration employed during this study, one
15      fine particle channel was operated with a Teflon filter backed by a nylon filter and preceded by a
16      sodium carbonate coated annular denuder; the second fine particle channel had a quartz filter
17      preceded by a citric acid-coated annular denuder; and the coarse particle channel had a
18      polycarbonate filter followed by a Zefluor filter for flow distribution. Differences in PM25 mass
19      concentrations between the samplers, although not large, were attributed to potential particle
20      nitrate losses, denuder losses, and losses of SVOC for some samplers.  Differences between
21      coarse paniculate mass concentrations,  on the other hand, varied widely between the
22      instruments.
23           In another intercomparison study, Tolocka et al. (200Ib) examined the magnitude of
24      potential sampling artifacts associated with the use of the FRM by collocating FRMs alongside
25      other chemical speciation samplers at four U.S. cities. The locations included a high nitrate and
26      carbon, low sulfate site (Rubidoux, CA); high crustal, moderate carbon and nitrate site
27      (Phoenix); high sulfate, moderate carbon, and low nitrate (Philadelphia); and low PM2 5 mass
28      (Research Triangle Park, NC). The use of Teflon and heat-treated quartz filters was also
29      examined in this study.  The Teflon filters collected less nitrate than the heat-treated quartz
30      filters.  Filters in samplers using denuders to remove organic gases collected less organic PM
31      than filters in samplers without denuders.

        June 2003                                  2-75         DRAFT-DO NOT QUOTE OR CITE

-------
 1          Peters et al. (2001b) compiled the results of several field studies in which a number of
 2     FRM and other PM2 5 samplers were intercompared. In addition to the FRM samplers listed in
 3     Table 2-3, other PM2 5 samplers included the Sierra Instruments dichotomous sampler, the
 4     Harvard impactor, the IMPROVE sampler, and the Air Metrics saturation monitor. Results were
 5     compiled from PM2 5 field studies conducted in Birmingham, Denver, Bakersfield, Phoenix,
 6     Research Triangle Park, Atlanta, and Rubidoux.  Limited studies on precision for the non-FRM
 7     samplers suggest that the Harvard Impactor and dichotomous samplers had the lowest coefficient
 8     of variations (CV), with both under 10%.  The CV for this study was calculated by dividing the
 9     sample standard deviation by the average concentration. The IMPROVE samplers had CVs
10     between 10 and 12%, and the Air Metrics samplers had the highest observed CVs, over 15%.
11     In intercomparisons with FRM samplers, the non-FRM samplers showed strong linear
12     relationships in comparison to the FRM sampler; however, none of the comparisons passed the
13     current EPA Subpart C equivalent method criteria, which EPA is in the process of revising.
14          Detailed information on precision of PM samplers used in monitoring networks may be
15     found in EPA's Technology Transfer Network website (U.S. Environmental Protection Agency,
16     2002a).
17
18     2.2.4.1.3   PM1025
19          Measurement techniques for PM10_25 are somewhat more complex than those for PM25 or
20     PM10 because, for PM10_2 5, it is necessary to isolate a size fraction between an upper 50% cut
21     point of 10 |im Da and a lower 50% cut point of 2.5 jim Da.  EPA is currently developing an
22     FRM for PM10_2 5.  Several candidate techniques are discussed below.
23
24          The Difference Method. One approach to measurement of PM10.2 5 is to make separate
25     measurements of PM10 and PM25 and take the difference of the resulting equilibrated masses.
26     One problem is that, if either the PM2 5 or the PM10 sampler fails, no PM10_2 5 measurement can be
27     obtained. In addition, errors in cut-point, flow rate, and filter weights (both before use and after
28     collection and equilibration of particles) and uncertainties due to loss of semivolatile
29     components of PM may occur for each size cut. Careful control of flow rate and equivalent
30     treatment of PM10 and PM25 filters in terms of pressure drop across the filter and temperature of
31     the filter during and after collection can improve precision and accuracy.  Allen et al. (1999a)

       June 2003                                2-76        DRAFT-DO NOT QUOTE OR CITE

-------
 1      summarized several sampling issues to consider in measuring coarse particulate mass by
 2      difference, including the use of identical instrumentation (except cutpoints), filter media, filter
 3      face velocity, and ambient-filter temperature differences; common flow measurement devices;
 4      use of higher sampler flow rates (10 L/min minimum for 24-h sample is recommended); and
 5      avoiding excessive filter loading.  The concern, expressed by Allen et al. (1999a), that the "pie-
 6      plate" inlet required by the final version of the PM2 5 FRM might have a different cut point than
 7      the "flat-top" inlet of the PM10 FRM, has been addressed by a wind tunnel study which found
 8      both to have an appropriate PM10 cut point (Tolocka et al., 2001a).
 9           Since the difference method requires weighing two filters, the key to obtaining high
10      precision in the coarse mass measurement is precise measurements of filter weights. Allen et al.
11      (1999a) discuss techniques for increasing the precision of the difference method by reducing
12      errors in filter weights.  These include proper temperature and humidity controls, use of a high
13      quality microbalance, 100% replicate weighings, control of static charge, aging of new filters,
14      weighing of a sufficient number of laboratory blank filters, and accounting for buoyancy errors
15      caused by variability in barometric pressure. Allen et al. (1999a) emphasize the necessity of
16      replicate weighing of filters and a third weighing if the difference between the first two weights
17      exceeds a specified minimum. Lawless et al. (2001) investigated the magnitude of uncertainties
18      attributed to fluctuations in some of these parameters (humidity, temperature, drafts, vibration,
19      and electrostatic charges) and recommended methods for improving their control. Koistinen
20      et al. (1999) and Hanninen et al. (2002) give a excellent discussion of the procedures developed
21      to overcome problems associated with gravimetric measurements of PM2 5 mass in the EXPOLIS
22      (The Air Pollution Exposure Distributions Within Adult Urban Populations in Europe) Study.
23      They discuss factors such as corrections for buoyancy, elimination of static charge, and increases
24      in the mass of blank filters with time.  The establishment of a temperature and humidity
25      controlled  room required for the equilibration and weighing of filters for the FRM is expensive.
26      Allen et al. (2001) describe a more cost-effective technique that uses a chamber with relative
27      humidity controlled at 34% relative humidity by a saturated aqueous solution of MgCl2.
28           Allen et al.  (1999a) recommend that, in reporting precision from collocated samplers both
29      the (CV) and the  square of the correlation coefficient (r2) be reported. For a study in Boston with
30      27 pairs of mass data from collocated PM10 and PM2 5 using standard weighing methods, they
31      obtained a  CV of 4.7% and an r2 of 0.991 for PM25, a CV of 4.4% and an r2 of 0.994 for PM10,

        June 2003                                 2-77        DRAFT-DO NOT QUOTE OR CITE

-------
 1      and a CV of 15% and an r2 of 0.88 for PM10_25.  By using duplicate weighings and other
 2      techniques suggested for improving precision, they obtained a CV of 1.3% and an r2 of 0.998 for
 3      PM2 5 in a study in Chicago with 38 collocated measurements. On the basis of the improvement
 4      in the CV for PM2 5, they estimate that use of the recommended techniques for PM10_2 5 by
 5      difference would have yielded a CV of 3.8% and an r2 of 0.98 if they had been applied in the
 6      Chicago study.
 7
 8          Multistage Impaction. A second technique involves the use of impaction to isolate the
 9      size fraction between 2.5 and 10 jim Da.  In the impaction process, the air stream is first
10      accelerated through a small hole (nozzle) or slit. The air stream is directed so that it "impacts"
11      on a surface. Depending on the velocity and pressure of the air stream, particles smaller than a
12      certain  size will follow the air stream around the impactor surface. Larger particles will  impact
13      on the surface. In practice, impactors have 50% cut points similar to those for the rejection of
14      larger particles in PM25 and PM10 samples (see  Figure 2-6).
15          Multistage impactors are used to separate particles into several size fractions for the
16      determination of mass and chemical composition as a function of size (Wang and John, 1988;
17      Marple et al.,  1991). The major problem with the use of impactors to separate the 10-2.5 jim Da
18      fraction of coarse particles (thoracic coarse PM) is bounce.  Coarse particles tend to be dry, solid
19      particles.  When they hit a hard  surface, they can bounce and be carried away with the air stream
20      (e.g., Dzubay  et al., 1976;  Wesolowski et al., 1977; Rao and Whitby, 1978; Cheng and Yeh,
21      1979; Wang and John, 1987; John and Sethi, 1993). Various techniques have been used to
22      reduce bounce. One technique is to use a porous substance such as a glass or quartz fiber filter
23      (Chang et al.,  1999) material or a polyurethane  foam (Breum, 2000; Kavouras and Koutrakis,
24      2001).  These techniques may result in less precise separation and yield a sample that must be
25      extracted before chemical  analyses can be performed.  Another technique is to coat the impactor
26      with a soft wax or grease (Rao and Whitby,  1977; Turner and Hering, 1987; Pak et al., 1992).
27      This can cause problems with weighing and chemical analyses. In addition, as the deposit of
28      particles builds up, incoming particles may not  hit the  soft surface, but instead hit a previously
29      collected hard particle and bounce off of it.  The WINS impactor discussed earlier uses a filter in
30      a well of low volatility oil  to ensure a wetted surface at all times.  However, such a technique,
        June 2003                                 2-78        DRAFT-DO NOT QUOTE OR CITE

-------
 1     while appropriate for removing unwanted particles, would not yield a particle sample suitable for
 2     weighing or for chemical analyses.
 3
 4           Virtual Imp action.  The problems of bounce and blow off of particles from impactors,
 5     especially for the collection of large quantities of particles, was addressed by aerosol scientists in
 6     the mid-1960s by the development of what is now known as "virtual" impaction (Hounam and
 7     Sherwood, 1965; Conner, 1966).
 8          In a virtual impactor, a hole is placed in the impaction plate just below the accelerating jet.
 9     Two controlled flows allow a fraction, e.g., 10% (or another predetermined fraction, typically 5 -
10     20%), of the air containing the coarse particles to go through the hole and through a filter (minor
11     flow).  A 10% minor flow gives a coarse channel enrichment factor of 10.  The remaining
12     fraction (e.g., 90% of the airflow) containing the fine particles follows a different path and goes
13     through a second filter (major flow). The upper cutpoint is usually set by the inlet (e.g.,
14     10 |im Da).  The flow rates, pressures, and distance  from the nozzle to the virtual impactor
15     surface can be varied to direct particles with an Da greater than the lower cutpoint (i.e.,
16     > 2.5 |im) to go through the hole and be collected on the first filter and to direct smaller particles
17     (i.e., < 2.5  |im) to flow around the impactor be collected on the second filter. Large particles
18     "impact" into the hole with a small amount of the air flow.  The smaller particles follow the
19     major air flow around the impactor plate. This technique overcomes the problem of bounce.
20     An example of the separation into fine and coarse particles is shown in Figure 2-24.
21          The usefulness of this technique for collecting samples of fine and coarse particles for
22     chemical analysis was recognized by EPA in the mid-1970s. A development program was
23     undertaken leading to the development of the now well known "dichotomous sampler" (a virtual
24     impactor that separates particles into two size fractions) and an associated XRF analyzer
25     (Dzubay and Stevens,  1975; Loo et al., 1976; Jaklevic et al., 1977; Dzubay et al., 1977). The
26     dichotomous sampler was originally developed for use in the Regional Air Monitoring Study
27     (RAMS), part of the Regional Air Pollution Study (RAPS), conducted in St. Louis, Missouri in
28     the mid-1970s. Dichotomous samplers were operated at 10 RAMS sites from March 1975 to
29     March 1977; and 33,695 filters were collected and analyzed by XRF with an overall sampling
30     efficiency of 97.25% (Strothmann and Schiermeier, 1979; Loo et al., 1976; Loo et al.,  1978;
31     Dzubay, 1980; Lewis and Macias, 1980). Dichotomous samplers  were a novel  concept at that

       June 2003                                 2-79        DRAFT-DO NOT QUOTE OR CITE

-------
       Figure 2-24.  Schematic diagram showing the principle of virtual impaction.  The initial
                     flow, Q0, is split into a minor flow, Qt, which carries the larger particles,
                     which impact into the hole, to the coarse particle filter and a major flow, Q2,
                     which carries the smaller particles, which can follow the airflow, to the fine
                     particle filter.
       Source: Loo etal. (1976).
 1     time. Concern over particle losses and other problems at cut point sizes smaller than 2.5 jim Da
 2     influenced the decision to choose 2.5 instead of 1.0 as the cut point diameter.
 3           Subsequent to the use of the dichotomous sampler in RAPS, considerable progress has
 4     been made in the theory and practice of designing virtual impactors, especially in how to reduce
 5     losses and provide a sharp cut (Masuda et al., 1979; Marple and Chien, 1980; Chen at al., 1985,
 6     1986; Loo and Cork,  1988). Now virtual impactors, with rectangular slits or round holes, are
 7     used (a) to provide cut point sizes as low as 0.15 jim Da and (b) to concentrate coarse,
 8     accumulation, and ultrafine mode particles for use in health studies (Solomon et al., 1983;
 9     Marple et al., 1990; Sioutas et al., 1994a,b,c).  Dichotomous samplers were also used in a
10     national network to measure PM2 5 and PM10_2 5 in the Harvard Six City Study (Spengler and
11     Thurston, 1983; Dockery et al., 1993) and the Inhalable Particulate Network (Suggs and Burton,
12     1983). A trichotomous high volume sampler has also been developed that provides samples of
13     particles less than 1.0 |im, 1.0 jim - 2.5 jim, and 2.5 jim (Marple and Olsen, 1995). This
       June 2003
2-80
DRAFT-DO NOT QUOTE OR CITE

-------
 1     sampler was intended for study of the composition of the intermodal mass between 1.0 and
 2     2.5|imDa.
 3          More recently, two dichotomous sequential PM air samplers were collocated with a manual
 4     Federal Reference Method (FRM) air sampler and operated for over one year at a waterfront site
 5     on Tampa Bay (Poor et al., 2002). The FRM sampler was alternately configured as a PM2 5, then
 6     as a PM10 sampler.  For the dichotomous sampler measurements, daily 24-h integrated PM2 5 and
 7     PM10_2 5 ambient air samples were collected at a total flow rate of 16.71 min"1. As was the case in
 8     earlier versions of the dichotomous sampler, a virtual impactor split the air into flow rates of
 9     1.67 and 15.01 min'1 and collected PM10.25 and PM25 on 47-mm diameter PTFE® filters.
10     Between the two dichotomous air samplers, the average concentration, relative bias and relative
                                                                                         10-2.5
11      precision for PM2 5 were 13.3 jig m , 0.02% and 5.2% concentrations (n = 282), and for PM
12      were 12.3 |ig m'3, 3.9% and 7.7% (n = 282). FRM measurements were alternate day 24-h
13      integrated PM2 5 or PM10 ambient air samples collected onto 47-mm diameter PTFE® filters at a
14      flow rate of 16.71 min"1. Between a dichotomous and a PM25 FRM air sampler, the average
15      concentration, relative bias and relative precision were 12.4 jig m"3, -5.6% and 8.2% (n = 43);
16      and between a dichotomous and a PM10 FRM air sampler, the average concentration, relative
17      bias and relative precision were 25.7 jig m"3, -4.0% and 5.8% (n = 102). The PM25 concentration
18      measurement standard errors for two dichotomous and one  FRM samplers were 0.95, 0.79 and
19      1.02 jig m"3; and for PM10, the standard errors were 1.06,  1.59, and  1.70 jig m"3. The authors
20      (Poor et al., 2002) concluded that their results indicate that  "the dichotomous samplers have
21      superior technical merit" and demonstrate "the potential for the dichotomous sequential air
22      sampler to replace the combination of the PM25 and PM10 FRM air samplers, offering the
23      capability of making simultaneous, self-consistent determinations of these particulate matter
24      fractions in a routine ambient monitoring mode."
25           The dichotomous sampler provides two separate samples.  However, a fraction of the
26      smaller particles, equal to the minor flow, will go through the virtual impaction opening with the
27      air stream and be collected on the coarse particle filter. In the dichotomous sampler used in the
28      RAPS program, 10% of the fine particles were collected with the coarse particles. Thus, in order
29      to determine the mass or composition of the coarse particles, it is necessary to determine the
30      mass and composition of the fine particles and subtract the  appropriate fraction from the mass or
31      composition of the particles collected on the coarse particle filter. Allen et al. (1999b) discuss

        June 2003                                 2-81         DRAFT-DO NOT QUOTE OR CITE

-------
 1      potential errors in the dichotomous sampler caused by uncertainties in the coarse mass channel
 2      enrichment factor. Virtual impactors have also been designed with a clean air jet in the center of
 3      the round nozzle. This makes possible lower contamination of coarse particles by fine particles
 4      but maintains low losses and sharp cuts (Chen and Yeh, 1986; Chien and Lundgren, 1993). The
 5      fine particle intrusion into the coarse particle sample can also be reduced by operating two
 6      virtual impactors in series (Dzubay and Stevens, 1975).
 7           Aerosol physicists have also conducted theoretical and experimental investigations of
 8      virtual impaction using slits instead of round holes (Forney et al., 1978, 1982; Ravenhall et al.,
 9      1978; Masuda and Nakasita, 1988; Sioutas et al., 1994b, c, d; Ding and Koutrakis, 2000).  The
10      slit virtual impactor permits much higher flow rate than round hole virtual impactors and
11      resolves problems that occur with multihole virtual impactors (Marple et al.,  1990; Fang et al.,
12      1991).  The slit technique has been used to develop virtual impaction systems for concentrating
13      particles in the size range 0.1  to 2.5 jim Da for exposure studies using animals and people
14      (Sioutas et al., 1995a, b).  The slit impactor can also be used to concentrate coarse particles for
15      measurement (Misra et al., 2001) or exposure studies (Chang et al., 2002).  It has also been
16      possible to concentrate ultrafme particles (> 0.1  jim) by first separating ultrafme particles from
17      larger particles, adding water vapor to saturate the air containing the ultrafme particles, cooling
18      the air to cause supersaturation and growth of the ultrafme particles into the 1.0 - 4.0 |im size
19      range, concentrating these particles with a slit virtual impactor,  and heating the air to return the
20      particles to their original size (Sioutas and Koutrakis, 1996; Sioutas et al., 1999; Sioutas et al.,
21      2000; Kim et al., 2001b,c; Geller et al., 2002).
22
23      2.2.5    Speciation Monitoring
24      Speciation Network and Monitoring
25           In addition to FRM sampling to determine compliance with PM standards, EPA requires
26      states to conduct chemical Speciation sampling primarily to determine source categories and
27      trends (Code of Federal Regulations, 200 Ib). Source category apportionment calculations are
28      discussed in Chapter 3.  A PM25 chemical Speciation Trends Network (STN) has been deployed
29      that consists of 54 core National Ambient Monitoring Stations (NAMS) and approximately
30      250 State and Local Air Monitoring Stations (SLAMS). In addition, over 100 IMPROVE
31      (Interagency Monitoring of Protected Visual Environments) samplers located at regional

        June 2003                                2-82        DRAFT-DO NOT QUOTE OR CITE

-------
 1      background and transport sites can be used to fulfill SLAMS requirements.  The overall goal of
 2      the speciation program is "to provide ambient data that support the Nation's air quality program
 3      objectives" (U.S. Environmental Protection Agency, 1999). Information and reports on EPA's
 4      speciation monitoring program may be found on EPA's Technology Transfer Network at
 5      http://www.epa.gov/ttn/amtic/pmspec.html.  The NAMS speciation sites will provide routine
 6      chemical speciation data that will be used to develop annual and seasonal aerosol
 7      characterization, air quality trends analysis, and emission control strategies. The SLAMS
 8      speciation sites will further support the NAMS network and provide information for
 9      development of State Implementation Plans (SIPs).
10           At both NAMs and SLAMs sites, aerosol samples will be collected for analysis of trace
11      elements, ions (sulfate, nitrate, ammonium, sodium, and potassium), and total carbon. The
12      NAMS speciation sites will operate on a  1 in 3 day schedule, with 10 of these sites augmented
13      with continuous speciation analyses for everyday operation. The SLAMS speciation sites will
14      generally operate on a 1 in 6 day basis; however, many sites may be operated on a 1 in 3 day
15      basis in locations where increased data collection is needed. There are several samplers that are
16      suitable for use in the NAMS/SLAMS network. These samplers include an inlet cutpoint with
17      size cut characteristics comparable to the WINS FRM; proven denuder technology for nitrate;
18      and sampler face velocity and sample volume similar to that of the FRM.  The current samplers
19      include three filters: (1) Teflon for equilibrated mass and elemental analysis by  energy
20      dispersive x-ray fluorescence (EDXRF), (2) a nitric acid denuded Nylon filter for ion analysis
21      (ion chromatography), (3)  a quartz fiber filter for elemental and organic carbon.  EC and OC are
22      determined by thermal optical analysis via a modification of the NIOSH (National Institute for
23      Occupational Safety and Health) method  5040 (Thermal Optical Transmission) [TOT]).
24      However, no corrections are made for positive artifacts caused by adsorption on organic gases or
25      the quartz filters or negative artifacts due to evaporation of semivolatile organic  compounds
26      from the collected particles.
27           Since 1987, the IMPROVE network has provided measurements of ambient PM and
28      associated light extinction  in order to quantify PM chemical components that affect visibility at
29      Federal Class 1 areas that include designated national parks, national monuments, and
30      wilderness areas.  Management of this network is a cooperative effort between U.S. EPA, federal
31      land management agencies, and state governments. The IMPROVE program has established

        June 2003                                 2-83        DRAFT-DO NOT QUOTE OR CITE

-------
 1     protocols for analysis of aerosol measurements that provide ambient concentrations for PM10,
 2     PM2 5, sulfates, nitrates, organic and elemental carbon, crustal material, and a number of other
 3     elements. Information on the IMPROVE program may be found at
 4     http://vista.cira.colostate.edu/ improve.
 5           IMPROVE aerosol monitoring consists of a combination of particle sampling and sample
 6     analysis. The IMPROVE sampler, which collects two 24-hour duration samples per week,
 7     simultaneously collects one sample of PM10 on a Teflon filter, and three samples of PM25 on
 8     Teflon, nylon, and quartz filters. PM10 mass concentrations are determined gravimetrically from
 9     the PM10 filter sample, while PM2 5 mass concentrations are determined gravimetrically from the
10     PM2 5 Teflon filter sample. The PM25 Teflon filter sample is also used to determine
11     concentrations of selected elements using particle-induced x-ray emission (PIXE), x-ray
12     fluorescence (XRF), and Proton Elastic Scattering Analysis (PESA). The PM25 nylon filter
13     sample, which is preceded by a denuder to remove acidic gases,  is analyzed to determine nitrate
14     and sulfate aerosol concentrations using Ion Chromatography (1C). Finally, the PM2 5 quartz
15     filter sample is analyzed for organic and elemental carbon using the Thermal Optical Reflectance
16     (TOR) method. Corrections are made for positive artifacts but not for negative artifacts.
17           The STN and the IMPROVE networks represent a major advance in the measurement of
18     nitrate since the combination of a denuder to remove metric acid vapor and a Nylon filter to
19     adsorb nitric acid vapor that volatilizes from the collected ammonium nitrate particles
20     overcomes the loss of nitrate from Teflon filters. However, the different techniques used for the
21     measurement of OC and EC lead to significant differences between OC and EC measurements
22     when the two techniques are intercompared (Chow et al.,  2001). IMPROVE yields higher EC
23     and lower OC  although there is good agreement for TC. Another difference arises from the
24     treatment of the positive artifact due to the absorption of organic gases by the quartz filters used
25     in IMPROVE and STN samplers. More information on these differences is given in Section
26     2.2.7 and details are discussed in Appendix 2B.
27           Several of the PM2 5 size selectors developed for use in the EPA National PM2 5 STN were
28     recently evaluated by comparing their penetration curves  under clean room experiments with
29     that of the WINS impactor (Peters et al., 2001c). The corresponding speciation monitors were
30     then compared to the FRM in four cities.  The PM2 5 inlets tested were the SCC 2.141 cyclone
31     (6.7 L/min) that is in the Met One Instruments SASS sampler; the SCC 1.829 cyclone

       June 2003                                 2-84        DRAFT-DO NOT QUOTE OR CITE

-------
 1      (5.0 L/min) that is proposed for use in the Rupprecht and Patashnik real-time sulfate/nitrate
 2      monitor; the AN 3.68 cyclone (24.0 L/min) that is in the Andersen RAAS; and the spiral
 3      separator (7.0 L/min) that was previously in the Met One SASS. The cutpoints of the SCC
 4      cyclones compared reasonably well with the WINS (2.52 and 2.44 micrometers for the SCC
 5      2.141 and SCC 1.829, respectively, at their design flowrates), but both demonstrated a tail
 6      extending into the coarse particle mode. The AN inlet had the sharpest cutpoint curve, but the
 7      50% cutpoint diameter was 2.7 jim Da at its design flowrate.  The spiral inlet had the shallowest
 8      cutpoint curve, and the 50% cut point was 2.69 and 2.67 jim Da for an ungreased and greased
 9      inlet, respectively.  The speciation samplers were also compared to the FRM sampler with WINS
10      inlet under ambient conditions in four cities. The Andersen RAAS equipped with the AN 3.68
11      cyclone compared well to the FRM in all four cities, when compared on the basis of PM2 5 mass,
12      sulfate, and crustal concentrations. Greasing the spiral inlet in the Met One sampler improved
13      the performance  of that sampler, which tended to give much higher PM25 concentrations than the
14      FRM in cities with high crustal particulate matter.
15
16      2.2.6    Inorganic Elemental Analyses
17          In addition to the lighter elements, hydrogen, carbon, oxygen and nitrogen, the following
18      40 heavier elements are commonly found in ambient air samples:  sodium, magnesium,
19      aluminum, silicon, phosphorus, sulfur, chlorine, potassium, calcium, titanium, vanadium,
20      chromium, manganese, iron, cobalt, nickel, copper, zinc, gallium, arsenic, selenium, bromine,
21      rubidium, strontium, yttrium, zirconium, molybdenum, palladium, silver,  cadmium, indium, tin,
22      antimony, barium, lanthanum, gold, mercury, thallium, lead, and uranium. These often indicate
23      air pollution sources and several of them are considered  to be toxic (transition metals,
24      water-soluble metals, and metals in certain valence  states [e.g., Fe(II), Fe(III), Cr(III), Cr(VI),
25      As(III), As(V)]). Measurement methods for inorganic elements are listed in Table 2-5. These
26      methods differ with respect to detection limits, sample preparation, and cost (Chow, 1995).
27      EDXRF and PIXE are the most commonly applied methods because they quantify more than
28      40 detectable elements, they are non-destructive, and they are relatively inexpensive.  Both were
29      discussed in the previous 1996 PM AQCD. TRXRF and S-XRF are newer techniques with
30      lower detection limits. AAS, ICP-AES, and ICP-MS are also appropriate for ionic
31      measurements of elements that can be dissolved. PESA provides a means for measuring

        June 2003                                 2-85        DRAFT-DO NOT QUOTE OR CITE

-------
              TABLE 2-5.  MEASUREMENT METHODS FOR INORGANIC ELEMENTS
         #   Acronym
Full Name
          Comments
          1.   EDXRF    energy dispersive x-ray fluorescence     heavier elements
         2.   S-XRF    synchrotron induced X-ray emission
                       heavier elements; lower detection limits
                       than EDXRF
         3.     PIXE     proton induced x-ray emission
                       heavier elements; lower detection limits
                       than EDXRF
         4.    PESA    proton (or particle) elastic scattering
                        analysis
                       lighter elements
         5.   TRXRF    total reflection X-ray fluorescence
                       heavier elements; lower detection limits
                       than EDXRF
         6.    INAA    instrumental neutron activation
                        analysis
                       many elements; sensitivity different than
                       EDXRF
         7.    AAS     atomic absorption spectrophotometry    many elements that can be dissolved
         8.  ICP-AES   inductively coupled plasma with
                        atomic emission spectroscopy
                       many elements that can be dissolved
         9.   ICP-MS    inductively coupled plasma with
                        mass spectroscopy
                       many elements that can be dissolved
        10.    SEM     scanning electron microscopy
                       heavier elements
 1     elements with lower atomic numbers from hydrogen to carbon.  More detailed information on
 2     each technique is given in Appendix 2B. 1.
 3
 4     2.2.7   Elemental and Organic Carbon in Particulate Matter
 5           Ambient particles from combustion  sources contain carbon in several chemically and
 6     optically distinct forms.  Health- and visibility-related studies of these particles require
 7     information about the relative contributions to total particle mass by these different forms of
 8     carbon. With the exception of carbonate-based carbon, however, a clear classification scheme
 9     has not yet been established to distinguish organic carbon, light-absorbing carbon, black carbon,
10     soot and elemental carbon.  The absence of clear, physically-based definitions results in
11     confusion in the interpretation of speciation data. For example, depending on the radiation
12     wavelength specified, "light absorbing" carbon can include compounds that volatilize without
13     oxidation.  "Black" carbon includes various mixtures containing "elemental" (graphitic) carbon;
       June 2003
              2-86
DRAFT-DO NOT QUOTE OR CITE

-------
 1      partially degraded, oxidized graphitic fragments; and partially-oxidized amorphous aromatic
 2      carbon. For studying visibility reduction, a measurement of light-absorbing carbon may be more
 3      useful than one of elemental carbon. For source apportionment by receptor models, several
 4      consistent but distinct fractions of carbon in both source and receptor samples are desired,
 5      regardless of their light-absorbing or chemical properties.  Differences in ratios of the carbon
 6      concentrations in these fractions form part of the source profiles that distinguish the contribution
 7      of one source from the contributions of other sources (Watson et al., 1994a,b).
 8           Three method-dependent operational classes of carbon are commonly measured in ambient
 9      aerosol samples collected on quartz-fiber filters:  (1) semi-volatile organic  or non-visible light-
10      absorbing carbon, termed "organic carbon (OC)"; (2) elemental carbon, soot, black carbon, or
11      light-absorbing carbon, termed "elemental carbon (EC)"; and (3) carbon present as K2CO3,
12      Na2CO3, MgCO3, CaCO3, termed "carbonate carbon (CC)." The sum of OC, EC,  and CC in PM
13      gives the total carbon (TC).
14           The thermal/optical reflectance (TOR), thermal/optical transmission (TOT), and thermal
15      manganese oxidation (TMO) methods are most commonly used for the analysis of organic and
16      elemental carbon in atmospheric PM. In thermal separation methods, OC is vaporized and the
17      EC remaining on the filter is then oxidized to CO2 and quantified by nondispersive infrared
18      detection, by electrochemical techniques or by reducing the CO2 to CH4 and quantifying CH4 via
19      flame ionization detection (FID). OC that does not vaporize below 550 C can pyrolyze at higher
20      temperatures to form additional black carbon. Thermal/optical methods must correct for this
21      effect in order to correctly distinguish OC from EC. The various methods give similar results for
22      TC, but not for EC or OC, due to differing assumptions regarding the thermal behavior of
23      ambient aerosol carbon. These methods are discussed in detail in Appendix 2B.2.
24           Carbonate carbon (i.e., K2CO3, Na2CO3, MgCO3, CaCO3) can be determined thermally, or
25      on a separate filter section by measurement of the carbon dioxide (CO2) evolved upon
26      acidification (Johnson et al., 1980). It is usually on the order of 5% or less of TC  for ambient
27      particulate samples collected in urban areas (Appel, 1993).
28           The forms of carbon present in natural materials that may be burnt to generate atmospheric
29      aerosol tend to be poorly defined.  Thus, the pyrolysis products of these materials during
30      thermal/optical  analysis cannot be predicted.  The Geochemical Society convened an
31      international  steering committee in  1999 to define a set of representative black carbon and black

        June 2003                                 2-87        DRAFT-DO NOT QUOTE OR CITE

-------
 1      carbon-containing benchmark materials to be used to support ambient aerosol sample analysis.
 2      These standard materials may be used to provide thermal/optical "fingerprints" for deducing
 3      primary aerosol sources, and to establish characteristic analytic interferences or artifacts
 4      associated with such sources. The materials recommended to date include n-hexane soot,
 5      lignocellulosic chars, soils, marine sediments, and the NIST urban dust standard reference
 6      material (SRM 1649a).  The committee has also recommended a set of standard materials that
 7      may potentially interfere with black carbon analyses: shale, natural organic matter, melanoidin
 8      (an amino acid-based material) and coals. These recommendations are discussed on the steering
 9      committee website: http://www.du.edu/~dwismith/bcsteer.html.
10
11      2.2.8    Ionic Species
12           Ion  chromatography (1C) is widely used for analyzing ionic species in the water-soluble
13      portion of suspended PM. 1C is the method of choice for the measurement of sulfate, nitrate,
14      ammonium, sodium, and potassium ions for the NAMS program. Aerosol strong acidity,  H+, is
15      determined by titration of a water solution of PM collected following a series of annular
16      denuders  to remove acid and basic gases with back-up filters to collect NH3 and HNO3 that
17      volatilize from the PM during collection.  The 1996 PM AQCD (U.S. Environmental Protection
18      Agency, 1996a) discussed measurement of ions by 1C (Section 4.3.3.1) and of strong acidity
19      (Sections 3.3.1.1 and 4.3.3.1); so, no further details will  be discussed here.
20
21      2.2.9    Continuous Monitoring
22           The EPA expects that many local environmental agencies will operate continuous PM
23      monitors. All currently available continuous measurements of suspended particle mass share the
24      problem of dealing with semivolatile PM components. So as not to include particle-bound water
25      as part of the mass, the particle-bound water must be removed by heating or dehumidification.
26      However, heating also causes loss of ammonium nitrate  and semivolatile organic components.
27      A variety of potential candidates for continuous measurement of particle mass and related
28      properties are listed in Table 2-6.  These techniques are discussed in more detail in Appendix
29      2B.3.
30
31

        June 2003                                2-88        DRAFT-DO NOT QUOTE OR CITE

-------
     TABLE 2-6.  METHODS FOR CONTINUOUS MEASUREMENT OF PM MASS,
                                      COMPONENTS, ETC.
#
1.
2.
3.
4.
5.
6.
7.
8.
9.
Acronym
TEOM
TEOM
with SES
—
FDMS
RAMS
CAMM
—
—
CCPM
Name
Tapered Element
Oscillating Microbalance
TEOM with Sample
Equilibration System
Differential TEOM
Filter Dynamics
Measurement System
Real-Time Total Ambient
Mass Sampler
Continuous Ambient
Mass Monitor
Beta gauge
Piezoelectric
microbalance
Continuous Coarse
Particle Monitor
Comments
Measures only nonvolatile components of PM.
By using dehumidification instead of heat for drying and a lower
temperature for stabilization, includes some of the semivolatile
components of PM.
Research instrument designed to measure both the nonvolatile
and semivolatile components of PM.
Commercial version of the differential TEOM.
Research instrument designed to measure total PM using
denuders and an adsorptive filter.
Measures total PM by pressure drop across a frequently
changed filter.
Measures PM mass by beta attenuation. Unless dried by heat or
dehumidification will also measure particle-bound water.
Measures mass by change in resonance frequency when particles
are deported on a crystal.
Virtual impaction is used to concentrate PM10_2 5 which is then
measured by a TEOM.
 10.
Semi-continuous EC/OC
Several commercially available instruments automate the thermal
technique for EC/OC and provide hourly measurements.
 11.
Semi-continuous nitrate
Collection of PM followed by flash vaporization and
determination of NOX provides 10 minute measurements of
paniculate nitrate.
 12.
Semi-continuous sulfate
Several techniques are available in which paniculate sulfate is
measured using flame ionization detection.
 13.       —
Continuous ion
chromatography of
water-soluble ions
Particles are grown by mixing with water vapor, collected in
water, and injected into an automatic ion chromatography.
 14.       —
Mass spectroscopy of
individual particles
Single particles are evaporated, ionized and the components
analyzed by mass spectroscopy. Several different systems are in
use in research studies.
 15.     BAD     Electrical Aerosol
                  Detector
                         This instrument measures charge collected by particles and gives
                         a continuous signal that is proportional to the integral of the
                         particle diameter.
 16.
Integrating nephelometer
Light scattering by suspended particles, collected over a large
solid angle, provides an indicator for particle mass including
particle-bound water unless the air sample is dried by heating or
dehumidification.
June 2003
                              2-89
                   DRAFT-DO NOT QUOTE OR CITE

-------
 1      2.2.10  Measurements of Individual Particles
 2           Recently, several researchers have developed instruments for real-time in situ analysis of
 3      single particles (e.g., Noble and Prather, 1996; Gard et al., 1997; Johnson and Wexler, 1995;
 4      Silva and Prather, 1997; Thomson and Murphy, 1994).  Although the technique varies from one
 5      laboratory to another, the underlying principle is to fragment each particle into ions, using either
 6      a high-power laser or a heated surface and, then, a time-of-flight mass spectrometer (TOFMS) to
 7      measure the ion fragments in a vacuum. Each particle is analyzed in a suspended state in the air
 8      stream (i.e., without collection), avoiding sampling artifacts associated with impactors and
 9      filters. The technique is called aerosol time-of-flight mass spectrometry (ATOFMS).
10      By measuring both positive and negative ions from the same particle, information can be
11      obtained about the composition, of individual particles of known aerodynamic diameter.  This
12      information is especially useful in determining sources of particles.  Noble and Prather (1996)
13      used ATOFMS to provide compositionally resolved particle-size distributions. Their instrument
14      is capable of analyzing size and chemical composition of 50 to 100 particles/min at typical
15      ambient concentrations and up to 600/min at high particle concentrations.  Four systems for
16      measurement of single particles using mass spectroscopy are reviewed by Middlebrook (2002).
17      An example of the type of information that can be determined is shown in Figure 2-25.
18           Because particles are analyzed individually, biases in particle sampling (the efficiency of
19      particle transmission into the sensor chamber as a function of size; particle size measurement,
20      and detection of particles prior to fragmentation) represent a major challenge for these
21      instruments.  Moreover, the mass spectrometer has a relatively large variability in ion yields
22      (i.e., identical samples would yield relatively large differences in mass spectrometer signals
23      [Thomson and Murphy, 1994]); therefore, quantitation is inherently difficult (Murphy and
24      Thomson, 1997).  Quantitation will be even more challenging for complex organic mixtures
25      because of the following two reasons:  (1) a large number of fragments are generated from each
26      molecule, and (2) ion peaks for organics can be influenced or obscured by inorganic ions
27      (Middlebrook et al., 1998). Nonetheless, scientists have been successful in using these
28      techniques to identify the presence of organics in atmospheric particles and laboratory-generated
29      particles (i.e., as contaminants in laboratory-generated sulfuric acid droplets) as well as the
30      identification of specific compound classes such as PAHs in combustion emissions (Castaldi and
31      Senkan,  1998; Hinz et al., 1994; Middlebrook et al., 1998; Murphy and Thomson, 1997;

        June 2003                                 2-90        DRAFT-DO NOT QUOTE OR CITE

-------
                 c
                 o
                 O
                 a»
                 "O
                 tr
                 CO
                 CL
                 0)
                 (B
                        0.2
                       0.3 0.4 0.5 0.6 0.70 80.91.0
                                Aerodynamic Diameter (|jm)
                                                                                Organic
                                                                              Soil
                                                                     3,0   4.0
       Figure 2-25.  Size distribution of particles divided by chemical classification into organic,
                     marine, and crustal.
       Source: Noble and Prather (1998).
 1
 2
 3
 4
 5
 9
10
11
12
13
Neubauer et al., 1998; Noble and Prather, 1998; Reilly et al., 1998; Silva and Prather, 1997).
A new multivariate technique for calibration of ATOFMS using microorifice impactors shows
promise for simplifying the calibration process (Fergenson et al., 2001).  This calibration
technique has been applied to gasoline and diesel particles to demonstrate the feasibility of using
this technique for the source apportionment of gasoline and diesel particles in an atmospheric
mixture (Song et al., 2001).
     Until recently, ATOFMS systems have only been able to characterize particles that are
larger than approximately 0.2 to 0.3 jim in diameter. Wexler and colleagues (Carson et al.,
1997; Ge et al., 1998) have developed an ATOFMS instrument that is able to size, count, and
provide chemical  composition on individual particles ranging in size from 10 nm to 2 |im.
       June 2003
                                          2-91
DRAFT-DO NOT QUOTE OR CITE

-------
 1      2.2.11  Low Flow Filter Samples for Multiday Collection of
 2              Particulate Matter
 3          For some purposes, such as demonstrating attainment of an annual standard or as an
 4      exposure indicator for epidemiologic studies of chronic health effects, 24-h measurements are
 5      not essential. Annual or seasonal averages may be adequate. Multiday sampling techniques can
 6      result in lower costs for weighing, chemical analysis, and travel time to change filters.  The
 7      multiday sampler serves a second purpose. Most commercially available samplers are optimized
 8      for collecting 24-h samples of the PM concentrations found in the U.S., Europe, or Japan. Many
 9      cities in other parts of the world have significantly higher PM concentrations. Under these
10      conditions, the  16.7 L/min flow through 37 or 47 mm diameter filters may overload the filter and
11      prevent the sampler from maintaining the prescribed flow rate for 24 h. A low flow sampler
12      with a 0.4 L/min flow rate and a 47 mm diameter filter has been designed by Aerosol Dynamics,
13      Inc. With this sampler, the sample collection time can be chosen to suit the ambient
14      concentration level. This sampler, with a one-week collection period, has been used to
15      characterize PM2 5 in Beijing, PRC (He et al., 2001). With a two-week collection period, it is
16      being used in a chronic epidemiologic study in southern California, USA (Gauderman, et al.,
17      2000).
18          The sampler, as described by He et al. (2001), has three PM25 channels. One channel
19      collects PM on a Teflon filter for gravimetric mass measurement and elemental analysis by XRF.
20      A second channel collects PM on a quartz filter for organic and elemental carbon analysis.
21      A denuder to remove organic gases and a backup filter to collect semivolatile organic
22      compounds may be added. The third channel uses a carbonate denuder to remove acid gases
23      (HNO3 and SO2), a Teflon filter to collect PM for analysis of ions by ion chromatography, and a
24      nylon filter to collect volatilized nitrate. The Teflon filter can also be weighed prior to
25      extraction. Thus, the multiday sampler can provide the information needed for source
26      apportionment by Chemical Element Balance techniques (Watson et al., 1990a,b; U.S.
27      Environmental  Protection Agency, 2002b).
28          For monitoring sites with high day-to-day variability in PM concentrations, a sample
29      integrated over a week may provide a more accurate measurement of the annual average than
30      can be obtained by l-in-3 or l-in-6 day sampling schedules. Daily PM data from Spokane, WA
31      were resampled to simulate common sampling schedules and the error due to less-than-everyday
32      sampling was computed (Rumburg et al., 2001). The error in the annual mean concentration for
        June 2003                                 2-92        DRAFT-DO NOT QUOTE OR CITE

-------
 1      PM25, expressed as a percentage difference from the everyday sampling mean, was 1.7, 3.4, and
 2      7.7% for l-in-2 day, l-in-3 day, and l-in-6 day sampling, respectively.
 3
 4
 5      2.3  SUMMARY AND KEY POINTS
 6      2.3.1   Atmospheric Physics and Chemistry of Particles
 7           Atmospheric particles originate from a variety of sources and possess a range of
 8      morphological, chemical, physical, and thermodynamic properties. The composition and
 9      behavior of aerosols are linked with those of the surrounding gas. Aerosol may be defined as a
10      suspension of solid or liquid particles in air and includes both the particles and all vapor or gas
11      phase components of air. However, the term aerosol often is used to refer to the suspended
12      particles only.
13           A complete description of the atmospheric aerosol would include an accounting of the
14      chemical composition, morphology, and size of each particle, and the relative abundance of each
15      particle type as a function of particle size. Recent developments in single particle analysis
16      techniques are bringing such a description closer to reality.
17           The diameter of a spherical particle may  be determined geometrically, from optical or
18      electron microscopy, by light scattering and Mie theory,  or by a particle's behavior (e.g.,
19      electrical mobility or its aerodynamic behavior).  However, the various types of diameters may
20      be different, and atmospheric particles  often are not spherical. Therefore, particle diameters are
21      described by an "equivalent" diameter. Aerodynamic diameter, Da (the diameter of a unit
22      density sphere that would have the same terminal settling velocity as the real particle),  and the
23      Stokes diameter, Dp (the diameter of a sphere of the same density as the particle that would have
24      the same aerodynamic resistance  or drag), are the most widely used equivalent diameters.
25           Atmospheric size distributions show that most atmospheric particles are quite small, below
26      0.1 |im; whereas most of the particle volume (and therefore most of the mass) is found  in
27      particles greater than 0.1  jim. An important feature of the mass or volume size distributions of
28      atmospheric particles is their multimodal nature.  Volume distributions, measured in ambient air
29      in the United States, are almost always found to have a minimum between 1.0 and 3.0 jim. That
30      portion of the size distribution that contains particles that are mostly larger than the minimum is
31      called "coarse" particles or the "coarse" mode. That portion of the size distribution that contains

        June 2003                                 2-93        DRAFT-DO NOT QUOTE OR CITE

-------
 1      particles that are mostly smaller than the minimum is called "fine" particles and includes several
 2      modes. "Accumulation-mode" refers to that portion of fine particles with diameters above about
 3      0.1 |im.  That portion of fine particles with diameters below 0.1 jim are usually called "ultrafme"
 4      by toxicologists and epidemiologists and "nanoparticles" by aerosol physicists and material
 5      scientists.  In the number distribution of ultrafme particles, particles in the size range below 0.01
 6      are called the  nucleation mode and particles between 0.01 and 0.1 are called the Aitken mode.
 7      The Aitken mode can be observed as a separate mode in mass or volume distributions only in
 8      clean or remote areas or near sources of new particle formation by nucleation.
 9           The aerosol community uses three different approaches or conventions in the classification
10      of particles by size:  (1) modes, based on the observed size distributions and formation
11      mechanisms; (2) cut point, usually based on the 50% cut point of the specific sampling device,
12      including legally specified, regulatory sizes for air quality standards; and (3) dosimetry or
13      occupational sizes, based on the entrance into various compartments of the respiratory system.
14      Over the years, the terms fine and coarse as applied to  particle sizes have lost the original precise
15      meaning of fine and coarse.  In any given article, therefore, the meaning of fine and coarse,
16      unless defined, must be inferred from the author's usage. In particular, PM2 5 and fine particles
17      are not equivalent.  PM25 refers to the aggregate sample of PM that is collected following a size-
18      selective inlet with a specified penetration as a function of size and a 50%  cutpoint of 2.5 jim Da.
19      It may also be used to refer to number (or other measure of particles suspended in the
20      atmosphere that would be collected by such a  sampler). PM10 is defined similarly. PM10_2 5
21      refers to the sample that would be collected if the PM2 5 component could be removed from a
22      PM10 sample.
23           Several processes influence the formation and growth of particles. New particles may be
24      formed by nucleation from gas phase material. Particles may grow by condensation as gas phase
25      material condenses onto existing particles. Particles may also grow by coagulation as two
26      particles combine to form one.  Gas phase material condenses preferentially on smaller particles,
27      and the rate constant for coagulation of two particles decreases as the particle size increases.
28      Therefore, nuclei mode particles grow into the accumulation mode, but growth of accumulation
29      mode particles into the coarse mode is rare.
30           The major constituents of atmospheric PM are sulfate, nitrate, ammonium, and hydrogen
31      ions; particle-bound water; elemental carbon;  a great variety of organic compounds; and  crustal

        June 2003                                 2-94        DRAFT-DO NOT QUOTE OR CITE

-------
 1      material.  Atmospheric PM contains a large number of elements in various compounds and
 2      concentrations and hundreds of specific organic compounds.  Particulate material can be primary
 3      or secondary. PM is called primary if it is in the same chemical form in which it was emitted
 4      into the atmosphere. PM is called secondary if it is formed by chemical reactions in the
 5      atmosphere.  Primary coarse particles are usually formed by mechanical processes; whereas
 6      primary fine particles are emitted from sources either directly as particles or as vapors that
 7      rapidly condense to form particles.
 8           Most of the sulfate and nitrate and a portion of the organic compounds in atmospheric
 9      particles are  secondary. Secondary aerosol formation depends on numerous factors including the
10      concentrations of precursors; the concentrations of other gaseous reactive species such as ozone,
11      hydroxyl  radical, peroxy radicals, and hydrogen peroxide; atmospheric conditions, including
12      solar radiation and relative humidity; and the interactions of precursors and preexisting particles
13      within cloud or fog droplets or on or in the liquid film on solid particles. As a result, it is
14      considerably more difficult to relate ambient concentrations of secondary species to  sources of
15      precursor emissions than it is to identify the sources of primary particles.
16           The lifetimes of particles vary with  particle size. Coarse particles can settle rapidly from
17      the atmosphere within minutes or hours and normally travel only short distances. However,
18      when mixed high into the atmosphere, as  in dust storms, the smaller-sized, coarse-mode particles
19      may have longer lives and travel greater distances.  Accumulation-mode particles are kept
20      suspended by normal air motions and have a lower deposition velocity than coarse-mode
21      particles.  They can be transported thousands of kilometers and remain in the atmosphere for a
22      number of days.  Accumulation-mode particles are removed from the atmosphere primarily by
23      cloud processes.  Dry deposition rates are expressed in terms of a deposition velocity that varies
24      with the particle size, reaching a minimum between 0.1 and 1.0 jim aerodynamic diameter.
25           PM is a factor in acid deposition. Particles serve as cloud condensation  nuclei  and
26      contribute directly to the acidification of rain. In addition, the gas-phase species that lead to dry
27      deposition of acidity are also precursors of particles.  Therefore, reductions in SO2 and NOX
28      emissions will decrease both acid deposition and PM concentrations.  Sulfuric acid, ammonium
29      nitrate, and organic particles also are deposited on surfaces by dry deposition and can contribute
30      to ecological effects.
31

        June 2003                                 2-95        DRAFT-DO NOT QUOTE OR CITE

-------
 1      2.3.2    Measurement of Atmospheric Particles
 2          The decision by the EPA to revise the PM standards by adding daily and yearly standards
 3      for PM2 5 has led to a renewed interest in the measurement of atmospheric particles and to a
 4      better understanding of the problems in developing precise and accurate measurements of
 5      particles. Unfortunately, it is very difficult to measure and characterize particles suspended in
 6      the atmosphere.
 7          PM monitoring is used to develop information to guide implementation of standards (i.e.,
 8      to identify sources of particles; to determine whether or not a standard has been attained; and to
 9      determine health, ecological, and radiative effects). Federal Reference Methods (FRM) specify
10      techniques for measuring PM10 and PM2 5.  Particles are collected on filters and mass
11      concentrations are determined gravimetrically. The PM10 FRM sampler consists of a PM10
12      inlet/impactor and a 47-mm Teflon filter with a particle collection efficiency greater than 99.7%.
13      The PM25 FRM is similar except that it includes a PM2 5 impactor with an oil-covered impaction
14      substrate to remove particles larger than 2.5 jim.  Both techniques provide relatively precise
15      (±10%) methods for determining the mass of material remaining on a Teflon filter after
16      equilibration. Despite considerable progress in measuring the atmospheric PM mass
17      concentration, numerous uncertainties continue to exist as to the relationship between the mass
18      and composition of material remaining on the filter as measured by the FRMs and the mass and
19      composition of material that exists in the atmosphere as suspended PM. There is no reference
20      standard for particles suspended in the atmosphere, nor is there an accepted way to remove
21      particle-bound water  without losing some of the semivolatile components of PM such as
22      ammonium nitrate and semivolatile organic compounds.  It also is difficult to cleanly separate
23      fine and coarse PM. As  a result, EPA defines accuracy for PM measurements in terms of
24      agreement of a candidate sampler with a reference sampler.  Therefore, intercomparisons of
25      samplers become very important in determining how well various samplers agree and how
26      various design choices influence what is actually measured.
27          Current filtration-based mass measurements lead to significant evaporative losses of a
28      variety of semivolatile components (i.e., species that exist in the atmosphere in dynamic
29      equilibrium between the condensed phase and gas phase) during and possibly after collection.
30      Important examples include ammonium nitrate, semivolatile organic compounds, and particle-
31      bound water. Loss of these components may significantly affect the quality of the measurement

        June 2003                                 2-96       DRAFT-DO NOT QUOTE OR CITE

-------
 1      and can lead to both positive and negative sampling artifacts. Negative artifacts resulting from
 2      loss of ammonium nitrate and semivolatile organic compounds may occur during sampling
 3      because of changes in temperature, relative humidity, or composition of the aerosol or because
 4      of the pressure drop across the filter. Negative artifacts also may occur during handling and
 5      storage because of evaporation.  Positive artifacts occur when gas-phase compounds (H2O,
 6      HNO3, SO2, and organic compounds) absorb onto or react with filter media or collected PM or
 7      when some particle-bound water is not removed.
 8           Sampling systems for semivolatile PM components make use of denuders to remove the
 9      gas-phase fraction and absorptive filters to remove the condensed phase and retain any material
10      that subsequently evaporates from the collected PM. The loss of particulate nitrate may be
11      determined by comparing nitrate collected on a Teflon filter to that collected on a nylon filter
12      (which absorbs nitric acid which evaporates from ammonium nitrate particles) preceded by a
13      denuder to remove gas-phase nitric acid.  In two studies in southern California, the PM2 5 mass
14      lost because of volatilization of ammonium nitrate was found to represent 10 to 20% of the total
15      PM25 mass and almost a third of the nitrate.  Denuder/absorptive filter sampling systems also
16      have been developed for measuring parti culate phase organic compounds. This technique is an
17      improvement over the filter/adsorbent collection method.  However, the denuder systems
18      currently discussed in the literature are not straightforward in their use, and the user must have a
19      thorough understanding of the technology. The FRM for PM25 will likely suffer loss of
20      particulate nitrates and semivolatile organic compounds, similar to the losses experienced with
21      other single filter collection systems.
22           It is generally desirable to collect and measure ammonium nitrate and semivolatile organic
23      compounds as part of particulate matter mass.  However, it is usually desirable to remove the
24      particle-bound water before determining the mass. In some situations, it may be important to
25      know how much of the suspended particle's mass or volume results from particle-bound water.
26      Calculations and measurements indicate that aerosol water content is strongly dependent on
27      relative humidity and composition. Particle-bound water can represent a  significant mass
28      fraction of the PM concentration at relative humidities above 60%. A substantial fraction of
29      accumulation-mode PM is hygroscopic or deliquescent. The more hygroscopic particles tend to
30      contain more sulfates, nitrates, and secondary organic compounds, while the less hygroscopic
31      particles tend to contain more elemental carbon, primary organic compounds, and crustal

        June 2003                                 2-97        DRAFT-DO NOT QUOTE OR CITE

-------
 1      components. Fresh, submicron-size soot particles may tend to shrink with increasing relative
 2      humidity because of a structural change. The effects of relative humidity on the sorption of
 3      semivolatile organic compounds on particles are not well understood. The amount of water
 4      sorbed to an atmospheric aerosol may be affected by the presence of an organic film on the
 5      particle, which may impede the transport of water across the surface.
 6           Fine and coarse particles differ not only in formation mechanisms and size, but also in
 7      sources; composition; and chemical, physical, and biological properties. Fine and coarse
 8      particles overlap in the intermodal size range (1-2.5 jim Da).  As relative humidity increases, fine
 9      particles grow into this size range; as relative humidity decreases, more coarse particles may be
10      suspended in this size range. It is desirable to measure fine PM and coarse PM separately in
11      order to properly allocate health effects to either fine PM or coarse PM and to correctly
12      determine sources by factor analysis or chemical mass balance. The selection of a cut point of
13      2.5 |im as a basis for EPA's 1997 NAAQS for fine particles (Federal Register, 1997) and its
14      continued use in many health effects studies reflects the  importance placed on more complete
15      inclusion of accumulation-mode particles, while recognizing that intrusion of coarse-mode
16      particles can occur under some conditions with this cut point.
17           In addition to FRM sampling of equilibrated mass to determine compliance with PM
18      standards, EPA requires states  to conduct speciation sampling primarily to determine source
19      categories and trends. The current speciation samplers collect PM25 on three filters:
20      (1) a Teflon filter for gravimetric determination of mass  and for analysis of heavy elements by
21      X-ray fluorescence; (2) a Nylon filter preceded by a nitric acid denuder for artifact-free
22      determination of nitrate and measurement of other ionic  species by  ion chromatography; and
23      (3) a quartz filter for measurement of elemental carbon (EC) and  organic carbon (OC). In
24      addition, IMPROVE (Interagency Monitoring of Protected Visual Environments) samplers
25      provide information on regional PM background and transport. IMPROVE samplers, in addition
26      to the three  types of filters collected by the speciation samplers, also collect a PM10 sample.  The
27      IMPROVE  and speciation networks use slightly different methods for determination of EC and
28      OC. The two methods agree on total carbon but differ in the split of total carbon into EC and
29      OC. The two methods also differ in their correction for positive artifacts due to absorption of
30      volatile organic compounds on the  quartz filters. Neither EC/OC method provides for any
        June 2003                                2-98        DRAFT-DO NOT QUOTE OR CITE

-------
 1      correction for negative artifacts due to evaporation of semivolatile organic compounds from the
 2      collected particles.
 3           The EPA expects that monitoring agencies will operate continuous PM monitors; and EPA
 4      is in the process of providing guidance regarding appropriate continuous monitoring techniques.
 5      All currently available techniques for continuous measurements of suspended particle mass, e.g.,
 6      the integrating nephelometer, the beta-absorption monitor, and the Tapered Element Oscillating
 7      Microbalance (TEOM), share the problem of dealing with semivolatile PM components: that is,
 8      in order not to include particle-bound water as part of the mass, the particle-bound water must be
 9      removed by heating or dehumidification; however, heating also causes ammonium nitrate and
10      semivolatile organic compounds to evaporate. The TEOM monitor operates at a constant, but
11      higher than ambient, temperature to remove particle-bound water; whereas, the FRM is required
12      to operate at no more than 5 °C above the ambient temperature. Subsequently, much of the
13      particle-bound water is removed during equilibration at 40% relative humidity.  This difference
14      in techniques for removal of particle-bound water causes differences in the measured mass
15      concentration between the TEOM and FRMs.
16           Several new techniques for continuous PM mass measurements are currently being field
17      tested. The Real-Time Total Ambient Mass Sampler (RAMS) measures the total mass of
18      collected particles including semivolatile species with a TEOM monitor using a "sandwich
19      filter." The sandwich contains a Teflon-coated particle-collection filter followed by a charcoal -
20      impregnated filter to collect any semivolatile species lost from the particles during sampling.
21      The RAMS uses a Nafion dryer to remove particle-bound water from the suspended particles and
22      a particle concentrator to reduce the quantity of gas phase organic compounds that must be
23      removed by the denuder.  The Continuous Ambient Mass Monitor (CAMM) estimates ambient
24      particulate matter mass by measurement of the increase in the pressure drop across a membrane
25      filter caused by particle loading. It also uses a Nafion dryer  to remove particle-bound water.
26      A new differential TEOM offers the possibility of measuring both nonvolatile and semivolatile
27      PM mass.  In addition to continuous mass measurement, a number of techniques for continuous
28      measurement of sulfate, nitrate, or elements are being tested.
29
30
31

        June 2003                                 2-99        DRAFT-DO NOT QUOTE OR CITE

-------
 1     2.3.3    Key Points
 2     Fine and Coarse Particles
 3          Particle size distributions show that atmospheric particles exist in two classes, fine
 4     particles and coarse particles.  Fine and coarse particles are defined primarily in terms of their
 5     formation mechanisms and size;  an they also differ in sources, chemical composition, and
 6     removal processes (see Table 2-1). Subsequent chapters will show that fine and coarse particles
 7     also differ in aspects of concentration, exposure, dosimetry, toxicology,  and epidemiology.
 8          These differences support the setting of separate standards for fine and coarse particles.
 9     Fine and coarse particles overlap in the size range between 1 and 3 jim aerodynamic diameter
10     where particulate matter (PM) concentrations are at a minimum. Coarse particles are generally
11     larger than this minimum and are generally formed by mechanical  processes. Coarse particles
12     and coarse-mode particles are  equivalent terms.  Fine particles are  generally smaller than the
13     minimum and are generally formed by coagulation and condensation of gases.  Fine particles
14     are subdivided into accumulation, Aitkin, and nucleation modes. In earlier texts, nuclei mode
15     referred to the size range now  split into the Aitkin and nucleation modes (see Figures 2-4
16     and 2-5).
17
18     Measurement of Mass and Composition
19     Nonvolatile PM. Analytical techniques exist for measurement of the mass and chemical
20     composition of PM retained on a filter (nonvolatile mass) in terms  of elements  (except carbon)
21     and certain key ions (sulfate, nitrate, hydrogen, and ammonium). Acceptable measurements can
22     be made of the total carbon retained on a filter.  However, the split into organic carbon and
23     elemental carbon depends on the operational details of the analytical methods and varies
24     somewhat among methods.  Determination of the various organic compounds in the organic
25     carbon fraction remains a challenge.
26
27     Semivolatile PM.  Important components of atmospheric PM (particle-bound water, ammonium
28     nitrate,  and many  organic compounds) are termed semivolatile because significant amounts of
29     both the gaseous and condensed  phases may exist in the atmosphere in equilibrium. Particle-
30     bound water is not considered a pollutant.  Most of the particle-bound water is  removed by
31     heating the particles or by equilibration of the collected particles at a low relative humidity

       June 2003                                2-100       DRAFT-DO NOT QUOTE OR CITE

-------
 1      (40%) for 24 hours. However, these processes also cause the loss of other semivolatile
 2      components. Semivolatile components also evaporate from the filter during sampling due to the
 3      pressure drop across the filter or due to a reduction in the atmospheric concentration during the
 4      sampling time.  Filter collection and equilibration techniques for PM, such as prescribed by the
 5      Federal Reference Methods, lose a fraction of the semivolatile PM.  Continuous methods must
 6      dry the PM to remove particle-bound water. If heating is used to dry the particles, more of the
 7      semivolatile components may be removed than are lost in filter sampling.  Collection and
 8      retention of ammonium nitrate and semivolatile organic compounds represents a major challenge
 9      in the effort to move to continuous measurement of PM mass. The use of diffusion dryers,
10      which dehumidify the air stream without heating, represents a promising approach. Uncertainty
11      in the efficiency of retention of ammonium nitrate and organic compounds on filters also
12      impacts source category attribution and epidemiologic studies.
13
14      Separation of Fine and Coarse PM
15           Techniques are available to separate fine particles from coarse particles and collect the fine
16      particles on a filter. No such technique exists for coarse particles.  As yet, no consensus exists
17      on the best technique for collecting a coarse particle sample for determination of mass  and
18      composition. Candidates include multistage impaction, virtual impaction, and difference
19      (subtracting PM25 mass or composition from PM10 mass or composition). Advances in the
20      theory and practice of virtual impaction suggest that it would be possible to design virtual
21      impactors with much less than the 10% of fine PM collected in the coarse PM sample as is now
22      the case for the dichotomous samplers used in air quality studies and with penetration curves as
23      sharp as those used in the current Federal Reference Method for PM2 5.
        June 2003                                 2-101       DRAFT-DO NOT QUOTE OR CITE

-------
         REFERENCES

  1       Aiginger, H. (1991) Historical development and principles of total reflection x-ray fluorescence analysis (TXRF).
  2             Spectrochim. ActaPartB 46: 1313-1321.
  3       Aiginger, H.; Streli, C. (1997) Total reflection x-ray fluorescence spectrometry (TXRF) for trace element analysis.
  4             Spectrosc. Eur. 9: 16-21.
  5       Allen, G.; Sioutas, C.; Koutrakis, P.; Reiss, R.; Lurmann, F. W.; Roberts, P. T. (1997) Evaluation of the TEOM®
  6             method for measurement of ambient paniculate mass in urban areas. J. Air Waste Manage. Assoc.
  7             47:682-689.
  8       Allen, G. A.; Oh, J. A.; Koutrakis, P.; Sioutas, C. (1999a) Techniques for high-quality ambient coarse paniculate
  9             mass measurements. J. Air Waste Manage. Assoc. 49(special issue): PM133-PM141.
10       Allen, J. O.; Paschkewitz, J. S.; Plummer, E. F.; Lafleur, A. L.; Sarofim, A. F.; Smith, K. A. (1999b) Absorption of
11             semi-volatile compounds in oiled impaction substrates: measurement of pyrene absorption. Aerosol Sci.
12             Technol. 30: 16-29.
13       Allen, G. A.; Lawrence, J.; Koutrakis, P. (1999c) Field validation of a semi-continuous method for aerosol black
14             carbon (aethalometer) and temporal patterns of summertime hourly black carbon measurements in
15             southwestern PA. Atmos. Environ. 33: 817-823.
16       Allen, R.; Box, M.; Larson, T. V.; Liu, L. J. (2001) A cost-effective weighing chamber for paniculate matter filters.
17             J. Air Waste Manage. Assoc. 51:  1650-1653.
18       American Conference of Governmental Industrial Hygienists. (1994) 1994-1995 threshold limit values for chemical
19             substances and physical agents and biological exposure indices. Cincinnati, OH: American Conference of
20             Governmental Industrial Hygienists.
21       Andrews, E.; Larson, S. M. (1993) Effect of surfactant layers on the size changes of aerosol particles as a function of
22             relative humidity. Environ. Sci. Technol. 27: 857-865.
23       Andrews, E.; Saxena, P.; Musarra, S.; Hildemann, L .M.; Koutrakis, P.; McMurry, P. H.; Olmez, I.; White, W. H.
24             (2000) Concentration and composition of atmospheric aerosols from the 1995 SEAVS experiment and a
25             review of the closure between chemical and gravimetric measurements. J. Air Waste Manage. Assoc.
26             50:648-664.
27       Appel, B. R. (1993) Atmospheric sample analysis and sampling artifacts. In: Willeke, K.; Baron, P. A., eds. Aerosol
28             measurement: principles, techniques, and applications. New York, NY: Van Nostrand Reinhold; pp.  233-259.
29       Appel, B. R.; Tokiwa, Y.; Haik, M. (1981) Sampling of nitrates in ambient air. Atmos. Environ. 15: 283-289.
30       Appel, B. R.; Tokiwa, Y.; Haik, M.; Kothny, E. L. (1984) Artifact paniculate sulfate and nitrate formation on filter
31             media. Atmos. Environ. 18: 409-416.
32       Arnott, W. P.; Moosmtiller, H.; Rogers, C. F.; Jin, T.; Bruch, R. (1999) Photoacoustic spectrometer for measuring
33             light absorption by aerosol; instrument description. Atmos. Environ. 33: 2845-2852.
34       Arnott, W. P.; Moosmtiller, H.; Walker, J. W. (2000) Nitrogen dioxide and kerosene-flame soot calibration of
3 5             photoacoustic instruments for measurement of light absorption by aerosols. Rev.  Sci. Instruments
36             71:4545-4552.
37       Ayers, G. P.; Keywood, M. D.; Gras, J. L. (1999) TEOM vs. manual gravimetric methods for determination of PM2 5
38             aerosol mass concentrations. Atmos. Environ. 33: 3717-3721.
39       Babich, P.; Davey, M.; Allen,  G.; Koutrakis, P. (2000) Method comparisons for paniculate nitrate, elemental carbon,
40             andPM25 mass in seven U.S. cities. J. Air Waste Manage. Assoc. 50: 1095-1105.
41       Bacon, J. R.; Chenery, S. R. N.; Ellis, A. T.; Fisher, A.; McMahon, A. W.; Potts, P. J.; Williams, J. G. (1995)
42             Atomic spectrometry update-atomic mass spectrometry and x-ray fluorescence spectrometry. J. Anal. At.
43             Spectrom. 10: 253R-310R.
44       Baldwin, D. P.; Zamzow, D. S.; D'Silva, A. P. (1994) Aerosol mass  measurement and solution standard additions for
45             quantitation in laser ablation-inductively coupled plasma atomic emission spectrometry. Anal. Chem.
46             66: 1911-1917.
47       Ballach, J.; Hitzenberger, R.; Schultz, E.; Jaeschke, W. (2001) Development of an improved optical transmission
48             technique for black carbon (BC) analysis. Atmos. Environ. 35: 2089-2100.
49       Baron, P. A.; Willeke, K., eds. (2002) Aerosol measurement: principles,  techniques, and applications. 2nd ed.
5 0             New York, NY: John Wiley & Sons.
51       Benner, R. L.; Stedman, D. H. (1989) Universal sulfur detection by chemiluminescence. Anal.  Chem.
52             61: 1268-1271.
53       Benner, R. L.; Stedman, D. H. (1990) Field evaluation of the sulfur chemiluminescence detector. Environ. Sci.
54             Technol. 24: 1592-1596.
         June 2003                                      2-102        DRAFT-DO NOT QUOTE OR CITE

-------
  1       Bergin, M. H.; Ogren, J. A.; Schwartz, S. E.; Mclnnes, L. M. (1997) Evaporation of ammonium nitrate aerosol in a
  2             heated nephelometer: implications for field measurements. Environ. Sci. Technol. 31: 2878-2883.
  3       Bertoni, G.; Febo, A.; Perrino, C.; Possanzini, M. (1984) Annular active diffuse sampler: a new device for the
  4             collection of organic vapours. Ann. Chim. (Rome) 74: 97-104.
  5       Billiet, J.; Dams, R.; Hoste, J. (1980) Multielement thin film standards for XRF analysis. X-Ray Spectrom.
  6             9:206-211.
  7       Binkowski, F. S.; Ching, J. K. S. (1995) Modeling fine paniculate mass and visibility using the EPA regional
  8             paniculate model. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Research
  9             and Development, National Exposure Research Laboratory; report no. EPA/600/A-95/143. Available from:
10             NTIS, Springfield, VA; PB96-139100.
11       Binkowski, F. S.; Shankar, U. (1995) The regional paniculate matter model. I. Model description and preliminary
12             results. J. Geophys. Res.  [Atmos.] 100: 26,191-26,209.
13       Birch, M. E.  (1998) Analysis of carbonaceous aerosols: interlaboratory comparison. Analyst (Cambridge, UK)
14             123:851-857.
15       Birch, M. E.; Gary, R. A. (1996) Elemental carbon-based method for occupational monitoring of paniculate diesel
16             exhaust: methodology and exposure issues. Analyst 121: 1183-1190.
17       Birmili, W.; Wiedensohler, A. (1998) The influence of meteorological parameters on ultrafine particle production at
18             a continental site. J. Aerosol  Sci. 29(suppl. 1):  S1015-S1016.
19       Bowers, W. D.;  Chuan, R. L. (1989) Surface acoustic-wave piezoelectnc crystal aerosol mass monitor. Rev. Sci.
20             Instrum. 60: 1297-1302.
21       Brauer, M.; Koutrakis, P.;  Keeler, G. J.; Spengler, J. D.  (1991) Indoor and outdoor concentrations of inorganic acidic
22             aerosols and gases.  J. Air Waste Manage. Assoc. 41: 171-181.
23       Breum, N. O. (2000) The dust holding capacity of porous plastic foam used in particle size-selective sampling.
24             J. Aerosol Sci. 31:379-385.
25       Brook, J. R.; Dann, T. F. (1999) Contribution of nitrate and carbonaceous species to PM25 observed in Canadian
26             cities. J. Air Waste Manage.  Assoc. 49: 193-199.
27       Brorstrom, E.; Grennfelt, P.; Lindskog, A. (1983) The effect of nitrogen dioxide and ozone on the decomposition of
28             particle-associated polycyclic aromatic hydrocarbons durtng sampling from the atmosphere.  Atmos. Environ.
29             17:601-605.
30       Bukowiecki, N.; Kittelson, D. B.; Watts, W. F.; Burtscher, H.; Weingartner, E.; Baltensperger, U. (2002) Real-time
31             characterization of ultrafine and accumulation mode particles  in ambient combustion aerosols. J. Aerosol Sci.
32             33:  1139-1154.
33       Bunz, H.; Koyro, M.; Mohler, O.; Saathoff, H. (1996) Temperature dependence of the redistribution mechanisms for
34             volatile components of atmospheric aerosol particles. J. Aerosol Sci. 27(suppl. 1): S653-S654.
35       Cachier, H.; Bremond, M.-P.; Buat-Menard, P. (1989) Determination of atmospheric soot carbon with a simple
36             thermal method. Tellus 41B: 379-390.
37       Cadle, S. H.; Groblicki, P. J. (1982) An evaluation of methods for the determination of organic and elemental carbon
38             in paniculate samples. In: Wolff, G. T.; Klimisch, R. L., eds. Paniculate carbon: atmospheric life cycle. New
39             York, NY: Plenum Press; pp. 89-109.
40       Cadle, S. H.; Groblicki, P. J.; Mulawa, P. A. (1983) Problems in the  sampling and analysis of carbon particles.
41             Atmos. Environ. 17: 593-600.
42       Cadle, S. H.; Mulawa, P.A. (1990) Atmospheric carbonaceous species measurement methods comparison study:
43             General Motors results. Aerosol Sci. Technol.  12: 128-131.
44       Cahill, T. A. (1981) Innovative aerosol sampling devices based upon PIXE capabilities. In: Johansson, S. A. E., ed.
45             Particle induced x-ray emission and its analytical applications: proceedings of the 2nd international
46             conference on PIXE and its analytical applications; June, 1980; Lund, Sweden. Amsterdam,  The Netherlands:
47             North-Holland Publishing Co. Nucl. Instr. Meth. Phys. Res. 181: 473-480.
48       Cahill, T. A. (1985) Proton-induced X-ray emission (PIXE). In: Res, J. P., ed. Metals Handbook: Materials
49             Characterization. Cleveland,  OH: American Society for Metals.
50       Cahill, T. A. (1990) Particle-induced x-ray emission. In: Mills, K., ed. Metals Handbook; v.  10, p. 102. 9th ed.
51             Cleveland, OH: Amencan Society for Metals.
52       Cahill, T. A.; Eldred, R. A.; Motallebi, N.; Malm, W. C. (1989) Indirect measurement of hydrocarbon aerosols
53             across the United States by nonsulfate hydrogen-remaining gravimetric mass correlations. Aerosol Sci.
54             Technol.  10: 421-429.
55       Cahill, T. A.; Wilkinson, K.; Schnell, R. (1992) Composition analyses of size-resolved aerosol samples taken from
56             aircraft downwind of Kuwait, spring 1991. J. Geophys. Res. [Atmos.] 97: 14,513-14,520.
         June 2003                                      2-103         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Cahill, T. A.; Wakabayashi, P.; James, T. A. (1996) Chemical states of sulfate at Shenandoah National Park during
  2             summer, 1991. Nucl. lustrum Methods Phys. Res. B 109/110: 542-547.
  3       Camp, D. C.; Stevens, R. K.; Cobourn, W. G.; Husar, R. B.; Collins, J. F.; Huntzicker, J. J.; Husar, J. D.; Jaklevic,
  4             J. M.; McKenzie, R. L.; Tanner, R. L.; Tesch, J. W. (1982) Intercomparison of concentration results from fine
  5             particle sulfur monitors. Atmos. Environ. 16: 911-916.
  6       Campbell, A. J.; Humayun, M. (1999) Trace element microanalysis in iron meteorites by laser ablation ICPMS.
  7             Anal. Chem. 71:939-946.
  8       Carson, P. G.; Johnston, M. V.; Wexler, A.  S. (1997) Laser desorption/ionization of ultra fine aerosol particles.
  9             Rapid Commun. Mass Spectrom.  11: 993-996.
10       Castaldi, M. J.; Senkan, S. M. (1998) Real-time, ultrasensitive monitoring of air toxics by laser photoionization
11             time-of-flight mass spectroscopy. J. Air Waste Manage. Assoc. 48: 77-81.
12       Chandramouli, B.; Kamens, R. (2001) The photochemical formation and gas-particle partitioning of oxidation
13             products of decamethyl cyclopentasiloxane and decamethyl tetrasiloxane in the atmosphere. Atmos. Environ.
14             35:87-95.
15       Chandramouli, B.; Jang, M.; Kamens, R. M. (2003) Gas-particle partitioning of semi-volatile organics on organic
16             aerosols using a predictive activity coefficient model: analysis of the effects of parameter choices on model
17             performance. Atmos. Environ. 37: 853-864.
18       Chang, M.; Kim, S.; Sioutas, C.  (1999) Experimental studies on particle impaction and bounce: effects of substrate
19             design and material. Atmos. Environ. 33: 2313-2322.
20       Chang, M. C.; Sioutas, C.; Kim, S.; Gong, H., Jr.; Linn, W. S. (2000) Reduction of nitrate losses from filter and
21             impactor samplers by means of concentration enrichment. Atmos. Environ. 34: 85-98.
22       Chang, M.-C.; Geller, M. D.; Sioutas, C.; Fokkens, P. H. B.; Cassee, F. R. (2002) Development and evaluation of a
23             compact, highly efficient coarse particle concentrator for lexicological studies. Aerosol Sci. Technol.
24             36:492-501.
25       Chein, H.; Lundgren, D. A. (1993) A virtual impactor with clean air core for the generation of aerosols with narrow
26             size distributions. Aerosol Sci. Technol. 18: 376-388.
27       Chen, B. T. (1993) Instrument calibration. In: Willeke, K.; Baron, P. A., eds. Aerosol measurement: principles,
28             techniques,  and applications. New York, NY: Van Nostrand Reinhold; pp. 493-520.
29       Chen, B. T.; Yeh, H. C. (1987) An improved virtual impactor: design and performance. J. Aerosol Sci. 18:  203-214.
30       Chen, B. T.; Yeh, H. C.; Cheng, Y. S. (1985) A novel virtual impactor: calibration and use. J. Aerosol Sci.  16:
31             343-354.
32       Chen, B. T.; Yeh, H. C.; Cheng, Y. S. (1986) Performance of a modified virtual impactor. Aerosol Sci. Technol.
33             5:369-376.
34       Cheng, Y.-S.; Yeh, H.-C. (1979) Particle bounce in cascade impactors. Environ. Sci. Technol. 13: 1392-1396.
3 5       Chow, J. C. (1995) Measurement methods to determine compliance with ambient air quality standards for suspended
36             particles. J.  Air Waste Manage. Assoc. 45: 320-382.
37       Chow, J. C.; Watson, J. G. (1998) Guideline on speciated paniculate monitoring (Draft 3). Research Triangle Park,
3 8             NC: U.S. Environmental Protection Agency; Office of Air Quality Planning and Standards. Available:
3 9             http://www.epa.gov/ttn/amtic/files/ambient/pm25/spec/drispec.pdf (20 March 2002).
40       Chow, J. C.; Watson, J. G. (1999) Ion chromatography in elemental analysis of airborne particles. In: Landsberger,
41             S.; Creatchman, M., eds. Elemental Analysis of Airborne Particles. Newark, NJ:  Gordon and Breach Science
42             Publishers; pp. 97-137.
43       Chow, J. C.; Watson, J. G.; Pritchett, L.  C.; Pierson, W. R.; Frazier, C. A.; Purcell, R. G. (1993) The DRI
44             thermal/optical reflectance carbon analysis system: description, evaluation and applications in U.S. air quality
45             studies. In: Lodge, J. P., Jr. Fourth international  conference  on carbonaceous particles in the atmosphere;
46             April 1991; Vienna, Austria. Atmos.  Environ. Part A 27: 1185-1201.
47       Chow, J. C.; Fujita, E. M.; Watson, J. G.; Lu, Z.; Lawson, D. R.; Ashbaugh, L. L. (1994a) Evaluation of filter-based
48             aerosol measurements during the  1987 Southern California Air Quality Study. Environ. Monit. Assess.
49             30:49-80.
50       Chow, J. C.; Watson, J. G.; Fujita, E. M.; Lu, Z.; Lawson, D. R.; Ashbaugh, L. L. (1994b)  Temporal and spatial
51             variations of PM2 5 and PM10 aerosol in the Southern California Air Quality Study. Atmos. Environ.
52             28:2061-2080.
53       Chow, J. C.; Watson, J. G.; Lu, Z.; Lowenthal, D. H.; Frazier, C. A.; Solomon, P. A.; Thuillier, R. H.; Magliano, K.
54             (1996) Descriptive analysis of PM2.5 and PM10 at regionally representative locations during
55             SJVAQS/AUSPEX. In: Parrish, D.; Trainer, M.; Rao, S. T.; Solomon, P. A., eds. A&WMA international
56             specialty conference on regional photochemical  measurements and modeling, part 2; November 1993;
57             San Diego,  CA.  Atmos. Environ.  30: 2079-2112.


         June 2003                                     2-104         DRAFT-DO  NOT QUOTE OR CITE

-------
  1       Chow, J. C; Zielinska, B.; Watson, J. G.; Fujita, E. M; Richards, H. W.; Neff, W. D.; Dietrich, D.; Hering, S. V.
  2             (1998) Northern Front Range Air Quality Study. Volume A: Ambient measurements. Fort Collins, CO:
  3             Colorado State University, Cooperative Institute for Research in the Atmosphere. Available:
  4             http://www.nfraqs.colostate.edu/dri.html [4 February 2002].
  5       Chow, J. C.; Watson, J.  G.; Crow, D.; Lowenthal, D. H.; Merrifield, T. (2000) Comparison of IMPROVE and
  6             NIOSH carbon measurements. Aerosol Sci. Technol. 34: 23-34.
  7       Chow, J. C.; Watson, J. G.; Crow, D.; Lowenthal, D. H.; Merrifield, T.  (2001) Comparison of IMPROVE and
  8             NIOSH carbon measurements. Aerosol Sci. Technol. 34: 23-34.
  9       Claiborn, C. S.; Finn, D.; Larson, T. V.; Koenig, J. Q. (2000) Windblown dust contributes to high PM2 s
10             concentrations. J. Air Waste Manage. Assoc. 50: 1440-1445.
11       Clark, W. E.; Whitby, K. T. (1975) Measurements of aerosols produced by the photochemical oxidation of SO2 in
12             air. J. Colloid Interface Sci. 51: 477-490.
13       Clarke, A. D. (1992) Atmospheric nuclei in the remote free-troposphere. J. Atmos. Chem. 14:  479-488.
14       Clarke, A. D.; Varner, J. L.; Eisele,  F.; Mauldin, R. L.; Tanner, D.; Litchy, M. (1998) Particle  production in the
15             remote marine atmosphere: cloud outflow and subsidence during ACE 1. J. Geophys. Res. A 103:
16             16,397-16,409.
17       Cobourn, W. G.; Husar, R. B.; Husar, J. D. (1978) Continuous in situ monitoring of ambient paniculate sulfur using
18             flame photometry and thermal analysis. In: Husar, R. B.; Lodge, J. P., Jr.; Moore, D. J., eds. Sulfur in the
19             atmosphere: proceedings of the international symposium; September 1977; Dubrovnik,  Yugoslavia. Atmos.
20             Environ. 12: 89-98.
21       Cobourn, W. G.; Husar, R. B.  (1982) Diurnal and seasonal patterns of paniculate sulfur and sulfuric acid in St.
22             Louis, July 1977-June 1978.  Atmos. Environ. 16: 1441-1450.
23       Code of Federal Regulations. (200 la) Appendix J to Part 50-reference method for the determination of paniculate
24             matter as PM10 in the atmosphere. C. F. R. 40: pt. 50, app. J.
25       Code of Federal Regulations. (200 Ib) Ambient air monitoring reference and equivalent methods. C. F. R. 40: §53.
26       Code of Federal Regulations. (200 Ic) Appendix L to Part 50-reference method for the determination of fine
27             paniculate matter as PM2 5 in the atmosphere. C. F. R. 40: pt. 50, app. L.
28       Code of Federal Regulations. (2001d) Ambient air quality surveillance. C. F. R. 40: §58.
29       Conner, W. D. (1966) An inertial-type particle separator for collecting large samples. J. Air Pollut. Control Assoc.
30             16:35-38.
31       Conner, T. L.; Norris, G. A.; Landis, M. S.; Williams, R. W. (2001) Individual particle analysis of indoor, outdoor,
32             and community samples from the 1998 Baltimore paniculate matter study. Atmos. Environ. 35: 3935-3946.
33       Gotham, W. E.; Bidleman, T. F. (1992) Laboratory investigations of the partitioning of organochlorine compounds
34             between the gas phase and atmospheric aerosols on glass fiber filters. Environ. Sci. Technol. 26: 469-478.
35       Countess, R. J. (1990) Interlaboratory analyses of carbonaceous aerosol samples. Aerosol Sci.  Technol. 12:  114-121.
36       Courtney, W. J.; Shaw, R.  W.; Dzubay, T. G. (1982) Precision and accuracy of a p gauge for aerosol  mass
37             determinations. Environ. Sci. Technol. 16: 236-239.
38       Coutant, R.  W.; Callahan, P. J.; Chuang, J. C.; Lewis, R. G. (1992) Efficiency of silicone-grease-coated denuders for
39             collection of polynuclear aromatic hydrocarbons. Atmos. Environ. Part A 26: 2831-2834.
40       Covert, D. S.; Kapustin, V. N.; Quinn, P. K.; Bates, T. S. (1992) New particle formation in the marine boundary
41             layer. J. Geophys. Res.  [Atmos.] 97: 20,581-20,589.
42       Covert, D. S.; Wiedensohler, A.; Aalto, P.; Heintzenberg, J.; McMurry, P. H.; Leek, C. (1996) Aerosol number size
43             distributions from 3 to 500 nm diameter in the arctic marine boundary layer during summer and autumn.
44             Tellus 48B: 197-212.
45       Cui, W.; Machir, J.; Lewis , L.; Eatough, D. J.; Eatough, N. L. (1997) Fine paniculate organic material at Meadview
46             during the project MOHAVE summer intensive study. J. Air Waste Manage. Assoc. 47: 357-369.
47       Cui, W.; Eatough, D. J.; Eatough, N. L. (1998) Fine paniculate organic material  in the Los Angeles Basin -1:
48             assessment of the high-volume Brigham Young University organic sampling system, BIG BOSS. J. Air
49             Waste Manage. Assoc.  48: 1024-1037.
50       Dagnall, R.  M.; Thompson, K. C.; West, T.  S. (1967) Molecular-emission in cool flames. Part I. the behaviour of
51             sulphur species in a hydrogen-nitrogen diffusion flame and in a shielded air-hydrogen flame. Analyst
52             92:506-512.
53       Dams, R.; Robbins,  J. A.; Rahn, K.  A.; Winchester, J. W. (1970) Nondestructive neutron activation analysis of air
54             pollution particulates. Anal. Chem. 42: 861-867.
55       Dane, A. D.; Timmermans, P.  A. M.; Van Sprang, H. A.; Buydens, L. M. C. (1996) A genetic  algorithm for
56             model-free x-ray fluorescence analysis of thin film. Anal. Chem. 68: 2419-2425.
         June 2003                                      2-105         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Davis, D. W.; Reynolds, R. L.; Tsou, G. C.; Zafonte, L. (1977) Filter attenuation corrections for x-ray fluorescence
  2             analysis of atmospheric aerosols. Anal. Chem. 49: 1990-1993.
  3       Davis, C. S.; Fellin, P.; Otson, R. (1987) A review of sampling methods for polyaromatic hydrocarbons in air.
  4             JAPCA 37: 1397-1408.
  5       DeBoer, D. K. G.; Leenaers, A. J. G.; Van Den Hoogenhof, W. W. (1995) Glancing-incidence x-ray analysis of
  6             thin-layered materials: a review. X-Ray Spectrom. 24: 91-102.
  7       Ding, Y.; Lee, M. L.; Eatough, D. J. (1998a) The determination of total nitrite and TV-nitroso compounds in
  8             atmospheric samples. Int. J. Environ. Anal. Chem. 69: 243-255.
  9       Ding, Y.; Cui, W.; Lee, M. L.; Eatough, D. J. (1998b) Fine paniculate 7V-nitroso and nitrite organic compounds in
10             the atmosphere. Appl. Occup. Environ. Hyg. 13: 432-438.
11       Ding, Y.; Koutrakis, P. (2000) Development of a dichotomous slit nozzle virtual impactor. J. Aerosol Sci.
12             31: 1421-1431.
13       Ding, Y.; Pang, Y.; Eatough, D. J.; Eatough, N. L.; Tanner, R. L. (2002) High volume diffusion denuder sampler for
14             the routine monitoring of fine paniculate matter: II. field evaluation of the PC BOSS. Aerosol Sci. Technol.:
15             36:383-396.
16       Dockery, D. W.; Pope, C. A., Ill; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. E.; Ferris, B. G., Jr.; Speizer, F. E.
17             (1993) An association between air pollution and mortality in six U.S. cities. N. Engl. J. Med. 329: 1753-1759.
18       Durham, J. L.; Wilson, W. E.; Bailey, E. B. (1978) Application of an SO2-denuder for continuous measurement of
19             sulfur in submicrometric aerosols. Atmos.  Environ. 12: 883-886.
20       Dzubay, T. G. (1980) Chemical element balance method applied to dichotomous sampler data. Ann. N. Y. Acad.
21             Sci. 338: 126-144.
22       Dzubay, T. G.; Stevens, R. K. (1975) Ambient air analysis with dichotomous sampler and X-ray fluorescence
23             spectrometer. Environ. Sci. Technol. 9: 663-668.
24       Dzubay, T. G.; Hines, L. E.; Stevens, R. K. (1976) Particle bounce errors in cascade impactors. Atmos. Environ.
25             10:229-234.
26       Dzubay, T. G.; Stevens, R. K.; Peterson, C. M. (1977) Application of the dichotomous sampler to the
27             characterization of ambient aerosols. In: Dzubay, T. G., ed. X-ray fluorescence analysis of environmental
28             samples. Ann Arbor, MI: Ann Arbor Science Publishers, Inc.
29       Easter, R. C.; Peters, L. K. (1994) Binary homogeneous nucleation: temperature and relative humidity fluctuations,
30             nonlinearity, and aspects of new particle production in the atmosphere. J. Appl. Meteorol. 33: 775-784.
31       Eatough, D. J. (1999) BOSS, the 5righam Young University Organic Sampling System: determination of paniculate
32             carbonaceous material using diffision denuder sampling technology. In: Lane, D., ed. Gas and particle phase
33             measurements of atmospheric organic compounds. New York, NY: Gordon and Breach Science Publishers.
34             (Advances in environmental, industrial and process control technologies: v. 2).
35       Eatough, D. J.; Wadsworth, A.; Eatough, D. A.; Crawford, J. W.; Hansen, L. D.; Lewis, E. A. (1993) A
3 6             multiple-system, multi-channel diffusion denuder sampler for the determination of fme-particulate organic
37             material in the atmosphere. Atmos. Environ. Part A 27: 1213-1219.
38       Eatough, D. J.; Tang, H.; Cui, W.; Machir, J. (1995) Determination of the size distribution and chemical composition
39             of fine paniculate semi-volatile organic material in urban environments using diffusion denuder technology.
40             In: Phalen, R. F.; Bates, D. V., eds. Proceedings of the colloquium on paniculate air pollution and human
41             mortality and morbidity, part II;  January 1994; Irvine, CA. Inhalation Toxicol. 7: 691-710.
42       Eatough, D. J.; Eatough, D. A.; Lewis,  L.; Lewis, E. A. (1996) Fine paniculate chemical composition and light
43             extinction  at Canyonlands  National Park using organic paniculate material concentrations obtained with a
44             multisystem, multichannel diffusion denuder sampler. J. Geophys. Res. [Atmos.] 101: 19,515-19,531.
45       Eatough, D. J.; Obeidi, F.; Pang, Y.; Ding, Y.; Eatough, N. L.; Wilson, W. E. (1999a) Integrated and real-time
46             diffusion denuder sampler for PM2 5. Atmos. Environ. 33: 2835-2844.
47       Eatough, D. J.; Pang, Y.; Eatough, N. L. (1999b)  Determination of PM2 s sulfate and nitrate with a PC-BOSS
48             designed for routine sampling for semi-volatile paniculate matter. J. Air Waste Manage. Assoc. 49(special
49             issue): PM69-75.
50       Eatough, D. J.; Eatough, N. L.; Obeidi, F.; Pang,  Y.; Modey, W.; Long, R. (2001) Continuous determination of
51             PM2 5 mass, including semi-volatile species. Aerosol Sci. Technol. 34: 1-8.
52       Fairchild, C. L; Wheat, L. D. (1984) Calibration and evaluation of a real-time cascade impactor. Am. Ind. Hyg.
53             Assoc. J. 45:205-211.
54       Falconer, R. L.; Bidleman, T. F.;  Gotham, W. E. (1995) Preferential sorption of non- and
55             mono-ortho-polychlorinated biphenyls to urban aerosols. Environ. Sci. Technol. 29:  1666-1673.
         June 2003                                      2-106         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Fang, C. P.; McMurry, P. H.; Marple, V. A.; Rubow, K. L. (1991) Effect of flow-induced relative humidity changes
  2             on size cuts for sulfuric acid droplets in the microorifice uniform deposit impactor (MOUDI). Aerosol Sci.
  3             Technol. 14: 266-277.
  4       Farwell, S. O.; Rasmussen, R. A. (1976) Limitations of the FPD and BCD in atmospheric analysis: a review. J.
  5             Chromatogr. Sci. 14: 224-234.
  6       Fassel, V. A.; Kniseley, R. N. (1974) Inductively coupled plasma-optical emission spectroscopy. Anal. Chem.
  7             46: 1110A-1120A.
  8       Federal Register.  (1987) Revisions to the national ambient air quality standards for paniculate matter. F. R. (July 1)
  9             52:24,634-24,669.
10       Federal Register.  (1997) National ambient air quality standards for paniculate matter; final rule. F. R.  (July 18)
11             62:38,652-38,752.
12       Federal Register.  (1998) National ambient air quality standards for paniculate matter and revised requirements for
13             designation of reference and equivalent methods for PM2 5 and ambient air quality surveillance  for paniculate
14             matter; correction. F. R. (February 17) 63: 7710-7718.
15       Fergenson, D. P.; Song, X.-H.; Ramadan, Z.; Allen, J. O.; Hughes, L. S.; Cass, G. R.; Hopke, P. K.; Prather, K. A.
16             (2001) Quantification of ATOFMS  data by multivariate methods. Anal. Chem. 73: 3535-3541.
17       Fernandez, F. J. (1989) Atomic absorption spectroscopy. In: Lodge, J. P., Jr., ed. Methods of Air Sampling and
18             Analysis. 3rd ed. Chelsea, MI: Lewis Publishers, Inc.; pp.  83-89.
19       Finn, D.; Rumburg, B.; Claiborn, C.; Bamesberger, L.; Siems, W. F.; Koenig, J.; Larson, T.; Norris, G. (2001)
20             Sampling artifacts from the use of denuder tubes with glycerol based coatings in the measurement of
21             atmospheric paniculate matter. Environ. Sci. Technol. 35:  40-44.
22       Fitz, D. R. (1990) Reduction of the positive organic artifact on quartz filters. Aerosol. Sci. Technol. 12: 142-148.
23       Flocchini, R. G.; Cahill, T. A.; Shadoan, D. J.; Lange, S. J.; Eldred, R. A.; Feeney, P. J.; Wolfe, G. W.; Simmeroth,
24             D. C.; Suder, J. K. (1976) Monitoring California's aerosols by size and elemental composition.  Environ. Sci.
25             Technol. 10: 76-82.
26       Forney, L. J.; Ravenhall, D. G.; Winn, D. S. (1978) Aerosol impactors: a study of a fluid jet impinging upon a void.
27             J. Appl. Phys. 49: 2339-2345.
28       Forney, L. J.; Ravenhall, D. G.; Lee, S. S.  (1982) Experimental and theoretical study of a two-dimensional virtual
29             impactor. Environ. Sci. Technol. 16: 492-497.
30       Friedlander, S. K. (1970) The characterization of aerosols distributed with respect to size and chemical composition.
31             J. Aerosol  Sci. 1:295-307.
32       Friedlander, S. K. (1977) Smoke, dust and haze: fundamentals of aerosol behavior. New York, NY: John Wiley &
33             Sons.
34       Friedlander, S. K.; Yeh, E. K. (1998) The submicron atmospheric aerosol as a carrier of reactive chemical species:
35             case of peroxides. Appl. Occup. Environ. Hyg.  13: 416-420.
36       Friedlander, S. K. (2000). Smoke, dust and haze: fundamentals of aerosol dynamics. 2nd ed. New York, NY:
3 7             Oxford University Press.
38       Fuchs, N. A. (1964) The mechanics of aerosols.  New York, NY: Pergamon Press.
39       Fuchs, N. A. (1989) The mechanics of aerosols.  New York, NY: Dover Publications.
40       Fung, K. (1990) Paniculate carbon speciationby MnO2 oxidation. Aerosol Sci. Technol. 12: 122-127.
41       Fung, K.; Chow, J. C.; Watson, J. G. (2002) Evaluation of OC/EC speciationby thermal manganese dioxide
42             oxidation and the IMPROVE method. J. Air Waste Manage. Assoc. 52: 1333-1341.
43       Gallorini, M. (1995) Trace element monitoring in atmospheric pollution processes by neutron activation analysis.
44             Microchem. J. 51: 127-137.
45       Card, E.; Mayer, J. E.; Morrical, B. D.; Dienes, T.; Fergenson, D. P.; Prather, K. A. (1997) Real-time  analysis of
46             individual  atmospheric aerosol particles: design and performance of a portable ATOFMS. Anal. Chem.
47             69:4083-4091.
48       Gauderman, W. J.; Mcconnell, R.; Gilliland, F.;  London, S.; Thomas, D.; Avol, E.; Vora, H.; Berhane, K.;
49             Rappaport, E. B.; Lurmann, F.; Margolis, H. G.; Peters, J.  (2000) Association between air pollution and lung
50             function growth in southern California children. Am. J. Respir. Crit. Care Med. 162:  1383-1390.
51       Ge, Z.; Wexler, A. S.; Johnston, M. V. (1998) Laser desorption/ionization of single ultrafine multicomponent
52             aerosols. Environ. Sci. Technol. 32: 3218-3223.
53       Geller, M. D.; Kim, S.; Misra, C.; Sioutas, C.; Olson, B. A.; Marple, V. A. (2002) Methodology for measuring
54             size-dependent chemical composition of ultrafine particles. Aerosol Sci. Technol. 36: 748-763.
55       Gong, H., Jr.; Linn, W. S.; Sioutas, C.; Terrell, S. L.; Clark, K. W.; Anderson, K. R.; Terrell, L. (2003) Controlled
56             exposures of healthy and asthmatic  volunteers to concentrated ambient fine particles  in Los Angeles.
57             Inhalation  Toxicol.  15: 305-325.


         June 2003                                       2-107         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Goss, K.-U.; Eisenreich, S. J. (1997) Sorption of volatile organic compounds to particles from a combustion source
  2             at different temperatures and relative humidities. Atmos. Environ. 31: 2827-2834.
  3       Gray, A. L.; Williams, J. G. (1987a) Oxide and doubly charged ion response of a commercial inductively coupled
  4             plasma mass spectrometry instrument. J. Anal. At. Spectrom. 2: 81-82.
  5       Gray, A. L.; Williams, J. G. (1987b) System optimisation and the effect on polyatomic, oxide and doubly charged
  6             ion response of a commercial inductively coupled plasma mass spectrometry instrument. J. Anal. At.
  7             Spectrom. 2: 599-606.
  8       Grosjean, D. (1983) Polycyclic aromatic hydrocarbons in Los Angeles air from samples collected on Teflon, glass
  9             and quartz filters. Atmos. Environ. 17: 2565-2573.
10       Grosjean, D.; Parmar, S. S. (1990) Interferences from aldehydes and peroxyacetyl nitrate when sampling urban air
11             organic acids on alkaline traps. Environ. Sci. Technol. 24: 1021-1026.
12       Grosjean, D.; Williams, E. L., II; Grosjean, E.; Novakov, T. (1994) Evolved gas analysis of secondary organic
13             aerosols. Aerosol Sci. Technol. 21: 306-324.
14       Gundel, L. A.; Lane, D. A. (1999) Sorbent-coated diffusion denuders for direct measurement of gas/particle
15             partitioning by semi-volatile organic compounds. In: Lane, D. A., ed. Gas and particle phase measurements of
16             atmospheric organic compounds. Amsterdam, The Netherlands: Gordon and Breach Science Publishers.
17             (Advances in environmental, industrial and process control technologies: v. 2).
18       Gundel, L. A.; Lee, V. C.; Mahanama, K. R. R.; Stevens, R. K.; Daisey, J. M. (1995) Direct determination of the
19             phase distributions of semi-volatile polycyclic aromatic hydrocarbons using annular denuders. Atmos.
20             Environ. 29: 1719-1733.
21       Hanninen, O. O.; Koistinen, K. J.; Kousa, A.; Keski-Karhu, J.; Jantunen, M. J. (2002) Quantitative analysis of
22             environmental factors in differential weighing of blank teflon filters. J. Air Waste Manage. Assoc.
23             52: 134-139.
24       Hansen, A.  D. A.; McMurry, P. H. (1990) An intercomparison of measurements of aerosol elemental carbon during
25             the 1986 carbonaceous species method comparison study. J.  Air Waste Manage. Assoc. 40: 894-895.
26       Hansen, A.  D. A.; Rosen, H.; Novakov, T. (1984) The aethalometer - an instrument for the real-time measurement of
27             optical absorption by aerosol particles. Sci. Total Environ. 36: 191-196.
28       Harman, J. N. (1989) Inductively coupled plasma emission spectroscopy. In: Lodge, J. P.,  Jr., ed. Methods of air
29             sampling and analysis. 3rd ed. Chelsea, MI: Lewis Publishers, Inc.; pp. 89-92.
30       Haupt, O.; Klane, B.; Schaefer, C.; Dannecker, W. (1995) Preparation of quartz fibre filter standards for x-ray
31             fluorescence analysis of aerosol samples. X-Ray Spectrom. 24: 267-275.
32       He, K.; Yang, F.; Ma, Y.; Zhang, Q.; Yao, X.; Chan, C. K.; Cadle, S.; Chan, T.; Mulawa, P. (2001) The
33             characteristics of PM25 in Beijing, China. Atmos. Environ. 35:  4959-4970.
34       Hegg, D. A.; Radke, L. F.; Hobbs, P. V. (1990) Particle production associated with marine clouds. J. Geophys. Res.
35             [Atmos.]  95: 13,917-13,926.
36       Hegg, D. A.; Radke, L. F.; Hobbs, P. V. (1991) Measurements of Aitken nuclei and cloud condensation nuclei in the
37             marine atmosphere and their relation to the DMS-cloud-climate hypothesis. J. Geophys. Res. [Atmos.]
38             96: 18,727-18,733.
39       Hering, S.; Cass, G. (1999) The magnitude of bias in the measurement of PM25 arising from volatilization of
40             paniculate nitrate from Teflon filters. J. Air Waste Manage. Assoc. 49: 725-733.
41       Hering, S. V.; Appel, B. R.; Cheng, W.; Salaymeh, F.;  Cadle, S. H.; Mulawa, P. A.; Cahill, T. A.; Eldred,  R. A.;
42             Surovik, M.; Fitz, D.; Howes, J. E.; Knapp, K. T.; Stockburger, L.; Turpin, B. J.; Huntzicker, J. J.; Zhang,
43             X.-Q.; McMurry, P. H. (1990) Comparison of sampling methods for carbonaceous aerosols in ambient air.
44             Aerosol Sci. Technol. 12: 200-213.
45       Hinds, W. C. (1999) Aerosol technology: properties, behavior, and  measurement of airborne particles. 2nd ed.
46             New York, NY: John Wiley & Sons.
47       Hinz, K.-P.; Kaufmann, R.; Spengler, B. (1994) Laser-induced mass analysis of single particles in the airborne state.
48             Anal. Chem. 66: 2071-2076.
49       Hitzenberger, R.; Dusek, U.; Berner, A. (1996) Black carbon measurements using an integrating sphere. J. Geophys.
50             Res.  [Atmos.] 101:  19,601-19,606.
51       Hitzenberger, R.; Berner, A.; Dusek, U.; Alabashi, R. (1997) Humidity-dependent growth of size-segregated aerosol
52             samples. Aerosol Sci. Technol. 27: 116-130.
53       Hitzenberger, R.; Jennings, S. G.; Larson, S. M.; Dillner, A.; Cachier, H.; Galambos, Z.; Roue, A.; Spain,  T. G.
54             (1999) Intercomparison of measurement methods for black carbon aerosols. Atmos. Environ. 33: 2823-2833.
55       Holyiiska, B.; Ptasinski, J.; Maenhaut, W.; Annegarn, H. J. (1997) Energy-dispersive x-ray fluorescence
56             spectrometer with capillary optics for the chemical analysis of atmospheric aerosols with high time resolution.
57             J. Aerosol Sci. 28: 1455-1463.


         June 2003                                      2-108         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Hoppel, W. A.; Frick, G. M; Fitzgerald, J. W.; Larson, R. E. (1994) Marine boundary layer measurements of new
  2             particle formation and the effects nonprecipitating clouds have on aerosol size distribution. J. Geophys. Res.
  3             [Atmos.] 99: 14,443-14,459.
  4       Houck, J. E.; Chow, J. C.; Watson, J. G.;  Simons, C. A.; Pritchett, L. C.; Goulet, J. M.; Frazier, C. A. (1989)
  5             Determination of particle size distribution and chemical composition of paniculate matter from selected
  6             sources in California: volume I and executive summary (final report). Sacramento, CA: California Air
  7             Resources Board.
  8       Hounam, R. F.; Sherwood, R. J. (1965) The cascade centripeter: a device for determining the concentration and size
  9             distribution of aerosols. Am. Ind. Hyg. Assoc. J. 26: 122-131.
10       Huntzicker, J. J.; Hoffman, R. S.; Ling, C.-S. (1978) Continuous measurement and speciation of sulfur-containing
11             aerosols by flame photometry. Atmos. Environ. 12: 83-88.
12       Huntzicker, J. J.; Johnson, R. L.; Shah, J. J.; Gary, R. A. (1982) Analysis of organic and elemental carbon in ambient
13             aerosols by a thermal-optical method. In: Wolff, G. T.; Klimisch, R. L., eds. Paniculate carbon - atmospheric
14             life cycle: proceedings of an international symposium; October 1980; Warren, MI. New York,  NY: Plenum
15             Press; pp. 79-88.
16       Husar, R. B.; Tratt, D. M.; Schichtel, B. A.; Falke, S. R.;  Li, F.; Jaffe, D.; Gasso, S.; Gill, T.; Laulainen, N. S.; Lu,
17             F.; Reheis, M. C.; Chun, Y.; Westphal, D.; Holben, B. N.; Gueymard, C.; McKendry, I.; Kuring, N.;
18             Feldman, G. C.; McClain, C.; Frouin, R. J.; Merrill, J.; DuBois, D.; Vignola, F.; Murayama,  T.; Nickovic, S.;
19             Wilson, W. E.; Sassen, K.; Sugimoto, N.; Malm, W. C. (2001) The Asian dust events of April  1998. J.
20             Geophys. Res. 106: 18,317-18,330.
21       Jackson, K. W.; Mahmood, T. M. (1994)  Atomic absorption, atomic emission, and flame emission spectrometry.
22             Anal. Chem. 66: 252R-279R.
23       Jaklevic, J. M.; Loo, B. W.; Goulding, F.  S. (1977) Photon-induced X-ray fluorescence analysis using
24             energy-dispersive detector and dichotomous sampler. In: Dzubay, T. G., ed. X-ray fluorescence analysis of
25             environmental samples. Ann Arbor, MI: Ann Arbor Science Publishers, Inc.; pp. 3-18.
26       Jaklevic, J. M.; Gatti, R. C.; Goulding, F. S.; Loo, B. W.  (1981a) A (3-gauge method applied to aerosol samples.
27             Environ. Sci. Technol.  15: 680-686.
28       Jaklevic, J. M.; Loo, B. W.; Fujita, T. Y.  (1981b) Automatic paniculate sulfur measurements with a dichotomous
29             sampler and on-line x-ray fluorescence analysis. Environ. Sci. Technol. 15: 687-690.
30       Jang, M.; Kamens, R. M. (1998) A thermodynamic approach for modeling partitioning of semivolatile organic
31             compounds on atmospheric paniculate matter: humidity effects. Environ. Sci. Technol. 32: 1237-1243.
32       Jang, M.; Kamens, R. M. (1999) Exploratory model  approach for adsorption partitioning of SOCs on  inorganic
33             particles. Environ. Sci. Technol.
34       Jang, M.; Kamens, R. M.; Leach, K. B.; Strommen, M. R. (1997) A thermodynamic approach using group
3 5             contribution methods to model the partitioning of semivolatile organic compounds on atmospheric paniculate
36             matter. Environ. Sci. Technol. 31:  2805-2811.
37       Jenkins, B. M.; Jones, A. D.; Turn, S. Q.; Williams, R. B. (1996) Particle concentrations, gas-particle partitioning,
3 8             and species intercorrelations for poly cyclic aromatic hydrocarbons (PAH) emitted during biomass burning.
39             Atmos. Environ. 30: 3825-3835.
40       Johansson, T. B.; Askelsson, R.; Johansson, S. A. E. (1970) X-ray analysis: elemental trace analysis at the 10"12g
41             level. Nucl. Instrum. Methods 84:  141-143.
42       John, W.; Sethi, V. (1993) Threshold for  resuspension by particle impaction. Aerosol  Sci. Technol. 19: 69-79.
43       John, W.; Wall, S. M.; Ondo,  J.  L. (1988) A new method for nitric acid and nitrate aerosol measurement using the
44             dichotomous sampler. Atmos. Environ. 22: 1627-1635.
45       Johnson, M V.; Wexler, A. S. (1995) MS of individual aerosol particles. Anal. Chem. 67: 721A-726A.
46       Johnson, R. L.; Shah, J. J.; Huntzicker, J. J. (1980) Analysis of organic, elemental, and carbonate carbon in ambient
47             aerosols. In: Verner,  S. S. Sampling and analysis of toxic organics in the  atmosphere: proceedings of a
48             symposium; August, 1979; Boulder, CO. Philadelphia, PA: American Society for Testing and Materials,
49             pp. 111-119. [ASTM Special Technical Publication 721].
50       Kamens, R. M.; Coe, D. L. (1997) A large gas-phase stripping device to investigate rates of PAH evaporation from
51             airborne diesel soot particles. Environ. Sci. Technol. 31: 1830-1833.
52       Kamens, R. M.; Jaoui, M. (2001) Modeling aerosol formation from a-pinene + NOX in the presence of natural
53             sunlight using gas phase kinetics and gas-particle partitioning theory. Environ.  Sci. Technol. 35: 1394-1405.
54       Kamens, R.; Odum, J.; Fan, Z.-H. (1995)  Some observations on times to equilibrium for semivolatile  polycyclic
55             aromatic hydrocarbons. Environ. Sci. Technol. 29: 43-50.
56       Kamens, R.; Jang, M.; Chien, C.-J.; Leach, K. (1999) Aerosol formation from the reaction of a-pinene and ozone
57             using a gas-phase kinetics-aerosol  partitioning model. Environ. Sci. Technol. 33: 1430-1438.


         June  2003                                      2-109         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Kao, A. S.; Friedlander, S. K. (1995) Frequency distributions of PM10 chemical components and their sources.
  2             Environ. Sci. Technol. 29: 19-28.
  3       Kavouras, I. G.; Koutrakis, P. (2001) Use of polyurethane foam as the impaction substrate/collection medium in
  4             conventional inertia! impactors. Aerosol Sci. Technol. 34: 46-56.
  5       Keeler, G. I; Brachaczek, W. W.; Gorse, R. A., Jr.; Japar, S. M; Pierson, W. R.  (1988) Effect of ambient humidity
  6             on dichotomous sampler coarse/fine ratios. Atmos. Environ. 22: 1715-1720.
  7       Keeler, G. I; Pirrone, N. (1996) Atmospheric transport and deposition of trace elements to Lake Erie from urban
  8             areas. Water Sci. Technol. 33: 259-265.
  9       Kegler, S. R.; Wilson, W. E.; Marcus, A. H. (2001) PM1, intermodal (PM2ri) mass, and the soil component of
10             PM25 in Phoenix, AZ,  1995-96. Aerosol Sci. Technol. 35: 914-920.
11       Keywood, M. D.; Ayers, G. P.; Gras, J. L.; Gillett, R. W.; Cohen, D. D. (1999) Relationships between size
12             segregated mass concentration data and ultrafine particle number concentrations in urban areas. Atmos.
13             Environ. 33:2907-2913.
14       Keywood, M. D.; Ayers, G. P.; Gras, J. L.; Gillett, R. W.; Cohen, D. D. (2000) Size distribution and sources of
15             aerosol in Launceston, Australia, during winter 1997. J. Air Waste Manage. Assoc. 50: 418-427.
16       Khlystov, A.; Wyers, G. P.; Slanina, J. (1995) The steam-jet aerosol collector. Atmos. Environ. 29: 2229-2234.
17       Kim, Y. J.; Boatman, J. F.; Gunter, R. L.; Wellman, D. L.; Wilkison, S. W. (1993) Vertical distribution of
18             atmospheric aerosol size distribution over south-central New Mexico. Atmos. Environ. Part A 27: 1351-1362.
19       Kim, B. M.; Lester, J.; Tisopulos, L.; Zeldin, M. D. (1999) Nitrate artifacts during PM25 sampling in the South Coast
20             Air Basin of California. J. Air Waste Manage. Assoc. 49(special issue): PM142-153.
21       Kim,B.M.; Cassmassi, I; Hogo, H.; Zeldin, M. D. (2001a) Positive organic carbon artifacts  on filter medium
22             during PM2 5 sampling in the South Coast Air Basin. Aerosol Sci. Technol. 34: 35-41.
23       Kim, S.; Jaques, P. A.; Chang, M.  C.; Froines, J. R.; Sioutas, C. (200Ib) Versatile aerosol concentration enrichment
24             system (VACES) for simultaneous in vivo and in vitro evaluation of toxic effects of ultrafine, fine and coarse
25             ambient particles Part I: development and laboratory characterization. J. Aerosol Sci. 32: 1281-1297.
26       Kim, S.; Jaques, P. A.; Chang, M.  C.; Barone, T.; Xiong, C.; Friedlander, S. K.; Sioutas, C. (2001c) Versatile
27             aerosol concentration enrichment system (VACES) for simultaneous in vivo and in vitro evaluation of toxic
28             effects of ultrafine, fine and coarse ambient particles Part II: field evaluation. J.  Aerosol Sci. 32: 1299-1314.
29       Kirchstetter, T. W.; Corrigan, C. E.; Novakov, T. (2001) Laboratory and field investigation of the adsorption of
30             gaseous organic compounds onto quartz filters. Atmos. Environ. 35: 1663-1671.
31       Kittelson, D. B. (1998) Engines and nanoparticles: a review. J. Aerosol Sci. 29: 575-588.
32       Kittelson, D. B.; McKenzie, R.; Vermeersch, M.; Dorman, F.; Pui, D.; Linne, M.; Liu, B.; Whitby, K. (1978) Total
3 3             sulfur aerosol concentration with an electrostatically pulsed flame photometric detector system. Atmos.
34             Environ. 12: 105-111.
35       Klouda, G. A.; Klinedinst, D. B.; Steel, E. B.; Benner, B. A., Jr.; Parish, H. J. (1996) Exploring a method to produce
36             an urban dust particle filter standard. J. Aerosol Sci. 27(suppl. 1): S351-S352.
37       Koistinen, K. J.; Kousa, A.; Tenhola, V.; Hanninen, O.; Jantunen, M. J.; Oglesby, L.; Kuenzli, N.; Georgoulis, L.
38             (1999) Fine particle (PM2 5) measurement methodology, quality assurance procedures and pilot results of the
39             EXPOLIS study. J. Air Waste Manage. Assoc. 49: 1212-1220.
40       Korhonen, P.; Kulmala, M.; Laaksonen, A.; Viisanen, Y.; McGraw, R.; Seinfeld, J. H. (1999) Ternary nucleation of
41             H2SO4, NH3, andH2O in the atmosphere. J. Geophys. Res. 104: 26,349-26,353.
42       Koutrakis, P.; Kelly, B. P. (1993) Equilibrium size of atmospheric aerosol sulfates as a function of particle acidity
43             and ambient relative humidity.  J. Geophys. Res. [Atmos.]  98: 7141-7147.
44       Koutrakis, P.; Wolfson, J. M.; Slater, J. L.; Brauer, M.; Spengler, J. D.; Stevens, R. K.; Stone, C. L. (1988a)
45             Evaluation of an annular denuder/filter pack system to collect acidic aerosols and gases. Environ. Sci.
46             Technol. 22: 1463-1468.
47       Koutrakis, P.; Wolfson, J. M.; Spengler, J. D. (1988b) An improved method for measuring aerosol strong acidity:
48             results from a nine-month study in St. Louis, Missouri and Kingston, Tennessee. Atmos. Environ.
49             22: 157-162.
50       Koutrakis, P.; Wolfson, J. M.; Spengler, J. D.; Stern, B.; Franklin, C. A. (1989) Equilibrium size of atmospheric
51             aerosol sulfates as a function of the relative humidity. J. Geophys. Res. [Atmos.] 94: 6442-6448.
52       Koutrakis, P.; Thompson, K. M.; Wolfson, J. M.; Spengler, J. D.; Keeler, G. J.; Slater, J. L. (1992) Determination of
53             aerosol strong acidity losses due to interactions of collected particles: results from laboratory and field
54             studies. Atmos. Environ. Part A 26: 987-995.
55       Koutrakis, P.; Sioutas, C.; Ferguson, S. T.; Wolfson, J. M.; Mulik, J. D.; Burton, R. M. (1993) Development and
56             evaluation of a glass honeycomb denuder/filter pack system to collect atmospheric gases and particles.
57             Environ. Sci. Technol. 27: 2497-2501.


         June 2003                                       2-110         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Koutrakis, P.; Wang, P.-Y.; Wolfson, J. M.; Sioutas, C., inventors. (1996) Method and apparatus to measure
  2             paniculate matter in gas. U.S. patent 5,571,945. November 5. Available: www.patents.ibm.com [1999,
  3             November 24].
  4       Krieger, M. S.; Hites, R. A. (1992) Diffusion denuder for the collection of semivolatile organic compounds. Environ.
  5             Sci. Technol. 26: 1551-1555.
  6       Kulmala, M.; Toivonen, A.; Makela, J. M.; Laaksonen, A. (1998) Analysis of the growth of nucleation mode
  7             particles observed in Boreal forest. Tellus 50B: 449-462.
  8       Kulmala, M.; Pirjola, L.; Makela, J. M. (2000) Stable sulphate clusters as a source of new atmospheric particles.
  9             Nature (London, U. K.) 404: 66-69.
10       Kusko, B. H.; Cahill, T. A.; Eldred, R. A.; Matsuda, Y.; Miyake, H. (1989) Nondestructive analysis of total
11             nonvolatile carbon by Forward Alpha Scattering Technique (FAST). Aerosol Sci. Technol. 10: 390-396.
12       Lammel, G.; Novakov, T. (1995) Water nucleation properties of carbon black and diesel soot particles. Atmos.
13             Environ. 29: 813-823.
14       Landsberger, S. (1988) Improved methodology for the determination of the seven elemental tracer long-distance
15             pollution signatures using thermal and epithermal neutron activation analysis. Anal. Chern. 60: 1842-1845.
16       Landsberger, S.; Wu, D. (1993) Improvement of analytical sensitivities for the determination of antimony, arsenic,
17             cadmium, indium, iodine, molybdenum, silicon and uranium in airborne paniculate matter by epithermal
18             neutron activation analysis. J. Radioanal. Nucl. Chem. 167: 219-225.
19       Landsberger, S.; Wu, D.; Vermette, S. J.;  Cizek, W. (1997) Intercompanson of IAEA airborne paniculate matter
20             reference material. J. Radioanal. Nucl. Chem. 215: 117-127.
21       Lane, D. A.; Johnson, N. D.; Barton, S. C.; Thomas, G. H. S.; Schroeder, W. H.  (1988) Development and evaluation
22             of a novel gas and particle sampler for semivolatile chlorinated organic compounds in ambient air. Environ.
23             Sci. Technol. 22: 941-947.
24       Lavanchy, V. M. H.; Gaggeler, H. W.; Nyeki, S.; Baltensperger, U. (1999) Elemental carbon (EC) and black carbon
25             (BC) measurements with a thermal method and an aethalometer at the high-alpine research station
26             Jungfraujoch. Atmos. Environ. 33: 2759-2769.
27       Lawless, P. A.; Rodes,  C. E.; Evans, G.; Sheldon, L.; Creason, J. (2001) Aerosol concentrations during the 1999
28             Fresno Exposure Studies as functions of size, season and meteorology. Aerosol Sci. Technol. 34: 66-74.
29       Lawrence, J. E.; Koutrakis, P. (1994) Measurement of atmospheric formic and acetic acids: methods evaluation and
30             results from field studies. Environ. Sci. Technol. 28: 957-964.
31       Lawrence, J.; Koutrakis, P. (1996a) Measurement and speciation of gas and paniculate phase organic acidity in an
32             urban environment: 1. analytical. J. Geophys. Res. [Atmos.] 101: 9159-9169.
33       Lawrence, J.; Koutrakis, P. (1996b) Measurement and speciation of gas and paniculate phase organic acidity in an
34             urban environment: 2. speciation. J. Geophys. Res. [Atmos.] 101: 9171-9184.
3 5       Lawson, D. R. (1990) The Southern California Air Quality Study. J. Air Waste Manage. Assoc. 40: 156-165.
36       Leach, K. B.; Kamens,  R. M.; Strommen, M. R.; Jang, M. (1999) Partitioning of semivolatile organic compounds in
37             the presence of a secondary organic aerosol in a controlled atmosphere. J. Atmos. Chem. 33: 241-264.
38       Lee, W.-M. G.; Shih, P.-M.; Wang, C.-S.  (1997) The influence of relative humidity on the size of atmospheric
39             aerosol. J. Environ. Sci. Health Part A: Environ. Sci. Eng. Toxic Hazard.  Subst. Control 32: 1085-1097.
40       Lewis, C. W.; Macias, E. S. (1980) Composition of size-fractionated aerosol  in Charleston, West Virginia. Atmos.
41             Environ. 14: 185-194.
42       Lewtas, J.; Pang, Y.; Booth, D.; Reimer, S.; Eatough, D. J.; Gundel, L. A. (2001) Compartson of sampling methods
43             for semi-volatile organic carbon associated with PM2 5. Aerosol Sci. Technol. 34: 9-22.
44       Li, W.; Montassier, N.; Hopke, P. K. (1992) A system to measure the hygroscopicity of aerosol particles. Aerosol
45             Sci. Technol. 17: 25-35.
46       Li, N.; Kim, S.; Wang,  M.; Froines, J. R.; Sioutas, C.; Nel, A. (2002) Use of a stratified oxidative stress model to
47             study the biological effects of ambient concentrated and diesel exhaust paniculate matter. Inhalation Toxicol.
48             14:459-486.
49       Li, N.; Sioutas, C.; Froines, J. R.; Cho,  A.; Misra, C.; Nel, A. (2003) Ultrafine paniculate pollutants induce oxidative
50             stress and mitochondrtal damage. Environ. Health Perspect. Ill: 455-460.
51       Liang, Z.; Chan, C. K. (1997) A fast technique for measurtng water activity of atmosphertc aerosols. Aerosol Sci.
52             Technol. 26: 255-268.
53       Liang, C.; Pankow, J. F.; Odum, J. R.; Seinfeld, J. H. (1997) Gas/particle partitioning of semivolatile organic
54             compounds to model inorganic, organic, and ambient smog aerosols. Environ. Sci. Technol. 31: 3086-3092.
55       Lin, J. J.; Noll, K. E.; Holsen, T. M. (1994) Dry deposition velocities as a function of particle size in the ambient
56             atmosphere. Aerosol Sci.  Technol.  20: 239-252.
         June 2003                                      2-111         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Lindskog, A.; Brorstrom-Lunden, E.; Alfheim, I; Hagen, I. (1987) Chemical transformation of PAH on airborne
  2             particles by exposure to NO2 during sampling: a comparison between two filter media.  Sci. Total Environ.
  3             61:51-57.
  4       Liousse, C.; Cachier, H.; Jennings, S. G. (1993)  Optical and thermal measurements of black carbon aerosol content
  5             in different environments: variation of the specific attenuation cross-section, sigma o. Atmos. Environ. Part A
  6             27: 1203-1211.
  7       Long, R. W.; Smith, R.; Smith, S.; Eatough, N. L.; Mangelson, N. F.; Eatough, D. I; Pope, C. A., Ill; Wilson, W. E.
  8             (2002) Sources of fine paniculate material along the Wasatch Front. Ener. Fuels 16: 282-293.
  9       Loo, B. W.; Cork, C. P. (1988) Development of high efficiency virtual impactors. Aerosol Sci. Technol. 9: 167-176.
10       Loo, B. W.; Jaklevic, J. M; Goulding, F. S. (1976) Dichotomous virtual impactors for large scale monitoring of
11             airborne paniculate matter. In: Liu, B. Y. H., ed. Fine particles: aerosol generation, measurement, sampling,
12             and analysis. New York, NY: Academic Press; pp. 311-350.
13       Loo, B. W.; French, W. R.; Gatti, R. C.; Goulding, F. S.; Jaklevic, J. M.; Llacer, J.; Thompson, A. C. (1978)
14             Large-scale measurement of airborne paniculate sulfur. Atmos. Environ. 12: 759-771.
15       Lundgren, D. A.; Burton, R. M. (1995) Effect of particle size distribution on the cut point between fine and coarse
16             ambient mass fractions. In: Phalen, R. F.; Bates, D. V., eds. Proceedings of the colloquium on paniculate air
17             pollution and human mortality and morbidity; January 1994; Irvine, CA. Inhalation Toxicol. 7: 131-148.
18       Lundgren, D. A.; Hausknecht, B. J.; Burton, R. M. (1984) Large particle size distribution in five U.S. cities and the
19             effect on a new ambient paniculate matter standard (PM10). Aerosol Sci. Technol. 7: 467-473.
20       Lynch, A. J.; McQuaker, N. R.; Brown, D. F. (1980) ICP/AES analysis and the composition of airborne and soil
21             materials in the vicinity of a lead/zinc smelter complex. J. Air Pollut. Control Assoc.  30: 257-260.
22       Mader, B. T.; Flagan, R. C.; Seinfeld, J. H. (2001) Sampling atmospheric carbonaceous aerosols using a particle trap
23             impactor/denuder sampler. Environ. Sci. Technol. 35: 4857-4867.
24       Mahanama, K. R. R.; Gundel, L. A.; Daisey, J. M. (1994) Selective fluorescence detection of poly cyclic aromatic
25             hydrocarbons in environmental tobacco smoke and other airborne particles. Int. J. Environ. Anal. Chem.
26             56:289-309.
27       Makela, J. M.; Aalto, P.; Jokinen, V.; Pohja,  T.;  Nissinen, A.; Palmroth, S.; Markkanen, T.;  Seitsonen, K.;
28             Lihavainen, H.; Kulmala, M. (1997) Observations of ultrafme aerosol particle  formation and growth in boreal
29             forest. Geophys. Res. Lett. 24: 1219-1222.
30       Malm, W. C.; Sisler, J. F.; Huffman, D.; Eldred, R. A.; Cahill, T. A. (1994) Spatial and seasonal trends in particle
31             concentration and optical extinction in the United States. J. Geophys. Res. [Atmos.] 99: 1347-1370.
32       Mamane, Y.; Willis, R.; Conner, T.  (2001) Evaluation of computer-controlled scanning electron microscopy applied
33             to an ambient urban aerosol sample. Aerosol Sci. Technol. 34: 97-107.
34       Marple,  V. A.; Chien, C. M. (1980)  Virtual impactors: a theoretical study. Environ. Sci. Technol. 14: 976-985.
35       Marple,  V. A.; Olson, B. A. (1995) A high volume PM10/PM2J/PMLO trichotomous sampler. In: Paniculate matter:
36             health and regulatory issues: proceedings of an international specialty conference; April; Pittsburgh, PA.
37             Pittsburgh, PA: Air & Waste Management Association; pp. 237-261. (A&WMA publication VIP-49).
38       Marple,  V. A.; Liu, B. Y. H.; Burton, R. M. (1990) High-volume impactorfor sampling fine and coarse particles. J.
39             Air Waste Manage. Assoc. 40: 762-767.
40       Marple,  V. A.; Rubow, K.  L.;  Behm, S. M. (1991) A microorifice uniform deposit impactor (MOUDI): description,
41             calibration, and use. Aerosol  Sci. Technol. 14: 434-446.
42       Masuda, H.; Nakasita, S. (1988) Classification performance of a rectangular jet virtual impactor-effects of nozzle
43             width ratio of collection nozzle to acceleration nozzle. J. Aerosol Sci. 19: 243-252.
44       Masuda, H.; Hochrainer, D.; Stober, W. (1979) An improved virtual impactor for particle classification and
45             generation of test aerosols with narrow size distributions. J. Aerosol Sci. 10: 275-287.
46       McGovern,  F. M. (1999) An analysis of condensation nuclei levels at Mace Head,  Ireland. Atmos. Environ.
47             33: 1711-1723.
48       McGovern,  F. M.; Jennings, S. G.; O'Connor, T. C.; Simmonds, P. G.  (1996) Aerosol and trace gas measurements
49             during the Mace Head experiment. Atmos. Environ. 30: 3891-3902.
50       Mclnnes, L. M.; Quinn, P. K.; Covert, D. S.; Anderson, T. L. (1996) Gravimetric analysis, ionic composition, and
51             associated water mass of the marine aerosol. Atmos. Environ. 30: 869-884.
52       McMurry, P. H. (2000) A review of atmospheric aerosol measurements. Atmos. Environ. 34: 1959-1999.
53       McMurry, P. H.; Friedlander,  S. K. (1979) New  particle formation in the presence of an aerosol. Atmos. Environ.
54             13: 1635-1651.
55       McMurry, P. H.; Stolzenburg, M. R. (1989) On the sensitivity of particle  size to relative humidity for Los Angeles
56             aerosols. Atmos. Environ. 23: 497-507.
         June 2003                                      2-112         DRAFT-DO NOT QUOTE OR CITE

-------
  1       McMurry, P. H.; Woo, K. S.; Weber, R.; Chen, D.-R.; Pui, D. Y. H. (2000) Size distributions of 3 to 10 nm
  2             atmospheric particles: implications for nucleation mechanisms. Philos. Trans. R. Soc. London Ser. A
  3             358:2625-2642.
  4       McQuaker, N. R.; Kluckner, P. D.; Chang, G. N. (1979) Calibration of an inductively coupled plasma-atomic
  5             emission spectrometer for the analysis of environmental materials. Anal. Chem. 51: 888-895.
  6       Meyer, A. S., Jr.; Boyd, C. M. (1959) Determination of water by titration with coulometrically generated Karl
  7             Fischer reagent. Anal. Chem. 31: 215-219.
  8       Meyer, M. B.; Rupprecht, E. G. (1996) Paniculate matter sampling methods: the importance of standardization.
  9             J. Aerosol. Sci. 27(suppl. 1):  S349-S350.
10       Meyer, M. B.; Rupprecht, E.; Patashnick, H. (1995) Considerations for the sampling and measurement of ambient
11             paniculate matter. In: Paniculate matter: health and regulatory issues: proceedings of an international
12             specialty conference; April; Pittsburgh, PA. Pittsburgh, PA: Air & Waste Management Association;
13             pp. 204-210. (A&WMA publication VIP-49).
14       Meyer, M. B.; Patashnick, H.; Ambs, J. L.; Rupprecht, E. (2000) Development of a sample equilibration system for
15             the TEOM continuous PM monitor. J. Air Waste Manage. Assoc. 50: 1345-1349.
16       Middlebrook, A. M.; Murphy, D.  M.; Thomson, D. S. (1998) Observations of organic material in individual marine
17             particles at Cape Grim during the first aerosol characterization experiment (ACE 1). J. Geophys. Res.
18             103: 16475-16483.
19       Middlebrook, A. M.; et al. (2002) A comparison of particle mass spectrometers during the 1999 Atlanta Supersites
20             project. J. Geophys. Res.: in press.
21       Mignacca, D.; Stubbs, K. (1999) Effects of equilibration temperature on PM10 concentrations from the TEOM
22             method in the Lower Fraser Valley. J. Air Waste Manage. Assoc. 49: 1250-1254.
23       Miller, F. J.;  Gardner, D. E.; Graham, J. A.; Lee, R.  E., Jr.; Wilson, W. E.; Bachmann, J. D. (1979) Size
24             considerations for establishing a standard for inhalable particles. J. Air Pollut. Control Assoc. 29: 610-615.
25       Misra, C.; Geller, M. D.; Shah, P.; Sioutas, C.; Solomon, P. A. (2001) Development and evaluation of a continuous
26             coarse (PM10-PM25) particle monitor. J. Air Waste Manage. Assoc. 51:  1309-1317.
27       Modey, W. K.; Pang, Y.; Eatough, N. L.; Eatough, D. J. (2001) Fine paniculate (PM25) composition in Atlanta,
28             USA:  assessment of the particle concentrator-Brigham Young University organic sampling system,
29             PC-BOSS, during the EPA supersite study. Atmos. Environ. 35: 6493-6502.
30       Moosmtiller, H.; Arnott, W. P.; Rogers, C. F. (1998) Photoacoustic and filter measurements related to aerosol light
31             absorption during the Northern Front Range Air Quality  Study (Colorado 1996.1997). J. Geophys. Res.
32             103:28,149-28,157.
33       Morawska, L.; Bofinger, N. D.; Kocis, L.; Nwankwoala, A. (1998) Submicrometer and supermicrometer particles
34             from diesel vehicle emissions. Environ. Sci. Technol. 32: 2033-2042.
3 5       Mueller, P. K.; Collins, J. F. (1980)  Development of a paniculate sulfate analyzer. Palo Alto, CA: Electric Power
36             Research Institute; report no. P-1382F.
37       Mueller, P. K.; Fung, K. K.; Heisler, S. L.; Grosjean, D.; Hidy,  G. M. (1982) Atmospheric paniculate carbon
3 8             observations in urban and rural areas of the United States. In: Wolff, G. T.; Klimisch, R. L., eds. Paniculate
39             carbon: atmospheric life cycle: proceedings of an international symposium; October 1980; Warren, MI.
40             New York, NY: Plenum Press; pp. 343-370.
41       Mueller, P. K.; Hansen, D. A.; Watson, J. G., Jr. (1986) The subregional cooperative electric utility, Department of
42             Defense, National Park Service, and EPA study (SCENES) on visibility: an overview. Palo Alto, CA: Electric
43             Power Research Institute; report no. EA-4664-SR.
44       Mukerjee, S.; Ellenson, W. D.; Lewis, R.  G.; Stevens, R. K.;  Somerville, M. C.; Shadwick, D. S.; Willis, R. D.
45             (1997) An environmental scoping study in the lower Rio Grande Valley of Texas-Ill, residential
46             microenvironmental monitoring for air, house dust, and soil. Environ. Int. 23: 657-673.
47       Munksgaard, N. C.; Parry, D. L. (1998) Lead isotope ratios determined by ICP-MS: monitoring of mining-derived
48             metal particulates in atmospheric fallout, Northern Territory, Australia.  Sci. Total Environ. 217: 113-126.
49       Murphy, D. M.; Thomson, D.  S. (1997) Chemical composition of single aerosol particles at Idaho Hill: negative ion
50             measurements. J. Geophys. Res. 102: 6353-6368.
51       Nadkarni, R. A. (1975) Multielement analysis of coal and coal fly ash standards by instrumental neutron activation
52             analysis. Radiochem. Radioanal. Lett. 21: 161-176.
53       Nam, S.-H.; Masamba, W. R. L.; Montaser, A. (1993)  Investigation of helium inductively coupled plasma-mass
54             spectrometry for the detection of metals and nonmetals in aqueous solutions. Anal. Chem. 65: 2784-2790.
55       NARSTO. (2003) Fine particle assessment (review draft). Oak Ridge, TN: U.S. Department of Energy, Oak Ridge
56             National Laboratory, North American Research Strategy for Tropospheric Ozone, Quality Systems Science
57             Center. Available: http://www.cgenv.com/Narsto/ (22  April 2003).


         June 2003                                      2-113         DRAFT-DO NOT QUOTE OR CITE

-------
  1       National Research Council. (1993) Protecting visibility in national parks and wilderness areas. Washington, DC:
  2             National Academy Press. 3v.
  3       Nelson, M. P.; Zugates, C. T.; Treado, P. I; Casucio, G. S.; Exline, D. L.; Schlaegle, S. F. (2000) Combining Raman
  4             chemical imaging scanning electron mocroscopy (SEM) to characterize ambient fine paniculate. Aerosol Sci.
  5             Technol. 34: 108-117.
  6       Neubauer, K. R.; Johnston, M. V.; Wexler, A. S. (1998) Humidity effects on the mass spectra of single aerosol
  7             particles. Atmos. Environ. 32: 2521-2529.
  8       Nham, T. T.  (1996) Cool plasma technique for the ultra-trace level determination of 56Fe, 40Ca, 39K, 7Li, 24Mg
  9             and 23Na for semi-conductor applications. Mulgrave, Victoria, Australia: Varian Australia Pty Ltd; ICP-MS
10             instruments at work no. ICP-MS-7. Available:
11             http://www.varianinc.com/cgi-bin/nav?varinc/docs/osi/icpms/atwork/icpms7&cid=680052 [18 January,
12             2002].
13       Noble, C. A.; Prather, K. A. (1996) Real-time measurement of correlated size and composition profiles of individual
14             atmospheric aerosol particles. Environ. Sci. Technol. 30: 2667-2680.
15       Noble, C. A.; Prather, K. A. (1998) Single particle characterization of albuterol metered dose inhaler aerosol in near
16             real-time. Aerosol Sci. Technol. 29: 294-306.
17       Noble, C. A.; Vanderpool, R. W.; Peters, T. M.; McElroy, F. F.; Gemmill, D. B.; Wiener, R. W.  (2001) Federal
18             reference and equivalent methods for measuring fine paniculate matter. Aerosol Sci. Technol. 34: 457-464.
19       Noel, M. A.; Topart, P. A. (1994) High-frequency impedance analysis of quartz crystal microbalances. 1. General
20             considerations. Anal. Chem. 66: 484-491.
21       Novakov, T.; Corrigan, C. E.; Penner, J. E.; Chuang, C. C.; Rosario, O.; Mayol Bracero, O. L.  (1997) Organic
22             aerosols in the Caribbean trade winds: a natural source? J. Geophys. Res. 102: 21307-21313.
23       O'Dowd, C. D.; Aalto, P.; Hameri, K.; Kulmala, M.; Hoffmann, T. (2002) Atmospheric particles from organic
24             vapours. Nature (London, U. K.) 416: 497-498.
25       Obeidi, F. D.; Eatough, D. J. (2002) Continuous measurement of semi-volatile fine paniculate mass in Provo, UT.
26             Aerosol Sci. Technol. 36: 191-203.
27       Obeidi, F.; Eatough, N. L.; Eatough, D. J. (2002) Use of the RAMS to measure semivolatile fine particulate matter at
28             Riverside and Bakersfield, California. Aerosol Sci. Technol. 36: 204-216.
29       Odum, J. R.; Yu, J.; Kamens, R. M. (1994) Modeling the mass transfer of semivolatile organics in combustion
30             aerosols. Environ. Sci. Technol. 28: 2278-2285.
31       Ohta, S.; Okita, T. (1990) A chemical characterization of atmospheric aerosol in Sapporo. Atmos. Environ. Part A
32             24A: 815-822.
33       Ohta, S.; Hori, M.; Yamagata, S.; Murao, N. (1998) Chemical characterization of atmospheric fine particles in
34             Sapporo with determination of water content. Atmos. Environ. 32: 1021-1025.
3 5       Olin, J. G.; Sem,  G. J. (1971) Piezoelectric microbalance for monitoring the  mass concentration of suspended
36             particles. Atmos. Environ. 5: 653-668.
37       Olmez, I. (1989) Trace element signatures in groundwater pollution. In: Watson, J. G., ed. Receptor models in air
38             resources management: an international specialty conference; February 1988; San Francisco, CA. Pittsburgh,
39             PA: Air & Waste Management Association; pp. 3-11. (APCA transactions series: no. 14).
40       Ondov, J. M. (1996) Particulate tracers for source attribution: potential for application to California's San Joaquin
41             Valley. J. Aerosol Sci. 27(suppl.  1): S687-S688.
42       Ondov, J. M.; Kelly, W. R.; Holland, J. Z.; Lin, Z. C.; Wight, S. A. (1992) Tracing fly ash emitted from a coal-fired
43             power plant with ennched rare-earth isotopes: an urban scale test. Atmos. Environ. Part B 26B: 453-462.
44       Ondov, J. M.; Divita, F., Jr. (1993) Size spectra for trace elements in urban aerosol particles by instrumental neutron
45             activation analysis. J. Radioanal. Nucl. Chem.  167: 247-258.
46       Ono, D. M.;  Hardebeck, E.; Parker, J.; Cox, B. G. (2000) Systematic biases in measured PM10 values with U.S.
47             Environmental Protection Agency-approved samplers at Owens Lake, California. J. Air Waste Manage.
48             Assoc. 50: 1144-1156.
49       Pak, S. S.; Liu, B. Y. H.; Rubow, K. L. (1992) Effect of coating thickness on particle bounce in inertia! impactors.
50             Aerosol Sci. Technol. 16: 141-150.
51       Pang, Y.; Ren, Y.; Obeidi, F.; Hastings, R.; Eatough, D. J.; Wilson, W. E. (2001) Semi-volatile species in PM25:
52             comparison of integrated and continuous samplers for PM2 5 research  or monitoring. J. Air Waste Manage.
53             Assoc. 51:25-36.
54       Pang, Y.; Eatough, N. L.; Wilson, J.; Eatough, D. J. (2002a) Effect of semivolatile material on PM2 5 measurement
55             by the PM2 5 federal reference method sampler at Bakersfield, California. Aerosol Sci. Technol. 36: 289-299.
56       Pang, Y.; Eatough, N. L.; Eatough, D. J. (2002b) PM2 5 semivolatile  organic  material at Riverside, CA: implications
57             for the PM2 5 federal reference method sampler. Aerosol Sci. Technol. 36: 277-288.


         June 2003                                       2-114         DRAFT-DO NOT  QUOTE OR CITE

-------
  1       Pankow, J. F. (1987) Review and comparative analysis of the theories on partitioning between the gas and aerosol
  2             paniculate phases in the atmosphere. Atmos. Environ. 21: 2275-2283.
  3       Pankow, J. F. (1994a) An absorption model of the gas/aerosol partitioning involved in the formation of secondary
  4             organic aerosol. Atmos. Environ. 28: 189-193.
  5       Pankow, J. F. (1994b) An absorption model of gas/particle partitioning of organic compounds in the atmosphere.
  6             Atmos. Environ. 28: 185-188.
  7       Pankow, J. F.; Bidleman, T. F. (1991) Effects of temperature, TSP and per cent non-exchangeable material in
  8             determining the gas-particle partitioning of organic compounds. Atmos. Environ. Part A 25: 2241-2249.
  9       Pankow, J. F.; Storey, J. M. E.; Yamasaki, H. (1993) Effects of relative humidity on gas/particle partitioning of
10             semivolatile organic compounds to urban paniculate matter. Environ. Sci. Technol. 27: 2220-2226.
11       Parmar, S. S.; Grosjean, D. (1990) Laboratory tests of KI and alkaline annular denuders. Atmos. Environ. Part A
12             24:2695-2698.
13       Patashnick, H.; Rupprecht, G.; Ambs, J. L.; Meyer, M. B. (2001) Development of a reference standard for
14             paniculate matter mass in ambient air. Aerosol Sci. Technol. 34: 42-45.
15       Pellizzari, E. D.; Krost, K. J.  (1984) Chemical transformations during ambient air sampling for organic vapors. Anal.
16             Chem. 56: 1813-1819.
17       Peng, C.; Chan, C. K. (2001) The water cycles of water-soluble organic salts of atmospheric importance.
18             Atmos. Environ. 35: 1183-1192.
19       Penner, J. E.; Novakov, T. (1996) Carbonaceous particles in the atmosphere: a historical perspective to the fifth
20             international conference on carbonaceous particles in the atmosphere. J. Geophys. Res. [Atmos.]
21             101: 19,373-19,378.
22       Perry, K. D.; Hobbs, P. V. (1994) Further evidence for particle nucleation in clear air adjacent to marine cumulus
23             clouds. J. Geophys. Res. [Atmos.] 99: 22,803-22,818.
24       Perry, K. D.; Cahill, T. A.; Schnell, R. C.; Harris, J. M. (1999) Long-range transport of anthropogenic aerosols to the
25             National Oceanic and Atmospheric Administration baseline  station at Mauna Loa Observatory, Hawaii.
26             J. Geophys. Res. Atmos. 104:  18,521-18,533.
27       Peters, A.; Liu, E.; Verrier, R. L.; Schwartz, J.; Gold, D.  R.; Mittleman, M.;  Baliff, J.; Oh, J. A.; Allen, G.;
28             Monahan, K.; Dockery, D. W. (2000) Air pollution and incidence of cardiac arrhythmia. Epidemiology
29             11: 11-17.
30       Peters, T. M.; Vanderpool, R. W.; Wiener, R. W. (200 la) Design and calibration of the EPA PM2 5 well impactor
31             ninety-six (WINS).  Aerosol Sci. Technol. 34: 389-397.
32       Peters, T. M.; Norris, G. A.; Vanderpool, R. W.; Gemmill, D. B.; Wiener, R. W.; Murdoch, R. W.; McElroy, F. F.;
33             Pitchford, M. (200 Ib) Field performance of PM25 federal reference method samplers. Aerosol Sci. Technol.
34             34:433-443.
35       Peters, T. M.; Gussman, R. A.; Kenny, L. C.; Vanderpool, R. W. (2001c) Evaluation of PM25 size selectors used in
36             speciation samplers. Aerosol Sci. Technol. 34: 422-429.
37       Petzold, A.; Niessner, R. (1996) Photoacoustic soot sensor for in-situ black carbon monitoring. Appl. Phys. B:
38             Lasers  Opt. 63: 191-197.
39       Petzold, A.; Kopp, C.; Niessner, R. (1997) The dependence  of the specific attenuation cross-section on black carbon
40             mass fraction and particle size. Atmos. Environ. 31: 661-672.
41       Pilinis, C.; Seinfeld, J. H.;  Grosjean, D. (1989) Water content of atmospheric aerosols. Atmos. Environ.
42             23: 1601-1606.
43       Piorek, S. (1994) Modern,  PC based, high resolution portable EDXRF analyzer offers laboratory performance for
44             field, in-situ analysis of environmental contaminants. Nucl. Instr. Meth. Phys. Res. A 353: 528-533.
45       Pirrone, N.; Keeler, G. J. (1996) The Rouge River watershed pollution by trace elements: atmospheric depositions
46             and emission sources.  Water Sci. Technol. 33: 267-275.
47       Plantz, M. (1996) Common molecular ion interferences in ICP-MS. Mulgrave, Victoria, Australia: Varian Australia
48             Pty Ltd; ICP-MS instruments at work no. ICP-MS-6.  Available:
49             http://www.varianinc.com/cgi-bin/nav?varinc/docs/osi/icpms/atwork/index&cid=680052 [18 January, 2002].
50       Poor, N.; Clark, T.; Nye, L.; Tamanini, T.; Tate, K.; Stevens, R.; Atkeson, T. (2002) Field performance of
51             dichotomous sequential PM air samplers. Atmos. Environ. 36: 3289-3298.
52       Pope, C. A., Ill; Thun, M. J.; Namboodiri,  M. M.; Dockery,  D. W.; Evans, J. S.; Speizer, F. E.; Heath, C. W., Jr.
53             (1995) Paniculate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am. J. Respir.
54             Crit. Care Med. 151: 669-674.
55       Potts, P. J.; Webb, P. C. (1992) X-ray fluorescence spectrometry. J. Geochem. Explor. 44: 251-296.
56       Radke, L. F.; Hobbs, P. V. (1991) Humidity and particle fields around some small cumulus clouds. J. Atmos. Sci.
57             48: 1190-1193.


         June 2003                                      2-115        DRAFT-DO NOT QUOTE OR CITE

-------
  1       Raes, F.; Van Dingenen, R.; Cuevas, E.; Van Velthoven, P. F. J.; Prospero, J. M. (1997) Observations of aerosols in
  2             the free troposphere and marine boundary layer of the subtropical Northeast Atlantic: discussion of processes
  3             determining their size distribution. J. Geophys. Res. [Atmos.]  102: 21315-21328.
  4       Ranweiler, L. E.; Moyers, J. L. (1974) Atomic absorption procedure for analysis of metals in atmospheric paniculate
  5             matter. Environ.  Sci. Technol. 8: 152-156.
  6       Rao, A. K.; Whitby, K.  T. (1977) Nonideal collection characteristics of single stage and cascade impactors. Am. Ind.
  7             Hyg. Assoc. J. 38: 174-177.
  8       Rao, A. K.; Whitby, K.  T. (1978) Non-ideal collection characteristics of inertial impactors-I.  single-stage impactors
  9             and solid particles. J. Aerosol Sci. 9: 77-86.
10       Ravenhall, D. G.; Forney, L. J.; Jazayeri, M. (1978) Aerosol sizing with a slotted virtual impactor. J. Colloid
11             Interface Sci. 65: 108-117.
12       Reid, J. S.; Cahill, T. A.; Wakabayashi, P. H.; Dunlap, M. R. (1994) Geometric/aerodynamic equivalent diameter
13             ratios of ash aggregate aerosols collected in burning Kuwaiti well fields. Atmos. Environ. 28: 2227-2234.
14       Reilly, P. T. A.; Gieray, R. A.; Whitten, W. B.; Ramsey, J. M. (1998) Real-time characterization of the organic
15             composition and size of individual diesel engine smoke particles. Environ. Sci. Technol. 32: 2672-2679.
16       Reist, P. C. (1984) Introduction to aerosol science. New York, NY: Macmillan Publishing Company.
17       Reist, P. C. (1993) Introduction to aerosol science. New York, NY: McGraw-Hill.
18       Risse, U.; Flammenkamp, E.; Kettrup, A. (1996) Determination of aromatic hydrocarbons in air using diffusion
19             denuders. Fresenius' J. Anal.  Chem. 356: 390-395.
20       Roberts, R. A.; Corkill,  J. (1998) Grass seed field smoke and its impact on respiratory health.  J. Environ. Health 60
21             (June): 10-16.
22       Rounds, S. A.; Pankow, J. F. (1990) Application of a radial diffusion model to describe gas/particle sorption
23             kinetics. Environ. Sci. Technol. 24:  1378-1386.
24       Rounds, S. A.; Tiffany,  B. A.; Pankow, J. F. (1993) Description of gas/particle sorption kinetics with an intraparticle
25             diffusion model:  desorption experiments. Environ. Sci. Technol. 27: 366-377.
26       Rumburg, B.; Alldredge, R.; Claiborn, C. (2001) Statistical distributions of paniculate matter  and the error
27             associated with sampling frequency. Atmos. Environ.  35: 2907-2920.
28       Rupprecht, G.; Patashnick, H.; Beeson, D. E.; Green, R. N.; Meyer, M. B. (1995) A new automated monitor for the
29             measurement of paniculate carbon in the atmosphere.  In: Paniculate matter: health and regulatory issues:
30             proceedings of an international specialty conference; April; Pittsburgh, PA. Pittsburgh, PA: Air & Waste
31             Management Association; pp. 309-319. (A&WMA publication VTP-49).
32       Sakata, K.; Kawabata, K. (1994) Reduction of fundamental polyatomic ions in inductively coupled plasma mass
33             spectrometry. Spectrochim. Acta49B: 1027-1038.
34       Saxena, P.; Hildemann, L. M. (1996) Water-soluble organics in atmosphenc  particles: a critical review of the
35             literature and applications of thermodynamics to identify candidate compounds. J. Atmos. Chem. 24: 57-109.
36       Saxena, P.; Hildemann, L. M.; McMurry, P. H.; Seinfeld, J. H.  (1995) Organics  alter hygroscopic behavior of
37             atmospheric particles. J. Geophys. Res. [Atmos.] 100: 18,755-18,770.
38       Schmid, H.; Laskus, L.; Abraham, J. H.; Baltensperger, U.; Lavanchy, V; Bizjak,  M.; Burba,  P.; Cachier, H.; Crow,
39             D.; Chow, J.; Gnauk,  T.; Even, A.; Ten Brink, H. M.;  Giesen, K.-P.; Hitzenberger, R.; Hueglin, C.;
40             Maenhaut, W.; Pio, C.; Carvalho, A.; Putaud, J.-P.; Toom-Sauntry, D.; Puxbaum, H. (2001) Results of the
41             "carbon conference" international aerosol carbon round robin test stage I. Atmos. Environ. 35: 2111-2121.
42       Schroder, F.; Strom, J. (1997) Aircraft measurements of submicrometer aerosol particles (> 7  nm)  in the midaltitude
43             free troposphere  and tropopause region. Atmos. Res. 44: 333-356.
44       Schwartz, J.; Dockery, D. W.; Neas, L. M.  (1996) Is daily mortality associated specifically with fine particles? J. Air
45             Waste Manage. Assoc. 46: 927-939.
46       Seinfeld, J. H.; Pandis, S. N. (1998) Atmosphenc chemistry and physics: from air pollution to climate change.
47             New York, NY: John Wiley & Sons, Inc.
48       Sem, G. J.; Tsurubayashi, K.; Homma, K.  (1977) Performance of the piezoelectric microbalance respirable aerosol
49             sensor. Am. Ind.  Hyg. Assoc. J. 38:  580-588.
50       Shah, J. J.; Rau, J. A. (1990) Carbonaceous methods comparison study: interlaboratory round robin interpretation
51             results. Sacramento, CA: California Air Resources Board. Available:
52             http://www.arb.ca.gov/research/apr/past/pm.htm (21 March 2002).
53       Silva, P. J.; Prather,  K. A. (1997) On-line characterization of individual particles from automobile  emissions.
54             Environ.  Sci. Technol. 31: 3074-3080.
55       Simon, P. K.; Dasgupta, P. K.  (1995) Continuous automated measurement of the soluble fraction of atmosphenc
56             paniculate matter. Anal. Chem. 67: 71-78.
         June 2003                                      2-116         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Sioutas, C.; Koutrakis, P. (1996) Inertial separation of ultrafine particles using a condensational growth/virtual
  2             impaction system. Aerosol Sci. Technol. 25: 424-436.
  3       Sioutas, C.; Koutrakis, P.; Burton, R. M. (1994a) Development of a low cutpoint slit virtual impactor for sampling
  4             ambient fine particles. J. Aerosol Sci. 25: 1321-1330.
  5       Sioutas, C.; Koutrakis, P.; Burton, R. M. (1994b) A high-volume small cutpoint virtual impactor for separation of
  6             atmospheric paniculate from gaseous pollutants. Part. Sci. Technol. 12: 207-221.
  7       Sioutas, C.; Koutrakis, P.; Olson, B. A. (1994c) Development and evaluation of a low cutpoint virtual impactor.
  8             Aerosol Sci. Technol. 21:223-235.
  9       Sioutas, C.; Koutrakis, P.; Wolfson, J. M. (1994d) Particle losses in glass honeycomb denuder samplers. Aerosol
10             Sci. Technol. 21: 137-148.
11       Sioutas, C.; Koutrakis, P.; Burton, R. M. (1995a) A technique to expose animals to concentrated fine ambient
12             aerosols. Environ. HealthPerspect. 103: 172-177.
13       Sioutas, D.; Koutrakis, P.; Ferguson,  S. T.; Burton, R. M. (1995b) Development and evaluation of a prototype
14             ambient particle concentrator for inhalation exposure studies. Inhalation Toxicol. 7: 633-644.
15       Sioutas, C.; Koutrakis, P.; Wolfson, J. M.; Azaroff, L. S.; Mulik, J. D. (1996a) Field evaluation of a glass
16             honeycomb denuder/filter pack system to collect atmospheric gases and particles. In: Gay, B. W.; Fuerst,
17             R. G.; Jayanty, R. K. M., eds. Measurement of toxic and related air pollutants: proceedings of a conference;
18             May, 1994; Durham, NC. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of
19             Research and Development; report no. EPA/600/R-94/136; pp. 426-435.
20       .Sioutas, C.; Wang, P. Y.; Ferguson, S. T.; Koutrakis, P.; Mulik, J. D. (1996b) Laboratory and field evaluation of an
21             improved glass honeycomb denuder/filter pack sampler.  Atmos. Environ. 30: 885-895.
22       Sioutas, C.; Koutrakis, P.; Wang, P.-Y.; Babich, P.; Wolfson, J. M. (1999) Experimental investigation of pressure
23             drop with particle loading in Nuclepore filters. Aerosol Sci. Technol. 30: 71-83.
24       Sioutas, C.; Kim, S.; Chang, M.; Terrell, L. L.; Gong, H., Jr. (2000) Field evaluation of a modified DataRAM MIE
25             scattering monitor for real-time PM2 5 mass concentration measurements. Atmos. Environ. 34: 4829-4838.
26       Smith, K. R.; Kim, S.; Recendez, J. J.; Sioutas, C.; Pinkerton, K. E.  (2003) Airborne particles of the California
27             Central Valley alter the lungs of health adult rats. Environ. Health Perspect.
28       Solomon, P. A.; Moyers, J. L.; Fletcher, R. A. (1983) High-volume dichotomous virtual impactor for the
29             fractionation and collection of particles according to aerodynamic size. Aerosol Sci. Technol. 2: 455-464.
30       Solomon, P. A.; Salmon, L. G.; Fall, T.; Cass, G. R. (1992) Spatial and temporal distribution of atmospheric nitric
31             acid and paniculate nitrate concentrations in the Los Angeles area. Environ. Sci. Technol. 26: 1596-1601.
32       Song, X.-H.; Faber, N. M.; Hopke, P. K.; Suess, D. T.; Prather, K. A.; Schauer, J. J.; Cass, G. R. (2001) Source
3 3             apportionment of gasoline and diesel by multivariate calibration based on single particle mass spectral data.
34             Anal. Chim. Acta 446: 329-343.
35       Speer, R. E.; Barnes, H. M.; Brown, R. (1997) An instrument for measuring the liquid water content of aerosols.
36             Aerosol Sci. Technol. 27: 50-61.
37       Spengler, J. D.; Thurston, G. D. (1983) Mass and elemental composition of fine and coarse particles in six U.S.
38             cities. J. AirPollut. Control Assoc. 33: 1162-1171.
39       Stahlschmidt, T.; Schulz, M.; Dannecker, W. (1997) Application of total-reflection X-ray fluorescence for the
40             determination of lead, calcium and zinc in size-fractionated marine aerosols. Spectrochim. Acta B
41             52:995-1001.
42       Stein, S. W.; Turpin, B. J.; Cai, X.; Huang, P.-F.; McMurry, P. H. (1994) Measurements of relative
43             humidity-dependent bounce and density for atmospheric particles using the DMA-impactor technique. Atmos.
44             Environ. 28: 1739-1746.
45       Stevens, R. K.; O'Keeffe, A. E.; Ortman, G. C. (1969) Absolute calibration of a flame photometric detector to
46             volatile sulfur compounds at sub-part-per-million levels. Environ. Sci. Technol. 3: 652-655.
47       Stevens, R. K.; Mulik,  J. D.; O'Keefe, A. E.; Krost, K. J. (1971) Gas chromatography  of reactive sulfur gases in air
48             at the parts-per-billion level. Anal. Chem. 43:  827-831.
49       Stolzenburg, M. R.; Hering, S. V. (2000) Method for the automated measurement of fine particle nitrate in the
50             atmosphere. Environ. Sci. Technol. 34: 907-914.
51       Streit, N.; Weingartner, E.; Zellweger, C.; Schwikowski, M.; Gaggeler, H. W.; Baltensperger, U. (2000)
52             Characterization of size-fractionated aerosol from the Jungfraujoch (3580 m asl) using total reflection x-ray
53             fluorescence (TXRF). Int. J. Environ. Anal. Chem. 76: 1-16.
54       Strobel, H. A.; Heineman, W. R. (1989) Chemical instrumentation: a systematic approach. 3rd ed. New York, NY:
55             Wiley and Sons, Inc.
         June 2003                                      2-117         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Strommen, M. R.; Kamens, R. M. (1997) Development and application of a dual-impedance radial diffusion model
  2             to simulate the partitioning of semivolatile organic compounds in combustion aerosols. Environ. Sci. Technol.
  3             31:2983-2990.
  4       Strothmann, J. A.; Schiermeier, F. A. (1979) Documentation of the regional air pollution study (RAPS) and related
  5             investigations in the St. Louis air quality control region. Research Triangle Park, NC: U.S. Environmental
  6             Protection Agency, Environmental Science Research Laboratory; report no. EPA-600/4-79-076.
  7       Suarez, A. E.; Caffrey, P. F.; Ondov, J. M.; Thaung, K. C. (1996) Size distribution of Ir-tagged soot from diesel
  8             sanitation trucks. J. Aerosol Sci. 27(suppl. 1): S697-S698.
  9       Suggs, J. C.; Burton, R.  M. (1983) Spatial characteristics of inhalable particles in the Philadelphia metropolitan area.
10             J. AirPollut. Control Assoc. 33: 688-691.
11       Suh, H. H.; Spengler, J.  D.; Koutrakis, P. (1992) Personal exposures to acid aerosols and ammonia. Environ. Sci.
12             Technol.  26:2507-2517.
13       Suh, H. H.; Koutrakis, P.;  Spengler, J. D. (1994) The relationship between airborne acidity and ammonia in indoor
14             environments. J. Exposure Anal. Environ. Epidemiol. 4: 1-23.
15       Sutherland, J. L.; Bhardwaja, P. S. (1987) Composition of the aerosol in northern Arizona and southern Utah. In:
16             Bhardwaja, P. S., ed. Transactions, Visibility Protection: Research and Policy Aspects. Pittsburgh, PA: Air
17             Pollution Control Association; pp. 373-385.
18       Tan, S. H.; Horlick, G. (1986) Background spectral features in inductively coupled plasma/mass spectrometry. Appl.
19             Spectrosc. 40: 445-460.
20       Tang, I. N. (1980) Deliquescence properties and particle size change of hygroscopic aerosols. In: Willeke, K., ed.
21             Generation of aerosols and facilities  for exposure experiments. Ann Arbor, MI: Ann Arbor Science
22             Publishers, Inc.; pp. 153-167.
23       Tang, H.; Lewis, E. A.; Eatough, D. J.; Burton, R. M.; Farber, R. J. (1994) Determination of the particle size
24             distribution and chemical composition of semi-volatile organic compounds in atmospheric fine particles with
25             a diffusion denuder sampling system. Atmos. Environ. 28: 939-947.
26       Tanner, R. L.; D'Ottavio, T.; Garber, R.; Newman, L. (1980) Determination of ambient aerosol sulfur using a
27             continuous flame photometric detection system. I. Sampling system for aerosol sulfate and sulfuric acid.
28             Atmos. Environ.  14: 121-127.
29       Thomson, D. S.; Murphy, D. M. (1994) Analyzing single aerosol particles in real time. CHEMTECH 24: 30-35.
30       Tobias, H. J.; Beving, D. E.; Ziemann, P. J.; Sakurai, H.; Zuk, M.; McMurry, P. H.; Zarling, D.; Wayrulonis, R.;
31             Kittelson, D. B. (2001) Chemical analysis of diesel engine nanoparticles using a nano-DMA/thermal
32             desorption particle beam mass spectrometer. Environ. Sci. Technol. 35: 2233-2243.
33       Tolocka, M. P.;  Peters, T. M.; Vanderpool,  R. W.; Chen, F. L.; Wiener, R. W. (2001a) On the modification of the
34             low flow-rate PM10 dichotomous sampler inlet. Aerosol Sci. Technol. 34: 407-415.
35       Tolocka, M. P.;  Solomon, P. A.; Mitchell, W.;  Norris, G. A.; Gemmill, D. B.; Wiener, R. W.; Vanderpool, R. W.;
36             Homolya, J. B.; Rice, J. (200Ib) East vs. west in the US: chemical characteristics of PM2 5 during the winter
37             of 1999. Aerosol  Sci. Technol. 34: 88-96.
38       Torok, S. B.; Labar, J.; Schmeling, M.; Van Grieken, R. E. (1998) X-ray spectrometry. Anal. Chem. 70: 495R-517R.
39       Tsai, C.-J.; Cheng, Y.-H. (1996) Comparison of two ambient beta gauge PM10 samplers. J. Air Waste Manage.
40             Assoc. 46: 142-147.
41       Tsai, C.-J.; Huang, H.-Y. (1995) Atmospheric aerosol sampling by an annular denuder system and a high-volume
42             PM10 sampler. Environ. Int. 21: 283-291.
43       Turner, D. B. (1994) Workbook of atmospheric dispersion estimates: an introduction to dispersion modeling. 2nd ed.
44             Boca Raton, FL: Lewis Publishers.
45       Turner, J. R.; Hering, S. V. (1987) Greased and oiled substrates as bounce-free impaction surfaces.  J. Aerosol Sci.
46             18:215-224.
47       Turpin, B. J.; Huntzicker, J. J. (1991) Secondary formation of organic aerosol in the Los Angeles basin: a descriptive
48             analysis of organic and elemental carbon concentrations. Atmos. Environ. Part A 25: 207-215.
49       Turpin, B. J.; Huntzicker, J. J. (1995) Identification of secondary organic aerosol episodes and quantitation of
50             primary and secondary organic aerosol concentrations during SCAQS.  Atmos. Environ. 29: 3527-3544.
51       Turpin, B. J.; Gary, R. A.;  Huntzicker, J. J.  (1990) An in situ, time-resolved analyzer for aerosol organic and
52             elemental carbon. Aerosol Sci. Technol. 12:  161-171.
53       Turpin, B. J.; Liu, S.-P.; Podolske, K. S.; Gomes, M. S. P.; Eisenreich, S. J.; McMurry, P. H. (1993) Design and
54             evaluation of a novel diffusion separator for measuring gas/particle distributions of semivolatile organic
55             compounds. Environ. Sci. Technol. 27: 2441-2449.
56       Turpin, B. J.; Huntzicker, J. J.; Hering, S. V. (1994) Investigation of organic aerosol sampling artifacts in the
57             Los Angeles basin.  Atmos. Environ.  28: 3061-3071.


         June  2003                                      2-118         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Turpin, B. I; Saxena, P.; Andrews, E. (2000) Measuring and simulating paniculate organics in the atmosphere:
  2             problems and prospects. Atmos. Environ. 34: 2983-3013.
  3       Tyler, G. (1992) ICP vs. AA: how do you choose? Environ. Lab. (October/November): 31-34.
  4       U.S. Court of Appeals for the District of Columbia. (1999) American Trucking Associations, Inc. vs. U.S.
  5             Environmental Protection Agency. 175 F.3d 1027 (D.C. Cir. 1999).
  6       U.S. Environmental Protection Agency. (1982)  Air quality criteria for particulate matter and sulfur oxides. Research
  7             Triangle Park, NC: Office of Health and  Environmental Assessment, Environmental Criteria and Assessment
  8             Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield, VA; PB84-156777.
  9       U.S. Environmental Protection Agency. (1996a) Air quality criteria for particulate matter. Research Triangle Park,
10             NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
11       U.S. Environmental Protection Agency. (1996b) Review of the national ambient air quality standards for particulate
12             matter: policy assessment of scientific and technical information. OAQPS staff paper. Research Triangle
13             Park, NC: Office of Air Quality Planning and Standards; report no. EPA/452/R-96-013. Available from:
14             NTIS, Springfield, VA; PB97-115406REB.
15       U.S. Environmental Protection Agency. (1998)  Monitoring PM2 5 in ambient air using designated reference of class I
16             equivalent methods. In: Quality Assurance  Handbook for Air Pollution Measurement Systems, V. II, Part II,
17             Section 2.12. Research Triangle Park, NC:  Research and Development, Environmental Monitoring Systems
18             Laboratory.
19       U.S. Environmental Protection Agency. (1999)  Particulate matter (PM25) speciation guidance. Final draft. Edition 1.
20             Research Triangle Park, NC: Office of Air  Quality Planning and Standards. Available:
21             www.epa.gov/ttn/amtic/files/ambient/pm25/spec/specfinl.pdf [2000, October 11].
22       U.S. Environmental Protection Agency. (2001)  List of designated reference and equivalent methods. Available:
23             http://www.epa.gov/ttn/amtic/files/ambient/criteria/repmlist.pdf [05 March 02].
24       U.S. Environmental Protection Agency. (2002a) Ambient monitoring technology information center (AMTIC).
25             Washington, DC: Office of Air Quality Planning and Standards; Monitoring and Quality Assurance Group.
26             Available: http://www.epa.gov/ttn/amtic/ (20 March 2002).
27       U.S. Environmental Protection Agency. (2002b) Support center for regulatory air models. Washington, DC: Office
28             of Air Quality Planning and Standards. Available: http://www.epa.gov/scram001 (21 March 2002).
29       Van Dingenen, R.; Raes, F.; Jensen, N. R. (1995)  Evidence for anthropogenic impact on number concentration and
3 0             sulfate content of cloud-processed aerosol particles over the North Atlantic. J. Geophys. Res.
31             100:21,057-21,067.
32       Van Loy, M.; Bahadori, T.; Wyzga, R.; Hartsell, B.; Edgerton, E. (2000) The Aerosol Research and Inhalation
33             Epidemiology Study (ARIES): PM25 mass and aerosol component concentrations and sampler
34             intercomparisons. J. Air Waste Manage.  Assoc. 50: 1446-1458.
35       Vanderpool, R. W.; Peters, T. M.; Natarajan, S.; Gemmill, D. B.; Wiener, R. W. (2001) Evaluation  of the loading
36             characteristics of the EPA WINS PM25 separator. Aerosol Sci. Technol. 34: 444-456.
37       Vasiliou, J. G.; Sorenson, D.; McMurry,  P. H. (1999) Sampling at controlled relative humidity with a cascade
38             impactor. Atmos. Environ. 30: 1049-1056.
39       Venkataraman, C.; Lyons, J. M.; Friedlander, S. K. (1994) Size distributions of polycyclic aromatic hydrocarbons
40             and elemental carbon.  1. Sampling, measurement methods, and source characterization. Environ. Sci.
41             Technol. 28: 555-562.
42       Vincent, J. H. (1989) Aerosol sampling:  science and practice. New York, NY: John Wiley & Sons.
43       Vincent, J. H. (1995) Aerosol science for industrial hygienists. Oxford, United Kingdom: Pergamon.
44       Wallace, J. C.; Kites, R. A. (1995) Computer-controlled low-volume air sampler for the measurement of semivolatile
45             organic compounds. Environ. Sci. Technol. 29: 2099-2106.
46       Wallace, L.; Howard-Reed, C. (2002) Continuous monitoring of ultrafine, fine, and coarse particles in a residence
47             for 18 months in 1999-2000. J. Air Waste Manage. Assoc. 52: 828-844.
48       Wang, H.-C.; John, W. (1987) Comparative bounce properties of particle materials. Aerosol Sci. Technol.
49             7:285-299.
50       Wang, H.-C.; John, W. (1988) Characteristics of the Berner impactor for sampling inorganic ions. Aerosol Sci.
51             Technol. 8:  157-172.
52       Wang, P. Y. (1997) Continuous aerosol mass measurement by flow obstruction. Boston, MA: Harvard University,
53             School of Public Health.
54       Ward, M. D.; Buttry, D. A. (1990) In situ interfacial mass detection with piezoelectric transducers. Science
55             (Washington, DC) 249: 1000-1007.
         June 2003                                      2-119        DRAFT-DO NOT QUOTE OR CITE

-------
  1       Watson, J. G.; Chow, J. C; Richards, L. W.; Neff, W. D.; Andersen, S. R.; Dietrich, D. L.; Houck, J. E.; Olmez, I.
  2             (1988a) The 1987-88 metro Denver brown cloud study volume II: measurements. Reno, NV: Desert Research
  3             Institute; final report no. 8810.1F2 (2 of 3).
  4       Watson, J. G.; Chow, J. C.; Richards, L. W.; Neff, W. D.; Andersen, S. R.; Dietrich, D. L.; Houck, J. E.; Olmez, I.
  5             (1988b) The 1987-88 metro Denver brown cloud study volume III: data interpretation. Reno, NV: Desert
  6             Research Institute; final report no. 8810.1F3 (3 of 3).
  7       Watson, J. G.; Robinson, N. F.; Chow, J. C.; Henry, R. C.; Kim, B. M; Pace, T. G.; Meyer, E. L.; Nguyen, Q.
  8             (1990a) The USEPA/DRI chemical mass balance receptor model,  CMB 7.0. Environ. Software 5: 38-49.
  9       Watson, J. G.; Robinson, N. F.; Chow, J. C.; Henry, R. C.; Kim, B.; Nguyen, Q. T.; Meyer, E. L.; Pace, T. G.
10             (1990b) Receptor model technical series, v. Ill (1989 revision):  CMB7 user's manual. Research Triangle Park,
11             NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards; report no.
12             EPA-450/4-90-004. Available: http://www.epa.gov/scram001/ [19 February 2002].
13       Watson, J. G.; Chow, J. C.; Lu, Z.; Gertler, A. W. (1993) Paniculate and gaseous organic receptor modeling forthe
14             Southern California Air Quality Study. Sacramento, CA: California Air Resource Board; report no. 8733. IF.
15             Available: http://www.arb.ca.gov/research/apr/past/atmospheric.htmMir%20Quality%20Study (5 June
16             2003).
17       Watson, J. G.; Chow, J. C.; Lu, Z.; Fujita, E. M.; Lowenthal, D. H.; Lawson, D. R.; Ashbaugh, L. L. (1994a)
18             Chemical mass balance source apportionment of PM10 during the Southern California Air Quality Study.
19             Aerosol Sci. Technol. 21: 1-36.
20       Watson, J. G.; Chow, J. C.; Lowenthal, D. H.; Pritchett, L. C.; Frazier, C. A.; Neuroth, G. R.; Robbins, R. (1994b)
21             Differences in the carbon composition of source profiles for diesel- and gasoline-powered vehicles. Atmos.
22             Environ. 28: 2493-2505.
23       Watson, J. G.; Fujita, E. M.; Chow, J. C.; Zielinska, B.; Richards, L. W.; Neff, W.; Dietrich, D. (1998) Northern
24             front range air quality study. Final report. Fort Collins, CO: Colorado State University, Cooperative Institute
25             for Research in the Atmosphere. Available: http://www.nfraqs.colostate.edu/index2.html (16 Jan 2002).
26       Watson, J. G.; Chow, J. C.; Frazier, C. A. (1999) X-ray fluorescence analysis of ambient air samples. In:
27             Landsberger, S.; Creatchman, M., eds. Elemental Analysis  of Airborne Particles. Newark, NJ: Gordon and
28             Breach Science Publishers; pp. 67-96.
29       Weber, R. J.; McMurry, P. H.; Eisele, F. L.; Tanner, D. J. (1995) Measurement of expected nucleation precursor
30             species and 3-500-nm diameter particles at Mauna Loa Observatory, Hawaii. J. Atmos. Sci. 52: 2242-2257.
31       Weber, R. J.; McMurry, P. H.; Mauldin, L.; Tanner, D. J.; Eisele,  F. L.; Brechtel, F. J.; Kreidenweis, S. M.; Kok, G.
32             L.; Schillawski, R. D.;  Baumgardner, D. (1998) A study of new particle formation and growth involving
33             biogenic and trace gas  species measured during ACE 1. J. Geophys. Res. A  103: 16,385-16,396.
34       Weber, R. J.; Marti, J. J.; McMurry, P. H.; Mauldin, L.; Tanner, D.; Eisele, F.; Brechtel, F.; Kreidenweis, S.; Kok,
35             G.; Schillawski, R.; Baumgardner, D. (1997) Measurements of new particle  formation and ultrafine particle
36             growth rates at a clean continental site. J. Geophys. Res.  102: 4375-4385.
37       Weber, R. J.; McMurry, P. H.; Mauldin, R. L., Ill; Tanner, D. J.; Eisele, F. L.; Clarke, A. D.; Kapustin,  V. N. (1999)
3 8             New particle formation in the remote troposphere: a comparison of observations at various sites.  Geophys.
39             Res. Lett. 26:307-310.
40       Weber, R. J.; Orsini, D.; Daun, Y.; Lee, Y.-N.; kotz, P. J.; Brechtel, F.  (2001) A particle-into-liquid collector for
41             rapid measurement of aerosol bulk chemical composition. Aerosol Sci. Technol.  35: 718-727.
42       Wedding, J. B.; Weigand, M.  A. (1993) An automatic particle sampler with beta gauging. J. Air Waste Manage.
43             Assoc. 43:475-479.
44       Weingartner, E.; Burtscher, H.; Baltensperger, U. (1997) Hygroscopic  properties of carbon and diesel soot particles.
45             Atmos. Environ. 31:  2311-2327.
46       Weschler, C. J.; Shields, H. C. (1999) Indoor ozone/terpene reactions as a source of indoor particles. Atmos.
47             Environ. 33:2301-2312.
48       Wesolowski, J. J.; John, W.; Devor, W.; Cahill, T. A.; Feeney, P.  J.; Wolfe, G.; Flocchini, R. (1977) Collection
49             surfaces of cascade impactors. In: Dzubay, T. G., ed. X-ray fluorescence analysis of environmental samples.
50             Ann Arbor, MI: Ann Arbor Science Publishers, Inc.; pp.  121-131.
51       Whitby, K. T. (1978) The physical characteristics of sulfur aerosols. Atmos. Environ. 12: 135-159.
52       Whitby, K. T.; Charlson, R. E.; Wilson, W. E.; Stevens, R. K. (1974) The size of suspended particle matter in air.
53             Science (Washington, DC) 183: 1098-1099.
54       Whitby, K. T.; Sverdrup, G. M. (1980) California aerosols: their physical and chemical characteristics. In: Hidy,
55             G. M.; Mueller, P. K.; Grosjean, D.; Appel, B. R.; Wesolowski, J. J., eds. The  character and origins of smog
56             aerosols: a digest of results from the California Aerosol Characterization Experiment (ACHEX).  New York,
57             NY: John Wiley & Sons, Inc.; pp. 477-517. (Advances in environmental science and technology: v. 9).


         June 2003                                       2-120         DRAFT-DO NOT QUOTE OR CITE

-------
  1       White, W. H. (1998) Statistical considerations in the interpretation of size-resolved paniculate mass data. J. Air
  2             Waste Manage. Assoc. 48: 454-458.
  3       Wiedensohler, A.; Hansson, H.-C.; Orsini, D.; Wendisch, M; Wagner, F.; Bower, K. N; Chourlarton, T. W.;
  4             Wells, M.; Parkin, M.; Acker, K.; Wieprecht, W.; Facchini, M. C.; Lind, J. A.; Fuzzi, S.; Arends, B. G.;
  5             Kulmala, M. (1997) Night-time formation and occurrence of new particles associated with orographic clouds.
  6             Atmos. Environ. 31:2545-2559.
  7       Willeke, K.; Baron, P. A., eds. (1993) Aerosol measurement: principles, techniques and applications. New York,
  8             NY: Van Nostrand Reinhold Publishers.
  9       Willeke, K.; Whitby, K. T. (1975) Atmosperic aerosols: size distribution interpretation. J. AirPollut. Control Assoc.
10             25: 529-534.
11       Williams, E. L., II; Grosjean, D. (1990) Removal of atmospheric oxidants with annular denuders. Environ. Sci.
12             Technol.  24: 811-814.
13       Williams, R.; Suggs, J.; Zweidinger, R.; Evans, G.; Creason, J.; Kwok, R.; Rodes, C.; Lawless, P.; Sheldon, L.
14             (2000) The 1998 Baltimore Paniculate Matter Epidemiology-Exposure Study: Part 1. Comparison of ambient,
15             residential outdoor, indoor and apartment paniculate matter monitoring. J. Exposure Anal. Environ.
16             Epidemiol. 10: 518-532.
17       Wilson, W. E. (1995) Aerosol exposure, physics, and chemistry. Inhalation Toxicol. 7: 769-772.
18       Wilson, W. E.; Suh, H. H. (1997) Fine particles and coarse particles: concentration relationships relevant to
19             epidemiologic studies. J. Air Waste Manage. Assoc. 47: 1238-1249.
20       Wilson, W. E.; Spiller, L. L.; Ellestad, T. G.; Lamothe, P. J.; Dzubay, T. G.; Stevens, R. K.; Macias, E. S.; Fletcher,
21             R. A.; Husar, J. D.; Husar, R. B.; Whitby, K. T.; Kittelson, D. B.; Cantrell, B. K. (1977) General Motors
22             sulfate dispersion experiment: summary of EPA measurements. J. AirPollut. Control Assoc. 27: 46-51.
23       Woo, K.-S.; Chen, D.-R.; Pui, D. Y. H.; McMurry, P. H. (2001a) Measurement of Atlanta aerosol size distributions:
24             observations of ultrafine particle events. Aerosol Sci. Technol. 34: 75-87.
25       Woo, K.-S.; Chen, D.-R.; Pui, D. Y. H.; Wilson, W. E. (2001b) Use of continuous measurements of integral aerosol
26             parameters to estimate particle surface area. Aerosol Sci. Technol. 34: 57-65.
27       Wu, C. C.; Suarez, A. E.; Lin, Z.; Kidwell, C. B.; Borgoul, P. V.; Caffrey, P. F.; Ondov, J. M.; Sattler, B. (1998)
28             Application of an Ir tracer to determine soot exposure to students commuting to  school on Baltimore public
29             buses. Atmos. Environ. 32: 1911-1919.
30       Yamasaki, H.; Kuwata, K.; Miyamoto, H. (1982) Effects of ambient temperature on aspects of airborne poly cyclic
31             aromatic  hydrocarbons. Environ. Sci. Technol. 16: 189-194.
32       Yatin, M.; Tuncel, S. G.; Tuncel, G.; Aras, N. K. (1994) Trace element composition of atmospheric aerosols in
33             Ankara, Turkey, determined by instrumental neutron activation analysis. J. Radioanal. Nucl. Chem.
34             181:401-411.
3 5       Zeng, X.; Wu, X.; Yao, H.; Yang, F.; Cahill, T. A. (1993) PIXE-induced XRF with transmission geometry. Nucl.
36             Instr. Meth. Phys.  Res. B 75: 99-104.
37       Zhang, X. Q.; McMurry,  P. H. (1987) Theoretical analysis of evaporative losses from impactor and filter deposits.
38             Atmos. Environ. 21:  1779-1789.
39       Zhang, X.; McMurry, P. H. (1992) Evaporative losses of fine paniculate nitrates during sampling. Atmos. Environ.
40             Part A 26: 3305-3312.
41       Zhang, X. Q.; McMurry,  P. H.; Hering, S. V.; Casuccio, G. S. (1993) Mixing characteristics and water content of
42             submicron aerosols measured in Los Angeles and at the Grand Canyon. Atmos.  Environ. Part A
43             27: 1593-1607.
44       Zhu, Y.; Hinds,  W. C.; Kim, S.; Shen, S.; Sioutas, C. (2002a) Study of ultrafine particles near a major highway with
45             heavy-duty diesel traffic. Atmos. Environ. 36: 4323-4335.
46       Zhu, Y.; Hinds,  W. C.; Kim, S.; Sioutas, C. (2002b) Concentration and size distribution of ultrafine particles near a
47             major highway. J.  Air Waste Manage. Assoc. 52: 1032-1042.
48       Zoller, W. H.; Gordon, G. E. (1970) Instrumental neutron activation analysis of atmospheric pollutants utilizing
49             Ge(Li) y-ray detectors. Anal. Chem. 42: 257-265.
50

51
         June 2003                                      2-121         DRAFT-DO NOT QUOTE OR CITE

-------
 1               APPENDIX 2A.  TECHNIQUES FOR MEASUREMENT OF
 2                       SEMIVOLATILE ORGANIC COMPOUNDS
 3
 4
 5      Use ofDenuder Systems To Measure Semivolatile Organic Compounds
 6          Phase distribution of semivolatile organic species has been the subject of several studies
 7      that have employed denuder technology (see Gundel et al., 1995; Gundel and Lane, 1999) to
 8      directly determine the phase distributions while avoiding some of the positive and negative
 9      sampling artifacts associated with using back-up quartz filters. In an ideal system with a
10      denuder that is 100% efficient, the gas phase would be collected in the denuder and the particle
11      phase would be the sum of the material collected on the filter and the adsorbent downstream.
12      Denuder collection efficiency depends on the denuder surface area (+), the diffusivity (+) and
13      vapor pressure (-) of the compound, the temperature (-) and flow rate (-) of the air stream, and
14      the presence of competing species (-), including water vapor (Cui et al., 1998; Kamens and Coe,
15      1997; Lane et al., 1988). (The + and - symbols in parentheses indicate qualitatively the effect
16      increasing each parameter would have on efficiency).  In a system with a denuder collection
17      efficiency less than 100%, the collection efficiency must be known to accurately attribute
18      adsorbed organics from denuder breakthrough to the gas phase and adsorbed organics volatilized
19      from collected particles to the particle phase. In calculating the overall phase  distributions of
20      SVOC PAH from a denuder system, the collection efficiency for each compound is needed.
21          The  efficiency of silicone-grease-coated denuders for the collection of polynuclear
22      aromatic hydrocarbons was examined by Coutant et al. (1992), who examined the effects of
23      uncertainties  in the diffusion coefficients and in the collisional reaction efficiencies on the
24      overall phase distributions of SVOC PAH calculated using denuder technology.  In their study,
25      they used  a single stage, silicone-grease-coated aluminum annular denuder with a filter holder
26      mounted ahead of the denuder and an XAD trap deployed downstream of the denuder.  In a
27      series of laboratory experiments, they spiked the filter with a mixture of perdeuterated PAH,
28      swept the  system with ultra-high purity air for several  hours, and then analyzed the filter and the
29      XAD.  They found that the effects of these uncertainties, introduced by using a single compound
30      as a surrogate PAH (in their case, naphthalene) for validation of the denuder collection
31      efficiency, are less significant than normal  variations because of sampling and analytical effects.
32      Results on field studies using their sampling system have not been published.
        June 2003                                2A-1       DRAFT-DO NOT QUOTE OR CITE

-------
 1          For measuring particulate phase organic compounds, the denuder-based sampling system
 2     represents an improvement over the filter/adsorbent collection method (Turpin et al., 1993).
 3     Some researchers, however, have reported that denuder coatings themselves can introduce
 4     contamination (Mukerjee et al., 1997) and that the adsorbed species may be difficult to remove
 5     from the coating (Eatough et al., 1993).
 6          In a study conducted in southern California (Eatough et al., 1995), the Brigham Young
 7     University Organic Sampling System (BOSS; Eatough et al., 1993) was used for determining
 8     POM composition, and a high-volume version (BIG BOSS; flow rate 200 L/min) was utilized
 9     for determining the particulate size distribution and  the chemical composition of SVOC in fine
10     particles. The BOSS, a multi-channel diffusion denuder sampling system, consists of two
11     separate samplers (each operating at 35 L/min).  The first sampler consists of a multi-parallel
12     plate diffusion denuder with charcoal-impregnated filter papers as the collection surfaces
13     followed by a two-stage quartz filter pack and a two-stage charcoal-impregnated filter pack.  The
14     second sampler operating in parallel with the first consists of a two-stage quartz  filter pack,
15     followed by the parallel plate denuder, followed  by the two-stage charcoal-impregnated filter
16     pack. The filter samples collected by the BOSS  sampler were analyzed by temperature-
17     programmed volatilization analysis. The second channel allows calculations of the efficiency of
18     the denuder in removing gas-phase specifics that would be absorbed by the charcoal impregnated
19     filter. Eatough et al. (1995) also operated a two-stage quartz filter pack alongside the BOSS
20     sampler. The BIG BOSS system (Tang et al., 1994) consists of four systems (each with a
21     flowrate of 200 L/min).  Particle size cuts of 2.5, 0.8, and 0.4 jim are achieved by virtual
22     impaction, and the sample subsequently flows through a denuder, then is split, with the major
23     flow (150 L/min) flowing through a quartz filter followed by an  XAD-II bed. The minor flow is
24     sampled through a quartz filter backed by a charcoal-impregnated filter paper. The samples
25     derived from the major flow (quartz filters and XAD-II traps) were extracted with organic
26     solvents and analyzed by gas chromatography (GC) and  GC-mass spectroscopy. The organic
27     material lost from the particles was found to represent all classes of organic compounds.
28          Eatough et al. (1996) operated the BOSS sampler for a year at the IMPROVE site at
29     Canyonlands National Park, UT, alongside the IMPROVE monitor and alongside a separate
30     sampler consisting of a two-stage quartz filter pack. They found that concentrations of
31     particulate carbon determined from the quartz filter  pack sampling system were low on average

       June 2003                                2A-2        DRAFT-DO NOT QUOTE OR CITE

-------
 1      by 39%, and this was attributed to volatilization losses of SVOC from the quartz filters.
 2      In another study conducted with the BOSS in southern California, losses of 35% of the POM, on
 3      average, were found and attributed to losses of the SVOC during sampling (Eatough et al.,
 4      1995).
 5           The denuder used in the various BOSS samplers consists of charcoal-impregnated cellulose
 6      fiber filter material.  Denuder collection efficiencies of greater than 95% have been reported for
 7      organic gases that adsorb on quartz and charcoal-impregnated filters (Eatough et al., 1999a; Ding
 8      et al., 2002; Lewtas et al.,  2001). However, because the mass concentration of gas phase species
 9      that adsorb on quartz and charcoal-impregnated filters is so much greater than the mass of
10      semivolatile organic material in the particulate phase, it is necessary to measure and account for
11      the inefficiency of the denuder in the BOSS samplers. To address this problem, Brigham Young
12      University (BYU) developed a particle-concentrator (PC)-BOSS system (Ding et al., 2002;
13      Eatough et al.,  1999b; Lewtas et al., 2001; Modey et al., 2001; Pang et al., 2001, 2002a,b). The
14      PC-BOSS includes a virtual impactor upstream of the denuder to improve the denuder collection
15      efficiency by removing a majority of the gases from the aerosol flow. With this system,  denuder
16      collection efficiencies of greater than 99% have been reported for organic gases, SO2(g),
17      HNO3(g) and other species that adsorb on quartz and charcoal-impregnated filters (Pang et al.,
18      2001).  Since the concentrations of semivolatile organic and other gases in the presence of the
19      concentrated particles is not altered by this process the virtual impaction concentration of
20      accumulation mode particles (except possibly by slight changes in pressure), it is anticipated that
21      the gas-particle distribution will not be significantly altered by the concentration process. The
22      virtual impactor has  a 50% cut point at 0.1 jim aerodynamic diameter. As a result, some
23      particles in the 0.05 to 0.2 |im diameter size range will be removed in the major flow along with
24      the majority of the gases.  Therefore, the mass collection efficiency of the virtual impactor
25      concentrator will be  a function of the particle size distribution in the 0.05 to 0.1 |im size  range.
26      This collection efficiency is measured by comparing the concentration of nonvolatile
27      components measured in the concentrated sample with that measured in an unconcentrated
28      sample.  The concentration efficiency varies from 50 to 75%.  It is relatively constant over
29      periods of weeks but varies by season and site, presumably as the particle size distribution
30      changes.  Previous studies at Harvard (Sioutas et al., 1995a,b) have shown that the composition
31      of the sampled aerosol is little changed by the  concentration process. The BYU studies listed

        June 2003                                2A-3         DRAFT-DO NOT QUOTE OR CITE

-------
 1      above have shown that the concentration efficiencies for sulfate, organic carbon (OC) and
 2      elemental carbon (EC) are comparable for a given sampling location.  Furthermore, the
 3      concentrations of these species and of fine particulate nitrate determined using the PC-BOSS
 4      have been shown to be comparable to those determined using more conventional samplers for
 5      sulfate or EC or using simpler denuder systems for OC and nitrate.
 6           Ding et al. (1998a) developed a method for the determination of total n-nitroso compounds
 7      in air samples and used the method to examine organic compounds formed from NOX chemistry
 8      in Provo, UT (Ding et al., 1998b).  In their method, n-nitroso compounds are selectively
 9      decomposed to yield nitric oxide, which is then detected using chemiluminescence. From the
10      samples from Provo, they found that the majority of the n-nitroso and nitrite organic compounds
11      that were present in fine particulate matter were semivolatile organic compounds that could be
12      evaporated from the particles during sampling. They found particulate n-nitroso compound
13      concentrations ranging between <1 and 3  nmoles/m3  and gas-phase n-nitroso compound
14      concentrations in the same range. Particulate organic nitrite concentrations were found in the
15      range of <1 to =5 nmoles/m3, and gas-phase concentrations as high as 10 nmoles/m3 were found.
16           Turpin et al. (1993) developed a sampling system that corrects for the loss of semivolatile
17      organic compounds during sampling by removal of most of the gas phase material from the
18      particles in a diffusion separator sampling system. Unlike the previously mentioned systems,
19      wherein the particulate phase is measured directly, in the system of Turpin et al. (1993) the
20      gas-phase is measured directly. In the laminar flow system, ambient, particle-laden air enters the
21      sampler as an annular  flow.  Clean, particle-free air is pushed through the core inlet of the
22      separator. The clean air and ambient aerosol join downstream of the core inlet section, and flow
23      parallel to each other through the diffusion zone. Because of the much higher diffusivities for
24      gases compared to particles, the SVOC in the ambient air diffuses to the clean, core flow. The
25      aerosol exits the separator in the annular flow, and the core flow exiting the separator now
26      contains a known fraction of the ambient  SVOC. Downstream of the diffusion separator, the
27      core exit flow goes into a polyurethane foam (PUF) plug, where the SVOC is collected.  The
28      adsorbed gas phase on the PUF plug is extracted with supercritical fluid CO2 and analyzed by
29      gas chromatography/mass-selective detection (GC/MSD).  The gas-phase SVOC is thus
30      determined.  Ultimately, to determine particulate phase SVOC concentrations, the total
31      compound concentration will also be measured and the particulate phase obtained  by difference.

        June 2003                                2A-4        DRAFT-DO NOT QUOTE OR CITE

-------
 1      The system was tested for the collection of PAH. The diffusional transport of gas-phase PAHs
 2      and particle concentrations agreed well with theory. Breakthrough was problematic for low
 3      molecular weight PAHs (MW < 160). Detection limits ranged from 20 to 50 pg of injected mass
 4      for all PAHs.
 5           Gundel et al. (1995) recently developed a technique for the direct determination of phase
 6      distributions of semivolatile polycyclic aromatic hydrocarbons using annular denuder
 7      technology. The method, called the integrated organic vapor/particle sampler (IOVPS), uses a
 8      cyclone inlet with a 50% cutpoint of 2.5 jim at a sampling rate of 10 L/min. The airstream then
 9      goes through two or three sandblasted glass annular denuders that are coated with ground
10      adsorbent resin material (XAD-4 was initially examined) that traps vapor-phase organics. The
11      airstream subsequently passes through a filter, followed by a backup denuder. The denuder
12      collection efficiency is high and compares well with predictions based on the diffusivity of the
13      compounds. The denuder can also be extracted to obtain gas-phase concentrations directly
14      (Gundel  and Lane, 1999). Particle-phase PAHs are taken to be the sum of material on the filter
15      and XAD adsorbent downstream after correction for denuder collection efficiency. The IOVPS
16      was tested for sampling semivolatile PAH in laboratory indoor air and in environmental tobacco
17      smoke (ETS).  After exposure, the denuders, filters, and sorbent traps were extracted with
18      cyclohexane (Gundel et al., 1995) and analyzed for PAHs from naphthalene to chrysene using
19      dual-fluorescence detection (Mahanama et al., 1994). Recoveries from both denuders and filters
20      were approximately 70% for 30 samples. Detection limits (defined as 3 times the standard
21      deviation of the blanks) for gas-phase SVOC PAHs ranged from 0.06 ng for anthracene to 19 ng
22      for 2-methylnaphthalene. The 95% confidence interval for reproduction of an internal standard
23      concentration was 6.5% of the mean value. Relative precision,  from a propagation of errors
24      analysis  or from the 95% confidence interval from  replicate analyses of standard reference
25      material  SRM 1649 (urban dust/organics), was 12% on average (8% for naphthalene to 22% for
26      fluorene).  Sources of error included sampling flow rate, internal standard concentration, and
27      co-eluting peaks.  Gundel and Lane (1999) reported that roughly two-thirds of particulate PAH
28      fluoranthene, pyrene, benz[a]anthracene, and chrysene were found on the postfilter denuders, so
29      that it is  likely that considerable desorption from the collected particles took place.
30           Solid adsorbent-based denuder systems have been investigated by other researchers as
31      well. Bertoni et al.  (1984) described the development of a charcoal-based denuder system for

        June 2003                                 2A-5        DRAFT-DO NOT QUOTE OR CITE

-------
 1      the collection of organic vapors.  Risse et al. (1996) developed a diffusion denuder system to
 2      sample aromatic hydrocarbons.  In their system, denuder tubes with charcoal coating and
 3      charcoal paper precede a filter pack for particulate collection and an adsorption tube to capture
 4      particle blow-off from the filter sample. Breakthrough curves for benzene, toluene, ortho-
 5      xylene, and meta-xylene were developed for 60-, 90-, and 120-cm denuder tubes.  The effects of
 6      relative humidity on the adsorption capacities of the denuder system were examined, and it was
 7      found that the capacity of the charcoal was not affected significantly by increases in relative
 8      humidity. The feasibility of outdoor air sampling with the system was demonstrated.
 9           Krieger and Kites (1992) designed a diffusion denuder system that uses capillary gas
10      chromatographic columns as the tubes for SVOC collection.  The denuder was followed by a
11      filter to collect particles, which in turn was followed by a PUF plug to collect organic material
12      volatilizing off the filter.  Denuder samples were analyzed by liquid solvent extraction (CH2C12)
13      followed by GC-mass spectrometric analysis. The PUF plugs and filters were extracted with
14      supercritical fluid extraction using supercritical N2O. Using this system, an indoor air sample
15      was found to contain primarily chlorinated biphenyls, ranging from trichlorobiphenyls (vapor
16      pressures 10"3 - 10"4 Torr at 25 °C) to octachlorobiphenyls (10"6 - 10"7  Torr).  This  demonstrated
17      that the sampler collects compounds with a wide range of volatility. They also found that on-
18      line desorption is successful in maintaining good chromatographic peak shape and resolution.
19      The entire method, from sample collection to the end of the chromatographic separation, took
20      2h.
21           Organic acids in both the vapor and particulate phases may be important contributors to
22      ambient acidity, as well as representing an important fraction of organic particulate matter.
23      Lawrence and Koutrakis (1996a,b) used a modified Harvard/EPA annular denuder system
24      (HEADS) to sample both gas and particulate phase organic acids in Philadelphia, PA, in the
25      summer of 1992.  The HEADS sampler inlet had a 2. l-|im cutpoint impactor (at 10 L/min),
26      followed by two denuder tubes, and finally a filter pack with a Teflon filter.  The first denuder
27      tube was coated with potassium hydroxide (KOH) to trap gas phase organic acids. The second
28      denuder tube was coated with citric acid to remove ammonia and thus to avoid neutralizing
29      particle phase acids collected on the filter.  The KOH-coated denuder  tube was reported to
30      collect gas phase formic and acetic acids at better than 98.5% efficiency and with precisions of
31      5% or better (Lawrence and Koutrakis, 1994).  It was noted that for future field measurements of

        June 2003                                 2A-6        DRAFT-DO NOT QUOTE OR CITE

-------
 1      particulate organic acids, a Na2CO3-coated filter should be deployed downstream of the Teflon
 2      filter to trap organic acids that may evaporate from the Teflon filter during sampling.
 3
 4      Role of the Collection Media
 5          The role of the collection media was recently examined in a study conducted in Seattle
 6      (Lewtas et al., 2001). In that study, the influence of denuder sampling methods and filter
 7      collection media on the measurement of SVOC associated with PM2 5 was evaluated.  Activated
 8      carbon and XAD collection media were used in diffusion denuders and impregnated back-up
 9      filters  in two different samplers, the Versatile Air Pollution Sampler (VAPS) and the PC-BOSS.
10      XAD-coated glass annular denuders and charcoal-impregnated cellulose fiber (GIF) filter
11      denuders also were used. GIF filters also were compared to XAD-coated quartz filters as backup
12      filter collection media.  Lewtas et al. (2001) found that the two denuder types resulted in an
13      equivalent measurement of parti culate organic carbon and particle mass. The carbon-coated
14      denuders in the BOSS sampler were more efficient than the XAD-coated denuders for the
15      collection of the more highly volatile organic compounds (MHVOC).  Lewtas et al. (2001)
16      concluded that this MHVOC that is collected in the carbon-coated BOSS denuder does not
17      contribute substantially to the particle mass or to the SVOC measured as OC on quartz filters.
18      However, this MHVOC would be captured in the carbon impregnated filters placed behind the
19      quartz filters, so that, in the XAD denuder configuration, the captured MHVOC would cause a
20      higher OC concentration and an overestimation of the SVOC.
21          Some of the recent research in denuder technology also has focused on reduction in the
22      size of the denuder, optimization of the residence time in the denuder, understanding the effect
23      of diffusion denuders on the positive quartz filter artifact, identifying changes in chemical
24      composition that occur during sampling, determining the effects of changes in temperature and
25      relative humidity, and identifying possible losses by absorption in coatings.
26
27      Reducing the Size of Denuders
28          The typical denuder configuration is an annular diffusion denuder tube of significant length
29      (e.g., 26.5  cm for 10 L/min [Koutrakis et al., 1988a,b]).  A more compact design based on a
30      honeycomb configuration was shown to significantly increase the capacity (Koutrakis et al.,
31      1993). However, in intercomparisons with an annular denuder/filter pack system (Koutrakis

        June 2003                                 2A-7        DRAFT-DO NOT QUOTE OR CITE

-------
 1      et al., 1988), significant losses of ammonia and nitric acid were observed for the honeycomb
 2      configuration and were attributed to the large inlet surface area and long sample residence time
 3      of the honeycomb design relative to the annular denuder system.  Sioutas et al. (1996a)
 4      subsequently designed a modified glass honeycomb denuder/filter pack sampler (HDS) with an
 5      inlet that minimizes vapor losses on the inlet surfaces. The modified HDS has reduced inlet
 6      surfaces and decreased residence time for sampled gases (NH3 and HNO3) compared to its
 7      predecessor (Sioutas et al., 1994d).  Sioutas et al. (1996b) also tested various inlet materials
 8      (glass, PFA, and polytetrafluoroethylene [PTFE]) in laboratory tests and found that a PTFE
 9      Teflon coated inlet minimized loss of sampled gases (1 to 8% loss of HNO3 observed, and -4 to
10      2% loss of NH3 observed). The highest inlet losses were observed for FINO3 lost to PFA
11      surfaces (14 to 25%).  The modified FIDS was tested in laboratory and field tests and found to
12      agree within 10% with the annular denuder system.
13
14      Residence Time in the Denuder
15          The efficiency of a diffusion denuder sampler for the removal of gas phase material can be
16      improved by increasing the residence time of the sampled aerosol in the denuder. However, the
17      residence time can only be increased within certain limits.  Because the diffusion denuder
18      reduces the concentration of gas-phase semivolatile organic material, semivolatile organic matter
19      present in the particles passing through the denuder will be in a thermodynamically unstable
20      environment and will tend to outgas SVOC during passage through the denuder.  The residence
21      time of the aerosol in the denuder, therefore, should be short enough to prevent significant loss
22      of particulate phase SVOC to the denuder. Various studies have suggested that the residence
23      time in the denuder should be less than about 2 s (Gundel and Lane, 1999; Kamens and Coe,
24      1997; Kamens et al., 1995).  The residence times in the various denuder designs described by
25      Gundel and Lane (1999) are from 1.5 to 0.2 s. The equilibria and evaporation rates are not as
26      well understood for organic components as they are for NH4NO3 (Zhang and McMurry, 1987,
27      1992; Hering and Cass, 1999).
28
29      Effect of Diffusion Denuders on the Positive Quartz Filter Artifact
30          The adsorption of organic compounds by a second quartz filter has been shown to be
31      reduced, but not eliminated, in samples collected in the Los Angeles Basin if a multi-channel

        June 2003                               2A-8       DRAFT-DO NOT QUOTE OR CITE

-------
 1      diffusion denuder with quartz filter material as the denuder collection surface preceded the
 2      quartz filters (Fitz, 1990). This artifact can be further reduced by the use of activated charcoal as
 3      the denuder surface and the use of a particle concentrator to reduce the amount of gas phase
 4      organic compounds relative to condensed phase organic compounds (Cui et al., 1997, 1998;
 5      Eatough, 1999). Other experiments (Cotham and Bidleman, 1992; Cui et al., 1998; Eatough
 6      et al., 1995, 1996) have shown that the quartz filter artifact can result both from the collection of
 7      gas phase organic compounds and from the collection of semivolatile organic compounds lost
 8      from particles during sampling. Thus, results available to date suggest that both a "positive" and
 9      a "negative" artifact can be present in the determination of particulate phase organic compounds
10      using two tandem quartz filters.
11           The importance of the adsorption of organic vapors on filters or PM relative to the
12      volatilization of organic compounds from PM collected on a filter continues to be a topic of
13      active debate. The relative importance of positive and negative artifacts will be different for
14      denuded and undenuded filters; will depend on face velocity, sample loading, and the vapor
15      pressures of the compounds of interest; and may vary with season and location because of
16      variations in the composition of volatile and semivolatile organic material.  Evidence exists for
17      substantial positive and negative artifacts in the collection of organic PM.
18
19      Changes in Chemical Composition During Sampling
20           The use of sampling systems designed to correctly identify the atmospheric gas and
21      parti culate phase distributions of collected organic material has been outlined above.
22      An additional sampling artifact that has received little consideration in the collection of
23      atmospheric samples is the potential alteration of organic compounds as a result of the sampling
24      process. These alterations appear to result from the movement of ambient air containing
25      oxidants and other reactive compounds past the collected particles.  The addition of NO2
26      (
-------
 1      of a deuterated amine on a filter to NOX (Pellizzari and Krost, 1984). When Tenax columns
 2      spiked with deuterated styrene and cyclohexene were exposed to ppm concentrations of ozone or
 3      halogens, oxygenated and halogenated compounds were formed (Pellizzari and Krost,  1984).
 4      Similar oxidation of aldehydes and peroxyacetylnitrate (PAN) during sampling has been
 5      observed (Grosjean and Parmar, 1990). Collected PAH compounds can be oxygenated or
 6      nitrated on a filter (Davis et al., 1987; Lindskog and Brorstrom-Lunden, 1987), but 1-nitropyrene
 7      has been shown to be resistant to additional nitration (Grosjean, 1983). These various  chemical
 8      transformations of collected organic compounds can be eliminated by removal of the gas phase
 9      oxidants, NOX, HNO3, etc., by reaction or adsorption prior to collection of the particles (Ding,
10      1998a,b; Grosjean and Parmar, 1990; Parmar and Grosjean, 1990; Pellizzari and Krost, 1984;
11      Williams and Grosjean, 1990).  The BOSS denuder should be effective in eliminating most of
12      the chemical transformation artifacts because reactive gases are removed by the charcoal
13      denuder that precedes the particle collection filter.
14
15      Temperature and Relative Humidity Effects
16           The problems of sampling artifacts associated with SVOC adsorption and evaporation are
17      compounded by temperature and relative humidity effects (Pankow and Bidleman, 1991;
18      Pankow et al., 1993; Falconer et al., 1995; Goss and Eisenreich, 1997). Effects of temperature
19      on the partitioning of PAH were examined by Yamasaki et al. (1982), who found that the
20      partition coefficient (PAHvapor/PAHpart) was inversely related to temperature and could be
21      described using the Langmuir adsorption concept. The dissociation of ammonium nitrate aerosol
22      is also a function of temperature.  Bunz et al. (1996) examined the dissociation and subsequent
23      redistribution of NH4NO3 within a bimodal distribution using a nine-stage low-pressure Berner
24      impactor followed by analysis by ion chromatography and found a strong temperature
25      dependency on the redistribution. Bunz et al. (1996) found that at lower temperatures  (below
26      10 °C) there was little change in the aerosol size distribution.  At temperatures between 25 and
27      45 °C, however, the lifetime of NH4NO3 particles decreases by more than a factor of 10, and size
28      redistribution, as measured by average ending particle diameter, increased more for higher
29      temperatures than for lower temperatures.
30           The effects of relative humidity on the sorption of SVOC on particles are not well
31      understood. In a series of laboratory experiments, Goss and Eisenreich (1997) examined the

        June 2003                                2A-10      DRAFT-DO NOT QUOTE OR CITE

-------
 1      sorption of both nonpolar (hydrocarbons and chlorinated hydrocarbons) and polar (ethyl ether
 2      and acetone) volatile organic compounds onto combustion soot particles as a function of
 3      temperature and relative humidity.  The soot particles used in their experiments were collected
 4      from oil furnaces and contained 60% (w/w) iron sulfate (water-soluble fraction) and 9% (w/w)
 5      elemental and organic carbon.  They found that, for all compounds, the sorption of VOC onto
 6      soot particles decreased with increasing relative humidity over the range of 10 to 95%. They
 7      also observed hysteresis in the relative humidity dependency, with sorption coefficients at a
 8      given relative humidity higher when the RH is being increased than when the RH is being
 9      decreased. The sorption coefficients were fit with an exponential function to the RH so that the
10      slope of the regression line would provide a  measure of the influence of relative humidity.
11      Based on the magnitude of the slope, they concluded that the RH-dependency of sorption was
12      stronger for water-soluble organic compounds.
13          In another study by Jang and Kamens (1998), humidity effects on gas-particle partitioning
14      of SVOC were examined using outdoor environmental chambers and the experimentally
15      determined partitioning coefficients were compared to theoretical values. They examined the
16      partitioning of SVOC onto wood soot, diesel soot, and secondary aerosols and concluded that
17      "the humidity effect on partitioning was most significant for hydrophobic compounds adsorbing
18      onto polar aerosols." Although these two studies seem to be contradictory, on closer
19      examination, it is difficult to compare the two studies for several reasons. The experiments
20      conducted by Jang and Kamens (1998) were conducted in outdoor chambers at ambient
21      temperatures and humidities.  Their model was for absorptive partitioning of SVOC on
22      liquid-like atmospheric particulate matter. In contrast, the results of Goss and Eisenreich (1997)
23      were obtained from a gas chromatographic system operated at 70 °C higher than ambient
24      conditions.  The model of Goss and Eisenreich (1997) was for adsorptive partitioning of VOC on
25      solid-like atmospheric parti culate matter. In the study of Jang and Kamens (1998),  calculated
26      theoretical values for water activity coefficients for diesel soot were based on an inorganic salt
27      content of 1 to 2%; whereas, the combustion particles studied by Goss and Eisenreich (1997)
28      contained 60% water-soluble, inorganic salt content. Jang and Kamens (1998) obtained their
29      diesel  soot from their outdoor chamber, extracted it with organic solvent (mixtures of hexane and
30      methylene chloride), and measured the organic fraction.  The resulting salt content of 2% of the
        June 2003                                2A-11       DRAFT-DO NOT QUOTE OR CITE

-------
 1      particulate matter studied in Jang and Kamens (1998) is enough to affect water uptake but
 2      presumably not to affect the sorption partitioning of organics.
 3
 4      Impactor Coatings
 5           Impactors are used as a means to achieve a size cutpoint and as particle collection surfaces.
 6      Particles collected on impactors are exposed to smaller pressure drops than filter-collected
 7      particles, making them less susceptible to volatile losses (Zhang and McMurry, 1987).
 8      However, size resolution can be affected by bounce when samples are collected at low
 9      humidities (Stein et al.,  1994). There are other sources of error inherent in some of the currently
10      acceptable practices that could potentially affect particulate mass concentration measurements
11      and that will surely become even more important as more emphasis is placed on chemical
12      speciation.  Allen et al. (1999a) reported that the practice of greasing impaction substrates may
13      introduce an artifact from the absorption of semivolatile species from the gas phase by the grease
14      because the grease could artificially increase the amount of PAHs and other organic compounds
15      attributed to the aerosol. Allen et al. (1999b) offer several criteria to ensure that this absorption
16      artifact is negligible, including selecting impaction oils in which analytes of interest are
17      negligibly soluble and ensuring that species do not have time to equilibrate between the vapor
18      and oil phases (criterion is met for nonvolatile species).  They recommend using oiled impaction
19      substrates only if the absorption artifact is negligible as determined from these criteria.
20      Application of greases and impaction oils for preventing or reducing bounce when sampling with
21      impactors is not suitable for carbon analysis because the greases contain carbon (Vasilou et al.,
22      1999).
23           Kavouras and Koutrakis (2001) investigated the use of polyurethane foam (PUF) as a
24      substrate for conventional inertial impactors. The PUF impactor substrate is not rigid like the
25      traditional impactor substrate so particle bounce and reentrainment artifacts are reduced
26      significantly. Kavouras and Koutrakis (2001) found that the PUF impaction substrate resulted in
27      a much smaller 50% cut point at the same flow rate and Reynolds number.  Moreover, the lower
28      50%  cut point was obtained  at a lower pressure drop than with the conventional substrate, which
29      could lead to a reduction of artifact vaporization of semivolatile components.
30
        June 2003                                2A-12        DRAFT-DO NOT QUOTE OR CITE

-------
 1                     APPENDIX 2B.  ANALYTICAL TECHNIQUES
 2
 3
 4     2B.1 INORGANIC ELEMENTS
 5     2B.1.1  Energy Dispersive X-Ray Fluorescence (EDXRF)
 6          EDXRF has usually been the method of choice for analysis of trace elements on filters.
 7     EDXRF is preferred for aerosol analysis over wavelength dispersive XRF because it allows fast
 8     and simultaneous analysis over the total spectrum, allowing for the analysis of numerous
 9     elements simultaneously. EDXRF can accommodate small sample sizes and requires little
10     sample preparation or operator time after the samples are placed into the analyzer. It also leaves
11     the sample intact after analysis; so, further analysis is possible. XRF irradiates a uniform
12     particle deposit on the surface of a membrane filter with 1 to 50 kev x-rays that eject inner shell
13     electrons from the atoms of each element in the sample (Dzubay and Stevens, 1975; Jaklevic et
14     al., 1977; Billiet et al., 1980; Potts and Webb, 1992; Piorek, 1994; Bacon et al., 1995; deBoer et
15     al., 1995; Holynska et al., 1997; Torok et al., 1998; Watson et al., 1999). When a higher energy
16     electron drops into the vacant lower energy orbital, a fluorescent x-ray photon is released.  The
17     energy of this photon is unique to each element, and the number of photons is proportional to the
18     concentration of the  element.  Concentrations are quantified by comparing photon counts for a
19     sample with those obtained from thin-film standards  of known concentration (Dane et al., 1996).
20     The previous  1996 PM AQCD included a detailed discussion of EDXRF.
21          Emitted x-rays with energies less than ~4 kev (affecting the elements sodium, magnesium,
22     aluminum, silicon, phosphorus, sulfur, chlorine, and potassium) are absorbed in the filter, in a
23     thick particle deposit, or even by large particles in which these elements are contained.  Very
24     thick filters also scatter much of the excitation radiation or protons, thereby lowering the
25     signal-to-noise ratio  for XRF and PIXE. For this reason, thin membrane filters with deposits in
26     the range of 10 to 50 jig/cm2 provide the best accuracy and precision for XRF and PIXE analysis
27     (Davis et al., 1977; Haupt et al., 1995).
28
29     2B.1.2  Synchrotron Induced X-ray Fluorescence (S-XRF)
30          S-XRF is a form of EDXRF in which the exciting x-rays are derived from a synchrotron.
31     Bremmstrahlung x-rays are generated when energetic electrons (generally several GeV in

       June 2003                               2B-1       DRAFT-DO NOT QUOTE OR CITE

-------
 1      energy) are forced by a magnetic field to make a bend in their path. The advantages of the
 2      technique are that an extremely high flux of x-rays may be obtained and that the x-rays are 100%
 3      polarized in the plane of the electron beam. The former allows for x-ray beams generally of
 4      50 to 500 |im in diameter.  However, the beams can be focused into x-ray microprobes, with spot
 5      sizes on the order of one |im diameter.  The x-ray polarization allows for removal of most of the
 6      background normally found under the characteristic x-ray peaks, greatly improving sensitivity
 7      compared to other XRF techniques. The primary disadvantages are the limited number of
 8      synchrotrons, and that few synchrotrons have S-XRF capabilities.  Thus, the technique has been
 9      relatively little used for PM, and then generally for special problems such as the smoke from the
10      Kuwaiti oil fires (Cahill et al., 1992, Reid et al., 1994). However, with the increasing
11      availability of S-XRF facilities dedicated to PM analysis, the first of which was the Advanced
12      Light Source opened at Lawrence Berkeley National Laboratory last year, utilization of S-XRF
13      for PM analysis is increasing.
14
15      2B.1.3  Proton (or Particle) Induced X-ray Emission (PIXE)
16           PIXE differs from XRF analysis in the excitation source for producing fluorescence.  The
17      filter deposit is bombarded with high-energy protons to remove  inner shell electrons and  the
18      resulting characteristic x-rays are analyzed as in XRF (Johansson, 1970, Cahill, 1981, 1985;
19      Zeng et al., 1993). Small accelerators, generally Van de Graaffs, generate intense beams of low
20      energy protons, generally of a few MeV in energy. These have the ability to remove electrons
21      from inner shells of atoms of any element.  Thus, PIXE can see a very wide range of elements in
22      a single analysis.  The cross section for producing x-rays using protons of a few MeV in energy
23      tends to favor lighter elements, Na through Ca, but sensitivities for equivalent PIXE and
24      multi-wavelength XRF analysis are roughly comparable.  The technique has been widely used in
25      the U.S. (Flocchini et al.,  1976, Malm et al., 1994) and around the world, as many universities
26      have the small accelerators needed for the method. Like S-XRF, the proton beams can be
27      focused into |im size beams, but these have been relatively little used for PM. However,  the mm
28      size  beams used in both S-XRF and PIXE are well suited to analyzing the limited mass and small
29      deposits that result from detailed particle size measurements by  impactors (Perry et al., 1999).
30
31

        June 2003                                2B-2        DRAFT-DO NOT QUOTE OR CITE

-------
 1     2B.1.4  Proton Elastic Scattering Analysis (PESA)
 2          With the routine availability of elemental analyses for all elements sodium and heavier,
 3     organic components remain the major unmeasured species for mass balance. For programs like
 4     IMPROVE (Malm et al., 1994), parallel filters are collected for separate organic and elemental
 5     carbon determinations.  Aerosol programs that use PIXE can directly measure hydrogen
 6     simultaneously by scattering protons from Teflon filters that lack hydrogen (Cahill et al., 1989,
 7     1992). Generally, results from organic matter by carbon combustion from quartz filters and
 8     organic matter by hydrogen from Teflon filters are in agreement, assuming certain assumptions
 9     about the chemical states of sulfates and nitrates are met (Malm et al., 1994, Cahill et al., 1996).
10
11     2B.1.5  Total Reflection X-ray Fluorescence (TRXRF)
12          One of the limitations of the EDXRF method is the minimum detection limit, which may
13     be high due to high background values (Streit et al.,  2000).  By implementation of x-ray optical
14     geometries that use the total reflection of the primary radiation on flat surfaces, scattering on the
15     substrate is reduced, so that detection limits can be reduced. This is the basis for the total
16     reflection x-ray fluorescence (TRXRF) method (Aiginger and Streli, 1997).  This modification to
17     the EDXRF technique improves detection limits and avoids the need to correct for matrix
18     effects.  Despite its apparent advantages, TRXRF has not yet become widely in use for
19     atmospheric aerosol analysis but has been used in the analysis of marine aerosol (Stahlschmidt et
20     al., 1997) and at a high elevation site (Streit et al., 2000). Streit et al. sampled ambient air at the
21     High Alpine Research Station (3580 m above sealevel) in the Bernese Alps, Switzerland, using a
22     nine-stage, single-jet, low-pressure, cascade impactor equipped with quartz impactor plates
23     coated with silicon oil diluted in 2-propanol.  The typical sample volume for a weekly sample
24     was 10 m3. The quartz plates were analyzed directly by TRXRF. Streit et al. reported that the
25     minimum detection limits, defined by the 3 a values of the blanks, ranged from 25  ng for S,
26     decreased monotonically with increasing atomic number down to 5 pg for Rb, and decreased
27     after that. The use of TRXRF is expected to increase as EDXRF users become aware of the
28     method.  A relatively low-cost, add-on unit has been developed that would allow EDXRF users
29     to test the TRXRF technique (Aiginger, 1991).
30
31

       June 2003                                2B-3        DRAFT-DO NOT QUOTE OR CITE

-------
 1      2B.1.6  Instrumental Neutron Activation Analysis (INAA)
 2           INAA irradiates a sample in the core of a nuclear reactor for few minutes to several hours,
 3      depending on the elements being quantified (Dams et al., 1970; Zoller and Gordon, 1970;
 4      Nadkarni, 1975; Landsberger, 1988; Olmez, 1989; Ondov andDivita, 1993). The neutron
 5      bombardment chemically transforms many elements into radioactive isotopes.  The energies of
 6      the gamma rays emitted by these isotopes identify them and, therefore, their parent elements.
 7      The intensity of these gamma rays is proportional to the amount of the parent element present in
 8      the sample.  Different irradiation times and cooling periods are used before counting with a
 9      germanium detector. In source apportionment studies, it is possible to use a combination of XRF
10      and INAA to develop a relatively complete set of elemental measurements. Between these two
11      analytical techniques, good sensitivity is possible for many elements, including most of the toxic
12      metals of interest.  In general, XRF provides better sensitivity for some metals (e.g., Ni, Pb, Cu,
13      and Fe); whereas INAA provides better sensitivity for others  (Sb, As, Cr, Co, Se, and Cd). Both
14      methods provide similar detection limits for still other elements (V, Zn, and Mn). INAA does
15      not quantify some of the abundant species in ambient particulate matter such as silicon, nickel,
16      tin, and lead. While INAA is technically nondestructive, sample preparation involves folding
17      the sample tightly and sealing it in plastic, and the irradiation process makes the filter membrane
18      brittle and radioactive. These factors limit the use of the sample for subsequent analyses.
19           INAA has been used to examine the chemical composition of atmospheric aerosols in
20      several studies either as the only method of analysis or in addition to XRF (e.g., Yatin et al.,
21      1994; Gallorini, 1995). INAA has higher sensitivity for many trace species,  and it is particularly
22      useful in analyzing for many trace metals. Landsberger and Wu (1993) analyzed air samples
23      collected near Lake Ontario for  Sb, As,  Cd, In, I, Mo, Si, and V using INAA. They
24      demonstrated that using INAA in conjunction with epithermal neutrons and Compton
25      suppression produces very precise values with relatively low  detection limits.
26           Enriched rare-earth isotopes have been analyzed via INAA and used to trace sources of
27      parti culate matter from a coal-fired power plant (Ondov et al., 1992), from various sources in the
28      San Joaquin Valley (Ondov,  1996), from intentially tagged (iridium) diesel emissions from
29      sanitation trucks (Suarez et al., 1996; Wu et al.,  1998), and from iridium-tagged emissions from
30      school buses (Wu et al., 1998).
        June 2003                                 2B-4        DRAFT-DO NOT QUOTE OR CITE

-------
 1           An intercomparison was conducted in which 18 pairs of filters were sent to participants in
 2     the Coordinated Research Program (CRP) on Applied Research on Waste Using Nuclear Related
 3     Analytical Techniques (Landsberger et al., 1997).  As part of that study, participants used PIXE,
 4     INAA, XRF, or AAS to analyze the samples. Many of the results for XRF and PIXE in the
 5     coarse fraction were observed to be biased low compared to INAA. The authors speculated that
 6     there is a systematic error because of self-attenuation of the x-rays resulting from the particle
 7     size effect.
 8
 9     2B.1.7  Atomic Absorption Spectrophotometry (AAS)
10           AAS is applied to the residue of a filter extracted in a strong solvent to dissolve the solid
11     material; the filter or a portion of it is also dissolved during this process (Ranweiler and Moyers,
12     1974; Fernandez, 1989; Jackson and Mahmood, 1994; Chow et al., 2000a). A few milliliters of
13     this extract are injected into a flame where the elements are vaporized. Elements absorb light at
14     certain wavelengths in the visible spectrum, and a light beam with wavelengths  specific to the
15     elements being measured is directed through the flame to be detected by a monochrometer. The
16     light absorbed by the flame containing the extract is compared with the absorption from known
17     standards to quantify the elemental concentrations.  AAS requires an individual  analysis for each
18     element, and a large filter or several filters are needed to obtain concentrations for a large variety
19     of elements.  AAS is a useful complement to other methods, such as XRF and PIXE, for species
20     such as beryllium, sodium, and magnesium which are not well-quantified by these methods.
21     Airborne particles are chemically complex and do not dissolve easily into complete solution,
22     regardless of the strength of the solvent. There is always a possibility that insoluble residues are
23     left behind and that soluble species may co-precipitate on them or on container walls.
24           AAS was used to characterize the atmospheric deposition of trace elements Zn,  Ni, Cr, Cd,
25     Pb, and Hg to the Rouge River watershed by particulate deposition (Pirrone and Keeler, 1996).
26     The modeled deposition rates were compared to annual emissions of trace elements that were
27     estimated from the emissions inventory for coal and oil combustion utilities, iron and steel
28     manufacturing, metal production, cement manufacturing, and solid waste and sewage sludge
29     incinerators.  They found generally good agreement between the trend observed in atmospheric
30     inputs to the river (dry + wet deposition) and annual emissions of trace elements, with R2s
31     varying from =0.84 to 0.98. Both atmospheric inputs and emissions were found to have

       June 2003                                2B-5       DRAFT-DO NOT QUOTE OR CITE

-------
 1     followed downward trends for Pb. For the period of 1987 to 1992, steady increases were
 2     observed for Cd (major sources are municipal solid waste incineration, coal combustion, sludge
 3     incineration, and iron and steel manufacturing), Cr and Ni (major sources are iron and steel
 4     production and coal combustion), and Hg (major sources are coal, the contribution from which
 5     had decreased from 53  to 45%, and municipal, solid, and medical waste incineration, the
 6     contribution from which has increased).
 7
 8     2B.1.8  Inductively Coupled Plasma with Atomic Emission Spectroscopy
 9              (ICP-AES)
10          ICP-AES introduces an extracted sample into an atmosphere of argon gas seeded with free
11     electrons induced by high voltage from a surrounding Tesla coil (Fassel and Kniseley, 1974;
12     McQuaker et al., 1979; Lynch et al., 1980; Harman, 1989; Tyler, 1992; Baldwin et al., 1994).
13     The high temperatures  in the induced plasma raise valence electrons above their normally stable
14     states.  When these electrons return to their stable states, a photon of light is emitted that is
15     unique to the element that was excited. This light is detected at specified wavelengths to identify
16     the elements in the sample.  ICP-AES acquires a large number of elemental concentrations using
17     small sample volumes with acceptable detection limits for atmospheric samples. As with AAS,
18     this method requires complete extraction and destruction of the sample.
19
20     2B.1.9  Inductively Coupled Plasma with Mass Spectroscopy (ICP-MS)
21          ICP-MS has  been applied in the analysis of personal exposure samples (Tan and Horlick,
22     1986; Gray and Williams, 1987a,b; Nam et al., 1993; Munksgaard and Parry, 1998; Campbell
23     and Humayun, 1999).  Ion species generated from ICP and from the sample matrix can produce a
24     significant background at certain masses resulting in formation of polyatomic ions that can limit
25     the ability of ICP-MS to determine some elements of interest.  Cool plasma techniques have
26     demonstrated the potential to detect elements at the ultra-trace level (Nham et al., 1996) and to
27     minimize common molecular ion interferences (Sakata and Kawabata, 1994;  Turner, 1994;
28     Plantz, 1996). Detection limits of ICP-MS using a one-second scan are typically in the  range of
29     10"3 ng/m3, which is an order of magnitude lower than  other elemental analysis methods. The
30     instrument can also be  set up to analyze a wide dynamic range of aerosol concentrations.
       June 2003                               2B-6       DRAFT-DO NOT QUOTE OR CITE

-------
 1      Isotope analysis can also be performed with ICP-MS. Intercomparison studies are needed to
 2      establish the comparability of ICP-MS with other non-destructive filter analysis methods.
 3           Keeler and Pirrone (1996) used ICP-MS to determine trace elements Cd, Mn, V, As, Se,
 4      and Pb in atmospheric fine particulate (PM2 5) and total suspended particulate samples collected
 5      in two Detroit sites. The results were used in a deposition model with additional measurements
 6      using AAS to estimate the dry deposition flux of trace elements to Lake Erie.
 7
 8      2B.1.10  Scanning Electron Microscopy (SEM)
 9           Mamane et al. (2001) investigated the use of computer-controlled scanning electron
10      microscopy (CCSEM) as a way of supplementing XRF analysis and providing automated
11      analysis of particle size, chemistry, and particle classification.  An ambient coarse particulate
12      sample from Baltimore was collected on a polycarbonate filter for this analysis. CCSEM
13      analyses were conducted for 2819 particles in 78 randomly selected fields of view during an
14      unattended 8-h run. Mamane et al. confirmed the stability of the CCSEM instrument over
15      several hours of operation. The physical properties of the sample such as particle diameter, mass
16      loading per field, and particle number per field were well represented by analyzing
17      approximately 360 particles with  little additional information gained by  analyzing more
18      particles. Teflon filters are not well suited for SEM analyses. Analysis  of fine PM is expected
19      to pose  analytical challenges not addressed in the present study (Mamane et al., 2001).
20           Nelson et al. (2000) applied Raman chemical imaging and SEM (Raman/SEM) to study the
21      size, morphology, elemental and molecular composition, and molecular  structure of fine
22      particulate  matter.  In their study, filter compatibility was examined, and Raman/SEM chemical
23      imaging was conducted for several standard materials as well as for ambient PM2 5 samples.
24      Polycarbonate was determined  to be a suitable substrate for both SEM and Raman chemical
25      imaging analysis.
26           Conner et al.  (2001) used CCSEM with individual X-ray analysis to study the chemical and
27      physical attributes of indoor and outdoor aerosols collected around a retirement home in
28      Baltimore.  The CCSEM technique was demonstrated to be capable of identifying spherical
29      particles typical of combustion or other high temperature (presumably industrial) processes as
30      well as pollens and spores. Indoor particles originating from cosmetics were also identified.
31

        June 2003                                2B-7        DRAFT-DO NOT QUOTE OR CITE

-------
 1     2B.2   ORGANIC AND INORGANIC CARBON
 2          Large scale efforts to characterize carbonaceous aerosol require cost effective methods that
 3     can analyze samples rapidly.  Commercial thermal/optical (TO) instruments were developed to
 4     serve this need. The US EPA IMPROVE and STN networks have employed these instruments
 5     to accumulate large datasets, including measurements taken over the past 18 years by the
 6     IMPROVE network and over the past 3 years by STN. In addition to the protocols developed for
 7     IMPROVE and STN networks, a number of alternative TO-based protocols and techniques have
 8     been employed by the academic research community. Protocols vary in temperature range and
 9     step size, in the duration of heating at each step, in the timing for introduction of oxygen for the
10     conversion of black carbon into CO2, and in the assignment of organic carbon (OC) and
11     elemental carbon (EC) fractions. Figure 2B-2 shows examples of two protocols in current use.
12     These operational differences have complicated efforts to compare and combine data sets  from
13     studies using different TO protocols.
14
15     Thermal-Optical Reflectance
16          The thermal optical reflectance (TOR) method of carbon analysis developed by Huntzicker
17     et al. (1982) has been adapted by several laboratories for the quantification of organic and
18     elemental carbon in PM collected on quartz-fiber filters.  Although the principle applied by these
19     laboratories is identical to that of Huntzicker et al. (1982), the details differ with respect to
20     calibration standards, analysis time, temperature ramping, and volatilization/combustion
21     temperature.  The IMPROVE network employs a version of the TOR method for its OC/EC
22     analyses.
23          In the most commonly used version of the TOR method (Chow et al., 1993), a punch from
24     the filter sample is heated to temperatures ranging from ambient to 550 °C in a pure helium
25     atmosphere. In principle, the organic carbon fraction of the PM contained by the filter punch
26     will vaporize, leaving behind only refractory elemental carbon. The organic carbon that evolves
27     at each temperature step is first oxidized to CO2, then converted to methane and finally
28     quantified with a flame ionization detector  (FID). The filter punch is incubated at 550 °C  for a
29     period sufficient to allow the flame ionization signal to return to its baseline value. The punch is
30     then exposed to a 2% oxygen and 98% helium atmosphere and heated from 550 °C to 800 °C
31     with several temperature ramping steps.  The reflectance from the deposit side of the filter punch

       June 2003                                2B-8        DRAFT-DO NOT  QUOTE OR CITE

-------
 1      is monitored throughout the analysis.  This reflectance decreases during OC volatilization in the
 2      helium atmosphere owing to the pyrolysis of involatile organic material present in the filter
 3      punch. When oxygen is added, the reflectance increases as light-absorbing carbon is combusted
 4      and removed. It is assumed that the first light-absorbing carbon compounds to combust are
 5      pyrolysis artifacts from the first phase of the analysis.  Therefore, the organic carbon mass is
 6      defined as that fraction that evolves up to the introduction of oxygen plus the oxidized carbon
 7      that evolves up to the point when the filter reflectance signal is restored to its pre-analysis value.
 8      Once the original reflectance level is re-attained, all further carbon evolving from the sample
 9      punch is assigned to the elemental carbon fraction. Accordingly, "organic carbon" (OC) is
10      carbon that does not absorb light at the laser wavelength (632.8 nm) typically used by TOR
11      instruments, and all other carbon is defined as "elemental carbon" (EC).
12
13      Thermal-Optical Transmission
14           The primary difference between TOR and thermal optical transmission methods is in the
15      choice of absorption detection — light transmission through the filter punch, rather than its
16      reflectance, is monitored throughout the analysis. The TOT method of Birch and Gary (1996)
17      also uses a pure helium atmosphere for volatilizing organic carbon, but the second stage involves
18      a higher oxygen/helium (10%) gas mixture to oxidize the black carbon remaining on the filter
19      punch.  The temperature is raised to approximately 820 °C in the helium phase, during which
20      both organic and carbonate carbon are volatilized from the filter. In the second stage, the oven
21      temperature is reduced then raised to about 860 °C. During this stage, pyrolysis correction and
22      the EC measurement is made. Figure 2B-1 is an example of a TOT thermogram, showing
23      temperature, transmittance, and FID response traces. The peaks that correspond to the
24      concentrations of CO2 that evolve from the filter punch during the course of the analysis are
25      assigned to OC, carbonate carbon (CC),  pyrolitic carbon (PC), and EC. The high temperature in
26      the first stage of the TOT thermal profile is included for the purpose of decomposing carbonate
27      carbon and for volatilization of very high-boiling organic compounds. However, many organic
28      carbon compounds will pyrolyze at this temperature to generate PC. The ability to quantify PC
29      is particularly important in high OC/EC  regions such as wood smoke-impacted air sheds.  Wood
30      smoke aerosol contains many complex compounds that generate substantial PC.  Significant
31      error in the EC fraction can result in the  absence of a careful PC correction.

        June 2003                                2B-9        DRAFT-DO NOT QUOTE OR CITE

-------
                                                        OC - EC split
                                            Time (min)

       Figure 2B-1. This thermogram, for a sample containing rock dust (a carbonate source)
                    and diesel exhaust, shows three traces that correspond to temperature, filter
                    transmittance, and FID detector response. Peaks correspond to organic
                    (OC), carbonate (CC), pyrolytic (PC), and elemental (EC) carbon. The final
                    peak is a methane calibration peak.
       Source: Birch and Gary (1996).
 1          Informal intercomparisons among different filter transmission methods have shown high
 2     correlations of absorption, but differences of up to a factor of two in absolute values (Watson
 3     et al., 1988a,b).  These differences are functions of the type of filter, filter loading, the chemical
 4     and physical nature of the deposit, the wavelengths of light used, calibration standards, and light
 5     diffusing methods.  At the current time, there is no agreement on which combination most
 6     accurately represents light absorption in the atmosphere.
 7          The National Institute for Occupational Safety and Health (NIOSH) Method 5040 is based
 8     on the TOT method (Birch and Gary, 1996).  The NIOSH protocol consists of a two-stage
 9     process with the first stage being conducted in a pure  helium atmosphere at temperatures of 250,
10     500, 650, and 850 °C for a total of 4.5 minutes and the second stage conducted in a 2%
11     oxygen/98% helium mix at temperatures of 650, 750,  850, and 940 °C for 4 minutes.
       June 2003
2B-10
DRAFT-DO NOT QUOTE OR CITE

-------
 1     A pyrolysis base correction is made based on the transmittance measurement. The U.S. EPA
 2     adopted a modified version of the NIOSH Method 5040 protocol for use in its Speciation Trends
 3     Network (STN).
 4
 5     IMPRO VE versus NIOSH protocols
 6          Although the IMPROVE protocol and the NIOSH protocol in use by the STN network
 7     yield closely comparable total carbon (TC) values, the differences in thermal profiles and
 8     operational definitions of organic versus black carbon result in very different mass assignments
 9     to these fractions. Other methodological differences may make data comparisons difficult,
10     including the different choice of absorption detection, e.g. reflectance versus transmission, the
11     different temperature ranges and incubation periods and the different approaches used to account
12     for background OC. Examples of thermograms obtained with the IMPROVE and NIOSH
13     protocols are shown in Figure 2B-2.
14          Using both the IMPROVE and NIOSH protocols, Chow et al. (2000) analyzed 60 quartz
15     filter samples on a prototype reflectance/transmittance analyzer that represented a wide variety
16     of aerosol compositions and concentrations. The two TC data sets possessed no statistically
17     significant differences. However, marked differences were found in the fraction of TC that is
18     attributed to EC as determined by the IMPROVE versus NIOSH thermal evolution protocols.
19     The  IMPROVE EC measurements were typically higher than the NIOSH EC measurements.
20     When the NIOSH protocol was modified to exclude the helium-only 850 °C temperature step,
21     however, the OC/EC ratios came into agreement between the two methods. Because OC and EC
22     are operationally defined parameters, Chow et al. (2000) pointed out that it is important to retain
23     ancillary information when reporting EC and OC by these analytical methods, so that
24     comparisons can be made among measurements taken at different sites using these two methods.
25          The NIOSH and IMPROVE protocols both require correction for positive organic artifacts
26     due to absorption of background organic vapor by the heat-treated quartz filters used for OC/EC
27     measurements. Both the IMPROVE and STN science teams have evaluated the presence of
28     carbon artifacts in their measurements.  The IMPROVE team has established that heat-treated
29     quartz filters adsorb organic carbon vapors up to a saturation threshold over the course of a few
30     days in the field. The STN science team has observed that positive carbon artifacts can vary
31     with sampler type.  Total carbon artifacts for the samplers used by the STN range from 9.5 to

       June 2003                               2B-11       DRAFT-DO NOT QUOTE OR CITE

-------
                                         IMPROVE Method
                               100% He
                                                         2% 02/88% He
                                       Temp
                                       Laser Trans   J
                                       Trans-Split    '
                  0  120 240  360  480 800  720  840 880  1080 1200 1320 1440 15BO 1880
                                           Time |sec)
                                                                        1800
                                          NiOSH Method
1000 i
900-
800
700
o
g 600 •
S
S 500-
E
I 400 •
300 -
200 •
100 -
n .
, . ~~ -~ , . ~ f ^, i- JU *_1 *~~ r~ - ,„ B-
...
F


**•*

t mum
*"^


«j*«,
V
i
* i / i
,11 "[[T^^'^'l**4*4**"^
/ ! I
.„..,/ t H— ECB — ->
i
: i
A i [ .......
\^;i^.;^V-T^
^>V/ \ «/|^
""f"*OC1 002 OC3 I OC4 JEqf EC2 | EG3 |

**
!-



	 FID
	 Temp
•»" Laser Reflect
**«* Laser Trans
**** ^ *•" ReHect-Split
1 	 - - - Trans-Spilt




V.
calib 1
                  0  120  240 380  480  600  720  840 i60  1080 1200 1320  1440 1580 1680 1800
Figure 2B-2. Examples of thermograms obtained by (a) the IMPROVE protocol, and by
             (b) the NIOSH protocol.  Flame ionization detection is used in both examples.
             FID denotes the observed signal intensity. The intervals denoted ECR/ECT,
             i.e., elemental carbon reflectance/transmittance, refer to the time and
             temperature intervals during which EC is oxidized and filter "blackness"
             is monitored by laser reflectance or transmission at the He-Ne wavelength,
             633 nm.

Source:  Birch and Gary (1996).
June 2003
2B-12
DRAFT-DO NOT QUOTE OR CITE

-------
 1      33.9% of the carbon collected during a typical ambient measurement.  Documents describing the
 2      issues regarding blank subtraction and the current protocols used by the IMPROVE and STN
 3      networks are available on the network websites:
 4
 5           IMPROVE - http://vista.cira.colostate.edu/improve
 6           STN - http://www.epa.gov/ttn/amtic/files/ambient/pm25/spec
 7
 8      Pyrolitic Carbon (PC) and Other Confounders
 9           In a methods comparison study, Countess (1990) determined that it is necessary to
10      minimize or correct for PC and that OC found in wood smoke and automobile exhaust samples
11      which pyrolyze to create interferences during the course of some thermal optical protocols.
12      Pyrolysis of organic compounds occurs during analysis by both TO methods, although the
13      greater temperature to which samples are exposed in the NIOSH protocol is likely to produce
14      larger quantities of pyrolysis-derived EC.  During the initial heating phase, pyrolysis is indicated
15      by an increase in optical density (blackness) by the filter sample.  Both methods distinguish
16      artifact pyrolysis-derived EC from ambient EC at the point when the transmittance or reflectance
17      signal is restored to the pre-heating level.  The assumption made is that heating does not alter the
18      absorption properties of the material collected on the filter. This assumption is reasonable if the
19      only light absorbing species of carbon is strictly a graphite-like elemental carbon that is unlikely
20      to undergo a change in its absorption properties over the temperature ranges used by TO
21      methods. The effects  of heat-induced changes to the light absorption and chemical properties of
22      atmospheric organic compounds on TO analysis  are being evaluated by NIST.  NIST has
23      identified three assumptions that must be met in order for TO methods to reliably measure EC:
24      (1) absorptivity of carbonaceous PM remains constant up to the point of pyrolysis; (2) once
25      formed, pyrolyzed carbon (char) absorbs at the analytic wavelength and its absorptivity remains
26      constant within the high temperature step; (3) pyrolyzed OC has the same absorptivity as EC that
27      is native to the sample. Using urban dust, forest  fire emissions and ambient laboratory aerosol,
28      they observed changes in the absorptivities of these materials during heating before and during
29      the formation of pyrolysis  artifacts, up to the OC/EC split point. NIST, therefore, recommends
30      that standard TOR/T protocols be developed that account for these changes.
        June 2003                                2B-13       DRAFT-DO NOT QUOTE OR CITE

-------
 1           Light-absorbing carbon includes a large number of compounds that may be altered, not
 2      only in their light-absorbing properties but also in their oxidation chemistry. The materials
 3      responsible for defining the original blackness of the filter sample may be altered during the
 4      initial heating phase of TO analysis, so that the mix present during the oxidation cycle may not
 5      be representative of the original, atmospherically-derived black carbon. An error in determining
 6      the mass of EC would arise, for example, if a heavy, but weakly light-absorbing compound were
 7      transformed into a material that possesses a higher absorption coefficient and higher resistance to
 8      oxidation than absorbing compounds  collected from the atmosphere. The pyrolized form of this
 9      compound would resist oxidation past the OC-EC split  point in the analysis, leading to a falsely
10      high EC fraction.
11           Chow et al. (2000) noted that neither the IMPROVE nor NIOSH methods were able to
12      accurately detect further blackening on optically dense, i.e., very black, filters that are typical of
13      source profile samples.  Predictions of artifact pyrolysis carbon by TOR and TOT differed
14      widely. Note that both transmittance  and reflectance methods rely on derivations of optics laws
15      (i.e., the Beer-Lambert and Lambert Laws) that predict a linear transmittance/reflectance signal
16      response with species concentration, but only for optically thin samples (Strobel and Heineman,
17      1989). Very black filters exceed this  limitation; thus, the signal response of these methods may
18      not be a linear or otherwise predictable function of black carbon concentration.
19           Another important source of error in any TO measurement of aerosol  OC/EC  arises when
20      samples contain transition metal oxides, such as iron oxide. Many transition metal  oxides are
21      found in crustal material. Fung et al.(2002) report that  such oxides can serve as oxidizing agents
22      for BC at high temperatures. The consequence of such  an effect is an elevation of the signal
23      usually assigned to OC and corresponding reduction in  apparent BC.
24
25      Thermal Manganese Oxidation
26           The thermal manganese oxidation (TMO) method (Mueller et al., 1982; Fung, 1990) uses
27      manganese dioxide in contact with the sample throughout the analysis as the oxidizing agent.
28      Temperature is relied upon to distinguish between organic and elemental carbon. Carbon
29      evolving at 525 °C is classified as organic carbon,  and carbon evolving at 850 °C is classified as
30      elemental carbon.  TMO does not correct for pyrolized  OC, which may lead to overestimation of
31      EC. This method has been used in the SCENES (Subregional Cooperative Electric Utility,

        June 2003                               2B-14       DRAFT-DO NOT QUOTE OR CITE

-------
 1      Department of Defense, National Park Services, and Environmental Protection Agency Study);
 2      (Sutherland and Bhardwaja, 1987; Mueller et al., 1986) visibility network, as well as in the
 3      SCAQS (Southern California Air Quality Study) (Chow et al., 1994a,b; Watson et al., 1993,
 4      1994a,b).
 5
 6      OC/EC Me thods Inter comparisons
 1           Chow et al. (1993) document several variations of the thermal (T), thermal/optical
 8      reflectance (TOR), thermal/optical transmission (TOT), and thermal manganese oxidation
 9      (TMO) methods for organic and elemental carbon.  Comparisons among the results of the
10      majority of these methods show that they yield comparable quantities of total carbon in aerosol
11      samples, but the distinctions between organic and elemental carbon are quite different (Cadle
12      and Groblicki, 1982; Cadle and Mulawa, 1990; Countess, 1990; Hering et al., 1990; Birch, 1998;
13      Schmid et al., 2001). For the sum of organic and elemental carbon, these  methods reported
14      agreement within 5% to 15% for ambient and source samples (Houck et al., 1989; Kusko et al.,
15      1989; Countess, 1990; Shah and Rau, 1990) and within 3% on carefully prepared standards.
16      Evaluation of these methods thus is a matter of assessing how they differentiate between organic
17      and elemental  carbon. The EC/TC ratio for samples is method dependent.
18           An international methods intercomparison study on the analysis of carbonaceous aerosols
19      on quartz fiber filters was organized by the Vienna University of Technology and involved
20      seventeen laboratories and nine different thermal and optical methods (Schmid et al., 2001).
21      All laboratories were sent punches from three 150-mm quartz fiber filters that had been exposed
22      for 24 h near a high traffic street in Berlin.
23           Five laboratories employed the German official standard VDI2465 methods. Two of these
24      laboratories used the VDI 2465/1 method that determines extractable organic carbon, non-
25      extractable organic carbon, and elemental carbon by way of a solvent-based extraction protocol.
26      Other laboratories participating in the intercomparison used variations of the VDI 2465 standard
27      that rely upon  differences in thermal stability to accomplish the separation of carbonaceous
28      aerosol fractions. A  range of thermal protocols, total carbon determination techniques and CO2
29      detection schemes were employed by the participating laboratories.
30           Good agreement of the TC results was obtained by all laboratories with only two outliers in
31      the complete data set. The relative standard deviation between laboratories for the TC results,

        June 2003                                2B-15        DRAFT-DO NOT QUOTE OR CITE

-------
 1     were 6.7, 10.6, and 8.8% for the three samples. In contrast, the EC results were much more
 2     variable.  The relative standard deviation between laboratories for the EC results, were 36.6,
 3     24.4, and 45.5% for the three samples. The VDI methods, especially the VDI 2465/2, were
 4     found to give generally higher amounts of EC than the thermal-optical methods.  This trend was
 5     detected for all samples. The authors recognized that uncorrected thermal methods are prone to
 6     positive artifacts by charring during pyrolysis.  They also noted that when using  solvent
 7     extraction methods, the dissolution of polymeric aerosol constituents may not be successful.
 8     Both of these effects would lead to overestimation of the EC fraction.  When the laboratories
 9     were grouped according to their methods, the relative standard deviation between laboratories
10     was much smaller. This study demonstrates that the TC measurement can yield similar results
11     from a variety of methods, but the EC measurement is highly dependent upon the method used.
12     The problems associated with the determination of EC are exacerbated by the lack of a standard
13     reference material and consistent definitions of EC.
14
15     Measuring Black Carbon (BC) Instead of EC
16          Light absorbing or black carbon (BC) can be measured by optical techniques (Penner and
17     Novakov, 1996). Both EC and BC define a similar fraction of aerosol; but EC is defined in
18     terms of both the thermal and light-absorption properties of the sample, whereas BC is based on
19     solely on its light-absorption properties. The aethalometer, the integrating sphere sunphotometer
20     and photoacoustic spectroscopy (described in Section 2B.3) are example techniques for
21     determining BC.
22          Hitzenberger et al. (1996) investigated the feasibility of using an integrating sphere
23     photometer as an adequate measurement system for the BC content and the absorption
24     coefficient.  In another study (Hitzenberger et al., 1999), the integrating sphere method was
25     compared to an aethalometer (Hansen et al., 1984), the thermal method of Cachier et al. (1989),
26     and the thermal/optical method of Birch and Gary (1996).  The absorption coefficients that were
27     obtained from both the integrating sphere and the aethalometer were comparable. The BC mass
28     concentrations obtained  from the aethalometer were 23% of those obtained from the integrating
29     sphere.  Compared to the thermal method, the integrating sphere overestimated the BC mass
30     concentrations by 21%.  Compared to the thermal/optical method, the integrating sphere was
31     within 5% of the 1:1 line. However, the data were not well correlated.

       June 2003                               2B-16       DRAFT-DO NOT QUOTE OR CITE

-------
 1          The Carbonaceous Species Methods Comparison Study (CSMCS), as mentioned in the
 2      1996 PM AQCD (U.S. Environmental Protection Agency, 1996a) was conducted in Los Angeles
 3      in 1986.  Hansen and McMurry (1990) compared the thermal manganese method with the
 4      aethalometer for aerosol elemental carbon.  The first involved collection of impactor samples
 5      backed by a quartz fiber after-filter followed by EC analysis by oxidation in helium over a MnO2
 6      catalyst; the other conducted real-time measurements using an aethalometer (an optical
 7      absorption technique).  They found good agreement between these two very different methods.
 8      The CSMCS interlaboratory precision for total carbon was 4.2% (Turpin et al., 2000). However,
 9      because the split between OC and EC is operationally defined, there was substantial
10      interlaboratory variability in OC and EC (e.g., 34% for EC [Turpin et al., 1990]).
11
12          EC/OC Summary. With the limitations and precautions described above, laboratory
13      analyses for the carbonaceous properties of collected particles have matured to the point where
14      they can be performed with commercially-available instruments following established standard
15      operating procedures. However, carbon analysis continues to be a subject of active research
16      within the atmospheric sciences community and EPA, and carbon speciation methods
17      comparisons are being undertaken during such studies as the Atlanta Supersite. The state of the
18      art for carbonaceous PM measurements continues to advance;  and, although progress is being
19      made, the definitions of OC, EC and BC continue to be operationally defined in reference to the
20      method employed. Reports of EC/OC measurements should therefore include mention of the
21      method with which the species were determined.  Finally, if possible, all ancillary data should be
22      retained, to allow later comparison to other methods.
23
24
25      2B.3  CONTINUOUS METHODS
26      2B.3.1  Continuous Measurement of Mass
27      Tapered Element Oscillating Microbalance (TEOM)
28          The advantages of continuous PM monitoring and the designation of the TEOM as an
29      equivalent method for PM10, have led to the deployment of the TEOM at a number of air
30      monitoring sites. The TEOM also is being used to measure PM2 5.  The TEOM differs from  the
31      federal reference methods for particulate mass in that it does not require equilibration of the

        June 2003                               2B-17       DRAFT-DO NOT QUOTE OR CITE

-------
 1      samples at a specified temperature and relative humidity. The TEOM samples at a constant
 2      temperature and is typically heated to some temperature higher than the ambient temperature
 3      (Meyer et al., 1995; Meyer and Rupprecht, 1996); whereas the FRM samples at the ambient
 4      temperature.  Thus, the TEOM does not provide data equivalent to the FRM because of losses of
 5      volatile species. Volatilization losses in the TEOM sampler can be reduced by operating the
 6      instrument at 30 °C, rather than the typical 50 °C specified, or by using a Nafion diffusion dryer
 7      instead of heating to dehumidify the particles (with a 30 °C temperature).
 8            This difference in operation and implications for fine particle measurements was
 9      examined by researchers at Commonwealth Scientific and Industrial Research Organisation
10      (CSIRO) Atmospheric Research in Australia (Ayers et al., 1999). That group compared 24-h
11      mean PM2 5 mass concentrations as determined by a TEOM and by two manual, gravimetric
12      samplers (a low-volume filter sampler and a MOUDI sampler) in four Australian cities, on 15
13      days in the winter half-year.  The TEOM was operated at 50  °C at one location and at 35 °C at
14      the other three locations.  A systematically low TEOM response in comparison to the integrated
15      gravimetric methods was  observed.  In a comprehensive study,  Allen et al. (1997) reported
16      results in which TEOM data collected at 10 urban sites in the United States and Mexico were
17      compared with 24 h integrated mass concentrations for both PM10 and PM2 5. They collected a
18      large data set that included both winter and summer seasons. Allen et al. (1997) concluded that,
19      especially for urban areas, a significant fraction of PM10 could be semivolatile compounds that
20      could be lost from the heated filter in the TEOM leading to a systematic difference between the
21      TEOM and the EPA FRM for PM10.  They suggested that this difference is likely to be larger for
22      PM2 5 than it is for PM10 (Allen et al., 1997).
23            In a similar study conducted in Vancouver, British Columbia, the effect of equilibration
24      temperature on PM10 concentrations from the TEOM was examined. Two collocated TEOM
25      monitors operated at 30 and 50 °C, respectively, were operated  in the Lower Fraser Valley in
26      British Columbia for a period of approximately 17 months (Mignacca  and Stubbs, 1999).
27      A third TEOM operating at 40 °C was operated for 2 months during this period.  They found
28      that, on average, the 1-h average PM10 from the TEOM operating at 30 °C was consistently
29      greater than that from the TEOM operated at 50 °C. For the period during which the third
30      TEOM was operated (at 40 °C), the PM10 from that instrument was between those values for the
31      other two instruments. They also found that the differences in masses were proportional to the

        June 2003                                2B-18       DRAFT-DO NOT QUOTE OR CITE

-------
 1     PM10 loading, and more strongly correlated to the PM10 from the TEOM operated at the lower
 2     temperature. They recommended that the TEOM monitors be operated at 40 °C as opposed to
 3     operating at 50 °C in summer and 30 °C in winter, in order to avoid introducing a methodological
 4     seasonal bias.
 5            A new sample equilibration system (SES) was developed to reduce losses of semivolatile
 6     species from the PM2 5 TEOM by conditioning the sample stream to lower humidity and
 7     temperature (Meyer et al., 2000). The SES utilizes humidity sensors and a Nafion dryer
 8     designed for low particle loss. The dryer fits between the flow splitter that follows the size-
 9     selective inlet and the sensor unit. A dry purge gas flows over the exterior of the Nafion tubing
10     and allows for self-regeneration. A TEOM with PM2 5 inlet and equipped with an SES was
11     operated at 30 °C alongside another TEOM operating at 50 °C without the SES in Albany, NY,
12     over a 6-day period during a summertime high-temperature, high-relative-humidity episode. The
13     SES maintained the sample air relative humidity under 30%, and the TEOM with the  SES
14     generally measured more mass than the other TEOM. The TEOM with SES also was operated
15     alongside an FRM-type sampler for the period of June 6 through September 25, 1999.  The
16     correlation between the FRM and TEOM/SES showed a slope of 1.0293 and R2 of 0.9352;
17     whereas the correlation between the FRM and the TEOM without SES and operating  at 50 °C
18     showed a slope of 0.8612 and R2 of 0.8209.  The SES can be installed on existing TEOM
19     monitors.
20
21     Beta-Gauge Techniques
22            The use of absorption of beta radiation as a indicator of particle mass has been used
23     effectively to measure the mass of equilibrated particulate matter collected on Teflon  filters
24     (Jaklevic et al., 1981a; Courtney et al., 1982). The technique also has been used to provide near
25     real-time measurements with time intervals on the order of an hour (Wedding and Weigand,
26     1993). However, real-time beta gauge monitors experience the same problems as other
27     continuous or near real-time particulate matter mass monitoring techniques.  Particle-bound
28     water must be removed to reduce the sensitivity of the indicated mass to relative humidity.
29     However, the simplest technique, mild heating, will remove a portion of the ammonium nitrate
30     and the semivolatile organic compounds as well as the particle-bound water.
       June 2003                               2B-19       DRAFT-DO NOT QUOTE OR CITE

-------
 1            An intercomparison study of two beta gauges at three sites indicated that the Wedding
 2      beta gauge and the Sierra Anderson (SA) 1200 PM10 samplers were highly correlated, r > 0.97
 3      (Tsai and Cheng, 1996). The Wedding beta gauge was not sensitive to relative humidity but
 4      yielded results approximately 7% lower. This suggests that the mild heating in the beta gauge
 5      causes losses comparable to those caused by equilibration, although the differences could result
 6      from slight differences in the upper cut points.  The Kimoto beta gauge that was operated at
 7      ambient temperature was sensitive to relative humidity yielding significantly higher mass
 8      concentrations relative to the SA 1200 for RH > 80% than for RH < 80% even though the
 9      correlation with the SA 1200 was reasonable (r = 0.94 for RH > 80% and 0.83 for RH < 80%).
10
11      Piezoelectric Microbalance
12            Piezoelectric crystals have mechanical resonances that can be excited by applying an
13      alternating electrical voltage to the crystal. As the resonance frequencies are well defined, such
14      crystals (quartz in particular) have  found applications as secondary time and frequency standards
15      in clocks and watches. As for all mechanical resonators,  the resonance frequency is a function
16      of mass. Therefore, by monitoring the resonance frequency in comparison with a second crystal,
17      one can continuously measure the mass deposited on the  crystal (Sem et al., 1977; Bowers and
18      Chuan, 1989; Ward and Buttry, 1990; Noel and Topart, 1994). Comparison with a second
19      crystal largely compensates for the effect of temperature  changes on the resonance frequency.
20            The piezoelectric principle  has been used to measure particle mass by depositing the
21      particles on the crystal surface either by electrostatic precipitation or by impaction (Olin and
22      Sem, 1971).  The collection efficiency of either mechanism has to be determined as function of
23      particle size to achieve quantitative measurements.  In addition, the mechanical coupling of large
24      particles to the crystal is uncertain. Both single and multi-stage impactors have been used (Olin
25      and Sem, 1971; Fairchild and Wheat, 1984). Quartz crystals have sensitivities of several
26      hundred hertz per microgram.  This sensitivity results in the ability to measure the mass
27      concentration of a typical 100  |ig/m3  aerosol to within a few percent in less than one minute
28      (Olin and Sem, 1971).
29
30
31

        June 2003                               2B-20      DRAFT-DO NOT QUOTE OR CITE

-------
 1      Coarse Particle Mass
 2            The RAMS and CAMM are only appropriate for fine particle measurements (PM2 5 or
 3      PMj).  However, the TEOM, beta gauge, and piezoelectric microbalance may be used to measure
 4      either PM2 5 or PM10 (or a sample with any specified upper 50% size cut). A pair of such
 5      samplers may be used to measure thoracic coarse PM mass concentration (PM10_2 5) by difference
 6      between the PM10 and PM2 5 concentrations. However, concerns have been raised concerning the
 7      quality of the data from such difference calculations and the resulting potential biases in
 8      exposure assessment and risk determinations (Wilson and Suh, 1997; White, 1998). Misra et al.
 9      (2001) describe the development and evaluation of a continuous coarse particle monitor (CCPM)
10      that may provide direct measurements of coarse mode PM mass concentrations at short time
11      intervals (on the order of 5-10 min). The basis of the CCPM is enrichment of the coarse particle
12      concentrations through use of virtual impaction while maintaining fine particle concentrations at
13      ambient levels. The resulting aerosol mixture is analyzed using a standard TEOM for which the
14      response is now dominated by the enriched coarse PM mass. The coarse PM concentrations
15      determined from the CCPM were compared to those obtained with a MOUDI, operating with
16      only the 10- and 2.5-micron cutpoint stages, and a Partisol dichotomous sampler. The CCPM
17      coarse particulate concentrations were highly correlated with both the MOUDI (R2 = 0.88) and
18      the Partisol (R2 = 0.88)  coarse PM concentrations.  By operating the CCPM at a coarse particle
19      enrichment factor of 25, the coarse PM concentration can be determined a priori without
20      determination of the fine particle concentration, so long as the fine-to-coarse particle
21      concentration ratios are not unusually high (i.e., 4-6). Misra et al. (2001) also concluded from
22      field experiments that the coarse particulate concentrations determined from the CCPM were
23      independent of the ambient fine-to-coarse particulate concentration ratio due to the decrease in
24      particle mass median diameter that should accompany fine-to-coarse particulate concentration
25      ratios during  stagnation conditions.
26
27      2B.3.2  Continuous Measurement of Organic and/or Elemental  Carbon
28            Testing and refinement of models that simulate aerosol concentrations from gas and
29      particle emissions require air quality measurements of approximately  1-h time resolution to
30      reflect the dynamics of atmospheric transport, dispersion, transformation, and removal. Below
31      instruments are described that have been used to collect and analyze atmospheric organic PM

        June 2003                                2B-21       DRAFT-DO NOT QUOTE OR CITE

-------
 1      with better than 2-h resolution.  These instruments were all present at the Atlanta Supersite
 2      experiment during the summer of 1999, and an intercomparison of results is underway.
 3
 4      Automated Carbon Analyses
 5            Turpin et al. (1990) describe an in situ, time-resolved analyzer for particulate organic and
 6      elemental carbon that can operate on a time cycle as short as 90 min. This analyzer collects
 7      particulate matter on a quartz fiber filter mounted in a thermal-optical transmittance carbon
 8      analyzer (Turpin et al., 1990).  A second quartz fiber filter behind a Teflon filter in a second
 9      sampling port may also be analyzed to provide an estimate of the positive sampling artifact (i.e.,
10      gas adsorption on the quartz sampling filter). The organic material in the collected PM is
11      thermally desorbed from the filter at 650 °C and oxidized at 1000 °C over a MnO2 catalyst bed.
12      The evolved CO2 is converted to methane over a nickel catalyst, and the methane is measured in
13      a flame ionization detector.  Then the elemental carbon is oxidized at 350 °C in a 98% He-2% O2
14      atmosphere. Correction is made for pyrolytic conversion of some of the organic particulate
15      matter. The instrument was operated with a 2-h resolution during SCAQS in 1987 (Turpin and
16      Huntzicker, 1991;1995),  as well as during CSMCS in 1986 (Turpin et al., 1990). By using
17      elemental carbon as a tracer for primary, combustion-generated organic carbon, these authors
18      estimated the contributions of primary sources (i.e., material  emitted in particulate form) and
19      secondary sources (i.e., particulate material formed in the atmosphere) to the total atmospheric
20      particulate organic carbon concentrations in these locations.
21            An automated carbon analyzer with 15-min to 1-h resolution is now commercially
22      available (Rupprecht et al., 1995) and has been operated in several locations including the
23      Atlanta Supersite.  It collects samples on a 0. l-jim impactor downstream of an inlet with a
24      2.5-|im cutpoint. Use of an impactor eliminates gas adsorption that must be addressed when
25      filter collection is used.  However, this collection system may experience substantial particle
26      bounce and loss of a sizable fraction of EC since some EC is in particles < 0.2  jim. It is possible
27      that ongoing research, in which particle size is increased by humidification prior to impaction,
28      may result in an improvement in collection efficiency.  In the analysis step, carbonaceous
29      compounds are removed  by heating in filtered ambient  air. Carbonaceous material removed
30      below 340 °C is reported as organic carbon, and material removed between 340 and 750  °C is
31      reported as elemental carbon. Turpin et al. (2000) comment that it would be more appropriate to

        June 2003                                2B-22       DRAFT-DO NOT  QUOTE OR CITE

-------
 1      report carbon values obtained by this method as "low-" and "high-temperature" carbon, because
 2      some organics are known to evolve at temperatures greater than 340 °C (e.g., organics from
 3      woodsmoke).
 4
 5      Aethalometer for EC
 6            As discussed earlier, black carbon (BC), a carbon fraction very similar to EC, is most
 7      commonly measured using an aethalometer, a commercially available, automated, time-resolved
 8      instrument (i.e., 5- to 15-min sample duration) that measures the light attenuation of aerosol
 9      particles collected on a filter tape (Hansen et al., 1984).  The concentration of elemental carbon
10      is derived from the light absorption measured on a filter using an estimate of the specific
11      absorption (m2/g) of elemental carbon on the filter; the specific absorption value is derived from
12      laboratory and atmospheric tests and is specified by the manufacturer.  The specific absorption
13      value could be expected to vary with location, season, and source mix.  Comparisons in
14      atmospheric experiments at some locations with EC values measured by thermal methods
15      confirm that the aethalometer provides a statistically meaningful estimate of EC concentration
16      (Allen et al., 1999c; Liousse et al., 1993). For instance,  Allen et al. (1999c) found the following
17      statistical relationship for Uniontown, PA, during summer 1990: black carbon (aethaometer)
18      = 0.95*EC (thermal) - 0.2 (r2 = 0.925, n not specified but appears to be >50, EC range from 0 to
19      9 |ig/m3). Another source of error in aethalometer measurements arises from the sampling
20      procedure.  Particles are trapped within a three-dimensional filter matrix. Therefore, scattering
21      of transmitted and reflected light may erroneously be attributed to absorption, thus causing errors
22      in the BC calculation. Ballach et al. (2001) investigated immersing the filter in oil of a similar
23      refractive index as a means to minimize the interferences due to light scattering effects from the
24      filter, similar to a procedure common in microscopy. BC measurements determined using the oil
25      immersion technique were compared to those from an integrating sphere, a polar photometer,
26      and Mie calculations. Aerosols tests included several pure carbon blacks from different
27      generating procedures that were used to calibrate the immersion technique, pure ammonium
28      sulfate aerosol, and external and internal mixtures of ammonium sulfate with varying amounts of
29      carbon blacks.  The oil immersion technique was also tested on ambient air samples collected at
30      two different sites in the cities of Frankfurt am Main and Freiburg, Germany. Optical
31      measurements,  both of blank and loaded filters, show that the oil immersion technique

        June 2003                                 2B-23       DRAFT-DO NOT QUOTE OR CITE

-------
 1     minimizes scattering losses. Ballach et al. (2001) found that site-related effects were reduced
 2     and that there was reasonably good agreement with the other optical techniques as well as with
 3     the Mie calculations.
 4
 5     Photoacoustic Measurement of EC
 6            Measurement of aerosol light absorption utilizing photoacoustic spectroscopy has been
 7     examined as a continuous method for measuring elemental carbon mass concentrations (Petzold
 8     and Niessner, 1996; Arnott et al., 1999; 2000).  Like the aethalometer, this method measures
 9     light absorption; however, unlike most other light absorption methods, the photoacoustic
10     technique does not require a filter.  The photoacoustic spectrometer of Arnott and coworkers was
11     demonstrated during the Northern Front Range Air Quality Study and compared to an
12     aethalometer (Moosmuller et al., 1998). Neither the aethalometer nor the photoacoustic
13     spectrometer measure elemental carbon mass directly.  Because the photoacoustic  spectrometer
14     measures the absorption coefficient directly, the specific absorption efficiency must be known or
15     assumed in order to determine elemental carbon mass.  Assuming a light absorption efficiency of
16     10 m2 g"1, Arnott et al. (1999) reported a lower detection limit for light absorption of 0.4 M m"1
17     corresponding to a mass concentration of elemental carbon of approximately 40 ngm"3.
18
19     2B.3.3  Continuous Measurements  of Nitrate and Sulfate
20     Nitrate
21            An integrated collection and vaporization cell was developed by Stolzenburg and Hering
22     (2000) that provides automated, 10-min resolution monitoring of fine particulate nitrate. In this
23     system, particles are collected by a humidified impaction process and analyzed in place by flash
24     vaporization and chemiluminescent detection of the evolved nitrogen oxides. In field tests in
25     which the system was collocated with two FRM samplers, the automated nitrate sampler results
26     followed the results from the FRM, but were offset lower.  The system  also was collocated with
27     a HEADS and a SASS speciation sampler (MetOne Instruments). In all these tests, the
28     automated sampler was well correlated to other samplers with slopes near 1 (ranging from 0.95
29     for the FRM to 1.06 for the HEADS) and correlation coefficients ranging from 0.94 to 0.996.
30            During the Northern Front Range Air Quality Study in Colorado (Watson et al., 1998),
31     the automated nitrate monitor captured the 12-minute variability in fine particle nitrate

       June 2003                               2B-24       DRAFT-DO NOT QUOTE OR CITE

-------
 1      concentrations with a precision of approximately ±0.5 |ig/m3 (Chow et al., 1998).  A comparison
 2      with denuded filter measurements followed by ion chromatographic analysis (Chow and Watson,
 3      1999) showed agreement within ±0.6 |ig/m3 for most of the measurements, but exhibited a
 4      discrepancy of a factor of two for the elevated nitrate periods. More recent intercomparisons
 5      took place during the 1997 Southern California Ozone Study (SCOS97) in Riverside, CA.
 6      Comparisons with 14 days of 24-hour denuder-filter sampling gave a correlation coefficient of
 7      R2 = 0.87 and showed no significant bias (i.e., the regression slope is not significantly different
 8      from 1). As currently configured, the system has a detection limit of 0.7 |ig/m3 and a precision
 9      of0.2|ig/m3.
10
11      Sulfate
12            Continuous methods for the quantification of aerosol  sulfur compounds first remove
13      gaseous sulfur (e.g.,  SO2, H2S) from the sample stream by a diffusion tube denuder followed by
14      the analysis of particulate sulfur (Cobourn et al., 1978; Durham et al., 1978; Huntzicker et al.,
15      1978; Mueller and Collins, 1980; Tanner et al.,  1980). Another approach is to measure total
16      sulfur and gaseous sulfur separately by alternately removing  particles from the sample stream.
17      Particulate sulfur is obtained as the difference between the total and gaseous sulfur (Kittelson
18      et al., 1978). The total sulfur content is measured by a flame photometric detector (FPD) by
19      introducing the sampling stream into a fuel-rich hydrogen-air flame (e.g., Stevens et al., 1969;
20      Farwell and Rasmussen, 1976) that reduces sulfur compounds and measures the intensity of the
21      chemiluminescence from electronically excited sulfur molecules (S2*).
22            Because formation of S2* requires two sulfur atoms, the intensity of the
23      chemiluminescence is theoretically proportional to the square of the concentration of molecules
24      that contain a single  sulfur atom. In practice, the exponent is between one and two and depends
25      on the sulfur compound being analyzed (Dagnall et al., 1967; Stevens et al., 1971). Calibrations
26      are performed using  both particles and gases as standards.  The FPD can also be replaced by a
27      chemiluminescent reaction with ozone that minimizes the potential for interference and provides
28      a faster response time (Benner and Stedman, 1989, 1990).
29            Capabilities added to the basic system include in situ  thermal analysis and sulfuric acid
30      speciation (Cobourn et al., 1978; Huntzicker et al., 1978; Tanner et al., 1980; Cobourn and
31      Husar, 1982) ). Sensitivities for particulate sulfur as low as 0.1 |ig/m3, with time resolution

        June 2003                               2B-25       DRAFT-DO NOT QUOTE OR CITE

-------
 1      ranging from 1 to 30 min, have been reported. Continuous measurements of particulate sulfur
 2      content have also been obtained by on-line x-ray fluorescence analysis with resolution of 30 min
 3      or less (Jaklevic et al., 1981b).  During a field-intercomparison study of five different sulfur
 4      instruments, Camp et al. (1982) reported four out of five FPD systems agreed to within ±5%
 5      during a one-week sampling period.
 6
 7
 8      2B.4  OTHER CONTINUOUS MEASUREMENTS
 9      Continuous Ion Chromatography of Water-Soluble Ions
10            Dasgupta and Slanina have independently developed particle collection systems that
11      grow particles by increasing the relative humidity and collect the particles in an aqueous solution
12      suitable for injection into an ion chromatography (Simon and Dasgupta, 1995; Khlystov et al.,
13      1995). Automation of these systems yield semi-continuous monitors for those ions that can be
14      determined using ion chromatography. A similar system suing a particle size magnifier has been
15      reported by Weber et al. (2001).
16
17      Determination of Aerosol Surface Area in Real Time
18            Aerosol surface area is an important aerosol property for health effects research.
19      However, methods for on-line measurement of surface area are not widely available.  Woo et al.
20      (200Ib) used three continuous aerosol sensors to determine aerosol surface area. They used a
21      condensation particle counter (CPC, TSI, Inc., Model 3020), an aerosol mass concentration
22      monitor (MCM, TSI, Inc., Model 8520), and an  electrical aerosol detector (BAD, TSI, Inc.,
23      Model 3070) for measuring particle charge concentration. The three sensor signals were
24      inverted to obtain the aerosol size distribution, using a log-normal size distribution model (by
25      minimizing the difference between the measured signals and the theoretical values based upon a
26      size distribution model, the  instrument calibration, and its theoretical responses). The log-
27      normal function was then integrated to calculate the total surface area concentration.  Woo et al.
28      (200Ib) demonstrated that this method can give  near real-time measurements of aerosol surface
29      area.
30
        June 2003                                2B-26       DRAFT-DO NOT QUOTE OR CITE

-------
 1      Light Scattering
 2            A variety of types of nephelometers that integrate aerosol light scattering over various
 3      solid angles are available (McMurry, 2000).  When used to measure visibility, e.g., to provide
 4      pilots with realtime data on visual range, it is desirable to include the light scattering due to
 5      particle-bound water. However, when used as an indicator of fine particle mass, it is desirable to
 6      exclude particle-bound water.  This is frequently done by heating the ambient aerosol to a low
 7      reference relative humidity of 40%. However, this heating has the potential of also causing the
 8      loss of semivolatile components of the aerosol. The evaporation of ammonium nitrate aerosol in
 9      a heated nephelometer has been examined. Bergin et al. (1997) conducted laboratory
10      experiments at low relative humidity (-10%) and as a function of temperature (27-47 °C), mean
11      residence time in the nephelometer, and initial particle  size distribution.  The evaporation of
12      ammonium nitrate aerosol was also modeled for comparison and was found to describe
13      accurately the decrease in aerosol scattering coefficient as a function of aerosol physical
14      properties and nephelometer operating conditions. Bergin et al. (1997) determined an upper
15      limit estimate of the decrease in the aerosol light scattering coefficient at 450 nm due to
16      evaporation for typical field conditions.  The model estimates for their worst-case scenario
17      suggest that the decrease in the aerosol  scattering coefficient could be roughly 40%.  Under most
18      conditions, however, they estimate that the decrease in aerosol scattering coefficient is generally
19      expected to be less than 20%.
20
        June 2003                                2B-27        DRAFT-DO NOT QUOTE OR CITE

-------
 i        3.  CONCENTRATIONS, SOURCES, AND EMISSIONS
 2             OF ATMOSPHERIC PARTICULATE  MATTER
 3
 4
 5     3.1    INTRODUCTION
 6          This chapter discusses topics covered in Chapter 5 (Sources and Emissions of Atmospheric
 7     Particles) and Chapter 6 (Environmental Concentrations) of the previous document, Air Quality
 8     Criteria for Particulate Matter or " 1996 PM AQCD" (U. S. Environmental Protection Agency,
 9     1996) and presents updates to these materials where available.
10          Information about concentrations, the composition, and the spatial and temporal variability
11     of ambient particles across the United States is presented in Section 3.2. Ambient concentration
12     data obtained during the first three years of operation of the recently deployed nationwide
13     network of Federal Reference Method PM25 monitors in twenty-seven metropolitan statistical
14     areas (MSAs) are presented and analyzed in Appendix 3A. Initial data from the pilot method
15     evaluation study for the national speciation network are presented in Appendix 3B.  Results of
16     field studies that have characterized the composition of organic compounds in ambient particles
17     are summarized in Appendix 3C to complement the data for the inorganic composition of
18     ambient particles presented in Appendix 6 A of the 1996 PM AQCD and Appendix 3B of this
19     document.  Data for characterizing the daily and seasonal variability of PM25 concentrations are
20     discussed in Section 3.2.1; the intra-day variability of PM25 concentrations is discussed in
21     Section 3.2.2; relationships among different size fractions are discussed in Section 3.2.3;
22     interrelationships and correlations among PM components are discussed in Section 3.2.4; and the
23     spatial variability of various PM components is discussed in Section 3.2.5.
24          Unlike gaseous criteria pollutants (SO2, NO2, CO, O3), which are well-defined chemical
25     entities, atmospheric parti culate matter (PM) is composed of a variety of particles differing in
26     size and chemical composition.  Therefore, sources of each component of the atmospheric
27     aerosol must be considered in turn. Differences in the composition  of particles emitted by
28     different sources also will lead to spatial and temporal heterogeneity in the composition of the
29     atmospheric aerosol. The nature of the sources and the composition of the emissions from these
30     sources are discussed in Section 3.3. The chemistry of formation of secondary PM from gaseous
31     precursors is discussed in Section  3.3.1. Estimates of contributions of various sources to

       June 2003                                3-1         DRAFT-DO NOT QUOTE OR CITE

-------
 1      ambient PM levels given by source apportionment studies also are presented in Section 3.3.2.
 2      More detailed information about the composition of emissions from various sources is given in
 3      Appendix 3D. The determination of "policy-relevant background" (PRB) concentrations,
 4      including contributions from natural local sources and the long-range transport of PM from
 5      sources outside the United States, is discussed in Section 3.3.3. Reviews of transport of PM and
 6      its precursors within the United States can be found in the North American Research Strategy for
 7      Tropospheric Ozone (NARSTO) Fine Particle Assessment (NARSTO, 2002).  More detailed
 8      information regarding sulfur and nitrogen species can be found in Hidy (1994). Further
 9      information about PM concentrations observed at relatively remote monitoring sites (RRMS),
10      i.e., monitoring sites located in relatively remote areas that are not obviously influenced by local
11      anthropogenic sources, is given in Appendix 3E. Because PM is composed of both primary and
12      secondary constituents, emissions of both the primary components and the gaseous precursors of
13      secondary PM must be considered. Nationwide emissions estimates of primary PM and
14      precursors to secondary PM are discussed in Section 3.3.4, and uncertainties in emissions
15      estimates in Section 3.3.5.
16           The organization of topics in this chapter (ambient measurements, source characterization
17      and apportionment, and emissions inventories) reflects,  in a broad sense, the order in which these
18      topics are addressed in scientific studies and,  arguably, the increasing levels of uncertainty that
19      are associated with these topics.
20
21
22      3.2   PATTERNS AND TRENDS IN AMBIENT PM CONCENTRATIONS
23           Considerable data exists for characterizing PM10 mass concentrations and trends, and those
24      available at the time were presented in the 1996 PM AQCD.  In contrast, data sets for
25      characterizing PM2 5 and PM10_2 5 mass or trends were not as extensive.  Sources of data for PM2 5
26      (fine) and PM10_25 (coarse) discussed in the  1996 PM AQCD include EPA's Aerometric
27      Information Retrieval System (AIRS; U.S. Environmental Protection Agency,  2000a);
28      Interagency Monitoring of Protected Visual Environments (IMPROVE; Eldred and Cahill, 1994;
29      Cahill, 1996); the California Air Resources Board (CARB) Data Base (California Air Resources
30      Board, 1995); the Harvard Six-Cities Data Base (Spengler et al., 1986; Neas, 1996); and the
31      Harvard-Philadelphia Data Base (Koutrakis, 1995). The Inhalable Particulate Network (IPN)

        June 2003                                 3-2        DRAFT-DO NOT QUOTE OR CITE

-------
 1      (Inhalable Paniculate Network, 1985; Rodes and Evans, 1985) provided TSP, PM15, and PM25
 2      data but only a small amount of PM10 data.
 3          New sources of PM data include the recently deployed nationwide PM2 5 compliance-
 4      monitoring network, which provides mass measurements using a Federal Reference Method
 5      (FRM). This section summarizes data obtained by this network from 1999 to 2001 and provides
 6      an approximate characterization of nationwide PM10_25 concentrations by comparing PM10 to
 7      PM2 5 measurements at sites where both types of compliance monitors are located.  Various
 8      aspects of these data are presented in greater detail in Appendix 3 A.  In addition, a small number
 9      of recent studies in which daily mass and composition measurements are available for extended
10      periods are discussed in this section. The results of quality-assured aerosol composition data
11      obtained by X-ray fluorescence (XRF) and by analyses of organic carbon (OC) and elemental
12      carbon (EC) for thirteen urban areas from the methods evaluation study for the national PM2 5
13      speciation network are presented in Appendix 3B.  The terms organic carbon (OC) and elemental
14      carbon (EC) are subject to some ambiguity, and the meanings of these terms discussed in Section
15      2.2.7 and Appendix 2.B.2 are applied here.
16          Organic compounds contribute from 10 to 70% of the dry fine particle mass in the
17      atmosphere (see Appendix 3C). However, concentrations and the composition of organic PM
18      are poorly characterized, and its formation mechanisms are poorly understood. Particulate
19      organic matter is an aggregate of hundreds of individual compounds  spanning a wide range of
20      chemical and thermodynamic properties (Saxena and Hildemann, 1996).  Some of the organic
21      compounds are "semivolatile" (i.e., they have atmospheric concentrations and saturation vapor
22      pressures such that both gaseous and condensed phases exist in equilibrium in the atmosphere).
23      The presence of semivolatile or multiphase organic compounds complicates the sampling
24      process. Organic compounds originally in the gas phase may be absorbed on glass or quartz
25      filter fibers and create a positive artifact. Conversely, semivolatile compounds originally present
26      in the condensed phase may evaporate from particles collected on glass, quartz, or Teflon filters
27      creating a negative artifact. In addition, no single analytical technique is currently capable of
28      analyzing the entire range of organic compounds present in atmospheric PM. Rigorous
29      analytical methods are able to identify only  10 to 20% of the organic PM mass on the molecular
30      level (Rogge et al., 1993), and only about 50% of the condensed phase compounds can be
31      identified in smog chamber studies of specific compounds (Forstner et al., 1997a,b).

        June 2003                                 3-3          DRAFT-DO NOT QUOTE OR CITE

-------
 1      Measurement techniques and associated artifacts are discussed in Section 2.2.3.2.  Information
 2      on the identification and concentration of the many different organic compounds identified in
 3      atmospheric samples obtained during the 1990s is given in Appendix 3C.
 4           Summary tables giving the results of 66 field studies that obtained data for the composition
 5      of particles in the PM2 5, PM10_25, or PM10 size ranges were presented in Appendix 6 A of the
 6      1996 PM AQCD.  The summary tables included data for mass, organic carbon, elemental
 7      carbon, nitrate, sulfate, and trace elements. Data from the studies were presented for the eastern,
 8      western, and central United States.  It should be noted that these studies took place at various
 9      times and lasted for various durations over a 20-year period, and there may have been significant
10      changes in the concentrations of many species between the times when these studies were
11      conducted and now. These changes resulted from a number of factors (e.g., pollution controls,
12      technological advances, land use changes, etc).
13           There were a number of discernible differences in the composition of particles across the
14      United States evident in the data sets listed in Appendix 6A in the 1996 PM AQCD (cf,
15      Figures 6-85a to 6-85c).  However, these differences can only be discussed in the context of the
16      uncertainties in the measurements of the main components (sulfate, organic carbon, elemental
17      carbon, crustal material, ammonium, and nitrate).  Sulfate, followed by crustal materials, has the
18      smallest uncertainty associated with its measurement among all the components listed. Sulfate
19      constituted about 38% of PM25 in the aerosol composition studies in the eastern United States
20      and was the major identifiable component of PM2 5. However, it constituted only about 11% of
21      PM2 5 in the studies listed for the western United States. The contribution of crustal materials to
22      PM2 5 ranged from about 4% in the East to about 15% in the West. The contribution of
23      unidentified material (possibly consisting mostly of water of hydration)  ranged from 23% in the
24      East to 0% in the West.  The contribution of elemental carbon to PM2 5 ranged from about 4% in
25      the East to about 15% in the West.  Organic compounds constituted about 21% of PM25 in the
26      eastern United States, reaching to about 39% for the studies listed in the western United States.
27      However, uncertainties for organic carbon, elemental carbon, ammonium, and nitrate are larger
28      than for sulfate and crustal material. Concentrations of organic carbon were multiplied by
29      a factor of 1.4 when calculating mass to account for the presence of oxygen, nitrogen, and
30      hydrogen in the organic compounds.  This factor may vary among different areas and may
31      represent the lowest reasonable estimate for an urban aerosol (Turpin and Lim, 2001).

        June  2003                                  3-4         DRAFT-DO NOT QUOTE OR CITE

-------
 1      In addition, the samples collected in the studies were subject to a number of sampling artifacts
 2      involving the adsorption of gases and the evaporation of volatile components that either formed
 3      on the filters or were present in the ambient particles.  The values reported for organic carbon
 4      and elemental carbon in filter samples depend strongly on the specific analysis method used
 5      (Chow et al., 2002) as discussed extensively in Chapter 2.
 6           Crustal materials constitute from 52% of PM10_25 in the eastern United States to 70% of
 7      PM10_25 in the western United States as indicated by the studies in Appendix 6 A of the 1996 PM
 8      AQCD.  The fraction of unidentified material in PM10_2 5 varied from 41% in the eastern United
 9      States to 27% in the western United States.  However, in the vast majority of these studies, no
10      attempt was made to characterize organic components or nitrate in the PM10_25 size fraction.
11      It should also be remembered that a small fraction (typically about 10%) of PM25 is entrained
12      into the flow of the channel of the dichotomous sampler that collects the PM10_2 5 sample and that
13      there may be errors invoked during the procedure used to account for this entrainment. Even if
14      analyses of total OC were available, they would not be able to distinguish between primary
15      biological particles (PBP), which include microorganisms and fragments of living things,  and
16      organic compounds of miscellaneous origin in surface deposits on filters. A clear distinction
17      should be made between PBP and primary OC emitted by organisms such as waxes coating the
18      surfaces of leaves,  seeds, fruits, stems, pollen, fungi, and on insects.  Indeed, the fields of view
19      of many photomicrographs of PM samples obtained by scanning electron microscopy, are
20      dominated by large numbers of pollens, plant and insect fragments, and microorganisms.
21      Bioaerosols such as pollens, fungal spores, and most bacteria are expected to be found mainly  in
22      the coarse size fraction. However, allergens from pollens can also be found in respirable
23      particles (Monn, 2001; Taylor et al., 2002). Matthias-Maser (1998) summarized information
24      about the size distribution of PBP in and around Mainz, Germany, in what is perhaps the most
25      complete study of this sort. Matthias-Maser found that PBP constituted up to 30% of total
26      particle number or volume in the size range from about 0.35 jim to about 50 jim on an annual
27      basis. Additionally, whereas the contribution of PBP to the total aerosol volume did not change
28      appreciably with season, the contribution of PBP to total particle number ranged from about 10%
29      in December and March to about 25% in June and October.
30           Data for the chemical composition of particles in a number of national parks and remote
31      areas have been collected for a number of years by the IMPROVE network.  Concentrations

        June 2003                                3-5         DRAFT-DO NOT QUOTE OR CITE

-------
 1      have been reported for sulfate, nitrate, light absorbing carbon, organic carbon, and soil
 2      components. With the collection of compositional data by the speciation network, more synoptic
 3      (i.e., concurrent) coverage will be obtained for these constituents from relatively remote to urban
 4      environments across the United States.
 5
 6      PM10 Concentrations and Trends
 1          Nationwide PM10 annual mean concentrations on a county-wide basis from the AIRS
 8      database for calendar years 1999, 2000, and 2001 are shown in Figure 3-la. Concentrations in
 9      most areas of the country were below the level of the PM10 annual standard of 50 |ig/m3 from
10      1999 through 2001.  Further information about the attainment status of different areas can be
11      found in the U.S. Environmental Protection Agency's Air Quality Trends Reports.  The median
12      annual mean PM10 concentration for this three year period was about 23 |ig/m3;  and the
13      95th percentile value was about 38 |ig/m3. Concentrations flagged as natural events (e.g.,
14      resulting form high winds, wildfires, volcanic eruptions) or exceptional events (e.g.,
15      construction, prescribed burning) are not included in the calculations. The procedures for
16      calculating the annual means at the site level follow 40 Code of Federal Regulations (CFR)
17      Part 50 Appendix K (requiring 75 percent completeness per quarter for all three years). The
18      98th percentile concentrations from the monitor showing the highest value in a given county over
19      the three year period are shown in Figure 3-lb.  In these and in similar maps for PM25 and
20      PM10_2 5, cut points were chosen at the 50th and 95th percentile concentrations. These cut points
21      were chosen as they represent the standard metrics for characterizing important  aspects of human
22      exposure used by the U.S. Environmental Protection Agency.  Of course, any other percentiles or
23      statistics that are believed to be useful for characterizing human exposures could also be used.
24      As shown by the blank areas on the maps, the picture is not complete because some monitoring
25      locations did not record valid data for all four quarters or recorded fewer than 11 samples in one
26      or more quarters or counties simply did not have monitors.  Similar considerations apply to the
27      maps shown later for PM2 5 and PM10_2 5.  It should also be noted that the area of counties can be
28      much greater in the West than in the East. As a result, the density of monitors may appear to be
29      greater in the West and air quality may appear to be worse over much larger areas in the West
30      than in the East. Concentrations are shown at the county level because this is the typical scale
        June 2003                                  3-6        DRAFT-DO NOT QUOTE OR CITE

-------
                 Concentration (ng/m3)
                                    0 < x < 23
                                                     23 < x < 38
                                                                     x>38
Figure 3-la.  1999-2001 county-wide average annual mean PMjo concentrations (ug/m3)
              for counties with PM10 monitors.

Source: Aerometric Information Retrieval System (AIRS; U.S. Environmental Protection Agency, 2002b).
                  Concentration (ng/m3)
                                    0 < x < 72
                                                 iSEs 72170
                                                                             10
Figure 3-lb.  1999-2001 highest county-wide 98th percentile 24-h average PM
              concentrations (ug/m3) for counties with PM10 monitors.

Source: Aerometric Information Retrieval System (AIRS; U.S. Environmental Protection Agency, 2002b).
June 2003
                                            5-7
DRAFT-DO NOT QUOTE OR CITE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
used in many health outcome studies.  MSA or multi-county scales have also been used in a
number of studies, e.g., Schwartz et al. (1996), or the NMMAPS study.
     Nationwide trends in annual mean PM10 concentrations from 1992 through 2001 (based on
data obtained at 119 rural sites, 297 suburban sites, and 316 urban sites reporting to AIRS) are
shown in Figure 3-2 (U.S. Environmental Protection Agency, 2002a).  Although average
concentration levels differ among types of sites with higher levels at urban and suburban sites
the nationwide data set shows a decrease of 14% that occurred mainly during the first half of the
record. PM10 concentrations basically leveled off during the last few years of the record but
there were indications of transient increases followed by decreases.  Figure 3-3 shows the annual
mean PM10 trend summarized by EPA region. Decreases in annual average PM10 concentrations
from 1990 to 1999 were largest in the Northwest (9.6 |ig/m3) and smallest in the south-central
United States (1.3 jig/m3).  Analyses of available TSP measurements obtained since 1950
indicate that mean TSP concentrations appear to have declined by about two- to three-fold in
urban areas between 1950 and 1980 (Lipfert, 1998).
                           30
                           25
                       I.  20
                re
                c
                o  10-
                o
                O
                                                                1992 - 2O01
                                                               Rural Sites (153}
                                                          ——— Suburban Sites (297)
                                                          	 Urban Sites (316)
                                I     I     I     I     I     I     I     I     I     \
                               92   93   94   95   96    97    98    99   00   01
                                                     YEAR
        Figure 3-2. Nationwide trend in ambient PM10 concentration from 1992 through 2001.
        Source: U.S. Environmental Protection Agency (2002a).
       June 2003
                                                      DRAFT-DO NOT QUOTE OR CITE

-------
                                                                               20.529
                                                                                       18.287
              The National Trend
                 27.745   23.907
                 1992
                         2001
                      14%
Alaska is in
EPA Region 10;
Hawaii, EPA Region
and PuertoRico,
EPA Region 2.
                                                                       Note: These trends are
                                                                       nfluenced by the distribution
                                                                      of monitoring locations in a
                                                                     given region and, therefore,
                                                                     can be driven largely by urban
                                                                     concentrations. For this reason,
                                                                       they are not indicative of
                                                                       background regional
                                                                         concentrations.
        Figure 3-3.  Trend in PM10 annual mean concentrations by EPA region, 1992 through
                     2001 (ug/m3).
        Source: U.S. Environmental Protection Agency (2002a).
 1
 2
 3
 4
 5
 6
 1
 8
 9
10
11
12
13
PM2 5 Concentrations and Trends
     Nationwide annual mean PM25 concentrations obtained from data collected during 1999,
2000, and 2001 are shown in Figure 3-4a; and 98th percentile concentrations are shown in
Figure 3-4b. Quantities shown in Figure 3-4a and 3-4b were calculated for individual counties.
Annual mean concentrations shown in Figure 3-4a were calculated on the basis of the latest
8 consecutive quarters containing at least 11 samples per quarter, and results from the highest
monitor were used to produce Figure 3-4b.  Data meeting these completeness criteria were
obtained at 897 sites in 592 counties. The median PM2 5 concentration nationwide was about
13 |ig/m3. Annual mean PM25 concentrations were above 17 |ig/m3 at 5% of the sites, mainly in
California and in the southeastern United States. The 98th percentile 24-h average concentrations
(as defined earlier for PM10) were below 47 |ig/m3 at 95% of the sites sampled. Most of the sites
with levels above this value are located in California.
        June 2003
                                             5-9
                                 DRAFT-DO NOT QUOTE OR CITE

-------
                  Concentration (ng/ms;
                                    017
Figure 3-4a.  1999-2001 county-wide average annual mean PM2 5 concentrations (ug/m3)
              for counties with PM2 5 monitors.

Source: Aerometric Information Retrieval System (AIRS; U.S. Environmental Protection Agency, 2002b).
                  Concentration (j.ig/m3)
                                    0 < x < 32
                                                 Um 32 < x < 47
                                                                    x>47
Figure 3-4b.  1999-2001 highest county-wide 98th percentile 24-h average PM2
                                                                            -2.5
              concentrations (ug/m3) for counties with PM2 5 monitors.

Source: Aerometric Information Retrieval System (AIRS; U.S. Environmental Protection Agency, 2002b).
June 2003
3-10
DRAFT-DO NOT QUOTE OR CITE

-------
 1           Annual average PM2 5 concentrations obtained as part of several health studies conducted
 2      in various locations in the United States and Canada from the late 1980s to the early 1990s are
 3      shown in Figure 3-5 (Bahadori et al., 2000a).  These studies include the Harvard six-cities study
 4      (Steubenville, OH; Watertown, MA; Portage, WI; Topeka, KS; St. Louis, MO; and Kingston-
 5      Harriman, TN); Particle Total Exposure Assessment Methodology (PTEAMS; Riverside, CA);
 6      Metropolitan Acid Aerosol Characterization Study (MAACS; Philadelphia, PA; Washington,
 7      DC; and Nashville, TN); South Boston Air Quality and Source Apportionment Study (Boston,
 8      MA); and NPMRMN (Phoenix, AZ). The remaining sites were part of the 24-cities study
 9      (Spengler et al.,  1996).
10           Sufficient data are not yet available to permit the calculation of nationwide trends of PM2 5
11      and PM10_2 5; however, some general emerging patterns can be discerned.  Darlington et al.
12      (1997) proposed that the consistent reductions in PM10 concentrations found in a wide variety of
13      environments ranging from urban to rural may have resulted from common factors or controls
14      that affected fine particles more strongly than  coarse particles. This is because fine particles
15      have longer atmospheric lifetimes than coarse particles and can be transported over longer
16      distances and, hence, can affect larger areas. Apart from the IMPROVE network of monitoring
17      sites located mainly in national parks, the longest time series of PM25 concentration and
18      composition data have been obtained by the California Air Resources Board (CARB). Their
19      data show that annual average PM2 5 concentrations decreased by about 50% in the South Coast
20      Air Basin, 35% in the  San Joaquin Valley, 30% in the San Francisco Bay Area, and  35% in the
21      Sacramento Valley from 1990 to 1995 (Dolislager and Motallebi, 1999).  PM25 data were
22      collected continuously from 1994 to 1998 as part of the children's health study in  12 southern
23      California communities (Taylor et al., 1998).  Data obtained at all sites show decreases in PM25
24      ranging from 2% at Santa Maria to 37% at San Dimas/Glendora from 1994 through  1998. These
25      decreases were accompanied by decreases in major components such as nitrate, sulfate,
26      ammonium, and acids.  Based on the analysis of PM2 5 data  sets collected prior to 1990, Lipfert
27      (1998) found that PM2 5 concentrations appear to have decreased by about 5% per year from
28      1970 to 1990 in a number of urban areas.  These declines were also found to be consistent with
29      decreases in emissions from combustion sources over that time period.
30
        June 2003                                 3-11        DRAFT-DO NOT QUOTE OR CITE

-------
to
O
o
H

6
o


o
H

O

O
H
W

O


O
HH
H
W
                                     PM25 Annual Standard
                                                   ' ^•^•^ A^^ £'<
                                                                                        -s ,.         ,

                    W4^'
              Figure 3-5.  Collection of annual distribution of 24-h average PM2 5 concentrations observed in U.S.

                         and Canadian health studies conducted during the 1980's and early 1990's.



              Source: Bahadori et al. (2000a).

-------
 1      PM10_2,5 Concentrations
 2           By using AIRS data for 1999, 2000, and 2001 obtained by the PM10 and PM2 5 compliance
 3      networks, it is possible to construct a picture of the distribution of coarse PM across the country.
 4      This is accomplished by pairing data from 208 compliance monitoring sites in  196 counties
 5      where PM10 and PM25 monitors are collocated and subtracting the mass concentrations of PM25
 6      from PM10.  Annual mean concentrations were calculated on the basis of the latest 8 consecutive
 7      quarters containing at least 11 samples per quarter. Nationwide annual mean PM10_2 5
 8      concentrations calculated by this approach are shown in Figure 3-6a.  Annual mean PM10_25
 9      concentrations ranged from about 1 to about 50 |ig/m3; the nationwide median  concentration was
10      about 10 |ig/m3; and 5% of the sites had mean concentrations greater than 28 |ig/m3. The higher
11      values occur mainly in the western United States, particularly in California.  The highest county -
12      wide 98th percentile PM10_2 5 concentrations based on this same data set are shown in Figure 3-6b.
13      Highest values in the western United States  are caused by dust raised locally either by natural
14      means or by anthropogenic activity.  It is not clear what the contribution of PBP to these values
15      may be.  Elevated dust levels are also found in southern Florida as the result of dust storms in
16      North Africa (Section 3.3.3) and trans-Atlantic transport.  In many areas, combined errors in the
17      PM2 5 and PM10 measurements may be similar to or even greater than the calculated PM10_2 5
18      concentrations.  Because of this and other potential problems with this approach (Section 3.2.1),
19      these results should be viewed with caution.
20
21      3.2.1   Seasonal Variability in PM Concentrations
22      PM25
23           Aspects of the spatial and temporal variability of PM2 5 concentrations  for 1999, 2000,
24      and 2001 in a number of metropolitan areas across the United States are presented in this and
25      following subsections.  Data for multiple sites in 27 urban areas across the United States have
26      been obtained from the AIRS data base and  analyzed for their seasonal variations and for their
27      spatial correlations and spatial uniformity in concentrations. Selection of these 27 MS As was
28      based on the criteria that data be available for at least 15 days in each calendar quarter of either a
29      three year period (1999, 2000, and 2001) or a two year period (2000 and 2001) at three or more
30      sites within that MSA.  In addition, a maximum of 11 sites per MSA were included for analysis.
31

        June 2003                                 3-13         DRAFT-DO NOT  QUOTE OR CITE

-------
Concentration (|ig/m3)
028
       Figure 3-6a.  1999-2000 estimated county-wide average annual mean PM10_25
                     concentrations (ug/m3) for counties with collocated PM2 5 and PM10 monitors.
       Source: Aerometric Information Retrieval System (AIRS; U.S. Environmental Protection Agency, 2002b).
 1     (In the Chicago and St. Louis MSAs, the 11 sites having the most observations were selected
 2     from a greater number of qualifying sites.) A number of aspects of the spatial and temporal
 3     variability of the 1999 PM2 5 data set were presented in Rizzo and Pinto (2001) based in part on
 4     analyses given in Fitz-Simons et al. (2000).
 5          Information regarding the seasonal variability in PM2 5 concentrations in four MSAs
 6     (Philadelphia,  PA; Cleveland, OH; Dallas, TX; and Los Angeles-Long Beach, CA) in the
 7     United States is summarized in Figures 3-7a through 3-7d. These four urban areas were chosen
 8     to illustrate some general features of the spatial and temporal variability found in the
 9     United States.  The figures show lowest, lower quartile, median, upper quartile, and highest
10     concentrations for each calendar quarter of 1999, 2000, and 2001 for the Cleveland, OH MSA
11     and for 2000 and 2001 for Philadelphia, Dallas and Los Angeles MSAs. For each monitoring
       June 2003
3-14
                                            DRAFT-DO NOT QUOTE OR CITE

-------
                      Alaska
                 Concentration (
0 < x<29
                                                           29 < x < 83
                  x>83
       Figure 3-6b.  1999-2000 estimated county-wide highest 98th percentile 24-h PM10_2 5
                     concentrations (ug/m3) for counties with collocated PM2 5 and PM10 monitors.
       Source: Aerometric Information Retrieval System (AIRS; U.S. Environmental Protection Agency, 2002b).
 1     site, the AIRS ID numbers, annual mean concentrations, the number of observations, and
 2     standard deviations are also shown. Data for multiple sites within these MSAs are shown to
 3     provide an indication of the degree of inter-site variability. Data for these MSAs and an
 4     additional twenty-three MSAs, criteria used for site selection, and additional descriptions of the
 5     data are given in Appendix 3 A.
 6          Annual mean PM2 5 concentrations at individual monitoring sites in the MSAs examined
 7     ranged from about 6 |ig/m3 to about 30 |ig/m3.  The lowest values were found in rural portions of
 8     the MSAs examined, typically near the perimeter of the MSA. The two highest mean
 9     concentrations were found in the Riverside and Los Angeles-Long Beach MSAs in southern
10     California while the three lowest means were found in the Northwest ( Portland, OR; Boise, ID;
11     Seattle, WA). MSAs situated along the Eastern seaboard (Washington, DC; Philadelphia, PA;
       June 2003
          3-15
DRAFT-DO NOT QUOTE OR CITE

-------
                                a. Philadelphia, PA (2000 - 2001}
                    ARiSID* 3400T1Q07 340155001 4201TOQ12 420450002 420910013 421010004 421010136
                      Mean
                       0»»
                       SO
                       to-i
                   CO
                   I
                    a.
WJ
19?
92
14.8
208
14.1
21?
85
             230
             S.6
14 3
221
89
1S.7
eio
                            i  i
                   I!
                           1734  1234  1234  1234   1234
                                              Quarter
                                 b.  Cleveland, OH (1999-2001)
               ARIStD* 3903S0013        390350060 390350K8 3S0S5Q066 390351002 390851001 380932003
fesn
Obs
SD
60-
Zfl —
0 —
163
388
9.6

ll
301
931
M.S

II
If I
184
353
9.8

II III!
i
1
i S
s

III
rs
40
.a

1


J
147 1
332 , 3
_B.4___j__J

l| 1
SO . 1*0 , 1$2
61 34? . 298
.2 8.4 8.8
' I I
I'l III I"'.
il! iilj Ml|
                      1234  1234  12S4  1i34   123*
                                              Quarter
                                                        1 2 » 4  1234   1S34
Figure 3-7a,b. Quarterly distribution of 24-h average PM2 5 concentrations for selected
               monitors in the (a) Philadelphia, PA and (b) Cleveland, OH. Values for the
               lowest, lower quartile, median, upper quartile and highest concentrations
               are shown in the figures. The AIRS site ID number, annual mean
               concentration (ug/m3), number of observations, and standard deviation of
               the data are shown above the figures for each site.
June 2003
               3-16
                     DRAFT-DO NOT QUOTE OR CITE

-------
                                 c.  Dallas, TX (2000 -2001)
                   ARIS ID# 480850005 481130020 481130035 481130050 481130057 481130069 481130087
                    Mean
                     Obs
                     SD
                     40-
                   0>
                   ^3.

                   if)
                   cvi

                   Q.
11.5
234
5.6
12.4
677
5.7
12.9
222
5.8
13.3
644
5.8
13.7
215
6.1
12.7
667
5.7
11.7
216
5.4
                         1234  1234  1234  1234  1234  1234  1234
                                            Quarter


                                 d.  Los Angeles, CA(2000 -2001)

                   ARISID* 060370002 0600371103  060371201  060371301  060372005  060374002
                     Mean
                     Obs
                      SD
                     100—1
                  n

                  ~
                  •=:  50—
 20.9
 641
 13.3
  22.5
  656
  13.5
   18.3
   217
   11.7
    23.7
    237
    13.4
     20.1
     220
     11.9
     20.3
     621
     12.4
                         I!'"

                           I'

      mil   i  w
                          1234  1234  1234   1234.  1234   1234
                                            Quarter
Figure 3-7c,d.  Quarterly distribution of 24-h average PM2 5 concentrations for selected
               monitors in the (c) Dallas, TX and (d) Los Angeles, CA MSAs.  Values
               for the lowest, lower quartile, median, upper quartile and highest
               concentrations are shown in the figures.  The AIRS site ID number,
               annual mean concentration (jig/m3), number of observations, and
               standard deviation of the data are shown above the figures for each site.
June 2003
               3-17
                     DRAFT-DO NOT QUOTE OR CITE

-------
 1      Norfolk, VA) tend to have lower mean PM2 5 concentrations than MSAs in the north-central
 2      United States (Steubenville, OH; Cleveland, OH; Pittsburgh, PA; Chicago, IL; Detroit, MI;
 3      Gary, IN; Appendix 3 A). In addition, average PM2 5 concentrations tended to be lower in 1999,
 4      2000, and 2001 in urban areas given in Appendix 3 A compared to the concentrations observed
 5      during pollution-health outcome studies conducted in those five urban areas where these overlap
 6      (Figure 3-5). It should be noted that there are no data demonstrating the comparability of the
 7      monitors used in the studies shown in Figure 3-5 and the FRM.
 8           The patterns of seasonal variability observed in the MSAs examined are complex.  In the
 9      Philadelphia, PA MSA, highest median concentrations occur at all but one site in the first quarter
10      (cf, Figure 3-7a). First quarter maxima are also observed at all sites in the Detroit, MI and
11      Chicago, IL MSAs (cf, Figures 3 A-l 1 and 3 A-14). The Cleveland, OH, MSA (Figure 3-7b) by
12      contrast does not have a clear seasonal pattern.  In  several MSAs examined in the  south and
13      eastern United States (Atlanta, GA; Baton Rouge, LA; Birmingham, AL; Columbia, SC;
14      and Washington, DC), highest median concentrations occur at all sites during the third calendar
15      quarter (i.e., summer months; cf, Appendix 3 A). Sites in Dallas, TX, (Figure 3-7c) as in the
16      other southern cities mentioned above, generally exhibit third quarter median concentration
17      maxima. Highest median concentrations occur during the fourth calendar quarter  in MSAs in the
18      western United States as in the Los Angeles,  CA MSA (Figure 3-7d) although there are
19      exceptions at individual sites  in the Riverside, CA MSA (Figure 3A-26).
20           Lowest median concentrations occur mainly during the first or fourth quarters at most sites
21      in the eastern United States as well as in Cleveland, OH and Dallas, TX (Figures 3-7b and 3-7c)
22      while some occur during the second quarter (Philadelphia, PA; Figure 3-7a). Moving westward,
23      the seasonal pattern is not as distinct: lowest median concentrations occur in any quarter, but
24      usually in the second or third  quarter as in the Chicago, Detroit, and Los Angeles-Long Beach
25      (Figure 3-7d) MSAs. With the exception of Los Angeles, CA and Riverside, CA,  sites in the
26      West show lowest median concentrations in the second quarter. In most of the MSAs examined,
27      seasonal variations follow a similar pattern at all of the sites within the MSA, but in a few MSAs
28      there are noticeable differences in the seasonal pattern between sites.  The large-scale differences
29      in seasonal variability between MSAs tend to follow differences in the major categories of PM
30      sources affecting the monitoring sites. Local heating by wood burning during the  colder months
31      is practiced more widely  in the western United  States than in the eastern United States.

        June 2003                                 3-18        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Prolonged winter stagnation events are also more common in western mountain valleys during
 2      winter than in sections of the eastern United States located in relatively flat topography.  Hence,
 3      winter maxima and greater variability in PM2 5 concentrations across sites are expected in the
 4      West due to the influence of the local sources. On the other hand, photochemical production of
 5      secondary PM, especially sulfate, occurs over wide areas in relatively homogeneous air masses
 6      during the summer months in the eastern United States.  Because sulfates (along with associated
 7      cations and water) constitute the major fraction of summertime PM2 5 in the East, there is greater
 8      uniformity in third quarter PM concentrations within eastern MSAs (cf, Appendix 3 A).
 9           Maximum twenty-four hour average concentrations shown in the box plots in Figures 3-7a
10      to 3-7d and in Figures 3A-1 to 3A-27 do not necessarily follow the same seasonal pattern as the
11      median concentrations.  There is no clear relation between the maximum and the median
12      concentrations evident in the Philadelphia, PA data set (Figure 3-7a). In Cleveland, OH
13      (Figure 3-7b), maximum concentrations occur during the second or fourth quarters, and highest
14      median concentrations generally occur during the third or first quarters. In Dallas, TX
15      (Figure 3-7c), maximum concentrations generally occur during the fourth quarter, but highest
16      median concentrations tend to  occur during the second or third quarter.  In the Los Angeles-Long
17      Beach MSA (Figure 3-7d), the maximum and highest median concentrations occur together in
18      the fourth quarter with the exception of one site. Peak individual concentrations likely reflect
19      the occurrence of transient events such as forest fires (mainly in the West) or episodes of
20      secondary PM production (mainly in the East).  However, chemical analyses of filter samples or
21      other evidence should be used  to determine specific causes in particular locations.
22           There have been a few studies that have characterized PM2 5 and PM10  concentrations in
23      major urban areas.  The Metropolitan Aerosol Acidity Characterization Study (MAACS)
24      (Bahadori et al., 2000b) characterized the levels and the spatial and temporal variability of PM25
25      PM10, and acidic sulfate concentrations in four cities in the eastern United States (Philadelphia,
26      PA; Washington, D.C.; Nashville, TN; and Boston, MA). Seasonal variations in PM2 5 and PM10
27      concentrations obtained during the course of this study are shown in Figure 3-8.  The data for the
28      four cities included in MAACS are presented in a box plot showing the lowest, lowest tenth
29      percentile, lowest quartile, median, highest quartile, highest tenth percentile, and highest PM2 5
30      and PM10 values.  Mean and highest PM2 5 and PM10 concentrations are found during the summer
31

        June 2003                                3-19        DRAFT-DO NOT QUOTE OR CITE

-------
/u-

00
E 60-
D)
r 50-
g
'^
(5
£ 40-
c
o
o
<= 30-
o
o
| 20-
3
O
t 10-
03
Q.
0J

Philadelphia Washington


O o
_


*



T
o
M

1
i

°1




o

o •

y O




i .

T T
1 *
O T


I 1
]

I I


o r




O


I
;
1
o

o
'









o| 0 *
o

o
o


1

-


0 1
J


I
i . J

I °i °
1 » ,1 " °
O


1

1

Nashvil













0

1

t

o






o






c
o



o
T


i
o J.
o

1 1 1
1

e Boston

I=IPM10
^PM2.5




o

o
o
o





•
T
1
O



o
T
1 | 0
^ -Or
! ffl
I Tl »
o y




o
•-i-]T
o| «|
0 "fi i ° "
o

1 1
1 1
                           SPSU F W  SPSU F  W  SPSU F  W
                                                  Season
                                                              SPSU F  W
       Figure 3-8.  Seasonal concentrations of PM2 5 and PM10 measured in the four MAACS
                   cities. The data show the lowest, lowest tenth percentile, lowest quartile,
                   median highest quartile, highest tenth percentile, and highest PM2 5 24-h
                   average values. The dashed line shows the level of the annual PM2 5 standard.
       Source: Bahadori et al. (2000b).
 1
 2
 3
 4
 5
 9
10
11
in all four cities (in contrast with Figure 3-7a) although the seasonal pattern in Boston appears to
be more nearly bimodal with an additional winter peak.
PM
   10-2.5
     Of the 27 MS As selected for analysis of PM25 data (Appendix 3 A), 17 have multiple sites
satisfying the PM10_25 data completeness criteria. A subset of 53 qualifying sites were selected
for analysis of PM10_2 5 data.  Each qualifying site has 11 or more observations per calendar
quarter for twelve consecutive quarters (1999-2001, 2 MS As), eight quarters (1999 + 2000 or
2000 + 2001, 7 MS As) or four quarters (2000 or 2001, 8 MS As).  In addition, data from sites
within the same MSA represent the same year (or years) of observations, so that intersite
comparisons within an MSA are based on the same reporting period. However, comparisons
       June 2003
                                         3-20
DRAFT-DO NOT QUOTE OR CITE

-------
 1     between different MSAs can involve different annual reporting periods. As can be seen by
 2     comparing Figures 3-7a,b,c,d and Figures 3-9a,b,c the number of observations used for
 3     summarizing PM10_2 5 data is much lower than that measured for PM2 5. Unfortunately, fewer
 4     qualifying MSAs for PM10_2 5 result in geographical coverage that is more spotty than with PM2 5.
 5     Five of the seven MSAs representing the eastern United States for PM25, for example, did not
 6     meet completeness criteria for PM10_2 5, leaving only Tampa, FL, and Columbia, SC, to represent
 7     the East.  Secondly, it can readily be seen, e.g., for Cleveland, OH (Figure 3-9a) and in
 8     Figures 3A-28 through 3A-44, that a number of PM10_25 concentrations are negative.
 9     For example, in 2001, approximately one-fourth of all PM10_25 concentrations at the three
10     qualifying sites in the  Gary, IN MSA were negative. (The negative estimates have been included
11     in the calculation of mean concentrations.)  There are a number of reasons for the negative
12     concentration estimates, many of which arise because the ratios of PM25 to PM10 are based on
13     two independent measurements. Measurement imprecision plays a role when the ratios are large
14     and concentrations are small. Differences in the behavior of semivolatile components in the two
15     samplers could occur; and the results may also reflect errors in sampler placement, field,
16     laboratory, or data processing procedures. For the reasons cited above, caution should be
17     exercised when attempting to interpret results for PM10_25 based on current network collocated
18     PM2 5 and PM10 monitors.
19          Annual mean PM10_2 5 concentrations at individual monitoring sites range from about
20     6 |ig/m3  (Portland, OR) to about 33 |ig/m3 (Riverside, CA).  (Gary, IN sites were excluded
21     because  of numerous negative PM10_25 concentrations, and one site in Riverside, CA, affected by
22     a local dust event was excluded.) The three highest annual  mean concentrations are observed in
23     Riverside, CA; Los Angeles-Long Beach, CA; and  Salt Lake City, UT; and the lowest
24     (excluding Gary) were observed in Portland, OR. For the remaining MSAs, there does not
25     appear to be a significant geographical trend associated with the annual mean PM10_2 5
26     concentration. Within MSAs, the lowest concentrations are frequently observed at sites near the
27     perimeter of the MSA although it must be noted that the number of sites is limited.
28          Within  each MSA, collocated PM2 5 and PM10_2 5 concentrations, averaged over the same
29     years, were compared. The mean PM2 5:PM10_25 ratio was calculated for each of the 17 MSAs
30     using as  many sites as possible.  The median mean PM25:PM10_25 ratio for the 17 MSAs was 1.2.
31     For eight MSAs (Columbia, SC; Chicago, IL; Louisville, KY; Gary, IN; Tampa, FL;

       June 2003                                 3-21         DRAFT-DO NOT QUOTE OR CITE

-------
                                  a, Cleveland, OH      - 2001J

                       AIRS ID*  390360013 380360038  390Jb004S 39Q3S0080 3KJ3WQ85  380861001
                         Mem
                          G&s
                          SO
                        3,
                       a.
  X*
  a*»
  17 «
    186
    »14
    116
188
112
80
11$
184
             72
             108
ll:
I'MilM,
                             1 2 3 * 1
                                       b. Dallas, TX( 2001)

                         AIRSIW  481130020  481130)35  481130050  481130057
                            Ota
                             SO
                         "ra
                         D-
     112
      60
      5.4
          SS
          67
        14,5
        56
        64
         19,1
          SS
         10.6
                                1234  1Z34  1234  1234
                                             Quarter
Figure 3-9a,b.  Quarterly distribution of 24-h average PM10_2 5 concentrations for selected
               sites in the (a) Cleveland, OH; (b) Dallas, TX MSAs. Values for the lowest,
               lower quartile, median, upper quartile and highest concentrations are
               shown in the figures. The AIRS site ID number, annual concentration,
               number and standard deviation are shown above the figures for each site.
June 2003
              3-22
                      DRAFT-DO NOT QUOTE OR CITE

-------
                                          c. Los Angeles, CA f2001)

                             AIRSID#  060370002 060371002  080371103  080374002
                                Mean
                                 Oba
                                  so
                              m
                              a.
                                  20-
                                  10
24.1
56
11.7
15.3
56
8.7
        21.4
         57
         8.7
16.1
53
6.6

                                                                f

                                     1234   1234  123
                                                   Quarter
                                                               1234
      Figure 3-9c.  Quarterly distribution of 24-h average PM10_2 5 concentrations for selected
                    sites in the Los Angeles, CA MSAs. Values for the lowest, lower quartile,
                    median, upper quartile and highest concentrations are shown in the figures.
                    The AIRS site ID number, annual concentration, number and standard
                    deviation are shown above the figures for each site.

      Source: AIRS Database.
1      Steubenville, WV; Portland, OR; Milwaukee, WI) the mean PM2 5 concentration exceeded the

2      mean PM10_2 5 concentration.  For an additional eight MSAs, the PM2 5 and PM10_2 5 concentration

3      means were the same (within one SD). Salt Lake City was the only MSA for which the mean

4      PM10_2 5 concentration exceeded the mean PM2 5 concentration.

5           Information regarding the seasonal variability in PM10_2 5 concentrations in three MSAs

6      (Cleveland, OH; Dallas, TX; Los Angeles-Long Beach, CA) is summarized in Figures 3-9a

1      through 3-9c.  The figures show lowest, lower quartile, median, upper quartile, and highest

8      concentrations for each calendar quarter of 2000 and 2001 for the Cleveland, OH MSA (six
      June 2003
3-23
                    DRAFT-DO NOT QUOTE OR CITE

-------
 1      sites); and of 2001 for the Dallas, TX and Los Angeles, CA MS As (four sites each). Data for
 2      these and an additional 13 MS As, as well as additional descriptions of the data, are given in
 3      Appendix 3 A.  The seasonal pattern for PM10_2 5 median concentrations is different from that of
 4      PM2 5 (Figures 3-7a,b,c,d, and Appendix 3A).  Most seasonal median maxima in PM10_2 5 occur
 5      during the second or third calendar quarters, i.e., spring and summer months (45% and 36% of
 6      all sites, respectively) as  seen, for example, at most sites in Figures 3-9. Contributions from
 7      bioaerosols during the spring blooming season, which mostly affect PM10_2 5 concentrations,
 8      might be a significant contributing factor in the second quarter PM10_2 5 median maxima in some
 9      regions of the United States.  Lowest median concentrations occur mainly in the first or fourth
10      quarters 62% and 23% of all sites, respectively). Cleveland, OH, (Figure 3-9a) and Tampa, FL,
11      where lowest medians generally are observed in the third quarter are exceptions (Figure 3 A-29).
12      At no  site was the lowest median PM10_2 5 concentration observed in the second  quarter. By
13      comparison with PM10_2 5, seasonal PM25 median maxima mostly occurred in the first or third
14      quarters and PM2 5 median minima are mostly observed in the second and fourth quarters. With
15      few exceptions, collocated PM25 and PM10_25 seasonal medians at individual sites peak in
16      different quarters. Likewise, at a given site, the lowest median concentrations for PM25 and
17      PM10_2 5 rarely occur in the same quarter. In MS As with several PM10_2 5 sites, the  seasonal
18      patterns are typically reproduced at all sites within the MSA. In the Dallas, TX MSA, the
19      maximum and minimum  estimated PM10_25 concentrations both occur in the first quarter at all
20      four sites.
21           The ratio of PM2 5/PM10_25 seasonal median concentrations peak in the first quarter for
22      MSAs in the central and north-central United States and tend to peak in the fourth quarter for
23      western states.  The largest ratios of PM2 5/PM10_2 5  seasonal median concentrations are observed
24      in the central and north-central MSAs (Chicago, Cleveland, Detroit,  and St. Louis); whereas
25      smaller ratios are found in the western and southern United States.
26           As can be seen from Figure 3-9a and Figures 3 A-40 and 3 A-43, the maxima for PM10_2 5
27      concentrations exceeded  100 |ig/m3 in Cleveland, OH, and Salt Lake City, UT,  and 500 |ig/m3 in
28      Riverside, CA.  This latter value is related to a dust storm. In several urban areas (Cleveland,
29      OH; Detroit, MI; Chicago, IL; Dallas, TX; and Riverside, CA) maxima in PM10_2 5 concentrations
30      were larger than those for PM2 5.  However, there is no clear geographic pattern.
        June 2003                                3-24        DRAFT-DO NOT QUOTE OR CITE

-------
 1           The results described above should be viewed with caution because of inherently large
 2      errors in the technique used to derive them.
 3
 4      Frequency Distributions for PM25 Data
 5           Frequency distributions for PM2 5 concentrations obtained in Philadelphia from 1992
 6      through 1995 are shown in Figure 3-10 (data obtained by Bahadori et al., 2000b). Also shown,
 7      are concentrations predicted from the log-normal distribution, using geometric mean values and
 8      standard deviations derived from the data. In Philadelphia, the highest PM2 5 values were
 9      observed when winds were from the southwest during sunny but hazy high pressure conditions.
10      In contrast, the lowest values were found after significant rainstorms during all seasons of the
11      year. Mean ± SD day-to-day concentration differences in the data set are 6.8 ± 6.5 |ig/m3 for
12      PM2 5 and 8.6 ± 7.5 |ig/m3 for PM10. Maximum day-to-day concentration differences are
13      54.7 |ig/m3 for PM25 and 50.4 |ig/m3 for PM10.
14           Different patterns are observed in data collected elsewhere in the United States.  PM2 5
15      concentrations obtained in Phoenix, AZ, from 1995 through 1997 (Zweidinger et al., 1998) are
16      summarized in Figure 3-11; and frequency distributions of PM2 5 concentrations obtained in
17      Phoenix are shown in Figure 3-12. Mean ±SD day-to-day concentration differences in this data
18      set are 2.9 ±3.0 |ig/m3; the maximum day-to-day concentration difference was 23 |ig/m3. PM25
19      and PM10_2 5 data were obtained with dichotomous samplers at a number of sites in California on
20      a sampling schedule of every 6 days from 1989 through 1998.  Histograms showing the
21      frequency distribution of the entire set of PM25 and PM10_25 concentrations obtained by the
22      CARB network of dichotomous samplers from 1989 to 1998 are shown in Figures 3-13 and
23      3-14. Also shown are log-normal distributions generated by using geometric means and
24      standard deviations derived from the data as input.  Although the data for both size fractions
25      appear to be reasonably well simulated by the function, data obtained at individual locations may
26      not be.  Data showing the seasonal variability of PM2 5 obtained at Riverside-Rubidoux are
27      summarized in box plot form in Figure 3-15. The frequency distribution of PM25 concentrations
28      obtained at Riverside-Rubidoux from  1989 to 1994 is shown in Figure 3-16. It can be seen that
29      the data are not as well  fit by a log-normal distribution as are data shown in Figure 3-10, partly
30      as the result of a significant number of days when PM2 5 concentrations exceed 100 |ig/m3.
31

        June 2003                                3-25         DRAFT-DO NOT QUOTE OR CITE

-------
                           350
                                                                  PM25
                                                         geometric mean = 15.2 |jg/m3
                                                                ag=1.69
                                              \
                                                                          t  •
                                     10     20    30    40     50    60     70    80
                                               Concentration (|jg/m3)
       Figure 3-10.  Frequency distribution of 24-h average PM25 concentrations measured at the
                     Presbyterian home (PBY) monitoring site in southwestern Philadelphia from
                     1992 to 1995. Log-normal distribution fit to the data shown as solid line.
       Source: Bahadori et al. (2000b).
 1          An examination of the data from the four MAACS cities, Phoenix, AZ, and Riverside, CA,
 2     indicates that substantial differences exist in aerosol properties between the eastern cities
 3     (MAACS) and the western cities (Phoenix, AZ; Riverside, CA).  Fine-mode particles account for
 4     most of the PM10 mass observed in the MAACS cities and appear to drive the daily and seasonal
 5     variability in PM10 concentrations in the East; whereas coarse-mode particles represent a larger
 6     fraction of PM10 mass in Phoenix and Riverside and drive the seasonal variability in PM10 seen in
 7     the West. The average ratio of PM25 to PM10 concentrations is much larger in the MAACS cities
 8     of Philadelphia, PA (0.72); Washington, DC (0.74); and Nashville, TN (0.63) than in either
 9     Phoenix, AZ (0.34)  or Riverside, CA (0.49). Differences  between median and maximum
10     concentrations in any size fraction are much larger at the Riverside site than at either the
11     MAACS or Phoenix sites. Many of these differences could reflect the more sporadic nature of
       June 2003
3-26
DRAFT-DO NOT QUOTE OR CITE

-------
              .g
              "-i—'
              TO
              -i—>

              I
              o
              O
                 40
                 30-
20-
                 10-
                             Phoenix, AZ
                                PM25
                          -..   (n = 876)
                   T
                                                T
                       Mar - May   June-Aug   Sep - Nov    Dec-Feb
Figure 3-11. Concentrations of 24-h average PM25 measured at the EPA site in Phoenix,
            AZ from 1995 to 1997. The data show the lowest, lowest tenth percentile,
            lowest quartile, median (black circles), highest quartile, highest tenth
            percentile, and highest PM2 5 values.

Source: Zweidinger et al. (1998).
           200
           150-
        Q.
        E
        5100
        M—
        O
        o
        z  50H
                                    PM2.5
                          geometric mean = 10.5 |jg/m3
                                   a = 1.70
               0      5     10     15     20     25    30    35    40
                               Concentration
Figure 3-12. Frequency distribution of 24-h average PM2 5 concentrations measured
            at the EPA site in Phoenix, AZ from 1995 to 1997.

Source: Zweidinger etal. (1998).
June 2003
                     3-27
DRAFT-DO NOT QUOTE OR CITE

-------
           3000
           2500-
                                                   PM
                                                     '2.5
                                         geometric mean = 12.8 |jg/m3
                                                  on = 2.29
                0  10  20  30  40  50  60  70  80  90  100  110  120 130 140 150
                                     Concentration (|jg/m3)
Figure 3-13.  Frequency distribution of 24-h average PM2 5 measurements obtained from
             all California Air Resources Board dichotomous sampler sites from 1989 to
             1998.
           3000
           2500-
          PM-iQ-25
geometric mean = 15.7 |jg/m3
        on = 2.26
                0  10  20  30  40  50   60  70  80  90  100  110 120 130 140 150
                                    Concentration (|jg/m3)
Figure 3-14.  Frequency distribution of 24-h average PM10_25 concentrations obtained from
             all California Air Resource Board Dichotomous sampler sites from 1989 to
             1998.
June 2003
 3-28
DRAFT-DO NOT QUOTE OR CITE

-------
iuu -

140-
CO
-i 120-
D)
o 100-
'-4— •
1 8°-
O
<5 60-
m
CN
2 40-
CL

20 -

Fine
(n = 382)























-p




-^ n
( A A fl
n rn

.
LJ
U TJ
^ T I V
l l l
Jan - Mar Apr-Jun Jul - Sept Oct - Dec
1st Qtr 2nd Qtr 3rd Qtr 4th Qtr
Figure 3-15.  Concentrations of 24-h average PM2 5 measured at the Riverside-Rubidoux
             site from 1989 to 1998. The data show the lowest, lowest tenth percentile,
             lowest quartile, median (black squares), highest quartile, highest tenth
             percentile, and highest PM2 5 values.
           100-
                                                       PM
                                                         2.5
                                             geometric mean = 26.6 |jg/m3
                                                     o =2.10
                    20   40
60    80   100   120   140   160   180   200
     Concentration (|jg/m3)
Figure 3-16.  Frequency distribution of 24-h average PM25 concentrations measured
             at the Riverside-Rubidoux site from 1989 to 1994.
June 2003
        3-29
DRAFT-DO NOT QUOTE OR CITE

-------
 1      dust suspension at Riverside. In addition, the seasonal variability of PM2 5 concentrations
 2      observed in Phoenix and Riverside appears to be different from that observed in the MAACS
 3      cities. These considerations demonstrate the hazards in extrapolating conclusions about the
 4      nature of variability in aerosol characteristics inferred at one location  to another.
 5
 6      3.2.2   Diurnal (Circadian) Variability in PM Concentrations
 7           The variability of PM concentrations on time scales shorter than a day can, in principle, be
 8      characterized by measurements made by continuous samplers (e.g, TEOMs and P-gauge
 9      monitors that are currently used to provide Air Quality Index [AQI] information to the public).
10      A description of these methods was provided in Section 2.2.9.  However, as shown in Chapter 2,
11      continuous methods are subject to artifacts in large part because of the heating of their inlets to
12      remove water, which results in the loss of components such as ammonium nitrate and
13      semivolatile organic compounds (Sections 2.2.2.1 and 2.2.3 for further details concerning the
14      chemistry of volatilizable components).  Consequently, caution should be used in interpreting
15      results obtained by these techniques.  It should be remembered that the Federal Reference
16      Methods (FRMs) are also subject to artifacts; therefore, caution should also be exercised in
17      interpreting results obtained by them.
18           The composite diurnal variation of PM25 concentrations obtained throughout the
19      continental United States by 31 TEOM and P-gauge monitors reporting to AIRS in 1999 is
20      shown in Figure 3-17. As can be seen, there is a distinct pattern with  maxima occurring during
21      the morning and evening.  Notable exceptions to this pattern occur in  California where broad
22      nighttime maxima and daytime minima occur which may be related to the use of p-gauges with
23      unheated inlets there. It should be noted in examining the diurnal variations shown in
24      Figure 3-17 that there is substantial day-to-day variability in the diurnal profile of PM2 5
25      measured at the same location that is  smoothed out after a suitably long averaging period is
26      chosen.  The large ratio of the interquartile range to the median values supports the view that
27      there is substantial variability in the diurnal profiles.
28           The diurnal variability of PM components is determined by interactions between variations
29      in emissions, the rates of photochemical transformations, and the vertical extent and intensity of
30      turbulent mixing near the surface. Wilson and Stockburger (1990) characterized the diurnal
31      variability of sulfate and lead in Philadelphia. At that time, Pb was emitted mainly by motor

        June 2003                                 3-30        DRAFT-DO NOT QUOTE OR CITE

-------
             30-
          o
         '-4—*
          CD

          §
          C
          O
         O
          IT)
          CN
      15-
              0-
                                       6685 6667
                                               6576 6571 6605 6640 6649
                  \   \    \   \   \   \   \   \   \   \   \    \   \   \   \   \   \   \   \   \   \   \    \   \
                  0  1   2   3   45   67   8  9  10  11  12  13 14 15 16 17 18 19  20  21  22 23
                                                    Hour
       Figure 3-17.  Intraday variability of hourly average PM2 5 concentrations across the United
                     States. Interquartile ranges, median and mean (+) values are shown. Values
                     above the box plots refer to the number of observations during 1999.
       Source: Fitz-Simons et al. (2000).
 1
 2
 3
 4
 5
 6
 1
 8
 9
10
11
12
13
vehicles. Pollutants emitted mainly by motor vehicles, such as carbon monoxide, show two
distinct peaks occurring during the morning and evening rush hours (see Chapter 3, U.S.
Environmental Protection Agency, 2000b).  Pollutants, such as sulfate, which are transported
long distances in the free troposphere (i.e., above the planetary boundary layer), tend to be mixed
downward and have their highest concentrations during the afternoon when the intensity and
vertical extent of turbulent mixing (and chemical oxidation) are greatest.  Secondary aerosol
components (such as secondary organic compounds) that are produced by photochemical
reactions may have a daily maximum in the afternoon, similar to ozone. PM produced by
residential heating (e.g., from wood burning), on the other hand, reach maximum levels during
the night when mixing heights are much lower than during the day.
     Although the interquartile ranges for hour-to-hour changes in PM2 5 concentrations shown
in Figure 3-17 encompass several |ig/m3, extreme values for the hour-to-hour variations can be
much larger (Fitz-Simons et al., 2000). The 98th percentile values for positive and negative
       June 2003
                                         3-31
DRAFT-DO NOT QUOTE OR CITE

-------
 1      excursions in concentration are all less than 20 |ig/m3. Maximum positive excursions were much
 2      larger, ranging from 27 |ig/m3 in the Northeast up to 220 |ig/m3 in the Southwest and with
 3      maximum excursions in other regions all less than 125 |ig/m3. It should be borne in mind that
 4      the hour-to-hour changes that are reported reflect the effects of a number of processes occurring
 5      during passage through the sampler inlets and on the TEOM measurement elements. These
 6      factors add uncertainty to the interpretation of the hour-to-hour changes that are observed, as
 7      discussed in Chapter 2. However, because of the tendency of these monitoring instruments to
 8      lose material by evaporation, the concentrations reported during excursions probably represent
 9      lower limits to the true values that were present.
10
11      3.2.3   Relations Among Particulate Matter in Different Size Fractions
12      Relations Among PM2^5, PM10_2^5, and PM10
13           Data obtained in 1999 by collocated PM2 5 and PM10 FRM monitors have been used to
14      calculate the ratio of PM2 5 to PM10 concentrations and correlations among PM25, PM10_25, and
15      PM10 concentrations. Results are shown in Table 3-1 for each of the seven aerosol characteristic
16      regions identified in Chapter 6 of the 1996 PM AQCD. As can be seen from the table, the ratio
17      of PM2 5 to PM10 concentrations tends to be higher in the eastern United States than in the
18      western United States.  This general pattern and the values are consistent with that found for the
19      studies included in Appendix 6A of 1996 PM AQCD. In that compilation based on the results of
20      studies using dichotomous samplers, the mean ratio of PM25 to PM10 was 0.75 in the East,
21      0.52 in the central United States, and 0.53 in the western United States. Although a large
22      number of paired entries have been included in Table 3-1, seasonal variations and annual
23      averages in a number of regions could not be determined from the data set because of data
24      sparseness mainly during the early part of 1999.  It also can be seen in Table 3-1 that the ratio of
25      PM2 5 to PM10  was greater than one for a few hundred measurements. There are a number of
26      reasons for these results, as mentioned in Section 3.2.1 in the discussion on PM10_25
27      concentrations.
28
29      Ultrafine Particle Concentrations
30           Data for characterizing the concentrations of ultrafme particles (< 0.10  jim Da) and the
31      relations between ultrafme particles and larger particles are sparse. Although ultrafme particles

        June  2003                                 3-32        DRAFT-DO NOT QUOTE OR CITE

-------
           TABLE 3-1. DISTRIBUTION OF RATIOS OF PM?. TO PM1(1 AND CORRELATIONS BETWEEN PM?. AND PM
                                                                         10
g PM2 5 AND PM10 2 5, AND PM
o
10 25 AND PM10 FOUND A
CHARACTERISTIC
1 COLLOC,
(EPA/HEI)
\IED MONITORING SITES IN SEVEN AEROSOL
REGIONS IN 1999
Percentiles
Region
Northeast
Southeast
Industrial Midwest
Upper Midwest
Southwest
Northwest
w Southern California

Mean
0.70
0.70
0.70
0.53
0.38
0.50
0.47
Total:
Sites
45
76
92
39
23
73
36
384
Values
1433
2823
4827
1446
701
3300
1813
16,343
95
0.97
1.27
1.09
0.92
0.51
0.67
0.70

90
0.95
1.06
0.88
0.84
0.51
0.65
0.57

75
0.77
0.74
0.78
0.62
0.47
0.56
0.55

50
0.67
0.63
0.68
0.49
0.40
0.49
0.48

25
0.60
0.54
0.59
0.44
0.31
0.44
0.44

10
0.51
0.46
0.51
0.34
0.23
0.39
0.31

5
0.48
0.43
0.47
0.24
0.23
0.36
0.24

PM25:PM10
0.72a
0.69a
0.71a
0.35a
0.63a
0.69a
0.70a

Correlations
PM25:PM1025
0.02
-0.04a
0.17a
-0.02
0.49a
0.07a
0.19a


PM10.2.5:PM10
0.71a
0.69a
0.81a
0.93a
0.99a
0.77a
0.83a

H

6
o


o
H

O

O
H
W

O


O
HH
H
W
        "Results considered to be significantly different from zero at the a = 0.01 level.


        Source: Aerometric Information Retrieval System (AIRS; U.S. Environmental Protection Agency, 2002b).

-------
 1     dominate particle number concentrations, they make very minor contributions to PM2 5 mass.
 2     For example, Cass et al. (2000) found that particles between 0.056 and 0.1 jim Da contributed
 3     only 0.55 -1.16 |ig/m3 at several sites in southern California. Perhaps the most extensive data
 4     set for ultrafine particle properties is that described by Woo et al. (2001) for a site located 10 km
 5     to the northwest of downtown Atlanta, GA.  Size distributions from 3 to 2000 nm were measured
 6     every 12 minutes for 24 months beginning in August 1998. Approximately 89% of the total
 7     number of particles were found to be smaller than 100 nm; whereas 26% were found to be
 8     smaller than 10 nm.  Concentrations tend to be lower during the summer than during the winter.
 9     No correlation was found between number concentration and either volume or surface area for
10     particle sizes up to 2 jim.  Because the total number of particles is concentrated in the smallest
11     size ranges, these results also indicate that fine particle mass does not correlate with the number
12     of ultrafine particles. The high time resolution of the measurements allows some inferences to
13     be made about the possible sources of the ultrafine particles.  The number of particles larger than
14     10 nm tends to peak during the morning rush hour (around 8 am) and then to decrease through
15     the day and to increase again after 6 pm, consistent with a traffic-related source. Particles
16     smaller than 10 nm tend to peak during the mid-afternoon, consistent with nucleation involving
17     products of active photochemistry (McMurry et al., 2000). More direct relations between
18     particle mass observed in different size ranges can be obtained using multi-stage impactors.
19     Keywood et al. (1999) found a correlation between PM25 and PM015 of about 0.7; whereas they
20     found correlations of about 0.96 between PMX and PM2 5 and between PM2 5 and PM10 based on
21     samples collected by MOUDIs (Multiple Orifice Uniform Deposit Impactors) in six Australian
22     cities.
23
24     3.2.4    Relations Between Mass and Chemical Component Concentrations
25          Time series of elemental composition data for PM2 5 based on X-ray fluorescence (XRF)
26     analyses have been obtained at a number of locations across the United States.  Time series of
27     components of the organic carbon fraction of the aerosol have not yet been obtained. The results
28     of XRF analyses for the composition of the inorganic fraction of PM2 5 and PM10_2 5 are presented
29     in Table 3-2 for Philadelphia, PA, and in Table 3-3 for Phoenix, AZ.  Frequency distribution for
30     PM2 5 concentration  data collected at these sites were shown in Figures 3-10 and 3-11. All XRF
31     analyses were performed at the same X-ray spectrometry facility operated by the U.S.

       June 2003                                3-34        DRAFT-DO NOT QUOTE OR CITE

-------

TABLE 3-2.  CONCENTRATIONS (in ng/m3) OF PM2 5, PM10 2 5, AND SELECTED ELEMENTS (ng/m3) IN THE
to
o
o
LtJ










OJ
UJ



o
§
H
6
O
0
H
O
O
H
W
O
ELEMENTS AND PM2 5 MASS IN PHILADELPHIA, PA*
n = 1105 Cone (ng/m3) ± SD (unc) r
PM25
Al
Si
P
S
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Co
Ni
Cu

Zn
As
Se
Br
Pb

1 17 ± 0.9 (0.8) x 103
4.0 ±56 (31)
116 ±107 (21)
8.6 ±14 (10)
2100 ± 1610(143)
5.1 ±35 (3.4)
60.4 ± 45 (4.7)
47 ± 33 (4.2)
4.9 ±5.2 (4.1)
8.8 ±8.7 (1.8)
0.7 ±1.1 (0.7)
3.1 ±2.2 (0.8)
109 ±71 (10.5)
0.1 ±1.8 (1.4)
7.3 ±8.4 (1.4)
4.8 ±4.9 (1.1)

36.9 ±44 (3.7)
0.6 ±1.4 (1.2)
1.5 ±1.3 (0.6)
5.0 ±11.7 (0.9)
17.6 ± 22 (2.5)

1.00
0.10
0.51
0.31
0.92
-0.01
0.50
0.39
0.44
0.37
0.15
0.39
0.50
0.04
0.22
0.25

0.21
0.18
0.63
0.11
0.19

n = 20 Cone (ng/m3) ± SD (unc) r
PM252 29.8 ± 14.7 (1. 1) x 103
Al 109 ±61 (21)
Si 191 ± 134 (26)
P 15 ± 4.3 (2.7)
S 3 190 ±1920 (207)
Cl 23 ± 28 (5.5)
K 68 ± 21 (6.4)
Ca 63 ± 33 (9.0)
Ti 8.7 ± 4.7 (9.0)
V 9.7 ±7. 1(2.9)
Cr 1.4 ±1.2 (2.9)
Mn 3.2 ±1.5 (1.6)
Fe 134 ±49 (0.5)
Co 0.8 ±0.7 (8.5)
Ni 8.5 ± 5.6 (0.3)
Cu 7.7 ± 3.8 (0.7)

Zn 56 ± 37 (4.8)
As 0.4 ±1.0 (1.0)
Se 1.3 ±0.8 (0.4)
Br 14 ±12 (1.3)
Pb 28 ± 24 (2.4)

'Data obtained at the Presbyterian home (PBY) site in Philadelphia from April 1992 to April
^ 2Data obtained at the Castor Avenue
O
HH
H
W

'Note:


Laboratory

Values in parentheses refer to analytical


, North Central Philadelphia, from July 25 to

1.00
0.15
0.22
0.72
0.91
0.19
0.31
-0.02
0.47
0.38
0.09
0.43
0.48
0.58
0.61
0.22

0.22
-0.02
0.65
0.21
0.26

1995 with
August 14

n = 20 Cone (ng/m3) ± SD (unc)
PM10.252 8.4 ± 2.9 (0.4) x 103
Al
Si
P
S
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Co
Ni
Cu

Zn
As
Se
Br
Pb

Harvard impactors.
, 1994 with a modified

325 ± 241 (99)
933 ±652 (231)
28 ±9.4 (7.1)
38 ±45 (71)
47 ± 48 (5.8)
100 ± 66 (10)
421 ±192 (31)
30 ±17 (5.6)
3.2 ±2.2 (1.5)
1.0 ±5.0 (0.9)
6.3 ±4. 1(0.6)
352 ± 156 (24)
-0.2 ±0.5 (0.3)
2.0 ± 1.4(0.3)
14 ± 12(1.1)

52 ± 43 (4.7)
0± 0.5 (0.5)
-0.1 ±0.2 (0.2)
3.0 ±2.5 (0.5)
13 ± 11 (1.3)


dichotomous sampler.

r
1.00
0.89
0.90
0.78
-0.15
-0.11
0.81
0.81
0.90
0.66
0.43
0.90
0.90
-0.10
0.08
-0.05

-0.03
0.07
-0.24
-0.10
0.10




uncertainty (unc) in X-ray fluorescence determinations.






-------
          TABLE 3-3. CONCENTRATIONS (in ng/m3) OF PM25, PM1025 AND SELECTED
             ELEMENTS IN THE PM25 AND PM1025 SIZE RANGE WITH STANDARD
            DEVIATIONS (SD) AND CORRELATIONS (r) BETWEEN ELEMENTS AND
                           PM2 5 AND PM10 2 5 MASS IN PHOENIX, AZ*
n = 164
PM25
Al
Si
P
S
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Co
Ni
Cu
Zn
As
Se
Br
Pb
Cone (ng/m3) ± SD (unc)
11.2 ±0.6 (0.6) x 103
125 ± 77 (30)
330 ±191 (48)
11 ±7.8 (5.7)
487 ± 254 (40)
19 ±44 (3.0)
110 ±63 (9.2)
129 ±72 (11)
11 ±7.1 (2.7)
0.7 ± 2.0 (2.2)
0.6 ±0.9 (0.7)
5.7 ±4.3 (0.7)
177 ±113 (16)
-0.4 ± 1.0(1.0)
0.6 ±0.9 (0.5)
5.2 ±6.1 (1.5)
17 ±14.7 (1.8)
1.9 ±3.2 (0.6)
0.4 ±0.8 (0.4)
3.8 ±2.0 (0.6)
6.6 ±6.6 (1.0)
r
1.00
0.23
0.35
0.52
0.16
0.13
0.67
0.51
0.44
-0.28
0.41
0.64
0.80
-0.01
0.38
0.69
0.64
0.50
0.40
0.57
0.69
n = 164
PM10.2.5
Al
Si
P
S
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Co
Ni
Cu
Zn
As
Se
Br
Pb
Cone (ng/m3) ± SD (unc)
27.6 ± 14.8 x 103
1879 ± 979 (547)
535 ± 2825 (1347)
37 ± 20 (17)
131 ±47 (26)
208 ± 204 (24)
561 ±298 (62)
1407 ± 755 (124)
130 ±71 (20)
2.0 ±2.0 (1.5)
2.6 ±1.7 (0.7)
29 ±16 (3.0)
1211 ±674 (133)
1.2 ±2.2 (1.9)
1.8 ±1.4 (0.7)
10.3 ±9.0 (1.5)
25 ±16 (3.2)
0.6 ±0.8 (0.6)
-0.02 ±0.3 (0.3)
0.8 ±0.6 (0.4)
4.6 ±3. 8 (1.1)
r
1.00
0.92
0.92
0.58
0.77
0.28
0.92
0.90
0.90
0.51
0.76
0.91
0.90
0.38
0.70
0.58
0.64
0.41
0.21
0.48
0.59
       * Values in parenthesis refer to analytical uncertainty (unc) in X-ray fluorescence determinations.

       Source: Calculations based on data obtained by Zweidinger et al. (1998).
1     Environmental Protection Agency in Research Triangle Park, NC. Data shown in the first

2     column of Table 3-2 are based on analyses of filters collected over three years (April 1992 to

3     April 1995, labeled a) at the PBY site in southwestern Philadelphia.  These data and data for
      June 2003
3-36
DRAFT-DO NOT QUOTE OR CITE

-------
 1      PM10 were collected using Harvard impactors. Data for PM2 5 and PM 10_2 5 shown in the second
 2      and third columns were collected at the Castor Avenue Laboratory operated by the City of
 3      Philadelphia from July 25 to August 14, 1994, using a modified dichotomous sampler (VAPS).
 4      The samples at the Phoenix site were collected in 1996 and 1997 using the same type of
 5      dichotomous sampler used in the shorter term study in Philadelphia. These data are shown to
 6      give an idea of the range of concentrations found in studies conducted more recently than those
 7      shown in Appendix 6A of the 1996 PM AQCD. The speciation network will at least provide
 8      more thorough coverage of the composition of particles in the PM25 size range across the
 9      United  States. Results from the pilot study for the speciation network are given in Appendix 3B.
10           As can be seen from inspection of Tables 3-2 and 3-3, the analytical uncertainty (given in
11      parentheses next to concentrations) as a fraction of the absolute concentration is highly variable.
12      It exceeds the concentration for a number of trace metals whose absolute concentrations are low;
13      whereas it is very small for abundant elements such as sulfur.
14           Sulfur is the major element analyzed in the PM2 5 size fraction in the two Philadelphia
15      studies  and is highly correlated with PM25; however its abundance is roughly two orders of
16      magnitude lower in the PM10_2 5 size range and is negatively correlated with PM10_25.
17      Concentrations of the crustal elements Al, Si, K, Ca, and Fe are much higher in the PM10_25 size
18      range than in the PM25 size range and are well correlated with PM10_25. A number of trace
19      elements (e.g., Cr, Co, Ni, Cu, Zn, As, Se and Pb) are detectable in the two PM25 data sets, and
20      the concentrations of many of these elements are much greater than the uncertainty in their
21      determination. Except for Co, As, and Se which are not detected in the PM10_25 samples, the
22      concentrations of many elements (Cr, Zn, and Pb) are comparable in the PM25 and PM10_25 size
23      ranges.  The concentration of Cu is significantly higher in the PM10_25 size  range, whereas the
24      concentration of Ni is smaller in the PM10_2 5 size range than in the PM2 5 size range.
25           There are a number of distinct differences between the PM2 5 sets for Philadelphia and
26      Phoenix. For instance, sulfate and associated cations and water that would be expected to
27      correspond to the measurement of S appear to constitute a major fraction of the composition of
28      the PM in the Philadelphia data set; whereas they appear to constitute a much  smaller fraction of
29      the PM in the Phoenix data set. The highest PM2 5 values were observed in Philadelphia during
30      episodes driven by high sulfate abundances; whereas those in Phoenix were driven by raised soil
31      dust. The concentration of S in Phoenix is much lower in the Phoenix PM2 5 data set than in

        June 2003                                 3-37        DRAFT-DO NOT QUOTE OR CITE

-------
 1      either Philadelphia PM2 5 data set, even though it represents the most abundant element and it is
 2      only weakly correlated with PM2 5. This is in marked contrast to the data shown for Philadelphia
 3      and also to data found at other eastern sites.  It is not clear what the reasons are for this finding.
 4      As in Philadelphia, the concentration of S in Phoenix is higher in the PM2 5 size range than in the
 5      PM10_2 5 size range. Trace metals (e.g., Cr, Co, Ni, Cu, Zn, As, and Pb) are not well correlated
 6      (0.04 < r < 0.25) with PM2 5 in the Philadelphia data set; whereas they are more variably
 7      correlated (0.01 < r < 0.69) with PM2 5 in the Phoenix data set. The uncertainty in the
 8      concentration measurement most probably plays a role in determining a species' correlation with
 9      PM2 5 especially when the analytical uncertainty is high relative to concentration as it is for a
10      number of elements in the data shown in Tables 3-2 and 3-3.  Concentrations of Al, Si, K, Ca,
11      and Fe are again much higher in the PM10_2 5 size range than in the PM2 5  size range and are
12      strongly correlated with PM10_2 5 in both data sets.
13           There are also similarities in the PM2 5 data sets for Philadelphia and Phoenix.  Crustal
14      elements are not as well correlated with PM25 as they are with PM10_25 in both data sets. The
15      concentrations of trace metals (Cr,  Ni, Cu, and Zn) in PM25 are similar in Philadelphia and
16      Phoenix. It can also be  seen that their concentrations are of the same order of magnitude in both
17      PM2 5 and PM10_2 5. Concentrations of Cu are noticeably higher in PM10_2 5 than in PM2 5 in both
18      Philadelphia and  Phoenix.  These results are consistent with those of many monitoring studies
19      shown in Appendix 6A of the 1996 PM AQCD, which also show that concentrations of these
20      metals are of the  same order of magnitude in both size fractions and that concentrations of Cu
21      tend to be higher in PM10_2 5 than in PM25.
22           One study suggests that the partitioning of trace metals between the fine and coarse
23      fractions varies with PM concentration.  Salma et al. (2002) determined the size distribution of a
24      number of trace elements at four sites characterizing environments ranging from the urban
25      background to an urban traffic tunnel in Budapest, Hungary.  S, K, V, Ni, Cu, Zn, As, and Pb
26      were found  mainly in the fine fraction at the urban background site; but their mass median
27      aerodynamic diameters increased with increasing PM concentrations until they were all found
28      mainly in the coarse fraction in the traffic tunnel. They also found that Na, Mg, Al, Si, P, Ca, Ti,
29      Fe, Ga, Sr, Zr, Mo, and Ba were concentrated mainly in the coarse fraction at all four sites and
30      that their mass median aerodynamic diameters increased with increasing PM concentrations.
        June 2003                                 3-38        DRAFT-DO NOT QUOTE OR CITE

-------
 1           The mean concentration of Pb observed in the methods evaluation study for the speciation
 2      network was only about 5 ng/m3 in Philadelphia during the first half of 2000 (Appendix 3B);
 3      whereas its concentration was about three times higher during the studies conducted during the
 4      early 1990s (Table 3-3). In a study conducted in the greater Philadelphia area during the
 5      summer of 1982, Dzubay et al. (1988) found concentrations  of Pb of about 250 ng/m3, or about
 6      fifty times higher than observed in 2000. The mean Pb concentration was about 3 ng/m3 at the
 7      Phoenix site included as part of the same methods evaluation study for the speciation network;
 8      however, the mean Pb concentration was 39 ng/m3 during an earlier study conducted during
 9      1989 and  1990 in Phoenix (Chow et al., 1991). These changes in Pb concentrations are
10      consistent with those in many  other urban areas for which monitoring studies have been
11      conducted during the late 1970s and 1980s (cf, Appendix 6A of the 1996 PM AQCD) and for
12      which there are data given in Appendix 3B. It should be remembered that the older studies were
13      conducted while Pb was still used as a gasoline additive. The ratio of Pb in PM2 5 to Pb in
14      PM10_2 5 was also much higher in the older studies than in the more recent ones, reflecting the
15      importance of combustion as its source. Smaller decreases are apparent in the concentrations of
16      other trace metals such as Cu,  Ni, and Zn between studies conducted in the early 1980s and in
17      the methods evaluation study for the speciation network conducted in 2000.
18           Some indication of the sources of metals such as Pb, Cu, Cd, and Zn in current, ambient
19      PM2 5 and PM10_2 5 samples can be obtained by examining their sources in urban runoff.  The
20      sources of these elements in urban runoff were found to be the weathering of building surfaces,
21      motor vehicle brake and tire wear, engine oil and lubricant leakage and combustion, and wet and
22      dry atmospheric deposition (Davis et al., 2001). Once deposited on the ground, these elements
23      can be resuspended with other material as PM2 5 and PM10_2 5 although research is needed into the
24      mechanisms of how this is accomplished. Wind-abrasion on building siding and roofs (coatings
25      such as Pb paint and building material such as brick, metal, and wood siding); brake wear (brake
26      pads contain significant quantities of Cu and Zn); tire wear (Zn is used as a filler in tire
27      production); and burning engine oil could all produce particles containing these metals,
28      especially Zn.
29           Data for the chemical composition of ambient ultrafine particles  are sparse.  In a study
30      conducted at several urban sites in Southern California, Cass et al. (2000) found that the
31      composition of ultrafine particles ranged from 32 to 67% organic compounds, 3.5 to 17.5%

        June 2003                                 3-39       DRAFT-DO NOT QUOTE OR CITE

-------
 1     elemental carbon, 1 to 18% sulfate, 0 to 19% nitrate, 0 to 9% ammonium, 1 to 26% metal oxides,
 2     0 to 2% sodium, and 0 to 2% chloride. Thus carbon, in various forms, was found to be the major
 3     contributor to the mass of ultrafme particles. However, ammonium was found to contribute 33%
 4     of the mass of ultrafme particles at one site in Riverside.  Iron was the most abundant metal
 5     found in the ultrafme particles.  Chung et al. (2001) found that carbon was the major component
 6     of the mass of ultrafme particles in a study conducted during January of 1999 in Bakersfield,
 7     CA. However, in the study of Chung et al., the contribution of carbonaceous species (OC and
 8     EC; typically 20 to 30%) was much lower than that found in the cities in Southern California.
 9     They found that calcium was the dominant cation, accounting for about 20% of the mass of
10     ultrafme particles in their samples. Sizable contributions from silicon (0 to 4%) and aluminum
11     (6 to 14%) were also found.  Further studies, including scanning electron microscopy, may be
12     needed to quantify the role of coarse particle bounce from the upper stages of their MOUDI
13     impactor.
14          Gone et al. (2000) measured the size distribution of trace elements from 0.056 jim to
15     1.8 |im Da in Pasadena, CA,  and in the Great Smoky Mountains National Park, TN. They found
16     that elements identified as being of anthropogenic origin had mass median diameters below
17     1 jim PM; whereas elements of crustal origin generally had a mass median diameter greater than
18     1 |im. Concentrations of trace metals were much higher in the accumulation mode than in the
19     ultrafme mode in both study areas. In PMX, 76% of Cr, 95% of Fe, 94% of Zn, 89% of As, and
20     79% of Cd at the Tennessee  site were found in the accumulation mode; and 70% of Fe, 85% of
21     Zn, 92% of As, and 84% of Cd were found in the accumulation mode in Pasadena.  Fe was the
22     most abundant metal found in the ultrafme particles. The abundance of crustal elements, such as
23     Al, declined rapidly with decreasing particle size at both locations; and Al in PMl probably
24     represented the lower tail of the coarse PM mode. However, on two days at Pasadena there were
25     increases in the concentration of Al in ultrafme particles that were associated with increases in
26     Sc and Sm. The latter two elements originate exclusively from crustal material (Gone et al.,
27     2000).
28
29
       June 2003                                3-40        DRAFT-DO NOT QUOTE OR CITE

-------
 1      3.2.5   Spatial Variability in Particulate Matter and its Components
 2      PM25
 3           Aspects of the spatial variability of PM2 5 concentrations on the urban scale are examined
 4      in this section.  Intersite correlation coefficients for PM2 5 can be calculated based on the results
 5      of FRM monitors placed at multiple sites within Metropolitan Statistical Areas (MSAs) across
 6      the United States.  Pearson correlation coefficients (r) calculated for pairs of monitoring sites in
 7      the Philadelphia, PA; Cleveland, OH; Dallas, TX; and Los Angeles, CA MSAs are shown in
 8      Table 3-4. The 90th percentile value, P90, of the absolute differences (in |ig/m3) between the two
 9      sites is shown in parentheses below r along with the coefficient of divergence (COD), and the
10      number of observations used in the calculation of r, P90 and COD is given on the third line.  The
11      COD was used by Wongphatarakul et al. (1998) as a measure of the degree of similarity between
12      aerosol data sets1.  The annual mean concentrations, the number of observations used to calculate
13      the annual average, and the standard deviation are shown directly beneath the correlation tables
14      for each site. These analyses  and those for another 23 MSAs are given along with maps in
15      Appendix 3 A. As the concentrations of PM25 at two sampling sites become more alike, the
16      COD approaches zero; as the  concentrations diverge, the COD approaches one.
17           The four MSAs shown in Table 3-4 were chosen to illustrate different patterns of spatial
18      variability across the United States.  In addition, air-pollution health-outcome studies have been
19      performed in a few of these MSAs. It can be seen from inspection of Table 3-4 that correlation
20      coefficients vary over a wide  range in the MSAs shown. Correlations between sites in the
21      Philadelphia, PA; Cleveland OH; and Dallas, TX MSAs are all high and span a relatively narrow
22      range (0.82 to 0.97).  However, correlations between sites in the Los Angeles-Long Beach MSA
23      are lower than in the three other MSAs and span a wider range of values (0.60 to 0.95). If the
24      monitoring site in Lancaster, CA, were included, correlations would be even lower.  This site
25      was omitted because it did not meet completeness criteria for 2001.  The extension of these
              'The COD for this purpose is defined as follows:
                                                                                           (3-D
        where x^ and % represent the 24-h average PM2 5 concentration for day i at site j and site k and p is the number of
        observations.

        June 2003                                 3-41        DRAFT-DO NOT QUOTE OR CITE

-------
        TABLE 3-4a-d. MEASURES OF THE SPATIAL VARIABILITY OF
       PM2 5 CONCENTRATIONS WITHIN SELECTED METROPOLITAN
                         STATISTICAL AREAS
(a) Philadelphia,
PA
Sitel.D. # 340071007 340155001 420170012
340071007

340155001

420170012

420450002
420910013
421010004
421010136
Mean
Obs
SD
1 0.91 0.93
(6.3,0.14) (5.2,0.15)
170 167
1 0.84
(7.5,0.19)
176
1



Key
Airs Site I.D. #
Pearson r
(90th %-tile difference in concentration,
coefficient of divergence)
number of observations

14.78 14.59 14.11
197 208 217
9.20 8.65 8.47

420450002
0.87
(6.9,0.19)
183
0.88
(7.4,0.18)
194
0.85
(7.5,0.16)
199
1



15.92
230
8.65

420910013
0.88
(5.0,0.16)
176
0.83
(7.1,0.18)
184
0.88
(4.7,0.13)
200
0.87
(6.1,0.15)
208
1
14.20
221
8.93

421010004
0.94
(4.6,0.15)
163
0.89
(7.4,0.17)
169
0.94
(4.9,0.11)
177
0.94
(5.1,0.11)
187
0.90
(4.6,0.11)
181
1
15.72
610
9.18

421010136
0.93
(5.1,0.14)
166
0.85
(6.9,0.18)
173
0.89
(5.3,0.13)
180
0.88
(4.0,0.12)
193
0.87
(4.5,0.11)
185
0.96
(3.3, 0.08)
550
1
15.81
616
9.26













(b) Cleveland, OH
Sitel.D. # 390350013 390350038 390350060
390350013

390350038

390350060

390350065

390350066
390351002
390851001
390932003
Mean
Obs
SD
1 0.91 0.96
(7.1,0.13) (3.3,0.12)
320 322
1 0.92
(6.9,0.14)
306
1





Key
Airs Site I.D. #
Pearson r
(90th %-tile difference in concentration,
coefficient of divergence)
number of observations

18.34 20.16 18.39
368 931 353
9.59 11.50 9.90
390350065
0.94
(5.4,0.10)
314
0.89
(9.4,0.15)
296
0.93
(5.1,0.14)
309
1





17.47
340
8.95
390350066
0.92
(7.2,0.16)
300
0.85
(13.1,0.21)
290
0.90
(8.15,0.19)
300
0.96
(4.7,0.14)
295
1

14.74
332
8.45
390351002
0.88
(9.0,0.18)
308
0.84
(12.9,0.21)
304
0.87
(8.6, 0.20)
310
0.91
(5.4,0.16)
310
0.88
(4.9,0.16)
304
1
15.0
351
8.16
390851001
0.89
(10.7,0.21)
308
0.84
(14.3, 0.23)
294
0.88
(10.8,0.22)
307
0.90
(7.8, 0.20)
306
0.91
(5.8,0.15)
295
0.89
(6.0,0.18)
303
1
14.0
342
8.44
390932003
0.92
(8.1,0.17)
265
0.892
(11.2,0.18)
256
0.90
(8.9,0.18)
256
0.91
(7.7,0.18)
264
0.91
(5.8,0.13)
247
0.87
(5.3,0.18)
261
0.90
(6.6,0.15)
275
1
15.22
298
8.80
June 2003
3-42
DRAFT-DO NOT QUOTE OR CITE

-------
     TABLE 3-4a-d (cont'd). MEASURES OF THE SPATIAL VARIABILITY OF
        PM2 5 CONCENTRATIONS WITHIN SELECTED METROPOLITAN
                         STATISTICAL AREAS.
(c) Dallas, TX
Site I.D. #
480850005
480850005 481130020 481130035 481130050 481130057
1 0.92 0.94
0.94 0.89
(3.5,0.11) (3.6,0.11) (4.3,0.13) (6.3,0.15)
220 204 213 195
481130020
1 0.95
0.94 0.92
(3.2,0.08) (3.3,0.09) (4.1,0.11)
212 603 205
481130035
1
0.97 0.93
(2.0, 0.06) (3.9, 0.09)
203 191
481130050
481130057
480850005
481130020
Mean
Obs
SD
(d) Los Angeles
Site I.D. #
060370002


Key
Airs Site I.D. #
Pearson r
(90th %-tile difference in concentration,
coefficient of divergence)
number of observations

11.54 12.42 12.83
234 677 222
5.62 5.69 5.79
CA
1 0.94
(2.7, 0.08)
199
1
13.34 13.67
644 215
5.79 6.14

060370002 060371103 060371201 060371301 060372005
1 0.87 0.76
0.68 0.95
(10.7,0.18) (14.6,0.23) (17.9,0.25) (6.2,0.14)
581 208 229 212
060371103
1 0.86
0.89 0.93
(12.8,0.20) (10.1,0.12) (7.1,0.11)
205 222 207
060371201
1
0.76 0.85
(18.1,0.24) (12.1,0.18)
212 197
060371301
060372005
060374002
Mean
Obs
SD
Key
Airs Site I.D. #
Pearson r
(90th %-tile difference in concentration,
coefficient of divergence)
number of observations

20.91 22.48 18.32
641 656 217
13.35 13.47 11.65
1 0.78
(13.2,0.18)
214
1
23.73 20.09
237 220
13.43 11.92
480850005
0.94
(3.7,0.10)
218
0.95
(2.5, 0.07)
635
0.97
(1.9,0.06)
207
0.98
(2.2, 0.06)
608
0.95
(3.1,0.08)
198
1
12.68
687
5.67

060374002
0.60
(18.1,0.26)
553
0.80
(13.6,0.17)
563
0.66
(18.2, 0.24)
197
0.95
(8.1,0.11)
216
0.62
(15,0.20)
204
1
20.31
621
12.42
481130020
0.94
(3.1,0.10)
189
0.97
(2.3, 0.07)
207
0.94
(3.6,0.10)
185
0.94
(4.2,0.12)
196
0.91
(5.5,0.14)
182
0.96
(3.0, 0.09)
198
1
11.67
216
5.43











June 2003
3-43
DRAFT-DO NOT QUOTE OR CITE

-------
 1      analyses to include the relevant CMSAs (consolidated MSA) would also produce a number of
 2      sites that are even less well correlated with each other in part because sites are located outside of
 3      urban airsheds.
 4           Correlation coefficients between pairs of sites in the other 23 MSAs given in Appendix 3 A
 5      for the most part fall within the range of values given in Table 3-4. In four MSAs (Columbia,
 6      SC; Norfolk, VA; Grand Rapids, MI; and Baton Rouge, LA), intersite correlations are all greater
 7      than 0.9.  In nine others (Cleveland, OH; Louisville, KY; Chicago, IL; Milwaukee, WI;
 8      Philadelphia, PA; Detroit, MI; Kansas City, KS-MO; Dallas, TX; and Salt Lake City, UT), they
 9      are all greater than 0.8.  Correlations between sites in the other MSAs examined tend to be lower
10      and span a broader range than for the MSAs mentioned above.
11           Seven pairs of collocated monitors in seven MSAs (Columbia, SC; Dallas, TX; Detroit,
12      MI; Grand Rapids, MI; Louisville, KY; Steubenville, OH; Washington, DC) provide an
13      indication of the performance of collocated monitors (see Table 3A-1). Mean values of r,  P90,
14      and COD for these seven pairs of monitors are 0.986, 1.63 |ig/m3, and 0.060 suggesting that
15      most of the intersite variability seen is not due to sampler imprecision.
16           There are no strong regional patterns evident in the data given in Appendix  3 A except that
17      correlations tend to be higher between monitoring sites in MSAs in the southeastern United
18      States than between monitoring sites in other regions.
19           A number of factors affect intersite correlations within MSAs. These  include field
20      measurement and laboratory analysis errors, placement of monitors close to active sources,
21      placement of monitors in outlying areas, placement of monitors in locations that are isolated
22      topographically from other monitors, placement of monitors in areas outside of local
23      atmospheric circulation regimes (e.g., land-sea breezes), and transient local  events
24      (thunderstorms, sporadic emissions). In several MSAs such as Atlanta, GA; Seattle,  WA;  and
25      Los Angeles-Long Beach, CA, there is at least one site that is remote from the others (by at least
26      100 km), is physically separated from them by mountains, and is really not part of the urban area
27      nor the urban airshed. Correlations between concentrations at these sites and others tend to be
28      lower than among the other sites, and concentration differences tend to be larger.  It should be
29      noted that outlying sites such as these are included in many epidemiologic time-series studies
30      without any weighting (e.g., with respect to the exposed population, spatial  differences in
31      susceptibility) or regard to compositional  differences. Although it is frequently the case that

        June 2003                                 3-44         DRAFT-DO NOT QUOTE OR CITE

-------
 1      distance between sites in urban areas is largely responsible for the spatial variability that is
 2      observed, there are a few instances for which correlations are higher and differences in
 3      concentrations lower for sites that are located farthest apart.  This situation may arise because
 4      these sites are influenced more by the regional background of secondary PM rather than by local
 5      sources, nor is there any set distance below which correlations and differences in concentrations
 6      tend towards some limiting values.  However, it is generally the case that outlying sites are
 7      characterized by lower annual mean concentrations.
 8           Indications of land use (commercial, industrial, residential, agricultural, forest) and
 9      location of sites (urban/city center, suburban, rural) are given in the AIRS data base. Categories
10      such as urban/city center can refer to very different conditions in Columbia, SC, and Chicago,
11      IL. Also, it should not be automatically assumed that concentrations measured at sites
12      categorized as industrial are dominated by local emissions. The PM25 monitoring sites are
13      generally deployed to capture potential population exposures in a variety of environments as
14      opposed to monitoring for compliance as  it exists around local sources. It should be
15      remembered that much of PM2 5 is secondary in origin. The widespread formation of secondary
16      PM coupled with the long lifetime of PM25 ensures some measure of uniformity in the
17      correlations of PM25 across urban areas.  Correlations between many site pairs classified as
18      industrial can be high even though they are separated by large distances, as in the Seattle MSA.
19           Some indication of the variability of primary PM25 produced by local sources can be
20      obtained by examining the variability of carbon monoxide (CO), which is produced mainly by
21      mobile sources (U.S. Environmental Protection Agency, 2000b), and by the variability in
22      elemental carbon (EC) concentrations (Kinney et al., 2000).  CO is relatively inert on the urban
23      scale, and its distribution is governed by the spatial pattern of its emissions and the subsequent
24      dispersion of these emissions not by photochemistry.  Carbon monoxide concentrations are at
25      least a factor of three higher near urban centers than in surrounding rural areas within the four
26      consolidated metropolitan statistical areas examined in the EPA document, Air Quality Criteria
27      for Carbon Monoxide (CO AQCD; U. S. Environmental Protection Agency, 2000b).
28      The correlations of CO within the urban areas examined in that document were all low to
29      moderate.  Therefore, it might be expected that primary PM2 5 produced by local traffic should be
30      at least as heterogeneous as CO in a given urban area. EC is a significant component of diesel
31      exhaust (cf, Appendix 3D). Kinney et al. (2000) measured EC and PM2 5 concentrations at four

        June 2003                                 3-45        DRAFT-DO NOT QUOTE OR CITE

-------
 1      sites located on sidewalks of streets characterized by varying exposures to diesel emissions in
 2      upper Manhattan (Harlem, NY). Whereas the mean PM2 5 concentrations varied by about one-
 3      third from 37 to 47 |ig/m3 at the four sites, mean EC concentrations varied by a factor of four
 4      from 1.5 to  6.2 |ig/m3.  The corresponding ratios of EC to PM25 ranged from 0.039 to 0.14.
 5      Although EC constituted a relatively small fraction of PM25 in this study, spatial variability in its
 6      sources (diesel and gasoline fueled vehicles, resuspended road dust,  and cooking) contributed, on
 7      average, about one-third of the spatial variability observed in PM25 concentrations.  Further
 8      analyses are needed to determine whether the remaining variability could be attributed to other
 9      local and city-wide sources. Because the effects of emissions from local point sources on
10      receptor sites depend strongly on wind direction, correlations involving contributions from local
11      sources can be much lower than from area sources (much as motor vehicle traffic) or from
12      regionally dispersed sources (such as the photochemical production  of secondary  organic PM
13      and sulfate).
14           The difference in mean PM2 5 concentrations between the site with the lowest and the site
15      with the highest mean  concentration range in all MS As included in Appendix 3 A  ranges from
16      0.4 |ig/m3 (Baton Rouge) to about 8 |ig/m3 (Pittsburgh). Six MS As (Chicago, Seattle, Cleveland,
17      St. Louis, Detroit, and  Pittsburgh) show maximum intersite differences in the annual mean larger
18      than 6 |ig/m3.  In the Seattle MSA, there is one monitoring site (Figure 3A-23a) that is separated
19      from the remaining sites by topography and has much lower mean PM25 concentrations, much
20      smaller seasonal variability in concentrations, and much lower maximum concentrations than
21      these other sites.  However, the annual mean concentrations at all the other sites within the
22      Seattle MSA are within 3  |ig/m3 of each other.  Differences in annual mean concentrations are
23      also larger between sites located in different MSAs but within the same CMSA. For example,
24      in the consolidated MSA of Los Angeles-Riverside the range of annual mean PM2 5
25      concentrations is extended from about 20 |ig/m3 in the urban area of Los Angeles county to
26      about 29 |ig/m3 in Riverside County. Large differences in annual mean concentrations within a
27      given area reflect differences in source or meteorological or unique topographic characteristics
28      affecting sites; whereas very small differences found in some areas may only be the result of
29      measurement imprecision.
30           Whereas high correlations of PM2 5 provide an indication of the spatial uniformity in
31      temporal variability (directions of changes)  in PM2 5 concentrations across urban areas, they do

        June 2003                                 3-46        DRAFT-DO NOT QUOTE OR CITE

-------
 1      not imply uniformity in the PM2 5 concentrations themselves.  The 90th percentile difference in
 2      concentrations (P90) and the coefficient of divergence (COD) are used here to give a more
 3      quantitative indication of the degree of spatial uniformity in PM2 5 concentrations across urban
 4      areas. A COD of zero implies that both data sets are identical, and a COD of one indicates that
 5      two data sets are completely different. The calculation of the Pearson correlation coefficient,
 6      P90,  and COD allows for distinctions between pairs of sites to be made based on various
 7      combinations of these parameters. Figure 3-18 shows examples of the varying degree of
 8      heterogeneity in concentrations between pairs of sites that are highly correlated (r > 0.9 for all
 9      three site pairs). The increase in the spread of concentrations between the chosen site-pairs is
10      reflected in increases in both P90 and COD. Pairs of sites showing high correlations and
11      CODs < 0.1 and P90's < 4 |ig/m3 (as in Columbia, SC, Figure 3-7a) indicate homogeneity in both
12      PM2 5 concentrations and in their temporal variations. Presumably, sites such as these are more
13      strongly affected by regional than local sources. Pairs of sites showing low correlations, values
14      of P90 > 10 |ig/m3 and CODs > 0.2, as in  Los Angeles, CA (Table 3-5), indicate heterogeneity in
15      both PM25 concentrations and in their temporal variations. Note that the extended urban area or
16      the CMSA includes Riverside County, as well as Los Angeles County. Even lower correlations
17      and a greater degree of heterogeneity in PM25 concentrations were found in the extended CMSA.
18      Pairs of sites showing high correlations (r > 0.9) and CODs > 0.2 and P90's > 10 |ig/m3 indicate
19      heterogeneity in concentrations but homogeneity in their day to day changes. Selected pairs of
20      sites in the Cleveland MSA show moderate to high correlations coupled with CODs > 0.2 and
21      P90's > 10 |ig/m3 (Table 3-4), suggesting  moderate homogeneity in day to day changes, but
22      significant spatial heterogeneity in concentrations.
23           The effect of local point sources on intersite variability can be seen at several sites among
24      those listed in Table 3-4 and Appendix 3A.  Sites 39-035-0038 (Cleveland, OH; Table 3-4 and
25      Figure 3A-8), 18-089-0022 (Gary, IN; Figure 3A-15), 55-079-0043 (Milwaukee, WI;
26      Figure 3A-13), and 17-119-0023 (St. Louis, MO; Figure 3A-17) are designated as "source
27      oriented" in the AIRS data base in contrast to the "population exposure" objective associated
28      with most of the MSA sites. PM25 concentrations at these sites are weakly correlated with other
29      sites within the MSA as evidenced by low correlation coefficients and  large P90s and CODs even
30      though some of the neighboring sites may be located short distances away. Other sites
31      designated as "source oriented" in Chicago, Milwaukee, and St. Louis  do not show clear

        June 2003                                  3-47        DRAFT-DO NOT QUOTE OR CITE

-------


(O
^
i
o
o
O
o

Q)
E
z
Columbia, SC 1999 & 2000
120 -

100 -

80 -

60 -

40 -

20 -
n
45-079-0007 vs 45-079-0019
r = 0.97
COD = 0.06
P9a = 2.7,«g/m3






1 1 . . _
                                    Chicago, IL 2000
(fl
8
Q)
3
O
o
•5

ID
E
z
30 -,
25 -

20 -

15 -

10 -

5 -
0 -

17-031-2001 vs 17-031-420
r = 0.94















COD = 0.1 4
P9D = 5.5/ig/m3





I.M .
^) ^f\ *\ O\ ^ ^X ^5 ^ O\ /^^
Ix V^ ^O ^J rx' ^^ ' K./ (%.^ *^_' r\.' "^
                                     Detroit, Ml 2000
8 12-
c
E1 10-
3 R
o o -
0
B 6-
Q) 4 -
! 2-
z n







|
























26-099-0009 vs 26-163-0033
r=0.93
COD = 0.22










P90= 12.7/ig/m3
_
1
1 •
nil. .1 . i
                              Concentration Difference
 Figure 3-18.  Occurrence of differences between pairs of sites in three MSAs.  The
              absolute differences in daily average PM2 5 concentrations between sites are
              shown on the x-axis and the number of occurrences on the y-axis. The MSA,
              years of observations, AIRS site I.D. numbers for the site pairs, Pearson
              correlation coefficients (r), coefficients of divergence (COD), 90th percentile
              (P90) difference in concentration between concurrent measurements are also
              shown.
 Source: Pinto et al. (2002).
June 2003
3-48
DRAFT-DO NOT QUOTE OR CITE

-------
         TABLE 3-5. MEASURES OF THE SPATIAL VARIABILITY OF
       PM102 5 CONCENTRATIONS WITHIN SELECTED METROPOLITAN
                         STATISTICAL AREAS
(a) Cleveland, OH
Site 390350013 390350038
390350013 1 0.67
(23.2, 0.26)
182
390350038 1

390350045
0.67
(28.5, 0.28)
95
0.65
(16.1,
0.22)
90
390350045


Key
390350060 Airs Site I.D.#
Pearson r
(90th %-tile difference in concentration
390350065 coefficient of divergence)
number of observations

390851001
MEAN 26.36 18.63
Obs 216 614
SD 17.38 11.60











16











76
112
8.96
390350060
0.73
(17.9,0.22)
97
0.73
(11.7,0.18)
93
0.66
(18.1,0.23)
94

1






21.35
113
16.39
390350065 390851001
0.62 0.41
(27.0,0.31) (40.0,0.60)
98 94
0.69 0.44
(13.9,0.62) (24.9,0.53)
90 89
0.71 0.49
(10.6,0.31) (19.9,0.50)
102 99

0.74 0.31
(15.4,0.38) (28.0,0.59)
93 94
1 0.22
(20.4, 0.55)
99
1
16.79 7.15
111 109
9.49 4.94
(b) Dallas, TX
Site 481130020 481130035
481130020 1 0.79
(4.5,0.17)
54

481130035 1
481130050
0.71
(9.3,
0.22)
55


0.69
(7.8,0.18)
50
481130050

481130057
MEAN 11.22 12.86
Obs 60 55
SD 5.35 6.66






14.46
56
6.44
481130057
0.66
(16.5,0.32)
54

0.60
(13.2,0.30)
50
069
(13.5,0.24)
50
1
19.12
55
10.55




Key
Airs Site I.D. #
Pearson r
(90th %-tile difference in
concentration,
coefficient of divergence)
number of observations




(c) Los Angeles, CA
Site 060370002 060371002
060370002 1 0.82
(19.0, 0.24)
49

060371002 1
060371103
0.63
(15.5,
0.18)
49


0.74
(11.5,0.21)
49
060371103




MEAN 24.10 15.33
Obs 56 56
SD 11.67 6.68





21





44
57
8.65
060374002
0.58
(17.3,0.27)
45

0.54
(11.5,0.25)
47
0.57
(12.5,0.22)
45


16.08
53
6.61




Key
Airs Site I D #
Pearson r
(90th %-tile difference in
concentration,
coefficient of divergence)
number of observations





June 2003
3-49
DRAFT-DO NOT QUOTE OR CITE

-------
 1      evidence that local sources are contributing to intersite variability. Conversely, in the Tampa,
 2      FL MSA pairs of sites are only moderately correlated (0.7 < r < 0.87), but the distribution of
 3      concentrations is rather homogeneous (COD < 0.14 and P90 < 5 |ig/m3; Figure 3 A-7). Thus, a
 4      number of different combinations of spatial uniformity in PM2 5 concentrations and correlations
 5      of these concentrations are found.
 6           Values of P90 for absolute differences in  concentrations between sites span a wide range in
 7      the data set given in Appendix 3 A.  In many instances they  can be quite low, only about a few
 8      ng/m3; these cases are found mainly in the eastern United States.  The largest P90 values were
 9      associated with a single site in Pittsburgh and reached as high as 21 |ig/m3 (Figure 3A-9).
10      Excluding this site, large P90 values are found mainly in the western United States.  Values of P90
11      > 18 |ig/m3 are found in the Riverside and Los-Angeles-Long Beach MS As. Maximum
12      differences in concentrations between sites can be much larger than shown in Figure 3-18 and
13      have been larger than 100 |ig/m3  on several occasions in the Atlanta, GA and Los Angeles-Long
14      Beach, CA MS As. Rizzo and Pinto (2001) and Fitz-Simons et al. (2000) examined correlations
15      between sites located even farther apart than those examined here based on the 1999 AIRS data
16      set for PM2 5. They found that in a number of MS As, PM25 concentrations are still well
17      correlated ® > 0.7) up to distances of 100 km or more.  Leaderer et al. (1999) found r = 0.49
18      between sites outside of homes and a regional  background monitor located from 1 to 175 km
19      away in southwestern Virginia. PM2 5 tends to be correlated over much larger areas  in the East
20      than in the West, mainly because the terrain tends to be flatter over wider areas in the East
21      (Rizzo and Pinto, 2001).  As a result,  there is a greater opportunity for mixing of emissions
22      among dispersed source regions.  Many large urban areas in the West are surrounded by
23      mountains.  The presence of more rugged terrain in the West leads to greater confinement of
24      emissions from large urban areas. Other factors such as differences in the composition and
25      amount of emissions of precursors and in the rates of photochemical oxidation of these
26      emissions in the atmosphere also play a role.
27           There is also evidence for inter-annual variability in the spatial variability in PM2 5
28      concentrations.  The median year-to-year changes in inter-site r (0.03), P90 (-0.75 |ig/m3), and
29      COD (-0.015) from 1999 to 2000 do  not differ significantly from zero for all the site pairs
30      considered in Appendix 3A.  The year-to-year changes in the  spatial variability of PM25
31      concentrations in a number of MS As  such as the Columbia, SC; Grand Rapids, MI; Milwaukee,

        June 2003                                 3-50         DRAFT-DO NOT QUOTE OR CITE

-------
 1      WI; Baton Rouge, LA; Kansas City, KS-MO; Boise, ID; and Portland, OR MSAs are similar and
 2      are smaller than those found in the Cleveland, OH; Salt Lake City, UT; and San Diego, CA
 3      MSAs. The ranges in these parameters are largest for a number of individual site-pairs,
 4      especially those involving sites that are remote from the others in their MSAs. In these MSAs
 5      (such as the Atlanta, GA; Los Angeles, CA; and Seattle, WA MSAs) there are sites that may be
 6      located in different airsheds from the remaining sites.  Year-to-year changes in parameters
 7      describing spatial variability in PM2 5 concentrations tend to be larger when sites in different
 8      counties within a given MSA are considered rather than when sites in the same county are
 9      considered. There are a number of factors that can account for inter-annual variability in these
10      parameters, such as changes in patterns in the emissions of primary PM25; in the transport and
11      rates of transformation of secondary PM2 5 precursors in field measurement; and analysis
12      procedures.
13           Some additional data for indicating the stability with respect to year-to-year changes in
14      spatial variability are available from earlier studies. For example, a comparison between data
15      obtained  during the summers of 1992 and 1993 (Wilson and Suh, 1997), shown in Figure 3-19,
16      and data obtained during the summer of 1994 (Pinto et al., 1995) (cf, Table 3-8) in Philadelphia,
17      PA, suggests that inter-site correlations of PM2 5 have remained high and have changed very little
18      between the two study  periods.
19
20      PM^s
21          Intersite correlations of PM10_25 concentrations obtained during the summers of 1992 and
22      1993 in Philadelphia, PA, (Wilson and Suh, 1997) are shown in Figure 3-19.  As can be seen,
23      correlations of PM10_2 5 are substantially lower than those for PM25.
24          Intersite correlation coefficients can also be calculated for PM10_2 5 based on the AIRS data
25      set as shown in Table 3-5 for the Cleveland, OH; Dallas, TX; and Los Angeles, CA MSAs.
26      However, data for analyzing the spatial variability of PM10_25 are more limited than for PM2 5;
27      therefore, fewer urban areas could be characterized in Appendix 3 A (Figures 3 A-28 to 3 A-44).
28      Whereas PM2 5 concentrations were found to be highly correlated between sites in the Detroit,
29      MI MSA  (Table 3-4), estimated PM10_25 concentrations are noticeably less well correlated.
30      Likewise, correlations of PM10.2 5 in the Chicago, IL MSA are also lower than those for PM2 5.
        June 2003                                3-51        DRAFT-DO NOT QUOTE OR CITE

-------
                    1.0
                    0.9
                    0.8
                    0.7
                  I
                  •5 0.6
                                  Correlations of PM Exposure Indicators
                                      Philidelphia, Summer, 1992-93, 7 Sites
O 0.5
g
03 0.4

8 0.3

  0.2

  0.1

  0.0

    V

y
                            *.
                             \







' /
/
y
/ /
j V






I!
\ / A
I \/\ XI f
Y \X ll\ ^—^

I ^
                                                                          Average r
                                                                      + PM2 5    0.90
                                                                        PM
                                                                          10
                                                   0.86
                                                                      4PM10-2.5  °-38
                                  8
                                                                    28
                                                                            32
                                        12      16     20     24
                                            Distance Between Sites (km)
                              ® Not Significant, all other correlations significant (P < 0.05)
        Figure 3-19.  Intersite correlation coefficients for PM25, PM10, and PM10_25.
        Source:  Wilson and Suh (1997).


 1      In contrast, correlations of PM10.2 5 concentrations between several pairs of sites in the
 2      Los Angeles-Long Beach partial MSA are higher than those for PM2 5.
 3           The interpretation of these results is not straightforward, as concentrations of PM10.25 are
 4      generated by taking the difference between collocated PM2 5 and PM10 monitors.  Consequently,
 5      caution must be exercised when viewing them.  Errors in the measurement of PM25 and PM10
 6      may play a large role in reducing apparent correlations of PM10.25 such that collocated PM10_25
 7      "measurements" may be expected to be poorly correlated (White, 1998). Indeed,  several
 8      estimates of concentrations are negative. Negative PM10_25 concentrations also lead to artifacts
 9      in the calculation of CODs. In cases where these artifacts cause a division by zero or a very
10      small number in the calculation of CODs, dashes are used in Table 3-5 and Figures 3A-28
11      through 3 A-44. These results imply that negative concentrations can be almost or identically
12      equal in absolute magnitude to positive concentrations in the same MSA.  The possible causes of
13      these errors are essentially the same as those discussed in Section 3.2.1 with regard to the
        June 2003
                      3-52
                                             DRAFT-DO NOT QUOTE OR CITE

-------
 1      occurrence of PM25 to PM10 ratios greater than one. There are also physical bases for expecting
 2      that PM10_2 5 concentrations may be more variable than those for PM2 5. PM10_2 5 is mainly
 3      primary in origin, and its emissions are spatially and temporally heterogenous.  Similar
 4      considerations apply to primary PM2 5, but much of PM2 5 is secondary, and sources of secondary
 5      PM are much less spatially and temporally variable. Dry deposition rates of particles depend
 6      strongly on particle size.  Whereas all particles may be brought to the surface by turbulent
 7      motions in the atmosphere; gravitational settling becomes more important with  increasing
 8      particle size. Gravitational settling can effectively limit the horizontal distance a particle can
 9      travel. For example, 10 jim Da particles suspended in a hypothetical 1 km deep planetary
10      boundary layer can be removed within a few hours, but 1 |im Da particles can remain suspended
11      in the atmosphere for up to 100 to 1,000 times longer before being dry deposited. (Estimated
12      atmospheric lifetimes were based on deposition velocities given in Lin et al. [1994] for typical
13      wind speeds.) The findings of larger correlations of PM10_25 between several site pairs in the
14      Los Angeles basin (cf, Figures 3A-25/26 and Figures 3A-42/43) and one other site pair in the
15      St. Louis, MO-IL MSA (cf, Figures 3A-17 and 3A-37) are anomalous in light of the discussion
16      above.  However, these findings could have resulted from differences between the spatial and
17      temporal behavior of sources of PM2 5 and PM10_2 5 in these locations.  Because of negative
18      values, CODs were not calculated.
19
20      PM Components
21           Three methods for comparing the chemical composition of aerosol  databases obtained at
22      different locations and times were discussed by Wongphatarakul et al. (1998). Log-log plots of
23      chemical concentrations obtained at pairs of sampling sites accompanied by the coefficient of
24      divergence (COD) were examined as a way to provide an easily visualized means of comparing
25      two data sets2.  Examples comparing downtown Los Angeles with Burbank and with
              2The COD for two sampling sites is defined as follows:
                                              1  P  I  v  — v  }
                                              ly    *?   **                              0-2)
                                              n^1  {  x  + x
                                              P i = \  \Xij + XikJ
        where x^ represents the average concentration for a chemical component i at site j, j and k represent two sampling
        sites, and p is the number of chemical components.

        June 2003                                 3-53        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Riverside-Rubidoux are shown in Figures 3-20 and 3-21.  As the composition of two sampling
 2      sites become more similar, the COD approaches zero; as their compositions diverge, the COD
 3      approaches one.  Correlation coefficients calculated between components can be used to show
 4      the degree of similarity between pairs of sampling sites.
 5           In addition to calculating correlation coefficients for total mass or for individual
 6      components, correlation coefficients for characterizing the spatial variation of the contributions
 7      from given source types can also be calculated by averaging the correlation coefficients of the
 8      set of chemical components that represent the source type. Correlation coefficients showing the
 9      spatial relationships among PM2 5 (total) and contributions from different source categories
10      obtained at various sites in the South Coast Air Basin (SoCAB) Study are shown in Table 3-6.
11      In Wongphatarakul et al. (1998), crustal material (crustal), motor vehicle exhaust (mv), residual
12      oil emissions (residual oil), and secondary PM (sec) were  considered as source categories.
13      Al, Si, Fe, and Ca were used as markers for crustal material (crustal); V and Ni were used as
14      markers for fuel oil combustion (residual oil); and Pb, Br,  and Mn were used as markers for
15      motor vehicle exhaust (mv), based on the lack of other, perhaps more suitable, tracers. NO3",
16      NH4+, and SO4"2 represent secondary PM components.  The average of the correlation
17      coefficients of marker elements within each source category are shown in Table 3-6. Values of
18      rsec and rmv are much higher than those for rcrustal and rresidualoil throughout the SoCAB, suggesting a
19      more uniform distribution of the contributions from secondary PM formation and automobiles
20      than from crustal material and localized stationary sources.
21           Correlation coefficients in Philadelphia air based on data obtained at four sites for PM2 5
22      (total),  crustal components (Al, Si, Ca, and Fe), the major secondary component (sulfate),
23      organic carbon (OC), and elemental carbon (EC) are shown in Table 3-7. Because these data
24      were obtained after Pb had been phased out of gasoline, a motor vehicle contribution could not
25      be estimated from the data. Pb also is emitted by  discrete point sources, such as the Franklin
26      smelter. Concentrations of V and Ni were often beneath detection limits; so, the spatial
27      variability in PM due to residual oil combustion were not  estimated. Sulfate in aerosol samples
28      collected in Philadelphia arises mainly from long-range transport from regionally dispersed
29      sources (Dzubay et al., 1988).  This conclusion is strengthened by the high correlations in sulfate
30      between different monitoring sites and the uniformity in sulfate concentrations observed among
31      the sites.  Widespread area sources (e.g., motor vehicle traffic) also may emit pollutants that are

        June 2003                                 3-54        DRAFT-DO NOT QUOTE OR CITE

-------
        102
        10°-
    c.
    CO
    .Q   A n-1
    S_
    13
    00
10  -
        io-2-
        10
          -3
                  COD=0.099
                                                                   TO"
            10
              -3
                                                   SO,
                                                      2-
                                                              nknown
                                                                NO;
                         T
                                           I
                                         10C
                                     T
10'2         10'1          10°          101

  Downtown Los Angeles (|jg/m3)
102
Figure 3-20. PM2 5 chemical components in downtown Los Angeles and Burbank (1986)
            have similar characteristics. The spread in the data is shown by the bars.


Source: Wongphatarakul et al. (1998).
correlated between sites provided that traffic patterns and emissions are similar throughout the

area under consideration.

     Landis et al. (2001) found relatively high correlations between PM2 5 (r = 0.97), sulfate

(r = 0.99), OC (r = 0.97),  EC (r = 0.83), NaCl (r = 0.83), and nitrate (r = 0.83) measured at two
June 2003
                               3-55
                         DRAFT-DO NOT QUOTE OR CITE

-------
              10
                -3
                  10
                     -3
10
                                  -2
                                 Downtown Los Angeles (|jg/m3)
      Figure 3-21.  Concentrations of PM25 chemical components in Rubidoux and downtown
                   Los Angeles (1986).  The diagram shows a significant spread in the
                   concentrations for the two sites compared with downtown Los Angeles and
                   Burbank (Figure 3-20).

      Source:  Wongphatarakul et al. (1998).
1     sites located several km apart in the Baltimore, MD area.  Concentrations of crustal material
2     (r = 0.63) and the sum of total metal oxides (r = 0.76) were not as well correlated.  These results
3     are consistent with those for another eastern city, Philadelphia, PA, given in Table 3-7.
      June 2003
               3-56
DRAFT-DO NOT QUOTE OR CITE

-------
   TABLE 3-6. CORRELATION COEFFICIENTS FOR SPATIAL VARIATION OF
       PM2 5 MASS AND DIFFERENT SOURCES FOR PAIRS OF SAMPLING
                SITES IN THE SOUTH COAST AIR BASIN (1986)

Hawthorne and Rubidoux
Long Beach and Rubidoux
Anaheim and Rubidoux
Downtown Los Angeles and Rubidoux
Burbank and Rubidoux
Hawthorne and Anaheim
Long Beach and Anaheim
Burbank and Anaheim
Downtown Los Angeles and Anaheim
Downtown Los Angeles and Hawthorne
Burbank and Hawthorne
Long Beach and Burbank
Long Beach and Hawthorne
Downtown Long Angeles and Long Beach
Downtown Los Angeles and Burbank
^total
-0.027
0.051
0.066
0.095
0.120
0.760
0.852
0.770
0.827
0.808
0.704
0.731
0.880
0.842
0.928
crustal





0.034
0.075
0.105
0.143
0.568
0.599
0.633
0.649
0.653
0.825
rsec





0.768
0.888
0.749
0.804
0.854
0.790
0.737
0.909
0.817
0.960
mv





0.492
0.504
0.579
0.556
0.669
0.688
0.714
0.861
0.719
0.871
residual oil





0.170
0.150
0.161
0.233
0.533
0.491
0.295
0.482
0.378
0.606
 Source: Wongphatarakuletal. (1998).
   TABLE 3-7. CORRELATION COEFFICIENTS FOR SPATIAL VARIATION OF
     PM25 MASS AND DIFFERENT COMPONENTS FOR PAIRS OF SAMPLING
                      SITES IN PHILADELPHIA (1994)

Castor Ave. and Roxboro
Castor Ave. and NE Airport
Castor Ave. and Broad St.
Roxboro and NE Airport
Roxboro and Broad St.
NE Airport and Broad St.
rtot
0.92
0.93
0.93
0.98
0.95
0.95
-'crustal
0.52
0.47
0.57
0.67
0.90
0.69
rsec
0.98
0.99
0.99
0.98
0.98
0.99
roc
0.88
0.88
0.85
0.83
0.86
0.84
TEC
0.84
0.77
0.89
0.82
0.79
0.63
rPb
0.43
-0.07
0.11
0.20
0.47
0.11
 Source: Pinto et al. (1995).
June 2003
3-57
DRAFT-DO NOT QUOTE OR CITE

-------
 1      The results presented above for Philadelphia, PA; Baltimore, MD; and Los Angeles, CA,
 2      indicate that secondary PM components are more highly correlated than primary components
 3      and may be more highly correlated than total PM2 5.  These results suggest that the correlation of
 4      PM concentrations across an urban area may depend on the relative proportions of primary and
 5      secondary components of PM at individual sites. Sampling artifacts affecting the measurement
 6      of nitrate and organic carbon can obscure these relations and may depress correlations between
 7      sites.
 8           Kao and Friedlander (1995) examined  the statistical properties of a number of PM
 9      components in the South Coast Air Basin (Los Angeles area).  They found that, regardless of
10      source type and location within their study area, the concentrations of nonreactive, primary
11      components of PM10 had approximately log-normal frequency distributions with constant values
12      of the geometric standard deviations (GSDs). However, aerosol constituents of secondary origin
13      (e.g., SO4"2, NH4+, and NO3") were found to have much higher GSDs. Surprisingly, the GSDs of
14      organic (1.87) and elemental (1.74) carbon were both found to be within 1 SD (0.14) of the mean
15      GSD (1.85) for nonreactive primary species, compared to GSD's of 2.1 for sulfate, 3.5 for
16      nitrate, and 2.6 for ammonium. These results suggest that most of the organic carbon seen in
17      ambient samples in the South Coast Air Basin was of primary origin. Pinto et al. (1995) found
18      similar results for data obtained during the summer of 1994 in Philadelphia. Further  studies are
19      needed to determine if these relations are valid at other locations and to what extent the results
20      might be influenced by sampling artifacts  such as the evaporation of volatile constituents during
21      or after sampling.
22           The use of correlations between OC  and EC and OC to EC ratios based on a comparison
23      between values measured in source emissions and ambient observations has also been suggested
24      as a means to distinguish between secondary; and primary sources of OC  (Turpin and
25      Huntzicker, 1995; Strader et al., 1999).  Ratios of OC to EC from combustion sources are
26      typically three or less and may even be less than one in diesel emissions (cf, Appendix 3D).
27      Cabada et al. (2002) concluded that secondary organic PM can contribute from 10 to 35% of
28      total organic PM on an annual basis, with values > 50% during the summer and 0% during the
29      winter months in Pittsburgh, PA, based on chemistry-transport model results and comparison
30      with emissions inventory values of OC to EC ratios.  All of these inferences are subject to
31      considerable uncertainty in the methods for measuring OC  (as discussed in Section 2).

        June 2003                                 3-58         DRAFT-DO NOT QUOTE OR CITE

-------
 1      Ambiguity also arises in the ratio method as the ratios may change due to chemical reactions
 2      occurring during aging of the particles.  However, not much work has been done on this issue.
 3      The ratio can be greater than nine in emissions from wildfires, and transport from distant fires
 4      can skew results unless this transport is taken into account (cf, Appendix 3D). Modeling studies
 5      that rely on OC to EC ratios in emissions inventories to predict the amount of secondary OC may
 6      be biased towards higher ratios of secondary OC because emissions of primary biologic particles
 7      are not included in the inventories.  Additional concerns arise from uncertainties in the
 8      mechanism of formation of secondary OC from gaseous biogenic and anthropogenic precursor
 9      emissions and the uncertainty in those emissions (Section 3.3.1).  It is clear, however, that
10      secondary organic PM is being formed in the atmosphere (Blando et al., 1998 and Appendix 3C).
11           Few studies have compared aerosol composition in urban areas to that in nearby rural
12      areas. One exception is Tanner and Parkhurst (2000), which indicates that sulfate constituted a
13      larger fraction of fine particle mass at rural sites in the Tennessee Valley PM25 monitoring
14      network than did  organic carbon. For urban sites, the situation was largely reversed:  organic
15      carbon constituted a larger fraction of aerosol mass than sulfate. Future systematic comparisons
16      of urban-rural differences in aerosol properties should be facilitated with implementation of the
17      national speciation network and continued operation of the IMPROVE network.
18
19
20      3.3   SOURCES  OF PRIMARY AND SECONDARY PARTICULATE
21            MATTER
22           Information about the nature and relative importance of sources of ambient PM is
23      presented in this section.  Table 3-8 summarizes anthropogenic and natural sources for the major
24      primary and secondary aerosol constituents of fine and coarse particles. Anthropogenic sources
25      can be further divided into stationary and mobile sources. Stationary sources include those such
26      as:  fuel combustion for electrical utilities, residential space heating, and industrial processes;
27      construction and demolition; metals, minerals, and petrochemicals; wood products processing;
28      mills and elevators used in agriculture; erosion from tilled lands; waste disposal and recycling;
29      and fugitive dust from paved and unpaved roads. Mobile or transportation-related sources
30      include direct emissions of primary PM and secondary PM precursors from highway and off-
31      highway vehicles and non-road sources. In addition to fossil fuel combustion, biomass in the
32      form of wood is burned for fuel. Vegetation is burned to clear new land for agriculture and for
        June 2003                                 3-59        DRAFT-DO NOT QUOTE OR CITE

-------
                     TABLE 3-8.  CONSTITUENTS OF ATMOSPHERIC PARTICLES AND THEIR MAJOR SOURCES1
to
O
o
Sources
Primary (PM < 2.5 um) Primary (PM > 2.5 um) Secondary PM Precursors (PM < 2.5 um)
Aerosol
species Natural Anthropogenic Natural
SO4"2 Sea spray Fossil fuel combustion Sea spray
Sulfate
N03- — — —
Nitrate
Anthropogenic Natural
— Oxidation of reduced sulfur
gases emitted by the oceans and
wetlands and SO2 and H2S
emitted by volcanism and forest
fires
— Oxidation of NO,, produced by
soils, forest fires, and lighting
Anthropogenic
Oxidation of SO2
emitted from fossil
fuel combustion
Oxidation of NOX
emitted from fossil
        Minerals
                                                                                                                                     fuel combustion and
                                                                                                                                     in motor vehicle
                                                                                                                                     exhaust
Erosion and
re-entrainment
Fugitive dust paved and
unpaved roads,
agriculture, forestry,
construction, and
demolition
Erosion and
re-entrainment
Fugitive dust, paved
and unpaved road dust,
agriculture, forestry,
construction, and
demolition
NH4+ —
Ammonium

Organic Wildfires
carbon (OC)
Elemental Wildfires
carbon
(EC)
Metals Volcanic
activity

Bioaerosols Viruses and
bacteria
—

Prescribed burning,
wood burning, motor
vehicle exhaust, and
cooking
Motor vehicle exhaust,
wood burning, and
cooking
Fossil fuel combustion,
smelting, and brake
wear
—

— —

Soil humic matter Tire and asphalt wear
and paved road dust
— Tire and asphalt wear
and paved road dust
Erosion, re-entrainment, —
and organic debris

Plant and insect —
fragments, pollen, fungal
spores, and bacterial
agglomerates
Emissions of NH3 from wild Emissions of NH3
animals, and undisturbed soil from animal
husbandry, sewage,
and fertilized land
Oxidation of hydrocarbons Oxidation of
emitted by vegetation (terpenes, hydrocarbons emitted
waxes) and wild fires by motor vehicles,
prescribed burning,
and wood burning

— —


— —

        'Dash (-) indicates either very minor source or no known source of component.

-------
 1      building construction, to dispose of agricultural and domestic waste, to control the growth of
 2      animal or plant pests, and to manage forest resources (prescribed burning).  Also shown are
 3      sources for precursor gases whose oxidation forms secondary particulate matter.  The
 4      atmospheric chemical processes producing secondary PM are described in Section 3.3.1.
 5           In general, the sources of fine PM are very different from those for coarse PM.  Some of
 6      the mass in the fine size fraction forms during combustion from material that has volatilized in
 7      combustion chambers and then recondensed before emission into the atmosphere. Some ambient
 8      PM2 5 forms in the atmosphere from photochemical reactions involving precursor gases.  PM
 9      formed by the first mechanism is referred to as primary, and PM formed by the second
10      mechanism is referred to as secondary.  PM10_2 5 is mainly primary in origin as it is produced by
11      the abrasion of surfaces or by the suspension of biological material. Because precursor gases
12      undergo mixing during transport from their sources, it is difficult to identify individual sources
13      of secondary constituents of PM. Transport and transformations of precursors can occur over
14      distances of hundreds of kilometers. The coarse PM constituents have shorter lifetimes in the
15      atmosphere, so their effects tend to be more localized. Only major sources for each constituent
16      within each broad category shown at the top of Table 3-8 are listed. Not all  sources are equal in
17      magnitude.  Chemical characterizations of primary particulate emissions for a wide variety of
18      natural and anthropogenic sources (as shown in Table 3-8) were given in Chapter 5 of the 1996
19      PM AQCD.  Summary tables of the composition of source emissions presented in the 1996 PM
20      AQCD and updates to that information are provided in Appendix 3D. The profiles of source
21      composition were based in large measure on the results of various studies that collected
22      signatures for use in source apportionment studies.
23           Natural sources of primary PM include windblown dust from undisturbed land,  sea spray,
24      and plant and insect debris. The oxidation of a fraction of terpenes  emitted by vegetation and
25      reduced sulfur species from anaerobic environments leads to secondary  PM formation.
26      Ammonium (NH4+) ions, which play a major role in regulating the pH of particles, are derived
27      from emissions of ammonia (NH3) gas.  Source categories for NH3 have been divided into
28      emissions from undisturbed soils (natural) and emissions that are related to human activities
29      (e.g., fertilized lands, domestic and farm animal waste).  There is ongoing debate about
30      characterizing emissions from wildfires (i.e., unwanted fire) as either natural or anthropogenic.
31      Wildfires have been listed in Table 3-8 as natural in origin, but land management practices and

        June  2003                                3-61        DRAFT-DO NOT QUOTE OR CITE

-------
 1      other human actions affect the occurrence and scope of wildfires.  For example, fire suppression
 2      practices allow the buildup of fire fuels and increase the susceptibility of forests to more severe
 3      and infrequent fires from whatever cause, including lightning strikes.  Similarly, prescribed
 4      burning is listed as anthropogenic, but can be viewed as a substitute for wildfires that would
 5      otherwise occur eventually on the same land.
 6           The transformations that gaseous precursors to secondary PM undergo after being emitted
 7      from the sources shown in Table 3-8 are described in Section 3.3.1. Aspects of the transport of
 8      primary PM and secondary PM, including the transport of material from outside the United
 9      States, are described in Section 3.3.2.  A brief introduction to the deposition of particles is also
10      given in Section 3.3.2, and a more detailed discussion of deposition processes is presented in
11      Chapter 4. Methods to infer contributions from different source categories to ambient PM using
12      receptor models and the results of these modeling efforts are given in Section 3.3.3. Estimates of
13      emissions of primary PM and precursors to  secondary PM from major sources are presented in
14      Section 3.3.4. A discussion of the uncertainties associated with these emissions is given in
15      Section 3.3.5.
16
17      3.3.1    Chemistry of Secondary PM Formation
18          Precursors to secondary PM have natural  and anthropogenic sources, just as primary PM
19      has natural and anthropogenic sources. The major atmospheric chemical transformations leading
20      to the formation of particulate nitrate and sulfate are relatively well understood; whereas those
21      involving the formation of secondary aerosol organic carbon are less so and are still subject to
22      much current investigation. A large number of organic precursors are involved; many of the
23      kinetic details still need to be determined; and many of the actual products of the oxidation of
24      hydrocarbons have yet to be identified.
25
26      Formation of Sulfates and Nitrates
27          A substantial fraction of the fine particle mass, especially during the warmer months of the
28      year, is secondary sulfate and nitrate formed as the result of atmospheric reactions.  Such
29      reactions involve the gas phase conversion of SO2 to H2SO4 (which forms liquid particles)
30      initiated by  reaction with OH radicals and aqueous-phase reactions of SO2 with H2O2, O3, or O2
31      (catalyzed by Fe and Mn).  These heterogeneous reactions may occur in cloud and fog droplets

        June 2003                                 3-62        DRAFT-DO NOT QUOTE OR CITE

-------
 1      or in films on atmospheric particles. NO2 can be converted to gaseous HNO3 by reaction with
 2      OH radicals during the day.  At night, NO2 also is oxidized to nitric acid by a sequence of
 3      reactions initiated by O3 that produce nitrate radicals (NO3) and dinitrogenpentoxide (N2O5) as
 4      intermediates. Both H2SO4 and HNO3 react with atmospheric ammonia (NH3).  Gaseous NH3
 5      reacts with gaseous HNO3 to form paniculate NH4NO3. Gaseous NH3 reacts with H2SO4 to form
 6      acidic HSO4  (in NH4 HSO4) as well as SO4"2 in (NH4)2SO4. In addition, acid gases such as SO2
 7      and HNO3 may react with coarse alkaline particles to form coarse secondary PM containing
 8      sulfate and nitrate. Examples include reactions with basic compounds resulting in neutralization
 9      (e.g., CaCO3 + 2 HNO3 - Ca (NO3)2 + H2CO3T) or with salts of volatile acids resulting in release
10      of the volatile acid (e.g., SO2 + 2NaCl + H2O - Na2SO3 + 2HC1T).
11           If particulate NH4NO3 coagulates with an acidic sulfate particle (H2SO4 or HSO4), gaseous
12      HNO3 will be released, and the NH3 will increase the neutralization of the acidic sulfate.  Thus,
13      in the eastern United States, where PM tends to be acidic, sulfate usually  constitutes a larger
14      fraction of PM mass than nitrate. However, in the western United States, where higher NH3 and
15      lower SO2 emissions permit complete neutralization of H2SO4, the concentration of nitrate could
16      be higher than that of sulfate as it is in areas such as the Los Angeles Basin and the San Joaquin
17      Valley. As SO2 concentrations in the atmosphere in the eastern United States are reduced, the
18      NH3 left in the atmosphere after neutralization of H2SO4 will be able to react with HNO3 to form
19      NH4NO3. Therefore, a reduction in SO2 emissions, especially without a reduction in NOX
20      emissions, could lead to an increase in NH4NO3 concentrations (West et al., 1999; Ansari and
21      Pandis, 1998).  Thus, possible environmental effects of NH4NO3 are of interest for both the
22      western and eastern United States.
23           Chemical reactions of SO2 and NOX within plumes are an important source of H+, SO4"2,
24      and NO3.  These conversions can occur by gas-phase and aqueous-phase mechanisms.
25      In power-plant or smelter plumes containing SO2 and NOX, the gas-phase chemistry depends on
26      plume dilution, sunlight, and volatile organic compounds either in the plume or in the ambient
27      air mixing into and diluting the plume.  For the conversion of SO2 to H2SO4 in the gas-phase in
28      such plumes  during summer midday conditions in the eastern United States, the rate typically
29      varies between 1 and 3% h"1 but in the  cleaner western United States rarely exceeds 1% h"1.
30      For the conversion of NOX to HNO3, the gas-phase rates appear to be approximately three times
        June 2003                                 3-63        DRAFT-DO NOT QUOTE OR CITE

-------
 1      faster than the SO2 conversion rates.  Winter rates for SO2 conversion are approximately an order
 2      of magnitude lower than summer rates.
 3          The contribution of aqueous-phase chemistry to particle formation in point-source plumes
 4      is highly variable, depending on the availability of the aqueous phase (wetted aerosols, clouds,
 5      fog, and light rain) and the photochemically generated gas-phase oxidizing agents, especially
 6      H2O2 for SO2 chemistry. The in-cloud conversion rates of SO2 to SO4"2 can be several times
 7      larger than the gas-phase rates given above. Overall, it appears that SO2 oxidation rates to SO4"2
 8      by gas-phase and aqueous-phase mechanisms may be comparable in summer, but aqueous-phase
 9      chemistry may dominate in winter. Further details concerning the chemistry of SO2 and NOX in
10      power plant plumes can be found in Hewitt (2001).
11          In the western United States, markedly higher SO2 conversion rates have been reported in
12      smelter plumes than in power plant plumes. The conversion occurs predominantly by a gas-
13      phase mechanism.  This result is attributed to the lower NOX in smelter plumes.  In power plant
14      plumes, NO2 depletes OH radicals and competes with SO2 for OH radicals.
15          In urban plumes, the upper limit for the gas-phase SO2 conversion rate appears to be about
16      5% h"1 under the more polluted conditions. For NO2, the rates appear to be approximately three
17      times faster than the SO2 conversion rates. Conversion rates of SO2 and NOX in background air
18      are comparable to the peak rates in diluted plumes.  Neutralization of H2SO4 formed by SO2
19      conversion increases with plume age and background NH3 concentration.  If the NH3
20      concentrations are more than sufficient to neutralize H2SO4 to (NH4)2SO4, the HNO3 formed
21      from NOX conversions may be converted to NH4NO3.
22
23      Formation of Secondary Organic Particulate Matter (SOPM)
24          Atmospheric reactions involving volatile  organic compounds such as alkanes, alkenes,
25      aromatics, cyclic olefins, and terpenes (or any reactive organic gas that contains at least seven
26      carbon atoms) yield organic compounds with low saturation vapor pressures at ambient
27      temperature.  Such reactions may occur in the gas phase, in fog or cloud droplets (Graedel and
28      Goldberg, 1983; Faust, 1994), or possibly in aqueous aerosols (Aumont et al., 2000).  Reaction
29      products from the oxidation of reactive organic gases also may nucleate to form new particles or
30      condense on existing particles to form secondary organic PM (SOPM). Organic compounds
31      with two double bonds or cyclic olefins may react to form dicarboxylic acids, which, with four

        June 2003                                3-64         DRAFT-DO NOT QUOTE OR CITE

-------
 1      or more carbon atoms, also may condense. Both biogenic and anthropogenic sources contribute
 2      to primary and secondary organic particulate matter (Grosjean, 1992; Hildemann et al., 1996;
 3      Mazurek et al., 1997; Schauer et al., 1996). Oxalic acid was the most abundant organic acid
 4      found in PM2 5 in California (Poore, 2000).
 5           Although the mechanisms and pathways for forming inorganic secondary particulate
 6      matter are fairly well known, those for forming SOPM are not as well understood.  Ozone and
 7      the OH radicals are thought to be the major initiating reactants.  However, HO2 and NO3 radicals
 8      also may initiate reactions; and organic radicals may be nitrated by HNO2, HNO3, or NO2.  Pun
 9      et al. (2000) discuss formation mechanisms for highly oxidized, multifunctional organic
10      compounds. The production of such species has been included in a photochemical model by
11      Aumont et al.  (2000), for example. Understanding the mechanisms of formation of secondary
12      organic PM is important because SOPM can contribute in a significant way to ambient PM
13      levels, especially during photochemical smog episodes. Experimental studies of the production
14      of secondary organic PM in ambient air have focused on the Los Angeles Basin. Turpin and
15      Huntzicker (1991,  1995) and Turpin et al. (1991) provided strong evidence that secondary PM
16      formation occurs during periods of photochemical ozone formation in Los Angeles and that as
17      much as 70% of the organic carbon in ambient PM was secondary in origin during a smog
18      episode in 1987. Schauer et al. (1996) estimated that 20 to 30% of the total organic carbon
19      PM in the < 2.1 |im size range in the Los Angeles airshed is  secondary in origin on an annually
20      averaged basis.
21           Pandis et al. (1992) identified three mechanisms for formation of SOPM:  (1) condensation
22      of oxidized end-products of photochemical reactions (e.g., ketones, aldehydes, organic acids, and
23      hydroperoxides), (2) adsorption of semivolatile organic compounds (SVOC) onto existing solid
24      particles (e.g., poly cyclic aromatic hydrocarbons), and (3) dissolution of soluble gases that can
25      undergo reactions in particles (e.g., aldehydes).  The first and third mechanisms are expected to
26      be of major importance during the summer when photochemistry is at its peak.  The second
27      pathway can be driven by diurnal and seasonal temperature and humidity variations at any time
28      of the year. With regard to the first mechanism, Odum et al. (1996) suggested that the products
29      of the photochemical oxidation of reactive organic gases are semivolatile and can partition
30      themselves onto existing organic carbon at concentrations below their saturation concentrations.
        June 2003                                 3-65        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Thus, the yield of SOPM depends not only on the identity of the precursor organic gas but also
 2      on the ambient levels of organic carbon capable of absorbing the oxidation products.
 3          Haagen-Smit (1952) first demonstrated that hydrocarbons irradiated in the presence of NOX
 4      produce light scattering aerosols. The aerosol-forming potentials of a wide variety of individual
 5      anthropogenic and biogenic hydrocarbons were compiled by Pandis et al. (1992) based mainly
 6      on estimates made by Grosjean and Seinfeld (1989) and on data from Pandis et al. (1991) for
 7      p-pinene and from Izumi and Fukuyama (1990) for aromatic hydrocarbons.  Zhang et al. (1992)
 8      examined the oxidation of cc-pinene. Pandis et al. (1991) found no aerosol products formed in
 9      the photochemical oxidation of isoprene, although they and Zhang et al. (1992) found that the
10      addition of isoprene to  reaction mixtures increased the reactivity of the systems studied. Further
11      details about the oxidation mechanisms and secondary organic PM yields from various reactive
12      organic gases are given in the above studies. Estimates of the production rate  of secondary
13      organic PM in the Los Angeles airshed are provided in the 1996 PM AQCD (U.S.
14      Environmental Protection Agency, 1996).
15          More recently, Odum et al. (1997a,b) have found that the aerosol-formation potential of
16      whole gasoline vapor can be accounted for solely by summing the contributions of the individual
17      aromatic compounds in the fuel.  In general, data for yields for secondary organic PM formation
18      can be broken into two distinct categories. The oxidation of toluene and aromatic compounds
19      containing ethyl or propyl groups (i.e., ethylbenzene, ethyltoluene, n-propylbenzene) produced
20      higher yields of secondary organic PM than did the oxidation of aromatic compounds containing
21      two or more methyl groups (i.e., xylenes, di-, tri-, tetra-methylbenzenes). Yields in the first
22      group ranged from about 7 to  10%; and in the second group, they ranged from 3 to 4% for
23      organic carbon concentrations between 13 and 100 |ig/m3. Reasons for the differences in
24      secondary organic PM yields found between the two classes of compounds are not clear.
25          There have been a few recent studies that have examined the composition of secondary
26      organic PM. Edney et al. (2001) carried out a smog chamber study to investigate the formation
27      of multi-functional oxygenates from photooxidation of toluene. The experiments were carried
28      out by irradiating toluene/propylene/NOx/air mixtures in  a smog chamber operated in the
29      dynamic mode and analyzing  the collected aerosol by positive chemical ionization GC-MS after
30      derivatization of the carbonyl  oxidation products. The results of the GC-MS analyses were
31      consistent with the formation  of semivolatile multi-functional oxygenates, including hydroxy

        June 2003                                 3-66        DRAFT-DO NOT QUOTE OR CITE

-------
 1      diones as well as triones, tetraones, and pentaones.  The authors also suggested that some of
 2      these compounds could be present in SOPM in the form of polymers.
 3           Jang and Kamens (200la) employed a number of analytical approaches, including GC-MS
 4      detection of volatile derivatives of carbonyl, hydroxy, and acid compounds in SOPM formed in
 5      the irradiation of toluene/NOx mixtures. A wide range of substituted aromatics, nonaromatic
 6      ring-retaining and ring-opening products were detected. Newly identified ring-opening
 7      oxycarboxylic acids detected included glyoxylic acid; methylglyoxylic acid; 4-oxo-2-butenoic
 8      acid; oxo-C5-alkenoic acids; dioxopentenoic acids; oxo-C7-alkadienoic acids; dioxo-C6-alkenoic
 9      acids; hydroxydioxo-C7-alkenoic acids; and hydroxytrioxo-C6-alkanoic acids.  Other newly
10      identified compounds included methylcyclohexenetriones; hydroxymethylcyclohexenetriones;
11      2-hydroxy-3-penten-l,5-dial, hydroxyoxo-C6-alkenals; hydroxy-C5-triones, hydroxydioxo-C7-
12      alkenals; and hydroxy-C6-tetranones. Included among these compounds were a number of the
13      hydroxy polyketones detected by Edney et al., (2001).
14           Recent laboratory and field studies support the concept that nonvolatile and semivolatile
15      oxidation products from the photooxidation of biogenic hydrocarbons contribute significantly to
16      ambient PM concentrations in both urban and rural environments. The oxidation of a variety of
17      biogenic hydrocarbons emitted by trees and plants, such as terpenes (
-------
 1      able to calculate formation rates with a kinetic model including formation mechanisms for
 2      O3 + cc-pinene reaction products.
 3           Griffin et al. (1999) introduced the concept of incremental aerosol reactivity, the change in
 4      the secondary organic aerosol mass produced (in |ig/m3) per unit change of parent organic
 5      reacted (in ppb), as a measure of the aerosol-forming capability of a given parent organic
 6      compound in a prescribed mixture of other organic compounds.  They measured the incremental
 7      aerosol reactivity for a number of aromatic and biogenic compounds for four initial mixtures.
 8      Incremental aerosol reactivity ranged from 0.133 to 10.352 jig m"3 ppb"1 and varied by almost a
 9      factor of two depending on the initial mixture.
10           A number of multifunctional oxidation products produced by the oxidation of biogenic
11      hydrocarbons have been identified in laboratory studies (Yu et al., 1998; Glasius et al., 2000;
12      Chri staffer sen et al., 1998; Koch et al., 2000; and Leach et al., 1999). Many of these  compounds
13      have subsequently been identified in field investigations (Yu et al., 1999; Kavouras et al., 1998,
14      1999a,b; Pio et al., 2001; and Castro et al., 1999). Most studies of the formation of secondary
15      organic aerosol formation from terpenes have focused on their reactions with ozone.  There have
16      been many fewer studies dealing with the oxidation of terpenes initiated by OH radicals. Larsen
17      et al. (2001) found that the major aerosol products produced ultimately from the reaction of OH
18      radicals with monoterpenes with endocyclic double bonds (cc-pinene, 3-carene) were C10
19      keto-carboxylic acids (such as pinonic and caronic acids); whereas the major products from the
20      oxidation of monoterpenes with exocyclic double bonds (p-pinene) were C9-dicarboxylic acids
21      (such as pinic acid), and the major product from the oxidation of limonene (which has both
22      endo-  and exocyclic double bonds) was 3-acetyl-6-oxo-heptanal (keto-limonaldehyde).  A large
23      number of related aldehydes, ketones and acids were  also found in their experiments.  However,
24      the total yields of condensable products are much lower than for the corresponding reactions
25      with ozone. For example, yields  of C9-dicarboxylic acids, C10-hydroxy-keto-carboxylic acids,
26      and C10-hydroxy-keto-aldehydes from the reaction of ozone with mono-terpenes with endocyclic
27      double bonds ranged form 3% to 9%; whereas they ranged only from 0.4 to 0.6% in the reaction
28      with OH radicals. Likewise, the reaction of monoterpenes with exocyclic double bonds with
29      ozone produced much higher yields (1% to 4%) of C8- and C9-dicarboxylic acids than did their
30      reaction with OH radicals (0.2% to 0.3%).  Apart from the complex products noted above, it
31      should be remembered that much simpler products, such as formaldehyde and formic acid, are

        June 2003                                3-68        DRAFT-DO NOT  QUOTE OR CITE

-------
 1      also formed in much larger yields from the same reactants (e.g., Winterhalter et al., 2000).
 2      Compounds such as these also contribute to the formation of secondary organic aerosol
 3      according to the mechanisms given in Pandis et al. (1992) and mentioned earlier in this section.
 4           It is worth noting that the dicarboxylic acids and hydroxy-keto-carboxylic acids have very
 5      low vapor pressures and may act as nucleating species in OH- and O3-terpene reactions (Larsen
 6      et al., 2001). The rate coefficient for reaction of cc-pinene with OH radicals is approximately a
 7      factor of 106 greater than for its reaction with O3, based on data given in Atkinson (1994).  The
 8      daytime average concentration of O3 is typically a factor of 106 greater than that for OH radicals
 9      in polluted boundary layers; whereas the above mentioned yields of aerosol products are roughly
10      a factor often greater in the O3-initiated reaction than in the corresponding OH radical reaction.
11      The foregoing analysis suggests that the O3-initiated reaction may be more important than the
12      OH-initiated reaction for the formation of aerosol products.  During the day, new particles may
13      be generated indoors through the infiltration of ambient ozone; and, because ambient ozone is
14      also present at night in lower concentrations, new particles may be generated under these
15      conditions at lower rates.  For example, Wainman et al. (2000) found that ozone can react with
16      limonene released by air fresheners in indoor environments to produce substantial quantities of
17      submicron particles. The corresponding reaction involving OH radicals outdoors at night is
18      expected to be negligible by comparison because of the very low OH concentrations present.
19      Sarwar et al. (2002) estimated indoor OH radical  concentrations and suggested that OH in indoor
20      environments  is produced mainly by reactions of ozone transported from outdoors  and terpenes
21      emitted from indoor sources.  They reported that indoor OH levels (1-5 x  10s OH/cm3) are
22      usually lower  than typical urban outdoor daytime OH levels (1-5 x 106 OH/cm3). However, they
23      can be greater than typical urban outdoor night time OH levels (1-5 x 104 OH/cm3). Although
24      much progress has been made in determining the importance of anthropogenic and biogenic
25      hydrocarbons  for the formation of secondary organic PM, further investigations are needed to
26      accurately assess their overall contributions to PM2 5 concentrations.
27           Reactions  of organic compounds either in particles or on the surface of particles have only
28      come under study during the past 20 years.  Tobias  and Ziemann (2000) reported evidence for
29      the formation  of relatively stable low volatility peroxy hemiacetals from reactions of
30      hydroperoxides  with aldehydes and ketones on the surface of secondary organic particles.
31      Not long after the publication of these results, Jang  and Kamens (200 la) suggested, based on

        June 2003                                 3-69         DRAFT-DO NOT QUOTE OR CITE

-------
 1      results of their outdoor Teflon chamber studies of SOPM formation from irradiation of
 2      toluene/propylene/NOx /air mixtures, that carbonyls and hydroxy compounds (either within or on
 3      the surface of aromatic SOPM) could react together to form larger and less volatile hemiacetals
 4      and acetals. They also proposed that dissolved carbonyls could undergo further reactions
 5      leading to the formation of a polymer, a mechanism that has also been suggested by Edney et al.
 6      (2001). Jang and Kamens (2001b) carried out a series of screening experiments to assess
 7      whether volatile carbonyl compounds absorbed into particles could undergo heterogeneous
 8      reactions forming low vapor pressure compounds.  Experiments were carried out in which
 9      aldehydes were introduced in Teflon bags in the dark in the presence of a seed aerosol containing
10      either ammonium sulfate or a mixture of ammonium sulfate and sulfuric acid.  The increase in
11      the aerosol volume was then measured using a scanning mobility particle sizer. The aldehydes
12      employed for the study included glyoxal, butanal, hexanal, octanal, and decanal.  Increased
13      organic aerosol yields were found in the presence of the ammonium sulfate seed aerosol for each
14      of the carbonyls; the highest yield was found for hexanal followed in decreasing order by
15      glyoxal and then octanal. The presence of the acidified sulfate salt significantly increased the
16      yields even further. In a number of other experiments,  1-decanol was added to the carbonyl-
17      aerosol system to investigate the possible formation of hemiacetals and/or acetals. Again, the
18      volume of aerosol increased in both  the presence of ammonium sulfate aerosol and the acidified
19      salt with a significantly larger yield found in the presence of acidity.
20           To explain their findings for acid-catalyzed carbonyl reactions, Jang and Kamens (2001a,b)
21      proposed a chemical mechanism in which the dissolved carbonyl first undergoes a protonization
22      reaction forming an adduct that can react with water to form its hydrate (1,1-dihydroxy gem-
23      diol). The adducts can then react with OH groups of the gem-diol forming higher molecular
24      weight and less volatile dimers that are subject to further reactions. In principal, this process,
25      which the authors refer to as a "zipping reaction" can lead to the formation of polymers.
26      However, because the individual reactions are reversible, the process can be reversed by an
27      unzipping reaction.  The zipping process could serve as an important mechanism for SOPM
28      formation by converting volatile oxidation products including glyoxal  and methyl glyoxal into
29      low volatility compounds.  On the other hand, the unzipping process that could take place during
30      the workup of the aerosol samples could be responsible for the detection of highly volatile
31      oxidation products in SOPM, including glyoxal and methyl glyoxal, that has been reported by

        June 2003                                 3-70        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Edney et al. (2001), Cocker et al. (2001), and Jang and Kamens (2001a). While these processes
 2      may take place in the absence of significant acidity, the experimental results suggest that the
 3      processes are likely enhanced by acid-catalyzed reactions. Further research is needed to
 4      determine the importance of the mechanisms proposed above for the ambient atmosphere.
 5           Sampling and characterizing PM in the ambient atmosphere and in important
 6      microenvironments is required to address important issues in exposure, toxicology, and
 7      compliance.  Currently, it is not possible to fully quantify the concentration, composition, or
 8      sources of the organic components. Many of the secondary organic aerosol components are
 9      highly oxidized, difficult to measure, multifunctional compounds.  Additional laboratory studies
10      are needed to identify such compounds; strategies need to be developed to sample and measure
11      such compounds in the atmosphere; and models of secondary organic aerosol formation need to
12      be improved and added to air quality models in order to address issues related to human
13      exposure.
14           A high degree of uncertainty is associated with all aspects of the calculation of secondary
15      organic PM concentrations. This is compounded by the volatilization of organic carbon from
16      filter substrates during and after sampling as well as potential positive artifact formation from
17      the absorption of gaseous hydrocarbon on quartz filters. Significant uncertainties  always arise in
18      the interpretation of smog chamber data because of wall reactions, sampling artifacts, and the
19      use of unrealistically high concentrations of reactants. Limitations also exist  in extrapolating the
20      results of smog chamber studies to  ambient conditions found in urban airsheds and forest
21      canopies. Concentrations  of terpenes and NOX are much lower in forest canopies (Altshuller,
22      1983) than the levels commonly used in smog chamber studies. The identification of aerosol
23      products of terpene oxidation has seldom been a specific aim of field studies, making it difficult
24      to judge the results of model calculations of secondary organic PM formation.
25           Uncertainties also arise because of the methods used to measure biogenic hydrocarbon
26      emissions. Khalil  and Rasmussen (1992) found much lower ratios of terpenes to other
27      hydrocarbons (e.g., isoprene) in forest air than were expected based on their relative emissions
28      strengths and rate coefficients for reaction with OH radicals and O3.  In many cases, reactions
29      with these species  are capable of reducing the concentrations of monoterpenes to beneath
30      minimum detection levels as has been found by others in a wide range of North American forest
31      ecosystems (Guenther et al., 1996, Helmig et al., 1998, Geron et al.,  2000). Thus, making

        June 2003                                 3-71        DRAFT-DO NOT QUOTE OR CITE

-------
 1     judgments about the importance of additional loss processes can be a highly problematic
 2     exercise given uncertainties in obtaining relevant OH radical and O3 concentrations and in the
 3     reaction rate coefficients. The ocimenes and sesquiterpenes are estimated to have half lives of
 4     seconds to minutes in the presence of ambient ozone levels while the pinenes and isoprene have
 5     lifetimes ranging from several hours to days. Reaction with hydroxyl radicals is the major sink
 6     for isoprene.  Khalil and Rasmussen offered two explanations for their findings:  (1) either the
 7     terpenes were being removed rapidly by some heterogeneous process or (2) emissions were
 8     enhanced artificially by feedbacks caused by the bag enclosures they used.  The isoprene
 9     emission rates observed by Khalil and Rasmussen (1992) are reasonably consistent with those
10     found by Geron et al. (2000a and 2001) given inevitable self-shading of much of the foliage
11     within the bag enclosures and also the lower position of the foliage within the accessible portion
12     of the sampled tree crowns. Temperatures were sufficiently warm (>30 °C) and sunny to induce
13     high isoprene emission although cloudiness from a passing thunderstorm could have reduced
14     these emissions from some of the species considerably. Disturbance or elevated temperatures
15     may also have induced elevated monoterpene emissions from several of the species sampled by
16     Khalil and Rasmussen because emissions rates they reported are in many cases 2-20 times higher
17     than those reported by other investigators, including those who performed enclosure studies
18     which did not cause elevated temperatures as well as micrometeorological flux studies which did
19     not disturb the forest canopy. Thus the somewhat reduced isoprene emissions, combined with
20     elevated monoterpene emissions, can indeed affect the comparison of ambient isoprene  versus
21     monoterpene emissions.  However, monoterpene compounds recently have been found to
22     undergo heterogenous reactions on the surface of acid aerosol particles.  Further work is needed
23     to assess the importance of these reactions on ambient monoterpene concentrations and  in the
24     rate of production of secondary organic PM in forest ecosystems.
25
26     3.3.2  Source Contributions to Ambient PM Determined by Receptor  Models
27           Receptor models are perhaps the primary means used to estimate the contributions of
28     different source categories to PM concentrations at individual monitoring sites.  Dispersion
29     models (i.e., three-dimensional chemistry and transport models) are formulated in a prognostic
30     manner (i.e., they attempt to predict species concentrations using a tendency equation that
31     includes terms based on emissions inventories, atmospheric transport, chemical transformations,

       June 2003                               3-72         DRAFT-DO NOT QUOTE  OR CITE

-------
 1      and deposition). Receptor models are diagnostic in their approach (i.e., they attempt to derive
 2      source contributions based either on ambient data alone or in combination with data from the
 3      chemical composition of sources). These methods have the advantage that they do not invoke all
 4      of the uncertainties inherent in emissions inventories or in parameterizing atmospheric transport
 5      processes in grid point models.
 6           There are two main approaches to receptor modeling. Receptor models such as the
 7      chemical mass balance (CMB) model (Watson et al., 1990a) relate source category contributions
 8      to ambient concentrations based on analyses of the composition of ambient particulate matter
 9      and source emissions samples. This technique has been developed for apportioning source
10      categories of primary parti culate matter and was not formulated to include the processes of
11      secondary particulate matter formation. In the second approach, various forms of factor analysis
12      are used that rely on the analysis of time series of compositional data from ambient samples to
13      derive both the composition of sources and the source contributions. Standard approaches such
14      as factor analysis or Principal Component Analysis (PCA) can apportion only the variance and
15      not the mass in an aerosol composition data set.  The other techniques described below, PMF and
16      UNMIX, do apportion mass, however. Positive matrix factorization (PMF) is a recently
17      developed multivariate technique (Paatero and Tapper, 1993; 1994) that overcomes many of the
18      limitations of standard techniques, such as PCA, by allowing for the treatment of missing data
19      and data near or below detection limits. This is accomplished by weighting elements inversely
20      according to their uncertainties.  Standard methods such as PCA weight elements equally
21      regardless of their uncertainty. Solutions also are constrained to yield non-negative factors.
22      Both the CMB and the PMF approaches find a solution based on least squares fitting and
23      minimize an object function. Both methods provide error estimates for the solutions based on
24      estimates of the errors in the input parameters. It should be remembered that the error estimates
25      often contain subjective judgments. For a complete  apportionment of mass, all of the major
26      sources affecting a monitoring site must be sampled for analysis by CMB; whereas there is no
27      such restriction in the use of PMF.
28           Among other approaches, the UNMIX model takes a geometric approach that exploits the
29      covariance of the ambient data to determine the number of sources, the composition and
30      contributions of the sources, and the uncertainties (Henry, 1997).  A simple example may help
31      illustrate the approach taken by UNMIX.  In a two-element scatter plot of ambient Al and Si, a

        June 2003                                 3-73         DRAFT-DO NOT QUOTE OR CITE

-------
 1      straight line and a high correlation for Al versus Si can indicate a single source for both species
 2      (soil) while the slope of the line gives information on the composition of the soil source. In the
 3      same data set, iron may not plot on a straight line against Si, indicating other sources of Fe in
 4      addition to soil.  More importantly, the Fe-Si scatter plot may reveal a lower edge. The points
 5      defining this edge represent ambient samples collected on days when the only  significant source
 6      of Fe was soil. Success of the UNMIX model hinges on the ability to find these "edges" in the
 7      ambient data from which the number of source types and the source compositions are extracted.
 8      UNMIX uses principal component analysis to find edges in m-dimensional space, where m is the
 9      number of ambient species. The problem of finding edges is more properly described as finding
10      hyperplanes that define a simplex.  The vertices at which the hyperplanes intersect represent pure
11      sources from which source compositions can be determined. However, there are measurement
12      errors in the ambient data that "fuzz" the edges making them difficult to find.  UNMIX employs
13      an "edge-finding" algorithm to find the best edges in the presence of error. UNMIX does not
14      make explicit use of errors or uncertainties in the ambient concentrations, unlike the methods
15      outlined above.  This is not to imply that the UNMIX approach regards data uncertainty as
16      unimportant, but rather that the UNMIX model results implicitly incorporate error in the ambient
17      data. The underlying philosophy is that the uncertainties are often unquantifiable,  and hence it is
18      best to make no  a priori assumptions about what they are.
19           In addition to chemical speciation data, Norris et al. (1999) showed that meteorological
20      indices could prove useful in identifying sources of particulate matter that are responsible for
21      observed health  effects (specifically asthma) associated with exposure to parti culate matter.
22      They examined meteorology associated with elevated pollution events in Spokane and Seattle,
23      WA, and identified a "stagnation index" that was associated with low wind speeds and increases
24      in concentrations of combustion-related pollutants.  Their factor analysis also identified a
25      meteorological index (low relative humidity and high temperatures) that was associated with
26      increases in soil-derived particulate matter as well as a third factor (low temperatures and high
27      relative humidity) that was associated with increases in concentrations of particulate sulfate and
28      nitrate species (Norris, 1998).
29           Ondov (1996) examined the feasibility of using sensitive isotopic and elemental tracer
30      materials to determine the contributions of petroleum-fueled sources of PM10 in the San Joaquin
31      Valley, CA.  Costs of these experiments are affected not only by the tracer materials cost, but

        June 2003                                 3-74         DRAFT-DO NOT QUOTE OR CITE

-------
 1      also by the sensitivities of the analytical methods for each, as well as the background levels of
 2      the tracers.  Suarez et al.  (1996) used iridium as a tracer to tag emissions from diesel-burning
 3      sanitation trucks in Baltimore and determined the size distribution of soot from the trucks.
 4           A number of specialty conference proceedings, review articles, and books have been
 5      published that provide greater detail about source category apportionment receptor models than
 6      described in the 1996 PM AQCD.  A review of the various methods used to apportion PM in
 7      ambient samples among its source categories was given in Section 5.5.2 of the  1996 PM AQCD.
 8      The collection of the source category characterization profiles shown in Appendix 3D has been
 9      motivated in many cases by the need to use them in receptor modeling applications.
10           The results of several source apportionment studies are discussed in this section to provide
11      an indication of the relative importance of different sources of particulate matter across the
12      United States. First, results obtained mainly by using the chemical mass balance (CMB)
13      approach for estimating contributions to PM25 from different source categories at monitoring
14      sites in the United States are discussed  and presented in Table 3-9.  More recent results using the
15      PMF approach are included for Phoenix, AZ.  Results obtained at a number of monitoring sites
16      in the central and western United States by using the CMB model for PM10 are shown in
17      Table 3-10. The sampling sites represent a variety of different source characteristics within
18      different regions of Arizona, California, Colorado, Idaho, Illinois, Nevada, and Ohio.
19      Definitions of source categories also vary from study to study. The results of the PM10 source
20      apportionment studies were given in the 1996 PM AQCD and are presented here to allow easy
21      comparison with results of PM25 source apportionment studies. Chow and Watson (2002)
22      present a detailed comparison of numerous studies using the CMB model performed mainly  after
23      1995.
24           There are several differences between the broadly defined source categories shown at the
25      tops of Tables 3-9 and 3-10. These differences reflect the nature of sources that are important
26      for producing fine and coarse particulate matter shown in Table 3-8.  They also are related to
27      improvements in the ability to distinguish between sources of similar nature (e.g., diesel and
28      gasoline vehicles,  meat cooking, and vegetation burning).  The use of organic tracers allows
29      motor vehicle emissions to be broken down into contributions from diesel and gasoline vehicles.
30      In studies where this distinction cannot be made, the source type is listed as 'total motor
31      vehicles' in the tables. The studies that were reported to be able to distinguish gasoline- from

        June 2003                                  3-75       DRAFT-DO NOT QUOTE OR CITE

-------
•— ^
3
ft>
to
OJ












OJ
a\




O
^
H
6
O
0
o

o
H
W
O
^
o
H
W




Sampling Site
Pasadena, CA 19821
Downtown LA, CA 19821
West LA, CA 19821
Rubidoux, CA 19821
Sacramento, CA
Winter 1991-962
Bakersfield, CA
Winter 19963
Fresno, CA Winter 19963
Philadelphia, PA Summer 19824
Camden, NJ Summer 19824
Clarksboro, NJ Summer 19824
Grover City, IL ENEj
1986-875
Grover City, IL SSW
1986-875
Grover City, IL WNWj 1986-875
Grover City, IL NNWj 1986-875

Reno, NV Summer 19986
Phoenix, AZ Summer 1995-987
Phoenix, AZ Winter 1995-987

'Schauer et al. (1996)
2Motallebi (1999)
3Magliano et al. (1998)
4Dzubayetal(1988)
5Gloveretal. (1991)
'Gillies et al. (2000)
'Ramadan et al. (2000)


TABLE 3-9. RECEPTOR MODEL i

Measured Total
PM2 5 Motor Gasoline Road Dust,
Concentration Vehicles Diesel Vehicles Soil
28.2 — 18.8 5.7 12.4
32.5 — 35.7 6.5 11.1
24.5 — 18.0 5.7 12.2
42.1 — 12.8 0.7 13.1
39.5 24.5' — — 1.2

52.0 16.0' — — <3

63.0 13.0' — — <3
27.0 8.5' — — 4.4
28.3 9.2' — — 3.2
26.0 5.8' — — 2.7
— — — 2.3

_ _ _ _

2.4' — 5.1
— — — 3.1

7.8 68.0' — — 14.5
8.3 — 10.9 36.2 1.8
13.8 — 14.5 38.9 1.1

"Secondary and other organic compounds
""Secondary ammonium
'Meat cooking
dVegetative detritus
'Value represents sum of diesel and gasoline
vehicle exhaust
'Including associated cations and water


SOURCE CONTRIBUTIONS TO
% Contribution

Vegetation Secondary Secondary Misc.
Burning Sulfate Nitrate Source 1
9.6 20.9 7.4 5.3"
5.8 20.3 9.2 3.7"
11.0 24.1 7.8 4.1"
1.2 13.8 24.7 4.5"
18.1 4.5 36.6 —

20.0 7.0 34.0 —

19.0 5.0 32.0 —
— 81. 9f — 2.28
— 81. 3f 0.4 2.5s
— 84.6f — 0.88
— 83.2f — 9.7k

— 59.0f — 11.6k

— 88.5f — 2.8k
— 86.6f — 3.41

4.0 11.0 2.0 0.6"
15.0 — — 20.8°
8.9 — — 9.5"

incinerators
hOil fly ash
'Fluidized catalyst cracker
jWind direction
kLead smelter
'Iron works
"Copper smelter


PM25


Misc. Misc. Misc.
Source 2 Source 3 Source 4
9.2b 8.5' l.ld
9.2b 5.2' 0.6d
9.4b 8.2' 1.6d
12. lb 4.5' 0.5d
	 	 	

_ _ _

— — —
1.9h 0.41 —
2.5h 0.71 —
1.5h 0.41 —
3.01 1.28 —

11.91 4. 18 4.6m

— — —
3.0" — —

— — —
4.9r 6.7s 3.6"
4.5r 18.7s 4.1"

"Coal power plant
"As ammonium sulfate
pAs ammonium nitrate
"Sea salt
"Wood burning
sNonferrous smelting






Total %
Allocated
98.9
107.3
102.1
87.9
84.9

<80

<85
99.3
99.8
95.8
99.4

91.2

98.8
96.0

100.1
99.9
100.2











-------
•— ^

to
O
OJ







OJ
^



O
H
6
o
0
H
O
O
H
W
O
O
HH
H
W
TABLE 3-10. RECEPTOR MODEL SOURCE CONTRIBUTIONS 1
0 PM10



% Contribution


Sampling Site
Craycroft, AZ
Winter 1989-19901
Hayden 1, AZ 19861
Hayden2,AZ 19861
Rillito, AZ 19882
Bakerfield, CA
1988-19893
Crows Landing, CA
1988-19893
Fellows, CA
1988-19893
Fresno, CA
1988-19893
Indio, CA4
Kern Wildlife
Refuge, CA
1988-19893
Long Beach, CA
19865
Long Beach, CA
Summer 19876
Long Beach, CA
Fall 19876
Riverside, CA 19887
Rubidoux, CA 19865
Rubidoux, CA
Summer 19876

Measured
PM10
Concentration
23.4
105.0
59.0
79.5
79.6
52.5
54.6
71.5
58.0

47.8

51.9
46.1
96.1
64.0
87.4
114.8

Primary Motor
Primary
Geological
55.6
4.8
35.6
53.7
53.9
61.3
53.1
44.5
56.9

31.6

39.9
24.1
11.8
50.9
49.3
30.4

Primary
Construction
0.0
1.9b
6.8b
17.4b
2.0
0.0
2.6
0.0
5.2

4.2

0.0
0.0
0.0
0.0
4.6
3.9

Vehicle
Exhaust
35.5
0.0
0.0
1.5f
9.7
4.2
3.8
9.5
7.6

4.6

9.81
13.7
44.5
10.9
6.41
15.1

Primary
Vegetative
Burning
0.0
0.0
0.0
0.0
8.2
6.5
6.2
7.1
12.2

8.4

0.0
0.0
0.0
0.0
0.0
0.0

Secondary
Ammonium
Sulfate
3.0
3.8
6.8
0.0
6.9
5.3
9.3
5.0
6.2

6.9

15.4
23.6
4.0
7.5
7.3
8.3

Secondary
Ammonium
Nitrate
2.6
0.0
0.0
0.0
16.0
12.4
13.7
14.5
7.1

3.1

17.7
1.7
24.1
33.4
24.4
23.9


Misc.
Source 1
5.T
70. 5C
47. 5C
14.6s
1.3m
1.0m
12. 8m
0.4m
0.3j

1.0m

0.2j
0.2j
0.0>
0.5j
0.3j
0.0>


Misc.
Source 2
0.0
4.8d
0.0
0.0
1.9"
1.9"
2.6"
1.9"
1.7"

3.1"

3.9h
4.8h
2.8h
2.0h
1.1"
4.4h


Misc.
Source 3
0.0
1.0e
1.7e
0.0
0.8k
2.3k
2.6k
O.lk
0.0

1.5k

12.3k
0.0
0.0
1.7°
6.8k
0.0


Misc.
Source 4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0

0.0

0.0
0.0
0.0
0.0
0.0
0.0


Total %
Allocated
101.8
86.6
98.4
87.2
100.7
94.9
106.7
83
97.2

64.4

63.2
68.1
87.2
106.9
100.2
86


-------
•— ^

to
O
OJ






OJ
oo


DRAFT-DO NOT QUOTE
O
O
H
W

Sampling Site
Rubidoux, CA
Fall 19876
Rubidoux, CA 19887
San Nicolas Island, CA
Summer 19876
Stockton, CA 19893
Pocatello, ID 1990s
S.Chicago, IL 19869
S.E. Chicago, IL
198810
Reno,NV 1986-87"
Sparks, NV 1986-87"
Follansbee, WV 199 112
Mingo, OH 199112
Steubenville, OH
199112
TABLE 3-10 (cont'd). RECEPr

Measured
PM10 Primary Primary
Concentration Geological Construction
112.0 17.1 14.4
87.0 55.2 0.0
17.4 9.2 0.0
62.4 55.1 0.8
100.0 8.3 7.5"
80.1 34.0 3.0
41.0 35.9V 0.0
30.0 49.7 0.0
41.0 36.8 0.0
66.0 15.2 0.0
60.0 20.0 0.0
46.0 18.0 0.0
'Chow et al. (1992a) 10Houck et al. (1992)
2Garfield; Ryan et al. (1988) "Hopke et al. (1988)
3Jail; Ryan et al. (1988) 12Vermette et al. (1992)
"Thanukos et al. (1992) "Chow et al. (1988)
5Chow et al. (1992b) "Skidmore et al. (1992)
6Kim et al. (1992) "Smelter background aerosol
'Gray et al. (1988) bCement plant sources, including
8Watson et al. (1994) kiln stacks, gypsum pile, and
'Chowetal. (1992c) kiln area






FOR MODEL SOUS

Primary Motor Primary
Vehicle Vegetative
Exhaust Burning
27.1 0.0
11.7 0.0
5.2 0.0
8.3 7.7
0.1 0.0
3.5 0.0
2.2f 0.0
33.3 6.3
28.3 32.7
53.0 0.0
23.3 6.8
30.4 1.7
LCE CONTRIBUTIONS T(
% Contribution
Secondary Secondary
Ammonium Ammonium Misc.
Sulfate Nitrate Source 1
1.9 28.2 Off
6.1 24.9 0.&
21.3 2.9 0.0s
5.0 11.2 l.lm
0.0 0.0 0.0
19.2s — 18.9'
18.8 — 2.0'
4.3 2.0 0.0
6.6 2.2 0.0
24.2 — 14.1'
25.0 — 5.7'
30.4 — 8.3'
'Copper ore 'Motor vehicle exhaust from
''Copper tailings diesel and leaded gasoline
'Copper smelter building ^Residual oil combustion
'Heavy-duty diesel exhaust kSecondary organic carbon
emission 'Biomass burning
background aerosol "Primary crude oil
hMarine aerosol, road salt, and "NaCl + NaNO3
sea salt plus sodium nitrate "Lime
pRoad sanding material






3PM10



Misc. Misc. Misc. Total %
Source 2 Source 3 Source 4 Allocated
1.0" 0.0 0.0
1.7h 6.6° 0.0
24.7h 0.0 0.0
2.9" 0.0k 0.0
0.0 84. lr 0.0
2.7" 0.0 0.0
0.7h 2.7" 18.8s
0.0 0.0 0.0
0.0 0.5k 0.0
0.0 0.0 0.0
18.3X 0.0 0.0
10.9X 0.0 0.0
89.7
106.8
63.3
92.1
100
81.3
81.1
95.6
107.1
106.5
99.1
99.7
'Asphalt industry
"Phosphorus/phosphate industry
sRegional sulfate
'Steel mills
"Refuse incinerator
"Local road dust, coal yard road
dust, and steel haul road dust
"Incineration
"Unexplained mass







-------
 1      diesel-fueled vehicles found that gasoline vehicles make significant, and sometimes the
 2      dominant, contributions to ambient PM2 5 concentrations. Meat cooking is also distinguished
 3      from vegetation burning in more recent studies although both are considered to be part of
 4      biomass burning. Vegetation burning consists of contributions from residential fuel wood
 5      burning, wildfires, prescribed burning, and burning of agricultural and other biomass waste.
 6      Miscellaneous sources of fine particles include contributions from combustion sources; whereas
 7      miscellaneous sources of coarse particles consist of contributions from soil and sea spray and
 8      industrial processing of geological material (e.g., cement manufacturing).  Although a large
 9      number of elements and chemical components are used to differentiate among source categories
10      and although there can be a large number of source types affecting a given site, only a few
11      broadly defined source types are needed to account for most of the mass of PM25 and PM10.
12      At any given site, < 5 source types account for > 65% of the mass of PM25 (Table 3-9); and
13      <5 source types account for > 65% of the mass of PM10 (Table 3-10).
14           Secondary sulfate is the dominant component of PM25 samples collected in the studies of
15      Dzubay et al. (1988) and Glover et al. (1991).  Both studies found that sulfate at their monitoring
16      site arose from regionally dispersed sources.  Sulfate also represents the major component of
17      PM25 found in monitoring studies in the eastern United States shown in Appendix 6 A of the
18      1996 PM AQCD. Primary and secondary organic components  also make major contributions to
19      PM25.  Contributions from road dust and soils are relatively minor, typically constituting less
20      than 10% of PM25 in the studies shown in Table 3-9. Studies in the western United States shown
21      in Table 3-9 have found larger contributions from motor vehicles, fugitive dust, and ammonium
22      nitrate. The most notable difference in the relative importance  of major source categories of
23      PM2 5 shown in Table 3-9 and PM10 shown in Table 3-10 involves crustal material, (e.g., soil,
24      road dust), which represents about 40% on average of the total mass of PM10 in the studies
25      shown in Table 3-10. The fraction is higher at sites located away from specific sources such as
26      sea spray or smelters. Emissions of crustal material are concentrated mainly in the PM10_25 size
27      range.
28          In Table 3-10, primary motor vehicle exhaust contributions are highly variable and can
29      account for over 40% of average PM10 at a few of the sampling sites.  Vehicle exhaust
30      contributions are also variable at different sites within the same study area. The mean value and
31      the variability of motor vehicle exhaust contributions reflects the proximity of sampling sites to

        June 2003                                 3-79        DRAFT-DO NOT QUOTE OR CITE

-------
 1      roadways and traffic conditions during the time of sampling. Many studies were conducted
 2      during the late 1980s when a portion of the vehicle fleet still used leaded gasoline.  Pb and Br in
 3      motor vehicle emissions facilitated the distinction of motor vehicle contributions from other
 4      sources. Vehicles using leaded fuels have higher emission rates than vehicles using unleaded
 5      fuels. Pb also poisons automobile exhaust catalysts and produces adverse human health effects.
 6      As a result, Pb has been eliminated from vehicle fuels.  Organic species such as n-pentacosane
 7      through n-nonacosene, cholestanes, ergostanes, sitostanes, and hopanes have replaced Pb as a
 8      source marker for motor vehicle emissions (e.g., Schauer and Cass, 2000).  In their
 9      comprehensive review of CMB modeling studies undertaken since 1995, Chow and Watson
10      (2002) note that in twenty-two studies fossil fuel combustion was found to be a large contributor
11      to PM25 and PM10 concentrations, with most of the contributions to primary PM originating from
12      the exhaust of diesel and gasoline vehicles.
13           Marine aerosol is found, as expected, at coastal sites such as Long Beach (average 3.8% of
14      total mass) and San Nicolas Island (25%). These contributions to PM10 are relatively variable
15      and are larger at the more remote sites. Individual values reflect proximity to local sources.
16      Of great importance are the contributions from secondary ammonium sulfate in the eastern
17      United  States and ammonium nitrate in the western United States. Secondary ammonium sulfate
18      is especially noticeable at sites in California's San Joaquin Valley (Bakersfield, Crows Landing,
19      Fellows, Fresno, and Stockton) and in the Los Angeles area.
20           Because many source apportionment studies address problems  in compliance with the
21      National Ambient Air Quality Standards and other air quality standards, samples selected for
22      chemical analysis are often biased toward the highest PM10 mass concentrations in the studies
23      shown in Table 3-10. Thus, the average  source contribution estimates shown in Table 3-10 are
24      probably not representative of annual averages and may not be representative of a large spatial
25      area for some source-dominated studies.  For example, the study by Motallebi (1999) considered
26      only days when the PM10 concentration was greater than 40 |ig/m3. Quoted uncertainties in the
27      estimated contributions of the individual sources shown in Tables 3-9 and 3-10 range from 10 to
28      50%. Errors can be much higher when the chemical source profiles for different sources are
29      highly uncertain or are too similar to distinguish one source from another.
30           Very few source apportionment studies using the CMB modeling technique have examined
31      the spatial variability of source  contributions at different sites within an urban area.  As can be

        June 2003                                  3-80        DRAFT-DO NOT QUOTE OR CITE

-------
 1      seen from Table 3-9, Dzubay et al. (1988) found a uniform distribution of sulfate at the NE
 2      Airport in Philadelphia, PA; downtown Camden, NJ; and Clarksboro, NJ, during the summer of
 3      1982.  The farthest distance between two monitoring sites (NE Airport and Clarksboro) was
 4      approximately 40 km. Magliano et al.  (1998) examined the spatial variability of PM10 source
 5      contributions at a number of sites in Fresno and Bakersfield, CA, during the winter of 1995-1996
 6      and reported values for 1 day, December 27, 1995. During that day, mobile sources contributed
 7      from 13.0 to 15.8 |ig/m3, vegetation burning contributed from 5.1 to 11.1 |ig/m3,  ammonium
 8      sulfate contributed 2.4 to 3.4 |ig/m3, and ammonium nitrate contributed 19.3 to 24.6 |ig/m3 to
 9      PM10 at the sites in Bakersfield. Mobile sources contributed 13.9 to 22.5 |ig/m3, vegetation
10      burning contributed 8.2 to 15.7 |ig/m3,  ammonium sulfate contributed  1.8 to 2.3  |ig/m3, and
11      ammonium nitrate contributed 14.5 to  18.9 |ig/m3 at the sites in Fresno.  All of these components
12      are expected to be found mainly in the  PM2 5 size fraction. As can be seen, source  contributions
13      at different sites varied by factors of 1.2 to 2.2 in Bakersfield and by factors of 1.3  to 1.9 in
14      Fresno on that day.
15          The receptor modeling methods outlined above do not explicitly include consideration of
16      the distances between PM sources and  the receptor site. Information about the relative
17      importance of sources as a function of  distance may be available from examination of data
18      obtained by continuous monitoring methods.  For example, concentration spikes are expected to
19      be the  result of transport from nearby sources because turbulent mixing in the atmosphere would
20      not allow them to persist for very long. Short duration spikes in the time series of  concentrations
21      are assumed to result from emissions from local sources (0.1 to 1 km away) in this method.
22      Contributions from sources located further away are determined by comparisons between
23      baselines measured at different sites. Details such as these are also lost in integrated 24-h
24      samples. Watson and Chow (2001) used time series of black carbon (BC) obtained by
25      aetholometers over five minute intervals to estimate the contributions from sources located
26      < 1 km away, 1 to 5 km away, and > 5  km away from a monitoring site in downtown Mexico
27      City. They found that most of the BC was produced by sources scattered throughout the city and
28      that sources located less than 1 km away from the site contributed only about 10%  to BC
29      concentrations even in the presence of  local sources such as buses and trucks.
30
31

        June 2003                                 3-81        DRAFT-DO NOT QUOTE OR CITE

-------
 1      3.3.3   Background Concentrations of PM in the United States
 2           This section contains information about the concentrations of "background" PM that are
 3      relevant for policy setting. For the purposes of this document (and consistent with the 1996
 4      PM AQCD), background concentrations are those that would result in the United States from
 5      emissions from natural sources worldwide plus anthropogenic sources outside of North America.
 6      In other words, they are the concentrations that would exist if anthropogenic emissions in North
 7      America were zero. Concentrations defined this way will be referred to here as policy relevant
 8      background (PRB) concentrations.  It is impossible to obtain estimates of PRB concentrations
 9      solely by examining measurements of PM25, PM10_25, or PM10 concentrations because of long-
10      range transport from anthropogenic source regions within North America.  Other additional
11      information must be used.
12           Annual average natural background concentrations of PM10 have been estimated to range
13      from 4 to 8 |ig/m3 in the western United States and 5 to 11 |ig/m3 in the eastern United States.
14      Corresponding PM2 5 levels have been estimated to range from 1 to 4 |ig/m3 in the western
15      United  States and from 2  to 5  |ig/m3 in the eastern United States; PM10_25 levels have been very
16      roughly estimated at 3 |ig/m3 in both the East and the West, with a range of 0 to 9 |ig/m3 in the
17      East and 0 to 7 |ig/m3 in the West (U.S. Environmental Protection Agency, 1996).  The estimated
18      natural  background concentrations given above do not include contributions from long-range
19      transport from sources outside North America. Values in the lowest 5th percentile annual mean
20      PM2 5 concentrations for specific sites in the AIRS data base range from 2.8 |ig/m3 to 6.9 |ig/m3.
21      This range is consistent with the range of annual mean PM2 5 concentrations at IMPROVE
22      network sites in the western United States (cf, Appendix E). However, PM2 5 concentrations are
23      much higher at sites in the eastern United States than at sites in the western United States.
24      At most IMPROVE sites  in the western United States, the annual mean concentration of PM10_25
25      is higher than that of PM2 5, and daily average PM25 concentrations are moderately correlated
26      (r = 0.72) with PM10_2 5 concentrations. In contrast, PM2 5 concentrations are higher than those of
27      PM10_2 5 at IMPROVE sites in  the eastern United States, and PM2 5 concentrations are only
28      weakly correlated (r = 0.26) with those of PM10_25.  Peak 24-h average natural background
29      concentrations may be substantially higher than the annual or seasonal average natural
30      background concentrations, especially within areas affected by wildfires and dust storms.
        June 2003                                 3-82        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Sources of PM that fit under the definition of PRB are located either within or outside of North
 2      America.
 3
 4      Long-Range Transport from Outside North America
 5           Windblown dust from dust storms in the Sahara desert has been observed in satellite
 6      images as plumes crossing the Atlantic Ocean and reaching the  southeast coast of the United
 7      States (e.g., Ott et al., 1991).  Dust transport from the deserts of Asia across the Pacific Ocean
 8      also occurs (Prospero, 1996). Most dust storms in the deserts of China occur in the spring
 9      following the passage of strong cold fronts after the snow has melted and before a surface
10      vegetation cover has been established.  Strong winds and unstable conditions result in the rapid
11      transport of dust to altitudes of several kilometers where it is transported by strong westerly
12      winds out over the Pacific Ocean (Duce, 1995). Satellite images were used to track the progress
13      of a dust cloud from the Gobi desert to the northwestern United States during the spring of 1998
14      (Husar et al., 2000).
15           Satellite images obtained at visible wavelengths cannot track mineral dust across the
16      continents because  of a lack of contrast between the plume and  the underlying surface. Other
17      means must be used to track the spread of North African dust through the eastern United States.
18      Perry et al. (1997) used two criteria (PM25 soil concentration >  3 jig m"3 and Al/Ca  > 3.8) to
19      distinguish between soil of local origin from soil  originating in North Africa in characterizing
20      the sources of PM in aerosol samples collected in the IMPROVE network.  North African dust
21      has been tracked as far north as Illinois (Gatz and Prospero,  1996) and Maine (Perry et al.,
22      1997).  The analysis of Perry et al. (1997) indicates that incursions of Saharan dust  into the
23      continental United States have occurred, on average, about three times per year from 1992 to
24      1995. These events persist for about 10 days on average, principally during the summer. Large
25      scale dust events typically cover from 15 to 30%  of the area of the continental United States and
26      result in increases of PM2 5 levels of 8.7 ± 2.3 jig  m"3 throughout the affected areas during these
27      events, with mean maximum dust contributions of 19.7 ± 8.4 jig m"3 and a peak contribution of
28      32 |ig m'3 to 24-h average PM2 5 levels.
29           As can be expected, the frequency of dust events is highest in the southeastern United
30      States.  About half of these events are observed only within the  state of Florida and are
31      associated with dense hazes in Miami (Figure 3-22) during the summer (Prospero et al.,  1987).

        June 2003                                  3-83       DRAFT-DO NOT QUOTE OR CITE

-------
                 74   76    78
                             80   82   84    86
                                            Year
        90   92    94    96
      Figure 3-22.  Monthly average Saharan dust components of the aerosol sampled in Miami,
                   FL, from 1974 to 1996. Approximately one-third to one-half of Saharan dust
                   is in the PM2 5 size range.
      Source: Prospero (1999).
1
2
3
4
5
North African dust is the dominant aerosol constituent in southern Florida during the summer;
whereas soil dust constitutes only a minor fraction of PM during the remainder of the year
(Prospero, 1999).  Approximately one-third to one-half of the mass of the particles reaching
southern Florida have aerodynamic diameters less than 2.5 micrometers (Prospero et al., 2001).
During episodes when daily total dust concentrations ranged up to 100 |ig/m3, it can be seen that
daily PM25 concentrations of up to 50 |ig/m3 could have resulted in Miami, FL.
     Husar et al. (2000) documented the transport of dust from the Gobi and Taklimakan deserts
to North America during April 1998. The PM10 concentration averaged over 150 stations in
Washington, Oregon, California, Nevada, and Idaho reporting data to AIRS was 65 |ig/m3
      June 2003
                                         3-84
DRAFT-DO NOT QUOTE OR CITE

-------
1
2
3
4
5
6
7
between April 26 and May 1, compared to about 20 |ig/m3 during the rest of April and May.
Data from several networks indicated that PM10 concentrations were over 100 |ig/m3 in central
British Columbia, Washington State, and Oregon. The highest PM concentrations observed
were 120 |ig/m3 for PM10 and 50 |ig/m3 for PM2 5 at Chilliwack Airport in northwestern
Washington State (Figure 3-23). Aircraft measurements made over the northwestern United
States were consistent with a mass median diameter of the dust being between 2 and 3 jim.
                 140
                 120-

              J100-
               o  80 H
               8  60 H
               c
               o
              o
              ^40-
                  20-
                    0
                   4/25/98  4/27/98   4/29/98    5/1/98    5/3/98    5/5/98
      Figure 3-23.  PM2 5 and PM10 concentrations measured at Chilliwack Airport, located in
                   northwestern Washington State, just before and during the Asian desert dust
                   episode of April and May 1998.
      Source: Aerometric Information Retrieval System (AIRS; U.S. Environmental Protection Agency, 2002b).
      June 2003
                                        3-85
DRAFT-DO NOT QUOTE OR CITE

-------
 1           Desert dust deposited over oceans provides nutrients to marine ecosystems (Savoie and
 2      Prospero, 1980).  Desert dust deposited on nutrient depleted soils also provides nutrients, as in
 3      Hawaiian rain forests (Chadwick et al., 1999). Microorganisms, including fungi and bacteria,
 4      have been found attached to African dust particles in the U.S. Virgin Islands (Griffin et al.,
 5      2001).  The fungus, Aspergillus sydowii, which has been connected to the death of coral reefs,
 6      has been identified in air samples collected in the Caribbean during African dust transport events
 7      (Smith et al., 1996; Shinn et al., 2000). Measurements of the composition of Saharan dust in
 8      Miami  indicate enhancements of nitrate, non-sea-salt sulfate, ammonium, and trace metals over
 9      concentrations expected for clean marine air, suggesting pollution emitted in Europe and North
10      Africa as sources (Prospero, 1999). It is likely that many other constituents will be found
11      associated with dust from outside North America as more measurements are made. It should be
12      noted that, as North African dust and associated material are transported northward through the
13      United States during the summer, they are  added to the mixture  of primary and secondary PM
14      generated domestically.
15           The transport of PM from uncontrolled biomass burning in Central America and southern
16      Mexico resulted in anomalously high PM levels observed in southern Texas and generally
17      elevated PM concentrations throughout the entire central and southeastern United States during
18      the spring and early summer of 1998.  Biomass burning for agricultural purposes occurs
19      normally during the spring  of each year in  Central America and southern Mexico. During the
20      spring of 1998, fires burned uncontrollably because of abnormally hot and dry conditions
21      associated with the intense  El Nino of 1997 to 1998.  PM10 concentrations observed in the
22      southern Rio Grande Valley were elevated substantially during the passage northward of the
23      biomass burning plume produced by these  fires as shown in Figure 3-24. Elevated PM10
24      concentrations also were found as far north as St. Louis, MO (Figure 3-25). As can be seen from
25      Figure  3-24 and Figure 3-25, the elevations in PM concentrations were limited  in duration.
26      Uncontrolled wildfires occur in the United States every year, but their effects on air quality
27      throughout the United States still need to be evaluated systematically.  These fires can be
28      widespread, and the frequency of their occurrence can vary markedly from year to year.
29      For example, approximately 26,000 km2 of forested land were consumed during 2000, but only a
30      small fraction of this area was burnt during 2001 in the western  United States.   Wildfires also
31      occur throughout the boreal forests of Canada.  Wotawa and Trainer (2000) suggested that the

        June 2003                                3-86        DRAFT-DO NOT QUOTE OR CITE

-------
                600
                  5/5 5/6 5/7 5/8 5/9 5/10 5/11 5/12 5/13 5/14 5/15 5/16 5/17 5/18 5/19 5/20 5/21 5/22 5/23 5/24 5/25 5/26 5/27 5/28
                                                  Time (days)
       Figure 3-24.  Time series of 24-h average PM10 concentrations observed in the Rio Grande
                     Valley during May 1998.
       Source: Aerometric Information Retrieval System (AIRS; U.S. Environmental Protection Agency, 2002b).
 1     plume from fires occurring in the Northwest Territories of Canada in early July 1995 may have
 2     extended throughout most of the eastern United States, resulting in elevated levels of CO and
 3     ozone.  Simple scaling of their calculated excess CO concentrations because of the fires, by the
 4     ratio of emission factors of PM2 5 to CO, indicates that the excess PM2 5 concentrations in the
 5     plume may have ranged from about 5 |ig/m3 in the Southeast and to close to 100 |ig/m3 in the
 6     northern Plains States.
 7
 8     Sources Within North America
 9           It is much more difficult to determine 24-h PRB concentrations in the absence of specific
10     events such as those given above because contributions from anthropogenic sources located
11     either nearby or elsewhere within North America can contribute substantially to observed values
12     and perhaps overwhelm the contributions from PRB sources. Source-apportionment modeling
13     techniques (described in Section 5.5 of the 1996 PM AQCD and updated in Section 3.3.2 of this
14     document) can be helpful for this purpose. Western sites are most useful because contributions
15     from pollution sources can overwhelm those from background sources in the East, and
       June 2003
3-87
DRAFT-DO NOT QUOTE OR CITE

-------
                          Q_
200
180-
160-
140-
120-
100-
 80-
 60-
 40-
 20-
   0
                                                                 Smoke
                                                                 Event
                                      PMin 24 hr Standard
                                        '10
                                        PM25 24 hr
                                    5  6  7  8  9 10  11 121314151617
                                                   May 1998
        Figure 3-25. PM10 concentrations observed in St. Louis, MO, during May 1998.
        Source: Aerometric Information Retrieval System (AIRS; U.S. Environmental Protection Agency, 2002b).
 1     background source contributions can be lost within the errors of their source contributions.
 2     However, it should be noted that source apportionment techniques such as PMF are not able to
 3     distinguish between bioaerosol and organic PM from whatever source, mainly because of
 4     analytical limitations. In addition, background PM source contributions are contaminated by
 5     contributions from pollution sources during transport from source to receptor monitoring site.
 6     Thus, background source contributions derived by these methods should be regarded as upper
 7     limits on the true values; however, the daily time series can provide more stringent limits about
 8     what the day to day variability in these contributions might be than by examining ambient PM2 5
 9     data alone.
10          Song et al. (2001) derived source contributions to PM25 concentrations measured at
11     Washington, DC, Brigantine, NJ, and Underbill, VT, using PMF. They found that wildfires
       June 2003
                            DRAFT-DO NOT QUOTE OR CITE

-------
 1      could contribute 0.93 ± 0.97 |ig/m3; soil could contribute 0.11 ± 0.22 |ig/m3; and sea spray could
 2      contribute 0.90 ±1.0 |ig/m3 on an annual basis at Brigantine, NJ throughout the sampling period
 3      from 1989 to 1999. They also derived contributions of 1.2 ± 0.9 |ig/m3 from wildfires; 0.32 ±
 4      0.61 |ig/m3 from soil; and 0.05 ± 0.05 |ig/m3 from sea spray at Underhill, VT from 1989 through
 5      1999. The "background" sources contribute about 7% to annual average PM25 concentrations at
 6      Brigantine and about 12% at Underhill. The daily time series at the NJ and VT sites show
 7      striking variability in background components, characterized by spikes.  Maximum daily values
 8      during these spikes are in the range of several  |ig/m3. Song et al. concluded these spikes in
 9      concentrations are likely caused by transient meteorological events such as storms or transport of
10      dust from remote regions such as northern Africa and by events such as wild fires. Contributions
11      from all of these sources should be regarded as upper limits because of entrainment of pollutant
12      emissions during transport from source to receptor of the background source emissions.
13           Pun et al. (2002) used a regional scale chemistry-transport model (CMAQ) to simulate O3
14      and PM2 5 concentrations arising from natural sources alone in model domains centered over
15      Tennessee and over the mid-Atlantic region for several days in July 1995. These calculations
16      were performed for meteorological conditions that resulted in high ambient ozone concentrations
17      in the eastern United States.  They found that natural sources contributed about 1.7 |ig/m3 to
18      Washington, DC, ( ranging from about 0.6 to 3.1 |ig/m3 in the mid-Atlantic domain) and about
19      1.2 |ag/m3 to Nashville, TN, on a 24 h average basis.  The formation of secondary organic PM
20      from biogenic precursors may be expected to be maximized for these conditions; however, their
21      contribution was  estimated to be at most 15% of natural PM25 or less than 1 |ig/m3. The largest
22      contributions in both cases came from natural PM2 5 that was advected in from other regions of
23      the United States. In addition to the sources considered above, contributions to both primary and
24      secondary PM from events such as volcanic eruptions or geothermal activity are highly sporadic.
25      The spatial and temporal distributions of secondary PM produced by background sources shown
26      in Table 3-8 still remain to be investigated.
27           It can be concluded from the above that 24-h policy relevant background concentrations
28      are highly variable both spatially and temporally. Contributions to PRB concentrations from
29      external sources (e.g., Asian and North African dust storms and Central American wildfires) can
30      be significant  on an episodic, but probably not on an  annual basis.  More local sources of
31      primary PRB PM are also likely to be episodic, reflecting the occurrence of volcanic eruptions,

        June 2003                                 3-89        DRAFT-DO NOT QUOTE OR  CITE

-------
 1      wildfires, and storms that raise dust and sea spray.  The influence from events such as these can
 2      be felt over thousands of square kilometers. Very little work has been done to quantify the
 3      magnitude and variability of contributions from the production of secondary PM. However, the
 4      one modeling study cited above found values < 1 |ig/m3 as secondary PM. Perhaps the greatest
 5      possibility for estimating these concentrations comes from the application of source
 6      apportionment techniques such as PMF (positive matrix factorization) to time series of species
 7      compositional data obtained at relatively remote monitoring sites (RRMS) to minimize
 8      interference from anthropogenic sources. In the absence of such results, some useful estimates
 9      may be obtained by examining the time series of PM concentrations at RRMS with screening to
10      eliminate days when concentrations are influenced by anthropogenic sources.
11           It is instructive in this regard to examine the variability observed at RRMS. Data from
12      eighteen IMPROVE sites are examined in Appendix 3E and analyzed in greater detail in Lefohn
13      et al. (2004). The ranges of annual average PM2 5, PM10_2 5, and PM10 concentrations are shown
14      in Table 3-E1 and the corresponding ranges for the 90th percentile concentrations are
15      summarized in Table 3-E2.  Information about the range of 24-h average PM25 and PM10_25
16      concentrations observed at selected sites in the IMPROVE network on a calendar quarter basis
17      are shown in Figures 3-E2a-d and Figures 3-E3a-d.
18           As mentioned earlier,  it is impossible to obtain estimates of PRB concentrations solely on
19      the basis of measurements of PM25, PM10_25, or PM10.  It is preferable to quantify contributions
20      from both background and non-background sources by using compositional data in techniques
21      such as source apportionment modeling.  However, of those measured throughout the United
22      States, the concentrations observed at several RRMS in the western United States probably come
23      closest to what PRB concentrations might be in the West. It must be recognized at the outset
24      that these concentrations will only provide  upper limits. However, at some  sites in the western
25      United States (e.g., Bridger, WY, and Yellowstone Park, WY) annual mean concentrations are
26      within the range of PRB values estimated in 1996 PM AQCD and given earlier in this section.
27      At other RRMS in the western United States, they are consistent with, although slightly above,
28      the annual average values defined earlier in the 1996 PM AQCD. Some screening should be
29      performed to rule out transport from urban  areas or other sources of anthropogenic PM on a day
30      by day basis.  Such  a procedure is impractical in the eastern United States as concentrations there
31      are heavily affected by anthropogenic emissions. Mean concentrations observed there are often

        June 2003                                3-90        DRAFT-DO NOT QUOTE OR CITE

-------
 1      several times those defined in the 1996 PM AQCD. It is important to note that there is much
 2      smaller relative variability in PM2 5 concentrations (from the minimum to the P90 level) on a
 3      year-to-year basis at the western IMPROVE sites than at the eastern IMPROVE sites (cf, Figure
 4      3E-4a-d). This may be related to a lack of discrete transport events affecting most samples at the
 5      western sites compared to the eastern sites and does not rule out the possibility that the western
 6      sites are receiving highly diluted contributions from many distant urban sources.  Further inquiry
 7      is needed to address these issues.  Likewise, data for PM10_2 5 concentrations at RRMS can be
 8      used for similar purposes.
 9           The above discussions do not explicitly consider the contributions from primary biological
10      particles (PBP). As mentioned earlier, one extensive study  conducted in Mainz, Germany,
11      (Mathias-Maser, 1998) found that PBP could contribute from about 17 to 20% of total aerosol
12      volume and from 9 to 30% of total particle number in the size range from 0.35 jim to 50 jim,
13      depending on season.  Pollen can at times represent the dominant contributors to PM in particles
14      larger than 10 jim.  These pollen grains can then fracture and appear as fragments in the smaller
15      size ranges.
16
17      3.3.4   Emissions Estimates for Primary Particulate Matter, and Precursors
18              to Secondary Particulate Matter (SO2, NOX, VOCs, and NH3) in the
19              United States
20           In principle, source contributions to ambient PM also  could be estimated on the basis of
21      predictions made by chemistry-transport models (CTM) or even on the basis of emissions
22      inventories alone. Uncertainties in emissions inventories have arguably been regarded as
23      representing the largest source of uncertainty in CTMs (Calvert et al., 1993).  Apart from
24      uncertainties in emission inventories, a number of other factors limit the ability of an emissions
25      inventory-driven CTM to determine the effects of various sources on particle samples obtained
26      at a particular location. CTM predictions represent averages over the area of a grid cell, which
27      in the case of CMAQ (Community Model for Air Quality) and MAQSIP (Multiscale Air Quality
28      Simulation Platform), ranges from 16 km2 (4 km x 4 km) to 1296 km2 (36 km x 36 km). CMAQ
29      and MAQSIP constitute the CTMs within the overall ModelsS framework, which also includes
30      emissions processors, the meteorological model, and modules for decision support.  The
31      contributions of sources to pollutant concentrations at a monitoring site are controlled strongly
32      by local conditions that cannot be resolved by an Eulerian grid-cell model. Examples would be

        June 2003                                 3-91         DRAFT-DO NOT QUOTE OR CITE

-------
 1     the downward mixing of tall-stack emissions and deviations from the mean flow caused by
 2     buildings.  The effect of local sources at a particular point in the model domain may not be
 3     predicted accurately because their emissions would be smeared over the area of a grid cell or if
 4     the local wind fields at the sampling point deviated significantly from the mean wind fields
 5     calculated by the model. CTMs also have problems in predicting pollutant concentrations
 6     because of uncertainties in vertical mixing and in predicting concentrations of pollutants from
 7     stationary combustion sources resulting from uncertainties in estimates of plume rise.  CTMs are
 8     an integral part of air quality management programs and are reviewed in the NARSTO Fine
 9     Particle Assessment (NARSTO, 2002).
10           Estimated emissions of primary PM25 from different sources in the United States are
11     summarized in Table 3-11, and estimated emissions of precursors to the formation of secondary
12     PM2 5 (SO2, NOX, VOCs, and NH3) are summarized in Table 3-12. These estimates are given to
13     provide a rough overview of the relative importance of major PM sources in the United States.
14     The emissions estimates are based on information presented in the EPA National Air Pollutant
15     Emission Trends Report, 1990-1999 (U.S. Environmental Protection Agency, 2001), to which
16     the reader is referred for detailed tables showing trends in PM2 5 emissions from a number of
17     source categories from 1990 to 1999.  Detailed descriptions of the methodology for constructing
18     emissions inventories for criteria pollutants, quality assurance procedures, and examples of
19     calculations of emissions can be found in U. S. Environmental Protection Agency (1999).
20     Although uncertainties associated with the estimates in the National Air Pollutant Emission
21     Trends Report are not given therein, a discussion of uncertainties in emissions estimates is given
22     in Section 3.3.5.
23           For the sake of completeness, an attempt was made to supplement the information given in
24     the emissions tables in the Trends Report, which concentrates mainly on anthropogenic
25     emissions,  with information about emissions from natural sources.  Details regarding the
26     composition of the emissions of primary PM2 5 from the source categories shown in Table 3-11
27     are summarized in Appendix 3D, where available. Fugitive dust emissions are estimated to
28     constitute over 50 percent of nationwide primary PM25 emissions according to Table 3-11.
29     However, there are a number of issues concerning the methods for obtaining relevant emissions
30     factor data for fugitive dust in field studies as discussed in Section 3.3.5. An estimate of the
31     production of PM2 5 from wind erosion on natural surfaces was not included in Table 3-11

       June 2003                                 3-92        DRAFT-DO NOT QUOTE OR CITE

-------
           TABLE 3-11. EMISSIONS OF PRIMARY PM2, BY VARIOUS SOURCES IN 1999
Source
On-road vehicle
exhaust
Non-road vehicle
exhaust
Fossil fuel
combustion
Industrial
processes
Biomass burning
Waste disposal
Fugitive dust
Windblown dust
Other
Total
Emissions
(10"kg/y)
0.21
0.37
0.36
0.35
1.2
0.48
3.3
NA1
0.02
6.2
Major PM
Components
Organic compounds,
elemental carbon
Organic compounds,
elemental carbon
Crustal elements,
trace metals
Metals, crustal material,
organic compounds
Organic compounds,
elemental carbon
Organic compounds,
trace metals
Crustal elements
Crustal elements
Organic compounds,
elemental carbon

Notes
Exhaust emissions from diesel (72%) and gasoline
vehicles (28%).
Exhaust emissions from off-road diesel (57%) and
gasoline vehicles (20%); ships and boats (10%);
aircraft (7%); railroads (6%).
Fuel burning in stationary sources such as power
plants (33%); industries (39%); businesses and
institutions (25%); residences (3%).
Metals processing (29%); mineral products (27%);
chemical mfg. (11%); other industries (33%).
Managed burning (47%); residential wood burning
(28%); agricultural burning (7%); wildfires (18%).
Open burning (91%); incineration (9%).
Dust raised by vehicles on paved (19%) and
unpaved roads (40%); construction (15%),
dust from raising crops (24%) and livestock (2%).
Dust raised by wind on bare land.
Structural fires.

        'NA = not available.
        Source:  Adapted from U. S. Environmental Protection Agency (2001).


 1     because this source is highly sporadic, occurs during periods of high winds, thus, the resulting
 2     emissions are too highly uncertain to be included.  As can be seen from a comparison of entries
 3     in Tables 3-11 and 3-12, estimates of emissions of potential precursors to secondary PM
 4     formation are considerably larger than those for estimates of primary PM25 emissions in the
 5     United States.  The emissions of SO2, NOX, and NH3 should be multiplied by factors of 1.5, 1.35,
 6     and 1.07, respectively, to account for their chemical form in the aerosol phase. Estimating a
 7     factor for VOCs  is somewhat less straight forward.  Turpin and Lim (2001) recommends a factor
 8     of 2 to account for the conversion of VOC precursors to oxygen- and nitrogen-containing
 9     compounds in the aerosol phase.  These factors are all greater than 1 and further underscore the
10     potential importance of secondary PM precursor emissions relative to primary PM emissions.
11     However, the emissions of precursors cannot be translated directly into rates of PM formation.
       June 2003
3-93
DRAFT-DO NOT QUOTE OR CITE

-------
         TABLE 3-12. EMISSIONS OF PRECURSORS TO SECONDARY PM2 s FORMATION
                                     BY VARIOUS SOURCES IN 1999
         Precursor
Emissions
(109 kg/y)
Secondary PM
Component
Notes
         SO,
         NO''2
         Anthropogenic
         VOCs
         Biogenic
         VOCs1
         NIL,
   17        Sulfate
   26        Nitrate
   16      Various mainly
           unidentified
           compounds of 'OC'
   44      Various mainly
           unidentified
           compounds of 'OC'

   45      Ammonium
                   Exhaust from on-road (2%) and non-road (5%) engines
                   and vehicles; fossil fuel combustion by electrical utilities,
                   industries, other sources (85%); various industrial
                   processes (7%); and other minor sources (1%).

                   Exhaust from on-road (34%) and non-road (22%) engines
                   and vehicles; fossil fuel combustion by electrical utilities,
                   industries, other sources (39%); lightning (4%); soils
                   (4%); and other minor sources i
                   Evaporative and exhaust emissions from on-road (29%)
                   and non-road (18%) vehicles; evaporation of solvents and
                   surface coatings (27%); biomass burning (9%); storage
                   and transport of petroleum and volatile compounds (7%);
                   chemical and petroleum industrial processes (5%); other
                   sources (5%).

                   Approximately 98% emitted by vegetation. Isoprene
                   (35%); monoterpenes (25%); all other reactive and
                   non-reactive compounds (40%).

                   Exhaust from on-road and non-road engines and vehicles
                   (5%); chemical manufacturing (3%); waste disposal,
                   recycling, and other minor sources (5%); livestock (82%);
                   and fertilizer application (18%).
         'Includes estimates of natural sources from Guenther et al. (2000).
         2Emissions expressed in terms of NO2.

         Source: Adapted from U. S. Environmental Protection Agency (2001).
 1      Dry deposition and precipitation scavenging of some of these gaseous precursors and their

 2      intermediate oxidation products occur before they are converted to PM in the atmosphere.

 3      In addition, some fraction of these gases are transported outside of the domain of the continental

 4      United States before being oxidized. Likewise, emissions of these gases from areas outside the

 5      United States can result in the transport of their oxidation products into the United States.

 6           As discussed in Section 3.3.1, the photochemical oxidation of sulfur dioxide leads to the

 7      production of sulfate; whereas that of nitrogen oxide leads ultimately to particulate-phase nitrite

 8      and nitrate.  Due to uncertainties it is difficult to calculate the rates of formation of secondary

 9      organic particulate matter (SOPM) from the emissions of VOC precursors. Smog chamber and

10      laboratory studies discussed in Section 3.3.1 indicate that anthropogenic aromatic compounds
        June 2003
                             3-94
                              DRAFT-DO NOT QUOTE OR CITE

-------
 1      and biogenic terpenoid compounds have the highest potential for forming secondary organic
 2      particulate matter; and as can be seen from Table 3C-1, the dominant compounds tend to be
 3      those derived from these categories.  Each of the source categories capable of emitting VOCs
 4      shown in Table 3-12 has components capable of forming SOPM, although in small yields
 5      (ranging typically up to several per cent, cf, Section 3.3.1).  The oxidation of lighter organic
 6      compounds leads ultimately to the formation of CO and CO2. As discussed by Pandis et al.
 7      (1991) and in Section 3.3.1, soluble gas phase compounds, such as formaldehyde (CH2O),  other
 8      aldehydes, organic acids, etc., formed during the oxidation of a wide variety of hydrocarbons,
 9      can be incorporated into suspended particles.  Although isoprene is a major component of
10      biogenic emissions, its oxidation has not been found to result in the formation of new particles;
11      whereas the oxidation of monoterpenes has.  However, it should be remembered that soluble gas
12      phase species such as  CH2O are formed during the oxidation of isoprene.
13           The emissions estimates shown in this section are based on annual totals.  However, annual
14      averages do not reflect the variability of a number of emissions categories on shorter time scales.
15      Residential wood burning in fireplaces and stoves, for example, is a seasonal practice that
16      reaches its peak during cold weather.  Cold weather also affects motor vehicle exhaust
17      particulate matter emissions, both in terms of chemical composition and emission rates (e.g.,
18      Watson et al., 1990b; Huang et al., 1994). Agricultural activities such as planting, fertilizing,
19      and harvesting are also seasonal. Forest fires occur mainly during the local dry season and
20      during periods of drought. Maximum dust production by wind erosion in the United States
21      occurs during the spring; whereas the minimum occurs during the summer (Gillette and Hanson,
22      1989). Efforts are being made to account for the seasonal variations of emissions in the
23      nationwide emissions inventories.  Techniques for calculating emissions of criteria pollutants on
24      a seasonal basis are given in U. S. Environmental Protection Agency (1999).
25           Trends in nationwide, annual average concentrations of PM10, and precursor gases (SO2,
26      NO2,  and VOC) over the 10 years from 1992 to 2001 are shown in Table 3-13.  As can be seen
27      from  Table 3-13, there have been substantial decreases in the ambient concentrations of PM10,
28      SO2, and NO2. Not enough data are available to define trends in concentrations of VOCs.  There
29      also have been substantial decreases in the emissions of all the species shown in Table 3-13,
30      except for NO2, although its average ambient concentration has decreased by 11%. These  entries
31      suggest that decreases in the average ambient concentration of PM10 could have been produced

        June 2003                                 3-95         DRAFT-DO NOT QUOTE OR CITE

-------
          TABLE 3-13.  NATIONWIDE CHANGES IN AMBIENT CONCENTRATIONS AND
               EMISSIONS OF PM10 AND GASEOUS PRECURSORS TO SECONDARY
                           PARTICULATE MATTER FROM 1992 TO 2001

PM10
PM2.5
SO2
NOX
VOC
% Change 1992-2001
Ambient Concentration Emissions
-14% -13%*
— -10%*
-35% (SO2) -24% (SO2)
-11%(NO2) -3%(NOX)
— -8%
         * Includes only primary PM.
         Source: U.S. Environmental Protection Agency (2002).
 1     by both decreases in emissions of primary PM10 and the formation of secondary PM10. The large
 2     reductions in ambient SO2 concentrations have resulted in reductions in sulfate formation that
 3     would have been manifest in PM2 5 concentrations on the regional scale in the East and Midwest
 4     where sulfate has constituted a larger fraction of PM2 5 than in the West. Likewise, reductions in
 5     NO2 concentrations would have had a more noticeable effect on PM2 5 concentrations in the West
 6     than in the East because nitrate is a larger component of the aerosol in the West.
 7           Trends in aerosol components (i.e., nitrate, sulfate, carbon, etc.) are needed for a more
 8     quantitative assessment of the effects of changes in emissions of precursors. Aerosol nitrate and
 9     sulfate concentrations obtained at North Long Beach and Riverside,  CA, tracked downward
10     trends in NOX concentrations.  SO2 and sulfate concentrations have both decreased; however, the
11     rate of decline of sulfate has been smaller than that of SO2, indicating the long range transport of
12     sulfate from outside the airshed may be an important source in addition to the oxidation of
13     locally generated SO2.  There are a number of reasons why pollutant concentrations do not track
14     estimated reductions in emissions. Some of these reasons are related to atmospheric effects such
15     as meteorological variability and secular changes in the rates of photochemical transformations
16     and deposition (U.S. Environmental Protection Agency, 2000c).  Other reasons  are related to
17     uncertainties in ambient measurements and in emissions inventories.
18
       June 2003                                 3-96         DRAFT-DO NOT QUOTE OR CITE

-------
 1      3.3.5   Uncertainties of Emissions Inventories
 2           As described in the 1996 PM AQCD, it is difficult to quantitatively assign uncertainties to
 3      entries in emissions inventories. Methods that can be used to verify or place constraints on
 4      emissions inventories are sparse.  In general, the overall uncertainty in the emissions of a given
 5      pollutant includes contributions from all of the terms used to calculate emissions (i.e., activity
 6      rates, emissions factors, and control device efficiencies). Additional uncertainties arise during
 7      the compilation of an emissions inventory because of missing sources and computational errors.
 8      The variability of emissions can cause errors when annual average emissions are applied to
 9      applications involving shorter time scales.
10           Activity rates for well-defined point sources (e.g., power plants) should have the smallest
11      uncertainty associated with their use because emissions are monitored continuously in many
12      cases and accurate production records need to be kept.  On the other hand, activity rates for a
13      number of very disperse fugitive sources are difficult to quantify. Emissions factors for easily
14      measured fuel components that are released quantitatively during combustion (e.g., CO2, SO2)
15      should be the most reliable.  Emissions of components formed during  combustion are more
16      difficult to characterize, as the emissions rates are dependent on factors specific to individual
17      combustion units and on combustion stage (i.e., smoldering or active). Although the AP-42
18      emissions factors (U.S. Environmental Protection  Agency, 1995) contain extensive information
19      for a large number of source types, these data are very limited in the number of sources sampled.
20      The efficiency of control devices is determined by their design, age, maintenance history,  and
21      operating conditions. It is virtually impossible to assign uncertainties in control device
22      performance because of these factors. It should be noted that the largest uncertainties occur for
23      those devices that have the highest efficiencies (>  90%). This occurs because the efficiencies are
24      subtracted from one, and small  errors in assigning efficiencies can lead to large errors in
25      emissions.
26           Ideally, an emissions inventory should include all major sources of a given pollutant. This
27      may be an easy task for major point sources. However, area sources of both primary PM and
28      precursors to secondary PM formation are more difficult to characterize than point sources; and,
29      thus, they require special emphasis when preparing emission inventories. Further research is
30      needed to better characterize the sources of pollutants to reduce this source of uncertainty.
31      Errors also can arise from the misreporting of data, and arithmetic errors can occur in the course

        June 2003                                 3-97        DRAFT-DO NOT QUOTE OR CITE

-------
 1      of compiling entries from thousands of individual sources. A quality assurance program is
 2      required to check for outliers and arithmetic errors.  Because of the variability in emissions rates,
 3      there can be errors in the application of inventories developed on an  annually averaged basis (as
 4      are the inventories shown in Tables 3-11 and 3-12) to episodes occurring on much shorter time
 5      scales. As an example, most modeling studies of air pollution episodes are carried out for
 6      periods of a few days.
 7           Uncertainties in annual emissions were estimated to range from 4 to 9% for SO2 and from
 8      6 to 11% for NOX in the 1985 NAPAP inventories for the United  States (Placet et al., 1991).
 9      Uncertainties in these estimates increase as the emissions are disaggregated both spatially and
10      temporally. The uncertainties quoted above are minimum estimates  and refer only to random
11      variability about the mean, assuming that the variability in emissions factors was adequately
12      characterized and that extrapolation of emissions factors to sources other than those for which
13      they were measured is valid. The estimates do not consider the effects of weather or variations
14      in operating and maintenance procedures.
15           Fugitive dust sources, as mentioned above, are extremely difficult to quantify; and  stated
16      emission rates may represent only order-of-magnitude estimates.  Although crustal dust
17      emissions constitute about 50% of the total primary PM25 inventory, they constitute less  than
18      about 15% of the source strengths inferred from the receptor modeling studies shown in
19      Table 3-9. However, it should be remembered that secondary components (sulfate, nitrate, and
20      some fractions of organic carbon) often account for most of the mass of ambient PM2 5 samples.
21           Although mineral dust sources represent the major category in  Table 3-11, their
22      contributions are distributed much more widely than are those from combustion sources.
23      Watson and Chow (2000) reexamined the methodology used to determine emissions of fugitive
24      dust. The standard methods use data obtained by particle monitors stacked at several elevations
25      from 1 to 2 m up to 7 to 10 m above the surface. However, small-scale turbulent motions and
26      variable winds characterize atmospheric flow patterns immediately adjacent to the surface in this
27      height range (Garratt, 1994). The depth of this turbulent layer is  determined by surface
28      roughness elements, and there is a high probability of particles being entrained in turbulent
29      eddies and redepositing on the ground within a very short distance. In addition to the source-
30      sampling problem referred to above, it should be remembered that dust often is raised in  remote
31      areas far removed from population centers. Precipitation or scavenging by cloud droplets and

        June 2003                                 3-98        DRAFT-DO NOT QUOTE OR CITE

-------
 1      dry deposition removes particles during transport from the source area.  In addition, gravitational
 2      settling can be an important loss mechanism for particles larger than a few micrometers in
 3      aerodynamic diameter.
 4           As rough estimates, uncertainties in emissions estimates could be as low as 10% for the
 5      best characterized source categories; whereas emissions figures for windblown dust should be
 6      regarded as order-of-magnitude estimates. The application of emissions inventories to the
 7      estimation of source contributions at monitoring sites is also limited by the effects of local
 8      topography and meteorology.  For example, Pinto et al. (1998) found that the contribution of
 9      power plants and residential space heating to PM2 5 concentrations in northwestern Bohemia are
10      comparable on the basis of CMB receptor modeling. However, according to the emissions
11      inventories, the contribution from power plants should have been roughly an order of magnitude
12      larger than that from residential space heating.  The difference between the two methods can be
13      explained by noting that mixing of the emissions from  the power plants  downward to the surface
14      is inhibited by strong surface inversions that develop during the winter season in this area.
15           There have been few field studies designed to test emissions inventories observationally.
16      The most direct approach would be to use aircraft to obtain cross-sections of pollutants upwind
17      and  downwind of major urban areas. The computed mass flux through a cross section of the
18      urban plume  can then be equated to emissions from the city chosen. This approach has been
19      attempted on a few occasions, but results have been ambiguous because of contributions from
20      fugitive sources, variable wind flows, and logistic difficulties.
21
22
23      3.4    SUMMARY AND KEY CONCLUSIONS
24           The recently deployed PM2 5 FRM network has returned data for a large number of sites
25      across the United States.  Annual mean PM2 5 concentrations range from about 5 |ig/m3 to about
26      30 |ig/m3.  In the eastern United States, the data from 1999 to 2001 indicate that highest
27      quarterly mean concentrations and maximum concentrations most often occur during the
28      summer. In the western United States, highest quarterly mean values and maximum values
29      occur mainly during the winter at a number of sites although there were exceptions to these
30      general patterns.  Sites affected strongly by sources of primary PM are expected to show winter
31      maxima. These findings are generally consistent with those based on longer term data sets such

        June 2003                                 3-99         DRAFT-DO NOT QUOTE OR CITE

-------
 1      as MAAQS in the eastern United States and the CARB network of dichotomous samplers in
 2      California. PM25 and PM10 concentrations in a number of urban areas have generally declined
 3      over the past few decades. However, they appear to have leveled off in the past few years.
 4           Differences in annual mean PM2 5 concentrations between monitoring sites in urban areas
 5      examined are typically less than 6 or 7 |ig/m3. However, on individual days, differences in 24-h
 6      average PM2 5 concentrations can be much larger.  Some  sites in metropolitan areas are highly
 7      correlated with each other but not to others due to the presence of local sources, topographic
 8      barriers, etc.  Although PM2 5 concentrations at sites within an MSA can be highly correlated,
 9      there still can be significant differences in their concentrations on any given day.  Consequently,
10      additional measures should be used to characterize the spatial variability of PM25 concentrations.
11      The degree of spatial uniformity in PM2 5 concentrations in  urban areas varies across the country.
12      These factors should be considered in using data obtained by the PM2 5 FRM network to
13      approximate community-scale human  exposure, and caution should be exercised in extrapolating
14      conclusions obtained in one urban area to another. PM2 5 to PM10 ratios were generally higher  in
15      the East than in the West, and values for this ratio are consistent with those found in numerous
16      earlier studies presented in the 1996 PM AQCD.
17           Data for PM10_2 5 are not as abundant as they are for PM2 5, and their interpretation  is
18      complicated by the difference method used to determine their concentrations.  The more
19      sporadic nature of sources of PM10_2 5 and its  shorter atmospheric lifetime tend to result in lower
20      spatial correlations for PM10_25 than for PM25 concentrations. Errors in measurement of PM25
21      and PM10 also result in calculations of lower spatial correlations of PM10_25. Calculated
22      concentrations of PM10_25 are occasionally negative as reflected by PM25 to PM10 ratios  greater
23      than one. Because analytical errors are generally larger for individual species than for total
24      mass, similar problems arise in their determination in PM10_2 5 samples by the difference
25      approach.  Some, but not all of these problems could be resolved by the use of dichotomous
26      samplers that also provide a direct sample of PM10_25 for compositional analyses.
27           Estimates of concentrations of individual species in PM10_25 samples used in this chapter
28      were limited to those obtained by dichotomous samplers. Generally, concentrations of most
29      elements differ for PM2 5 and PM10_2 5.  However, the available data suggest that concentrations of
30      many metals are of the same order of magnitude in both size fractions.  This is in marked
31      contrast to the situation twenty years ago when uncontrolled combustion sources were prevalent.

        June 2003                                 3-100        DRAFT-DO NOT QUOTE  OR CITE

-------
 1      At that time, concentrations of many metals, especially lead, were much higher than today in
 2      fine-mode particles, and their concentrations were much higher in the fine-mode than in the
 3      coarse-mode.  No substantive conclusions about contemporary concentrations and composition
 4      of ultrafme particles (0.1 jim < Da) can be drawn for the nation as a whole because of a lack of
 5      data.
 6           Ambient PM contains both primary and secondary components. The results of ambient
 7      monitoring studies and receptor modeling studies indicate that PM2 5  is dominated by secondary
 8      components in the eastern United States. General statements about the origin of OC in ambient
 9      PM2 5 samples cannot yet be made and so the contribution of secondary components throughout
10      the rest of the United States is still highly uncertain. Primary constituents represent smaller but
11      still important component of PM25. Crustal materials, which are primary constituents, constitute
12      the largest measured fraction of PM10_25 throughout the United States. Data for the concentration
13      of bioaerosols in both the PM25 and PM10_25 size ranges are sparse. Data collected in several
14      airsheds, including the Los Angeles Basin, Bakersfield and Fresno, CA;  and Philadelphia, PA,
15      suggest that secondary PM components are more uniformly distributed than are primary PM
16      components. Compositional data obtained at multiple sites in other urban areas are sparse.
17           Because of the complexity of the composition of ambient PM25 and PM10_25, sources are
18      best discussed in terms of individual constituents of both primary and secondary PM25 and
19      PM10_25. Each of these constituents can have anthropogenic and natural sources as shown in
20      Table 3-8. The distinction between natural and anthropogenic sources is not always obvious.
21      Although windblown dust might seem to be the result of natural processes, highest emission
22      rates are associated with agricultural activities in areas that are susceptible to periodic drought.
23      Examples include the dust bowl region of the midwestern United States and the Sahel of Africa.
24      There is also ongoing debate about characterizing wildfires as either  natural or anthropogenic.
25      Land management practices and other human actions affect the occurrence and scope of
26      wildfires. Similarly, prescribed burning can be viewed as anthropogenic or as a substitute for
27      wildfires that would otherwise eventually occur on the same land.
28           Over the past decade,  a significant amount of research has been carried out to improve the
29      understanding of the atmospheric chemistry of secondary organic PM formation.  Although
30      additional sources of SOPM might still be identified, there appears to be a general consensus that
31      biogenic compounds (monoterpenes, sesquiterpenes) and aromatic compounds (toluene,

        June 2003                               3-101       DRAFT-DO NOT QUOTE OR CITE

-------
 1      ethylbenzene) are the most significant SOPM precursors. A large number of compounds have
 2      been detected in biogenic and aromatic SOPM although the chemical composition of these two
 3      categories has not been fully established, especially for aromatic SOPM. Transformations that
 4      occur during the aging of particles are still inadequately understood.  There are still large gaps in
 5      the current understanding of a number of key processes related to the partitioning of semivolatile
 6      compounds between the gas phase and ambient particles containing organic compounds, liquid
 7      water, and inorganic salts and acids. In addition, there is a general lack of reliable analytical
 8      methods for measuring multifunctional oxygenated compounds in the gas and aerosol phases.
 9          The results of receptor modeling studies throughout the United  States indicate that the
10      combustion of fossil and biomass fuels is the major source of measured ambient PM2 5.  Fugitive
11      dust, found  mainly in the PM10_2 5 range size, represents the largest source of measured ambient
12      PM10 in many locations in the western United States.  Quoted uncertainties in the  source
13      apportionment of constituents in ambient aerosol samples typically range from  10 to 50%.  It is
14      apparent that a relatively small number of source categories, compared to the total number of
15      chemical species that typically are measured in ambient monitoring-source receptor model
16      studies, are  needed to account for the majority of the observed mass of PM in these studies.
17          The application of any of the source apportionment techniques is still limited by the
18      availability  of source profile data. Whereas the CMB approach relies directly on  source profile
19      data, solutions from the PMF technique yield profiles for the factors that contribute to PM.
20      However, there is some rotational ambiguity present in the solutions.  Source profile data
21      obtained by PMF must still be verified by comparison with data and these data can be used in
22      techniques such as PMF to improve the solutions.  Serious limitations still exist with regard to
23      source profiles for organic compounds. The complexity of reactions involving  organic
24      compounds in particles adds to the difficulties of finding stable species that could be used as
25      tracers.
26          As seen in Table  3-8, emissions of mineral dust, organic debris, and sea spray are
27      concentrated mainly in the coarse fraction of PM10 (> 2.5 jim aerodynamic diameter). A small
28      fraction of this material is in the PM25 size range (< 2.5  jim aerodynamic diameter). Still, PM25
29      concentrations of crustal material can be appreciable, especially during dust events.  It also
30      should be remembered that from one-third to one-half of the Saharan dust reaching the United
        June 2003                                3-102       DRAFT-DO NOT QUOTE OR CITE

-------
 1      States is in the PM2 5 size range. Emissions from combustion sources (mobile and stationary
 2      sources and biomass burning) are also predominantly in the PM2 5 size range.
 3           A number of sources contribute to policy relevant background (PRB) concentrations.  Data
 4      obtained at relatively remote monitoring sites (RRMS) in the western United States could be
 5      used to place reasonable upper limits on policy relevant background concentrations. More
 6      definitive results for both annual average and daily average concentrations could potentially be
 7      obtained from the application of source-receptor models and/or the  application of large-scale
 8      chemistry transport models. Many areas in the East are affected by dust transported from
 9      northern Africa, and it has  recently become apparent that many areas, especially, but not limited
10      to the Northwest, are affected by dust transported from the deserts of Asia. In addition to crustal
11      material, pollutants and primary biological aerosol particles (PBP) are also transported during
12      intercontinental transport events.  Many areas are also affected by smoke from wildfires either
13      within the United States or in Canada, Mexico and Central America.  Storms, in which the winds
14      can suspend material from  the surface of the land or seas, contribute soil, sea spray and PBP.
15      Contributions of primary PM from natural sources and sources outside Northern America as
16      given above are all episodic. Because the concentration of PBP is so poorly quantified, even
17      though it can constitute a significant fraction of the organic fraction of the atmospheric aerosol,
18      estimates of policy relevant background concentrations will remain highly uncertain. Estimates
19      of annually averaged PRB  concentrations or their range have not changed from the previous PM
20      assessment document.
21           Key points derived from the findings summarized above can be  highlighted as follows:
22          •  Spatial Variability in PM2; Concentrations.  Although PM2 5 concentrations within an
              MSA can be highly  correlated between sites, there can still be significant differences in
              their concentrations. The degree of spatial uniformity in PM2 5 concentrations and the
              strength of site to site correlations in urban areas varies across the country. These
              factors should be considered in using data obtained by the PM2 5 FRM network to
              approximate community-scale human exposures, and caution should be exercised in
              extrapolating conclusions as to spatial uniformity or correlations obtained in one urban
              area to another.  Limited information also suggests that the spatial variability in urban
              source contributions is likely to be larger than for regional source contributions to PM2 5
              and for PM2 5, itself.

        June 2003                                 3-103        DRAFT-DO NOT QUOTE OR CITE

-------
      PMin_2 ^ Concentrations.  Data for PM10_2 5 concentrations are not as abundant as they are
      for PM25. The difference method used in their derivation is subject to the effects of
      uncertainties in both PM10 and PM2 5.  As a result, estimates of PM10_2 5 concentrations, at
      times, come out as negative values, based on currently available data (e.g., in the EPA
      AIRS Database).
      Evaluating Source Contributions.  The use of organic compounds in source
      apportionment studies could potentially result in the attribution of PM to many more
      source categories than is possible using only trace elements. However, in the relatively
      few studies of the composition of the organic fraction of ambient particles that have been
      performed, typically only about  10 to 20 % of organic compounds have been quantified.
      The separation of contributions from diesel- and gasoline-fueled vehicles using organic
      marker compounds is still somewhat problematic. Additional efforts to develop
      protocols for extraction and analysis of organic markers are needed to fully realize their
      potential.
      Policy Relevant Background (PRB) Concentrations. Recent but limited information
      about PRB concentrations have not provided sufficient evidence to warrant any changes
      in estimates of the annual average background concentrations given in the 1996 PM
      AQCD.  These are:  1 to 4 |ig/m3 in the West and 2 to 5 |ig/m3 in the East for PM2 5; and
      approximately 3 |ig/m3 in both the East and the West for PM10_2 5, with a range of 0 to 9
      |ig/m3 in the West and 0 to 7 |ig/m3 in the East.  PRB concentration are likely to be
      highly variable both spatially and temporally. Further information regarding the
      frequency distribution of 24-hour concentrations based on analyses of observations at
      relatively remote monitoring sites  and on source apportionment analyses has become
      available and can be used for selected sites.
June 2003                                3-104        DRAFT-DO NOT QUOTE OR CITE

-------
REFERENCES

Altshuller, A. P. (1983) Review: natural volatile organic substances and their effect on air quality in the United
      States. Atmos. Environ. 17: 2131-2165.
Ansari, A. S.; Pandis, S. N. (1998) Response of inorganic PM to precursor concentrations. Environ. Sci. Technol.
      32: 2706-2714.
Atkinson, R. (1994) Gas-phase tropospheric chemistry of organic compounds. Washington, DC: American Chemical
      Society. (J. Phys. Chem. Ref. Data Monograph no. 2).
Aumont, B.; Madronich, S.; Bey, I.; Tyndall, G. S. (2000) Contribution of secondary VOC to the composition of
      aqueous atmospheric particles: a modeling approach. J. Atmos. Chem. 35: 59-75.
Bahadori, T.; Suh, H.; Koutrakis, P. (2000a) A cross-study analysis of ambient concentrations of fine mass. J. Air
      Waste Manage. Assoc.: in press.
Bahadori, T.; Suh, H.; Koutrakis, P. (2000b) PM2 5 and PM10 concentrations in four metropolitan cities in eastern
      United States. J. Air Waste Manage. Assoc.: in press.
Blando, J. D.; Porcja, R. J.; Li, T.-H.; Bowman, D.; Lioy, P. J.; Turpin, B. J. (1998) Secondary formation and the
      Smoky Mountain organic aerosol: an examination of aerosol polarity and functional group composition
      during SEAVS Environ. Sci. Technol. 32: 604-613.
Cabada, J. C.; Pandis Spyros, N.; Robinson, A. L. (2002) Sources of atmospheric carbonaceous paniculate matter in
      Pittsburgh, Pennsylvania. J. Air Waste Manage. Assoc. 52: 732-741.
Cahill, T. A. (1996) [Data from IMPROVE data base.] Fort Collins, CO: Air Resource Specialists, Inc.
California Air Resources Board. (1995)  [Database provided by JohnR. Holmes, Chief, Research Division.]
      Sacramento, CA: California Air Resources Board.
Calvert, J. G.; Heywood, J. B.; Sawyer, R. F.; Seinfeld, J. H. (1993) Achieving acceptable air quality: some
      reflections on controlling vehicle emissions. Science (Washington, DC) 261: 37-45.
Cass, G. R. ; Hughes, L. A.; Bhave, P.; Kleeman, M. J.; Allen, J. O.; Salmon, L. G. (2000) The chemical
      composition of atmospheric ultrafine particles. Philos. Trans. R. Soc. London Ser. A 358: 2581-2592.
Castro, L. M.; Pio, C. A.; Harrison, R. M.; Smith, D. J. T. (1999) Carbonaceous  aerosol in urban and rural European
      atmospheres: estimation of secondary organic carbon concentrations. Atmos. Environ. 33: 2771-2781.
Chadwick, O. A.; Deny, L. A.; Vitousek, P. M.; Huebert, B. J.; Hedin, L. O. (1999) Changing sources of nutrients
      during four million years of ecosystem development. Nature (London, UK)  397: 491-497.
Chow, J. C.; Watson, J. G. (2002) Review of PM2 5 and PM10 apportionment for  fossil fuel combustion and other
      sources by the chemical mass balance receptor model. Energy Fuelsl6: 222-260.
Chow, J. C.; Watson, J. G.; Frazier, C. A.; Egami, R. T.; Goodrich, A.; Ralph, C. (1988) Spatial and temporal source
      contributions to PM10 andPM25 in Reno, NV. In: Mathai, C. V.; Stonefield, D. H., eds. PM-10:
      implementation of standards, an APCA/EPA international specialty conference; February; San Francisco, CA.
      Pittsburgh, PA: Air Pollution Control Association; pp. 438-457. (APCA transactions series TR-13).
Chow, J. C.; Watson, J. G.; Richards, L. W.; Haase, D. L.; McDade, C.; Dietrich, D. L.; Moon, D.; Sloane, C. (1991)
      The 1989-90 Phoenix PM10 study. Volume II:  source apportionment. Final report. Phoenix, AZ: Arizona
      Department of Environmental Quality; Desert Research Institute document no.  8931.6F1.
Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Frazier, C. A.; Hinsvark, B. A.; Prichett, L. C.; Neuroth, G. R.
      (1992a) Wintertime PM10 and PM2 5 chemical compositions and source contributions in Tucson, Arizona.
      In: Chow, J. C.; Ono, D. M., eds. PM10 standards and nontraditional paniculate source controls, an
      A&WMA/EPA international specialty conference, v. I; January; Scottsdale, AZ. Pittsburgh, PA: Air & Waste
      Management Association;  pp. 231-243. (A&WMA transactions series no. 22).
Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Solomon, P. A.; Magliano, K. L.; Ziman, S. D.; Richards, L. W.
      (1992b) PM10 source apportionment in California's San Joaquin Valley. Atmos. Environ. Part A
      26: 3335-3354.
Chow, J. C.; Liu, C. S.; Cassmassi, J.; Watson, J. G.; Lu, Z.; Pritchett, L.  C. (1992c) A neighborhood-scale study of
      PM10 source contributions  inRubidoux, California. Atmos. Environ. Part  A  26: 693-706.
Chow, J. C.; Watson, J. G.; Edgerton, S. A.; Vega, E.; Ortiz, E. (2002) Spatial differences in outdoor PM10 mass and
      aerosol composition in Mexico City. J. Air Waste Manage. Assoc.  52: 423-434.
Christoffersen, T. S.; Hjorth, J.; Horie, O.; Jensen, N. R.; Kotzias, D.;  Molander, L. L.; Neeb, P.; Ruppert, L.;
      Winterhalter, R.; Virkkula, A.; Wirtz, K.; Larsen, B. R. (1998)  cis-Pinic acid, a possible precursor for organic
      aerosol formation from ozonolysis of a-pinene. Atmos. Environ. 32: 1657-1661.
Chung, A.; Herner, J. D.; Kleeman, M. J. (2001) Detection of alkaline ultrafine atmospheric particles at Bakersfield,
      California. Environ.  Sci. Technol. 35: 2184-2190.
June 2003                                      3-105         DRAFT-DO NOT QUOTE OR CITE

-------
Cocker, D. R., Ill; Clegg, S. L.; Flagan, R. C.; Seinfeld, J. H. (2001) The effect of water on gas-particle partitioning
      of secondary organic aerosol. Part I: a-pinene/ozone system. Atmos. Environ. 35: 6049-6072.
Darlington, T. L.; Kahlbaum, D. F.; Heuss, J. M; Wolff, G. T. (1997) Analysis of PM10 trends in the United States
      from 1988 through 1995. J. Air Waste Manage. Assoc. 47: 1070-1078.
Davis, A. P.; Shokouhian, M.; Ni, S. (2001) Loading estimates of lead, copper, cadmium, and zinc in urban runoff
      from specific sources. Chemosphere 44: 997-1009.
Dolislager, L. J.; Motallebi, N. (1999) Characterization of paniculate matter in California. J. Air Waste Manage.
      Assoc. 49: PM-45-56.
Duce, R. A. (1995) Sources, distributions, and fluxes of mineral aerosols and their relationship to climate.
      In: Charlson, R. J.; Heintzenberg, J., eds. Aerosol forcing of climate: report of the Dahlem workshop on
      aerosol forcing of climate; April  1994; Berlin, Federal Republic of Germany. Chichester, United Kingdom:
      John Wiley & Sons,  Ltd.; pp. 43-72.
Dzubay, T.  G.; Stevens, R.  K.; Gordon, G. E.; Olmez, I.; Sheffield, A. E.; Courtney, W. J. (1988) A composite
      receptor method applied to Philadelphia aerosol. Environ. Sci. Technol. 22: 46-52.
Edney, E. O.; Driscoll, D. J.; Weathers, W. S.; Kleindienst, T. E.; Conver, T. S.; Mclver, C. D.; Li, W. (2001)
      Formation of polyketones in irradiated toluene/propylene/NOx/air mixtures. Aerosol Sci. Technol.
      35: 998-1008.
Eldred, R. A.; Cahill, T. A. (1994)  Trends in elemental concentrations of fine particles at remote sites in the
      United States of America. Atmos. Environ. 28: 1009-1019.
Faust, B. C. (1994) Photochemistry of clouds, fogs, and aerosols. Environ. Sci. Technol. 28: 217A-222A.
Fitz-Simons, T.  S.; Mathias, S.; Rizzo, M. (2000) Analyses of 1999 PM  data for the PMNAAQS review. Research
      Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards;
      November 17. Available: http://www.epa.gov/oar/oaqps/pm25/analyses.html [2 April, 2002].
Forstner, H. J. L.; Flagan, R. C.; Seinfeld, J. H. (1997a) Molecular speciation of secondary organic aerosol from
      photooxidation of the higher alkenes: 1-octene and 1-decene. Atmos. Environ. 31: 1953-1964.
Forstner, H. J. L.; Flagan, R. C.; Seinfeld, J. H. (1997b) Secondary organic aerosol from the photooxidation of
      aromatic  hydrocarbons: molecular composition. Environ. Sci. Technol. 31: 1345-1358.
Garratt, J. R. (1994) The atmospheric boundary layer. Cambridge, United Kingdom: Cambridge University Press.
      (Cambridge atmospheric and space science series).
Gatz, D. F.; Prospero, J. M. (1996) A large silicon-aluminum aerosol plume in central Illinois: North African Desert
      dust? Atmos. Environ. 30: 3789-3799.
Geron, C.; Rasmussen, R.;  Arnts, R. R.; Guenther, A. (2000) A review and synthesis of monoterpene speciation
      from forests in the United States. Atmos. Environ. 34: 1761-1781.
Geron, C. D.; Harley, P.; Guenther, A. (2001) Isoprene emission capacity for US tree species. Atmos. Environ.
      35: 3341-3352.
Gillette, D. A.; Hanson, K.  J. (1989) Spatial and temporal variability of dust production caused by wind erosion in
      the United States. J.  Geophys. Res. [Atmos.] 94: 2197-2206.
Gillies, J. A.;  Gertler, A. W.; Lowenthal, D.; Abu-Allaban, M.;  Jennison, B.; Goodrich, A. (2000) Enhanced PM25
      source apportionment using chemical mass balance receptor modeling and scanning electron microscopy.
      Aerosol Sci. Technol.: submitted.
Glasius, M.; Lahaniati, M.; Calogirou, A.; Di Bella, D.; Jensen, N. R.; Hjorth, J.; Kotzias, D.; Larsen, B. R. (2000)
      Carboxylic acids in secondary aerosols from oxidation of cyclic monoterpenes by ozone. Environ. Sci.
      Technol.  34: 1001-1010.
Glover, D. M.; Hopke, P. K.; Vermette, S. J.; Landsberger, S.; D'Auben, D. R. (1991) Source apportionment with
      site specific source profiles. J. Air Waste Manage. Assoc. 41: 294-305.
Gone, J.-K.; Olmez, I.; Ames, M. R. (2000) Size distribution and probable sources of trace elements in submicron
      atmospheric paniculate material. J. Radioanal. Nucl. Chem. 244:  133-139.
Graedel, T.  E.; Goldberg, K. I. (1983) Kinetic studies of raindrop chemistry: 1. Inorganic and organic processes.
      J. Geophys. Res. C:  Oceans Atmos. 88: 10,865-10,882.
Gray, H. A.; Landry, B.; Liu, C. S.; Henry, R. C.; Cooper, J. A.; Sherman, J. R. (1988) Receptor modeling for PM10
      source apportionment in the South Coast Air Basin of California.  In: Mathai, C. V.; Stonefield, D. H., eds.
      PM.10: implementation of standards, an APCA/EPA international  specialty conference; February; San
      Francisco, CA. Pittsburgh, PA: Air Pollution Control Association; pp. 399-418. (APCA transactions series
      TR-13).
Griffin, R. J.; Cocker, D. R., Ill; Seinfeld, J. H. (1999) Incremental aerosol reactivity: application to aromatic and
      biogenic  hydrocarbons. Environ. Sci. Technol. 33: 2403-2408.
June 2003                                      3-106         DRAFT-DO NOT QUOTE OR CITE

-------
Griffin, D. W.; Garrison, V. H.; Herman, J. R.; Shinn, E. A. (2001) African desert dust in the Caribbean atmosphere:
      microbiology and public health. Aerobiologia 17: 203-213.
Grosjean, D. (1992) In situ organic aerosol formation during a smog episode: estimated production and chemical
      functionality. Atmos. Environ. Part  A 26: 953-963.
Grosjean, D.; Seinfeld, J. H. (1989) Parameterization of the formation potential of secondary organic aerosols.
      Atmos. Environ. 23: 1733-1747.
Guenther, A.; Zimmerman, P.; Klinger, L.; Greenberg, J.; Ennis, C.; Davis, K.; Pollock, W.; Westberg, H.; Allwine,
      G.; Geron, C. (1996) Estimates of regional natural volatile organic compound fluxes from enclosure and
      ambient measurements. J. Geophys. Res. 101: 1345-1360.
Guenther, A.; Geron, C.; Pierce, T.; Lamb, B.; Harley, P.; Fall, R. (2000) Natural emissions of non-methane volatile
      organic compounds, carbon monoxide, and oxides of nitrogen from North America. Atmos. Environ.
      34: 2205-2230.
Haagen-Smit, A. J.  (1952) Chemistry and physiology of Los Angeles smog. Ind. Eng.  Chem. 44: 1342-1346.
Helmig, D.; Greenberg,  J.; Guenther, A.; Zimmerman, P.; Geron, C. (1998) Volatile organic compounds and
      isoprene oxidation products at a temperate deciduous forest site.  J. Geophys. Res. 103(D17): 22,397-22,414.
Henry, R. C. (1997) History and fundamentals of multivariate air quality receptor models. Chemom. Intell. Lab.
      Syst. 37: 37-42.
Hewitt, C. N. (2001) The atmospheric chemistry of sulphur and nitrogen in power station plumes. Atmos. Environ.
      35: 1155-1170.
Hidy, G. M. (1994) Atmospheric sulfur and nitrogen oxides: eastern North American source-receptor relationships.
      San Diego, CA: Academic Press, Inc.
Hildemann, L. M.; Rogge, W. F.; Cass, G.  R.; Mazurek, M. A.; Simoneit, B. R. T. (1996) Contribution of primary
      aerosol emissions from vegetation-derived sources to fine particle concentrations in Los Angeles. J. Geophys.
      Res. [Atmos.] 101: 19,541-19,549.
Hoffmann, T.; Odum, J.  R.; Bowman, F.; Collins, D.; Klockow, D.; Flagan, R. C.; Seinfeld, J. H. (1997) Formation
      of organic aerosols from the oxidation of biogenic hydrocarbons. J. Atmos. Chem. 26: 189-222.
Hopke, P. K.; Wlaschin, W.; Landsberger,  S.; Sweet, C.; Vermette, S. J. (1988) The source apportionment of PM10
      in south Chicago. In: Mathai, C. V.; Stonefield, D. H., eds. PM-10: implementation of standards, an
      APCA/EPA international specialty conference; February; San Francisco, CA. Pittsburgh, PA: Air Pollution
      Control Association; pp. 484-494. (APCA transactions series TR-13).
Houck, J. E.; Rau, J. A.; Body, S.; Chow, J. C.  (1992) Source apportionment: Pocatello, Idaho PM10 nonattainment
      area. In: Chow, J. C.; Ono, D. M., eds. PM10 standards and nontraditional paniculate source controls, an
      A&WMA/EPA international specialty conference, v. I; January;  Scottsdale, AZ. Pittsburgh, PA: Air & Waste
      Management Association; pp. 219-230. (A&WMA transactions series no. 22).
Huang, X.; Olmez, L; Aras, N.  K.; Gordon, G. E. (1994) Emissions of trace elements from motor vehicles: potential
      marker elements and source composition profile. Atmos. Environ. 28: 1385-1391.
Husar, R. B.; Schichtel,  B. A.; Falke, S.  R.; Li, F.; Wilson, W. E.; Pinto, J.; Malm, W. C.; Fox, D. G.; Feldman,
      G. C.; McClain, C.; Kuring, N.; Holben, B. N.; Vermote, E. F.; Herman, J. R.; Elvidge, C. D. (2000) The
      impact of the 1998 Central American smoke on the atmospheric environment of eastern North America.
      J. Geophys. Res.: submitted.
Inhalable Paniculate Network.  (1985) [Inhalable Paniculate Network Database.] Research Triangle Park,  NC:
      U.S. Environmental Protection Agency, National Exposure Research Laboratory; Jose M. Sune, contact.
Izumi, K.; Fukuyama, T. (1990) Photochemical aerosol formation from aromatic hydrocarbons in the presence of
      NOX. Atmos. Environ. Part A 24: 1433-1441.
Jang, M.; Kamens, R. M. (1999) Newly  characterized products and composition of secondary aerosols from the
      reaction of a-pinene with ozone. Atmos. Environ. 33: 459-474.
Jang, M.; Kamens, R. M. (200la) Atmospheric secondary aerosol formation by heterogeneous reactions of
      aldehydes in the presence of a sulfuric acid aerosol catalyst. Environ. Sci. Technol. 35: 4758-4766.
Jang, M.; Kamens, R. M. (200 Ib) Characterization of secondary aerosol from the photooxidation of toluene in the
      presence of NOxand 1-propene. Environ. Sci. Technol. 35: 3626-3639.
Kamens, R.; Jang, M.; Chien, C.-J.; Leach, K. (1999) Aerosol formation from the reaction of a-pinene and ozone
      using a gas-phase kinetics-aerosol partitioning model. Environ. Sci. Technol. 33: 1430-1438.
Kao, A.  S.; Friedlander,  S. K. (1995) Frequency distributions of PM10 chemical components and their sources.
      Environ. Sci. Technol. 29: 19-28.
Kavouras, I. G.; Mihalopoulos, N.; Stephanou, E. G. (1998) Formation of atmospheric particles from organic acids
      produced by forests. Nature (London) 395: 683-686.
June 2003                                      3-107         DRAFT-DO NOT QUOTE OR CITE

-------
Kavouras, I. G.; Mihalopoulus, N.; Stephanou, E. G. (1999a) Formation and gas/particle partitioning of monoterpene
      photooxidation products over forests. Geophys. Res. Lett. 26: 55-58.
Kavouras, I. G.; Mihalopoulus, N.; Stephanou, E. G. (1999b) Secondary organic aerosol formation vs primary
      organic aerosol emission: in situ evidence for the chemical coupling between monoterpene acidic
      photooxidation products and new particle formation over forests. Environ. Sci. Technol. 33: 1028-1037.
Keywood, M. D.; Ayers, G. P.; Gras, J. L.; Gillett, R. W.; Cohen, D. D. (1999) Relationships between size
      segregated mass concentration data and ultrafine particle number concentrations in urban areas. Atmos.
      Environ.  33:2907-2913.
Khalil, M. A. K.; Rasmussen, R. A. (1992) Forest hydrocarbon emissions: relationships between fluxes and ambient
      concentrations. J. Air Waste Manage. Assoc. 42: 810-813.
Kim, B. M.; Zeldin, M. D.; Liu, C. S. (1992) Source apportionment study for state implementation plan development
      in the Coachella Valley. In: Chow, J. C.; Ono, D. M., eds. PM10 standards and nontraditional paniculate
      source controls: an A&WMA/EPA international specialty conference, v. II; January; Scottsdale, AZ.
      Pittsburgh, PA: Air & Waste Management Association; pp. 979-991.  (A&WMA transactions series no. 22).
Kinney, P. L.; Aggarwal, M.; Northridge, M. E.; Janssen, N. A. H.; Shepard, P. (2000) Airborne concentrations of
      PM2.5 and diesel exhaust particles on Harlem sidewalks: a community-based pilot study. Environ. Health
      Perspect. 108: 213-218.
Koch, S.; Winterhalter, R.; Uherek, E.; Kolloff, A.; Neeb, P.; Moortgat, G. K.  (2000) Formation of new particles in
      the gas-phase ozonolysis of monoterpenes. Atmos. Environ. 34: 4031-4042.
Koutrakis, P. (1995) [PM25 andPM10 data for Philadelphia, 1992-1995]. Boston, MA:  Harvard School of Public
      Health.
Landis, M.  S.; Norris, G. A.; Williams, R. W.; Weinstein, J. P. (2001) Personal exposures to PM2 5 mass and trace
      elements in Baltimore, MD, USA. Atmos. Environ. 35: 6511-6524.
Larsen, B. R.; Di Bella, D.; Glasius, M.; Winterhalter, R.; Jensen, N. R.; Hjorth, J. (2001) Gas-phase OH oxidation
      of monoterpenes: gaseous and paniculate products. J. Atmos. Chem. 38: 231-276.
Leach, K. B.; Kamens, R. M.; Strommen, M. R.; Jang, M. (1999) Partitioning of semivolatile organic compounds in
      the presence of a secondary organic aerosol in a controlled atmosphere. J. Atmos. Chem. 33: 241-264.
Leaderer, B. P.; Naeher, L.; Jankun, T.; Balenger, K.; Holford, T. R.; Toth, C.; Sullivan, J.; Wolfson, J. M.;
      Koutrakis, P. (1999) Indoor, outdoor, and regional summer and winter concentrations of PM10, PM2 5, SO42",
      H+, NH4+, NO3", NH3, and nitrous acid in homes with and without kerosene space heaters. Environ. Health
      Perspect. 107:223-231.
Lefohn, A.  S.; Pinto, J. P.; Shadwick, D.; Ziman, S. D. (2004) The variation of background paniculate matter at
      'clean sites'  in the United States. J. Air Waste Manage. Assoc.: submitted.
Lefohn, A.  S.; Pinto, J. P.; Shadwick, D.; Ziman, S. D. (2004) The variation of background paniculate matter at
      'clean sites'  in the United States. J. Air Waste Manage. Assoc.: submitted.
Lin, J. J.; Noll, K. E.; Holsen, T. M. (1994) Dry deposition velocities as a function of particle size in the ambient
      atmosphere. Aerosol Sci. Technol. 20: 239-252.
Lipfert, F. W. (1998) Trends in airborne paniculate matter in the United States. Appl. Occup. Environ. Hyg.
      13: 370-384.
Magliano, K. L.; Ranzieri, A.; Solomon, P.; Watson, J. (1998)  Chemical mass balance modeling of data from the
      1995 integrated monitoring study. Sacramento, CA: California Air Resources Board; California Regional
      PM10/PM2 5  Air Quality Study.
Matthias-Maser, S. (1998) Primary biological aerosol particles: their significance, sources, sampling methods and
      size distribution in the  atmosphere. In: Harrison, R. M.; Van Grieken, R., eds. Atmospheric Particles.
      New York, NY: John Wiley & Sons; pp. 349-368. (IUPAC series on Analytical and Physical Chemistry of
      Environmental Systems, v. 5).
Mazurek, M.; Masonjones, M. C.; Masonjones, H. D.; Salmon, L. G.; Cass,  G. R.; Hallock, K. A.; Leach, M. (1997)
      Visibility-reducing organic aerosols in the vicinity of Grand Canyon National Park: properties observed by
      high resolution gas chromatography. J. Geophys. Res. [Atmos.] 102: 3779-3793.
McMurry, P. H.; Woo, K. S.;  Weber, R.; Chen, D.-R.; Pui, D. Y. H. (2000) Size distributions of 3 to 10 nm
      atmospheric particles: implications for nucleation mechanisms. Philos. Trans. R. Soc. London Ser. A
      358: 2625-2642.
Monn, C. (2001) Exposure assessment of air pollutants: a review on spatial heterogeneity  and
      indoor/outdoor/personal exposure to suspended paniculate matter, nitrogen dioxide and ozone.
      Atmos. Environ. 35: 1-32.
Motallebi, N. (1999) Wintertime PM2 5 and PM10 source apportionment at Sacramento, California. J. Air Waste
      Manage.  Assoc. 49: PM-25-34.


June 2003                                     3-108          DRAFT-DO NOT QUOTE OR CITE

-------
NARSTO. (2002) Fine particle assessment (2nd review draft). Available: http://www.cgenv.com/Narsto/
      (13 February 2002).
Neas, L. M. (1996) [Data provided for 1986 from Harvard six cities data base.] Boston, MA: Harvard School of
      Public Health.
Norris, G. (1998) Air pollution and the exacerbation of asthma in an arid, western, US city [dissertation]. Seattle,
      WA: University of Washington. Available from: University Microfilms International, Ann Arbor, MI;
      publication no. 9836228.
Norris, G.; Young-Pong, S. N.; Koenig, J. Q.; Larson, T. V.; Sheppard, L.; Stout, J. W. (1999) An association
      between fine particles and asthma emergency department visits for children in Seattle. Environ. Health
      Perspect. 107: 489-493.
Odum, J. R.; Hoffmann, T.; Bowman, F.; Collins, D.; Flagan, R. C.; Seinfeld, J. H. (1996) Gas/particle partitioning
      and secondary organic aerosol yields. Environ.  Sci. Technol. 30: 2580-2585.
Odum, J. R.; Jungkamp, T. P. W.; Griffin, R. J.; Flagan, R. C.; Seinfeld, J. H. (1997a) The atmospheric
      aerosol-forming potential of whole gasoline vapor. Science (Washington, DC) 276: 96-99.
Odum, J. R.; Jungkamp, T. P. W.; Griffin, R. J.; Forstner, H. J. L.; Flagan, R. C.; Seinfeld, J. H. (1997b) Aromatics,
      reformulated gasoline, and atmospheric organic aerosol formation. Environ.  Sci. Technol. 31:  1890-1897.
Ondov, J. M. (1996) Paniculate tracers for source attribution: potential for application to California's San Joaquin
      Valley. J. Aerosol Sci. 27(suppl. 1): S687-S688.
Ott, S.-T.; Ott, A.; Martin, D. W.; Young, J. A. (1991) Analysis of a trans-Atlantic  Saharan dust outbreak based on
      satellite and GATE data. Mon. Weather Rev. 119:  1832-1850.
Paatero, P.; Tapper, U. (1993) Analysis of different modes of factor analysis as least squares fit problems. Chemom.
      Intelligent Lab. Syst.  18: 183-194.
Paatero, P.; Tapper, U. (1994) Positive matrix factorization: a non-negative factor model with optimal utilization of
      error estimates of data values. Environmetrics 5: 111-126.
Pandis, S. N.; Paulson, S. E.; Seinfeld, J. H.; Flagan, R. C. (1991) Aerosol formation in the photooxidation of
      isoprene and (3-pinene. Atmos. Environ. Part A 25: 997-1008.
Pandis, S. N.; Harley, R. A.; Cass,  G. R.; Seinfeld, J. H. (1992) Secondary organic aerosol formation and transport.
      Atmos. Environ. Part A 26: 2269-2282.
Perry, K. D.; Cahill, T. A.; Eldred,  R. A.; Dutcher, D.  D.; Gill, T. E. (1997) Long-range transport of North African
      dust to the eastern United States. J. Geophys. Res.  [Atmos] 102: 11,225-11,238.
Pinto, J. P.; Pleasant, M.; Kellogg,  R.; Torok, S.; Stiles, D.; Willis, R. D.; Koutrakis, P.; Allen, G. A. (1995)
      Chemical characterization of the ambient aerosol in Philadelphia. Presented at: Paniculate matter: health and
      regulatory issues: an A&WMA international specialty conference; April; Pittsburgh, PA. Pittsburgh, PA:
      Air & Waste Management Association.
Pinto, J. P.; Stevens, R. K.; Willis,  R. D.; Kellogg, R.; Mamane, Y.; Novak, J.; Santroch, J.; Benes, L; Lenicek, J.;
      Bures, V. (1998) Czech air quality monitoring and receptor modeling study.  Environ. Sci. Technol. 32:
      843-854.
Pinto, J. P.; Lefohn, A. S.; Shadwick, D. S. (2002) Aspects of the spatial variability of PM25 concentrations within
      urban areas of the United States. J. Air Waste Manage. Assoc.: submitted.
Pio, C.; Alves, C.; Duarte, A. (2001) Organic components of aerosols in aforested area of central Greece. Atmos.
      Environ. 35: 389-401.
Placet, M.; Battye, R. E.; Fehsenfeld, F. C.; Bassett, G. W. (1991) Emissions involved in acidic deposition
      processes. In: Irving, P. M.,  ed. Acidic deposition: state of science and technology, volume I: emissions,
      atmospheric processes, and deposition. Washington, DC: The U.S. National Acid Precipitation Assessment
      Program. (State of science and technology report no. 1).
Poore, M. W. (2000) Oxalic acid in PM2 5 paniculate matter in California. J. Air Waste Manage. Assoc.
      50: 1874-1875.
Prospero, J. M. (1996) The atmospheric transport of particles to the ocean. In: Ittekkot, V.; Honjo, S.; Depetris, P.  J.,
      eds. Particle flux in the ocean. New York, NY:  John Wiley & Sons, Ltd.; pp. 19-52. (SCOPE report no. 57).
Prospero, J. M. (1999) Long-range transport of mineral dust in the global atmosphere: impact of African dust on the
      environment of the southeastern United States. Proc. Natl. Acad. Sci. U.S.A. 96: 3396-3403.
Prospero, J. M.; Nees, R. T.; Uematsu, M. (1987) Deposition rate of paniculate and dissolved aluminum derived
      from Saharan dust in precipitation at Miami, Florida. J. Geophys. Res. [Atmos.] 92: 14,723-14,731.
Prospero, J. M.; Olmez, I.; Ames, M. (2001) Al and Fe in PM25 and PM10 suspended particles in south-central
      Florida: the impact of the long range transport of African mineral dust. J. Water Air Soil Pollut. 125: 291-317.
Pun, B. K.; Seigneur, C.; Grosjean, D.; Saxena, P. (2000) Gas-phase formation of water-soluble organic compounds
      in the atmosphere: a retrosynthetic analysis. J. Atmos. Chem. 35: 199-223.


June 2003                                      3-109         DRAFT-DO NOT QUOTE OR CITE

-------
Pun, B. K.; Wu, S.-Y.; Seigneur, C. (2002) Contribution of biogenic emissions to the formation of ozone and
      paniculate matter in the eastern United States. Environ. Sci. Technol. 36: 3586-3596.
Ramadan, Z.; Song, X.-H.; Hopke, P. K. (2000) Identification of sources of Phoenix aerosol by positive matrix
      factorization. J. Air Waste Manage. Assoc. 50: 1308-1320.
Rizzo, M; Pinto, J. P. (2001) Initial characterization of fine paniculate matter (PM25) collected by the National
      Federal Reference Monitoring Network. Presented at: 94th annual conference & exhibition of the Air &Waste
      Management Association; June, Orlando, FL. Pittsburgh, PA: Air & Waste Management Association.
Rodes, C. E.; Evans, E.  G. (1985) Preliminary assessment of 10 um paniculate sampling at eight locations in the
      United States. Atmos. Environ. 19: 293-303.
Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993) Sources of fine organic
      aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ. Sci.
      Technol. 27:636-651.
Ryan, W. M.; Badgett-West, C. R.; Holtz, D. R.; Peters, T. A.; Cooper, J. A.; Ono, D. (1988) Reconciliation of
      receptor and dispersion modeling impacts of PM10 in Hayden, Arizona. In: Mathai, C. V.; Stonefield, D. H.,
      eds. PM-10: implementation of standards, an APCA/EPA international specialty conference; February;
      San Francisco, CA. Pittsburgh, PA: Air Pollution Control Association; pp. 419-429. (APCA transactions
      series TR-13).
Salma, L; Maenhaut, W.; Zaray, G. (2002) Comparative study of elemental mass size distributions in urban
      atmospheric aerosol. J. Aerosol Sci. 33: 339-356.
Sarwar, G.; Corsi, R.; Kumura, Y.; Allen, D.; Weschler, C. J. (2002) Hydroxyl radicals in indoor environments.
      Atmos. Environ.  36: 3973-3988.
Savoie, D. L.; Prospero, J. M. (1980) Water-soluble potassium, calcium, and magnesium in the aerosols over the
      tropical North Atlantic. J. Geophys. Res. 85: 385-392.
Saxena, P.; Hildemann, L. M. (1996) Water-soluble organics in atmospheric particles: a critical review of the
      literature and applications of thermodynamics to identify candidate compounds. J. Atmos. Chem. 24: 57-109.
Schauer, J. J.; Cass, G. R. (2000) Source apportionment of wintertime gas-phase and  particle-phase air pollutants
      using organic compounds as tracers. Environ. Sci. Technol. 34: 1821-1832.
Schauer, J. J.; Rogge, W. F.; Hildemann, L. M.; Mazurik, M. A.; Cass, G. R.  (1996) Source apportionment of
      airborne paniculate matter using organic compounds as tracers. Atmos. Environ. 30: 3837-3855.
Shinn, E. A.; Smith, G. W.; Prospero, J. M.; Betzer, P.; Hayes, M. L.; Garrison, V.; Barber, R. T. (2000) African
      dust and the demise of Caribbean coral reefs. Geophys. Res. Lett. 27: 3029-3032.
Skidmore, L. W.; Chow, J. C.; Tucker, T. T. (1992) PM10 air quality assessment for the Jefferson County, Ohio air
      quality control region. In: Chow, J. C.; Ono, D. M., eds. PM10 standards and nontraditional paniculate source
      controls, an A&WMA/EPA international specialty conference, v. II; January; Scottsdale,  AZ. Pittsburgh, PA:
      Air and Waste Management Association; pp.  1016-1031. (A&WMA transactions series no. 22).
Smith, G. T.; Ives, L. D.; Nagelkerken, I. A.; Ritchie, K. B. (1996) Caribbean sea fan mortalities. Nature (London)
      383: 487.
Song, X.-H.; Polissar, A. V.; Hopke, P.  K. (2001) Sources of fine particle composition in the northeastern US.
      Atmos. Environ.  35: 5277-5286.
Spengler, J. D.; Briggs,  S. L. K.; Ozkaynak, H. (1986) Relationship between TSP measurements and
      size-fractionated particle mass measurements in six cities participating in the Harvard Air Pollution Health
      Study. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning
      and Standards; EPA cooperative  agreement no. CR-812667-01-0; December 5.
Spengler, J. D.; Koutrakis, P.; Dockery, D. W.; Raizenne, M.;  Speizer, F. E. (1996) Health effects of acid aerosols
      on North American children: air pollution exposures. Environ. Health Perspect. 104: 492-499.
Strader, R.; Lurmann, F.; Pandis, S. N. (1999) Evaluation of secondary organic aerosol formation in winter.
      Atmos. Environ.  33: 4849-4863.
Suarez, A. E.; Caffrey, P. F.; Ondov, J. M.; Thaung, K. C. (1996) Size distribution of Ir-tagged soot from diesel
      sanitation trucks. J. Aerosol Sci. 27(suppl.  1): S697-S698.
Tanner, R. L.; Parkhurst, W. J. (2000) Chemical composition of fine particles in the Tennessee Valley region. J. Air
      Waste Manage. Assoc. 50:  1299-1307.
Taylor, C. A., Jr.;  Stover, C. A.; Westerdahl, F. D. (1998)  Speciated fine particle (<2.5 um aerodynamic diameter)
      and vapor-phase acid concentrations in southern California. Presented at: 91st  annual meeting & exhibition of
      the Air & Waste Management Association; June; San Diego, CA. Pittsburgh, PA: Air & Waste Management
      Association; paper no. WA7401.
Taylor, P. E.; Flagan, R. C.; Valenta, R.; Glovsky, M. M. (2002) Release of allergens as respirable aerosols: a link
      between grass pollen and asthma. J. Allergy Clin. Immunol. 109: 51-56.


June 2003                                      3-110         DRAFT-DO NOT QUOTE OR CITE

-------
Thanukos, L. C.; Miller, T.; Mathai, C. V.; Reinholt, D.; Bennett, J. (1992) Intercomparison of PM10 samplers and
      source apportionment of ambient PM10 concentrations at Rillito, Arizona. In: Chow, J. C.; Ono, D. M, eds.
      PM10 standards and nontraditional paniculate source controls, an A&WMA/EPA international specialty
      conference, v. I; January; Scottsdale, AZ. Pittsburgh, PA: Air & Waste Management Association;
      pp. 244-261. (A&WMA transactions series no. 22).
Tobias, H. J.; Ziemann, P. J. (2000) Thermal desorption mass spectrometric analysis of organic aerosol formed from
      reactions of 1-tetradecene and O3 in the presence of alcohols and carboxylic acids. Environ. Sci. Technol.
      34:2105-2115.
Turpin, B. J.; Huntzicker, J. J. (1991) Secondary formation of organic aerosol in the Los Angeles basin: a descriptive
      analysis of organic and elemental carbon concentrations. Atmos.  Environ. Part A 25: 207-215.
Turpin, B. J.; Huntzicker, J. J. (1995) Identification of secondary organic aerosol episodes and quantitation of
      primary and secondary organic aerosol concentrations during SCAQS. Atmos. Environ. 29: 3527-3544.
Turpin, B. J.; Lim, H.-J. (2001) Species contributions to PM2 5 mass concentrations: revisiting common assumptions
      for estimating organic mass. Aerosol Sci. Technol. 35:  602-610.
Turpin, B. J.; Huntzicker, J. J.; Larson, S. M.; Cass, G. R. (1991) Los Angeles summer midday paniculate carbon:
      primary and secondary aerosol. Environ. Sci. Technol.  25: 1788-1793.
U.S. Environmental Protection Agency. (1995) Compilation of air pollutant emission factors. Volume I: stationary
      point and area sources. 5th ed. Research Triangle Park, NC: Office of Air Quality Planning and Standards;
      report no. AP-42-ED-5. Available from: NTIS, Springfield, VA;  PB95-196028.
U.S. Environmental Protection Agency. (1996) Air quality criteria for particulate matter. Research Triangle Park,
      NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
U.S. Environmental Protection Agency. (1999) Getting started: emissions inventory methods for PM-2.5. Research
      Triangle Park, NC: Office of Air Quality Planning and Standards, Emission Factor and Inventory Group; EIIP
      document series - volume IX. Available: http://www.epa.gov/ttn/chief/eiip/techreport/volume09/index.html
      [7 May 2003].
U.S. Environmental Protection Agency. (2000a) Aerometric Information Retrieval System (AIRS). Research
      Triangle Park, NC: Office of Air Quality Planning and Standards. Available: www.epa.gov/ttn/airs/airsaqs
      [7 May 2003].
U.S. Environmental Protection Agency. (2002b) Aerometric Information Retrieval System (AIRS). Research
      Triangle Park, NC: Office of Air Quality Planning and Standards. Available: www.epa.gov/ttn/airs/airsaqs
      [2 June 2003].
U.S. Environmental Protection Agency. (2000c) National air pollutant emission trends, 1900-1998. Research
      Triangle Park, NC: Office of Air Quality Planning and Standards; report no. EPA-454/R-00-002. Available:
      http://www.epa.gov/ttn/chief/trends/trends98/ [7 May 2003].
U.S. Environmental Protection Agency. (2000d) National air quality and emissions trends report, 1998. Research
      Triangle Park, NC: Office of Air Quality Planning and Standards; report no. EPA/454/R-00-003. Available
      from: NTIS, Springfield, VA; PB2000-104960.
U.S. Environmental Protection Agency. (2001) National air quality and  emissions  trends report, 1999. Research
      Triangle Park, NC: Office of Air Quality Planning and Standards; report no. EPA 454/R-01-004. Available:
      http://www.epa.gov/oar/aqtrnd99/ [7 May 2003].
U.S. Environmental Protection Agency. (2002) Latest findings on  national air quality: 2001 status and trends.
      Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and
      Standards; report no. EPA-454/K-02-001.
Vermette, S. J.; Williams, A. L.; Landsberger, S. (1992) PM10 source apportionment using local surface dust
      profiles: examples from Chicago. In: Chow, J.  C.; Ono, D. M., eds. PM10 standards and nontraditional
      particulate source controls, an A&WMA/EPA international specialty conference, v. I; January; Scottsdale,
      AZ. Pittsburgh, PA: Air & Waste Management Association; pp. 262-278. (A&WMA transactions series
      no. 22).
Wainman, T.; Zhang, J.; Weschler, C. J.; Lioy, P. J. (2000) Ozone and limonene in indoor air: a source of submicron
      particle exposure. Environ. Health Perspect. 108: 1139-1145.
Watson, J. G.; Chow, J. C. (2000) Reconciling urban fugitive  dust emissions inventory and ambient source
      contribution estimates: summary of current knowledge and  needed  research. Reno, NV: Desert Research
      Institute; DRI document no. 6110.4F.
Watson, J. G.; Chow, J. C. (2001) Estimating middle-, neighborhood-, and urban-scale contributions to elemental
      carbon in Mexico City with a rapid response aethalometer. J. Air Waste Manage. Assoc. 51: 1522-1528.
Watson, J. G.; Robinson, N. F.; Chow, J. C.; Henry, R. C.; Kim, B. M.;  Pace, T. G.; Meyer, E. L.; Nguyen, Q.
      (1990a) The USEPA/DRI chemical mass balance receptor model, CMB 7.0. Environ. Software 5: 38-49.


June 2003                                      3-111          DRAFT-DO NOT QUOTE OR CITE

-------
Watson, J. G.; Chow, J. C.; Pritchett, L. C.; Houck, J. A.; Ragazzi, R. A.; Burns, S. (1990b) Chemical source
      profiles for paniculate motor vehicle exhaust under cold and high altitude operating conditions. Sci. Total
      Environ. 93: 183-190.
Watson, J. G.; Chow, J. C.; Lu, Z.; Fujita, E. M; Lowenthal, D. H.; Lawson, D. R.; Ashbaugh, L. L. (1994)
      Chemical mass balance source apportionment of PM10 during the Southern California Air Quality Study.
      Aerosol Sci. Technol. 21:  1-36.
West, J. J.; Ansari, A. S.; Pandis, S. N. (1999) Marginal PM25: nonlinear aerosol mass response to sulfate  reductions
      in the eastern United States. J. Air Waste Manage. Assoc. 49:  1415-1424.
White, W. H. (1998) Statistical considerations in the interpretation of size-resolved paniculate mass data. J. Air
      Waste Manage. Assoc. 48: 454-458.
Wilson, W. E.; Stockburger, L. (1990) Diurnal variations in aerosol composition and concentration. In: Masuda, S.;
      Takahashi, K., eds. Aerosols: science, industry, health and environment: proceedings of the third international
      aerosol conference, v. 2; September; Kyoto, Japan. Oxford, United Kingdom: Pergamon Press; pp.  962-965.
Wilson, W. E.; Suh, H. H. (1997) Fine particles and coarse particles:  concentration relationships relevant to
      epidemiologic  studies. J. Air Waste Manage. Assoc. 47: 1238-1249.
Winer, A. M.; Arey, J.; Atkinson, R.; Aschmann, S. M.; Long, W. D.; Morrison, C. L.; Olszyk, D. M. (1992)
      Emission rates of organics from vegetation in California's Central Valley. Atmos. Environ. Part A 26:
      2647-2659.
Winterhalter, R.; Neeb, P.; Grossmann, D.; Kolloff, A.; Horie, O.; Moortgat, G. (2000) Products and mechanisms of
      the gas phase reaction of ozone with p-pinene. J. Atmos. Chem. 35: 165-197.
Wongphatarakul, V.;  Friedlander, S. K.; Pinto, J. P. (1998) A comparative study of PM25 ambient aerosol chemical
      databases. Environ. Sci. Technol. 32: 3926-3934.
Woo, K.-S.; Chen, D.-R.; Pui, D. Y. H.; McMurry, P. H. (2001) Measurement of Atlanta aerosol size distributions:
      observations of ultrafine particle events. Aerosol Sci. Technol. 34: 75-87.
Wotawa, G.; Trainer,  M. (2000) The influence of Canadian forest fires on pollutant concentrations in the United
      States. Science (Washington, DC) 288:  324-328.
Yu, J.; Cocker, D. R., Ill; Griffin, R. J.; Flagan, R. C.; Seinfeld, J. H. (1998) Gas-phase ozone oxidation of
      monoterpenes: gaseous and paniculate products. J. Atmos. Chem. 34: 207-258.
Yu, J.; Griffin, R. J.; Cocker, D. R., Ill; Flagan, R. C.; Seinfeld, J. H. (1999) Observation of gaseous and paniculate
      products of monoterpene oxidation in forest atmospheres. Geophys. Res. Lett. 26: 1145-1148.
Zhang, S.-H.; Shaw, M.; Seinfeld, J. H.; Flagan, R. C. (1992) Photochemical aerosol formation from a-pinene and
      P-pinene. J. Geophys. Res. [Atmos.] 97: 20,717-20,729.
Zweidinger, R.; Purdue, L.; Fitzgerald, K.; Carmichael, L.; Kellog, R. (1998) NERL PM research monitoring
      platforms: Baltimore, Fresno and Phoenix, data report. Research Triangle Park, NC: U.S. Environmental
      Protection Agency, National Exposure Research Laboratory; report no. EPA/600/R-98/138. Available from:
      NTIS, Springfield, VA; PB99-105660.
June 2003                                      3-112         DRAFT-DO NOT QUOTE OR CITE

-------
 i                                     APPENDIX 3A
 2
 3        Spatial and Temporal Variability of the Nationwide AIRS
 4                            PM2 5 and PM10_2 5 Data Sets
 7          Aspects of the spatial and temporal variability of 24-h average PM25 concentrations for
 8      1999, 2000, and 2001 in a number of metropolitan statistical areas (MSAs) across the United
 9      States are presented in this Appendix. PM2 5 data for multiple sites in 27 urban areas have been
10      obtained from the AIRS data base and analyzed for their seasonal variations, for their spatial
11      correlations, and for their spatial uniformity. A number of aspects of the spatial and temporal
12      variability of the PM25 data set from 1999 were presented in Rizzo and Pinto (2001) based in
13      part on analyses given in Fitz-Simons et al. (2000).  An analysis of the data obtained during the
14      first two years of operation of the Federal Reference Method PM2 5 network can be found in
15      Pinto et al. (2003).
16          Information about seasonal and spatial variability in PM25 concentrations within 27 MSAs
17      across the United States are provided in the accompanying figures (Figures 3A-1 to 3A-27).
18      Underneath the value for r, the 90th percentile values of the absolute difference in PM25
19      concentrations (in |ig/m3) and the coefficient of divergence (COD)  are given in parentheses.
20      Beneath these two measures of spatial variability, the number of observations used in the
21      calculations of the statistics in part c of each figure is given.
22          Quality assured measurements for at least fifteen days during each calendar quarter for
23      1999, 2000, and 2001 (preferably) or for 2000 and 2001 at a minimum of four monitoring  sites
24      in a given MSA were required for their inclusion in  the analyses given in this appendix. The
25      Columbia, SC and Baton Rouge, LA MSAs, which had only three sites meeting this criterion,
26      are exceptions.  Typically, at least 200 measurements were available for each monitoring site
27      chosen.  Monitoring sites were chosen without consideration of the land use type used to
28      characterize their locations.
29          Because of changes in monitoring strategies, funding levels etc., there were year to year
30      changes in monitoring sites meeting the above criteria in a number  of MSAs.  Data for the
31      Philadelphia, PA; Norfolk, VA; Pittsburgh, PA; Detroit, MI; Chicago, IL; Louisville,  KY;

        June 2003                               3A-1        DRAFT-DO NOT QUOTE OR CITE

-------
 1      St. Louis, MO; and the Dallas, TX MS As have been analyzed only for 2000 and 2001 because of
 2      a lack of consistent coverage in 1999.
 3           Information about seasonal and spatial variability in PM10_2 5 concentrations within
 4      17 MS As across the United States are provided in the accompanying figures (Figures 3A-28 to
 5      3A-44). Underneath the value for r, the 90th percentile values of the absolute difference in PM25
 6      concentrations (in  |ig/m3) and the coefficient of divergence (COD) are given in parentheses.
 7      Beneath these two measures of spatial variability, the number of observations used in the
 8      calculations of the statistics in part c of each figure is given.  In order to maximize coverage,
 9      data were calculated for a number of sampling periods.  Only Milwaukee, WI, and Salt Lake
10      City, UT, had enough data for a 3 year average (1999 to 2001). Tampa, FL; Cleveland, OH;
11      Steubenville, OH;  Baton Rouge, LA; Portland, OR; and Riverside, CA had data for a 2 year
12      average (2000 to 2001), as did Chicago, IL and Pittsburgh, PA (1999 to 2000).  Other MSAs had
13      only one year data (2000 or 2001).
14           The COD was defined mathematically and used earlier in Chapter 3 as a measure of the
15      degree of similarity between two data sets.  A COD of zero implies that values in both data sets
16      are identical, and a COD of one indicates that two data sets are completely different. Values of
17      P90 provide a measure in absolute terms of differences in concentrations between sites, and
18      CODs provide a relative measure of these differences. The maximum number of days of
19      coincident data from paired sites  were used to calculate correlation coefficients, values for P90,
20      and CODs.  The correlation coefficients were also calculated by using only concurrent
21      measurements obtained at all of the monitoring sites within urban areas meeting the above
22      selection criteria.  The correlation coefficients that were calculated differed only in the third
23      significant figure between the two methods.
24           Metrics used above for characterizing differences between separated monitors are applied
25      to collocated monitors in Table 3A-1.
26           Information about the spatial and temporal variability of 24-h average PM10_25
27      concentrations is summarized in Figures 3A-28 to 3A-44. Data are shown for a subset of MSAs
28      included in the analyses for PM2 5. Not all MSAs could be included because of a lack of data.
29      A schematic map showing locations of sampling sites within each MSA is given in part a, at the
30      top of each figure.  Also included in the map are major highways and a distance scale. A key
31      giving the AIRS site ID #'s is shown alongside each map. Box plots showing lowest, lower

        June 2003                                3 A-2        DRAFT-DO NOT QUOTE OR CITE

-------
 1      quartile, median, upper quartile and highest PM2 5 concentrations for each calendar quarter are
 2      shown in part b of each figure. AIRS site ID #'s, annual mean concentrations, the number of
 3      observations, and the standard deviation of the data are shown above the box plots.  Finally, in
 4      part c of each figure, statistics characterizing the spatial variability in PM2 5 concentrations are
 5      given. For each site-pair, the Pearson correlation coefficient (r) is provided. Underneath each
 6      value for r, the 90th percentile  of the absolute difference in PM10_25 concentrations, the COD,  and
 7      number  of observations is given.  In some cases, because of negative concentration values, the
 8      COD may not be calculated.  Dashes are shown for these cases.
 9
10
11      REFERENCES
12      Fitz-Simons, T. S.; Mathias, S.; Rizzo, M. (2000) Analyses of 1999 PM data for the PM NAAQS review. Research
13            Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards;
14            November 17. Available: http://www.epa.gov/oar/oaqps/pm25/analyses.html [2 April, 2002].
15      Pinto, J. P.; Lefohn, A. S.; Shadwick, D. S. (2003) Aspects of the spatial variability of PM25 concentrations within
16            urban areas of the United States.  J. Air Waste Manage. Assoc.: submitted.
17      Rizzo, M.; Pinto, J. P. (2001) Initial characterization of fine paniculate matter (PM25) collected by the National
18            Federal  Reference Monitoring Network. Presented at:  94th annual conference & exhibition of the Air &
19            Waste Management Association; June, Orlando, FL. Pittsburgh, PA: Air & Waste Management Association.
20
        June 2003                                   3 A-3        DRAFT-DO NOT QUOTE OR CITE

-------
                              Philadelphia, PA MSA
                a.
              c.
/*C H AIRS Site ID
jf B\ TL
S\ \\ jSw Site A
SiteB
X^y^Cj* \ site c
^CjgdPrv \ Site D
'jScV. >s. 1 Site E
< ^-\ X^X. I SiteF
l-Xxr^ s|teG

1 f !
0 50 S3GSW
A B C D E F
Mean
Qbs
SO
83 —
40 —
E

~U)
3
u>
•e"

°- *_




j-*~
MB ! 148 1+1 I 160 142 157
187 1 208 217 i 230 221 610
92 ; 88 8,5 | 86 89 92
i

t ' ! I
° ,
! I ' '1
' ' ' ' '

' 1 ' • ' ;
i ;
i i
lili lih Hi! 1 'in
(II p! i|H li i|ij i |l
,*'i ••' TIT
' , !
i i i ' !
i ' I
, • 1

1234 1234 1234 1 2 » 4 1234 » 2 3 4
Quarter
Site A 8 C D E F
A I (I'M OOJ QS7 OPS PM
,H I 014s .li ,*, (1 lE-i rhV (I 181 ('. U U tl't |4h 0 1!
IT-'i 1f T !BJ »7i> U 1
Q 1 0 H4 0 ISB U ftl '1 8li
(7 5 0 19i i~4 1) 18i 17 1 (I 18! .74017
t7o !'I4 1H4 It-1-
Q 1 0«5i Hd" MS*4
34-007-1007
34-015-5001
42-017-0012
42-045-0002
42-091-0013
42-101-0004
42-101-0136


G
f 15 S
816
| 93
i ' !





i

i
I
i
s.
j
i



















t 2 3 4

G
o9->
} Aid Mi
1%
oar-
1 i6n 0 181
i/2
Ob'.'
|TO 0 )<-,) ,4 7 n P| i49 « 11) lS3 0 1?j
W8 2W 177
n 1 cio" 004
1BO
0«
(f. 1 II 1r'l it t 11 111 14 0 I! 1><
201 18"
C 1 0 90
i4(j 0 11
181
F 1

G
IBS
087
l (4 ') 0 »-n
1B5
Iffi
]3J OUsi

Figure 3A-1.  Philadelphia, PA-NJ MSA.  (a) Locations of sampling sites by AIRS ID#;
             (b) Quarterly distribution of 24-h average PM2 5 concentrations for
             2000-2001; (c) Intersite correlation statistics, for each data pair, the
             correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
5A-4
DRAFT-DO NOT QUOTE OR CITE

-------
                              Washington, DC MSA
AIRS Site m
Site A
SiteB
SiteC
Site D
SlteE
SiteF
11-001-0041
11-001-0043
51-013-0020
51-059-0030
51-107-1005
54-003-0003
!
« Mean
Obs
SO
125 —
100—
n
| 75-
O)
in
| «.

25-
0—
1 I
S6 i«te^
A B C D E F
167 1S.5 14.7 14.1 13.8 16.1
859 926 337 848 324 314
9.8 87 8.0 7.4 8,2 9,3




,
' ' ' i
I ! , i \''.\ f>|1 '<:? f'l* "' I
                         1234  1234  1234  1234  1234  1234
                                         Quarter
                    Site    A
B
D
A


B


C


D


E


F
1 083 091
?5 1,018) (57.0 15)
784 273
1 0.84
(5 1 0 16)
2S7
1









082
(74,0 17!
891
087
(62,0 16)
731
096
(35,0081
276
1






084
(77.0 19)
267
079
(750 20}
284
094
(49, 0 11)
312
091
(45,011)
286
1



077
(86 020)
262
o?r
i76, 019)
28!
085
|8 3 0' 16)
289
082
|9 1 0 18)
248
087
|85 0 17)
284
1
Figure 3A-2.  Washington, DC MSA. (a) Locations of sampling sites by AIRS ID#;
             (b) Quarterly distribution of 24-h average PM2 5 concentrations for
             1999-2001; (c) Intersite correlation statistics, for each data pair, the
             correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
      5A-5
     DRAFT-DO NOT QUOTE OR CITE

-------
                                  Norfolk, VA
AIRS Site ID
Site A
Site B
SiteC
SiteD
SiteE
51-550-0012
51-650-0004
51-700-0013
51-710-0024
51-810-0008
                  b.
                  C.

Mean
Ote
SO
50-
40-

I "~
10
31 20—
a.
10—
0—


Site
A


B


C


D


E
ABC
13.6 13.S 12,8
S98 208 222
6.9 7Jt 6.7
[ ^
t
i
t
i ,

1
III ill i! !

1234 1234 1234
Quarter
ABC
1 0 94 0 91
(3"?, 009) (50 011:
16? 180
1 096
,26 009)
191
1






D E
13.7 13.C
3ZT 232
7.0 7,0





.
HI I!

1234 12C

D E
C 96 0 9'
;3 0 0 08) .3 5. 0
182 1fta
086 09
13 1 OO.'i (40 0
194 igj
0 96 0 9,
(3 6, 0 09) (4 3, 0
206 215
1 09
\3 6 0
215
1









[

4


i
OSj

3
09s

?
11)

3
10!


Figure 3A-3.  Norfolk, VA MSA. (a) Locations of sampling sites by AIRS ID#;
             (b) Quarterly distribution of 24-h average PM2 5 concentrations for
             2000-2001; (c) Intersite correlation statistics, for each data pair, the
             correlation coefficient, (P90, coefficient of divergence) and number of
             measurements are given.
June 2003
5A-6
DRAFT-DO NOT QUOTE OR CITE

-------
                                Columbia, SC MSA
            a.
                                                Site A
                                                SiteB
                                                SiteC
1 I
t »
Obs
SO
so—
50 —

40 —
e*>
£
a.
„ 30-
S
CL
20 —
10 —
o_
I
««tei
A
15
34
6.S


}





\
B
7 ! 14,7
J : 337
> : 8,8








III










I1
C
15.5
341
6.8








I










I
                                 1234   1234   1?34
                                          Quarter
C.    Site
                                   A
B
A


B


C
1 0.93
(33 008)
318
1



0.95
(32 007)
319
0.97
(2 8 0,06)
313
1
                                                          AIRS Site ID
                                       45-063-0008
                                       45-079-0007
                                       45-079-0019
Figure 3A-4.  Columbia, SC MSA. (a) Locations of sampling sites by AIRS ID#;
             (b) Quarterly distribution of 24-h average VM2S concentrations for
             1999-2001; (c) Intersite correlation statistics, for each data pair, the
             correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
                   5A-7
        DRAFT-DO NOT QUOTE OR CITE

-------
                                 Atlanta, 6A MSA
AIRS Site ID
Site A
SiteB
SiteC
SiteD
SiteE
SlteF
SiteG
13-063-0091
13-089-0002
13-089-2001
13-121-0032
13-121-0039
13-121-1001
13-223-0003
                                B
                  O)
                  S
                  a,
toad
Ote
SO

12S—
100—

75 —




,-
192 ' 183 . 188
31S , WO 918
94 P? 100









I!





I

1
J
^jlf
i





I
1




^

1

HI
187 !1,2 i
8S» 317
98 106 1

















1
i
s
s
i
'I1


1
1





1
184 i
M4
9.3 1









!!









il









il
j





!|
168
313
11 7









!!






J
                               234  1234  1134  1 J 3 4  1234
C. Site
A


B


C


n
Quarter
A B C D
i a ^7 o 8" o s:
;«.? f .u) ;i>4 o u ;» .- n IP,
256 ^73 2Sy
t 0 ,'i 1 B
;B4 f! 15. (7 1 n tfi;
'60 731)
1 0/i
i5 ! i" I3i
76"
1
E
07S
./:> u IM
267
Hfll
'9 ,1 (I 17;
26U
n (*/
{7 1 0 12;
^ra
IMI!
F
085
\?2 IS Ills
26U
O,'/
iao u u-ti
JliU
082
IT1) 0 17.
2BO
0?3
G
059
(11 i W1>
2, '2
Ob'
no1, r ??i
AS
u w
ilU i '^01
27!
O'i'i
                                               S 7, n t5j  ;7
                     E


                     F
              u 'f.
            110 8 0 IHi
                    (IV.
                    16 a i
                    ifiC

                     i
Figure 3A-5.  Atlanta, GA MSA. (a) Locations of sampling sites by AIRS ID#;
             (b) Quarterly distribution of 24-h average VM2S concentrations for
             1999-2001; (c) Intersite correlation statistics, for each data pair, the
             correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
5A-8
DRAFT-DO NOT QUOTE OR CITE

-------
                              Birmingham, AL MSA
a, j
xC

j 1 %,!<
Y*"^ _J^fi£w«
^ t^^ — **^

i f
Q 50
i*
D. Mean
Obs
SO
80 —
60 —
CO
i
O
a.
"~* 40 —
iq
Q.
20 —
0 —










I

1

C. Site
A


B


C


D


J AIRS Site ID
4^H Site A 01-073-0023
r^* 1 SiteB 01-073-1005
T ^J SiteC 01-073-2003
%CT? Site D 01-073-2008
]^ Site E 01-073-5002
/
1

100 fern
A B C D E
218
1056
11.1
{






ll
I'll
Ifl
1 i*

18,7
360
8,4


1

,


ill!

20.0
1046
10,3

|
\
j


j
i |
!'"i

17.6 16,6
346 356
8.5 8.2



}



1 1
li'i :i'i
; I T
1 i 1













234 1234 1234 1234 1234
Quarter
A B C D E
1 080 oee 079 oso
1145,020) (98,015) (137020) (15.2,021)
358 1011 342 353
1 079 088 086
(102 0 18) (780 1C) (6.7,0 15)
348 334 345
1 078 078
(99,0 181 (103 018!
335 344
1 086
(76,0.15)
329
Figure 3A-6. Birmingham, AL MSA. (a) Locations of sampling sites by AIRS ID#;
            (b) Quarterly distribution of 24-h average VM2S concentrations for
            1999-2001; (c) Intersite correlation statistics, for each data pair, the
            correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
5A-9
DRAFT-DO NOT QUOTE OR CITE

-------
                                  Tampa, FL MSA
                                              Site A
                                              SiteB
                                              SiteC
                                              SIteD
                                        B
                      S
                      £L
                               1234  1234  1234
                                          Quarter
                  C.
Site     A
B
                                                        AIRS Site ID
                                 12-057-0030
                                 12-057-1075
                                 12-103-0018
                                 12-103-1008
y©m
Ote
SD
10§— ,
SO —
60 —
40 —
20 —
o_





12,7
987
5.?



I!
*




,
i




I
12,3
100S
8,0



J
11.8
1024
6.0



III




1
11.1
337
6.0



lijl
                                                     1234
D
A


B


C


D
1 079
(3.6,0.10)
919
1






08?
(40,0 11)
920
070
(46, 0 14)
93S
1



087
(4 3, 0 12)
308
on
(50.013!
314
0.82
(3 1,0 10)
325
1
Figure 3A-7. Tampa, FL MSA.  (a) Locations of sampling sites by AIRS ID#; (b) Quarterly
             distribution of 24-h average PM25 concentrations for 1999-2001; (c) Intersite
             correlation statistics, for each data pair, the correlation coefficient, (P90,
             COD) and number of measurements are given.
June 2003
              3 A-10
            DRAFT-DO NOT QUOTE OR CITE

-------
                               Cleveland, OH MSA
AIRS Site ID
Site A
SiteB
SlteC
SiteD
SiteE
SiteF
SiteG
SiteH
39-035-0013
39-035-0038
39-035-0060
39-035-0065
39-035-0066
39-035-1002
39-085-1001
39-093-2003
                            B
                                               H
              E
              s
              Q-
felfS
OfJS
SD
60—
§0 —


40 —



fl-
ies
saa
98



'




III!
202
S31
115


j



1

Hi
184
S53
§9



i
!

•

"h
175 147 1
340 333 3
90 84 6


i
, .
) ' '•
I i
I ' i :
'Ml 1
i
''Mil1
50 140
31 342
2 64
i


,

'
1 i

w
* 152
298
ss



I
. '
*
i '

I1!1
1 '• «
                     1234  1 2 3 4
          c.
Site
1234  1SJ4
   Quarter

 D     E
                                                          34  5234
       G
A


B


C


D


E


F

G

H
1 II 91 OWS U*J4 i' V
\T \ C 11" M ~ 3 12' 15 » ll 'Oi (7 1 1 1fii
J20 -U2 4!4 JOO
i n'u tie-1 081
(h9 0 14, (940 15' 13 1 1)211
ire .jy-3 
-------
                               Pittsburgh, PA MSA
      C.












Maafi
Ote
SO

75 ~

1 -
g5~



Site
A
B
C
D
E
F
G
H
1
J
K
j.™™™™,
" C
®^JL__, ^ Site A
XTT^S V Site B
^"*^-7 SiteC
Jr-->®/L © / SiteD
=^&=^^^^^yr Site E
f**^***c^ ^r Site F
j c Site G

I I i Site 1
0 50 109M «., ,
Site J
SiteK
A B C D E F G
SM MT MS 3IF »1 1«T ill
»5 1 P3 1 "46 i S? 1 »5 1 » 1 07
i i
I ' * '
i , • ! i
' ', • * '
, ' ; j i i
i . * 1 ' ' ! ' i ' i '
1 f '
•li _ I 1 1 i 1' • * • i •••I illl! • 1 1
"T"1 Tf"?;1 l!iMMl|i:!|M,»f?'
1234 12S4 1234 1234 1 Z 3 4 1Z34 1Z3H
Quarter
A B C D t F G
1 096 0 fit OS" OS4 09? 09^
l,'6 00,"i ,159 PIS' (300071 (4",OG9i ^SOtCj )h<<008
171 ill! 16* 1ti,* 1,'.S 194
1 07? 09S 092 OUI OSS
H80 021, (34,00ft iGSOIQl n'00131 18(0(0
tSS 180 t71 179 1,C
1 ISO OS! 086 07?
i!65 020)1!" t 020MU5 0 15\ l'$ t,U?
•da iw (86 ta;
1 0 sC 0 E^1 0 94
(5? 0 (I> (600 til i57 0 11
(69 i7.' 131
1 C 91 00-)
t6i,0131 153 Oil
1«j 128
1 09'
i? 1,0 13
106
5





AIRS Site (D
42-003-0008
42-003-0021
42-003-0064
42-003-0116
42-003-1008
42-003-1301
42-007-0014
4?-1?*vflOO'5

42-125-0200
42-125-5001
42-129-0008
H t J K
I 1S« | 1S6 I «§ [ 180
\ ZH \ m 1 tat \ tvi
\ as I >2 1 r? 1 is
I
1 . '
' S
! *
| , . j
; h|-| -, ' ;M
. i i *•• ji*" 1 1 ^ i
'< "*t T III' •"" ll •
, f , T *•
1Z34 1234 1294 »234
H I j K
Oi« 093 091 0«7
t (bllJlH \SIOIOi 166 0 I* (54011.
183 185 531 17-!
09? 0*t OSM 08"
} ^42,00©i i48009j 
186 184 179 PS
i 094 oas ow
(4 f 009i 14 T 014< 1*7 (j )2,
2(9 ?01 200
( 0 94 05$
(56 012! S63 0 121
( OS4
(T 3 0 16)
195
t
Figure 3A-9.  Pittsburgh, PA MSA. (a) Locations of sampling sites by AIRS ID#;
             (b) Quarterly distribution of 24-h average PM2 5 concentrations for
             2000-2001; (c) Intersite correlation statistics, for each data pair, the
             correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
3A-12
DRAFT-DO NOT QUOTE OR CITE

-------
                                Stcubcnvillo, OH MSA
                                                           AIRS Site ID
                                                 Site A
                                                 SiteB
                                                 SiteC
                                                 SiteD
                                                 SiteE
                              39-081-0016
                              39-081-1001
                              54-009-0005
                              54-029-0011
                              54-029-1004
                                50km
                                      B
                        Mean
                         Obs
                          SD
                         80 -«
                     •3
                      
-------
                                     Detroit, MI MSA
                                                            AIRS Site ID
                                                  Site A
                                                  SiteB
                                                  SiteC
                                                  SiteO
                                                  SiteE
                                                  SiteF
                                                  SiteG
                                                  SiteH
                                                  Site!
                                                  SiteJ
                                             26-09i-OOOi
                                             26-115-0005
                                             26-125-0001
                                             26-147-0005
                                             26-163-0001
                                             26-163-0015
                                             26-163-0016
                                             26-163-0033
                                             26-163-0036
            b.
                  Ot>8
                  SD
                  80-4
               5E  SD-
               SL
                   0-
          IS 2isi    '41
          ?2»   188    514
          90   94    92
165
«37
92
     1ST
     853
100   94
146   19ft    17 ft
MS   t2S    ?18
9Z   106    10 T
                      !
            C.
    1234


Site  A
                                     1234 1234  1234  1 2 J 4  1234 1234
                                            Quarter
A


B


C


D


E


F


G


H


1


J
t OB9 085 Oa"5 094
)?f Ot8!(50 013t-54 0'4) 16 ' 0 16.
i10 1"2 20S 202
1 09s QS^ 09S
tss 0 14, is" n»i ise o io>
tW 190 TOO
; o sr i > a«
C64 0 'bt i**0 0 Hi
161 '6fi
t C89
•36 0 I9i
*^?
1















09;
(95 019)
21s-
091
|75 0 IS)
214
093
i?5 016>
! 74
069
HO 1 02li
203
096
<46 009,
210
1












095
iSB 0 Ml
202
n 91
isa o 121
j?C^f
090
093
ISO OISi
JOS
091
i69,0 I3i
"'Ol
094
143 0 10; (500 in
tea
093
|75 OO
*93
094
(46 0 lOi
SSQ
09"
l46 0 ID
210
1









164
090
i6D OtSi
190
084
i!>9 Otji
200
090
i9 1 Oir,
'str
094
(64 Pt?)
?04
1






089
it? 9 02JI
;os
OSS
111 9 0 !91
?02
(191
111 " 0 101
IK
084
.(58 0»l
193
OSS
ttOB 0 1b)
195
090
18 7 0 13)
210
OSM
it04 0 Ifil
201
O.S6
((28 021)
IS"
1



oso
I'J 6 0 t8i
;co
090
iT 01*
''34
OSS
183 0 Ki
tb7
os:
(il 8 0221
l<»
OSS
600 10)
;S2
091
r«o or,
204
081
!™ 1 Q 12*
IS
087
i»r o t7>
191
080
lT7 C 13;
ISi
1
Figure 3A-11.  Detroit MI MSA. (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average VM2S concentrations for
               2000-2001; (c) Intersite correlation statistics, for each data pair, the
               correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
                         3A-14
            DRAFT-DO NOT QUOTE OR CITE

-------
          a.
                                      Rapids, Ml MSA
                                                          AIRS Site ID
                                                Site A
                                                SIteB
                                                SlteC
                                                SiteD
                            26-005-0003
                            26-081-0020
                            26-121-0040
                            26-139-0005
                                       B
                         0—i
                                                      1234
                       Site
A
B
D
A


B


C


D
1 0,94
(5,7, 0,15)
9S8
1






0,93
(44,013)
357
093
<6 1,0 16)
359
1



095
(4.4, 0 12)
337
098
(3 1,0 10)
332
094
(52. 0 13)
324
1
Figure 3A-12. Grand Rapids, MI MSA. (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average VM2S concentrations for
              1999-2001; (c) Intersite correlation statistics, for each data pair, the
              correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
        3 A-15
            DRAFT-DO NOT QUOTE OR CITE

-------
                                           Wl MSA

.




==aa=i^=
€


0
fa
JU« Mean
OH
SO
1QQ-
80—
C 60-
1
IO
ss* 4B~
0.
20-
0—


C. Site
A


B


C


D


E


F


G

H


1 £=f
W Site A
j| Site B
	 "m Site C
Jk SIte °
Site E
' i^ «S sttefr
1 J SiteG

50 tOStan
A B C D E F
140 135 '144 1J1 141 141
tOSS 1012 328 3SJ 342 3S6
S9 »$ . 90 84 B2 91
,
,
' • ' • " ' ' , ' i I
! • ; I , i
; ' ' ' , i |
t * " l ' i
l|ii SHI 'is! in M :!in

1234 1Z34 1234 1234 1134 tJ34
Quarter
A B C D E F
1 JU1 OH4 It 06 tl'li' DK,
.?i Jin v*»ui..i r.tiott ,3 MO io) 114 omtj
VT4 J1s? jj 3,'.L* J4>>
1 0 95 U IS 0 96 0 U7
," ( ' V i11 •! l«;i i,\ 1 U tit i '5 'I IHl
-11 ">!5 K4 3j7
1 o«i4 OHB 097
4-> 0 13i C. a OOS'i iJ4 0 OPi
"15 •>« j'rt
1 I) 4ti LI W-.
09 0 12; (450 12i
Jut jt'i
t DM,
i>fi DUB)
019
1







AIRS Site ID
55-079-0010
55-079-0026
55-079-0043
55-079-0051
55-079-0059
55-079-0099
55-133-0027
OO-l4a-utW4

G H
138 131
811 34i
83 83

I .
! '
'

III! in;

1234 1234

G H
U'J( OB.'
U, 3 li 10 ^4 I) 11)
:'&> 438
ft * OS1
i , h ittr^ ^ h n H
M'l jji'
It 53 0 141
IIS Ki>
nt' n V
13 "5 0 IP) .IB 0 T,
11 J JIB
u yt> n u~
it) 4, 0 O^s e * 0 01^1
32C j!?
U 97 0 92
Hi' II Id) (Ml (I 12>
3JS 33"
1 no:
1-8 n 101
1
Figure 3A-13.  Milwaukee, WI MSA. (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average VM2S concentrations for
              1999-2001; (c) Intersite correlation statistics, for each data pair, the
              correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
3 A-16
DRAFT-DO NOT QUOTE OR CITE

-------
                                     Chicago, IL MSA
                                                                   AIRS Site ID
                        so
                              100km
                                                        Site A
                                                        SiteB
                                                        SiteC
                                                        SiteD
                                                        SiteE
                                                        SiteF
                                                        SiteG
                                                        SiteH
                                                        Site I
                                                        SiteJ
                                                        SiteK
                                                         17-031-0022
                                                         17-031-0052
                                                         17431-0057
                                                         17-031-0076
                                                         17-031-1016
                                                         17-031-2001
                                                         17-031-3301
                                                         17-031-4201
                                                         17-043-4002
                                                         17-089-0003
                                                         17-197-1002
                                                                         K
                ttean
                 Ob*
                 SO
              01
           17*
           22i
           85
188
87 1
9S
18?
22«
183
220
20,8
2Z4
100
170
222
81
(66
233
SO
145
63S
85
164
22S
81
147
223
S3
180
22S
83

                              ijriii 'il

                                      ii
                               Jll,,
                               UP  '
                    1294 1234  1234  1234  1234  1234 < 2 3 4 1 S 3 4  (234  1234  '234
                                             Quarter
C«    Site   A     B
                                                         H     i
                 A

                 B


                 C

                 D

                 E

                 F

                 G


                 H




                 J

                 K
                        > M <• i. s^i «f a i"i I'n i n, rt i o 121 i"j * u 'pi ro c iu ri " »t
                          .OJ   >'     .'(>'   ^"o   '»>    ' >r    '7*1
                                             «' ' n -,i (•!»  11 ft > n Ml
Figure 3A-14. Chicago, IL MSA.  (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average PM2 5 concentrations for
               2000-2001; (c) Intersite correlation statistics, for each data pair, the
               correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
                                 3 A-17
                              DRAFT-DO NOT QUOTE OR CITE

-------
          a.
                                     Gary. IN MSA
                  b.
                         Mean
                          Qte
                          SO
                          80—i
                          60 —
                     "B»
                     s
                     Q.
                          40 —
                          20 —
15.7

8,S

                                I
                                 I
                              1234
C.   Site
                                A
                                                            AIRS Site ID
                                                  Site A
                                                  SiteB
                                                  SiteC
                                                  SiteD
                              18-089-0006
                              18-089-0022
                              18-089-1016
                              18-127-0024
                                         B
17.6
827
10,1
                               16.2
                               965
                               8.3
14.0
312
8.0
                                         734   1234   1234
                                            Quarter
         B
                  D
A
B
C
D
1 065
(9 1,0.19)
823
1


0,93
(42, 013)
822
0.59
(11.3,020)
841
1

091
(50,0.1?)
277
0,56
(10.9,023)
277
092
(6.4,0.17)
281
1
Figure 3A-15.  Gary, IN MSA.  (a) Locations of sampling sites by AIRS ID#; (b) Quarterly
               distribution of 24-h average PM25 concentrations for 1999-2001;
               (c) Intersite correlation statistics, for each data pair, the correlation
               coefficient, (P90, COD) and number of measurements are given.
June 2003
         3 A-18
             DRAFT-DO NOT QUOTE OR CITE

-------
                                Louisville, KY MSA
                 a.
AIRS Site ID
Site A
SiteB
SiteC
SiteD
SiteE
18-019-0005
18-043-1004
21-029-0006
21-111-0044
21-111-0048
                        I
                        50
                  C.
                                    B
Mean
Obs
SD
80-
80-
1
0»
a.
"~Z, 40-
01
S
Q.
20-
0—


Site
A


B


C


D


E
17,4 1S.7 18.1 17,2
331 296 323 1011
8.? S.3 7.8 8,8







•
1 .'
I!1!!!"-! '!!'!
f 1 H'S-i i 1" 1

1234 1234 1234 1234
Quarter
A B C D
1 t)St> 040 091
153 0 13t (600 121 (39 0 HI)
27» 28!) 315
1 0 87 0 89
(600 IS) (590 14|
260 282
1 OP3
(54 0 12)
3or
i



17.0
315
8,5

l






ill.

1234

E
GM
(38, 003)
273
091
(53,0 13)
250
090
(56 0 12|
266
090
(42 011)
302
1
Figure 3A-16.  Louisville, KY MSA.  (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average PM2 5 concentrations for
              2000-2001; (c) Inter site correlation statistics, for each data pair, the
              correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
3 A-19
DRAFT-DO NOT QUOTE OR CITE

-------
                                    St. Louis, MO MSA
      c.
                                                                 AIRS Site ID
                                                      Site A
                                                      Site 6
                                                      SiteC
                                                      SiteD
                                                      Site E
                                                      SiteF
                                                      SiteG
                                                      SiteH
                                                      Site!
                                                      SiteJ
                                                      SiteK
                                                      17-119-0023
                                                      17-119-1007
                                                      17-119-2009
                                                      17-119-3007
                                                      29-099-0012
                                                      29-183-1002
                                                      29-189-2003
                                                      29-189-5001
                                                      29-510-0085
                                                      29-510-0086
                                                      29-510-0087
           3E  »-
           DL
                  ?QJ  I  173
                  2Z6    230
                  t;  I  T,?.
               _C    D  _ _E    JF    G     H     I     J     K
                is » "T'tss"  ; " tij  T~"«s~" T"" 145r'p's" T" i" T'wi"
                ZM    2§4   '  239    238    236    239    719    681
                7.9    7J   :  «J>   .. 7B  .  7i   .  7.3.  1 ?t  J  .7.*...
Site
                 1234  1234  12S4  1234  123H  1234  1J34
                                                                 34  1234 1434
A
B
C
D
E
K
G
H
I
08! 0 "3 0 "4
|»8 0 tSi 
20? (93 JOT
I 0 ofr 0 «6
(88 P12, 16 7 Olli
!9o .'10
i 036
i4 5 0 13)
195
!





067
218
084
r j oi4i
^17
3S«
>59 nili
08J

22 i
O.W
(67 0 Hi
08t
ISZ 0 til
086
(5J Ot4|
095
(37 008,
234
003
iSt 0 Hi
213
090
(3 / 0 1 1 1
!31
095
14 J 0 10)
234
1
089
II50 023,
I~S 0151
091
|54 012}
0*7
(SS U 14)
218
095
|3S 0091
218
14 7 0 101
318
090
(3 1 0101
215
035
(29,0091
osa
!?? 00,1
068
21"
tU'b
tS5 0 .21
ISO Bill
OS4
tSS OMi
225
,193
U^l OO&f
2»
091
04 0 12)
?29
• S3 OC)
227
230
ai o'osi
                                                                            S9T
                                                                            096
                                                                           33 OC8)

                                                                            t
Figure 3A-17. St. Louis, MO MSA. (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average PM2 5 concentrations for
               2000-2001; (c) Intersite correlation statistics, for each data pair, the
               correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
                             3A-20
DRAFT-DO NOT QUOTE OR CITE

-------
                                      Rouge, LA MSA
                      b.
                            Mean
                            Ota
                             SO
                            SO—I
                       *»
                        fi
                       s
                       Q.
                             30 —
                             0—1
14,5
331
7.1
                       C,   Site
A
                                                             AIRS Site ID
                                                   Site A
                                                   SiteB
                                                   SifeC
                           22-033-0002
                           22-033-0009
                           22-121-0001
                                             B
14.5
1067
6.6
B
14.1
1031
6,6
                                 1234    1234   1234
                                            Quarter
A


B


C
1 093
(2.7, 0.08)
326
1



093
(2.9, 0 09)
318
097
(2 5, 0 07)
1006
1
Figure 3A-18. Baton Rouge, LA MSA.  (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average VM2S concentrations for
              1999-2001; (c) Intersite correlation statistics, for each data pair, the
              correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
     3A-21
        DRAFT-DO NOT QUOTE OR CITE

-------
         a.
                                      City, MO MSA

	
A in
"^ jj •jjjj*. 4f"%ll\,1

5 Site ID
"vAv J& *, Site A 20-091-0008
i*"^bw^>-rz/*' :"~~*^ ^'^e ^ 20-091-0009
^^^^^^2==^^^^,^^ Site C 29-047-0005
jrp) one u £a \
f^f__ 	 Site E 29-(
)47-0041
^ • SiteF 29-095-2002
'-i - - 	


-^_-J
1 i i
0 SO 163 Km
0. Mean

SO
SO —
40 —


*****
C*3
E
"3)
to
s x
a.
10 —

0_
A B C D E F
11? 1"4 116 129 1?4 131
330 336 341 1033 988 354
8,0 60 92 64 82 67
s
«
I j
I , i !
! 'ill.
1 ' 1
i i
' ! ' , | 1 i
I ' ! * !
' i l ,X
|,| 'I "1 'lll'll |
I i . T ' f i ' | '
1 i i '
1
















1
i
i


















1234 1234 1234 1234 1234 1234
Quarter
C. Site A B C D E F
A 1 OM 0*>2 OaO 093 Ut?t
tlo.OOPl (42, n i;, (43013) I3B010! (59,0


!•>)
320 3C4 318 2% 420
g i 090 089 090 087
(400 1di (4 1 0 14) i.l 6 0 121 16^0
312 322 300 328
1/1

r 1 096 096 091}
|3 ! 0091 f28 0 101 (550
327 300 329
>6)

n i 095 095
(29.0,09) 140,0,111
940 338
E 1 094
151,0
314
F 1
12,


Figure 3A-19. Kansas City, KS-MO MSA. (a) Locations of sampling sites by AIRS ID#;
             (b) Quarterly distribution of 24-h average PM2 5 concentrations for
             1999-2001; (c) Intersite correlation statistics, for each data pair, the
             correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
3A-22
DRAFT-DO NOT QUOTE OR CITE

-------
           a.
                                              TX MSA
                                                      Site A
                                                      SifeB
                                                      SiteC
                                                      SiteE
                                                      SiteF
                                                      SiteG
                                                                 AIRS Site ID
                                             48-085-0005
                                             48-113-0020
                                             48-113-0035
                                             48-113-0057
                                             48-113-0069
                                             48-113-0087
                                  B
                    Moan
                     Ob*
                     SD
                 3L

                  W



                 Q.
     11.5
     ?34
     SB
124     12.9
V?     ?J2
5?     S8
133     13.?
«44     JS15
58     81
12:7
687
S,7
11.7
218
5.4
                         1234  1 2 3 «  1 2 3 <  1234  « a 3 4  1  J 4 4  1  2 J 4
                                              Quarter
                    Site    A
             B
               D
A


B




D


E

F


G
                                 P?2    PSi4    094    'Jflt
                                (35.0'ti  iJOUMt  i'1301.«  In i  ft ('•
                                       (32 I1
                                               Q 34
                                                ocn
                             1M     C £«,
                             0 ,01  I < i 0 10'i
                             ?1S     111
                                                                     ,'JT
                                               i) !1;'     S> ;• i     a ,>     i'' a
                                              (aoocei  i? co osi  rsoopi  i1 en io>
                                               2l*i      t .t     ,J/      !S5
                                                             OX    U6t
                                                            I > I  (Mi!!l  ti 'a (i <4l
                                                                    C 36
                                                                   IJ J UtWi
Figure 3A-20. Dallas, TX MSA. (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average VM2S concentrations for
               2000-2001; (c) Intersite correlation statistics, for each data pair, the
               correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
                       3A-23
                        DRAFT-DO NOT QUOTE OR CITE

-------
                                          ID MSA
         a.
AIRS Site ID
Site A
SiteB
SlteC
SiteD
16-001-0011
16-001-0017
16-027-0004
16-027-0005
                          so
                                    100k
                                        B
                         Obs
                         SO
                         BO-
                    "m
                         20 —
                          o—4
        9?
        602
        9,0
        8.7
        318
        7.6
        9.8
        358
        9.1
10.3
328
9.4
      III  I
                              1234   1234   1234
                                           Quarter
                c.
Site
A
B
 D
A


B


C


D
1 091 095
(4,3 0 16) (3 8, 0 13}
304 344
1 085
(60,0 19)
313
1



082
(5 1 0,16}
315
079
(88,023}
317
096
(390 12)
322
1
Figure 3A-21. Boise, ID MSA.  (a) Locations of sampling sites by AIRS ID#; (b) Quarterly
              distribution of 24-h average PM25 concentrations for 1999-2001; (c)
              Intersite correlation statistics, for each data pair, the correlation coefficient,
              (P90, COD) and number of measurements are given.
June 2003
                3A-24
                     DRAFT-DO NOT QUOTE OR CITE

-------
                               Salt Lake City, UT MSA
AIRS Sit© ID
Site A
SiteB
SiteC
SiteD
SiteE
SiteF
49-011-0001
49-035-0003
49-035-0012
49-035-3006
49-035-3007
49-057-0007
                               I
                               100 i,T
                           A
             B
             D
                      so
                      100-1
                   o>
                  S
                  ft.
                      SO-
                      80-
                      40
                       0-
                           S,0
                           355
                           9.2
             121
             349
             122
      136
      346
      11?
11,3
9B3
117
118
333
114
               c.
Site    A
B
 D
88
328
74
                          1234  1234  1234  1234   1234  1Z34
                                          Quarter
A


B


C


D


E


F
1 0 89 0 90
(880 22; (990 27)
33-* 331
1 033
(7 0 0 19)
327
1









064
(690 19)
322
092
(49 0 17|
316
093
(71 01&I
314
1






090
(82 022)
315
089
I fi 1 0 17!
313
096
(48 017)
308
094
if 3 0 161
300
1



094
(44,0 !5t
306
088
(9 3, 0 20l
307
086
(114 0 24)
302
092
(870 18j
297
089
(08 020)
288
1
Figure 3A-22. Salt Lake City, UT MSA. (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average PM2 5 concentrations for
              1999-2001; (c) Intersite correlation statistics, for each data pair, the
              correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
                   3A-25
                  DRAFT-DO NOT QUOTE OR CITE

-------
         a.
                                  Seattle, WA MSA
                                                           AIRS Site ID
                                                 Site A
                                                 Site 6
                                                 SiteC
                                                 SiteD
                                                 SiteE
                     53-033-0017
                     53-033-0021
                     53-033-0057
                     53-033-0080
                     53-061-1007
                                     &
D. Mean
Ot»
SO
80 —
50-
40 	
W
1
01
3
«o BO-
S'
0.
20-
10 —
0 —

6,8
10,9 • 11.9 ] 8.9
316










I
!


3,6









\











1

1057 : 1054 ! 808
7.3 : 70









I









|l









1

J 57









|









!l|

11.4
I 357
I 8,6









1









1










„
'1










1

234 1234 1234 1234 1234
Quarter
C, Site
A

A
1
BCD
031 028 045
E

0.37
(150.0.39) (15.7,0.431 (88,0.32) (179,039)
303 304 291
B

1 0 96 0 92
298
081
(36 014) (68,0.181 (6,2,0 17)
1021 774
C


D



1
091
344
0,79
(7.5, 0.20! (8,2, 0.20)






779
1
344
075
(8,5,020)
327
E





1

Figure 3A-23. Seattle, WA MSA. (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average PM2 5 concentrations for
              1999-2001; (c) Intersite correlation statistics, for each data pair, the
              correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
3A-26
DRAFT-DO NOT QUOTE OR CITE

-------
                                 Portland, OR MSA
             a.
AIRS Site ID
Site A
Site B
SiteC
SiteD
41-009-0004
41-051-0080
41-051-0244
41-067-0111
                                        B
                          Mean
                           O&S
                           SO
                      Z
                      Q.
                          100 —
                           76
                           50
                           25*
                           0—
       8.3
       487
       4.4
9.1
1042
7.2
8,7      7.3
1037   ,'   503
5.4   i   5.9
                               1234  1234   1234   1234
                                          Quarter
                 C,
Site    A
B
        D
A


B


C


D
1 0,79
(6 5, 0 24}
433
1






0 90 0,81
(5.1 0.20) (45,0.19)
429 427
077 089
14,1,0.14) (4.3,0.17)
986 477
1 083
(4.5, 0 19)
472
1
Figure 3A-24. Portland, OR MSA.  (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average VM2S concentrations for
              1999-2001; (c) Intersite correlation statistics, for each data pair, the
              correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
               3A-27
             DRAFT-DO NOT QUOTE OR CITE

-------
           a.
                               Los Angeles, CA MSA
AIRS Site ID
Site A
SiteS
SiteC
SiteD
SiteE
SiteF
06-037-0002
06-037-1103
06-037-1201
06-037-1301
06-037-2005
06-037-4002
             C.
                                 B
                   Mean
                    Obs
                    SO
                    100-
                 rn

                a.
                    75—
                    so-
                    25 —
                     0 —
      20.9
      641
      133
22.S
656
13.5
18,3
217
11.7
237
237
134
20.1
220
11 i
203
621
124
                                   11!  l

                        1234  1734   1234   1234  1234
                                          Quarter
                                                            1234
Site    A
 B
        D
A


B


C


D


E


F
t 087 OTb 068
HOT 018) (14 R 021i (171,025)
581 208 >29
1 Q 86 0 89
i'28 02tn (10 1.0 12)
205 222
1 076
(18 1, 024)
212
1






095
(82,0 14)
212
0 93
{? 1,011)
20?
085
[12 1,0 18)
197
078
(132,0 181
214
1



080
(18 1,026)
553
080
(13,t> 0 17)
S63
066
1182,024)
197
095
(81 0 111
216
082
(18,0201
204
1
Figure 3A-25. Los Angeles-Long Beach, CA MSA. (a) Locations of sampling sites by
              AIRS ID; (b) Quarterly distribution of 24-h average VM2S concentrations
              for 1999-2001; (c) Intersite correlation statistics, for each data pair, the
              correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
                    3A-28
                   DRAFT-DO NOT QUOTE OR CITE

-------
                                 Riverside. CA MSA
AIRS Site ID
Site A
SiteB
SiteC
SiteD
SiteE
06-065-1003
06-065-8001
06-071-0025
06-071-2002
06-071-9004
                b.
                             A
             B
                    o>
                     to
                     oj
                    a.
Mean
Obs
SO
60-
60-


40-


20-
0-








1
26.9 30.0
327 766
16.1 18.0


|
j
j
i

Hi! |l:l








i,
25,4
320
14.6







r
25.0
34?
14.9



i


I
i
III"
• llTI
! t
25.7
307
18.4



|
. i
f '
I
H'l
I
t
                             234   1234   1234  1234  123
                                          Quarter
                c.
Site   A
B
D
A
B
C
D
E
1 0 94 0 83
(66.0 101 (14,3,021*
294 289
1 081
(17 6, 0,23!
29Q
1


0,93
1106,0 13)
308
093
|133,0 14)
313
086
(11 8,020i
302
1

090
{105 013}
278
091
(11 S. 0 131
275
078
069 022)
26S
094
(89, 0 11)
260
1
Figure 3A-26.  Riverside-San Bernadino, CA MSA.  (a) Locations of sampling sites by
              AIRS ID#; (b) Quarterly distribution of 24-h average VM25 concentrations
              for 1999-2001; (c) Intersite correlation statistics, for each data pair, the
              correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
               3A-29
               DRAFT-DO NOT QUOTE OR CITE

-------
                                San Diego, CA MSA
         a,
AIRS Site ID
Site A
Site B
SiteC
Site D
06-073-0001
06-073-0003
06-073-1002
06-073-1007
                           50
                                   100 ten
                 ft.
                B
                                                       D
                        Maan
                         Ote
                         SO
                         80 —
                     «

                     w
                      CM
                     £
                     a.
                         20-
                          0 —
       14,6
       313
       7.2
16,5
938
B.I
17,
886
9,2
18,8
879
9.3

                             1234   1234   1234
                                          Quarter
                                                      1234
                 C,
Site
B
         D
A


B


C


D
1 0-76 0.73
(10.0,0.18) (10.0.0.19)
270 253
1 0.85
(6.3 0.13}
773
1



0.83
(7,6. 0.18)
255
0.78
(9.7, 0 18)
769
0.73
(11,0,0.20)
728
1
Figure 3A-27. San Diego, CA MSA. (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average PM2 5 concentrations for
              1999-2001; (c) Inter site correlation statistics, for each data pair, the
              correlation coefficient, (P90, COD) and number of measurements are given.
June 2003
                3A-30
             DRAFT-DO NOT QUOTE OR CITE

-------
            TABLE 3A-1. PERFORMANCE METRICS FOR PM2 5 FROM
     COLLOCATED SAMPLERS (Mean concentrations, the number of samples, the
      standard deviation at each site, and the Pearson correlation coefficient and the
  coefficient of divergence for the site pair are shown; concentrations are given in ug/m3).
Columbia, SC

MEAN
N
SD
r
COD
Dallas, TX

MEAN
N
SD
r
COD
Detroit, MI

MEAN
N
SD
r
COD
Grand Rapids, MI

MEAN
N
SD
r
COD
Louisville, KY

MEAN
N
SD
r
COD
Steubenville, OH

Mean
Obs
SD
r
COD
Washington DC

Mean
Obs
SD
r
COD
Sampler 1
45-079-0019
15.3
226
6.5


Sampler 1
48-113-0069
12.7
687
5.7


Sampler 1
26-163-0001
16.5
637
9.2


Sampler 1
26-081-0020
14.0
1050
8.6


Sampler 1
18-043-1004
15.8
196
8.2


Sampler 1
54-029-0011
16.5
342
10.2


Sampler 1
11-001-0041
16.8
600
9.7







0.995
0.022





0.996
0.032





0.986
0.041





0.992
0.059





0.997
0.027





0.985
0.087





0.953
0.150
Sampler 2
45-079-0019
15.5
211
6.5


Sampler 2
48-113-0069
13.1
116
6.0


Sampler 2
26-163-0001
16.2
111
8.8


Sampler 2
26-081-0020
14.3
181
8.3


Sampler 2
18-043-1004
16.0
104
7.8


Sampler 2
54-029-0011
16.6
325
10.3


Sampler 2
11-001-0041
17.5
132
11.1


June 2003
3A-31
DRAFT-DO NOT QUOTE OR CITE

-------
                                 Columbia, SC MSA
                               Mean
                                Gbs
                                 SO

                                 40 H
                                 30-

                                 *"
                                 ,.
                                  0-
                                -10-
                                                 Site A
                                                 SiteB
                                       A
          B
7.4
56
4.3
9,6
53
6.2
                                     1234   1234
                                           Quarter
                              Site     A
          B
A


B
1 0,70
(8,0. 0.3?)
49
1
                                                           AIRS Site ID
                     45-063-0005
                     45-079-0019
Figure 3A-28. Columbia, SC MSA. (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
              for 2000; (c) Intersite correlation coefficients, number of observations,
              90th percentile differences in concentrations, and CODs.
June 2003
3A-32
   DRAFT-DO NOT QUOTE OR CITE

-------
                                   Tampa, FL MSA
                                               Site A
                                               SiteB
                                                          AIRS Site ID
                            12-057-0030
                            12-103-0018
                                                  B
                                Mean
                                 Obs
                                 SO
                                  so-1
                                  40—
                       c.
                                  30-,
                              0.   20-
                                  10-
                                  0-4
         11.3
         112
10.1
104
5.2
                                                      i
                                      1234   1234
                                            Quarter
Site
 B
                                A
                                 B
                   0,81
                 (5,3, 0.17)
                    95

                    1
Figure 3A-29.  Tampa, FL MSA (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
               for 2000-2001; (c) Intersite correlation coefficients, number of observations,
               90th percentile differences in concentrations, and CODs.
June 2003
         3A-33
   DRAFT-DO NOT QUOTE OR CITE

-------
                                   Cleveland, OH MSA
                  O.
                  C.
                        Mwn
                        Obs
                         SO
                      OS
                      0.
      284
      216
      17,4
                                                       Site A
                                                       SiteB
                                                       SiteD
                                                       SiteE
                                                       SiteF
                                                   D
18.6
814
11.6
16.8
112
9,0
21.3
MS
16,4
18.8
111
i-5
                                                    =
                                                                   AIRS Site ED
                                             39-035-0013
                                             39-035-0038
                                             39-035-0045
                                             39-035-0060
                                             39-035-0065
                                             39-085-1001
7,2
109
4.9
                            1234  1234  1234  1234 1234  J234
                                              Quarter
Site    A      B      C      D
                         A      1


                         R


                         r>


                         n


                         p



                         F
                                                                 o si
              18?

              1
        95

       065
                            073
                          *!" (1 1&
                            b.<
                      1      081'     D <*1     040
                          C« 1 n MI (!"«• 'i jn 1169 ,' ^'"J>
                            04      10!     P9
                             1
                                   o "4     p ji
                                 
-------
                                Stcubenvillc, OH MSA
           a.
                                                 Site A
                                                 SiteB
                                                 SfteC
                                                 SiteD
                 Dm
SO ton

A
                                          B
 Mean
  Obs
  SO
  100-

                           75 -
                           »-
                        *?
                        o
12.7
10?
8,6
14.3
190
10,9
10.2
211
6.3
                                        1234  123
                                             Quarter
                 C.
Site
          B
                                                            AIRS Site ID
                                     39-081-0016
                                     39-081-1001
                                     54-009-0005
                                     54-029-1004
                                   D
13.0
114
10.4
                  D
A


B


C


D
1 0.84 0.69
(10.8,0.77) (11.6.-)
83 91
1 0.54
(14,7, -)
188
1



068
(11 3,-)
100
0.48
(185,-)
88
089
(12 8 ~)
9?
1
Figure 3A-31.  Steubenville, OH MSA (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
               for 2000-2001; (c) Intersite correlation coefficients, number of observations,
               90th percentile differences in concentrations, and CODs.
June 2003
                 3A-35
                      DRAFT-DO NOT QUOTE OR CITE

-------
                                    Detroit, MI MSA
                                                           AIRS Site ID
                                                Site A
                                                SiteB
                                                SiteC
                        26-163-0001
                        26-163-0015
                        26-163-0025
Jb.
                                              B
                             Mean
                              Obs
                              SO
                          w
                           o
                          af
                               80-
                               40-
                               20-
                                0-
                              -ao-
11.5
56
10.3
19.4
58
15,6
 7.3
 56
 7,6
                                   1234  1234  1234

                                            Quarter
                       C,    Site    A
          B
A
B
C
1 058
(29,4, -}
53
1

0,54
(15,7 0.54)
50
0.39
(34,9, 0.79)
51
1
Figure 3A-32.  Detroit, MI MSA (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
               for 2000; (c) Intersite correlation coefficients, number of observations,
               90th percentile differences in concentrations, and CODs.
June 2003
    3A-36
DRAFT-DO NOT QUOTE OR CITE

-------
                        C,
                                 Milwaukee, Wl MSA
                                                  Site A
                                                  SiteB
                                        A
                   B
                                Mean
                                 Obs
                                  SO
                                  so-
                                  40
                                  30
                              ts
                              0.
                                  20-
                                  10
                                   0-
                                  -10-J
         7.9
         160
         5.7
9,1
175
7.4
                                      1234   1234
                                            Quarter
Site
 B
                                 B
                   0.65
                 (9.2, 0.53)
                   150

                    1
                                                            AIRS Site ID
                              55-079-0059
                              55-133-002?
Figure 3A-33.  Milwaukee, WI MSA (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
               for 1999-2001; (c) Intersite correlation coefficients, number of observations,
               90th percentile differences  in concentrations, and CODs.
June 2003
         3A-37
   DRAFT-DO NOT QUOTE OR CITE

-------
                                 Chicago, !L MSA
          a.
• u

•JSft >- * ,S\
*"•— J--""'^ljSS ;f\
tfl %J*^I 1
/..J^Wfg
"""'" * y " 1
L-f—1
0 SO 100km

JL
D, Mean
Obs
SD
100—
7S~



«?*
E
|, 50-
ff
o
"ri 2S*™"
0~
-20 —


c. Site
A


B


C

Site A
SiteB
SiteC





ABC

14.7 ' 12.8 T 16.1
163 ! 109 112
19.0 '• 10.7 12.2
i :
i
f i
i
i
i
i
i ,


; i r
,1; 1 III!

1234 1234 1234
Quarter
ABC
1 0.68 0.53
(20.0, --} (24,8, --}
97 101
1 0.82
(11.1.0.40)
103
1
AIRS Site ID
17-031-1016
17-031-2001
17-031-3301






























Figure 3A-34.  Chicago, IL MSA (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
              for 1999-2000; (c) Intersite correlation coefficients, number of observations,
              90th percentile differences in concentrations, and CODs.
June 2003
3A-38
DRAFT-DO NOT QUOTE OR CITE

-------
                                     Gary, IN MSA
          a.
                                                            AIRS Site ID
                                                  Site A
                                                  SiteB
                                                  SiteC
                                 18-089-0006
                                 18-089-1016
                                 18-127-0024
                                              B
Mean
 Obs
 SD

  30 H
                           5
                           0.
                               20-
                               10-
                                0-
                              -10-
                              -20-
                                     3.9
                                     54
                                     S.6
                 5.1
                 55
                 8.4

- 1| -
                     c.
Site
                   3.4
                   56
                   3.9
1234   1234  1234
         Quarter


  ABC
A


B


C
1 079
(7 8, -)
49
1



063
(6 3, -)
49
060
(8.0. 0.83)
50
1
Figure 3A-35.  Gary, IN MSA (a) Locations of sampling sites by AIRS ID#; (b) Quarterly
               distribution of 24-h average PM10_25 concentrations (ug/m3) for 2001;
               (c) Intersite correlation coefficients, number of observations, 90th percentile
               differences in concentrations, and CODs.
June 2003
            3A-39
                   DRAFT-DO NOT QUOTE OR CITE

-------
                                 Louisville, KY MSA
                                  I
                                 100km
                                          A
                                   Mean
                                    Qbs
                                     SD

                                     §0-1
                                     30-
                                  w  10
                                  s
                                  0.
                                      0-
              9.1
              55
              5.0

                                                 Site A
                                                 SiteB
                      B
7.6
51
4.0
mm —
                                        1234  1234
                                            Quarter
C.     Site    A
                                                  B
A


B
1 0.65
(5.5,
0.48)
46
1
                                                            AIRS Site ID
                                 18-019-0005
                                 21-029-0006
Figure 3A-36.  Louisville, KY MSA (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
               for 2001; (c) Intersite correlation coefficients, number of observations,
               90th percentile differences in concentrations, and CODs.
June 2003
             3A-40
    DRAFT-DO NOT QUOTE OR CITE

-------
                                  St. Louis, MO MSA
                                         &
                                                                    Site ID
                                                    Site A
                                                    Site 6
                                                    SiteC
                            17-119-0023
                            17-119-3007
                            17-163-0010
                                    A
          B
                            Mean
                            Obs
                             SD
                             100-
                         o>

                          "I
                          oi
                          o
                          v"
                         5
                         a.
                             75 —
                             80-
                              0 —
                             -25—I
22,5
106
17.4
12.1
S7
13,7
15,5
 52
14,2
                                  1334
                                           1234
                                            Qyarttr
                                                        234
                           Site
           B
A
B
C
1 0,70
(27,2, -)
51
1

0.73
(26.2, 0.76)
4?
0,82
(13,0,0,91)
50
1
Figure 3A-37.  St. Louis, MI MSA (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
               for 2000; (c) Intersite correlation coefficients, number of observations,
               90th percentile differences in concentrations, and CODs.
June 2003
     3A-41
        DRAFT-DO NOT QUOTE OR CITE

-------
                               Baton Rouge, LA. MSA
                                                             AIRS Site tD
                                                  Site A
                                                  SiteB
                      22-033-0002
                      22-121-0001
                            I
                            90
                                        A
          B
                                Moan
                                 Obs
                                 SD

                                  W-
                                  40-
                              Oi

                              10
                              cii
                              o

                              a"
                              Q_
                                 -20—I
12.8
107
8.9
                          19.1
                          112
                          10.5
                                      1234  1234
                                           Quarter
c.
                               Site
           B
                                A
                                B
 1        0,40
       (22,4, 0,43)
           96
Figure 3A-38. Baton Rouge, LA MSA (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
              for 2000-2001; (c) Intersite correlation coefficients, number of observations,
              90th percentile differences in concentrations, and CODs.
June 2003
3A-42
                             DRAFT-DO NOT QUOTE OR CITE

-------
                                    Dallas, TX MSA
          a.
                                                             AIRS Site ID
                                                  Site A
                                                  SiteB
                                                  SiteC
                                                  SiteD
                     48-113-0020
                     48-113-0035
                     48-113-0050
                     48-113-0057
                                 A
 B
                                         D
Mean
 Obs
 SD
 50-1
                           30-
                       I
                           10
                           0-
                                 11.2
                                 60
                                 6,4
12.9
65
6.7
                                14.5
                                56
                                6,4
191
S5
10.5
                               1Z34  1234  1234   1234
                                            Quarter
C.    Site     A
                                          B
                  D
A


B


C


D
1 079
(4.5,0.17)
54
1






071
(9,3. 0.22}
55
0.89
(7,3.0,18)
50
1



0.66
(16.5, 0.32)
54
060
(13.2,0.301
50
0.69
(13.5,0.24)
50
1
Figure 3A-39.  Dallas, TX MSA (a) Locations of sampling sites by AIRS ID#; (b) Quarterly
               distribution of 24-h average PM10_25 concentrations (ug/m3) for 2001;
               (c) Intersite correlation coefficients, number of observations, 90th percentile
               differences in concentrations, and CODs.
June 2003
3A-43
                                   DRAFT-DO NOT QUOTE OR CITE

-------
                               Salt Lake City, UT MSA
                                                          AIRS Site ID
                                                Site A
                                                SiteB
                                                SiteC
                    49-035-0003
                    49-035-0012
                    49-035-3006
                                   A
     B
                  C.
Mean
Obs
SD
180-
100-
1
5
IA SO-





|
I





'll

5.3
310
9.7


•
•TT

234 1234 1234

A
1
Quarter
B
0.72



c
0.74
(28.7. -) {9.8, -)





283
1




264
0.70
(27.


8. 0.47)
274
1
Figure 3A-40. Salt Lake City, UT MSA (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
              for 1999-2001; (c) Intersite correlation coefficients, number of observations,
              90th percentile differences in concentrations, and CODs.
June 2003
3A-44
DRAFT-DO NOT QUOTE OR CITE

-------
                                 Portland, OR MSA
             a.
             ©
                       b.
                              "a™**
                              m
                              o.
                                                  Site A
                                                  SiteB
                                                  B
Mean
Qte
SD
40-
30-
20-
>
>
•4
>
10-
0-

5.7 6.7
320 113
3.5 4.3



I I
§l

|
1
;
                                      1234   1234
                                           Quarter
                                                                  Site ID
Site
A
A
1
B
0,69
(5.1,-~)
10?
                      41-051-0080
                      53-011-0013
                        C.
                                B                 1

Figure 3A-41. Portland, OR MSA (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
              for 2000-2001; (c) Intersite correlation coefficients, number of observations,
              90th percentile differences in concentrations, and CODs.
June 2003
3A-45
DRAFT-DO NOT QUOTE OR CITE

-------
           a.
                               Los Angeles, GA MSA
                                ©
AIRS Site ID
Site A
SiteB
SiteC
SiteD
06-037-0002
06-037-1002
06-037-1103
06-037-4002
                  O«
                                A
         B
                         Mean
                         Obs
                          SO
                      I

                      3,
                      0.
                          40-
                          30-
                          20-
                          10-
24.1
58
11.7
15.3
S8
8.7
21.4
67
8.7
                                        234   1234
                                           Quarter
                  C.    Site
 A
 B
                 D
16.1
53
8,8
                                                       1234
         D
A


B


C


D
1 0.32 0,63
(19,0,0,24) (15.5,0.18)
49 49
1 0.74
(11.5,0,21)
49
1



058
(17,3,0.27)
46
054
(11 5,0.25)
4?
057
(12.5, 0,221
45
1
Figure 3A-42. Los Angeles, CA MSA (a) Locations of sampling sites by AIRS ID#;
              (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (jig/m3)
              for 2001; (c) Intersite correlation coefficients, number of observations,
              90th percentile differences in concentrations, and CODs.
June 2003
        3A-46
            DRAFT-DO NOT QUOTE OR CITE

-------
                                  Riverside, CA MSA
AIRS Sits ID
Site A
SiteB
SiteC
SiteD
06-065-2002
06-065-8001
06-071-0025
06-071-2002
                  b.
                                          B
                           D
                         Mean
                          Obs
                           SD

                          600-
                       o>
                       **»!*
                        10
                        cj
                        o
                          400-
                          300-
                        « 200-
                       I
                          100-
                           0-
                         -100-
46.2
210
46.3
32,1
200
20.4
2S.S
108
15,4
28.3
112
1S4
                               1234  1234  1234   1234
                                            Quarter
                  C.     Site     A
          B
                  D
A
B
C
D
1 0.36 0.32
(428.0.38) (368,0.39)
184 102
1 082
(25.9, 0.32)
104
1

0.45
139.0, 0,38}
107
0 79
(182,0,33)
108
0.80
(13.3.028)
98
1
Figure 3A-43.  Riverside, CA MSA (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
               for 2000-2001; (c) Intersite correlation coefficients, number of observations,
               90th percentile differences in concentrations, and CODs.
June 2003
         3A-47
             DRAFT-DO NOT QUOTE OR CITE

-------
         a.
                                 San Diego, CA MSA
AIRS Site ID
Site A
SiteB
SiteC
SiteD
06-073-0001
06-073-0003
06-073-1008
06-073-1002
                           50
                                   100 km
                                         B
                          D
                         Mean
                         Obs
                          SO
                          40-
                       eh
                          20-
                          10-
                           0-
                         -10-J
13.3
55
4.0
19,4
51
8.2
11.6
57
5.1
12.9
S8
7.7
                              1234  1234   1234
                                           Quarter
                                                       1234
                        Site    A
         B
                  D
A
B
C
D
1 0.38 0,01
(14.4, 0.25) (9.5, 0.34)
48 51
1 0.85
(12.6,0,28)
48
1

009
(11.7,0.32)
50
0.63
(14.7, 0.34)
46
0,70
(83,045)
51
1
Figure 3A-44.  San Diego, CA MSA (a) Locations of sampling sites by AIRS ID#;
               (b) Quarterly distribution of 24-h average PM10_2 5 concentrations (ug/m3)
               for 2001; (c) Intersite correlation coefficients, number of observations,
               90th percentile differences in concentrations, and CODs.
June 2003
        3A-48
            DRAFT-DO NOT QUOTE OR CITE

-------
 i                                     APPENDIX 3B
 2
 3     Aerosol Composition Data from the  Speciation Network
 4
 5
 6          The United States Environmental Protection Agency (EPA), working with state and local
 7     air quality agencies, began implementing an ambient air monitoring network in 1999/2000 to
 8     provide a consistent data set for the characterization and evaluation of trends in PM components
 9     (chemical species). The network was designed to include about 52 core trends sites across the
10     United States and to provide a stable ongoing national perspective. In 1999, an initial thirteen
11     sites were installed and operated to be used as a model for the deployment of the more
12     comprehensive network consisting of the 52 core trends sites and roughly two hundred
13     additional locally relevant sites.
14          Data from the initial thirteen sites were presented in the Air Quality Criteria for Particulate
15     Matter (3rd External Review Draft) by the EPA in 2001 for public comment and CAS AC review.
16     These sites were designed and operated to evaluate the suitability of various aerosol sampling
17     devices for obtaining PM25  composition data. These data were summarized in Appendix 3B of
18     the 2001 3rd External Review Draft.  Three types of collocated aerosol sampling devices were
19     used in this study, which lasted from February 2000 through July 2000; data were obtained from
20     the three sampling devices shown for each site. A complete description of the data, techniques
21     used to analyze the filters, and the results of the evaluation of the performance of the sampling
22     devices (including a number of caveats regarding the data) can be found in Coutant and Stetzer
23     (2001) and the analyses of data in Coutant et al. (2001).
24          More recent measurements of ambient levels of PM25 constituents are provided in the
25     tables below. Table 3B-1 presents the locations and sample type for a different set of
26     13 monitoring sites representing a cross section of the country.  Many of these sites were chosen
27     because they are located in MSAs (or close to MSAs) in which risk assessments are to be
28     performed. Sacramento, CA, and Riverside,  CA, are located near San Francisco and
29     Los Angeles, respectively. The data reported here are from the period October 2001 to
30     September 2002. For this time period, a total of 51 sites (this includes both "Trends" and
31     "non-Trends" sites) across the country have complete data (as defined by 50% of observations

       June 2003                               3B-1        DRAFT-DO NOT QUOTE  OR CITE

-------
 1      available for every quarter for each of the major chemical species:  sulfate, ammonium, nitrate,
 2      elemental carbon, organic carbon, and the five trace elements that go into the calculation of the
 3      crustal contribution to PM2 5 (Al, Fe, Ca, Ti, and Si).
 4           Summary statistics for concentrations of PM25 and chemical species are given for each of
 5      the 13 sites in Table 3B-2. The number of samples (n) and the AIRS site code are given above
 6      each table. Entries in the tables give for each component the mean, minimum and maximum
 7      component concentrations, and minimum detection limits. Minimum detection levels (MDL)
 8      differ among the various sampling methods; these limits were estimated by Research Triangle
 9      Institute in July of 2001 and are subject to review, revision, and reinterpretation. Anions and
10      cations (ammonium, nitrate, sodium, potassium, and sulfate) were determined by ion
11      chromatography; carbonaceous species were determined by the thermal optical transmittance
12      method (NIOSH method); and trace elements (aluminum through zirconium) were determined
13      by X-ray fluorescence spectrometry. The sulfate (calculated) entry is based on the XRF
14      determination of S. In general, relatively good agreement is found between the reconstructed
15      mass and the PM25 concentration measured by the collocated FRM monitor at each site.
16      However, there are exceptions at  several locations as can be seen from inspection of Table 3B-2.
17      Somewhat different sampling trains are used in the different sampling systems. These
18      differences can result in differences in performance metrics and in differences in the entities that
19      are measured. All samplers use a denuder in front of the filter for species to be analyzed by ion
20      chromatography.  In the RAAS and SASS samplers the denuder is followed by a nylon filter;
21      whereas in the MASS samplers (Chicago, Houston, Seattle) a teflon filter is followed by a nylon
22      filter. Particulate nitrate is collected on the teflon filter and is referred to as nonvolatile nitrate.
23      However, there may be volatilization of nitrate containing compound from the teflon filter which
24      is then adsorbed on the nylon filter. Nitrate extracted from the nylon filter is referred to as
25      volatile nitrate.
26           Organic carbon (OC) concentrations are multiplied by a factor of 1.4 when calculating
27      mass to account for the presence of H, N, and O in organic compounds on all samplers.
28      Carbonate carbon has never been detected in any of the samples. Field blank corrections that
29      could be applied to elemental carbon (EC) and OC concentrations are shown in Table 3B-3 for
30      different samplers. Blank corrections for OC and EC  shown in Table 3B-3 were applied to
31      concentrations shown in Table 3B-2. However, subtracting blank corrections from OC

        June 2003                                 3B-2       DRAFT-DO NOT QUOTE OR CITE

-------
 1      concentrations results in negative values in several cases.  A possible cause for results such as

 2      these could have been that not enough blank samples were obtained to fully characterize the

 3      blank levels. Although the concentrations of 47 elements could be obtained by X-ray

 4      fluorescence spectrometry, the concentrations of many of these elements are beneath MDLs and

 5      are not shown. The same elements shown in Appendix 6 A of the 1996 PM AQCD are shown

 6      here.  The usual practice of denoting table entry values below MDL by (—) is followed here.

 7      Missing data for PM2 5 are also indicated by (—).
10      REFERENCES
11
12      Coutant,B.; Stetzer, S. (2001) Evaluation of PM25 speciation sampler performance and related sample collection and
13           stability issues: final report. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of
14           Air Quality Planning and Standards; report no. EPA-454/R-01-008. Available:
15           http://www.epa.gov/ttn/amtic/pmspec.html [5 April, 2002].
16      Coutant, B.; Zhang, X.; Pivetz, T. (2001) Summary statistics and data displays for the speciation minitrends study:
17           final report. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality
18           Planning and Standards; contract no. 68-D-98-030.
19
        June 2003                                   3B-3        DRAFT-DO NOT QUOTE OR CITE

-------
  TABLE 3B-1. PM25 SPECIATION SAMPLERS BY LOCATION:  SITES SELECTED
                             FOR PM CD SUMMARY
Sampler
Type
SASS
SASS
RAAS
SASS
MASS
RAAS
MASS
SASS
SASS
SASS
MASS
SASS
SASS
Location_Name
Burlington
Philadelphia
(AMS Laboratory)
Atlanta
(South DeKalb)
Detroit (Allen Park)
Chicago (Com ED)
St. Louis (Blair Street)
Houston (Deer Park)
Minneapolis (Philips)
Boulder
(Commerce City)
Phoenix (Supersite)
Seattle (Beacon Hill)
Sacramento
(Del Paso Manor)
Riverside-Rubidoux
State
VT
PA
GA
MI
IL
MO
TX
MN
CO
AZ
WA
CA
CA
Began
Operation
12/14/2000
9/10/2001
3/2/2001
12/14/2000
5/22/2001
2/9/2000
2/9/2000
12/14/2000
2/18/2001
2/21/2000
2/9/2000
2/9/2000
5/13/2001
AIRS_Code
000500070012
000421010004
000130890002
000261630001
000170310076
000295100085
000482011039
000270530963
000080010006
000040139997
000530330080
000060670006
000060658001
POC
5
7
5
5
5
6
6/7
5
5
7
6
5
5/6
Reason for
Schedule Inclusion
l-in-3
l-in-3
l-in-3
l-in-3
Alt l-in-3
l-in-3
l-in-3
Alt l-in-3
Alt l-in-3
l-in-3
l-in-3
l-in-3
l-in-3
o
J
3
2
1
3
1
o
3
2
o
J
1
3
2
1
 1 = Completeness of geographic coverage
 2 = PM10 risk assessment city
 3 = PM2 5 and PM10 risk assessment city
June 2003
3B-4
DRAFT-DO NOT QUOTE OR CITE

-------
               TABLE 3B-2a. BURLINGTON, VT SUMMARY DATA
                        (October 2001 to September 2002).
       All concentrations are given in ug/m3; n = 201; AIRS Site Code: 500070012
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
10.0
10.9
2.6
2.8
1.1
0.22
0.035
1.3
—
—
0.26
—
2.5
0.020
—
—
0.002
—
0.035
—
0.002
0.002
0.041
—
0.014
0.002
0.003
0.002
0.005
0.041
0.001
0.001
0.069
0.10
0.001
0.88
0.013
0.004
0.001
0.008
Max Min
61.0 1.0
63.0 1.6
17.7 0.001
18.8 0.17
6.8 —
2.1 —
0.38 —
8.8 —
— —
— —
0.84 —
— —
31.4 -0.43
0.28 —
0.005 —
0.12 —
0.007 —
0.014 —
0.19 —
0.12 —
0.060 —
0.014 0.001
0.390 0.007
0.016 —
0.19 —
0.007 —
0.012 —
0.022 0.001
0.092 —
0.41 —
0.003 —
0.005 —
0.54 —
0.40 —
0.004 —
5.9 —
0.033 —
0.023 —
0.006 —
0.057 0.007
MDL
—
—
—
0.012
0.017
0.030
0.014
0.008
—
—
0.15
0.15
0.15
0.011
0.002
0.059
0.002
0.011
0.003
0.006
0.002
0.001
0.002
0.006
0.018
0.002
0.005
0.001
0.006
0.003
0.002
0.002
0.008
0.051
0.003
0.007
0.018
0.002
0.002
0.002
June 2003                            3B-5       DRAFT-DO NOT QUOTE OR CITE

-------
      TABLE 3B-2b. PHILADELPHIA, PA SUMMARY DATA (October 2001 to
           September 2002). All concentrations are given in ug/m3; n = 262;
                          AIRS Site Code: 421010004
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
14.2
16.0
4.3
4.4
2.0
0.21
0.042
2.1
0.84
0.61
0.66
—
3.1
0.019
—
—
0.004
—
0.037
0.011
0.002
0.004
0.084
0.005
0.014
0.002
—
0.006
—
0.053
—
0.002
0.086
0.087
—
1.45
—
0.006
0.004
0.015
Max Min
86.8 2.1
239.0 0.0
29.0 0.004
30.5 0.020
11.4 —
1.6 —
0.83 —
8.7 0.050
4.0 0.030
5.6 0.020
2.3 —
— —
50.5 -1.13
0.54 —
0.006 —
0.12 —
0.013 —
0.017 —
0.18 —
0.68 —
0.018 —
0.025 0.003
0.53 —
0.025 —
0.30 —
0.009 —
0.012 —
0.13 0.001
0.039 —
1.1 —
0.003 —
0.005 —
1.1 —
0.63 —
0.016 —
9.6 —
0.035 —
0.049 —
0.035 —
0.095 0.007
MDL
—
—
—
0.012
0.017
0.030
0.014
0.008
—
—
0.15
0.15
0.15
0.011
0.002
0.059
0.002
0.011
0.003
0.006
0.002
0.001
0.002
0.005
0.018
0.002
0.005
0.001
0.006
0.003
0.002
0.002
0.008
0.051
0.003
0.007
0.018
0.002
0.002
0.001
June 2003
3B-6
DRAFT-DO NOT QUOTE OR CITE

-------
TABLE 3B-2c. ATLANTA,
All concentrations are
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Gold
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
GA SUMMARY DATA (October 2001 to September
given in jig/m3; n = 183; AIRS Site Code: 130890002
Mean
—
16.3
4.8
4.8
1.3
0.27
0.044
0.70
—
—
0.90
—
4.3
0.028
0.001
0.018
0.003
0.002
0.037
0.003
—
0.002
—
0.084
0.003
0.008
0.002
—
—
—
0.060
0.005
0.001
0.11
0.056
0.001
1.6
—
0.006
0.001
0.008
Max
—
40.0
14.5
15.2
4.2
1.7
0.31
3.5
—
—
3.5
—
11.2
0.65
0.014
0.054
0.009
0.006
0.28
0.049
0.003
0.011
0.004
0.47
0.008
0.16
0.011
0.003
0.002
0.017
0.36
0.002
0.004
1.4
0.31
0.005
4.8
0.019
0.043
0.003
0.034
Min
—
4.7
0.27
0.88
—
—
—
0.16
—
—
—
—
0.8
—
—
0.012
—
—
0.006
—
—
—
—
0.013
—
—
—
—
—
—
0.015
—
0.000
0.010
—
0.001
0.089
—
—
—
0.001
2002).
MDL
—
—
—
0.011
0.015
0.028
0.013
0.008
—
—
0.13
0.13
0.13
0.004
0.001
0.024
0.001
0.004
0.001
0.002
0.001
0.001
0.002
0.001
0.002
0.007
0.001
0.002
0.001
0.003
0.001
0.001
0.001
0.003
0.021
0.001
0.003
0.007
0.001
0.001
0.001
June 2003
3B-7
DRAFT-DO NOT QUOTE OR CITE

-------
  TABLE 3B-2d. DETROIT, MI SUMMARY DATA (October 2001 to
       All concentrations are given in ug/m3; n = 189; AIRS Site Code:
                        September 2002).
                         261630001
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
16.6
18.0
4.4
4.6
2.2
0.27
0.061
3.1
—
—
0.68
—
3.2
0.025
0.002
0.043
0.003
—
0.069
0.017
0.002
0.006
0.12
0.006
0.020
0.004
—
0.002
—
0.078
—
0.002
0.11
0.098
—
1.5
—
0.007
0.002
0.025
Max
53.1
52.9
22.0
24.4
9.3
1.7
1.2
15.2
—
—
3.7
—
11.2
0.53
0.010
0.11
0.011
0.014
0.33
0.65
0.033
0.043
0.60
0.034
0.30
0.025
0.013
0.022
0.028
1.3
0.003
0.011
0.84
0.53
0.021
7.3
0.032
0.079
0.012
0.19
Min
2.6
3.2
0.010
0.040
0.009
0.015
0.007
0.004
—
—
—
—
-0.46
—
—
—
—
—
—
—
—
0.001
0.003
—
—
—
—
0.001
—
—
—
—
—
—
—
—
—
—
—
0.001
MDL

—

0.012
0.017
0.030
0.014
0.008
—
—
0.15
0.15
0.15
0.011
0.002
0.059
0.002
0.011
0.003
0.006
0.002
0.001
0.002
0.005
0.018
0.002
0.005
0.001
0.006
0.003
0.002
0.002
0.008
0.051
0.003
0.007
0.018
0.002
0.002
0.001
June 2003
3B-8
DRAFT-DO NOT QUOTE OR CITE

-------
  TABLE 3B-2e. CHICAGO,
       All concentrations are
IL SUMMARY DATA (October 2001 to
given in ug/m3; n = 139; AIRS Site Code:
             September 2002).
              170310076
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
15.7
15.0
4.3
4.2
1.9
0.052
0.068
2.0
0.71
1.3
0.61
—
2.8
0.03
0.001
0.018
0.003
—
0.058
0.017
0.001
0.004
0.091
0.006
0.017
0.003
—
0.001
0.002
0.085
—
0.001
0.11
0.048
0.001
1.4
—
0.004
0.001
0.023
Max
42.7
44.5
22.8
21.2
8.8
0.26
3.1
10.2
3.6
9.9
1.6
—
7.4
1.1
0.006
0.22
0.011
0.007
0.55
0.67
0.005
0.056
0.53
0.040
0.69
0.014
0.006
0.007
0.015
3.1
0.001
0.004
2.2
0.58
0.061
7.6
0.023
0.070
0.002
0.093
Min
3.6
3.4
0.64
0.56
0.17
—
—
0.12
0.04
0.03
—
—
0.31
—
0.001
0.012
—
—
0.10
—
—
—
0.014
—
—
—
—
—
—
0.005
—
0.001
0.010
0.010
0.001
0.21
—
—
—
0.001
MDL
—
—
—
0.005
0.007
0.012
0.006
0.003
0.003
0.003
0.059
0.059
0.059
0.004
0.001
0.024
0.001
0.004
0.001
0.002
0.001
0.001
0.001
0.002
0.007
0.001
0.002
0.001
0.003
0.001
0.001
0.001
0.003
0.021
0.001
0.003
0.007
0.001
0.001
0.001
June 2003
           3B-9
DRAFT-DO NOT QUOTE OR CITE

-------
 TABLE 3B-2f. ST. LOUIS, MO SUMMARY DATA (October 2001 to September 2002).
       All concentrations are given in ug/m3; n = 324; AIRS Site Code: 295100085
Parameter
PM2 5 (FRM Mass)
PML5_ (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
15.6
16.4
4.3
4.3
1.9
0.19
0.069
2.3
—
—
0.72
—
3.5
0.044
0.002
0.024
0.004
—
0.13
0.027
0.002
0.018
0.24
0.014
0.011
0.014
—
0.002
0.004
0.099
—
0.001
0.17
0.045
0.001
1.4
0.008
0.008
0.001
0.033
Max
52.1
53.2
22.5
20.6
7.2
1.6
5.9
14.0
—
—
2.7
—
11.5
2.5
0.011
0.38
0.16
0.010
1.1
0.57
0.034
0.72
1.3
0.09
0.79
0.13
0.007
0.04
0.095
5.6
0.002
0.006
4.0
0.43
0.12
7.5
0.028
0.14
0.011
0.48
Min
3.8
2.8
0.61
0.64
0.14
0.014
0.007
0.20
—
—
—
—
—
—
0.001
—
—
—
0.010
—
—
—
0.012
—
0.004
—
—
—
—
0.016
—
—
0.011
0.010
0.001
0.20
—
—
—
0.002
MDL
—
—
—
0.011
0.015
0.028
0.013
0.008
—
—
0.13
0.13
0.13
0.004
0.001
0.024
0.001
0.004
0.001
0.002
0.001
0.001
0.001
0.002
0.007
0.001
0.002
0.001
0.003
0.001
0.001
0.001
0.003
0.021
0.001
0.003
0.007
0.001
0.001
0.001
June 2003                            3B-10      DRAFT-DO NOT QUOTE OR CITE

-------
 TABLE 3B-2g. HOUSTON,
       All concentrations are
TX SUMMARY DATA (October 2001 to
given in ug/m3; n = 229; AIRS Site Code:
              September 2002).
              482011039
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
12.4
11.0
3.4
3.5
1.2
0.22
0.051
0.68
0.37
0.32
0.30
—
2.2
0.092
0.001
—
0.003
0.002
0.055
0.050
0.001
0.003
0.073
0.002
0.020
0.002
0.002
0.002
0.002
0.072
—
0.001
0.23
0.17
0.001
1.1
0.006
0.007
0.003
0.006
Max
21.9
48.0
24.5
26.0
8.8
1.1
0.38
5.8
3.5
4.2
1.1
—
12.1
1.6
0.006
0.059
0.022
0.008
0.44
0.70
0.009
0.024
0.83
0.007
0.38
0.013
0.022
0.047
0.021
0.64
0.002
0.001
2.8
1.1
0.009
8.2
0.019
0.091
0.011
0.030
Min
5.9
0.0
0.004
0.030
—
—
—
0.040
0.010
0.020
—
—
—
—
0.001
—
—
—
0.001
—
—
—
—
—
—
—
—
—
—
0.001
—
0.001
—
—
0.001
—
—
—
—
—
MDL
—
—
—
0.005
0.007
0.012
0.006
0.003
0.003
0.003
0.059
0.059
0.059
0.004
0.001
0.024
0.001
0.004
0.001
0.002
0.001
0.001
0.001
0.002
0.007
0.001
0.002
0.001
0.003
0.001
0.001
0.001
0.003
0.021
0.001
0.003
0.007
0.001
0.001
0.001
June 2003
          3B-11
DRAFT-DO NOT QUOTE OR CITE

-------
       TABLE 3B-2h. MINNEAPOLIS, MN SUMMARY DATA (October 2001 to
           September 2002). All concentrations are given in ug/m3; n = 163;
                          AIRS Site Code: 270530963
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
10.3
12.4
2.3
2.4
1.2
0.27
0.059
2.3
—
—
0.39
—
2.6
0.028
0.002
—
0.002
—
0.071
0.008
0.002
0.003
0.065
0.005
0.018
0.002
—
0.002
0.005
0.068
—
—
0.12
0.093
—
0.78
—
0.006
0.002
0.008
Max
33.4
37.7
8.6
8.9
6.0
3.7
1.7
16.4
—
—
1.8
—
9.9
0.67
0.012
0.15
0.008
0.028
0.46
0.17
0.016
0.045
0.37
0.066
0.52
0.009
0.011
0.014
0.11
1.8
0.002
0.004
1.3
0.69
0.026
2.9
0.034
0.043
0.021
0.039
Min
1.8
2.2
0.12
0.24
—
—
—
0.098
—
—
—
—
-0.63
—
—
—
—
—
—
—
—
0.001
0.016
0.003
0.009
—
—
0.001
—
0.005
—
—
—
—
—
0.040
—
—
—
0.001
MDL
—
—
—
0.012
0.017
0.030
0.014
0.008
—
—
0.15
0.15
0.15
0.011
0.002
0.059
0.002
0.011
0.003
0.006
0.002
0.001
0.002
0.005
0.018
0.002
0.005
0.001
0.006
0.003
0.002
0.002
0.008
0.051
0.003
0.007
0.018
0.002
0.002
0.001
June 2003
3B-12
DRAFT-DO NOT QUOTE OR CITE

-------
  TABLE 3B-21. BOULDER, CO SUMMARY DATA (October 2001 to
       All concentrations are given in ug/m3; n = 161; AIRS Site Code:
                         September 2002).
                         080010006
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
9.5
11.3
1.5
1.6
0.79
0.24
0.032
1.5
—
—
1.0
—
3.2
0.092
—
—
0.002
—
0.12
0.019
0.002
0.004
0.13
0.005
0.020
0.003
—
0.001
0.006
0.063
—
—
0.28
0.075
—
0.49
—
0.010
—
0.023
Max
26.9
28.2
5.3
5.9
4.3
1.7
0.29
7.9
—
—
4.4
—
13.8
0.41
0.005
0.12
0.011
0.016
0.55
0.66
0.009
0.013
0.47
0.036
0.15
0.015
0.010
0.010
0.096
0.33
0.003
0.004
0.97
0.28
0.005
1.8
0.033
0.031
0.004
0.53
Min
2.1
3.7
0.31
0.29
0.009
0.015
0.007
0.14
—
—
0.15
—
0.28
0.005
—
—
—
—
0.008
—
—
0.001
0.016
—
—
—
—
0.001
—
—
—
—
0.027
—
—
0.10
—
—
—
0.001
MDL
—
—
—
0.012
0.017
0.030
0.014
0.008
—
—
0.15
0.15
0.15
0.011
0.002
0.059
0.002
0.011
0.003
0.006
0.002
0.001
0.002
0.005
0.018
0.002
0.005
0.001
0.006
0.003
0.002
0.002
0.008
0.051
0.003
0.007
0.018
0.002
0.002
0.001
June 2003
3B-13
DRAFT-DO NOT QUOTE OR CITE

-------
  TABLE 3B-2J. PHOENIX, AZ SUMMARY DATA (October 2001 to
       All concentrations are given in ug/m3; n = 275; AIRS Site Code:
                        September 2002).
                         040139997
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
9.9
12.0
1.2
1.3
0.5
0.28
0.062
1.1
—
—
0.75
—
4.0
0.12
0.002
—
0.003
—
0.15
0.047
0.002
0.006
0.17
—
0.025
0.004
—
0.003
0.006
0.11
—
—
0.36
0.10
0.003
0.39
0.014
0.013
—
0.009
Max
32.1
158.0
2.9
3.8
2.6
1.9
0.64
6.9
—
—
5.0
—
21.0
0.87
0.009
0.13
0.009
0.024
0.82
0.40
0.021
0.038
0.95
0.020
0.38
0.027
0.011
0.13
0.13
0.69
0.003
0.005
2.2
0.52
0.10
0.98
0.038
0.058
0.005
0.080
Min
3.3
0.8
0.001
0.006
0.009
0.015
0.007
0.004
—
—
—
—
-0.66
0.005
—
—
—
—
0.005
—
—
0.001
0.003
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
0.001
MDL
—
—
—
0.012
0.017
0.030
0.014
0.008
—
—
0.15
0.15
0.15
0.011
0.002
0.059
0.002
0.011
0.003
0.006
0.002
0.001
0.002
0.005
0.018
0.002
0.005
0.001
0.006
0.003
0.002
0.002
0.008
0.051
0.003
0.007
0.018
0.002
0.002
0.001
June 2003
3B-14
DRAFT-DO NOT QUOTE OR CITE

-------
 TABLE 3B-2k. SEATTLE, WA SUMMARY DATA (October 2001 to
       All concentrations are given in ug/m3; n = 314; AIRS Site Code:
                        September 2002).
                         530330080
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
8.2
8.0
1.4
1.3
0.47
0.16
0.045
0.67
0.25
0.41
0.60
—
2.6
0.015
0.001
—
0.002
—
0.029
0.055
0.002
0.003
0.053
0.004
0.014
0.003
0.002
0.002
—
0.055
—
0.001
0.049
0.15
0.001
0.45
0.007
0.003
0.003
0.009
Max
29.1
28.5
4.9
4.6
2.0
0.76
2.3
2.9
1.6
2.5
2.7
—
11.0
0.24
0.007
0.15
0.014
0.007
0.20
0.84
0.016
0.045
0.29
0.078
0.19
0.024
0.011
0.020
0.028
2.3
0.002
0.008
0.52
0.84
0.044
1.6
0.030
0.026
0.029
0.048
Min
1.9
1.5
0.057
0.14
—
0.012
—
0.12
0.010
0.020
—
—
—
—
0.001
—
—
—
0.001
—
—
—
0.001
—
—
—
—
—
—
0.007
—
0.001
—
—
0.001
0.019
—
—
—
—
MDL
—
—
—
0.005
0.007
0.012
0.006
0.003
0.003
0.003
0.059
0.059
0.059
0.004
0.001
0.024
0.001
0.004
0.001
0.002
0.001
0.001
0.001
0.002
0.007
0.001
0.002
0.001
0.003
0.001
0.001
0.001
0.003
0.021
0.001
0.003
0.007
0.001
0.001
0.001
June 2003
3B-15
DRAFT-DO NOT QUOTE OR CITE

-------
TABLE 3B-21. SACRAMENTO, CA SUMMARY DATA (October 2001 to September
2002). All concentrations are given in ng/m3; n = 265; AIRS Site Code: 060670006
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Silver
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
9.4
15.0
1.2
1.3
0.74
0.38
0.077
2.3
—
—
0.66
—
5.2
0.040
0.002
—
0.002
—
0.043
0.052
0.002
0.006
0.079
—
0.021
0.002
—
0.010
0.006
0.10
—
—
0.12
—
0.19
—
0.41
—
—
—
0.005
Max
78.0
120.0
5.5
4.1
7.4
3.5
3.6
23.4
—
—
8.4
—
54.1
0.52
0.006
0.26
0.013
0.018
0.56
1.8
0.043
0.16
0.64
0.044
0.88
0.013
0.015
0.61
0.11
3.3
0.003
0.004
1.1
0.015
1.7
0.077
1.8
0.035
0.052
0.006
0.11
Min
3.0
3.0
0.14
0.06
0.009
0.015
0.007
0.004
—
—
—
—
0.18
—
—
—
—
—
0.005
—
—
0.001
0.005
—
—
—
—
0.001
—
—
—
—
0.011
—
—
—
0.045
0.009
—
—
—
MDL
—
—
—
0.012
0.017
0.030
0.014
0.008
—
—
0.15
0.15
0.15
0.011
0.002
0.059
0.002
0.011
0.003
0.006
0.002
0.001
0.002
0.005
0.018
0.002
0.005
0.001
0.006
0.003
0.002
0.002
0.008
0.010
0.010
0.051
0.003
0.007
0.018
0.002
0.002
June 2003
3B-16
DRAFT-DO NOT QUOTE OR CITE

-------
  TABLE 3B-2m. RIVERSIDE-RUBIDOUX, CA SUMMARY DATA (October 2001 to
           September 2002). All concentrations are given in ug/m3; n = 161;
                          AIRS Site Code:  060658001
Parameter
PM2 5 (FRM Mass)
PM2 5 (Reconstructed Mass)
Sulfate (Calculated)
Sulfate (by I.C.)
Ammonium (by I.C.)
Sodium Ion (by I.C.)
Potassium (by I.C.)
Nitrate
Volatile Nitrate
Nonvolatile Nitrate
Elemental Carbon
Carbonate Carbon
Organic Carbon
Aluminum
Arsenic
Barium
Bromine
Cadmium
Calcium
Chlorine
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Phosphorous
Potassium
Rubidium
Selenium
Silicon
Silver
Sodium
Strontium
Sulfur
Tin
Titanium
Vanadium
Zinc
Mean
28.6
30.5
3.6
3.7
4.8
0.49
0.075
12.3
—
—
1.2
—
6.0
0.057
0.002
—
0.006
—
0.17
0.072
0.003
0.006
0.17
0.006
0.032
0.004
—
0.002
0.007
0.11
—
—
0.20
—
0.20
0.002
1.2
0.014
0.012
0.006
0.023
Max
78.2
79.2
9.6
10.0
16.8
1.8
1.3
40.3
—
—
4.3
—
13.0
0.32
0.006
0.17
0.016
0.017
1.0
0.71
0.024
0.037
0.58
0.026
0.31
0.015
0.024
0.014
0.067
1.31
0.003
—
0.88
0.015
0.94
0.026
3.2
0.081
0.041
0.017
0.21
Min
2.6
3.1
0.23
0.28
—
—
—
0.18
—
—
0.12
—
0.38
—
—
—
—
—
0.012
—
—
0.001
0.019
—
—
—
—
0.006
—
0.017
—
—
0.023
—
—
—
0.078
—
—
—
0.001
MDL
—
—
—
0.012
0.017
0.030
0.014
0.008
—
—
0.15
0.15
0.15
0.011
0.002
0.059
0.002
0.011
0.003
0.006
0.002
0.001
0.002
0.005
0.018
0.002
0.005
0.001
0.006
0.003
0.002
0.006
0.008
0.010
0.051
0.003
0.007
0.018
0.002
0.002
0.001
June 2003
3B-17
DRAFT-DO NOT QUOTE OR CITE

-------
     TABLE 3B-3. BLANK CORRECTIONS FOR ELEMENTAL CARBON (EC),
            ORGANIC CARBON (OC), AND TOTAL CARBON (TC)
                    IN THE SPECIATION NETWORK
Sampler
Type
URGMASS
R & P 2300
Anderson RAAS
R & P 2025
MetOne SASS
Elemental
EC Mass
(jigC/filter)
0.63
3.21
0.97
1.67
1.03
Carbon
EC Cone
OigC/m3)
0.03
0.22
0.09
0.07
0.11
Organic
OC Mass
OigC/filter)
7.08
12.93
12.54
18.42
13.75
Carbon
OC Cone
OigC/m3)
0.29
0.90
1.19
0.77
1.42
Total
TC Mass
OigC/filter)
7.71
16.13
13.51
19.91
14.78
Carbon
TC Cone
OigC/m3)
0.32
1.12
1.29
0.83
1.53
June 2003
3B-18
DRAFT-DO NOT QUOTE OR CITE

-------
 i                                    APPENDIX 3C
 2
 3                 Organic Composition of Particulate Matter
 4
 5
 6          Although organic compounds typically constitute approximately 10 to 70% of the total dry
 7     fine particle mass in the atmosphere, organic PM concentrations, composition, and formation
 8     mechanisms are poorly understood. This is because particulate organic matter is an aggregate of
 9     hundreds of individual compounds spanning a wide range of chemical and thermodynamic
10     properties (Saxena and Hildemann, 1996). The presence of multiphase or "semivolatile"
11     compounds complicates collection of organic particulate matter. Furthermore, no single
12     analytical technique currently is capable of analyzing the entire range of compounds present.
13     Rigorous analytical methods frequently identify only 10 to 20% of the organic mass on the
14     molecular level (Rogge et al., 1993). The data shown in Appendix 3C are meant to complement
15     the data given for the inorganic components of particles in Appendix 6 A of the 1996 PM AQCD
16     (U. S. Environmental Protection Agency, 1996). Table 3C-1 lists a number of recent urban and
17     some rural measurements of particulate organic and elemental carbon in jig of carbon/m3 (jig
18     C/m3).  Emphasis is placed on measurements published after 1995. The analysis method and
19     artifact correction procedure, if any, are indicated.  Table 3C-2 presents information on recent
20     (post-1990) studies concerning concentrations (in ng C/m3) of particulate organic compounds
21     found at selected U.S. sites.
       June 2003                               3C-1        DRAFT-DO NOT QUOTE OR CITE

-------
3
TABLE 3C-1. PARTICULATE ORGANIC AND ELEMENTAL CARBON CONCENTRATIONS (in ug C/m3)
                      BASED ON STUDIES PUBLISHED AFTER 1995
to
o
o
OJ











OJ
O
to


o
1

o

0
H
O
n
s.**
H
W
O



Reference
URBANPM25
Offenberg and Baker
(2000)

Allen etal. (1999)

Pedersen et al. (1999)





IMPROVE (2000)

Lewtasetal. (2001)
Khwaja(1995)
Christoforou et al.
(2000)




Turpin and Huntzicker
(1995)






Location

Chicago, IL


Uniontown, PA

Boston, MA
Reading, MA (suburban)
Quabbin, MA (rural)
Rochester, NY (urban)
Brockport, NY (rural)

Washington, DC
Seattle, WA
Seattle, WA
Schenectady, NY
Azusa, CA
Long Beach, CA
Central, LA
Rubidoux, LA
San Nicolas, LA

Claremont, CA
Long Beach, CA






Dates

July 1994;
Jan 1995

July-Aug 1990

Jan-Dec 1995





1994-1998

Apr-May 1999
Oct 24-26, 1991
Jan-Dec 1993





Jun-Sept
Nov-Dec 1987





OC Mean
(Max)




(0.8-8.4)3

5.8
4.0
2.8
3.3
2.7

3.4
1.8
8.0

9.4
8.9
12.3
9.7
1.6

na (29.4)
na (62.6)





EC Mean TC Mean
(Max) (Max)

2.2 (3.8)

1.7
(0.4-3. 5)a
1.3(3.1)
1.7
0.7
0.5
0.7
0.5

1.1
0.3
1.4
23.2 (49.9)
1.3
1.8
2.7
1.5
1.5

na (9.0)
na (24.6)





Avg.
Time Notes

12 h PM12; Imp; TOT

PMM; Imp; TOT
3 h PM25; DQQ; TORb
10 min Aeth
24 h PM2 0; Q; TOT





24 h PM25; QQ; TOR

23 h PM25; DQA; EGAC
6 h PMj 0; Q; Th
24 h PM21; Q; TOR





2h PM25;Q+TQ;TOTd
2-6 h



O

-------
c
3

to
o
o
OJ
        TABLE 3C-1 (cont'd). PARTICULATE ORGANIC AND ELEMENTAL CARBON CONCENTRATIONS (in jig C/m3)

                                  BASED ON STUDIES PUBLISHED AFTER 1995
OJ

O
O
O


O
H

O

O
H
W

O


O
HH

s
Reference
RURAL PM2 5
Klinedinst and Currie
(1999)
Andrews et al. (2000)


Malm and Gebhart
(1996)
IMPROVE (2000)






Heggetal. (1997)


Cuietal. (1997)
Chow etal. (1996)














Location Dates

Welby, CO Dec 1996- Jan
Brighton, CO 1997
Look Rock, Smoky Mountains, July-Aug 1995
TN

Tahoma Wood, WA June-Aug 1990

Three Sisters Wilderness, OR 1994-1998
Rocky Mountains, CO
Brigantine, NJ
Acadia, MA
Jefferson: James River Face
Wilderness, VA
Glacier, MT
1 50 km East of Mid- Atlantic July 1 996
Coast
(0.02-4 km altitude)
Meadview, AZ Aug 6- 1 5, 1 992
Point Reyes, CA July-Aug 1 990
Altamont Pass, CA
Pacheco Pass, CA
Crows Landing, CA
Academy, CA
Button- Willow, CA
Edison, CA
Caliente, CA
Sequoia, CA
Yosemite, CA





OC Mean
(Max)

5.6(13.4)
3.6 (6.4)
2.2
2.7
1.2
2.6 (7.4)

0.9
1.0
2.0
1.2
3.8

2.4




1.5(2.7)
4.8 (7.2)
3.2(6.1)
7.4
(12.7)
5.9 (8.7)
6.4
(10.6)
10.0
(12.8)
7.4
(10.7)
5.3 (7.0)
12.1
(25.8)
EC Mean TC Mean
(Max) (Max)

3.3(8.1)
1.9(3.6)
0.4
0.1
0.2
0.7 (2.2)

0.2
0.2
0.5
0.2
0.7

0.4
2.9 (5.4)


3.0
0.4 (0.6) 5-7
2.6 (3.9) hPM25;
1.0(1.3) Q+TQ;
1.8(2.5) TOR8
1.4(2.4)
1.9(2.7)
2.9(4.1)
3.3 (4.4)
1.6(3.0)
1.9(3.5)





Avg.
Time Notes

6 h PM25; Q; TOR

12 h (day) PM21;QQ;TORe
PM21; Q+TQ; TORd
PML8; Imp; TMO
12 h PM25; QQ; TORf

24 h PM25; QQ; TOR






PMLO; QQ ; EGA6


12 h PM25; VDQA; EGA0
















-------
to
O
o
             TABLE 3C-1 (cont'd). PARTICULATE ORGANIC AND ELEMENTAL CARBON CONCENTRATIONS (in jig C/m3)
                                                       BASED ON STUDIES PUBLISHED AFTER 1995
p
-k
fe
H
6
o
o
H
O
O
H
W
O
O
Reference
RURAL PM2 5
Malm and Day (2000)
PM10
Omar etal. (1999)
Gertleretal. (1995)
Chow etal. (1996)





Lioy and Daisey (1987)






Location

Grand Canyon, AZ

Bondville, IL
Bullhead City, AZ
Santa Barbara, CA (urban)
Santa Maria, CA (urban)
Santa Ynez, CA (airport)
Gaviota, CA (rural SB)
Watt Road, CA (rural SB)
Anacapa Island, CA

Newark, NJ

Elizabeth, NJ

Camden, NJ

Dates

July-Aug 1998

Jan-Dec 1994
Sept 1988-Oct 1989
Jan-Dec 1989





1982:
Summer
Winter
Summer
Winter
Summer
Winter
OC Mean
(Max)

1.1(1.6)

2.6
6.0 (16.0)







4.1
5.9
2.1
7.1
2.2
5.2
EC Mean TC Mean Avg.
(Max) (Max) Time Notes

0. 10 (0.3) 24 h PM25; QQ; TORf

0.2 24-48 h PM10; Q; TOR
1.9 (4.0) 24 h PM10; Q; TOR
8.8 24 h PM10; Q; TOR
4.6
3.5
3.4
2.1
3.1
PM15; Q
3.0
3.3
1.7
2.3
1.3
2.0
         A limited amount of rural data is presented. In some cases, total carbon (TC = OC + EC) is reported. OC concentrations must be multiplied by the average molecular weight per
         carbon weight to convert to mass of particulate organic compounds. The location and dates over which sampling occurred are provided. Averaging time refers to the sampling
         duration. Sampling method: Q - quartz fiber filter; QQ - two quartz fiber filters in series; Q+TQ - a quartz fiber filter in one port and a Teflon followed by a quartz filter in a
         parallel port; Imp - cascade impactor; DQQ - denuder followed by two quartz fiber filters; DQA - denuder followed by quartz fiber filter and adsorbent; VDQA - virtual
         impactor inlet followed by denuder, quartz filter, and adsorbent. Analysis method is reported as follows:  TOR - thermal optical reflectance; TOT - thermal optical
         transmittance; TMO - thermal MnO2 oxidation; EGA - evolved gas analysis; Th - Thermal analysis; Aeth - Aethalometer. na - data not available.

         aRange is provided. It should be noted that samples were collected only during elevated pollution episodes and are not representative of average concentrations.
         bParticulate OC was considered to be the sum of front and back quartz fiber filters.
         cSum of adsorbent and filter after correction for inlet losses and denuder efficiency.
         ''Corrected for adsorption by subtracting the Teflon-quartz back-up filter.
         'Reported concentrations are corrected for adsorption by subtracting the quartz (TQ or QQ) back-up filter.
         'Sampler contained two quartz fiber filters in series, but publication did not indicate whether the quartz back-up filter was subtracted to correct for adsorption.
         8Corrected for adsorption using Micro-Orifice Uniform Deposit Impactor (MOUDI) data from a collocated sampler.

-------
TABLE 3C-2. PARTICULATE ORGANIC COMPOUND CONCENTRATIONS (in ng C/m3) BASED ON STUDIES
                        PUBLISHED AFTER 1990 AT SELECTED SITES
Rogge et al. (1993)a
Jan-Dec 1982
(annual average)
PM7,

n-Alkanes
n-tricosane
n-tetracosane
n-pentacosane
n-hexacosane
n-heptacosane
n-octacosane
n-nonacosane
n-triacontane
n-hentriacontane
n-dotriacontane
n-tritriacontane
n-tetratriacontane
Total n-alkanes
n-Alkanoic Acids
n-nonanoic acid
n-decanoic acid
n-undecanoic acid
n-dodecanoic acid
n-tridecanoic acid
n-tetradecanoic acid
n-pentadecanoic acid
n-hexadecanoic acid
(palmitic acid)
n-heptadecanoic acid
n-octadecanoic acid
(stearic acid)
n-nonadecanoic acid
n-eicosanoic acid
n-heneicosanoic acid
Los Angeles,
CA

6.7
6.4
11.2
8.2
6.7
3.1
7.1
2.7
12.6
1.5
2.1
0.58
68.9

6.6
2.0
2.8
5.3
4.3
19.7
5.3
140.5

4.7
59.2

1.1
5.1
2.1
Pasadena,
CA

5.4
4.7
9.5
4.3
5.6
2.5
4.7
2.5
9.6
1.5
2.3
0.68
53.3

5.3
2.4
6.0
7.0
4.9
22.2
6.1
127.4

5.2
50.0

1.1
6.1
2.3
Schauer and Cass (2000) Khwaja (1995) Allen et al. (1997)
Dec 26-28, 1995 Veltkamp et al. (1996) October 1991 Summer 1994 Fraser et al. (1998)
(pollution episode) July 24-Aug 4, 1989 (semiurban) (urban) Sept 8-9, 1993
PM, , no precut no precut PM, , (urban)
Fresno,
CA


42.3
41.2
29.9
25.0
12.3
33.8
7.39
16.1
2.61
5.02

215.6


0.711

0.905
6.17
9.42
33.7
166

13.6
60.0

10.7
41.2
20.8
Bakersfleld,
CA


12.7
14.2
10.7
10.8
5.24
23.6
4.27
9.66
3.50
3.31

98.0


0.164

0.803
1.78
4.01
5.63
54.4

3.77
24.1

2.58
10.4
6.46
Niwot Ridge, Schenectady, Kenmore Square, Los Angeles Basin,
CO NY Boston, MA CA

19.23(57.7)
6.04(21.1)
7.77(21.3)
2.08(12.7)
5.62(15.1)
1.26(9.0)
7.70 (20.6)
0.76 (4.6)
5.24(17.9)
0.41(2.1)
1.49(5.5)

57.9

















-------
< — I
ft
to
o
o
OJ








OJ
O
ON

O
£
H
6
0
0
H
O
0
H
W
O
O
HH
s


TABLE 3C-2 (cont'd). PARTICULATE ORGANIC COMPOUND CONCENTRATIONS (in ng C/m3) BASED ON STUDIES
PUBLISHED AFTER 1990 AT SELECTED SITES
Rogge et al. (1993)a
Jan-Dec 1982
(annual average)
PM21

n-Alkanoic Acids
(cont'd)
n-docosanoic acid
n-tricosanoic acid
n-tetracosanoic acid
n-pentacosanoic acid
n-hexacosanoic acid
n-heptacosanoic acid
n-octacosanoic acid
n-nonacosanoic acid
n-triacontanoic acid
Total n-alcanoic acids
n-Alkenoic Acids
n-9-hexadecenoic acid
n-9-octadecenoic acid
n-9, 12-octadecane-
dienoic acid
Total n-alkenoic acids
n-Alkanals
1 -octanal
n-nonanal
n-decanal
n-dodecanal
n-tridecanal
n-tetradecanal
n-pentadecanal
n-hexadecanal
n-heptadecanal
n-octadecanal
Total n-alkanals
Los Angeles,
CA


8.7
2.0
11.8
1.3
5.6
0.49
2.7
0.33
1.0
292.6

24.8
24.8

5.7




5.7
Pasadena,
CA


9.9
2.5
16.5
1.6
9.3
0.81
4.9
0.57
2.2
294.3

26.0
26.0

9.5




9.5
Schauer and Cass (2000)
Dec 26-28, 1995
(pollution episode)
PM25
Fresno,
CA


160
32.1
205
15.4
174
2.56
21.3
1.46
4.32
979.3

18.8
27.1
13.6
59.5

19.4




19.4
Bakersfield,
CA


43.1
9.71
78.0
6.59
81.3
2.38
9.65
2.11
5.79
352.7

3.96
3.96
1.83
9.75

3.01




3.01
Khwaja (1995) Allen et al. (1997)
Veltkamp et al. (1996) October 1991 Summer 1994 Fraser et al. (1998)
July 24- Aug 4, 1989 (semiurban) (urban) Sept 8-9, 1993
no precut no precut PMl 9 (urban)
Niwot Ridge, Schenectady, Kenmore Square, Los Angeles Basin,
CO NY Boston, MA CA














3.26(14.4)
29.01 (62.8)
23.58(71.2)
6.01 (16.4)
6.50(25.8)
9.62 (30.7)
12.47(113.6)
17.45 (49.3)
24.09 (88.9)
1.84(11.7)
133.8

-------
to
O
o
      TABLE 3C-2 (cont'd). PARTICULATE ORGANIC COMPOUND CONCENTRATIONS (in ng C/m3) BASED ON STUDIES

                                  PUBLISHED AFTER 1990 AT SELECTED SITES
O
fe
H

6
o


o
H

O


O
H
W

O


O

n-Alkanols
1-decanol
1-dodecanol
1 -tetradecanol
1 -pentadecanol
1 -hexadecanol
Total n-alkanok
Aliphatic Dicarboxylic
Acids
oxalic acid (C2)
malonic acid
(propanedioic)
methylmalonic acid
(methylpropanedioic)
malonic acid
(2-butenedioic)
succinic acid
(butanedioic)
methylsuccinic acid
(methylbutanedioic)
glutaric acid
(pentanedioic)
methylglutaric acid
(methylpentanedioic)
hydroxybutanedioic acid
adipic acid
(hexanedioic)
pimelic acid
(heptanedioic)
suberic acid
(octanedioic)
axelaic acid
(nonanedioic)
Total aliphatic
dicarboxylic acids
Rogge et al. (1993)"
Jan-Dec 1982
(annual average)
PM,,
Los Angeles, Pasadena,
CA CA









32.7 44.4

0.66 1.3

66.5 51.2
18.0 15.0
32.3 28.3
19.3 16.6
14.3 16.0
14.1 14.1

3.4 4.1

29.0 22.8

230.3 213.8
Schauer and Cass (2000) Khwaja (1995) Allen et al. (1997)
Dec 26-28, 1995 Veltkamp et al. (1996) October 1991 Summer 1994 Fraser et al. (1998)
(pollution episode) July 24-Aug 4, 1989 (semiurban) (urban) Sept 8-9, 1993
PM, , no precut no precut PM, Q (urban)
Fresno, Bakersfield, Niwot Ridge, Schenectady, Kenmore Square, Los Angeles Basin,
CA CA CO NY Boston, MA CA

8.66(64.1)
21.29(61.7)
13.59(41.4)
4.50(30.1)
27.42(141.1)
75.5

198(360)
84(107)
2.13 nd


102(167)
24.0 8.80
21.3 10.5


3.39 3.07
2.22 1.03
4.41 13.4

19.9 8.22

77.4 45.0 384

-------
»
O
 TABLE 3C-2 (cont'd). PARTICULATE ORGANIC COMPOUND CONCENTRATIONS (in ng C/m3) BASED ON STUDIES
                                              PUBLISHED AFTER 1990 AT SELECTED SITES
o
o
(UJ>




Ketocarboxylic Acids
pyruvic acid (C3)
glyoxylic acid (C2)
Total ketocarboxylic acids
Rogge et al. (1993)a
Jan-Dec 1982
(annual average)
PM21
Los Angeles, Pasadena,
CA CA




Schauer and Cass (2000)
Dec 26-28, 1995
(pollution episode)
PM25
Fresno, Bakersfield,
CA CA




Veltkamp et al. (1996)
July24-Aug4, 1989
no precut
Niwot Ridge,
CO




Khwaja (1995)
October 1991
(semiurban)
no precut
Schenectady,
NY

59(103)
44 (68)
103
Allen et al. (1997)
Summer 1994
(urban)
PML9
Kenmore Square,
Boston, MA




Fraser et al. (1998)
Sept 8-9, 1993
(urban)
Los Angeles Basin,
CA




(UJ>
p
oo>
a
a
o
x
o
a
o
>H
W
O
"7®
O
Diterpenoid/Resin Acids
dehydroabietic acid              23.6
abietic acid
13-isopropyl-5a-                 0.63
podocarpa-6,8,11,13-
tetraen-16-oic acid
8,15-pimaradien-18-oic           0.44
acid
pimaric acid
isopimaric acid
7-oxodehydroabietic acid
abieta-6,8,11,13,15-
pentaen-18-oic acid
abieta-8,11,13,15-tetraen-
18-oic acid
sandaracopimaric acid            1.6
Total diterpenoid acids          33.3
Aromatic Polycarboxylic
Acids
1,2-benzene-dicarboxylic         60.0
acid (phthalic acid)
1,3-benzene-dicarboxylic          3.4
acid
                                                    22.6
 1.2
                                                     0.57
 2.2
37.6
                                                    55.7

                                                     2.9
                                                                98.5
                                                                30.4
                                                                 0.48
  2.62

  8.91
296.4
             9.16

             3.41
                         8.01
                         0.784
                                                                            0.03
2.3
1.3
3.4

4.8
2.3
4.1

9.97
127
6.68
11.8
0.735
7.95
1.43
2.43
 0.251

 0.525
22.15
             6.78

             1.98

-------
•— ^
3

to
O
o
OJ
TABLE 3C-2 (cont
'd). PARTIC
ULATE O
RGANIC C
IMPOUND CONCENTRATIONS (in ng C/m3) BASED
ON STUDIES
PUBLISHED AFTER 1990 AT SELECTED SITES




Rogge et al.
Jan-Dec

(1993)a
1982
(annual average)













O
1
VO



O
£
H
6
0
0
H
O
0
H
W
O
O
HH
w




Aromatic Polycarboxylic
Acids (cont'd)
1 ,4-benzene-dicarboxylic
acid
benzene tricarboxylic acids
4-methyl-l,2-
benzenedicarboxylic acid
1 ,2,4-benzene-tricarboxylic
acid (trimellitic acid)
1,3,5-benzene-tricarboxylic
acid (trimesic acid)

1,2,4,5-benzene-
tetracarboxylic acid
(pyromellitic acid)
Total aromatic
polycarboxylic acids

Poly cyclic Aromatic
Hydrocarbons
retene
fluoranthene
acephenanthrylene
pyrene
Cr202 MW PAH
C2-202 MW PAH
benz [ajanthracene
cyclopenta[c
-------
to
O
o
       TABLE 3C-2 (cont'd). PARTICULATE ORGANIC COMPOUND CONCENTRATIONS (in ng C/m3) BASED ON STUDIES

                                   PUBLISHED AFTER 1990 AT SELECTED SITES
o
I

O
fe
H
6
o

o
H
O

o
H
W
O

O
Rogge et al. (1993)" Schauer and Cass (2000)
Jan-Dec 1982 Dec 26-28, 1995 Veltkamp et al. (1996)
(annual average) (pollution episode) July 24-Aug 4, 1989
PM, , PM, < no precut
Los Angeles, Pasadena, Fresno, Bakersfield, Niwot Ridge,
CA CA CA CA CO
Poly cyclic Aromatic
Hydrocarbons (cont'd)
Cr228MWPAH 17.6 5.35
C2-228 MW PAH
benz [ejacephen-anthrylene
benzo[/t]fluoranthene 1.15 1.20 8.69 2.13
benzo[6]fluoranthene 1.23 0.85 10.7 2.48
benzo[/]fluoranthene 3.62 0.499
benzo[e]pyrene 0.97 0.93 7.20 1.98
benzo[a]pyrene 0.42 0.44 8.23 1.77
perylene 1.50 0.246
methyl-substituted 252 MW
PAH
mdeno[l,2,3-cd\-pyrene 0.37 0.42 6.84 2.56
mdeno[l,2,3-cd\- 1.05 1.09 1.36 0.764
fluoranthene
benzo[g/»]perylene 4.47 4.43 9.75 3.49
anthanthrene 0.180 0.131
coronene
Total polycyclic aromatic 11.66 11.10 139.57 34.40
hydrocarbons
Oxygenated PAHs/
Polycyclic Aromatic
Ketones/Quinones
1 ,4-naphthoquinone
1 -acenaphthenone
9-fluorenone
1,8-naphthalic anhydride
phenanthrenequinone
phenalen-9-one
anthracene-9, 1 0-dione
methylanthracene-9, 10-
dione
1 lH-benzo[a]fluoren-l 1-
one
Khwaja (1995) Allen et al. (1997)
October 1991 Summer 1994 Fraser et al. (1998)
(semiurban) (urban) Sept 8-9, 1993
no precut PM, „ (urban)
Schenectady, Kenmore Square, Los Angeles Basin,
NY Boston, MA CA


0.34(2.16)
0.09 (0.46)
0.20(1.00)
0.22(1.07)

0.02(0.10)
0.22(1.00)
0.14(0.80)
0.05(0.51)
0.10(0.88)

0.29(1.38)
0.10(0.46)

0.77 (4.23)


3.77





0.26
2.07 0.29(1.04)
1.77 0.41(1.65)
0.43
0.53 (2.23)
0.36(1.14)
0.09 (0.24)

1.03


-------
 TABLE 3C-2 (cont'd). PARTICULATE ORGANIC COMPOUND CONCENTRATIONS (in ng C/m3) BASED ON STUDIES
                                              PUBLISHED AFTER 1990 AT SELECTED SITES
Rogge et al. (1993)'
Jan-Dec 1982
(annual average)
PM21
Los Angeles, Pasadena,
CA CA
Schauer and Cass (2000)
Dec 26-28, 1995
(pollution episode)
PM25
Fresno, Bakersfield,
CA CA
Veltkamp et al. (1996)
July24-Aug4, 1989
no precut
Niwot Ridge,
CO
Khwaja (1995)
October 1991
(semiurban)
no precut
Schenectady,
NY
Allen et al. (1997)
Summer 1994
(urban)
PM19
Kenmore Square,
Boston, MA
Fraser et al. (1998)
Sept 8-9, 1993
(urban)
Los Angeles Basin,
CA
Oxygenated PAHs/
Poly cyclic Aromatic
Ketones/Quinones (cont'd)
7H-benzo[c]fluoren-7-one
1 lH-benzo[6]fluoren-l 1-
one
1 H-phenalen-1 -one
benzanthrone
5,12-naphthacene-quinone
7H-benz[fife]-anthracen-7-          0.81
one
benz [fife]anthracene-7-dione
benz[a]anthracene-7,12-            0.21
dione
cyclopenta[fife/]phen-
anthrone
benzo[c
-------
to
O
o
       TABLE 3C-2 (cont'd).  PARTICULATE ORGANIC COMPOUND CONCENTRATIONS (in ng C/m3) BASED ON STUDIES

                                   PUBLISHED AFTER 1990 AT SELECTED SITES
O
i

to
fe
H

6
o


o
H

O

o
H
W

O


O
Rogge et al. (1993)" Schauer and Cass (2000) Khwaja (1995) Allen et al. (1997)
Jan-Dec 1982 Dec 26-28, 1995 Veltkamp et al. (1996) October 1991 Summer 1994 Fraser et al. (1998)
(annual average) (pollution episode) July 24-Aug 4, 1989 (semiurban) (urban) Sept 8-9, 1993
PM21 PM25 no precut no precut PM19 (urban)

Guaiacol and Substituted
Guaiacols
guaiacol
4-methylguaiacol
trans-isoeugenol
vanillin
acetovanillone
guaiacyl acetone
coniferyl aldehyde
Total guaiacol and
substituted guaiacols
Syringol and Substituted
Syringols
syringol
4-methylsyringol
4-ethylsyringol
4-propylsyringol
4-propenylsyringol
syringaldehyde
acetosyringone
acetonylsyringol
propionylsyringol
butyrylsyringol
sinapyl aldehyde
Total syringol and
substituted syringols
Sugars
levoglucosan
other sugars
Total sugars
Los Angeles, Pasadena, Fresno,
CA CA CA


0.889
0.606
1.45
26.8
3.23
10.8
47.0
90.78

1.16
1.72
2.28
0.871
4.38
135
171
406
32.1
15.3
15.9
785.7

7590
1070
8660
Bakersfield, Niwot Ridge, Schenectady, Kenmore Square, Los Angeles Basin,
CA CO NY Boston, MA CA


0.832
0.387
1.04
6.05
0.705
4.29
nd
13.30

0.845
1.77
2.39
nd
1.40
44.5
55.7
68.1
16.2
6.18

197.1

1100
171
1271

-------
to
O
o
              TABLE 3C-2 (cont'd).  PARTICULATE ORGANIC COMPOUND CONCENTRATIONS (in ng C/m3) BASED ON
                                            STUDIES PUBLISHED AFTER 1990 AT SELECTED SITES
Rogge et al. (1993)a Schauer and Cass (2000) Veltkamp et al.
Jan-Dec 1982 Dec 26-28, 1995 (1996)
(annual average) (pollution episode) July 24- Aug 4, 1989
PM2 ! PM2 5 no precut

Other Compounds
divanillyl
divanillyl methane
vanillylmethylguaiacol
Total other
Los Angeles, Pasadena, Fresno,
CA CA CA

19.4
2.39
3.24
25.0
Bakersfield, Niwot Ridge,
CA CO

3.18
nd
0.568
3.75
Khwaja (1995) Allen et al. (1997) Fraser et al.
October 1991 Summer 1994 (1998)
(semiurban) (urban) Sept 8-9, 1993
no precut PMj 9 (urban)
Schenectady, Kenmore Square, Los Angeles
NY Boston, MA Basin, CA





o
 I
UJ
fe
H
6
o
o
H
O
o
H
W
O
O
         N-Containing Compounds
         3-methoxypyridine
         isoquinoline
         1 -methoxypyridine
         1,2-dimethoxy-4-nitro-
         benzene
         dihydroxynitrobenzene
         Total N-containing
         compounds
         Total Quantified Organic
         Compound Mass
         Total Organic Compound
         Mass
         Percent of Organic Mass
         Quantified
         Percent of Organic Mass
         Extractable and Elutable
   0.86
   1.1
   0.27
   1.8
   4.03
                                      789
1.4
1.1
0.24
3.9
                  6.64
                                                   764
8-15% (a)     8-15% (a)

45-60% (a)    45-60% (a)
          11410

          55700

           20%

           30%
                                                                            2075

                                                                            18700

                                                                            11%

                                                                            21%
                                                                                            267
                                                                                                             487
                                                                                                        1.62(10.52)
                                                                                                        1.62
                                                                                                                              10
         Mean values are provided with maximum concentrations in parentheses.

         "Rogge et al. (1993) summarized these percentages for all four Los Angeles Basin sampling sites (West LA, Downtown LA, Pasadena, and Rubidoux). Only Downtown LA and Pasadena
         data are
         shown here.

-------
  1       REFERENCES

  2       Allen, J. O.; Dookeran, N. M; Taghizadeh, K.; Lafleur, A. L.; Smith, K. A.; Sarofim, A. F. (1997) Measurement of
  3             oxygenated polycyclic aromatic hydrocarbons associated with a size-segregated urban aerosol. Environ. Sci.
  4             Technol. 31:2064-2070.
  5       Allen, G. A.; Lawrence, J.; Koutrakis, P. (1999) Field validation of a semi-continuous method for aerosol black
  6             carbon (aethalometer) and temporal patterns of summertime hourly black carbon measurements in
  7             southwestern PA. Atmos. Environ. 33: 817-823.
  8       Andrews, E.; Saxena, P.; Musarra, S.; Hildemann, L .M.; Koutrakis, P.; McMurry, P. H.; Olmez, L; White, W. H.
  9             (2000) Concentration and composition of atmospheric aerosols from the 1995 SEAVS experiment and a
10             review of the closure between chemical and gravimetric measurements. J. Air Waste Manage. Assoc.
11             50: 648-664.
12       Chow, J. C.; Watson, J. G.; Lu, Z.; Lowenthal, D. H.; Frazier, C. A.; Solomon, P. A.; Thuillier, R. H.; Magliano, K.
13             (1996) Descriptive analysis of PM2 5 and PM10 at regionally representative locations during
14             SJVAQS/AUSPEX. In: Parrish, D.; Trainer, M.; Rao, S. T.; Solomon, P. A., eds. A&WMA international
15             specialty conference on regional photochemical measurements and modeling, part 2; November 1993;
16             San Diego, CA. Atmos. Environ. 30: 2079-2112.
17       Christoforou, C. S.; Salmon, L. G.; Hannigan, M. P.; Solomon, P. A.; Cass, G. R. (2000) Trends in fine particle
18             concentration and chemical composition in southern California. J. Air Waste Manage. Assoc. 50: 43-53.
19       Cui, W.; Machir, J.; Lewis , L.; Eatough, D. J.; Eatough, N. L. (1997) Fine paniculate organic material at Meadview
20             during the project MOHAVE summer intensive study. J. Air Waste Manage. Assoc. 47: 357-369.
21       Fraser, M. P.; Cass, G. R.; Simoneit, B. R. T.; Rasmussen, R. A. (1998) Air quality model evaluation data for
22             organics. 5. C6-C22 nonpolarand semipolar aromatic compounds. Environ. Sci. Technol. 32: 1760-1770.
23       Gertler, A. W.; Lowenthal, D. A.; Coulombe, W. G. (1995) PM10 source apportionment study in Bullhead City,
24             Arizona. J. Air Waste Manage. Assoc. 45: 75-82.
25       Hegg, D. A.; Livingston, J.; Hobbs, P. V.; Novakov, T.; Russell, P. (1997) Chemical apportionment of aerosol
26             column optical depth off the mid-Atlantic coast of the United States. J. Geophys. Res. 102: 25,293-25,303.
27       IMPROVE: interagency monitoring of protected visual environments [database]. (2000) [Data on paniculate organic
28             and elemental carbon concentrations after 1995]. Fort Collins, CO: National Park Service (NFS);  Cooperative
29             Institute for Research in the Atmosphere (CIRA). Available at: http://vista.cira.colostate.edu/improve/ [2001,
30             January 26].
31       Khwaja, H. (1995) Atmospheric concentrations of carboxylic acids and related compounds at a semiurban site.
32             Atmos. Environ. 29: 127-139.
33       Klinedinst, D. B.; Currie, L. A. (1999) Direct quantification of PM2 5 fossil and biomass carbon within the northern
34             front range air quality study's domain. Environ. Sci. Technol. 33: 4146-4154.
35       Lewtas, J.; Pang, Y.; Booth, D.; Reimer,  S.; Eatough, D. J.; Gundel, L. A. (2001) Comparison of sampling methods
36             for semi-volatile organic carbon associated with PM2 5. Aerosol. Sci. Technol. 34: 9-22.
37       Lioy, P. J.; Daisey, J. M. (1987) Toxic air pollution: a comprehensive study of non-criteria air pollutants. Chelsea,
38             MI: Lewis Publishers.
39       Malm, W. C.; Day, D. E. (2000) Optical properties of aerosols at Grand Canyon National Park. Atmos. Environ.
40             34: 3373-3391.
41       Malm, W. C.; Gebhart, K. A. (1996) Source apportionment of organic and light-absorbing carbon using receptor
42             modeling techniques. Atmos. Environ. 30: 843-855.
43       Offenberg, J. H.; Baker, J. E. (2000) Aerosol size distributions of elemental and organic carbon in urban and
44             over-water atmospheres. Atmos. Environ. 34: 1509-1517.
45       Omar, A. H.; Biegalski, S.; Larson, S. M.; Landsberger, S. (1999) Paniculate contributions to light extinction and
46             local forcing at a rural Illinois  site. Atmos. Environ. 33: 2637-2646.
47       Pedersen, D.  U.; Durant, J. L.; Penman, B. W.; Crespi, C. L.; Hemond, H. F.; Lafleur, A. L.; Cass, G. R. (1999)
48             Seasonal and spatial variations in human cell mutagenicity of respirable airborne particles in the northeastern
49             United States. Environ. Sci. Technol. 33: 4407-4415.
50       Rogge, W. F.; Mazurek, M. A.; Hildemann, L. M.; Cass, G. R.; Simoneit, B. R. T. (1993) Quantification of urban
51             organic aerosols at a molecular level: identification, abundance and seasonal variation. Atmos. Environ.
52             Part A 27: 1309-1330.
53       Saxena, P.; Hildemann, L. M. (1996) Water-soluble organics in atmospheric particles: a critical review of the
54             literature and applications of thermodynamics to identify candidate compounds. J. Atmos. Chem. 24: 57-109.
55
         June 2003                                      3C-14        DRAFT-DO NOT QUOTE OR CITE

-------
 1       Schauer, J. I; Cass, G. R. (2000) Source apportionment of wintertime gas-phase and particle-phase air pollutants
 2            using organic compounds as tracers. Environ. Sci. Technol. 34: 1821-1832.
 3       Turpin, B. I; Huntzicker, J. J. (1995) Identification of secondary organic aerosol episodes and quantitation of
 4            primary and secondary organic aerosol concentrations during SCAQS. Atmos. Environ. 29: 3527-3544.
 5       U.S. Environmental Protection Agency. (1996) Air quality criteria for particulate matter. Research Triangle Park,
 6            NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
 7       Veltkamp, P. R.; Hansen, K. J.; Barkley, R. M; Sievers, R. E. (1996) Principal component analysis of summertime
 8            organic aerosols at Niwot Ridge, Colorado. J. Geophys. Res. [Atmos.] 101: 19,495-19,504.
 9
10
         June 2003                                    3C-15         DRAFT-DO NOT QUOTE OR CITE

-------
 i                                     APPENDIX 3D
 2
 3           Composition of Particulate Matter Source Emissions
 4
 5
 6          This appendix includes discussions of the elemental composition of emissions from various
 7     source categories discussed in Table 3-8. Discussions in this appendix incorporate material
 8     dealing with the inorganic components of source emissions from Chapter 5 of the 1996 PM
 9     AQCD (U. S. Environmental Protection Agency, 1996), updates to that material, and material
10     describing the composition of organic components in source emissions. Primary emphasis is
11     placed in the discussions on the composition of PM25 sources.
12
13     Soil and Fugitive Dust
14          The compositions of soils and average crustal material are shown in Table 3D-1 (adapted
15     from Warneck, 1988). Two entries are shown as representations of average crustal material.
16     Differences from the mean soil composition shown can result from local  geology and climate.
17     Major elements in both soil and crustal profiles are Si, Al, and Fe, which are found in the form
18     of various minerals. In addition, organic matter constitutes a few percent, on average, of soils.
19     In general, the soil profile is similar to the crustal profiles, except for the depletion of soluble
20     elements such as Ca, Mg, Na, and K.  It should be noted that the composition of soils from
21     specific locations can vary considerably from these global averages, especially for elements like
22     Ca, Mg, Na, and K.
23          Fugitive dust  emissions arise from paved and unpaved roads, building construction and
24     demolition, parking lots, mining operations, storage piles, feed lots, grain handling, and
25     agricultural tilling, in addition to wind erosion. Figure 3D-1 shows examples of size
26     distributions in dust from paved and unpaved roads, agricultural soil, sand and gravel, and
27     alkaline lake bed sediments, which were measured in a laboratory resuspension chamber as part
28     of a study in California (Chow et al., 1994). This figure shows substantial variation in particle
29     size among some of these fugitive dust sources. The PMX 0 abundance (6.9%) in the total
30     suspended PM (TSP) from alkaline lake bed dust is twice its abundance in paved and unpaved
31     road dust. Approximately 10% of the TSP is in the PM2 5 fraction and approximately 50% of

       June 2003                               3D-1        DRAFT-DO NOT QUOTE OR CITE

-------
   TABLE 3D-1. AVERAGE ABUNDANCES OF MAJOR ELEMENTS IN SOIL AND
  	CRUSTAL ROCK	
                               Elemental Abundances (ppmw)
Element
Si
Al
Fe
Ca
Mg
Na
K
Ti
Mn
Cr
V
Co
Soil
(1)
330,000
71,300
38,000
13,700
6,300
6,300
13,600
4,600
850
200
100
8

(2)
277,200
81,300
50,000
36,300
20,900
28,300
25,900
4,400
950
100
135
25
Crustal Rock
(3)
311,000
77,400
34,300
25,700
33,000
31,900
29,500
4,400
670
48
98
12
 Source: (1) Vinogradov (1959); (2) Mason (1966); (3) Turekian (1971), Model A; as quoted in Warneck (1988).
               100
                      Paved      Unpaved     Agricultural    Soil/Gravel    Alkaline
                     Road Dust   Road Dust      Soil                 Lake Bed
                          CZ]<1.0|jm
                                                      m TSP
Figure 3D-1.  Size distribution of particles generated in a laboratory resuspension
              chamber.

Source: Chow etal. (1994).
June 2003
3D-2
DRAFT-DO NOT QUOTE OR CITE

-------
 1      TSP is in the PM10 fraction. The sand/gravel dust sample shows that 65% of the mass is in
 2      particles larger than the PM10 fraction.  The PM2 5 fraction of TSP is approximately 30 to 40%
 3      higher in alkaline lake beds and sand/gravel than in the other soil types. The tests were
 4      performed after sieving and with a short (< 1 min) waiting period prior to sampling. It is
 5      expected that the fraction of PMj 0 and PM25 would increase with distance from a fugitive dust
 6      emitter, as the larger particles deposit to the surface faster than do the smaller particles.
 7           The size distribution of samples of paved road dust obtained from a source characterization
 8      study in  California is shown in Figure 3D-2. As might be expected, most of the emissions are in
 9      the coarse size mode.  The chemical composition of paved road dust obtained in Denver, CO,
10      during the winter of 1987-1988 is shown in Figure 3D-3. The chemical composition of paved
11      road dust consists of a complex mixture of particulate matter from a wide variety of sources.
12      Hopke et al. (1980) found that the inorganic composition of urban roadway dust in samples from
13      Urbana,  IL, could be described in terms of contributions from natural soil, automobile exhaust,
14      rust, tire wear, and salt. Automobile contributions arose from exhaust emissions enriched in Pb;
15      from rust as Fe; tire wear particles enriched in Zn; brake linings enriched in Cr, Ba, and Mn; and
16      cement particles derived from roadways by abrasion.  In addition to organic compounds from
17      combustion and secondary sources, road dust also contains biological material such as pollen
18      and fungal spores.
19           Very limited data exist for characterizing the composition in organic compounds found in
20      resuspended paved road dust and  soil dust. The only reported measurements are from Rogge
21      et al. (1993a) and Schauer and  Cass (2000), which consist of data for the fine particle fraction.
22      The resuspended road dust sample analyzed by Rogge et al. (1993a) was collected in Pasadena,
23      CA, during May of 1988. The  sample analyzed by Schauer and Cass (2000) is a composite
24      sample collected at several sites in the Central Valley of California in 1995.  In both cases, road
25      dust samples were resuspended in the laboratory. Samples were drawn through a PM2 0 cyclone
26      upstream of the collection substrate to remove particles with aerodynamic diameters greater than
27      2.0 |im.  It is unclear if these samples are representative of road dust in other locations of the
28      United States. Table 3D-2 summarizes the organic compounds measured in these road dust
29      samples.
30
        June 2003                                 3D-3        DRAFT-DO NOT QUOTE OR CITE

-------
                        100-i
                         80-
                         60-

                         40-
                         20-
                                 52.3%
                                \(<10u)
                                 110.7%
                                 |(<2.5u)
                                           92.8%
                                            10u)
                                           82.7%
                                            :2.5|J)
                                           81.6%
 95.8%
 93.1%
 (<2.5u)
 92.4%
                                                             96.2%
      99.2%

     I 97.4%
      (<2.5u)

     I 87.4%
                                                                               34.9%
                                                                               5.8%
                                                                               (<2.5M)
                                                                               4.6%
                            Road and   Agricultural  Residential    Diesel     Crude Oil   Construction
                            Soil Dust    Burning     Wood      Truck    Combustion    Dust
                                              Combustion   Exhaust
                              Code:  I   I >1 Ou
                                                  2.5u - 1 0u
                                                                1U-2.5M
       Figure 3D-2.  Size distribution of California source emissions, 1986.
       Source: Houcketal. (1989, 1990).


1      Stationary Sources
2           The elemental composition of primary particulate matter emitted in the fine fraction from a
3      variety of power plants and industries in the Philadelphia area is shown in Table 3D-3 as a
4      representative example of emissions from stationary fossil combustion sources (Olmez et al.,
5      1988). Entries for the coal fired power plant show that Si and Al, followed by sulfate, are the
6      major primary constituents produced by coal combustion; whereas fractional abundances of
7      elemental carbon were much lower and organic carbon species were not detected.  Sulfate is the
       June 2003
3D-4
DRAFT-DO NOT QUOTE OR CITE

-------

                                   Chemical Species
Figure 3D-3. Chemical abundances for PM2 5 emissions from paved road dust in Denver,
            CO. Solid bars represent fractional abundances, and the error bars
            represent variability in species abundances. Error bars represent detection
            limits when there are no solid bars.

Source: Watson and Chow (1994).
     TABLE 3D-2. SUMMARY OF PARTICLE-PHASE ORGANIC COMPOUNDS
               PRESENT IN FINE PARTICLE ROAD DUST SAMPLE
Source
Pasadena Road Dust
(Rogge et al., 1993a)





San Joaquin Valley
Road Dust (Schauer
and Cass, 2000)
Contribution to Dominant Contributors to
Compound Class Paniculate Mass (%) Emissions of Compound Class
n-Alkanes
n-Alkanoic acids
n-Alkenoic acids
Petroleum biomarkers
PAH
n-Alkanals
n-Alkanols
n-Alkanes
n-Alkanoic acids
n-Alkenoic acids
0.13
0.37
0.028
0.017
0.0059
0.046
0.021
0.023
0.23
0.095
Cn, C19, C21
Palmitic acid and stearic acid
Oleic acid and linoleic Acid
Hopanes and steranes
No dominant compounds
Octacosanol and triacontanal
Hexacosanol and octacosanol
No dominant compounds
Palmitic acid and stearic acid
Oleic acid, linoleic acid, and
hexadecenoic acid
June 2003
3D-5
DRAFT-DO NOT QUOTE OR CITE

-------
•— ^
3
to
o
o
OJ







OJ
O
1
ON

O
!>
H
O
!2
O
H
O
O
H
W
O
O
HH
H
W

TABLE 3D-3. <
Species Eddystone Coal-
(Units) Fired Power Plant
C-v (%)
C-e (%)
NH4 (%)
Na (%)
Al (%)
Si (%)
P (%)
S (%)

c/~\ /o/x
oV-^4 ^ /o)
Cl (%)

K (%)
Ca (%)
Sc (ppm)
Ti (%)

V (ppm)

Cr (ppm)
Mn (ppm)
Fe (%)

Co (ppm)
Ni (ppm)
ND
0.89 ±0.12
1.89 ±0.19
0.31 ±0.03
14 ±2
21. 8 ±1.6
0.62 ±0.10
3.4 ±0.6

11.9±1.2
0.022 ±0.11

1.20 ±0.09
1.4 ±0.5
42 ±2
1.1 ±0.2

550 ± 170

390 ± 120
290 ± 15
7.6 ±0.4

93 ±10
380 ±50
COMPOSITION OF FINE PARTICLES RELEASED BY VARIOUS STATIONARY
SOURCES IN THE PHILADELPHIA AREA
Oil-Fired Power Plants
: N

3
3
3
3
9
9
9

3
3

9
3
3
3

3

3
3
3

3
9
Eddystone
2.7 ±1.2
7.7 ±1.5
3.5 ±1.6
3.0 ±0.8
0.45 ± 0.09
1.9 ±0.6
1.5 ±0.4
11±2

40 ±4
0.019 ±0.009

0.16 ±0.05
3.6 ±1.0
0.17 ±0.02
0.040 ± 0.044

11500 ±3000

235 ± 10
380 ± 40
1.6 ±0.2

790 ± 150
15000 ± 5000
N
3
3
3
3
3
9
9
9

3
2

9
3
3
9

3

3
3
3

3
9
Schuylkill
0.75 ±0.63
0.22 ±0.17
3.7 ±1.7
3. 3 ±0.8
0.94 ±0.08
2.6 ±0.4
1.0 ±0.2
13 ± 1

45 ±7
ND

0.21 ±0.03
2.3 ±1.0
0.47 ± 0.02
0.12 ±0.02

20,000 ± 3000

230 ± 70
210 ±50
1.7 ±0.4

1100 ±200
19000 ± 2000
N
4
4
4
3
3
11
11
11

4


11
3
3
11

3

3
3
3

3
11
Secondary
Al Plant
1.6 ±1.5
0.18±0.10
2.2 ±0.9
16.3 ±0.8
1.74 ±0.09
3.1 ±2.2
0.45 ±0.27
3±4

5.9 ±2
21±4

10.9 ± 1.5
0.12 ±0.09
0.092 ±0.039
0.024 ± 0.003

36 ±7

410 ±20
120 ±15
0.31 ±0.02

13 ±2
300 ± 100
N
2
2
2
1
1
2
2
2

2
1

2
2
1
2

1

1
1
1

1
2
Fluid Cat.
Cracker
ND
0.16 ±0.05
0.43 ±0.22
0.38 ±0.05
6.8 ±1.2
9.8 ±20.0
ND
4.2 ± 12.6

38 ±4
ND

0.031 ±0.005
0.030 ± 0.004
2.7 ±0.4
0.38 ±0.1

250 ± 70

59 ±8
14 ±3
0.20 ±0.03

15 ±2
220 ± 30
N

3
3
3
3
9

9

3


9
9
3
3

3

3
3
9

3
9
Municipal
Incinerator
0.57 ±0.26
3.5 ±0.2
0.36 ±0.07
6.6 ±3. 5
0.25 ±0.10
1.7 ±0.3
0.63 ±0.12
2.9 ±0.8

6.8 ±2.3
29 ±5

7.6 ±2.3
0.23 ±0.10
0.11 ±0.02
0.030 ±0.015

8.6 ±5.3

99 ±31
165 ± 40
0.22 ±0.05

3.7 ±0.8
290 ± 40
N
4
4
4
3
3
10
10
10

4
3

10
10
1
10

2

3
3
3

3
10

-------

TABLE 3D-3 (cont'd). COMPOSITION OF FINE PARTICLES RELEASED BY VARIOUS STATIONARY
                       SOURCES IN THE PHILADELPHIA AREA
l-^
o
o
OJ






OJ
O
•^

O
H
O
0
H
O
O
H
W
O
O
H
W

Species
(units)
Cu (ppm)
Zn (%)
As (ppm)
Se (ppm)
Br (ppm)
Rb (ppm)
Sr (ppm)
Zr (ppm)
Mo (ppm)
Ag (ppm)
Cd (ppm)
In (ppm)
Sn (ppm)
Sb (ppm)
Cs (ppm)
Ba (ppm)
La (ppm)
Ce (ppm)
Nd (ppm)
Sm (ppm)
Eddy stone
Coal-Fired
Power Plant
290 ± 20
0.041 ±0.005
640 ± 80
250 ± 20
35 ±8
190 ± 80
1290 ± 60
490 ± 190
170 ± 60
ND
ND
0.71 ±0.04
ND
a
9.2 ±0.9
ND
120 ±10
180 ± 10
80 ±26
23 ±2
Oil-Fired Power Plants
N
9
3
3
3
3
1
9
9
2


2
2

3
2
3
3
Eddy stone
980 ± 320
1.3 ±0.3
33 ±6
26 ±9
90 ±60
ND
160 ± 50
140 ±180
930 ±210
ND
ND
ND
320 ±230
370 ±410
ND
1960 ± 100
130 ±30
89 ±23
28 ±5
3.7 ±0.7
N
9
3
1
3
9

9
9
3



9
3
3
3
3
2
3
Schuylkill
1100 ±500
0.78 ±0.30
50 ± 16
23 ±7
45 ± 17
ND
280 ± 70
100 ± 120
1500 ± 300
ND
ND
ND
200 ± 80
1020 ± 90
ND
2000 ± 500
450 ± 30
360 ± 20
230 ± 20
20.5 ±1.5
N
11
3
3
3
11

11
11
3



11
3
3
3
3
3
3
Secondary
Al Plant
450 ± 200
0.079 ± 0.006
15 ±6
66 ±3
630 ± 70
97 ±38
ND
ND
ND
ND
ND
ND
550 ± 540
6 100 ±300
ND
ND
19 ±2
ND
ND
ND
N Fluid Cat. Cracker N
2 14 ± 8 9
1 0.0026 ± 0.0007 3
1 ND
1 15 ±1 3
2 5.6 ±1.8 9
1 ND
36 ±6 9
130 ±50 2
ND
ND
ND
ND
2 ND
1 7.7 ±1.5 3
ND
290 ±90 2
1 3300 ±500 3
2700 ± 400 3
1800 ± 250 3
170 ±20 3
Municipal
Incinerator
1300 ± 500
10.4 ±0.5
64 ±34
42 ± 16
2300 ± 800
230 ± 50
87 ±14
ND
240 ± 130
71 ± 15
1200 ± 700
4.9 ±1.4
6700 ± 1900
1300 ± 1000
5.9 ±3.0
ND
1.1 ±0.5
ND
ND
ND
N
3
3
3
3
10
2
10

10
3
3
3
10
3
3

1



-------

TABLE 3D-3 (cont'd). COMPOSITION OF FINE PARTICLES RELEASED BY VARIOUS
           STATIONARY SOURCES IN THE PHILADELPHIA AREA
l-^
o
o
OJ





OJ
O
oo

o
H
^
O
0
H
O
O
H
W
O
O
H
W
Oil-Fired Power Plants
(units) Fired Power Plant N Eddystone
Eu(ppm) 5.1 ±0.5 3 ND
Gd (ppm) ND ND
Tb(ppm) 3. 3 ±0.3 3 ND
Yb(ppm) 10.3 ±0.5 1 ND
Lu (ppm) ND ND
Hf(ppm) 5. 8 ±0.8 3 0.39 ±0.07
Ta (ppm) ND ND
W(ppm) 20 ±8 1 60 ±5
Au(ppm) ND 0.054 ±0.017
Pb(%) 0.041 ±0.004 9 1.8 ±0.6

Th(ppm) 24 ±2 3 1.9 ±0.5
%mass 24 ±2 6 93.5 ±2.5
aOmitted because of sample contamination.
N = Number of samples.
ND = Not detected.
The "% mass" entries give the average percentage of the
lppm=10-4%
Source: Adapted from Olmez et al. (1988).
N Schuylkill
0.65 ±0.23
ND
0.90 ±0.29
ND
ND
1 ND
ND
2 ND
2 ND
9 1.0 ±0.2

2 ND
6 96 ±2


total emitted mass found



N Al Plant N Cracker N
3 ND 4.9 ±0.7 3
ND 71 ±10 3
3 ND 8.9 ±1.3 3
ND 3.7 ±0.4 3
ND 0.59 ±0.17 3
ND 0.99 ±0.08 3
ND 0.56 ±0.10 3
ND ND
ND ND
11 0.081 ±0.014 2 0.0091 ±0.0021 9

ND 6.2 ±0.7 3
6 81 ±10 2 97 ±2 7


in the fine fraction.


Municipal
Incinerator N
ND
ND
ND
ND
ND
ND
ND
ND
0.56 ±0.27 3
5.8 ±1.2 10

ND
89 ±2 7






-------
 1      major particulate constituent released by the oil fired power plants examined in this study; and,
 2      again, elemental and organic carbon are not among the major species emitted.  Olmez et al.
 3      (1988) also compared their results to a number of similar studies and concluded that their data
 4      should have much wider applicability to receptor model studies in other areas with some of the
 5      same source types.  The high temperature of combustion in power plants results in the almost
 6      complete oxidation  of the carbon in the fuel to CO2 and very small amounts of CO.  Combustion
 7      conditions in smaller boilers and furnaces allow the emission of unburned carbon and sulfur in
 8      more reduced forms such as thiophenes  and inorganic sulfides.  A number of trace elements are
 9      greatly enriched over crustal abundances in different fuels, such as Se in coal and V, Zn, and Ni
10      in oil. In fact, the higher V content of the fuel oil than in coal could help account for the higher
11      sulfate seen in the profiles from the oil-fired power plant compared to the coal-fired power plant
12      because V at combustion temperatures found in power plants is known to catalyze the oxidation
13      of reduced sulfur species.  During combustion at lower temperatures, the emission of reduced
14      sulfur species also occurs. For example, Huffman et al. (2000) identified sulfur species emitted
15      by the combustion of several residual fuels oil (RFO) in a fire tube package boiler that is meant
16      to simulate conditions in small institutional and industrial  boilers. They found that sulfur was
17      emitted not only as  sulfate (26 to 84%),  but as thiophenes  (13 to 39%) with smaller amounts  of
18      sulfides and elemental S.  They also found that Ni, V, Fe,  Cu, Zn, and Pb are present mainly  as
19      sulfates in emissions.  Linak et al. (2000) found, when burning RFO, that the fire tube package
20      boiler produced particles with a bimodal size distribution in which about 0.2% of the mass was
21      associated with particles smaller than 0.1 jim AD, with the rest of the mass lying between
22      0.5 and 100 |im. Miller et al. (1998) found that larger particles consisted mainly of cenospheric
23      carbon; whereas trace metals and sulfates were found concentrated in the smaller particles in a
24      fire tube package boiler. In contrast, when RFO was burning in a refractory-lined combustor
25      that is meant to simulate combustion conditions in a large utility residual oil fired boiler, Linak
26      et al.  (2000) found that particles were distributed essentially unimodally, with a mean diameter
27      of about 0.1 |im.
28           Apart from emissions in the combustion of fossil fuels, trace elements are emitted as the
29      result of various industrial processes such as steel and iron manufacturing and nonferrous metal
30      production (e.g., for Pb, Cu, Ni, Zn, and Cd). As may be expected, emissions factors for the
31      various trace elements are highly source-specific (Nriagu and Pacyna, 1988). Inspection of

        June 2003                                 3D-9        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Table 3D-3 reveals that the emissions from the catalytic cracker and the oil-fired power plant are
 2      greatly enriched in rare-earth elements (such as La) compared to other sources.
 3           Emissions from municipal waste incinerators are heavily enriched in Cl, arising mainly
 4      from the combustion of plastics and metals that form volatile chlorides. The metals can
 5      originate from cans or other metallic objects, and some metals such as Zn and Cd are also
 6      additives in plastics or rubber.  Many elements such as S, Cl, Zn, Br, Ag, Cd, Sn,  In, Sb, and Pb
 7      are enormously enriched compared to their crustal abundances. A comparison of the trace
 8      elemental composition  of incinerator emissions in Philadelphia, PA (shown in Table 3D-3) with
 9      those in Washington  DC, and Chicago, IL (Olmez et al., 1988) shows agreement for most
10      constituents to within less than a factor of two.
11           Very limited data exist for characterizing the chemical composition of organic compounds
12      present in particulate emissions from industrial-scale stationary fuel combustion.  Oros and
13      Simoneit (2000) have presented the abundance and distribution of organic constituents in coal
14      smokes that have been burned under laboratory conditions. This work provides the basis for
15      further investigation  addressing the emissions of coal fired boilers.
16           Rogge et al. (1997a) measured the composition of the organic constituents in the
17      particulate matter emissions from a 50 billion kj/h boiler that was operating at 60% capacity and
18      was burning number  2 distillate fuel oil. The fine carbon particulate matter emissions from this
19      boiler over five tests  were composed of an average of 14% organic carbon and 86% elemental
20      carbon (Hildemann et al., 1991). Significant variability in the distribution of organic compounds
21      present in the emissions from two separate tests was observed. Most of the  identified organic
22      mass consisted of n-alkanonic acids, aromatic acids, n-alkanes, PAH, oxygeanted PAH, and
23      chlorinated compounds. It is unclear if these emissions are representative of typical fuel oil
24      combustion units in the United States.  Rogge et al. (1997b) measured the composition of hot
25      asphalt roofing tar pots, and Rogge et al. (1993b) measured the composition of emissions from
26      home appliances that use natural gas.
27
28      Motor Vehicles
29           Exhaust emissions of particulate matter from gasoline powered motor vehicles and diesel
30      powered vehicles have  changed significantly over the past 25 years (Sawyer and Johnson, 1995;
31      Cadle et al., 1999). These changes have resulted from reformulation of fuels, the wide
32      application of exhaust-gas treatment in gasoline-powered motor vehicles, and changes in engine
        June 2003                                3D-10       DRAFT-DO NOT QUOTE OR CITE

-------
 1      design and operation.  Because of these evolving tailpipe emissions, along with the wide
 2      variability of emissions between vehicles of the same class (Hildemann et al., 1991; Cadle et al.,
 3      1997; Sagebiel et al., 1997; Yanowitz et al., 2000), well-defined average emissions profiles for
 4      the major classes of motor vehicles have not been established. Two sampling strategies have
 5      been employed to obtain motor vehicle emissions profiles:  (1) the measurement of exhaust
 6      emissions from vehicles operating on dynamometers and (2) the measurement of integrated
 7      emissions of motor vehicles driving through roadway tunnels. Dynamometer testing can be used
 8      to measure vehicle emissions operating over an integrated driving cycle and allows the
 9      measurement of emissions from individual vehicles. However, dynamometer testing requires
10      considerable resources and usually precludes testing a very large number of vehicles.  In
11      contrast, a large  number of vehicles can be readily sampled in tunnels; however, vehicles driving
12      through tunnels operate over limited driving conditions, and the measurements represent
13      contributions from a large number of vehicle types. As a result, except in a few cases, tunnel
14      tests have not been effective at developing chemically speciated particulate matter emissions
15      profiles for individual motor vehicle classes. Rather, several studies have measured the
16      contribution of both organic and elemental carbon to the particulate matter emissions from
17      different classes of motor vehicles operating on chassis dynamometers.
18          The principal components emitted by diesel and gasoline fueled vehicles are organic
19      carbon (OC) and elemental carbon (EC) as shown in Tables 3D-4a and 4b. As can be seen, the
20      variability among entries for an individual fuel type is large and overlaps that found between
21      different fuel types. On average,  the abundance of elemental carbon is larger than that of
22      organic carbon in the exhaust of diesel vehicles; whereas organic carbon is the dominant species
23      in the exhaust of gasoline fueled vehicles. Per vehicle mile, total carbon emissions from light
24      and heavy duty diesel vehicles can range from 1 to 2 orders of magnitude higher than those from
25      gasoline vehicles.
26          As might be expected, most of the PM emitted by motor vehicles is in the PM25 size range.
27      Particles in diesel exhaust are typically trimodal (consisting of a nuclei mode, an accumulation
28      mode, and a coarse mode) and are log-normal in form (Kittelson, 1998).  More than 90% of the
29      total number of particles are in the nuclei mode, which contains only about 1 to 20% of the
30      particle mass with a mass median diameter of about 0.02 jim; whereas the accumulation mode
31      (with a mass median diameter of  about 0.25 jim) contains most of the mass with a smaller

        June 2003                                3D-11        DRAFT-DO NOT QUOTE OR CITE

-------
         TABLE 3D-4a.  ORGANIC AND ELEMENTAL CARBON FRACTIONS OF DIESEL
                  AND GASOLINE ENGINE PARTICIPATE MATTER EXHAUST

Heavy-duty diesel engines*
Heavy-duty diesel engines (SPECIATE)b
Light-duty diesel engines0
Light-duty diesel engines (SPECIATE)b
Gasoline engines (hot stabilized)8
Gasoline engines ("smoker" and "high emitter"/'0
Gasoline engines (cold start/
Organic Carbon
19 ± 8%
21 - 36%
30 ± 9%
22 - 43%
56 ± 11%
76 ± 10%
46 ± 14%
Elemental Carbon
75 ± 10%
52 - 54%
61 ± 16%
51-64%
25 ± 15%
7 ± 6%
42 ± 14%
        Tujita et al. (1998) and Watson et al. (1998).
        bSPECIATE database (U.S. Environmental Protection Agency, 1999).
        °Norbeck et al. (1998).
        Source:  U.S. Environmental Protection Agency (2002).
 1     fraction (5 to 20%) contained in the coarse mode.  Kerminin et al. (1997), Bagley et al. (1998),
 2     and Kleeman et al. (2000) also have shown that gasoline and diesel fueled vehicles produce
 3     particles that are mostly less than 2.0 jam in diameter. Cadle et al. (1999) found that 91% of PM
 4     emitted by in-use gasoline vehicles in the Denver area was in the PM2 5 size range, which
 5     increased to 97% for "smokers" (i.e., light-duty vehicles with visible smoke emitted from their
 6     tailpipes) and 98% for light-duty diesels. Durbin et al. (1999) found that about 92% of the PM
 7     was smaller than 2.5 jim for smokers and diesels.  The mass median diameter of the PM emitted
 8     by the gasoline vehicles sampled by Cadle et al. (1999) was about 0.12 jam and increased to
 9     0.18 |im for smokers and diesels. Corresponding average emissions rates of PM25 found by
10     Cadle et al. (1999) were 552 mg/mile for diesels; 222 mg/mile for gasoline smokers; and
11     38 mg/mile for other gasoline vehicles.  The values for gasoline smokers and for diesels appear
12     to be somewhat lower than those given in Table 3D-5; whereas the value for other gasoline
13     vehicles falls in the range given for low and medium gasoline vehicle emissions.
14          Examples of data for the trace elemental composition of the emissions from a number of
15     vehicle  classes obtained December 1997 in Colorado, as part of the North Frontal Range Air
16     Quality Study (NFRAQS), are shown in Table 3D-5.  As can be seen from Table 3D-5,
17     emissions of total carbon (TC), which is equal to the sum of organic carbon (OC) and elemental
       June 2003
3D-12
DRAFT-DO NOT QUOTE OR CITE

-------
           TABLE 3D-4b. CONTRIBUTION OF ORGANIC CARBON TO PARTICULATE
          MATTER CARBON EMISSIONS IN MOTOR VEHICLE EXHAUST COLLECTED
                  FROM VEHICLES OPERATED ON CHASSIS DYNAMOMETERS

Year of Tests
Test Cycle
Number of
Vehicles
OC % of
Total Carbon
Notes
GASOLINE POWERED VEHICLES
Light-duty vehicles
High-CO/VOC-emitting smokers
High-CO/VOC-emitting nonsmokers
Catalyst-equipped vehicles
Noncatalyst vehicles
DIESEL VEHICLES
Light-duty diesel vehicles
Medium-duty diesel vehicles
Heavy-duty diesel vehicles
Heavy-duty diesel vehicles
1996-97
1994
1994
Mid-1980s
Mid-1980s

1996-1997
1996
1992
Mid-1980s
FTP
IM-240
DVI-240
FTP
FTP

FTP
FTP
C
C
195a
7
15
7
6

195a
2
6
2
70
91
76
69
89

40
50b
42
45
A
B
B
C
C

A
D
E
C
        Notes:
        A. From Cadle et al. (1999). Average of summer and winter cold start emissions.
        B. From Sagebiel et al. (1997).  Hot start testing of vehicles identified as either high emitters of carbon monoxide
           or volatile organic compounds (VOCs).
        C. From Hildemann et al. (1991). Cold start tests.
        D. From Schauer et al. (1999). Hot start tests of medium duty vehicles operating on an FTP cycle.
        E. From Lowenthal et al. (1994). Only includes measurement of vehicles powered by diesel fuel operated
           without an exhaust paniculate trap.

        aA total of 195 light duty vehicles were tested that include both gasoline powered vehicles and diesel powered
         vehicles.
        bFraction of paniculate matter consisting of organic carbon was measured with and without an organics denuder
         upstream of paniculate filter. Results reported here represent measurement without an organics denuder for
         consistency with other measurements. Using an organics denuder, the organic carbon comprised 39% of the
         paniculate matter carbon.
        Thriving cycle comprised of multiple idle, steady acceleration, constant speed, deceleration steps (see reference
         for more details).
1      carbon (EC), from gasoline vehicles are highly variable.  Gillies and Gertler (2000) point out that

2      there is greater variability in the concentrations of trace elements and ionic species than for OC

3      and EC among different source profiles (e.g., SPECIATE [U.S. Environmental Protection

4      Agency, 1999], Lawson and Smith [1998], Norbeck et al. [1998]).  They suggest that this may

5      arise because emissions of trace elements are not related only to the combustion process, but also

       June 2003                                  3D-13        DRAFT-DO NOT QUOTE OR CITE

-------
 TABLE 3D-5. EMISSION RATES (mg/mi) FOR CONSTITUENTS OF PARTICIPATE
            MATTER FROM GASOLINE AND DIESEL VEHICLES
Gasoline Vehicles























1
2
3
4
5

TC
OC
EC
NO3
S04=
Na
Mg
Al
Si
P
S
Cl
K
Ca
Fe
Ni
Cu
Zn
Br
Ba
Pb
Source:
Low
9.07 ± 0
6.35 ±0
2.72 ± 0

.75
.54
.52
0.039 ± 0.027
0.158 ±0
0.060 ± 0
0.036 ±0
0.083 ± 0
0.066 ± 0
0.035 ±0
0.085 ± 0
0.024 ± 0
0.010 ±0
0.060 ± 0
0.143±0
0.001 ±0
0.002 ± 0
0.048 ± 0
0.001 ±0
0.013 ±0
0.007 ± 0
.036
.063
.022
.016
.008
.004
.006
.012
.009
.010
.004
.004
.004
.003
.002
.136
.006
Medium
41.30 ± 1.68
26.02 ±1.31
15.28 ±0.99
0.057 ±0.028
0.518 ±0.043
0.023 ±0.111
0.068 ± 0.027
0.078 ±0.016
0.279 ±0.011
0.152 ±0.007
0.442 ± 0.009
0.038 ±0.012
0.019 ±0.009
0.212 ±0.011
0.756 ±0.005
0.005 ± 0.004
0.016 ±0.003
0.251 ±0.004
0.016 ±0.002
0.009 ±0.138
0.085 ±0.005
High
207.44 ± 7.29
95.25 ±
4.28
112.19 ±5.82
0.141 ±
0.651 ±
0.052 ±
0.041 ±
0.057 ±
0.714 ±
0.113 ±
0.822 ±
0.081 ±
0.03 1±
0.210 ±
1.047±
0.011 ±
0.021 ±
0.265 ±
0.079 ±
0.011 ±
0.255 ±
0.031
0.052
0.092
0.033
0.014
0.012
0.007
0.022
0.020
0.035
0.030
0.010
0.005
0.005
0.023
0.003
0.299
0.008
Smoker
456.38
350.24
± 16.80
± 15.27
106. 14 ±5.42
0.964
2.160
0.000
0.000
0.000
0.000
0.000
2.515
0.140
0.033
0.362
2.438
0.008
0.071
0.188
0.047
0.380
0.345
±0.051
±0.137
± 0.000
± 0.000
± 0.000
± 0.000
± 0.000
±0.116
±0.117
±0.386
±0.250
±0.054
±0.017
±0.018
± 0.272
±0.012
±2.175
±0.032
Diesel Vehicles
Light Duty
373.43 ±13.75
132.01 ±5.82
241.42 ±12.11
1.474 ±0.071
2.902 ±0.165
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
2.458 ±0.124
0.228 ±0.114
0.000 ± 0.426
0.150 ±0.304
0.515 ±0.057
0.014 ±0.018
0.024 ±0.021
0.000 ± 0.299
0.003 ±0.014
0.428 ±2.390
0.153 ±0.033
Heavy
Duty
1570.69 ±58.24
253.94 ±16. 12
1316.75 ±55.33
1.833±
3.830±
1.288 ±
1.061 ±
0.321 ±
8.018 ±
0.407 ±
3.717 ±
0.881 ±
0.064 ±
0.716 ±
0.376 ±
0.002 ±
0.001 ±
0.707 ±
0.012 ±
0.493 ±
0.008 ±
1.285
1.286
2.160
0.729
0.543
0.221
0.136
0.111
0.221
0.248
0.107
0.055
0.057
0.062
0.032
0.050
3.108
0.154
Lawson and Smith (1998).
to their abundances in
operation. Emissions
vehicles
. Thus,
(Sagebiel et al.,
gaseous
older,
1997;
different fuels
and lubricants
from gasoline smokers
and to
wear and tear during vehicle
are comparable to those from light-duty diesel
poorly maintained gasoline vehicles
Lawson and Smith, 1998), in
pollutants (e.g., Calvert et al
., 1993).
could be significant sources of PM25
addition to being significant
Durbin et al. (1999) point out that
sources of
although


June 2003
3D-14
DRAFT-DO NOT QUOTE OR CITE

-------
 1      smokers constitute only 1.1 to 1.7% of the light-duty fleet in the South Coast Air Quality
 2      Management District in California, they contribute roughly 20% of the total PM emissions from
 3      the light-duty fleet. In general, motor vehicles that are high emitters of hydrocarbons and carbon
 4      monoxide also will tend to be high emitters of PM (Sagebiel et al.,1997; Cadle et al., 1997).
 5      Particle emission rates, even in newer vehicles, also are correlated with vehicle acceleration; and
 6      emissions occur predominantly during periods of heavy acceleration (Maricq et al., 1999).
 7           Although the data shown in Table 3D-5 indicate that S (mainly in the form of sulfate) is a
 8      minor component of PM25 emissions, S may be the major component of the ultrafine particles
 9      that are emitted by either diesel or internal combustion engines (Gertler et al., 2000). It is not
10      clear what the source of the small amount of Pb seen in the auto exhaust profile is. It is
11      extremely difficult to find suitable tracers for automotive exhaust because Pb has been removed
12      from gasoline.  However, it also should be remembered that restrictions in the use of leaded
13      gasoline have resulted in a dramatic lowering of ambient Pb levels.
14           Several tunnel studies have measured the distribution of organic and elemental carbon in
15      the integrated exhaust of motor vehicle fleets comprising several classes of motor vehicles
16      (Pierson and Brachaczek, 1983; Weingartner et al., 1997a; Fraser et al., 1998a).  The study by
17      Fraser et al. (1998a) found that organic carbon constituted 46% of the carbonaceous PM
18      emissions from the vehicles operating in the Van Nuys tunnel in Southern California in the
19      Summer of 1993. Although diesel vehicles constituted only 2.8% of the vehicles measured by
20      Fraser et al. (1998a), the contribution of the organic carbon to the total paniculate carbon
21      emissions obtained in the Van Nuys tunnels is in reasonable agreement with the dynamometer
22      measurements shown in Table 3D-4b.
23           Very few studies have reported comprehensive analyses of the organic composition of
24      motor vehicle exhaust. The measurements by Rogge et al. (1993c) are the most comprehensive
25      but are not expected to be the best representation of current motor vehicle emissions because
26      these measurements were made in the mid-1980s.  Measurements reported by Fraser et al.
27      (1999) were made in a tunnel study conducted in 1993 and represent integrated diesel and
28      gasoline powered vehicle emissions. In addition, exhaust emissions from two medium-duty
29      diesel vehicles operating over an FTP cycle were analyzed by Schauer et al. (1999). A unique
30      feature of both the measurements by Faser et al. (1999) and Schauer et al. (1999) is that they
31      include the quantification of unresolved complex mixture (UCM), which comprises aliphatic and
32      cyclic hydrocarbons that cannot be resolved by gas chromatography (Schauer et al., 1999).
        June 2003                                3D-15        DRAFT-DO NOT QUOTE OR CITE

-------
1 Schauer et al. (1999) have shown that all of the organic compound mass in their diesel exhaust
2 samples could be extracted and eluted by CG/MS techniques even though not all of the organic
3 compound mass can identified on a single compound basis. Table 3D-6 summarizes the
4 composition of motor
5
6
TABLE 3D-6.


Source
Gasoline and diesel-
powered vehicles
driving through the
Van Nuys Tunnel
(Fraser et al., 1999)a




Medium-duty diesel
vehicles operated over
an FTP Cycle
(Schauer et al., 1999)




vehicle exhaust measured


by Fraser et al. (1999) and Schauer et al. (1999).


SUMMARY OF PARTICLE-PHASE ORGANIC COMPOUNDS
EMITTED FROM

Compound Class
n-Alkanes
Petroleum biomarkers
PAH
Aromatic acids
Aliphatic acids
Substituted aromatic
UCMb
n-Alkanes
Petroleum biomarkers
PAH
Aliphatic acids
Aromatic acids
Saturated cycloalkanes
UCMb
MOTOR VEHICLES
Contribution to Dominant Contributors to
Paniculate Mass (%) Emissions of Compound Class
0.009 C21 through C29
0.078 Hopanes and steranes
0.38 No dominant compound
0.29 Benzenedicarboxylic acids
0.21 Palmitic and stearic acids
0.042 No dominant compound
23.0
0.22 C20 through C28
0.027 Hopanes and steranes
0.54 No dominant compound
0.24 n-Octadecanoic acid
0.014 Methylbenzoic acid
0.037 C21 through C25
22.2
        "Includes emissions of brake wear, tire wear, and resuspension of road dust associated with motor vehicle traffic.
        bUnresolved complex mixture.
1           Several studies have measured the distribution of poly cyclic aromatic hydrocarbons
2      (PAHs) in motor vehicles exhaust from on-road vehicles (Westerholm et al., 1991; Lowenthal
3      et al., 1994; Venkataraman et al., 1994; Westerholm and Egeback, 1994; Reilly et al., 1998;
4      Cadle et al.,  1999, Weingartner et al., 1997b; Marr et al., 1999).  Cadle et al. (1999) found high
       June 2003
3D-16
DRAFT-DO NOT QUOTE OR CITE

-------
 1      molecular weight PAHs (PAHs with molecular weights greater than or equal to 202 g/mole)
 2      constitute 0.1 to 7.0% of the particulate matter emissions from gasoline powered and diesel
 3      powered light duty vehicles. It is important to note, however, that PAHs with molecular weights
 4      of 202 (fluoranthene, acephenanthrylene, and pyrene), 226 (benzo[ghi]fluoranthene and
 5      cyclopenta[cd]pyrene), and 228 (benz[a]anthracene, chrysene, and triphenylene) exist in both the
 6      gas-phase and particle-phase at atmospheric conditions (Fraser et al., 1998b) although those with
 7      molecular weight of 228 are predominantly associated with particles, with only traces in the
 8      gas-phase (Arey et al., 1987). Excluding these semivolatile PAHs, the contribution of
 9      nonvolatile PAHs to the particulate matter emitted from the light-duty vehicles sampled by
10      Cadle et al. (1999) ranges from 0.013 to 0.18%. These measurements are in good agreement
11      with the tunnel  study conducted by Fraser et al. (1999) and the heavy-duty diesel truck and bus
12      exhaust measurements by Lowenthal et al. (1994), except that the nonvolatile PAH emissions
13      from the heavy  duty diesel vehicles tested by Lowenthal et al. (1994) were moderately higher,
14      making up approximately 0.30% of the particulate matter mass emissions.
15
16      Biomass Burning
17           In contrast to the mobile and stationary sources discussed earlier, emissions from biomass
18      burning in wood stoves and forest fires are strongly seasonal and can be highly episodic within
19      their peak emissions seasons.  The burning of fuelwood is confined mainly to the winter months
20      and is acknowledged to be a major source of ambient air particulate matter in the northwestern
21      United States during the heating season. Forest fires occur primarily during the driest seasons of
22      the year in different areas of the country and are especially prevalent during prolonged droughts.
23      PM produced by biomass burning outside the United States (e.g., in Central America during the
24      spring of 1988) also can affect ambient air quality in the United States.
25           An example of the composition of fine particles (PM25) produced by wood stoves is shown
26      in Figure 3D-4. These data were obtained in Denver during the winter of 1987-1988 (Watson
27      and Chow, 1994).  As was the case for motor vehicle emissions, organic and elemental carbon
28      are the major components of particulate emissions from wood burning. It should be remembered
29      that the relative amounts shown for organic carbon and elemental carbon vary with the type of
30      stove, the stage of combustion, and the type and condition of the fuelwood. Fine particles are
31      dominant in smoke studies of wood burning emissions. For instance, the mass median diameter

        June 2003                                3D-17        DRAFT-DO NOT QUOTE OR CITE

-------
                       10
                       10
                       10
                                             Chemical Species
       Figure 3D-4.  Chemical abundances for PM2 5 emissions from wood burning in Denver,
                    CO.  Solid bars represent fractional abundances, and the error bars
                    represent variability in species abundances.  Error bars represent detection
                    limits when there are no solid bars.
       Source:  Watson and Chow (1994).
 1     of wood particles was found to be about 0.17 jim in a study of the emissions from burning
 2     hardwood, softwood, and synthetic logs (Dasch, 1982).
 3          Kleeman et al. (1999) showed that the particles emitted by the combustion of wood in
 4     fireplaces are predominately less than 1.0 jim in diameter, such that the composition of fine PM
 5     (PM25) emitted from fireplace combustion of wood is representative of the total particulate
 6     matter emissions from this source.  Hildemann et al. (1991) and McDonald et al. (2000) reported
 7     that smoke from fireplace and wood stove combustion consists  of 48% to 71% OC and 2.9% to
 8     15% EC. Average elemental and organic carbon contents for these measurements are shown in
 9     Table 3D-7. It should be noted that the two methods used for the measurements shown in
10     Table 3D-7 have been reported to produce different relative amounts of OC and EC for wood
11     smoke samples but show good agreement for total  carbon (OC + EC) measurements (Chow
12     etal., 1993).
       June 2003
3D-18
DRAFT-DO NOT QUOTE OR CITE

-------
             TABLE 3D-7. MASS EMISSIONS, ORGANIC CARBON, AND ELEMENTAL
              CARBON EMISSIONS FROM RESIDENTIAL COMBUSTION OF WOOD

Wood Type
Softwood
Softwood
Hardwood
Hardwood
Hardwood

Combustion
Type
Fireplace
Fireplace
Fireplace
Fireplace
Wood Stove
Average Mass
Emission Rate
(gkg1 of wood
burned)
13.0
5.14
5.28
5.66
3.96

Number
of Tests
2
5
3
5
8
Percent
Organic
Carbon3
48.4
58.5
48.4
63.2
71.2
Percent
Elemental
Carbon3
5.2
15.0
2.9
7.0
9.0

References
Hildemann et al. (1991)
McDonald et al. (2000)
Hildemann etal. (1991)
McDonald et al. (2000)
McDonald et al. (2000)
        aHildemann et al. (1991) used the method described by Birch and Gary (1996) to measure EC and McDonald
         et al. (2000) used the method reported by Chow et al. (1993) to measure OC.
 1          Hawthorne et al. (1988) and Hawthorne et al. (1989) measured gas-phase and particle-
 2     phase derivatives of guaiacol (2-methoxyphenol), syringol (2,6-dimethoxyphenol), phenol, and
 3     catechol (1,2-benzenediol) in the downwind plume of 28 residential wood stoves and fireplaces.
 4     Rogge et al. (1998) reported a broad range of particle-phase organic compounds in the wood
 5     smoke samples collected by Hildemann et al. (1991), which include n-alkanes, n-alkanoic acids,
 6     n-alkenoic acids, dicarboxylic acids, resin acids, phytosterols, polycyclic aromatic hydrocarbons
 7     (PAH), and the compounds reported by Hawthorne et al. (1989).  Supplementing these
 8     measurements, McDonald et al. (2000) reported the combined gas-phase and particle-phase
 9     emissions of PAH and the compounds quantified by Hawthorne et al. (1989).  The measurements
10     by Rogge et al. (1998), which represent a comprehensive data set of the organic compounds
11     present in wood smoke aerosol, are summarized in Table 3D-8. It should be noted, however,
12     that these nearly 200 compounds account for only approximately  15 to 25% of the organic
13     carbon particle mass emitted from the residential combustion of wood. Simoneit et al. (1999)
14     have shown that levoglucosan constitutes a noticeable portion of the organic compound mass not
15     identified by Rogge et al. (1998). In addition, Elias et al. (1999) used high-temperature gas
16     chromatography/mass spectrometry (HTGC-MS) to measure high-molecular-weight organic
17     compounds in smoke from South American leaf and stem litter biomass burning. These
18     compounds cannot be measured by the analytical techniques employed by Rogge et al. (1998)
19     and, therefore, are strong candidates to make up some of the unidentified organic mass in the
       June 2003
3D-19
DRAFT-DO NOT QUOTE OR CITE

-------
            TABLE 3D-8.  SUMMARY OF PARTICLE-PHASE ORGANIC COMPOUNDS
               EMITTED FROM THE COMBUSTION OF WOOD IN FIREPLACES*
Biomass Type
Fireplace
combustion of
softwood







Fireplace
combustion of
hardwood







Contribution to Paniculate Dominant Contributors to Emissions
Compound Class Mass (%) of Compound Class
n-Alkanes
n-Alkanoic acids
n-Alkenoic acids
Dicarboxylic acids
Resin acids
Substituted phenols
Phytosterols
PAH
Oxygenated PAH
n-Alkanes
n-Alkanoic acids
n-Alkenoic acids
Dicarboxylic acids
Resin acids
Substituted phenols
Phytosterols
PAH
Oxygenated PAH
0.039
0.45
0.12
0.36
1.28
3.30
0.37
0.092
0.019
0.044
1.33
0.049
0.42
0.11
8.23
0.21
0.13
0.020
C21 through C31
Cl6, C18, C20, C21, C22, C24
Oleic and linoleic acid
Malonic acid
Abietic, dehydroabietic, isopimaric,
pimaric, and sandaracopimaric acids
Benzenediols and guaiacols
p-Sitosterol
Fluoranthene and pyrene
IH-phenalen- 1 -one
C21 through C29
c c c c
M6> ^22; ^24; ^26
Oleic and linoleic acid
Succinic acid
Dehydroabietic acid
Benzediols, guaiacols, and syringols
p-sitosterol
No dominant compounds
IH-phenalen- 1 -one
       *Note: Measurements were made using a dilution sampler and no semivolatile organic compound sorbent.
       Source: Rogge et al. (1998).


1     wood smoke samples analyzed by Rogge et al. (1998). These compounds, which include
2     triterpenyl fatty acid esters, wax esters, triglycerides, and high-molecular-weight n-alkan-
3     2-ones,are expected to be present in North American biomass smoke originating from
4     agricultural burning, forest fires, grassland fires, and wood stove/fireplace smoke.
      June 2003
3D-20
DRAFT-DO NOT QUOTE OR CITE

-------
 1           Measurements of aerosol composition, size distributions, and aerosol emissions factors
 2      have been made in biomass burning plumes, either on towers (Susott et al., 1991) or aloft on
 3      fixed-wing aircraft (e.g., Radke et al., 1991) or on helicopters (e.g., Cofer et al., 1988).  As was
 4      found for wood stove emissions, the composition of biomass burning emissions is strongly
 5      dependent on the stage of combustion (i.e., flaming, smoldering, or mixed), and the type of
 6      vegetation (e.g., forest, grassland, scrub). Over 90% of the dry mass in particulate  biomass
 7      burning emissions is composed of organic carbon (Mazurek et al., 1991).  Ratios of organic
 8      carbon to elemental carbon are highly variable, ranging from 10:1 to 95:1, with the highest ratio
 9      found for smoldering conditions and the lowest for flaming conditions. Emissions  factors for
10      total parti culate emissions increase by factors of two to four in going from flaming to smoldering
11      stages in the individual fires studied by Susott et al. (1991).
12           Particles in biomass burning plumes from a number of different fires were found to have
13      three distinguishable size modes: (1) a nucleation mode, (2) an accumulation mode, and
14      (3) a coarse mode (Radke et al., 1991). Based on an average of 81 samples, approximately 70%
15      of the mass was found in particles <3.5 jim in aerodynamic diameter. The fine particle
16      composition was found to be dominated by tarlike, condensed hydrocarbons; and the particles
17      were usually spherical in shape. Additional information for the size distribution of particles
18      produced by vegetation burning is shown in Figure 3D-2.
19           An example of ambient data for the composition of PM2 5 collected at a tropical site that
20      was heavily affected by biomass burning is shown in Table 3D-9.  The samples were collected
21      during November of 1997 on the campus of Sriwijaya University, which is located in a rural
22      setting on the island of Sumatra in Indonesia (Pinto et al., 1998). The site was subjected
23      routinely to levels of PM25 well in excess of the U.S. NAAQS as a result of the Indonesian
24      biomass fires from the summer of 1997 through the spring of 1998.  As can be seen from a
25      comparison of the data shown in Table 3D-9 with those shown in Figure 3D-4, there are a
26      number of similarities and differences (especially with regard to the heavy metal content) in the
27      abundances of many species.  The abundances of some crustal elements (e.g., Si, Fe) are higher
28      in Table 3D-9 than in Figure 3D-4, perhaps reflecting additional contributions of entrained soil
29      dust.
30           Limited emissions  data that include organic compound speciation information have been
31      reported for agricultural  burning (Jenkins et al., 1996), forest fires (Simoneit, 1985), and
32      grassland burning (Standley and Simoneit, 1987).  Jenkins et al. (1996) present PAH emissions
        June 2003                                3D-21        DRAFT-DO NOT QUOTE OR CITE

-------
               TABLE 3D-9. MEAN AEROSOL COMPOSITION AT TROPICAL SITE
                 (SRIWIJAYA UNIVERSITY, SUMATRA, INDONESIA) AFFECTED
                          HEAVILY BY BIOMASS BURNING EMISSIONS3
Component
OC
EC
so4-2
Al
Si
Cl
K
Ca
Ti
V
Abundance (%)
76
1.2
11
BDb
9.3 x 1C'2
4.4
0.7
4.5 x 1C'2
4.2 x 1Q-3
BDb
Component
Cr
Mn
Fe
Ni
Cu
Zn
As
Se
Br
Pb
Abundance (%)
BDb
BDb
3.9 x 1C'2
<3.8 x 1Q-5
4.8 x 1C'4
3.1 x 1Q-3
6.4 x 1C'4
2.8 x 1C'4
3.6 x 1C'2
3.1 x 1Q-3
        The mean PM2 5 concentration during the sampling period (November 5 through 11, 1997) was 264 ug/m3.
        bBeneath detection limit.
        Source: Pinto etal. (1998).
 1     factors for the combustion of cereals (barley, corn, rice, and wheat), along with PAH emissions
 2     factors for wood burning. Profiles of organic compounds in emissions from meat cooking
 3     (Rogge et al., 1991) and cigarette smoke (Rogge et al., 1994) also have been obtained.
 4
 5     Natural Sources
 6          Although sea-salt aerosol production is confined to salt water bodies, it is included here
 7     because many marine aerosols can exert a strong influence on the composition of the ambient
 8     aerosol in coastal areas. In some respects, the production of sea-salt aerosols is like that of
 9     windblown dust, in that both are produced by wind agitation of the surface.  The difference
10     between the two categories arises because sea-salt particles are produced from the bursting of air
11     bubbles rising to the sea surface.  Air bubbles are formed by the entrainment of air into the water
12     by breaking waves. The surface energy of a collapsing bubble is converted to kinetic energy in
13     the form of a jet of water that can eject drops above the sea surface. The mean diameter of the
       June 2003
3D-22
DRAFT-DO NOT QUOTE OR CITE

-------
 1     jet drops is about 15% of the bubble diameter (Wu, 1979). Bubbles in breaking waves range in
 2      size from a few |im to several mm in diameter.  Field measurements by Johnson and Cooke
 3      (1979) of bubble size spectra show maxima in diameters at around  100 jam, with the bubble size
 4      distribution varying as (d/d0)"5 with d0 = 100 |im.
 5           Because sea-salt particles receive water from the surface layer, which is enriched in
 6      organic compounds, aerosol drops are composed of this organic material in addition to sea salt
 7      (about 3.5% by weight in seawater). Na+ (30.7%), Cl' (55.0%),  SO;2 (7.7%), Mg2+ (3.6%), Ca2+
 8      (1.2%), K+ (1.1%), HCCV (0.4%), and Br' (0.2%) are the major ionic species by mass in
 9      seawater (Wilson, 1975).  The composition of the marine aerosol also reflects the occurrence of
10      displacement reactions that enrich sea-salt particles in SO4"2 and NO3" while depleting them of
11      Cr and Br'.
12           Sea salt is concentrated in the coarse  size mode with a mass median diameter of about
13      7 |im for samples collected in Florida, the Canary Islands, and Barbados (Savoie and Prospero,
14      1982). The size distribution of sulfate is distinctly bimodal.  Sulfate in the coarse mode is
15      derived from sea water, but sulfate in the submicron aerosol arises from the oxidation of
16      dimethyl sulfide (CH3SCH3) or DMS.  DMS is  produced during the decomposition of marine
17      micro-organisms. DMS is oxidized to methane sulfonic acid (MSA), a large fraction of which
18      is oxidized to sulfate (e.g., Hertel et al., 1994).
19           Apart from sea spray, other natural sources of particles include the suspension of organic
20      debris and volcanism. Profiles of organic compounds in vegetative detritus have been obtained
21      by Rogge et al. (1993d).  Particles are released  from plants in the form of seeds, pollen, spores,
22      leaf waxes, and resins, ranging in size from 1 to 250 jim (Warneck, 1988). Fungal spores and
23      animal debris, such as insect fragments, also are to be found in ambient aerosol samples in this
24      size range. Although material  from all the foregoing categories  may exist as individual particles,
25      bacteria usually are found attached to other dust particles (Warneck, 1988).  Smaller bioaerosol
26      particles include viruses, individual bacteria, protozoa, and algae (Matthias-Maser and Jaenicke,
27      1994). In addition to natural sources, other sources of bioaerosol include industry (e.g., textile
28      mills), agriculture, and municipal waste disposal (Spendlove, 1974).  The size distribution of
29      bioaerosols has not been characterized as well as it has for other categories of airborne particles.
30           Trace metals are emitted to the atmosphere from a variety  of sources such as sea spray,
31      wind-blown dust, volcanoes, wildfires and biotic sources (Nriagu, 1989). Biologically mediated
32      volatilization processes (e.g., biomethylation) are estimated to account for 30 to 50% of the
        June 2003                                3D-23       DRAFT-DO NOT QUOTE OR CITE

-------
1     worldwide total Hg, As, and Se emitted annually; whereas other metals are derived principally
2     from pollens, spores, waxes, plant fragments, fungi, and algae.  It is not clear, however, how
3     much of the biomethylated species are remobilized from anthropogenic inputs. Median ratios of
4     the natural contribution to globally averaged total sources for trace metals are estimated to be
5     0.39 (As), 0.15 (Cd), 0.59 (Cr), 0.44 (Cu), 0.41 (Hg), 0.35 (Ni), 0.04 (Pb), 0.41 (Sb), 0.58 (Se),
6     0.25 (V), and 0.34 (Zn), suggesting a significant natural source for many trace elements.
7     It should be noted, however, that these estimates are based on emissions estimates that have
8     uncertainty ranges of an order of magnitude.
9
      June 2003                                3D-24       DRAFT-DO NOT QUOTE OR CITE

-------
  1       REFERENCES
  2
  3       Arey, I; Zielinska, B.; Atkinson, R.; Winer, A. M. (1987) Polycyclic aromatic hydrocarbon and nitroarene
  4             concentrations in ambient air during a wintertime high-NO^ episode in the Los Angeles basin.
  5             Atmos. Environ. 21: 1437-1444.
  6       Bagley, S. T.; Gratz, L. D.; Johnson, J. H.; McDonald J. F. (1998) Effects of an oxidation catalytic converter and a
  7             biodiesel fuel on the chemical, mutagenic, and particle size characteristics of emissions from a diesel engine.
  8             Environ. Sci. Technol. 32: 1183-1191.
  9       Birch, M. E.; Gary, R. A. (1996) Elemental carbon-based method for occupational monitoring of paniculate diesel
10             exhaust: methodology and exposure issues. Analyst 121: 1183-1190.
11       Cadle, S. H.; Mulawa, P. H.; Ball, J.; Donase, C; Weibel, A.; Sagebiel, J. C; Knapp, K. T.; Snow, R.  (1997)
12             Paniculate emission rates from in-use high-emitting vehicles recruited in Orange county, California. Environ.
13             Sci. Technol. 31:3405-3412.
14       Cadle, S. H.; Mulawa, P.; Hunsanger, E. C.; Nelson, K.; Ragazzi, R. A.; Barrett, R.; Gallagher, G. L.;  Lawson,
15             D. R.; Knapp, K.  T.; Snow, R. (1999) Light-duty motor vehicle exhaust paniculate matter measurement in the
16             Denver, Colorado, area. J. Air Waste Manage. Assoc. 49: PM-164-174.
17       Calvert, J.  G.; Heywood, J. B.; Sawyer, R. F.; Seinfeld, J. H. (1993) Achieving acceptable air quality:  some
18             reflections on controlling vehicle emissions. Science (Washington, DC) 261: 37-45.
19       Chow, J. C.; Watson, J. G.; Pritchett, L. C.; Pierson, W. R.; Frazier, C. A.; Purcell, R. G. (1993) The DRI
20             thermal/optical reflectance carbon analysis system: description, evaluation and applications in U.S. air quality
21             studies. In: Lodge, J. P., Jr. Fourth international conference on carbonaceous particles in the atmosphere;
22             April 1991; Vienna, Austria. Atmos. Environ. Part A 27: 1185-1201.
23       Chow, J. C.; Watson, J. G.; Houck, J. E.; Pritchett, L. C.; Rogers, C. F.; Frazier,  C. A.; Egami, R. T.; Ball, B. M.
24             (1994) A laboratory resuspension chamber to measure fugitive dust size distributions and chemical
25             compositions. Atmos. Environ. 28: 3463-3481.
26       Cofer, W. R., Ill; Levine, J. S.; Sebacher, D. I.; Winstead, E. L.; Riggin, P. J.; Brass, J. A.; Ambrosia, V. G. (1988)
27             Paniculate emissions from a mid-latitude prescribed chaparral fire. J. Geophys. Res. [Atmos.] 93: 5207-5212.
28       Dasch, J. M. (1982) Paniculate and gaseous emissions from wood-burning fireplaces. Environ. Sci.  Technol.
29             16:639-645.
30       Durbin, T. D.; Smith, M. R.; Norbeck, J. M.; Truex, T. J. (1999) Population density, paniculate emission
31             characterization, and impact on the paniculate inventory of smoking vehicles in the South Coast Air Quality
32             Management District. J. Air Waste Manage. Asssoc.  49: 28-38.
33       Elias, V. O.; Simoneit, B. R. T.; Pereira, A. S.; Cabral, J. A.; Cardoso, J. N. (1999) Detection of high molecular
34             weight organic tracers in  vegetation smoke samples by high-temperature gas chromatography-mass
35             spectrometry. Environ. Sci. Technol. 33: 2369-2376.
36       Fraser, M. P.; Cass, G. R.; Simoneit, B. R. T. (1998a) Gas-phase and particle-phase organic compounds emitted
37             from motor vehicle traffic in a Los Angeles roadway tunnel. Environ. Sci. Technol. 32: 2051-2060.
38       Fraser, M. P.; Cass, G. R.; Simoneit, B. R. T.; Rasmussen, R. A. (1998b) Air quality model evaluation data for
39             organics. 5. C6-C22 nonpolar and semipolar aromatic compounds. Environ. Sci. Technol. 32: 1760-1770.
40       Fraser, M. P.; Cass, G. R.; Simoneit, B. R. T. (1999) Paniculate organic compounds emitted from motor vehicle
41             exhaust and in the urban atmosphere. Atmos. Environ. 33: 2715-2724.
42       Fujita, E.; Watson, J. G.; Chow, J.  C.; Robinson, N. F.; Richards, L. W.; Kumar, N. (1998) Northern front range air
43             quality study, volume C:  source apportionment and simulation methods and evaluation. Reno, NV: Desert
44             Research Institute; prepared for Colorado State University, Cooperative Institute for Research in the
45             Atmosphere.
46       Gertler, A. W.; Abu-Allaban, M.; Coulombe, W.; Gilles, J. A.; Pierson, W. R.; Rogers, C. F.; Sagebiel, J. C.;
47             Tarnay, L. (2000) Preliminary results of a tunnel study to characterize mobile source paniculate emissions.
48             Presented at: PM2000: paniculate matter and health—the scientific basis for regulatory decision-making,
49             specialty conference & exhibition; January;  Charleston, SC. Pittsburgh, PA: Air & Waste Management
50             Association.
51       Gillies, J. A.; Gertler, A. W. (2000) Comparison and evaluation of chemically speciated mobile source PM2 5
52             paniculate matter profiles. J. Air Waste Manage. Assoc. 50: 1459-1480.
53       Hawthorne, S. B.; Miller, D. J.; Barkley, R. M.;  Krieger, M. S. (1988) Identification of methoxylated phenols as
54             candidate tracers for atmospheric wood smoke pollution. Environ. Sci. Technol. 22: 1191-1196.
         June 2003                                      3D-25         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Hawthorne, S. B.; Krieger, M. S.; Miller, D. J.; Mathiason, M. B. (1989) Collection and quantitation of
  2             methoxylated phenol tracers for atmospheric pollution from residential wood stoves. Environ. Sci. Technol.
  3             23:470-475.
  4       Hertel, O.; Christensen, J.; Hov, 0. (1994) Modelling of the end products of the chemical decomposition of DMS in
  5             the marine boundary layer. Atmos. Environ. 28: 2431-2449.
  6       Hildemann, L. M.; Markowski, G. R.; Jones, M. C.; Cass, G. R. (1991) Submicrometer aerosol mass distributions of
  7             emissions from boilers, fireplaces, automobiles, diesel trucks, and meat-cooking operations. Aerosol Sci.
  8             Technol. 14: 138-152.
  9       Hopke, P. K.; Lamb, R. E.; Natusch, D. F. S. (1980) Multielemental characterization of urban roadway dust.
10             Environ. Sci. Technol. 14: 164-172.
11       Houck, J. E.; Chow, J.  C.; Watson, J. G.; Simons, C. A.; Pritchett, L. C.; Goulet, J. M.; Frazier, C. A. (1989)
12             Determination of particle size distribution and chemical composition of paniculate matter from selected
13             sources in California: volume I and executive summary (final report). Sacramento, CA: California Air
14             Resources Board.
15       Houck, J. E.; Goulet, J. M.; Chow, J. C.; Watson, J. G.; Pritchett, L. C. (1990) Chemical characterization of emission
16             sources contributing to light extinction. In: Mathai, C. V., ed. Visibility and fine particles: an A&WMA/EPA
17             international specialty conference; October 1989; Estes Park, CO. Pittsburgh, PA: Air & Waste Management
18             Association; pp. 437-446. (A&WMA transactions series no.  TR-17).
19       Huffman, G. P.; Huggins, F. E.; Shah, N.; Huggins, R.; Linak, W. P.; Miller, C. A.; Pugmire, R. J.; Meuzelaar, H. L.
20             C.; Seehra, M. S.; Manivannan, A. (2000) Characterization of fine paniculate matter produced by combustion
21             of residual fuel oil. J. Air Waste Manage. Assoc. 50: 1106-1114.
22       Jenkins, B. M.; Jones, A.  D.; Turn, S. Q.; Williams, R. B. (1996) Particle concentrations, gas-particle partitioning,
23             and species intercorrelations for polycyclic aromatic hydrocarbons (PAH) emitted during biomass burning.
24             Atmos. Environ. 30: 3825-3835.
25       Johnson, B. D.; Cooke, R. C. (1979) Bubble populations  and spectra in coastal waters: a photographic approach.
26             J. Geophys. Res. C: Oceans Atmos. 84: 3761-3766.
27       Kerminen, V. M.; Makela, T. E.; Ojanen, C. H.; Hillamo, R.  E.; Vilhunen, J. K.; Rantanen, L.; Havers, N.;
28             Von Bohlen, A.; Klockow, D. (1997) Characterization of the paniculate phase in the exhaust from a diesel
29             car. Environ. Sci. Technol. 31: 1883-1889.
30       Kittelson, D. B. (1998) Engines and nanoparticles: a review.  J. Aerosol Sci. 29: 575-588.
31       Kleeman, M. J.; Schauer, J. J.; Cass, G. R. (1999) Size and composition distribution of fine paniculate matter
32             emitted from wood burning, meat charbroiling, and cigarettes. Environ. Sci. Technol. 33: 3516-3523.
33       Kleeman, M. J.; Schauer, J. J.; Cass, G. R. (2000) Size and composition distribution of fine paniculate matter
34             emitted from mo tor vehicles. Environ. Sci. Technol. 34: 1132-1142.
3 5       Lawson, D. R.; Smith, R. E. (1998) The northern front range air quality  study: a report to the Governor and General
36             Assembly. Fort Collins, CO: Colorado State University; December.
37       Linak, W. P.; Miller, C. A.; Wendt, J. O. L. (2000) Comparison of particle size distributions and elemental
3 8             partitioning from the combustion of pulverized coal and residual fuel oil. J. Air Waste Manage. Assoc.
39             50: 1532-1544.
40       Lowenthal, D. H.; Zielinska, B.; Chow, J. C.; Watson, J.  G.;  Gautam, M.; Ferguson, D. H.; Neuroth, G.  R.; Stevens,
41             K. D. (1994) Characterization of heavy-duty diesel vehicle emissions. Atmos. Environ. 28: 731-743.
42       Maricq, M. M.; Podsiadlik, D. H.; Chase, R. E. (1999) Gasoline vehicle particle size distributions: comparison of
43             steady state, FTP, and US06 measurements. Environ. Sci. Technol. 33: 2007-2015.
44       Marr, L. C.; Kirchstetter,  T. W.; Harley, R.  A.; Miguel, A. H.; Hering, S. V; Hammond, S. K. (1999)
45             Characterization of polycyclic aromatic hydrocarbons in motor vehicle fuels and exhaust emissions.
46             Environ. Sci. Technol. 33: 3091-3099.
47       Mason, B. (1966) Principles of geochemistry. 3rd ed. New York, NY: John Wiley & Sons, Inc.
48       Matthias-Maser, S.; Jaenicke, R. (1994) Examination of atmospheric bioaerosol particles with radii > 0.2 um.
49             J. Aerosol Sci. 25: 1605-1613.
50       Mazurek, M. A.; Cofer, W. R., Ill; Levine, J. S. (1991) Carbonaceous aerosols from prescribed burning of a boreal
51             forest ecosystem. In: Levine, J. S., ed. Global biomass burning: atmospheric, climatic, and biospheric
52             implications. Cambridge, MA: MIT Press; pp. 258-263.
53       McDonald, J. D.; Zielinska, B.; Fujita, E. M.; Sagebiel, J. C.; Chow, J. C.; Watson, J. G. (2000) Fine particle and
54             gaseous emission rates from residential wood combustion. Environ. Sci. Technol. 34: 2080-2091.
55       Miller, C. A.; Linak, W. P.; King, C.; Wendt, J. O. L. (1998) Fine particle emissions from heavy fuel oil combustion
56             in a firetube package boiler. Combust. Sci. Technol. 134: 477-502.
         June 2003                                      3D-26        DRAFT-DO NOT QUOTE OR CITE

-------
  1       Norbeck, J. M; Durbin, T. D.; Traex, T. J. (1998) Measurement of primary paniculate matter emissions from
  2             light-duty motor vehicles. Riverside, CA: University of California, College of Engineering, Center for
  3             Environmental Research and Technology; prepared for Coordinating Research Council, Inc. and South Coast
  4             Air Quality Management District.
  5       Nriagu, J. O. (1989) A global assessment of natural sources of atmospheric trace metals. Nature (London)
  6             338:47-49.
  7       Nriagu, J. O.; Pacyna, J. M. (1988) Quantitative assessment of worldwide contamination of air, water and soils by
  8             trace metals. Nature (London) 333: 134-139.
  9       Olmez, I.; Sheffield, A. E.; Gordon, G. E.; Houck, J. E.; Pritchett, L. C.; Cooper, J. A.; Dzubay, T. G.; Bennett, R. L.
10             (1988) Compositions of particles from selected sources in Philadelphia for receptor modeling applications.
11             JAPCA 38: 1392-1402.
12       Oros, D. R.; Simoneit, B. R. T. (2000) Identification and emission rates of molecular tracers in coal smoke
13             paniculate matter. Fuel 79: 515-536.
14       Pierson, W. R.; Brachaczek, W. W. (1983) Paniculate matter associated with vehicles on the road. II. Aerosol Sci.
15             Technol. 2: 1-40.
16       Pinto, J. P.; Grant, L. D.; Hartlage, T. A. (1998) Report on U.S. EPA air monitoring of haze from S.E. Asiabiomass
17             fires. Research Triangle Park, NC: U.S. Environmental Protection Agency, National Center for
18             Environmental Assessment; report no. EPA/600/R-98/071.
19       Radke, L. F.; Hegg, D. A.; Hobbs, P. V.; Nance, J. D.; Lyons, J. H.; Laursen, K. K.; Weiss, R. E.; Riggan, P. J.;
20             Ward, D. E. (1991) Paniculate and trace gas emissions from large biomass fires in North America. In: Levine,
21             J. S., ed. Global biomass burning: atmospheric, climatic, and biospheric implications. Cambridge, MA: MIT
22             Press; pp. 209-224.
23       Reilly, P. T. A.; Gieray, R. A.; Whitten, W. B.; Ramsey, J. M. (1998) Real-time characterization of the organic
24             composition and size of individual diesel engine smoke particles. Environ. Sci.  Technol. 32: 2672-2679.
25       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1991) Sources of fine organic
26             aerosol. 1. Charbroilers and meat cooking operations. Environ. Sci.  Technol. 25: 1112-1125.
27       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993a) Sources of fine organic
28             aerosol. 3. Road dust, tire debris, and organometallic brake lining dust: roads as sources and sinks.
29             Environ. Sci. Technol. 27: 1892-1904.
30       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993b) Sources of fine organic
31             aerosol. 5. Natural gas home appliances. Environ. Sci. Technol. 27:  2736-2744.
32       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993c) Sources of fine organic
33             aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ. Sci.
34             Technol. 27:636-651.
35       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993d) Sources of fine organic
3 6             aerosol. 4. Paniculate abrasion products from leaf surfaces of urban plants. Environ. Sci. Technol.
37             27:2700-2711.
38       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1994) Sources of fine organic
39             aerosol. 6. Cigarette smoke in the urban atmosphere.  Environ. Sci. Technol. 28: 1375-1388.
40       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1997a) Sources of fine organic
41             aerosol. 8. Boilers burning no. 2 distillate fuel oil. Environ.  Sci. Technol. 31: 2731-2737.
42       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1997b) Sources of fine organic
43             aerosol. 7. Hot asphalt roofing tar pot fumes. Environ. Sci. Technol. 31: 2726-2730.
44       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1998) Sources of fine organic
45             aerosol. 9. Pine, oak, and synthetic log combustion in residential fireplaces. Environ. Sci. Technol. 32:  13-22.
46       Sagebiel, J. C.; Zielinska, B.; Walsh, P. A.; Chow, J. C.; Cadle, S.  H.; Mulawa, P. A.;  Knapp, K. T.; Zweidinger,
47             R. B.; Snow, R. (1997) PM-10 exhaust samples collected during IM-240 dynamometer tests of in-service
48             vehicles in Nevada. Environ. Sci. Technol. 31: 75-83.
49       Savoie, D. L.; Prospero, J. M. (1982) Particle size distribution of nitrate and sulfate in the marine  atmosphere.
50             Geophys. Res. Lett. 9: 1207-1210.
51       Sawyer, R. F.; Johnson, J. H. (1995) Diesel emissions and control technology. Diesel exhaust: a critical analysis of
52             emissions, exposure, and health effects. Cambridge, MA: Health Effects Institute.
53       Schauer, J. J.; Cass, G. R. (2000) Source apportionment of wintertime gas-phase and particle-phase air pollutants
54             using organic compounds as tracers. Environ. Sci. Technol. 34: 1821-1832.
55       Schauer, J. J.; Kleeman, M. J.; Cass, G. R.;  Simoneit, B. R.  T. (1999) Measurement of emissions from air pollution
56             sources. 2. Q through C30 organic compounds from medium duty diesel trucks. Environ. Sci. Technol.
57             33: 1578-1587.


         June 2003                                      3D-27         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Simoneit, B. R. T. (1985) Application of molecular marker analysis to vehicular exhaust for source reconciliations.
  2             Int. J. Environ. Anal. Chem. 22: 203-233.
  3       Simoneit, B. R. T.; Schauer, J. J.; Nolte, C. G.; Oros, D. R.; Elias, V. O.; Fraser, M. P.; Rogge, W. F.; Cass, G. R.
  4             (1999) Levoglucosan, a tracer for cellulose in biomass burning and atmospheric particles. Atmos. Environ.
  5             33: 173-182.
  6       Spendlove, J. C. (1974) Industrial, agricultural, and municipal microbial aerosol problems. In: Current problems in
  7             aerobiology: proceedings of the thirtieth general meeting of the Society for Industrial Microbiology; August
  8             1973; Evanston, IL. Dev. Ind. Microbiol. 15: 20-27.
  9       Standley, L.  J.; Simoneit, B. R. T. (1987) Characterization of extractable plant wax, resin, and thermally matured
10             components in smoke particles from prescribed burns. Environ. Sci. Technol. 21: 163-169.
11       Susott, R. A.; Ward,  D. E.; Babbitt, R. E.; Latham, D. J. (1991) The  measurement of trace emissions and combustion
12             characteristics for a mass fire. In: Levine, J. S., ed. Global biomass burning: atmospheric, climatic, and
13             biospheric implications. Cambridge, MA: MIT Press; pp, 245-257.
14       Turekian, K. K. (1971) Elements, geochemical distribution of. In: Encyclopedia of science and technology, v. 4.
15             New York, NY: McGraw-Hill Book Company; pp. 627-630.
16       U.S. Environmental Protection Agency. (1996) Air quality criteria for paniculate matter. Research Triangle Park,
17             NC: National  Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
18       U.S. Environmental Protection Agency. (1999) SPECIATE. Washington, DC: Technology Transfer Network-
19             Clearinghouse for Inventories & Emissions Factors. Available:
20             http://www.epa.gov/ttn/chief/software/speciate/index.html (2  June 2003).
21       U.S. Environmental Protection Agency. (2002) Health assessment document for diesel engine exhaust. Washington,
22             DC: Office of Research and Development; report no. EPA/600/8-90/057.
23       Venkataraman, C.; Lyons, J. M.; Friedlander, S. K. (1994) Size distributions of polycyclic aromatic hydrocarbons
24             and elemental carbon. 1. Sampling, measurement methods, and source characterization. Environ. Sci.
25             Technol. 28: 555-562.
26       Vinogradov, A. P. (1959) The geochemistry of rare and dispersed chemical elements in soils. 2nd ed. New York,
27             NY: Consultants Bureau, Inc.
28       Warneck, P.  (1988) Chemistry of the natural atmosphere. New York, NY: Academic Press, Inc.
29       Watson, J. G.; Chow, J. C. (1994) Clear sky visibility as a challenge for society. Annu. Rev. Energy Environ.
30             19:241-266.
31       Watson, J. G.; Fujita, E. M.; Chow, J. C.; Zielinska, B.; Richards, L. W.; Neff, W.; Dietrich, D. (1998) Northern fron
32             range air quality study final report.  Reno, NV: Desert Research Institute; prepared for Colorado State
3 3             University, Cooperative Institute for Research in the Atmosphere.
34       Weingartner, E.; Burtscher, H.; Baltensperger, U. (1997a) Hygroscopic properties of carbon and diesel soot
3 5             particles. Atmos. Environ. 31: 2311-2327.
36       Weingartner, E.; Keller, C.; Stahel, W. A.; Burtscher, H.; Baltensperger, U. (1997b) Aerosol emission in a road
37             tunnel. Atmos. Environ. 31: 451-462.
38       Westerholm, R.; Egeback, K.-E. (1994) Exhaust emissions from light- and heavy-duty vehicles: chemical
3 9             composition, impact of exhaust after treatment, and fuel parameters. In: Symposium of risk assessment of
40             urban air: emissions, risk identification, and risk quantitation; May-June  1992; Stockholm,  Sweden. Environ.
41             HealthPerspect. 102(suppl. 4) 13-23.
42       Westerholm, R. N.; Almen, J.; Li, H.; Rannug, J. U.; Egeback, K.-E.; Gragg, K. (1991) Chemical  and biological
43             characterization of paniculate-, semivolatile-, and gas-phase-associated compounds in diluted heavy-duty
44             diesel exhausts: a comparison of three different semivolatile-phase samplers. Environ. Sci.  Technol.
AC             1C>'5'51'5'5O
45             25: 332-338.
46       Wilson, T. R. S. (1975) Salinity and the major elements of sea water. In: Riley,  J. P.; Skirrow, G., eds. Chemical
47             oceanography: v. 1. 2nd ed. London, United Kingdom: Academic Press Inc.; pp. 365-413.
48       Wu, J. (1979) Spray  in the atmospheric surface layer: review and analysis of laboratory and oceanic results.
49             J. Geophys. Res. C: Oceans Atmos. 84: 1693-1704.
50
         June 2003                                      3D-28        DRAFT-DO NOT QUOTE OR CITE

-------
 i                                     APPENDIX 3E
 2
 3       Variability Observed in PM25 and PM10_25 Concentrations
 4                                  at IMPROVE Sites
 5
 6
 7          In this appendix, the temporal variability of PM25, PM10_25 and PM10 concentrations is
 8     characterized on various time scales based on daily values in the PM data set obtained at 18 sites
 9     in the IMPROVE network.  The purpose of these analyses is to provide concentrations of PM2 5
10     and PM10_2 5 that could be used for characterizing human and ecosystem exposures to PM under
11     the least polluted conditions that are found in the United States and to provide limits on policy
12     relevant background PM concentrations (cf, Section 3.3.3). This is accomplished by analyzing
13     PM2 5 and PM10_2 5 concentrations and their variability at relatively remote monitoring sites that
14     are not obviously influenced by local pollution sources in the United States. The potential for
15     receiving contributions from anthropogenic sources exists at every monitoring site. However,
16     we are interested in characterizing data obtained at the sites with lowest concentrations. Only
17     those sites in which anthropogenic sources do not contribute extensively to the observations are
18     considered. These sites will be referred to as relatively remote monitoring sites (RRMS). Thus,
19     virtually all sites in the eastern United States would not fit into this definition.  Data from these
20     sites are included to permit  comparison with sites in the West. More detailed analyses of the
21     data are available in Lefohn et al. (2004).
22          Data stored in the IMPROVE  database as of December 4, 2002 from 18 IMPROVE
23     network sites were downloaded from the following internet address:
24     (http://vista.cira.colostate.edu/IMPROVE/Data/IMPROVE/IMPLoctable_Data.asp?SortColumn
25     =Site_Code).  The locations of the sites chosen are shown in Figure 3E-1. Data at many sites are
26     available from 1989, data at other sites are available for shorter periods of time. In addition,
27     sites have been relocated within various national parks. The monitoring site in Yellowstone
28     National Park was relocated in 1996; thus, the designations for the two locations are
29     Yellowstone National Parkl and Yellowstone National Park2. The Voyageurs National Park site
30     was also relocated.  The designations for the two locations are similar to those used for
31     Yellowstone National Park.  The 24-hour average data were summarized on seasonal and annual

       June 2003                                3E-1        DRAFT-DO NOT QUOTE OR CITE

-------
                                       IMPROVE SITES
                              SELECTED FOR  THE STUDY
                                                        Boundary Waters
                                                           noe Area
                    I  • Glacier         Voyageu
                    \National Park       National Park
                     H    Yellowstone National Park
                                  Bridger Wilderness
                  Redwood National Park
                     J_assen Vole anic
                                                        Doll* Sods/Ottei Creek
                       ational Pt irk Lon< i Peak Wild 3 mess
                               -Bate Canyon National Park
                                    Gila Wi derness
                                                                    Brigantine
                                                                   lional Wildlife
                                                                     Refuge
                              Not shown:
                          Denali National Park, AK
       Figure 3E-1. Locations of the rural PM IMPROVE sites selected for the study.
 1
 2
 3
 4
 5
 9
10
11
12
13
14
15
16
17
bases.  The data capture requirement for the annual analysis required that there had to be at least
13 valid days of data for a particular constituent for each quarter of a calendar year in order for
the information to be included.  For the seasonal analysis, all 24-hour average data were used.
The percentile information was organized by minimum, 10th percentile concentration (P10),
25th percentile concentration (P25), 50th percentile concentration (P50), 75th percentile
concentration (P75), 90th percentile concentration (P90), 95th percentile concentration (P95), 99th
percentile concentration (P99), and maximum concentration.
     Information about the range of annual mean PM2 5, PM10_2 5 and PM10 concentrations is
summarized in Table 3E-1, and the range of 90th percentile concentrations is summarized in
Table 3E-2.  As can be seen from Table 3E-1, the lowest average annual mean PM25
concentrations among all sites examined were recorded at Denali National Park (AK), with
concentrations there ranging from an annual mean of 1.15 |ig/m3 to 3.14 |ig/m3.  On the other
hand, the highest annual mean PM2 5 concentrations in the 12 western sites studied were
observed in Glacier National Park (MT); values there ranged from 4.92 to 6.36 |ig/m3. The
highest average annual mean PM2 5 concentrations in the continental United States were
observed at Dolly Sods/Otter Creek Wilderness (WV), those ranging from 10.03 |ig/m3 to
14.24 |ig/m3. In general, the lowest average annual mean PM25 concentration increased in going
       June 2003
                                         3E-2
DRAFT-DO NOT QUOTE OR CITE

-------
      TABLE 3E-1. THE RANGE OF ANNUAL MEAN PM2 s, PM10 2 5, AND PM10
   CONCENTRATIONS AT IMPROVE PARTICIPATE MATTER MONITORING
                          SITES (all units in





















1
2
3
4
5
6
1
Site Name
*Denali National Park, AK
* Glacier National Park, MT
*Bridger Wilderness, WY
* Yellowstone National Parkl, WY
*Yellowstone National Park2, WY
*Bryce Canyon National Park, UT
*Canyonlands National Park, UT
*Lone Peak Wilderness, UT
*Three Sisters Wilderness, OR
*Redwood National Park, CA
*Lassen Volcanic National Park, CA
*Gila Wilderness, NM
Boundary Waters Canoe Area, MN
Voyageurs National Parkl, MN
Voyageurs National Park2, MN
Dolly Sods/Otter Creek Wilderness, WV
Acadia National Park, ME
Lye Brook Wilderness, VT
Brigantine National Wildlife Refuge, NJ
* Western monitoring sites
PM25 PM1025 PM10
1.15-3.14 1.09-5.37 2.36-8.19
4.92-6.36 4.31-9.66 9.89-15.25
2.15-3.05 2.36-4.75 4.22-7.39
2.79-4.68 3.98-7.72 6.80-10.55
2.33-3.15 1.91-2.81 4.21-5.71
2.68-3.65 2.48-5.74 5.23-8.66
2.73-4.03 3.25-8.05 6.12-11.79
4.13-5.81 3.25-6.19 7.37-11.60
2.63-4.19 1.92-4.43 4.57-7.74
3.62-5.43 3.33-6.50 7.20-12.19
2.10-5.47 1.86-6.47 4.42-8.68
3.46-4.86 2.55-4.94 5.91-9.17
4.28-5.88 2.24-3.72 7.19-9.35
5.01-6.60 3.98-4.22 10.08-13.25
4.48-7.82 2.71-3.03 6.93-7.14
10.03-14.24 2.29-5.3 13.26-18.92
4.23-8.50 1.84-6.04 7.28-15.03
5.33-8.43 1.58-4.88 6.84-13.16
10.20 - 12.63 7.90 - 14.87 18.49 - 26.63

to the midwestern and then to the eastern United States, and the range of annual average
concentrations tends to be lower at western
observational record, a substantial range in
than at eastern sites. Over the period of the
annual mean concentrations in all three size ranges is
observed, indicating the importance of examining a number of years in attempting to
characterize long term averages. Inspection
PM10_2 5 concentrations was generally larger
of Table 3E-1 shows that the range in annual average
than for PM2 5. However, it should be noted that
PM10_2 5 concentrations are obtained by a difference technique and so is more strongly affected by
errors in the determination of both PM25 and PM10.
June 2003                            3E-3       DRAFT-DO NOT QUOTE OR CITE

-------
    TABLE 3E-2. THE RANGE OF 24-HOUR PM2 5, PM10 2 5, AND PM10 AVERAGE
      CONCENTRATIONS AT THE 90™ PERCENTILE LEVEL AT IMPROVE
        PARTICIPATE MATTER MONITORING SITES (all units in




















1
2
3
4
5
6
7
8
9
Site Name
*Denali National Park, AK
* Glacier National Park, MT
*Bridger Wilderness, WY
* Yellowstone National Parkl, WY
*Yellowstone National Park2, WY
*Bryce Canyon National Park, UT
*Canyonlands National Park, UT
*Lone Peak Wilderness, UT
*Three Sisters Wilderness, OR
*Redwood National Park, CA
*Lassen Volcanic National Park, CA
*Gila Wilderness, NM
Boundary Waters Canoe Area, MN
Voyageurs National Parkl, MN
Dolly Sods/Otter Creek Wilderness, WV
Acadia National Park, ME
Lye Brook Wilderness, VT
Brigantine National Wildlife Refuge, NJ
*Western monitoring sites.
As can be seen from Table 3E-2
PM
2.18-
8.05-
3.94-
4.67-
4.06-
4.40-
4.48-
6.74-
5.82-
6.25-
4.00-
5.58-
8.11-
7.72-
15.57 -
8.80-
11.44-
16.20 -

[2S
7.44
11.00
5.77
6.58
6.04
5.92
6.38
11.07
9.97
9.93
9.05
10.57
12.14
13.57
29.04
17.13
16.46
22.72


1.
8.
4.
8.
4
5.
6.
6.
3
5.
3.
4
4
9.
4
3.
3
13

PM
92-
36-
80-
78-
.22-
03-
30-
55-
.97-
94-
74-
.22-
.77-
04-
.35-
58-
.31-
.60-

10-2.5
11.65
27.60
12.26
16.53
-5.94
11.99
14.68
10.87
-9.63
12.97
14.73
-9.66
-8.84
17.38
-9.20
12.02
-5.85
•32.19

, concentrations of PM in all three size
PM10
4.31-
17.66 -
8.32-
12.34-
7.42-
8.93-
10.89 -
13.42-
8.51-
13.21 -
8.54-
10.25 -
13.64-
17.89 -
21.30-
12.74 -
15.66-
31.37-

fractions at
17.94
36.82
15.14
21.34
10.82
13.60
18.46
21.52
16.47
23.73
15.95
18.84
18.85
26.46
34.63
26.04
22.88
45.28

the
90th percentile level are often several times higher than the mean concentrations at all sites.
In addition, the range of values at the
90th percentile level is
levels. These findings indicate that extreme events
distribution of PM that is observed. Wildfires play
variability at specific sites, especially
in the West.
much greater than that at lower
are important
a major
Notable
for
shaping
the frequency
role in defining the year-to-year
examples
fire in 1988 in Yellowstone National Park and those that occurred
states. Dust storms also play a role in
include the massive forest
in 2000 in
many western
arid climates. The variability in annual average PM10_2 5
concentrations is again related to several factors. Variability in factors that govern the
June 2003
3E-4
DRAFT-DO NOT QUOTE OR CITE

-------
 1      production of soil dust from the surface and the production of primary biological aerosol
 2      particles may be largely responsible.  Wildfires also produce PM in this size range (cf,
 3      Appendix 3D).
 4           The interannual variability in the percentile distributions of PM25 concentrations for 1997
 5      through 2001 is shown in Figures 3E-2(a-d) for four sites. The corresponding interannual
 6      variability in PM10_2 5 concentrations at these sites is shown in Figures 3E-3(a-d). As can be seen,
 7      at concentrations less than the P90 level, year-to-year differences can be less than 1 |ig/m3 at
 8      sites such as Bridger, WY. Differences at the P90 level are much larger, both on an absolute and
 9      on a relative basis at eastern sites.
10           Percentile distributions by season are shown in Figure 3E-4(a-d) for PM25 and
11      Figures 3E-5(a-d) for PM10_2 5.  For most sites, the variability in concentrations on a quarterly
12      basis increases substantially beyond the 90th percentile level, as can be seen from inspection of
13      Figures 3E-4(a-d) for PM2 5 and Figures 3E-5(a-d) for PM10_25. Perhaps the most striking
14      features seen in Figures 3E-4(a-d) and Figures 3E-5(a-d) are the concentration changes
15      associated with the P95 and P99 events, which represent extreme value events. Most of these
16      events occur in the third calendar quarter; however, at some sites high concentrations occur
17      during the second and fourth calendar quarters. If locally derived climatologic seasons were
18      used instead of calendar quarters, a more accurate depiction of the seasonal variability of these
19      events may have been obtained. In most cases, there is consistency in the behavior of PM25 and
20      PM10, which suggests that the episodes of higher concentrations could be associated with sources
21      that produce mainly PM25 such as wildfires and/or anthropogenic combustion sources. Wildfires
22      are limited to hotter and drier times of the year, but anthropogenic sources can contribute to high
23      concentrations during other seasons.  Additional factors which would tend to produce a 3rd
24      quarter maximum include: the enhanced production of secondary particulate matter from
25      anthropogenic and biogenic precursors during summer months; wildfires that are located in the
26      East or elsewhere during summer; and surface dust produced locally  and/or in northern Africa.
27
28
29      REFERENCES
30      Lefohn, A. S.; Pinto, J. P.; Shadwick, D.; Ziman, S. D. (2004) The variation of background particulate matter at
31           'clean sites' in the United States. J. Air Waste Manage. Assoc.: submitted.
        June 2003                                 3E-5        DRAFT-DO NOT QUOTE OR CITE

-------
   30
fT 25
 £
 "
r 20
o
15
I 15
o
c
o
0 10
         Min
                          Bridger Wilderness WY
                 10
       P75     P90
             P95    P99
                   Max
         Min
                       Yellowstone National Park WY
                                     PM2.5
                        25
50
75
90
95
99
                                  Max
Figure 3E-2a,b.  Interannual variability in 24-h average PM2 5 concentrations observed at
               selected IMPROVE sites: (a) Bridger Wilderness, WY; (b) Yellowstone
               National Park, WY; (c) Dolly Sods/Otter Creek Wilderness, WV; and
               (d) Brigantine National Wildlife Refuge, NJ.
June 2003
       3E-6
           DRAFT-DO NOT QUOTE OR CITE

-------
                 Dolly Sods / Otter Creek Wilderness WV
                                    PMa.5
         Mln
                          Max
   45

   40

I  35
O)
f  30
o
1  25
-4-*
|  20
o
0  15

I  10

    5

    0
                  Brigantine National Wildlife Refuge NJ
                                    PM2.5
         Min     P10     P25     P50     P75     P90     P95    P99    Max
Figure 3E-2c,d.  Interannual variability in 24-h average PM2 5 concentrations observed at
               selected IMPROVE sites: (a) Bridger Wilderness, WY; (b) Yellowstone
               National Park, WY; (c) Dolly Sods/Otter Creek Wilderness, WV; and
               (d) Brigantine National Wildlife Refuge, NJ.
June 2003
3E-7
DRAFT-DO NOT QUOTE OR CITE

-------
    80



-------
16

14

12

10

 8

 6

 4

 2

 0
                  Dolly Sods / Otter Creek Wilderness WV
                                   PWho-2.5
TO
 E
 c
 o
 (J
 E
 O
 o
 in
 cji
 o
                 10     '25     '50     '75    '90     '95    " 99
                                                                          -1998
                                                                          -1999
                                                                          - 2000
                                                                          -2001
         Min
                   Brigantine National Wildlife Refuge NJ
                        25
                           50
                                      75
90
95
Pgg    Max
Figure 3E-3c,d. Interannual variability in 24-h average PM10_2 5 concentrations observed at
               selected IMPROVE sites: (a) Bridger Wilderness, WY; (b) Yellowstone
               National Park, WY; (c) Dolly Sods/Otter Creek Wilderness, WV; and
               (d) Brigantine National Wildlife Refuge, NJ.
June 2003
                                   3E-9
                                                 DRAFT-DO NOT QUOTE OR CITE

-------
                                Bridger Wilderness WY
                                   PM2.5 by Season
         Min
                           25
        50
         75
        90
        95
        99
                                                 Max
               Quarter 1
      Quarter 2
                Quarters
                         Quarter 4
                            Yellowstone National Park2 WY
                                   PM2.5  by Season
         Min
                   10
25
50
75
90
95
99
                                                 Max
               Quarter 1
       Quarter 2
                Quarter3
                         Quarter 4
Figure 3E-4a,b.  Seasonal variability in 24-h average PM2 5 concentrations observed at
                selected IMPROVE sites:  (a) Bridger Wilderness, WY; (b) Yellowstone
                National Park, WY; (c) Dolly Sods/Otter Creek Wilderness, WV; and
                (d) Brigantine National Wildlife Refuge, NJ.
June 2003
           3E-10
              DRAFT-DO NOT QUOTE OR CITE

-------
          Min
                        Dolly Sods / Otter Creek Wilderness WV
                                   PM2.5 by Season
                           25
 50
75
90
95
99
                                          Max
               Quarter 1
Quarter 2
        Quarters
                 Quarter 4
                         Brigantlne National Wildlife Refuge NJ
                                   PM2.5  by  Season
          Min
                           25
 50
75
90
95
99
                                          Max
               Quarter 1
Quarter 2
        Quarters
                 Quarter 4
Figure 3E-4c,d. Seasonal variability in 24-h average PM2 5 concentrations observed at
               selected IMPROVE sites: (a) Bridger Wilderness, WY; (b) Yellowstone
               National Park, WY; (c) Dolly Sods/Otter Creek Wilderness, WV and
               (d) Brigantine National Wildlife Refuge, NJ.
June 2003
    3E-11
      DRAFT-DO NOT QUOTE OR CITE

-------
                                Bridger Wilderness WY
                                 PM 10-2.5 by Season
          Min
                            25
  50
 75
90
 95
 99
                                          Max
               Quarter 1
Quarter 2
        Quarters
                 Quarter 4
CL
                            Yellowstone National Park 2 WY
                                 PMio-2.5 by Season
         Min
                           25
  50
75
90
95
99
                                         Max
               Quarter 1
Quarter 2
    A  Quarters
                 Quarter 4
Figure 3E-5a,b. Seasonal variability in 24-h average PM10_2 5 concentrations observed at
               selected IMPROVE sites: (a) Bridger Wilderness, WY; (b) Yellowstone
               National Park, WY; (c) Dolly Sods/Otter Creek Wilderness, WV; and
               (d) Brigantine National Wildlife Refuge, NJ.
June 2003
    3E-12
       DRAFT-DO NOT QUOTE OR CITE

-------
                         Dolly Sods / Otter Creek Wilderness WV
                                  PM 10-2.5 by Season
           Min
                             25
                                  50
        75
        90
        95
        99
                                                                         Max
                Quarter 1
                                Quarter 2
           A   Quarter 3
                         Quarter 4
   160
                         Brigantine National Wildlife Refuge NJ
                                  PM 10-2.5 by Season
 E
 "at
140
 - 120

 1 100
 o
 c
 o
 O
 CL
    80
           Min
                    10
                         25
50
75
90
95
99
                                                                         Max
                Quarter 1
                                Quarter 2
                Quarters
                         Quarter 4
Figure 3E-5c,d. Seasonal variability in 24-h average PM10_2 5 concentrations observed at
               selected IMPROVE sites:  (a) Bridger Wilderness, WY; (b) Yellowstone
               National Park, WY; (c) Dolly Sods/Otter Creek Wilderness, WV; and
               (d) Brigantine National Wildlife Refuge, NJ.
June 2003
                                    3E-13
              DRAFT-DO NOT QUOTE OR CITE

-------
 i           4.  ENVIRONMENTAL EFFECTS  OF AIRBORNE
 2                           PARTICULATE MATTER
 3
 4
 5     4.1  INTRODUCTION
 6          This chapter assesses information providing inputs to U.S. EPA decision making on
 7     secondary National Ambient Air Quality Standards (NAAQS) aimed at protecting against
 8     welfare effects of ambient airborne particulate matter (PM).  Specifically, it assesses the effects
 9     of atmospheric PM on the environment, including: (a) direct and indirect effects on vegetation
10     and natural ecosystem integrity; (b) effects on visibility; and (c) effects on man-made materials,
11     as well as (d) relationships of atmospheric PM to climate change processes. The 1997 EPA
12     revisions to the U.S. PM NAAQS, discussed in Chapter 1 (Introduction), included establishment
13     of PM2 5 secondary standards identical to the primary PM2 5 NAAQS set at that time. The 1997
14     FR notice promulgating these standards noted "The new secondary standards, in conjunction
15     with a regional haze program, will provide appropriate protection against PM-related public
16     welfare effects including soiling, material damage, and visibility impairment."
17
18
19     4.2  EFFECTS OF AMBIENT AIRBORNE PM  ON VEGETATION AND
20          NATURAL ECOSYSTEMS
21     Introduction
22          The effects of airborne particles are manifested via physical and chemical effects exerted at
23     the individual  plant level.  However, plants are key members of ecosystems, structurally
24     complex communities comprised of populations of plants, animals (including humans), insects,
25     and microorganisms that interact with one another and with their non-living (abiotic) chemical
26     and physical environment in which they exist (Odum, 1989;  U.S. Environmental Protection
27     Agency, 1993). All life on Earth is dependent on  chemical energy in the form of carbon
28     compounds to sustain their life processes. Terrestrial vegetation, via the process of
29     photosynthesis, provides approximately half of the carbon that annually cycles between the Earth
30     and the atmosphere (Chapin and Ruess, 2001).
       June 2003                                4-1         DRAFT-DO NOT QUOTE OR CITE

-------
 1           Ecosystems respond to stresses through their constituent organisms. The responses of
 2      plant species and populations to environmental perturbations (such as those caused by
 3      atmospheric PM) depend on their genetic constitution (genotype), their life cycles, and the
 4      microhabitats in which they are growing. Stresses that produce changes in their physical and
 5      chemical environment apply selection pressures on individual organisms (Treshow, 1980). The
 6      changes that occur within populations and plant communities reflect these new and different
 7      pressures.  A common response in a community under stress is the elimination of the more
 8      sensitive populations and an increase in abundance of species that tolerate or are favored by
 9      stress (Woodwell,  1970, Guderian et al., 1985).
10           The present section is organized to discuss:  (1) factors affecting deposition of airborne PM
11      on plants and ecosystems and then (2) the effects of PM deposition on individual plants, plant
12      populations, forest trees and terrestrial ecosystems. As such, the  section is organized to follow,
13      in rough outline, the Framework for Assessing and Reporting on  Ecological Condition
14      recommended in a report by the Ecological Processes  and Effects Committee (EPEC) of EPA's
15      Science Advisory Board (Science Advisory Board, 2002), which states "The purpose of this
16      report is to provide the Agency with a sample framework that may serve as a guide for designing
17      a system to assess, and then report on, ecological condition at local, regional, or national scale.
18      The sample framework is intended as an organizing tool that may help the Agency decide what
19      ecological attributes to measure and how to aggregate those measurements into an
20      understandable picture of ecological integrity." This framework  is not actually a risk assessment
21      per se, but it can be used to "construct a report of ecological condition" that characterizes the
22      ecological integrity of an ecosystem based on "the relationship between common anthropogenic
23      stressors and one or more of the six Essential Ecological Attributes." It nevertheless does
24      provide a useful approach for organizing discussions of stressor effects on ecosystem
25      components at successive levels of complexity.
26
27      4.2.1  Ecological Attributes
28           The EPEC Framework provides a checklist of generic ecological attributes that should be
29      considered when evaluating the integrity of ecological systems (see Table 4-1). The six generic
30      ecological attributes, termed Essential Ecological Attributes (EEA), represent groups of related
31      ecological characteristics (Science Advisory Board, 2002; Harwell et al., 1999) and include:

        June 2003                                  4-2        DRAFT-DO NOT QUOTE OR CITE

-------
                      TABLE 4-1.  ESSENTIAL ECOLOGICAL ATTRIBUTES AND
                                        REPORTING CATEGORIES
         Landscape Condition
           • Extent of Ecological System/Habitat Types
           • Landscape Composition
           • Landscape Pattern and Structure

         Biotic Condition
           • Ecosystems and Communities
            - Community Extent
            - Community Composition
            - Trophic Structure
            - Community Dynamics
            - Physical Structure
           • Species and Populations
            - Population Size
            - Genetic Diversity
            - Population Structure
            - Population Dynamics
            - Habitat Suitability
           • Organism Condition
            - Physiological Status
            - Symptoms of Disease or Trauma
            - Signs of Disease

         Chemical and Physical Characteristics
         (Water, Air, Soil, and Sediment)
           • Nutrient Concentrations
           - Nitrogen
           - Phosphorus
           - Other Nutrients
           • Trace Inorganic and Organic Chemicals
           - Metals
           - Other Trace Elements
           - Organic Compounds
           • Other Chemical Parameters
           -pH
           - Dissolved Oxygen
           - Salinity
           - Organic Matter
           - Other
           • Physical Parameters
Ecological Processes
  • Energy Flow
    - Primary Production
    - Net Ecosystem Production
    - Growth Efficiency
  • Material Flow
    - Organic Carbon Cycling
    - Nitrogen and Phosphorus Cycling
    - Other Nutrient Cycling

Hydrology and Geomorphology
  • Surface and Groundwater flows
    - Pattern of Surface flows
    - Hydrodynamics
    - Pattern of Groundwater flows
    - Salinity Patterns
    - Water Storage
  • Dynamic Structural Characteristics
    - Channel/Shoreline Morphology, Complexity
    - Extent/Distribution of Connected Floodplain
    - Aquatic Physical Habitat Complexity
  • Sediment and Material Transport
    - Sediment Supply/Movement
    - Particle Size Distribution Patterns
    - Other Material Flux

Natural Disturbance Regimes
  • Frequency
  • Intensity
  • Extent
  • Duration
         Source: Science Advisory Board (2002).
1      Chemical and Physical Characteristics; Biotic Conditions; Landscape Conditions; Ecological

2      Processes; Hydrology and Geomorphology; and Natural Disturbance Regimes.  All of the EEAs

3      are interrelated (i.e., changes in one EEA may directly or indirectly affect other EEAs).
       June 2003
          DRAFT-DO NOT QUOTE OR CITE

-------
 1           The first three ecological attributes listed in Table 4-1 are primarily "patterns," whereas the
 2      last three are "processes." Ecological science has used "patterns" and "processes" as terms to
 3      describe features of ecological systems for many years (e.g., Bormann and Likens, 1979).  Of
 4      main concern in this chapter are relationships between a certain class of diverse airborne
 5      stressors from anthropogenic sources, termed particulate matter (PM), and one or more of the
 6      EEAs.  Changes in patterns resulting from responses of vegetation and ecosystems to the effects
 7      of fine and coarse PM deposition, along with known or possible effects on ecological processes
 8      associated with changes in the patterns, are discussed in the subsections that follow.
 9           The reader is also referred to several other sources for more detailed discussions of several
10      topics only briefly alluded to or addressed here.  For example, an extensive discussion of various
11      types of effects of acidic deposition is presented in the U.S. National Acid Precipitation
12      Assessment Program (NAPAP) Biennial Report to Congress: An Integrated Assessment
13      Program (National  Scientific and Technology Council, 1998). Additionally, ecological effects
14      of acidic precipitation and nitrate deposition on aquatic  systems are discussed in the EPA Air
15      Quality Criteria Document for Nitrogen Oxides (U.S. Environmental Protection Agency, 1993);
16      and sulfate deposition and effects, as related to wetlands and aquatic habitats,  are discussed in
17      U.S. Environmental Protection Agency (1982). Effects of lead on crops, vegetation, and
18      ecosystems are assessed in the EPA document, Air Quality Criteria for Lead (U.S.
19      Environmental Protection Agency, 1986).  Lastly, effects of "certain pesticides, metal
20      compounds, chlorinated organic compounds, and nitrogen compounds" are discussed in
21      Deposition of Air Pollutants to the Great Waters, Third Report to Congress (U.S. Environmental
22      Protection Agency, 2000a).
23
24      4.2.2  Ecosystem Exposures - Particle Deposition
25           Airborne particles, their precursors, and their transformation products are removed from
26      the atmosphere by wet and dry deposition processes.  This atmospheric cleansing process
27      fortunately lowers the long-term buildup of lethal concentrations of these pollutants in the air
28      and moderates the potential for direct human health effects caused by their inhalation.
29      Unfortunately, these deposition processes also mediate the transfer of PM pollutants  to other
30      environmental media where they can and do alter the structure, function, diversity, and
31      sustainability of complex ecosystems.

        June 2003                                  4-4        DRAFT-DO NOT QUOTE OR CITE

-------
 1           The potential effects of PM deposition on vegetation and ecosystems encompass the full
 2      range, scales, and properties of biological organization listed under Biotic Condition, Table 4-1.
 3      Exposure to a given mass concentration of airborne PM, however, may lead to widely differing
 4      responses, depending on the particular mix of deposited particles.  Particulate matter is not a
 5      single pollutant, but rather a heterogeneous mixture of particles differing in size, origin, and
 6      chemical composition.  This heterogeneity exists across individual particles within samples from
 7      individual sites and, to an even greater extent, between samples from different sites. Thus far,
 8      atmospheric PM has been defined, for regulatory purposes, mainly by size fractions and less
 9      clearly so in terms of chemical nature, structure, or source. While size is related to the mode and
10      magnitude of deposition to vegetated landscapes and may be a useful surrogate for chemical
11      constitution, PM size classes do not necessarily have specific differential relevance for
12      vegetation effects (Whitby, 1978; U.S. Environmental Protection Agency, 1996a); that is, both
13      fine- and coarse-mode particles may affect plants. Much of the burden of sulfates, nitrates,
14      ammonium salts, and hydrogen ions resides in the atmosphere either dissolved in fog water or as
15      liquid or solid aerosols.  Therefore, assessment of atmospheric PM deposition and effects on
16      vegetation unavoidably include discussion of nitrates and sulfates and associated compounds
17      involved in acidic and acidifying deposition. Other important issues relate to trace elements and
18      heavy metals often found in ambient airborne PM.
19
20      4.2.2.1  Fine and Coarse Particulate Matter
21           Particulate matter in current U.S. regulatory standards is classified as fine PM (PM25;
22      < 2.5 |im aerodynamic diameter) and coarse PM (2.5-10 jim).  These combined fractions
23      constitute PM10 (U.S. Environmental Protection Agency, 1996a,b).
24           Fine and coarse PM have a number of contrasting properties that affect their impact on
25      vegetated systems (see Chapter 2, Table 2-1  of this document). The model results of Wiman and
26      Agren (1985) and the measurements of Lovett and Lindberg  (1993) addressing the complexity of
27      deposition processes in patchy forested landscapes and vertical stratification within canopies
28      reveal clear distinctions between the deposition behavior of fine and coarse particles. For one,
29      coarse particles settle nearer their site of formation than do fine particles.  Also, the chemical
30      constitution of individual particles is strongly correlated with size (i.e., most S and much N is
31      present on fine particles, whereas much  of the base cation and heavy metal burden is present on

        June 2003                                 4-5         DRAFT-DO NOT QUOTE OR CITE

-------
 1      coarse particles) and influences the predicted landscape loading of specific elements.
 2      Atmospheric PM may also act as a carrier for other directly phytotoxic materials (e.g.,
 3      herbicides). Fine PM dominates the surface area of particles suspended in the atmosphere, while
 4      coarse PM dominates the mass of such airborne particles.  Surface area may become more
 5      central to ecological impact assessment as recognition of the oxidizing capacity of fine particles,
 6      their interactions with other pollutants such as ozone, and their adsorption of phytoactive organic
 7      compounds such as herbicides become more fully appreciated. Fine and coarse particles respond
 8      to changes in atmospheric humidity, precipitation, and wind through different mechanisms,
 9      differentially altering their deposition characteristics.
10           Fine PM may exhibit similar mass concentrations at different sites (Figure 4-1) and yet be
11      composed of very different constituents.  In eastern North America, sulfate typically is the major
12      component of this fraction, in contrast to the West where nitrate is a key component (Figure 4-1;
13      cf., the eastern Appalachian site and the western California site). On the other hand, in the urban
14      center of Mexico City (Hidy et al., 2000), an environment more similar to the western than
15      eastern U.S., concentrations of fine PM of about 300 jig m"3 are found, and sulfate concentrations
16      are 3 times that  of nitrate. In contrast to sulfur and nitrogen, the contributions of organic and
17      elemental carbon to the eastern and western U.S. sites were similar (Figure 4-1); soil-derived
18      geologic material was greater at the more arid western site.
19           Fine PM is typically more  diverse than coarse PM and is secondary in nature,  having
20      condensed from the vapor phase or been formed by chemical reaction from gaseous precursors
21      in the atmosphere (see Chapter 2). Fine PM derives from atmospheric gas-to-particle conversion
22      reactions involving nucleation, condensation, and coagulation, and from evaporation of water
23      from contaminated fog and cloud droplets. Sulfur and nitrogen oxides (SOX and NOX) are often
24      oxidized to their respective acids and neutralized with ammonium cations as particulate salts.
25      Fine PM may also contain condensates of volatile organic compounds, volatilized metals, and
26      products of incomplete combustion (see Chapter 3).
27           Coarse PM, in contrast, is mainly primary in nature, having been emitted from area or
28      point sources as fully formed particles derived from abrasion and crushing processes, soil
29      disturbances, desiccation of marine aerosol from bursting bubbles, hygroscopic fine PM
30      expanding with  humidity to coarse mode, and/or gas condensation directly onto pre-existing
31      coarse particles  (see Chapters 2 and 3).  Suspended primary coarse PM may contain iron, silica,

        June 2003                                  4-6          DRAFT-DO NOT QUOTE OR CITE

-------
                                                         Southern California
                                 SO42
                              %
                           % Organic
                         % Geologic
                           % Carbon

                              % SO
                              %
                           % Organic
                         % Geologic
                           % Carbon
                                            10    20    30   40    50    60
       Figure 4-1.  The diversity of fine PM from sites in the western and eastern U.S.
       Modified from Sisler and Malm (2000).
 1     aluminum, and base cations from soil, plant and insect fragments, pollen, fungal spores, bacteria,
 2     and viruses, as well as fly ash, brake linings, debris, and automobile tire fragments. Coarse-
 3     mode particles can be altered by chemical reactions and/or physical interactions with gaseous or
 4     liquid contaminants.
 5          The coating of coarse particles with semivolatile materials can substantially affect their
 6     deposition and potential for biological effects. For example, nitrogen exhibits a strongly
 7     bimodal size distribution: the peak above 1 |im can be attributed to HNO3 adsorption onto coarse
 8     alkaline particles; and that below 1 |im can be attributed to gas phase condensation of ammonia
 9     with either sulfuric or nitric acid yielding either (NH4)2SO4 or NH4NO3 aerosol.  HNO3 has an
10     extremely high deposition velocity, nearly independent of the physiology of the surface.
       June 2003
4-7
DRAFT-DO NOT QUOTE OR CITE

-------
 1      Therefore, formation of ammonium nitrate reduces nitrogen deposition, because the deposition
 2      velocity of these particles is much less than that of HNO3 gas.
 3           Similarly, anthropogenic emissions of sulfur are mostly as sulfur dioxide (SO2), which is
 4      hydrophilic, rapidly hydrated, and subsequently oxidized to sulfate (SO4"2), which is about
 5      30-fold less phytotoxic than SO2. The ratio of SO4"2: SO2 increases with aging of the air mass
 6      and, therefore, with distance from the source.  Sulfate is thus a widespread regional/global
 7      pollutant and is sufficiently hygroscopic, that in humid air it exists significantly in the coarse PM
 8      fraction.  It is unusual for injurious levels of particulate sulfate to be deposited upon vegetation,
 9      while direct injury due to SO2 is commonly observed near uncontrolled point sources. Gas-to-
10      particle conversion in this case is of benefit to vegetation.  The chemical composition of gaseous
11      precursors of parti culate matter and the formation of sulfates and nitrates is discussed in Section
12      2.1.3 of Chapter 2.
13           Since enactment of the 1990 Clean Air Act Amendments, the atmospheric mix of PM
14      precursors in the United States has changed substantially.  That is, as emissions of SO2 have
15      declined, emissions of oxides of nitrogen (NOX) have remained about the same, but emissions of
16      cations have increased. This is almost certainly due to increased suspension of wind-borne
17      geologic material  from exposed soils.
18           For characterization of tropospheric chemistry, deposition of O3, NOX, peroxides, and
19      ammonia are first-order problems, followed by deposition of organics, SO2, and parti culate
20      sulfate and nitrate (Wesely and Hicks, 2000). For impact on vegetation, however, the order may
21      be different and may include different species — notably SO2, fluoride where it still exists as a
22      problem, parti culate heavy metals, base cations, sulfate and nitrate. In spite of the current
23      regulatory focus on non-speciated PM, exposure to a given mass concentration of PM may lead
24      to widely differing phytotoxic outcomes depending upon the particular mix of PM constituents
25      involved.  This variability has not been characterized adequately.  Though effects of specific
26      chemical fractions of PM have been described, there has been relatively little research aimed at
27      defining the effects of unspeciated PM on plants or ecosystems.
28
29      4.2.2.2  Diversity of Deposition Modes
30           Atmospheric deposition of particles to ecosystems takes place via both wet and dry
31      processes, through three major routes: (1) wet, by precipitation scavenging in which particles

        June 2003                                  4-8        DRAFT-DO NOT QUOTE OR CITE

-------
 1      are deposited in rain and snow; (2) much slower dry deposition; and (3) occult deposition (so
 2      named because it was hidden from measurements which determined the previous two types of
 3      deposition) by fog, cloud-water, and mist interception (Table 4-2).  Unlike gaseous dry
 4      deposition, neither the solubility of the particle nor the physiological activity of the surface are
 5      likely to be of first order importance in determining particulate dry deposition velocity (Vd).
 6      Factors that contribute to surface wetness or stickiness may be critical determinants of deposition
 7      efficiency.  Available tabulations of deposition velocity are highly variable and suspect. Recent
 8      evidence indicates that all three modes of deposition (wet, occult, and dry) must be considered in
 9      determining inputs to watersheds or ecosystems because each may dominate over specific
10      intervals of time or space and ultimately, by interception and impaction onto vegetation or other
11      rough elements in the landscape.
12           The distribution of deposition between wet,  dry, and occult modes varies substantially
13      between locations for both nitrogen and sulfur (Figure 4-2).  Clearly, rainfall and snowfall will
14      determine the magnitude of wet deposition.  Precipitation events clean the air so that dry
15      deposition is eliminated or reduced during subsequent dry periods.  Occult deposition depends
16      upon landscape interception of the cloud base (Cape, 1993). This may occur at high elevation
17      sites, in coastal areas subject to onshore advection, or in low-lying interior areas subject to
18      radiation fogs. Thus, ecosystem exposure is determined by the mode, and to some extent the
19      magnitude, of deposition. Total deposition particularly for nitrogen, among mountain sites is
20      related to the magnitude of the occult deposition, particularly for nitrogen (Figure 4-2).
21      Topography and vegetation characteristics influence the deposition modes differently
22      (Table 4-3). In general, dry deposition is the most sensitive and wet deposition is the least
23      sensitive to features of the vegetated surface.
24           Comparison of micrometeorological and other methods for estimating particle deposition
25      velocity (Erisman et al., 1997) suggests that there is little discrepancy between contrasting
26      methodologies for estimating particle deposition and that this conclusion holds for both anions
27      and cations, with the  exception of nitrogenous species, which appear to interact with foliage in
28      more complex ways.  These comprehensive studies in the Speulder forest in the Netherlands
29      indicated that aerosol deposition represents a considerable fraction of total deposition to the
30      landscape.  At this location, occult deposition was relatively insignificant, but dry deposition
31      accounted for about one-fourth of alkaline-earth cation deposition.

        June 2003                                  4-9         DRAFT-DO NOT QUOTE OR CITE

-------
    TABLE 4-2.  TYPES AND DETERMINANTS OF PARTICIPATE DEPOSITION
                            AND IMPACT TO VEGETATION
 Type of Deposition       Determinant of Deposition
                               Quantifiable Factors
 Dry Deposition
 Wet Deposition
Ambient Concentration
                         Atmospheric Conditions
                         Aerosol Properties
                         Surface Roughness
 Occult Deposition
Vegetation Condition




Ambient Concentration


Atmospheric Conditions




Aerosol Properties



Surface Roughness




As Above
Distance from Source
Emission Strength

Wind Speed
Stability
Mixing Height
Temperature
Humidity
Dew Formation

Chemical Reactivity
Particle Solubility
Aerodynamic Diameter
Biological Availability
Hygroscopicity

Terrain Discontinuity
Leaf Pubescence
Leaf Shape
Plant Density
Branch Spacing
Tissue Flexibility

Surface Wetness
Salt Exudates
Organic Exudates
Insect Excreta

Distance from Source
Emission Strength

Mixing Height
Timing of Precipitation
Intensity of Precipitation
Duration of Precipitation

Chemical Reactivity
Particle Solubility
Biological Availability

Terrain Discontinuity
Leaf Pubescence
Leaf Area Index
Nature of Exposed Bark and Stem

Combination of Above Factors
June 2003
                   4-10
DRAFT-DO NOT QUOTE OR CITE

-------
80-
60-
40 -
20 -
o -1
140-i
120 -
100-
80-
60-
40 -
20 -
^aaaaa^
ggg=j

1
r~l

5?
•fr.
3?
i
w
5$
1 HKf








K»I=^
w^^
>O^^3
	 »«=j
= m====j
= ff^^
= ^>pa ^
1
1 1 1 1

H
I
^
1
1
xxx:
^
wX
1
P
>&x
1
1
JOOC
kSoc
m
m
m
RXX:
S88:
I
m
OO&<
p
I
m.
Total
n
E3
B Whitetop


™
sll
1
I n_ 1
I Qcqult I Dfy I Wet '
(kg / har1) %
Figure 4-2. Relative importance of three modes of deposition of nitrate (A) and sulfate (B)
           at high elevation sites (Unsworth and Wilshaw, 1989; Fowler et al., 1989;
           Mueller, 1988; Aneja and Murthy, 1994).
June 2003
4-11
DRAFT-DO NOT QUOTE OR CITE

-------
            TABLE 4-3.  RELATIVE IMPORTANCE OF WET, DRY, PARTICIPATE, AND
                          TOTAL DEPOSITION TO THREE FOREST SITES3
Deposition
Total Nitrogenb
Site
Duke Forest
Gary Forest
Austin Forest
Wet
(%)
75
71
71
Dry
(%)
25
20
29
Particle
(%)
0.11
0.94
0.58
Total
(kg ha'1)
9.87
5.80
6.57
Wet
(%)
64
76
83
Total Sulfur0
Dry
(%)
33
20
13
Particle
(%)
2.7
4.2
4.3
Total
(kg ha'1)
17.20
7.60
7.79
         "Data from Allen et al. (1994). Sampling was by triple filter pack so that fine-mode particles could be sampled
         preferentially. An average particle deposition velocity of 0.9 cm s"1 was derived as in Hicks et al. (1987).
         bWet nitrogen consists of NO3 and NH4+; dry nitrogen consists of vapor phase HNO3 and NO2; and paniculate
         nitrogen consists of NO3".
         °Wet sulfur consists of SO4"2, dry sulfur consists of vapor phase SO2, and paniculate sulfur consists
 1     Wet Deposition
 2           Wet deposition results from the incorporation of atmospheric particles and gases into cloud
 3     droplets and their subsequent precipitation as rain or snow, or from the scavenging of particles
 4     and gases by raindrops or snowflakes as they fall (Lovett, 1994). Precipitation scavenging, in
 5     which particles are incorporated in hydrometers and deposited in the resulting rain and snow,
 6     includes rainout (within-cloud incorporation by nucleation) and washout (below-cloud
 7     scavenging by impaction). Wet deposition generally is confounded by fewer factors than dry or
 8     occult deposition and has been easier to quantify.  Total inputs from wet deposition to vegetative
 9     canopies can be significant (Table 4-3), although not all wet deposition involves particle
10     scavenging because gaseous pollutants also dissolve in raindrops during precipitation events
11     (Lovett, 1994). This contribution is obscured during measurements because wet deposition is
12     measured simply by chemical analysis  of total precipitation collected in clean, non-reactive
13     buckets. Exclusion of dry deposited material (as opposed to dissolved gaseous species) requires
14     closure or covering of the vessels except during precipitation.
15           Wet deposition is largely a function of precipitation amount and ambient pollutant
16     concentrations. It is not affected by surface properties (Table 4-2) as much as dry or occult
17     deposition although leaves (depending  on their surface properties of wettability,  exposure, and
18     roughness) retain liquid and solubilized PM.  Extensive vegetative canopies typically develop
        June 2003
4-12
DRAFT-DO NOT QUOTE OR CITE

-------
 1      leaf area indices (LAI; ratio of proj ected leaf area to ground area) much greater than 1.  Thus any
 2      material deposited via precipitation to the upper stratum of foliage likely will be intercepted by
 3      several foliar surfaces before reaching the soil.
 4           Landscape characteristics may also be important.  Forested hillsides receive four- to
 5      six-fold greater inputs of wet deposition than short vegetation in nearby valleys. This is due to a
 6      variety of orographic effects (Unsworth and Wilshaw, 1989) and closer aerodynamic coupling to
 7      the atmosphere of tall forest canopies than of the shorter canopies in the valleys. This leads to
 8      more rapid foliar drying, which reduces the residence time but concentrates the solubilized
 9      particulate materials available for foliar uptake on the cuticular surface more quickly; and
10      concentration increases the thermodynamic driving force for foliar uptake (Fowler et al.,  1991;
11      Unsworth, 1984; Schonherr and Huber, 1977). Following wet deposition, humidity and
12      temperature conditions strongly influence the extent of biological effects, because of the
13      competing effects of drying versus concentrating the solutions, and influence the rate of
14      metabolic uptake of surface solutes (Swietlik and Faust, 1984). The net consequence of these
15      factors on direct physical effects of wet deposited PM on leaves is not known.
16           Rainfall both introduces new wet  deposition and redistributes throughout the canopy
17      previously dry-deposited particulate material, particularly coarse particles which are
18      preferentially deposited in the upper foliage (Peters and Eiden,  1992). Both effects scale the
19      likelihood of foliar contact and potential direct PM effects on vegetation nearly linearly with
20      canopy leaf area.  The concentrations of suspended and dissolved materials are typically highest
21      at the onset and decline with duration of individual precipitation events  (Lindberg and
22      McLaughlin,  1986;  Hansen et al., 1994).  Sustained rainfall removes much of the accumulation
23      of dry-deposited PM from foliar  surfaces, reducing direct foliar effects and combining the
24      associated chemical burden with the wet-deposited material (Lovett and Lindberg, 1984; Lovett,
25      1994) for transfer to the soil. Intense rainfall may contribute substantial total particulate inputs
26      to vegetated land surfaces, mostly via the soil, but is less effective as a source of directly
27      bioavailable or injurious pollutants to foliar surfaces. This washing effect, combined with
28      differential foliar uptake and foliar leaching of different chemical constituents of PM alters the
29      composition of the rainwater that reaches the soil. Low intensity precipitation events, in
30      contrast, may deposit significantly more particulate pollutants to foliar-surfaces than high
31      intensity precipitation events.  Because of the short duration and limited atmospheric cleansing,

        June 2003                                  4-13        DRAFT-DO NOT QUOTE OR CITE

-------
 1      the concentration of PM in the final precipitation that remains in contact with foliar surfaces may
 2      be high. Additionally, such events may facilitate foliar uptake by hydrating some previously
 3      dry-deposited particles without removing them. This combination of dry deposition to foliage
 4      and subsequent wet removal increases the potential for PM to exert effects via soil pathways:
 5      first by enhancing dry deposition relative to adjacent unvegetated surfaces; and second by
 6      accelerating passage along with wet deposited material of the deposited PM by throughfall and
 7      stemflow to the soil.  In the soil, PM may affect important ecosystem-level biogeochemical
 8      cycles of major, minor, and trace elements.
 9
10      Dry Deposition
11           Dry deposition of atmospheric particles to plant and soil is a much slower process than wet
12      or occult deposition, but it acts nearly continuously and affects all exposed surfaces (Hicks,
13      1986).  In dry deposition, particles at the large end of the spectrum (i.e., > 5  jim diameter) are
14      deposited mainly by gravitational sedimentation and inertial impaction. Smaller particles,
15      especially those with diameters between ~ 0.2 and 2 jim, are not readily dry-deposited and tend
16      to travel long distances in the atmosphere until their eventual deposition, most likely by
17      incorporation into precipitation. This long-distance transport of fine aerosols is largely
18      responsible for the regional nature of acid deposition (Lovett, 1994). A major conclusion from
19      atmospheric deposition research is the realization that dry deposition is usually a significant and,
20      in some cases, a dominant portion of total atmospheric deposition to an ecosystem (Lovett,
21      1994).  Plant parts of all types, including those not currently physiologically active, along with
22      exposed soil and water surfaces, receive steady deposits of dry dusts, elemental carbon
23      encrustations, grease films, tarry acidic coatings, and heterogeneous secondary particles formed
24      from gaseous precursors (U.S. Environmental Protection Agency, 1982).  The range of particle
25      sizes, the diversity  of canopy surfaces, and the variety of chemical constituents in airborne PM
26      have slowed progress in both prediction and measurement of dry particulate deposition.
27      Particulate deposition is a complex, poorly characterized process controlled primarily by
28      atmospheric stability, macro- and micro-surface roughness, particle diameter, and surface
29      characteristics (Table 4-2; Hosker and Lindberg, 1982).  Deposition of particles suspended
30      regionally and throughout the full depth of the planetary boundary layer (PEL) is controlled by
31      different mechanisms within the three distinct atmospheric transport zones above the surface.

        June 2003                                 4-14        DRAFT-DO NOT QUOTE OR CITE

-------
 1      In the lower atmosphere, fine particles are transported by turbulent eddies of mechanical and
 2      convective origin. In the relatively unstirred, laminar boundary layer surrounding individual
 3      surface elements, Brownian diffusion dominates.  Near the surface, actual deposition and contact
 4      with the surface is mediated by impaction (El-Shobokshy, 1985).
 5           Deposition fluxes may be calculated from measurements, estimates, or modeled values of
 6      mass concentration (C) at a specified measurement height and the total conductance or
 7      deposition velocity (Vd) from this height to the surface (Eq. 4-1; Hicks et al., 1987). These
 8      modeling techniques are closely allied with the micrometeorological techniques used to measure
 9      such fluxes. The flux (F) may be inferred as:
10
11                                     F = Vd * (Cz - C0),                                 (4-1)
12
13      where F is flux to the surface, Cz is the particle concentration at measurement height z, C0 is the
14      particle concentration at receptor sites in the canopy (usually assumed equal to 0), and Vd is the
15      overall deposition velocity. The flux is controlled by Vd and Cz.
16           Vertical transport of particles through the lower atmosphere to the vicinity of the
17      vegetation elements is by turbulence and sedimentation, such that:
18
19                                        Vd = Vt + Vs                                    (4-2)
20
21      in which V, (inner, left hand pathway of Figure 4-3) is a turbulent diffusion term, and Vs is a
22      sedimentation term that dominates deposition of very coarse particles (Figure 4-4) and increases
23      with particle size (Figure 4-5; dotted line).  Sedimentation may be considered a pathway parallel
24      to turbulent transport (Figure 4-4), but this is an over simplification. Vs affects the concentration
25      of particles near the surface where eddy transport may occur and also governs the redeposition of
26      some fraction of the particles lost to resuspension or rebound following deposition by impaction.
27      For this reason, Vs is included (Figure 4-3) in the  composite surface resistance term (RaR^VJ  as
28      well as in the parallel sedimentation term.
29           For submicron particles for which sedimentation is negligible (Hicks et al., 1987; Monteith
30      and Unsworth, 1990; Wesely,  1989), the Ohm's Law Analogy (resistance catena) analogous to
31      that used to describe transport of heat, momentum, or gases may be adequate, as:
        June 2003                                4-15        DRAFT-DO NOT QUOTE OR CITE

-------
                             v;
                                                  Particles
                                                 Atmospheric
                                                    Source
                                                    Particles
                                                            Vs
       Figure 4-3. A simplified resistance catena representing the factors controlling deposition
                   of particles to the surface. Vegetation-specific parameters are not explicitly
                   considered.  Modified after Hicks et al. (1987).
 1
 2
 3
 4
 5
                             Vd = Vt = [r. + rb + rj-1,
                             (4-3)
 9
10
11
where V, is the deposition velocity due to turbulent transport of particles or other entities through
the atmosphere; rais aerodynamic resistance (inverse of conductance or velocity) associated with
the efficiency of turbulent transport above the canopy; rbis the boundary layer resistance
associated with diffusional transport through the still air layer immediately adjacent to canopy
elements; and rcis canopy resistance associated with physiological control of leaf porosity via
stomata in the leaf surface.  Significant departures from the analogy arise near the surface
(Chamberlain, 1975; Sehmel, 1980), as particles that were transported efficiently by turbulent
eddies are slowed substantially in the laminar boundary layer that reduces the efficiency of
impaction.  The preservation of momentum in this zone declines with decreasing diameter;
       June 2003
                                          4-16
DRAFT-DO NOT QUOTE OR CITE

-------
                           8
                           7-
                       E  6-
£  5H
o
o^
•a>  4H
I  3H
o
Q.  0
(D  2
Q
                                Zn^.--' -1-
                                          Mn
                                         Cu
                                                                          Cl
                                                         Fe
                             0
                1234
                Coarse/Fine Ratio (M9/M9)
       Figure 4-4. The relationship between deposition velocity of selected particulate materials
                   and the distribution of the material between the coarse- and fine-aerosol
                   fractions. Data from Foltescu et al. (1994). Ranges for Mn and Fe are from
                   Davidson and Wu (1989).
 1     however, this is offset by an increase in Brownian diffusivity with decreasing diameter
 2     (Figure 4-3).  Aerodynamic streamlines are parallel to the surface of each roughness element, so
 3     that deposition ultimately depends on diffusion to the surface. The transition from impaction to
 4     diffusion is likely blurred in the presence of leaf pubescence extending beyond the boundary
 5     layer.  These conflicting trends lead to a broad range over which empirical measurements of Vd
 6     and particle size are relatively independent (Figure 4-3), further demonstrating the importance of
 7     the quasilaminar boundary layer (Lamaud et al., 1994; Shinn,  1978).
 8          The aerodynamic term (ra)  decreases with increasing wind speed, turbulence, and friction
 9     velocity and increases with measurement height and atmospheric stability. It describes the
10     capacity of turbulent eddies to transport material, momentum, and heat between the
11     measurement height and the roughness height of the surface. Coarse particles may not be carried
12     efficiently by the high frequency eddies near the surface and may fall more rapidly than they
       June 2003
                          4-17
DRAFT-DO NOT QUOTE OR CITE

-------

                O
                o
                c
                o
             1,000  -

               100  J

                 10

                  1

                0.1
                w   0.01
                o
                Q.
                S  0.001
                    0.000
                                        Stokes Law
                                        Brownian Diffusion
                                        Peters and Eiden (1992)
                                        Little and Wiffen (1977)

                                  I          r
                    0.001     0.01      0.1
                                                                      10
                     100
                                        Particle Diameter
      Figure 4-5. The relationship between particle diameter and deposition velocity for
                  particles. Values measured in wind tunnels by Little and Wiffen (1977) over
                  short grass with wind speed of 2.5 m s"1 closely approximate the theoretical
                  distribution determined by Peters and Eiden (1992) for a tall spruce forest.
                  These distributions reflect the interaction of Brownian diffusivity (descending
                  dashed line), which decreases with particle size and sedimentation velocity
                  (ascending dotted line from Stokes Law), which increases with particle size.
                  Intermediate-sized particles (»0.1 to 1.0 um)  are influenced strongly by both,
                  and deposition is  independent of size.
1

2

3

4

5

6

7
diffuse by either Brownian or turbulent processes. Thus, the relevance of ra breaks down as Vs

increases. Indeed because Vs (Eq. 4-2) is independent of a concentration gradient, the electrical

analogy is a theoretically flawed approximate approach (Venkatram and Pleim, 1999).


Deposition Velocity

     Because the final stage of deposition for particles involves either impaction following
deceleration through a quasi-laminar boundary layer or diffusion through this boundary layer, its
      June 2003
                                         4-18
DRAFT-DO NOT QUOTE OR CITE

-------
 1      effective depth is a critical determinant of Vd (Wiman et al., 1985; Peters and Eiden, 1992). The
 2      term corresponding to the boundary layer resistance for gases (rb; equation 4-3) incorporates the
 3      absence of form drag for gases.  This parameter decreases with increasing turbulence and
 4      particle diffusivity but is poorly characterized for gases,  depending critically on canopy
 5      morphology, vertical wind profiles, and gust penetration, and is of extremely limited usefulness
 6      for particles.
 7           Once delivered by turbulent transport or sedimentation to the vicinity of vegetative surface
 8      elements, a variety  of particle size-dependent mechanisms come into play, some differing
 9      substantially from those governing gaseous deposition.  The concepts of rb (the still air or
10      boundary layer resistance) and rc (the canopy or surface resistance) are not generally applicable
11      to deposition of poly disperse particles. Because of the roles of momentum and bounce-off and
12      complication by reentrainment back into the airstream following deposition of a particle to the
13      surface, the factors determining the effective rb and rc for particle deposition are not as
14      independent as for gases.  They are replaced in some resistance formulations (e.g., Hicks et al.,
15      1987) by the term, rcp, that combines near-surface and surface effects and by a mathematically
16      derived composite term, RaRcpVs, that combines atmospheric, surface, and sedimentation effects
17      (Figure 4-4). This latter term was insignificant for the submicron sulfate component considered
18      originally in its derivation (Hicks et al.,  1987), but it scales with the square of particle diameter,
19      so that its general applicability to poly disperse particles is unclear. In general, transport between
20      the turbulent air column and the leaf surface through the laminar boundary layer remains
21      difficult to describe (Lindberg and McLaughlin,  1986).
22           Current estimates of regional particulate dry deposition (e.g., Edgerton et al., 1992; Brook
23      et al., 1999) infer fluxes from the product of (variable and uncertain) measured or modeled
24      particulate concentrations and (even more variable and uncertain) measured or modeled
25      estimates of dry deposition velocity parameterized for a variety of specific surfaces (e.g., Brook
26      et al., 1999). However, even for specific sites and well defined particles, uncertainties in F are
27      largest in the values of Vd, which are typically characterized by the large ranges and variances
28      described in other sources (e.g., Bytnerowicz et al.,  1987a,b,  Hanson and Lindberg, 1991, for
29      nitrogen-containing particles; McMahon and Denison, 1979, Hicks et al.,  1987, for general
30      treatment).  The nature of the vegetative cover to which particulate deposition occurs has a
31      moderate to substantial effect on the components of Vd.  The surface resistance (Hicks et al.,

        June 2003                                 4-19        DRAFT-DO NOT QUOTE OR CITE

-------
 1      1987) is a significant and highly site-specific component of total resistance that is difficult to
 2      predict along with site, seasonal, and diurnal effects on the atmospheric components of total
 3      resistance.
 4           Early models of dry particulate deposition to vegetation (e.g., U.S. Environmental
 5      Protection Agency, 1982; Chamberlain, 1975; Davidson and Friedlander,  1978; Garland, 1978;
 6      Little and Wiffen, 1977; McMahon and Denison, 1979; Sehmel, 1980; Sehmel and Hodgson,
 7      1976; and Slinn, 1977, 1978) used this paradigm (e.g., Eq. 4-3) to deal with transport to the near-
 8      surface regime explicitly including conventional micrometeorological and particle size
 9      considerations. Alternative modeling treatments have attempted to parameterize the geometry of
10      vegetative receptor surfaces and within-canopy micrometeorology (Wiman and Agren, 1985;
11      Peters and Eiden, 1992).  Chemical reactivity, particle shape and density,  rates of physiological
12      sequestration, and reentrainment by gusts of wind remain to be addressed. Modeling the
13      deposition of particles to vegetation is at a relatively early stage of development, and it is not
14      currently possible to identify  a best or most generally applicable modeling approach. These
15      approaches have been further elaborated with canopy-specific choices among the available
16      models and with specific incorporation of capture efficiencies by Brook et al. (1999).
17
18      Methods of Measuring Dry  Deposition
19           Methods of measuring dry deposition  of particles are more restricted than for gaseous
20      species and fall into two major categories (Davidson and Wu, 1990).  Surface extraction or
21      washing methods characterize the accumulation of particles on natural receptor surfaces of
22      interest or on experimental surrogate surfaces.  These techniques rely on methods designed
23      specifically to remove only surface-deposited material (Lindberg and Lovett, 1985). Total
24      surface rinsate may be equated to accumulated deposition or to the difference in concentrations
25      in rinsate between exposed and control (sheltered) surfaces and may be used to refine estimates
26      of deposition (John et al., 1985; Dasch, 1987).  In either case, foliar extraction techniques may
27      underestimate deposition to leaves because  of uptake and translocation processes that remove
28      pollutants from the leaf surface (Taylor et al.,  1988; Garten and Hanson, 1990). Foliar extraction
29      methods also cannot distinguish sources of chemicals (e.g., N) deposited as gases from those
30      deposited as particles (e.g., nitric acid [HNO3] or nitrate [NO3"] from nitrogen dioxide [NO2], or
31      ammonia [NH3] from ammonium [NH4+]; Bytnerowicz et al.,  1987a,b; Dasch, 1987; Lindberg

        June 2003                                  4-20       DRAFT-DO NOT QUOTE OR CITE

-------
 1      and Lovett, 1985; Van Aalst, 1982). Despite these limitations, these methods are often used in
 2      the development of in-canopy deposition models (McCartney and Aylor, 1987).
 3           Deposition of pollutants by wet deposition is relatively straightforward to determine
 4      through analysis of precipitation samples.  Dry deposition of pollutants, on the other hand, is
 5      more difficult to measure. A significant limitation on current capacity to estimate regional
 6      impacts of PM is inadequate knowledge of the mechanism and factors governing particle dry
 7      deposition to diverse surfaces. This has hindered efforts to develop robust measurement
 8      techniques for particle deposition  (distinct from atmospheric concentration) and has
 9      compromised efforts to develop generally applicable deposition models for particles.  The
10      National Dry Deposition Network (NDDN) was established in 1986 to document the magnitude,
11      spatial variability, and trends in dry deposition across the United States.  Currently, the network
12      operates as a component of the Clean Air Status and Trends Network (CASTNet; Clarke et al.,
13      1997).
14           Dry deposition is not measured directly, but  is determined by a inferential  approach (i.e.,
15      fluxes are calculated as the product of measured ambient concentration and a modeled deposition
16      velocity).  This method is appealing and widely used because atmospheric concentrations are
17      relatively easy to measure when compared to dry deposition fluxes, and models  have been
18      developed to calculate deposition  velocities (Lovett, 1994). Ambient pollutant concentrations,
19      meteorological conditions, and land use data required for the inferential model are routinely
20      collected at CASTnet dry deposition sites. Monitored chemical species include  ozone, sulfate,
21      nitrate, ammonium, sulfur dioxide, and nitric acid.  The temporal resolution for the ambient
22      concentration measurements and dry deposition flux calculations is hourly for ozone and weekly
23      for the other chemical  substances  (Clarke et al., 1997).  Isotopic labeling of dry deposited PM
24      (e.g., sulfate with 35S) prior to experimental surface exposures and extractions (Garten et al.,
25      1988) can provide more precise differentiation between the deposition rates of related chemical
26      species (e.g., sulfate [SO4"2] from  sulfur dioxide [SO2]).
27           At the whole-canopy level, natural surface washing by rainfall may be used to estimate dry
28      deposition of PM and gases during the preceeding dry period  (Cape et al., 1992; Davidson and
29      Wu, 1990; Draaijers and Erisman, 1993; Erisman,  1993; Fahey et al., 1988; Lindberg and Lovett,
30      1992; Lovett and Lindberg, 1993; Reiners and Olson, 1984; Sievering, 1987). Collection and
31      analysis of stem flow and throughfall provides useful  estimates of particulate deposition when

        June 2003                                  4-21        DRAFT-DO NOT QUOTE OR CITE

-------
 1      compared to directly sampled precipitation.  The method is most precise for strictly PM
 2      deposition when gaseous deposition is a small component of the total dry deposition and when
 3      leaching or uptake of compounds of interest out of or into the foliage (i.e., N, S, base cations) is
 4      not a significant fraction of the total deposit!onal flux (Davidson and Wu,  1990; Draaijers and
 5      Erisman,  1993; Lindberg and Lovett, 1992; Lovett and Lindberg,  1993). Throughfall sampling
 6      of sulfate deposition (Garten et al., 1988; Lindberg and Garten, 1988; Lindberg et al., 1990)
 7      often suggests substantial foliar exchange. Other throughfall studies (e.g., Erisman, 1993; Fahey
 8      et al., 1988) may lack sufficient specificity for dry particle deposition.
 9           Careful timing of throughfall and stemflow measurements after prolonged dry periods,
10      with simultaneous direct measurement of (unchanged) precipitation, can be used to determine
11      the magnitude of dry deposition between precipitation events.  Indeed, this foliar washing
12      technique, whether using subsequent precipitation or experimental lavage, is one of the best
13      currently  available methods to determine dry deposition of PM to vegetated ecosystems. Major
14      limitations are the extreme site specificity of the measurements, the substantial labor
15      requirement that normally precludes regional coverage, and the restriction to elements that are
16      conserved within the vegetative system (thereby excluding semivolatile organics, ammonium
17      salts, and gases  such as ozone).
18           Deposition to surrogate surfaces deployed in extensive plant canopies provides a measure
19      of particle deposition to the surrounding foliage or soil surfaces. For example, a uniform
20      population of 0.8 jim gold colloid particles were deposited similarly to leaves ofPhaseolus
21      vulgaris and to upward facing inert surfaces (Klepper and Craig, 1975).  However, comparison
22      of dry deposition of particles to foliage and to inert surrogate surfaces (polycarbonate Petri
23      dishes) in a deciduous forest demonstrated greater accumulation on the inert surfaces; with both
24      surfaces having accumulated particles of a similar range of sizes (Lindberg and Lovett, 1985).
25      These differences in deposition/accumulation remain to be fully characterized and hinder efforts
26      to use these surrogate techniques  to provide quantitative estimates of deposition. Surrogate
27      surfaces have not been found that can adequately replicate essential features of natural surfaces;
28      and therefore the surfaces currently used do not produce reliable estimates of particle deposition
29      to the landscape.
30           Micrometeorological methods employ an eddy covariance, eddy accumulation, or flux
31      gradient protocol in contrast to washing or extracting of receptor surfaces to quantify dry

        June 2003                                 4-22        DRAFT-DO NOT QUOTE OR CITE

-------
 1      deposition.  These techniques require measurements of PM concentrations and of atmospheric
 2      transport processes.  They are currently well developed for ideal conditions of flat,
 3      homogeneous, and extensive landscapes and for chemical species for which accurate and rapid
 4      sensors are available. Additional studies have expanded the range of such species and extended
 5      these techniques to more complex terrain (McMillen, 1988; Hicks et al., 1984; Wesely and
 6      Hicks, 1977).
 7           The eddy covariance technique measures vertical fluxes of gases and fine particles directly
 8      from calculations of the mean covariance between the vertical component of wind velocity and
 9      pollutant concentration (Wesely et al., 1982).  It is particularly limited by a requirement for
10      sensors capable of acquiring concentration data at 5-20 Hz. For the flux-gradient or profile
11      techniques, vertical fluxes are calculated from a concentration difference and an eddy exchange
12      coefficient determined at discrete heights (Erisman et al., 1988; Huebert et al., 1988). Businger
13      (1986), Baldocchi (1988), and Wesely and Hicks  (1977) evaluate the benefits and pitfalls of
14      these micrometeorological flux measurements for gases. Most measurements of eddy transport
15      of PM have used chemical sensors (rather than mass or particle counting) to focus on specific
16      PM components. These techniques have not been well developed for generalized particles and
17      may be less suitable for coarse PM10 transported efficiently in high frequency eddies (Gallagher
18      et al., 1988) for the same reasons that limit mathematical description of particle deposition.
19
20      Factors Affecting Dry Deposition
21           Ambient Concentration. The ambient concentration of particles (Cz; Eq. 4-1), the
22      parameter for which there is the most data (for example, see Chapter 3, this document), is at best
23      an indicator of exposure. However,  it is the amount of PM actually entering the immediate  plant
24      environment that determines the biological effect. The linkage between ambient concentration
25      and delivery to vegetation is the deposition velocity (Vd; Eq.  4-1). Cz is determined by regional
26      and local emission sources, regional circulation, and weather. It may be locally sensitive to
27      removal from the atmosphere by deposition, but the effect is  generally small.  Average annual
28      concentrations for NO3" exhibit much more variability than SO4"2 and a definite pattern of higher
29      concentrations in the midwest than elsewhere. The highest concentrations are observed (i.e, > 2
30      jig m"3) in the agricultural areas of the midwest, while the lowest are seen at forested sites in
31      New England and  the southern Appalachian Mountains.  Annual average concentrations SO4"2 of

        June 2003                                 4-23        DRAFT-DO NOT QUOTE OR CITE

-------
 1      5.0 jig m"3 are observed over nearly the entire eastern U.S. from New York and Michigan to
 2      northern Mississippi and Alabama (Edgerton et al., 1992).
 3           Deposition increases linearly with the concentration of many materials over a broad range.
 4      This allows atmospheric cleansing to take place and accounts for the greater surface impact of
 5      pollutants during pollution episodes. A serious limitation of the Vd formulation used to infer
 6      deposition of specific chemical species that exist in a range of particle sizes is an appropriate
 7      specification of their concentration.  Most sulfur emissions are readily oxidized to sulfite,
 8      bisulfite,  and sulfate. In the presence of atmospheric ammonia, paniculate ammonium sulfate is
 9      formed. However, this material is hygroscopic and will increase in mass and diameter in the
10      presence  of high humidity and alter its deposition behavior.  Similarly, coalescence of small
11      particles into larger aggregates and adsorption of gaseous pollutants onto existing coarse
12      particles complicate the association of particle size with concentration of individual chemical
13      species.
14           Distance and the resulting residence time in the atmosphere control the relative
15      concentrations of surface reactive materials (NO, SO2) of secondary particles that take some time
16      to form in the atmosphere and of coarse particles that exhibit high rates  of deposition by
17      sedimentation near the source.  These interacting processes affect the time required for the
18      formation of secondary particles by gas-to-particle conversion reactions and result in a greater
19      ratio of dry to wet deposition near emission sources where gaseous sulfur dioxide (gSO2)
20      deposition predominates than at greater distances where rainout of particulate  SO4"2 (pSO4"2) may
21      dominate (Barrie et al., 1984) and where dry deposition of pSO4"2 may be greater than that of
22      gSO2. The effect of gas-to-particle conversion on dry deposition of a specific  chemical species
23      can be substantial because Vd for SO2 is approximately 0.33 ± 0.17 cm s"1; whereas it is
24      approximately 0.16 ± 0.08 cm s"1 for SO4"2. These phase conversions affect both Cz and the
25      effective  Vd which together control dry deposit!onal fluxes (Eq. 4-1).  The neutralization of
26      acidic gaseous and particulate species by alkaline coarse particles has been described in arid
27      regions, but it may be more prevalent in urban New York where coarse  particles are observed to
28      be neutral because alkaline cations approximately balance gaseous acidic species (Lovett et al.,
29      2000). The deposition of the acidic materials in the urban environment  is likely enhanced by
30      incorporation into these previously formed coarse particles.
        June 2003                                 4-24        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Similarly, the ratio of coarse to fine particle concentrations determines the effective Vd for
 2      chemically speciated particles (Figure 4-4). This reflects the size-dependent deposition
 3      processes that govern delivery of PM to receptor surfaces (Fig. 4-5). For example, SO4"2 was
 4      found predominantly on fine submicron particles; whereas potassium ion (K+), calcium (Ca+2),
 5      and nitrate (NO3") were associated most often with coarse particles larger than 2 jim (Lindberg
 6      and Lovett, 1985).  However, concentrations of particulate S and K+ within a coniferous canopy
 7      were strongly correlated (Wiman and Lannefors, 1985), suggesting a primary source of
 8      coarse-mode sulfur particles.  However, many researchers reported mass-size distributions of
 9      NO3" were bimodal with one peak in the fine mode and the other peak in coarse mode (Wu and
10      Okada, 1994).  The behavior of NO3" in ambient aerosols depends not only on the concentrations
11      of gaseous nitric acid and ammonia, but also on the chemical composition of the particles and
12      atmospheric conditions (Tang, 1980). When sea-salt particles were transported from the Pacific
13      Ocean in the Nagoya, Japan, area, the amount of nitrate in the coarse particle size range
14      increased.  Coarse particle formation on sea salt under these conditions becomes a major
15      pathway for nitrate. The heterogeneous reaction of NaCl with gaseous HNO3 is considered to be
16      the dominating path (Wu and Okada, 1994).  As a result, marine and continental particle size
17      spectra for both N and S differ substantially: a peak in the coarse mode is generally apparent
18      near marine sources (Milford and Davidson,  1987). The issue for NO3" is further confounded by
19      uncertain discrimination between gaseous and parti culate species in current sampling methods.
20      The substantial effect of particle size on Vd (Figure 4-5) implies a need for size resolution as
21      well as chemically speciated ambient parti culate concentrations even within the PM10 fraction.
22
23           Particle Effects on Vd. Particle size is a key determinant of Vd as noted above; but,
24      unfortunately, the size spectra may be quite complex.  The particles in the study of Lindberg and
25      Lovett (1985) at Walker Branch Watershed had median diameters ranging from 3 to 5 jim; but
26      approximately 25% of the particles had diameters less than 1 |im (0.2 to 0.3 jim), and 5 to 20%
27      of the particles were much larger aggregates. The aggregated particles are significant in that
28      chemically they reflect their fine particle origins, but physically they behave like large particles
29      and deposit by sedimentation. Direct observations with SEM demonstrate that particle
30      morphology can be highly variable. Many submicron particles can be observed on trichomes
31      (leaf hairs), although most particles are in the 5 to 50 jim diameter range. Large aggregated

        June 2003                                  4-25         DRAFT-DO NOT QUOTE OR CITE

-------
 1      particles in excess of 100 jim also are seen, and carbonaceous aggregate particles are especially
 2      common (Smith, 1990a). Trichomes are efficient particle receptors; however, they are reduced
 3      in size by "weathering" and occasionally are completely broken off during the growing season.
 4           In the size range around 0.1 to 1.0 |im,  where Vd is relatively independent of particle
 5      diameter (Fig. 4-5), deposition is controlled by macroscopic roughness properties of the surface
 6      and by the stability and turbulence of the atmospheric surface layer. The resistance catena
 7      (Figure 4-3) is less useful in this size range and, in some treatments, has been abandoned entirely
 8      (e.g., Erisman et al.,  1994; Eq. 4-4). Impaction and interception dominate over diffusion, and the
 9      Vd is considerably (up to two orders of magnitude; Figure 4-2) lower than for particles either
10      smaller or larger (Shinn, 1978).  The deposition velocity may be parameterized in this size range
11      as a function of friction velocity,
12
13                                       Vd = (a/b)u*,                                    (4-4)
14
15      where a depends on atmospheric stability and b depends on surface roughness (Wesely et al.,
16      1985; Erisman et al., 1994). Similar formulations have been presented in terms of turbulence
17      (standard deviation of wind direction) and wind speed (e.g.,  Wesely et al., 1983), both
18      determinants of u*.
19           Deposition of particles between 1 and 10 jim diameter, including the coarse mode of PM10,
20      is strongly dependent on particle size (Shinn, 1978).  Larger particles within this size range are
21      collected more efficiently at typical wind speeds than are smaller particles (Clough, 1975),
22      suggesting the importance of impaction.  Impaction is related to wind speed, the square of
23      particle diameter, and the inverse of receptor diameter as a depositing particle fails to follow the
24      streamlines of the air in which it is suspended around the receptor. When particle trajectory
25      favors a collision, increasing wind speed or ratio of particle size to receptor cross section makes
26      collision nearly certain; and, as these parameters become very small, the probability of collision
27      becomes negligible.  However, the shape parameter for the more common range of situations
28      between these extremes remains poorly characterized (Peters and Eiden, 1992; Wiman and
29      Agren,  1985).
30           As particle size increases above 1 |im, deposition is governed increasingly by
31      sedimentation (Figure 4-5) and decreasingly  by turbulence and impaction.  Particles between

        June  2003                                 4-26       DRAFT-DO NOT QUOTE OR CITE

-------
 1      -10 and 24 jim (Gallagher et al., 1988) are both small enough to be transported efficiently by
 2      turbulent eddies to the surface and large enough to impact with sufficient momentum to
 3      overcome boundary layer effects. These particles deposit highly efficiently and relatively
 4      independently of particle size. Deposition of the largest suspended particles (e.g., > 50 jim) is
 5      governed, through sedimentation and the corresponding terminal settling velocity (Vs), almost
 6      entirely by size. These particles are not transported efficiently by small-scale eddies near the
 7      surface.
 8           Theoretically based models for predicting particle deposition velocities have been
 9      published by Bache (1979a,b), Davidson et al.  (1982), Noll and Fang (1989), Slinn (1982), and
10      Wiman (1985).  These models deal primarily with low canopies or individual elements of
11      canopy surfaces. Wiman and Agren (1985) have developed an aerosol deposition model that
12      specifically treats the problem of particle deposition to forests where turbulence plays a
13      particularly important role, especially at roughness transitions such as forest edges.  They found
14      that deposition of supermicron particles is controlled by complex interactions among particle
15      size and concentration, forest structure, and aerodynamics; whereas deposition of fine particles
16      (submicron) is controlled by particle concentration and forest structure.
17           At the present time, empirical  measurements of Vd for fine particles under wind tunnel and
18      field conditions are often several-fold greater than predicted by available theory (Unsworth and
19      Wilshaw, 1989). A large number of transport phenomena, including streamlining of foliar
20      obstacles, turbulence structure near  surfaces, and various phoretic transport mechanisms remain
21      poorly parameterized in current models. The discrepancy between measured and predicted
22      values of Vd may reflect such model limitations or experimental limitations in specification of
23      the effective size and number of receptor obstacles, as suggested by Slinn (1982).
24           Available reviews (Davidson and Wu,  1990; McMahon and Denison, 1979; Nicholson,
25      1988; Sehmel, 1980;  Slinn, 1982; U.S. Environmental Protection Agency, 1982,  1996a) suggest
26      the following generalizations: (1) particles larger than 10 jim exhibit a variable Vd between =0.5
27      and 1.1 cm s"1 depending on friction velocities; whereas a minimum particle Vd of 0.03 cm s"1
28      exists for particles in the size range 0.1 to 1.0 |im; (2) the Vd of particles is approximately a
29      linear function of friction velocity; and (3) deposition of particles from the atmosphere to a
30      forest canopy is from 2 to 16 times greater than deposition in adjacent open terrain  (i.e.,
31      grasslands or other vegetation of low stature).

        June 2003                                  4-27        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Leaf Surface Effects on Vd.  The term rc (Equation 4-3) reflects the chemical, physical, or
 2      physiological characteristics of the surface that govern its ability to capture, denature, or
 3      otherwise remove particulate material from the atmospheric surface layer.  For gases, relevant
 4      surface properties involve the physiological state of the vegetation, including stomatal opening
 5      and mesophyll antioxidant activity, and the chemical reactivity of the exposed surface with the
 6      specific gas. For particles, relevant surface properties involve stickiness, microscale roughness,
 7      and cross-sectional area. These properties determine the probability of impaction and bounce
 8      (e.g., Shinn, 1978).  The chemical composition of PM is not usually a primary determinant of
 9      deposition velocity.  At the microscopic scale where Van der Waals forces may determine
10      particle bounce and reentrainment, the chemical  properties of both surface and particle may be
11      significant but remain poorly understood.
12           Stickiness may itself depend on previous deposition of deliquescent particles that prolong
13      leaf wetness, on the wettability of foliar surfaces, and on the presence of sticky residues such as
14      honeydew deposited by aphids. These factors increase deposition by decreasing bounce-off,
15      wind reentrainment, and, to some extent, wash-off by precipitation.
16           The distribution of particles on and the efficiency of deposition to vegetation also varies
17      based on leaf shape and plant  part.  Particles are  more prevalent on the adaxial (upper facing of
18      the twig) surface than on the abaxial (lower away from the twig) surface.  Peripheral leaf areas
19      tend to be the cleanest most particles accumulate in the midvein, central portion of leaves.  The
20      rough area surrounding the stomatal pores was not found to be a preferential site for particle
21      deposition or retention (Smith and Staskawicz, 1977). Most particles were located near veins
22      with smaller particles localized on the trichomes. The greatest particulate loading on
23      dicotyledonous leaves is frequently on the adaxial (upper) surface at the base of the blade, just
24      above the petiole junction. It  is probable that precipitation washing plays an important role in
25      this distribution pattern.  Lead particles accumulated to a larger  extent on older than younger
26      needles and twigs of white pine, indicating that wind and rain were insufficient to fully wash the
27      foliage.  Fungal mycelia (derived from windborne spores) were  frequently observed in intimate
28      contact with other particles on other leaves (Smith and Staskawicz, 1977), which may reflect
29      shelter by the particles minimizing reentrainment of the  spores,  mycelia development near
30      sources of soluble nutrients provided by the particles, or codeposition. This pattern is significant
31      and could yield further insight into deposition mechanisms.

        June 2003                                 4-28         DRAFT-DO NOT QUOTE OR CITE

-------
 1           Leaves with complex shapes collect more particles than those with regular shapes. Conifer
 2      needles are more effective than broad leaves in accumulating particles.  The edge to area ratio
 3      (Woodcock,  1953) is also a key determinant of salt deposition to individual artificial leaves.
 4      A strong negative correlation was observed under wind tunnel conditions between the area of
 5      individual leaves and deposition of coarse particles (Little, 1977).  Small twigs and branches
 6      were more effective particle collectors than were large branches and trunks of trees (Smith,
 7      1984). Lead particles accumulated 20-fold more on woody stems than on leaves of white pine
 8      (Pinus strobus\ even though leaves displayed a 10-fold greater total area (Heichel and Hankin,
 9      1976). Deposition is heaviest at tips of individual leaves.
10           Rough, pubescent broadleaf discs collected coarse (5.0-|im) particles up to sevenfold more
11      efficiently than glabrous leaf discs (Little, 1977).  Laminae, petioles, and stems, all differed in
12      collection efficiency. Pubescent leaves of sunflower (Helianthus annuus) collected coarse
13      particles nearly an order of magnitude more efficiently than the glabrous leaves of tulip poplar
14      (Liriodendron tulipifera) under wind tunnel conditions (Wedding et al., 1975). Rough pubescent
15      leaves of nettle (Urtica dioica)  were more effective in capturing coarse PM10 than were the
16      densely tomentose leaves of poplar (Populus alba) or smooth leaves of beech (Fagus sylvatica).
17           Conifer needles are more  efficient than broad leaves in  collecting particles by impaction.
18      This reflects the small cross section of the needles relative to larger leaf laminae of broadleaves
19      and the greater penetration of wind into conifer than broadleaf canopies. Conifers were more
20      effective in removing coarse (=20 jim) particles of ragweed pollen from the atmosphere than
21      were broadleaf trees (Steubing and Klee,  1970) and in intercepting the even coarser particles of
22      rain (Smith, 1984). Conifers are also more effective in retaining and accumulating particles
23      against reentrainment by wind and removal by rain, particularly on  foliar surfaces where they are
24      likely to be most biologically active.  Seedlings of white pine (Pinus strobus) and oak (Quercus
25      rubrd) initially retained between a quarter (pine) and a third (oak) of very coarse (88 to 175 jim)
26      134Cs-tagged quartz particles applied under field conditions (Witherspoon and Taylor,  1971).
27      After 1 h, the pine retained over 20% of the 134Cs-tagged particles; whereas the oak retained only
28      approximately 3%. Long-term  retention of the particles was  concentrated at the base of the
29      fascicles in pine and near the surface roughness caused by the vascular system on leaves of oak.
30      The sheltered locations available in the conifer foliage contribute substantially to greater
        June 2003                                 4-29        DRAFT-DO NOT QUOTE OR CITE

-------
 1      retention of particles. For similar reasons, grasses also are efficient particle collectors (Smith
 2      and Staskawicz, 1977) with long-term retention mostly in the ligule and leaf sheath.
 3           Because of the strong relationship between particle size and deposition, the sharply
 4      increasing humidity gradient near transpiring foliar surfaces may cause hygroscopic particles to
 5      behave, at the immediate surface, as larger particles than reflected in ambient measurements at
 6      reference heights.  This needs to be further considered (Wesely and Hicks, 2000).  Recent
 7      deposition models (Ruijgrok et al., 1997; Zhang et al., 2001) have introduced the role of ambient
 8      humidity but lack sufficient emphasis on the role of vegetation itself in modifying humidity near
 9      the surface. This may be significant, as the size of a dry, 0.5-|im particle of ammonium sulfate
10      may increase  to about 3.5 jim at saturated humidity (Ruijgrok et al., 1997). Kinetic analyses
11      have suggested that full (95% to 100%) equilibration to the new diameter will occur during the
12      deposition process.
13           Wind tunnel studies also demonstrated equivalent deposition  properties of 3.36-|im
14      particles of dense lead chloride and 6.77-|im particles of less dense uranine dye. These particles
15      were shown to be aerodynamically equivalent, substantiating the use of aerodynamic diameter as
16      a classification parameter for particle deposition.
17
18           Canopy Surface Effects on Vd.  In general, surface roughness contributes to greater
19      particulate deposition. As a result, Vd is typically greater for a forest than for a field or
20      nonwoody wetland and greater for a field than for a water surface.  The contrasting transport
21      properties and deposition velocities of different size particles lead to predictable patterns of
22      deposition. For coarse particles, the upwind leading edges of forests, hedge rows, and individual
23      plants, as well as of individual leaves, are primary sites of deposition.  Impaction at high wind
24      speed and the sedimentation that follows the reduction in wind speed and carrying capacity of
25      the air in these areas lead to preferential deposition of larger particles.
26           Air movement is slowed in proximity to vegetated surfaces. Resulting log profiles of wind
27      and pollutant  concentrations in the near-surface turbulent boundary layer above canopies reflect
28      surface characteristics of roughness length, friction velocity, and displacement height.  Plasticity,
29      streamlining,  and oscillations of foliar elements also alter the aerodynamic roughness and the
30      level of within-canopy turbulence. Canopies of uneven age or with a diversity of species are
31      typically aerodynamically rougher and receive larger inputs of pollutants than do smooth, low, or

        June 2003                                 4-30        DRAFT-DO NOT QUOTE OR CITE

-------
 1      monoculture vegetation (Garner et al., 1989; Wiman and Agren, 1985).  Canopies on slopes
 2      facing the prevailing winds and individual plants on the windward edges of discontinuities in
 3      vegetative cover over which roughness increases receive larger inputs of pollutants than more
 4      sheltered, interior canopy regions. For example, some 80% of coarse particulate sea salt was
 5      deposited on the upwind edge of a hedgerow (Edwards and Claxton, 1964), and the
 6      concentration of ragweed (Ambrosia spp.} pollen was reduced by 80% within 100 m of the
 7      leading edge of a forest (Neuberger et al., 1967).
 8           Beier et al. (1992) and Beier (1991) discussed two methods for estimating the dry
 9      deposition of base cations to forest edges: (1) a difference method between measured
10      precipitation and throughfall concentrations of base cations, and (2) a calculation method based
11      on known ratios of Na+ deposition in wet and dry forms (Ulrich, 1983).  A combination of these
12      two approaches produced the best estimates of SO4"2, Ca+2, Mg+2, and K+ particle deposition.
13      Using these methods, particulate SO4"2 (Beier, 1991) and particulate Ca+2, Mg+2, and K+ (Beier
14      et al., 1992) were found to decrease by an order of magnitude from the forest edge to the forest
15      interior. A number of authors also have shown that particle deposition is elevated at forest edges
16      when compared to a uniform forest canopy (Draaijers et al., 1988; Grennfelt, 1987; Lindberg and
17      Owens, 1993),  and Draaijers et al. (1992) reported that differences are likely to exist between
18      forest types because of variable canopy structure. Draaijers et al. (1988) further emphasized that
19      enhanced particle deposition at or near forest edges is strongly dependent on the velocity and
20      wind direction  during observations.
21           The factors leading to horizontal gradients are confounded by time- and distance-related
22      sedimentation.  For example, geologic dust (mostly around 7 jim aerodynamic diameter) collects
23      on stems of wild oats (Avena spp.; Davidson and Friedlander, 1978) and on eastern white pine
24      (Pinus strobus; Heichel and Hankin, 1972; Smith 1973) downwind of roadways. Rapid
25      sedimentation of coarse crustal particles  suggests that potential direct effects may be restricted to
26      roadway margins, forest edges, and, because of the density of unpaved roads in agricultural
27      areas, crop plants.
28           Simulated deposition to an ecologically complex, mixed canopy was considerably higher
29      than to a pure spruce stand in which most of the leaf area was concentrated in regions of low
30      wind speed. Limitations to the application of these models to predict deposition over large
31      regions include a limited understanding both of the nature  of microscopic particle-surface

        June 2003                                  4-31        DRAFT-DO NOT  QUOTE OR CITE

-------
 1      interactions and of the effects of complex terrain and species composition on macroscopic
 2      transport processes.
 3           Macroscopic turbulent transport processes, related to ra at successive layers through the
 4      canopy can be separated from microscopic processes, related to rb and rc (or rcp) at each
 5      deposition surface (e.g., Peters and Eiden, 1992; Wiman and Lannefors, 1985).  The
 6      macroscopic approach deals with deposition as the product of a turbulent diffusion coefficient
 7      and a concentration (Cz) at each canopy layer, both of which vary with particle size and with
 8      height (Z) in the canopy.  The microscale parameters involve those factors that determine
 9      absorption of a particle at each surface as captured imperfectly by rc. Shelter effects caused by
10      the crowding of foliar elements within the canopy can be ignored if the wind speed within each
11      canopy layer is specified. This approach requires knowledge of the vertical distribution of
12      particle concentration and foliage density in the canopy airspace along with profiles of wind
13      speed or turbulence.
14           Once introduced into a forest canopy, elements associated with course particles tend to
15      decrease markedly with canopy depth; whereas  elements associated with fine particles do not
16      (Lovett and Lindberg, 1992). Trace elements and alkaline-earth elements are enriched below the
17      canopies of both southern (Lindberg et al., 1986) and northern (Eaton et al., 1973) hardwood
18      forests. Vertical gradients in concentration of coarse particles and of elements associated with
19      coarse particles were observed in a mixed conifer/birch forest canopy (Wiman and Lannefors,
20      1985; Wiman et al.,  1985) and in a mixed oak forest (e.g., Ca+2, Figure 4-6A; Lovett and
21      Lindberg, 1992). The highly reactive gas HNO3 also exhibited a vertical gradient but with a
22      steep decline at the top of the canopy (Figure 4-6B).  Lovett and Lindberg (1992) studied
23      concentration profiles of various gases and particles within an closed canopy forest and
24      concluded that coarse particle concentrations associated with elements like Ca+2 would decrease
25      markedly with depth in the canopy, but they found only minor reductions with depth in the
26      concentrations of fine aerosols containing SO4"2, NH4+, and FT.  These data suggest that all foliar
27      surfaces within a forest canopy are not exposed equally to particle deposition: upper canopy
28      foliage would receive maximum exposure to coarse and fine particles, but foliage within the
29      canopy would receive primarily fine aerosol exposures. Fine-mode particles (e.g.,  sulfate,
30      Figure 4-6C) and unreactive gases typically do not exhibit such vertical profiles, suggesting that
31      uptake is smaller in magnitude and more evenly distributed throughout the canopy. In multilayer

        June 2003                                 4-32        DRAFT-DO NOT QUOTE OR CITE

-------
                     D)
                    '0
                    I
                     D)
                    "0
                    I
25
20
15
10
 5
 0

25
20
15
10
 5
 0
                                            A = 34%
                              i      i      i      i
                            0.6   0.7    0.8    0.9
         A = 43%
                  \
                  5
                                              so;2
                                            A = 6%
                                                       D
           K*
         A = 0%
                                9.5
                      10.0       0.08    0.10    0.12    0.14
                   Concentration (|jg/m3)
      Figure 4-6.  Vertical stratification of diverse, chemically speciated particles in a mixed oak
                   forest.  Nitric acid vapor, as a highly reactive, nonparticulate species, is shown
                   for comparison. The horizontal dotted line indicates canopy height, and A
                   indicates the percent depletion from above to below the canopy. Data from
                   Table 2-4 of Lovett and Lindberg (1992).
1      canopies, simultaneous reentrainment and deposition may effectively uncouple deposition from
2      local concentration. Poly disperse size distributions of many chemical species effectively prevent
3      the use of a single estimate of Vd for any element if highly accurate results are required.
4          Although gradients (Figures 4-6 and 4-7) may be related to local Vd within the canopy
5      (Bennett and Hill, 1975), the absence of a gradient may reflect either low rates of deposition or
6      very high rates relative to turbulent replenishment from above the canopy (Tanner,  1981).
7      Below- or within-canopy emissions may confound interpretation of vertical gradients.  Linear
8      gradients of the gaseous pollutants hydrogen fluoride (HF) and ozone (O3) reflected large uptake
      June 2003
                        4-33
DRAFT-DO NOT QUOTE OR CITE

-------
c
.g
-t— '
03
•£
 04-
^
O3
o
o^
"" 0.2-


n
.IIIIIIIIMMIIIIH 	 Illlllllllllll///,.
^ \
•^ "Zr
i \
= ^
~ s
~ ^
= E
E E
= =
1 §
=
|
%
?•
•^.
"^.
%//(mX



                                          0.01              1
                                              Particle Diameter (|jm)
                         100
       Figure 4-7. The relationship between particle size and concentration below a spruce
                   canopy with wind velocity at a height of 16.8 m equaling 5 m s"1. Modified
                   from Peters and Eiden (1992).
 1     rates; whereas small gradients in NO suggested little uptake by foliage (Bennett and Hill, 1973,
 2     1975). However, soil efflux of NO could have complicated the latter interpretation.  The lack of
 3     a vertical gradient and a peak near the top of the active canopy in particulate K+ (Figure 4-6D)
 4     was interpreted as evidence for a biogenic source within the deciduous forest canopy with
 5     moderate rates of deposition (Lindberg et al.,  1986; Lovett and Lindberg, 1992).
 6           The size dependence of this vertical stratification of parti culate concentration (see
 7     Figure 4-7) is reflected in current simulation models (Wiman et al., 1985; Peters and Eiden,
 8     1992). The model of Wiman and Agren (1985) predicts a uniform vertical distribution of
 9     fine-mode particles and a pronounced vertical gradient of coarse-mode particles which is in
10     agreement with observations (Lovett and Lindberg, 1992).
11           Simulation of the horizontal deposition patterns at the windward edge of a spruce forest
12     downwind of an open field with the canopy between 1 and 25 m above the ground indicated that
13     deposition was maximal at the forest edge where wind speed and impaction were greatest.
14     Simulation of the vertical deposition pattern was more complex.  Deposition was not greatest at
       June 2003
4-34
DRAFT-DO NOT QUOTE OR CITE

-------
 1      the top of the canopy where wind speed was highest, but at z = 0.75 h where the balance between
 2      leaf area (obstacles for impaction) and wind speed (momentum for impaction) was optimal,
 3      although neither parameter alone was maximal. Simulated deposition in this spruce forest
 4      increased considerably with increasing LAI at the forest edge, where wind speed was insensitive
 5      to LAI but the number of obstacles increased. Inside the forest, where both wind speed and
 6      impaction increasingly were attenuated by increasing LAI, deposition increased only marginally
 7      in spite of the increase in obstacle frequency.
 8           To scale surface-specific measurements of particle deposition to forest or crop canopies,
 9      conversions of the following type have been suggested:
10
11                               Vdcanopy = VAsurface *  scaling factor,                            (4-5)
12
13      with empirical scaling factors proposed by Lindberg et al. (1988).
14           To appropriately scale surface-specific measurements of particle deposition to landscapes,
15      one must consider the complexity of grassland, crop, and forest canopies in order to avoid
16      serious over- or under-estimates of particle deposition. Individual species exposed to similar
17      ambient concentrations may receive a range of particulate loading that is more closely related to
18      foliar damage than the ambient concentration (Vora and Bhatnagar, 1987).
19           Both uptake and release of specific constituents of PM may occur within a single canopy
20      (e.g., K+;  Lovett and Lindberg,  1992). The leaf cuticular surface is a region of dynamic
21      exchange processes through leaching and uptake. Exchange occurs with epiphytic
22      microorganisms and bark and through solubilization and erosion of previously deposited PM.
23      Vegetation emits a variety of parti culate and parti culate precursor materials. Terpenes and
24      isoprenoids predominate and, on oxidation, become condensation nuclei for heterogeneous
25      particle formation.  Salts and exudates on leaves and other plant parts continually are  abraded
26      and suspended as particles, as are plant constituents from living and dead foliage (Rogge et al.,
27      1993a). Soil minerals, including radioactive strontium, nutrient cations and anions, and trace
28      metals are transferred to the active upper foliage and then to the atmosphere in this way.
29      Although not representing a net addition to an ecosystem, particle release from vegetation is a
30      mechanism for redistributing chemical pollutants derived from the soil or prior deposition within
31      a canopy, potentially enhancing direct effects and confounding estimates of Vd.

        June 2003                                4-3 5         DRAFT-DO NOT QUOTE OR CITE

-------
 1      Range of Deposition Velocity
 2           As noted in an earlier criteria document (U.S. Environmental Protection Agency, 1982)
 3      and in McMahon and Denison (1979), estimates of Vd for PM10 particles to vegetation are
 4      variable and suggest a minimum between 0.1 and 1.0 jim as predicted from first principles
 5      (Monteith and Unsworth, 1990; Sehmel, 1980).  Determinations in wind tunnels with passive
 6      collectors and micrometeorological methods tend to converge in this range.  The range of Vd for
 7      sulfate from passive collectors was found to be from 0.147 to 0.356 cm s"1; and, from eddy
 8      covariance techniques, a mean Vd of 0.27 cm s"1  was observed (Dolske and Gatz, 1984).
 9      Micrometeorological techniques over grass (Wesely et al., 1985); indirect, inert collector
10      techniques within an oak forest (Lindberg and Lovett, 1985); and many  other empirical
11      determinations (e.g., McMahon and Denison, 1979; Table 4-4) generally support this range.
12      Over aerodynamically smooth snow (Duan et al., 1988; Table 4-5), measurements of Vd were an
13      order of magnitude smaller.  Very coarse particles, often non-size-specified primary geologic
14      material, frequently exhibit Vd greater than 1.0 cm s"1 (e.g., Clough, 1975).  The increase in Vd
15      with decreasing size below 0.1 |im is probably hidden in most empirical determinations of Vd
16      because the total mass in this fraction is very small despite the large number of individual
17      particles. Table 4-6 shows published estimates of Vd with variability estimates for fine particles
18      of specified aerodynamic diameters dominated by a range of chemical species.
19           Ibrahim et al. (1983) evaluated the deposition of ammonium sulfate particles to a range of
20      surfaces and found that particles having a mean diameter of 0.7 jim had  deposition velocities
21      ranging from 0.039 to 0.096 cm s"1. Larger particles (having mean diameters of 7 |im) had
22      greater deposition velocities (between 0.096 and 0.16 cm s"1). The authors further concluded that
23      the hygroscopic nature of the sulfate particle could increase its size and  enhance deposition near
24      sources of water (e.g., snow). Using eddy correlation approaches, Hicks et al.  (1989) found a
25      mean daily Vd for sulfur-containing PM to be 0.6 cm s"1. However, they suggested that the Vd
26      value could be as high as 1 cm s"1 during the day and near zero at night.
27           Lindberg et al. 1990 found a wide discrepancy between deposition velocities for NO3"
28      between study sites in Oak Ridge, TN (~ 2 cm s"1) and Gottingen, Germany (~  0.4 cm s"1). They
29      suggest that the increased Vd at Oak Ridge could be explained by the primary occurrence of NO3"
30      in coarse particles that exhibit greater Vd than fine particles (Davidson et al., 1982).  Large
        June 2003                                4-36        DRAFT-DO NOT QUOTE OR CITE

-------
  TABLE 4-4.  REPORTED MEAN DEPOSITION VELOCITIES (Vd) FOR SULFATE,
  CHLORINE, NITRATE, AND AMMONIUM AND ION-CONTAINING PARTICLES
Chemical Species/ Surface
SO42
Inert plates
Inert plates
Inert plates
Inert bucket
Foliage
Chaparral
Grass canopy
Grass canopy
Pine foliage
Plant canopies
Grass canopy
cr
Inert bucket
Inert bucket
Beech canopy
Spruce canopy
NO3
Inert plates
Ceanothus
Glycine max
Ligustmm
Quercus
Quercus
summer
winter
Quercus
summer
winter
Pinus
Pasture
Ulmus
NH4+
Calluna/Molina
Ceanothus
Kalmia
Pinus
V^cms-1)"

0.13
=0.14
0.14
0.51
0.29
0.15
0.10
0.07
0.07
0.50
0.22

3.1
5
1
1.9

0.4-2
0.4
0.24
0.1-0.5
0.7-1.1

0.55
0.7

0.3
0.1
0.5-1.3
0.7-0.8
1.1

0.18
0.4
0.03-0.14
0.01-0.06
Method

Extraction
Extraction
Extraction
Extraction
Extraction
Extraction
Gradient
Gradient
Extraction
Gradient
Eddy covariance

Extraction
Extraction
Throughfall
Throughfall

Extraction
Extraction
Extraction
Extraction
Extraction

Throughfall
Throughfall

—
—
Extraction
Gradient
Extraction

Gradient
Extraction
Extraction
Extraction
Reference

Lindberg and Lovett (1985)
Lindbergetal. (1990)
Davidson and Wu (1990)a
Davidson and Wu (1990)a
Davidson and Wu (1990)a
Bytnerowicz et al. (1987a)
Allen etal. (1991)
Nicholson and Davies (1987)
Wiman(1981)
Davidson and Wu (1990)a
Weseley etal. (1985)

Dasch and Cadle (1985)
Dasch and Cadle (1986)
Hofken etal. (1983)
Hofken etal. (1983)

Lindbergetal. (1990)
Bytnerowicz et al. (1987a)
Dolske (1988)
John etal. (1985)
Dasch (1987)

Lovett and Lindberg (1984)
Lovett and Lindberg (1984)

Lovett and Lindberg (1986)
Lovett and Lindberg (1986)
Dasch (1987)
Huebert etal. (1988)
Dasch (1987)

Duyzer etal. (1987)
Bytnerowicz et al. (1987a)
Tjepkema etal. (1981)
Dasch (1987)
 "These data represent the mean of data by measurement technique as reported in the cited reference. The reader is referred to the
 referenced articles for information on the specific cations contributing to the means.
June 2003
4-37
DRAFT-DO NOT QUOTE OR CITE

-------
 TABLE 4-5. REPRESENTATIVE EMPIRICAL MEASUREMENTS OF DEPOSITION
             VELOCITY (Vd) FOR PARTICIPATE DEPOSITION
vd
x ± SE (cm s'1)
0.034 ±0.014
0.021 ±0.005
0.1 ±0.03
0.22 ± 0.06
0.13 ±0.02
0.75 ±0.24
l.liO.l
0.9
2.5
9.4
Particle Size
fam)
0.15-0.30
0.5 - 1.0
0-2
0.1 -2.0
(SO/2)
(K+)
(Ca+2)
2.75
5.0
8.5
Method
Eddy covariance with optical counter, flat snow
surface
Profile, fine SO4"2, short grass
Eddy covariance with flame photometer plus
denuder, 40-cm grass, fine SO4"2
Inert surface collectors (petri dish) in oak forest
Wind tunnel to pine shoots; polystyrene beads;
within-canopy wind speed, 2.5 m s"1
Reference
Duanetal. (1988)
Allen etal. (1991)
Weselyetal. (1985)
Lindberg and Lovett
(1985)
Chamberlain and
Little (1981)
          TABLE 4-6. REPORTED MEAN DEPOSITION VELOCITIES
          FOR POTASSIUM, SODIUM, CALCIUM, AND MAGNESIUM
                 BASE CATION CONTAINING PARTICLES
Chemical Species/Surface
K+
Inert plates
Inert bucket
Na+
Inert bucket
Inert plate
Ca+2
Inert plates
Inert plates
Inert bucket
Inert bucket
Vd (cm s-1) Method

0.75 Extraction
0.51 - 2.4 Extraction

1.7-2.9 Extraction
0.8 - 8.2 Extraction

1.1 Extraction
~2 Extraction
1.7-3.2 Extraction
1.1-2.7 Extraction
Reference

Lindberg and Lovett (1985)
Dasch and Cadle (1985)

Dasch and Cadle (1985)


Lindberg and Lovett (1985)
Lindberg etal. (1990)
McDonald etal. (1982)
Dasch and Cadle (1985)
June 2003
4-38
DRAFT-DO NOT QUOTE OR CITE

-------
 1      values of Vd for base-cation-containing particles (> 1 cm s"1) suggest their occurrence in coarse
 2      particles (Lindberg and Lovett, 1985).
 3           The several attempts to estimate Vd for SO4"2, NO3", and NH4+ with the throughfall mass
 4      balance approach (Davidson and Wu, 1990; Gravenhorst et al., 1983; Hofken and Gravenhorst,
 5      1982) have produced higher Vd values that are considered suspect.  They have not been included
 6      in Tables 4-4 and 4-6.  Overestimates of Vd for SO4"2 and NO3" particles derived from throughfall
 7      mass balance approaches may be the result of gaseous SO2 and HNO3 gaseous deposition to
 8      foliar surfaces (Lindberg and Lovett, 1985).  A similar contribution of NH3 deposition may lead
 9      to erroneously high Vd values for NH4+ when the throughfall method is attempted in areas of
10      high NH3 concentrations. Dolske's (1988) reported Vd values for NO3" deposition to soybean
11      ranged from 0.4 to 31 with a mean of 0.24 cm s"1. However, because Dolske's leaf extraction
12      measurements included a component of HNO3 vapor, the Vd values may represent more than
13      deposition caused by aerosol nitrate alone.
14           The quantitative importance of dry particulate deposition depends upon the chemical
15      species, topography, precipitation regime, and surface characteristics, including vegetation
16      properties. Across the diverse landscapes of the Integrated Forest Study (Johnson and Lindberg,
17      1992a), the relative contribution of dry deposition for Ca+2 ranged from about 0% to nearly 90%
18      (Figure 4-8). In contrast, for S the total range was from just over 0% to about 30%.  An average
19      for these forest systems demonstrates that deposition of (usually coarse) base cations was nearly
20      50% by dry parti culate deposition (Figure 4-9). Both N and S were around 15%. While the
21      relative significance of dry particle deposition varies from site to site, it cannot be excluded from
22      the analysis at any site.
23           In a recent review, Wesely  and Hicks (2000) concluded that  a comprehensive
24      understanding of parti culate deposition remains a distant goal.  In general, there is only modest
25      confidence in available particulate deposition parameterizations at this time, although recent
26      experimental and theoretical efforts to improve this situation have been made (e.g., Erisman
27      et al., 1997, and companion articles; Zhang et al., 2001; Kim et al., 2000).
28           The successful treatment of dry deposition of gaseous pollutants (e.g.,  SO2 and O3) linking
29      turbulent deposition to surface physiological properties has allowed incorporation of these
30      models into landscape and regional scale models. This has allowed gaseous deposition to be
31      scaled up for purposes of atmospheric chemistry and vegetation damage assessment.  These

        June 2003                                 4-39       DRAFT-DO NOT QUOTE OR CITE

-------
IUU
on
OU
-t-*
C Gf\
 <®
Q"S 9Km

CO ^S *w
3 ''^ ' 15"
o
s 1 n -
i_ IU"
(O
a 5-
0_










--• •







^•ww^t;
"•*" h?C^'™
{.'JiT'^"?5""
fe^.l










	 -- •
1







PS

w» ;" ' *****











I
|"'«"^;..',;
J*.-.lJt.,.-
>ll~ • ,r
,-^r'>
.:i- «"--;•'
i "-".--" '•
IT.it .,!>«'
V 1 / ' v












	

                                      N
               Base Cations
      Figure 4-9.  Contribution of particulate deposition to total deposition of nitrogen, sulfur,
                  and base cations (after Johnson and Lindberg, 1992a; Lovett, 1992; Lovett
                  and Lindberg, 1993; Lindberg et al., 1990; Kelly and Meagher, 1986).
1     advances have not been matched by progress in deposition of PM. A serious remaining

2     impediment is the lack of suitable techniques for measuring deposition of heterogeneous and

3     polydisperse particles, such as the gradient and eddy covariance techniques that are used for
      June 2003
4-40
DRAFT-DO NOT QUOTE OR CITE

-------
 1      gaseous pollutant species.  As with gaseous pollutants, parameterization of particle deposition in
 2      hilly terrain, to patchy surfaces (small agricultural fields, forest edges), and under extremely
 3      windy conditions, remains to be fully developed.  These limitations must be addressed before a
 4      full accounting of regional PM effects on vegetation in natural and managed ecosystems can be
 5      achieved.
 6
 7      Occult Deposition
 8           Gaseous pollutant species may dissolve in the suspended water droplets of fog and clouds.
 9      The stability of the atmosphere and persistence of the droplets often allow a gas/liquid phase
10      equilibrium to develop. This permits the use of air mass history or ambient concentrations of
11      specific pollutants to estimate fog or cloud water concentrations. Further estimates of the
12      deposition velocity of the polluted droplets allows calculation of deposit!onal fluxes.
13      Unfortunately, interception of fog or cloud droplets by plant parts or other receptor surfaces
14      remains difficult to predict and to measure. Fog formation influences the total atmospheric
15      burden and deposition of particulate matter (Pandis and Seinfeld, 1989) by accreting and
16      removing particles from the air, by  facilitating particle growth through aqueous  oxidation
17      reactions, and by enhancing deposition as noted.  Aqueous condensation may occur onto
18      preexisting fine particles, and such  particles may coalesce or dissolve in fog or cloud droplets.
19      Material transported in fog and cloud water and intercepted by vegetation escapes detection by
20      measurement techniques designed to quantify either dry or wet deposition; hence it is hidden
21      (i.e., "occult") from the traditional measurements.
22           Occult deposition of fog and cloud droplets  is by impaction and gravitational settling in
23      concert with the instantaneous particle (droplet) size.  This mode of deposition may be
24      significant at high elevation sites, particularly near the base of orographic clouds.  While coastal
25      fog may be generally more pristine than high-elevation continental fogs, this is not true in areas
26      subject to inland radiation fog (e.g., the increasingly polluted San Joaquin Valley of Central
27      California) and in areas where the marine layer is advected through a highly polluted urban area
28      (e.g., the Los Angeles basin of Southern California). Fog water in these areas may be at least as
29      contaminated as that occurring at higher elevations.
30           Low elevation radiation fog has different formation and deposition characteristics than
31      high elevation cloud or coastal fog water droplets. A one-dimensional deposition model has

        June 2003                                 4-41        DRAFT-DO NOT QUOTE OR CITE

-------
 1      recently been described for a radiation fog episode (Von Glasow and Bott, 1999).
 2      A substantially greater concentration of key polluting species (e.g., NO3", SO4"2, organics) may
 3      be observed in smaller than in larger droplets in fog (Collett et al., 1999). Acidity differences
 4      exceeding 1 pH unit were also observed in the San Joaquin Valley winter radiation fog, with
 5      smaller particles being more acidic. This has implications for aqueous phase oxidation of sulfur
 6      and nitrogen compounds, in particular, while sulfur oxidation by ozone (the dominant reaction in
 7      this environment even during winter) is well known in typically acidic fog droplets.  However,
 8      the alkaline larger droplets in the San Joaquin Valley could lead to greater nitrate production
 9      through aqueous ozonation reactions (Collett et al., 1999).  The size class distinctions have
10      substantial implications for deposition of particulate pollutant species in the fog droplets due to
11      the larger Vd for impaction and occult deposition of the larger fog particles.
12           Acidic cloud water deposition has been associated with forest decline in industrialized
13      areas of the world (Anderson et al., 1999). Clouds can contain high concentrations of acids and
14      other ions. The four most prevalent ions found in cloud water samples, in decreasing order of
15      concentration, were usually sulfate (SO4"2), hydrogen (H+), ammonium (NH4+), and nitrate
16      (NO3").  The concentrations of these major ions tend to co-vary within cloud events, and
17      typically there was an inverse relationship between liquid water content (LWC) of the cloud and
18      ionic concentration of the cloud water.  Cloud water typically is 5 to 20 times more acid than
19      rain water. This can increase pollutant deposition and exposure of vegetation and soils at
20      high-elevation  sites by more than 50% when compared with rainfall and dry deposition.
21           The widespread injury to mountain forests documented in West Germany since the 1970s
22      and other parts of Europe and more recently  in the Appalachian Mountains has been attributed to
23      exposure to the cloud water reducing cold tolerance of red spruce. Forest injury also has been
24      attributed to increased leaching of cations and amino acids and increased deposition  of nitrogen
25      and aluminum toxicity resulting from acidic  deposition and the combined effect of acidic
26      precipitation, acid fog, oxidants, and heavy metals (Anderson et al.,  1999). The Mountain Acid
27      Deposition Program (MADPro) was initiated in 1993 as part of the Clean Air Status  and Trends
28      Network (CASTnet). MADPro monitoring efforts focused on the design and implementation of
29      an automated cloud water collection system in combination with continuous measurement of
30      cloud liquid water content (LWC) and meteorological parameters relevant to the cloud
31      deposition process.

        June 2003                                4-42        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Results from the MADPro automated cloud-water collectors at three selected mountain
 2      sites (Whiteface Mt, NY; Whitetop Mt, Va; and Clingman's Dome, TN) taken hourly from
 3      nonprecipitating clouds during non-freezing seasons of the year from 1994 to 1997 were
 4      promptly analyzed for pH, conductivity, and concentration of dissolved ions give an indication
 5      of exposures at the three sites. Cloud LWC was measured at each site. The mean cloud water
 6      frequencies and LWC were higher at Whiteface Mountain, NY, that in the southern
 7      Appalachians.  The four most prevalent ions found in cloud water samples, in order of
 8      decreasing concentrations, usually were sulfate (SO4"2), hydrogen (H+), ammonium (NH4+), and
 9      nitrate (NO3").  The concentrations of these ions tended to co-vary within cloud events and
10      typically there was an inverse relationship between LWC of the cloud and ionic concentration of
11      the cloud water. Highest ionic concentrations were seen in mid-summer. Ionic concentrations
12      of samples from southern sites were significantly higher than samples from Whiteface Mountain,
13      but further analysis indicated that this was due at least in part to North to South differences in the
14      LWC of clouds (Anderson et al., 1999).
15           Several factors make occult deposition particularly effective for delivery of dissolved and
16      suspended materials to vegetation. Concentrations of particulate-derived materials are often
17      many-fold higher in cloud or fog water than in precipitation or ambient air in the same area due
18      to orographic effects and from gas-liquid partitioning coefficients of specific chemical species.
19      Fog and cloud water deliver PM in a hydrated and, therefore, bioavailable form to foliar
20      surfaces. Previously dry-deposited PM may also become hydrated through delinquence or by
21      dissolution in the film of liquid water from fog deposition. The presence of fog itself maintains
22      conditions of high relative humidity and low radiation, thus reducing evaporation and
23      contributing to the persistence of these hydrated particles on leaf surfaces. Deposition of fog
24      water is very efficient (Fowler et al., 1991) with a Vd (fog 10-24 |im; Gallagher et al., 1988)
25      essentially equal to the aerodynamic conductance for momentum transfer (rj"1.  This greatly
26      enhances deposition by sedimentation and impaction of submicron aerosol particles that exhibit
27      very low Vd prior to fog droplet formation (Fowler et al., 1989). The near equivalence of Vd and
28      (ra)4 simplifies calculation of fog water deposition and reflects the absence of vegetative
29      physiological control over surface resistance.  Fog particles outside this size range may exhibit
30      Vd below (rj"1. For  smaller particles, this decline reflects the increasing influence of still air and
31      boundary layer effects on impaction as particle size and momentum decline.  For larger particles,

        June 2003                                 4-43        DRAFT-DO NOT QUOTE OR CITE

-------
 1      momentum is sufficient to overcome these near surface limitations, but Vd may decline as
 2      turbulent eddy transport to the surface becomes inefficient with increasing inertia (Gallagher
 3      et al., 1988). The deposition to vegetation for PM in fog droplets is directly proportional to wind
 4      speed, droplet size, concentration, and fog density (liquid water content per volume air) although
 5      the latter two may be inversely related. In some areas, typically along foggy coastlines or at high
 6      elevations, occult deposition represents a substantial fraction of total deposition to foliar surfaces
 7      (Fowler et al., 1991, Figure 4-2).
 8
 9      4.2.2.3  Magnitude of Deposition
10           Dry deposition of PM is most effective for coarse particles including primary geologic
11      material and for elements such as iron and manganese. Wet deposition is most effective for fine
12      particles of atmospheric (secondary) origin (e.g., nitrogen and sulfur, Table 4-6) and elements
13      such as cadmium, chromium, lead, nickel, and vanadium (Reisinger, 1990; Smith, 1990a,b,c;
14      Wiman and Lannefors,  1985). The occurrence of occult deposition is more restricted. The
15      relative magnitudes of the different deposition modes varies with ecosystem type, location,
16      elevation, and chemical burden of the atmosphere. For the Walker Branch Watershed, a
17      deciduous forest in rural eastern Tennessee, dry deposition constituted a major fraction of the
18      total annual atmospheric input of cadmium and zinc (=20%), lead (=55%), and manganese
19      (=90%). Whereas wet deposition fluxes during precipitation events exceeded dry deposition
20      fluxes by one to four orders  of magnitude (Lindberg and Harriss, 1981), dry deposition was
21      nearly continuous.  Immersion of high-elevation forests in cloud water may occur for 10% or
22      more of the year, significantly enhancing transfer of PM and dissolved gases to the canopy.
23      Occult deposition in the Hawaiian Islands dominated total inputs of inorganic N (Heath and
24      Huebert, 1999). Much of this N was volcanically derived during the generation of volcanic fog
25      in part through reactions with seawater.  In this humid climate, the dominance of occult rather
26      than wet deposition is notable.
27           High-elevation forests receive larger particulate deposition loadings than equivalent low
28      elevation sites. Higher wind speeds enhance the rate of aerosol impaction. Orographic  effects
29      enhance rainfall intensity and composition and increase the duration of occult deposition.
30      Coniferous species in these areas with needle-shaped leaves also enhance impaction and
31      retention of PM delivered by all three deposition modes (Lovett, 1984).

        June 2003                                 4-44        DRAFT-DO NOT QUOTE OR CITE

-------
 1           In more arid regions, such as the western United States, the importance of dry deposition
 2      may be larger. In the San Gabriel Mountains of southern California, for example, while annual
 3      deposition of SO4"2 (partly of marine origin) was dominated by wet deposition (Fenn and Kiefer,
 4      1999), deposition of NO3" was dominated by dry deposition,  as was that of NH4+ at two of three
 5      sites. Similarly, at a series of low elevation sites in southern California (Padgett et al.,  1999),
 6      dry deposition of NO3" was dominated by dry deposition. In both cases, however, the
 7      contribution of gaseous HNO3 was probably substantial.
 8
 9      Nitrates, Sulfates and Cations
10           Much particulate sulfate and nitrate is found on particles in the 0.1- to 1.0-|im size range
11      (U.S. Environmental Protection Agency, 1982).  However, most sulfate and nitrate, base cation
12      and heavy metal inputs to forested ecosystems results from the deposition of larger particles
13      (Chapter 2; Lindberg and Lovett, 1985; Lindberg et al., 1982).  The influence of aerodynamic
14      diameter is particularly critical for nitrogen species, because they exist as a wide range of
15      particle sizes in the atmosphere (Milford and Davidson,  1987). For example, at many sites in
16      North America, NO3" is characterized by a bimodal size distribution with modes above and
17      below 1 |im.  The supermicron particles are often the result of reactions between HNO3 and
18      coarse alkaline aerosols (Wolff, 1984) as, for example, in the San Joaquin Valley of California
19      (Lindberg et al., 1990). Although the annual deposition of NH4+ is distributed similarly among
20      the fine and coarse particles, particulate NO3" is found predominantly in the coarse-particle
21      fraction (Table 4-7).  Similar to the pattern for NH4+, the estimated annual deposition of SO4"2
22      particles occurs in both the fine- and coarse-particulate fractions (Table 4-8), while base cation
23      deposition is virtually restricted to contributions from coarse particles (Table 4-9).
24           Although the annual chemical inputs to ecosystems from particle deposition is  significant
25      by itself, it is important to compare it with the total chemical inputs from all sources of
26      atmospheric deposition (i.e., precipitation, particles, and gaseous dry  deposition).  Figure 4-10
27      shows the mean percentage contribution of NO3" andNH4+, SO4"2, and base cation-containing
28      particles to the total nitrogen, sulfur, and base cation deposition load to forest ecosystems
29      (derived from Tables 4-7 through 4-9).  Although the mean contribution of particulate deposition
30      to cumulative nitrogen and sulfur deposition is typically less than 20% of annual inputs from all
        June 2003                                 4-45        DRAFT-DO NOT QUOTE OR CITE

-------

to
o
o
OJ
     TABLE 4-7. MEAN ANNUAL NITROGEN DEPOSITION (equivalents/ha/yr) FROM FINE AND

COARSE PARTICLES COMPARED TO TOTAL NITROGEN DEPOSITION FROM ALL SOURCES TO A

                          VARIETY OF FOREST ECOSYSTEMS
Dry Particle Deposition
Fine







^
Oi


o

H
6
o

0
H
O
o
w
o

Region/Forest Type
North America
Douglas fir
Loblolly pine
Loblolly pine
Loblolly pine
Loblolly pine
Slash pine
White pine
Red spruce
Red spruce

Red spruce
Alder
Maple/beech

Oak
Oak

Europe
Norway spruce

Norway spruce
Location

Washington
Georgia
North Carolina
Tennessee
Tennessee
Florida
North Carolina
Maine
North Carolina

New York
Washington
New York

Tennessee
Tennessee


Norway

Germany
NO3

2
2
3
0.8
0.2
6
I
2
3

1
1
0.3


1


8

21
NH4+

9
27
35
18
14
16
22
14
74

9
5
5


36


21

62
Coarse
NO3

58
36
89
27
8
105
23
64
133

5
58
37


83


52

56
NH4+

27
37
66
7
1
4
23
102
43

2
27
13


8


29

4
Total

96
102
193
53
23
131
69
182
253

17
91
55

307
128


110

143
Annual Total
Nitrogen3

345
647
997
699
410C
431
510
545
1,939

1,136
339
567

857
720


775

1,250C
References'"

1,2
1,2
1,2
1,2
3
1,2
1,2
1,2
1,2

1,2
1,2
1,2

4
5


1,2

3
O
HH
H
W

-------

               TABLE 4-8. MEAN ANNUAL SULFATE DEPOSITION (equivalents/ha/yr) FROM FINE AND COARSE

            PARTICLES COMPARED TO TOTAL SULFUR DEPOSITION FROM ALL SOURCES TO A VARIETY OF
to
o
o
OJ







i
^

O
H
6
o
0
H
O
FOREST ECOSYSTEMS
Dry Particle Deposition
Region/Forest Type
North America
Douglas fir
Loblolly pine
Loblolly pine
Loblolly pine
Loblolly pine
Slash pine
White pine
Red spruce
Red spruce
Red spruce
Alder
Maple/beech
Oak forest
Europe
Norway spruce
Norway spruce
aTnHiirlp<; Hpnnsitinn firm
Location

Washington
Georgia
North Carolina
Tennessee
Tennessee
Florida
North Carolina
Maine
North Carolina
New York
Washington
New York
Tennessee
Germany
Norway
i nrprinitatinn rasps anH n
Fine

12
47
57
27
22
25
33
34
135
10
6
1
70
58
19
artiHps
Coarse

62
74
59
69
44
129
55
115
161
4
62
32
190
97
79

Total

74
121
116
96
66
154
88
149
296
14
68
39
260
155
98

Annual Sulfur Deposition3

320
776
1,050
941
570C
514
552
585
2,214
1,096
325
488
1,600
1,100C
663

References'"

1,2
1,2
1,2
1,2
3
1,2
1,2
1,2
1,2
1,2
1,2
1,2
4
3
1,2

        bl = Johnson and Lindberg (1992a), 2 = Lindberg (1992) and Lindberg and Lovett (1992), 3 = Lindberg et al. (1990); 4 = Lindberg et al. (1986).
        Includes only the growing season from April to October 1987.
O
HH
H
W

-------

TABLE 4-9. MEAN ANNUAL BASE CATION DEPOSITION (equivalents/ha/yr) FROM FINE AND COARSE
    PARTICLES COMPARED TO TOTAL BASE CATION DEPOSITION FROM ALL SOURCES TO
to
o
o
LtJ








r\
r\
oo


o
£5
H
6
o
0
H
O
a,
o
A VARIETY OF FOREST ECOSYSTEMS
Dry Base Cation Particles Deposition3
Region/Forest Type
North America
Douglas fir
Loblolly pine
Loblolly pine
Loblolly pine
Mixed Hardwood
Slash pine
White pine
Red spruce
Red spruce
Red spruce
Alder
Maple/beech
Oak forest

Europe
Spruce

Spruce

"Includes K+, Na+, Ca+2, and
Location

Washington
Georgia
North Carolina
Tennessee
Tennessee
Florida
North Carolina
Maine
North Carolina
New York
Washington
New York
Tennessee


Germanyd

Norway

Mg+2.
Fine

9
17
30
6.1
9.5d
17
4
5
25
2
4
1
lle




6


Coarse

180
130
340
220
121d
600
150
150
440
36
180
84
312e




180


Total

189
147
370
226
131d
617
154
155
465
38
184
85
323e




186


Total Annual Depositionb

670
300
780
410
899d
1,200
280
240
1,500
230
660
200
452e




390


References0

1,2
1,2
1,2
1,2
3
1,2
1,2
1,2
1,2
1,2
1,2
1,2
3


3

1,2


g Includes deposition from precipitation, gases, and particles.
0

O
H
W
cl = Ragsdaleetal. (1992),
Includes H+, K+, Na+, Ca+2,
"Includes only Ca+2 and K+.


2 = Johnson and Lindberg (1992a)
Mg+2 for the growing season from



, 3 = Lindberg et
April to October



al. (1986).
1987.



















-------
                         o
                        :•= 40 H
                         o
                         Q.
                         
-------
 1      aquatic or terrestrial ecosystem. As a result, ground-level concentrations of an air pollutant
 2      depend on the proximity to the sources, prevailing meteorology, and nature and extent of
 3      atmospherical reactions between the source and the receptor (Holland et al., 1999).  A more
 4      direct relationship exists between source strength and downwind ambient concentrations for
 5      primary air pollutants (e.g., SO2) than for secondary pollutants (e.g., sulfate, SO4"2). Interaction
 6      of the chemical and physical atmospheric processes and source locations for all of the pollutants
 7      have a tendency to produce data patterns that show large spatial and temporal variability.
 8           Holland et al. (1999) analyzed CASTnet monitoring data and using generative additive
 9      models (GAM) estimated the form and magnitude of trends of airborne concentrations of SO2
10      SO42", and nitrogen from 1989 to 1995 at 34 rural long-term CASTnet monitoring sites in the
11      eastern United States. These models provide a highly flexible method for describing potential
12      nonlinear relationships between concentrations, meteorology, seasonality, and time (e.g., how
13      weekly SO2 varies as a function of temperature). For most of the 34 sites in the eastern United
14      State, estimates of change in SO2 concentrations showed a decreasing functional form in 1989-
15      1990, followed by a relatively stable period during 1991-1993), then a sharper decline beginning
16      in 1994 (Holland  et al., 1999).
17           Regional trends of seasonal and annual wet deposition and precipitation-weighted
18      concentrations (PWCs) of sulfate in the United States over the period 1980-1995 were developed
19      by Shannon (1999) from monitoring data and scaled to a mean of unity.  In order to reduce some
20      effects  of year to year climatological variability, the unitless regional deposition and PWC trends
21      were averaged (hereafter referred to a CONCDEP).  During the 16 year period examined in the
22      study, estimated aggregate emissions of SO2 in the United States and Canada fell approximately
23      12% from about 1980 to 1982, it remained roughly level for a decade and then fell
24      approximately another  15% from 1992 to 1995 — for an overall decrease of about 18%.  Eastern
25      regional trends of sulfate concentrations and deposition and their average CONCDEP, also
26      exhibited patterns of initial decrease, near steady state, and final decrease with year to-to-year
27      variability.  The overall relative changed in CONCDEPs are greater than the changes in SO2
28      emissions.
29           Concentrations and calculated deposition (concentration times amount of water) of SO4"2 at
30      the Hubbard Brook Experiment Forest (HBEF) in the White Mountains of  central New
31      Hampshire have been measured since June of 1964 (Likens et al., 2001). These measurements

        June 2003                                 4-50        DRAFT-DO NOT QUOTE OR CITE

-------
 1      represent the longest continuous record of precipitation chemistry in North America. The
 2      long-term measurements generally concur with those of Shannon (1999) discussed above.  Major
 3      declines in emissions of SO2 have been observed during recent decades in the eastern United
 4      States and have been correlated with significant decreases in SO4"2 concentrations in
 5      precipitation (Shannon, 1999).
 6           Deposition of sulfates and nitrates are very clearly linked to emissions. Reduction in
 7      emissions must occur before concentrations can be reduced below current levels (Likens et al.,
 8      2001).  Deposition is the key variable as sensitive ecosystems in the eastern North America have
 9      not yet shown improvement in response to decreased emissions of SO2 (Driscoll et al., 1989;
10      Likens et al., 1996).  Clearly, additions of other chemicals, such as nitric acid and base cations,
11      must be considered in addition to sulfur when attempting to resolve the acid rain problem
12      (Likens et al., 1996,  1998).  The effects of sulfur and nitrogen deposition on ecosystems are
13      discussed in Section 4.2.2.2.
14           The long-term record indicates that a reduction in the deposition of basic cations (Ca+2,
15      Mg2+, K+, Na+) in bulk precipitation was associated with significant declines in sulfate deposition
16      cited above for the HBEF region (Driscoll et al., 1989). Decreases in streamwater
17      concentrations of basic cations have decreased simultaneously, suggesting that streamwater
18      concentrations of basic cations are relatively responsive to changes in atmospheric inputs.
19      Regardless of the cause, the decline in atmospheric influx of basic cations could have important
20      effects on nutrient availability as well as on the acid/base status of soil  and drainage water
21      (Driscoll etal., 1989).
22
23      Trace Elements
24           Deposition velocities for fine particles to forest surfaces have been reported  in the range of
25      1 to 15 cm s"1 (Smith, 1990a). For example, total, annual heavy metal deposition amounts are
26      highly variable depending on specific forest location and upwind source strength (Table 4-10).
27      Lindberg et al. (1982) quantified the dry deposition of heavy metals to inert surfaces and to
28      leaves of an upland oak forest. As noted for other chemical species, Vd was highly dependent on
29      particle size and chemical species (Table 4-11) with the larger particles depositing more
30      efficiently.
31

        June 2003                                 4-51        DRAFT-DO NOT QUOTE OR CITE

-------
       TABLE 4-10.  MEAN (±SE) PARTICLE SIZE, DEPOSITION RATES, AND DERIVED
          DEPOSITION VELOCITIES (Vd) FOR HEAVY METAL DEPOSITION TO THE
         UPPER CANOPY (INERT PLATES OR LEAVES) OF AN UPLAND OAK FOREST
Metal
Manganese
Cadmium
Zinc
Lead

Particle Size (um)
3.4 ±0.7
1.5 ±0.7
0.9 ±0.2
0.5
Deposition Rate
(pg cm 2 h'1)
91 ±23
0.3 ±0.1
6± 1
23 ±8
vd
(cm s"1)
6.4 ±3.6
0.37 ±0.18
0.38±0.1
0.06 ±0.01
       Source: Lindbergetal. (1982).
             TABLE 4-11.  TOTAL HEAVY METAL DEPOSITION TO TEMPERATE
                                     LATITUDE FORESTS
                                                      Forest Deposition kg ha ~l year ~l
                     Heavy Metal                                 (Range)
                       Cadmium                                0.002 - 0.02
                        Copper                                 0.016 - 0.24
                         Lead                                  0.099 - 1000
                        Nickel                                 0.014-0.15
       	Zinc	0.012-0.178	
       Source: Smith (1990c).
1          Preferential association of heavy metals with fine particles results in reduced control in
2     emission control systems. Metal removal efficiencies for baghouse filters are typically 95 to
3     99% for all but mercury, but fine particle capture is much less efficient. Wet scrubber efficiency
4     varies with design and pressure drop, typically 50 to 90% (McGowan et al., 1993). Fine
5     particles also have the longest  atmospheric residence times and, therefore, can be carried long
6     distances.  Depending on climate conditions and topography, fine particles may remain airborne
7     for days to months and may be transported 1,000 to 10,000 km or more from their source. This
8     long-distance transport and subsequent deposition qualify heavy metals as regional- and global-

      June 2003                               4-52        DRAFT-DO NOT QUOTE OR CITE

-------
 1      scale air pollutants. Ecosystems immediately downwind of major emissions sources (such as
 2      power generating, industrial, or urban complexes) may receive locally heavy inputs. Mass
 3      balance budgets (inputs and outputs) of seven heavy metals (cadmium, copper, iron, lead,
 4      manganese, nickel, and zinc) have been determined at the Hubbard Brook Experimental Forest
 5      (White Mountain National Forest) in New Hampshire. This forest is about 120 km northwest of
 6      Boston and relatively distant from major sources of heavy metal emissions. However,
 7      continental air masses that have passed over centers  of industrial and urban activity also
 8      frequently follow storm tracks over northern New England. Resulting annual inputs for the
 9      seven heavy metals at Hubbard Brook for  1975 to 1991 are presented in Figure 4-11. Note that
10      the 44-fold decrease in lead deposition is correlated with removal of lead from motor vehicle
11      fuels.
12           Trace element investigations conducted in roadside, industrial, and urban environments
13      have shown that impressive burdens of particulate heavy metals accumulate on vegetative
14      surfaces.  Lead deposition to roadside vegetation (prior to its removal from fuel) was 5 to 20,
15      50 to 200, and 100 to 200 times lead deposition to agricultural crops, grasses, and trees,
16      respectively, in non-roadside environments.  In an urban setting, it has been estimated that the
17      leaves and twigs of a 30-cm (12-in.) diameter sugar maple remove 60, 140, 5800, and 820 mg of
18      cadmium, chromium, lead, and nickel, respectively, during the course of a single growing season
19      (Smith, 1973).
20
21      Semivolatile Organics
22           Organic compounds partition between gas and particle phases, and particulate deposition
23      depends largely on the particle sizes available for adsorption (Pankow, 1987; Smith and Jones,
24      2000). Dry deposition of organic materials (e.g., dioxins, dibenzofurans, polycyclic aromatics)
25      is often dominated by the coarse fraction, even though mass loading in this size fraction may be
26      small (Lin et al., 1993) relative to the fine  PM fraction.  For example, measurements in Bavaria
27      in both summer and winter revealed that > 80% of organics were in the fine (< 1.35 jim) fraction
28      (Kaupp and McLachlan, 1999). Nevertheless, in most cases, calculated values of dry deposition
29      were dominated by the material adsorbed to coarse particles.  Wet deposition, in contrast, was
30      dominated by the much larger amount of material associated with fine particles.  In this
31      environment (where monthly precipitation is about 50 mm in winter and summer), wet

        June 2003                                4-53        DRAFT-DO NOT QUOTE OR CITE

-------
         s
                        x1
                     x10
                          -1
              x10
                          -2
Cd
Cu
Ni
Pb
Mn
Fe |
1
(
jJlljiJilslLdll.J^^ :
gjpr^--
" *" ' r-tV-ft t-
!L^Lii*«2»i&£i££^


x' ^A** , >, 'V"1
Sma'Y'ft -4""-"1^l^V-!
BBBBBB
*• ' "" ' k i '4
^^^^^JW^^jLy^
™1
|^^^^^._-_^^^
i*T^«??pT^r^
BBBBBBI

•j^%^Tj}iJ|SkW
LBBBV
•'' • ' * - *j ^ '


Tr"n " " " ' "'""-W *" * "^
ju'*' ^liiSj'tj^iL^JrSiife; 'jiif'^. ^S^
T iiiiiiiiiibiiiiiiiiiiiiiiiiiiiiiiiii

,"s ' ' " °", ^ 7 ! • -1" ^ ' -^ i-""
$,&.' jA-^iiliSSJ :

-? i ' ' f 'vs -v ^» A^,


^^,^^-^Z^^^
yggrv"— ~»^*^~^.
. 1
- i ; ^ - J ^ T., > £
^j^23 :
r^ 1
szn ^


^^j
!^3

.*£»«&$

^^ - ^,, - ^ ^^ i

T^^T™!
S

»* ' 5 * ,4l :
1975-79
1980-84
1985-89
1990-91
1 1 1 1 1
J 0.005 0.01 0.015 0.02 0.025
       Figure 4-11. Annual total deposition of heavy metals to Hubbard Brook Experimental
                    Forest, NH (Smith, 1990a).
 1
 2
 3
 4
 5
 6
 1
 8
 9
10
deposition dominated, with dry deposition accounting for only 14 to 25% of total deposition
(Kaupp and McLachlan, 1999). Lower relative contents of more volatile species in summer than
winter (Kaupp and McLachlan, 1999) indicate the critical importance of gas-particle phase
interconversions in determining deposition.
     Effective deposition of PM is required before biological effects on plants or ecosystems
can occur. It is clear that substantially improved techniques for monitoring and predicting
deposition will be required to characterize with certainty quantitative relationships between rates
and quantities of PM deposition associated with such effects.
       June 2003
                                         4-54
DRAFT-DO NOT QUOTE OR CITE

-------
 1      4.2.3  Assessment of Atmospheric PM Deposition Effects
 2      Introduction
 3           The discussion in the pages that follow assesses and characterizes the overall ecological
 4      condition or integrity of the ecosystems within the United States affected by the deposition of the
 5      anthropogenic stressors associated with PM and indicates their status. The six Essential
 6      Ecological Attributes (EEAs) - the Landscape Condition, Biotic Condition, and Chemical/
 7      Physical Characteristics,  Ecological Processes, Hydrology/Geomorphology, and Natural
 8      Disturbance Regimes (Table 4-1) provide a hierarchical framework for determining ecosystem
 9      status.  Measurable characteristics  related to structure, composition, or functioning of ecological
10      systems may be determined by the use of endpoints or ecological indicators of condition that are
11      significant either ecologically or to society (Harwell et al., 1999).
12           The relationships among the  EEAs are complex because all are interrelated (i.e., changes in
13      one EEA may affect, directly or indirectly, every other EEA).  The ecological processes create
14      and maintain patterns, which consist of elements in the system and the way they are arranged;
15      whereas the patterns in turn affect  how the processes are expressed (Science Advisory Board,
16      2002).  Changes in patterns or processes result in changes in the status and functioning of an
17      ecosystem.  The information in the sections that follow discusses changes in Landscape and
18      Biotic patterns and in Ecological and Chemical/Physical Processes resulting from the stressors in
19      PM deposition (Figure 4-12).
20           The elements of Biotic Condition are organized as nested hierarchy with several levels.
21      These include the structural and composition aspects (patterns) of the biota within landscape,
22      ecosystem or  ecological  community, species/population, organism and genetic/molecular level
23      (Science Advisory Board, 2002). Within these biological levels of organization, changes in the
24      biodiversity, composition, and structural elements relate  directly to functional integrity (such as
25      trophic status or structural integrity within habitats). Changes in biodiversity  are of particular
26      significance in altering the functioning of ecosystems.
27           As previously stated, ecosystems are dynamic, self-adjusting, self-maintaining, complex,
28      adaptive systems, in which patterns at the higher levels of biotic organization  emerge from
29      interactions and selection processes at localized levels (Levin, 1998).  Ecosystems components
30      must have an adequate supply of energy, mineral nutrients, and water to maintain themselves
31      and function properly. During the  ecological processes of energy and material flow, the energy

        June 2003                                  4-55        DRAFT-DO NOT QUOTE OR CITE

-------
         o
         m
         V)
        CO
Hydrolie Alteration
Habitat Conversion
Habitat Fragmentation
Climate Change
invasive Non-native Species
Turbidity/Sedimentation
Pesticides
Disease/Pest Outbreaks
Nutrient Pulses
Metals
Dissolved Oxygen  Depletion
Ozone (Tropospherie)
Nitrogen Oxides
Nitrates
                                              to
   Hydrolic Alteration
   Habitat Conversion
   Habitat Fragmentation
   Climate Change
   Over-Harvesting Vegitation
   Large-Scale Invasive
      Species Introduction
   Large-Scale Disease/Pest
      Outbreaks
                OT
                O
               CO
                                Hydrolic Alteration
                                Habitat Conversion
                                   Climate Change
                         Over-Harvesting Vegttatton
                           Disease/Pest Outbreaks
                               Altered Fire Regimt
                             Altered Flood Regime
                                                      (A
                                                      O
                                                      w
                                                      in
       Hydrolic Alteration
       Habitat Conversion
       Climate Change
       Turbidity/Sedimentation
       Pesticides
       Nutrient Pulses
       Metals
       Dissolved Oxygen Depletion
       Ozone (Tropospherie)
       Nitrogen Oxides
       Nitrates
       Sulfatss
       Salinity
       Acidic Deoosition
O
m
                                               (0
Hydrolic Alteration
Habitat Conversion
Climate Change
Pesticides
Disease/Pest Outbreaks
Nutrient Pulses
Dissolved Oxygen Depletion
Nitrogen Oxides
Nitrates
Sulfates
                              Hydrolic Alteration
                             Habitat Conversion
                           Habitat Fragmentation
                                Climate Change
                         Turbidity/Sedimentation
                                                                                                O
                                                                                                m
                                                                                                0)
                                                                                                0
                                                                                               +-<
                                                                                               w
       Figure 4-12.  Sample stressors and the essential ecological attributes they affect
                      (after Science Advisory Board, 2002).
1      obtained by plants (the producers) from sunlight during photosynthesis (primary production) and

2      chemical nutrients (e.g., nitrogen, phosphorus, sulfur cycling) taken up from the soil are

3      transferred to other species (the consumers) within the ecosystem through food webs.  The
       June 2003
                                         4-56
                    DRAFT-DO NOT QUOTE OR CITE

-------
 1      movement of chemical nutrients (materials) through an ecosystem is cyclic, as the nutrients are
 2      used or stored and eventually returned to the soil by decomposer organisms. Energy, on the
 3      other hand, is transferred from organism to organism through an ecosystem in food webs and,
 4      finally, is dissipated into the environment as heat (Odum, 1993).
 5           Ecosystem and community patterns  are characterized by the interaction of their component
 6      species, the ecosystems processes of energy flow, nutrient flux, water and material flow and by
 7      the effect their activities have on the physical and chemical environment.  The flows of energy
 8      and nutrient cycling provide the interconnectedness among the elements of the biotic hierarchy
 9      and transform the community from random  collection of species into an integrated whole,
10      an ecosystem. Elucidating these interactions across scales is fundamental to understanding the
11      relationships between biodiversity and ecosystem functioning (Levin, 1998).
12           Human existence on this planet depends on the life-support services provided by the
13      interaction of the different EEAs. Both ecosystem structure (biotic condition) and functions
14      (ecological processes) play an essential role in providing ecosystem goods (products) and
15      services (Table 4-12; Daily, 1997). Ecosystem processes maintain clean water, clean air, a
16      vegetated Earth, and a balance of organisms, the functions that enable humans to survive.  The
17      benefits they impart include absorption and  breakdown of pollutants, cycling of nutrients,
18      binding of soil, degradation of organic waste, maintenance of a balance of gases in the air,
19      regulation of radiation balance and climate,  and the fixation of solar energy (World Resources,
20      2000-2001; Westman, 1977; Daily,  1997). Economic benefits  and values associated with
21      ecosystem functions, goods and services,  and the need to preserve them because of their value to
22      human life are discussed by Costanza et al. (1997) and (Pimentel et al.,  1997). Goods have
23      market value; whereas services usually are not considered to have market value.
24           Attempts have been made to calculate the value of biodiversity and the world's ecosystem
25      services and natural capital (Goulder and  Kennedy, 1997; Pimentel et al., 1997;  Costanza et al.,
26      1997).  The majority of these have been controversial because of a lack of agreement on the
27      philosophical basis for placing a value on ecosystem services.  Costanza et al. (1997) state that it
28      may never be possible to make a precise estimate of the services provided by ecosystems;
29      however, their estimates indicate the relative importance of ecosystem services, not their true
30      value considering that the loss of ecosystem services can affect human existence.  One problem
31      with traditional economic analysis is that  following consumption, the material and energy

        June 2003                                4-57       DRAFT-DO NOT QUOTE OR CITE

-------
   TABLE 4-12. PRIMARY GOODS AND SERVICES PROVIDED BY ECOSYSTEMS
 Ecosystem
Goods
Services
 Agroecosvstems
 Coastal ecosystems
 Forest ecosystems
 Freshwater
 Grassland
 ecosystems
Food crops
Fiber corps
Crop genetic resources
Fish and shellfish
Fishmeal (animal feed)
Seaweeds (for food and industrial
use)
Salt
Genetic resources
Timber
Fuelwood
Drinking and irrigation water
Fodder
Nontimber products (vines,
bamboos, leaves, etc.)
Food (honey, mushrooms fruit,
and other edible plants; game)
Genetic resources
Drinking and irrigation water
Fish
Hydroelectricity
Genetic resources
Livestock (food, game, hides, and
fiber)
Drinking and irrigation water
Genetic resources
Maintain limited watershed functions (infiltration,
flow control, and partial soil protection)
Provide habitat for birds, pollinators, and soil
organisms important to agriculture
Sequester atmospheric carbon
Provide employment

Moderate storm impacts (mangroves, barrier islands)
Provide wildlife (marine and terrestrial) habitat and
breeding areas/hatcheries/nurseries
Maintain biodiversity
Dilute and treat wastes
Provide harbors and transportation routes
Provide human and wildlife habitat
Provide employment
Contribute aesthetic beauty and provide recreation

Remove air pollutants, emit oxygen
Cycle nutrients
Maintain array of watershed functions (infiltration,
purification, flow control, soil stabilization)
Maintain biodiversity
Sequester atmospheric carbon
Moderate weather extremes and impacts
Generate soil
Provide employment
Provide human and wildlife habitat
Contribute aesthetic beauty and provide recreation

Buffer water flow (control timing and volume)
Dilute and carry away wastes
Cycle nutrients
Maintain biodiversity
Provide aquatic habitat
Provide transportation corridor
Provide employment
Contribute aethetic beauty and provide recreation

Maintain array of watershed functions (infiltration,
purification, flow control, and soil stabilization)
Cycle nutrients
Remove air pollutants and emit  oxygen
Maintain biodiversity
Generate soil
Sequester atmospheric carbon
Provide human and wildlife habitat
Provide employment
Contribute aesthetic beauty and provide recreation
 Source: World Resources (2000-2001).
June 2003
                            4-58
         DRAFT-DO NOT QUOTE OR CITE

-------
 1      embedded in economic goods disappear from the universe of economic analysis although the
 2      laws of thermodynamics conservation guarantee their persistence (Moomaw, 2002).
 3      An industrial ecology approach considers the full life cycle of materials and energy and
 4      recognizes that economic wastes still need to be accounted for following consumption of goods.
 5      For example, nitrogen incorporated in fertilizer does not disappear after fuel is burned to produce
 6      smog; it continues to cascade through the biosphere. The degraded ecosystem continues to
 7      provide fewer ecosystem services for the indefinite future. Some nitrogen in the form of nitric
 8      acid may rain down on crops and forests, providing nutrients in degraded form.  This diluted
 9      nitrogen has the economic equivalent of "scrap value," even if humans do not find it expedient to
10      recover it, given present techniques (Moomaw, 2002).
11           Heal (2000), on the other hand, feels that attempts to value ecosystems and their services
12      are misplaced:  "Economics cannot estimate the importance of natural environments to society:
13      only biology can do that" (Heal, 2000).  The role of economics is to help design institutions that
14      will provide incentives to the public and policy makers for the conservation of important natural
15      systems and for mediating human impacts on the biologically diverse ecosystems and the
16      biosphere so that they are sustainable. The approach of Harwell et al. (1999) and the report by
17      the Ecological Processes and Effects Committee of the SAB (Science Advisory Board, 2002)
18      also deals with the need to understand human effects on ecosystems so that ecosystem
19      management can define what ecological conditions are desired. Further, they state that the
20      establishment of ecological goals involves a close linkage between scientists and decision
21      makers in which science informs decision makers and the public by characterizing the ecological
22      conditions that are achievable under particular management regimes. Decision makers then can
23      make choices that reflect societal values including issues of economics, politics, and culture.
24      For management to achieve their goals, the general public, scientific community, resource
25      managers, and decision makers need to be routinely apprised of the condition or integrity of
26      ecosystems so that ecological goals may be established (Harwell et al., 1999).
27           Biodiversity, the variety of life, encompasses all levels of biological organization,
28      including species individuals, populations, and ecosystems (Figure 4-1; Wilson, 1997).
29      Human-induced changes in biotic diversity and alterations in the structure and functioning of
30      ecosystems are the two most dramatic ecological trends in the past century (Vitousek et al.,
31      1997).  There are few ecosystems on Earth today that are not influenced by humans

        June 2003                                4-59        DRAFT-DO NOT QUOTE OR CITE

-------
 1      (Freudenburg and Alario, 1999; Vitousek et al., 1997; Matson et al., 1997; Noble and Dirzo,
 2      1997).  The scientific literature is filled with publications discussing the importance of
 3      ecosystem structure (patterns) and function (processes). The deposition of particulate matter
 4      from the atmosphere has the potential to alter ecosystem structure and function by altering
 5      nutrient cycling and changing biodiversity.  There is a need, therefore, to understand how
 6      ecosystems respond to both natural and anthropogenic stresses in order to determine whether or
 7      not anthropogenic stresses are affecting ecosystem services and products (Table 4-12).
 8           Concern has risen in recent years regarding the consequences of changing the biological
 9      diversity of ecosystems (Tilman, 2000; Ayensu et al., 1999; Wall, 1999; Hooper and Vitousek,
10      1997; Chapin et al., 1998).  The concerns arise because human activities are creating
11      disturbances that are decreasing biodiversity; altering the complexity and  stability of
12      ecosystems; and producing changes in ecological processes and the structure, composition and
13      function of ecosystems (Figure 4-13; Pimm, 1984; Levin, 1998; Chapin et al., 1998; Peterson
14      et al., 1998; Tilman, 1996; Tilman and Downing, 1994; Wall, 1999; Daily and Ehrlich, 1999).
15      The above changes can affect the ecosystem services vital to human life.
16
17      4.2.3.1  Effects on Vegetation and Ecosystems
18           Exposure to a given mass concentration of airborne PM may lead to widely differing
19      phytotoxic responses, depending on the particular mix of deposited particles. Effects of
20      particulate deposition on individual plants or ecosystems are difficult to characterize because of
21      the complex interactions among biological, physicochemical, and climatic factors.  Most direct
22      effects, other than regional effects associated with global changes, occur in the severely polluted
23      areas surrounding industrial point sources, such as limestone quarries, cement kilns, and metal
24      smelting facilities.  Fine particles are more widely distributed from their source. Experimental
25      applications of PM constituents to foliage typically elicit little response at the more common
26      ambient concentrations.  The diverse chemistry and size characteristics of ambient PM and the
27      lack of clear distinction between effects attributed to phytotoxic particles  and to other air
28      pollutants further confound understanding of the direct effects on foliar surfaces. The majority
29      of documented toxic effect of particles on vegetation reflect their chemical content (acid/base,
30      trace metal, nutrient), surface properties, or  salinity.  Studies of direct effects of particles on
31      vegetation have not yet advanced to the stage of reproducible exposure experiments.  Difficulties

        June 2003                                  4-60         DRAFT-DO NOT QUOTE OR CITE

-------
                          Food supply
                          and demand
                                           Water use and
                                           nutrient loss
            Hydroiogic
            CO2 and
            temperature
            changes
                                                  Freshwater
                                              supply and demand
                              Water availability
                       \                 r
                           Land     Preaprtation/Erosion
                           transfer-  and temp- / changes
                           mation    erature   /  jnwa|erflow
                                                and temperature
                        N, CH4,N20
                        emissions
                                                                        Forest product
                                                                      supply and demand
climate change
                                Loss of crop
                                genetic diversity
                                                                 traamentation
                                               Habitat loss
                                 Change in
                                 transpiration
                                    " albedo

                                            Biodiversity loss
          Habitat
          change
       Figure 4-13.  Linkages among various ecosystem goods and services (food, water,
                     biodiversity, forest products) and other driving forces (climate change)
                     (modified from Ayensu et al., 1999).
 1     in experimental application of ambient particles to vegetation have been discussed by Olszyk
 2     et al. (1989).  It is now clear that many phytotoxic gases are deposited more readily, assimilated
 3     more rapidly, and lead to greater direct injury to vegetation than do most common particulate
 4     materials (Guderian, 1986).  The dose-specific responses (dose-response curves) obtained in
 5     early experiments following the exposure of plants to phytotoxic gases generally have not been
 6     observed following the application of particles.
 7          Unlike gaseous dry deposition, neither the solubility of the particles nor the physiological
 8     activity of the surface is likely to be of first order of importance in determining deposition
 9     velocity (Vd). Factors that contribute to surface wetness and stickiness may be critical
10     determinants  of sticking efficiency. Available tabulation of deposition velocities are highly
11     variable and suspect.  High-elevation forests receive larger particle deposition loadings than
12     equivalent lower elevations sites because of higher wind speeds and enhanced rates of aerosol
       June 2003
                               4-61
DRAFT-DO NOT QUOTE OR CITE

-------
 1      impaction; orographic effects on rainfall intensity and composition; increased duration of occult
 2      deposition; and, in many areas, the dominance of coniferous species with needle-shaped leaves
 3      (Section 4.2.2.3; Lovett, 1984). Recent evidence indicates that all three modes of deposition
 4      (wet, occult, and dry) must be considered in determining inputs to ecosystems or watersheds,
 5      because each may dominate over specific intervals of space.
 6           Atmospheric PM may affect vegetation directly following deposition on foliar surfaces or
 7      indirectly by changing the soil chemistry or through changes in the amount of radiation reaching
 8      the Earth's surface through PM-induced climate change processes.  Indirect effects, however, are
 9      usually the most significant because they can alter nutrient cycling and inhibit plant nutrient
10      uptake.
11
12      Direct Effects of PM Deposition
13           Coarse and fine PM deposition affects both (a) patterns in the elements of EEA categories
14      subsumed under Biotic Condition and (b) processes subsumed under Chemical/Physical and
15      Ecological Processes in the EPEC framework described earlier. Measurable responses have
16      been observed as reductions in photosynthesis  and in salinity of the soil and foliar effects of
17      nitrate, sulfate, and acidic and heavy metal  deposition.
18           Particles transferred from the atmosphere to foliar surfaces may reside on the leaf, twig, or
19      bark surface for extended periods; be taken up  through the leaf surface; or be removed from the
20      plant via resuspension to the atmosphere, washing by rainfall, or litter-fall with subsequent
21      transfer to the soil.  Any PM deposited on above-ground plant parts may exert physical or
22      chemical  effects. The effects of "inert" PM are mainly physical; whereas those of toxic particles
23      are both chemical and physical.  The effects of dust deposited on plant surfaces or soil are more
24      likely to be associated with their chemistry  than simply with the mass of deposited particles and
25      may be more important than any physical effects (Farmer,  1993).  Nevertheless, vegetative
26      surfaces represent filtration and reaction/exchange sites (Tong,  1991; Youngs et al., 1993).
27
28      Direct Effects of Coarse Particles
29           Coarse particles, ranging in size from 2.5 to 100 jim, are chemically diverse, dominated by
30      local sources, and typically deposited near their source because of their sedimentation velocities.
31      Airborne coarse particles are derived from the  following sources:  road, cement kiln, and

        June 2003                                 4-62        DRAFT-DO NOT QUOTE OR CITE

-------
 1      foundry dust; fly ash; tire particles and brake linings; soot and cooking oil droplets; biogenic
 2      materials (e.g, plant pollen, fungal spores, bacteria and viruses); abraded plant fragments; sea
 3      salt; and hydrated deliquescent particles of otherwise fine aerosol. In many rural areas and some
 4      urban areas, the majority of mass in the coarse particle mode is in the elements silicon,
 5      aluminum, calcium, and iron, suggesting a crustal origin as fugitive dust from disturbed land,
 6      roadways, agriculture tillage, or construction activities. Rapid sedimentation of coarse particles
 7      tends to restrict their direct effects on vegetation largely to roadsides and forest edges.
 8
 9           Physical Effects — Radiation. Dust can cause physical and chemical effects. Deposition
10      of inert PM on above-ground plant organs sufficient to coat them with a layer of dust may result
11      in changes in radiation received, a rise in leaf temperature, and the blockage of stomata.
12      Increased leaf temperature and heat stress; reduced net photosynthesis; and leaf chlorosis,
13      necrosis, and abscission were reported by Guderian (1986).  Road dust decreased the leaf
14      temperature on Rhododendron catawbiense by ca., 4 °C (Eller, 1977); whereas foundry dust
15      caused an 8.7 °C increase in leaf temperature of black poplar (Populus nigra) under the
16      conditions of the experiment (Guderian, 1986). Deciduous (broad) leaves exhibited larger
17      temperature increases because of particle loading than did conifer (needle) leaves, a function of
18      poorer coupling to the atmosphere. Inert road dust caused a three- to four-fold increase in the
19      absorption coefficient  of leaves of English ivy (Eller, 1977; Guderian, 1986) for near infrared
20      radiation (NIR; 750 to 1350 nm). Little change in absorption occurred for photosynthetically
21      active radiation (PAR; 400 to 700 nm).  The increase in NIR absorption was equally at the
22      expense of reflectance and transmission in these wavelengths. The amount of energy entering
23      the leaf increased by ca., 30% in the dust-affected leaves. Deposition of coarse particles
24      increased leaf temperature and contributed to heat stress; reduced net photosynthesis; and caused
25      leaf chlorosis, necrosis, and abscission (Dassler et al.,  1972; Parish, 1910; Guderian, 1986;
26      Spinka,  1971).
27           Starch storage in dust-affected leaves increased with dust loading under high (possibly
28      excessive) radiation, but decreased following dust deposition when radiation was limiting.
29      These modifications of the radiation environment had a large effect on single-leaf utilization of
30      light.  The boundary layer properties, determined by leaf morphology and environmental
31      conditions, strongly influenced the direct effects of particle deposition on radiation heating

        June 2003                                 4-63        DRAFT-DO NOT QUOTE OR CITE

-------
 1      (Eller, 1977; Guderian, 1986) and on gas exchange. Brandt and Rhoades (1973) attributed the
 2      reduction in growth of trees to crust formation from limestone dust on the leaves. Crust
 3      formation reduced photosynthesis and the formation of carbohydrates needed for normal growth,
 4      induced premature leaf-fall, damaged leaf tissues, inhibited growth of new tissue, and reduced
 5      starch storage.  Dust may decrease photosynthesis, respiration, and transpiration; and it may
 6      allow penetration of phytotoxic gaseous pollutants, thereby causing visible injury symptoms and
 7      decreased productivity.  Permeability of leaves to ammonia increased with increasing dust
 8      concentrations  and decreasing particle size (Farmer, 1993).
 9           Dust also has been reported to physically block stomata (Krajickova and Mejstfik, 1984).
10      Stomatal clogging by particulate matter from automobiles, stone quarries, and cement plants was
11      also studied by Abdullah and Iqbal (1991). The percentage of clogging was low in young leaves
12      when compared with old and mature leaves and the amount of clogging varied with species and
13      locality. The maximum clogging of stomata observed was about 25%.  The  authors cited no
14      evidence that stomatal clogging inhibited plant functioning.  The heaviest deposit of dust is
15      usually on the upper surface of broad-leaved plants; whereas the majority of the stomata are on
16      the lower surface where stomatal clogging would be less likely.
17
18           Chemical Effects.  The chemical composition of PM is usually the key phytotoxic factor
19      leading to plant injury. Cement-kiln dust on hydration liberates calcium hydroxide that can
20      penetrate  the epidermis and enter the mesophyll; and, in some cases, this  has caused the leaf
21      surface alkalinity to reach a pH of 12. Lipid hydrolysis, coagulation of the protein compounds,
22      and ultimately plasmolysis of the leaf tissue result in reduction in the growth and quality of
23      plants (Guderian, 1986).  In experimental studies, application of cement kiln dust of known
24      composition for 2 to 3 days yielded dose-response curves between net photosynthetic inhibition
25      or foliar injury  and dust application rate (Darley, 1966). Lerman and Darley (1975) determined
26      that leaves must be misted regularly to produce large effects. Alkalinity was probably the
27      essential phytotoxic property of the applied dusts.
28
29           Salinity.  Parti culate matter enters the atmosphere from oceans following the mixing of air
30      into the water column and the subsequent bursting of bubbles at the surface.  The effervescence
31      of bubbles on the surface of the ocean forcefully ejects droplets of sea water into the air. These

        June  2003                                4-64        DRAFT-DO NOT QUOTE OR CITE

-------
 1      droplets, concentrated by evaporation, are carried inland by wind and deposited on the seaward
 2      side of coastal plants (Boyce, 1954). This occurs largely in the surf line (i.e., near land and
 3      potentially sensitive terrestrial receptors). This process can be a significant source of sulfate,
 4      sodium, chloride, and trace elements (as well as living material) in the atmospheric aerosol
 5      impacting coastal vegetation.  Sea-spray particles (Taback et al., 1979) are approximately 24%
 6      greater in size than 10 jim, and 54% are between 3 and 10 jim.  Thus, only about 20% are fine
 7      (0 to 2.5 |im) particles; and deposition by sedimentation and impaction is concentrated near the
 8      coast, whereas particle size distribution shifts toward the fine fraction over longer inland
 9      transport distances.  Airborne concentrations of this marine PM decrease quickly with distance
10      inland from the surfline both by deposition and dilution within atmospheric mixed layer (McKay
11      et al., 1994; Nelis et al., 1994). Near-shore sediments with associated pollutants present in
12      coastal runoff may be suspended in the surf and reentrained into the air. This can be a
13      substantial source of microorganisms and of radionuclides to coastal vegetation (Nelis et al,
14      1994; McKay et al., 1994).
15           Sea-salt particles can serve as nuclei for the absorption and subsequent reaction of other
16      gaseous and particulate air pollutants. Both nitrate and sulfate from the atmosphere have been
17      found to associate with coarse and fine sea-salt particles (Wu and Okada, 1994). Direct effects
18      on vegetation reflect these inputs, as well as classical salt injury caused by the sodium and
19      chloride that constitute the bulk of these particles. Foliar accumulation of airborne salt particles
20      may lead to foliar injury, thusly affecting the species composition in coastal environments
21      (Smith, 1984).
22           The effects of winds and sea spray on coastal vegetation has been reported in the literature
23      since the early 1800s (Boyce, 1954). However, there has been a difference of opinion as to
24      whether the injury to coastal vegetation resulted from windblown aerial salts or from mechanical
25      injury (i.e., sand blasting) due to wind alone. Though the significance of sea water dashed on
26      fore dunes and rocky coasts had been recognized by several authors, Wells and  Shunk (1937,
27      1938) and Wells (1939) were the first to recognize the importance of salt spray in coastal
28      ecology. Wells and Shunk (1937) reported that salt spray carried over dunes was the most
29      important factor influencing growth form, zonation, and succession in coastal dunes.  Salt spay
30      injury was recorded 1.25 miles inland on the North Carolina coast.  On the basis of observations
31      in the Cape Fear area, they determined that the shape of coastal "wind form" shrubs were the

        June 2003                                 4-65        DRAFT-DO NOT QUOTE OR CITE

-------
 1      result of sea spray carried by high winds. They found injury on shrubs only near the coast while
 2      those at a greater distances inland showed no injury whatsoever after a strong southeast wind
 3      that persisted for a period of nineteen hours during cloudy weather and abundant soil moisture.
 4           To determine the cause of injury, injured and uninjured shoots were titrated for chlorides.
 5      A marked difference was observed between the injured and uninjured shoots (Wells and Shunk,
 6      1937, 1938). Experimental spraying of shoots of woody plants with seawater resulted in a
 7      pattern of injury similar to the injury observed on seaside shrubs.  The absence of the more
 8      inland species, such as persimmon (Diosporos virginiana L.), turkey oak (Quercus laevis Walt.),
 9      longleaf pine (Pinuspalustris Mill., P. australis Michx.), and wire grass (Aristida stricta
10      Michx.), was explained on the basis of intolerance of these species to salt spray. The dominance
11      of live oak (Quercus virginiana Mill.), as a practically pure stand on Smith Island (also known
12      as Bald Head Island), NC, and along the eastern and southern NC coast,  was determined by
13      Wells (1939) to be due to the tree's tolerance to  salt spray. He termed the long term stabilization
14      of the live oak stand as a new type of climax, the "salt spray climax."  The later work of Costing
15      and Billings (1942) near Beaufort, NC, corroborated the findings of Wells and Shunk, 1937,
16      1938).
17           The report by Boyce (1954) is probably  the most extensive on salt-spray communities.
18      Dune sands in many coastal  areas have been shown to have extremely low concentrations of
19      dissolved salts. Studies have indicated that the salt content of the coastal dunes of Virginia,
20      Massachusetts, and California did not exceed the maximum occurring in ordinary cultivated
21      soils. Costing and Billings (1942) found no correlation between soil salinity and plant
22      distribution on the North Carolina coast.  Surface crusts of sand dunes have been shown to have
23      high concentrations of chlorides which could be attributed to sea spray, while concentrations of
24      chlorides in underlying layers was low.  The surface layer, however, varied with exposure of the
25      dunes to oceanic winds (Boyce, 1954).
26           Boyce (1954), Wells (1939), and Wells and Shunk (1938) concluded on the basis of their
27      studies that necrosis  and death of plant tissues results from the high deposition of salt spray and
28      high accumulation of the chloride ion in the plant tissues.  Very little salt is taken up by plant
29      roots; most enters through the aerial organs. Leaves of plants exposed to salt spray show
30      a distinct pattern of injury (Wells and Shunk,  1938).  Necrotic areas first appear at the leaf tips
31      and upper margins and then  progress slowly in an inverted "V" toward the petiole. This leaf

        June 2003                                 4-66        DRAFT-DO NOT QUOTE OR CITE

-------
 1      injury pattern was verified experimentally. Mechanical injury resulting from leaves and twigs
 2      beating against each another in the wind causes the formation of small lesions through which salt
 3      can enter.  After entry into the plant, the chloride ion is rapidly translocated to the apices of the
 4      leaves and twigs where it accumulates to injurious concentrations and results in the death of only
 5      a portion of the plant.  The differential deposition and translocation of the chloride ion results in
 6      the death of the  seaward leaves and twigs. The result is the continued growth of the uninjured
 7      branches in an inland direction.  As a result, the canopy angle varies with the intensity of the
 8      spray (Boyce, 1954).
 9           Little or no mineral ions are available in the silicate sands of the of coastal dunes.
10      Consequently, plants obtain mineral ions needed for growth from the salt spray.  Seawater
11      contains all of the mineral ions required for growth except nitrogen and phosphorus.  The
12      amount of nitrogen and phosphorus in seawater varies over a wide range (Boyce, 1954).
13      Experiments indicated that available nitrogen in sea spray was a conditioning factor.  Low
14      nitrogen availability increased the tolerance of dune species to salt spray.  Increasing the
15      availability of nitrogen resulted in a different pattern of plant zonation and distribution.
16      Dicotyledonous species were restricted to areas of lower spray intensity. The severity of
17      chloride injury was associated more with the amount of available potassium than with the
18      concentration of chlorides within the limits of 280-360 mg Cl/liter (Boyce, 1954).
19           Other sources of phytotoxic saline PM include aerosols from cooling towers and roadway
20      deicing salt. Cooling towers used to dissipate waste heat from steam-electric power generating
21      facilities may emit salt if brackish water or saltwater is used as a coolant (McCune et al., 1977;
22      Talbot, 1979). Foliar injury is related to salt droplets deposited  by sedimentation or impaction
23      from cooling tower drift.  The distance of the salt drift determines the amount of deposition and
24      location of injury. Environmental conditions most conducive to injury were absence of
25      precipitation, which can wash salt off leaves, and high relative humidity (RH; Talbot, 1979).
26      Increased injury is associated with wind speed and salt concentrations.  Typical toxicity
27      symptoms from acute exposures include marginal foliar necrosis and lesions; shoot-tip  dieback;
28      leaf curl; and interveinal necrosis (McCune et al., 1977).  Based on experimental data, Grattan
29      et al. (1981) observed that, to cause injury, salt deposited on leaf surfaces must dissolve and be
30      absorbed into leaf tissue.  Their work also indicated the importance of RH in foliar uptake.
        June 2003                                 4-67        DRAFT-DO NOT QUOTE OR CITE

-------
 1      If RH remained below 70%, even heavy deposition of salt did not induce injury in peppers,
 2      soybeans, and tomatoes.
 3           Injury to vegetation from the application of deicing salt was related to salt spray blown or
 4      drifting from the highways by Hofstra and Hall (1971) and Viskari and Karenlampi (2000).  The
 5      most severe injury was observed nearest to the highways.  The results presented in these studies
 6      agrees with that of Wyttenbach et al. (1989), who observed that conifers planted near roadway
 7      margins in the eastern United States often exhibit foliar injury due to toxic levels of saline
 8      aerosols deposited from deicing solutions. Piatt and Krause (1974) demonstrated that road and
 9      site factors influence the spread of deicing salt into forested areas.  The slope away from the road
10      influenced the distance from the road where injury was observed and the percent slope was
11      correlated with distance.
12
13      Effects of Fine Particles
14           Fine PM in rural areas is generally secondary in nature, having condensed from the vapor
15      phase or been formed by chemical reaction from gaseous precursors in the atmosphere, and is
16      generally smaller than 1 to 2.5 jim. Nitrogen and sulfur oxides, volatile organic gases,
17      condensates of volatilized metals, and products of incomplete combustion are common
18      precursors for fine PM. Reactions of many of these materials with an oxidizing atmosphere
19      contribute to high secondary PM concentrations during summer months in many U.S. areas or
20      during late fall and winter in areas with high nitrate concentrations.  The conclusion that
21      sufficient data were not available for adequate quantification of dose-response functions for
22      direct effects of fine aerosols on vegetation, reached in the  1982 PM/SOX AQCD (U.S.
23      Environmental Protection Agency, 1982), continues to be true today.  Only a few studies on the
24      direct effects of acid aerosols have been completed (U.S. Environmental Protection Agency,
25      1982).  The major effects are indirect and occur through the soil (Section 4.2.3.2).
26
27           Nitrogen. Nitrate is observed in both fine and coarse particles. Nitrates from atmospheric
28      deposition represent a substantial fraction of total nitrogen  inputs to southeastern forests (Lovett
29      and Lindberg, 1986).  However, much of this is  contributed by gaseous nitric acid vapor; and a
30      considerable amount of the particulate nitrate is taken up indirectly through the soil. Garner
31      et al. (1989) estimated deposition of nitrogen to forested landscapes in eastern North America at

        June 2003                                  4-68        DRAFT-DO NOT QUOTE OR CITE

-------
 1      10 to 55 kg/ha/year for nitrate and 2 to 10 kg/ha/year for ammonium.  About half of these values
 2      were ascribed to dry deposition.
 3           Driscoll et al. (2003) estimated anthropogenic nitrogen inputs to eight large watersheds in
 4      the Northeastern United States for the year 1997.  Inputs of total nitrogen deposition ranged from
 5      14 kg/ha/year in the Casco Bay watershed, ME, to 68 kg/ha/year in the Massachusetts Bay
 6      watershed.  Atmospheric deposition of nitrogen was the second largest source ranging in
 7      amounts from 5 to 10 kg/ha/year (11-36% of the total; Driscoll et al., 2003). Nitrogen deposition
 8      in the western United States ranges from 1 to 4 kg/ha/year over much of the region to as high as
 9      30 kg/year downwind of major urban and agricultural areas.  An unknown amount of nitrogen
10      deposited to the West Coast originates in Asia (Fenn et al., 2003a).
11           Atmospheric additions of particulate nitrogen in excess of vegetation needs are lost from
12      the system, mostly as leachate from the soil as nitrate. Managed agricultural ecosystems may be
13      able to utilize deposited parti culate nitrogen more efficiently than native ecosystems although
14      many cultivated systems also lose considerable nitrogen as nitrate in runoff, deep drainage, or
15      soil water.  It has proven difficult to quantify direct foliar fertilization by uptake of nitrogen from
16      ambient particles.
17           There is no doubt that foliar uptake of nitrate can occur, as clearly shown by the efficacy of
18      foliar fertilization in horticultural systems.  Potassium nitrate was taken up by leaves of
19      deciduous fruit trees (Weinbaum and Neumann, 1977) and resulted in increased foliar nitrogen
20      concentrations.  Not all forms of nitrogen are absorbed equally, nor are all equally benign.
21      Following foliar application of 2600  ppm of nitrogen as Ca(NO3)2, (NH4)2SO4, or (NH2)2CO to
22      apple canopies (Rodney, 1952; Norton and Childers,  1954), leaf nitrogen levels were observed to
23      increase to similar levels;  but calcium nitrate and ammonium sulfate caused visible foliar injury;
24      whereas urea did not. Urea is generally the recommended horticultural foliar fertilizer.
25           The mechanism of uptake of foliarly deposited nitrate is not well established. Nitrate
26      reductase is generally a root-localized enzyme:  it is generally not present in leaves, but is
27      inducible there. Induction typically occurs when the soil is heavily enriched in NO3".  As the
28      root complement of nitrate reductase becomes overloaded, unreduced nitrate reaches the leaves
29      through the transpiration stream.  Nitrate metabolism has been demonstrated in  leaf tissue
30      (Weinbaum and Neumann, 1977) following foliar fertilization. Residual nitrate reductase
31      activity in leaves may be adequate to assimilate typical rates of paniculate nitrate deposition.

        June  2003                                 4-69        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Uptake of nitrate may be facilitated by codeposited sulfur (Karmoker et al., 1991; Turner and
 2      Lambert, 1980).
 3           Nitrate reductase is feedback-inhibited by its reaction product NH4+. The common
 4      atmospheric aerosol, NH4NO3, therefore may be metabolized in two distinct biochemical steps:
 5      first the ammonium (probably leaving nitric acid) and then the nitrate. Losses of nitric acid by
 6      volatization during this process, if they occur, have not been quantified.
 7           Direct foliar effects of particulate nitrogen have not been documented. Application of a
 8      variety of fine nitrogenous aerosol particles (0.25 jim) ranging from 109 to 244 |ig/m3 nitrogen
 9      with or without 637 |ig/m3 sulfur caused no consistent short-term (2- to 5-h) effect on gas
10      exchange in oak, maize, or soybean leaves (Martin et al., 1992).
11           Although no evidence  exists for the direct transfer of nutrient parti culate aerosols into
12      foliage, a few studies give insights into the potential for ammonium and nitrate transfer into
13      leaves. Fluxes of both NO3" and NH4+, measured in wet deposition and in throughfall plus
14      stemflow in forests, commonly indicate higher fluxes of nitrogen above the canopy (Parker,
15      1983; Lindberg et al., 1987;  Sievering et al., 1996), and imply net foliar uptake. Lovett and
16      Lindberg (1993) reported a linear relationship between inorganic nitrogen fluxes in deposition
17      and throughfall, suggesting that uptake may be considered passive to some extent.
18           Garten and Hanson (1990) studied the movement of 15N-labeled nitrate and ammonium
19      across the cuticles of red maple (Acer rubruni) and white oak (Quercus alba) leaves when
20      applied as an artificial rain mixture. Brumme et al. (1992), Bowden et al. (1989), and Vose and
21      Swank (1990) have published similar data for conifers. These studies show the potential for
22      nitrate and ammonium to move into leaves, where it may contribute to normal physiological
23      processes (e.g., amino acid production; Wellburn, 1990).  Garten (1988) showed that internally
24      translocated 35S was not leached readily from tree leaves of yellow poplar (Liriodendron
25      tulipiferd) and red maple (Acer rubrum), suggesting that SO4"2 would not be as mobile as the
26      nitrogen-containing ions discussed by Garten and Hanson (1990).  Further, when the foliar
27      extraction method is used, it is not possible to distinguish sources of chemicals deposited as
28      gases  or particles (e.g., nitric acid [HNO3], nitrogen dioxides [NO2], nitrate [NO3"]), or sources of
29      ammonium (deposited as ammonia [NH3] or ammonium ion [NH4+]; Garten and Hanson, 1990).
30           Particle deposition contributes only a portion of the total atmospheric nitrogen deposition
31      reaching vegetation; but, when combined with gaseous and precipitation-derived sources, total

        June 2003                                 4-70        DRAFT-DO NOT QUOTE OR CITE

-------
 1      nitrogen deposition to ecosystems has been identified as a possible causal factor leading to
 2      changes in natural ecosystems (See Section 4.2.2).
 3
 4           Sulfur. Anthropogenic sulfur emissions are > 90% SO2. Most of the remaining emission
 5      of sulfur is directly as sulfate (U.S. Environmental Protection Agency,  1996a). Sulfur dioxide is
 6      hydrophilic and is rapidly hydrated and oxidized to sulfite and bisulfite and then to sulfate,
 7      which is approximately 30-fold less phytotoxic.  The ratio of SO4"2/SO2 increases with aging of
 8      the air mass  and, therefore, with distance from the source. Sulfate is sufficiently hygroscopic in
 9      humid air that it may exist significantly in the coarse particulate fraction. Because dilution of
10      both SO2 and particulate SO4"2 occurs with distance from the source, it is unusual for damaging
11      levels of particulate sulfate to be deposited.  Gas to particle conversion in this case is of benefit
12      to vegetation.
13           Sulfur  is an essential plant nutrient.  Low dosages of sulfur serve  as a fertilizer, particularly
14      for plants growing in sulfur-deficient soil (Hogan et al., 1998). However, current levels of
15      sulfate deposition reportedly exceed the capacity of most vegetative canopies to immobilize the
16      sulfur (Johnson,  1984). Nitrogen uptake in forests may be regulated loosely by sulfur
17      availability,  but sulfate additions in excess of needs do not typically lead to injury unless
18      deposited in acidic precipitation (Turner and Lambert, 1980).
19           There are few field demonstrations of foliar sulfate uptake (Krupa and Legge,  1986, 1998).
20      Sulfate in throughfall is often enriched above levels in precipitation.  The relative importance of
21      foliar leachate and prior dry-deposited sulfate particles remains difficult to quantify  (Cape et al.,
22      1992).  Leaching rates are not constant and may respond to levels of other pollutants, including
23      acids. Uptake and foliar retention of gaseous and particulate sulfur are confounded by variable
24      rates of translocation and accessibility of deposited materials to removal and quantification by
25      leaf washing. Following soil enrichment with 35SO4"2 in a Scots pine forest, the apparent
26      contribution of leachate to throughfall was only a few percent following an initial burst of over
27      90% because of extreme disequilibrium in labeling of tissue sulfate pools (Cape et al., 1992).
28           Olszyk et al. (1989) provide information on the effects of multiple pollutant exposures
29      including particles (NO3; 142 |ig/m3; NH4+,  101 |ig/m3; SO/2,  107 |ig/m3). They found that only
30      gaseous pollutants produced direct (harmful) effects on vegetation for the concentrations
31      documented, but the authors hypothesized that long-term accumulation of the nitrogen and sulfur

        June 2003                                 4-71         DRAFT-DO NOT QUOTE OR CITE

-------
 1      compounds contributed from particle deposition might have effects on plant nutrition over long
 2      periods of time. Martin et al. (1992) exposed oak (Quercus macrocarpa), soybean (Glycine
 3      max), and maize (Zea mays) plants to acute exposures (2 to 5 h) of aerosols (0.25 jim) containing
 4      only nitrate (109 |ig/m3), ammonium and nitrate (244 and  199 |ig/m3), or ammonium and sulfate
 5      (179 and 637 |ig/m3). They found that these  exposures, which exceeded the range of naturally
 6      occurring aerosol concentrations, had little effect on foliar photosynthesis and conductance.
 7      Martin et al. (1992) concluded that future investigations should focus on the effects of particles
 8      on physiological characteristics of plants following chronic exposures.
 9           Acidic Deposition. The effects of acidic deposition  have been accorded wide attention in
10      the media and elsewhere (Altshuller and Linthurst,  1984; Hogan et al., 1998). Probably the most
11      extensive assessment of acidic deposition processes and effects is the NAPAP Biennial Report to
12      Congress: An Integrated Assessment (National Science and Technology Council, 1998).
13      Concern regarding the effects of acidic deposition on crops and forest trees has resulted in
14      extensive monitoring and research. Exposures to acidic rain or clouds can be divided into
15      "acute" exposures to higher ionic concentrations (several  |imol/L) and "chronic" long-term
16      repeated exposures to lower concentrations (Cape, 1993).  Pollutant concentrations  in rainfall
17      have been shown to have little capacity for producing direct effects on vegetation (Altshuller and
18      Linthurst, 1984; Hogan et al., 1998); however, fog and clouds, which may contain solute
19      concentrations up to 10 times those found in  rain, have the potential to cause direct  effects.
20      More than 80% of the ionic composition of most cloud water is made up of four major pollutant
21      ions: H+, NH4+, NO3", and SO4"2.  Ratios of hydrogen to ammonium and sulfate to nitrate vary
22      from site to site with all four ions usually present in approximately equal concentrations.
23      Available data from plant effect studies  suggest that hydrogen and sulfate ions are more likely to
24      cause injury than ions containing nitrogen (Cape, 1993).
25           The possible direct effects of acidic precipitation on forest trees have been evaluated by
26      experiments on seedlings and young trees. The size of mature trees makes experimental
27      exposure difficult, therefore necessitating extrapolations from experiments on seedlings and
28      saplings; however, such extrapolations must  be used with  caution (Cape, 1993). Both conifers
29      and deciduous species have shown significant effects on leaf surface structures after exposure to
30      simulated acid rain or acid mist at pH 3.5.  Some species have shown subtle effects  at pH 4 and
31      above.  Visible lesions have been observed on many species at pH 3 and on sensitive species at

        June 2003                                4-72         DRAFT-DO NOT QUOTE OR CITE

-------
 1      pH 3.5 (Cape, 1993). The relative sensitivities of forest vegetation to acidic precipitation based
 2      on macroscopic injury have been ranked as follows:  herbaceous dicots > woody dicots >
 3      monocots > conifers (Percy, 1991).
 4           Huttunen (1994) described the direct effects of acid rain or acidic mist on epicuticular
 5      waxes whose ultrastructure  is affected by plant genotype and phenotype. The effects of air
 6      pollutants on epicuticular waxes of conifers have received greater study than the waxes of other
 7      species. Leaf age and the shorter life span of broad-leaved trees make them less indicative of the
 8      effects of acid precipitation. Many experimental studies indicate that epicuticular waxes that
 9      function to prevent water loss from plant leaves can be destroyed by acid rain in a few weeks
10      (Huttunen, 1994). This function is crucial in conifers because of their longevity and evergreen
11      foliage. Microscopic observations of epicuticular wax structures have, for a long time,
12      suggested links between acidic deposition and aging. In Norway spruce (Picea abies), acid rain
13      causes not only the aging of needles (which in northern conditions normally last from
14      11 to  14 years) to be shortened, but also accelerates the erosion rate of the waxes as the needles
15      age.
16           The effects of acidic precipitation and fog on red spruce (Picea rubens) have been studied
17      extensively (Schier and Jensen, 1992). Visible foliar injury of the needles in the form of a
18      reddish-brown discoloration has been observed on red spruce seedlings experimentally exposed
19      to acidic mist, but this visible symptom has not been observed in the field.  Ultrastructural
20      changes in the epicuticular wax were observed both experimentally and on spruce growing at
21      high elevations. Laboratory studies indicate that visible injury usually does not  occur unless the
22      pH is  3 or less (Schier and Jensen, 1992). Cape (1993) reported that, when compared with other
23      species, red spruce seedlings appeared to be more sensitive to acid mist. From studies of
24      conifers and a review of the literature, Huttunen (1994) concluded that acidic precipitation
25      causes direct injury to tree foliage and indirect effects through the soil. The indirect effects of
26      acidic precipitation are discussed in Section 4.2.3.2.
27           Based on a review of the  many studies in the literature involving field and controlled
28      laboratory experiments on crops, Cape (1993) drew a number of conclusions concerning the
29      direct effects of acidic precipitation on crops:
30       • foliar injury and growth reduction occur below pH 3;
31       • allocation of photosynthate is altered, with increased shoot to root ratios;

        June 2003                                  4-73        DRAFT-DO NOT QUOTE OR CITE

-------
 1       •  expanded and recently expanded leaves are most susceptible, and injury occurs first to
           epidermal cells;
 2       •  leaf surface characteristics such as wettability, buffering capacity, and transport of material
           across the leaf surface contribute to susceptibility and differ among species;
 3       •  data obtained from experiments in greenhouses or controlled environmental chambers
           cannot be used to predict effects on plants grown in the field;
 4       •  quantitative data from experimental exposures cannot be extrapolated to field exposures
           because of differences and fluctuations in concentrations, durations, and frequency  of
           exposure;
 5       •  there are large differences in response within species;
 6       •  timing of exposure in relation to phenology is of utmost importance;
 7       •  plants may be able to recover from or adapt to injurious exposures; and
 8       •  sequential exposure to acidic precipitation and gaseous pollutants is unlikely to be more
           injurious than exposure to individual  pollutants.
 9           Studies by Chevone et al.  (1986), Krupa and Legge (1986), and Blaschke (1990) differ
10      with the last conclusion of Cape listed above. Their studies indicate that interactions between
11      acidic deposition and gaseous pollutants do occur.  Acidity affects plant responses to both O3 and
12      SO2.  Chevone et al. (1986) observed increased visible injury on soybean and pinto bean when
13      acid aerosol exposure preceded O3 exposure; whereas linear decreases in dry root weight of
14      yellow poplar occurred as acidity increased with simultaneous exposures to O3 and simulated
15      acid rain.  Krupa and Legge (1986) also noted increased visible injury to pinto bean plants when
16      aerosol exposure preceded O3 exposure.  In none of the studies cited above did acid rain per se
17      produce significant growth changes.  In contrast, Blaschke (1990) observed a decrease in
18      ectomycorrhizal frequency and  short root distribution caused by acid rain exposure in
19      combination with either SO2 or O3.
20
21           Trace Elements. All but  10 of the 90 elements that comprise the inorganic fraction of the
22      soil occur at concentrations of less than 0.1% (1000 |ig/g) and are termed "trace" elements.
23      Trace elements with a density greater than 6 g cm"3, referred to as "heavy metals," are of
24      particular interest because of their potential toxicity for plant and animals. Although some trace

        June 2003                                 4-74        DRAFT-DO NOT QUOTE OR CITE

-------
 1      metals are essential for vegetative growth or animal health, they are all toxic in large quantities.
 2      Combustion processes produce metal chlorides that tend to be volatile and metal oxides that tend
 3      to be nonvolatile in the vapor phase (McGowan et al., 1993). Most trace elements exist in the
 4      atmosphere in particulate form as metal oxides (Ormrod,  1984). Aerosols containing trace
 5      elements derive predominantly from industrial activities (Ormrod, 1984).  Generally, only the
 6      heavy metals cadmium, chromium, nickel, and mercury are released from stacks in the vapor
 7      phase (McGowan et al., 1993).  Concentrations of heavy metals in incinerator fly ash increase
 8      with decreasing particle size.
 9           Vegetational surfaces, especially the foliage, present a major reaction and filtration surface
10      to the atmosphere and act to accumulate particles deposited via wet and dry  processes described
11      in Section 4.2.2 (Tong, 1991; Youngs et al., 1993).  The chemical constituents of particles
12      deposited on foliar surfaces may be taken up through the leaf surface. The greatest particle
13      loading is usually on the adaxial (upper) leaf surface where particles accumulate in the mid-vein,
14      center portion of the leaves. Additionally, the mycelium of fungi becomes particularly abundant
15      on leaf surfaces as the growing season progresses and is in intimate association with  deposited
16      particles (Smith, 1990c).
17           Investigations of trace elements present along roadsides and in industrial and urban
18      environments indicate that impressive burdens of parti culate heavy metals can accumulate on
19      vegetative surfaces. Foliar uptake of available metals could result in metabolic effects in above-
20      ground tissues.  Only a few metals, however, have been documented to cause direct
21      phytotoxicity in field conditions. Copper, zinc, and nickel toxicities have been observed most
22      frequently.  Low solubility, however, limits foliar uptake and direct heavy metal toxicity
23      because trace metals must be brought into solution before they can enter into leaves or bark of
24      vascular plants. In those instances when trace metals are absorbed, they are frequently bound in
25      leaf tissue and are lost when the leaf drops off (Hughes, 1981).  Trace metals in mixtures may
26      interact to cause a different plant response when compared with a single element; however, there
27      has been little research on this aspect (Ormrod, 1984).  In experiments using chambers,
28      Marchwinska and Kucharski (1987) studied the effects of SO2 alone and in combination with
29      PM components (Pb, Cd, Zn, Fe, Cu, and Mn) obtained from a zinc smelter  bag filter.  The
30      combined effects of SO2 and PM further increased the reduction in yield of beans caused by SO2;
        June 2003                                4-75        DRAFT-DO NOT QUOTE OR CITE

-------
 1      whereas the combination, though severely injuring the foliage, produced little effect on carrots
 2      and parsley roots except after long-term exposures (when there was a decrease in root weight).
 3           Trace metal toxicity of lichens has been demonstrated in relatively few cases. Nash (1975)
 4      documented zinc toxicity in the vicinity of a zinc smelter near Palmerton, PA.  Lichen species
 5      richness and abundance were reduced by approximately 90% in lichen communities at Lehigh
 6      Water Gap near the zinc smelter when compared with those at Delaware Water Gap.  Zinc,
 7      cadmium, and SO2 were present in concentrations toxic to some species near the smelter;
 8      however, toxic zinc concentrations were detected farther away than the detectable limits of SO2
 9      (Nash, 1975).  Experimental data suggest that lichen tolerance to Zn and Cd falls between
10      200 and 600 ppm (Nash, 1975).
11           Though there has been no direct evidence of a physiological association between tree
12      injury and exposure to metals, heavy metals have been implicated because their deposition
13      pattern is correlated with forest decline. The role of heavy metals has been indicated by
14      phytochelatin measurements. Phytochelatins are intracellular metal-binding peptides that act as
15      specific indicators of metal stress. Because they are produced by plants as a response to
16      sublethal concentrations of heavy metals, they can indicate that heavy metals play a role in forest
17      decline (Gawel et al., 1996).  Concentrations of heavy metals increased with altitude, as did
18      forest decline, and increased concentrations across the study region that show increased levels of
19      forest injury, as well.
20           Phytochelatin concentrations were measured in red spruce and balsam fir {Abies balsamed)
21      needles throughout the 1993 growing season at 1000 m on Whiteface Mountain in New York.
22      Mean foliar concentrations in red spruce were consistently higher than in balsam fir from June
23      until August, with the greatest and most significant difference occurring at the peak of the
24      growing season in mid-July. In July, the phytochelatin concentrations were significantly higher
25      than at any other time measured. Balsam fir did exhibit this peak, but maintained a consistently
26      low level throughout the season. Both the number of dead red spruce trees and phytochelatin
27      concentrations increased sharply with elevation (Gawel et al., 1996).  The relationship between
28      heavy metals and the decline of forests in northeastern United States was further tested by
29      sampling red spruce stands showing varying degrees of decline at  1000 m on nine mountains
30      spanning New Hampshire, Vermont, and New York. The collected samples indicated a
31      systematic and significant increase in phytochelatin concentrations associated with the extent of

        June 2003                                 4-76        DRAFT-DO NOT QUOTE OR CITE

-------
 1      tree injury.  The highest phytochelatin concentrations were measured during 1994 from sites
 2      most severely affected by forest decline in the Green Mountains, VT, and the Adirondack
 3      Mountains, NY. These data strongly imply that metal stress causes tree injury and contributes to
 4      forest decline in the northeastern United States (Gawel et al., 1996).
 5           One potential direct effect of heavy metals is on the activity of microorganisms and
 6      arthropods resident on and in the leaf surface ecosystem.  The fungi and bacteria living on and in
 7      the surfaces of leaves play an important role in the microbial succession that prepares leaves for
 8      decay and litter decomposition after their fall (U.S. Environmental Protection Agency, 1996b).
 9      Numerous fungi were consistently isolated from foliar surfaces at various crown positions from
10      London plane trees growing in roadside environments in New Haven, CT. Those existing
11      primarily as saprophytes included Aureobasidium pullulans, Chaetomium sp., Cladosporium sp.,
12      Epicoccum sp., and Philaphora verrucosa. Those existing primarily as parasites included
13      Gnomoniaplatani, Pestalotioposis sp., and Pleurophomella sp.  The following cations were
14      tested in vitro for their ability to influence the growth of these fungi: cadmium, copper,
15      manganese, aluminum, chromium, nickel, iron, lead, sodium, and zinc. Results indicated
16      variable fungal response with no correlation between saprophytic or parasitic activity and
17      sensitivity to heavy metals. Both linear extension and dry weight data indicated that the
18      saprophytic Chaetomum sp. was very sensitive to numerous metals.  Aureobasidium pullulans,
19      Epicoccum sp., and especially P.verrucosa, on the other hand, appeared to be much more
20      tolerant. Of the parasites, G. platani appeared to be more tolerant than Pestalotiopsis sp. and
21      Pleurophomella sp. Metals exhibiting the broadest spectrum growth suppression were iron,
22      aluminum, nickel, zinc, manganese,  and lead (Smith and Staskawicz, 1977; Smith, 1990c).
23      These in vitro studies employed soluble compounds containing heavy metals. Trace metals
24      probably occur naturally on leaf surfaces as low-solubility oxides, halides, sulfates, sulfides, or
25      phosphates (Clevenger et al., 1991; Koslow et al., 1977).  In the event of sufficient solubility and
26      dose, however, changes  in microbial community structure on leaf surfaces because of heavy
27      metal accumulation are possible.
28
29           Organic Compounds.  Volatile organic compounds in the atmosphere are partitioned
30      between the gas and particle phases, depending on the liquid phase  vapor pressure at the ambient
31      atmospheric temperature, the surface area of the particles per unit volume of air, the nature of the

        June 2003                                 4-77        DRAFT-DO NOT QUOTE OR CITE

-------
 1      particles and of the chemical being adsorbed; and they can be removed by wet and dry
 2      deposition (McLachlan, 1996a).  Materials as diverse as DDT, polychlorinated biphenyls
 3      (PCBs), and polynuclear aromatic hydrocarbons (PAHs) are being deposited from the
 4      atmosphere on rural as well as urban landscapes (Kylin et al., 1994). Motor vehicles emit
 5      particles to the atmosphere from  several sources in addition to the tailpipe. Rogge et al. (1993b)
 6      inventoried the organic contaminants associated with fine particles (diameter <2.0 jim) in road
 7      dust, brake-lining-wear particles, and tire tread debris. More than 100 organic compounds were
 8      identified in these samples, including n-alkanols, benzoic acids, benzaldehydes, polyalkylene
 9      glycol ethers, PAHs, oxy-PAH, steranes, hopanes, natural resins, and other compound classes.
10      A large number of PAHs, ranging from naphthalene (C10H8) to 5- and 6-ring and higher PAHs,
11      their alkyl-substituted analogues, and their oxygen- and  nitrogen-containing derivatives are
12      emitted from motor vehicle sources (Seinfeld, 1989).
13          Plants may be used as environmental monitors to compare the deposition of PAH,
14      persistent organic pollutants (POPs), or semivolatile organic components (SOCs) between sites
15      (e.g., urban versus rural; Wagrowski and Kites, 1997; Ockenden et al., 1998; McLachlan, 1999).
16      Vegetation can be used qualitatively to indicate organic  pollutant levels as long as the
17      mechanism of accumulation is considered. The substance may enter the plant via the roots or, as
18      mentioned above, be deposited as a particle onto the waxy cuticle of leaves or be taken up
19      through the stomata.  The pathways are a function of the chemical and physical properties  of the
20      pollutant such as  its lipophilicity, water solubility, vapor pressure (which controls the
21      vapro-particle partitioning) and Henry's law constant; environmental conditions such as ambient
22      temperature and the organic content of the soil; and the plant species, which controls the surface
23      area and lipids available for accumulation (Simonich and Kites, 1995).  Ockenden et al. (1998)
24      have observed that, for lipophilic POPs, atmospheric transfer to plant has been the main avenue
25      of accumulation.  Plants can differentially accumulate POPs.  Results have shown differences
26      between species with higher concentrations in the lichen (Hypogymniaphysiodes) than in Scots
27      pine needles (Pinus sylvestris). Even plants of the same species, because they have different
28      growth rates and different lipid contents (depending on the habitat in which they are growing),
29      have different rates of sequestering pollutants.  These facts confound data interpretations and
30      must be taken into account when considering their use as passive samplers.
        June 2003                                4-78        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Vegetation itself is an important source of hydrocarbon aerosols. Terpenes, particularly
 2      cc-pinene, p-pinene, and limonene, released from tree foliage may react in the atmosphere to
 3      form submicron particles. These naturally generated organic particles contribute significantly to
 4      the blue haze aerosols formed naturally over forested areas (Geron et al., 2000).
 5           The low water solubility with high lipoaffinity of many of these organic xenobiotics
 6      strongly control their interaction with the vegetative components of natural ecosystems.  The
 7      cuticles of foliar surfaces are covered with a wax layer that helps protect plants from moisture
 8      and short-wave radiation stress.  This epicuticular wax, consisting mainly of long-chain esters,
 9      polyesters, and paraffins, has been demonstrated to accumulate lipophilic compounds. Organic
10      air contaminants in the particulate or vapor phase are absorbed to and accumulate in the
11      epicuticular wax of vegetative surfaces (Gaggi et al., 1985; Kylin et al., 1994). Direct uptake of
12      organic contaminants through the cuticle or the vapor-phase uptake through the stomates are
13      characterized poorly for most trace organics. The phytotoxicity and toxicity of organic
14      contaminants to soil microorganisms is not well studied (Foster, 1991).
15
16      4.2.3.2  Ecosystem Response to Stresses
17           Ecosystem response to stress begins within the Biotic Condition Attribute at the population
18      level with changes in patterns resulting from the  response of sensitive individual plants or
19      animals. Ecosystem response to pollutant deposition is a direct function of the ecosystem's
20      ability to ameliorate resulting change (Strickland et al., 1993). Plant responses, changes in both
21      structural and compositional patterns, and functional ecological processes must be scaled in both
22      time and space and be propagated from the individual to the more complex levels of community
23      interaction to produce observable changes in an ecosystem (see Figure 4-14). Among ecosystem
24      biota, at least three levels of biological interaction are involved:  (1) the individual plant and its
25      environment; (2) the population and its environment; and (3) the biological community
26      composed of many species and its environment (Billings, 1978). Individual organisms within a
27      population vary in their ability to withstand the stress of environmental change. The response of
28      individual organisms within a population is based on their genetic constitution (genotype), stage
29      of growth at time of exposure to stress, and the microhabitat in which they are growing (Levin,
30      1998).  The range within which these organisms  can exist and function determines the ability of
31      the population to survive. Those able to cope with the stresses survive and reproduce.

        June 2003                                 4-79        DRAFT-DO NOT QUOTE OR CITE

-------
N. Reactic
_evel of.
Organization
Leaf
(cm2)
Branch
(cm2)
Tree
(nf)
Stand
(ha)
n Time
Minute
•—



Day





Year

	 ^
Ni 	 •
fc3
• •
as i
t^3
^
•
•



Decade
2
•3
—4
in CD 1^0005
iilii
s
^
\\



Century


-10
11
-12
13
^14
.^ VI

16
For a given level, the dot associated with a line begins with
and ends with the associated structure (e.g., the needle).
Injury Symptom
Needle necrosis
and abscission
Branch length,
bifurcation ratio,
and ring-width
growth altered
Reduction in
diameter and death
of tree
Decreases in
stand productivity,
increases in mortality
and alterations in
regeneration patterns
Key Changes in Processes
Reduced carbon assimilation
because of reduced radiation
Reduced carbon available for foliage
replacement and branch growth/
export Synergistic interaction
between mistletoe and tephra
deposition
Reduced carbon available for
height, crown, and stem growth
Influence of crown class on initial
impact and subsequent recovery
Interaction between stand
composition and recovery
a process (e.g. photosynthesis for#1 under leaf)
 Evaluating Impacts Within a Level of Organization
 Leaf Level     Carbon exchange-1
              Carbon pools-2
              Needle number and size-3
              Needle retention/abscission-4

 Branch Level   Carbon allocation-5
              Branch growth-6
              Branch morphology-7
              Branch vigor-8
              Branch retention-9
Tree Level     Height and diameter growth-10
             Crown shape and size-11
             Tree vigor-12
             Mortality-13

Stand Level    Productivity-14
             Mortality-15
             Species composition-16
 Evaluating Interactions Between Different Levels of Organization
              The diagonal arrow indicates the interaction between any two levels of organization.
              The types of interaction are due to the properties of variability and compensation.
              A- Refers to the interaction between the leaf and branch levels, where, for example,
                  variability at the branch level determines leaf quantity, and compensation at the leaf
                  level in photosynthesis may compensate for the reduction in foliage amount.
              B - Refers to the interaction between the branch and the tree, where variability in branches
                 determines initial interception, branch vigor, and branch location in the crown;
                 compensation may be related to increased radiation reaching lower branches.
              C - Refers to the interaction between the tree and the stand. Both genetic and
                 environmental variability, inter- and intraspecific compensations, and tree historical
                 and competitive synergisms are involved.
Figure 4-14.  Effects of environmental stress on forest trees are presented on a hierarchial
                scale for the leaf, branch, tree, and stand levels of organization.  The
                evaluation of effects within a level of organization are indicated by horizontal
                arrows.  The evaluation of interactions between different levels of
                organization are indicated by diagonal arrows.

Source:  Hinckley et al. (1992).
June 2003
4-80
DRAFT-DO NOT QUOTE OR CITE

-------
 1      Competition among the different species results in succession (community change over time)
 2      and, ultimately, produces ecosystems composed of populations of plant species that have the
 3      capability to tolerate the stresses (Rapport and Whitford, 1999; Guderian, 1985).
 4           The number of species in a community usually increases during succession in unpolluted
 5      atmospheres. Productivity, biomass, community height, and structural complexity increase.
 6      Severe stresses, on the other hand, divert energy from growth and reproduction to maintenance
 7      and return succession to an earlier stage (Waring and Schlesinger, 1985).  Ecosystems are
 8      subject to natural periodic stresses, such as drought, flooding, fire, and attacks by biotic
 9      pathogens (e.g., fungi, insects). Extremely severe natural perturbations return succession to an
10      earlier stage; reduce ecosystem structure and functions (i.e., produce a scarcity of life forms and
11      extinguish symbiotic interactions); disrupt the plant processes of photosynthesis and nutrient
12      uptake, carbon allocation, and transformation that are directly related to energy flow and nutrient
13      cycling; shorten food chains; and reduce the total nutrient inventory (Odum, 1993).  This
14      transformation, however, sets the stage for recovery that permits the perturbed ecosystem to
15      adapt to changing environments (Holling,  1986). Therefore, these perturbations are seldom
16      more than a temporary setback, and recovery  can be rapid (Odum, 1969).
17           In contrast, anthropogenic stresses usually are severe, debilitating stresses.  Severely
18      stressed ecosystems do not recover readily, but may be further degraded (Odum, 1969; Rapport
19      and Whitford, 1999).  Anthropogenic stresses can be classified into four main groups:
20      (1) physical restructuring (e.g., changes resulting from land use); (2) introduction of exotic
21      species; (3) over harvesting; and (4) discharge of toxic substances into the atmosphere, onto
22      land, and into water. Ecosystems usually lack the capacity to adapt to the above stresses and
23      maintain their normal structure and functions unless the stress is removed (Rapport and
24      Whitford, 1999). These stresses result in a process of degradation marked by a decrease in
25      biodiversity, reduced primary and  secondary production, and a lower capacity to recover and
26      return to its original state. In addition, there is an increased prevalence of disease, reduced
27      nutrient cycling, increased dominance of exotic species, and increased dominance by smaller,
28      short-lived opportunistic species (Odum, 1985; Rapport and Whitford, 1999).  Discharge of toxic
29      substances into the atmosphere, onto land, and into water can cause acute and chronic stresses;
30      and, once the stress is removed, a process of succession begins that can ultimately return the
31      ecosystem to a semblance of its former structure.  Air pollution stresses, if acute, are usually

        June 2003                                 4-81        DRAFT-DO NOT QUOTE OR CITE

-------
 1      short-term and the effects soon visible. Chronic stresses, on the other hand, are long-term
 2      stresses whose effects occur at different levels of ecosystem organization and appear only after
 3      long-term exposures, as in the case of acidic deposition in the northeast or ozone in California
 4      (Shortle and Bondietti, 1992; U.S. Environmental Protection Agency, 1996b).
 5           The possible effects of air pollutants on ecosystems have been categorized by Guderian
 6      (1977) as follows:
 7       (1)   accumulation of pollutants in the plant and other ecosystem components (such as soil
              and surface- and groundwater),
 8       (2)   damage to consumers as a result of pollutant accumulation,
 9       (3)   changes in species diversity because of shifts in competition,
10       (4)   disruption of biogeochemical cycles,
11       (5)   disruption of stability and reduction in the ability of self-regulation,
12       (6)   breakdown of stands and associations, and
13       (7)   expansion of denuded zones.
14           How changes in these functions can result from PM deposition and influence ecosystems is
15      discussed in the following text. It should be remembered that, although the effects of PM are
16      being emphasized, the vegetational components of ecosystems also are responding to multiple
17      stresses from multiple sources.
18
19      Response to Direct Effects of PM
20           In the previous section, it was noted that PM affects patterns in the EEA Biotic Condition
21      and processes in the Chemical/Physical and Ecological Processes categories.  The presence of
22      PM in the atmosphere may affect vegetation directly, following physical contact with the foliar
23      surface (Section 4.2), but in most cases, the more significant  effects are indirect.  These effects
24      may be mediated by suspended PM (i.e., through effects on radiation and climate) and by
25      particles that pass through the vegetative canopies to the soil.
26           The majority of studies dealing with direct effects of particulate dust and trace metals on
27      vegetation have  focused on responses of individual plant species and were conducted in the
28      laboratory or in  controlled environments (Saunders and Godzik, 1986). A few have considered
29      the effects of particles on populations, communities, and ecosystems.  Most of these focused on

        June  2003                                 4-82        DRAFT-DO NOT QUOTE OR CITE

-------
 1      ecosystems in industrialized areas heavily polluted by deposits of both chemically inert and
 2      active dusts. Effects can result from direct deposition or indirectly by deposition onto the soil.
 3      Reductions in growth, yield, flowering, and reproductive processes of plants from particulate
 4      deposition have been reported (Saunders and Godzik, 1986).  Sensitivities of individual species
 5      have been associated with changes in composition and structural patterns of natural ecosystems.
 6           Evidence from studies of effects of PM deposition, specifically chemically inert and active
 7      dusts indicates that, within a population, plants exhibit a wide range of sensitivity, which is the
 8      basis for the natural selection of tolerant individuals. Rapid evolution of certain populations of
 9      tolerant species at sites with heavy trace element and nitrate deposition has been observed
10      (Saunders and Godzik, 1986).  Tolerant individuals present in low frequencies in populations
11      when growing in unpolluted areas have been selected for tolerance at both the seedling and adult
12      stages when exposed to trace metal or nitrate deposition (Ormrod, 1984; U.S. Environmental
13      Protection Agency,  1993). Chronic pollutant injury to a forest community may result in the loss
14      of sensitive species, loss of tree canopy, and maintenance of a residual cover of pollutant-
15      tolerant herbs or shrubs that are recognized as success!onal species (Table 4-13; Smith, 1974).
16      These changes in forest patterns result from altered ecological processes.
17           Responses of ecosystems to stresses (unless severe or catastrophic) are difficult to
18      determine because the changes are subtle (Garner, 1991).  This is particularly true of responses
19      to particles.  Changes in the soil may not be observed until accumulation of the pollutant has
20      occurred for 10 or more years, except in the severely polluted areas around heavily industrialized
21      point sources (Saunders and Godzik, 1986). In addition, the presence of other co-occurring
22      pollutants makes it difficult to attribute the effects to PM alone. In other words, the potential for
23      alteration of ecosystem function and structure exists but is difficult to quantify, especially when
24      there are other pollutants present in the ambient air that  may produce additive or synergistic
25      responses even though PM concentrations may not be elevated.
26
27      Physical Effects
28           The direct effects of limestone dust on plants and ecosystems has been known for many
29      years. Changes have been observed in both ecosystem patterns and processes. Long-term
30      changes in the structure and composition of the seedling-shrub and sapling strata of an
31      experimental site near limestone quarries and processing plants in Giles  County in southwestern

        June 2003                                  4-83         DRAFT-DO NOT QUOTE OR CITE

-------
             TABLE 4-13. ECOSYSTEM FUNCTIONS IMPACTED BY AIR POLLUTION
                         EFFECTS ON TEMPERATE FOREST ECOSYSTEMS
         Forest Soil and Vegetation: Activity and Response
           Ecosystem Consequence and Impact
         (1) Forest tree reproduction, alteration, or inhibition

         (2) Forest nutrient cycling, alteration
            a. Reduced litter decomposition
            b. Increased plant and soil leaching and soil
               weathering
            c. Disturbance of microbial symbioses

         (3) Forest metabolism
            a. Decreased photosynthesis
            b. Increased respiration
            c. Altered carbon allocation

         (4) Forest stress, alteration
            a. Phytophagous insects, increased or decreased
               activity
            b. Microbial pathogens, increased or decreased
               activity
            c. Foliar damage increased by direct air pollution
               influence
        (1) Altered species composition

        (2) Reduced growth, less biomass
        (3) Reduced growth, less biomass
        (4) Altered ecosystem stress:
            increased or decreased insect
            infestations; increased or decreased
            disease epidemics; and reduced growth,
            less biomass, and altered species
            composition
         Source:  Smith (1974).
 1      Virginia were reported by Brandt and Rhoades (1972, 1973).  Dominant trees in the control area,

 2      a part of the oak-chestnut association of the eastern deciduous forests of eastern North America,
 3      were chestnut oak (Quercusprinus), red oak (Q. mbra), and red maple (Acer rubrum). An

 4      abundance of uniformly distributed saplings and seedlings were visible under the tree canopy,

 5      and herbs appeared in localized areas in canopy openings. Chestnut oak dominated the area, and

 6      the larger trees were 60 to 80 years old. The dusty site was dominated by white oak (Q. alba);

 1      whereas red oak and tulip poplar (Liriodendron tulipifera) were subcodominants. The largest

 8      trees were 100 years old and had necrotic leaves, peeling bark, and appeared to be in generally

 9      poor condition except for tulip poplar (which thrived in localized areas). The site contained a

10      tangled growth of seedlings and shrubs, a few saplings, and a  prevalence of green briar (Smilax

11      spp.) and grape (Vitis spp). The sapling strata in the area was represented by red maple, hickory

12      (Carya spp.), dogwood (Cornus florida), and hop-hornbeam (Ostrya virginiana). Saplings of

13      none of the leading dominant trees were of importance in this stratum. The most obvious form
        June 2003
4-84
DRAFT-DO NOT QUOTE OR CITE

-------
 1      of vegetation in the seedling-shrub stratum, because of their tangled appearance, were dogwood,
 2      hop-hornbeam, redbud (Cercis canadensis), and sugar maple (Acer saccarum).
 3           Crust formation reduced photosynthesis, induced premature leaf fall and destruction of leaf
 4      tissues, inhibited growth of new tissue, and reduced the formation of carbohydrate needed for
 5      normal growth and storage (Brandt and Rhoades, 1973). The authors (Brandt and Rhoades,
 6      1972), citing Odum (1969), also stated that one result of the accumulation of toxic pollutants in
 7      the biosphere (as the result of human activities) is the simplification of both plant and animal
 8      communities. In plant communities, structure is determined by sampling various strata within
 9      the community. Each stratum comprises a particular life form (e.g., herbs, seedlings, saplings,
10      trees). Dust accumulation favored growth of some species and limited others. For example,
11      sugar maple was more abundant in all strata of the dusty site when compared with the control
12      site where it was present only as a seedling.  The growth of tulip poplar, dogwood, hop-
13      hornbeam, black haw (Viburnum prunifolium), and redbud (C. canadensis) appeared to be
14      favored by the dust.  Growth of conifers and acidophiles such as rhododendron (Rhododendron
15      maximum)., however, was limited.  Although dust accumulation began in 1945, the heaviest
16      accumulation occurred between 1967 and 1972 during the time of the study.
17           Changes in community composition were associated closely with changes in the growth of
18      the dominant trees. Decrease in density of seedlings and saplings and in mean basal area, as well
19      as lateral growth of red maple,  chestnut oak, and red oak, occurred in all strata. On the other
20      hand, all of these characteristics increased in tulip poplar, which was a subordinate species
21      before dust accumulation began but had assumed dominance at the time of the study.  Reduction
22      in growth of the dominant trees had apparently given tulip poplar competitive advantage because
23      of its ability to tolerate dust.  Changes in soil  alkalinity occurred because of the heavy deposition
24      of limestone dust; however, the facilities necessary for critical analysis of the soils were not
25      available. From the foregoing, it is obvious that PM physical effects in the vicinity of limestone
26      quarries and processing plants can affect ecosystems.
27           Changes in ecosystem structure resulting from exposures to sea salt were cited previously
28      (Section 4.3.1.1).  The dominance of live oak (Quercus virginiand) as a practically pure stand on
29      Smith Island (Bald Head), NC and along the eastern and southern coast of North Carolina has
30      been explained as due to its tolerance to salt spray. The absence of more inland species is
        June 2003                                 4-85        DRAFT-DO NOT QUOTE OR CITE

-------
 1      attributed to their intolerance to salt spray.  Wells (1939) termed the long-term stabilization of
 2      live oak as "salt spray climax," a new type of climax.
 3
 4      Acidic Deposition
 5           The effects of acidic deposition have been discussed in several previous reports. The 1982
 6      EPA document, Air Quality Criteria for P articulate Matter and Sulfur Oxides, devoted a chapter
 7      to the effects of acidic deposition (U.S. Environmental Protection Agency, 1982). In 1984, EPA
 8      published The Acidic Deposition Phenomenon and Its Effects (Altshuller and Linthurst, 1984),
 9      and, in 1991, NAPAP published the result of its extensive study, Acidic Deposition: State of
10      Science and Technology (Irving, 1991).  The major effects of acidic deposition occur through the
11      soil and are discussed under indirect effects. However, included among the direct responses of
12      forest trees to acidic deposition are increased leaching of nutrients from foliage; accelerated
13      weathering of leaf cuticular surfaces; increased permeability of leaf surfaces to toxic materials,
14      water, and disease agents; and altered reproductive processes (Altshuller and Linthurst, 1984).
15
16      Trace Elements
17           Possible direct responses of trace elements on vegetation result from their deposition and
18      residence on the phyllosphere (i.e., leaf surfaces). Fungi and other microorganisms living on the
19      leaves of trees and other vegetation play an important role in leaf decomposition after litterfall
20      (Miller and McBride, 1999; Jensen, 1974; Millar, 1974). Possible effects of heavy metals on
21      nutrient cycling and their effects on leaf microflora appear not to have been studied.
22           A trace metal must be brought into solution before it can enter into the leaves or bark of
23      vascular plants.  Low solubility limits entry. In those instances when  trace metals are absorbed,
24      they frequently are bound in the leaf tissue and then are lost from the plant when the leaf drops
25      off (Hughes, 1981) are transferred to the litter layer where they can affect litter decomposition,
26      an important source of soil nutrients.  Changes in litter decomposition processes influence
27      nutrient cycling in the soil and limit the supply of essential nutrients.  Both Cotrufo et al. (1995)
28      and Niklinska et al. (1998) point out that heavy metals affect forest litter decomposition.
29      Cotrufo et al. (1995) observed that decomposition of oak leaves containing Fe, Zn, Cu, Cr, Ni,
30      and Pb was influenced strongly during the early stages by metal contamination. Fungal
31      mycelium was significantly less abundant in litter and soil in contaminated sites when compared

        June 2003                                  4-86        DRAFT-DO NOT QUOTE OR CITE

-------
 1      with control sites.  Nikliiiska et al. (1998) stated that toxic effects of heavy metals on soil
 2      respiration rate have been reported by many scientists, and that, in polluted environments, this
 3      results in accumulation of undecomposed organic matter.  However, they state that results of
 4      experiments should identify the most important "natural" factors affecting soil/litter sensitivity
 5      because the effects of heavy metals on respiration rates depend on the dose of heavy metals, the
 6      type of litter, types of metals deposited,  and the storage time before respiration tests are made.
 7           Trace metals, particularly heavy metals (e.g., cadmium, copper, lead, chromium, mercury,
 8      nickel, zinc) have the greatest potential for influencing forest growth (Smith, 1991).
 9      Experimental data indicate that the broadest spectrum of growth suppression of foliar microflora
10      resulted from iron, aluminum, and zinc.  These three metals also inhibited spore formation,  as
11      did cadmium, chromium, manganese, and nickel  (see Smith, 1990c).  In the field, the greatest
12      injury occurs from pollution near mining, smelting, and other industrial sources (Ormrod, 1984).
13      Direct metal phytotoxicity can occur only if the metal can move from the surface into the leaf or
14      directly from the soil into the root.
15
16      Organic Compounds
17           Primary and  secondary organic compounds formed in the atmosphere, the effects of some
18      of which are discussed below, have been referred to under the following terms: toxic substances,
19      pesticides, hazardous air pollutants (HAPS), air toxics, semivolatile organic compounds (SOCs),
20      and persistent organic pollutants (POPS).  Again, it should be remembered that chemical
21      substances denoted by such headings are not criteria air pollutants controlled by the National
22      Ambient Air Quality Standards under Section 109 of the Clean Air Act (CAA), but rather are
23      controlled under Sect. 112, Hazardous Air Pollutants (U.S. Code, 1991). Their possible effects
24      on humans and ecosystems are discussed in a number of government documents and in many
25      other publications. They are mentioned here because many of the chemical compounds are
26      partitioned between gas and particle phases in the atmosphere. As particles, they can become
27      airborne, be distributed over a wide area, and affect remote ecosystems. Some of the chemical
28      compounds are of concern because they may reach toxic levels in food  chains of both animals
29      and humans; whereas others tend to decrease or maintain the  same toxicity as they move through
30      the food chain.  Some examples of movement through  food chains are provided below.
        June 2003                                 4-87        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Many chemical compounds from a variety of anthropogenic sources are released into the
 2      ambient air (see Section 4.2.1). In the atmosphere, the emitted compounds initially go through a
 3      mixing process, and the airborne particles then are distributed over a wide area and ultimately
 4      deposited on ecosystem components. Atmospheric deposition of poly chlorinated dibenzo-p-
 5      dioxins and dibenzofurans (PCDD/Fs), as an example, can be divided into three different forms:
 6      (1) dry gaseous, (2) dry particle-bound, and (3) wet deposition.  Dry particle-bound deposition
 7      occurs when the PM containing the pollutant is deposited on the plant surface; whereas wet
 8      deposition ranges from hail to rain to fog and dew fall (McLachlan, 1996b).
 9           Human exposure to PCDD/Fs has been demonstrated to be caused almost exclusively by
10      the ingestion of animal fat from fish, meat, and dairy products. Almost half of human exposure
11      to PCDD/Fs is caused by consumption of beef and dairy products (McLachlan, 1996b). Cattle
12      obtain most of their PCCD/Fs though grass.  Therefore, the grass - cattle - milk/beef pathway is
13      critical for human exposure.  It has been shown that root uptake/translocation is an insignificant
14      pathway of PCDD/Fs to aerial plant parts.  Wet and dry particle deposition are the most
15      important for the accumulation of the higher chlorinated cogeners in vegetation. The persistence
16      of PCDD/Fs in plants has not been investigated extensively;  however, biodegradation probably
17      does not occur in that these compounds are found primarily in the lipophilic cuticle and are very
18      resistant to microbial degradation (McLachlan, 1996b). Feed contaminated with soil containing
19      the pollutant can be another source of exposure of beef and dairy cattle, as well as chickens. The
20      PCDD/Fs are near a steady state in milk cows and laying hens; however, animals raised for meat
21      production (such as beef cattle and pigs) may accumulate them. The beef cattle and pigs cannot
22      excrete the contaminants in a lipid-rich matrix such as milk or eggs.  Thus, all of the PCDD/Fs,
23      ingested are stored in the body. In agricultural food chains, there is a biodilution of PCDD/Fs,
24      with the fugacity decreasing by up to three orders of magnitude between the air and cows milk
25      (McLachlan, 1996b). Fiirst et al.  (1993), based on surveys to determine the factors that influence
26      the presence of PCDD/PCDF in cows milk, earlier concluded that regardless of which pathway,
27      soil - grass - cow or air - grass - cow, it was the congener of the chemical that was most
28      important.
29           Chlorinated persistent organic pollutants (POPS), such as PCBs, PCDFs, and PCDDs, can
30      be transported as particles through the atmosphere from industrial and agricultural sources; be
31      brought down via wet and dry deposition in remote regions, such as the Arctic; and have been

        June 2003                                4-88       DRAFT-DO NOT QUOTE OR CITE

-------
 1      detected in all levels of the Arctic food chain (Oehme et al., 1995). High concentrations of PCB
 2      (1 to 10 ppm) were found in seals, but the concentrations increased to 10 to 100 ppm in polar
 3      bears. The polar bear is the top predator in the Arctic and feeds preferentially on ringed seals
 4      and, to a lesser extent, on other seal species. Bioconcentration factors of organochlorines in the
 5      Arctic food web, reaching 107 for fish and seals, are biomagnified in polar bears (Oehme et al.,
 6      1995). Poly chlorinated dibenzo-p-dioxins and poly chlorinated dibenzofurans have also been
 7      found in seals (Oehme et al., 1995).  Milk taken from anaesthetized polar bears was also found
 8      to contain PCDD/PCDF. Very little is known regarding the intake of milk by polar bear cubs.
 9      However, estimates of the intake of milk containing detectable levels of PCDD/PCDF and PCB
10      and the additional consumption of seal blubber confirm that these pollutants are passed on to the
11      next generation (Oehme et al.,  1995).
12           Section 112 of the CAA provides the  legislative basis for U.S. hazardous air pollutant
13      (HAP) programs. In response to mounting  evidence that air pollution contributes to water
14      pollution, Congress included Section 112m (Atmospheric Deposition to Great Lakes and Coastal
15      Waters) in the 1990 CAA Amendments that direct the EPA to establish a research program on
16      atmospheric deposition of HAPS to the "Great Waters."
17           Actions taken by EPA and others to evaluate and control sources of Great Waters
18      pollutants of concern appear to have positively affected trends in pollutant concentrations
19      measured in air,  sediment, and biota.  Details concerning these effects may be found in
20      "Deposition of Air Pollutants to the Great Waters," Third Report to Congress (U.S.
21      Environmental Protection Agency, 2000a).  The Third Report (EPA-453/R-00-005, June 2000),
22      like the First and Second Reports to Congress, focuses on 15 pollutants of concern, including
23      pesticides, metal compounds, chlorinated organic compounds, and nitrogen compounds. The
24      new scientific information in the Third Report supports and builds on three broad conclusions
25      presented in the  previous two EPA Reports to Congress:
26       (1)   Atmospheric deposition from human activities can be a significant contributor of toxic
              chemicals and nitrogen compounds to the Great Waters.  The relative importance of
              atmospheric loading  for a particular  chemical in  a water body depends on many factors
              (e.g., characteristics of the water body, properties of the chemical, and the kind and
              amount of atmospheric deposition versus water discharges).
        June 2003                                 4-89        DRAFT-DO NOT QUOTE OR CITE

-------
 1       (2)   A plausible link exists between emissions of toxic pollutants of concern into the air
              above the Great Waters; the deposition of these pollutants (and their transformation
              products); and the concentrations of these pollutants found in the water, sediments, and
              biota, especially fish and shellfish.  For mercury, fate and transport modeling and
              exposure assessments predict that the anthropogenic contribution to the total amount of
              methylmercury in fish is, in part, the result of anthropogenic mercury releases from
              industrial and combustion sources increasing mercury body burdens (i.e.,
              concentrations) in fish. Also, the consumption offish is the dominant pathway of
              exposure to methylmercury for fish-consuming humans and wildlife.  However, what is
              known about each stage of this process varies with each pollutant (for instance, the
              chemical species of the emissions and its transformation in the atmosphere).
 2       (3)   Airborne emissions from local as well as distant sources, from both within and outside
              the United States, contribute pollutant loadings to waters through atmospheric
              deposition. Determining the relative roles of particular sources — local, regional,
              national, and possibly global, as well  as anthropogenic, natural, and reemission of
              pollutants — contributing to specific water bodies is complex, requiring  careful
              monitoring, atmospheric modeling, and other analytical techniques.
 3
 4      Response to Indirect Effects of PM
 5           The presence of PM in the atmosphere directly affects vegetation following physical
 6      contact with foliar surfaces (as discussed above), but in many cases the more significant effects
 7      are indirect.  These effects may be mediated by suspended PM (i.e., through effects on radiation
 8      and climate) and by particles that pass through vegetative canopies to reach the soil. Effects
 9      mediated through the atmosphere are considered briefly below and in greater detail later, under
10      Section 4.5.
11           The major indirect plant responses are  chiefly soil-mediated and depend primarily on the
12      chemical composition of the individual stressors deposited in PM.  The chemical stressors must
13      be bioavailable in order to produce an effect. The effects  of exposures may result in changes in
14      biota patterns and in chemical/physical soil conditions that affect ecological processes, such as
15      nutrient cycling and uptake by plants.
        June 2003                                 4-90        DRAFT-DO NOT QUOTE OR CITE

-------
 1           The soil environment, composed of mineral and organic matter, water, air, and a vast array
 2      of bacteria, fungi, algae, actinomycetes, protozoa, nematodes, and arthropods, is one of the most
 3      dynamic sites of biological interactions in nature (Wall and Moore, 1999; Alexander, 1977).
 4      The quantity of organisms in soils varies by locality. Bacteria and fungi are usually most
 5      abundant in the rhizosphere, the soil around plant roots that all mineral nutrients must pass
 6      through.  Bacteria and fungi benefit from the nutrients in the root exudates (chiefly sugars) in the
 7      soil and,  in turn, they play an essential role by making mineral nutrients available for plant
 8      uptake (Wall and Moore, 1999; Rovira and Davey, 1974). Their activities create chemical and
 9      biological changes in the rhizosphere by decomposing organic matter and making inorganic
10      minerals available for plant uptake. Bacteria are essential in the nitrogen and sulfur cycles and
11      make these elements available for plant uptake and growth (see Section 4.3.3). Fungi are
12      directly essential to plant growth.  Attracted to the roots by the exudates, they develop
13      mycorrhizae, a mutualistic,  symbiotic relationship, that is integral in the uptake of the mineral
14      nutrients (Allen, 1991).  The impact in ecosystems of PM, particularly nitrates, sulfates, and
15      metals, is determined by their effects  on the growth of the bacteria involved in nutrient cycling
16      and the mycorrhizal fungi involved in plant nutrient uptake.
17
18           Atmospheric Turbidity: Effects on Direct Beam and Photosynthetically Active
19      Radiation.  Photosynthetic processes underlie the contribution of vegetative surfaces to nutrient
20      and energy cycling. The characteristics and net  receipts of environmental radiation determine
21      the rates  of both photosynthesis and the heat-driven process of water cycling. Atmospheric
22      turbidity due to particulate loading can substantially alter the characteristics and net receipts of
23      solar radiation.  One measure of atmospheric turbidity, Linke's turbidity factor, T, can be
24      derived as a direct function  of light extinction by solid particles. It is defined as the ratio of the
25      total extinction coefficient and the extinction due exclusively to gases:
26
27                                T = o/og =  1 + wow/og + sos/og                             (4-6)
28
29      where s and w are the relative concentrations of dust and water vapor in the atmosphere, and os
30      and ow are the wavelength-dependent scattering  coefficients for solid, dry particles  and water
31      vapor.  The scattering coefficients are in units of inverse distance, such as km"1 (Rosenberg,

        June 2003                                  4-91        DRAFT-DO NOT QUOTE OR CITE

-------
 1      1974).  According to this expression, a clean atmosphere would have a turbidity value of 1.
 2      Given that turbidity and visibility are both functions of light scattering, the trends in and physical
 3      processes underlying reduced visibility discussed in Section 4.3 are directly relevant to the
 4      discussion of radiative effects on vegetation due to particulate matter.
 5           Turbidity, as defined above, describes the degree of scattering occurring in the atmosphere
 6      due to particles and gases.  Total, particle-based extinction, however, is the sum of both
 7      scattering and absorption.  Absorption of short-wavelength solar radiation reduces the amount of
 8      radiation reaching the Earth's surface and leads to atmospheric heating. If the absorbing
 9      particles re-radiate in the infrared range,  some of this energy is lost as long-wave re-radiation to
10      space.  The balance of this energy is captured at the surface as down-welling infrared radiation.
11      Canopy temperature and transpirational water use by vegetation are particularly sensitive to
12      long-wave, infrared radiation. Atmospheric heating by particles reduces vertical temperature
13      gradients, potentially reducing the intensity of atmospheric turbulent  mixing.  The magnitude of
14      such potential effects on turbulent transport within canopies remains unknown although the
15      damping of eddy transport might inhibit canopy gas exchange. Suppressed tropospheric mixing
16      could also intensify local temperature inversions and increase the severity of pollution episodes
17      (Pueschel, 1993) with direct inhibitory effects on photosynthetic processes.
18           Atmospheric turbidity increases the intensity of diffuse (sky) radiation (Hoyt, 1978).  In a
19      clear atmosphere, diffuse radiation may be on the order of 10% of total solar radiation
20      (Choudhury, 1987). However, in highly  turbid, humid conditions,  this fraction may increase, to
21      as much as 100% of the incident solar intensity in extreme cases. The direct-to-diffuse-radiation
22      ratio is highest at solar noon and lowest near dawn or dusk when the path length through the
23      atmosphere is longest.  The wavelength dependence of particle scattering induces an enrichment
24      of PAR with respect to total or direct beam radiation.
25           Aerosols produced by incomplete combustion, from forest fires to specifically
26      anthropogenic processes such as diesel fuel combustion, contain significant fractions of black
27      carbon which absorbs across the solar and terrestrial radiation spectra. The presence  of
28      absorbing aerosols reduces the ratio of photosynthetically active radiation to total radiation
29      received at the surface, potentially reducing photosynthetic water uptake efficiency. The net
30      effect of aerosol absorption on the surface depends on the relative magnitudes of the particulate
31      absorption coefficients  in the visible and infrared area and on the albedo of the Earth's surface.

        June 2003                                  4-92         DRAFT-DO NOT QUOTE OR CITE

-------
 1           The greater effect of particulate loading on visibility and turbidity is due to scattering.
 2      Non-absorbing, scattering aerosols raise the overall albedo of the atmosphere and reduce the
 3      amount of radiation reaching the surface by the amount reflected or scattered back into space.
 4      Analysis of data collected by a global network of thermopile pyranometers operated by the
 5      World Meteorological Organization show a 50-year global trend of a 2.7% per decade reduction
 6      in the amount of solar radiation reaching the Earth's surface. This has been associated with an
 7      increasing global albedo caused by an increasing abundance of atmospheric particles.
 8      By evaluating the WMO data set with four different approaches to the statistical analyses,
 9      Stanhill and Cohen (2001) have estimated that average global solar radiation receipts have
10      declined by 20 Wm"2 since 1958.  Examples of individual measurement sites include Barrow,
11      Alaska (71 °N), where the average solar irradiance from  1963 - 1993 was 100.9 W m"2 and the
12      estimated linear trend was -0.23% per year; and Jerusalem, Israel (32 °N), where the average
13      solar irradiance from  1954 - 1994 was 244.2 W m"2 and the estimated linear trend was -0.37%
14      per year.
15
16           Increased Turbidity and Altered Radiative Flux: Effects on Vegetative Processes. In a
17      detailed canopy-response model (e.g., Choudhury, 1987), radiation is considered in its direct and
18      diffuse components.  Foliar interception by canopy elements is considered for both up- and
19      down-welling radiation (a two-stream approximation). In this case, the effect of atmospheric
20      PM on turbidity influences canopy processes both by radiation attenuation and by influencing
21      the efficiency of radiation interception throughout the canopy through conversion of direct to
22      diffuse radiation (Hoyt, 1978). Diffuse radiation is more uniformly distributed throughout the
23      canopy and increases canopy photosynthetic productivity by distributing radiation to lower
24      leaves. The treatment of downwelling direct-beam radiation in the two-stream approach remains
25      an elaboration of the simplified Beer's Law analogy with solar angle, leaf area distribution, and
26      orientation individually parameterized (Choudhury, 1987). Diffuse downwelling radiation is a
27      function of diffuse and direct radiation at the top of the canopy and penetration within the
28      canopy according to cumulative leaf area density and foliage orientation.  Diffuse upwelling
29      radiation results from scattering and reflectance of both direct and diffuse downwelling radiation
30      within the canopy and by the soil.
        June 2003                                 4-93        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Rochette et al. (1996) conducted simultaneous measurements of radiation and water use
 2      efficiencies by maize and found that, in the absence of water stress and with adequate
 3      fertilization, 90% of all variation in crop net photosynthesis (P-n) could be explained by
 4      variations in PAR.  Alternatively, an evaluation of the available experimental literature and
 5      statistics on crop yields by Stanhill and Cohen (2001) indicate that plant productivity is more
 6      affected by changes in evapo-transpiration induced by changes in the amount of solar radiation
 7      plants receive than by changes in the amount of PAR plants receive.
 8           The enrichment in PAR present in diffuse radiation appears,  however, to offset a portion of
 9      the effect of an increased atmospheric albedo due to atmospheric particles. An observational and
10      theoretical study by Bange et al. (1997) of the level of radiation use efficiency (RUE) of
11      sunflowers indicated a degree of compensation for reduced incident radiation by a proportional
12      increase in diffuse radiation.  Variables measured by Bange et al. (1997) included biomass,
13      phenology, leaf area, canopy light extinction, grain size, and harvest index. Crops subject to
14      reduced direct beam/increased diffuse radiation produced biomass, phenology, leaf area and
15      canopy light extinction at leaves similar to unshaded crops, but yielded smaller grains and a
16      lower harvest index. RUE was also seen to improve for soybeans  and maize with a proportional
17      increase in diffuse radiation with respect to direct beam (Sinclair et al., 1992; Healey et al.,
18      1998) although the effect on harvest index was not indicated. Following a comparison of the
19      relative efficiencies of canopy photosynthesis to diffuse and direct PAR for a Scots pine forest,
20      an aspen forest, a mixed deciduous forest, a tall grass prairie and a winter wheat crop, Gu et al.
21      (2002) concluded (1) diffuse radiation over direct radiation results in higher light use efficiencies
22      by plant canopies; (2) diffuse radiation has much less tendency to cause canopy photosynthetic
23      saturation; (3) the advantages of diffuse radiation over direct radiation increase with radiation
24      level; (4) temperature as well as vapor pressure deficit can cause different responses in diffuse
25      and direct canopy photosynthesis, indicating that their effects on terrestrial ecosystem carbon
26      assimilation may depend upon radiation regimes, thus sky conditions.
27           The potentially significant effect of regional haze on the yield of crops because of
28      reduction in solar radiation has been examined by Chameides et al. (1999). Using a case study
29      approach, Chameides et al. (1999), studied the effects of regional haze on  crop production in
30      China where regional haze is especially severe. A rudimentary assessment of the direct effect of
31      atmospheric aerosols on agriculture suggests that yields of approximately  70% of crops are being

        June 2003                                 4-94        DRAFT-DO NOT QUOTE OR CITE

-------
 1      depressed by at least 3 to 5% by regional scale air pollution and its associated haze (Chameides
 2      etal., 1999).
 3
 4           Effects of Nitrogen Deposition. Nitrogen is required by all organisms.  It is a major
 5      constituent of the nucleic acids that determine the genetic character of all living things and the
 6      enzyme proteins that drive the metabolic machinery of every living cell (Galloway, 1998;
 7      Galloway and Cowling, 2002). Though nitrogen composes 80% of the total mass of the Earth's
 8      atmosphere, it is not biologically available. Nitrogen fixation is accomplished in nature by
 9      certain unique organisms that have developed the capability of converting N2 to biologically
10      active reduced forms of nitrogen such as ammonia, amines, and amino acids which are the
11      structural constituents  of proteins and nucleic acids (Galloway and Cowling, 2002).
12           Nitrogen has long been recognized as the nutrient most important for plant growth. It is of
13      overriding importance in plant metabolism and, to a large extent, governs the utilization of
14      phosphorus, potassium, and other nutrients. Most of the nitrogen in soils is associated with
15      organic matter. Typically, the availability of nitrogen via the nitrogen cycle controls net primary
16      productivity, and possibly, the decomposition rate of plant litter. Photosynthesis is influenced by
17      nitrogen uptake in that ca., 75%  of the nitrogen in a plant leaf is used during the process of
18      photosynthesis.  The nitrogen-photosynthesis relationship is, therefore, critical to the growth of
19      trees and other plants (Chapin et al., 1987). Plants usually obtain nitrogen directly from the soil
20      through their roots by absorbing NH4+ or NO3", or it is formed by symbiotic organisms (bacteria,
21      blue-green algae) in the roots.
22           Because nitrogen is not readily available and is usually in short supply, it is the chief
23      element in agricultural fertilizers. The realization of the importance of nitrogen in crop
24      production resulted in a search for natural nitrogen such as guano and nitrate deposits. The
25      invention of the Haber-Bosch process in 1913 made reactive nitrogen (Nr) available for use in
26      food production, and more than half of the food eaten by peoples of the world today is produced
27      using fertilizer produced by this  process (Galloway and Cowling, 2002).
28           Nitrogen in nature may be  divided into two groups: nonreactive (N2) and reactive (Nr).
29      Reactive Nr includes all biologically, photochemically,  and radioactively active nitrogen
30      compounds in the Earth's  atmosphere and biosphere (Galloway et al., 2003). Among those
31      included are: the inorganic reduced forms of nitrogen (e.g., ammonia [NH3] and ammonium

        June 2003                                 4-95        DRAFT-DO NOT QUOTE OR CITE

-------
 1      [NH4+]), inorganic oxidized forms (e.g., nitrogen oxide [NOX], nitric acid [HNO3], nitrous oxide
 2      [N2O], and nitrate [NO3"]) , and organic compounds (e.g., urea, amine, proteins, and nucleic
 3      acids)]).
 4           Food production continues to account for most of the newly Nr created. However, since
 5      approximately 1965 the magnitude of Nr created by humans began to exceed natural terrestrial
 6      creation of Nr and its conversion back to N2by denitrification.  The overall increase in global Nr
 7      is the result of three main causes:  (1) widespread cultivation of legumes, rice and other crops
 8      that promote conversion of N2 to organic nitrogen through biological nitrogen fixation;
 9      (2) combustion of fossil fuels, which converts both atmospheric N2 and fossil nitrogen to
10      reactive NOX; and (3) the Haber-Bosch process, which converts nonreactive NH3 to sustain food
11      production and some industrial activities (Galloway and Cowling,  2002; Galloway et al., 2003).
12           Reactive nitrogen is now accumulating in the environment on all spatial scales - local,
13      regional and global (2002:Galloway and Cowling, 2002; Galloway et al. 2003). As a result, Nr
14      is accumulating in various environmental reservoirs, e.g., the atmosphere, soils and waters
15      (Galloway and Cowling, 2002). The accumulation of Nr in the environment has effects on
16      humans and ecosystems Rabelais,  2002; van Egmond et al., 2002;  Galloway, 1998).
17           Large uncertainties regarding the rates of Nr accumulation in the various reservoirs limits
18      our ability to determine the temporal and spatial distribution of environmental effects. These
19      uncertainties are of great significance because of the sequential nature of Nr on environmental
20      processes. (Galloway and Cowling, 2002).  The sequence of transfers, transformations, and
21      environmental effects is referred to as the nitrogen cascade (Figure 4-15; Galloway and Cowling,
22      2002; Galloway et al, 2003).  A single atom of new NHX or NOX can alter a wide array of
23      biogeochemical processes and exchanges among environmental reservoirs.
24           The results of the Nr cascade in the global system and the wide variety of changes in the
25      nitrogen cycle are both beneficial and detrimental to humans and to ecosystems (Galloway and
26      Cowling, 2002; Galloway et al., 2003).  Though the synthetic fertilizers used in cultivation and
27      the cultivation-induced bacterial nitrogen fertilization (BNF) sustain a large portion of the
28      world's population, there are consequences: (1) the wide dispersal of Nr by  hydrological and
29      atmospheric transport; (2) the accumulation of Nr in the environment because the rates of its
30      creation are greater than the rates of its removal through denitrification to nonreactive N2; (3) Nr
31      creation and accumulation is projected to continue to increase in the future as per capita use of

        June 2003                                 4-96        DRAFT-DO NOT QUOTE OR CITE

-------

NOX
Energy
Production
|
1
Ozi
Efff
t
AAAAAAAAAAA
A
                                                    Atmosphere
                Food
              Production
               People
            (Food; Fiber)
          Human Activities
                                                                   Terrestrial
                                                                  Ecosystems
NH
         The Nitrogen
            Cascade
          Indicates denitrification potential
                                                                               N20
      Figure 4-15.  Illustration of the nitrogen cascade showing the movement of the human-
                  produced reactive nitrogen (Nr) as it cycles through the various
                  environmental reservoirs in the atmosphere, terrestrial ecosystems, and
                  acquatic ecosystems.

      Modified from Galloway and Cowling (2002).
1     resources by human populations increases; and (4) Nr accumulation contributes to many
2     contemporary environmental problems (Galloway et al., 2003)
3          Among the contemporary environmental problems are the following:

4      • increases in Nr lead to production of tropospheric ozone and aerosols and the associated
         human health problems (Wolfe and Patz, 2002);
      June 2003
              4-97
DRAFT-DO NOT QUOTE OR CITE

-------
 1       •  productivity increases in forests and grasslands and then decreases wherever atmospheric
           Nr deposition increases significantly and critical thresholds are exceeded; Nr additions
           probably also decrease biodiversity in many natural habitats (Aber et al., 1995);
 2       •  reactive nitrogen in association with sulfur is responsible for acidification and loss of
           biodiversity in lakes and streams in many regions of the world (Vitousek et al., 1997);
 3       •  reactive nitrogen is responsible for eutrophication, hypoxia, loss of biodiversity, and habitat
           degradation in coastal ecosystems. It is now considered the biggest pollution problem in
           coastal waters (Rabelais, 2002);
 4       •  reactive nitrogen contributes to global climate change and stratospheric ozone depletion,
           both of which have an effect on the health of humans and ecosystems (Cowling et al.,
           1998).
 5           The effect of increasing nitrogen inputs (e.g., NOX, nitrates, nitric acid) on the nitrogen
 6      cycle in forests, wetlands, and aquatic ecosystems is discussed in detail elsewhere (U.S.
 7      Environmental Protection Agency, 1993, 1997'a; Garner, 1994; World Health Organization,
 8      1997).
 9           The deposition of nitrogen in the United States from human activity has doubled between
10      1961 and 1997 due mainly to the use of inorganic nitrogen fertilizers and the emissions of
11      nitrogen oxides (NOX) from fossil fuel emissions with the largest increase occurring in the 1960s
12      and 1970s (Howarth et al.,  2002). Among the most important effects of chronic nitrogen
13      deposition are changes in the composition of plant communities, disruptions in nutrient cycling,
14      increased emissions from soil of nitrogenous greenhouse gases and accumulation of nitrogen
15      compounds resulting in the enhanced availability of nitrate or ammonium, the soil-mediated
16      effects of acidification, and increased susceptibility to stress factors (Fenn et al., 1998; Bobbink
17      et al., 1998). A major concern is "nitrogen saturation," the result of the atmospheric deposition
18      of large amounts of particulate nitrates. Nitrogen saturation results when additions to soil
19      background nitrogen (nitrogen loading) exceeds the capacity of plants and soil microorganisms
20      to utilize and retain nitrogen (Aber et al., 1989,  1998; Garner, 1994; U.S. Environmental
21      Protection Agency, 1993).  Under these circumstances, disruptions of ecosystem functioning
22      may result (Hornung and Langan, 1999).
23           Possible ecosystem responses to nitrate saturation,  as postulated by Aber and coworkers
24      (Aber et al., 1989), include (1) a permanent increase in foliar nitrogen and reduced foliar

        June 2003                                 4-98        DRAFT-DO NOT QUOTE OR CITE

-------
 1      phosphorus and lignin caused by the lower availability of carbon, phosphorus, and water;
 2      (2) reduced productivity in conifer stands because of disruptions of physiological function;
 3      (3) decreased root biomass and increased nitrification and nitrate leaching; and (4) reduced soil
 4      fertility, resulting from increased cation leaching, increased nitrate and aluminum concentrations
 5      in streams, and decreased water quality. Saturation implies that some resource other than
 6      nitrogen is limiting biotic function.
 7           Water and phosphorus for plants and carbon for microorganisms are the resources most
 8      likely to be the secondary limiting factors  (Aber et al., 1989).  The appearance of nitrogen in soil
 9      solution is an early symptom of excess nitrogen.  In the final stage, disruption of forest structure
10      becomes visible (Garner, 1994).
11           Changes in nitrogen supply can have a considerable effect on an ecosystem's nutrient
12      balance (Waring, 1987). Large chronic additions of nitrogen influence normal nutrient cycling
13      and alter many plant and soil processes involved in nitrogen cycling (Aber et al., 1989).
14      Among the processes affected are (1) plant uptake and allocation, (2) litter production,
15      (3) immobilization (includes ammonification [the release of ammonia] and nitrificatrion
16      [conversion of ammonia to nitrate during decay of litter and soil organic matter]), and (4) nitrate
17      leaching and trace gas emissions (Figure 4-16; Aber et al.,  1989; Garner 1994).
18           Subsequent studies have shown that, although there was an increase in nitrogen
19      mineralization initially (i.e., the conversion of soil organic  matter to nitrogen in available form
20      [see item 3 above]), nitrogen mineralization rates were reduced under nitrogen-enriched
21      conditions.  Aber et al.(1998) hypothesize that mycorrhizal assimilation and exudation, using
22      photosynthate from the host plant as the carbons source, is  the dominant process involved in
23      immobilization of nitrogen. In addition, studies suggested  that soil microbial communities
24      change from predominantly fungal (mycorrhizal) communities to those dominated by bacteria
25      during  saturation (Aber et al., 1998).
26           The growth of most forests in North America is limited by the nitrogen supply. Severe
27      symptoms of nitrogen saturation, however, have been observed in high-elevation,  nonaggrading
28      spruce-fir ecosystems in the Appalachian Mountains, as well as in the eastern hardwood
29      watersheds at Fernow Experimental Forest near Parsons, WV.  Mixed conifer forests and
30      chaparral watersheds with high smog exposure in the Los Angeles Air Basin also are nitrogen
31      saturated and exhibit the highest stream water NO3" concentrations for wildlands in North

        June 2003                                  4-99        DRAFT-DO NOT QUOTE OR CITE

-------
                      Deposition
f
Plant
Utilization

Photosynthesis


X
Animal
Proteins
                                                                    Process altered by
                                                                    nitrogen saturation
       Figure 4-16.  Nitrogen cycle (dotted lines indicate processes altered by nitrogen saturation).
       Source:  Garner (1994).
1
2
3
4
5
6
7
America (Bytnerowicz and Fenn, 1996; Fenn et al., 1998). Forests in southern California, the
southwestern Sierra Nevada in Central California, and the Front Range in northern Colorado
have all been exposed to highly elevated nitrogen deposition, and nitrogen saturated watersheds
have been reported in the above mentioned areas. Annual nitrogen additions through deposition
(6-11 kg ha"1 y"Jas through fall) in the southwestern Sierra Nevada are similar to nitrogen storage
(4 kg ha"1 y"1) in vegetation growth increment of western forests suggesting that current nitrogen
deposition rates may be near the assimilation capacity of the overstory vegetation. Ongoing
urban expansion will increase the potential for nitrogen saturation of forests from urban sources
       June 2003
                                          4-100
DRAFT-DO NOT QUOTE OR CITE

-------
 1      (e.g., Salt Lake City, Seattle, Tucson, Denver, central and southern California) unless there are
 2      improved emission controls (Fenn et al., 1998).
 3           Atmospherically deposited nitrogen also can act as a fertilizer in soil low in nitrogen. Not
 4      all plants, however, are capable of utilizing extra nitrogen as plants vary in their ability to absorb
 5      ammonium and nitrate (Chapin, et al., 1987). Inputs of nitrogen to natural ecosystems that
 6      alleviate deficiencies and increase growth of some plants can alter competitive relationships and
 7      alter species composition and diversity (Ellenberg, 1987; Kenk and Fischer, 1988; U.S.
 8      Environmental Protection Agency, 1993).
 9           Not all forest ecosystems react in the same manner to nitrogen deposition.  High-elevation
10      alpine watersheds in the Colorado Front Range (Bowman, 2000) and a deciduous forest in
11      Ontario, Canada, also are naturally saturated even though nitrogen deposition has been moderate
12      (~ 8 kg ha"1 y"1).  The nitrogen saturated forests in North America, including estimated inputs and
13      outputs, are shown in Table 4-14 (Fenn et al., 1998). The Harvard Forest hardwood stand in
14      Massachusetts, however, has absorbed > 900 kg N/ha without significant NO3" leaching during a
15      nitrogen amendment study  of 8 years (Table 4-14;  Fenn et al.,1998).  Nitrate leaching losses
16      were high, on the other hand, in the Harvard Forest pine sites suggesting that deciduous forest
17      may have a greater capacity for nitrogen retention. During the 8-year experimental study
18      (1988-96), nitrate leaching  was observed in the pine stand after the first year (1989) in the high
19      nitrogen plots. Further increase was observed in 1995 and  1996, while the hardwood stand
20      showed no significant increases in nitrate leaching until 1996. The sharp contrast in response of
21      pine and hardwood stands indicates that the mosaic of community types across the landscape
22      must be considered when determining regional scale response to nitrogen deposition (Magill
23      et al., 2002).  Johnson et al. (1991a) reported that measurements showing the leaching of nitrates
24      and aluminum (Al+3) from  high elevation forests in the Great Smoky Mountains indicate that
25      these forests have reached saturation.
26           Because the competitive equilibrium of plants in any  community is finely balanced, the
27      alteration of one of a number of environmental parameters, (e.g., continued nitrogen additions)
28      can change the vegetation structure of an ecosystem (Bobbink, 1998; Skeffington and Wilson,
29      1988).  Increases in soil nitrogen play  a selective role. When nitrogen becomes more readily
30      available, plants adapted to living in an environment of low nitrogen availability will be replaced
31      by plants capable of using increased nitrogen because they  have a competitive advantage.

        June 2003                                 4-101        DRAFT-DO NOT QUOTE OR CITE

-------
g1
to
o
o
OJ







f>
o
to
INCLUDING ESTIMATED N INPUTS AND OUTPUTS
Location
Adirondack Mts. northeastern New York
Catskill Mts., southeastern New York
Turkey Lakes Watershed,
Ontario, Canada
Whitetop Mt, southwestern Virginia
Fernow, West Virginia
Great Smoky Mts. National Park,
Tennessee
Great Smoky Mts. National Park,
Becking Site, North Carolina
Great Smokey Mts. National Park,
Tower Site, North Carolina
Front Range, Colorado
San Dimas, San Gabriel Mts.,
Forest Type
Northern hardwoods or hardwood/
conifer mix
Mainly hardwood; some eastern
hemlock
Sugar maple and yellow birch
Red spruce
Mixed hardwood
American beech
Red spruce
Red spruce
Alpine tundra, subalpine conifer
Chapparral and grasslands
Elevation
(in)
396-661
335-675
350-400
1650
735-870
1600
1800
1740
3000-4000
580-1080
N Input
(kg ha ! year !)
9.3"
10.2"
7.0-7.7
(as throughfall)
32C
15-20
3.1"
10.3d
26.6
7.5-8.0
23.3e
N Output
(kg ha1 year !)
Stage 1 N lossb
Stage 1 and
2 N lossb
17.9-23.6
47°
6.1
2.9
19.2
20.3
7.5
0.04-19.4
Reference
Driscoll and Van Dreason (1993)
Stoddard (1994)
Foster et al. (1989); Johnson and Lindberg
(1992a)
Joslin and Wolfe (1992); Joslin et al. (1992)
Gilliam et al. (1996); Peterjohn et al. (1996)
Johnson and Lindberg (1992b)
Johnson etal. (199 la)
Johnson etal. (199 la)
Williams et al. (1996)
Rigganetal. (1985)
 H
 O
 O

 o
 H
O
 O
 H
 W
 O
 O
 HH
 H
 W
Camp Paivika, San Bernadino Mts.,
 southern California

Klamath Mts, northern California

Thompson Forest, Cascade Mts.,
Washington
                                                     Mixed conifer


                                                     Western coniferous

                                                     Red alder
1600


NA

220
                                                                                                      30
Mainly geologic8

4.7 plus > 100 as
   N, fixation
7-26f           Fenn etal. (1996)


NA8            Dahlgren (1994)

38.9            Johnson and Lindberg (1992b)
"Estimated total N deposition from wet deposition data is from Driscoll et al. (1991) for the Adirondack?, and from Stoddard and Murdoch (1991) for the Catskills.  Total deposition
was estimated based on the wet deposition/total N deposition ratio (0.56) at Huntingdon Forest in the Adirondacks (Johnson and Lindberg, 1992b). Nitrogen deposition can be higher
in some areas, especially at high-elevation sites such as Whiteface Mountain (15.9 kg ha"1 year"1; Johnson and Lindberg, 1992b).
""Stage 1 and 2 of N loss according to the watershed conceptual model of Stoddard (1994). Nitrogen discharge (kg ha"1 year"1) data are not available; only stream water NO3"
concentration trend data were collected.
"Values appear high compared to other sites, especially N leaching losses. Joslin and Wolfe (1992) concede that "there is considerable uncertainty associated with the estimates of
atmospheric deposition and leaching fluxes." However, elevated NO3" concentrations  in soil solution, and lack of a growth response to N fertilization (Joslin and Wolfe, 1994)
support the hypothesis that the forest at Whitetop Mountain is N saturated.
•"Estimated total N deposition from throughfall data. Total deposition was estimated based on the throughfall/total N deposition ration (0.56) from the nearby Smokies Tower site
(Johnson and Lindberg,  1992b).
'Annual throughfall deposition to the chaparral ecosystem.
fNitrogen output is from unpublished streamwater data (Fenn and Poth, 1999). The low value represents a year of average precipitation, and the high value is for 1995, when
precipitation was nearly double the long-term average.  Nitrogen output includes N export in streamwater and to groundwater.
8Annual input and output data are not known, although N deposition in this forest is probably typical for much of the rural western United States (2-3 kg N ha"1 year"1 (Young et al.,
1988).  Excess N is from weathering of ammonium in mica schist bedrock.  The ammonium was rapidly nitrified, leading to high NO3" concentrations  in soil solution (Dahlgren, 1994).

-------
 1           Plant succession patterns and biodiversity are affected significantly by chronic nitrogen
 2      additions in some North American ecosystems (Figure 4-17).  The location of nitrogen saturated
 3      ecosystems in North America, and the steps leading to nitrogen saturation, are indicated on the
 4      map in Figure 4-16. Conceptual models of regional nitrogen saturation indicate saturation in
 5      New England, in the Colorado alpine ecosystems and in California forests.  Fenn et al. (1998)
 6      reported that long-term  nitrogen fertilization studies in both New England and Europe, as well,
 7      suggest that some forests receiving chronic inputs of nitrogen may decline in productivity and
 8      experience greater mortality. Long-term fertilization experiments  at Mount Ascutney, VT,
 9      suggest that declining coniferous forest stands with slow nitrogen cycling may be replaced by
10      deciduous fast-growing forests that cycle nitrogen rapidly (Fenn et al., 1998).
11           Atmospheric nitrogen deposition in the northeastern United States is largely a regional
12      problem (Driscoll et al., 2001). In contrast, in the western United States, vast acres of land
13      receive low levels of atmospheric nitrogen deposition that are interspersed with hot spots of
14      elevated nitrogen deposition downwind of large expanding metropolitan centers or large
15      agricultural operations (Fenn et al., 2003).
16           Fenn et al. (1998) have documented the major effects of Nr deposition in terrestrial and
17      aquatic ecosystems in the western United States. Primarily these effects are in response to
18      nitrogen enrichment of  systems that are naturally nitrogen limited. Included in these effects are
19      greenhouse gas emissions, higher nitrogen concentrations in plant tissues, and increased
20      nitrification rates and nitrate (NO3") levels in soils, streams, and lakes (Fenn et al., 2003b). A
21      result of chronic Nr enrichment has resulted in important community changes in vegetation,
22      lichens, mycorrhizae, and phytoplankton, occasionally at relatively low levels of nitrogen
23      deposition (3 to 8 kgN/ha/year; Baron et al., 2000).
24           Developments in recent decades in the Colorado Front range have resulted in increased
25      nitrogen deposition since the 1980s at high-elevation  sites. Total deposition values currently
26      range from 4-8 kgN/ha/year (Baron et al., 2000). Competition among species resulting in
27      changes in community composition is one of the most notable responses to environmental
28      change (Bowman, 2000). Nitrogen saturation, the result of increased deposition in the alpine
29      tundra of Niwot Ridge in the Front Range of the Southern Rockies in Colorado, has changed
30      nitrogen cycling and provided the potential for replacement in plant species by more
31      competitive, faster growing species (Bowman and Steltzer, 1998; Bowman, 2000; Baron et al.,

        June 2003                                 4-103        DRAFT-DO NOT QUOTE OR CITE

-------
                                                                            N-Saturated Ecosystems
                                                                            in North America
 Review of Ecosystem Effects
 and Responses to Excess N
                                1. Nitrogen Inputs:
                                  > Atmospheric deposition, N2 fixation, fertilization
                                  Nitrogen Retention:
                                  > In plant biomass and soil organic matter
                                  > The role of soil microbes and woody residues
                                  > Abiotic retention
                                  Nitrogen Outputs:
                                  »Hydrologic transport, gaseous emissions from soil
                                  > Removal in harvest, fire emissions, and soil erosion
            2, Characteristics Predisposing Forests
              to N Saturation:
              >• Stand vigor and succession, forest type
              > Previous land use—stand history
              *• Soil N accumulation
              >• Topography and climate
              > Nitrogen deposition
 3. Ecosystem Responses to Excess Nitrogen:
   > Nitrate leaching and export
   > Eutrophicationof estuaries
   >• Toxicity of surface waters
   >• Foliar nutrient responses
   > Nitrogen mineralization and nitrification
   > Effects on soil organic matter
   > Soil acidification, cation depletion, Al toxicity
   >• Foliar nutrient responses
   >• Greenhouse gas fluxes
                                4. Regional N Saturation Conceptual Models:
                                  > New England forests
                                  >• California forests
                                  + Colorado alpine ecosystems
Figure 4-17. Diagrammatic overview of excess nitrogen (N) in North America.


Source:  Fennetal. (1998).
June 2003
4-104
DRAFT-DO NOT QUOTE OR CITE

-------
 1      2000).  Plants growing in an alpine tundra, as is true of other plants growing in low resource
 2      environments (e.g., infertile soil, shaded understory, deserts), have been observed to have certain
 3      similar characteristics: a slow grow rate, low photosynthetic rate, a low capacity for nutrient
 4      uptake  and low soil microbial activity (Bowman and Steltzer, 1998; Bowman, 2000).
 5      An important feature of such plants is that they continue to grow slowly and tend to respond
 6      even less when provided with an optimal supply and balance of resources (Pearcy et al., 1987;
 7      Chapin, 1991). Plants adapted to cold, moist environments grow more leaves than roots as the
 8      relative availability of nitrogen increases; however, other nutrients may soon become limiting.
 9      These patterns of vegetative development affect their capacity to respond to variation in
10      available resources and to environmental stresses such as frost, high winds, and drought.
11      Preformation of buds 3-4 years in advance of emergence, reduced cell numbers, and high
12      biomass allocation to belowground organs also limits the ability of many alpine plants to
13      respond to variations in their environment (Bowman, 2000). However, significant interspecific
14      genetic variation influences the capacity of the alpine species to respond to changes in resource
15      availability.  The capacity of subalpine and boreal species in particular, and gymnosperms in
16      general, to reduce nitrates in either roots or leaves appears to be limited. In addition, the ability
17      of trees to use nitrogen varies with the age of the tree and the density of the stand (Waring,
18      1987).
19           In experimental studies of nitrogen deposition conducted by Wedin and Tilman (1996)
20      over a 12-year period on Minnesota grasslands, plots dominated by native warm-season grasses
21      shifted  to low-diversity mixtures dominated by cool-season grasses at all but the lowest rates of
22      nitrogen addition. Grasslands with high nitrogen retention and carbon storage rates were the
23      most vulnerable to loss of species and major shifts in nitrogen cycling. The shift to low-diversity
24      mixtures was associated with the decrease in biomass carbon to nitrogen (C:N) ratios, increased
25      nitrogen mineralization, increased soil nitrate, high nitrogen losses, and low carbon storage
26      (Wedin and Tilman, 1996). Naeem et al. (1994) experimentally demonstrated under controlled
27      environmental conditions that the loss of biodiversity, genetic resources, productivity, ecosystem
28      buffering against ecological perturbation, and loss of aesthetic and commercially valuable
29      resources also may alter or impair ecosystems services.
30           The long-term effects of increased nitrogen deposition have been studied in several
31      western and central European plant communities these include lowland heaths, species-rich

        June 2003                                4-105        DRAFT-DO NOT QUOTE OR  CITE

-------
 1      grasslands, mesotrophic fens, ombrotrophic bogs, upland moors, forest-floor vegetation, and
 2      freshwater lakes (Bobbink, 1998). Large changes in species composition have been observed in
 3      regions with high nitrogen loadings or infield experiments after years of nitrogen addition
 4      (Bobbink et al., 1998).  The increased input of nitrogen gradually increased the availability of
 5      nitrogen in the soil, and its retention because of low rates of leaching and denitrification resulted
 6      in faster litter decomposition and rate of mineralization.  Faster growth and greater height of
 7      nitrophilic species enables these plants to shade out the slower growing species, particularly
 8      those in oligotrophic or mesotrophic conditions (Bobbink, 1998; Bobbink et al., 1998). Excess
 9      nitrogen inputs to unmanaged heathlands in the Netherlands has resulted in nitrophilous grass
10      species replacing slower growing heath species (Roelofs et al., 1987; Garner, 1994).
11      Van Breemen and Van Dijk (1988) noted that over the past several decades the composition of
12      plants in the forest herb layers has been shifting toward species commonly found on nitrogen-
13      rich areas. It also was observed that the fruiting bodies of mycorrhizal fungi had decreased in
14      number.
15           Other studies in Europe point out the effects of excessive nitrogen deposition on mixed-oak
16      forest vegetation along  a deposition gradient largely controlled by soil acidity, nitrogen supply,
17      canopy composition, and location of sample plots (Brunet et al., 1998; Falkengren-Grerup,
18      1998). Results of the study, using multivariate methods, suggest that nitrogen deposition has
19      affected the field-layer vegetation directly by increased nitrogen availability and, indirectly, by
20      accelerating soil acidity. Time series studies indicate that 20 of the 30 field-layer species
21      (nonwoody plants) that were associated most closely with high nitrogen deposition increased in
22      frequency in areas with high nitrogen deposition during the past decades. Included in the field-
23      layer species were many generally considered nitrophilous; however, there were several acid
24      tolerant species (Brunet et al, 1998). In an experimental study involving 15 herbs and
25      13 grasses, Falkengren-Grerup  (1998), observed that species with a high nitrogen demand and a
26      lesser demand for other nutrients were particularly competitive in areas with acidic soils and
27      high nitrogen deposition. The grasses grew better than herbs with the addition of nitrogen.  It
28      was concluded that, at the highest nitrogen deposition, growth was limited for most species by
29      the supply of other nutrients; and, at the intermediate nitrogen concentration, the grasses were
30      more efficient than the herbs in utilizing nitrogen. Nihlgard (1985) suggested that excessive
31      nitrogen deposition may contribute to forest decline in other specific regions of Europe.

        June 2003                                4-106        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Additionally, Schulze (1989), Heinsdorf (1993), and Lamersdorf and Meyer (1993) attribute
 2      magnesium deficiencies in German forests, in part, to excessive nitrogen deposition.
 3           The carbon to nitrogen (C:N) ratio of the forest floor can also be changed by nitrogen
 4      deposition over time. This change appears to occur when the ecosystem becomes nitrogen
 5      saturated (Gundersen et al., 1998a). Long-term changes in C:N status have been documented in
 6      Central Europe and indicate that nitrogen deposition has changed the forest floor.  In Europe,
 7      low C:N ratios coincide with high deposition regions (Gundersen et al., 1998a).  A strong
 8      decrease in forest floor root biomass has been observed with increased nitrogen availability.
 9      Roots and the associated mycorrhizae appear to be an important factor in the accumulation of
10      organic matter in the forest floor at nitrogen-limited sites. If root growth and mycorrhizal
11      formation are impaired by nitrogen deposition, the stability of the forest floor may be affected by
12      stimulating turnover and decreasing the root litter input to the forest floor and thus decrease the
13      nitrogen that can be stored in the forest floor pool (Gundersen et al., 1998b).  Nitrogen-limited
14      forests have  a high capacity for deposited nitrogen to be retained by plants and microorganisms
15      competing for available nitrogen (Gundersen et al., 1998b). Nitrate leaching has been correlated
16      significantly with nitrate status but not with nitrate depositions. Forest floor C:N ratio has been
17      used as a rough indicator of ecosystem nitrogen status in mature coniferous forests and the risk
18      of nitrate leaching; analyses of European databases indicated an empirical relationship between
19      forest floor C:N ratio and nitrate leaching (Gundersen  et al., 1998a). Nitrate leaching was
20      observed when the deposition received was more than 10 kg N/ha.  All of the data sets supported
21      a threshold at which nitrate leaching seems to increase at a C:N ratio of 25. Therefore, to predict
22      the rate of changes in nitrate leaching, it is necessary to be able to predict the rate of changes in
23      the forest floor C:N ratio. Decreased foliar and soil nitrogen and soil C:N ratios, as well as
24      changes in nitrogen mineralization rates, have been observed when comparing responses to
25      nitrogen deposition in forest stands east and west of the Continental Divide in the Colorado
26      Front Range (Baron et al., 2000; Rueth and Baron, 2002). Understanding the variability in forest
27      ecosystem response to nitrogen input is essential in assessing pollution risks (Gundersen et al.,
28      1998a).
29           The plant root is an important region of nutrient dynamics.  The rhizosphere  includes the
30      soil that surrounds  and is influenced by plant roots (Wall and Moore, 1999). The mutualistic
31      relationship between plant roots, fungi,  and microbes is critical for the growth of the organisms

        June 2003                                4-107       DRAFT-DO NOT QUOTE OR CITE

-------
 1      involved. The plant provides shelter and carbon; whereas the symbiont provides access to
 2      limiting nutrients such as nitrogen and phosphorus. As indicated above, changes in soil nitrogen
 3      influence the mycorrhizal-plant relationship. Mycorrhizal fungal diversity is associated with
 4      above-ground plant biodiversity, ecosystem variability, and productivity (Wall and Moore,
 5      1999).  Aber et al. (1998) showed a close relationship between mycorrhizal fungi and the
 6      conversion of dissolved inorganic nitrogen to soil nitrogen. During nitrogen saturation, soil
 7      microbial communities change from being fungal, and probably being dominated by
 8      mycorrhizae, to being dominated by bacteria. The loss of mycorrhizal function has been
 9      hypothesized as the key process leading to increased nitrification and nitrate mobility. Increased
10      nitrate mobility leads to increased cation leaching and soil acidification (Aber et al., 1998).
11           The interrelationship of above- and below-ground flora is illustrated by the natural
12      invasion of heathlands by oaks (Quercus robur).  The soil-forming factors under the heath have
13      been vegetation typed during the last 2000 years; whereas the invasion by oaks has been taking
14      place for only a few decades. Clearly changes in the ground floor and soil morphology takes
15      place when trees colonize heath (Nielsen et al., 1999). The distribution of roots also changed
16      under the three different vegetation types.  Under both heather and the Sitka spruce plantation,
17      the majority of roots are confined to the uppermost horizons; whereas under oak, the roots are
18      distributed more homogeneously. There was also a change in the C:N ratio when heather was
19      replaced by oaks.  Also, the spontaneous succession of the heath by oaks changed the biological
20      nutrient cycle into a deeper vertical cycle when compared to the  heath where the cycle is
21      confined to the upper soil horizons. Soils similar to those described in this study (Jutland,
22      Denmark) with mainly an organic buffer system seem to respond quickly to changes in
23      vegetation (Nielsen et al., 1999).
24           The affects of changes in root to shoot relationships in plants were observed in studies of
25      the coastal sage scrub (CSS) community in southern California which is composed of the
26      drought-deciduous shrubs Artemisia californica, Enceliafarmosa,  and Eriogonumfasciculatum.
27      The CSS in California  has been declining in land area and in shrub density over the past 60 years
28      and is being replaced in many areas by Mediterranean annual grasses (Allen et al., 1998; Padgett
29      et al., 1999; Padgett and Allen, 1999). Nitrogen deposition was considered as a possible cause.
30      Up to 45 kg/ha/yr are deposited in the Los Angeles Air Basin (Bytnerowicz and Fenn, 1996).
31      Tracts of land set aside as reserves, which in many cases in southern California are surrounded

        June 2003                                4-108        DRAFT-DO NOT QUOTE OR CITE

-------
 1      by urbanization, receive large amounts of nitrogenous compounds from polluted air.  The CSS is
 2      of particular interest because some 200 sensitive plant species and several federally listed animal
 3      species are found in the area (Allen et al., 1998). Because changes in plant community structure
 4      often can be related to increases in the availability of a limiting soil nutrient or other resource,
 5      experiments were conducted to determine whether increased nitrogen availability was associated
 6      with the significant loss in native shrub cover.  Studies indicated that the three native perennial
 7      shrubs (Artemisia californica, Eriogonum fasciculatum, and Encelia farinosa tended to be more
 8      nitrophilous than the two exotic annual grasses (Bromus mbens, Avena fatua) and the weedy pod
 9      mustard (Brassica geniculatd).  These results contrast with most models dealing with the
10      adaptation of perennial species  to stressful environments (Padgett and Allen,  1999).  If nitrogen
11      were the only variable between the invasive annuals and native shrubs, neither shrubs nor
12      grasses have a particular advantage.  Although CSS shrubs are able to take up nitrogen at high
13      rates, native grasses have a denser seedbank and earlier germination than native species. The
14      native seedlings are not able to  compete with dense stands of exotic grasses, and thus are
15      gradually replaced by the grasses following disturbance such as frequent fire (Eliason and Allen,
16      1997; Clone et al., 2002; Yoshida and Allen, 2001).  In  addition, nitrogen-induced changes in
17      arbuscular mycorrhizal fungi may also affect the growth of native seedlings. Nitrogen
18      enrichment of the soils induced a shift in the arbuscular mycorrhizal community composition.
19      Larger-spored fungal species (Scutellospora and Gigaspora), due to a failure  to sporulate,
20      decreased in number with a concomitant proliferation of small-spored species of Glomus
21      aggregatum, G. leptotichum, and G. geospomm, indicating a strong selective  pressure for the
22      smaller spores species of fungi  (Edgerton-Warburton and Allen, 2000). These results
23      demonstrate that nitrogen enrichment of the soil significantly alters the arbuscular mycorrhizal
24      species composition and richness and markedly decreases the overall diversity of the  arbuscular
25      mycorrhizal community. The decline in coastal sage scrub species can, therefore, directly be
26      linked to the decline of the arbuscular mycorrhizal community (Tidgerton-Warburton  and Allen,
27      2000).
28           In addition to excess nitrogen deposition effects on terrestrial ecosystems of the types
29      noted above (e.g., dominant species shifts and other biodiversity impacts), direct atmospheric
30      nitrogen deposition and increased nitrogen inputs via runoff into streams, rivers, lakes, and
31      oceans can noticable affect aquatic ecosystems as well (Figure 4-15).  Estuaries are among the

        June 2003                                4-109        DRAFT-DO NOT QUOTE OR CITE

-------
 1      most intensely fertilized ecosystems on Earth, receiving far greater nutrient inputs than other
 2      systems.  Chesapeake Bay is a prime example (Fenn et al., 1998).  Another illustrative example
 3      is recently reported research (Paerl et al., 2001) characterizing the effects of nitrogen deposition
 4      on the Pamlico Sound, NC,  estuarine complex, which serves as a key fisheries nursery
 5      supporting an estimated 80% of commercial and recreational finfish and shellfish catches in the
 6      southeastern U.S. Atlantic coastal region.  Such direct atmospheric nitrogen deposition onto
 7      waterways feeding into the Pamlico Sound or onto the sound itself and indirect nitrogen inputs
 8      via runoff from upstream watersheds contribute to conditions of severe water oxygen depletion;
 9      formation of algae blooms in portions of the Pamlico Sound estuarine complex; altered fish
10      distributions, catches, and physiological states; and increases in the incidence of disease.  Under
11      extreme conditions of especially high rainfall rate events (e.g., hurricanes) affecting watershed
12      areas feeding into the sound, the effects of nitrogen runoff (in combination with excess loadings
13      of metals or other nutrients) can be massive — e.g., creation of the widespread "dead-zone"
14      affecting large areas of the Pamlico Sound for many months after hurricane Fran in 1996 and
15      hurricanes Dennis, Floyd, and Irene in 1999 impacted eastern North Carolina.
16           The primary pathways of nitrogen loss from forest ecosystems are hydrological transport
17      beyond the rooting zone into groundwater or stream water, or surface flows of organic nitrogen
18      as nitrate and nitrogen loss associated with soil erosion (Fenn et al., 1998). Stream water nitrate
19      concentrations have been related to forest success!onal  stage in the eastern United States.
20      Logging history and fire history of an area are major variables determining the capability  of a
21      forest stand to  retain nitrogen. Nitrogen concentrations were high in manure ecosystems after
22      disturbances such as clearcutting, but lower in mid-successional forests.
23           Nitrogen saturation of a high elevation watershed in the southern Appalachian Mountains
24      was observed to affect stream water chemistry.  High nitrate concentrations have been observed
25      in streams draining undisturbed watersheds in the Great Smoky Mountains National Park in
26      Tennessee and North Carolina. Nitrate concentrations were highest at higher elevations and in
27      areas around old-growth forests that had never been logged (Fenn  et al., 1998).
28           In the Northeast, nitrogen is the element most responsible for eutrophication in coastal
29      waters of the region (Jaworski et al., 1997).  There has been a 3 to 8-fold increase in nitrogen
30      flux from 10 watersheds in the Northeastern United States since the early 1900s. These
31      increases are associated with nitrogen oxide emissions from combustion which have increased

        June 2003                                4-110        DRAFT-DO NOT QUOTE OR CITE

-------
 1      5-fold. Riverine nitrogen fluxes have been correlated with atmospheric deposition onto their
 2      landscapes and also with nitrogen oxides emissions into their airsheds. Data from 10 benchmark
 3      watersheds with good historical records, indicate that ca., 36-80% of the riverine total nitrogen
 4      export, with an average of 64%, was derived directly or indirectly from nitrogen oxide emissions
 5      (Jaworskietal., 1997).
 6           Nitrogen saturation of a high elevation watershed in the southern Appalachian Mountains
 7      was observed to affect stream water chemistry. The Great Smoky Mountains in the southeastern
 8      United States receive high total atmospheric deposition of sulfur and nitrogen (2,200 Eq/ha/yr of
 9      total sulfur and approximately 1,990 Eq/ha/yr of total nitrogen). A major portion of the
10      atmospheric loading is from dry and cloud deposition. Extensive surveys conducted in October
11      1993 and March 1994 indicated that stream pH values were near or below pH 5.5 and that the
12      acid neutralizing capacity (ANC) was below 50 |ieq/L at high elevations.  Analysis of
13      streamwater indicated that nitrate was the dominant anion (Flum and Nodvin, 1995; Nodvin
14      et al., 1995).  The  study was expanded to the watershed scale with monitoring of precipitation,
15      thoughfall, stream hydrology, and stream chemistry. Nitrogen saturation of the watershed
16      resulted in extremely high exports of nitrate and promoted both chronic and episodic stream
17      acidification in which the nitrate was the dominant ion. Significant exports of base cation was
18      also observed. Nitrification of the watershed soils resulted in elevations of soil  solution
19      aluminum concentrations to levels known to inhibit calcium uptake in red spruce (Nodvin et al.,
20      1995).
21           Excessive nitrogen loss is a symptom of terrestrial ecosystem dysfunction and results in the
22      degradation of water quality and potentially deleterious effects on terrestrial and aquatic
23      ecosystems (Fenn and Poth, 1999).  Data from a number of hydrologic, edaphic, and plant
24      indicators indicate that the mixed conifer forests and chaparral systems directly exposed to air
25      pollution from greater Los Angeles are nitrogen saturated. Preliminary data suggests that
26      symptoms of nitrogen saturation are  evident in mixed conifer or chaparral sites receiving
27      atmospheric deposition of 20 to 25 kg/N/ha/y (Fenn et al, 1996).  Available data clearly indicate
28      that ecosystems with a Mediterranean climate have a limited capacity to retain nitrogen within
29      the terrestrial system (Fenn and Poth, 1999). A 3-year study of streamwater NO3" concentrations
30      along nitrogen deposition gradients in the San Bernardino Mountains in southern California
31      evaluated streamwater quality and whether the streamwater concentrations covaried with

        June 2003                                4-111        DRAFT-DO NOT QUOTE OR CITE

-------
 1      nitrogen deposition across pollution gradients in the San Bernardino Mountains. Streamwater
 2      NO3" concentrations at Devil Canyon in the San Gabriel Mountains northeast of Los Angeles are
 3      the highest reported in North America for forested watersheds (Fenn and Poth, 1999). Five of
 4      the six streams monitored maintained elevated NO3" throughout the year.  Peak nitrate
 5      concentrations ranged from 40 to 350 |imol/L. In the San Gorgonio Wilderness, an area of low
 6      to moderate deposition where 12 streams were sampled, only the five that had the greatest air
 7      pollution exposure had high NO3" concentrations.  The results of the study suggested a strong
 8      association between levels of NO3" export in streamwater and the severity of chronic nitrogen
 9      deposition to the terrestrial watersheds. However, nitrogen processing within terrestrial and
10      aquatic systems, even in areas with high nitrogen deposition, determine streamwater NO3"
11      concentrations (Fenn and Poth,  1999). The Fernow Experimental Forest in West Virginia, the
12      Great Smoky Mountains National park in Tennessee,  and watersheds in southwestern
13      Pennsylvania are the only undisturbed forested sites in North America known to have
14      streamwater NO3" concentrations within the range  of values found at Devil Canyon (Fenn and
15      Poth, 1999).
16
17           Effects of Sulfur Deposition.  Sulfur is a major component of plant proteins and as such is
18      an essential plant nutrient. The most important source of sulfur is sulfate taken up from the soil
19      by plant roots even though plants can utilize atmospheric SO2 (Marschner, 1995).  The
20      availability  of organically bound sulfur in soils depends largely on microbial decomposition, a
21      relatively slow process. The major factor controlling  the movement of sulfur from the soil into
22      vegetation is the rate of release  from the organic to the inorganic compartment (May et al., 1972;
23      U.S. Environmental Protection Agency,  1982; Marschner, 1995).  Sulfur plays a critical role in
24      agriculture as an essential component of the balanced fertilizers needed to grow and increase
25      worldwide food production (Ceccotti and Messick, 1997). Atmospheric deposition is an
26      important component of the sulfur cycle.  This is true not only in polluted areas where
27      atmospheric deposition is very high, but also in areas  of low sulfur input.  Additions of sulfur
28      into the soil in  the form of SO4"2 could alter the important organic-sulfur/organic-nitrogen
29      relationship involved in protein formation in plants. The biochemical relationship between
30      sulfur and nitrogen in plant proteins and the regulatory coupling of sulfur and nitrogen
31      metabolism indicate that neither element can be assessed adequately without reference to the

        June 2003                                4-112       DRAFT-DO NOT QUOTE OR CITE

-------
 1      other. Sulfur deficiency reduces nitrate reductase and, to a similar extent, also glutamine
 2      synthetase activity. Nitrogen uptake in forests, therefore, may be loosely regulated by sulfur
 3      availability, but sulfate additions in excess of needs do not necessarily lead to injury (Turner and
 4      Lambert, 1980; Hogan et al.,  1998).
 5           Only two decades ago, there was little information comparing sulfur cycling in forests with
 6      other nutrients, especially nitrogen. With the discovery of deficiencies in some unpolluted
 7      regions (Kelly and Lambert, 1972; Humphreys et al., 1975; Turner et al., 1977; Schnug, 1997)
 8      and excesses associated with  acidic deposition in other regions of the world (Meiwes and
 9      Khanna, 1981;  Shriner and Henderson, 1978; Johnson et al.,  1982a,b), interest in sulfur nutrition
10      and cycling in forests has heightened. General reviews of sulfur cycling in forests have been
11      written by Turner and Lambert (1980), Johnson (1984), Mitchell et al. (1992a,b), and Hogan
12      et al. (1998). The salient elements of the sulfur cycle as it may be affected by changing
13      atmospheric deposition are summarized by Johnson and Mitchell (1998). Sulfur has become the
14      most important limiting factor in European agriculture because of the desulfurization of
15      industrial emissions (Schnug, 1997).
16           Most studies dealing with the effects of sulfur deposition on plant communities have been
17      conducted in the vicinity of point sources and  have investigated above-ground effects of SO2 or
18      acidifying effects of sulfate on soils (Krupa and Legge, 1998; Dreisinger and McGovern, 1970;
19      Legge, 1980; Winner and Bewley, 1978a,b; Laurenroth and Michunas, 1985; U.S.
20      Environmental  Protection Agency, 1982).  Krupa and Legge (1986), however, observed a
21      pronounced increase in foliar sulfur concentrations in all age classes of needles of the hybrid
22      pine lodgepole  x jack pine (Pinus contorta x P. banksiana).  This vegetation had been exposed to
23      chronic low concentrations of sulfur dioxide (SO2) and hydrogen sulfide (H2S) for more than 20
24      years and, then, to fugitive sulfur aerosol. Observations under the microscope showed no sulfur
25      deposits on the needle surfaces and led to the conclusion that the sulfur in the needles was
26      derived from the soil. The oxidation of elemental sulfur and the generation of protons is well
27      known for the soils of Alberta, Canada. This process is mediated by bacteria of the Thiobacillus
28      sp. As elemental sulfur gradually is converted to protonated SO4, it can be leached downward
29      and readily taken up by plant roots. The activity of Thiobacillus sp. is stimulated by elemental
30      sulfur additions (Krupa and Legge, 1986).
31

        June 2003                                4-113       DRAFT-DO NOT QUOTE OR CITE

-------
 1           Effects of Acidic Deposition on Forest Soils. Acidic deposition over the past quarter of a
 2      century has emerged as a critical environmental stress that affects forested landscapes and
 3      aquatic ecosystems in North America, Europe, and Asia (Driscoll et al., 2001). Acidic
 4      deposition can originate from transboundary air pollution and affect large geographic areas.  It is
 5      composed of ions, gases, particles derived from gaseous emissions of sulfur dioxide (SO2),
 6      nitrogen oxides (NOX), ammonia (NH3), and particulate emissions of acidifying and neutralizing
 7      compounds and is highly variable across space and time. It links air pollution to diverse
 8      terrestrial and aquatic ecosystems and alters the interactions of the hydrogen ion (H+) and many
 9      elements (e.g., sulfur, nitrogen, calcium, magnesium, aluminum, and mercury). Acidic
10      deposition contributes directly and indirectly to biological stress and the degradation of
11      ecosystems and has played a major role in recent acidification of soil in some areas of Europe
12      and, to a more limited extent, eastern North America (Driscoll et al., 2001).
13           Substantial  and previously unsuspected changes in soils have been observed in polluted
14      areas of eastern North America, the United Kingdom, Sweden, and Central Europe and in less
15      polluted regions of Australia and western North American (reviewed by Johnson et al., 1999 and
16      by Huntington, 2000).  In some cases, trends were toward more acidic soils (e.g., Markewitz
17      et al.,  1998), and, in others, there were no consistent trends, with some soils showing increases
18      and some showing decreases at different sampling times, and some showing no change (e.g.,
19      Johnson and Todd, 1998; Trettin et al.,  1999; Yanai et al., 1999).
20           Significant changes in soil chemistry have occurred at many sites in the eastern United
21      States during recent decades.  Patterns of change in tree ring chemistry, principally at high
22      elevations sites in the eastern United  States, reflect the changing inputs of regional pollutants to
23      forests. A temporal sequence of changes in uptake patterns, and possibly in tree growth, would
24      be expected if significant base cation mobilization and depletion of base cations from eastern
25      forest soils has occurred. Temporal changes in the chemistry of tree rings of red spruce were
26      examined as indicators of historical changes in the chemical environment of red spruce.
27           Analysis of changes in wood chemistry from samples across several sites indicated that
28      there have been substantial departures from the expected linear decreases in calcium
29      accumulation patterns in wood. A region-wide calcium increase above expected levels followed
30      by decreasing changes in wood calcium suggest that calcium mobilization began possibly 30 to
31      40 years ago and has been followed by  reduced accumulation rates in wood, presumably

        June 2003                                4-114        DRAFT-DO NOT QUOTE OR CITE

-------
 1      associated with decreasing calcium availability in soil (Bondietti and McLaughlin, 1992). The
 2      period of calcium mobilization coincides with a region-wide increase in growth rate of red
 3      spruce; whereas the period of decreasing levels of calcium in wood corresponds temporally with
 4      patterns of decreasing radial growth at high elevation sites throughout the region during the past
 5      20 to 30 years. The decline in wood calcium suggests that calcium loss may have increased to
 6      the point at which base saturation of soils has been reduced. Increases in aluminum and iron
 7      typically occur as base cations are removed from the soils by tree uptake (Bondietti and
 8      McLaughlin, 1992). The changes are spatially and temporally consistent with changes in the
 9      emissions of SO2 and NO2 across the region and suggest that increased acidification of soils has
10      occurred.
11           Studies by Shortle and Bondietti (1992) support the view that changes in soil chemistry in
12      eastern North America forest sites occurred many decades ago, "before anybody was looking."
13      Sulfur and nitrogen emissions began increasing in eastern North America in the 1920s and
14      continued to increase into the 1980s when sulfur began to decrease but nitrogen emissions did
15      not (Garner et al., 1989). Shortle and Bondietti (1992) present evidence that, from the late 1940s
16      into the 1960s, the mor humus (organic) layer of acid-sensitive forest sites in eastern North
17      America underwent a significant change that resulted in the loss of exchangeable essential base
18      cations and interrupted the critical base nutrient cycles between mature trees and the root-humus
19      complex. The timing of the effect appears to have coincided with the period when the SOX and
20      NOX emissions in eastern North America subject to long-range transport were increasing the
21      most rapidly (see above; Shortle and Bondietti, 1992). Although  forest ecosystems other than
22      the high-elevation spruce-fir forests are not currently manifesting symptoms of injury directly
23      attributable to acid  deposition, less sensitive forests throughout the United States are
24      experiencing gradual losses of base cation nutrients, which in many cases will reduce the quality
25      of forest nutrition over the long term (National Science and  Technology Council, 1998).
26      In some cases it may not even take decades because these forests  already have been receiving
27      sulfur and nitrogen deposition for many years.  The current  status of forest ecosystems in
28      different U.S. geographic regions varies, as does their sensitivity to nitrogen and sulfur
29      deposition. Variation in potential future forest responses or sensitivity are caused,  in part, by
30      differences in deposition of sulfur and nitrogen, ecosystem sensitivities to sulfur and nitrogen
        June 2003                                4-115        DRAFT-DO NOT QUOTE OR CITE

-------
 1      additions, and responses of soils to sulfur and nitrogen inputs (National Science and Technology
 2      Council, 1998).
 3           Acidic deposition has played a major role in recent soil acidification in some areas of
 4      Europe and, to a more limited extent, eastern North America. Examples include the study by
 5      Hauhs (1989) at Lange Bramke, Germany, which indicated that leaching was of major
 6      importance in causing substantial reduction in soil-exchangeable base cations over a 10-year
 7      period (1974-1984).  Soil acidification and its effects result from the deposition of nitrate (NO3")
 8      and sulfate (SO4"2) and the associated hydrogen (H+) ion. The effects of excessive nitrogen
 9      deposition on soil acidification and nutrient imbalances have been well established in Dutch
10      forests (Van Breemen et al., 1982; Roelofs et al., 1985; Van Dijk and Roelofs,  1988).
11      For example, Roelofs et al. (1987) proposed that NH3 /NH4+ deposition leads to heathland
12      changes via two modes:  acidification of the soil and the loss of cations K+, Ca+2, and Mg+2; and
13      nitrogen enrichment that results in "abnormal" plant growth rates and  altered competitive
14      relationships.  Nihlgard (1985) suggested that excessive nitrogen deposition may contribute to
15      forest decline in other specific regions of Europe. Falkengren-Grerup (1987) noted that, during
16      about 50 years, unexpectedly large increases in growth of beech (Fagus sylvatica L.) were
17      associated with decreases in pH and  exchangeable cations in some sites in southernmost Sweden.
18           Likens et al. (1996,  1998) suggested that soils are changing at the Hubbard Brook
19      Watershed, NH, because of a combination of acidic deposition and reduced base cation
20      deposition.  They surmised, based on long-term trends in streamwater data, that large amounts of
21      calcium and magnesium have been lost from the soil-exchange complex over a 30-year period
22      from approximately 1960 to 1990. The authors speculate that the declines in base cations in
23      soils may be the cause of recent slowdowns in forest growth at Hubbard Brook. In a follow-up
24      study, however, Yanai et al. (1999) found no significant decline in calcium and magnesium
25      concentrations in forest floors at Hubbard Brook over the period 1976 to 1997.  They also found
26      both gains and losses in forest floor calcium and magnesium between  1980 and 1990 in a
27      regional survey. Thus, they concluded that "forest floors in the region are not currently
28      experiencing rapid losses of base cations, although losses may have preceded the onset of these
29      three studies." The biogeochemistry of calcium at Hubbard Brook is discussed in detail by
30      Likens etal. (1998).
        June 2003                                 4-116       DRAFT-DO NOT QUOTE OR CITE

-------
 1           Hydrogen ions entering a forest ecosystem first encounter the forest canopy, where they
 2      are often exchanged for base cations that then appear in throughfall (Figure 4-18 depicts a model
 3      of FT sources and sinks).  Base cations leached from the foliage must be replaced through uptake
 4      from the soil, or foliage cations will be reduced by the amounts leached.  In the former case, the
 5      acidification effect is transferred to the soil where FT is exchanged for a base cation at the
 6      root-soil interface. Uptake of base cations or NH4+ by vegetation or soil microorganisms causes
 7      the release of FT in order to maintain charge balance; uptake of nutrients in anionic form (NO3",
 8      SO4"2, PO4"3) causes the release of OH" in order to maintain charge balance. Thus, the net
 9      acidifying effect of uptake is the difference between cation and anion uptake. The form  of ions
10      taken up is known for all nutrients but nitrogen because either NH4+ or NO3" can be utilized.
11      In that nitrogen is a nutrient taken up in greatest quantities, the uncertainty in the ionic form of
12      nitrogen utilized creates great uncertainty in the overall H+ budget for soils (Johnson 1992).
13           The cycles of base cations differ from those of N, P, and S in several respects. The fact
14      that calcium, potassium, and magnesium exist primarily as cations in solution, whereas N, P, and
15      S exist primarily as anions, has major implications for the cycling of the nutrients and the effects
16      of acid deposition on these cycles. The most commonly accepted model of base cation cycling
17      in soils is one in which base cations are released by weathering of primary minerals to cation
18      exchange sites where they are available for either plant uptake or leaching (Figure 4-18). The
19      introduction of H + by atmospheric deposition or by internal processes will affect the fluxes of
20      Ca, K, and Mg via cation  exchange or weathering processes. Therefore, soil leaching is  often of
21      major importance in cation cycles, and many forest ecosystems show a net loss of base cations
22      (Johnson, 1992).
23           Two basic types of soil change are involved: (1) a short-term intensity type change
24      resulting from the concentrations of chemicals in soil water and (2) a long-term capacity change
25      based on the total content of bases, aluminum, and iron stored in the soil (Reuss and Johnson,
26      1986; Van Breemen et al., 1983). Changes in intensity factors can have a rapid affect on the
27      chemistry of soil solutions.  Increases in the amounts of sulfur and nitrogen in acidic deposition
28      can cause immediate increases in acidity and mobilization of aluminum in soil solutions.
29      Increased aluminum concentrations and an increase in the Ca/Al ratio in soil solution have been
30      linked to a significant reduction in the availability of essential base  cations to plants, an increase
31

        June 2003                                4-117        DRAFT-DO NOT QUOTE OR CITE

-------
                                              Deposition
                                                H+
                     Soil
                   Organism
                    Uptake
                               CO2 + H2O
                               Carbonic Acid Formation
                                R-COOH
                                Organic Acid Formation
                                                      Leaching
       Figure 4-18.  Schematic of sources and sinks of hydrogen ions in a forest (from Taylor
                     et al., 1994).
 1     in plant respiration, and increased biochemical stress (National Science and Technology Council,
 2     1998).
 3           Rapid changes in intensity resulting from the addition of increased amounts of nitrogen or
 4     sulfur in acidic deposition can have a rapid effect on the chemistry of soil solutions by increasing
 5     the acidity and mobilizing aluminum.  Increased concentrations of aluminum and an increase in
 6     the ratio of calcium to aluminum in soil solution have been linked to a significantly reduced
 7     availability of essential cations to plants.
 8           Capacity changes are the result of many factors acting over long time periods. The content
 9     of base cations (calcium, magnesium, sodium, and potassium) in soils results from additions
10     from the atmospheric deposition, decomposition of vegetation, and geologic weathering.  Loss of
       June 2003
4-118
DRAFT-DO NOT QUOTE OR CITE

-------
 1      base cations may occur through plant uptake and leaching.  Increased leaching of base cations
 2      may result in nutrient deficiencies in soils as has been happening in some sensitive forest
 3      ecosystems (National Science and Technology Council, 1998).
 4           Aluminum toxicity is a possibility in acidified soils.  Atmospheric deposition (or any other
 5      source of mineral anions) can increase the concentration of Al, especially A13+, in soil solution
 6      without causing significant soil acidification (Johnson and Taylor, 1989). Aluminum can be
 7      brought into soil solution in two ways: (1) by acidification of the soil and (2) by an increase in
 8      the total anion and cation concentration of the soil solution. The introduction of mobile, mineral
 9      acid anions to an  acid soil will cause increases in the concentration of aluminum in the soil
10      solution, but extremely acid soils in the absence of mineral acid anions will not produce a
11      solution high in aluminum. An excellent review of the relationships among the most widely
12      used cation-exchange equations and their implications for the mobilization of aluminum into soil
13      solution is provided by Reuss (1983).
14           A major concern has been that soil acidity would lead to nutrient deficiency.  Calcium is
15      essential for root  development and the formation of wood, and it plays a major role in cell
16      membrane integrity and cell wall structure.  Aluminum concentrations in the soil can influence
17      forest tree growth in regions where acidic deposition and natural acidifying processes increase
18      soil acidity. Acidic deposition mobilizes calcium and magnesium, which are essential for root
19      development and stem growth.  Mobilized aluminum can also bind to fine root tips of red spruce,
20      further limiting calcium and magnesium  uptake (Shortle and Smith, 1988; Shortle et al., 1997).
21           There is abundant evidence that aluminum is toxic to plants. Upon entering tree roots, it
22      accumulates in root tissues (Thornton et  al., 1987; Vogt et al., 1987a, b).  Reductions in calcium
23      uptake have been associated with increases in aluminum uptake (Clarkson and Sanderson, 1971).
24      A number of studies suggest that the toxic effect of aluminum on forest trees could be caused by
25      Ca+2 deficiency (Shortle and Smith,  1988; Smith, 1990a). Mature trees have a high calcium
26      requirement relative to agriculture crops  (Rennie, 1955).  Shortle and Smith (1988) attributed the
27      decline of red spruce in eight stands across northern New England from Vermont to Maine to an
28      imbalance  of A13+ and Ca+2 in fine root development.
29           To be taken up from the soil by roots, calcium must be dissolved in soil water (Lawrence
30      and Huntington, 1999). Aluminum in soil solution reduces calcium uptake by competing for
31      binding sites in the cortex of fine roots. Tree species may be adversely affected if high

        June 2003                                 4-119        DRAFT-DO NOT QUOTE OR CITE

-------
 1      aluminum to nutrient ratios create a nutrient deficiency by limiting uptake of calcium and
 2      magnesium (Shortle and Smith, 1988; Garner, 1994).  Acid deposition, by lowering the pH of
 3      aluminum-rich soil, can increase aluminum concentrations in soil water through dissolution and
 4      ion exchange processes. Aluminum is more readily taken up than is calcium because of its
 5      greater affinity for negatively charged surfaces. When present in the forest floor, aluminum
 6      tends to displace adsorbed calcium and causes it to be more readily leached. The continued
 7      buildup of aluminum in the forest floor layer, where nutrient uptake is greatest, can lower
 8      efficiency of calcium uptake when the ratio of calcium to aluminum in soil water is less than one
 9      (Lawrence and Huntington, 1999). Reduction in calcium uptake suppresses cambial growth and
10      reduces the rate of wood (annual ring) formation,  decreases the amount of functional sapwood
11      and live crown, and predisposes trees to disease and injury from stress agents when the
12      functional sapwood becomes less that 25% of cross-sectional stem area (Smith, 1990a). A 1968
13      Swedish report to the United Nations postulated a decrease in forest growth  of ca.,  1.5% per year
14      when the ratio of calcium to aluminum in soil water is less than one (Lawrence and Huntington,
15      1999). The concern that acidification and nutrient deficiency may result in forest decline
16      remains today.
17           Acidic deposition has been firmly implicated as a causal factor in the northeastern high-
18      elevation decline of red spruce (DeHayes et al., 1999).  The frequency of freezing injury of red
19      spruce has increased over the past 40 years, a period that coincides with increase emissions of
20      sulfur and nitrogen oxides and acidic deposition (DeHayes et al., 1999). Studies indicate that
21      there is a significant positive association between cold tolerance and foliar calcium in trees
22      exhibiting deficiency in foliar calcium.  Most of the calcium in conifer needles is insoluble
23      calcium oxalate and pectate crystals, which are of little physiological importance.  It is the labile
24      calcium ions in equilibrium within the plasma membrane that are of major physiological
25      importance (DeHayes et al.,  1999). The membrane-associated pool of calcium (mCa), although
26      a relatively small fraction of total foliar ion pools, strongly influences the response of cells to
27      changing environmental conditions. The plant plasma membrane plays a critical role in
28      mediating cold acclimation and low-temperature injury. Leaching of calcium associated with
29      acidic deposition is considered to be the result of  cation exchange due to exposure to the H+ ion.
30      The studies of DeHayes et al. (1999) demonstrate that the direct deposition of acidic deposition
31      on needles represents a unique environmental stress, in that it preferentially removes mCa which

        June 2003                                4-120       DRAFT-DO NOT QUOTE OR CITE

-------
 1      is not readily replaced in autumn.  They propose that direct deposition on red spruce foliage
 2      preferentially displaces calcium ions specifically associated with plasma membranes of
 3      mesophyll cells resulting in the reduction of mCa and the destabilizing of plasma membranes
 4      and depletion of messenger calcium. Further, DeHayes et al. (1999) state that their studies raise
 5      the strong possibility that acid rain alteration of the mCa and membrane integrity is not unique to
 6      red spruce but has been demonstrated in many other northern temperate forest tree species
 7      including yellow birch (Betula alleghaniensis\ white spruce (Picea glaucus), red maple (Acer
 8      rubrum) eastern white pine (Pinus strobus), and sugar maple (Acer saccharum).  Assessments of
 9      mCa, membrane integrity,  and the effects of other secondary stresses have not yet been made for
10      these species.
11           Seasonal and episodic acidification of surface waters have been observed in the eastern
12      United States, Canada and  Europe (Hyer et al., 1995). In the Northeast, the Shenandoah
13      National Park in Virginia, and the Great Smoky Mountains, episodic acidification has been
14      associated with the nitrate ion (Driscoll et al., 2001; Hyer et al., 1995 ; Eshleman et al., 1995).
15      The short-term acid episodes occur during spring snowmelts and large precipitation events
16      (Driscoll et al., 2001). Episodic acidification of surface waters has usually been considered to be
17      a transient loss of acid neutralizing capacity associated with snowmelt/rainfall runoff and, as
18      such, represents short-term (hours to weeks) effects considered to be distinguishable from
19      chronic long-term (years to centuries) changes in acidity.  Studies of both episodic and chronic
20      acidification of surface waters indicate that acidification can have long-term adverse effects on
21      fish populations, declines of species richness, abundance of zooplankton, and macroinvertebrates
22      (Driscoll et al., 2001; Eshleman et al., 1995).  Nitrogen saturation of soils and the slow release of
23      nitrates inhibited the recovery of acid sensitive systems (Driscoll et al., 2001). The acidification
24      of aquatic ecosystems and  effects on aquatic biota are discussed in more detail in the EPA
25      document Air Quality Criteria for Nitrogen Oxides (U.S. Environmental Protection Agency,
26      1993).
27           Air pollution is not the sole cause of soil change. High rates of acidification are occurring
28      in less polluted  regions of the western United States and Australia because of internal  soil
29      processes, such as tree uptake of nitrate and nitrification associated with excessive nitrogen
30      fixation (Johnson et al., 1991b). Many studies have shown that acidic deposition is not a
31      necessary condition for the presence of extremely acidic soils, as evidenced by their presence in

        June 2003                                 4-121        DRAFT-DO NOT QUOTE OR CITE

-------
 1      unpolluted, even pristine, forests of the northwestern United States and Alaska (Johnson et al.,
 2      1991b).  Soil can become acidic when H+ ions attached to NH4+ or HNO3 remain in the soil after
 3      nitrogen is taken up by plants. For example, Johnson et al. (1982b) found significant reductions
 4      in exchangeable K + over a period of only 14 years in a relatively unpolluted Douglas fir
 5      Integrated Forest Study (IFS) site in the Washington Cascades. The effects of acid deposition at
 6      this site were negligible relative to the effects of natural leaching (primarily carbonic acid) and
 7      nitrogen tree uptake (Cole and Johnson, 1977).  Even in polluted regions, numerous studies have
 8      shown the importance of tree uptake of NH4+ and NO3" in soil acidification. Binkley et al. (1989)
 9      attributed the marked acidification (pH decline of 0.3 to 0.8 units and base saturation declines of
10      30 to 80%) of abandoned agricultural soil in South Carolina over a 20-year period to NH4+ and
11      NO3" uptake by a loblolly pine plantation.
12           An interesting example of uptake effects on  soil acidification is that of Al uptake and
13      cycling (Johnson et al.,  1991b). Aluminum accumulation in the leaves of coachwood
14      (Ceratopetalum apetalum) in Australia has been found to have a major effect on the distribution
15      and cycling of base cations (Turner and Kelly, 1981). The presence of C.  apetalum as a
16      secondary tree layer beneath brush box (Lophostemon confertus) was found to lead to increased
17      soil exchangeable Al3+ and decreased soil exchangeable Ca2+ (Turner and Kelly, 1981). The
18      constant addition of aluminum-rich litter fall obviously has had a substantial effect on soil
19      acidification, even if base cation uptake is not involved directly.
20           Given the potential importance of particulate deposition for base cation status of forest
21      ecosystems, the findings of Driscoll et al. (1989, 2001) and Hedin et al. (1994) are especially
22      relevant.  Driscoll et al. (1989, 2001) noted a decline in both SO4"2 and base cations in both
23      atmospheric deposition and stream water over the past two decades at Hubbard Brook
24      Watershed, NH. The decline in SO4"2 deposition was attributed to a decline in emissions, and the
25      decline in stream water SO4"2 was attributed to the decline in sulfur deposition. Hedin et al.
26      (1994) reported a steep decline in atmospheric base cation concentrations in both Europe and
27      North America over the past 10 to 20 years. The reductions in SO2  emissions in Europe and
28      North America in recent years have not been accompanied by equivalent declines in net acidity
29      related to sulfate in precipitation.  These current declines in sulfur deposition have, to varying
30      degrees, been offset by declines in base cations  and may be contributing "to the increased
31      sensitivity of poorly buffered systems."  Analysis  of the data from the IFS supports the authors'

        June 2003                                4-122        DRAFT-DO NOT QUOTE OR CITE

-------
 1      contention that atmospheric base cation inputs may seriously affect ecosystem processes.
 2      Johnson et al. (1994) analyzed base cation cycles at the Whiteface Mountain IPS site in detail
 3      and concluded that Ca losses from the forest floor were much greater than historical losses,
 4      based on historical changes in forest floor Ca observed in an earlier study. Further, the authors
 5      suggested that the difference between historical and current net loss rates of forest floor Ca may
 6      be caused by sharply reduced atmospheric inputs of calcium after about 1970 and may be
 7      exacerbated by sulfate leaching (Johnson et al., 1994b).
 8           The calcium/aluminum molar ratio has been suggested as a valuable ecological indicator of
 9      an approximate threshold beyond which the risk of forest injury from Al stress and nutrient
10      imbalances increases (Cronan and Grigal, 1995). The Ca/Al ratio also can be used as an
11      indicator to assess forest ecosystem changes over time in response to acidic deposition, forest
12      harvesting,  or other process that contribute to acid soil infertility. This ratio, however, may not
13      be a reliable indicator of stress in areas with both high atmospheric deposition of ammonium and
14      magnesium deficiency via antagonism involving ammonium rather than aluminum and in areas
15      with soil solutions with calcium concentrations greater than 500 micromoles per liter (National
16      Science and Technology Council,  1998).  Cronan and Grigal (1995), based on a review of the
17      literature, have made the following estimates for determining the effects of acidic deposition on
18      tree growth or nutrition:
19          •  forests have a 50% risk of adverse effects if the Ca/Al ration is 1.0,
20          •  the risk is 75% if the ratio is 0.5, and
21          •  the risk approaches 100% if the ratio is 0.2.
22      The Ca/Al ratio of soil solution provides only an index of the potential for Al stress.  Cronan and
23      Grigal (1995) state that the overall uncertainty of the  Ca/Al ratio associated with a given
24      probability  ratio is considered to be approximately ±50%.  Determination of thresholds for
25      potential forest effects requires the use of the four successive measurement endpoints in the soil,
26      soil solution, and plant tissue listed below.
27          (1)  Soil base saturation less than  15% of effective cation exchange capacity,
28          (2)  Soil solution Ca/Al molar ratio less than 1.0 for 50% risk,
29          (3)  Fine roots tissue Ca/Al molar ratio less than 0.2 for 50% risk, and
30          (4)  Foliar tissue  Ca/Al molar ratio less than  12.5 for 50% risk.

        June 2003                                  4-123        DRAFT-DO NOT QUOTE OR CITE

-------
 1      The application of the Ca/Al ratio indicator for assessment and monitoring of forest health risks
 2      has been recommended for sites or in geographic regions where the soil base saturation < 15%.
 3
 4           Critical Loads. The critical loads framework for assessing the effects  of atmospheric
 5      deposition originated in Europe where the concept has generally been accepted as the basis for
 6      abatement strategies to reduce or prevent injury to the functioning and vitality of forest
 7      ecosystems caused long-range transboundary chronic acidic deposition (Lokke  et al., 1996). The
 8      critical load has been defined as a "quantitative estimate of an exposure to one or more
 9      pollutants below which significant harmful effects on specified sensitive elements of the
10      environment do not occur according to present knowledge" (Lokke et al., 1996).
11           The concept is useful for estimating the amounts of pollutants that sensitive ecosystems
12      can absorb on a sustained basis without experiencing measurable degradation.  The response to
13      pollutant deposition of an ecosystem is a direct function of the level of sensitivity of the
14      ecosystem to the pollutant and its capability to ameliorate change. The estimation of ecosystem
15      critical loads requires an understanding of how an ecosystem will respond to different loading
16      rates in the long term.  The approach can be of special value for ecosystems  receiving chronic
17      deposition of pollutants such as nitrogen and sulfur. A program was designed to develop and
18      evaluate a framework for setting critical loads of nitrogen and sulfur in the United States in 1989
19      (Strickland et al., 1993). A flexible six step approach has been outlined for use with the critical
20      load framework (Figure 4-19).  These are (1) selection of ecosystem components, indicators, and
21      characterization of the resource; (2) definition of the functional subregions; (3)  characterization
22      of deposition within each of the subregions; (4) definition of an assessment endpoint;
23      (5) selection and application of models; and (6) mapping projected ecosystem responses.  The
24      approach permits variability in ecosystem characteristic  and data availability (Strickland et al.,
25      1993).
26           Ecological endpoints or indicators are measurable  characteristics related to the structure,
27      composition, or functioning of ecological systems (i.e., indicators of condition). One or more
28      measurable endpoints are associated with each element in Table 4-1.  These assessment
29      endpoints represent a formal expression of the environmental value that is to be protected.  If the
30      assessment endpoint is to be used as a regulatory limit, it should be socially  relevant.  Selection
31      of a specific ecosystem for study will depend on the severity of the problem of concern for a

        June 2003                                 4-124       DRAFT-DO NOT QUOTE OR CITE

-------
                    IDENTIFY REGIONS,
                        Subregions
                        Populations
                        Ecosystems
                    Response Hindcasting
CHARACTERIZE RESOURCE
         ISSUE
     Water Quality?
     Human Health?
      Biodiversity?
                                         CHARACTERIZE DEPOSITION
                                        >•  MODEL SELECTION  -4
     . SELECT ENDPOINTSJ
      ! Nitrate < 0,1 mgIL
       ^ Species Richness,
          pH>5.3
                          Response Forecasting
                               ECOSYSTEM RESPONSE PRESENTATION (MAPPING)
        Figure 4-19.  Key elements of proposed framework for determining critical loads for
                     nitrogen and sulfur in the United States.
        Source:  Strickland etal. (1993).
 1      region. Time scales o  response must be considered in selecting and evaluating ecosystem
 2      response(s) to changes in atmospheric deposition. Responses of aquatic ecosystems to
 3      depositions can occur quickly.  Surface water acidification associated with nitrate leaching
 4      should respond to decreases in nitrogen loading in a short period of time. However, changes in
 5      growth responses of vegetation resulting from soil nutrient imbalances may require years or
 6      decades to detect.  The focus of concern should be the populations within an ecosystem that are
 7      sensitive to nitrogen and sulfur deposition (Hunsaker et al., 1993).
 8           Biogeochemicals as indicators for monitoring forest nitrogen status have been proposed by
 9      Fenn and Poth  (1998).  Because nitrogen is a major constituent of all forms of life and is cycled
10      through a complex web of processes involving many biotic and abiotic mechanisms, evaluating
11      forest nitrogen status is a challenge. Indicators of ecosystems at risk of nitrogen saturation
12      should include those that can be identified when nitrogen availability exceeds biotic demand.
        June 2003
        4-125
DRAFT-DO NOT QUOTE OR CITE

-------
 1      Such indicators typically should monitor parameters that are normally at background or low
 2      levels in nitrogen-limited systems and should be those that commonly respond to excess nitrogen
 3      in a wide range of ecosystems (Fenn and Poth,  1998). Such indicators include foliar nitrogen;
 4      nutrient ratios (N:P,N:cation); foliar nitrate; foliar 615 N; arginine concentration; soil C:N ratio;
 5      NO3" in soil extracts or in soil solution; and flux rates of nitrogenous trace gasses from soil (Fenn
 6      et al., 1998).  The cardinal indicator or manifestation of nitrogen saturation in all ecosystem
 7      types, including California forests and chaparral, is increased and prolonged NO3" loss below the
 8      main rooting zone and in stream water.  Seasonal patterns of stream water nitrate concentrations
 9      are especially good indicators of watershed nitrogen status (see sections on nitrate, sulfur and
10      acidic deposition; Fenn and Poth,  1998).
11           In Europe, the elements used in the critical load concept are:  a biological indicator, a
12      chemical criterion, and a critical value.  The biological indicator is the organism used to indicate
13      the status of the receptor ecosystem; the chemical criterion is the parameter that results in harm
14      to the biological indicator; and the critical value is the value of the chemical criterion below
15      which no significant harmful response occurs to the biological indicator (Lokke et al., 1996).
16      Trees, and sometimes other plants, are used as the biological indicators in the case of critical
17      loads for forests.  The critical load calculation using the current methodology, is essentially an
18      acidity/alkalinity mass balance calculation.  The chemical criterion must be expressible in terms
19      of alkalinity.  Initially, the Ca/Al ratio was used; but, recently, the (Ca+Mg+K)/Al ratio has been
20      used (Lokke etal., 1996).
21           Ideally, changes in acidic deposition should result in changes in  the status of the biological
22      indicator used in the critical load calculation. However, the biological indicator is the integrated
23      response to a number of different stresses.  Furthermore, there are other organisms more
24      sensitive to acid deposition than trees. At high concentrations, Al+3 is known to be toxic to
25      plants, inhibiting root growth and, ultimately, plant growth and performance (Lokke et al., 1996;
26      National Science and Technology Council, 1998).  Sensitivity to Al varies considerably between
27      species and within species because of changes in nutritional demands  and physiological status
28      that are related to age and climate. Experiments have shown that there are large variations in Al
29      sensitivity, even among  ecotypes.
30           Mycorrhizal fungi  as possible biological indicators have been suggested by Lokke et al.
31      (1996) because they are  intimately associated with tree roots, depend on plant assimilates, and

        June 2003                                 4-126       DRAFT-DO NOT QUOTE OR CITE

-------
 1      play an essential role in plant nutrient uptake influencing the ability of their host plants to
 2      tolerate different anthropogenically generated stresses. Mycorrhizas and fine roots are an
 3      extremely dynamic component of below-ground ecosystems and can respond rapidly to stress.
 4      They have a relatively short life span, and their turnover appears to be strongly controlled by
 5      environmental factors. Changes in mycorrhizal species composition or the loss of dominant
 6      mycorrhizal species in areas where diversity is already low may lead to increased susceptibility
 7      of plant to stress (Lokke et al., 1996). Stress affects the total amount of carbon fixed by plants
 8      and modifies carbon allocation to biomass, symbionts, and secondary metabolites. The
 9      physiology of carbon allocation has also been suggested as an indicator of anthropogenic stress
10      (Andersen and Rygiewicz, 1991). Because mycorrhizal fungi are dependent for their growth on
11      the supply of assimilates from the host plants, stresses that shift the allocation of carbon reserves
12      to the production of new leaves at the expense of supporting tissues will be reflected rapidly in
13      decreased fine root and mycorrhizzal biomass (Winner and Atkinson, 1986). Decreased carbon
14      allocation to roots also affects soil carbon and rhizosphere organisms.  Soil dwelling animals are
15      important for  decomposition,  soil aeration, and nutrient redistribution in the soil.  They
16      contribute to decomposition and nutrient availability mainly by increasing the accessibility of
17      dead plant material to microorganisms. Earthworms decrease in abundance and in species
18      number in acidified soils (Lokke et al., 1996).
19
20           Effects of Wet and Dry Deposition on Biogeochemical Cycling — The Integrated Forest
21      Study. The Integrated Forest Study (IPS; Johnson and Lindberg, 1992a) has provided the most
22      extensive data set available on wet and dry deposition and deposition effects on the cycling of
23      elements in forest ecosystems. The overall patterns of deposition and cycling have been
24      summarized by Johnson and Lindberg (1992a), and the reader is referred to that reference for
25      details. The following is a summary  of particulate deposition, total deposition, and leaching in
26      the IPS sites.
27           Particulate deposition in the IPS was separated at the 2-|im level; a decision was made to
28      include total parti culate deposition in this analysis and may include the deposition of particles
29      larger than 10 jim. Particulate deposition contributes considerably to the total impact of base
30      cations to most of the IPS  sites. On average, paniculate deposition contributes 47% to total
31      calcium deposition (range: 4  to 88%), 49% of total potassium deposition (range:  7 to 77%),

        June  2003                                4-127        DRAFT-DO NOT QUOTE OR CITE

-------
 1      41% to total magnesium deposition (range:  20 to 88%), 36% to total sodium deposition (range:
 2      11 to 63%), and 43% to total base cation deposition (range: 16 to 62%).  Of total particulate
 3      deposition, the vast majority (> 90%) is > 2 jim.
 4           Figures 4-20 through 4-23 summarize the deposition and leaching of calcium, magnesium,
 5      potassium, and total base cations for the IPS sites. As noted in the original synthesis (Johnson
 6      and Lindberg, 1992a), measurements indicated annual gains of base cations for some sites (i.e.,
 7      total deposition > leaching), some losses (total deposition < leaching), and some are
 8      approximately in balance. Not all cations follow the same pattern at each site. For example, a
 9      net accumulation of calcium occurs at the Coweeta, TN, Durham (Duke), NC, and Florida sites
10      (Figure 4-20), whereas accumulation of potassium was noted at the Duke, Florida, Thompson,
11      WA, Huntington Forest, NY, and White Face Mountain, NY, sites (Figure 4-22).  Magnesium
12      accumulated only at the Florida sites (Figure 4-21), and only at the Florida site is there a clear
13      net accumulation of total base  cations (Figure 4-23).
14           As noted previously, the  factors affecting net calcium accumulation or loss include the
15      soil-exchangeable cation composition; base cation deposition rate; the total leaching pressure
16      because of atmospheric sulfur  and nitrogen inputs, as well as natural (carbonic and organic)
17      acids; and biological demand (especially for potassium).  At the Florida site, which has a very
18      cation-poor, sandy soil (derived from marine sand), the combination of all these factors leads to
19      net base cation accumulation from  atmospheric deposition (Johnson and Lindberg, 1992a). The
20      site showing the greatest net base cation losses, the red alder stand in Washington state, is one
21      that is under extreme leaching  pressure by nitrate produced because  of excessive fixation by that
22      species (Van Miegroet and Cole, 1984).  In the red spruce site in the Smokies, the combined
23      effects of SO4"2 and NO3" leaching are even greater than in the red alder site (Figure 4-24), but a
24      considerable proportion of the  cations leached from this extremely acid soil consist of FT and
25      Al+3 rather than of base cations (Johnson and Lindberg, 1992a). Thus, the red spruce site in the
26      Smokies is approximately in balance with respect to calcium and total base cations, despite the
27      very high leaching pressure at  this  site (Figures 4-20 and 4-23).
28           The relative importance of particulate-base-cation deposition varies widely with site and
29      cation and is not always related to the total deposition rate. The proportion of calcium
30      deposition in parti culate form ranges from a low of 4% at the Whiteface Mountain site to a high
31      of 88% at the Maine site (Figure 4-20). The proportion of potassium deposition as particles

        June 2003                                4-128        DRAFT-DO NOT QUOTE OR CITE

-------
i,uuu -
500-
S n
0) U
'cc
-£ -500 -
CD
CO
o- -1,000-
LU
-1,500-
2r\r\r\
63%


^





41%


i
m



Pe
55%




| > 2 pirn
| < 2|JI71
^ Wet
^ Leaching
*—
CP

DL


rcent c
65%



—
I
I





)f total
40%


n
H





GS LP
FS
depos
26%


—
1






DF
ition a;
25%





s

1
I
I





5 partic
49%


^^^
^





RA
	 k. ^ 	
NS
;les:
58%



1





88%

M



4%



I




FF MS WF
t


16°/
1
1
I
3


ST
fe.
      Figure 4-20.  Calcium deposition in > 2-um particles, < 2-um particles, and wet forms
                   (upper bars) and leaching (lower bars) in the Integrated Forest Study sites.
                   CP = Pinus strobus, Coweeta, TN; DL = Pinus taeda, Durham (Duke), NC;
                   GS = Pinus taeda, B. F. Grant Forest, GA; LP = Pinus taeda, Oak Ridge, TN;
                   FS = Pinus eliottii, Bradford Forest, FL; DF = Psuedotsuga menziesii,
                   Thompson, WA; RA = Alnus rubra; Thompson WA; NS = Picea abies,
                   Nordmoen, Norway; HF = northern hardwood, Huntington Forest, NY;
                   MS = Picea rubens, Howland, ME; WF = Picea rubens, Whiteface Mountain,
                   NY;  and ST = Picea rubens, Clingman's Dome, NC.
1     ranges from 7% at the Smokies site to 77% at the Coweeta site (Figure 4-22), and the proportion

2     of total base cation deposition ranges from 16% at the Whiteface site to 62% at the Maine site

3     (Figure 4-23). Overall, paniculate deposition at the site in Maine accounted for the greatest
4     proportion of calcium, magnesium, potassium, and base cation deposition (88, 88, 57, and 62%,

5     respectively) even though total deposition was relatively low. At some sites, the relative

6     importance of particulate deposition varies considerably by cation.  At the Whiteface Mountain

7     site, parti culate deposition accounts for 4, 20, and 40% of calcium, magnesium, and potassium
      June 2003
4-129
DRAFT-DO NOT QUOTE OR CITE

-------
i,uuu -
500 -
'L_
CO
'to
-£ -500 -
CD
CO
o- -1,000-
LLJ
-1 ,500 -
.9 nnn -
63%


^





41%


i
1



Pe
55%




^ Wet
^ Leaching


rcent c
65%



—
1
I





)f total
40%


i
H





depos
26%


—
1





ition a;
25%





H

§
i
i





5 partic
49%


^_^_^
n





;les:
58%



n





88%

i



4%



i





i



16°/
1
1
I
3


             CP    DL    GS   LP     FS
            	Warmer Sites —
DF    RA    NS
 FF   MS   WF
-Colder Sites—
                                                                                       ST
       Figure 4-21.  Magnesium deposition in > 2-um particles, < 2-um particles, and wet forms
                     (upper bars) and leaching (lower bars) in the Integrated Forest Study sites.
                     See Figure 4-20 for site abbreviations.
 1
 2
 3
 4
 5
 9
10
11
12
deposition, respectively. At the red spruce site in the Smokies, particulate deposition accounts
for 46, 26%, 7% of calcium, magnesium, and potassium deposition, respectively.
     As indicated in the IPS synthesis, SO4"2 and NOj leaching often are dominated by
atmospheric sulfur and nitrogen (Johnson and Lindberg, 1992a). The exceptions to this are in
cases where  natural nitrogen inputs are high (i.e.,the nitrogen-fixing red alder stand), as are NO3"
leaching rates even though nitrogen deposition is low, and where soils adsorb much of the
atmospherically deposited SO4"2 thus reducing SO4"2 leaching compared to atmospheric sulfur
input.
     Sulfate and NO3" leaching have a major effect on cation leaching in many of the IPS sites
(Johnson and Lindberg, 1992a). Figure 4-24 shows the total cation leaching rates of the IPS sites
and the degree to which cation leaching is balanced by SO4"2 + NO3" deposition. The SO4"2 and
NO3" fluxes are subdivided further into that proportion potentially derived from particulate sulfur
       June 2003
                                         4-130
         DRAFT-DO NOT QUOTE OR CITE

-------
4UU
300-
200-
^ra 100-
0)
"ro
-| -200-
CT
LU
-300-
-400-
-500
77%




H
1




76%

H!
fc
1





| < 2|jm
P Wet
^ Leaching
CP
<4
Percent of total
61% 48% 54%


H
1












i 	 1
1









^





deposition a
40% 40%


1
Ed






DL GS LP FS DF




^
fi.
I






s partk
68%





•








RA NS
h. ^
elates
23%








57%


R
1




40%


R





7%







1
I
%
"/,
I
-


~\ZA
FF MS WF ST
       Figure 4-22.  Potassium deposition in > 2-um particles, < 2-um particles, and wet forms
                     (upper bars) and leaching (lower bars) in the Integrated Forest Study sites.
                     See Figure 4-20 for site abbreviations.
 1     and nitrogen deposition (assuming no ecosystem retention, a maximum effect) and other sulfur
 2     and nitrogen sources (wet and gaseous deposition, internal production).
 3           As noted in the IPS synthesis, total SO4"2 and NO3" inputs account for a large proportion
 4     (28 to 88%) of total cation leaching in most sites. The exception is the Georgia loblolly pine site
 5     where there were high rates of HCO3" and Cl" leaching (Johnson and Lindberg, 1992a).  The role
 6     of paniculate sulfur and nitrogen deposition in this leaching is generally very small (< 10%),
 7     however, even if it is assumed that there is no ecosystem sulfur or nitrogen retention.
 8           It was noted previously in this chapter that the contribution of particles to total deposition
 9     of nitrogen and sulfur at the IPS sites is lower than that for base cations. On average, particulate
10     deposition contributes 18% to total nitrogen deposition (range:  1 to 33%) and 17% to total
11     sulfur deposition (range:  1 to 30%). Particulate deposition contributes only a small amount to
12     total H+ deposition (average = 1%; range: 0 to 2%). (It should be noted, however,
13     that particulate FT deposition in the > 2 jim fraction was neglected.)
       June 2003
4-131
DRAFT-DO NOT QUOTE OR CITE

-------
o,uuu -
2,000 -
L 1,000-
CD
cb 0-
JZ
-i-i
c
1 -1,000-
>
'D
U -2,000 -
-3,000 -
A nnn -
53%

1

47%


//
t/L


^ Wet
^ Leaching
P
48%

1




ercent
62%

1
1




of tota
49%
^_^_^
1




1 depo
28%


I




sition i
28%

--


1
I
I




is part
47%

1




culate
44%

I



s:
62%


1




16%

1




31%




1
1
1





                     CP    DL   GS    LP    FS   DF   RA    NS    FF    MS   WF    ST
                                Warmer Sites	^	Colder Sites
       Figure 4-23.  Base cation deposition in > 2-um particles, < 2-um particles, and wet forms
                     (upper bars) and leaching (lower bars) in the Integrated Forest Study sites.
                     See Figure 4-20 for site abbreviation.
 1           Based on the IPS data, it appears that the parti culate deposition has a greater effect on base
 2     cation inputs to soils than on base cation losses associated with inputs of sulfur, nitrogen, and H+.
 3     It cannot be determined what fraction of the mass of these particles are < 10 jim, but only a very
 4     small fraction is < 2 jim.  These inputs of base cations have considerable significance, not only
 5     to the base cation status of these ecosystems, but also to the potential of incoming precipitation
 6     to acidify or alkalize the soils in these ecosystems. As noted above, the potential of precipitation
 7     to acidify or alkalize soils depends on the ratio of base cations to H+ in deposition, rather than
 8     simply on the inputs of H+ alone.  In the case of calcium, the term "lime potential" has been
 9     applied to describe this ratio; the principle is the same with respect to magnesium and potassium.
10     Sodium is a rather special case,  in that it is a poorly absorbing cation and leaching tends to
11     balance input over a relatively short term.
12           Net balances of base cations tell only part of the story as to potential effects on soils; these
13     net losses or gains must be placed in the perspective of the soil pool size.  One way  to express
       June 2003
4-132
DRAFT-DO NOT QUOTE OR CITE

-------
             7,000
             6,000
                   P: 4 %
                   0:28%
        •^   5,000 -
         CO
rro   4,000 -
         76   3,000 H
             2,000 -
             1,000
                        % of total cation leaching balanced by SO J" and NO 5 from Particles (P) and other (O) sources
 7%
55%
                         1%
                         6%
 8%
78%
 9%
30%
 10%
.42%. ,.88%.
                        Other Anions
                        Particulate Sulphur and Nitrogen
                        Other Sulphur and Nitrogen Sources
10%
 8%
55%
 6%
73%
18%
69%
 1%
77%
 9%
69%
                                                               I
                           DL    GS
             LP
                                      FS
             DF    RA   NS    HF    MS    WF    ST
       Figure 4-24.  Total cation leaching (total height of bar) balanced by sulfate and nitrate
                     estimated from particulate deposition (assuming no ecosystem retention,
                     particulate sulfur and nitrogen) and by other sources (both deposition and
                     internal) of sulfate and nitrate (other sulfur and nitrogen sources) and by
                     other anions in the Integrated Forest Study sites. See Figure 4-20 for site
                     abbreviations.
 1     this perspective is to simply compare soil pool sizes with the net balances. This comparison is
 2     made for exchangeable pools and net balances for a 25-year period in Figures 4-25 to 4-27.
 3     It readily is seen that net leaching losses of cations pose no threat in terms of depleting
 4     soil-exchangeable Ca+2, K+, or Mg+2 within 25 years at the Coweeta, Duke, Georgia, Oak Ridge,
 5     or Douglas-fir sites. However, there is a potential for significant depletion at the red alder,
 6     Whiteface Mountain (magnesium), and Smokies red spruce sites.
 7           The range of values for soil-exchangeable turnover is very large, reflecting variations in
 8     both the size of the exchangeable pool and the net balance of the system.  Soils with the highest
 9     turnover rates are those most likely to experience changes in the shortest time interval, other
10     things being equal.  Thus, the Whiteface Mountains, Smokies, and Maine red spruce sites; the
11     Thompson red alder site; and the Huntington Forest northern hardwood site appear to be most
       June 2003
                      4-133
                                                     DRAFT-DO NOT QUOTE OR CITE

-------
oou.uuu •
300,000
CD
>, 250,000
'co
•£. 200,000
,? 150,000
'co
f! 100,000
c
> 50,000
'D
cr
m o
-?;n nnn .






Q








-









i — i

—









~











u






Q






,






--C3-


Q Soil Exchangeable
• (Dep-Leaching)*25





a





°





=





-•=•
                   CP   DL   GS   LP    FS    DF   RA   NS   HF   MS   WF   ST
                •4	Warmer Sites	^--^	Colder Sites	>•
Figure 4-25.  Soil exchangeable Ca+2 pools and net annual export of Ca+2 (deposition minus
             leaching times 25 years) in the Integrated Forest Study sites. See Figure 4-20
             for site abbreviations.
iuu,uuu •
80,000 -
1_
CD
£ 60,000 -
CD
.C
~ 40,000 -
CD
CD
'g. 20,000 -
0 -
-9n nnn -




0





i — 1
-







I — I



—






	





II







y




-




1




-


Q Soil Exchangeable
| (Dep - Leaching)*25



—



-



•



H
                  CP   DL   GS    LP   FS   DF   RA   NS   HF   MS   WF   ST
                             Warmer Sites	>• -4	Colder Sites
Figure 4-26.  Soil exchangeable Mg+2 pools and net annual export of Mg+2 (deposition
             minus leaching times 25 years) in the Integrated Forest Study sites.
             See Figure 4-20 for site abbreviations.
June 2003
4-134
DRAFT-DO NOT QUOTE OR CITE

-------
IUU.UUU •
140,000 -
120,000 -
1_
8 100,000 -
^^i
ro 80,000 -
.c
§5 60,000 -
1
•5 40,000 -
CT
LU
20,000 -
0 -
20 000 •







II








[]









1 — 1

















































D

















D



Q Soil Exchangeable
H(Dep - Leaching)*25






i — i







n







_JHL







n
•
CP DL GS LP FS DF RA NS HF MS WF ST
       Figure 4-27.  Soil exchangeable K2+ pools and net annual export of K2+ (deposition minus
                     leaching times 25 years) in the Integrated Forest Study sites. See Figure 4-20
                     for site abbreviations.
 1      sensitive to change. The actual rates, directions, and magnitudes of changes that may occur in
 2      these soils (if any) will depend on weathering inputs and vegetation outputs, in addition to
 3      deposition and leaching. It is noteworthy that each of the sites listed above as sensitive has a
 4      large store of weatherable minerals; whereas many of the other soils, with larger exchangeable
 5      cation reserves, have a small store of weatherable minerals (e.g., Coweeta white pine, Duke
 6      loblolly pine, Georgia loblolly pine, and Oak Ridge loblolly pine; Johnson and Lindberg, 1992a;
 7      April and Newton, 1992).
 8           Base cation inputs are especially important to the Smokies red spruce site because of
 9      potential aluminum toxicity and calcium and magnesium deficiencies.  Johnson et al. (199la)
10      found that soil- solution aluminum concentrations occasionally reached levels found to inhibit
11      calcium uptake and cause changes in root morphology in solution culture studies of red spruce
12      (Raynal et al., 1990).  In a follow-up study, Van Miegroet et al. (1993) found a slight but
13      significant growth response to calcium and magnesium fertilizer in red spruce saplings near the
14      Smokies red  spruce site. Joslin et al. (1992) reviewed soil and solution characteristics of red
15      spruce in the southern Appalachians, and it appears that the IPS site is rather typical.
       June 2003
4-135
DRAFT-DO NOT QUOTE OR CITE

-------
 1           Wesselink et al. (1995) reported on the complicated interactions among changing
 2      deposition and soils at this site (including repeated sampling of soil exchangeable base cation
 3      pools) from 1969 to 1991 and compared these results with those of a simulation model. They
 4      identified three basic stages of change in this ecosystem. During Stage 1, there was increased
 5      deposition of sulfur and constant deposition of base cations, causing increased base cation
 6      leaching and reduced base saturation in the soils.  During Stage II, sulfur deposition is reduced;
 7      soil solution sulfate and base cation leaching decline accordingly, but base saturation continues
 8      to decrease.  During Stage III, two alternative  scenarios are introduced:  (a) sulfur deposition
 9      continues to decline, whereas base cation deposition says constant; or (b) both sulfur and base
10      cation deposition decline.  Under Stage Ill-a, sulfate and base cation leaching continue to
11      decline, and base saturation begins to increase as base cations displace exchangeable aluminum
12      and cause it to transfer to the gibbsite pool.  Under Stage Ill-b, this recovery in base saturation is
13      over-ridden by the reduction in base cation deposition.
14           The IPS project, for the first time, accurately quantified atmospheric deposition inputs to
15      nutrient cycles using state-of-the-art techniques to measure wet and dry deposition. The
16      principal aim of the project was to determine the effects of atmospheric deposition on the
17      nutrient status of a variety of forest ecosystems and to determine if these effects are in any way
18      related to current or potential forest decline. Acidic deposition is having a significant effect on
19      nutrient cycling in most of the forest ecosystems studied in the IPS project.  The exceptions were
20      the relatively unpolluted Douglas fir, red alder, and Findley Lakes in Washington state. The
21      nature of the effects, however,  varies from one location to another (Johnson, 1992).  In all but the
22      relatively unpolluted Washington sites,  atmospheric deposition was having a significant, often
23      overwhelming effect on cation leaching from the soils. In general, nutrient budget data from IPS
24      and the literature suggest that the susceptibility of southeastern sites to base-cation-depletion
25      from soils and the development of cation deficiencies by that mechanism appears to be greater
26      than in northern sites (Johnson, 1992).
27           Atmospheric deposition may have affected significantly the nutrient status of some US
28      sites through the mobilization of Al. Soil-solution Al levels in the Smokies sites approach and
29      sometimes exceed levels noted to impede cation uptake in solution culture studies.  It is therefore
30      possible that the rates of base cation uptake  and cycling in these sites have been reduced because
31      of soil solution Al. To the extent that atmospheric deposition has contributed to these elevated

        June 2003                                4-136        DRAFT-DO NOT QUOTE OR CITE

-------
 1      soil-solution Al levels, it has likely caused a reduction in base cation uptake and cycling rates at
 2      these sites.  Nitrate and sulfate are the dominant anions in the Smokies sites, and nitrate pulses
 3      are the major cause of Al pulses in soil solution (Johnson, 1992). The connection between Al
 4      mobilization and forest decline is not clear. The decline in red spruce certainly has been more
 5      severe in the Northeast than in the Southeast, yet all evidence indicates that Al mobilization is
 6      most pronounced in the southern Appalachians. However, at the Whiteface Mountain site
 7      selected for study because it was in a state of decline, and soil solution levels were lower than in
 8      the Smokies, which are not in a visibly obvious state of decline (there was no dieback other than
 9      the fir killed by the balsam wooly adelgid, no needle yellowing). Thus, Al mobilization
10      constitutes a situation worthy of further study (Johnson, 1992).
11           The simple calculations shown above give some idea of the importance of particulate
12      deposition in these forest ecosystems, but they  cannot account for the numerous potential
13      feedbacks between vegetation and soils nor for the dynamics through time that can influence the
14      ultimate response. One way to examine some of these interactions and dynamics is to use
15      simulation modeling. The nutrient cycling model (NuCM) has been developed specifically for
16      this purpose and has been used to explore the effects of atmospheric deposition, fertilization, and
17      harvesting on some of the IPS sites (Johnson et al., 1993). The NuCM model is a stand-level
18      model that incorporates all major nutrient cycling processes  (uptake, translocation, leaching,
19      weathering, organic matter decay, and accumulation).
20           Johnson et al. (1999) used the NuCM model to simulate the effects of reduced S, N, and
21      base cation (CB) deposition on nutrient pools, fluxes, soil, and soil solution chemistry in two
22      contrasting southern Appalachian forest ecosystems:  the red spruce and Coweeta hardwood sites
23      from the IPS project.  The scenarios chosen for these simulations included "no change"; 50% N
24      and S  deposition; 50% CB deposition; and 50% N, S, and CB deposition (50% N, S, CB). The
25      NuCM simulations suggested that, for the extremely acid red spruce site, S and N deposition is
26      the major factor affecting soil-solution Al concentrations and CB deposition is the major factor
27      affecting soil solution CB concentrations.  The effects of S and N deposition were largely through
28      changes in soil solution SO4"2 and NO3" and, consequently, mineral  acid anion (MAA)
29      concentrations rather than through changes in soils. This is illustrated in Figures 4-28a,b and
30      4-29a,b, which show simulated soil-solution mineral-acid anions, base cations, Al, and soil base
31      saturation in B horizon from in the red spruce site. The 50% S and N scenario caused reductions

        June 2003                                 4-137       DRAFT-DO NOT QUOTE OR CITE

-------
to
o
o
oo
H
6
o

o
H
O
              o
             O
             E
             0)
             D)
             CD
             .C
             O
             M—
             O
             ro
             "c
             O
             c
             o
             o
                                                Red Spruce
                500 —
                400 —
                300 —
             O 200-
100 —
                   0'
                                                        Al
                        — No Change
                        - 50% N,S
                        -- 50% BC
                         \
                         2
                               \
                               4
                     \
                     6
8
10
 i      i
12     14
Year
16    18
 \
20
22    24
     Figure 4-28a. Simulated soil solution mineral acid anions and base cations in the red spruce site with no change, 50% N
                 and S deposition, and 50% base cation deposition. Redrawn from Johnson et al. (1999).
o
HH
H
W

-------
to
O
o
J^.
 I


VO
H
6
o

o
H
O
              O
              E
              E
                 200
                  160
              E?  120
              (0
              .c
              O
              o
              E
              .2
              ll_l
              m
              Hh->
              c
              0)
              o
              c
              o
              O
                   80
                   40
                    0
                                                         Spruce
                                                         Cations
                          — No Change
                          — 50% N,S
                          ~ 50% BC
                                i
                                4
                                       i
                                       6
i
8
10
 i
 12
Year
14
16
18
20
 i
22
24
      Figure 4-28b. Simulated soil solution mineral acid anions and base cations in the red spruce site with no change, 50% N
                  and S deposition, and 50% base cation deposition. Redrawn from Johnson et al. (1999).
o
HH
H
W

-------
to
O
o
J^.
I

o
H
6
o

o
H
O
               co
               0)
               o
               c
               o
              O
                                                  Red Spruce
                o
               O
               E
               0)
               D)
               OJ
               .c
               O
               M—
               O
                  500 —
                  400 —
                  300 —
               O  200-
                  100 —
                                                         Al
                          — No Change
                          - 50% N,S
                          -- 50% BC
                           \
                          2
                                 \
                                 4
\
6
\
8
10
 i      i
12     14
Year
 \
16
18    20    22    24
       Figure 4-29a. Simulated soil solution Al and soil base saturation in the red spruce site with no change, 50% N and S
                   deposition, and 50% base cation deposition. Redrawn from Johnson et al. (1999).
o
HH
H
W

-------
to
O
o
H

6
O


o
H

O
              __

              O
              cp
              S>
              cs
              JZ
              O
              c
              _g
              v^
              2
              ^_*


              I
              o
              O
                        — No Change
                        — 50%N,S

                        -- 50% BC
                                                Red  Spruce
                                                     Saturation
                        i
                        2
                               i
                               4
                                                                                     V,  -V
                                            •''  ,-   A  '

                                        - .   v/V/V
                                       A  ,   /',  '^
                                    i •  "'    I '  ' V
                            *  *t  A  ' I / I A  , \*
                            i >  "  i ,''''/»  i ^
                            ' >  M  ,M ' l '  i / i ,
                                                                               11
                                                                               
-------
 1      in soil solution SO4"2, NO3" and, therefore, MAA concentrations, as expected.  This, in turn,
 2      caused short-term reductions in base cation concentrations. However, by the end of the 24-year
 3      simulation, base cations in the 50% S, N scenario were nearly as high as in the no change
 4      scenario because base saturation had increased and the proportion of cations as Al decreased.
 5      The 50% CB scenario had virtually no effect on soil solution SO4"2, NO3" and, therefore, MAA
 6      concentrations, as expected, but did cause a long-term reduction in base cation concentrations.
 7      This was caused by a long-term reduction in base saturation (Figure 4-29a,b).  Thus, the effects
 8      of CB deposition were solely through changes in soils rather than through changes in soil solution
 9      MAA, as postulated by Driscoll et al. (1989).  In the less acidic Coweeta soil, base saturation
10      was high and little affected by the scenario (cited above); Al was unimportant; and S and
11      N deposition had a much greater effect than CB deposition in all respects (Figure 4-30a,b).
12           In summary, Johnson et al. (1999) found that the results of the red spruce simulations
13      support the hypothesis of Driscoll et al. (1989) in part: CB deposition can have a major effect on
14      CB leaching through time in an extremely acid system. This effect occurred through changes in
15      the soil exchanger and not through changes in soil solution MAA concentration.  On the other
16      hand, S and N deposition had a major effect on Al leaching at the Noland Divide site.  This
17      occurred  primarily because of changes in soil  solution MAA  concentration. At the less acidic
18      Coweeta  site, CB deposition had a minor effect on soils and soil solutions; whereas S and N
19      deposition had delayed but major effects on CB leaching because of changes in SO4"2 and MAA
20      concentrations.
21
22      Effects of Trace Elements
23           Trace metals are natural elements that are ubiquitous in small (trace) amounts in soils,
24      ground water, and vegetation. Many are essential elements required for growth by plants and
25      animals as micronutrients. Naturally occurring surface mineralizations can produce metal
26      concentrations in soils and vegetation that are as high, or higher, than those in the air and
27      deposited near man-made sources (Freedman  and Hutchinson, 1981).  The occurrence and
28      concentration of trace metals in any ecosystem component depend on the sources of the metal
29      (i.e., via the soil or as a particulate).  Even when air pollution is the primary source, continued
30      deposition can result in the accumulation of trace metals in the soil (Martin and Coughtrey,
        June 2003                                4-142        DRAFT-DO NOT QUOTE OR CITE

-------
to
O
o
    100
o

E
CO


6
c
.o
".J3

s
"c

o
c
o
o
                  80 -
                  60
                  40
                  20
                                               Co we eta
                                           Mineral Acid Anions
                                                                               No Change

                                                                               50% N,S

I
2
i
4
i
6
i
8
i
10
\
12
i
14
i
16
i
18
i
20
\
22
2
                                                    Year
      Figure 4-30a.  Simulated soil solution mineral acid anions and base cations in the Coweeta site with no change, 50% N and S

                  deposition, and 50% base cation deposition. Redrawn from Johnson et al. (1999).

-------
to
O
o
H
6
o

o
H
O
                                Coweeta
                100
              p
             o

             E,

             0)
             E1
             CO

            6
             c
            .2
            v-»
             m



             I
             o
            O
80 -
60 -
                                              Base Cations
                 40
                                                              No Change

                                                              50% N,S

                                                            - 50% BC
                       i
                       2
                        8
10
 i
12
14    16
 i
18
 \
20
 i
22
24
                                                    Year
      Figure 4-30b. Simulated soil solution mineral acid anions and base cations in the Coweeta site with no change, 50% N and S

                 deposition, and 50% base cation deposition. Redrawn from Johnson et al. (1999).
o
HH
H
W

-------
 1      1981).  Many metals are deposited into soils by chemical processes and are not available to
 2      plants (Saunders and Godzik, 1986).
 3           When aerial deposition is the primary source of metal particles, both the chemical form and
 4      particle size deposited determine the heavy metal concentration in the various ecosystem
 5      components (Martin and Coughtrey, 1981). Human activities introduce heavy metals into the
 6      atmosphere and have resulted in the deposition of antimony, cadmium, chromium, copper, lead,
 7      molybdenum, nickel, silver, tin, vanadium, and zinc (Smith, 1990c). Extensive evidence
 8      indicates that heavy metals deposited from the atmosphere to forests accumulate either in the
 9      richly organic forest floor or in the soil layers immediately below, areas where the activity in
10      roots and soil is greatest. The greater the depth of soil, the lower the metal concentration.  The
11      accumulation of metal in the soil layers where the biological activity is greatest, therefore, has
12      the potential to be toxic to roots and soil  organisms and to interfere with nutrient cycling (Smith,
13      1990c). The shallow-rooted plant species are those most likely to take up metals from the soil
14      (Martin and Coughtrey, 1981). Though all metals can be directly toxic at high levels, only
15      toxicity from copper, nickel, and zinc have been documented frequently. Toxicity of cadmium,
16      cobalt, and lead has been seen  only under unusual conditions (Smith, 1990c).  Exposures at
17      lower concentrations have the potential, over the long-term, to interfere with the nutrient-cycling
18      processes when they affect mycorrhizal function.
19           Biological accumulation  of metals through the plant-herbivore and litter-detrivore chains
20      can occur. A study of the accumulation of cadmium, lead, and zinc concentrations in
21      earthworms suggested that cadmium and zinc were concentrated, but not lead. Studies indicate
22      that heavy metal deposition onto the soil, via food chain accumulation, can cause excessive
23      levels and toxic effects in certain animals.  Cadmium appears to be relatively mobile within
24      terrestrial food  chains; however, the subsequent mobility of any metal after it is ingested by a
25      herbivorous animal depends on the site of accumulation within body tissues.  Although food
26      chain accumulation may not in itself cause death, it can reduce the breeding potential in a
27      population (Martin and Coughtrey, 1981).
28           In actual case studies, it was observed that the deposition of copper and zinc particles
29      around a brassworks resulted in an accumulation of incompletely decomposed litter.  In one
30      study, litter accumulation was reported up to 7.4 km from the stack of a primary smelter in
31      southeastern Missouri.  Similar results were reported around a metal smelter at Avonmouth,

        June  2003                                4-145        DRAFT-DO NOT QUOTE OR CITE

-------
 1      England.  In the latter case, litter accumulation was associated closely with concentrations
 2      specifically of cadmium, as well as with those of lead, copper, and zinc (Martin and Coughtrey,
 3      1981). Experimental data (using mesh bags containing litter) supports the hypothesis that
 4      reduced decomposition occurs close to heavy metal sources.
 5           Accumulations of metals emitted in particulate matter also were reported in soil litter close
 6      to a metal smelter at Palmerton, PA, in 1975 and 1978. The continued presence of cadmium,
 7      lead,  zinc, and copper in the upper soil horizons (layers) were observed 6 years after the smelter
 8      terminated operation in 1980. Metal levels were highest near the smelter. The relationship of
 9      decreasing amounts of metal in body tissues also held true for amphibians and mammals. Levels
10      of cadmium in kidneys and liver of white-tailed deer (Odocoileus virginaus) were five times
11      higher at Palmerton than in those collected 180 km southwest downwind. The abnormal
12      amounts of metal in the tissues of terrestrial vertebrates and the absence or low abundance of
13      wildlife at Palmerton indicated that ecological processes within 5 km of the  zinc smelter
14      continued to be markedly influenced even 6 years after its closing (Storm et al., 1994).
15           Accumulation of heavy metals in litter presents the greatest potential for interference with
16      nutrient cycling. Microorganisms are essential for decomposition of organic matter and soil
17      fertility.  Toxic effects on the microflora can be caused by Zn, CD, and Cu.  Addition of a few
18      mg/kg of soil of Zn can inhibit the more sensitive microbial processes (van Beelen and Doelman,
19      1997). Experiments by Kandeler et al. (1996) indicated that microbial biomass and enzyme
20      activities decreased with increasing heavy metal concentrations. The amount of decrease varied
21      among the enzymes with those involved in carbon cycling least affected, whereas the activities
22      of the enzymes involved in the cycling of N, P and S, especially arylsulfatase and phosphatase
23      were  dramatically affected.
24           Accumulation of metals in the litter occurs chiefly around brass works and lead and zinc
25      smelters.  There is some evidence that invertebrates inhabiting soil litter do accumulate metals.
26      Earthworms from roadsides were shown to contain elevated concentrations of cadmium, nickel,
27      lead,  and zinc; however, interference with earthworm activity was not cited  (Martin and
28      Coughtrey, 1981).  It has been shown, however, that when soils are acidic, earthworm abundance
29      decreases, and bioaccumulation of metals from soil may increase exponentially with decreasing
30      pH (Lokke et al, 1996). Organisms that feed on earthworms living in soils with elevated levels
31      of Cd, Ni, Pb, and Z for extended periods could accumulate lead and zinc to toxic levels (Martin

        June 2003                                 4-146       DRAFT-DO NOT QUOTE OR CITE

-------
 1      and Coughtrey, 1981). Increased concentrations of heavy metals have been found in a variety of
 2      small mammals living in areas with elevated heavy metal concentrations in the soils.
 3           Studies by Babich and Stotsky (1978) support the concept that increased accumulation of
 4      litter in metal-contaminated areas is due to the effects on the microorganismal populations.
 5      Cadmium toxicity to microbial populations was observed to decrease and prolong logarithmic
 6      rates of microbial population increase, to reduce microbial respiration and fungal spore
 7      formation and germination, to inhibit bacterial transformation, and to induce abnormal
 8      morphologies.  Additionally, the effects of cadmium, copper, nickel, and zinc on the symbiotic
 9      activity of fungi, bacteria, and actinomycetes were reported by Smith (1991).  The formation of
10      mycorrhizae by Glomus mosseae with onions was reduced when zinc, copper, nickel, or
11      cadmium was added to the soil.  The relationship of the fungus with white clover, however, was
12      not changed. It was suggested that the effect of heavy metals on vesicular-arbuscular
13      mycorrhizal fungi will vary from host to host (Gildon and Tinker, 1983).  Studies with ericoid
14      plants indicated that, in addition to Calluna vulgaris, mycorrhizae also protect Vaccinium
15      macrocarpa and Rhodendronponticum from heavy metals (Bradley et al., 1981). Heavy metals
16      tend to accumulate in the roots, and shoot toxicity is prevented.
17           The effects of sulfur deposition on litter decomposition in the vicinity of smelters also must
18      be considered.  Metal smelters emit SO2 as well as heavy metals. Altered litter decomposition
19      rates have been well documented near SO2 sources (Prescott and Parkinson, 1985). The
20      presence of sulfur in litter has been associated with reduced microbial activity (Bewley and
21      Parkinson, 1984).  Additionally, the effects on symbiotic activity of fungi, bacteria, and
22      actinomycetes were reported by Smith (1990d).
23           The potential pathways of accumulation of trace metals in terrestrial ecosystems, as well as
24      the possible consequences of trace metal deposition on ecosystem functions, is summarized in
25      Figure 4-31. The generalized trophic levels found in an ecosystem and the various physiological
26      and biological processes that could be affected by trace metals are shown in the figure.
27      Reduction in physiological processes can affect productivity, fecundity, and mortality (Martin
28      and Coughtrey, 1981). Therefore, any effects on structure and function of an ecosystem are
29      likely  to occur through the soil and litter (Tyler, 1972).
30           Certain species of plants are tolerant of metal-contaminated soils (e.g., soils from mining
31      activities) (Antonovics et al., 1971). Certain species of plants also have been used as

        June 2003                                 4-147       DRAFT-DO NOT QUOTE OR CITE

-------
         1. Wet/dry deposition
                                                                               . Retranslocation
                                   Atmosphere
         3. Litterfall, resuspension,
           deposition, leaching,
           stem flow
Plant Surface
Phyllosphere
                                                                       Ground
                                                                      Storage,
                                                                     Metabolism
                                                                           4. Translocation
          Biologically
          Unavailable
  Biologically
  Available
IX.
Soil Organic

X.
Primary
Minerals
11. Mineralization
•»-
%
^-
12. Weathering
IV
Upper Soil
<, f7. Leac
VII.
Lower Soil
77
                                                5. Mass flow,
                                                  diffusion
V.
Rhizosphere
Rhizoplane

•w
6. Root"
uptake
                                                5. Mass flow,
                                                  diffusion
                                         7. Leaching
                                      VIII.
                                   Groundwater
jo. Root
    turnover
                                                                                   VI.
                                                                               Root Storage
                                                                                Metabolism
      Figure 4-31. Relationship of plant nutrients and trace metals with vegetation.
                   Compartments (roman numerals) represent potential storage sites;
                   whereas arrows (arabic numerals) represent potential transfer routes.
1     bioindicators of metals (e.g., Astragalus is an accumulator of selenium). The sources of both
2     macroelements and trace metals in the soil of the Botanical Garden of the town of Wroclow,
3     Poland, were determined by measuring the concentrations of the metals in Rhododendron
4     catawbiense, Ilex aquifolium, and Mahonia aquifolium growing in a garden and comparing the
5     results with the same plant species growing in two botanical gardens in nonpolluted areas. Air
6     pollution deposition was determined as the source of metals in plants rather than the soil
7     (Samecka-Cymerman and Kempers, 1999).
8           The effects of lead in ecosystems are discussed in the EPA document Air Quality Criteria
9     for Lead (U.S. Environmental Protection Agency, 1986).  Studies have shown that there is cause
      June 2003
             4-148
               DRAFT-DO NOT QUOTE OR CITE

-------
 1      for concern in three areas in which ecosystems may be extremely sensitive to lead: (1) delay of
 2      decomposition because the activity of some decomposer microorganisms and invertebrates is
 3      inhibited by lead, (2) subtle shifts toward plant populations tolerant of lead, and (3) lead in the
 4      soil and on the surfaces of vegetation which may circumvent the processes of biopurification.
 5      The problems cited above arise because lead is deposited on the surface of vegetation,
 6      accumulates in the soil, and is not removed by the surface and ground water of the ecosystem
 7      (U.S. Environmental Protection Agency,  1986).
 8
 9      4.2.4 Urban Ecosystems
10           Humans dominate Earth's ecosystems.  Their influence on the environment has been
11      pervasive for thousands of years.  Evidence has been accumulating from anthropological and
12      archeological research that human influence has been pervasive for thousands of years (Grimm
13      et al., 2000).  Major human effects on the environment probably began as  early as 12,000 to
14      15,000 years ago and continue to be a major influence on all natural ecosystems. Nowhere has
15      human action been more intense than in cities, suburbs, exurbs and in the supporting hinterlands
16      (Grimm et al., 2000). This fact has lead to much recent interest in the study of urban ecological
17      systems.
18           Vitousek et al. (1997) point out that understanding a human-dominated planet requires that
19      the human dimensions of global change — the social, cultural, and other drivers of human
20      actions — need to be included within ecological analyses. Therefore, humans must be integrated
21      into models for a complete understanding of extant ecological systems.  Development of more
22      realistic models for ecological systems will lead to greater success in finding solutions to
23      environmental problems.
24           In the past, ecological plant or animal studies conducted in urban settings used traditional
25      ecological approaches and considered humans as agents of disturbance. Although the term urban
26      ecosystem has been used to describe human-dominated ecosystems, it does not adequately take
27      into account the developmental history, sphere of influence, and potential  impacts required in
28      order to understand the true nature of an urban ecosystem (Mclntyre, et al., 2000). Because
29      urbanization is both an ecological and a social phenomenon, urban ecology implicitly recognizes
30      the role humans play in developing unique systems.  Therefore, if urban ecology is to be a truly
        June 2003                                4-149       DRAFT-DO NOT QUOTE OR CITE

-------
 1      interdisciplinary field, it is crucial that it integrate both social and natural sciences into the study
 2      of urban ecosystems (Mclntyre, et al., 2000).
 3           Although the study of ecological phenomena in urban environments is not a new area of
 4      science, the concept of the city as an ecosystem is relatively new for the field of ecology (Grimm
 5      et al., 2000). There is a wealth of information on the terrestrial components of urban ecological
 6      systems.  However, much of it is organized from the perspective of ecology in cities while the
 7      more comprehensive perspective identified as ecology of cities is needed (Pickett et al., 2001).
 8      The basic questions addressed by the literature of ecology in cities are how do ecological
 9      patterns and processes differ in cities as compared with other environments? What is the effect
10      of the city (i.e., a concentration of human population and activities) on the ecology of organisms
11      inside and outside of its boundary and influence?  The concept of ecology o/"cities has to do with
12      how aggregated parts make up the whole (i.e., how cities process energy or matter relative to
13      their surroundings; Grimm et al., 2000). The latter concept includes primary production, species
14      richness,  biogeophysical budgets, ecosystem patterns and processes, and an open definition of
15      urban ecosystems that incorporates the exchanges of materials and influence between cities and
16      surrounding landscapes (Pickett et al., 2001). If ecosystems are to be understood, there is a need
17      for a new integrative ecology that explicitly incorporates human decisions,  culture, institutions,
18      and economic systems (Grimm et al., 2000).  This fact makes an ecological approach to land use
19      planning  not only necessary but essential to maintain long-term sustainability of ecosystem
20      benefits,  services, and resources (Zipperer et al., 2000).  The ecological  and social effects of
21      "edge city" need to be studied as they may be greater than the previous patterns of
22      suburbanization. The classical ecosystem approach and  a patch dynamic approach are needed
23      to understand and manage the dynamics of urban and urbanizing ecosystems (Zipperer et al.,
24      2000).
25           There has been little work on the rates of atmospheric deposition to urban ecosystems
26      despite the large body of knowledge on concentrations and chemical reactions of air pollutants in
27      cities. A search of the abundant literature produced no references that dealt with the effects of
28      PM deposition.  Lovett et al. (2000), however, reported that urban ecosystems are likely to be
29      subjected to large rates of deposition of anthropogenic pollutants. Decades of research on urban
30      air quality indicate that cities are often sources of nitrogen oxides, sulfur oxides,  and dust,
31      among many other pollutants. Some of these air pollutants are major plant nutrients (e.g.,

        June 2003                                 4-150       DRAFT-DO NOT QUOTE OR CITE

-------
 1      nitrogen) and may be affecting nutrient cycles in plant-dominated areas in and around cities.
 2      The gases and particles in urban air can increase atmospheric deposition within and downwind of
 3      the city.  Studying the deposition rates of atmospheric pollutants in urban areas can provide a
 4      quantitative estimate of the amounts of gaseous and particulate air pollutants that are removed by
 5      urban vegetation.
 6           To determine the patterns of atmospheric deposition and throughfall in the vicinity of a
 7      large city, Lovett et al. (2000) measured bulk deposition, oak forest throughfall, and paniculate
 8      dust at  sites along a transect within and to the north of New York City. They observed that
 9      concentrations and fluxes of NO3", NH4+, Ca+2, Mg+2,  SO4"2,  and Cl" in throughfall all declined
10      significantly with distance from the city, while hydrogen ion concentration and flux increased
11      significantly with distance from the city.  Most of the change in concentrations and fluxes
12      occurred  within 45  km of the city.  Additionally, it was observed that throughfall nitrogen was
13      twice as high in the urban areas when compared with the suburban and rural sites.  Most of the
14      dry deposition of nitrate was from gaseous nitrogen oxides.  As mentioned above, the effects of
15      the atmospheric deposition of the parti culate pollutants was not mentioned.
16           McDonnell et al. (1997) in a 10-year study of ecosystem processes along an urban-rural
17      gradient included plant litter dynamics and nitrogen cycling of two key components of a forest
18      ecosystem:  litter decomposition and heavy metal levels in soil and foliar litter. Foliar litter
19      decomposition integrates many features  of the abiotic and biotic environment. It is an important
20      site of heavy metal incorporation into ecosystems and provides a both a habitat and a resource
21      for fungi, bacteria,  and invertebrates.  Litter decomposition  integrates the effects of resource
22      quality, environmental factors, and activities of decomposer organisms on nutrient cycling and
23      serves as  an easily measured indicator of the effect of urbanization on an important ecosystem
24      function.  McDonnell et al. (1997) noted that levels of heavy metals in the foliar litter in urban
25      forest soils were higher than in rural forest soils. The levels in urban forest stands  approached  or
26      exceeded the levels reported to affect soil invertebrates, macrofungi, and soil microbial
27      processes. The urban forests exhibited reduced fungal biomass and microarthropod densities
28      when compared to rural stands. These results supported the concept that urban forests have
29      depauperate  communities because of anthropogenic stress resulting from poor air quality due to
30      high levels of SO2,  sulfate, ozone, and nitrogen; elevated levels of soil- and forest-floor-heavy
31      metals; and low water availability such as those caused by hydrophobic soils (McDonnell

        June 2003                                 4-151        DRAFT-DO NOT QUOTE OR CITE

-------
 1      et al.,1997).  Thus, forests at the urban end of the gradient exhibited reduced fungal and
 2      microarthropod populations and poorer leaf quality than the more rural forests. The potential
 3      effect of these conditions on the ecosystem processes of decomposition and nitrogen cycling in
 4      urban forests appeared to be ameliorated by two other anthropogenic factors:  increased average
 5      temperatures caused by the heat island effect and the introduction and successful colonization of
 6      earthworms in the urban forests (McDonnell et al., 1997).
 7           McDonnell et al. (1997) observed that the changes in forest nitrogen dynamics were related
 8      to increased  anthropogenic nitrogen deposition in an urban environment.  The studies of Aber
 9      et al. (1989)  in the northeastern United States on forest nitrogen dynamics demonstrated that
10      elevated nitrogen deposition over many years results in increased nitrification and the
11      mineralization of more nitrogen than can be taken up by plants and microorganisms.
12      Nitrification can precipitate decreases in fine root biomass and increases in nitrate leaching
13      below the root zone. These effects of nitrogen deposition were not related to inputs from a
14      specific source such as PM.
15           There have also been studies of heavy metal deposition in or near cities; but the studies do
16      not cite the effects of metals in the soil. Pouyat and McDonnell (1991) discuss heavy metal
17      accumulations in forest soils along an urban-rural gradient in southeastern New York.
18      Variations in the amounts of Zn, Cu, Ni, and Cd appeared indicative of a pattern of atmospheric
19      deposition near point sources (Section 4.3.2.6). The concentrations of heavy metals in forest
20      floor and soils corresponded closely with the urban-rural land use gradient.  Again, as in the
21      study by Lovett et al. (2000), the pollutants were highest near the urban end of the gradient and
22      declined toward rural sites, with Pb, Ni and Cu highest near the urban end.
23           The air quality of the region around East  St. Louis has been of concern due to industries in
24      the area (Kaminski and Landsberger, 2000a), which include ferrous and nonferrous metal
25      smelters (Pb, Zn, Cu, and Al), coal-fired power plants, producers of organic and inorganic
26      chemicals, municipal waste incinerators, and petroleum refineries.  The city is also in the path of
27      diverse plumes from refineries to the north,  coal-fired power plants to the west, and nonferrous
28      smelters to the  south. Concentrations of heavy metals and metalloids (As, Cd, Cu, Hg, Pb, Sb,
29      Zn) in the soil provided a basis for analysis (Kaminski and Landsberger, 2000b). These studies
30      of the extent of long-term metal deposition on the soil surface and depth of soil contamination,
31      as well  as the leaching dynamics of heavy metals were made to determine possible effects on

        June 2003                                 4-152       DRAFT-DO NOT QUOTE OR CITE

-------
 1      biota uptake or groundwater contamination. The effects on biota are not mentioned; however,
 2      the soils in the area acted as a sink and there was little groundwater mobility (Kaminski and
 3      Landsberger, 2000b).
 4           Overall, the above assessment of new information leads to the clear conclusion that
 5      atmospheric PM at levels currently found in the United States have the potential to alter
 6      ecosystem structure and function in ways that may reduce their ability to meet societal needs.
 7      The possible direct effects of airborne PM on individual plants were discussed in Section 4.2.2.1.
 8
 9
10      4.3  AIRBORNE PARTICLE EFFECTS ON VISIBILITY
11      4.3.1  Introduction
12           Visibility may be thought of as the degree to which the atmosphere is transparent to visible
13      light (National Research Council, 1993). The beauty of scenic vistas in many parts of the U.S. is
14      often diminished by haze that reduces contrast, washes out colors, and renders distant landscape
15      features indistinct or invisible. This degradation of visibility is due primarily to the scattering
16      and absorption of light by fine particles suspended in the atmosphere. One quantitative measure
17      of visibility, used traditionally by meteorologists, is the visual range, defined as the farthest
18      distance at which a large black object can be distinguished against the horizon sky (U.S.
19      Environmental Protection Agency,  1979).
20           In August 1977, Congress amended the Clean Air Act (CAA) to establish as a national goal
21      "the prevention of any future and remedying of any existing impairment of visibility in
22      mandatory Class I Federal areas (many national parks and wilderness areas), which impairment
23      results from manmade air pollution" (Title  I Part C Section 169A, U.S. Code [1990]).  The 1977
24      Amendments also included provisions requiring applicants for new major source permits to
25      assess the potential for their projects to cause adverse effects on air quality-related values,
26      including visibility, in nearby Class I  areas. In 1980, the EPA established regulatory
27      requirements under Section 169A to address Class I protection from "reasonably attributable"
28      visibility impairment; i.e., visibility impairment attributable to a single source or small group of
29      sources.
30           The CAA, as amended in 1990 (section 169B),  required the U.S. Environmental Protection
31      Agency to conduct research on regional visibility impairment and to establish the Grand Canyon

        June 2003                                4-153       DRAFT-DO NOT QUOTE OR CITE

-------
 1      Visibility Transport Commission (GCVTC).  The GCVTC was charged with assessing and
 2      providing recommendations to help preserve clear days and to improve visibility in the 16
 3      national parks and wilderness areas located on the Colorado Plateau. The GCVTC also was
 4      mandated to provide recommendations to the U.S. Environmental Protection Agency for the
 5      reduction of visibility impairment due to regional haze, described as any perceivable change in
 6      visibility (light extinction, visual range, contrast, or coloration) from that which would have
 7      existed under natural conditions that is caused predominantly by a combination of many
 8      anthropogenic sources over a wide geographical area (U.S. Environmental Protection Agency,
 9      1999a). In July 1999, the U.S. Environmental Protection Agency published the Regional Haze
10      Rule (Federal Register, 1999).  The regulation established a program for the improvement and
11      protection of visibility in the 156 protected Class I parks and wilderness areas and included the
12      establishment of baseline and current visibility conditions and the tracking of changes in
13      visibility conditions over time. Implementation of the regional haze regulations is supported by
14      the U.S. Environmental Protection Agency's PM25 monitoring network and  an expanded
15      Interagency Monitoring of Protected Visual Environments (IMPROVE) network. The PM25
16      monitoring network and the IMPROVE network are described briefly later in this section and in
17      more detail  elsewhere (National Park Service, 1998; Evans and Pitchford, 1991; U.S.
18      Environmental Protection Agency, 2000b; U.S. Environmental Protection Agency, 2001).
19           The objective of the visibility discussion in this section is to provide a brief description of
20      the fundamentals of atmospheric visibility and to summarize the linkage between particulate
21      matter and visibility. Visibility is an effect of air quality and, unlike the particulate matter
22      concentration, is not a property of an element of volume in the atmosphere.  Visibility can be
23      quantified only for a sight path and depends on the illumination of the atmosphere and the
24      direction of view.  However, the  concentration of particles in the atmosphere plays a key role in
25      determining visibility. Therefore, visibility impairment may be controlled by control of particle
26      concentrations.  The relationships between particles, other factors, and visibility impairment are
27      described in this section.  For a more detailed discussion on visibility, the reader is referred to
28      the 1996 Air Quality Criteria for Particulate Matter  (1996 PM AQCD; U.S. Environmental
29      Protection Agency, 1996a); the Recommendations of the Grand Canyon Visibility Transport
30      Commission (Grand Canyon Visibility Transport Commission, 1996); the National Research
31      Council (National Research Council,  1993); the National Acid Precipitation Assessment

        June 2003                                4-154       DRAFT-DO NOT QUOTE OR CITE

-------
 1      Program (Trijonis et al., 1991); Interim Findings on the Status of Visibility Research (U.S.
 2      Environmental Protection Agency, 1995a); Visibility: Science and Regulation (Watson, 2002),
 3      and reports summarizing visibility science and data from the IMPROVE visibility monitoring
 4      network (Malm et al., 2000; Sisler, 1996; Sisler et al., 1993).
 5
 6      4.3.2  Factors Affecting Atmospheric Visibility
 7           The visual perception of a distant object is influenced by a large number of factors
 8      including human vision (the eye), the brain's response to signals received from the eye, the
 9      interaction of light with the atmosphere (e.g., atmospheric illumination, path and transmitted
10      radiance, contrast, and optical properties), and atmospheric pollution from natural and
11      anthropogenic sources.  Detailed discussion of this full range of topics can be found in the 1996
12      PM AQCD (U.S. Environmental Protection Agency, 1996a) and other general references (Malm,
13      1999, Watson, 2002). This section focuses only on those topics that have been addressed by
14      more recent research, including atmospheric illumination, the optical properties of gases and
15      particles in the atmosphere, and the effects of relative humidity on the optical properties of
16      particles.
17
18      4.3.2.1  Optical Properties  of the Atmosphere and Atmospheric Particles
19           Atmospheric particles and gases attenuate image-forming light as it travels from a viewed
20      object to an observer. The fractional attenuation of light per unit distance is known as the light
21      extinction coefficient.  The light extinction coefficient, bext, is expressed in units of one over
22      length, for example inverse kilometers (km"1) or inverse megameters (Mm"1).  The light
23      extinction coefficient can be expressed as the sum of the light scattering and light absorption
24      coefficients of particles and gases.
25
26                                bext = bap + bag + bsg  + bsp                            (4-7)
27
28      where the subscripts/* and g signify particles and gases, and s and a signify scattering and
29      absorption.
30           The light extinction coefficient can be measured with a reasonable degree of accuracy or
31      can be calculated with the size, composition, shape, and the orientation of the particles.  The

        June 2003                                 4-155        DRAFT-DO NOT QUOTE OR CITE

-------
 1      light extinction coefficient is influenced by meteorological conditions and optical properties
 2      along the sight path.
 3
 4      Relationship Between Light, Targets, and Objects in a Sight Path
 5           The appearance of a distant object is determined by light from two sources: the light
 6      reflected from the object itself (initial radiance) and the light reflected by the intervening
 7      atmosphere (path radiance). Human vision and the brain's response to signals received from the
 8      eye distinguishes between objects by contrast or differences in the radiance of adjacent objects.
 9      Light reflected by objects is attenuated by scattering and absorption as it travels through the
10      atmosphere toward the observer.  The portion that reaches the observer is the transmitted
11      radiance. During the daytime, the sight path is illuminated by the direct rays of the sun, diffuse
12      skylight, light that has been reflected from the surface of the Earth, etc. Some of this
13      illumination is scattered toward the observer by the air molecules and paniculate matter in the
14      sight path. The accumulation of the light scattered into the sight path is the path radiance or air
15      light. The path radiance significantly influences the light transmitted by the object being
16      viewed. As the path radiance increases, the light transmitted by an object decreases.
17           The transmitted radiance carries the information about the object; the path radiance only
18      carries  information about the intervening atmosphere and is often quite featureless. When the
19      transmitted radiance is dominant, visibility is good.  Conversely, when the path radiance is
20      dominant, visibility is poor. In a dense fog, the transmitted radiance from nearby objects can be
21      seen, but the transmitted radiance from more distant objects is completely overwhelmed by the
22      path radiance (i.e., the light scattered by the fog).  Distant objects are lost in the white (or gray)
23      of the fog (Gazzi et al., 2001).
24           Figure 4-32 illustrates the radiance seen by an observer looking at a hillside or through the
25      aperture of a measurement instrument.  The radiance that enters the eye of the observer (or the
26      aperture of a measurement instrument)  is known as the apparent radiance (i.e., the sum  of the
27      transmitted and path radiance). The competition between the transmitted radiance and the path
28      radiance determines visibility.
29
30
        June 2003                                4-156        DRAFT-DO NOT QUOTE OR CITE

-------
       Figure 4-32.  Light reflected from a target toward an observer. The intervening
                     atmosphere scatters a portion of this light out of the sight path and scatters
                     light from the sun into the sight path. Some particles and gases also absorb a
                     portion of the light from the target. The light scattered into the sight path
                     increases with distance from the target, whereas the light transmitted from
                     the target decreases with distance from the target. The visual range is the
                     closest distance between the target and the observer at which the transmitted
                     light no longer can be distinguished from the light scattered into the sight
                     path.

       Source: Watson and Chow (1994).
 1     Light Absorption and Scattering by Gases

 2          In the ambient atmosphere the only visible-light-absorbing gas of any consequence is

 3     nitrogen dioxide (NO2), which primarily absorbs blue light and, if present in sufficient
 4     concentration across a sight path, causes yellow or brown color seen in urban hazes. Usually the

 5     absorption by NO2 is much smaller than the scattering by particles that are typically present in

 6     polluted environments, such as urban areas. The most common exception to this situation of

 7     relatively  small NO2 absorption is in effluent plumes from combustion facilities where the

 8     particles are effectively removed but the nitrogen oxide (NO), which can convert rapidly to NO2,

 9     is not removed. Except for such particle-depleted NO plumes, the light absorption coefficient

10     for gases is usually ignored in determinations of the light extinction.
       June 2003
4-157
DRAFT-DO NOT QUOTE OR CITE

-------
 1           Scattering by gases in the atmosphere is described by the Rayleigh scattering theory (van
 2      de Hulst, 1981) and is referred to as Rayleigh scattering.  The magnitude of the Rayleigh
 3      scattering depends on the gas density of the atmosphere and varies from about 9 Mm"1
 4      to 11 Mm"1 for most locations of interest, depending primarily on site elevation. To simplify
 5      comparisons of light extinction coefficient values among  sites at a variety of elevations, a
 6      standard value of 10 Mm"1 is often used for the Rayleigh scattering component (Malm, 2000).
 7
 8      Light Absorption and Scattering by Particles
 9           Particle scattering tends to dominate light extinction except under pristine atmospheric
10      conditions when Rayleigh scattering by gas molecules is the largest contributor. If the particle
11      size, refractive index, and shape are known, the extinction coefficient can be calculated. For
12      particles of sizes similar to the wavelength of visible light, Mie equations for homogeneous
13      spheres  can be used to calculate the scattering and absorption of individual particles.
14           Absorption by particles is primarily caused by elemental  carbon (also referred to as soot or
15      light-absorbing carbon) generated by the incomplete combustion of fossil fuels. Some minerals
16      in crustal particles also absorb light and can be a significant factor during fugitive dust episodes.
17
18           Most particle absorption data are determined by measuring light transmission or reflection
19      of particles captured on filter media. Absorption estimates made in this way are sensitive to the
20      filter substrate used, the optical configuration of the transmission measurement, particle loading
21      on the filter, and particle scattering albedo with the result that there are  significant uncertainties
22      for measurements of filtered particles (Horvath, 1993).  Another approach to estimating aerosol
23      light absorption is by subtracting concurrent light scattering measurements made with a
24      nephelometer from light extinction measurements made with a transmissometer. Substantial
25      uncertainty in this difference approach results from the assumption that the point measurement
26      of scattering is representative of the scattering over a long path (1 to 10 km) that is typically
27      required for transmissometers measurements. A recently  field-tested prototype photoacoustic
28      spectrometer designed to determine absorption of suspended aerosol and an enclosed-folded path
29      transmissometer offers hope of resolving the problems of the filter-based and difference
30      approaches to the measurement of light absorption by particles (Arnott et al., 1999).
        June 2003                                 4-158       DRAFT-DO NOT QUOTE OR CITE

-------
 1           The relationship between elemental carbon concentration and particle absorption can be
 2      calculated using Mie equations for particles with known size distribution, particle density, index
 3      of refraction, shapes, and for various internal mixtures with non-absorbing aerosol materials
 4      (Fuller et al., 1999). Mie equations are used to determine the efficiency factors for extinction
 5      Qext Qscat and Qabs. The efficiency factors represent the fraction of light falling on a circle with
 6      the same diameter and index of refraction as the particle.  While such application of this theory
 7      can provide a range of absorption efficiencies for various model aerosol distributions, it is rare
 8      that sufficiently detailed particle characterization data for ambient aerosols are available. Also,
 9      although elemental carbon is the strongest and most common of the absorbing particles, light
10      absorption by elemental  carbon particles can be reduced when the particle is covered by other
11      chemical  species (Dobbins et al., 1994) or may be enhanced when coated with a non-absorbing
12      refractive material such as  ammonium sulfate (Fuller et al., 1999).
13           More commonly, estimates of elemental carbon absorption efficiency are empirically
14      determined from the ratios of or the slopes of regression analysis fits to absorption coefficient
15      and corresponding elemental carbon concentration measurements. Use of the regression
16      approach permits the inclusion of crustal component concentrations as a second dependent
17      parameter, so that crustal absorption can also be estimated. Uncertainties in the absorption
18      efficiency determined empirically are a combination of the measurement uncertainties for the
19      absorption coefficients, elemental carbon concentrations, and where used, the crustal
20      concentrations.  In reviews of estimates of elemental carbon light absorption mass efficiency
21      (i.e., the absorption coefficient per carbon mass concentration), Horvath (1993) and Liousse et
22      al. (1993)  found values ranging from 2 to 17 m2/g.  Moosmiiller et al. (1998)  showed that by
23      limiting the absorption coefficient estimates to those using photoacoustic  methods, the
24      absorption efficiency shows a wavelength dependence with highest values (17 m2/g) at the
25      shortest wavelength used (A = 0.42  jim) and lowest values (3 m2/g) at the  longest wavelengths
26      used (A = 0.8 |im).  The center of the visible light wavelength (A = 0.53 jim) yielded elemental
27      carbon absorption efficiencies values of approximately 10 m2/g, a commonly used value for
28      elemental  carbon absorption efficiency. Fuller et al. (1999) suggested that isolated spheres of
29      light absorbing carbon have a specific absorption of less than 10 m2/g. Light absorption by
30      carbon particles will be greater than 10 m2/g only if the particles are internally mixed  and the
        June 2003                                 4-159       DRAFT-DO NOT QUOTE OR CITE

-------
 1      occluding particles are sufficiently large. Absorption values for graphitic and amorphous carbon
 2      for primary sizes typical of diesel soot are approximately 5 m2/g.
 3           Particle scattering tends to dominate light extinction except under pristine atmospheric
 4      conditions when Rayleigh scattering by gas molecules is the largest contributor.  Light-scattering
 5      by particles has been reported to account for 68 to 86% of the total extinction coefficient in
 6      several cities in California (Eldering et al., 1994). When light-scattering increases, visibility is
 7      impaired because of a decrease in the transmitted radiance and an increase in the path radiance.
 8      The single most important factor that determines the amount of light scattered by a particle is its
 9      size, as shown in Figure 4-33 (based on Mie calculations).  The maximum single-particle
10      scattering efficiency (i.e., scattering per cross-sectional area of a particle) is associated with
11      particles with diameters of about the wavelength of visible light (centered at 0.53 jim).
12      For particles that are small compared to the wavelength of light, the single- particle scattering
13      efficiency is low. For particles larger than the wavelength, the single particle scattering
14      efficiency initially decreases with diameter and then  fluctuates around a value of two as size
15      increases. However, a larger particle always scatters more light than a smaller particle because
16      particle cross-sectional area increases faster with diameter than does the decrease in single-
17      particle scattering efficiency at any point on the scattering efficiency curve.  The mass scattering
18      efficiency (i.e., the scattering per mass concentration) peaks for particles that are about 0.5 jim to
19      0.8 |im in diameter. Smaller particles are much less  efficient at scattering light, while larger
20      particles have mass that increases with particle size faster than the increase in the amount of light
21      they scatter.
22           Use of the Mie equation to calculate light scattering or the light scattering efficiency of
23      particles in the atmosphere is severely limited by the general lack of sufficiently detailed particle
24      characterization data.  At a minimum, size-resolved particle composition data (e.g., aerosol
25      collected on an 8-stage impactor) are needed to permit meaningful Mie scattering calculations.
26      The chemical composition provides clues to the appropriate particle density and index of
27      refraction, while the size distribution is inferred by fitting a distribution function to the
28      concentration for each stage. Assumptions are still necessary to address the particle component
29      mixture characteristics of the aerosol. Resulting scattering calculations can be compared to
30      directly measured particle extinction to assess the reasonableness of the Mie calculations.
31

        June 2003                                 4-160        DRAFT-DO NOT QUOTE OR CITE

-------
                                            Total MIE Scattering Coefficient
                                                    ForR=1.50
                     1  2  3  4  5  6  7  8  9  10  11  12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
                                                Size Parameter, a

       Figure 4-33.  Light-scattering efficiency factor (per cross sectional area), Q, for a
                     homogeneous sphere with an index of refraction of 1.50 as a function of the
                     size parameter, a  =
       Source: Penndorf (1958).
 1           Reported calculated scattering efficiencies for sulfates range from 1.2 to 5.6 m2/g.  Sulfate
 2      scattering efficiencies have been reported to increase by a factor of two when the size
 3      distribution increased from 0.15 to 0.5 jim (McMurry et al., 1996).  Calculated scattering
 4      efficiencies for carbon particles ranged from 0.9 to 8.1 m2/g.  A scattering efficiency of 1.0 and
 5      0.6 m2/g was reported for soil and coarse mass, respectively (U.S. Environmental Protection
 6      Agency, 1996a; Sisler and Malm, 2000).
 7           Figure 4-34 gives the volume-specific light scattering efficiency in units of jim"1 as a
 8      function of particle diameter. The light scattering coefficient is derived by multiplying the
 9      volume-specific light scattering efficiency factor by the volume concentration. The mass-
10      specific light scattering efficiency can be obtained by dividing the values for the curves by the
11      density of the particulate matter.
12           Similar results have been produced in field nephelometer measurements of ambient particle
13      scattering. A variety of nephelometer configurations, unrestricted or size selective inlets and the
14      control of sample air temperature and relative humidity, permit the composite scattering
15      properties of ambient aerosol to be directly observed (Day et al., 1997). When sample-
16      controlled nephelometer  data are combined with collocated particle speciation data, composite
       June 2003
4-161
DRAFT-DO NOT QUOTE OR CITE

-------
                                           Particle Diameter (|jm)
       Figure 4-34.  Volume-specific light-scattering efficiency as a function of particle diameter
                    Dp. The calculations were performed for the indicated indices of refraction
                    and a wavelength of 550 nm. For large particle diameters the scattering
                    efficiencies tend toward a value of 3/Dp.  Mass-specific light-scattering
                    efficiencies (in units of m2/g) can be obtained by dividing the values of the
                    curves by the particle density (in units of g/cm3).

       Source: U.S. Environmental Protection Agency (1996a).
1

2

3

4

5
particle scattering efficiency values for ambient aerosol can be empirically derived (Malm et al.,
2000).

     The scattering efficiency for particles have been reported by White et al. (1994) for dry
particles less than 2.5 jim (2.4 and 2.5 m2/g) and coarse particles 0.34 to 0.45 m2/g).  Other

reported values for coarse particles include 0.4 and 0.6 m2/g (White and Macias, 1990 and

Trijonis and Pitchford, 1987). Nephelometer measurements for light scattering by coarse

particles is underestimated (White et al., 1994). Chow et al. (2002a) reported scattering

efficiencies of 2 to 3 m2/g but in some cases > 5 m2/g for dry particles less than 2.5 |im.
       June 2003
                                          4-162
DRAFT-DO NOT QUOTE OR CITE

-------
 1      4.3.2.2  Relative Humidity Effects on Particle Size and Light-Scattering Properties
 2           The ability of some commonly occurring chemical components of atmospheric aerosol to
 3      absorb water from the vapor phase has a significant effect on particle light scattering.
 4      Hygroscopic participate materials, which typically include sulfuric acid, the various ammonium
 5      sulfate salts, ammonium nitrate, and sodium chloride, change size by the accumulation and loss
 6      of water as they maintain equilibrium with the vapor phase as a function of changes in relative
 7      humidity. For some materials (e.g., sulfuric acid), the growth is continuous and reversible over
 8      the entire range of relative humidity.  For other materials, water absorption begins abruptly for  a
 9      dry particle at a specific relative humidity known as the deliquescent point (e.g., -80% for
10      ammonium sulfate) and continues as relative humidity increases. There is a hysteresis effect
11      with these materials in that, once wet, the relative humidity can be reduced below the
12      deliquescent point until crystallization occurs at a substantially lower relative humidity (e.g.,
13      -30% for ammonium sulfate). Figure 4-35 shows the water vapor growth curve for ammonium
14      sulfate.
15           Water growth behavior for hygroscopic materials commonly found in atmospheric aerosol
16      in pure form or in some mixtures is generally well known as a result of laboratory measurements
17      (Tang and Munkelwitz, 1994; Tang, 1997).  Models that calculate water growth of mixtures
18      from known solubility  properties of many common water-soluble chemicals have long been
19      available (Zdanovskii,  1948) and have been successfully applied to determine growth for
20      particles with known composition (Saxena and Peterson, 1981; Pilinis et al., 1995;  Saxena et al.,
21      1993).
22           The water growth of individual ambient particles can be directly measured using a
23      humidity-controlled tandem differential mobility analyzer or TDMA (McMurry and
24      Stolzenburg, 1989; Zhang et al., 1993). Inferences can be made about the mixtures of soluble
25      and insoluble  particle components by comparing TDMA measured growth and size-resolved
26      aerosol composition data with water growth model predictions  (Pitchford and McMurry, 1994;
27      Zhang et al., 1993; Saxena et al.,  1995). A practical limitation  of TDMA measurements in
28      investigating aerosol optical properties is that particles with diameter greater than 0.5 jim are not
29      well measured by this approach.
30           Accounting for water growth of atmospheric aerosols is important in determining visibility
31      because particles containing hygroscopic or deliquescent materials change size, index of

        June 2003                                4-163       DRAFT-DO NOT QUOTE OR CITE

-------
                            1.5-
                         Q
                         Q^
                         g
                         '-i—»
                         05
                         o
                         (D
 90%).  The magnitude of the water growth effect on light scattering for ambient aerosols can
 5     be directly measured with humidity-controlled nephelometer measurements (Day et al., 1997).
 6     Measurements of water growth effects on scattering are compared to results of water growth and
 7     Mie scattering models applied to size-resolved composition data using various mixture
 8     assumptions to infer average mixture and other aerosol characteristics (Malm et al., 2000).
 9          While the importance of inorganic hygroscopic particles is well understood, the role of
10     organic compounds in particle water growth has been the subject of recent investigations.
11     In their interpretation of TDMA and particle composition data from two locations, Saxena et al.
12     (1995) made the case that organic components of the aerosol enhanced water absorption by
       June 2003
                        4-164
                        DRAFT-DO NOT QUOTE OR CITE

-------
 1      particles at a remote desert location and retarded water absorption at an urban location.  They
 2      speculated that the latter might be due to hydrophobic organic material coatings on inorganic
 3      hygroscopic particles.
 4           While some of the thousands of organic compounds that are in atmospheric aerosols are
 5      known to be hygroscopic and while a significant fraction of the organic aerosol material is
 6      known to be water soluble, there is a lack of water absorption data for most organic compounds.
 7      The incomplete water solubility data combined with incomplete data on the abundance of the
 8      numerous organic compounds in ambient aerosols means that organic water growth model
 9      calculations are not a reasonable approach to assessing the importance of water growth by
10      organic aerosol components in the atmosphere. To overcome this constraint, Saxena et al.
11      (1995) compared organic concentration to the difference between total aerosol water measured
12      by TDMA and model-estimated water for the inorganic hygroscopic aerosol components. In
13      contrast, Pitchford and McMurry (1994) used the same remote location data set and showed that
14      on six of the eight sampling days water uptake by the sulfates and nitrates could account for all
15      of the measured water absorption.
16           Swietlicki et al. (1999) made TDMA measurements in northern England and found that
17      growth takes place in two modes, one mode being less hygroscopic that the other. They
18      concluded that growth could be attributed to the inorganic content of the aerosol. Cocker et al.
19      (2001) measured hygroscopic properties of Pasadena, California aerosol and concluded that
20      growth factors increased when forest fires were present. McDow et al. (1995) measured water
21      uptake by diesel soot, automobile exhaust, and wood smoke particles.  They found  all three
22      emission types absorbed water: the wood smoke sample weight increased by about 10% as
23      sample relative humidity was increased; whereas diesel soot sample weight increased by only
24      2% to 3%.  Chughtai et al. (1999) examined hydration characteristics of a number of
25      anthropogenic and natural organic materials.  They found surface water adsorption  increased
26      with age and surface oxidation. Hemming and Senfield (2001) evaluated the relative
27      hygroscopicity of different organics, the differences in the amount of water taken up by
28      mixtures, and the individual components of the mixtures in  their pure state using the UNIFAC.
29      They found that mixtures take up less water than the individual components in the pure state.
30      The relative hygroscopicity of atmospheric organics was diacids > monoacids > alcohols
31      > carbonyls. Analysis of humidity-controlled and size-resolved chemistry data from Great

        June 2003                                4-165        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Smoky Mountains and Grand Canyon National Parks (Malm et al., 1997; Malm and
 2      Kreidenweis, 1996; Malm et al., 2000) show that ambient organic aerosol are at most weakly
 3      hygroscopic to within the measurement uncertainty and modeling assumptions.
 4           A more detailed discussion of the effects of relative humidity on the size distribution of
 5      ambient particles appears in Chapter 2 of this document.
 6
 7      4.3.3  Relationships Between Particles and Visibility
 8           Visibility, referring to the appearance of scenic elements in an observer's line of sight,
 9      depends on more than the optical characteristics of the  atmosphere.  Numerous scene and
10      lighting characteristics are important to this broad definition of visibility. However, under a
11      variety of viewing conditions, visibility reduction or haziness is directly related to the extinction
12      coefficient.
13           Light extinction, the sum of the light scattered and absorbed by particles and gases, is
14      frequently used to estimate the effect of air pollution on visibility. Light extinction is usually
15      quantified using the light extinction coefficient, that is  the sum of the light scattering and
16      absorption coefficients for gases and particles (see Section 4.3.2.1).
17           The influence of particles on visibility degradation is dependent on the particle size,
18      composition, and solubility (Pryor and Steyn, 1994). Fine particles (particles with mass mean
19      diameters less than or equal to 2.5 jim) scatter more light than coarse particles. Fine particle
20      species included sulfates (assumed to be  ammonium  sulfate), nitrates (assumed to be ammonium
21      nitrate), organics,  light- absorbing carbon and soil (Malm et al., 1994).  Of the fine particle
22      species, sulfates and nitrates are the most hygroscopic and require the use of a relative humidity
23      adjustment factor. The effect of particle  light extinction can be determined by totaling the
24      scattering and absorption of light by multiplying the mass-specific efficiency values and the
25      mass concentration for each of the particle species.  The effect of relative humidity and the
26      relative humidity adjustment factors are discussed in Section 4.3.2.2. Visibility is measured by
27      human observation, the light extinction coefficient (light scattering and absorption by particles
28      and gases), and parameters related to the light extinction coefficient (visual range, deciview) and
29      fine particle mass  concentrations.  Using the equation for the light extinction coefficient, light
30      extinction by particles can be expressed as
31

        June 2003                                 4-166        DRAFT-DO NOT QUOTE OR CITE

-------
 1           bext = bap(bEC)  +  bsp(bs04  +  bN03  +  boc  +   bsoil  +  bcoarse+  bsg)       (4-8)
 2
 3      where
 4            bap     = 10 [mass of elemental carbon]
 5            bso     = 3[mass concentration of (NH^ 2SO 4]f(RH)
 6            bNO     = 3[mass concentration of(NH4)2SOJf(RH)
 1            boc     = 4[OC]
 8            bsoll     = If Soil]
 9            bsg     = 10 [Rayleigh scattering by gases]
10
11           The visual range method of visibility measurement, commonly taken to be the greatest
12      distance that a large dark object (e.g., a mountain in shadow) can be seen against the background
13      sky (Middleton, 1952), was developed for and continues to function well as an aid in military
14      operations  and transportation safety.  Visual range is inversely proportional to the light
15      extinction.  Visual range can be calculated from a point measurement of light, assuming that the
16      atmosphere and the illumination over the sight path is uniform and the threshold contrast is 2%
17      of the extinction coefficient
18
19                                    Visual Range = K/bext                                (4-9)
20
21      where visual range is in kilometers, &ext is in km"1, and a threshold contrast of 2% is assumed.
22      If bext is in  Mm"1, the Koschmieder constant becomes 3,912.
23           A new index of haziness, expressed in deciview (dv) units, is also very simply related to
24      the light extinction coefficient (Pitchford and Malm, 1994).
25
26                            Haziness (dv) = 10 ln( bjl 0 Mm1)                       (4-10)
27
28      An important characteristic of this visibility index is that it  is more nearly linearly related to
29      perceived changes in haze level than either visual range or light extinction.  A change of 1 or

        June  2003                                 4-167        DRAFT-DO NOT QUOTE OR CITE

-------
 1     2 dv in uniform haze under many viewing conditions will be seen as a small but noticeable
 2     change in the appearance of a scene regardless of the initial haze condition.
 3           Figure 4-36 illustrates the relationship of light extinction in Mm"1, deciview index, and
 4     visual range in kilometers.  Although the deciview is related to extinction, it is scaled in such a
 5     way that is perceptually correct (Fox et al., 1999).
 6
 7
              Extinction (Mm1)
              Deciviews   (dv)
10      20    30   40  50   70 100    200    300   400  500  700 1000
                                                        I II
I
0
I
I
7
I
I
11
I
I
14
I
I
16
I
i mi
19 23
i mi
i
30
I
I
34
I
I
37
I
I
39
I
I I III
42 46
I I III
           Visual Range   (km)   400     2oo    130  100  so   eo 40      20    13    10   s    e  4

       Figure 4-36. Comparison of extinction (Mm"1) and visual range (km).
       Source: Fox etal. (1999).
 1           Several early studies have demonstrated a relationship between enclosed nephelometer
 2     light scattering measurements and fine particle mass collected on a filter (Waggoner et al., 1981;
 3     Samuels et al.,  1973). However, the relationship between fine particle mass and light scattering
 4     may differ between locations and for different times of the year. The relationship is improved by
 5     using the same size selective inlet on both the nephelometer and filter sampler (White et al.,
 6     1994) and by minimizing the effect of high relative humidity on filter specimens and in the
 7     neplelometer scattering chamber. When particle speciation data for the major aerosol
 8     components are available, the relationship between particles and light extinction can be further
 9     improved by treating the individual major components separately. A recent study by Chow et al.
10     (2002a) also suggested that continuous light scattering measurements may be useful as an
11     indicator of short-term variations in PM2 5 mass concentrations when the measurements are made
12     under dry conditions.  Figure 4-37 shows the relationship between fine particle mass and
13     calculated light extinction.
14           Most routine aerosol monitoring programs and many special study visibility
15     characterization programs were designed to measure the major aerosol components (Malm et al.,
16     1994; Tombach and Thurston,  1994; Watson et al.,  1990); they were not designed to determine

       June 2003                                 4-168        DRAFT-DO NOT QUOTE OR CITE

-------
                         1600-1

                         1400-

                         1200-
                       01
                       i" 1000-
                       fS
                          800-
                       2  600H
                       o»  400-
                       o.
                       «  20
-------
 1           Sloane (1983, 1984, 1986), Sloane and Wolff (1985), and more recently Lowenthal et al.
 2      (1995) and Malm and Kreidenweiss (1997) have shown that differences in estimated specific
 3      scattering between external and internal model assumptions are usually less than about 10%.
 4      In the absence of detailed microphysical and chemical information of ambient particles, the
 5      above studies demonstrate that a reasonable estimate of aerosol extinction can be achieved by
 6      assuming each species is externally mixed.
 7           The latest IMPROVE Program report (Malm, 2000) includes calculated aerosol light
 8      extinction for each of the five major fine fraction particle (PM25) components, coarse fraction
 9      mass (PM10_25), and Rayleigh scattering by gases and sums them for an estimate of total light
10      extinction in Mm"1 using the following algorithm:
11
12                                 bext = (3)f(RH) [SULFATE] +
13                                    (^)f(RH) [NITRATE]
14                                  +(4) [ORGANIC CARBON]
15                            +(10) [LIGHT'ABSORBING CARBON]                        (4-12)
16                                         +(l)[SOIL]
17                                    +(0.6) [COARSE PM]
18                             +10 (for Rayleigh scattering by gases)
19
20      where each PM term is the product of a constant dry extinction efficiency for that species, the
21      mass concentration of the species, and, for sulfate and nitrate, an adjustment factor that is a
22      function of relative humidity to account for their hygroscopic behavior. The relative humidity
23      adjustment term for sulfate and nitrate, shown in Figure 4-38, is based upon the ammonium
24      sulfate growth curve, shown in Figure 4-28, smoothed between the upper and lower curves of the
25      hysteresis loop for the relative humidity range of 30-80%.  The extinction efficiencies for soil
26      and coarse mass used in this algorithm are taken from a literature review by Trijonis et al.
27      (1987). The extinction efficiency for light absorbing (elemental) carbon of 10 m2/g  is consistent
28      with the value reported by Moosmiiller et al. (1998) corresponding to A = 0.53 in the middle of
29      the visible light spectrum.  The dry extinction efficiencies of 3 m2/g  for sulfate and nitrate
30      species and 4 m2/g for organic species are  based on literature reviews by Trijonis et  al. (1991)
31      and by White (1991).  Trijonis' best estimate for sulfates is 2.5 m2/g with an uncertainty of a
        June 2003                                4-170        DRAFT-DO NOT QUOTE OR CITE

-------
                        I
                        tr
                        03
                        LL
                        D)
                                      30    40    50   60    70    80   90   100
                                             Relative Humidity (percent)
       Figure 4-38.  Relative humidity adjustment factor, f(RH), for ammonium sulfate as a
                     function of relative humidity.
       Source: Malm et al. (2000).
 1
 2
 3
 4
 5
 9
10
11
12
13
14
factor of 2, while White's average low and high estimates for the rural West are 3.0 and
3.7 m2/g. For organics, Trijonis estimates a dry extinction efficiency of 3.75 m2/g with an
uncertainty of a factor of 2, and White's range for the rural West is 1.8 to 4.1 m2/g. Malm et al.
(1996) and Malm (2000) used this algorithm to successfully reconstruct scattering at a total of
eleven IMPROVE monitoring sites.
     Malm (2000) used additional sophisticated aerosol size, composition, and microphysical
data from a special study at the Great Smoky Mountain National Park to compare the
performance of a number of models for calculating light extinction. He found that the simplist
approach adequately predicted for periods of low light scattering but under-predicted by about
30% during periods of high sulfate concentration.  The greatest improvement over the simple
model was obtained by including the degree of sulfate ammoniation in the model. This
produced better estimates of the extinction coefficient over the entire range.  Table 4-15 lists
various visibility metrics and methods for visibility measurement.
       June 2003
                                         4-171
DRAFT-DO NOT QUOTE OR CITE

-------
                                           TABLE 4-15.  VISIBILITY MEASUREMENT TECHNIQUES
ft>
to
O
o
Visibility Metric and Method
Measurement Principle
J^.

to
        Visual Range




        Light extinction (6ext)

        Long-path transmissometer


        Short-path transmissometer
        Contrast transmittance:
         Teleradiometer
         Photographs and time-lapse
         film
         Particle scattering (6scatp)

         Integrating nephelometer
                               Human observation of prevailing visibility. Targets are selected at known distances from an observer. Nighttime targets
                               require lights and may differ from daytime targets.  Each hour, the observer records the distance (i.e., VR) corresponding to
                               the furthest target that is visible. This method provides the longest history of visibility measurements in the United States,
                               as it was used at most U.S. airports from 1948 to 1995.

                               Directly measures the radiance of a constant light source (transmitter) after the light travels over a finite atmospheric path.

                               The transmittance of the path is calculated by dividing the measured radiance by the calibrated initial intensity of the light
                               source.  The average extinction of the path is calculated from the transmittance and length of the path.

                               Starting in the early 1990s, many sites replaced human observations with automated sensors  (i.e., Automated Surface
                               Observing System [ASOS], Automated Weather Observing System [AWOS]) to measure 1-min-average light extinction.
                               The visibility sensor measures forward scattering using a xenon flash lamp source. Instruments can provide measurements
                               up to 48 km, but they are not recorded beyond 16 km.

                               A telescope is focused on a distant target and the background.  Changes in radiance are measured by photodiode detectors.
                               Measurements can be made at different wavelengths (e.g., 450, 550, and 630 nm) using narrow band filters. Measures
                               effects of particles of all sizes.  Sensitive to variations in inherent contrast (e.g., bright or dark cloud behind the target) and
                               nonuniform illumination conditions.  Because path radiance depends on how the atmosphere is illuminated, contrast
                               transmittance represents human perception better than air quality.

                               A densitometer  measures light transmitted through different portions of a color slide.  The film's light-response function
                               (the gamma curve) determines target-sky contrast. Computerized photographic simulations can change contrast
                               transmittance for different meteorological and atmospheric conditions, and can be used to judge how people react to these
                               changes.

                               Air is drawn into a nephelometer chamber that is illuminated with white or filtered (typically 500-550 nm) light.  Light is
                               detected at 90°  to the direction of illumination to measure the amount scattered out of the light path.  Chamber dimensions
                               limit the integrated arc to ~ 10-170° instead of a full 0-180°, which results in some forward scattering from coarse particles
                               being underestimated. Nephelometers are  calibrated with gases of known indices of refraction. Particles (especially
                               hygroscopic and volatile species) may be modified as they pass through the chamber, which is  inadvertently heated by the
                               illumination source.

-------
                                      TABLE 4-15 (cont'd). VISIBILITY MEASUREMENT TECHNIQUES
to
o
o
       Visibility Metric and Method
                                             Measurement Principle
       Particle absorption (b,^r):
         Aethalometer or particle
         soot absorption photometer


         Photoacoustic spectroscopy
         Filter transmittance,
         reflectance
         Suspension of insoluble
         elemental carbon
Particles are collected on a quartz-fiber filter tape. The change in transmittance at selected wavelengths (e.g., 880 nm)
across the filter before and after sampling is measured or compared with the reference area. When the filter spot darkens, a
new portion of the tape is moved into the sampling position. Assumes a constant relationship between BC mass and quartz
filter transmittance of ~19m2/g. Assumes a relationship of lOmVg between BC absorption and concentration.

Particles absorb energy from a modulated laser (-514.5 nm) and transfer heat to the surrounding air.  Expansion of the
heated gas produces sound waves (acoustic signals) that are proportional to the amount of absorbed energy. These are
detected by a high-sensitivity microphone. Absorption appears to vary with illumination wavelength.

Uses densitometer, integrating plate, or integrating sphere spectrophotometer to measure how much light is transmitted or
reflected. The difference in the logarithms of light transmission through the filter before and after sampling is proportional
to the particle deposit. Light transmission of particles collected on a filter may overestimate light absorption because some
of the incident light is scattered within the filter and by other particles in the deposit.  There are no absolute calibration
standards for densitometry, integrating plate, or sphere methods.

Particles collected on a Nuclepore polycarbonate-membrane filter are extracted in 30% isopropanol/70% distilled deionized
water to form a suspension of insoluble EC particles. Using a spectrophotometer, light transmission (e.g., 400-650 nm,
peaking at 575 nm) is measured through the liquid extract.
         Component extinction (6ext)    The sum of clean air scattering estimated from temperature and pressure, NO2 absorption estimated from NO2
         (sum of 6scatp, 6scat>g, 6absp, and
        Chemical extinction (b^t):
         Filter measurements for
         SO4'2, NO3', organics, EC,
         fine soil, and coarse mass,
         plus clear air scattering
concentrations, particle scattering measured by nephelometer, and particle absorption (babsp) measured by one of the b
methods. Measurements are at a single location rather than along a sight path.

Six aerosol chemical components are used to calculate chemical extinction.
                                                                                                                                                abs,p
        Adapted from:  Watson (2002).

-------
 1      4.3.4  Photographic Modeling of Visibility Impairment
 2           None of the visibility indices communicate visibility associated with various aerosol
 3      conditions as well as directly seeing their effects on a scene. However, photographic modeling
 4      for the representation of haze can be useful in portraying changes in visibility specifically due to
 5      changes in air pollutant concentrations. Photographic modeling holds constant the effects of sun
 6      angle, cloud cover, and relative humidity and is a cost-effective method of evaluating various air
 7      quality  scenarios.  Photographic modeling is difficult to do with actual photographs because of
 8      the range of possible conditions in the same scene over multiple days; and, over time,
 9      photographs can be expensive to produce. Another limitation in using photographic models for
10      representation of haze is that haze is assumed to be uniformly distributed throughout the scene
11      and selected conditions are idealized, so the full range of conditions that occur in a scene are not
12      represented.
13           Eldering et al. (1996) proposed the use of a model that uses simulated photographs from
14      satellite and topographic images to evaluate the effect of atmospheric aerosols and gases on
15      visibility. Use of this model requires ground-based photography and  data concerning the size
16      distribution and chemical composition of atmospheric aerosols, NO2 concentration, temperature,
17      and relative humidity for a clear day. Light extinction and sky color are then calculated based on
18      differences in aerosol size distribution, NO2 concentration, temperature, and relative humidity.
19      The images created represent natural landscape elements.
20           Molenar et al. (1994) provides a discussion of existing visual air quality simulation
21      methods based on techniques under development for the past 20 years. A photograph taken on a
22      very clean, cloud-free day serves as the base image.  The photograph  is taken during the season
23      and at the same time of day as the scene to be modeled.  The light extinction represented by the
24      scene is derived from aerosol and optical data associated with the day the  image was taken, or it
25      is estimated from contrast measurements of features in the image. The image is then digitized to
26      assign an optical density to each picture element (pixel) for the wavelength bands of interest.
27      A detailed topographic map and an interactive image-processing display system is used to
28      determine the specific distance, elevation angle, and azimuth angle for each element in the
29      picture  with respect to the observer's position.
30           Various models are employed to allow the presentation of different air quality scenarios.
31      The output from atmospheric aerosol models (e.g., extinction, scattering coefficients, single

        June 2003                                4-174        DRAFT-DO NOT QUOTE OR CITE

-------
 1      scattering albedo, and scattering phase matrix) is incorporated into radiative transfer models to
 2      calculate the changes in radiant energy (path radiance, image radiance, sky radiance, terrain
 3      radiance) caused by scattering and absorption by gases and particles as it passes through the
 4      atmosphere.  Atmospheric aerosol models are also use to model the effect of relative humidity
 5      on the visual air quality (Molenar et al., 1994).
 6           Molenar et al. (1994) has developed a system call WinHaze that permits the viewing of
 7      computer-generated uniform hazes superimposed on digitized scenic photographs of both remote
 8      and urban scenes.  The program simulates changes in visual air quality imagery from user-
 9      specified changes in optical parameters (e.g., <7ext, visual range, or deciview values) or aerosol
10      species concentrations.  WinHaze includes imaging for various Class I national parks and
11      wilderness areas and Boston, MA; Dallas, TX; Denver, CO; Fort Collins, CO; Phoenix, AZ;
12      and Tucson, AZ. The computer software is available through the IMPROVE website
13      (http://vista.cira.colostate.edu/ improve/).
14
15      4.3.5  Visibility Monitoring Methods and Networks
16           Visibility monitoring studies measure the properties of the atmosphere either at the sampler
17      inlets (point measurements), as is the case with air quality measurements, or by determining the
18      optical properties of a sight path through the atmosphere (path measurements). Instrumental
19      methods for measuring visibility are generally of three types:  (1) direct measurement of light
20      extinction of a sight path using a transmissometer, (2) measurement of light scattering at one
21      location using an integrating nephelometer, and (3) measurement of ambient aerosol mass
22      concentration and composition (Mathai, 1995).
23           The largest instrumental visibility monitoring network in the United States is the
24      Automated Surface Observing System (ASOS), commissioned by the National Weather Service,
25      Federal Aviation Administration, and Department of Defense at more than 900 airports.  The
26      system is designed to objectively measure the clarity of the air versus the more subjective
27      evaluations of human observations.  The system provides real-time data for airport visibility.
28           The visibility sensor, instead of measuring how far one can see, measures the clarity of the
29      air using a forward-scatter visibility meter.  The clarity is then converted to what would be
30      perceived by the human eye using a value called Sensor Equivalent Visibility (SEV). Values
31      derived from the sensor are not affected by terrain, location, buildings, trees, lights, or cloud

        June  2003                                 4-175       DRAFT-DO NOT QUOTE OR CITE

-------
 1      layers near the surface.  The amount of moisture, dust, snow, rain, and particles in the light beam
 2      will affect the amount of light scattered.  The sensor transmits 1-min values based on rolling
 3      10-min periods. Hourly visibility range data are available only at a quantized resolution of
 4      18 binned ranges with a visual range of up to 10 miles. The value provides a generally accurate
 5      and representative visibility measurement within a 2 to 3 mile radius of the site.  The forward
 6      scatter meter was found to correlate fairly well with extinction coefficient measurements from
 7      the Optec Transmissometer (National Weather Service, 1998).
 8           Visibility data from the ASOS network is reported in terms of visual  range in increments of
 9      1/4 to 1 statute mile.  Visual range conditions exceeding  10 miles are truncated to 10 miles for
10      real-time reporting purposes. Data is not extensively archived at ASOS locations; however,
11      researchers are able to download the raw data directly from certain sites. In addition,  since 1998,
12      the raw visibility data (including light extinction measurements corresponding to visual ranges
13      exceeding 10 miles) have been archived for a number of sites. Visual range measurements
14      beyond 10 miles may be used to derived particulate matter concentrations except in clean
15      environments.  These data are available from the National Climatic Data Center.
16           The largest monitoring network that includes both visibility and aerosol conditions is the
17      Interagency Monitoring of Protected Visual Environments (IMPROVE) network. This network
18      was formed in  1987 as a collaborative effort between Federal, regional, and state organizations
19      responsible for protection of visibility in the 156 mandatory Class I Federal areas (national parks
20      and wilderness areas) and other areas of interest to land management agencies, states, tribes, and
21      other organizations (National Park Service, 1998; U.S. Environmental Protection Agency,
22      1996a; U.S. Environmental Protection Agency, 1995b; Eldred et al., 1997; Perry et al., 1997;
23      Sisler and Malm, 2000;  U.S. Environmental Protection Agency, 1999b).  It is predominantly a
24      rural-based network with more than 140  sites across the country. The primary monitoring
25      objectives of the IMPROVE program are to document current visibility conditions in the
26      mandatory Class I areas, to identify anthropogenic chemical species and emission sources of
27      visibility impairment  through the collection of speciated PM2 5 data, and to document long-term
28      trends for assessing progress towards elimination of anthropogenic visibility impairment.  The
29      IMPROVE network has also been involved in visibility related research, including the
30      advancement of visibility monitoring instrumentation and analysis techniques and visibility
        June 2003                                4-176        DRAFT-DO NOT QUOTE OR CITE

-------
 1      monitoring and source attribution field studies (National Park Service, 1998; Evans and
 2      Pitchford, 1991).
 3           Visibility monitoring under the IMPROVE network can be divided into three categories:
 4      aerosol, optical, and scene.  Twenty-four hour PM2 5 and PM10 aerosol samples are collected by
 5      filters at least every third day. The PM2 5 samples are analyzed to determined the mass
 6      concentration of the major parti culate constituents (sulfates, nitrates, organic carbon compounds,
 7      elemental carbon, chlorides, and crustal elements) and for elements that indicate sources of
 8      visibility-impairing particles (trace elements and ions).  Optical monitoring provides a direct
 9      measurement of light scattering and absorption.  Color photographic imaging documents the
10      appearance of the scene under a variety of air quality and illumination conditions (U.S.
11      Environmental Protection Agency, 1999b). It is anticipated that all data generated by the
12      IMPROVE network will be added to the AIRS database.
13           The U.S. Environmental Protection Agency has deployed a new national monitoring
14      network (Federal Reference Method Monitoring network) designed to assess PM2 5
15      concentrations and composition. There are over 1,000 monitoring sites in operation and many
16      sites report data to the Aerometric Informational Retrieval System (AIRS). Analyses of these
17      data are expected to provide a more complete understanding of visibility conditions, in particular
18      urban visibility, across the country. The PM25 monitoring effort has been coordinated with
19      visibility monitoring efforts currently in place to maximize the benefits of all of the monitoring
20      programs (U.S. Environmental Protection Agency, 1997b; U.S. Environmental Protection
21      Agency, 2000b; U.S. Environmental Protection Agency, 2001).
22           The Northeast States for Coordinated Air Use Management (NESCAUM) has established a
23      real-time visibility monitoring network (CAMNET) using digital photographic imaging.
24      Currently, there is digital photographic imaging for five urban locations (Boston, MA;
25      Burlington, VT; Hartford, CT; Newark, NJ; and New York City, NY), and two rural locations
26      (Acadia National Park, ME  and Mt. Washington, NH).  The visibility images are updated every
27      15 minutes.  Near real-time  air pollution and meteorological data are updated every hour.
28      Archived images will be available for studies of the visual effects of paniculate matter air
29      pollution in the Northeast.  CAMNET may be accessed  at www.hazecam.net (Northeast States
30      for Coordinated Air Use Management, 2002; Leslie, 2001).
        June 2003                                4-177        DRAFT-DO NOT QUOTE OR CITE

-------
 1           The Midwest Regional Planning Organization, in cooperation with a number of other
 2      groups, has also developed a real-time visibility camera network (hazecam). The camera
 3      network includes several urban (Chicago, IL; Indianapolis, IN; and Cincinnati, OH) and rural
 4      locations (Seney NWR, MI; Mayville, WI; and Isle Royale National Park, Ml/Grand Portage,
 5      MN).  The Midwest hazecam can be found at http://www.mwhazecam.net.
 6
 7      4.3.6  Visibility Impairment: Trends and Current Conditions
 8           In the United States, visibility impairment is caused by particles primarily composed of
 9      sulfates, nitrates, organic compounds, carbon soot, and crustal dust. Visibility is best in Alaska
10      and the western Great Basin.  Moderate levels of light extinction are common on the Pacific
11      Coast, including the western slopes of the Sierra Nevadas in California and the Cascade Range in
12      Oregon and Washington. Visibility is most impaired in the  areas encompassing and adjacent to
13      the Ohio and Tennessee River Valleys.  Visibility gradually improves along the Atlantic
14      seaboard northeast of New York City (Watson, 2002).  Estimates of natural visibility for the
15      eastern and western United States are 26 ± 7 Mm"1 natural extinction,  12 Mm"1 for clean air and
16      17 ± 2.5 Mm"1 natural extinction, 11 Mm"1 for clean air, respectively (Watson, 2002).
17
18      4.3.6.1  Trends in Visibility Impairment
19           Trends in visibility impairment or haziness often are used as  indicators of trends in fine
20      particles mass.  Observations of visual range, obtained by the National Weather Service and
21      available through the National Climatic Data Center of the National Oceanic and Atmospheric
22      Administration provide one of the few truly long-term,  daily records of impairment related to air
23      pollution.  After some manipulation including correction for relative humidity effects, the visual
24      range data can be used as an indicator of fine mode particle  pollution. The data reduction
25      process and analyses of resulting trends have been reported  by Schichtel et al. (2001), Husar
26      et al. (1994), Husar and Wilson (1993),  and Husar et al. (1981).
27           There are many statistical approaches to estimating trends. These  approaches include
28      simple correlation and regression analyses, time series analyses, and methods based on
29      non-parametric statistics. A discussion  and comparison of the methods for the detection of linear
30      trends is provided in Hess et al. (2001).  Schimek (1981) provides  a discussion of nonlinear
31      trends. In its annual air quality trends report, the U.S. Environmental  Protection Agency

        June 2003                                4-178       DRAFT-DO NOT QUOTE OR CITE

-------
 1      characterized trends using a non-parametric regression analysis approach commonly referred to
 2      as the Theil test (U.S. Environmental Protection Agency, 1998; Hollander and Wolfe, 1973).
 3
 4      Regional Trends and Class I Areas
 5           The two largest contributors to visibility impairment are sulfates and carbon-based
 6      particles. In the east, sulfates are responsible for 60 to 86% of the visibility impairment. The
 7      sulfate contribution decreases further west but is still responsible for between 25 to 50% of the
 8      visibility impairment. Carbon-based particles are responsible for 10 to 18% of the visibility
 9      impairment in the East and 25 to 40% in the West. Nitrates account for only 7 to 16% of the
10      light extinction in the East but are responsible for between 5 to 45% of the light extinction in the
11      West. Crustal material can be a major contributor in the West, accounting for 5 to 25%  of the
12      light extinction. Elemental carbon also is contributor to light extinction but to a much lesser
13      degree (U.S. Environmental Protection Agency, 2001).
14           The U.S. Environmental Protect Agency designated five regional groupings as part of the
15      regional haze program.  The regions are Northeast (Mid-Atlantic/Northeast  Visibility Union),
16      Southeast (Visibility Improvement State and Tribal Association of the Southeast), Central
17      (Central States Regional and Air Partnership), Midwest (Midwest Regional  Planning
18      Organization), and West (Western Regional Air Partnership). The regional  groupings serve  as
19      concensus organizations comprised  of states, tribes, and federal agencies coordinating the
20      implementation of the regional haze rule.
21           Using hourly prevailing daytime visibility data from human observations at weather
22      stations, Schichtel et al. (2001) observed that haziness declined approximately 10% across the
23      United States between 1980 and 1995.  The decrease in haziness was highest in the southeastern
24      United States with a 20% decrease in the 90th percentile light extinction and a 12% decrease  in
25      the 75th percentile over the  15 year period. There was a 17% decrease in the 90th percentile and a
26      9% decrease in the 75th percentile over the eastern United States. Over the eastern United States,
27      haziness was greatest during the summer months. The greatest visibility impairment was
28      adjacent to the Appalachian Mountains in Tennessee and the Carolinas (extinction coefficient of
29      > 0.2 km"1 equivalent to 6 miles). During the cold season, elevated haze (extinction coefficient
30      of > 0.2 km"1) was seen between the Great Lakes and the Ohio River Valley, over the Gulf States
        June 2003                                4-179        DRAFT-DO NOT QUOTE OR CITE

-------
 1      between Texas and Florida, along the coast from North Carolina to New Jersey, and along the
 2      Pacific coast, particularly central and south California.
 3           Visibility impairment or haziness in the southeastern United States from sulfate emissions
 4      is greatest in the humid summer months because of the ability of sulfate to absorb atmospheric
 5      water vapor. Summer haziness increased in the southeastern United States from the 1950s to
 6      1980 along with increasing SO2 emissions. A statistically significant increase in summer sulfate
 7      concentrations was noted in two Class I areas in the eastern United States (Shenandoah and the
 8      Great Smoky Mountains) from  1982 to 1992 (Eldred et al., 1993; Cahill et al., 1996).  During
 9      that time, the majority of the Southwest showed decreasing sulfur concentrations (Eldred et al.,
10      1993; Eldred and Cahill, 1994). Increasing summer sulfate concentrations were later  shown at
11      those two locations by Iyer et al. (2000).
12           Limited visibility measurements are available for the upper Midwest region (Illinois,
13      Indiana, Michigan, Ohio, and Wisconsin and the tribal lands located in the states). The Midwest
14      Regional Planning Organization conducted an initial assessment of the regional haze problem in
15      this region using existing reports and available air quality data for four major urban areas (St.
16      Louis, Chicago, Detroit, Cincinnati) and Class I areas. The "worst" and "best" visibility days
17      occur throughout the year.  Particulate sulfates were the major contributors to light extinction
18      during the summer months, and nitrates dominated on the worst visibility days during the winter
19      and fall in both urban and Class I areas. Organics also were significant contributors to light
20      extinction in urban areas. Higher PM2 5 concentrations were correlated with poorer visibility in
21      the southern portion of the  regional (Midwest Regional Planning Organization, 2001).
22           The U.S. Environmental Protection Agency's National Air Quality and Emission Trends
23      Report summarized an estimated of the regional trends and current conditions in 35 Class I areas
24      and one urban area (Washington, DC), using chemical concentrations data from the IMPROVE
25      network (U.S. Environmental Protection Agency, 2001). The visibility trends analysis is an
26      aggregate of 10 eastern Class I areas and 26 western Class I areas.  Trends were presented for
27      annual  average values for the clearest ("best") 20% , middle ("typical") 20%, and haziest
28      ("worst") 20% of the  days monitored each year. The visibility trends, given in changes in
29      deciview values, for the eastern and western sites are illustrated in Figures 4-39a and 4-39b.
30      From the figures it can be seen that the haziest days in the West are equivalent to the best days in
31      the East.  In the East,  there was a 16% (1.5 deciview) improvement in haziness on the clearest

        June 2003                                4-180        DRAFT-DO NOT QUOTE OR CITE

-------
                    35
                CD
                O
                CD
 CD   20
 l_
 '2.   15
 E

 i*   10

 IP    5
                     0
                                              Haziest 20-percent

                                              Clearest 20-percent
                      92     93    94     95    96     97    98     99

                                            Year


Figure 4-39a. Aggregate visibility trends (in deciviews) for 10 eastern Class 1 areas.
                   35
                   30
               o
               CD
CD  ^-u
E
CD  15
Q.

T:  10
               f")    C
               '«    5

                    0
                                              Haziest 20-percent
                                             Jypical 20-percent_
                               Clearest'20:percenf
                     90   91    92   93   94   95   96  97   98   99

                                            Year



Figure 4-39b.  Aggregate visibility trends (in deciviews) for 26 western Class 1 areas.

Source: U.S. Environmental Protection Agency (2001).
June 2003
                         4-181       DRAFT-DO NOT QUOTE OR CITE

-------
 1      days since 1992.  Improvements in visibility were noted in the East for the haziest days;
 2      however, based on monitoring data for 1999, visibility remains significantly impaired, with a
 3      visual range of 23 km for the haziest days compared to a mean visual range of 84 km for the
 4      clearest days.  A 25% and a 14% improvement in visibility impairment were seen for the clearest
 5      and middle days in the West, respectively; whereas conditions for the haziest days degraded by
 6      18.5% (1.7 deciviews) between 1997 and 1999, but were relatively unchanged compared to 1990
 7      conditions (U.S. Environmental Protection Agency, 2001).
 8           Figures 4-40a and 4-40b illustrate aggregate trends in aerosol light extinction, including
 9      trends by major aerosol component for the haziest 20% of days monitored for the 10 eastern
10      Class I areas from 1992 to 1999 and the haziest 20% of days monitored for the 26 western
11      Class I areas from 1990 to 1999. The report also includes a number of maps characterizing
12      aerosol light extinction and key components at 36 IMPROVE sites (all rural except Washington,
13      DC) for 1997 through 1999 (U.S. Environmental Protection Agency, 2001).
14
15      Urban Trends
16           Most of the available visibility measurements, with the exception of the airport visual
17      range measurements, focus on the impact of pollution on visibility in scenic vistas and regional
18      haze (Class I areas). Many urban metropolitan areas are monitoring daily visibility conditions.
19      These findings are generally not available in a published form and may not distinguish between
20      pollution- and weather-related effects. Although the EPA Regional Haze Rule addresses
21      visibility impairment in Class I areas and calls for states to establish goals for improving
22      visibility in these areas and to develop long-term strategies for reducing emissions of air
23      pollutants that cause visibility impairment, the steps states take to implement the regulation will
24      also improve visibility and health in broad areas across the country.
25           Kleeman et al. (2001), citing  previously published studies, provided an historical
26      description of visibility conditions  in Southern California from the early 1930s. Based on airport
27      observation data for 1932 to 1949,  visibility conditions began to decrease in Los Angeles with
28      the advent of industrialization and population growth.  Visibility conditions were worse during
29      the  1940s than the 1930s, and the lowest visibility conditions occurring between 1944 and 1947.
30      During this period, there was nearly a complete loss of extremely good visibility days.  Between
        June 2003                                4-182        DRAFT-DO NOT QUOTE OR CITE

-------
              200
                       Sulfate   '•. ^ Organic Carbon
                      I Nitrate    '!.,. Elemental Carbon
                j
               150-
           _  100 H
          T™

           I   50
           g
           13
           g    0
           5      92    93    94    95    96     97    98    99
               100-
           o
           w
           8   75-
                50-
                  0
                   90   91   92  93  94  95  96   97   98   99
                                        Year
Figure 4-40. (a) Eastern class I area aggregate trends in aerosol light extinction on the
           20% haziest days, including trends by major aerosol component; (b) Western
           class I area aggregate trends in aerosol light extinction on the 20% haziest
           days, including trends by major aerosol component.

Source: U.S. Environmental Protection Agency (2001).
June 2003
4-183
DRAFT-DO NOT QUOTE OR CITE

-------
 1      1943 and 1947, the number of extremely good visibility days during the summer season dropped
 2      to 0.2% from 21%.
 3           Between 1950 and 1961, deteriorating visibility conditions extended eastward from
 4      Los Angeles along the corridor adjacent to the foothills of the San Gabriel and San Bernardino
 5      Mountains.  The visual range in the areas nearest Los Angeles was <3 miles for more than
 6      140 days per year when the relative humidity was < 70%. Further east of Los Angeles, past
 7      Ontario and San Bernardino, the visibility was < 3 miles for 110 days per year during the same
 8      time period.
 9           Improvements in visibility conditions have been made since the 1970s. The largest have
10      been made in the western Los Angeles Basin. In Ontario, the average number of days per year
11      when visual range was greater than 10 miles was 99 between 1976 and 1978 and increased to
12      113 between 1988 and 1990 (Kleeman et al., 2001).  In contrast to the Los Angeles Basin and
13      Ontario, Denver and surrounding areas have experienced high pollution episodes since the
14      1970s. Climate changes trap cooler air under a cover of warm air causing the pollution to remain
15      stagnant over the area producing a brown cloud comprised of a variety pollutants including
16      nitrogen and sulfur oxides, and the grit and dust.
17           Debate over the cause of the increasing pollution in the Denver area and the controversy
18      over converting coal-fired power plants to a cleaner natural gas  system led to initiation of the
19      1973 Denver Air Pollution Study, 1978 Denver Winter Haze Study, 1987-1988 Metro Denver
20      Brown Cloud Study, and the 1993 Denver Brown Cloud Modeling Study. In 1990, in an effort
21      to improve air quality, Denver adopted a visibility standard of 0.076/km (units of atmospheric
22      extinction per kilometer; 20.1 deciviews) averaged over 4 h. While this is a step towards
23      reducing air pollution, the Denver region still exceeds the visibility standard 50 to 80 times per
24      year (Lloyd, 2002).
25           A major air quality study was conducted in Phoenix, AZ, during the fall and winter of 1988
26      to 1990 to address degrading visibility conditions in Phoenix and other urban areas. The
27      objectives were to (1) develop a data base of visibility, air quality, and meteorological
28      measurements;  (2) establish quantitative relationships between light extinction and emission
29      sources; and (3) evaluate measurement systems for short-term and long-term monitoring in
30      Phoenix. The major contributors to light extinction in Phoenix were residual wood burning,
31      primary  mobile source emissions, and secondary  ammonium nitrate (Chow et al., 1990).

        June 2003                               4-184        DRAFT-DO NOT QUOTE OR CITE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
     The Arizona Department of Environmental Quality has conducted optical measurements of
visibility in Tucson since 1993 and in Phoenix since 1994.  The measurements are divided into
the mean of the "dirtiest" 20% of all hours, the mean of all hours, and the mean 20% "cleanest"
of all hours for the entire day and for the 5:00 to 11:00 a.m. period. Figures 4-41a,b represent
the trends in visibility conditions from 1993 to 2001 for Tucson and from 1994 to 2001 for
Phoenix. Visibility on the dirtiest days in the Phoenix metropolitan area has not changed since
visibility monitoring started; but, visibility on the best days has significantly degraded. There is
a seasonally related effect on visibility in Phoenix:  the mean and the dirtiest 20% of all hourly
light extinction are more pronounced during the winter and fall months (Arizona Department of
Environmental Quality,  2002). Daytime visibility is worst during the morning. Samples taken
from 1994 to 1996 showed that organic and elemental carbon dominated visibility impairment
during the dirtiest and cleanest days, whereas ammonium nitrate was an important contributor
during the 20% dirtiest days (Neuroth and Heisler, 2000).
                140
                        1993   1994   1995
                       - dirtiest 20%-all
                       - dirtiest20%-5-11
                                      1996    1997   1998
                                      -*- mean-all
                                      -±- mean-5-11
        1999   2000   2001
       -  cleanest 20%-all
       ^  cleanest 20%-5-11
       Figure 4-41a. Light extinction trends in Tucson, Arizona from 1993 to 2002.
       Source: Arizona Department of Environmental Quality (2002).
       June 2003
                                        4-185
DRAFT-DO NOT QUOTE OR CITE

-------
           GO
        140
        120-
        100-
         80-
         60-
         40-
                20
                       1994    1995   1996
                      F-  dirtiest 20%-all
                      —  dirtiest 20%-5-11
                                         1997    1998    1999    2000    2001
                                        *- mean-all     -^*- cleanest 20%-all
                                        •*r- mean-5-11   -^- cleanest 20%-5-1
       Figure 4-41b. Light extinction trends in Phoenix, Arizona from 1994 to 2001.
       Source: Arizona Department of Environmental Quality (2002).
 1
 2
 3
 4
 5
 9
10
11
12
13
     The State of Virginia compared visibility trends at five locations both with and without
pre-1948 airport visual range data. Average annual airport visual-range-data from Roanoke
(1936 to 1998), DC Reagan National (1930 to 1998), Richmond (1942 to 1998), Lynchburg
(1935 to 1998), and Elkins,  WV (1936 to 1994) were used in the analysis. Their findings
indicate that trends in visibility conditions were dependent on the baseline year used for
visibility measurements. When airport visual range data from 1948 are used as the baseline data
for visibility conditions, the trend in visual range shows declining conditions for all sites. When
pre-1948 data are included in the analysis, visibility conditions basically are unchanged or
improved (except for Elkins).  When 1948 data are used as the baseline for visibility trends
measurements, visual range decreases at all locations. Average annual visual ranged varied
between 7+ miles and 8+ miles for all sites (Virginia Climate Advisory, 2000).
       June 2003
                                        4-186
DRAFT-DO NOT QUOTE OR CITE

-------
 1      4.4.6.2  Current Conditions
 2           Current visibility conditions have been well-characterized for Class I areas using updated
 3      data from the IMPROVE network (U.S. Environmental Protection Agency, 2001; Malm et al.,
 4      2000; IMPROVE, 1998). During recent decades, daytime visibility conditions at all major
 5      airports throughout the United States were recorded hourly by human observation. These data
 6      were used to determine current visibility conditions and visibility trends in the United States,
 7      as well as the spatial distribution of visibility conditions (Trijonis et al., 1991).  The use of
 8      human observation is being replaced by an automated observing system, the Automated Surface
 9      Observing System (ASOS).  More than 900 airports are currently commissioned.  In addition,
10      the U.S.  Environmental Protection Agency has deployed a new national monitoring network to
11      assess PM2 5 concentrations and composition.
12           More detailed information on visibility conditions for urban and suburban areas will
13      become more widely available as data from the national PM25 speciation monitoring network
14      and the ASOS airport visibility network are further analyzed.  Efforts are currently underway to
15      develop  a web-based system to allow the use of the high resolution ASOS data in air quality
16      monitoring and assessment programs.  The objectives are to collect and quality control an
17      archive of the ASOS visibility data, to deliver processed hourly  visibility data to public and air
18      quality communities, and to use the web-based system to support the acquisition and
19      dissemination of visibility data (Falke, 2001).
20
21      4.3.7  Economics of Particulate Matter Visibility Effects
22           Society recognizes the need to impose remedies for repairing and preventing further
23      anthropogenic pollutant-related effects on visibility conditions.  Various methods have been
24      utilized to help determine the economic valuation of changes in  visibility.  Where possible,
25      direct economic valuation can be determined using marketplace cost estimates.  Avoided-cost
26      methods estimate the costs of pollution by using the expenditures that are made necessary by
27      pollution damage.  As an example, if ambient levels of PM result in increased frequency of
28      building cleaning or repainting, the appropriately  calculated increase in these costs is a
29      reasonable estimate of true economic damage. Benefits associated with reductions in the
30      pollution levels are then represented by the avoided costs of these damages.
        June 2003                                4-187       DRAFT-DO NOT QUOTE OR CITE

-------
 1           Estimating the benefits of clear skies is a more difficult and less precise exercise because,
 2      although the public values aesthetic views, they are not directly bought and sold in the
 3      marketplace. However, there are several methods available to economists to estimate the
 4      economic impact of these kinds of changes in environmental conditions (Freeman, 1993).  These
 5      methods include hedonic valuation or pricing, contingent valuation and contingent choice, and
 6      travel cost (Johnson and Desvousges, 1997; Hanley and Spash, 1993). The primary methods
 7      used to date for valuation of visibility have been the hedonic price and contingent valuation
 8      methods (Hanley and Spash, 1993); but this is not an exact science, and there are still issues and
 9      limitations associated with each of these methods.
10           Hedonic pricing can be used to estimate economic valuations for environmental effects that
11      have a direct effect on market values. It relies on the measurement of differentials in property
12      values under various environmental quality conditions including air pollution and environmental
13      amenities, such as aesthetic views. The hedonic method works by analyzing the way that market
14      prices change with changes in environmental quality.  Part of the economic costs imposed by
15      PM-related reductions in visibility can be estimated by looking at the differences in sales price
16      between otherwise identical houses that have different degrees of visibility impairment.
17           The contingent valuation method (CVM) is the most widely used method for estimating
18      value changes in both visibility and ecosystem functions (Hanley and Spash, 1993; Chestnut,
19      1997; Watson and Chow, 1993). The CVM creates hypothetical markets for goods and services
20      that have no market-determined price. For determination of visibility evaluation, individuals are
21      shown photographs with perceivable differences in visibility levels.  Carefully structured surveys
22      are administered to estimate the amount of compensation equivalent to a given change in
23      environmental  quality or, equivalently, how much an individual would be willing to pay (WTP)
24      for improvements in environmental quality or willing to accept (WTA) existing conditions
25      without further deterioration.  There is an extensive scientific literature and body of practice on
26      both this theory and technique; however, there are still concerns about the use of this technique
27      for quantitative purposes.
28           The travel-cost method estimates can be used to estimate the value of recreational benefits
29      of an ecosystem based on the environmental quality at the site.  The travel-cost method uses
30      information on actual behavior rather than responses to hypothetical scenarios.  The time and
31      travel expenses incurred to visit a site represents the price of access to the site.  The willingness

        June 2003                                4-188       DRAFT-DO NOT QUOTE OR CITE

-------
 1      to pay to maintain the site is determined by the number of times the individual visits the site at
 2      different travel costs.
 3           The effects of PM on visibility may differ widely between urban residential and
 4      recreational areas.  Therefore, separate estimates are needed to account for impact associated
 5      with changes in visibility in residential and recreational (Class I) areas.  Chestnut and Dennis
 6      (1997) compared the findings of the more recent studies on the economic impact of changes in
 7      regional haze using the contingent valuation method in residential areas in several eastern cities
 8      and in Los Angeles and San Francisco and using the hedonic value method in Los Angeles and
 9      San Francisco.  The findings  of the contingent valuation studies are discussed in Table 4-16.
10      Findings using the contingent and hedonic methods of valuation for Los Angeles and San
11      Francisco are compared in Table 4-17.
12           Using the contingent valuation method, Chestnut and Rowe (1990) found that 83% of those
13      individuals responding to a survey on visibility were willing to pay to improve visibility in the
14      National Parks. Survey participants were selected from California, Arizona, Missouri,
15      New York, and Virginia.  The National Parks from three regions (California, southwestern
16      United States, and southeastern United States) were considered in different versions of the
17      survey.  The survey included questions on past and future visitations to national parks, potential
18      pollution effects from human activities outside of the park, three hypothetical visibility
19      scenarios,  socioeconomic characteristics, and various photographic presentations of visibility
20      conditions within the parks. Higher responses were noted for residents residing in the state or
21      region where the national park was located; responses for males and the elderly were generally
22      lower and there was a direct correlation between household income and the response.
23           Using the results from the Chestnut and Rowe (1990) study, Chestnut and Dennis (1997)
24      calculated an extinction coefficient of 85 for in-state residents and 50 for out-of-state residents.
25      These extinction coefficients were suggested to represent an annual willingness to pay per
26      household of $15 and $9 for a 20% improvement in visual range.
27
28
29
        June 2003                                4-189       DRAFT-DO NOT QUOTE OR CITE

-------
       TABLE 4-16. RESIDENTIAL VISIBILITY CONTINGENT VALUATION
                               STUDY RESULTS
City
Atlanta
Chicago

Chicago

Atlanta


Boston


Mobile


Washington, DC


Cincinnati


Miami


Cincinnati
Mean WTP
($1994)
Unadj. $44
Partial $28
Full $20
-$361
$346
$430
-$301
$289
$432
-$222
$212
$262
-$240
$257
$302
-$356
$366
$465
-$88
$87
$98
-$152
$136
$160
$198
Starting-Ending
Visual Range (VR)
(miles)
17.6-20

9-4
9-18
9-30
12-7
12-22
12-32
18- 13
18-28
18-38
10-5
10-20
10-30
15- 10
15-25
15-35
9-4
9-19
9-29
13-8
13-19
13-29
11.4-16.4
WTP for 20%
Extinction Change in VR
Coefficient ($) Reference
346 $63
McClelland et al.
222 (1993)
159
416 $76 Tolleyetal.
(1986)

469 $86


422 $77


312 $57


635 $116


120 $22


256 $47


602 $110 Rae(1983)
 Source: Chestnut and Dennis (1997).
June 2003
4-190
DRAFT-DO NOT QUOTE OR CITE

-------
          TABLE 4-17. RESIDENTIAL VISIBILITY VALUATION STUDY RESULTS FOR
                              LOS ANGELES AND SAN FRANCISCO
City
Los
Angeles

San
Francisco

Method
CVM
Property
Value
CVM
Property
Valiip
Starting-Ending
Mean Visual Range
WTP (VR) (miles)
$130 2-12
$333 2-28
$183 12-28

$211 18.6-16.3
$124 16.3 - 18.3

Extinction WTP for 20%
Coefficient Change in VR Reference
119 $22 Brookshire
etal. (1979)
1328 $242
$245-647 Trijonis et al.
(1984)
Loehman et al.
(1985)
$496 - 552 Trijonis et al.
riQ8/n
        Adapted from: Chestnut and Dennis (1997).


 1     4.4   PARTICULATE MATTER EFFECTS ON MATERIALS
 2           Effects of air pollution on materials are related to both aesthetic appeal and physical
 3     damage.  Studies have demonstrated that particles, primarily consisting of carbonaceous
 4     compounds, cause soiling of commonly used building materials and culturally important items,
 5     such as statutes and works of art. Physical damage from the dry deposition of air pollutants,
 6     such as PM (especially sulfates and nitrates) and SO2, and the absorption or adsorption of
 7     corrosive agents on deposited particles also can result in the acceleration of naturally occurring
 8     weathering processes of man-made building and cultural materials.
 9           In the atmosphere, PM may be "primary," existing in the same form in which it was
10     emitted, or "secondary," formed by the chemical reactions of free, absorbed, or dissolved gases.
11     The major constituents of atmospheric PM are sulfate,  nitrate, ammonium, and hydrogen ions;
12     particle-bound water; elemental carbon; a great variety of organic compounds; and crustal
13     material. A substantial fraction of the fine particle mass, particularly during the warmer months,
14     is secondary sulfate and nitrate. Sulfates may be formed by the gas-phase conversion of SO2 to
15     H2SO4 by OH radicals and aqueous-phase reactions of SO2 with H2O2, O3, or O2.  During the day,
16     NO2 may be converted to nitric acid (HNO3) by reacting with OH radicals. Nitrogen dioxide
       June 2003
4-191
DRAFT-DO NOT QUOTE OR CITE

-------
 1      also can be oxidized to HNO3 by a sequence of reactions initiated by O3.  A more detailed
 2      discussion of the atmospheric chemistry of PM appears in Chapter 2 of this document.
 3
 4      4.4.1  Corrosive Effects of Particles and Sulfur Dioxide on
 5             Man-Made Surfaces
 6           Limited new studies have been published that better define the role of air pollution in
 7      materials damage. This section briefly summarizes information on exposure-related effects on
 8      materials and sulfur-containing pollutants (formed by the chemical reactions of SO2 with other
 9      atmospheric pollutants) addressed in the 1996 PM AQCD (U.S. Environmental Protection
10      Agency, 1996a) and presents relevant information published since completion of that document.
11      The effects of nitrates on man-made building materials and naturally occurring cultural materials
12      were discussed in the earlier EPA Nitrogen Oxides Criteria Document (U.S. Environmental
13      Protection Agency, 1993).
14
15      4.4.1.1   Metals
16           Metals undergo natural weathering processes in the absence of environmental pollutants.
17      The additive effect of pollutants on the natural weathering processes depend on the nature of the
18      pollutant, the deposition rate (the uptake of a pollutant by the material's surface), and the
19      presence of moisture.  The influence of the metal-protective corrosion film, the presence of other
20      surface electrolytes, the orientation of the metal surface, the presence of surface moisture, and
21      the variability in the electrochemical reactions will also contribute to the effect of pollutant
22      exposure on metal surfaces.
23           Several studies demonstrate the importance of the duration of surface wetness (caused by
24      dew and fog condensation and rain) on metals.  Surface moisture facilitates the deposition of
25      pollutants, especially SO2, and promotes corrosive electrochemical reactions on metals (Haynie
26      and Upham, 1974; Sydberger and Ericsson, 1977). Of critical importance is the formation of
27      hygroscopic salts on the metal that increases the duration of surface wetness and, thereby,
28      enhances the corrosion process.
29           Pitchford and McMurry (1994) and Zhang et al. (1993) demonstrated particle-size-related
30      effects of relative humidity.  The effect of temperature on the rate of corrosion is complex.
31      Under normal temperature conditions, temperature would not have an effect on the rate of
32      corrosion; but when the temperature decreases, the relative humidity increases and the diffusivity
        June 2003                                4-192       DRAFT-DO NOT QUOTE OR CITE

-------
 1      decreases. The corrosion rate decreases as the temperature approaches freezing because ice
 2      prohibits the diffusion of SO2 to the metal surface and minimizes electrochemical processes
 3      (Haynie, 1980; Biefer, 1981; Sereda, 1974).
 4            The metal protective corrosion film (i.e., the rust layer on metal surfaces) provides some
 5      protection against further corrosion.  The effectiveness of the corrosion film in slowing down the
 6      corrosion process is affected by the solubility of the corrosion layer and the concentration and
 7      deposition rate of pollutants. If the metal-protective corrosion film is insoluble, it may add some
 8      protection against acidic pollutants. An atmospheric corrosion model that considers the
 9      formation and dissolution of the corrosion film on galvanized steel was proposed by Spence
10      et al. (1992).  The model considers the effects of SO2, rain acidity, and duration of wetness on
11      the rate of corrosion. Although the model does not specifically characterize particle effects, the
12      contribution of particulate sulfate was considered in model development.
13            Whether suspended particles actually enhance the corrosion of metals is not clear.  Several
14      studies suggest that suspended particles will promote the corrosion of metals (Goodwin et al.,
15      1969; Barton, 1958; Sanyal and Singhania, 1956; Baedecker et al., 1991); however, other studies
16      have not demonstrated a correlation between particle exposure and metal corrosion  (Mansfeld,
17      1980; Edney et al., 1989).  Walton et al. (1982) suggested that catalytic species within several
18      species in fly ash promote the oxidation of SOX to a corrosive state.  Still other researchers
19      indicate that the catalytic effect of particles is not significant and that the corrosion rate is
20      dependent on the conductance of the thin-film surface electrolytes during periods of wetness.
21      Soluble particles likely increase the solution conductance (Skerry et al., 1988; Askey et al.,
22      1993).        The corrosion of most ferrous metals (iron, steel, and steel alloys)  is  increased by
23      increasing SO2 exposure. Steels are susceptible to corrosion when exposed to SO2 in the absence
24      of protective organic or metallic coatings. Studies on the corrosive effects of SO2 on steel
25      indicate that the rate of corrosion increases with increasing SO2 and is dependent on the
26      deposition rate of the SO2 (Baedecker et al., 1991; Butlin et al., 1992a). The corrosive effects of
27      SO2 on aluminum is exposure-dependent, but appears to be insignificant (Haynie, 1976; Fink
28      et al., 1971; Butlin et al., 1992a). The rate of formation of the patina (protective covering) on
29      copper can take as long as five years and is dependent on the SO2 concentration, deposition rate,
30      temperature, and relative humidity (Simpson and Horrobin, 1970). Further corrosion is
31      controlled by the availability of copper to react with deposited pollutants (Graedel et al., 1987).

        June 2003                                 4-193       DRAFT-DO NOT QUOTE OR CITE

-------
 1      Butlin et al. (1992a), Baedecker et al. (1991), and Cramer et al. (1989) reported an average
 2      corrosion rate of 1 |im/year for copper; however, less than a third of the corrosion was attributed
 3      to SO2 exposure, suggesting that the rate of patina formation was more dependent on factors
 4      other than SO2.  A report by Strandberg and Johansson (1997) showed relative humidity to be
 5      the primary factor in copper corrosion and patina formation. The results of the studies on
 6      particles and SO2 corrosion of metals are summarized in Table 4-18.
 7
 8      4.4.1.2  Painted Finishes
 9           Exposure to air pollutants affects the durability of paint finishes by promoting
10      discoloration, chalking, loss of gloss, erosion, blistering, and peeling. Evidence exists that
11      indicates particles can damage painted finishes by serving  as carriers for corrosive pollutants
12      (Cowling and Roberts, 1954) or by staining and pitting of the painted surfaces (Fochtman and
13      Langer, 1957; Wolff et al., 1990).
14           The erosion rate of oil-based house paint has been reported to be enhanced by exposure to
15      SO2 and high humidity. In a study by Spence et al. (1975), an erosion rate of
16      36.71 ± 8.03 jam/year was noted for oil-based house paint samples exposed to SO2 (78.6 |ig/m3),
17      O3 (156.8 |ig/m3), and NO2 (94 |ig/m3), and low humidity (50%). The erosion rate increased
18      with increased SO2 and humidity. The authors concluded that SO2 and humidity accounted for
19      61% of the erosion. Acrylic coil coating and vinyl coil coating shows less pollutant-related
20      erosion.  Erosion rates range from 0.7 to 1.3 jim/year and 1.4 to 5.3 jim/year, respectively.
21      Similar findings on SO2-related erosion of oil-based house paints and coil coatings have been
22      reported by other researchers (Davis et al., 1990; Yocom and Grappone, 1976; Yocom and
23      Upham, 1977; Campbell  et al., 1974). Several studies suggest that the effect of SO2 is caused by
24      its reaction with extender pigments such as calcium carbonate and zinc oxide (Campbell et al.,
25      1974; Xu and Balik,  1989; Edney, 1989; Edney et al., 1988, 1989).  However, Miller et al.
26      (1992) suggested that calcium carbonate acts to protect paint substrates. Another  study indicated
27      that exposure to SO2 can  increase the drying time of some  paints by reacting with  certain drying
28      oils and will compete with the auto-oxidative curing mechanism responsible for crosslinking the
29      binder (Holbrow, 1962).
30
31

        June 2003                                4-194        DRAFT-DO NOT QUOTE OR CITE

-------
               TABLE 4-18. CORROSIVE EFFECTS OF PARTICIPATE MATTER AND SULFUR DIOXIDE ON METALS
to
O
o
Metal
Exposure Conditions
Comments
                                                                                                                                       Source
fe
H
6
o
o
H
O
O
H
W
O
O
HH
H
W
         Mild Steel
         Galvanized Steel
         Zinc
         Zinc
         Carbon Steel
         Weathering Steel
         Aluminum


         Aluminum
                     Specimens exposed to SO2 and O3 under natural
                     and artificial conditions, and to NO2 under natural
                     conditions.  SO2 concentrations ranged from 2.1 to
                     60 ng/m3. Annual average concentrations were
                     about 20 ng/m3. Meteorological conditions were
                     unaltered.  Specimens exposed at 29 sites for
                     2 years for mild steel and 1 y for galvanized steel.
                     Rolled zinc specimens exposed at various sites
                     around the country (rural, industrialized, marine)
                     for up to 20 years.  Actual pollutant exposures not
                     reported.

                     Specimens exposed at 5 sites for 1 to 5 years.
                     Average SO2 concentrations ranged from 2 ± 4 to
                     15 ± 17 ppb (5.2 ± 10.4 to 39.3 ± 44.5 ng/m3).
                     PM concentrations ranged from 14 to 60 (ig/m3.
                     Highest pollutant concentrations recorded at 1 year
                     exposure site.

                     See Baedecker et al. (1991) above for exposure
                     conditions.
                     See Baedecker et al. (1991) above for exposure
                     conditions.

                     See Butlin et al. (1992a) above for exposure
                     conditions.
                                   Steel corrosion was dependent on long-term SO2
                                   exposure. The corrosion rate was about 50 urn/year
                                   for mild steel specimens for most industrial sites, but
                                   ranged from 21 to 71 urn/year.  The corrosion rate
                                   ranged from 1.45 to 4.25 urn/year for galvanized
                                   steel.  The authors concluded that rainfall also may
                                   have a significant effect on galvanized steel based on
                                   a corrosion rate of 3.4 nm/year seen at a very wet site.

                                   The highest corrosion rates were associated with
                                   industrialized environments and marine environments
                                   in direct contact with salt spray.
                                   Average corrosion rate ranged from 0.63 to
                                   1.33 um/y. The highest corrosion was noted in the
                                   most industrialized area. However, the corrosion
                                   rates did not differ significant regardless of the SO2
                                   concentration, suggesting that SO2 exposure may not
                                   be the dominant factor in zinc corrosion.

                                   Average corrosion rate for samples exposed for
                                   5 years ranged from 6.6 to 12.8 urn/year for carbon
                                   steel and 3.7 to 5.0 ^rn/year for weathering steel.
                                   Highest corrosion rate noted for samples exposed for
                                   1 year.

                                   Corrosion rate was very low at all sites and ranged
                                   from 0.036 to 0.106 urn/year.

                                   Corrosion greater on the under side of specimens,
                                   possibly because of lack of washoff and increased PM
                                   in area. Maximum corrosion rate was 0.85 urn/year.
                                   Pit depths of up to 72 um were noted after 2 years of
                                   exposure.
                               Butlin etal. (1992a)
                                Showak and Dunbar
                                (1982)
                               Baedecker et al. (1991)
                               Cramer etal. (1989)
                               Baedecker et al. (1991)
                               Cramer etal. (1989)
                               Baedecker etal. (1991)
                               Butlin etal. (1992a)

-------
          TABLE 4-18 (cont'd). CORROSIVE EFFECTS OF PARTICIPATE MATTER AND SULFUR DIOXIDE ON METALS
to
O
o
Metal
Exposure Conditions
Comments
                                                                                                                                     Source
fe
H
6
o
o
H
O
O
H
W
O
O
HH
H
W
         Copper
         Copper
         Copper
         Copper
         Iron
                    See Baedecker et al. (1991) above for exposure
                    conditions.
                    See Butlin et al. (1992a) above for exposure
                    conditions.
                    Specimens exposed to 4 to 69 ppb (10.4 to
                    180.7 ug/m3) and 1.0 ppm (2,618.7 ug/m3) SO2 for
                    20h at various relative humidities.
                    Specimens exposed artificially to 0.49 ± 0.01 ppm
                    (187 ± 3.8 ug/m3) SO2 for 4 weeks at 70 and 90%
                    relative humidity.

                    Specimens from restorations of Acropolis
                    monuments over many years. The oldest
                    specimens were 142 years old. Other specimens
                    used for monument restoration after 1950.
                                  Average corrosion rate for 3- and 5-year exposures
                                  was about 1 urn/year but the soluble portion was less
                                  than a third of that which could be contributed to SO2
                                  exposure.  Dry deposition of SO2 was not as important
                                  in patina formation as wet deposition of H+

                                  Majority of test sites showed a corrosion rate of
                                  1 ± 0.2 urn/year.  The corrosion rate was
                                  1.48 urn/year at the site receiving the most rainfall.
                                  The lowest corrosion rate, 0.66 urn/year, was
                                  associated with low rainfall, low SO2.

                                  SO2 had no effect on copper when relative humidity
                                  was -<75%. Increasing relative humidity increases
                                  patina formation in presence of trace SO2.
                                  No SO2-related effects were noted on copper
                                  specimens exposed to high SO2 regardless of the
                                  percent relative humidity.

                                  Corrosive effect of SO2 on copper increased with
                                  increasing relative humidity.
                                  Specimens used after 1950s had an oxidation rate
                                  25% greater than those specimens before 1950. The
                                  accelerated oxidation had a negative impact on the
                                  structure, producing a quicker formation of cracks
                                  inside the marble.
                               Baedecker etal. (1991)
                               Butlin etal. (1992a)
                               Strandberg and
                               Johansson (1997)
                               Eriksson etal. (1993)
                               Zuburtikudis and
                               Triantafyllou (2001)

-------
 1      4.4.1.3  Stone and Concrete
 2           Numerous studies suggest that air pollutants can enhance the natural weathering processes
 3      on building stone.  The development of crusts on stone monuments has been attributed to the
 4      interaction of the stone's surface with sulfur-containing pollutants, wet or dry deposition of
 5      atmospheric particles, and dry deposition of gypsum particles from the atmosphere. Because of
 6      a greater porosity and specific surface, mortars have a greater potential for reacting with
 7      environmental pollutants (Zappia et al., 1998). Details on these studies are discussed in
 8      Table 4-19. The stones most susceptible to the deteriorating effects of sulfur-containing
 9      pollutants are the calcareous stones (limestone, marble, and carbonated cement). Exposure-
10      related damage to building stones result from the formation of salts in the stone that are
11      subsequently washed away during rain events, leaving the stone surface more susceptible to the
12      effects of pollutants. Dry deposition of sulfur-containing pollutants promotes the formation of
13      gypsum  on the stone's surface.  Gypsum is a gray to black crusty material comprised mainly of
14      calcium  sulfate dihydrate from the reaction of calcium carbonate (calcite) in the stone with
15      atmospheric SO2 and moisture (relative humidities exceeding 65%) according to the following
16      reaction.
17
18                     CaCO3 (marble) + SO3 + 2H2O -» CaCO4-2H2O + CO2                 (4-13)
19
20      The sulfate anions formed in the moist air reacts with the Ca+2 through diffusion processes
21      forming  the gypsum (Zuburtikudis and Triantafyllou, 2001). Approximately 99% of the sulfur
22      in gypsum is sulfate because of the sulfonation process caused by the deposition of SO2 aerosol.
23      Sulfites  also are present in the gypsum layer as an intermediate product (Sabbioni et al.,  1996;
24      Ghedini  et al., 2000; Gobbi et al., 1998; Zappia et al., 1998).  Gypsum is more soluble than
25      calcite and is known to form on limestone, sandstones, and marble when exposed to SO2.
26      Gypsum also has been reported to form on granite stone by replacing silicate minerals with
27      calcite (Schiavon et al., 1995).  Gypsum occupies a larger volume than the original stone,
28      causing the stone's surface to become cracked and pitted.  The rough surface serves as a site for
29      deposition of airborne particles. As the gypsum grows, it becomes loose and falls apart
30      (Zuburtikudis and  Triantafyllou, 2001).
31

        June 2003                                 4-197       DRAFT-DO NOT QUOTE OR CITE

-------
                  TABLE 4-19.  CORROSIVE EFFECTS OF PARTICIPATE MATTER AND SULFUR DIOXIDE ON STONE
to
O
o
Stone
              Exposure Conditions
                                                                                                               Comments
Source
vo
oo
 fe
 H
 6
 o
 o
 H
O
 O
 H
 W
 O
 O
 HH
 H
 W
          Vermont marble          Runoff water was analyzed from seven summer
                                  storms. SO2 concentration stated to be low.

          Marble sandstone         Analysis of runoff water for five slabs test exposed
                                  to ambient conditions at a angle of 30° to horizontal.
          Limestone
Portland limestone
White Mansfield
dolomitic sandstone
Monk's Park limestone

Sandstones (calcite and
noncalcite stones)
          Limestones
          Sandstones
          Marble
          Granite
          Basalt
          Portland limestone
          Massangis Jaime Roche
          limestone
          White Mansfield
          dolomitic
Ambient air conditions. Exposure ranged from 70 to
1065 days.  Averaged pollutant exposure ranged
from 1.4 to 20.4 ppb (3.7 to 53.4 |ig/m3) SO2; 4.1 to
41.1 ppbNOx; 2.4 to 17.4 ppb (4.5 to 32.7 |ig/m3)
NO2; 10.1 to 25.6 ppb (19.8 to  50.2 |ig/m3) O3.

Experimental tablets exposed under sheltered and
unsheltered ambient air conditions. Exposure for
1 and 2 years.
                                  Ambient air; low concentrations of sulfates, SO2,
                                  and nitrates; RH sufficient to produce condensation
                                  on stones rarely occurred.
                        Ambient air; urban and rural locations in
                        Mediterranean.
                        Samples exposed to SO2, NO2, and NO at 10 ppmv,
                        both with and without O3 and under dry (coming to
                        equilibrium with the 84% RH) or wetted with
                        CO2-equilibrated deionized water conditions.
                        Exposure was for 30 days.
                                                                          Between 10 to 50% of calcium in runoff water estimated from              Schuster et al.
                                                                          gypsum formation from dry deposition of SO2.                           (1994)

                                                                          Pollutant exposure related erosion was primarily caused by dry             Baedecker
                                                                          deposition of SO2 and nitric acid between rain events and wet deposition    et al. (1992)
                                                                          of hydrogen ion. Recession estimates ranged from 15 to 30 urn/year for
                                                                          marble and 25 to 45 jim/year for limestone. A large portion of the
                                                                          erosion results from the reaction of CO2 with the calcium in the stone.

                                                                          Increased stone weight loss with increased SO2.  Rainfall did not           Webb et al.
                                                                          significantly affect stone degradation. Stone loss associated with           (1992)
                                                                          SO2 exposure estimated to be 24 urn/year.  Slight trend in decreasing
                                                                          stone loss with increasing length of exposure.
                                                                                   Significant correlations existed between the mean annual SO2              Butlin et al.
                                                                                   concentration, rainfall volume, and hydrogen ion loading and the weight    (1992b)
                                                                                   changes.
                                                  Insignificant differences in erosion rate found between calcite             Petuskey et al.
                                                  and noncalcite sandstone. Moisture affected the rate of pollutant          (1995)
                                                  deposition and enhanced susceptibility to pollutant related erosion. Rain
                                                  events given as primary factor affecting stone erosion. Pollutant related
                                                  erosion judged to be insignificant.

                                                  Crusts on stones were found to contain two layers; top layer, usually        Garcia-Valles
                                                  black in color, composed of gypsum between 40 and 400 |im thick.         et al. (1998)
                                                  Innermost layer, ranging from brown to orange in color, primarily
                                                  consisted of calcite, between 10 and 600 |im thick. Gypsum-rich
                                                  layer thought to be the result of sulfation of the calcitic layer by
                                                  atmospheric pollutants or dry or wet deposition of atmospheric dust.

                                                  In the absence of moisture, little reaction is seen. SO2 is oxidized to        Haneef et al.
                                                  sulfates in the presence of moisture. The effect is enhanced in the         (1993)
                                                  presence of O3. Massangis Jaune Roche limestone was the least affected
                                                  by the pollutant exposure. Crust lined pores of specimens exposed to
                                                  SO,.

-------
>
            TABLE 4-19 (cont'd). CORROSIVE EFFECTS OF PARTICIPATE MATTER AND SULFUR DIOXIDE ON STONE
          Stone
                Exposure Conditions
                                                                                                     Comments
                                                                                                                                                        Source
to
o
o
to
vo
vo
 fe
 H
 6
 o
 o
 H
O
 O
 H
 W
 O
 O
 HH
 H
 W
Monk's Park Portland
limestone
         Carrara marble
         Travertine
         Tranistone
         Carrara marble
         Georgia marble
         Carrara marble
         Monk's Park
         limestone
         Portland limestone
         Lime mortar
         Pozzolan mortar
         Cement mortar
         Limestone
         Travertine marble
Samples exposed for 2 mo under both sheltered and
unsheltered conditions. Mean daily atmospheric SO2
concentration was 68.7 ug/m3 and several heavy rainfalls.
Sample exposed in laboratory to 3 ppm SO2 and 95% RH at
25 °C for 150 days. Samples were coated with three
carbonaceous particle samples from combustion sources,
and with activated carbon and graphite.

Samples exposed in sheltered ambient environment for 6,
12, or 20 mo.
Samples exposed for 6 mo (cold and hot conditions) in
ambient environment. PM concentrations ranged from 57.3
to 116.7 ug/m3 (site 1) and 88 to 189.8 ug/m3 (site 2).
Some exposures also were associated with high SO2, NO,
andNO2.

Samples artificially exposed to fly-ash containing
1309.3 ug/m3 SO2 (0.5 ppm) at 95% RH and 25 °C for
81 or 140 days.  Fly-ash samples from five different
sources were used in the study.

Samples exposed to 7,856 ug/m3 (3 ppm) SO2 at 100% RH
and 25 °C for 30, 60, or 90 days; samples  sprayed with
bidistilled water every 7 days to simulate rainfall.
Samples exposed under actual ambient air conditions at
two locations in Rome. Monitoring data obtained for SO2,
NO, NO2, and total suspended particulates (TSP) but not
reported.  Exposure was for four seasons.
                                                                                       Significant amounts of gypsum were noted on the Portland stone.     Viles (1990)
                                                                                       Sheltered stones also showed soiling by carbonaceous particles
                                                                                       and other combustion products. Etch holes and deep etching was
                                                                                       noted in some of the exposed unsheltered samples.

                                                                                       Exposure to particles from combustion processes enhanced          Sabbioni et al.
                                                                                       sulfation of calcareous materials by SO2 because of metal content     (1996)
                                                                                       of particles.
                                                                             Carrara marble found to be more reactive with SO2 than Georgia     Yerrapragada
                                                                             marble possibly because of the compactness of the Georgia          et al. (1994)
                                                                             marble. Greater effects noted when samples were also exposed
                                                                             to N02.

                                                                             Pollutant exposed samples showed increased weight gain over       Realini et al.
                                                                             that expected from natural weathering processes. There was a       (1995)
                                                                             blackening of stone samples exposed to carbonaceous rich
                                                                             particulate matter.
                                                                             Exposure to fly-ash did not enhance oxidation of SO2 to sulfates.     Hutchinson
                                                                             Mineral oxides in fly ash contributed to sulphation of CaCO3.        et al. (1992)
                                                                             Exposure to SO2 produced significant quantities of calcium sulfite    Zappia et al.
                                                                             and calcium sulfate on specimens; however, the amount produced    (1994 )
                                                                             was dependent of the porosity, specific surface, and alkalinity of
                                                                             the sample.

                                                                             TSP exposure increased the cleaning frequency for stone            Lorusso et al.
                                                                             monuments.  Monuments are soiled proportionately overtime,        (1997)
                                                                             based on brightness values. Horizontal surfaces showed higher
                                                                             graying values because of particle sediment.

-------
            TABLE 4-19 (cont'd). CORROSIVE EFFECTS OF PARTICIPATE MATTER AND SULFUR DIOXIDE ON STONE
to
O
o
Stone
             Exposure Conditions
                                                                                                            Comments
Source
to
o
o
fe
H
6
o
o
H
O
O
H
W
O
O
HH
H
W
         Limestone
         Quartz-cemented
           sandstone
         Calcite-cemented
           sandstone
         Granite
         Brick

         Carrara marble
Limestone
Sandstone
         Carrara marble
         Travertine marble
         Irani limestone
         Portland limestone
         Lime mortar
         Pozzolan mortar
         Cement mortar

         Lime mortar
         Carrara marble
                    Samples from structures exposed for varying
                    periods of time under ambient air conditions.
                    Samples selected because of black layer on surface.
                    Samples from structures taken from monuments in
                    Venice Italy.  Samples included those shielded from
                    running water (unwashed), areas at the interface of
                    the washed and unwashed areas, and washed areas.
                    Samples of fog and rain near the monument were
                    also taken for analysis.
Samples of ancient grey crust formed between
1180 and 1636 on the Church of Saint Trophime in
Arks and formed between 1530 and 1187 on the
Palazz d'Accursio in Bolonga.

Samples of the stones and mortars were
representative of those used in the past and currently
for new construction and restorations. Samples
were exposed for 6, 12, and 24 mo under ambient
conditions in Milan.
                    Sample of black crust taken from Zamboni
                    Tower Gate.
                    Samples of crust removed from Milan General
                    Hospital, built around 1937.
                                                Black layers were found to be primarily comprised of iron compounds, quartz,    Nord and
                                                silicate, soot, and dirt.                                                     Ericsson
                                                                                                                        (1993)
                                                                               Sulfate was higher in fog compared to other ions. The average concentration     Fassina et al.
                                                                               of chlorides and sulfates were 7 to 16 times higher in fog than rain. The degree   (2001)
                                                                               of sulphonation on stone samples from areas shielded from running water was
                                                                               < 40%.  At the interface between washed and unwashed surfaces and the
                                                                               washed surfaces, sulphonation is > 40%. Dendrite shaped crust from the
                                                                               transformation of calcium carbonate into gypsum was found on samples
                                                                               shielded from rain.  Samples from the washed areas displayed superficial
                                                                               granular disaggregation, a natural process of deterioration ascribed to natural
                                                                               agents; however, the process was accelerated due to the ambient air sulfates.

                                                                               Crust samples contained calcite, soil dust, carbonaceous particles, and gypsum    Ausset et al.
                                                                               crystals.                                                                (1998)
                                                                     Mortars were more reactive than the stones. Of the mortars, cement and         Zappia et al.
                                                                     pozzolan mortar were more reactive than the lime mortar. Carrara marble was    (1998)
                                                                     the least reactive of the stones. The maximum amount of degradation was
                                                                     found in areas sheltered from rain.
                                                Exposure to environmental pollutants caused the formation of two separate       Sabbioni et al.
                                                layers on the mortar: An outer thin surface black crust composed of gypsum      (1998)
                                                and carbonaceous particles and the inner composed of products from the
                                                dissolution and sulphation of the carbonate matrix in the mortar.

                                                Gypsum main component of crust followed by carbonaceous particles and iron    Bugini et al.
                                                oxides.  Estimated rate of crust formation was 2-5 urn/year.  Total amount of      (2000)
                                                gypsum formed over the lifetime of exposure was 5 to 13 mg/cm2, an estimated
                                                0.2 mg/cm2/year.

-------
 1           The dark colored gypsum is caused by surface deposition of carbonaceous particles
 2      (noncarbonate carbon) from combustion processes occurring in the area (Sabbioni, 1995;
 3      Saiz-Jimenez, 1993; Ausset et al., 1998; Hermosin and Saiz-Jimenez, 2000), trace metals
 4      contained in the stone, dust, and numerous other anthropogenic pollutants.  After analyzing
 5      damaged layers of several stone monuments, Zappia et al. (1993) found that the dark-colored
 6      damaged surfaces contained 70% gypsum and 20% noncarbonate carbon.  The lighter colored,
 7      damaged layers were exposed to rain and contained 1% gypsum and 4% noncarbonate carbon. It
 8      is assumed that rain removes reaction products, permitting further pollutant attack of the stone
 9      monument and likely redeposits some of the reaction products at rain runoff sites on the stone.
10      Following sulfur compounds, carbon was reported to be the next highest element in dark crust on
11      historical monuments in Rome.  Elemental carbon and organic  carbon accounted for 8 and 39%
12      of the total carbon in the black crust samples.  The highest percentage of carbon, carbonate
13      carbon, was caused by the carbonate matrix in the stones. The  high ratio of organic carbon to
14      elemental carbon indicates the presence of a carbon source other than combustion processes
15      (Ghedini et al., 2000). Cooke and Gibbs (1994) suggested that stones damaged during times of
16      higher ambient pollution exposure likely would continue to exhibit a higher rate of decay,
17      termed the "memory effect," than newer stones exposed under lower pollution conditions.
18      Increased stone damage also has been associated with the presence of sulfur-oxidizing bacteria
19      and fungi on stone surfaces (Garcia-Valles et al.,  1998; Young, 1996; Saiz-Jimenez, 1993;
20      Diakumaku et al., 1995).
21           Dissolution of gypsum  on the stone's surface initiates structural changes in the crust layer.
22      Garica-Valles et al. (1998) proposed a double mechanism: the  dissolution of the gypsum, in the
23      presence of sufficient moisture, followed by recrystallization inside fissures or pores. In the
24      event of limited moisture,  the gypsum is dissolved and  recrystallizes at its original location.
25      According to the authors, this would explain the gypsum-rich crustal materials on stone surfaces
26      sheltered from precipitation.
27           Moisture was found to be the dominant factor in  stone deterioration for several sandstones
28      (Petuskey et al.,  1995).  Dolske (1995) reported that the deteriorative effects of sulfur-containing
29      rain events, sulfates, and SO2 on marble were largely dependent on the shape of the monument
30      or structure rather than the type of marble. The author  attributed the increased fluid turbulence
31      over a non-flat vertical surface versus a flat surface to the increased erosion. Sulfur-containing

        June 2003                                4-201       DRAFT-DO NOT QUOTE OR CITE

-------
 1      particles also have been reported to enhance the reactivity of Carrara marble and Travertine and
 2      Trani stone to SO2 (Sabbioni et al., 1992). Particles with the highest carbon content had the
 3      lowest reactivity.
 4           The rate of stone deterioration is determined by the pollutant and the pollutant
 5      concentration, the stone's permeability and moisture content, and the pollutant deposition
 6      velocity.  Dry deposition of SO2 between rain events has also been reported to be a major
 7      causative factor in pollutant-related erosion of calcareous stones (Baedecker et al.,  1991; Dolske,
 8      1995; Cooke and Gibbs,  1994; Schuster et al., 1994; Hamilton et al., 1995; Webb et al., 1992).
 9      Sulfur dioxide deposition increases with increasing relative humidity (Spiker et al., 1992), but
10      the pollutant deposition velocity is dependent on the stone type (Wittenburg and Dannecker,
11      1992), the porosity of the stone, and the presence of hygroscopic contaminants.
12           Although it is clear from the available information that gaseous pollutants (dry deposition
13      of SO2 in particular) will promote the decay of some types  of stones under the specific
14      conditions, carboneous particles (noncarbonate carbon) may help to promote the decay process
15      by aiding in SO2 transformation to a more acidic species (Del Monte and Vittori, 1985).  Several
16      authors have reported enhanced sulfation of calcareous material by SO2 in the presence of
17      particles containing metal oxides (Sabbioni et al.,  1996; Hutchinson et al., 1992).
18
19      4.4.2   Soiling and Discoloration of Man-Made Surfaces
20           Ambient particles can cause soiling of man-made surfaces. Soiling has been defined as
21      the deposition of particles of less than 10 jim on surfaces by impingement.  Soiling generally is
22      considered an optical effect, that is, soiling changes the reflectance from opaque materials and
23      reduces the transmission of light through transparent materials. Soiling can represent a
24      significant detrimental effect requiring increased frequency of cleaning of glass windows and
25      concrete structures, washing and repainting of structures, and, in some cases, reduction in the
26      useful life of the object.  Particles, in particular carbon, also may help catalyze chemical
27      reactions that result in the deterioration of materials during exposure.
28           It is difficult to determine the accumulated particle levels that cause an increase in soiling.
29      Soiling is dependent on the particle concentration  in the ambient environment, particle size
30      distribution, the deposition rate, and the horizontal or vertical orientation and texture of the
31      surface being exposed (Haynie, 1986).  The chemical composition and morphology of the

        June 2003                                4-202        DRAFT-DO NOT QUOTE OR CITE

-------
 1      particles and the optical properties of the surface being soiled will determine the time at which
 2      soiling is perceived (Nazaroff and Cass, 1991). Carey (1959) reported that the average observer
 3      could observe a 0.2% surface coverage of black particles on a white background.  Work by
 4      Bellan et al. (2000) indicates that it would take a 12% surface coverage by black particles before
 5      there is 100% accuracy in identifying soiling.  Sharpies et al. (2001) studied the effect of air
 6      pollution, moisture, and the function of the room/building on glazing daylight transmittance for a
 7      number of building windows. They found that the direct soiling load to a window was
 8      dependent on the immediate external and internal environment.  For instance, there was only a
 9      10% reduction in daylight transmittance for windows from an office building that had not been
10      cleaned for 5 years compared to clean windows. The reduction in transmittance for windows in
11      a swimming pool complex was in excess of 20% due to soiling of the interior surface. For most
12      office buildings, there was a reduction of glazing transmittance ranging from 3 to  10%, with
13      most windows showing about a 3% reduction. The rate at which an object is soiled increases
14      linearly with time; however, as the soiling level increases, the rate of soiling decreases. The
15      buildup of particles on a horizontal surface is counterbalanced by an equal and opposite
16      depletion process. The depletion process is based on the scouring and washing effect of wind
17      and rain (Schwar, 1998).
18
19      4.4.2.1  Stones and Concrete
20           Most of the research evaluating the effects of air pollutants on stone structures has
21      concentrated on gaseous pollutants. The deposition of the sulfur-containing pollutants is
22      associated with the formation of gypsum on the stone (see Section 4.5.1.3). The dark color of
23      gypsum is attributed to soiling by carbonaceous particles from nearby combustion processes.
24      A lighter gray colored crust is attributed to soil dust and metal deposits (Ausset et al., 1998;
25      Camuffo, 1995; Moropoulou  et al., 1998).  Realini et al. (1995) recorded the formation of a dark
26      gypsum layer and a loss of luminous reflection in Carrara marble structures exposed for 1 year
27      under ambient air conditions.  Dark areas of gypsum were found by McGee and Mossitti (1992)
28      on limestone and marble specimens exposed under ambient air conditions for several years. The
29      black layers of gypsum were located in areas shielded from rainfall; whereas particles of dirt
30      were concentrated around the edges of the gypsum formations.  Lorusso et al. (1997) attributed
31      the need for frequent cleaning and restoration of historic monuments in Rome to exposure to

        June 2003                                 4-203       DRAFT-DO NOT QUOTE OR CITE

-------
 1      total suspended particulates. They also concluded that, based on a decrease in brightness
 2      (graying), surfaces are soiled proportionately over time; however, graying is higher on horizontal
 3      surfaces because of sedimented particles. Davidson et al. (2000) evaluated the effects of air
 4      pollution exposure on a limestone structure on the University of Pittsburgh campus using
 5      estimated average TSP levels in the 1930s and 1940s and actual values for the years 1957 to
 6      1997.  Monitored levels of SO2 were available for the years 1980 to 1998. Based on the
 7      available data concerning pollutant levels and photographs, it was thought that soiling began
 8      while the structure was under construction. With decreasing levels of pollution, the soiled areas
 9      have been slowly washed away, the process taking several decades, leaving a white, eroded
10      surface.
11
12      4.4.2.2  Household and Industrial Paints
13           Few studies are available that evaluate the soiling effects of particles on painted surfaces.
14      Particles composed of elemental carbon, tarry acids, and various other constituents are
15      responsible for soiling of structural painted surfaces. Coarse-mode particles (> 2.5 jim) initially
16      contribute more soiling of horizontal and vertical painted surfaces than do fine-mode particles (<
17      2.5 |im), but are more easily removed by rain (Haynie and Lemmons, 1990). The accumulation
18      of fine particles likely promotes remedial action (i.e., cleaning of the painted surfaces); whereas
19      coarse-mode particles are primarily responsible for soiling of horizontal surfaces.  Rain interacts
20      with coarse particles, dissolving the particle and leaving  stains on the painted surface (Creighton
21      et al., 1990; Haynie and Lemmons, 1990). Haynie and Lemmons (1990) proposed empirical
22      predictive equations for changes in surface reflectance of gloss-painted surfaces that were
23      exposed protected and unprotected from rain while oriented horizontally or vertically.
24           Early studies by Parker (1955) and Spence and Haynie (1972) demonstrated an association
25      between particle exposure and increased frequency of cleaning of painted surfaces. Particle
26      exposures also caused physical damage to the painted surface (Parker, 1955).  Unsheltered
27      painted surfaces are initially more soiled by particles than sheltered surfaces but the effect is
28      reduced by rain washing. Reflectivity is decreased more rapidly on glossy paint than on flat
29      paint (Haynie and Lemmons, 1990).  However, surface chalking of the flat paint was reported
30      during the exposure. The chalking interfered with the reflectance measurements for particle
31      soiling.  Particle composition measurements that were taken during exposure of the painted

        June 2003                                4-204        DRAFT-DO NOT QUOTE OR CITE

-------
 1      surfaces indicated sulfates to be a large fraction of the fine mode and only a small fraction of the
 2      coarse mode. Although no direct measurements were taken, fine mode particles likely also
 3      contained large amounts of carbon and possibly nitrogen or hydrogen (Haynie and Lemmons,
 4      1990).
 5
 6
 7      4.5  ATMOSPHERIC PARTICULATE MATTER, CLIMATE CHANGE,
 8           AND EFFECTS ON SOLAR UVB RADIATION TRANSMISSION
 9           Atmospheric particles alter the amount of solar radiation transmitted through the Earth's
10      atmosphere. The absorption of solar radiation by atmospheric particles, together with trapping
11      of infrared radiation emitted by the Earth's surface by certain gases, enhances heating of the
12      Earth's surface and lower atmosphere (i.e., the widely-known "greenhouse effect") potentially
13      leading to "global warming" impacts on human health and the environment. Lesser impacts of
14      airborne particles include alterations in the amount of ultraviolet solar radiation (especially
15      UV-B) penetrating through the Earth's atmosphere and reaching its surface where UV radiation
16      can exert various effects on human health, plant and animal biota, and other environmental
17      components.
18           The effects of atmospheric PM on the transmission of electromagnetic radiation emitted
19      by the sun at ultraviolet and visible wavelengths and by the Earth at infrared wavelengths depend
20      on radiative properties (extinction efficiency, single scattering albedo, and asymmetry
21      parameter) of the particles, which depend, in turn, on the size and shape of the particles, the
22      composition of the particles, and the distribution of components within individual particles.
23      In general, the radiative properties of particles are size- and wavelength-dependent.  In addition,
24      the extinction cross-section tends to be at a maximum when the particle radius is similar to the
25      wavelength of the incident radiation. Thus, fine particles present mainly in the accumulation
26      mode would be expected to exert a greater influence on the transmission of electromagnetic
27      radiation than would coarse particles.  The composition of particles can be crudely summarized
28      in terms of the broad classes identified in Chapter 2 of this document. These include fine
29      particles consisting mainly of (a) nitrate, sulfate, mineral dust, elemental carbon, organic carbon
30      compounds (e.g., PAHs), and (b) metals derived from high temperature combustion or smelting
31      processes. The major sources  of these components are shown in Table 3-9 of Chapter 3 in this
32      document.
        June 2003                                4-205        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Knowledge of the effects of PM on the transfer of radiation in the visible and infrared
 2      spectral regions is needed for assessing relationships between particles and global warming
 3      processes and associated environmental and biological changes. Knowledge of the factors
 4      controlling the transfer of solar radiation in the ultraviolet spectral range is needed to assess
 5      potential the biological and environmental effects associated with exposure to UV-B radiation
 6      (290 to 315 nm). Climate change processes, their potential to affect human and environmental
 7      health, and their potential relationships to atmospheric PM are discussed below.  Subsequently,
 8      aspects related to solar ultraviolet radiation processes and effects are summarized and the roles
 9      of atmospheric PM are discussed.
10
11      4.5.1  Atmospheric Particle Interactions with  Solar and Terrestrial Radiation
12            Related to Climate Change
13      4.5.1.1   The Projected Impacts of Global Climate Change
14           The study of atmospheric processes involved in mediating global climate change and its
15      potential consequences for human health and global ecosystems is an area of active research.
16      The most thorough evaluation of current scientific understanding of climate change available at
17      this time is the Third Assessment Report (TAR) of the Intergovernmental Panel on Climate
18      Change (IPCC, 200la).  Earlier assessments include those conducted by the United Nations
19      Environment Program (UNEP, 1986), the World Meteorological Organization  (WHO 1988), the
20      U.S. Environmental Protection Agency (U.S. Environmental Protection Agency,  1987), and
21      others (e.g., Patz et al., 2000a,b). The reader is referred to these documents for a complete
22      discussion of climate change science.  An abbreviated list of the IPCC conclusions, to date, and a
23      short discussion of the potential impacts of climate change on human  health and welfare is
24      provided here to serve as the context for the discussion of the role of particulate matter in
25      climate.
26           The IPCC TAR (200la) notes that the increasing body of observations indicates that the
27      Earth is warming and that other climate changes are underway.  These observations include the
28      global surface temperature record assembled since the year 1860, the  satellite temperature record
29      begun in 1979, recorded changes in snow and ice cover since the 1950's, sea level measurements
30      taken throughout the 20th century, and sea surface temperature observations since the 1950's.
31      Other evidence includes a marked increase over the past 100 years in the frequency, intensity
32      and persistence of the zonal atmospheric circulation shifts known as the El Nino-Southern
        June 2003                               4-206      DRAFT-DO NOT  QUOTE OR CITE

-------
 1      Oscillation (ENSO). ENSO events occur when the tropical ocean has accumulated a large,
 2      localized mass of warm water which interrupts cold surface currents along South America,
 3      altering precipitation and temperature patterns in the tropics, sub-tropics and the mid-latitudes.
 4           Atmospheric concentrations of greenhouse gases (GHGs) which trap solar energy within
 5      the climate system, are continuing to increase due to human activities.  These activities will
 6      continue to result changes in the composition of the atmosphere throughout the 21st century. The
 7      IPCC TAR describes the scientific evidence that ties this increase in GHGs over the past 50
 8      years to human activities (IPCC 200Ib).
 9           The IPCC (1998, 2001b) reports also describe the results of general circulation model
10      (GCM) studies that indicate that human activities will alter the climate system in a manner that
11      will likely to lead to marked global and regional changes in temperature, precipitation  and other
12      climate properties.  This is expected to increase global mean sea level; increase the number of
13      extreme weather events including floods, and droughts; and induce changes in soil moisture.
14      These changes will directly impact human health, ecosystems, and global economic sectors, e.g.,
15      hydrology and water resources,  food and fiber production, etc., (IPCC 1998, 200Ib). Table 4-20
16      summarizes these projected impacts. Wide variations in the course and net impacts  of climate
17      change in different geographic areas can be expected.  In general, projected climate  change
18      impacts can be expected to represent additional stresses on those  natural ecosystems and human
19      societal systems already impacted by increasing resource demands, unsustainable resource
20      management practices, and pollution — with wide variation likely across regions and nations in
21      their ability to cope with consequent alterations in ecological balances, in availability of
22      adequate food, water, and clean air, and in human health and safety.  However, although many
23      regions are likely to experience  severe adverse impacts (some possibly irreversible) of climate
24      change, some climate change impacts may be locally beneficial in some regions. For example,
25      sectors or subregions may benefit from warmer temperatures or increased CO2 fertilization (e.g.,
26      west coast coniferous forests; some western rangelands; reduced  energy costs for heating in
27      northern latitudes; reduced road salting and snow-clearance costs; longer open-water seasons in
28      northern channels and ports; and agriculture in the northern latitudes, the interior West, and the
29      west coast).  The IPCC report, "The Regional Impacts of Climate Change" (IPCC, 1998),
30      describes the projected effects of human-induced climate change  on the different regions of the
31      globe, including Africa, the Arctic and Antarctic, the Middle East and arid Asia, Australasia,

        June 2003                                 4-207       DRAFT-DO NOT QUOTE OR CITE

-------
        TABLE 4-20.  EXAMPLES OF IMPACTS RESULTING FROM PROJECTED
                            CHANGES IN EXTREME CLIMATE EVENTS
  Projected changes during the 21st Century in
  Extreme Climate Phenomena and their Likelihood"
Representative Examples of Projected Impacts'1
(all high confidence of occurrence in some areas")
 Simple Extremes
  Higher maximum temperatures; more hot days and heat
  waves'1 over nearly all land areas (very likely")
  Increased incidence of death and serious illness in older age groups
  and urban poor
  Increased heat stress in livestock and wildlife
  Shift in tourist destinations
  Increased risk of damage to a number of crops
  Increased electric cooling demand and reduced energy supply
  reliability
  Higher (increasing) minimum temperatures; fewer cold
  days, frost days, and cold waves'1 over nearly all land
  areas (very likely")
• Decreased cold-related human morbidity and mortality
• Decreased risk of damage to a number of crops, and increased risk to
  others
• Extended range and activity of some pest and disease vectors
• Reduced heating energy demand
  More intense precipitation events (very likely" over many
  years)
  Increased flood, landslide, avalance, and mudslide damage
  Increased soil erosion
  Increased flood runoff could increase recharge of some floodplain
  aquifers
  Increased pressure on government and private flood insurance systems
  and disaster relief
  Complex Extremes
  Increased summer drying over most mid-latitude
  continental interiors and associated risk of drought
  (likely")
  Decreased crop yields
  Increased damage to building foundations caused by ground shrinkage
  Decreased water resource quantity and quality
  Increased risk of forest fire
  Increase in tropical cyclone peak wind intensities, mean
  and peak precipitation intensities (likely" over some
  areas)'
  Increased risk to human life, risk of infections, disease epidemics,
  and many other risks
  Increased coastal erosion and damage to coastal buildings and
  infrastructure
  Increased damage to coastal ecosystems such as coral reefs and
  mangroves
  Intensified droughts and floods associated with El Nino
  events in many different regions (likely") (see also under
  droughts and intense precipitation events)
  Decreased agricultural and rangeland productivity in drought- and
  flood-prone regions
  Decreased hydro-power potential in drought-prone regions
  Increased Asian summer monsoon precipitation
  variability (likely")
  Increased flood and drought magnitude and damages in temperate
  and tropical Asia
  Increased intensity of mid-latitude storms
  (little agreement between current models)11
  Increased risks to human life and health
  Increased property and infrastructure losses
  Increased damage to coastal ecosystems
  "Likelihood refers to judgmental estimates of confidence used by TAR WGI: very likely (90-99% chance); likely (66-90% chance).
  Unless otherwise stated, information on climate phenomena is taken from the Summary for Policymakers, TAR WGI.  TAR
  WGI = Third Assessment Report of Working Group 1 (IPCC, 2001a).
  bThese impacts can be lessened by appropriate response measures.
  'High confidence refers to probabilities between 67 and 95%.
  ''information from TAR WGI, Technical Summary.
  'Changes in regional distribution of tropical cyclones are possible but have not been established.

  Source: IPCC (2001b).
June 2003
  4-208
DRAFT-DO NOT QUOTE OR CITE

-------
 1      Europe, Latin America, North America, the small island nations, temperate Asia, and tropical
 2      Asia. For further details about the projected effects of climate change on a US-regional scale,
 3      the reader is also referred to several regionally-focused reports (MARAT, 2000; Yarnal et al.,
 4      2000; NERAG, 2001; GLRAG, 2000), as well as a report on potential impacts upon human
 5      health due to climate change (Bernard et al., 2001).
 6           It must be borne in mind that while current climate models are successful in simulating
 7      present annual mean climate and the seasonal cycle on continental scales, they are less
 8      successful at regional scales. Clouds and humidity, essential factors in defining local and
 9      regional ("sub-grid") climate, are significantly uncertain (IPCC 200la).  Due to modeling
10      uncertainties, both in reproducing regional climate and in predicting future economic activity,
11      the projected impacts discussed above are also uncertain.
12           Findings from the IPCC TAR  (2001a), Regional Impacts Assessment (1998) and other
13      regional assessments cited above illustrate well the considerable uncertainties and difficulties in
14      projecting likely climate change impacts on regional or local scales. The findings also reflect the
15      mixed nature of projected potential climate change impacts (combinations of mostly deleterious,
16      but other possible beneficial effects) for U.S. regions and their variation across the different
17      regions. Difficulties in projecting region-specific climate change impacts are complicated by the
18      need to evaluate potential effects of local- or regional-scale changes in key air pollutants not
19      only on global scale temperature trends but also in terms of potentially more local- or regional -
20      scale impacts on temperature and precipitation patterns.
21
22      4.5.1.2  Airborne  Particle Relationships to Global Warming and Climate Change
23           Atmospheric particles both scatter and absorb incoming solar radiation.  Visibility
24      reduction is caused by  particle scattering in all directions; whereas climate effects result mainly
25      from scattering in the upward direction. Upward scattering of solar radiation reduces the total
26      amount of energy received by the Earth system,  leading to surface cooling. The effect on
27      climate due to upward  scattering and to absorption of radiation by aerosol can be roughly
28      quantified as a "radiative forcing" (Houghton et al., 1990).  Global and regional climate (at
29      equilibrium) is defined by the balance between a large number of "positive" and "negative"
30      forcings induced by different components of the Earth system. The Earth system responds to
31      these forcings in a potentially complex way due to feedback mechanisms that are theorized but

        June 2003                                4-209       DRAFT-DO NOT QUOTE OR CITE

-------
 1      very difficult to define.  In the absence of information about climate feedbacks, radiative forcing
 2      values for the many components of the climate system are estimated as a tool for comparing, to
 3      first order, their relative importance in climate change. Forcing estimates for various classes of
 4      atmospheric particles are derived on the basis of climate modeling studies and reported by the
 5      IPCC (IPCC 200Ib).
 6          Particulate matter appears to play a significant role in defining climate on both a global and
 7      regional scale. Haywood et al. (1999) found that the inclusion of anthropogenic aerosols results
 8      in a significant improvement between calculations of reflected sunlight at the top of the
 9      atmosphere and satellite observations in oceanic regions close to sources of anthropogenic PM.
10      Significant reductions over the past 50 years in solar radiation received at the Earth's surface on
11      a globally averaged basis correlate with increases in atmospheric aerosol (Stanhill and Cohen,
12      2001).  While this correlation seems clear, quantifying the cooling and warming effects of
13      aerosol in relation to greenhouse gas-related warming is difficult. Aerosol complicates the
14      interpretation of climate change due to its spatial  and temporal inhomogeneities and uncertain
15      radiative properties.  However, to first order, inclusion of modeled atmospheric sulfate
16      concentrations substantially improved the agreement between modeled and observed surface
17      temperatures (Kiehl and Briegleb, 1993).  On a regional scale, the suspected influence of
18      aerosols upon climate relates to regional hydrological cycles. Evidence is accumulating that
19      pollution aerosols reduce precipitation frequency by clouds, potentially leading to drought in
20      some parts of the world (Ramanathan et al, 2001).
21
22      Greenhouse Gases, Particulate Matter and the Earth's Radiative Equilibrium
23          According to simple radiative transfer theory, at thermal equilibrium, the Earth's surface
24      should be near -15 °C. This is the temperature of a theoretical "black body" that is receiving and
25      then re-emitting 342.5 Wm"2 (i.e., the globally averaged amount of solar radiation absorbed and
26      then re-emitted by the Earth as infrared terrestrial radiation).  In fact, satellite observations
27      indicate that the Earth's average planetary temperature is remarkably close to the theoretical
28      black body value at -18 °C, a temperature at which liquid water ordinarily does not  exist.
29          At its surface, however, the Earth's average temperature is + 15 °C. The 33 °C
30      temperature differential between Earth's planetary temperature and its surface temperature is due
31      to the existence of infrared radiation-absorbing components in the atmosphere, i.e. GHGs,

        June 2003                                4-210       DRAFT-DO NOT QUOTE OR CITE

-------
 1      including carbon dioxide, methane, several other trace gases and some types of particles and
 2      clouds. The phenomenon of planetary surface warming due to atmospheric absorption and re-
 3      emission of infrared radiation is popularly known as the "greenhouse effect" (Arrhenius, 1896;
 4      Schneider, 1992). Radiation trapped by the Earth's atmosphere is reflected back to its surface,
 5      with some small fraction absorbed by dark atmospheric particles.  The fraction of this radiation
 6      that is not directly re-emitted as long-wave terrestrial radiation transforms into heat energy that
 7      drives the atmospheric processes that form the basis of weather and climate. Eventually this
 8      energy is transformed a second time, to terrestrial radiation, and is re-emitted as part of the
 9      process that maintains Earth's radiative equilibrium.
10           Satellite measurements of the Earth's radiation budget suggest that the Earth is, however,
11      out of equilibrium. These measurements indicate that from 1985 through 1990,  the Earth
12      absorbed 238 W/m2 and emitted 235  Wm"2 - an imbalance of 3 Wm"2.  If these measurements are
13      correct, they suggest that continuously changing atmospheric concentrations of radiatively active
14      species along with other alterations to the climate system due to human activities may be
15      responsible for preventing the  re-establishment of the Earth's radiative equilibrium.
16           Radiatively active gases  in the atmosphere are largely responsible for the greenhouse
17      effect, although some light absorbing particles and clouds contribute to atmospheric and surface
18      warming (IPCC 200la).  The majority of clouds and particles play a role in counteracting the
19      greenhouse effect by increasing the degree to which the Earth is able to reflect solar radiation,
20      i.e., its "albedo."   Successful modeling of the Earth's climate  and, therefore, assessment of the
21      degree of human-induced climate change and development of appropriate policy depends on the
22      high quality information on the relative efficiencies, amounts, spatial and temporal distributions
23      of the various radiatively active components of the atmosphere at absorbing and/or reflecting
24      solar and terrestrial radiation.
25
26      "Forcing" and the Earth's Radiative Balance
27           A measure of the relative influence of a given component of the climate system on the
28      Earth's radiative balance is its "radiative forcing." Radiative forcing, in Wm"2, is a quantity that
29      was developed by the climate modeling community as a first order-only means of estimating
30      relative effects of anthropogenic and natural processes on the surface-troposphere  system.  No
31      more precise metric has yet been found to  replace radiative forcing as a measure of impact of

        June 2003                                 4-211        DRAFT-DO NOT QUOTE OR CITE

-------
 1      upon climate (IPCC, 2001). The convention for this quantity assigns a negative forcing to
 2      climate system components that reflect solar radiation back into space and positive forcing for
 3      those which enhance the greenhouse effect, or otherwise act to enhance the heating absorbing
 4      capacity of the Earth system. Purely reflective atmospheric aerosol, snow-covered land surfaces
 5      and dense sea ice provide a negative forcing, while highly absorbing atmospheric aerosol,
 6      greenhouse gases and dark ocean surfaces positively force the climate system.
 7           The IPCC reports estimated values for forcing by the individual radiatively active gas and
 8      particle-phase components of the atmosphere.  These estimates are derived primarily through
 9      expert judgment incorporating the results of peer-reviewed modeling studies. Uncertainty
10      ranges are assigned that reflect the range of modeled values reported in these studies. According
11      to the available research on climate forcing by aerosol, the panel has provided estimates for
12      sulfate, organic, black carbon, biomass burning, and mineral dust aerosol. The current estimate
13      of forcing due to long-lived, well-mixed, greenhouse gases accumulated in the atmosphere from
14      the pre-industrial era (ca., 1750) through the year 2000 is + 2.4 Wm"2 (IPCC, 2001). In contrast,
15      forcing due to a sulfate  aero sol-related increase in planetary albedo has been assigned a value of
16      - 0.4 Wm"2. Biomass burning and fossil-fuel-related organic aerosol also increase the Earth's
17      reflectivity  and are estimated to contribute a -0.2 Wm"2 and -0.1 Wm"2 forcing, respectively.
18      Fossil-fuel black carbon is expected to warm the atmosphere, resulting in an estimated
19      + 0.2 Wm"2 forcing. No estimate for forcing by nitrate aerosol has been proposed due to wide
20      discrepancies in current global modeling results and the difficulties associated with obtaining
21      accurate ambient samples of nitrate concentrations and size distributions. Likewise, no specific
22      estimate has been offered for forcing by mineral  dust aerosol introduced into the atmosphere due
23      to human activities beyond the assignment of a tentative range of - 0.6 to + 0.4 Wm"2.  The
24      estimated forcing and associated uncertainty for each aerosol type is shown in relation to forcing
25      estimates for the known greenhouse gases along with an indication of the level of confidence in
26      each of these estimates  in Figure 4-42.
27           The relationship between perturbations to the Earth's radiative balance and climate is
28      complicated by various feedbacks within the climate system. An example would be the positive
29      feedback associated with melting sea ice. As sea ice melts with increasing surface temperatures,
30      the dark ocean surface is revealed which absorbs, rather than reflects, solar radiation. Such a
31      feedback increases the rate of surface warming.  The role of feedbacks  in determining the

        June 2003                                 4-212        DRAFT-DO NOT QUOTE OR CITE

-------

E

l_ O)
^ E
O" >=
w 5
0) 1~
Q.
c/)
s
> 0
c
o g.
£ ° -1-
0) O
"ro
CD
ri
1







c
Halocarbons
N,O

CH4 (—
C02
Tropospheric
ozone
rh



Aerosols

Black
carbon from




"i

f'Jgl' Mineral Aviation-induced
burning Dust
r^i
T~^ |T|
stratospheric Organic 1 1
ozone 1 carbon Biomass
Sulphate ™m burning



fuel
burning

s^ Solar
Contrails cirrus i-T-i
- - L|J
Land-
Aerosol use
ndirect (albedo)
effect only
1

                         High Medium Medium Low   Very
                                               Low
     Very Very  Very  Very   Very Very Very
     Low Low  Low  Low   Low Low  Low
                                         Level of Scientific Understanding
       Figure 4-42.  Estimated global mean radiative forcing exerted by gas and various particle
                     phase species for the year 2000, relative to 1750.
       Source: IPCC(2001a).
 1      sensitivity of climate to changes in radiative forcing is described in detail by the IPCC in its third
 2      assessment report (IPCC, 200la).
 3           One possible feedback of interest in the discussion of the role of aerosols in determining
 4      climate may result from the possible sensitivity of aerosol number and mass to atmospheric
 5      temperature (Hemming and Seinfeld, 2001). Increasing atmospheric temperatures may result in
 6      a reduction of aerosol as semivolatile organic and inorganic aerosol constituents evaporate,
 7      leading to a change in aerosol forcing. As described in Chapter 2 of this document, and further
 8      discussed below, ambient aerosols are known to contain complex chemical mixtures of both
 9      scattering and absorbing materials. The feedback that may result from this phenomenon will
10      depend upon whether the aerosols become more absorbing or more reflective upon the loss of
11      semivolatile material. Research is presently underway to evaluate both the role of temperature in
12      determining aerosol mass and in defining the link between air quality and climate, but no
13      literature presently exists to support an assessment of these effects. The following discussion,
       June 2003
4-213
DRAFT-DO NOT QUOTE OR CITE

-------
 1      therefore, will focus solely on the relative forcing properties of aerosols, for which a body of
 2      scientific research is available for consideration.
 3           The physical and chemical properties of atmospheric aerosols, and their regional
 4      distribution and temporal nature, all play a role in determining the degree to which they force
 5      climate. These details are described, below, followed by a description of the "indirect" effect of
 6      aerosols due to changes in cloud properties; and brief discussions of the sources of uncertainty in
 7      determining aerosol-related climate forcing and of the link to human health and the environment
 8      then follow.
 9
10      The Physics of PM and Its Climate Effects
11           Four wavelength-dependent physical properties of ambient aerosols are needed to calculate
12      optical depth and thus their radiative properties: the mass light-scattering efficiency (asp), the
13      functional dependence of light-scattering on relative humidity f(RH), the single-scattering albedo
14      (o)0), and the scattering asymmetry  parameter (g); (Charlson et al., 1992; Penner et al., 1994a).
15      Direct forcing by aerosols is especially sensitive to single scattering albedo. Depending on the
16      color of the underlying surface, small changes in co0 can change the sign of the calculated aerosol
17      forcing (Hansen  et al., 1997).
18           The wavelength-dependent phase function and scattering and absorption coefficients are
19      calculated using  Mie scattering theory. These values are then used to calculate mass light-
20      scattering efficiency, single-scattering albedo, and the asymmetry factor.  Mie calculations
21      require the ratio of particle size versus wavelength and the complex refractive index of the
22      particle - a composition-dependent property (Salby,  1996).  The relationship between light-
23      scattering and relative humidity likewise depends upon composition as water absorption depends
24      upon the presence of hygroscopic material within the particle. Therefore, according to the
25      current understanding of aerosol optics, good information about composition and size
26      distribution is needed to successfully predict aerosol-related forcing.
27
28      The Chemistry  of PM and Its Climate Effects
29           Although forcing estimates are reported for specific aerosol classes (i.e.,  sulfate, black
30      carbon, dust, etc.) it is understood that shortly after emission, primary aerosols undergo chemical
31      transformation in the atmosphere. These transformations occur through partitioning of gas phase

        June 2003                                 4-214        DRAFT-DO NOT QUOTE OR CITE

-------
 1      compounds, by coagulation with other aerosol or a combination of both processes (Pandis and
 2      Seinfeld, 1998). While the radiative properties of non-absorbing aerosol, such as sulfates or
 3      nitrates, are primarily sensitive to particle size, the radiative properties of aerosols containing
 4      absorbing constituents (i.e., black carbon and mineral aerosols) are also sensitive to chemical
 5      composition and mixing state. Studies of refractive index changes as a function of composition
 6      have shown that the type of mixing present both in the population of aerosols (i.e.,  "internal"
 7      versus "external" mixtures) and the extent of mixing within individual aerosols influence aerosol
 8      optical properties when absorbing material is present (Fuller et al., 1999).  Modeled estimates for
 9      radiative forcing by black-carbon-containing aerosols do, in fact, range widely depending on
10      whether the aerosol population is assumed to be internally or externally mixed and whether the
11      absorbing carbon is uniformly distributed in the particle or whether it exists as a core surrounded
12      by non-absorbing material. The IPCC Third Assessment report (2001) provides a tabulation of
13      studies and their forcing estimates for black carbon existing in different mixing states. For
14      example, Haywood and Shine (1997)  calculated that externally mixed fossil-fuel black carbon
15      forces climate by + 0.2 W/m2. Jacobson (2000) assumed that if black carbon exists as a solid
16      core contained within an otherwise non-absorbing droplet, the global forcing will be + 0.54
17      W/m2.
18           Chemically and physically detailed models at high spatial resolution are required to
19      accurately represent the chemical transformation and size evolution of aerosol within the
20      ambient atmosphere.  Such models, however, are impractical for global scale climate modeling.
21      Parameterizations are formulated to represent processes occurring at spatial and temporal scales
22      that are too fine for climate models (i.e., "sub-grid" processes). Several important fundamental
23      chemical and physical processes, however, are not yet well-enough defined to assure reasonable
24      parameterization in large scale climate models.  Prediction of the organic and black carbon
25      content of ambient  aerosols and their associated radiative properties, as described above, and the
26      aerosol-induced changes in cloud properties, remain especially weak.
27           Modeling of the effect of black carbon aerosol on climate, to date, has been done on the
28      basis of limited and poor quality data  regarding total global emissions, aerosol composition, and
29      the mixing state of  ambient aerosols.  Appendix 2B.2 of Chapter 2 in this document describes
30      the many problems associated with the most commonly used organic carbon/elemental carbon
31      measurement method (thermal optical reflectance and transmittance). For example, elemental

        June 2003                                4-215        DRAFT-DO NOT QUOTE OR CITE

-------
 1      carbon comprises a relatively small fraction of total carbon in biomass burning-related aerosol
 2      which is, nevertheless, an important source of black carbon in the atmosphere due to the
 3      extensive use of biofuels in developing countries and to the prevalence of natural and
 4      agriculture-related open biomass burning (Ludwig et al., 2003).  The concentrations of elemental
 5      carbon in these aerosols are close to the detection limits of thermal optical measurement
 6      systems, adding a high degree of uncertainty to the reported values. Thermal optical
 7      measurements also do not provide information about the specific chemical composition of
 8      carbonaceous aerosol that is needed to calculate refractive indices. Furthermore, the technique is
 9      based on filter collection, which only provides an average concentration of organic and
10      elemental carbon and no information about mixing states.
11           The magnitude of the errors introduced into estimates of black carbon-related radiative
12      forcing due to the lack of detailed information about optical properties of ambient black carbon
13      aerosol cannot, yet, be estimated. The forcing estimate of + 0.2 Wm"2 provided by the IPCC is
14      based upon a summary of the available modeling estimates,  some of which are founded on
15      simplistic assumptions about the composition and mixing state of ambient BC aerosol.
16           New methods, such as photo-acoustic spectroscopy, hold promise for detecting quantities
17      of light absorbing materials in streams of ambient aerosol (Arnott et al., 1999; Moosmuller et al.,
18      2001). The photo-acoustic method is capable of identifying the absolute fraction of a given
19      ambient aerosol sample that absorbs radiation at a given wavelength at high temporal resolution.
20      The method does not alter the composition of the aerosol  sample through heating and oxidation,
21      thus eliminating chemical artifacts.  Present photo-acoustic instruments detect absorption at
22      wavelengths  of interest for visibility studies, but the measurement principle can be adapted to
23      climate studies through the selection of climate-forcing relevant wavelengths.  Specific chemical
24      speciation  at  the organic  and inorganic compound level coupled with photo-acoustic
25      measurements of the light absorbing properties of ambient aerosol would provide the best
26      available test of atmospheric chemistry models attempting to simulate the evolution in aerosol
27      radiative properties between emission and atmospheric removal.
28
29      The Regional and Temporal Dependence of Aerosol Forcing
30           In addition to information about the composition and size distributions of ambient aerosol,
31      details regarding atmospheric concentrations on a spatial and temporal basis are needed for

        June 2003                                4-216       DRAFT-DO NOT QUOTE OR CITE

-------
 1      estimating climate forcing. Unlike the long-lived greenhouse gases, aerosols have lifetimes
 2      averaging only a week or less, leading to inhomogeneity in regional and global scale
 3      concentrations.  Industrial processes and other human activities that produce air pollution
 4      aerosols also vary on a seasonal, monthly, weekly and diurnal basis, introducing other
 5      complications into forcing estimates (IPCC, 2001). Peak aerosol concentrations, along with the
 6      greatest temporal variation, exist near emissions sources.
 7           Deviations from global mean forcing estimates can be very large on the regional scale.
 8      For instance, Tegen et al. (1996) found that local radiative forcing exerted by dust raised from
 9      disturbed lands ranges from -  2.1 W m"2 to 5.5 W m"2 over desert areas and their adjacent seas.
10      The largest regional values of radiative forcing caused by anthropogenic sulfate are about
11      -3 W m"2 in the eastern United States, south central Europe, and eastern China (Kiehl and
12      Briegleb, 1993).  These regional maxima in aerosol forcing are at least a factor of 10 greater than
13      their global mean values shown in Figure 4-41. By comparison, regional maxima in forcing by
14      the well-mixed greenhouse gases are only about 50% greater than their global mean value (Kiehl
15      and Briegleb, 1993). Thus, the estimates of local radiative forcing by particles also are large
16      enough to completely cancel the effects of greenhouse gases in many regions and to cause a
17      number of changes in the dynamic structure of the atmosphere that still need to be evaluated.
18      A number of anthropogenic pollutants whose distributions are highly variable are also effective
19      greenhouse absorbers.  These gases include O3 and, possibly, HNO3, C2H4, NH3, and SO2, all of
20      which are not commonly considered in radiative forcing calculations (Wang  et al., 1976). High
21      ozone values are found downwind of urban areas and areas where there is biomass burning.
22      However, Van Borland et al. (1997) found that there may not be much cancellation between the
23      radiative effects for ozone and for sulfate because both species have different seasonal cycles
24      and show significant differences in their spatial distribution.
25
26      "Indirect" Effects of PM on  Climate
27           Aerosols directly affect climate by scattering and absorbing solar and terrestrial radiation.
28      Depending on chemical composition, they can also nucleate new cloud droplets. For a given
29      total  liquid water content (LWC), increasing cloud droplet number means smaller droplets that
30      both  scatter solar radiation more effectively, reduce the  amount of precipitation from the cloud,
31      and consequently increase cloud lifetime - and the cloud's ability to scatter solar radiation.

        June 2003                                4-217        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Observational evidence exists of this impact of hygroscopic aerosol on both warm and ice cloud
 2      properties.  Increased albedo and increased cloud lifetime effects are treated separately by the
 3      climate modeling community.  Particle-induced increases in cloud albedo are referred to as the
 4      "first" indirect effect, and changes to the cloud lifetime due to reduced precipitation are referred
 5      to as the "second" indirect effect.  The highly uncertain estimated forcing due to the effects of
 6      particles on cloud albedo is given as a range from 0 to + 2.0 W m"2 (IPCC, 200la). A further
 7      effect, referred to as the "semi-indirect effect" is the reduction in cloud reflectivity due to the
 8      inclusion of black carbon-containing aerosol within the cloud drops themselves or as interstitial
 9      aerosol (Hansen et al., 2000).
10           Sulfate aerosols, especially those larger than 50 nm in diameter, are believed to alter clouds
11      to the largest extent due to their efficient nucleation of cloud drops and ice crystals (Twomey,
12      1974).  Organic aerosols that contain highly oxidized carbon compounds may be similarly
13      efficient in nucleating cloud droplets (Novakov and Penner, 1993).  Both satellite and in-situ
14      aircraft observations  reinforce the hypothesis that pollutant aerosols increase cloud reflectivity
15      and lifetime (Ramanathan et al., 2001).
16           An important consequence of this property of aerosols on regional climate includes
17      suppression of rain over polluted regions. Satellite observations show that precipitation occurs
18      only outside of pollution tracks, while clouds within pollution tracks show a reduction in
19      effective cloud drop radius to below the precipitation threshold (D. Rosenfeld, 1999).  Desert
20      dust also appears to alter the microphysical properties of clouds, suppressing precipitation from
21      warm clouds while nucleating ice crystals in cold clouds.
22           While climate models are not yet equipped for modeling the effect of aerosols on regional
23      and global hydrological cycles, it has been proposed that aerosols will reduce precipitation
24      efficiency over land where anthropogenic activities provide a major source of cloud
25      condensation  nuclei.  Several studies using coupled ocean-atmosphere general circulation
26      models support the possibility of a "spin down" effect upon hydrological cycles due only to a
27      reduction in surface radiation receipts from  sulfate aerosol scattering. When  indirect effects  are
28      included, the reduction in precipitation rates from clouds was large enough to reverse the effect
29      of greenhouse gas-related forcing (Ramanathan et al., 2001).
30
31

        June 2003                                 4-218        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Sources of Uncertainty in Aerosol Forcing Estimates
 2           Uncertainties in calculating the direct effect of airborne particles arise from a lack of
 3      knowledge of their vertical and horizontal variability, their size distribution, chemical
 4      composition, and the distribution of components within individual particles. For instance,
 5      gas-phase sulfur species may be oxidized to form a layer of sulfate around existing particles in
 6      continental environments or they may be incorporated in sea-salt particles (e.g., Li-Jones and
 7      Prospero,  1998). In either case, the radiative effects of a given mass of the sulfate will be much
 8      lower than if pure sulfate particles were formed. It also must be stressed that the overall
 9      radiative effect of particles at a given location is not simply determined by the sum of effects
10      caused by individual classes of particles because of radiative interactions between particles with
11      different radiative characteristics and with gases.
12           Calculations of the indirect effects of particles on climate are subject to much larger
13      uncertainties than are calculations of their direct effects, reflecting uncertainties in a large
14      number of chemical and microphysical processes that determine aerosol chemistry, size
15      distribution, and the number of droplets within a cloud.  A complete assessment of the radiative
16      effects of PM will require computationally intensive calculations that incorporate the spatial and
17      temporal behavior of particles of varying composition that have been emitted from (or formed
18      from precursors emitted from) different sources. Refining values of model input parameters
19      (such as improving emissions estimates) may be as important as improving the models per se in
20      calculations of direct radiative forcing (Pan et al., 1997) and indirect radiative forcing (Pan et al.,
21      1998). However, uncertainties associated with the calculation of radiative effects of particles
22      will likely remain much larger than those associated with well-mixed greenhouse gases.
23
24      Aerosol-Related Climate Effects, Human Health and the Environment
25           Given the present difficulty in accurately modeling aerosol physical, chemical and
26      temporal properties, its regionally-dependent atmospheric concentration levels, combined with
27      difficulties in projecting location-specific increases or decreases in anthropogenic  emissions of
28      atmospheric particles (or their precursors), the specific impacts on human health and the
29      environment due to aerosol effects on the climate system can not be  calculated with confidence.
30      However,  substantial qualitative information, from observation and modeling, indicates that
31      aerosol forces climate both positively and negatively, both globally and regionally, and may be

        June 2003                                4-219       DRAFT-DO NOT QUOTE OR CITE

-------
 1      negatively impacting hydrological cycles on a regional scale.  The global and other regional
 2      scale impacts are suspected, on the basis of current, though uncertain, modeling studies suggest
 3      that climate change in general can have positive and negative effects on human health, human
 4      welfare and the environment.
 5
 6      4.5.2   Atmospheric Particulate Matter Effects on  the Transmission of Solar
 7             Ultraviolet Radiation Transmission:  Impacts on Human Health and
 8             the Environment
 9      4.5.2.1  Potential Effects of Increased Ultraviolet Radiation Transmission
10          The transmission of solar UV-B radiation through the Earth's atmosphere is controlled by
11      ozone, clouds, and particles.  The depletion of stratospheric ozone caused by the release of
12      anthropogenically produced chlorine (Cl)- and bromine (Br)-containing compounds has resulted
13      in heightened concern about potentially serious increases in the amount of solar UV-B radiation
14      (SUVB) reaching the Earth's surface. SUVB is also responsible for initiating the production of
15      OH radicals that oxidize a wide variety of volatile organic compounds, some of which can
16      deplete stratospheric ozone (e.g., CH3C1, CH3Br), absorb terrestrial infrared radiation (e.g., CH4),
17      and contribute to photochemical smog formation (e.g., C2H4,  C5H8).
18          Increased penetration of SUVB to the Earth's surface as the result of stratospheric ozone
19      depletion  continues to be of much concern because of projections of consequent increased
20      surface-level SUVB exposure and associated potential negative impacts on human health, plant
21      and animal biota, and man-made materials. Several summary overviews (Kripke, 1989; Grant,
22      1989; Kodama and Lee, 1994; Van der Leun et al., 1995, 1998) of salient points related to
23      stratospheric ozone depletion and bases for concern provide a concise introduction to the subject.
24      Only a brief summary will be given here. Stratophospheric ozone depletion results from
25      (a) anthropogenic emissions of certain trace gases having long atmospheric residence times, e.g.,
26      chlorofluorocarbons (CFCs), carbon tetrachloride (CC14), and Halon 1211 (CF2C1 Br) and 1301
27      (CF3Br) — which have atmospheric residence times of 75 to 100 years, 50 years, 25 years,  and
28      110 years, respectively; (b) their tropospheric accumulation and gradual transport, over decades,
29      up to the stratosphere, where (c) they photolyze to release Cl and Br that catalyze ozone
30      destruction; leading to (d) stratospheric ozone depletion.  Such ozone depletion is most marked
31      over Antarctica during spring in the Southern Hemisphere, to a less marked but still significant
32      extent over the Arctic Polar Region during late winter and spring in the Northern Hemisphere,

        June 2003                               4-220       DRAFT-DO NOT QUOTE OR CITE

-------
 1      and to a lesser extent, over mid-latitude regions during any season.  Given the long time frame
 2      involved in transport of such gases to the stratosphere and their long residence times there, any
 3      effects already seen on stratospheric ozone are likely caused by the atmospheric loadings of trace
 4      gases from anthropogenic emissions over the past few decades. Ozone-depleting gases already
 5      in the atmosphere will continue to affect stratospheric ozone concentrations well into the 21st
 6      century. Shorter-lived gases, such as CH3Br, also exert significant ozone depletion effects.
 7          The main types of deleterious effects hypothesized as likely to result from stratospheric
 8      ozone  depletion and consequent increased SUVB penetration through the Earth's atmosphere
 9      include the following:
10         (1)  Direct Human Health Effects, such as skin damage (sunburn), leading to more rapid
                aging and increased incidence of skin cancer; ocular effects  (retinal damage and
                increased cataract formation possibly leading to blindness);  and suppression of some
                immune system components (contributing to skin cancer induction and spread to
                nonirradiated skin  areas, as well as possibly increasing susceptibility to certain
                infectious diseases).
11         (2)  Agricultural/Ecological Effects, mediated largely through altered biogeochemical
                cycling resulting in consequent damaging impacts on terrestrial plants (leading to
                possible reduced yields of rice, other food crops, and commercially important trees, as
                well as to biodiversity shifts in natural terrestrial ecosystems); and deleterious effects
                on aquatic  life (including reduced ocean zooplankton and phytoplankton, as important
                base components of marine food-chains supporting the existence of commercially
                important,  edible fish and other seafood, as well as to other aquatic ecosystem shifts).
12         (3)  Indirect Human Health and Ecological Effects, mediated through increased
                tropospheric ozone formation (and consequent exacerbation of surface-level, ozone-
                related health and ecological impacts) and alterations in the  concentrations of other
                important trace species,  most notably the  hydroxyl radical and acidic aerosols.
13         (4)  Other  Types of Effects, such as faster rates of polymer weathering because of
                increased UV-B radiation and other effects  on man-made commercial materials and
                cultural artifacts, secondary to climate change or exacerbation of air pollution
                problems.
        June 2003                                4-221        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Extensive qualitative and quantitative characterizations of stratospheric ozone depletion
 2      processes and projections of their likely potential impacts on human health and the environment
 3      have been the subjects of periodic (1988,  1989, 1991, 1994, 1998) international assessments
 4      carried out under WMO and UNEP auspices since the 1987 signing of the Montreal Protocol on
 5      Substances that Deplete the Ozone Layer. For more detailed up-to-date information, the reader
 6      is referred to recent international assessments of (a) processes contributing to stratospheric ozone
 7      depletion and the status of progress towards ameliorating the problem (WMO, 1999) and
 8      (b) revised qualitative and quantitative projections of potential consequent human health and
 9      environmental effects (UNEP, 1998, 2000).
10           Of considerable importance is the growing recognition, as reflected in these newer
11      assessments, of impacts of enhanced solar radiation on biogeochemical cycles (see, for example,
12      Zepp et al.,  1998). As noted in that paper, the effects of UV-B radiation (both in  magnitude and
13      direction) on trace gas (e.g., CO) emissions and mineral nutrient cycling are species specific and
14      can affect a variety of processes.  These include, for example, changes in the chemical
15      composition of living plant tissue, photodegradation of dead plant matter (e.g., ground litter),
16      release of CO from vegetation previously charred by fire, changes in  microbial decomposer
17      communities, and effects on nitrogen-fixing microorganisms and plants. In addition, changes in
18      the amount and composition of organic matter, caused by enhanced UV-B penetration, affect the
19      transmission of solar ultraviolet and visible radiation through the water column. These changes
20      in light quality broadly impact the effects of UV-B on aquatic biogeochemical cycles. Enhanced
21      UV-B levels cause both positive and negative effects on microbial activities in aquatic
22      ecosystems that can  affect nutrient cycling and the uptake or release of greenhouse gases.  Thus,
23      there are emerging complex issues regarding interactions and feedbacks between climate change
24      and changes in terrestrial and marine biogeochemical cycles because  of increased UV-B
25      penetration to the Earth's surface.
26           In contrast to the above types of negative impacts projected as likely to be associated with
27      increased UV-B penetration to Earth's surface, some research results are suggestive of possible
28      beneficial effects of increased UV-B radiation. For example, a number of U.S. and international
29      studies have focused on the protective effects of UV-B radiation with regard to non-skin cancer
30      incidence. One of the first of these studies investigated potential relationships between  sunlight,
31      vitamin D, and colon cancer (Garland  and Garland, 1980).  More recent studies continue to

        June 2003                                 4-222        DRAFT-DO NOT QUOTE OR CITE

-------
 1      provide suggestive evidence that UV-B radiation may be protective against several types of
 2      cancer and some other diseases.  For example, Grant (2002) conducted a number of ecologic-
 3      type epidemiologic studies, which suggest that UV-B radiation, acting through the production of
 4      vitamin D, is a risk-reduction factor for mortality due to several types of cancer, including cancer
 5      of the breast, colon, ovary, and prostate, as well as non-Hodgkin lymphoma.  Other related
 6      studies that provide evidence for protective effects of UV-B radiation include Gorham et al.
 7      (1989); Gorham et al. (1990); Garland et al. (1990); Hanchette and Schwartz (1992); Ainsleigh
 8      (1993); Lefkowitz and Garland (1994); Hartge et al. (1996); and Freedman et al. (1997).
 9           As noted in the above detailed international assessments, since the signing of the Montreal
10      Protocol, much progress has been made in reducing emissions of ozone depleting gases, leading
11      to estimates that the maximum extent of stratospheric ozone depletion has likely leveled off
12      during recent years, and this is expected to be followed by gradual lessening of the problem and
13      its impacts during the next half-century. However, the assessments also note that the modeled
14      projections are subject to considerable uncertainty  (see, for example, UNEP, 2000). Varying
15      potential roles of atmospheric particles, discussed below, are among salient factors complicating
16      predictive modeling efforts.
17
18      4.5.2.2  Airborne Particle Effects on Atmospheric Transmission of Solar Ultraviolet
19              Radiation
20           A given amount of ozone in the lower troposphere has been shown to absorb more solar
21      radiation than an equal amount of ozone in the stratosphere because of the increase in its
22      effective optical path produced by Rayleigh scattering in the lower atmosphere (Briihl and
23      Crutzen, 1988).  The effects of particles are more complex. The impact of particles on the
24      SUVB flux throughout the boundary layer are highly sensitive to the altitude of the particles and
25      to their single scattering albedo.  Even the sign of the effect can reverse as the composition of the
26      particle mix changes from scattering to absorbing types (e.g.,  from sulfate to elemental carbon or
27      PAHs; Dickerson et al., 1997). In addition, scattering by particles also may increase the
28      effective optical path of absorbing molecules, such as ozone, in the lower atmosphere.
29           The effects of particles present in the lower troposphere on the transmission of SUVB have
30      been examined both by field measurements and by radiative transfer model calculations. The
31      presence of particles in urban areas modifies the spectral distribution of solar irradiance  at the
32      surface.  Shorter wavelength radiation (i.e., in the ultraviolet) is attenuated more than visible

        June 2003                                 4-223        DRAFT-DO NOT QUOTE OR CITE

-------
 1      radiation (e.g., Peterson et al., 1978; Jacobson, 1999). Wenny et al. (1998) also found greater
 2      attenuation of SUVB than SUVA (315 to 400 nm). However, this effect depends on the nature
 3      of the specific particles involved and, therefore, is expected to depend strongly on location.
 4      Lorente et al. (1994) observed an attenuation of SUVB ranging from 14 to 37%, for solar zenith
 5      angles ranging from about 30° to about 60°, in the total (direct and diffuse) SUVB reaching the
 6      surface in Barcelona during cloudless conditions on very polluted days (aerosol scattering
 7      optical depth at 500 nm, 0.46 ^ T500nm ^ 1.15) compared to days on which the turbidity of urban
 8      air was similar to that for rural  air (T500nm ^ 0.23).
 9           Particle concentrations that can account for these observations can be estimated roughly by
10      combining Koschmieder's relation for expressing visual range in terms of extinction coefficient
11      with one for expressing the mass of PM2 5 particles in terms of visual range (Stevens et al.,
12      1984). By assuming a scale height (i.e., the height at which the concentration of a substance
13      falls off to 1/e of its value at the surface) of 1 km for PM2 5, an upper limit of 30 jig/ m3 can be
14      derived for the clear case and between 60 and 150 |ig/m3 for the polluted case. Estupifian et al.
15      (1996) found that summertime  haze under clear sky conditions attenuates SUVB between 5 and
16      23% for a solar zenith angle of 34°compared to a clear sky day in autumn. Minis (1996)
17      measured a decrease in SUVB by about 80% downwind of major biomass burning areas in
18      Amazonia in 1995.  This decrease in transmission corresponded to optical depths at 340 nm
19      ranging from three to four. Justus and Murphey (1994) found that SUVB reaching the surface
20      decreased by about 10% because of changes in aerosol loading in Atlanta, GA, from 1980 to
21      1984.  In addition, higher particle levels in Germany (48 °N) may be responsible for greater
22      attenuation of SUVB than in New Zealand (Seckmeyer and McKenzie, 1992).
23           In a study of the effects of nonurban haze on  SUVB transmission, Wenny et al. (1998)
24      derived a very simple  regression relation between the measured aerosol optical depth at 312 nm,
25
26           ln( SUVB transmission at solar noon) = - 0.1422 T312nm - 0.138, R2 = 0.90,        (4-14)
27
28      and the transmission of SUVB  to the surface. In principle, values of T312nm could be found from
29      knowledge of the aerosol optical properties and visual range values.  Wenny et al. (1998) also
30      found that absorption by particles accounted for 7 to 25% of the total (scattering + absorption)
31      extinction. Relations such as the above one are strongly dependent on local conditions and

        June 2003                                4-224       DRAFT-DO NOT QUOTE OR CITE

-------
 1      should not be used in other areas without knowledge of the differences in aerosol properties.
 2      Although all of the above studies reinforce the idea that particles play a major role in modulating
 3      the attenuation of SUVB, none included measurements of ambient PM concentrations, so direct
 4      relations between PM levels and SUVB transmission could not be determined.
 5           Vuilleumier et al. (2001) concluded that variations in aerosol scattering and absorption
 6      were responsible for 97% of the variability in the optical depth measured at seven wavelengths
 7      from 300 to 360 nm at Riverside, CA, from 1 July to 1 November, 1997.  Similar measurements
 8      made at nearby Mt. Wilson, located above the main  surface haze layer, showed that 80% of the
 9      variations in optical depth were  still driven by variations in aerosol scattering and absorption.
10      The remainder of the variability in optical depth was attributed mainly to variability in ozone
11      under clear-sky conditions.  However, these  results cannot be extrapolated to other locations
12      because these effects are coupled and non-linear and are not straightforward.  They depend on
13      the concentrations of these species and on the physical and chemical characteristics of the
14      particles. Hence, any quantitative statements regarding the relative importance of particles and
15      ozone will be location-specific.
16           Liu et al. (1991) roughly estimated the overall  effects on atmospheric transmission of
17      SUVB of increases of anthropogenic airborne particles that have occurred since the beginning of
18      the industrial revolution.  Based on (a) estimates of the reduction in visibility from about 95 km
19      to about 20 km over nonurban areas in the eastern United States and in Europe, (b) calculations
20      of optical properties of airborne particles found in rural areas to extrapolate the increase in
21      extinction at 550 to 310 nm, and (c) radiative transfer model calculations, Liu et  al. concluded
22      that the amount of SUVB reaching Earth's surface likely has decreased from  5 to 18% since the
23      beginning of the industrial revolution.  This was attributed mainly to scattering of SUVB back to
24      space by sulfate-containing particles.  Radiative transfer model calculations have not been done
25      for urban particles.
26           Although aerosols are expected to decrease the flux of SUVB reaching the  surface,
27      scattering by particles is expected to result in an increase in the actinic flux within and above the
28      aerosol layer. However, when the particles significantly absorb SUVB, a decrease in the actinic
29      flux is expected.  Actinic flux is the radiant energy integrated over all directions  at a given
30      wavelength incident on a point in the atmosphere and is the quantity needed to calculate rates of
31      photolytic reactions in the atmosphere. Blackburn et al. (1992) measured attenuation of the

        June 2003                                 4-225        DRAFT-DO NOT QUOTE OR CITE

-------
 1      photolysis rate of ozone and found that aerosol optical depths near unity at 500 nm reduced
 2      ozone photolysis rate by as much as a factor of two. Dickerson et al. (1997) showed that the
 3      photolysis rate for NO2, a key parameter for calculating the overall intensity of photochemical
 4      activity, could be increased within and above a scattering aerosol layer extending from the
 5      surface although it would be decreased at the surface.  This effect is qualitatively similar to what
 6      is seen in clouds, where photolysis rates are increased in the upper layers of a cloud and above
 7      the cloud (Madronich, 1987). For a simulation of an ozone episode that occurred during July
 8      1995 in the Mid-Atlantic region, Dickerson et al. (1997) calculated ozone increases of up to
 9      20 ppb compared to cases that did not include the radiative effects of particles in urban airshed
10      model (UAM-IV) simulations.  In contrast, Jacobson (1998) found that particles may have
11      caused a 5 to 8% decrease in O3 levels during the Southern California Air Quality Study in 1987.
12      Absorption by organic compounds and nitrated inorganic compounds was hypothesized to
13      account for the reductions in UV radiation intensity.
14           The photolysis of ozone in the Hartley bands  also leads to production of electronically
15      excited oxygen atoms, O(JD) that then react with water vapor to form OH radicals. Thus,
16      enhanced photochemical production of ozone is accompanied by the scavenging of species
17      involved in greenhouse warming and stratospheric  depletion. However, these effects may be
18      neutralized or even reversed by the presence of absorbing material in the particles.  Any
19      evaluation of the effects of particles on photochemical activity therefore will depend on the
20      composition of the particles and will also be location-specific.
21           Further complicating any straightforward evaluation of UV-B penetration to specific areas
22      of the Earth's surface are the influences of clouds, as discussed by Erlick et al. (1998), Frederick
23      et al. (1998), and Soulen and Fredrick (1999). The transmission of solar UV and visible
24      radiation is highly sensitive to cloud type, cloud amount, and the extent of their external or
25      internal mixing with cloud droplets. Even in situations of very low atmospheric PM (e.g., over
26      Antarctica), interannual variations in cloudiness over specific areas can be as important as ozone
27      levels in determining UV surface  irradiation, with net impacts varying from a month or season to
28      another (Soulen and Fredrick,  1999).  Evaluations of the effects of changes in the transmission of
29      solar UV-B radiation to the surface have been performed usually for cloud-free or constant
30      cloudiness conditions.
        June 2003                                4-226        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Given the above considerations, quantification of projected effects of variations in
 2      atmospheric PM on human health or the environment because of the effects of particles on the
 3      transmission of solar UV-B radiation would require location-specific evaluations, that take into
 4      account (a) composition, concentration, and internal structure of the particles; (b) temporal
 5      variations in atmospheric mixing height and depths of layers containing the particles; and (c) the
 6      abundance of ozone and other absorbers within the planetary boundary layer and the free
 7      troposphere. The outcome of such modeling effects would likely vary from location to location
 8      in terms of increased or decreased surface level UV-B exposures because of location-specific
 9      changes in atmospheric PM concentrations or composition.  For example, to the extent that any
10      location-specific scattering by airborne PM were to affect the directional characteristics of UV
11      radiation at ground level, and thereby enhance radiation incident from low angles (Dickerson,
12      1997), the biological effectiveness (whether deleterious or beneficial) of resulting ground-level
13      UV-B exposures could be enhanced. Airborne PM also can reduce the ground-level ratio of
14      photorepairing radiation (UV-A and short-wavelength visible) to damaging UV-B radiation.
15      Lastly, PM deposition is a major source of PAHs in certain freshwater lakes and coastal areas,
16      and the adverse effects of solar UV are enhanced by the uptake of PAHs by aquatic organisms.
17      Thus, although airborne PM may, in general, tend to reduce ground-level UV-B,  its net effect in
18      some locations may be to increase UV damage to certain aquatic and terrestrial organisms, as
19      discussed by Cullen and Neale (1997).
20
21
22      4.6  SUMMARY AND KEY CONCLUSIONS
23      4.6.1  Particulate Matter Effects on Vegetation and Ecosystems
24           The first section of this chapter assesses and characterizes the overall ecological integrity
25      and indicate the status of ecosystems within the United States affected by the deposition of the
26      anthropogenic stressors associated with PM. There are six Essential Ecological Attributes
27      (EEAs) — Landscape Condition, Biotic Condition, and Chemical/ Physical Characteristics, and
28      Ecological Processes, Hydrology/Geomorphology, and Natural Disturbance Regimes that can be
29      used to provide a hierarchical framework for determining ecosystem status. The  first three can
30      be separated into "patterns" and the last three into "processes." The ecological processes create
31      and maintain the ecosystem elements in patterns.  The patterns in turn affect how the ecosystem

        June 2003                                4-227        DRAFT-DO NOT QUOTE OR CITE

-------
 1      processes are expressed. Patterns at the higher level of biological organization emerge from the
 2      interactions and selection processes at localized levels. Changes in patterns or processes result
 3      in changes in the status and functioning of an ecosystem. The relationships among the EEAs are
 4      complex because all are interrelated (i.e., changes in one EEA may affect, directly or indirectly,
 5      every other EEA).
 6           Human existence on Earth depends on the life-support services provided by the interaction
 7      of the different EEAs. Ecosystem processes and patterns provide the functions that maintain
 8      clean water,  clean air, a vegetated earth, and a balance of organisms that enable humans to
 9      survive. The benefits they impart include the absorption and breakdown of pollutants, cycling of
10      nutrients, binding of soil, degradation of organic waste, maintenance of a balance of gases in the
11      air, regulation of radiation balance, climate, and fixation of solar energy. Concern has arisen in
12      recent years  regarding biodiversity and the integrity of ecosystems.  There are few ecosystems
13      on Earth today whose activities are not influenced by humans. For this reason, understanding
14      the changes in biodiversity and nutrient cycling resulting from PM deposition are of great
15      importance.
16           The PM whose effects on vegetation and ecosystems is discussed in this section is not a
17      single pollutant but represents a heterogeneous mixture of particles differing in origin, size and
18      chemical constituents.  The effects of exposure to a given mass concentration of PM of particular
19      size may, depending on the particular mix of deposited particles, lead to widely differing
20      phytotoxic responses.
21           The deposition of PM onto vegetation and soil, depending on its  chemical composition
22      (acid/base, trace metal, or nutrients, e.g., nitrates or sulfates),  can produce direct or indirect
23      responses within an ecosystem.  Ecosystem response to pollutant deposition is a direct function
24      of the level of sensitivity of the ecosystem and its ability to ameliorate resulting change.
25      Changes in ecosystem structural patterns and the functioning of ecological processes, must be
26      scaled in both time and space and propagated to the more complex levels of community
27      interaction to produce observable ecosystem changes.
28           The nitrates and sulfates deposited in PM whose indirect effects occur via the soil are the
29      stressors of greatest environmental significance.  Upon entering the soil environment, they can
30      alter the ecological processes of energy flow and nutrient cycling, inhibit nutrient uptake, change
31      ecosystem structure, and affect ecosystem biodiversity.  The soil environment is one of the most

        June 2003                                 4-228        DRAFT-DO NOT QUOTE OR CITE

-------
 1      dynamic sites of biological interaction in nature. It is inhabited by microbial communities of
 2      bacteria, fungi and actinomycetes. Bacteria are essential participants in the nitrogen and sulfur
 3      cycles that make these elements available for plant uptake. Fungi in association with plant roots
 4      form mycorrhizae, a mutualistic symbiotic relationship that is integral in mediating plant uptake
 5      of mineral nutrients. Changes in the soil environment that influence the role of the bacteria in
 6      nutrient cycling and fungi in making minerals available for plant utilization, determine plant and
 7      ultimately ecosystem response.
 8           The affects on the growth of plants resulting from the deposition of nitrates and sulfates
 9      and the acidifying effect of the associated FT ion in wet and dry deposition in PM are the most
10      important environmentally.  Nitrogen is of overriding importance in plant metabolism and, to a
11      large extent governs the utilization of phosphorus, potassium, and other nutrients.  Typically, the
12      availability of nitrogen via nitrogen cycle controls net primary productivity, and possibly, the
13      decomposition rate of plant  litter. Plants usually obtain nitrogen directly from the  soil by
14      absorbing NH4+ or NO3" through their roots, or it is formed by symbiotic organisms in their roots.
15      Plants vary in their ability to absorb ammonium and nitrate from the soil.
16           Although nitrogen as molecular nitrogen (N2) is the most abundant element in the
17      atmosphere, it only becomes available to support the growth  of plants after it is converted into
18      reactive forms.  In nature, nitrogen may be  divided into two groups:  reactive (Nr)  and
19      nonreactive (N2).  Reactive Nr includes the inorganic reduced forms of nitrogen (e.g., ammonia
20      [NH3] and ammonium  [NH4+]), inorganic oxidized forms (e.g., nitrogen oxides [NOJ, nitric acid
21      [HNO3], nitrous oxide  [N2O], and nitrate [NO3"]), and organic compounds (e.g., urea, amine,
22      proteins, and nucleic acids).
23           Reactive nitrogen is now accumulating in the environment on all spatial  scales - local,
24      regional and global. The three main causes of the increase in global Nr is the result of the (1)
25      widespread cultivation of legumes, rice and other crops that promote conversion of N2 to organic
26      nitrogen through biological  nitrogen fixation; (2) combustion of fossil fuels, which converts both
27      atmospheric N2 and fossil nitrogen to reactive NOX; and the (3) Haber-Bosch process, which
28      converts nonreactive NH3 to sustain food production and some industrial activities. The major
29      changes in the nitrogen cycle due to the cited causes can be both beneficial and detrimental to
30      the health and welfare  of humans and ecosystems.
        June 2003                                 4-229       DRAFT-DO NOT QUOTE OR CITE

-------
 1           Reactive nitrogen can be widely dispersed and accumulate in the environment when the
 2      rates of its formation exceed the rates of removal via denitrification.  Nr creation and
 3      accumulation is projected to increase as per capita use of resources by human populations
 4      increases.  The cascade of environmental effects of increases in Nr include the following:
 5      (1) production of tropospheric ozone and aerosols that induce human health problems;
 6      (2) increases in the productivity in forests and grasslands followed by decreases wherever
 7      deposition increases significantly and exceeds critical thresholds; Nr additions probably also
 8      decrease biodiversity in many natural habitats; (3) in association with sulfur is responsible for
 9      acidification and loss of biodiversity in lakes and streams in many regions of the world;
10      (4) eutrophication, hypoxia, loss of biodiversity, and habitat degradation in coastal ecosystems.
11      [Eutrophication is now considered the biggest pollution problem in coastal waters.]
12      (5) contributes to global climate change and stratospheric ozone depletion, which can in turn
13      affect ecosystems and human health (See Table 4-21).
14           Changes in nitrogen supply can have a considerable effect on an ecosystem's nutrient
15      balance. Large chronic additions of nitrogen influence normal nutrient cycling and alter many
16      plant and soil processes involved in nitrogen cycling. "Nitrogen saturation" results when Nr
17      concentrations exceed  the capacity of a system to utilize it.  Saturation implies that some
18      resource other than nitrogen is limiting biotic function.  Water and phosphorus for plants and
19      carbon for microorganisms are most likely to be the secondary limiting factors. The appearance
20      of nitrogen in soil solution is an early symptom of excess nitrogen. In the final stage, disruption
21      of ecosystem structure becomes visible.
22           Possible ecosystem responses to nitrogen saturation include (1) a permanent increase in
23      foliar nitrogen and reduced foliar phosphorus and lignin caused by the lower availability of
24      carbon, phosphorus, and water; (2) reduced productivity in conifer stands because of disruptions
25      of physiological function; (3) decreased root biomass and increased nitrification and nitrate
26      leaching; and (4) reduced soil fertility, resulting from increased cation leaching, increased nitrate
27      and aluminum concentrations in streams, (5) decreased water quality and (6) changes in soil
28      microbial communities from predominantly fungal (mycorrhizal) communities to those
29      dominated by bacteria  during saturation.
30           Although soils of most North American forest ecosystems are nitrogen limited, there are
31      some that exhibit severe symptoms of nitrogen saturation (See Table 4-14). They include the

        June 2003                                4-230        DRAFT-DO NOT QUOTE OR CITE

-------
                           TABLE 4-21.  EFFECTS OF REACTIVE NITROGEN
         Direct effects ofNr on ecosystems include:
            Increased productivity of Nr-limited natural ecosystems.
            Ozone-induced injury to crop, forest, and natural ecosystems and predisposition to attack by pathogens
              and insects.
            Acidification and eutrophication effects on forests, soils, and freshwater aquatic ecosystems.
            Eutrophication and hypoxia in coastal ecosystems.
            N saturation of soils in forests and other natural ecosystems.
            Biodiversity losses in terrestrial and aquatic ecosystems and invasions by N-loving weeds.
            Changes in abundance of beneficial soil organisms that alter ecosystem functions.

         Indirect effects ofNr on other societal values include:

            Increased wealth and well being of human populations in many parts of the world.
            Significant changes in patterns of land use.
            Regional hazes that decrease visibility at scenic vistas and airports.
            Depletion of stratospheric ozone by N2O emissions.
            Global climate change induced by emissions of N2O and formation of tropospheric ozone.
            Damage to  useful materials and cultural artifacts by ozone, other oxidants, and acid deposition.
            Long-distance transport ofNr which causes harmful effects in countries distant from emission sources
              and/or increased background concentrations of zone and fine paniculate matter.

         In addition to these effects, it is important to recognize that:

            The magnitude of Nr flux often determines whether effects are beneficial or detrimental.
            All of these effects are linked by biogeochemical circulation pathways ofNr.
            Nr is easily transformed among reduced and oxidized forms in many systems.  Nr is easily distributed by
              hydrologic and atmospheric transport processes.
 1      high-elevation, non-aggrading spruce-fir ecosystems in the Appalachian Mountains; the eastern

 2      hardwood watersheds at the Fernow Experimental Forest near Parsons, WV; forests in southern

 3      California, the southwestern Sierra Nevada in Central California; and the Front Range in

 4      northern Colorado. The mixed conifer forest and chaparral watershed with high smog exposure

 5      in the Los Angeles Air Basin exhibit the highest stream water NO3" concentrations for wild-lands

 6      in North America;

 7            Increases in soil nitrogen play a selective role in ecosystems. Plants adapted to living in an

 8      environment of low nitrogen availability will be replaced by nitrophilic plants capable of using

 9      increased nitrogen because they have a competitive advantage when nitrogen becomes more

10      readily available. Plant succession patterns and biodiversity are affected significantly by chronic

11      nitrogen additions in some North American ecosystems.  Long-term nitrogen fertilization studies

12      in both New England and Europe suggest that some forests receiving chronic inputs of nitrogen

13      may decline in productivity and experience greater mortality.  Declining coniferous forest stands


        June 2003                                   4-231        DRAFT-DO NOT QUOTE OR CITE

-------
 1      with slow nitrogen cycling may be replaced by deciduous fast-growing forests that cycle
 2      nitrogen.
 3           Linked to the nitrogen cascade is the deposition of Nr and sulfates and the associated
 4      hydrogen ion is acidic precipitation, a critical environmental stress that affects forest landscapes
 5      and aquatic ecosystems in North America, Europe, and Asia.  Composed of ions, gases, and
 6      particles derived from gaseous emissions of sulfur dioxide (SO2), nitrogen oxides (NOX),
 7      ammonia (NH3) and particulate emissions of acidifying and neutralizing compounds, acidic
 8      precipitation is highly variable across time and space. Its deposition and the resulting soil
 9      acidity can lead to nutrient deficiencies and to high aluminum-to-nutrient ratios that limit uptake
10      of calcium and magnesium and create a nutrient deficiency.  Aluminum accumulation in root
11      tissue can reduce calcium uptake and causes  Ca2+deficiencies.  Tree species can be adversely
12      affected if altered Ca/Al ratios impair calcium or magnesiums uptake.  Calcium is  essential in
13      the formation of wood and the maintenance of the primary plant tissues necessary  for tree
14      growth.  Studies suggest that the decline of red spruce stands in Vermont may be related to the
15      Ca/Al ratio (Johnson and Lindberg, 1992b).
16           The evidence of the effects of wet and dry particulate deposition SO4"  and Nr species on
17      nutrient cycling in forest ecosystems is provided by the Integrated Forest Study (IPS).  The
18      deposition data from the study illustrates  several important aspects of the atmospheric exposure
19      characteristics across a wide elevational gradient and over a wide spatial scale. Atmospheric
20      deposition plays a significant role in the biogeochemical cycles at all IPS  sites, but is most
21      important in the east at high-elevation sites.  The flux of the sulfate ion, Nr compounds, and FT
22      ions in throughfall at all sites is dominated by atmospheric deposition.  Atmospheric deposition
23      may have significantly affected the nutrient status of some IPS sites through the mobilization of
24      aluminum by impeding cation uptake. Nitrates and sulfate are the dominant anions in the
25      Smokies and pulses of nitrates are the major  causes of aluminum pulses in soil solutions.
26      However, the connection between aluminum mobilization and forest decline is not clear, hence,
27      aluminum mobilization presents a situation worthy of further study.
28           Notable impacts of excess nitrogen  deposition  also have been observed with regard to
29      aquatic systems.  For example, atmospheric nitrogen deposition into soils in watershed areas
30      feeding into estuarine sound complexes (e.g., the Pamlico Sound of North Carolina) appear to
31      contribute to excess nitrogen flows in runoff (especially during and after heavy rainfall events

        June 2003                                4-232        DRAFT-DO NOT QUOTE OR CITE

-------
 1      such as hurricanes) from agricultural practices or other uses (e.g., fertilization of lawns or
 2      gardens), massive influxes of such nitrogen into watersheds and sounds can lead to dramatic
 3      decreases in water oxygen and increases in algae blooms that can cause extensive fish kills and
 4      damage to commercial fish and sea food harvesting.
 5           The critical loads concept is useful for estimating the amounts of pollutants (e.g., Nr and
 6      acidic precipitation) that sensitive ecosystems can absorb on a sustained basis without
 7      experiencing measurable degradation. The estimation of ecosystem critical loads requires an
 8      understanding of how an ecosystem will respond to different loading rates in the long term and
 9      can be of special value for ecosystems receiving chronic deposition of Nr and sulfur
10      independently and as acidic deposition when in combination.
11           Changes in the soil can result from the deposition of heavy metals. Exposures to heavy
12      metals are highly variable depending whether deposition is by wet or dry processes.  Few (e.g.,
13      copper, nickel, zinc) have been documented to have direct phytoxicity under field conditions.
14      Exposure to coarse particles of natural origin and elements such as iron and manganese are more
15      likely to occur via dry deposition while fine particles of atmospheric origin and elements such as
16      cadmium, chromium, lead nickel, and vanadium. Ecosystems immediately downwind of major
17      emissions sources such as power generating, industrial, or urban complexes can receive locally
18      heavy inputs.  Heavy metal accumulation by affecting litter decomposition presents the greatest
19      potential for influencing nutrient cycling. Microbial populations decreased and logarithmic rates
20      of microbial increase were prolonged as a result of cadmium toxicity.  Additionally, the presence
21      of cadmium, copper and nickel were observed to affect the symbiotic activity of fungi, bacteria,
22      and actinomycetes.
23           Phytochelatins produced by plants as a response to sublethal concentrations of heavy
24      metals, are indicators of metal stress and can be used to indicate that heavy metals are involved
25      in forest decline. Increasing concentrations of phytochelatins with increase in altitude and their
26      increase across regions showing increased levels of forest injury implicated them in forest
27      decline.
28           The ambient concentration of particles, the parameter for which there is most data (Chapter
29      3), is at best an indicator of exposure. The amount of PM entering the immediate plant
30      environment, deposited onto the plant surfaces or soil in the vicinity of the roots, determines the
31      biological effect. Three major routes are involved during the wet and dry deposition processes:

        June 2003                                4-233        DRAFT-DO NOT QUOTE OR CITE

-------
 1      (1) precipitation scavenging in which particles are deposited in rain and snow; (2) occult (fog,
 2      cloud water, and mist interception); and (3) dry deposition, a much slower, yet more continuous
 3      removal to place surfaces.
 4           Deposition of PM on above-ground plant parts can have either a physical and/or chemical
 5      effect.  Particles transferred from the atmosphere to plant surfaces may cause direct effects if
 6      they (1) reside on the leaf, twig, or bark surface for an extended period; (2) are taken up through
 7      the leaf surface; or (3) are removed from the plant via resuspension to the atmosphere, washing
 8      by rainfall, or litter-fall with subsequent transfer to the soil.  Ecosystem effects have been
 9      observed only in the neighborhood limestone quarries.
10           Secondary organics formed in the atmosphere have been variously subsumed under the
11      following terms: toxic substances, pesticides, hazardous  air  pollutants (HAPS), air toxics,
12      semivolatile organic compounds (SOCs), and persistent organic pollutants (POPS).  The
13      substances alluded to under the above headings are controlled under CAA Sect. 112, Hazardous
14      Air Pollutants not as criteria pollutants controlled by NAAQS under CAA Sections 108 and 109
15      (U.S. Code, 1991).  Their possible effects on humans and ecosystems are discussed in many
16      other government documents and publications.  They are noted in this chapter because, in the
17      atmosphere, many of the chemical compounds are partitioned between gas and particle phases
18      and are deposited as particulate matter.  As particles, they become airborne and can be
19      distributed over a wide area and affect remote ecosystems. Some of the chemical compounds
20      are of concern to humans because they may reach toxic levels in food chains of both animals and
21      humans; whereas others tend to decrease or maintain the  same toxicity as they move through the
22      food chain.
23           An important characteristic of fine  particles is their ability to affect flux of solar radiation
24      passing through the atmosphere directly, by scattering and absorbing solar radiation, and
25      indirectly, by acting as cloud condensation nuclei that, in turn, influence the optical properties of
26      clouds. Regional haze has been estimated to diminish surface solar visible radiation by
27      approximately 8%.  Crop yield have been reported as being sensitive to the amount of sunlight
28      receive, and crop losses have been attributed to increased airborne particle levels in some area of
29      the world.
30           From  among all of the various effects of PM on ecosystems discussed above, the effects of
31      excess nitrogen added to the environment via emissions to the atmosphere and especially

        June 2003                                 4-234       DRAFT-DO NOT QUOTE OR CITE

-------
 1      increases in Nr due to food production has resulted in the accumulation in the environmental at
 2      all spatial scales, local, regional and global.  The results of the nitrogen cascade as Nr moves
 3      through the different environmental compartments has effects both beneficial and detrimental to
 4      humans and ecosystems.  Table 4-21 summarizes a variety of effects (both and indirect)
 5      associated with the presence of Nr as it circulates through the various environmental
 6      compartments. These effects are associated with the long-term chronic deposition of Nr on the
 7      terrestrial and aquatic ecosystems.
 8
 9      4.6.2  Particulate Matter-Related Effects on Visibility
10           Visibility is defined as the degree to which the  atmosphere is transparent to visible light
11      and the clarity and color fidelity of the atmosphere. Visual range is the farthest distance a black
12      object can be distinguished against the horizontal sky. Visibility impairment is  any humanly
13      perceptible change in visibility. For regulatory purposes, visibility impairment, characterized by
14      light extinction, visual range, contrast, and coloration, is classified into two principal forms:
15      (1) "reasonably attributable" impairment, attributable to a single source or small group of
16      sources, and (2) regional haze, any perceivable change in visibility caused by a combination of
17      many sources over a wide geographical area.
18           Visibility is measured by human observation, light scattering by particles, the light
19      extinction-coefficient, and parameters related to the light-extinction coefficient (visual range and
20      deciview scale),  and fine PM mass concentrations.
21           The air quality within a sight path will affect the illumination of the sight path by scattering
22      or absorbing solar radiation before it reaches the Earth's surface. The rate of energy loss with
23      distance from a beam of light is the light extinction coefficient. The light extinction coefficient
24      is the sum of the coefficients for light absorption by gases (oag), light scattering by gases (osg),
25      light absorption by particles (oap), and light scattering by particles (osp). Corresponding
26      coefficients for light scattering and absorption by fine and coarse particles are osfp and oafp and
27      oscp and oacp, respectively. Visibility within a sight path longer than approximately 100 km  (60
28      mi) is affected by the change in the optical properties of the atmosphere over the length of the
29      sight path.
30           Visual range was developed for and continues to be used as an aid in military operations
31      and to a lesser degree in transportation safety.  Visual range is commonly taken to be the greatest

        June 2003                                 4-23 5        DRAFT-DO NOT QUOTE OR CITE

-------
 1      distance a dark object can be seen against the background sky. The deciview is an index of
 2      haziness. A change of 1 or 2 deciviews is seen as a noticeable change in the appearance of a
 3      scene.
 4           Under certain conditions, fine particle mass concentrations may be used as a visibility
 5      indicator. However, the relationship may differ between locations and for different times of the
 6      year. Also, measurement should be made under dry conditions.
 7           Visibility impairment is associated with airborne particle properties, including size
 8      distributions (i.e., fine particles in the 0.1- to 1.0-|im size range) and aerosol chemical
 9      composition, and with relative humidity. With increasing relative humidity, the amount of
10      moisture available for absorption by particles increases, thus causing the particles to increase in
11      both size and volume. As the particles increase in size and volume, the light scattering potential
12      of the particles also generally increases.  Visibility impairment is greatest in the eastern United
13      States and Southern California.  In the eastern United States, visibility impairment is caused
14      primarily by light scattering by sulfate aerosols and, to a lesser extent, by nitrate particles and
15      organic aerosols, carbon soot, and crustal dust. Up to 86% of the haziness in the eastern United
16      States is caused by atmospheric sulfate. Further West, scattering contributions to visibility
17      impairment decrease to from 25 to 50%.  Light scattering by nitrate aerosols is the major cause
18      of visibility impairment in southern California. Nitrates contribute about 45% to the total light
19      extinction in the West and up to 17%  of the total extinction in the East.  Organic particles are the
20      second largest contributors to light extinction in most U.S. areas. Organic carbon is the greatest
21      cause of light extinction in the West, accounting for up to 40% of the total extinction and up to
22      18% of the visibility impairment in the East.  Coarse mass and soil, primarily considered
23      "natural extinction," is responsible for some of the visibility impairment in the West, accounting
24      for up to 25% of the light extinction.
25
26      4.6.3  Particulate Matter-Related Effects on Materials
27           Building materials (metals, stones,  cements, and paints) undergo natural weathering
28      processes from exposure to environmental elements (wind, moisture, temperature fluctuations,
29      sun light, etc.).  Metals form a protective film of oxidized metal (e.g., rust) that slows
30      environmentally induced corrosion. On the other hand, the natural process of metal corrosion
        June 2003                                 4-236       DRAFT-DO NOT QUOTE OR CITE

-------
 1      from exposure to natural environmental elements is enhanced by exposure to anthropogenic
 2      pollutants, in particular SO2, that render the protective film less effective.
 3           Dry deposition of SO2 enhances the effects of environmental elements on calcereous stones
 4      (limestone, marble, and cement) by converting calcium carbonate (calcite) to calcium sulfate
 5      dihydrate (gypsum). The rate of deterioration is determined by the SO2 concentration, the
 6      stone's permeability and moisture content, and the deposition rate; however, the extent of the
 7      damage to stones produced by the pollutant species apart from the natural weathering processes
 8      is uncertain.  Sulfur dioxide also has been found to limit the life expectancy of paints by causing
 9      discoloration and loss of gloss and thickness of the paint film layer.
10           A significant detrimental effect of particle pollution is the soiling of painted surfaces and
11      other building materials. Soiling changes the reflectance of a material from opaque and reduces
12      the transmission of light through transparent materials. Soiling is a degradation process that
13      requires remediation by cleaning or washing, and, depending on the soiled surface, repainting.
14      Available data on pollution exposure indicates that particles can result in increased cleaning
15      frequency  of the exposed surface and may reduce the usefulness of the soiled material. Attempts
16      have been  made to quantify the pollutants exposure levels at which materials damage and soiling
17      have been  perceived.  However, to date, insufficient data are available to advance our knowledge
18      regarding perception thresholds with respect to pollutant concentration, particle size, and
19      chemical composition.
20
21      4.6.4  Effects of Atmospheric Particulate  Matter on Global Warming
22             Processes and Transmission of Solar Ultraviolet Radiation
23           The physical processes (i.e., scattering and absorption) responsible for airborne particle
24      effects on transmission of solar visible and ultraviolet radiation are the same as those responsible
25      for visibility degradation. Scattering of solar radiation back to space and absorption of solar
26      radiation determine the effects of an aerosol layer on solar radiation.
27           Atmospheric particles greatly complicate projections of future trends in global warming
28      processes because of emissions of greenhouse gases; consequent increases in global mean
29      temperature; resulting changes in regional and local weather patterns; and mainly deleterious
30      (but some beneficial) location-specific human health and environmental  effects. The body of
31      available evidence, ranging from satellite to in situ measurements of aerosol effects on radiation
32      receipts and cloud properties, is strongly indicative  of an important role in climate for aerosols.
        June 2003                                 4-237       DRAFT-DO NOT QUOTE OR CITE

-------
 1      This role, however, is poorly quantified.  No significant advances have been made in reducing
 2      the uncertainties assigned to forcing estimates provided by the IPCC for aerosol-related forcing,
 3      especially for black carbon-containing aerosol.  The IPCC characterizes the scientific
 4      understanding of greenhouse gas-related forcing as "high" in contrast to that for aerosol, which it
 5      describes as "low" to "very low."
 6           Quantification of the effect of anthropogenic aerosol on hydrological cycles requires more
 7      information than is presently available regarding ecosystems responses to reduced solar radiation
 8      and other changes occurring in the climate system. However, several global scale studies
 9      indicate that aerosol cooling alone can slow down the hydrological cycle, while cooling plus the
10      nucleation of additional cloud droplets can dramatically reduce precipitation rates.
11           In addition to direct climate effects through the scattering and absorption of solar radiation,
12      particles also  exert indirect effects on climate by serving as  cloud condensation nuclei, thus
13      affecting the abundance and vertical distribution of clouds.  The direct and indirect effects of
14      particles appear to have significantly offset global warming effects caused by the buildup of
15      greenhouse gases on a globally-averaged basis.  However, because the lifetime of particles is
16      much shorter than that required for complete mixing within the Northern Hemisphere, the
17      climate effects of particles generally are felt much less homogeneously than are the effects of
18      long-lived greenhouse gases.
19           Any effort to model the impacts of local alterations in particle concentrations on projected
20      global climate change or consequent local and regional weather patterns would be subject  to
21      considerable uncertainty.
22           Atmospheric particles also complicate estimation of potential future impacts on human
23      health and the environment projected  as possible to occur because of increased transmission of
24      solar ultraviolet radiation (UV-B) through the Earth's atmosphere, secondary to stratospheric
25      ozone depletion due to anthropogenic emissions of chlorofluorcarbons (CFCs), halons, and
26      certain other gases. The transmission of solar UV-B radiation is affected strongly by
27      atmospheric particles. Measured attenuations of UV-B under hazy conditions range up to  37%
28      of the incoming solar radiation.  Measurements relating variations in PM mass directly to UV-B
29      transmission are lacking.  Particles also can affect the rates of photochemical reactions occurring
30      in the atmosphere, e.g., those involved in catalyzing tropospheric ozone formation. Depending
31      on the amount of absorbing substances in the particles,  photolysis rates either can be increased or

        June 2003                                 4-23 8        DRAFT-DO NOT QUOTE OR CITE

-------
1      decreased. Thus, atmospheric particle effects on UV-B radiation, which vary depending on size
2      and composition of particles, can differ substantially over different geographic areas and from
3      season to season over the same area. Any projection of effects of location-specific airborne PM
4      alterations on increased atmospheric transmission of solar UV radiation (and associated potential
5      human health or environmental effects) due to stratospheric ozone-depletion would, therefore,
6      also be subject to considerable uncertainty.
7
       June 2003                                4-239        DRAFT-DO NOT QUOTE OR CITE

-------
  1       REFERENCES

  2       Abdullah, U. M; Iqbal, M. Z. (1991) Response of automobile, stone and cement paniculate matters on stomatal
  3             clogging of plants. Geobios (Jodhpur, India) 18: 196-202.
  4       Aber, J. D.; Nadelhoffer, K. J.; Steudler, P.; Melillo, J. M. (1989) Nitrogen saturation in northern forest ecosystems:
  5             excess nitrogen from fossil fuel combustion may stress the biosphere. Bioscience 39: 378-386.
  6       Aber, J.; McDowell, W.; Nadelhoffer, K.; Magill, A.; Berntson, G.; Kamakea, M.; McNulty, S.; Currie, W.;
  7             Rustad, L.; Fernandez, I. (1998) Nitrogen saturation in temperate forest ecosystems. BioScience 48: 921-934.
  8       Ainsleigh, H. G. (1993) Beneficial effects of sun exposure on cancer mortality. Prev. Med. 22: 132-140.
  9       Alexander, M. (1977) Introduction to soil microbiology. 2nd ed. New York, NY: John Wiley & Sons; pp. 225-230;
10             239-246; 272-286; 355-356.
11       Allen, M. F. (1991) The ecology of mycorrhizae. Cambridge, MA: Cambridge University Press.
12       Allen, A. G.; Harrison, R. M.; Nicholson, K. W. (1991) Dry deposition of fine aerosol to a short grass surface.
13             Atmos. Environ. Part A 25: 2671-2676.
14       Allen, E. R.; Cabrera, N.; Kim, J.-C. (1994) Atmospheric deposition studies in rural forested areas of the
15             southeastern United States. Presented at: 87th annual meeting & exhibition of the Air & Waste Management
16             Association;  June; Cincinnati, OH. Pittsburgh, PA: Air & Waste Management Association; paper no.
17             94-WP88.02.
18       Allen, E. B.; Padgett, P.  E.; Bytnerowicz, A.; Minich, R. (1998) Nitrogen deposition effects on coastal sage
19             vegetation of southern California. USDA Forest Service Gen. Tech. Rep. PSW-GTR-166, pp. 131-139.
20       Altshuller, A. P.; Linthurst, R. A., eds.  (1984) Acidic deposition phenomenon and its effects: critical assessment
21             review papers; volume I, atmospheric sciences; volume II, effects sciences. Washington, DC: U.S.
22             Environmental Protection Agency, Office of Research and Development; EPA  report nos.
23             EPA/600/8-83/016aF-bF. Available from: NTIS, Springfield, VA; PB85-100030 and PB85-100048.
24       Andersen, C. P.; Rygiewicz,  P. T. (1991) Stress interactions and mycorrhizal plant response: understanding carbon
25             allocation priorities. Environ. Pollut. 73: 217-244.
26       Anderson, J. B.; Baumgardner, R. E.; Mohnen, V. A.; Bowser, J.  J. (1999) Cloud chemistry in the eastern United
27             States, as sampled from three high-elevation sites along the Appalachian Mountains. Atmos. Environ.
28             33:5105-5114.
29       Aneja, V. P.; Murthy, A. B. (1994) Monitoring deposition of nitrogen-containing compounds in a high-elevation
30             forest canopy. Air Waste 44: 1109-1115.
31       Antonovics, J.; Bradshaw, A. D.; Turner, R. G. (1971) Heavy metal tolerance in plants. In: Cragg, J. B., ed.
32             Advances in ecological research. Volume 7. London, United Kingdom: Academic Press; pp. 1-85.
33       April, R.; Newton, R.  (1992) Mineralogy and mineral weathering. In: Johnson, D. W.; Lindberg, S. E., eds.
34             Atmospheric deposition and forest nutrient cycling: a synthesis of the integrated forest study. New York, NY:
35             Springer-Verlag; pp. 378-425. (Billings, W. D.; Golley, F.; Lange, O. L.; Olson,  J. S.; Remmert, H., eds.
36             Ecological studies: analysis and synthesis: v. 91).
37       Arizona Department of Environmental  Quality. (2002) Air quality report. Phoenix, AZ: Arizona Department of
38             Environmental Quality; Air Quality Assessment Section. A.R.S. "section" 49-424.10. Available:
39             http://www.adeq.state.az.us/comm/pubs/ar.html (28 May 2003).
40       Arnott, W. P.; Moosmtiller, H.; Rogers, C. F.; Jin, T.; Bruch, R. (1999) Photoacoustic spectrometer for measuring
41             light absorption by aerosol; instrument description. Atmos. Environ. 33: 2845-2852.
42       Arrhenius, S. (1896) On the influence of carbonic acid in the air upon the temperature of the ground. Philos. Mag.
43             41:237-276.
44       Askey, A.; Lyon, S. B.; Thompson, G.  E.; Johnson, J. B.; Wood, G. C.;  Sage, P. W.; Cooke, M. J. (1993) The effect
45             of fly-ash particulates on the atmospheric corrosion of zinc and mild steel. Corros. Sci. 34: 1055-1081.
46       Ausset, P.; Bannery, F.; Del Monte, M.; Lefevre, R. A. (1998) Recording of pre-industrial atmospheric environment
47             by ancient crusts on stone monuments. Atmos. Environ. 32: 2859-2863.
48       Ayensu, E.; Van R.  Claasen, D.; Collins, M.; Dearing, A.; Fresco, L.; Gadgil, M.; Gitay, H.; Glaser, G.; Juma, C.;
49             Krebs, J.; lenton, R.; Lubchenco, J.; McNeely, J. A.; Mooney, H. A.; Pinstrup-Andersen, P.; Ramos, M.;
50             Raven, P.; Reid, W. V; Samper, C.; Sarukhan, J.; Schei, P.; Tundisi, J. G.; Watson, R. T.; Guanhua, X.;
51             Zakri, A. H. (1999) International ecosystem assessment. Science  286: 685-686.
52       Babich, H.; Stotzky, G. (1978) Effects of cadmium on the biota: influence of environmental factors. In: Perlman, D.,
53             ed. Advances in applied microbiology, volume 23. New York, NY: Academic Press; pp. 55-117.
54       Bache, D. H. (1979a) Particle transport within plant canopies -1. a framework for analysis. Atmos. Environ.
55             13: 1257-1262.
         June 2003                                      4-240         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Bache, D. H. (1979b) Paniculate transport within plant canopies - II. prediction of deposition velocities.
  2             Atmos. Environ. 13: 1681-1687.
  3       Baedecker, P. A.; Edney, E. O.; Moran, P. I; Simpson, T. C.; Williams, R. S. (1991) Effects of acidic deposition on
  4             materials. In: Irving, P. M, ed. Acidic deposition: state of science and technology, volume III: terrestrial,
  5             materials, health and visibility effects. Washington, DC: The U.S. National Acid Precipitation Assessment
  6             Program. (State of science and technology report no. 19).
  7       Baedecker, P. A.; Reddy, M. M.; Reimann, K. J.; Sciammarella, C. A. (1992) Effects of acidic deposition on the
  8             erosion of carbonate stone-experimental results from the U.S. National Acid Precipitation Assessment
  9             Program (NAPAP). Atmos. Environ. Part B 26:  147-158.
10       Baldocchi, D. (1988) A multi-layer model for estimating sulfur dioxide deposition to a deciduous oak forest canopy.
11             Atmos. Environ. 22: 869-884.
12       Bange, M. P.; Hammer, G. L.; Rickert, K. G. (1997) Effect of radiation environment on radiation use efficiency and
13             growth of sunflower. Crop Sci. 37: 1208-1214.
14       Baron, J. S.; Rueth, H. M.; Wolfe, A. M.; Nydick, K. R.; Allstott, E. J.; Minear, J. T.; Moraska, B. (2000) Ecosystem
15             responses to nitrogen deposition in the Colorado front range. Ecosystems 3: 352-368.
16       Barone, J. B.; Cahill, T. A.; Eldred, R. A.; Flocchini, R. G.; Shadoan, D. J.; Dietz, T. M. (1978) A multivariate
17             statistical analysis of visibility degradation at four California cities. Atmos. Environ. 12: 2213-2221.
18       Barrie, L. A.; Anlauf, K.; Wiebe, H. A.; Fellin, P. (1984) Acidic pollutants in air and precipitation at selected rural
19             locations in Canada. In: Hicks, B. B., ed. Deposition both wet and dry: [papers presented at the American
20             Chemical Society acid rain symposium]; March 1982; Las Vegas, NV. Boston, MA: Butterworth Publishers;
21             pp. 15-35. (Teasley, J. L, ed. Acid precipitation series: v. 4).
22       Barton, K. (1958) The influence of dust on atmospheric corrosion of metals. Werkst. Korros. 8/9: 547-549.
23       Beier, C. (1991) Separation of gaseous and paniculate dry deposition of sulfur at a forest edge in Denmark. J.
24             Environ. Qual. 20: 460-466.
25       Beier, C.; Gundersen, P.; Rasmussen, L. (1992) A new method for estimation of dry deposition of particles based on
26             throughfall measurements in a forest edge. Atmos. Environ. Part A 26: 1553-1559.
27       Bellan, L. M.; Salmon, L. G.; Cass, G. R. (2000) A study on the human ability to detect soot deposition onto works
28             of art. Environ.  Sci. Technol. 34: 1946-1952.
29       Bennett, J. H.; Hill, A. C. (1973) Absorption of gaseous air pollutants by a standardized plant canopy. J. AirPollut.
30             Control Assoc. 23: 203-206.
31       Bennett, J. H.; Hill, A. C. (1975) Interactions of air pollutants with canopies of vegetation. In: Mudd, J. B.;
32             Kozlowski, T. T., eds. Responses of plants to air pollution. New York, NY: Academic Press; pp. 273-306.
33       Bernard, S. M.; McGeehin, M. G.; Patz, J. A., eds. (2001) Health: the potential consequences of climate variability
34             and change.  Environ. Health Perspect. 109(suppl. 2):  175-233.
3 5       Bewley, R. J. F.; Parkinson, D. (1984) Effects of sulphur dioxide pollution on forest soil microorganisms. Can. J.
36             Microbiol. 30: 179-185.
37       Biefer, G. J. (1981) Atmospheric corrosion of steel in the Canadian Arctic. Mater. Perform. 20: 16-19.
38       Billings, W. D. (1978) Plants and the ecosystem. 3rded. Belmont, CA: Wadsworth Publishing Company, Inc.;
39             pp. 1-62, 83-108.
40       Binkley, D.; Valentine, D.; Wells, C.; Valentine, U. (1989) An empirical analysis of the factors contributing to
41             20-year decrease in soil pH in an old-field plantation of loblolly pine. Biogeochemistry 8: 39-54.
42       Blackburn, T. E.; Bairai, S. T.; Stedman, D. H. (1992) Solar  photolysis of ozone to singletD oxygen atoms.
43             J. Geophys. Res. 97: 10,109-10,117.
44       Blaschke, H. (1990) Mycorrhizal populations and fine root development on Norway spruce exposed to controlled
45             doses of gaseous pollutants and simulated acidic rain treatments. Environ. Pollut. 68: 409-418.
46       Bobbink, R. (1998) Impacts of tropospheric ozone and airborne nitrogenous pollutants on natural and semi-natural
47             ecosystems:  a commentary. NewPhytol. 139: 161-168.
48       Bobbink, R.; Hornung, M.; Roelofs, J. G. M. (1998) The effects of air-borne nitrogen pollutants on species diversity
49             in natural and semi-natural European vegetation. J. Ecol. 86: 717-738.
50       Bondietti, E. A.; McLaughlin, S. B. (1992) Evidence of historical influences of acidic deposition on wood and soil
51             chemistry. In: Johnson, D. W.; Lindberg, S. E., eds. Atmospheric deposition and forest nutrient cycling:
52             a synthesis of the integrated forest study. New York,  NY: Springer-Verlag: pp. 358-377. (Billings, W. D.;
53             Golley, F.; Lange, O. L.; Olson, J. S.; Remmert, H., eds.  Ecological studies analysis and synthesis: v. 91).
54       Bormann, F. H.; Likens, G. E. (1979) Catastrophic disturbance and the steady state in northern hardwood forests.
55             Am. Sci. 67: 660-669.
56       Bowden, R. D.; Geballe, G. T.; Bowden, W. B. (1989) Foliar uptake of 15N from simulated cloud water by red
57             spruce (Picea rubens) seedlings. Can. J. For. Res. 19: 382-386.


         June 2003                                      4-241         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Bowman, W. D. (2000) Biotic controls over ecosystem response to environmental change in alpine tundra of the
  2             Rocky Mountains. Ambio 29: 396-400.
  3       Bowman, W. D.; Steltzer, H. (1998) Positive feedbacks to anthropogenic nitrogen deposition in Rocky Mountain
  4             alpine tundra. Ambio 27: 514-517.
  5       Boyce, S. G. (1954) The salt spray community. Ecol. Monogr. 24: 29-67.
  6       Bradley, R.; Burt, A. I; Read, D. J. (1981) Mycorrhizal infection and resistance to heavy metal toxicity in Calluna
  1             vulgaris. Nature (London) 292: 335-337.
  8       Brandt, C. J.; Rhoades, R. W. (1972) Effects of limestone dust accumulation on composition of a forest community.
  9             Environ. Pollut. 3: 217-225.
10       Brandt, C. J.; Rhoades, R. W. (1973) Effects of limestone dust accumulation on lateral growth of forest trees.
11             Environ. Pollut. 4: 207-213.
12       Brook, J. R.; Zhang, L.; Di-Giovanni, F.; Padro, J. (1999) Description and evaluation of a model of deposition
13             velocities for routine estimates of air pollutant dry deposition over North America. Part I: model development.
14             Atmos. Environ. 33: 5037-5051.
15       Brookshire, D. S.; d'Arge, R. C.; Schulze, W. D.; Thayer, M. D.  (1979) Methods development for assessing air
16             pollution control benefits: volume II.  alternative benefit measures of air pollution control in the south coast air
17             basin of southern California. Washington, DC: U.S. Environmental Protection Agency, Office of Research
18             and Development; report no. EPA-600/6-79-00Ib.
19       Briihl, C.; Crutzen,  P. J. (1988) On the disproportionate role of tropospheric ozone as a filter against solar UV-B
20             radiation. Geophys. Res. Lett. 16: 703-706.
21       Brumme, R.; Leimcke, U.; Matzner, E. (1992) Interception and uptake of NH4 and NO3 from wet deposition by
22             above-ground parts of young beech (Fagus silvatica L.) trees. Plant Soil 142: 273-279.
23       Brunei, J.; Diekmann, M.; Falkengren-Grerup, U. (1998) Effects of nitrogen deposition on field layer vegetation in
24             south Swedish oak forests. Environ. Pollut. (Oxford, U.K.) 102(suppl. 1): 35-40.
25       Bugini, R.; Tabasso, M. L.; Realini, M.  (2000) Rate of formation of black crusts on marble. A case study. J. Cult.
26             Heritage 1: 111-116.
27       Businger, J. A. (1986) Evaluation of the accuracy with which dry deposition can be measured with current
28             micrometeorological techniques.  J. Clim. Appl.  Meteorol. 25:  1100-1124.
29       Butlin, R. N.; Coote, A. T.; Devenish, M.; Hughes, I. S. C.; Hutchens, C. M.; Irwin, J. G.; Lloyd, G. O.; Massey,
30             S. W.; Webb, A. H.; Yates, T. J.  S. (1992a) Preliminary results from the analysis of metal samples from the
31             National Materials Exposure Programme (NMEP). Atmos. Environ. Part B 26:  199-206.
32       Butlin, R. N.; Coote, A. T.; Devenish, M.; Hughes, I. S. C.; Hutchens, C. M.; Irwin, J. G.; Lloyd, G. O.; Massey,
33             S. W.; Webb, A. H.; Yates, T. J.  S. (1992b) Preliminary results from the analysis of stone tablets from the
34             National Materials Exposure Programme (NMEP). Atmos. Environ. Part B 26:  189-198.
35       Bytnerowicz, A.; Fenn, M.  E. (1996) Nitrogen deposition in California forests:  a review. Environ. Pollut.
36             92:  127-146.
37       Bytnerowicz, A.; Miller, P. R.; Olszyk, D. M. (1987a)  Dry deposition of nitrate, ammonium and sulfate to a
38             Ceanothus crassifolius canopy and surrogate surfaces. Atmos. Environ. 21: 1749-1757.
39       Bytnerowicz, A.; Miller, P. R.; Olszyk, D. M.; Dawson, P. J.; Fox, C. A. (1987b)  Gaseous and paniculate air
40             pollution in the San Gabriel Mountains of southern California. Atmos. Environ. 21: 1805-1814.
41       Cahill, T. A.; Eldred, R. A.; Wakabayashi, P. H. (1996) Trends in fine particle concentrations at Great Smoky
42             Mountains National Park. For presentation at: Annual meeting of the Air & Waste Management Association;
43             June; Nashville, TN. Pittsburgh, PA: Air & Waste Management Association; submitted paper no.
44             96-MP1A.05.
45       Campbell, G. G.; Schurr, G. G.; Slawikowski, D. E.; Spence, J. W. (1974) Assessing air pollution damage to
46             coatings. J. Paint Technol. 46: 59-71.
47       Camuffo, D. (1995) Physical weathering of stones. In:  Saiz-Jimenez, C., ed. The deterioration of monuments:
48             proceedings  of the 2nd international symposium on biodeterioration and biodegradation; January 1994;
49             Sevilla,  Spain. Sci. Total Environ. 167: 1-14.
50       Cape, J. N. (1993) Direct damage to vegetation caused by acid rain and polluted cloud: definition of critical levels
51             for forest trees. Environ. Pollut. 82: 167-180.
52       Cape, J. N.; Sheppard, L. J.; Fowler, D.; Harrison, A. F.; Parkinson, J. A.; Dao, P.; Paterson, I.  S. (1992)
53             Contribution of canopy leaching to sulphate deposition in a Scots pine forest. Environ. Pollut. 75: 229-236.
54       Carey, W. F. (1959) Atmospheric deposits in Britain: a study of dinginess. Int. J. Air Pollut. 2:  1-26.
55       Ceccotti, S. P.; Messick, D. L. (1997) A global review of crop requirements, supply, and environmental impact on
56             nutrient sulphur balance.  In: Cram, W. J.; De Kok, L. J.; Stulen, L; Brunold, C.; Rennenberg, H., eds. Sulphur
57             metabolism in higher plants. Leiden, The Netherlands: Backhuys Publishers; pp. 155-163.


         June 2003                                      4-242         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Chamberlain, A. C. (1975) The movement of particles in plant communities. In: Monteith, J. L., ed. Vegetation and
  2             the atmosphere: v. 1, principles. London, United Kingdom: Academic Press; pp. 155-203.
  3       Chamberlain, A. C.; Little, P. (1981) Transport and capture of particles by vegetation. In: Grace, J.; Ford, E. D.;
  4             Jarvis, P. G., eds. Plants and their atmospheric environment: the 21st symposium of the British Ecological
  5             Society; 1979; Edinburgh, United Kingdom. Oxford, United Kingdom:  Blackwell Scientific Publications;
  6             pp. 147-173.
  7       Chameides, W. L.; Yu, H.; Liu, S. C.; Bergin, M; Zhou, X.; Mearns, L.; Wang, G.; Kiang, C. S.; Saylor, R. D.;
  8             Luo, C.; Huang, Y.; Steiner, A.; Giorgi, F. (1999) Case study of the effects of atmospheric aerosols and
  9             regional haze on agriculture: an opportunity to enhance crop yields in China through emission controls?
10             Proc. Natl. Acad. Sci. U. S. A. 96:  13626-13633.
11       Chapin, F. S., III. (1991) Integrated responses of plants to stress: a centralized system of physiological responses.
12             Bioscience  41: 29-36.
13       Chapin, F. S., Ill; Ruess, R. W. (2001) The roots of the matter. Nature (London) 411: 749-752.
14       Chapin, F. S., Ill; Bloom, A. J.; Field, C. B.; Waring, R. H. (1987) Plant responses to multiple environmental
15             factors. BioScience 37: 49-57.
16       Chapin, F. S., Ill; Walker, B. H.; Hobbs, R. J.; Hooper, D. U.; Lawton, J. H.; Sala, O. E.; Tilman, D. (1997) Biotic
17             control over functioning ecosystems. Science 277:  500-504.
18       Chapin, F. S., Ill; Sala, O. E.; Burke, I. C.; Grime, J. P.; Hooper, D. U.; Lauenroth, W. K.; Lombard, A.; Mooney,
19             H. A.; Mosier, A. R.; Naeem, S.; Pacala, S. W.; Roy, J.; Steffen, W. L.; Tilman, D. (1998) Ecosystem
20             consequences of changing biodiversity. BioScience 48: 45-52.
21       Charlson, R. J.; Lovelock, J. E.; Andreae, M. O.; Warren, S. G. (1987) Oceanic phytoplankton, atmospheric
22             sulfphur, cloud albedo and climate. Nature (London) 326: 655-661.
23       Charlson, R. J.; Schwartz, S. E.; Hales, J.  M.; Cess, R. D.; Coakley, J. A., Jr.;  Hansen, J. E.; Hofmann, D. J. (1992)
24             Climate forcing by anthropogenic aerosols. Science (Washington, DC)  255: 423-430.
25       Chestnut, L. (1997) Methodology for estimating values for changes in visibility at national parks [draft
26             memorandum]. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality
27             Planning and Standards; contract no. 68-D-98-001; EPA air docket A-96-56, document no. VI-B-09-(ooo).
28       Chestnut, L. G.; Dennis, R. L. (1997) Economic benefits of improvements in visibility: acid rain. Provisions of the
29             1990 clean air act amendments. J. Air Waste Manage.  Assoc. 47: 395-402.
30       Chestnut, L. G.; Rowe, R. D. (1990) Preservation values for visibility protection at the national parks (draft final
31             report). Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning
32             and Standards, Economic Analysis Branch.
33       Chevone, B. I.; Herzfeld, D.  E.; Krupa, S. V.; Chappelka, A.  H. (1986) Direct effects of atmospheric sulfate
34             deposition on vegetation. J. AirPollut. Control Assoc. 36: 813-815.
3 5       Choudhury, B. J. (1987) Relationships between vegetation indices, radiation absorption, and net photosynthesis
36             evaluated by a sensitivity analysis.  Remote Sens. Environ. 22: 209-233.
37       Chow, J. C.; Watson, J. G.; Pritchett, L.; Lowenthal, D.; Frazier, C.; Neuroth,  G.; Evans, K.; Moon, D. (1990)
3 8             Wintertime visibility in Phoenix, Arizona. Presented at: the 83rd annual meeting & exhibition; June;
39             Pittsburgh, PA. Pittsburgh, PA: Air & Waste Management Association; paper no. 90-66.6.
40       Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Richards, L. W. (2002a) Comparability between PM25 and particle
41             light scattering measurements. Environ. Monit. Assess. 79: 29-45.
42       Chow, J. C.; Bachmann, J. D.; Wierman, S. S. G.; Mathai, C. V.; Malm, W. C.; White, W. H.; Mueller, P. K.;
43             Kumar, N.; Watson, J. G. (2002b) Visibility: science and regulation. J.  Air Waste Manage. Assoc.
44             52: 973-999.
45       Chughtai, A. R.; Williams, G. R.; Atteya,  M. M. O.; Miller, N. J.; Smith,  D. M. (1999) Carbonaceous particle
46             hydration. Atmos. Environ. 33: 2679-2687.
47       Cione, N. K.; Padgett, P. E.;  Allen, E. B. (2002) Restoration of a native shrubland impacted by exotic grasses,
48             frequent fire and nitrogen deposition in southern California. Restor. Ecol. 10: 376-384.
49       Clarke, J. F.; Edgerton, E. S.; Martin, B. E. (1997) Dry deposition calculations for the clean air status and trends
50             network. Atmos. Environ. 31: 3667-3678.
51       Clarkson, D. T.; Sanderson, J. (1971) Inhibition of the uptake and long-distance transport of calcium by aluminum
52             and other polyvalent cations. J. Exp. Bot. 22: 837-851.
53       Clevenger, T. E.; Saiwan, C.; Koirtyohann, S. R. (1991) Lead speciation of particles on air filters collected in the
54             vicinity of a lead smelter. Environ.  Sci. Technol. 25: 1128-1133.
55       Clough, W. S. (1975) The deposition of particles on moss and grass surfaces. Atmos. Environ. 9: 1113-1119.
56       Cocker, D. R., Ill; Whitlock, N. E.; Flagen, R. C.; Seinfield, J. H. (2001)  Hygroscopic properties of Pasadena,
57             California aerosol. Aerosol Sci. Technol. 35: 637-647.


         June 2003                                       4-243         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Cole, D. W.; Johnson, D. W. (1977) Atmospheric sulfate additions and cation leaching in a Douglas fir ecosystem.
  2             Water Resour. Res. 13:  313-317.
  3       Collett, J. L., Jr.; Hoag, K. J.; Sherman, D. E.; Bator, A.; Richards, L. W. (1999) Spatial and temporal variations in
  4             San Joaquin Valley fog chemistry. Atmos. Environ. 33: 129-140.
  5       Cooke, R. U.; Gibbs, G. B. (1994) Crumbling heritage?  Studies of stone weathering in polluted atmospheres. Atmos.
  6             Environ. 28: 1355-1356.
  7       Costanza, R.; d'Arge, R.; De Groot, R.; Farber, S.; Grasso, M; Harmon, B.; Limburg, K.; Naeem, S.; O'Neill, R. V.;
  8             Paruelo, J.; Raskin, R. G.; Sutton, P.; Van Den Belt, M. (1997) The value of the world's ecosystem services
  9             and natural capital. Nature (London) 387: 253-259.
10       Cotrufo, M. F.; De Santo, A. V; Alfani,  A.; Bartoli, G.;  De Cristofaro, A. (1995) Effects of urban heavy metal
11             pollution on organic matter decomposition in Quercus ilex L. woods. Environ. Pollut. 89: 81-87.
12       Cowling, J. E.; Roberts, M. E.  (1954) Paints, varnishes,  enamels, and lacquers. In: Greathouse,  G. A.; Wessel, C. J.,
13             eds. Deterioration of materials:  causes and preventive techniques. New York, NY: Reinhold Publishing
14             Corporation; pp. 596-645.
15       Cramer, S. D.; McDonald, L. G.; Bhagia, G.; Flinn, D. R.; Linstrom, P. J.; Carter, J. P. (1989) Effects of acid
16             deposition on the atmospheric corrosion of structural metals. Research Triangle Park, NC: U.S.
17             Environmental Protection Agency; internal report. Incorporated in: Irving, P. M., ed. Acidic Deposition: State
18             of Science and Technology. Volume III: Terrestrial, Materials, Health and Visibility Effects; report 19,
19             section 2. Washington, DC: National Acid Precipitation Assessment Program.
20       Creighton, P. J.; Lioy, P. J.; Haynie, F. H.; Lemmons, T. J.; Miller, J. L.; Gerhart, J. (1990) Soiling by atmospheric
21             aerosols in an urban industrial area. J. Air Waste Manage. Assoc. 40:  1285-1289.
22       Cronan, C. S.;  Grigal, D. F.  (1995) Use of calcium/aluminum ratios as indicators of stress in forest ecosystems.
23             J. Environ. Qual. 24:  209-226.
24       Cullen, J. J.; Neale, P. J. (1997) Biological weighting functions for describing the effects of ultraviolet radiation on
25             aquatic  ecosystems. In:  Hader, D.-P., ed. The effects of ozone depletion on aquatic ecosystems. Austin, TX:
26             Academic Press, pp. 97-118.
27       Dahlgren, R. A. (1994) Soil acidification and nitrogen saturation from weathering of ammonium-bearing rock.
28             Nature 368:  838-841.
29       Daily, G. C. (1997) Introduction: what are ecosystem services? In: Daily, G. C., ed. Nature's services: societal
30             dependence  on natural ecosystems. Washington, DC: Island Press; pp. 1-10.
31       Daily, G. C.; Ehrlich, P. R. (1999) Managing Earth's  ecosystems: an  interdisciplinary challenge. Ecosystems
32             2:277-280.
33       Darley, E. F. (1966) Studies on the effect of cement-kiln dust on vegetation.  J. Air Pollut. Control Assoc.
34             16: 145-150.
3 5       Dasch, J. M. (1987) Measurement of dry deposition to surfaces in deciduous and pine canopies. Environ. Pollut.
36             44:261-277.
37       Dasch, J. M.; Cadle, S.  H. (1985) Wet and dry deposition monitoring in southeastern Michigan. Atmos. Environ.
38             19:789-796.
39       Dasch, J. M.; Cadle, S.  H. (1986) Dry  deposition to snow in an urban area. Water Air Soil Pollut. 29: 297-308.
40       Dassler, H.-G.; Ranft, H.; Rehn, K.-H. (1972) Zur Widerstandsfahigkeit von Geholzen gegentiber
41             Fluorverbindungen und Schwefeldioxid [The susceptibility of woody  plants exposed to fluorine compounds
42             and SO2]. Flora (Jena) 161: 289-302.
43       Davidson, C. L; Friedlander, S. K. (1978) A filtration model for aerosol dry deposition: application to trace metal
44             deposition from the atmosphere. J. Geophys. Res. 83: 2343-2352.
45       Davidson, C. L; Wu, Y.-L.  (1989) Dry deposition of trace elements. In: Pacyna, J.M.; Ottar, B., eds. Control and fate
46             of atmospheric trace metals: proceedings of the NATO advanced research workshop on fate and control of
47             toxic metals in the atmosphere;  September 1988;  Oslo, Norway. Boston, MA: Kluwer Academic Publishers;
48             pp. 147-202. (NATO ASI series C, mathematical  and physical sciences: v. 268).
49       Davidson, C. L; Wu, Y.-L.  (1990) Dry deposition of particles and vapors. In: Lindberg, S. E.; Page, A. L.; Norton,
50             S. A., eds. Acidic precipitation: v. 3, sources, deposition,  and  canopy interactions. New York, NY:
51             Springer-Verlag; pp.  103-216.
52       Davidson, C. I.; Miller, J. M.; Pleskow, M. A. (1982) The influence of surface structure on predicted particle dry
53             deposition to natural grass canopies. Water Air Soil Pollut. 18: 25-43.
54       Davidson, C. I.; Tang, W.; Finger, S.; Etyemezian, V.; Striegel, M. F.; Sherwood, S. I. (2000) Soiling patterns on a
55             tall limestone building:  changes over 60 years. Environ. Sci. Technol. 34: 560-565.
56       Davis, G. D.; Shaw, B.  A.; Arah, C.  O.; Fritz, T. L.; Moshier, W. C.; Simpson, T. C.; Moran, P. J.; Zankel, K. L.
57             (1990) Effects of SO2 deposition on painted steel  surfaces. Surf. Interface Anal. 15: 107-112.


         June 2003                                      4-244         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Day, D. E.; Malm, W. C.; Kreidenweis, S. M; Tree, R. M. (1997) Aerosol light scattering measurements:
  2             a comparison of differently configured optec nephelometers. In: Visual air quality: aerosols and global
  3             radiation balance: proceedings of a specialty conference sponsored by the Air & Waste Management
  4             Association and the American Geophysical Union; September; Bartlett, NH. Pittsburgh, PA: Air & Waste
  5             Management Association; pp. 952-962. (A&WMA publication VTP-76).
  6       DeHayes, D. H.; Schaberg, P. G.; Hawley, G. I; Strimbeck, G. R. (1999) Acid rain impacts on calcium nutrition and
  7             forest health. Bioscience 49: 789-800.
  8       Del Monte, M.; Vittori, O. (1985) Air pollution and stone decay: the case of Venice. Endeavour 9: 117-122.
  9       Diakumaku, E.; Gorbushina, A. A.; Krumbein, W. E.; Panina, L.; Soukharjevski, S.  (1995) Black fungi in marble
10             and limestones~an aesthetical, chemical and physical problem for the conservation of monuments.
11             In: Saiz-Jimenez, C., ed. The deterioration of monuments: proceedings of the 2nd international symposium on
12             biodeterioration and biodegradation; January 1994; Sevilla, Spain. Sci. Total Environ.  167: 295-304.
13       Dickerson, R. R.; Kondragunta, S.; Stenchikov, G.; Civerolo, K. L.; Doddridge, B. G.; Holben, B. N. (1997)
14             The impact of aerosols on solar ultraviolet radiation and photochemical smog. Science (Washington, DC)
15             278:827-830.
16       Dobbins, R. A.; Mulholland, G. W.; Bryner, N. P. (1994) Comparison of a fractal smoke optics model with light
17             extinction measurements. Atmos. Environ. 28: 889-897.
18       Dolske, D. A. (1988) Dry deposition of airborne sulfate and nitrate to soybeans. Environ. Pollut.  53: 1-12.
19       Dolske, D. A. (1995) Deposition of atmospheric pollutants to monuments, statues, and buildings. Sci. Total Environ.
20             167: 15-31.
21       Dolske, D. A.; Gatz, D. F. (1984) Field intercomparison of sulfate dry deposition monitoring and measurement
22             methods: preliminary results. In: Hicks, B. B., ed.  Deposition both wet and dry: [papers presented at the
23             American Chemical Society acid rain symposium]; March 1982; Las Vegas, NV. Boston,  MA: Butterworth
24             Publishers; pp. 121-131. (Teasley, J. L, ed. Acid precipitation series: v. 4).
25       Draaijers, G. P. J.; Erisman, J.  W. (1993) Atmospheric sulphur deposition to forest stands: throughfall estimates
26             compared to estimates from inference.  Atmos. Environ. Part A 27: 43-55.
27       Draaijers, G. P. J.; Ivens, W. P. M.  F.; Bleuten, W. (1988) Atmospheric deposition in forest edges measured by
28             monitoring canopy throughfall. Water Air Soil Pollut. 42: 129-136.
29       Draaijers, G. P. J.; van Ek, R.;  Meijers, R. (1992) Research on the impact of forest stand structure on atmospheric
30             deposition. Environ. Pollut. 75: 243-249.
31       Dreisinger, B. R.; McGovern, P. C. (1970) Monitoring atmospheric sulphur dioxide  and correlating its effects  on
32             crops and forests in the  Sudbury area. In: Linzon,  S. N, ed. Impact of air pollution on vegetation conference;
33             April; Toronto, ON, Canada. Pittsburgh, PA: Air Pollution Control Association.
34       Driscoll, C. T.; Van Dreason, R. (1993) Seasonal and long-term temporal patterns in the  chemistry of Adirondack
3 5             lakes. Water Air Soil Pollut. 67: 319-344.
36       Driscoll, C. T.; Wyskowski, B. J.; DeStaffan, P.; Newton, R. M. (1989) Chemistry and transfer of aluminum in a
3 7             forested watershed in the Adirondack region of New York, USA. In: Lewis, T. E., ed. Environmental
38             chemistry and toxicology of aluminum. Chelsea, MI: Lewis Publishers, Inc.; pp. 83-105.
39       Driscoll, C. T.; Newton, R. M.; Gubala, C. P.; Baker, J. P.; Christensen, S. W. (1991) Adirondack mountains.
40             In: Charles, D. F., ed. Acidic deposition and aquatic ecosystems: regional case studies. New York, NY:
41             Springer-Verlag; pp. 13 3 -202.
42       Driscoll, C. T.; Lawrence, G. B.; Bulger, A. J.; Butler,  T. J.; Cronan, C. S.; Eagar, C.; Lambert, K. F.; Likens,  G. E.;
43             Stoddard, J. L.; Weathers, K. C. (2001) Acidic deposition in the northeastern United States: sources and
44             inputs, ecosystem effects, and management strategies. BioScience 51: 180-198.
45       Driscoll, C. T.; Whitall, D.; Aber, J.; Boyer, E.; Castro, M.; Cronan, C.; Goodale, C. L.; Groffman, P.; Hopkinson,
46             C.; lambert, K.; Lawrence, G.; Ollinger, S. (2003) Nitrogen pollution in the northeastern United States:
47             sources, effects, and management options. Bioscience 53: 357-374.
48       Duan, B.; Fairall, C. W.; Thomson, D. W. (1988) Eddy correlation measurements of the dry deposition of particles
49             in wintertime. J. Appl. Meteorol. 27: 642-652.
50       Duyzer, J. H.; Bouman, A. M.  M.; Van Aalst, R. M.; Diederen, H. S. M. A. (1987) Assessment of dry deposition
51             fluxes of NH3 and NH4+ over natural terrains. In: Asman, W. A. H.; Diederen, H. S. M. A., eds. Ammonia and
52             acidification: proceedings  of a symposium of the European Association for the Science of Air Pollution
53             (EURASAP); April; Bilthoven, The Netherlands. European Association for the  Science of Air Pollution;
54             pp. 97-106.
55       Eaton, J. S.; Likens, G. E.; Bormann, F. H. (1973) Throughfall and stemflow chemistry in a northern hardwood
56             forest. J. Ecol. 61: 495-508.
         June 2003                                      4-245        DRAFT-DO NOT QUOTE OR CITE

-------
  1       Edgerton, E. S.; Lavery, T. F.; Boksleitner, R. P. (1992) Preliminary data from the USEPA dry deposition network:
  2             1989. Environ. Pollut. 75: 145-156.
  3       Edgerton-Warburton, L. M; Allen, E. B. (2000) Shifts in arbuscular mycorrhizal communities along an
  4             anthropogenic gradient nitrogen deposition gradient. Ecol. Appl. 10: 484-496.
  5       Edney, E. O. (1989) Paint coatings: controlled field and chamber experiments. Research Triangle Park, NC:
  6             U.S. Environmental Protection Agency, Atmospheric Research and Exposure Assessment Laboratory; report
  7             no. EPA/600/3-89/032. Available from: NTIS, Springfield, VA; PB89-189849.
  8       Edney, E. O.; Stiles, D. C; Corse, E. W.; Wheeler, M. L.; Spence, J. W.; Haynie, F. H.; Wilson, W. E., Jr. (1988)
  9             Field study to determine the impact of air pollutants on the corrosion of galvanized steel. Mater. Perform.
10             27:47-50.
11       Edney, E. O.; Cheek, S. F.; Corse, E. W.; Spence, J.  W.; Haynie, F. H. (1989) Atmospheric weathering caused by
12             dry  deposition of acidic compounds. J. Environ. Sci. Health Part A 24: 439-457.
13       Edwards, R. S.; Claxton,  S. M. (1964) The distribution of air-borne salt of marine origin in the Aberystwyth area.
14             J. Appl. Ecol. 1:253-263.
15       El-Shobokshy, M. S. (1985) The dependence of airborne paniculate deposition on atmospheric stability and surface
16             conditions. Atmos. Environ. 19: 1191-1197.
17       Eldering, A.; Cass,  G. R.; Moon, K. C. (1994) An air monitoring network using continuous particle size distribution
18             monitors: connecting pollutant properties to visibility via Mie scattering calculations. Atmos. Environ.
19             28:2733-2749.
20       Eldering, A.; Hall, J. R.; Hussey, K. J.; Cass, G. R. (1996) Visibility model based on satellite-generated landscape
21             data. Environ. Sci. Tech.  30: 361-370.
22       Eldred, R.  A.; Cahill, T. A. (1994) Trends in elemental concentrations of fine particles at remote sites in the
23             United States of America. Atmos. Environ. 28: 1009-1019.
24       Eldred, R.  A.; Cahill, T. A.; Feeney, P. J. (1993) Comparison of independent measurements of sulfur and sulfate in
25             the IMPROVE network. Presented at: 86th annual meeting and exhibition of the Air & Waste Management
26             Association; June; Denver, CO. Pittsburgh, PA: Air & Waste Management Association; paper no.
27             93-RA-l 10.02.
28       Eldred, R.  A.; Cahill, T. A.; Flocchini, R. G. (1997)  Composition of PM2 5 and PM10 aerosols in the IMPROVE
29             network. In:  Special issue on aerosols and atmospheric optics-part I, [published papers from the international
30             specialty conference; September  1994; Snowbird, UT]. J. Air Waste Manage. Assoc. 47: 194-203.
31       Eliason, S. A.; Allen, E. B. (1997) Exotic grass competition in suppressing native shrubland re-establishment.
32             Restor. Ecol. 5:245-255.
33       Ellenberg, H. (1987) Floristic changes due to eutrophication. In: Asman, W. A. H.; Diederen, S. M. A., eds.
3 4             Ammonia and acidification: proceedings of a  symposium of the European Association for the Science of Air
3 5             Pollution (EURASAP); April; Bilthoven, The Netherlands. European Association for the Science of Air
36             Pollution; pp. 301-308.
37       Eller, B. M. (1977) Road dust induced increase of leaf temperature. Environ. Pollut. 13: 99-107.
38       Eriksson, P.; Johansson, L.-G.; Strandberg, H. (1993) Initial stages of copper corrosion in humid air containing SO2
39             and NO2. J. Electrochem. Soc.  140: 53-59.
40       Erisman, J. W. (1993) Acid deposition onto nature areas in the Netherlands; part II. Throughfall measurements
41             compared to deposition estimates. Water Air Soil Pollut. 71: 81-99.
42       Erisman, J.-W.; Vermetten, A. W. M.; Asman, W. A. H.; Waijers-IJpelaan, A.; Slanina, J. (1988) Vertical
43             distribution of gases and aerosols: the behaviour of ammonia and related components in the  lower
44             atmosphere.  Atmos. Environ. 22: 1153-1160.
45       Erisman, J. W.; Van Pul,  A.; Wyers, P. (1994) Parametrization of surface resistance for the quantification of
46             atmospheric  deposition of acidifying pollutants and ozone. Atmos. Environ. 28: 2595-2607.
47       Erisman, J. W.; Draaijers, G.; Duyzer, J.; Hofschreuder, P.; VanLeeuwen, N.; Romer, F.; Ruijgrok, W.;  Wyers, P.;
48             Gallagher, M. (1997) Particle deposition to forests-summary of results and application. Atmos. Environ.
49             31:321-332.
50       Erlick, C.; Frederick, J. E.; Saxena, V. K.; Wenny, B. N. (1998) Atmospheric transmission in the ultraviolet and
51             visible: aerosols in cloudy atmospheres. J. Geophys. Res. 103: 31,541-31,556.
52       Eshleman, K. N.; Miller-Marshall, L. M.; Webb, J. R. (1995) Long-term changes in episodic acidification of streams
53             in Shenandoah National Park, Virginia (U.S.A.). Water Air Soil Pollut. 85: 517-522.
54       Estupinan, J. B.; Raman,  S.; Crescenti, G. H.; Streicher, J. J.; Barnard, W. F. (1996) Effects of clouds and haze on
55             UV-B radiation. J. Geophys. Res. 101: 16,807-16,816.
         June 2003                                      4-246        DRAFT-DO NOT QUOTE OR CITE

-------
  1       Evans, E. G.; Pitchford, M. (1991) Federal visibility monitoring in the eastern United States. Presented at:
  2             84th annual meeting and exhibition of the Air & Waste Management Association; June; Vancouver, BC,
  3             Canada. Pittsburgh, PA: Air & Waste Management Association; paper no. 91-49.6.
  4       Fahey, T. J.; Yavitt, J. B.; Joyce, G. (1988) Precipitation and throughfall chemistry mPinus contorta ssp. latifolia
  5             ecosystems, southeastern Wyoming. Can. J. For. Res. 18: 337-345.
  6       Falke, S. (2001) Web-based visibility information system. Available:
  7             http://capita.wustl.edu/datawarehouse/datasets/ASOS/WebVIS.htm (28 May 2003).
  8       Falkengren-Grerup, U. (1987) Long-term changes in pH of forest soils in southern Sweden. Environ. Pollut.
  9             43:79-90.
10       Falkengren-Grerup, U. (1998) Nitrogen response of herbs and graminoids in experiments with simulated acid soil
11             solution. Environ. Pollut. 102(suppl.  1): 93-99.
12       Farber, R. J.; Welsing, P. R.; Rozzi, C. (1994) PM10 and ozone control strategy to improve visibility in the
13             Los Angeles Basin. Atmos. Environ. 28: 3277-3283.
14       Farmer, A. M. (1993) The effects of dust on vegetation~a review. Environ. Pollut. 79: 63-75.
15       Fassina, V.; Favaro, M.; Crivellari, F.; Naccari, A. (2001) The  stone decay of monuments in relation to atmospheric
16             environment. Ann. Chim. 91: 767-774.
17       Federal Register. (1999) Regional haze regulations; final rule. F. R. (July 1) 64: 35,713-35,774.
18       Fenn, M. E.; Keifer, J. W.  (1999) Throughfall deposition of nitrogen and sulfur in a Jeffrey pine forest in the
19             San Gabriel Mountains, Southern California. Environ. Pollut. 104: 179-187.
20       Fenn, M. E.; Poth, M. A. (1998) Indicators of nitrogen status in California forests. U.S.D.A. Forest Service Gen.
21             Tech. Rep. PSW-GTR-166.
22       Fenn, M. E.; Poth, M. A. (1999) Temporal and spatial trends in streamwater nitrate concentrations in the
23             San Bernardino Mountains, southern California. J. Environ. Qual. 28: 822-836.
24       Fenn, M. E.; Poth, M. A.; Johnson, D. W. (1996) Evidence for nitrogen saturation in the San Bernardino Mountains
25             in southern California. For. Ecol. Manage. 82: 211-230
26       Fenn, M. E.; Poth, M. A.; Aber, J. D.; Baron, J. S.; Bormann, B. T.; Johnson, D. W.; Lemly, A. D.; McNulty, S. G.;
27             Ryan, D. F.; Stottlemyer, R. (1998) Nitrogen excess in North American ecosystems: predisposing factors,
28             ecosystem responses, and management strategies. Ecol.  Appl. 8: 706-733.
29       Fenn, M. E.; Haeuber, R.; Tonnesen, G. S.; Baron, J. S.; Grossman-Clarke, S.; Hope, D.; Jaffe, D. A.; Copeland, S.;
30             Geiser, L.; Rueth, H. M.; Sickman, J. O. (2003a) Nitrogen emissions, deposition, and monitoring in the
31             western United States. Bioscience 53: 391-403.
32       Fenn, M. E.; Baron, J. S.; Allen, E. B.; Rueth, H. M.; Nydick, K. R.; Geiser, L.; Bowman, W. D.; Sickman, J. O.;
33             Meixner, T.; Johnson, D. W.; Neitlich, P. (2003b) Ecological effects of nitrogen deposition in the western
34             United States. Bioscience 53: 404-420.
35       Fink, F. W.; Buttner, F. H.; Boyd, W. K. (1971) Final report on technical-economic evaluation of air-pollution
36             corrosion costs on metals in the U.S. Columbus, OH: Battelle Memorial Institute, Columbus Laboratories;
37             EPA contract no. CPA 70-86; EPA report no. APTD-0654. Available from: NTIS, Springfield, VA;
38             PB-198-453.
39       Flum, T.; Nodvin, S. C. (1995) Factors affecting streamwater chemistry in the Great Smoky Mountains, USA. Water
40             Air Soil Pollut. 85:  1707-1712.
41       Fochtman, E. G.; Langer, G. (1957) Automobile paint damaged by airborne iron particles. J. Air Pollut. Control
42             Assoc. 6: 243-247.
43       Foltescu, V. L.; Isakson, J.; Selin, E.; Stikans, M. (1994) Measured fluxes of sulphur, chlorine and some
44             anthropogenic metals to the Swedish west coast. Atmos. Environ. 28: 2639-2649.
45       Foster, J. R. (1991) Effects of organic chemicals in the atmosphere on terrestrial plants. In: Moser, T. J.; Barker,
46             J. R.; Tingey, D. T., eds. Ecological exposure and effects of airborne toxic chemicals: an overview. Corvallis,
47             OR: U.S. Environmental Protection Agency, Environmental Research Laboratory; pp. 60-89; EPA report no.
48             EPA-600/3-91-001. Available from: NTIS, Springfield,  VA; PB91-148460.
49       Foster, N. W.; Nicolson, J. A.; Hazlett, P. W. (1989) Temporal variation in nitrate and nutrient cations in drainage
50             waters from a deciduous forest. J. Environ. Qual. 18: 238-244.
51       Fowler, D.; Cape, J. N.; Unsworth, M. H. (1989) Deposition of atmospheric pollutants on forests. Philos. Trans. R.
52             Soc. London B 324: 247-265.
53       Fowler, D.; Duyzer, J. H.; Baldocchi, D. D. (1991) Inputs of trace gases, particles and cloud droplets to terrestrial
54             surfaces. Proc. R. Soc. Edinburgh Sect. B: Biol. Sci. 97: 35-59.
55       Fox, D. G.; Malm, W. C.; Mitchell, B.; Fisher, R. W. (1999) Where there's fire, there's smoke: fine paniculate and
56             regional haze. EM November:  15-24.
         June 2003                                      4-247         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Frederick, J. E.; Qu, Z.; Booth, C. R. (1998) Ultraviolet radiation at sites on the Antarctic coast. J. Photochem.
  2             Photobiol. B 68: 183-190.
  3       Freedman, B.; Hutchinson, T. C. (1981) Sources of metal and elemental contamination of terrestrial environments.
  4             In: Lepp, N. W., ed. Effect of heavy metal pollution on plants. Volume 2: metals in the environment. London,
  5             United Kingdom: Applied Science Publishers; pp. 35-94.
  6       Freeman, A. M, III. (1993) The measurement of environmental and resource values: theory and methods.
  7             Washington, DC: Resources for the Future; pp. 1-27.
  8       Freedman, D. M.; Zahm, S. H.; Dosemeci, M. (1997) Residential and occupational exposure to sunlight and
  9             mortality from non-Hodgkin's lymphoma: composite (threefold) case-control study. Br. Med. J. BMJ
10             314:  1451-1455.
11       Freudenburg, W. R.; Alario, M. (1999) What ecologists can learn from nuclear scientists. Ecosystems 2: 286-291.
12       Fuller, K. A.; Malm, W. C.; Kreidenweis, S. M. (1999) Effects of mixing on extinction by carbonaceous particles.
13             J. Geophys. Res. 104: 15,941-15,954.
14       Ftirst, P.; Krause, G.  H. M.; Hein, D.; Delschen, T.; Winners, K. (1993) PCDD/PCDF in cow's milk in relation to
15             their levels in grass and soil. Chemosphere 27: 1349-1357.
16       Gaggi, C.;Bacci,E.; Calamari, D.; Fanelli, R. (1985) Chlorinated hydrocarbons in plant foliage: an indication of the
17             tropospheric contamination level. Chemosphere  14: 1673-1686.
18       Gallagher, M. W.; Choularton, T. W.; Morse, A. P.; Fowler, D. (1988) Measurements of the size dependence of
19             cloud droplet deposition at a hill site. Q. J. R. Meteorol.  Soc. 114: 1291-1303.
20       Galloway, J. N. (1998) The global nitrogen cycle: changes and consequences. Environ. Pollut. 102(suppl. 1): 15-24.
21       Galloway, J. N.; Cowling, E. B. (2002) Reactive nitrogen and the world: 200 years of change. Ambio 31: 64-71.
22       Galloway, J. N.; Aber, J. D.; Erisman, J. W.; Seitzinger, S. P.; Howarth, R. W.; Cowling, E. B.; Cosby, B. J. (2003)
23             The nitrogen cascade. Bioscience 53: 341-356.
24       Garcia-Valles, M.; Vendrell-Saz, M.; Molera,  J.; Blazquez, F. (1998) Interaction of rock and atmosphere: patinas on
25             Mediterranean monuments. Environ. Geol. 36(1-2): 137-149.
26       Garland, J. A. (1978) Dry and wet removal of sulphur from the atmosphere. Atmos. Environ. 12: 349-362.
27       Garland, C.; Garland, F. (1980) Do sunlight and vitamin D reduce the likelihood of colon cancer? Int. J. Epidemiol.
28             9:227-231.
29       Garland, F.  C.; Garland, C. F.; Gorham, E. D.; Young, J. F. (1990) Geographic variation in breast cancer mortality in
30             the United States: a hypothesis involving exposure to solar radiation. Prev. Med. 19: 614-622.
31       Garner, J. H. B. (1991) Ozone exposure and nitrogen loading in the forests of eastern North America. In: Berglund,
32             R. L.; Lawson, D. R.; McKee, D. J., eds. Tropospheric ozone and the environment: papers from an
33             international conference; March 1990; Los Angeles, CA. Pittsburgh, PA: Air & Waste Management
34             Association; pp. 289-310. (A&WMA transactions series no. TR-19).
35       Garner, J. H. B. (1994) Nitrogen oxides, plant metabolism, and forest ecosystem response. In: Alscher, R. G.;
36             Wellburn, A. R., eds. Plant responses to the gaseous environment: molecular, metabolic and physiological
37             aspects, [3rd international symposium on air pollutants and plant metabolism]; June 1992; Blacksburg, VA.
3 8             London, United Kingdom: Chapman & Hall; pp. 301-314.
39       Garner, J. H. B.; Pagano, T.; Cowling, E. B. (1989) An evaluation of the role of ozone, acid deposition, and other
40             airborne pollutants in the forests of eastern North America. Asheville, NC: U.S. Department of Agriculture,
41             Forest Service, Southeastern Forest Experiment Station; general technical report SE-59.
42       Garten, C. T., Jr. (1988) Fate and distribution of sulfur-35 in yellow poplar and red maple trees. Oecologia
43             76:43-50.
44       Garten, C. T., Jr.; Hanson, P. J. (1990) Foliar retention of 15N-nitrate and 15N-ammonium by red maple
45             (Acer rubrum) and white oak (Quercus alba) leaves from simulated rain. Environ. Exp. Bot. 30: 333-342.
46       Garten, C. T., Jr.; Bondietti, E. A.; Lomax, R.  D. (1988) Contribution of foliar leaching and dry deposition to sulfate
47             in net throughfall below deciduous trees. Atmos. Environ. 22: 1425-1432.
48       Gawel, J. E.; Ahner, B. A.; Friedland, A. J.; Morel, F. M. M. (1996) Role for heavy metals in forest decline
49             indicated by phytochelatin measurements. Nature 3 81: 64-65.
50       Gazzi, M.; Georgiadis, T.; Vicentini, V. (2001) Distant contrast measurements through fog and thick haze.
51             Atmos. Environ. 35: 5143-5149.
52       Geron, C.; Rasmussen, R.; Arnts, R. R.; Guenther, A. (2000) A review and synthesis of monoterpene speciation
53             from forests in the United States. Atmos. Environ. 34: 1761-1781.
54       Ghedini, N.; Gobbi, G.; Sabbioni, C.; Zappia,  G. (2000) Determination of elemental and organic carbon on damaged
55             stone monuments. Atmos. Environ. 34: 4383-4391.
         June 2003                                      4-248         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Gildon, A.; Tinker, P. B. (1983) Interactions of vesicular-arbuscular mycorrhizal infection and heavy metals in
  2             plants: I. the effects of heavy metals on the development of vesicular-arbuscular mycorrhizas. New Phytol.
  3             95:247-261.
  4       Gilliam, F. S.; Adams, M. B.; Yurish, B. M. (1996) Ecosystem nutrient responses to chronic nitrogen inputs at
  5             Fernow Experimental Forest, West Virginia. Can. J. For. Res. 26: 196-205.
  6       Gobbi, G.; Zappia, G.; Sabbioni, C. (1998) Sulphite quantification on damaged stones and mortars. Atmos.  Environ.
  7             32:783-789.
  8       Goodwin, J. E.; Sage, W.; Tilly, G. P. (1969) Study of erosion by solid particles. Proc. Inst. Mech. Eng.
  9             184:279-292.
10       Gorham, E. D.; Garland, C. F.; Garland, F. C. (1989) Acid haze air pollution and breast and colon cancer mortality
11             in 20 Canadian cities. Can. J. Public Health 80: 96-100.
12       Gorham, E. D.; Garland, F. C.; Garland, C. F. (1990) Sunlight and breast cancer incidence in the USSR. Int. J.
13             Epidemiol. 19: 820-824.
14       Goulder, L. H.; Kennedy, D. (1997) Valuing ecosystem services: philosophical bases and empirical methods. In:
15             Daily, G. C. Nature's Services: Societal Dependence on Natural Ecosystems. Washington, DC: Island Press;
16             pp. 23-47.
17       Graedel,  T. E.; Nassau, K.; Franey, J. P. (1987) Copper patinas formed in the atmosphere—I. introduction.
18             Corros. Sci. 27: 639-657.
19       Grand Canyon Visibility Transport Commission. (1996) Report of the Grand Canyon Visibility Transport
20             Commission to the United States Environmental Protection Agency: recommendations for improving western
21             vistas. Denver, CO: Grand Canyon Visibility Transport Commission.
22       Grant, L. D. (1989) Health effects issues associated with regional and global air pollution problems. In: Proceedings
23             of the changing atmosphere: implications for global security; June, 1988; Toronto, Canada.  Geneva,
24             Switzerland: World Meteorological Organization; report no. WMO-No. 710.
25       Grant, W. B. (2002) An estimate of premature cancer mortality in the U.S. due to inadequate doses of solar
26             ultraviolet-B radiation. Cancer 6: 1867-1875.
27       Grattan, S. R.; Maas, E. V.; Ogata, G. (1981) Foliar uptake and injury from saline aerosol. J. Environ. Qual.
28             10:406-409.
29       Gravenhorst, G.;  Hoefken, K. D.; Georgii, H. W. (1983) Acidic input to a beech and spruce forest. In: Beilke, S.;
30             Elshout, A. J., eds. Acid deposition: proceedings of the CEC workshop organized as part of the concerted
31             action "Physico-chemical behaviour of atmospheric pollutants;" September 1982; Berlin, Federal Republic of
32             Germany.  Dordrecht, The Netherlands: D. Reidel Publishing Company; pp. 155-171.
33       Great Lakes Regional Assessment Group (GLRAG).  (2000) Preparing for a changing climate: the potential
34             consequences of climate variability and change. Great Lakes overview. Washington, DC: U.S. Environmental
35             Protection Agency; Office of Research and Development; U.S. Global Change Research Program (USGCRP)
36             and Ann Arbor, MI: University of Michigan; Atmospheric, Oceanic and Space Sciences Department.
37       Greenberg, B. M.; Wilson, M. I.; Huang, X.-D.; Duxbury,  C. L.; Gerhardt, K. E.; Gensemer, R.  W. (1997)
38             The effects of ultraviolet-B radiation on higher plants. In: Wang, W.; Gorsuch, J. W.; Hughes, J. S., eds.
39             Plants for  Environmental Studies. New York, NY: Lewis Publishers; pp.  1-35.
40       Grennfelt, P. (1987) Deposition processes for acidifying compounds. Environ. Technol. Lett. 8:  515-527.
41       Grimm, N. B.; Grove, J. M.; Pickett, S.  T. A.; Redman, C.  L. (2000) Integrated approaches to long-term studies of
42             urban ecological systems. Bioscience 50: 571-584.
43       Gu, L. H.; Baldocchi, D.; Verma, S. B.; Black, T. A.; Vesala, T.; Falge, E. M.; Dowty, P. R. (2002) Advantages of
44             diffuse radiation for terrestrial ecosystem productivity. J. Geophys. Res. (Atmos.) 107(D6):
45             10.1029/2001JDOO1242.
46       Guderian, R. (1977) Accumulation of pollutants in plant organs. In: Air pollution: phytotoxicity of acidic gases and
47             its significance in air pollution control. Berlin, Germany: Springer-Verlag; pp. 66-74. (Ecological studies:
48             v.  22).
49       Guderian, R., ed. (1985)  Air pollution by photochemical oxidants: formation, transport, control, and effects on
50             plants. New York, NY: Springer-Verlag. (Billings, W. D.; Golley, F.; Lange, O. L.; Olson, J. S.;
51             Remmert,  H., eds. Ecological studies: analysis and synthesis: v. 52).
52       Guderian, R.; Tingey, D. T.; Rabe, R. (1985) Effects  of photochemical oxidants on plants. In: Guderian, R., ed. Air
53             pollution by photochemical oxidants: formation, transport, control, and effects on plants.  Berlin, Federal
54             Republic of Germany: Springer-Verlag; pp. 127-333. (Billings, W. D.; Golley, F.; Lange, O. L.; Olson, J.  S.;
55             Remmert,  H., eds. Ecological studies: analysis and synthesis, v. 52).
         June 2003                                      4-249         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Guderian, R. (1986) Terrestrial ecosystems: paniculate deposition. In: Legge, A. H.; Krupa, S. V., eds.
  2             Air pollutants and their effects on the terrestrial ecosystem. New York, NY: John Wiley & Sons; pp. 339-363.
  3             (Advances in environmental science and technology: v. 18).
  4       Gundersen, P.; Callesen, I.; De Vries, W. (1998a) Nitrate leaching in forest ecosystems is related to forest floor C/N
  5             ratios. Environ. Pollut. 102(suppl.  1): 403-407.
  6       Gundersen, P.; Emmett, B. A.; Kjonaas, O. I; Koopmans, C. I; Tietema, A.  (1998b) Impact of nitrogen deposition
  7             on nitrogen cycling in forests: a synthesis of NITREX data. For. Ecol. Manage. 101: 37-55.
  8       Hamilton, R. S.; Revitt, D. M; Vincent, K. I; Butlin, R. N. (1995) Sulphur and nitrogen paniculate pollutant
  9             deposition on to building surfaces. In: Saiz-Jimenez, C., ed. The deterioration of monuments: proceedings of
10             the 2nd international symposium on biodeterioration and biodegradation; January 1994; Sevilla, Spain. Sci.
11             Total Environ. 167: 57-66.
12       Hanchette, C. L.; Schwartz, G.  G. (1992) Geographic patterns of prostate cancer mortality. Cancer 70: 2861-2869.
13       Haneef, S. J.; Johnson, J. B.; Thompson, G. E.; Wood, G. C. (1993) Simulation of the degradation of limestones and
14             dolomitic sandstone under dry deposition conditions. In: Corrosion control for low-cost reliability:
15             preceedings [of the] 12th international corrosion congress. Volume 2:  process industries plant operation.
16             Houston, TX: National Association of Corrosion Engineers; pp. 700-710.
17       Hanley, N.; Spash, C. L. (1993) Cost-benefit analysis and the environment. Hants, United Kingdom: Edward Elgar.
18       Hansen, K.; Draaijers, G. P. J.; Ivens, W. P. M. F.; Gundersen, P.; vanLeeuwen, N. F. M. (1994) Concentration
19             variations in rain and canopy throughfall collected sequentially during individual rain events. Atmos. Environ.
20             28: 3195-3205.
21       Hansen, J. E.; Sato, M.; Ruedy, R. (1997) Radiative forcing and climate response. J. Geophys. Res. 102: 6831-6864.
22       Hanson, P. J.; Lindberg, S. E. (1991) Dry deposition of reactive nitrogen compounds: a review of leaf, canopy and
23             non-foliar measurements. Atmos. Environ. Part A 25: 1615-1634.
24       Hartge, P.; Devessa,  S. S.; Graumen, D.; Fears, T. R.; Fraumeni, J. F. (1996) Non-Hodgkin's lymphoma and
25             sunlight. J. Natl. Cancer Inst. 88: 298-300.
26       Harwell, M. A.; Myers, V; Young, T.; Bartuska, A.; Gassman, N.; Gentile, J. H.; Harwell, C. C.; Appelbaum, S.;
27             Barko, J.; Causey, B.; Johnson, C.; McLean, A.; Smola, R.; Templet, P.; Tosini, S. (1999) A framework for
28             an ecosystem integrity report card. BioScience 49: 543-556.
29       Hauhs, M. (1989) Lange Bramke: an ecosystem study of a forested catchment. In: Adriano, D. C.; Havas, M., eds.
30             Acidic precipitation: volume 1, case studies. New York, NY: Springer-Verlag; pp. 275-305. (Adriano, D. C.;
31             Salomons, W., eds. Advances in environmental science).
32       Haynie, F. H. (1976) Air pollution effects on stress induced intergranular corrosion of 7005-T53 aluminum alloy. In:
33             Stress corrosion-new approaches. Philadelphia, PA: American Society for Testing and Materials; pp. 32-43.
34             (ASTM special technical publication 610).
3 5       Haynie, F. H. (1980) Theoretical air pollution and climate effects on materials confirmed by zinc corrosion data.
36             In: Sereda, P. J.; Litvan, G.  G., eds. Durability of building materials and components:  proceedings of the
37             1st international conference; August 1978; Ottawa, ON, Canada. Philadelphia, PA: American Society for
38             Testing and Materials; pp. 157-175. (ASTM special technical publication 691).
39       Haynie, F. H. (1986) Environmental  factors affecting corrosion of weathering steel. In: Baboian, R., ed. Materials
40             degradation caused by acid rain: developed from the 20th state-of-the-art symposium of the American
41             Chemical Society; June 1985; Arlington, VA. Washington, DC: American Chemical Society; pp. 163-171.
42             (ACS symposium series 318).
43       Haynie, F. H.; Lemmons, T. J. (1990) Paniculate matter soiling of exterior paints at a rural site. Aerosol Sci.
44             Technol. 13: 356-367.
45       Haynie, F. H.; Upham, J. B. (1974) Correlation between corrosion behavior of steel and atmospheric pollution data.
46             In: Coburn, S. K., ed. Corrosion in natural environments: presented at the 76th annual meeting American
47             Society for Testing and Materials;  June 1973; Philadelphia, PA. Philadelphia, PA: American Society for
48             Testing and Materials; pp. 33-51. (ASTM special technical publication 558).
49       Haywood, J. M.; Shine, K. P. (1995) The effect of anthropogenic sulfate and soot aerosol on the clear sky planetary
50             radiation budget. Geophys. Res. Lett. 22: 603-606.
51       Haywood, J. M.; Ramaswamy, V.; Soden, B. J. (1999) Tropospheric aerosol  climate forcing in clear-sky satellite
52             observations over the oceans.  Science (Washington, DC) 283: 1299-1303.
53       Heal, G. (2000) Valuing ecosystem services. Ecosystems 3: 24-30.
54       Healey, K. D.; Rickert, K. G.; Hammer, G. L.; Bange, M. P. (1998) Radiation use efficiency increases when the
55             diffuse component of incident radiation is enhanced under shade. Aust. J. Agric. Res. 49: 665-672.
56       Heath, J. A.; Huebert, B. J. (1999) Cloudwater deposition as a source of fixed nitrogen in a Hawaiian montane
57             forest. Biogeochemistry 44: 119-134.


         June 2003                                     4-250         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Hedin, L. O.; Granat, L.; Likens, G. E.; Buishand, T. A.; Galloway, J. N.; Butler, T. I; Rodhe, H. (1994)
  2             Steep declines in atmospheric base cations in regions of Europe and North America. Nature (London)
  3             367: 351-354.
  4       Heichel, G. H.; Hankin, L. (1972) Particles containing lead, chlorine, and bromine detected on trees with an electron
  5             microprobe. Environ. Sci. Technol. 6: 1121-1122.
  6       Heichel, G. H.; Hankin, L. (1976) Roadside coniferous windbreaks as sinks for vehicular lead emissions. J. Air
  7             Pollut. Control Assoc. 26: 767-770.
  8       Heinsdorf, D. (1993) The role of nitrogen in declining Scots pine forests (Pinus sylvestris) in the lowland of
  9             East Germany. Water Air Soil Pollut. 69: 21-35.
10       Hemming, B. L.; Seinfeld, J. H. (2001) On the hygroscopic behavior of atmospheric organic aerosols. Ind. Eng.
11             Chem. Res. 40: 4162-4171.
12       Hermosin, B.; Saiz-Jimenez, C. (2000) Characterization of the organic compounds deposited on monument surfaces.
13             Fresenius' Environ. Bull. 9: 322-327.
14       Hess, A.; Iyer, H.; Malm, W. (2001) Linear trend analysis: a comparison of methods. Atmos. Environ.
15             35:5211-5222.
16       Hicks, B. B. (1986) Differences in wet and dry particle deposition parameters between North America and Europe.
17             In: Lee, S. D.; Schneider, T.; Grant, L.  D.; Verkerk, P.  J., eds. Aerosols: research, risk assessment and control
18             strategies; proceedings of the  second U.S.-Dutch international symposium; May 1985; Williamsburg, VA.
19             Chelsea, MI: Lewis Publishers; pp.  973-982.
20       Hicks, B. B.; Matt, D. R.; McMillen, R. T.; Womack, J. D.; Shelter, R. E. (1984) Eddy fluxes of nitrogen oxides to a
21             deciduous forest in complex terrain. In: Samson, P. J., ed. Transactions: the meteorology of acid deposition,
22             an APCA specialty conference; October 1983; Hartford, CT. Pittsburgh, PA: Air Pollution Control
23             Association; pp. 189-201.
24       Hicks, B. B.; Baldocchi, D. D.; Meyers, T. P.; Hosker, R. P., Jr.; Matt, D. R. (1987) A preliminary multiple
25             resistance routine for deriving dry deposition velocities from measured quantities. Water Air Soil Pollut.
26             36:311-330.
27       Hicks, B. B.; Matt, D. R.; McMillen, R. T.; Womack, J. D.; Wesely, M. L.; Hart, R. L.; Cook, D. R.; Lindberg,
28             S. E.; de Pena, R. G.; Thomson, D. W.  (1989) A field investigation of sulfate fluxes to a deciduous forest.
29             J. Geophys. Res.  [Atmos.] 94: 13,003-13,011.
30       Hidy, G. M.; Hales, J. M.; Roth, P. M.; Scheffe, R. (2000) Fine particles and oxidant pollution: developing an
31             agenda for cooperative research. J. Air Waste Manage. Assoc. 50: 613-632.
32       Hinckley, T.; Ford, D.; Segura, G.; Sprugel, D. (1992) Key processes from tree to stand level. In: Wall, G., ed.
33             Implications of climate change for Pacific Northwest forest management. Waterloo, ON, Canada: University
34             of Waterloo, Department of Geography; pp. 33-43. (Department of Geography publication series: occasional
35             paper no. 15).
36       Hofken, K. D.; Gravenhorst, G. (1982) Deposition of atmospheric aerosol particles to beech- and spruce forest. In:
37             Georgii, H.-W.; Pankrath, J., eds. Deposition of atmospheric pollutants: proceedings of a colloquium;
38             November 1981;  Oberursel/Taunus, Federal Republic of Germany. Dordrecht, The Netherlands: D. Reidel
39             Publishing Company; pp. 191-194.
40       Hofken, K. D.; Meixner, F. X.; Ehhalt, D. H. (1983) Deposition of atmospheric trace constituents onto different
41             natural surfaces. In: Pruppacher, H. R.; Semonin, R. G.; Slinn, W. G. N., eds. Precipitation scavenging,  dry
42             deposition, and resuspension:  v. 2, dry  deposition and resuspension: proceedings of the fourth international
43             conference; November-December 1982; Santa Monica, CA. New York, NY: Elsevier; pp. 825-835.
44       Hofstra, G.;Hall,R. (1971) Injury on roadside trees: leaf injury on pine and white cedar in relation to foliar levels of
45             sodium and chloride. Can. J. Bot. 49: 613-622.
46       Hogan, G. D.; Rennenberg, H.; Fink, S. (1998) Role and effect of sulfur in tree biology. In: Maynard, D. G., ed.
47             Sulfur in the environment. New York, NY:  Marcel Dekker, Inc.; pp.  173-217. (Books in soils, plants, and the
48             environment: v. 67).
49       Holbrow, G. L. (1962) Atmospheric  pollution: its measurement and some effects on paint. J. Oil Colour Chem.
50             Assoc. 45: 701-718.
51       Holland, D. M.; Principe, P. P.; Sickles, J. E., II. (1999) Trends in atmospheric sulfur and nitrogen species in the
52             eastern United States for 1989-1995. Atmos. Environ. 33: 37-49.
53       Hollander, M.; Wolfe, D. A. (1973) Nonparametric statistical methods. New York: John Wiley & Sons.
54       Holling, C. S. (1986) The resilience of terrestrial ecosystems:  local surprise and global change. In: Clark, W. C.;
55             Munn, R. E., eds. Sustainable development of the biosphere. Cambridge, UK: Cambridge University;
56             pp. 292-317.
         June 2003                                      4-251         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Hooper, D. U.; Vitousek, P. M. (1997) The effects of plant composition and diversity on ecosystem processes.
  2             Science 277: 1302-1305.
  3       Hornung, M.; Langan, S. J. (1999) Nitrogen deposition, sources, impacts and responses in natural and semi-natural
  4             ecosystems. In: Bangan, S. J., ed. Impact of nitrogen deposition on natural ecosystems and semi-natural
  5             ecosystems. Dordrect, Netherlands: Kluwer Academic Publishers; pp. 1-14. [Environmental Pollution, no.3].
  6       Horvath, H. (1993) Atmospheric light absorption-a review. Atrnos. Environ. Part A 27: 293-317.
  7       Hosker, R. P., Jr.; Lindberg, S. E. (1982) Review: atmospheric deposition and plant assimilation of gases and
  8             particles. Atmos. Environ. 16: 889-910.
  9       Houghton, J. T.; Jenkins, G. J.; Ephraums, J. J., eds. (1990) Climate change: the IPCC scientific assessment.
10             Cambridge, MA: Cambridge University Press; p. 55.
11       Hoyt, D. V. (1978) A model for the calculation of solar global insolation. Sol. Energy 21: 27-35.
12       Huebert, B. J.; Luke, W. T.; Delany, A. C.; Brost, R. A. (1988) Measurements of concentrations and dry surface
13             fluxes of atmospheric nitrates in the presence of ammonia. J. Geophys. Res. [Atmos.] 93: 7127-7136.
14       Hughes, M. K. (1981) Cycling of trace metals in ecosystems. In: Lepp, N. W., ed. Effect of heavy metal pollution on
15             plants. Volume 2: metals in the environment. London, United Kingdom: Applied Science Publishers;
16             pp. 95-118.
17       Humphreys, F. R.; Lambert, M. J.; Kelly, J. (1975)  The occurrence of sulphur deficiency in forests. In: McLachlan,
18             K. D., ed. Sulphur in Australasian agriculture. Sydney, Australia: Sydney University Press; pp. 154-162.
19       Hunsaker,  C.; Graham, R.; Turner, R. S.; Ringold, P. L.; Holdren, G.  R., Jr.; Strickland, T.  C. (1993)  A national
20             critical loads framework for atmospheric deposition effects assessment: II. Defining assessment end points,
21             indicators, and functional subregions. Environ. Manage. 17: 335-341.
22       Huntington, T.G. (2000) The potential for calcium depletion in forest ecosystems of southeastern United States:
23             review and analysis. Global Biogeochem. Cycles 14: 623-638.
24       Husar, R. B.; Wilson, W. E. (1993) Haze and sulfur emission trends in the eastern United States. Environ. Sci.
25             Technol. 27: 12-16.
26       Husar, R. B.; Holloway, J. M.; Patterson, D. E.; Wilson, W. E. (1981) Spatial and temporal pattern of eastern U.S.
27             haziness: a summary. Atmos. Environ.  15: 1919-1928.
28       Husar, R. B.; Elkins, J. B.; Wilson, W. E. (1994) U.S. visibility trends, 1960-1992, regional and national. Presented
29             at: 87th annual meeting of the Air & Waste Management Association; June; Cincinnati, OH. Pittsburgh, PA:
30             Air & Waste Management Association; paper no. 94-MP3.05.
31       Hutchinson, A. J.; Johnson, J. B.; Thompson,  G. E.; Wood, G. C.; Sage, P. W.; Cooke, M. J. (1992) The role of
32             fly-ash paniculate material and oxide catalysts in stone degradation. Atmos. Environ. 26A: 2795-2803.
33       Huttunen, S. (1994) Effects of air pollutants on epicuticular wax structure. In: Percy, K. E.; Cape, J. N.; Jagels, R.;
34             Simpson, C. J., eds. Air pollutants and  the leaf cuticle. New York, NY: Springer-Verlag; pp. 81-96.
35             (NATO ASI series).
36       Hyer, K. E.; Webb, J. R.; Eshleman, K. N. (1995) Episodic acidification of three streams in Shenandoah National
37             Park, Virginia, USA. Water Air Soil Pollut. 85:  523-528.
38       Ibrahim, M.; Barrie, L. A.; Fanaki, F. (1983) An experimental and theoretical investigation of the  dry deposition of
39             particles to snow, pine trees and artificial collectors. Atmos. Environ. 17: 781-788.
40       IMPROVE: interagency monitoring of protected visual environments [database]. (1998) [Data on gross chemical
41             composition of PM2.5 particles  in rural areas of the eastern and western United States as of 1998]. Fort
42             Collins, CO: National Park Service (NFS); Cooperative Institute for Research in the Atmosphere (CIRA).
43             Available at: http://vista.cira.colostate.edu/improve/ [2001, January 26].
44       Intergovernmental Panel on Climate Change (IPCC). (1998) The regional impacts of climate change:  an assessment
45             of vulnerability. Cambridge, United Kingdom: Cambridge University Press.
46       Intergovernmental Panel on Climate Change (IPCC). (2001a) Climate change 2001: the scientific basis. Contribution
47             of working group I to the third assessment report of the Intergovernmental Panel on  Climate Change.
48             Cambridge, United Kingdom: Cambridge University Press.
49       Intergovernmental Panel on Climate Change (IPCC). (200 Ib) Climate change 2001: impacts, adaptation, and
50             vulnerability. Contribution of working  group II to the third assessment report of the Intergovernmental Panel
51             on Climate Change. Cambridge, United Kingdom: Cambridge  University Press.
52       Irving, P. M., ed. (1991) Acidic deposition: state of science and technology:  volumes I-IV.  Washington, DC:
53             The U.S. National Acid Precipitation Assessment Program.
54       Iyer, H.; Patterson, P.; Malm, W. C. (2000) Trends  in the extremes of sulfur concentration distributions. J. Air Waste
55             Manage. Assoc. 50: 802-808.
56       Jacobson, M. Z. (1998) Studying the effects of aerosols on vertical photolysis rate coefficient and temperature
57             profiles over an urban airshed. J. Geophys. Res. [Atmos.] 103: 10,593-10,604.


         June 2003                                       4-252         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Jacobson, M. Z. (1999) Isolating nitrated and aromatic aerosols and nitrated aromatic gases as sources of ultraviolet
  2             light absorption. J. Geophys. Res. 104 (D3): 3527-3542.
  3       Jacobson, M. Z. (2001) Global direct radiative forcing due to multicomponent anthropogenic and natural aerosols.
  4             J. Geophys. Res. 106: 1551-1568.
  5       Jaworski, N. A.; Howarth, R. W.; Hetling, L. J. (1997) Atmospheric deposition of nitrogen oxides onto the landscape
  6             contributes to coastal eutrophication in the northeast United States. Environ. Sci. Technol. 31:  1995-2004.
  7       Jensen, V. (1974) Decomposition of angiosperm tree leaf litter. In: Dickinson, C. H.; Pugh, G. J. F., eds. Biology of
  8             plant litter decomposition: v. I. London, United Kingdom: Academic Press; pp. 69-104.
  9       John, W.; Wall, S. M.; Ondo, J. L. (1985) Dry acid deposition on materials and vegetation: concentrations in
10             ambient air. Berkeley, CA: California Department of Health Services, Air and Industrial Hygiene Laboratory;
11             report no. CA/DOH/AIHL/SP-34. Available from: NTIS, Springfield, VA; PB85-241206.
12       Johnson, D. W. (1984) Sulfur cycling in forests. Biogeochemistry 1: 29-43.
13       Johnson, D. W. (1992) Base cation distribution and cycling. In: Johnson, D. W.; Lindberg, S. E., eds. Atmospheric
14             deposition and forest nutrient cycling: a synthesis of the integrated forest study. New York, NY:
15             Springer-Verlag, Inc.; pp. 275-333. (Billings, W. D.; Golley, F.; Lange, O. L.; Olson, J. S.; Remmert, H., eds.
16             Ecological studies, analysis and synthesis:  v. 91).
17       Johnson, F. R.; Desvousges, W. H. (1997) Estimating stated preferences with rated-pair data: environmental, health,
18             and employment effects of energy programs. J. Environ. Econ. Manage. 34: 79-99.
19       Johnson, D. W.; Lindberg, S. E., eds. (1992a) Atmospheric deposition and forest nutrient cycling: a synthesis of the
20             integrated forest study. New York, NY: Springer-Verlag, Inc. (Billings, W. D.; Golley, F.; Lange, O. L.;
21             Olson, J. S.; Remmert, H., eds. Ecological studies: analysis and synthesis: v. 91).
22       Johnson, D. W.; Lindberg, S. E., eds. (1992b) Nitrogen chemistry, deposition, and cycling in forests.  In: Johnson,
23             D. W.; Lindberg, S. E., eds. Atmospheric deposition and forest nutrient cycling: a synthesis of the integrated
24             forest study. New York, NY: Springer-Verlag,  Inc.; pp.  150-213. (Billings, W. D.; Golley, F.; Lange, O. L.;
25             Olson, J. S.; Remmert, H., eds. Ecological studies: analysis and synthesis: v. 91).
26       Johnson, D. W.; Mitchell, M. J. (1998) Responses of forest ecosystems to changing sulfur inputs. In: Maynard,
27             D. G., ed. Sulfur in the environment.  New  York, NY: Marcel Dekker, Inc.; pp. 219-262. [Books in Soils,
28             Plants, and the Environment, v. 67].
29       Johnson, D. W.; Taylor, G. E. (1989) Role of air pollution in forest decline in eastern North America. Water Air Soil
30             Pollut. 48: 21-43.
31       Johnson, D. W.; Todd, D. E., Jr. (1998) Harvesting effects on long-term changes in nutrient pools of mixed oak
32             forest. Soil Sci. Soc. Amer. J. 62: 1725-1735.
33       Johnson, D. W.; Henderson, G. S.; Huff, D.  D.; Lindberg,  S. E.; Richter, D. D.; Shriner, D. S.; Todd, D. E.;
34             Turner, J. (1982a) Cycling of organic and inorganic sulphur in a chestnut oak forest. Oecologia 54: 141-148.
35       Johnson, D. W.; Cole, D. W.; Bledsoe, C. S.; Cromack, K.; Edmonds, R. L.; Gessel, S. P.; Grier, C. C.; Richards,
36             B. N.; Vogt, K. A. (1982b) Nutrient cycling in forests of the Pacific Northwest. In: Edmonds, R. L., ed.
3 7             Analysis of coniferous forest ecosystems in the Western United States. Stroudsburg, PA: Hutchinson Ross
38             Publishing Company; pp. 186-232. (US/IBP synthesis series: v. 14).
39       Johnson, D. W.; Van Miegroet, H.; Lindberg, S. E.; Todd, D. E.; Harrison, R. B.  (1991a) Nutrient cycling in red
40             spruce forests of the Great Smoky Mountains. Can. J. For. Res. 21: 769-787.
41       Johnson, D. W.; Cresser, M. S.; Nilsson, S. I.; Turner, J.; Ulrich, B.; Binkley, D.; Cole, D. W. (1991b) Soil changes
42             in forest ecosystems: evidence for and probable causes. Proc. R. Soc. Edinburgh Sect.  B: Biol. Sci.
43             97B: 81-116.
44       Johnson, D. W.; Swank, W.  T.; Vose, J. M.  (1993) Simulated effects of atmospheric sulfur deposition on nutrient
45             cycling in a mixed deciduous forest. Biogeochemistry 23: 169-196.
46       Johnson, A. H.; Friedland, A. J.; Miller, E. K.; Siccama, T. G. (1994) Acid rain and soils of the Adirondacks. III.
47             Rates of soil acidification in a  montane spruce-fir forest at Whiteface Mountain, New York. Can. J. For. Res.
48             24:663-669.
49       Johnson, D.W.; Susfalk, R.B.; Brewer, P.,F.; Swank, W.T. (1999) Simulated effects of reduced sulfur, nitrogen, and
50             base cation deposition on soils and solutions in southern Appalachian forests. J. Environ. Qua! 28: 1336-1346.
51       Joslin, J. D.; Wolfe, M. H. (1992) Red spruce soil chemistry and root distribution across  a cloud water deposition
52             gradient. Can. J. For. Res. 22:  893-904.
53       Joslin, J. D.; Wolfe, M. H. (1994) Foliar deficiencies of mature Southern Appalachian red spruce determined from
54             fertilizer trials. Soil Sci. Soc. Am. J. 58:  1572-1579.
55       Joslin, J. D.; Kelly, J. M.; Van Miegroet, H. (1992) Soil chemistry and nutrition of North American spruce-fir
56             stands: evidence for recent change. J. Environ.  Qual. 21: 12-30.
         June 2003                                      4-253         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Justus, C. G.; Murphey, B. B. (1994) Temporal trends in surface irradiance at ultraviolet wavelengths. J. Geophys.
  2             Res. 99 (Dl): 1389-1394.
  3       Kaminski, M. D.; Landsberger, S. (2000a) Heavy metals in urban soils of East St. Louis, IL, part I: total
  4             concentration of heavy metals in soils. J. Air Waste Manage. Assoc. 50: 1667-1679.
  5       Kaminski, M. D.; Landsberger, S. (2000b) Heavy metals in urban soils of East St. Louis, IL, part II: leaching
  6             characteristics and modeling. J. Air Waste Manage. Assoc. 50: 1680-1687.
  7       Karmoker, J. L.; Clarkson, D. T.; Saker, L. R.; Rooney, J. M.; Purves, J. V. (1991) Sulphate deprivation depresses
  8             the transport of nitrogen to the xylem and the hydraulic conductivity of barley (Hordeum vulgare Z.) roots.
  9             Planta 185: 269-278.
10       Kaupp, H.; McLachlan, M. S. (1999) Atmospheric particle size distributions of polychlorinated dibenzo-p-dioxins
11             and dibenzofurans (PCDD/Fs) and polycyclic aromatic hydrocarbons (PAHs) and their implications for wet
12             and dry deposition. Atmos. Environ. 33: 85-95.
13       Kelly, J.; Lambert, M. J. (1972) The relationship between sulphur and nitrogen in the foliage of Pinus radiata.
14             Plant Soil 37: 395-407.
15       Kelly, J. M.; Meagher, J. F. (1986) Nitrogen input/output relationships for three forested watersheds in eastern
16             Tennessee.  In: Correll, D. L., ed. Watershed research perspectives. Washington, DC: Smithsonian Institution
17             Press; pp. 360-391.
18       Kenk, G.; Fischer, H. (1988) Evidence from nitrogen fertilisation in the forests of Germany. Environ. Pollut.
19             54:  199-218.
20       Kiehl, J. T.; Briegleb, B. P. (1993) The relative roles of sulfate aerosols and greenhouse gases in climate forcing.
21             Science (Washington, DC) 260: 311-314.
22       Kim, E.; Kalman, D.; Larson, T. (2000) Dry deposition of large airborne particles on a surrogate surface. Atmos.
23             Environ. 34: 2387-2397.
24       Kleeman, M. J.; Eldering, A.; Hall, J. R.; Cass, G. R. (2001) Effect of emissions control programs on visibility in
25             southern California. Environ. Sci. Technol. 35: 4668-4674.
26       Klepper, B.; Craig, D. K. (1975) Deposition of airborne particulates onto plant leaves. J. Environ. Qual. 4: 495-499.
27       Kodama, Y.; Lee,  S. D., eds. (1994) The 13th UOEH international symposium and the 2nd pan Pacific cooperative
28             symposium on impact of increased UV-B  exposure on human health and ecosystem;  October 1993;
29             Kitakyushu, Japan. Kitakyushu, Japan: University of Occupational and Environmental Health.
30       Koslow, E. E.; Smith, W. H.; Staskawicz, B.  J. (1977) Lead-containing particles on urban leaf surfaces.
31             Environ. Sci. Technol.  11: 1019-1021.
32       Krajickova, A.; Mejstfik, V. (1984) The effect of fly-ash particles on the plugging of stomata. Environ. Pollut.
33             36:83-93.
34       Kripke, M. L. (1989) Health effects of stratospheric ozone depletion: an overview. In: Schneider, T.; Lee,  S. D.;
35             Wolters, G. J. R.; Grant, L. D. eds. Atmospheric ozone research and its policy implications: proceedings of
36             the Third US-Dutch International Symposium; May, 1988; Nijmegen,  The Netherlands. Amsterdam, The
37             Netherlands: Elsevier, pp. 795-802. [Studies in Environmental Science 35].
38       Krupa, S. V.; Legge, A. H. (1986) Single or joint effects of coarse and fine paniculate sulfur aerosols and ozone on
39             vegetation.  In: Lee, S. D.; Schneider, T.; Grant, L. D.; Verkerk, P. J., eds. Aerosols: research, risk assessment
40             and control strategies, proceedings of the second U.S.-Dutch international symposium; May 1985;
41             Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 879-887.
42       Krupa, S. V.; Legge, A. H. (1998) Sulphur dioxide, paniculate sulphur and its impacts on a boreal forest ecosystem.
43             In: Ambasht, R. S., ed.  Modern trends in ecology and environment. Leiden, The Netherlands: Backhuys
44             Publishers;  pp. 285-306.
45       Kylin, H.;  Grimvall, E.; Ostman, C. (1994) Environmental monitoring of polychlorinated biphenyls using pine
46             needles as passive samplers. Environ.  Sci. Technol. 28: 1320-1324.
47       Lamaud, E.; Chapuis, A.; Fontan, J.; Serie, E. (1994) Measurements and parameterization of aerosol dry deposition
48             in a semi-arid area. Atmos. Environ. 28: 2461-2471.
49       Lamersdorf, N. P.; Meyer, M. (1993) Nutrient cycling and acidification of a northwest German forest site with high
50             atmospheric nitrogen deposition. For. Ecol. Manage. 62: 323-354.
51       Laurenroth, W. K.; Milchunas, D. G. (1985) SO2 effects on plant community function. In: Winner, W. E.; Mooney,
52             H. A.; Goldstein, R. A., eds. Sulfur dioxide and vegetation: physiology, ecology, and policy issues. Stanford,
53             CA: Stanford University Press; pp. 454-477.
54       Lawrence, G. B.; Huntington, T. G. (1999) Soil-Calcium depletion linked to acid rain and forest growth in the
55             eastern United States. Reston, VA: U.S. Geological Survey. Available: http://btdqs.usgs.gov/acidrain [1999,
56             November 23].
         June 2003                                      4-254         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Lefkowitz, E. S.; Garland, C. F. (1994) Sunlight, vitamin D, and ovarian cancer mortality rates in U.S. women. Int.
  2             J. Epidemiol. 23: 1133-1136.
  3       Legge, A. H. (1980) Primary productivity, sulfur dioxide, and the forest ecosystem: an overview of a case study.
  4             In: Miller, P. R., ed. Proceedings of symposium on effects of air pollutants on Mediterranean and temperate
  5             forest ecosystems; June; Riverside, CA. Berkeley, CA: U.S. Department of Agriculture, Pacific Southwest
  6             Forest and Range Experiment Station; pp. 51-62; Forest Service general technical report no. PSW-43.
  7             Available from: NTIS, Springfield, VA; PB81-133720.
  8       Lerman, S. L.; Darley, E. F.  (1975) Particulates. In: Mudd, J. B.; Kozlowski, T. T., eds. Responses of plants to air
  9             pollution. New York, NY: Academic Press; pp. 141-158.
10       Leslie, M. (2001) Having a bad air day. Science 294: 2255.
11       Levin, S. A. (1998) Ecosystems and the biosphere as complex adaptive systems. Ecosystems 1: 431-436.
12       Li-Jones, X.; Prospero, J. M. (1998) Variations in the size distribution of non-sea-salt sulfate aerosol in the marine
13             boundary layer at Barbados: impact of African dust. J. Geophys. Res.  103: 16,073-16,084.
14       Likens, G. E.; Driscoll, C. T.; Buso, D. C. (1996) Long-term effects of acid rain: response and recovery of a forest
15             ecosystem. Science (Washington, DC) 272: 244-246.
16       Likens, G. E.; Driscoll, C. T.; Buso, D. C.; Siccama, T. G.; Johnson, C. E.; Lovett, G. M.; Fahey, T. J.; Reiners,
17             W. A.; Ryan, D. F.; Martin, C. W.; Bailey, S. W. (1998) The biogeochemistry of calcium at Hubbard Brook.
18             Biogeochemistry 41:  89-173.
19       Likens, G. E.; Butler, T. J.; Buso, D. C. (2001) Long- and short-term changes in sulfate deposition:  effects of the
20              1990 Clean Air Act Amendments. Biogeochemistry 52: 1-11.
21       Lin, J.-M.; Fang, G.-C.; Holsen, T. M.; Noll, K. E. (1993) A comparison of dry deposition modeled from size
22             distribution data and measured with a smooth surface for total particle mass, lead and calcium in Chicago.
23             Atmos. Environ. Part A 27: 1131-1138.
24       Lindberg, S. E. (1992) Atmospheric deposition and canopy interactions of sulfur. In: Johnson, D. W.; Lindberg,
25             S. E., eds. Atmospheric deposition and forest nutrient cycling: a synthesis of the integrated forest study.
26             New York, NY: Springer-Verlag, Inc.; pp.  74-90.  (Billings, W. D.; Golley, F.; Lange, O. L.;  Olson, J. S.;
27             Remmert, H., eds. Ecological studies: analysis and synthesis: v. 91).
28       Lindberg, S. E.; Garten, C. T., Jr.  (1988) Sources  of sulphur in forest canopy throughfall.  Nature (London)
29             336: 148-151.
30       Lindberg, S. E.; Harriss, R. C.  (1981) The role of atmospheric deposition in an eastern U.S. deciduous forest.
31             Water Air Soil Pollut. 16: 13-31.
32       Lindberg, S. E.; Lovett, G. M.  (1985) Field measurements of particle dry deposition rates to foliage and inert
33             surfaces in a forest canopy. Environ.  Sci. Technol. 19: 238-244.
34       Lindberg, S. E.; Lovett, G. M.  (1992) Deposition  and forest canopy interactions of airborne sulfur: results from the
35             integrated forest study. Atmos. Environ. Part A 26: 1477-1492.
36       Lindberg, S. E.; McLaughlin, S. B. (1986) Air pollutant interactions with vegetation: research needs in data
37             acquisition and interpretation. In: Legge, A. H.; Krupa, S. V., eds. Air pollutants and their effects on the
38             terrestrial ecosystem.  New York, NY: John Wiley & Sons, Inc.; pp. 449-503. (Nriagu, J. O., ed. Advances in
39             environmental science and technology: v. 18).
40       Lindberg, S. E.; Owens, J. G. (1993) Throughfall studies of deposition to forest edges and gaps in montane
41             ecosystems. Biogeochemistry 19: 173-194.
42       Lindberg, S. E.; Harriss, R. C.; Turner, R. R. (1982) Atmospheric deposition of metals to  forest vegetation. Science
43             (Washington, DC) 215: 1609-1611.
44       Lindberg, S. E.; Lovett, G. M.; Richter, D. D.; Johnson, D. W. (1986)  Atmospheric deposition and canopy
45             interactions of major ions in a forest.  Science (Washington, DC) 231:  141-145.
46       Lindberg, S. E.; Lovett, G. M.; Meiwes, K.-J. (1987) Deposition and forest canopy interactions of airborne nitrate.
47             In: Hutchinson, T. C.; Meema, K. M., eds.  Effects of atmospheric pollutants on forests, wetlands and
48             agricultural ecosystems. Berlin, Federal Republic  of Germany: Springer-Verlag; pp.  117-130. (NATO as
49              1 series: v. G16).
50       Lindberg, S. E.; Lovett, G. M.; Schaefer, D. A.; Bredemeier, M.  (1988) Dry deposition velocities and
51             surface-to-canopy scaling factors for aerosol calcium from forest canopy throughfall. J. Aerosol Sci.
52              19: 1187-1190.
53       Lindberg, S. E.; Bredemeier, M.; Schaefer, D. A.; Qi, L.  (1990) Atmospheric concentrations and deposition of
54             nitrogen and major ions in conifer forests in the United States and Federal Republic of Germany. Atmos.
55             Environ.  Part A 24: 2207-2220.
         June 2003                                      4-255         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Liousse, C.; Cachier, H.; Jennings, S. G. (1993) Optical and thermal measurements of black carbon aerosol content
  2             in different environments: variation of the specific attenuation cross-section, sigma (o). Atmos. Environ.
  3             Part A 27: 1203-1211.
  4       Little, P. (1977) Deposition of 2.75, 5.0 and 8.5 |im particles to plant and soil surfaces. Environ. Pollut. 12: 293-305.
  5       Little, P.; Wiffen, R. D. (1977) Emission and deposition of petrol engine exhaust Pb~I. deposition of exhaust Pb to
  6             plant and soil surfaces. Atmos. Environ. 11: 437-447.
  7       Liu, S. C.; McKeen, S. A.; Madronich, S. (1991) Effect of anthropogenic aerosols on biologically active ultraviolet
  8             radiation. Geophys. Res. Lett. 18: 2265-2268.
  9       Lloyd, J. (2002) After years of smog, a shroud lifts. The  Christian Science Monitor. June 4: l(s. 1).
10       Loehman, E.; Boldt, D.; Chaikin,K. (1985) Measuring the benefits of air quality improvements in the SanFrancisco
11             Bay area: property value and contingent valuation studies, volume IV of methods development for
12             environmental control benefits assessment. Washington, DC: U.S. Environmental Protection Agency; Office
13             of Policy, Planning and Evaluation; report no. EPA-230-12-85-022. Available:
14             http://yosemite.epa.gov/EE/epa/eerm.nsf/vwRepNumLookup/EE-0278D7OpenDocument (28 May 2003).
15       Lekke, H.; Bak, J.; Falkengren-Grerup, U.; Finlay, R. D.; Ilvesniemi, H.; Nygaard, P. H.; Starr, M. (1996)
16             Critical loads of acidic deposition for forest soils: is the current approach adequate. Ambio 25: 510-516.
17       Lorente, J.; Redano, A.; de Cabo, X. (1994) Influence of urban aerosol on spectral solar irradiance. J. Appl.
18             Meteorol. 33:406-415.
19       Lorusso, S.; Marabelli, M.; Troili, M. (1997) Air pollution and the deterioration of historic monuments. J. Environ.
20             Pathol. Toxicol. Oncol. 16: 171-173.
21       Lovett, G. M. (1984) Atmospheric deposition to forests.  In: Breece, L.; Hasbrouck, S., eds. Forest responses to
22             acidic deposition: proceedings of the U. S.-Canadian conference; August 1983; Orono, ME. Orono, ME:
23             University of Maine, Orono, Land and Water Resources Center; pp. 7-18.
24       Lovett, G. M. (1992) Atmospheric deposition and canopy interactions of nitrogen. In: Johnson, D. W.; Lindberg,
25             S. E., eds. Atmospheric deposition and forest nutrient cycling: a synthesis of the  integrated forest study.
26             New York, NY: Springer-Verlag, Inc.; pp.  152-166. (Billings, W. D.; Golley, F.; Lange, O. L.; Olson, J. S.;
27             Remmert, H., eds. Ecological studies: analysis and synthesis: v. 91).
28       Lovett, G. M. (1994) Atmospheric deposition of nutrients and pollutants in North America: an ecological
29             perspective. Ecol. Appl. 4: 629-650.
30       Lovett, G. M.; Lindberg, S. E. (1984) Dry deposition and canopy exchange in a mixed oak forest as determined by
31             analysis of throughfall. J. Appl. Ecol. 21: 1013-1027.
32       Lovett, G. M.; Lindberg, S. E. (1986) Dry deposition of nitrate to a deciduous forest. Biogeochemistry 2: 137-148.
33       Lovett, G. M.; Lindberg, S. E. (1992) Concentration and deposition of particles and vapors in a vertical profile
34             through a forest canopy. Atmos. Environ. Part A 26: 1469-1476.
35       Lovett, G. M.; Lindberg, S. E. (1993) Atmospheric deposition and canopy interactions of nitrogen in forests. Can. J.
36             For. Res. 23:  1603-1616.
37       Lovett, G. M.; Traynor, M. M.; Pouyat, R. V.; Carreiro, M. M.; Zhu, W.-X.; Baxter, J. W. (2000) Atmospheric
3 8             deposition to oak forest along an urban-rural gradient. Environ. Sci. Technol. 34: 4294-4300.
39       Lowenthal, D. H.; Rogers, C. F.; Saxena, P.; Watson, J. G.; Chow, J. C. (1995) Sensitivity of estimated light
40             extinction coefficients to model assumptions and measurement errors. Atmos. Environ. 29: 751-766.
41       Ludwig, J.; Marufu, L. T.; Huber, B.; Andreae, M. O.; Helas, G. (2003) Domestic combustion of biomass fuels in
42             developing countries: a major source of atmospheric pollutants. J. Atmos. Chem. 44: 23-37.
43       Madronich, S. (1987) Photodissociation in the atmosphere. 1. Actinic flux and the effects of ground reflections and
44             clouds. J. Geophys. Res. [Atmos.] 92: 9740-9752.
45       Magill, A. H.; Aber, J. D.; Berntson, G. M.; McDowell, W. H.; Nadelhoffer, K.  J.; Melillo, J. M.; Steudler, P. (2000)
46             Long-term nitrogen additions and nitrogen saturation in two temperate forests. Ecosystems 3: 238-253.
47       Malm, W. C. (1989) Atmospheric haze: its sources and effects on visibility in rural areas of the continental United
48             States. Environ. Monit. Assess. 12: 203-225.
49       Malm, W. C. (1999) Introduction to visibility. Fort Collins, CO: Colorado State  University; Cooperative Institute for
50             Research in the Atmosphere (CIRA); NFS Visibility Program. Available:
51             http://www.epa.gov/air/visibility/introvis.pdf [2 April 2002].
52       Malm, W. C. (2000) Spatial and seasonal patterns and temporal variability of haze and its constituents in the United
53             States. Report III. Fort Collins, CO: Cooperative Institute for Research in the Atmosphere, Colorado State
54             University. Available: http://vista.cira.colostate.edu/improve/Publications/Reports/2000/2000.htm [22 March,
55             2002].
56       Malm, W. C.; Kreidenweis, S. M. (1997) The effects of models of aerosol hygroscopicity on the apportionment of
57             extinction. Atmos. Environ. 31: 1965-1976.


         June 2003                                       4-256        DRAFT-DO  NOT QUOTE OR CITE

-------
  1       Malm, W. C.; Sisler, J. F.; Huffman, D.; Eldred, R. A.; Cahill, T. A. (1994) Spatial and seasonal trends in particle
  2             concentration and optical extinction in the United States. J. Geophys. Res. [Atmos.] 99: 1347-1370.
  3       Malm, W. C.; Molenar, J. V.; Eldred, R. A.; Sisler, J. F. (1996) Examining the relationship among atmospheric
  4             aerosols and light scattering and extinction in the Grand Canyon area. J. Geophys. Res. [Atmos.]
  5             101: 19,251-19,265.
  6       Malm, W. C.; Day, D.; Kreidenweis, S. M. (1997) Comparison of measured and reconstructed scattering during an
  7             intensive field study at Great Smoky Mountains National Park. Presented at: 90th annual meeting and
  8             exhibition of the Air & Waste Management Association; June; Toronto, ON, CA. Pittsburgh, PA: Air &
  9             Waste Management Association; 97-WA70.02.
10       Malm, W. C.; Day, D. E.; Kreidenweis, S. M. (2000) Light scattering characteristics of aerosols as a function of
11             relative humidity: part I-a comparison of measured scattering and aerosol concentrations using the theoretical
12             models. J. Air Waste Manage. Assoc. 50: 686-700.
13       Malm, W. C.; Day, D. E.; Kreidenweis, S. M. (2000) Light scattering characteristics of aerosols as a function of
14             relative humidity: Part Il-a comparison of measured scattering and aerosol concentrations using the statistical
15             models. J. Air Waste Manage. Assoc. 50: 701-709.
16       Mansfeld, F. (1980) Regional air pollution study: effects of airborne sulfur pollutants on materials. Research
17             Triangle Park, NC: U.S. Environmental Protection Agency, Environmental Sciences Research Laboratory;
18             EPA report no. EPA-600/4-80-007. Available from: NTIS, Springfield, VA; PB81-126351.
19       Marchwiiiska, E.; Kucharski, R. (1987) The combined influence of SO2 and heavy metal-containing particulates on
20             beans, carrots and parsley. Environ. Monit. Assess. 8: 11-25.
21       Markewitz, D.; Richter, D.D.; Allen, L. H.; Urrego, J. B. (1998) Three decades of observed soil acidification in the
22             Calhoun Experimental Forest: Has acid rain made a difference? Soil Sci. Soc. Amer. J. 62: 1428-1439.
23       Marschner, H. (1995) Sulfur. In: Mineral nutrition of higher plants. 2nd ed. London, United Kingdom: Academic
24             Press; pp. 255-265.
25       Martin, M. H.; Coughtrey, P. J. (1981) Impact of metals on ecosystem function and productivity. In: Lepp, N. W.,
26             ed. Effect of heavy metal pollution on plants: volume 2, metals in the environment. Barking, United
27             Kingdom: Applied Science Publishers; pp. 119-158. (Mellanby, K., ed. Pollution monitoring  series).
28       Martin, C. E.; Gravatt, D. A.; Loeschen, V. S. (1992) Photosynthetic responses of three species to acute exposures of
29             nitrate- and sulphate-containing aerosols. Atmos. Environ. Part A 26: 381-391.
3 0       Mathai, C. V. (1995) The Grand Canyon visibility.  EM (December): 20-31.
31       Matson, P. A.; Parton, W. J.; Power, A. G.; Swift, M. J. (1997) Agricultural intensification and ecosystem properties.
32             Science (Washington, DC) 277: 504-509.
33       May, P. F.; Till, A. R.; Gumming, M. J. (1972) Systems analysis of "sulphur kinetics in pastures grazed by sheep.
34             J. Appl. Ecol. 9: 25-49.
3 5       McCartney, H. A.; Aylor, D. E. (1987) Relative contributions of sedimentation and impaction to deposition of
36             particles in a crop canopy. Agric. For. Meteorol. 40: 343-358.
37       McClelland, G.; Schulze, W.; Waldman, D.; Schenk, D.; Irwin, J.; Stewart, T.; Deck, L.; Thayer, M. (1993) Valuing
38             eastern visibility: a field test of the contingent valuation method. Washington, DC: U.S. Environmental
39             Protection Agency, Office of Policy, Planning and Evaluation; cooperative  agreement no. CR-815183.
40       McCune, D. C.; Silberman, D. H.; Mandl, R. H.; Weinstein, L. H.; Freudenthal, P.  C.; Giardina, P. A. (1977) Studies
41             on the effects of saline aerosols of cooling tower origin on plants. J. Air Pollut. Control Assoc. 27:  319-324.
42       McDonald, R. L.; Unni, C. K.; Duce, R. A. (1982) Estimation of atmospheric sea salt dry deposition: wind speed
43             and particle size dependence. J. Geophys. Res. 87: 1246-1250.
44       McDonnell, M.  J.; Pickett, S. T. A.; Groffman, P.; Bohlen, P.; Pouyat, R. V.; Zipperer, W. C.; Parmelee, R. W.;
45             Carreiro, M. M.; Medley, K. (1997) Ecosystem processes along an urban-to-rural gradient. Urban Ecosyst.
46             1:21-36.
47       McDow,  S. R.; Vartiainen, M.; Sun, Q.; Hong, Y.; Yao, Y.; Kamens, R. M. (1995) Combustion aerosol water
48             content and its effect on polycyclic aromatic hydrocarbon reactivity. Atmos. Environ. 29: 791-797'.
49       McGee, E. S.; Mossotti, V. G. (1992) Gypsum accumulation on carbonate stone. Atmos. Environ. Part B
50             26:249-253.
51       McGowan, T. F.; Lipinski, G. E.; Santoleri, J. J. (1993) New rules affect the handling of waste fuels. Chem. Eng.
52             (March):  122-128.
53       Mclntyre, N. E.; Knowles-Yanez, K.; Hope, D. (2000) Urban ecology as an indisciplinary field: differences  in the
54             use of "urban" between the social and natural sciences. Urban Ecosyst. 4: 5-24.
55       McKay, W. A.;  Garland, J. A.; Livesley, D.; Halliwell, C. M.; Walker, M. I. (1994) The characteristics of the
56             shore-line sea spray aerosol and the landward transfer of radionuclides discharged to coastal sea water.
57             Atmos. Environ. 28: 3299-3309.


         June 2003                                      4-257        DRAFT-DO NOT QUOTE OR CITE

-------
  1       McLachlan, M. S. (1996a) Bioaccumulation of hydrophobic chemicals in agricultural food chains. Environ. Sci.
  2             Tech. 30: 252-259.
  3       McLachlan, M. S. (1996b) Biological uptake and transfer of polychlorinated dibenzo-p-dioxins and dibenzofurans.
  4             In: Hester, R. E.; Harrison, R. M., eds. Chlorinated organic micropollutants. Cambridge, United Kingdom:
  5             Royal Society of Chemistry; pp. 31-52. (Issues in environmental science and technology).
  6       McLachlan, M. S. (1999) Framework for the interpretation of measurements of SOCs in plants. Environ. Sci.
  7             Technol. 33: 1799-1804.
  8       McMahon, T. A.; Denison, P. J. (1979) Empirical atmospheric depositionparameters-a survey. Atmos. Environ.
  9             13:571-585.
10       McMillen, R. T. (1988) An eddy correlation technique with extended applicability to non-simple terrain. Boundary
11             Layer Meteorol. 43: 231-245.
12       McMurry, P. H.; Stolzenburg, M. R.  (1989) On the sensitivity of particle size to relative humidity for Los Angeles
13             aerosols. Atmos. Environ. 23: 497-507.
14       McMurry, P. H.; Zhang, X.; Lee, C.-T. (1996) Issues in aerosol measurement for optics assessments. J.  Geophys.
15             Res.  [Atmos.]  101: 19,189-19,197.
16       Meiwes, K.  J.; Khanna, P. K. (1981) Distribution and cycling of sulphur in the vegetation of two forest ecosystems
17             in an acid rain environment. Plant Soil 60: 369-375.
18       Mid-Atlantic Regional Assessment Team (MARAT). (2000) Preparing for a changing climate: the potential
19             consequences of climate variability and change. Mid-Atlantic overview.  Washington, DC: U.S.
20             Environmental Protection Agency; Office of Research and Development; U.S. Global Change Research
21             Program (USGCRP) and University Park, PA: Pennsylvania State University.
22       Middleton, W. E. K. (1952) Vision through the atmosphere. Toronto, ON, Canada: University of Toronto Press.
23       Midwest Regional Planning Organization. (2001) Regional haze and visibility in the upper Midwest. Des Plaines,
24             IL: Midwest Regional Planning Organization.
25       Milford, J. B.; Davidson, C. I. (1987) The sizes of paniculate sulfate and nitrate in the atmosphere-a review. JAPCA
26             37: 125-134.
27       Millar, C. S. (1974) Decomposition of coniferous leaf litter. In: Dickinson, C. H.;  Pugh, G. J. F., eds. Biology of
28             plant litter decomposition: volume I. New York, NY: Academic Press; pp.  105-128.
29       Miller, P. R.; McBride, J. R., eds. (1999) Oxidant air pollution impacts in the Montane forests of southern
30             California: a case study of the San Bernardino Mountains. New York, NY:  Springer-Verlag.
31             (Ecological Studies: v. 134.)
32       Miller, W. C.; Fornes, R. E.; Gilbert, R. D.; Speer, A.; Spence, J. (1992) Removal of CaCO3 extender in residential
3 3             coatings by atmospheric acidic deposition. In:  Measurement of toxic and related air pollutants: proceedings of
34             the 1992 U.S. EPA/A&WMA international symposium. Pittsburgh, PA:  Air & Waste Management
3 5             Association; pp. 129-134. (A&WMA publication VIP-25).
36       Miller, E. K.; Panek, J. A.; Friedland, A. J.; Kadlecek, J.; Mohnen, V. A. (1993) Atmospheric deposition to a
37             high-elevation forest at Whiteface Mountain, New York, USA. Tellus Ser.  B 45: 209-227.
3 8       Mims, F. M., III. (1996) Significant reduction of UVB caused by smoke from biomass burning in Brazil.
39             Photochem. Photobiol. 64: 814-816.
40       Mitchell, M. J.; Harrison, R. B.; Fitzgerald, J. W.; Johnson, D. W.; Lindberg, S. E.; Zhang, Y.; Autry, A. (1992a)
41             Sulfur distribution and cycling in forest ecosystems. In: Johnson, D. W.; Lindberg, S. E., eds. Atmospheric
42             deposition and forest nutrient cycling: a synthesis of the integrated forest study. New  York, NY:
43             Springer-Verlag; pp. 90-129. (Billings, W. D.; Golley, F.; Lange, O. L.; Olson, J. S.; Remmert, H., eds.
44             Ecological studies: analysis and synthesis: v. 91).
45       Mitchell, M. J.; Foster, N. W.; Shepard, J. P.; Morrison,  I. K. (1992b) Nutrient cycling in Huntington Forest and
46             Turkey Lakes deciduous stands: nitrogen and sulfur. Can. J. For. Res. 22: 457-464.
47             Molenar, J. V.; Malm, W. C.;  Johnson, C. E. (1994) Visual air quality simulation techniques. Atmos. Environ.
48             28: 1055-1063.
49       Monteith, J. L.; Unsworth, M. H. (1990) Principles of environmental physics. 2nd ed. London, United Kingdom:
50             Edward Arnold; pp. 165-175.
51       Moomaw, W. R. (2002) Energy, industry and nitrogen: strategies for decreasing reactive nitrogen emisisons. Ambio
52             31: 184-189.
53       Moosmtiller, H.; Arnott, W. P.; Rogers, C. F. (1998) Photoacoustic and filter measurements  related to aerosol light
54             absorption during the Northern Front Range Air Quality Study (Colorado 1996/1997). J. Geophys. Res.
55             103:28,149-28,157.
56       Moropoulou, A.; Bisbikou, K.; Torts, K.; Van Grieken, R.; Zezza, F.; Macri, F. (1998) Origin and growth of
57             weathering crusts on ancient marbles in industrial atmosphere. Atmos. Environ. 32: 967-982.


         June 2003                                      4-258         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Mueller, S. F. (1988) Chemical deposition to high-elevation spruce-fir forests in the eastern United States. Muscle
  2             Shoals, AL: Tennessee Valley Authority, Air Quality Branch.
  3       Naeem, S.; Thompson, L. I; Lawler, S. P.; Lawton, J. H.; Woodfin, R. M. (1994) Declining biodiversity can alter
  4             the performance of ecosystems. Nature 368: 734-737.
  5       Nash, T. H., III. (1975) Influence of effluents from a zinc factory on lichens. Ecol. Monogr. 45: 183-198.
  6       National Assessment Synthesis Team (NAST). (2000) Climate change impacts on the United States: the potential
  7             consequences of climate variability and change. Overview. Washington, DC: U.S. Global Change Research
  8             Program.
  9       National Park Service. (1998) IMPROVE program goals and objectives. Washington, DC: National Park Service.
10             Available: www2.nature.nps.gov/ard/impr/index.htm [24 November 1999].
11       National Research Council. (1993) Protecting visibility in national parks and wilderness areas. Washington, DC:
12             National Academy Press. 3v.
13       National Science  and Technology Council. (1998) National acid precipitation assessment program biennial report to
14             Congress:  an integrated assessment; executive summary. Silver Spring, MD: U.S. Department of Commerce,
15             National Oceanic and Atmospheric Administration.  Available:
16             www.nnic.noaa.gov/CENR/NAPAP/NAPAP_96.htm [24 November 1999].
17       National Weather Service. (1998) Visibility. Washington, DC: National Weather Service, National Oceanic and
18             Atmospheric Administration. Available: www.nws.noaa.gov/asos/vsby.htm [24 November 1999].
19       Nazaroff, W. W.; Cass, G. R. (1991) Protecting museum collections from soiling due to the deposition of airborne
20             particles. Atmos. Environ. Part A 25: 841-852.
21       Nelis, P. M.; Branford, D.; Unsworth, M. H. (1994) A model of the transfer of radioactivity from sea to land in sea
22             spray. Atmos. Environ. 28: 3213-3223.
23       Neuberger, H.; Hosier, C. L.; Kocmond, W. C. (1967) Vegetation as aerosol filter In: Tromp, S. W.; Weihe, W. H.,
24             eds. Biometeorology, v. 2. pp. 693-702.
25       Neuroth, G.; Heisler, S. (2000) The nature and sources of visibility impairment in Phoenix. Presented at: Governor's
26             Brown Cloud Summit; August; Phoenix, AZ. Phoenix, AZ: Arizona Department of Environmental Quality.
27             Available: http://www.adeq.state.az.us/environ/air/browncloud/download/gary.pdf (28 May 2003).
28       New England Regional Assessment Group (NERAG). (2001) Preparing for a changing climate: the potential
29             consequences of climate variability and change. New England regional overview. Washington, DC: U.S.
30             Environmental Protection Agency; Office of Research and Development; U.S. Global Change Research
31             Program (USGCRP) and Durham, NH: University of New Hampshire; Institute for the Study of Earth,
32             Oceans, and Space.
33       Nicholson, K. W. (1988) The dry deposition of small particles: a review of experimental measurements. Atmos.
34             Environ. 22: 2653-2666.
35       Nicholson, K. W.; Davies, T. D. (1987) Field measurements of the dry deposition of paniculate sulphate.  Atmos.
36             Environ. 21: 1561-1571.
37       Nielsen, K. E.; Ladekarl, U. L.; Nornberg, P. (1999) Dynamic soil processes on heathland due to changes in
3 8             vegetation to oak and Sitka spruce. For. Ecol. Manage. 114: 107-116.
39       Nihlgard, B.  (1985) The ammonium hypothesis-an additional explanation to the forest dieback in Europe. Arnbio
40             14:2-8.
41       Nikliiiska, M.; Laskowski, R.; Maryaiiski, M. (1998) Effect of heavy metals and storage time on two  types of forest
42             litter: basal respiration rate and exchangeable metals. Ecotoxicol. Environ.  Safety 41: 8-18.
43       Noble, I. R.; Dirzo, R. (1997) Forests as human-dominated ecosystems. Science (Washington, DC) 277: 522-525.
44       Nodvin, S. C.; VanMiegroet, H.; Lindberg, S. E.; Nicholas, N. S.;  Johnson, D. W. (1995) Acidic deposition,
45             ecosystem processes, and nitrogen saturation in a high elevation southern Appalachian watershed. Water Air
46             Soil Pollut. 85: 1647-1652.
47       Noll, K. E.; Fang, K. Y. P. (1989) Development of a dry deposition model for atmospheric coarse particles. Atmos.
48             Environ. 23:585-594.
49       Nord, A. G.; Ericsson, T. (1993) Chemical analysis of thin black layers on building stone. Stud. Conserv.  38: 25-35.
50       Northeast States for Coordinated Air Use Management (NESCAUM). (2002) CAMNET: realtime air pollution and
51             visibility monitoring. Boston, MA: NESCAUM. Available at: http://www.hazecam.net. (23 January 2002).
52       Norton, R. A.; Childers,  N. F. (1954) Experiments with urea sprays on the peach.  Proc. Am. Soc. Hortic. Sci.
53             63:23-31.
54       Novakov, T.; Penner, J. E. (1993) Large contribution of organic aerosols to cloud-condensation-nuclei
55             concentrations. Nature (London) 365: 823-826.
         June 2003                                      4-259         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Ockenden, W. A.; Stinnes, E.; Parker, C.; Jones, K. C. (1998) Observations on persistent organic pollutants in
  2             plants: implications for their use as passive air samplers and for POP cycling. Environ. Sci. Technol.
  3             32:2721-2726.
  4       Odum, E. P. (1969) The strategy of ecosystem development: an understanding of ecological succession provides a
  5             basis for resolving man's conflict with nature. Science (Washington, DC) 164: 262-270.
  6       Odum, E. P. (1985) Trends expected in stressed ecosystems. BioScience 35: 419-422.
  7       Odum, E. P. (1989) Ecology and our endangered life-support systems. Sunderland, MA: Sinauer Associates, Inc.;
  8             pp. 58-66, 108-118.
  9       Odum, E. P. (1993) The ecosystem. In: Ecology and our endangered life-support systems. 2nd ed. Sunderland, MA:
10             Sinauer Associates, Inc.; pp. 38-67.
11       Oehme, M.; Biseth, A.; Schlabach, M.; Wiig, 0. (1995) Concentrations of polychlorinated dibenzo-p-dioxins,
12             dibenzofurans and non-ortho substituted biphenyls in polar bear milk from Svalbard (Norway). Environ.
13             Pollut. 90: 401-407.
14       Olszyk, D. M.; Bytnerowicz, A.; Takemoto, B. K. (1989) Photochemical oxidant pollution and vegetation: effects of
15             gases, fog, and particles. Environ. Pollut. 61:  11-29.
16       Oosting, H. J.; Billings, W. D. (1942) Factors effecting vegetational zonation on coastal dunes. Ecology 23:
17             131-142.
18       Ormrod, D. P. (1984) Impact of trace element pollution on plants. In: Treshow, M., ed. Air pollution and plant life.
19             Chichester, United  Kingdom: John Wiley & Sons; pp. 291-319.
20       Ouimette, J. R.; Flagan, R. C. (1982) The extinction coefficient of multicomponent aerosols. Atmos. Environ.
21             16:2405-2420.
22       Padgett, P.; Allen, E. B. (1999) Differential responses to nitrogen fertilization in native shrubs and exotic annuals
23             common to Mediterranean coastal sage  scrub  of California. Plant Ecol.  144: 93-101.
24       Padgett, P. E.; Allen, E. B.; Bytnerowicz, A.; Minich, R. A. (1999) Changes in soil inorganic nitrogen as related to
25             atmospheric nitrogenous pollutants in southern California. Atmos. Environ. 33: 769-781.
26       Paerl, H. W.; Bales, J. D.; Ausley, L.  W.; Buzzelli, C. P.;  Crowder, L. B.; Eby, L. A.; Go, M.; Peierls, B. L.;
27             Richardson, T. L.; Ramus, J. S. (2001) Ecosystem impacts of three sequential hurricanes (Dennis, Floyd, and
28             Irene) on the United States' largest lagoonal estuary, Pamlico Sound, NC. Proc. Nat. Acad. Sci. U. S. A.
29             98:5655-5611.
30       Pan, W.; Tatang, M. A.; McRae, G. J.; Prinn, R. G. (1997) Uncertainty analysis of direct radiative forcing by
31             anthropogenic sulfate aerosols. J. Geophys. Res. 102 (D18): 21,915-21,924.
32       Pan, W.; Tatang, M. A.; McRae, G. J.; Prinn, R. G. (1998) Uncertainty analysis of indirect radiative forcing by
33             anthropogenic sulfate aerosols. J. Geophys. Res. 103 (D4): 3815-3823.
34       Pandis, S. N.; Seinfeld, J.  H. (1989) Mathematical modeling of acid deposition due to radiation fog. J. Geophys. Res.
35             [Atmos.] 94: 12,911-12,923.
36       Pankow, J. F. (1987) Review and comparative analysis of the theories on partitioning between the gas and aerosol
37             particulate phases in the atmosphere. Atmos. Environ. 21: 2275-2283.
38       Parish, S. B. (1910) The effect of cement dust on citrus trees. Plant World 13: 288-291.
39       Parker, A. (1955) The destructive effects of air pollution on materials. Colliery Guardian 191: 447-450.
40       Parker, G. G. (1983) Throughfall and stemflow in the forest nutrient cycle. In: Macfadyen, A.; Ford, E. D., eds.
41             Advances in ecological research, v.  13;  pp. 57-120.
42       Patz, J. A.; Engelberg, D.; Last, J. (2000a)  The effects of changing weather on public health. Ann. Rev. Public
43             Health 21: 271-307.
44       Patz, J. A.; McGeehin, M. A.; Bernard, S. M.;  Ebi, K. L.;  Epstein, P. R.; Grambsch, A.; Gubler,  D. J.; Reiter, P.;
45             Romieu, L; Rose, J. B.; Samet, J. M.; Trtanj, J. (2000b) The potential health impacts of climate variability and
46             change for the United States: executive  summary of the report of the health sector of the U.S. national
47             assessment. Environ. Health Perspect. 108: 367-376.
48       Pearcy, R. W.; Bjorkman, O.; Caldwell, M. M.; Keeley, J. E.; Monson, R. K.;  Strain, B. R. (1987) Carbon gain by
49             plants in natural environments. BioScience 37: 21-29.
50       Penndorf, R. B. (1958) An approximation method to the Mie theory for colloidal spheres. J. Phys. Chem.
51             62:1537-1542.
52       Penner, J. E.; Charlson, R. J.; Hales, J. M.;  Laulainen, N. S.; Leifer, R.; Novakov, T.; Ogren, J.;  Radke, L. F.;
53             Schwartz, S. E.; Travis, L. (1994) Quantifying and minimizing uncertainty of climate forcing by
54             anthropogenic aerosols. Bull. Am. Meteorol. Soc. 75: 375-400.
55       Percy, K. E. (1991) Effects of acid rain on forest vegetation: morphological and non-mensurational growth effects.
56             In: Rennie, P. J.; Robitaille, G., eds.  Effects of Acid Rain on Forest Resources: Proceedings of the Conference
57             ; June, 1983; Sainte-Foy, Quebec. Ottawa, Canada: Canadian Forestry Service; pp. 97-111.


         June 2003                                      4-260        DRAFT-DO NOT QUOTE OR CITE

-------
  1       Perry, K. D.; Cahill, T. A.; Eldred, R. A.; Dutcher, D. D.; Gill, T. E. (1997) Long-range transport of North African
  2             dust to the eastern United States. J. Geophys. Res. [Atmos] 102: 11,225-11,238.
  3       Peterjohn, W. T.; Adams, M. B.; Gilliam, F. S. (1996) Symptoms of nitrogen saturation in two central Appalachian
  4             hardwood forest ecosystems. Biogeochemistry 35: 507-522.
  5       Peters, K.; Eiden, R. (1992) Modelling the dry deposition velocity of aerosol particles to a spruce forest. Atmos.
  6             Environ. Part A 26: 2555-2564.
  7       Peterson, J. T.; Flowers, E. C.; Rudisill, J. H. (1978) Urban-rural solar radiation and atmospheric turbidity
  8             measurements in the Los Angeles basin. J. Appl. Meteorol. 17: 1595-1609.
  9       Peterson, G.; Allen, C. R.; Holling, C. S. (1998) Ecological resilience, biodiversity, and scale. Ecosystems 1: 6-18.
10       Petuskey, W.  T.; Richardson, D. A.; Dolske, D. A. (1995) Aspects of the deterioration of sandstone masonry in
11             Anasazi dwelling ruins at Mesa Verde National Park, Colorado, USA. Sci. Total Environ. 167: 145-159.
12       Piatt, J. R.; Krause, P. D. (1974) Road and site characteristics that influence road salt distribution and damage to
13             roadside aspen trees. Water Air Soil Pollut. 3: 301-304.
14       Pickett, S. T. A.; Cadenasso, M. L.; Grove, J. M.; Nilon, C. Y.; Pouyat, R. V.; Zipperer, W. C.; Costanza, R. (2001)
15             Urban ecological systems: linking terrestrial ecological, physical, and socioeconomic components of
16             metropolitan areas. Annu. Rev. Ecol.  Syst. 32: 127-157.
17       Pilinis, C.; Pandis, S. N.; Seinfield, J. H. (1995) Sensitivity of direct climate forcing by atmospheric aerosols to
18             aerosol size and composition. J. Geophys. Res. [Atmos.] 100: 18739-18754.
19       Pimentel, D.;  Wilson, C.; McCullum, C.; Huang, R.; Dwen, P.; Flack, J.; Tran, Q.; Saltman, T.; Cliff, B. (1997)
20             Economic and environmental benefits of biodiversity. BioScience 47: 747-757'.
21       Pimm, S. L. (1984) The complexity and stability of ecosystems. Nature (London) 307: 321-326.
22       Pitchford, M.  L.; Malm, W. C. (1994) Development and applications of a standard visual index. Atmos. Environ.
23             28: 1049-1054.
24       Pitchford, M.  L.; McMurry, P. H. (1994) Relationship between measured water vapor growth and chemistry of
25             atmospheric aerosol for Grand Canyon, Arizona, in winter 1990. Atmos. Environ. 28: 827-839.
26       Pouyat, R. V.; McDonnell, M.  J. (1991) Heavy metal accumulations in forest soils along an urban-rural gradient in
27             southeastern New York, USA. Water Air Soil Pollut. 57/58: 797-808.
28       Prescott,  C. E.; Parkinson, D. (1985) Effects of sulphur pollution on rates of litter decomposition in a pine forest.
29             Can. J. Bot. 63: 1436-1443.
30       Pryor, S.; Steyn, D. (1994) Visibility and ambient aerosols in southwestern British Columbia  during REVEAL
31             (REgional Visibility Experimental Assessment in the Lower Fraser Valley): application of analytical
32             techniques for source apportionment and related visibility impacts of the REVEAL fine paniculate optical and
33             scene data. Victoria, BC, Canada: British Columbia Ministry of Environment, Lands and Parks. Available
34             from: NTIS, Springfield, VA; MIC-96-003 WINS.
35       Pueschel, R. F. (1993) Potential climatic effects of anthropogenic aerosols.  In:  Jennings, S. G., ed. Aerosol effects on
36             climate. Tucson, AZ: University of Arizona Press; pp.  110-132.
37       Rae, D. A. (1983) The value to visitors of improving visibility at Mesa Verde and Great Smoky National Parks. In:
38             Rowe,  R. D.; Chestnut, L. G., eds. Managing air quality and scenic resources at national parks and wilderness
39             areas. Boulder, CO: Westview Press; pp. 217-234.
40       Ragsdale, H. L.; Lindberg, S. E.; Lovett, G. M.; Schaefer, D. A. (1992) Atmospheric deposition and throughfall
41             fluxes of base cations. In: Johnson, D. W.; Lindberg, S. E., eds. Atmospheric deposition and forest nutrient
42             cycling: a synthesis of the integrated forest study. New York, NY: Springer Verlag; pp. 235-253. (Billings,
43             W. D.; Golley, F.; Lange, O. L.; Olson, J. S.; Remmert, H., eds. Ecological studies: analysis and synthesis:
44             v.  91).
45       Ramanathan, V.; Crutzen, P. J.; Kiehl, J. T.; Rosenfeld, D. (2001) Aerosols, climate and the hydrological cycle.
46             Science 294: 2119-2124.
47       Rapport, D. J.; Whitford, W. G. (1999) How ecosystems respond to stress: common properties of arid and aquatic
48             systems. BioScience 49: 193-203.
49       Raynal, D. J.; Joslin, J. D.; Thornton, F. C.; Schaedle, M.; Henderson, G. S. (1990) Sensitivity of tree seedlings to
50             aluminum: III. Red spruce and loblolly pine. J. Environ. Qual. 19: 180-187.
51       Realini, M.; Negrotti, R.; Appollonia, L.; Vaudan, D. (1995) Deposition of paniculate matter on stone surfaces:
52             an experimental verification of its effects on Carrara marble. In: Saiz-Jimenez, C., ed. The deterioration of
5 3             monuments: proceedings of the 2nd international symposium on biodeterioration and biodegradation;
54             January 1994; Sevilla, Spain. Sci. Total Environ. 167: 67-72.
55       Reiners, W. A.; Olson, R. K. (1984) Effects of canopy components on throughfall chemistry:  an experimental
56             analysis. Oecologia 63: 320-330.
         June 2003                                       4-261         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Reisinger, L. M. (1990) Analysis of airborne particles sampled in the southern Appalachian Mountains. Water Air
  2             SoilPollut. 50: 149-162.
  3       Rennie, P. J. (1955) The uptake of nutrients by mature forest growth. Plant Soil 7: 49-95.
  4       Reuss, J. O. (1983) Implications of the calcium-aluminum exchange system for the effect of acid precipitation on
  5             soils. J. Environ. Qual. 12: 591-595.
  6       Reuss, J. O.; Johnson, D. W. (1986) Acid deposition and the acidification of soils and waters. New York, NY:
  7             Springer-Verlag. (Billings, W. D.; Golley, F.; Lange, O. L.; Olson, J. S.; Remmert, H., eds. Ecological
  8             studies: analysis and synthesis: v. 59).
  9       Riggan, P. J.; Lockwood, R. N.; Lopez, E. N. (1985) Deposition and processing of airborne nitrogen pollutants in
10             Mediterranean-type ecosystems of southern California. Environ. Sci. Technol. 19: 781-789.
11       Rochette, P.; Desjardins, R. L.; Pattey, E.; Lessard, R. (1996) Instantaneous measurement of radiation and water use
12             efficiencies of a maize crop. Agron. J. 88: 627-635.
13       Roderick, M. L.; Farquhar, G. D.; Berry, S. L.; Noble, I. R. (2001) On the direct effect of clouds and atmospheric
14             particles on the productivity and structure of vegetation. Oecologia 129: 21-30.
15       Rodney, D. R. (1952) The entrance of nitrogen compounds through the epidermis of apple leaves. Proc. Am. Soc.
16             Hortic. Sci. 59: 99-102.
17       Roelofs, J. G. M.; Kempers, A. J.; Houdijk, A. L. F. M.; Jansen, J. (1985) The effect of air-borne ammonium
18             sulphate on Pinus nigra var. maritima in the Netherlands. Plant Soil 84: 45-56.
19       Roelofs, J. G. M.; Boxman, A. W.; VanDijk, H. F. G. (1987) Effects of airborne ammonium on natural vegetation
20             and forests. In: Asman, W. A. H.; Diederen, H. S. M. A., eds. Ammonia and acidification: proceedings [of a]
21             symposium of the European Association for the Science of Air Pollution (EURAS AP); April; Bilthoven,
22             The Netherlands. European Association for the Science of Air Pollution; pp. 266-276.
23       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993a) Sources of fine organic
24             aerosol. 4. Paniculate abrasion products from leaf surfaces of urban plants. Environ. Sci. Technol.
25             27:2700-2711.
26       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993b) Sources of fine organic
27             aerosol. 3. Road dust, tire debris, and organometallic brake lining dust: roads as sources and sinks.
28             Environ. Sci. Technol. 27: 1892-1904.
29       Rosenberg, N. J.; Blad, B. L.; Verma, S. B. (1983) Microclimate: the biological environment. 2nd ed. New York,
30             NY: John Wiley & Sons.
31       Rosenfeld, D. (1999) TRMM observed first direct evidence of smoke from forest fires inhibiting rainfall. Geophys.
32             Res. Lett. 26:3105-3108.
33       Rovira, A. D.; Davey, C. B. (1974) Biology of the rhizosphere. In: Carson, E. W., ed. The plant root and its
34             environment: proceedings of an institute; July, 1971; Blacksburg, VA. Charlottesville, VA: University Press
35             of Virginia; pp. 153-204.
36       Rueth, H.; Baron, J. S. (2002) Differences in Englemann spruce forest biogeochemistry east and west of the
37             Continental Divide in Colorado, USA. Ecosystems  5: 45-57.
38       Ruijgrok, W.; Tieben, H.; Eisinga, P. (1997) The dry deposition of particles to a forest canopy: a comparison of
3 9             model and experimental results. Atmos. Environ. 31:399-415.
40       Sabbioni, C. (1995) Contribution of atmospheric deposition to the formation of damage layers. In: Jimenez-Saiz, C.,
41             ed. The deterioration of monuments: proceedings of the 2nd international symposium on biodeterioration and
42             biodegradation; January 1994; Sevilla, Spain. Sci. Total Environ. 167: 49-56.
43       Sabbioni, C.; Zappia, G.; Gobbi, G. (1992) Carbonaceous particles on carbonate building stones in a simulated
44             system. In: Proceedings of the 1992 European aerosol conference; September; Oxford, United Kingdom.
45             J. Aerosol Sci. 23(suppl. 1): S921-S924.
46       Sabbioni, C.; Zappia, G.; Gobbi, G. (1996) Carbonaceous particles and stone damage in a laboratory exposure
47             system.  J. Geophys. Res. [Atmos.]  101: 19,621-19,627.
48       Sabbioni, C.; Zappia, G., Ghedini, N.;  Gobbi, G.; Favoni, O. (1998) Black crusts on ancient mortars.
49             Atmos. Environ. 32: 215-223.
50       Saiz-Jimenez, C. (1993) Deposition of airborne organic pollutants on historic buildings. Atmos. Environ. Part B
51             27:77-85.
52       Salby, M. L. (1996) Fundamentals of atmospheric physics. New York, NY: Academic Press. (The International
53             Geophysics Series, Textbooks and Monographs, v. 61).
54       Samecka-Cymerman, A.; Kempers, A.  J. (1999) Bioindication of heavy metals in the town Wroclaw (Poland) with
55             evergreen plants. Atmos. Environ. 33: 419-430.
         June 2003                                      4-262         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Samuels, H. I; Twiss, S.; Wong, E. W. (1973) Visibility, light scattering and mass concentration of paniculate
  2             matter: report of the California Tri-City Aerosol Sampling Project. Sacramento, CA: State of California, Air
  3             Resources Board.
  4       Sanyal, B.; Singhania, G. K. (1956) Atmospheric corrosion of metals: part I. J. Sci. Ind. Res. Sect. B 15: 448-455.
  5       Saunders, P. J. W.; Godzik, S. (1986) Terrestrial vegetation-air pollutant interactions: nongaseous air pollutants.
  6             In: Legge, A. H.; Krupa, S.  V., eds. Air pollutants and their effects on the terrestrial ecosystem. New York,
  7             NY: John Wiley & Sons; pp. 389-394. (Advances in environmental science and technology: v. 18).
  8       Saxena, P.; Peterson, T. W. (1981) Thermodynamics of multicomponent electrolytic aerosols. J. Colloid Interface
  9             Sci. 79: 496-510.
10       Saxena, P.; Mueller, P. K.; Kim, Y. P.; Seinfeld, J. H.; Koutrakis, P. (1993) Coupling thermodynamic theory with
11             measurements to characterize acidity of atmospheric particles. Aerosol Sci. Technol. 19: 279-293.
12       Saxena, P.; Hildemann, L. M; McMurry, P. H.; Seinfeld, J. H. (1995) Organics alter hygroscopic behavior of
13             atmospheric particles. J. Geophys. Res. [Atmos.] 100: 18,755-18,770.
14       Schiavon, N.; Chiavari, G.; Schiavon, G.; Fabbri, D. (1995) Nature and decay effects of urban soiling on granitic
15             building stones. In: Jimenez-Saiz, C., ed. The deterioration of monuments: proceedings of the 2nd
16             international symposium on biodeterioration and biodegradation; January 1994; Sevilla, Spain. Sci. Total
17             Environ.  167: 87-101.
18       Schichtel, B. A.; Husar, R. B.; Falke, S. R.; Wilson, W. E. (2001) Haze trends over the United States, 1980-1995.
19             Atmos. Environ. 35:  5205-5210.
20       Schier, G.  A.; Jensen, K. (1992) Atmospheric deposition effects on foliar injury and  foliar leaching in red spruce. In:
21             Eager, C.; Adams, M. B., eds. Ecology and decline of red spruce in the eastern United States. New York, NY:
22             Springer-Verlag; pp. 271-294. [Billings, W. D.; Golley, F.; Lange, O. L.; Olson, J. D.; Remmert, H., eds.
23             Ecological Studies, Analysis and Synthesis Ecological Studies, v. 96].
24       Schimek, M. G. (1981) Smoothing and regression. New York, NY: Wiley.
25       Schneider, S. H. (1992) Introduction to climate modeling. In: Trenberth, K. E. Climate system modeling. New York,
26             NY: Cambridge University  Press; pp.  1-26.
27       Schnug, E. (1997) Significance of sulphur for the quality of domesticated plants. In:  Cram, W. J.; De Kok, L. J.;
28             Stulen, L; Brunold, C.; Rennenberg, H., eds. Sulphur metabolism in higher plants: molecular,
29             ecophysiological and nutritional aspects. Leiden, The Netherlands: Backhuys  Publishers; pp. 109-130.
30       Schonherr, J.; Huber, R. (1977) Plant cuticles are polyelectrolytes with isoelectric points around three. Plant Physiol.
31             59:145-150.
32       Schulze, E.-D. (1989) Air pollution and forest decline in a spruce (Picea abies) forest. Science (Washington, DC)
33             244:776-783.
34       Schuster, P. F.; Reddy, M. M.; Sherwood, S. I. (1994) Effects of acid rain and sulfur dioxide on marble  dissolution.
3 5             Mater. Perform. 33: 76-80.
36       Schwar, M. J. R. (1998) Nuisance  dust deposition and soiling rate measurements. Environ. Technol. 19: 223-229.
37       Science Advisory Board.  (2002) A framework for assessing and reporting on ecological condition:  an SAB report.
38             Young, T. F.;  Sanzone, S., eds. Washington, DC: U.S. Environmental Protection Agency, Ecological
39             Processes and Effects Committee; report no. EPA-SAB-EPEC-02-009. Available at:
40             http://www.epa.gov/sciencel/index.html (27 January 2003).
41       Seckmeyer, G.; McKenzie, R. L. (1992) Increased ultraviolet radiation in New Zealand (45 degrees S) relative to
42             Germany (48 degrees N). Nature 359: 135-137.
43       Sehmel, G. A. (1980) Particle and gas dry deposition: a review. Atmos. Environ. 14: 983-1011.
44       Sehmel, G. A.; Hodgson, W. H. (1976) Predicted dry deposition velocities. In: Atmosphere-surface exchange of
45             paniculate and gaseous pollutants (1974): proceedings of a symposium; September 1974; Richland, WA. Oak
46             Ridge, TN: Energy Research and Development Administration; pp. 399-422.  (ERDA symposium series: 38).
47             Available from: NTIS, Springfield, VA; CONF-740921.
48       Seinfeld, J. H. (1989) Urban air pollution: state of the science. Science 243: 745-752.
49       Seinfeld, J. H.; Pandis, S. N. (1998) Atmospheric chemistry and physics: from air pollution to climate change.
50             New York, NY: John Wiley & Sons, Inc.
51       Sereda, P.  J. (1974) Weather factors affecting corrosion of metals. In: Corrosion in natural environments: three
52             symposia presented at the seventy-sixth annual meeting [of the] American Society for Testing and Materials;
53             June 1973; Philadelphia, PA. Philadelphia, PA: American Society for Testing and Materials; pp. 7-22.
54             (ASTM special technical publication 558).
55       Shannon, J. D. (1999) Regional trends in wet deposition of sulfate in the United States and SO2 emissions from 1980
56             through 1995. Atmos. Environ. 33:  807-816.
         June 2003                                      4-263         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Sharpies, S.; Stewart, L.; Tregenza, P. R. (2001) Glazing daylight transmittances: afield survey of windows in urban
  2             areas. Build. Environ. 36: 503-509.
  3       Shinn, J. H. (1978) A critical survey of measurements of foliar deposition of airborne sulfates and nitrates. Presented
  4             at: 71st annual meeting of the Air Pollution Control Association; June; Houston, TX. Pittsburgh, PA: Air
  5             Pollution Control Association; paper no. 78-7.2.
  6       Shortle, W. C.; Bondietti, E. A. (1992) Timing, magnitude, and impact of acidic deposition on sensitive forest sites.
  7             Water Air Soil Pollut. 61: 253-267.
  8       Shortle, W. C.; Smith, K. T. (1988) Aluminum-induced calcium deficiency syndrome in declining red spruce.
  9             Science (Washington, DC) 240: 1017-1018.
10       Shortle, W. C.; Smith, K. T.; Minocha, R.; Lawrence, G. B.; David, M. B. (1997) Acidic deposition, cation
11             mobilization, and biochemical indicators of stress in healthy red spruce. J. Environ. Qual. 26: 871-876.
12       Showak, W.; Dunbar, S. R. (1982) Effect of 1 percent copper addition on atmospheric corrosion of rolled zinc after
13             20 years' exposure. In: Dean, S. W.; Rhea, E. C., eds. Atmospheric corrosion of metals: a symposium;
14             May 1980; Denver, CO. Philadelphia, PA: American Society for Testing and Materials; pp. 135-162.
15             (ASTM special technical publication 767).
16       Shriner, D. S.; Henderson, G. S. (1978) Sulfur distribution and cycling in a deciduous forest watershed. J. Environ.
17             Qual. 7: 392-397.
18       Sievering, H. (1987) Dynamics of sulfur exchange at the air/forest canopy interface: a review of throughfall inferred
19             deposition rates. Global Biogeochem. Cycles 1: 233-249.
20       Sievering, H.; Rusch, D.; Marquez, L. (1996) Nitric acid, paniculate nitrate and ammonium in the continental free
21             troposphere: nitrogen deposition to an alpine tundra ecosystem. Atmos. Environ. 30: 2527-2537.
22       Simonich; S. L.; Kites, R. A. (1995) Organic pollutant accumulation in vegetation. Environ. Sci. Technol.
23             29:2905-2914.
24       Simpson, J. W.; Horrobin, P. J., eds. (1970) The weathering and performance of building materials. New York, NY:
25             Wiley-Interscience.
26       Sinclair, T. R.; Shiraiwa, T.; Hammer, G. L. (1992) Variation in crop radiation-use efficiency with increased
27             diffuse-radiation. Crop Sci. 32: 1281-1284.
28       Sisler, J. F. (1996) Spatial and seasonal patterns and long term variability of the composition of the haze in the
29             United States: an analysis of data from the IMPROVE network. Fort Collins, CO: Colorado State University,
30             Cooperative Institute for Research in the Atmosphere; EPA air docket A096-56, document no. VI-B-09-(ee).
31       Sisler, J. F.; Cahill, T. A. (1993) Spacial and temporal patterns and the chemical composition of the haze and its
32             impact on visibility in Alaska. In: Proceedings [of the] 86th annual meeting & exhibition of the Air & Waste
33             Management Association. Volume 8: general environmental topics; June; Denver, CO. Pittsburgh, PA: Air &
34             Waste Management Association; 93-MP-4.03.
3 5       Sisler, J. F.; Malm, W. C. (2000) Interpretation of trends of PM2 5 and reconstructed visibility from the IMPROVE
36             network. J. Air Waste Manage. Assoc. 50: 775-789.
37       Sisler, J. F.; Huffman, D.; Lattimer, D. A.; Malm, W. C.; Pitchford, M. L. (1993) Spatial and temporal patterns and
3 8             the chemical composition of the haze in the United States: an anlysis of data from the IMPROVE network,
39             1988-1991. Fort Collins, CO: Colorado State University, Cooperative Institute for Research in the
40             Atmosphere (CIRA).
41       Skeffington, R. A.; Wilson, E. J. (1988) Excess nitrogen deposition: issues for consideration. Environ. Pollut.
42             54: 159-184.
43       Skerry, B. S.; Alavi, A.; Lindgren, K. I. (1988) Environmental and electrochemical test methods for the evaluation of
44             protective organic coatings. J. Coat.  Technol. 60: 97-106.
45       Slinn, W. G. N. (1977) Some approximations for the wet and dry removal of particles and gases from the
46             atmosphere.  Water Air Soil Pollut. 7: 513-543.
47       Slinn, W. G. N. (1978) Parameterizations for resuspension and for wet and dry deposition of particles and gases for
48             use in radiation dose calculations. Nucl. Saf. 19: 205-219.
49       Slinn, W. G. N. (1982) Predictions for particle deposition to vegetative canopies. Atmos. Environ. 16: 1785-1794.
50       Sloane, C. S. (1983) Optical properties of aerosols-comparison of measurements with model calculations.  Atmos.
51             Environ. 17: 409-416.
52       Sloane, C. S. (1984) Optical properties of aerosols of mixed composition. Atmos. Environ. 18: 871-878.
53       Sloane, C. S. (1986) Effect of composition on aerosol light scattering efficiencies. Atmos. Environ. 20: 1025-1037.
54       Sloane, C. S.; Wolff, G. T. (1985) Prediction of ambient light scattering using a physical model responsive to
55             relative humidity: validation with measurements from Detroit. Atmos. Environ. 19: 669-680.
56       Smith, W. H. (1973) Metal contamination of urban woody plants. Environ. Sci. Technol. 7: 631-636.
         June 2003                                      4-264         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Smith, W. H. (1974) Air pollution - effects on the structure and function of the temperate forest ecosystem.
  2             Environ. Pollut. 6: 111-129.
  3       Smith, W. H. (1984) Pollutant uptake by plants. In: Treshow, M, ed. Air pollution and plant life. New York, NY:
  4             John Wiley & Sons, Ltd.; pp. 417-450.
  5       Smith, W. H. (1990a) Forests as sinks for air contaminants: soil compartment. In: Air pollution and forests:
  6             interactions between air contaminants and forest ecosystems. 2nd ed. New York, NY: Springer-Verlag;
  7             pp. 113-146. (Springer series on environmental management).
  8       Smith, W. H. (1990b) Forests as sinks for air contaminants: vegetative compartment. In: Air pollution and forests:
  9             interactions between air contaminants and forest ecosystems. 2nd ed. New York, NY: Springer-Verlag;
10             pp. 147-180. (Springer series on environmental management).
11       Smith, W. H. (1990c) Forest nutrient cycling: toxic ions. In: Air pollution and forests: interactions between air
12             contaminants and forest ecosystems. 2nd ed. New York, NY: Springer-Verlag; pp. 225-268. (Springer series
13             on environmental management).
14       Smith, W. H. (1990d) Forest biotic agent stress: air pollutants and disease caused by microbial pathogens. In: Air
15             pollution and forests: interactions between air contaminants and forest ecosystems. 2nd ed. New York, NY:
16             Springer-Verlag; pp. 366-397. (Springer series on environmental management).
17       Smith, W.H. (1991) Air pollution and forest damage. Chem. Eng. News 69(45): 30-43.
18       Smith, K. E. C.; Jones, K. C. (2000) Particles and vegetation: implications for the transfer of particle-bound organic
19             contaminants to vegetation. Sci. Total Environ. 246: 207-236.
20       Smith, W. H.; Staskawicz, B. J. (1977) Removal of atmospheric particles by leaves and twigs of urban trees:
21             some preliminary observations and assessment of research needs. Environ. Manage. (N. Y.) 1: 317-330.
22       Soulen, P. E.; Frederick, J. E. (1999) Estimating biologically active UV irradiance from satellite radiance
23             measurements: a sensitivity study. J. Geophys. Res. 104: 4117-4126.
24       Spence, J. W.; Haynie, F. H. (1972) Paint technology and air pollution: a survey and economic assessment. Research
25             Triangle Park, NC: U.S. Environmental Protection Agency,  Office of Air Programs; publication no. AP-103.
26             Available from: NTIS, Springfield, VA; PB-210736.
27       Spence, J. W.; Haynie, F.; Upham, J. B. (1975) Effects of gaseous pollutants on paints: a chamber study. J. Paint
28             Technol. 47: 57-63.
29       Spence, J. W.; Haynie, F. H.; Lipfert, F.  W.; Cramer, S. D.; McDonald, L. G. (1992) Atmospheric corrosion model
30             for galvanized steel structures. Corrosion 48:  1009-1019.
31       Spiker, E. C.; Comer, V J.; Hosker, R. P.; Sherwood, S. I. (1992) Dry deposition of SO2 on limestone and marble:
32             the role of humidity. In: Rodrigues, J. D.; Henriques, F.; Jeremias, F. T., eds. Proceedings of the 7th
33             international congress on deterioration and conservation of stone; June; Lisbon, Portugal. Lisbon, Portugal:
34             Laboratorio Nacional de Engenharia Civil; pp. 397-406.
35       Spinka, J. (1971) Vliv znecisteneho ovzdusi na ovocne stromy a zeleninu [Effects of polluted air on fruit trees and
36             legumes]. Ziva 19:  13-15.
37       Stanhill, G.; Cohen, S. (2001) Global dimming: a review of the evidence for a widespread and significant reduction
38             in global radiation with discussion of its probable causes and possible agricultural consequences.  Agric. For.
39             Meteorol. 107: 255-278.
40       Steubing, L.; Klee, R. (1970)  Vergleichende Untersuchungen zur Staubfilterwirkung von Laub- und Nadelgeholzen
41             [Comparative investigations on the dust-filter effect of broad-leaved and coniferous trees]. Angew. Bot.
42             44:73-85.
43       Stevens, R. K.; Dzubay, T. G.; Lewis, C. W.; Shaw, R. W., Jr. (1984) Source apportionment methods applied to the
44             determination of the origin of ambient aerosols that affect visibility in forested areas. Atmos. Environ.
45             18:261-272.
46       Stoddard, J. L.  (1994) Long-term changes in watershed retention of nitrogen: its causes and aquatic consequences.
47             In: Baker, L. A., ed. Environmental chemistry of lakes and reservoirs. Washington, DC: American Chemical
48             Society; pp. 223-284. (Advances in chemistry series no. 237).
49       Stoddard, J. L.; Murdoch, P. S.  (1991) Catskill Mountains. In: Charles, D. F., ed.  Acidic deposition and aquatic
50             ecosystems: regional case studies. New York, NY: Springer-Verlag; pp. 237-271.
51       Storm, G. L.; Fosmire, G. J.; Bellis, E. D. (1994) Persistence of metals in soil and selected vertebrates in the vicinity
52             of the Palmerton zinc smelters. J. Environ. Qual. 23: 508-514.
53       Strandberg, H.; Johansson, L.-G. (1997)  The formation of black patina on copper in humid air containing traces of
54             SO2. J. Electrochem. Soc. 144: 81-89.
55       Strickland, T. C.; Holdren, G. R., Jr.; Ringold, P. L.; Bernard, D.; Smythe, K.; Fallen, W. (1993) A national critical
56             loads framework for atmospheric deposition effects assessment: I. method summary.  Environ. Manage.
57             17:329-334.


         June 2003                                      4-265         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Suzuki, D. T. (1997) Barriers to perception: from a world of interconnection to fragmentation. In: Raven, P. H., ed.
  2             Nature and Human Society: the quest for a sustainable world. Washington, DC: National Academy Press;
  3             pp. 11-21.
  4       Swietlicki, E.; Zhou, J. C.; Berg, O. H.; Martinsson, B. G.; Frank, G.; Cederfelt, S. I.; Desek, U.; Berner, A.;
  5             Birmilli, W.; Wiedensohler, A.; Yuskiewicz, B.; Bower, K. N. (1999) A closure study of sub-micrometer
  6             aerosol particle hygroscopic behaviour. Atmos. Res. 50: 205-240.
  7       Swietlik, D.; Faust, M. (1984) Foliar nutrition of fruit crops. Hortic. Rev. 6: 287-355.
  8       Sydberger, T.; Ericsson, R.  (1977) Laboratory testing of the atmospheric corrosion of steel. Werkst. Korros.
  9             28: 154-158.
10       Taback, H. J.; Brienza, A. R.; Macko, J.; Brunetz, N. (1979) Fine particle emissions from stationary and
11             miscellaneous sources in the south coast air basin: appendix. Sacramento, CA: California Air Resources
12             Board; contract no. A6-191-30.
13       Talbot, J. J. (1979) A review of the potential biological impacts of cooling tower salt drift. Atmos. Environ.
14             13:395-405.
15       Tang, I. N. (1980) Deliquescence properties and particle size change of hygroscopic aerosols. In: Willeke, K., ed.
16             Generation of aerosols and facilities for exposure experiments. Ann Arbor, MI: Ann Arbor Science
17             Publishers, Inc.; pp.  153-167.
18       Tang, I. N. (1997) Thermodynamic and optical properties of mixed-salt aerosols of atmospheric importance.
19             J. Geophys. Res. [Atmos.] 102: 1883-1893.
20       Tang, I. N.; Munkelwitz, H. R. (1994) Water activities, densities, and refractive indices of aqueous sulfates and
21             sodium nitrate droplets of atmospheric  importance. J. Geophys. Res. [Atmos.] 99: 18,801-18,808.
22       Tanner, R. L. (1981) An ambient experimental study of phase equilibrium in the atmospheric system: aerosol H+,
23             NH4+, SO42", NO3" ~NH3(g), HNO3(g).  In: Fogler, H. S., ed. Chemical reactors: based on a symposium
24             sponsored by the Division of Industrial and Engineering Chemistry at the second chemical congress of the
25             North American continent (180th ACS national meeting);  August 1980; Las Vegas, NV. Washington, DC:
26             American Chemical  Society; pp. 527-532. (ACS symposium series 168).
27       Taylor, G. E., Jr.; Hanson, P. J.; Baldocchi, D. D. (1988) Pollutant deposition to individual leaves and plant
28             canopies: sites of regulation and relationship to injury. In:  Heck,  W.  W.; Taylor, O. C.; Tingey, D. T., eds.
29             Assessment of crop loss from air pollutants. New York, NY: Elsevier Applied Science; pp. 227-257.
30       Taylor, G. E., Jr.; Johnson,  D. W.; Andersen,  C. P. (1994) Air pollution and forest ecosystems: a regional to global
31             perspective. Ecol. Appl. 4: 662-689.
32       Tegen, I.; Lacis, A. A.; Fung, I. (1996) The influence on climate  forcing of mineral aerosols from disturbed soils.
33             Nature (London) 380: 419-422.
34       Thornton, F. C.; Schaedle, M.; Raynal, D. J. (1987) Effects of aluminum on red spruce seedlings in solution culture.
35             Environ. Exp. Bot. 27: 489-498.
36       Tilman, D. (1996) Biodiversity: population versus ecosystem stability. Ecology 77: 350-363.
37       Tilman, D. (2000) Causes, consequences and  ethics of biodiversity. Nature 405: 208-211.
38       Tilman, D.; Downing, J. A.  (1994) Biodiversity and stability in grasslands.  Nature (London) 367: 363-365.
39       Tjepkema, J. D.;  Cartica, R. J.; Hemond, H. F. (1981) Atmospheric concentration of ammonia in Massachusetts and
40             deposition on vegetation. Nature (London) 294: 445-446.
41       Tolley, G. A.; Randall, A.; Blomquist, G.; Fabian, R.; Fishelson,  G.; Frankel, A.; Hoehn, J.; Krumm, R.; Mensah, E.;
42             Smith, T. (1986) Establishing and valuing the effects of improved visibility in eastern United States.
43             Washington, DC: U.S. Environmental Protection Agency.
44       Tombach, I.; Thurston, S. A. (1994) The quality of the SCENES  measurements: the roles of data quality goals and
45             evolving technology. In: Aerosols and  atmospheric optics: radiative balance and visual air quality:
46             proceedings of the international specialty conference, volume A; September; Snowbird, UT. Pittsburgh, PA:
47             Air & Waste Management Association; pp. 21-32. (A&WMA publication VTP-41).
48       Tong, S. T. Y. (1991) The retention of copper and lead paniculate matter in plant foliage and forest soil.
49             Environ. Int. 17:31-37.
50       Treshow, M. (1980) Pollution effects on plant distribution. Environ. Conserv. 7: 279-286.
51       Trettin, C.C.; Johnson, D.W.; Todd, D.E., Jr.  (1999) Forest nutrient and carbon pools at Walker Branch watershed:
52             changes during a 21-year-period.  Soil Sci. Soc. Amer. J. 63: 1436-1448.
53       Trijonis, J.; Shapland, D. (1979) Existing visibility levels in the United States: isopleth maps of visibility in
54             suburban/nonurban areas during 1974-76. Research Triangle Park, NC: U.S. Environmental  Protection
55             Agency, Office of Air Quality Planning and Standards; report no. EPA-450/5-79-010. Available from:
56             NTIS, Springfield, VA;  PB80-156177.
         June 2003                                       4-266        DRAFT-DO NOT QUOTE OR CITE

-------
  1       Trijonis, I; Thayer, M.; Murdoch, I; Hagemen, R. (1984) Air quality benefits analysis for Los Angeles and
  2             San Francisco based on housing values and visibility. Sacramento, CA: California Air Resources Board.
  3       Trijonis, J. C.; Pitchford, M.; McGown, M. (1987) Preliminary extinction budget results from the RESOLVE
  4             program. In: Bhardwaja, P. S., ed.  Visibility protection: research and policy aspects, an APCA international
  5             specialty conference; September 1986; Grand Teton National Park, WY. Pittsburgh, PA: Air Pollution
  6             Control Association; pp. 872-883.  (APCA transactions series no. TR-10).
  7       Trijonis, J. C.; Malm, W. C.; Pitchford, M.; White, W. H.; Charlson, R.; Husar, R. (1991) Visibility: existing and
  8             historical conditions-causes and effects. In: Irving, P. M., ed. Acidic deposition: state of science and
  9             technology, volume III: terrestrial, materials, health and visibility effects. Washington, DC: The U.S. National
10             Acid Precipitation Assessment Program. (State of science and technology report no. 24).
11       Turner, J.; Kelly, J. (1981) Relationships  between soil nutrients  and vegetation in a north coast forest, New South
12             Wales. Aust. For. Res. 11: 201-208.
13       Turner, J.; Lambert, M. J. (1980) Sulfur nutrition of forests. In: Shriner, D. S.; Richmond, C. R.; Lingberg, S. E.,
14             eds. Atmospheric sulfur deposition: environmental impact and health effects, proceedings of the second life
15             sciences symposium; October 1979; Gatlinburg, TN. Ann Arbor, Ml: Ann Arbor Science Publishers, Inc.;
16             pp. 321-333.
17       Turner, J.; Lambert, M. J.; Gessel, S. P. (1977) Use of foliage sulphate concentrations to predict response to urea
18             application by Douglas-fir. Can. J. For. Res. 7: 476-480.
19       Twomey, S. (1974) Pollution and the planetary albedo. Atmos. Environ. 8: 1251-1256.
20       Twomey, S. (1977) Atmospheric aerosols. Amsterdam, The Netherlands: Elsevier Scientific Publishing Company;
21             p. 237.
22       Tyler, G. (1972) Heavy metals pollute nature, may reduce productivity. Ambio 1: 52-59.
23       U.S. Code. (1990) Clean Air Act, §169B, visibility. U. S. C. 42: §7492.
24       U.S. Code. (1991) Clean Air Act, §108, air quality criteria and control techniques, §109, national ambient air quality
25             standards. U. S. C. 42: §§7408-7409.
26       U.S. Environmental Protection Agency. (1979) Protecting visibility: an EPA report to Congress. Research Triangle
27             Park, NC: Office of Air Quality Planning and Standards;  EPA report no. EPA-450/5-79-008. Available from:
28             NTIS, Springfield, VA; PB80-220320.
29       U.S. Environmental Protection Agency. (1982) Air quality criteria for paniculate matter and sulfur oxides. Research
30             Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
31             Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield, VA; PB84-156777.
32       U.S. Environmental Protection Agency. (1986 ) Air quality criteria for lead. Research Triangle Park, NC: Office of
33             Health and Environmental Assessment, Environmental Criteria and Assessment Office; EPA report no.
34             EPA-600/8-83/028aF-dF. 4v. Available from: NTIS, Springfield, VA; PB87-142378.
35       U.S. Environmental Protection Agency. (1987) Assessing the risks of trace gases that can modify the stratosphere.
36             Washington, DC: U.S. Environmental Protection Agency, Office of Air Radiation; report no. EPA
37             400/1-87/001A-H.
38       U.S. Environmental Protection Agency. (1993) Air quality criteria for oxides of nitrogen. Research Triangle Park,
3 9             NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment Office; report
40             nos. EPA/600/8-9!/049aF-cF. 3v.  Available from: NTIS, Springfield, VA; PB95-124533, PB95-124525, and
41             PB95-124517.
42       U.S. Environmental Protection Agency. (1995a) Interim findings on the status of visibility research. Research
43             Triangle Park, NC: Office of Research and Development; report no. EPA/600/R-95/021.
44       U.S. Environmental Protection Agency. (1995b) Testing of meteorological and dispersion models for use in regional
45             air quality modeling. Research Triangle Park,  NC: Office of Air Quality Planning and Standards; report no.
46             EPA-454/R-95-005. Available from: NTIS, Springfield, VA; PB95-215638.
47       U.S. Environmental Protection Agency. (1996a) Air  quality criteria for paniculate matter. Research Triangle Park,
48             NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
49       U.S. Environmental Protection Agency. (1996b) Air  quality criteria for ozone and related photochemical oxidants, v.
50             I-III. Research Triangle Park, NC:  Office of Research and Development; report no. EPA/600/P-93/004aF-cF.
51             Available from: NTIS, Springfield, VA; PB96-185582. PB96-185590, and PB96-185608. Available:
52             www.epa.gov/ncea/ozone.htm [25  February 2002].
53       U.S. Environmental Protection Agency. (1997a) Nitrogen oxides: impacts on public health and the environment.
54             Washington, DC: Office of Air and Radiation; August. Available:
55             www.epa.gov/ttncaaal/tl/reports/noxrept.pdf [1999, November 24].
56       U.S. Environmental Protection Agency. (1997b) Summary of the paniculate monitoring program. Washington, DC:
5 7             Office of Air and Radiation.


         June 2003                                      4-267         DRAFT-DO NOT QUOTE OR CITE

-------
  1       U.S. Environmental Protection Agency. (1998) Monitoring PM2 5 in ambient air using designated reference of class I
  2             equivalent methods. In: Quality Assurance Handbook for Air Pollution Measurement Systems, V. II, Part II,
  3             Section 2.12. Research Triangle Park, NC: Research and Development, Environmental Monitoring  Systems
  4             Laboratory.
  5       U.S. Environmental Protection Agency. (1999a) Protection of visibility. C. F. R. 40: §51.300-51.309.
  6       U.S. Environmental Protection Agency. (1999b) Visibility monitoring guidance. Research Triangle Park, NC:
  7             Emissions, Monitoring, and Analysis Division.
  8       U.S. Environmental Protection Agency. (2000a) Deposition of air pollutants to the great waters. Third report to
  9             Congress. [Executive Summary]. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office
10             of Air Quality Planning and Standards; report no. EPA-453/R-00-005.
11       U.S. Environmental Protection Agency. (2000b) Latest findings on national air quality: 1999 status and trends.
12             Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and
13             Standards; report no. EPA-454/F-00-002.
14       U.S. Environmental Protection Agency. (2001) National air quality and emissions trends report, 1999. Research
15             Triangle Park, NC: Office of Air Quality Planning and Standards; report no.  EPA/454/R-01-004. Available:
16             http://www.epa.gov/oar/aqtrnd99/ [11 March, 2002].
17       Ulrich, B. (1983) Interaction of forest canopies with atmospheric constituents: SO2, alkali and earth alkali  cations
18             and chloride. In: Ulrich, B.; Pankratn, J., eds. Effects of accumulation of air pollutants in forest ecosystems:
19             proceedings of a workshop; May 1982; Gottingen, Federal Republic of Germany. Dordrecht, The
20             Netherlands: D. Reidel Publishing Company; pp. 33-45.
21       United Nations Environment Programme (UNEP). (1986) Report of the international conference on the assessment
22             of the role of carbon dioxide and of other greenhouse gases in climate variations and associated impacts;
23             October 1985; Villach, Austria. Geneva, Switzerland: World Meteorological Organization; WMO no. 661.
24       United Nations Environment Programme (UNEP). (1998) Environmental effects of ozone  depletion: 1998
25             assessment. J. Photochem. Photobiol. B 46: 1-4.
26       United Nations Environment Programme (UNEP). (2000) Environmental effects of ozone  depletion: interim
27             summary. Available at: http://www.gcrio.org/ozone/unep2000summarv.html [9 April 2002].
28       Unsworth, M. H. (1984) Evaporation from forests in cloud enhances the effects of acid deposition. Nature (London)
29             312:262-264.
30       Unsworth, M. H.; Wilshaw, J. C. (1989) Wet, occult and dry deposition of pollutants on forests. Agric. For.
31             Meteorol. 47: 221-238.
32       Van Aalst, R.  M. (1982) Dry deposition of NOX. In: Schneider, T.; Grant, L., eds. Air pollution by nitrogen oxides.
33             Amsterdam, The Netherlands: Elsevier Scientific Publishing Company; pp. 263-270.
3 4       Van Breemen, N.; Van Dijk, H. F. G. (1988) Ecosystem effects of atmospheric deposition of nitrogen in The
35             Netherlands. In: Dempster, J. P.; Manning, W. J., eds. Excess nitrogen deposition. Environ. Pollut.
36             54:249-274.
37       Van Breemen, N.; Burrough, P. A.; Velthorst, E. J.; Van Dobben, H. F.; De Wit, T.; Ridder, T. B.; Reijnders, H. F.
38             R. (1982) Soil acidification from atmospheric ammonium sulphate in forest canopy throughfall. Nature
39             (London) 299: 548-550.
40       Van Breemen, N.; Mulder, J.; Driscoll, C. T. (1983) Acidification and alkalinization of soils. Plant Soil 75: 283-308.
41       Van de Hulst, H. C. (1957, reprint 1981) Light scattering by small particles. New York, NY: Wiley. (Structure of
42             matter series).
43       Van derLeun, J. C.; Tang, X.; Tevini, M. (1995) Environmental effects of ozone depletion: 1994 assessment. Ambio
44             24:  138.
45       Van derLeun, J. C.; Tang, X.; Tevini, M., eds. (1998) Environmental effects of ozone depletion: 1998 assessment.
46             J. Photochem. Photobiol. B: Biology 46(1-3): 1-108.
47       Van Dijk, H. F. G.; Roelofs, J. G. M. (1988) Effects of excessive ammonium deposition on the nutritional status and
48             condition of pine needles. Physiol. Plant. 73: 494-501.
49       Van Dorland,  R.; Dentener, F. J.; Lelieveld, J. (1997) Radiative forcing due to tropospheric ozone and sulfate
50             aerosols. J. Geophys. Res. [Atmos.] 102: 28,079-28,100.
51       Van Miegroet, H.; Cole, D. W. (1984) The impact of nitrification on soil acidification and cation leaching in red
52             alder ecosystem. J. Environ. Qual. 13: 586-590.
53       Van Miegroet, H.; Johnson, D. W.; Todd, D. E. (1993) Foliar response of red spruce saplings to fertilization with Ca
54             and Mg in the Great Smoky Mountains National Park. Can. J. For. Res. 23: 89-95.
55       Venkatram, A.; Pleim, J. (1999) The electrical analogy does not apply to modeling dry deposition of particles.
56             Atmos. Environ. 33:  3075-3076.
         June 2003                                      4-268         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Viles, H. A. (1990) The early stages of building stone decay in an urban environment. Atmos. Environ.
  2             24A: 229-232.
  3       Virginia Climate Advisory. (2000) A closer look at visibility. Charlottesville, VA: University of Virginia, Virginia
  4             State Climatology Office. (Virginia Climate Advisory: 00/8). Available:
  5             http://climate.virginia.edu/advisory/list.htm (28 May 2003).
  6       Viskari, E.-L.; Karenlampi, L. (2000) Roadside Scots pine as an indicator of deicing salt use - a comparative study
  7             from two consecutive winters. Water Air Soil Pollut. 122: 405-419.
  8       Vitousek, P. M; Mooney, H. A.; Lubchenco, I; Melillo, J. M.  (1997) Human domination of Earth's ecosystems.
  9             Science (Washington, DC) 277: 494-499.
10       Vogt, K. A.; Dahlgren, R.; Ugolini, F.; Zabowski, D.; Moore, E. E.; Zasoski, R. (1987a) Aluminum, Fe, Ca, Mg, K,
11             Mn, Cu, Zn and P in above- and belowground biomass. I. Abies amabilis and Tsuga mertensiana.
12             Biogeochemistry 4: 277-294.
13       Vogt, K. A.; Dahlgren, R.; Ugolini, F.; Zabowski, D.; Moore, E. E.; Zasoski, R. (1987b) Aluminum, Fe, Ca, Mg, K,
14             Mn, Cu, Zn and P in above- and belowground biomass. II. Pools, and circulation in a subalpine Abies
15             amabilis stand. Biogeochemistry 4: 295-311.
16       Von Glasow, R.; Bott, A. (1999) Interaction of radiation fog with tall vegetation. Atmos. Environ. 33: 1333-1346.
17       Vora, A. B.; Bhatnagar, A. R. (1987) Comparative study of dust fall on the leaves in high pollution and low
18             pollution area of Ahmedabad: V Caused foliar injury. J. Environ. Biol. 8: 339-346.
19       Vose, J. M.; Swank, W. T. (1990) Preliminary estimates of foliar absorption of 15N-labeled nitric acid vapor (HNO3)
20             by mature eastern white pine (Pinus strobus).  Can. J. For. Res. 20: 857-860.
21       Vuilleumier, L.; Harley, R.  A.; Brown, N. J.; Slusser, J. R.; Kolinski, D.; Bigelow, D. S. (2001) Variability in
22             ultraviolet total optical depth during the Southern California Ozone Study (SCOS97). Atmos. Environ.
23             35: 1111-1122.
24       Waggoner, A. P.; Weiss, R. E.; Ahlquist, N.  C.; Covert, D. S.; Will, S.; Charlson, R. J. (1981) Optical characteristics
25             of atmospheric aerosols. In: White, W. H.; Moore, D. J.; Lodge, J. P., Jr., eds. Plumes and visibility:
26             measurements and model components: proceedings of the symposium; November 1980; Grand Canyon
27             National Park, AZ. Atmos. Environ. 15: 1891-1909.
28       Wagrowski, D. M.; Hites, R. A. (1997) Polycyclic aromatic hydrocarbon accumulation in urban, suburban, and rural
29             vegetation. Environ.  Sci. Technol. 31: 279-282.
30       Wall, D. H. (1999) Biodiversity and ecosystem functioning. BioScience 49: 107-108.
31       Wall, D. H.; Moore, J. C. (1999) Interactions underground: soil biodiversity, mutualism, and ecosystem processes.
32             Bioscience 49: 109-117.
33       Walton, J. R.; Johnson, J. B.; Wood, G. C. (1982) Atmospheric corrosion initiation by  sulphur dioxide and
34             paniculate matter: II. characterisation and corrosivity of individual paniculate atmospheric pollutants.
35             Br. Corros. J. 17: 65-70.
36       Wang, W. C.; Yung, Y. L.;  Lacis, A. A.; Mo, T.; Hansen, J. E.  (1976) Greenhouse effects due to man-made
37             perturbations of trace gases. Science (Washington, DC)  194: 685-690.
3 8       Waring, R. H. (1987) Nitrate pollution: a particular danger to boreal and subalpine coniferous forests. In:
3 9             Fujimori, T.; Kimura, M., eds. Human impacts and management of mountain forests: [proceedings of a
40             symposium]. Ibaraki, Japan: Forestry  and Forest Products Research Institute; pp. 93-105.
41       Waring, R. H.;  Schlesinger, W. H. (1985) The carbon balance of trees. In: Forest ecosystems: concepts and
42             management. Orlando, FL: Academic Press, Inc.; pp. 7-37.
43       Watson, J. G. (2002) Visibility: science and regulation. J. Air Waste Manage. Assoc. 52: 628-713.
44       Watson, J. G.; Chow, J. C. (1993) Ambient air sampling. In: Willeke, K.; Baron, P. A., eds. Aerosol measurement:
45             principles, techniques, and applications. New  York, NY: Van Nostrand Reinhold; pp. 622-639.
46       Watson, J. G.; Chow, J. C. (1994) Clear sky  visibility as a challenge for society. Annu. Rev. Energy Environ.
47             19:241-266.
48       Watson, J. G.; Robinson, N. F.; Chow, J. C.; Henry,  R. C.; Kim, B. M.; Pace, T. G.; Meyer, E. L.; Nguyen, Q.
49             (1990) The USEPA/DRI chemical mass balance receptor model, CMB 7.0. Environ.  Software 5: 38-49.
50       Webb, A. H.; Bawden, R. J.; Busby, A. K.; Hopkins, J. N. (1992) Studies on the effects of air pollution on limestone
51             degradation in Great Britain. Atmos. Environ. Part B 26: 165-181.
52       Wedding, J. B.; Carlson, R. W.; Stukel, J. J.; Bazzaz, F. A. (1975) Aerosol deposition on plant leaves. Environ. Sci.
53             Technol. 9:  151-153.
54       Wedin, D. A.; Tilman, D. (1996) Influence of nitrogen loading and species composition on the carbon balance of
55             grasslands. Science 274: 1720-1723.
56       Weinbaum, S. A.; Neumann, P. M.  (1977) Uptake and metabolism of 15N-labeled potassium nitrate by French prune
57             (Primus domestica L.) leaves and the effects of two surfactants. J. Am. Soc. Hortic. Sci. 102: 601-604.


         June 2003                                       4-269         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Wellburn, A. R. (1990) Tansley review no. 24: Why are atmospheric oxides of nitrogen usually phytotoxic and not
  2             alternative fertilizers? New Phytol. 115: 395-429.
  3       Wells, B. W. (1939) A new forest climax: the salt spray climax of Smith Island, North Carolina. Torrey Bot. Club
  4             Bull. 66: 629-634.
  5       Wells, B. W.; Shunk, I. V. (1937) Seaside shrubs: wind forms vs. spray forms. Science 85: 499.
  6       Wells, B. W.; Shunk, I. V. (1938) Salt spray: an important factor in coastal ecology. Torrey Bot. Club Bull.
  7             65:485-492.
  8       Wenny, B. N.; Schafer, J. S.; DeLuisi, J. I; Saxena, V. K.; Barnard, W. F.; Petropavlovskikh, I. V.; Vergamini, A. J.
  9             (1998) A study of regional aerosol radiative properties and effects on ultraviolet-B radiation. J. Geophys. Res.
10             [Atmos.] 103: 17,083-17,097.
11       Wesely, M. L. (1989) Parameterization of surface resistances to gaseous dry deposition in regional-scale numerical
12             models. Atmos. Environ. 23: 1293-1304.
13       Wesely, M. L.; Hicks, B. B. (1977) Some factors that affect the deposition rates of sulfur dioxide and similar gases
14             on vegetation. J. AirPollut. Control Assoc. 27: 1110-1116.
15       Wesely, M. L.; Hicks, B. B. (2000) A review of the current status of knowledge on dry deposition. Atmos. Environ.
16             34:2261-2282.
17       Wesely, M. L.; Eastman, J. A.; Stedman, D. H.; Yalvac, E. D. (1982) An eddy-correlation measurement of NO2 flux
18             to vegetation and comparison to O3 flux. Atmos. Environ. 16: 815-820.
19       Wesely, M. L.; Cook, D. R.; Hart, R. L. (1983) Fluxes of gases and particles above a deciduous forest in wintertime.
20             Boundary Layer Meteorol. 27:  237-255.
21       Wesely, M. L.; Cook, D. R.; Hart, R. L.; Speer, R. E. (1985) Measurements and parameterization of paniculate
22             sulfur dry deposition over grass. J. Geophys. Res.  [Atmos.] 90: 2131-2143.
23       Wesselink, L. G.; Meiwes, K.-J.; Matzner, E.; Stein, A. (1995) Long-term changes in water and soil chemistry in
24             spruce and beech forests, Soiling, Germany. Environ. Sci. Technol. 29: 51-58.
25       Westman, W. E. (1977) How much are nature's services worth? Measuring the social benefits of ecosystem
26             functioning is both controversial and illuminating. Science (Washington, DC) 197: 960-964.
27       Whitby, K. T. (1978) The physical characteristics of sulfur aerosols. Atmos. Environ. 12: 135-159.
28       White, W. H. (1991) Contributions to light extinction. In: Irving, P. M., ed. Acidic deposition: state of science and
29             technology, v. Ill: terrestrial, materials, health and visibility effects; report 24. Washington,  DC: National
30             Acid Precipitation Assessment Program; pp. 85-102.
31       White, W. H.; Macias, E. S. (1990) Light scattering by haze and dust at Spirit Mountain, Nevada. In: Mathai, C. V.,
32             ed. Visibility and fine particles: an A&WMA/EPA international specialty conference;  October 1989; Estes
33             Park, CO. Pittsburgh, PA; Air & Waste Management Association; pp. 914-922. (A&WMA transactions series
34             no. TR-17).
35       White, W. H.; Macias, E. S.; Nininger, R. C.; Schorran, D. (1994) Size-resolved measurements of light scattering by
36             ambient particles in the southwestern U.S. A. Atmos. Environ. 28: 909-921.
37       Williams, M.  W.; Baron, J. S.; Caine, N.; Sommerfeld, R.; Sanford, R. (1996) Nitrogen saturation in the
38             Rocky Mountains. Environ. Sci. Technol. 30: 640-646.
39       Wilson, E. O. (1997) The creation of biodiversity. In: Raven, P. H., ed. Nature and human society:  the quest for a
40             sustainable world: proceedings of the 1997 forum  on biodiversity; October; sponsored by the National
41             Research Council. Washington, DC: National Academy Press.
42       Wiman, B. (1981) Aerosol collection by Scots pine seedlings: design and application of a wind tunnel method.
43             Oikos 36: 83-92.
44       Wiman, B. L. B. (1985) A non-stationary model for sulphur aerosol depletion and deposition in a young spruce
45             stand. Tellus Ser. B 37: 230-242.
46       Wiman, B. L. B.; Agren, G. I. (1985)  Aerosol depletion and deposition in forests~a model analysis. Atmos. Environ.
47             19:335-347.
48       Wiman, B. L. B.; Lannefors, H. O. (1985) Aerosol characteristics in a mature coniferous forest-methodology,
49             composition, sources and spatial concentration variations. Atmos. Environ. 19: 349-362.
50       Wiman, B. L. B.; Agren, G. I.; Lannefors, H. O. (1985) Aerosol concentration profiles within a mature coniferous
51             forest—model versus field results. Atmos. Environ. 19: 363-367.
52       Winner, W. E.; Atkinson, C. J. (1986) Absorption of air pollution by plants, and consequences for growth.
53             Trends Ecol. Evol. 1: 15-18.
54       Winner, W. E.; Bewley, J. D. (1978a) Terrestrial mosses as bioindicators of SO2 pollution stress: synecological
55             analysis and the index of atmospheric purity. Oecologia 35: 221-230.
56       Winner, W. E.; Bewley, J. D. (1978b) Contrasts between bryophyte and vascular plant synecological responses in an
5 7             SO2-stressed white spruce association in central Alberta. Oecologia 33:311-325.


         June 2003                                      4-270         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Witherspoon, J. P.; Taylor, F. G., Jr. (1971) Retention of 1-44 [i simulated fallout particles by soybean and sorghum
  2             plants. Health Phys. 21: 673-677.
  3       Wittenburg, C.; Dannecker, W. (1992) Dry deposition and deposition velocity of airborne acidic species upon
  4             different sandstones. In: Proceedings of the 1992 European aerosol conference; September; Oxford,
  5             United Kingdom. J. Aerosol Sci. 23(suppl. 1): S869-S872.
  6       Wolff, G. T. (1984) On the nature of nitrate in coarse continental aerosols. Atmos. Environ. 18: 977-981.
  7       Wolff, G. T.;  Collins, D. C.; Rodgers, W. R.; Verma, M. H.; Wong, C. A. (1990) Spotting of automotive finishes
  8             from the interactions between dry deposition of crustal material and wet deposition of sulfate. J. Air Waste
  9             Manage. Assoc. 40: 1638-1648.
10       Woodcock, A. H. (1953) Salt nuclei in marine air as a function of altitude and wind force. J. Meteorol. 10: 362-371.
11       Woodwell, G. M. (1970) Effects of pollution on the structure and physiology of ecosystems: changes in natural
12             ecosystems caused by many different types of disturbances are similar and predictable. Science (Washington,
13             DC) 168: 429-433.
14       World Health Organization. (1997) Nitrogen oxides. 2nd ed. Geneva, Switzerland: World Health Organization.
15             (Environmental health criteria 188).
16       World Meteorological Organization. (1988) Developing policies for responding to climatic change: a summary of
17             the discussions and recommendations of workshops; September-October 1987; Villach, Austria; and
18             November 1987; Bellagio, Austria. Geneva, Switzerland: World Meteorological Organization; report no.
19             WMO/TD;  no. 225. [World Climate Impact Programme series report no. WCIP-1 ].
20       World Meteorological Organization. (1999)  Scientific assessment of ozone depletion: 1998. Geneva, Switzerland:
21             World Meteorological Organization, Global Ozone and Monitoring Project; report no. 44.
22       World Resources Institute. (2000) World resources 2000-2001: people and ecosystems: the fraying web of life.
23             Washington, DC: World Resources Institute.
24       Wu, P.-M.; Okada, K. (1994) Nature of coarse nitrate particles in the atmosphere-a single particle approach. Atmos.
25             Environ. 28: 2053-2060.
26       Wyttenbach, A.; Tobler, L.; Bajo, S. (1989) Na, Cl and Br in needles of Norway spruce (P. abies) and in the aerosol
27             adhering to  the needles. Toxicol. Environ. Chem. 19: 25-33.
28       Xu, J. R.; Balik, C. M. (1989) Infrared attenuated total reflectance study of latex paint films exposed to aqueous
29             sulfur dioxide. J. Appl. Polym. Sci. 38: 173-183.
30       Yanai, R. D.;  Siccama, T.  G.; Arthur, M.. A.; Federer, C. A.; Friedland, A. J. (1999) Accumulation and depletion of
31             base cations in forest floors in the northeastern United States. Ecology 80: 2774-2787.
32       Yarnel, B.; Kalkstein, L. S.; Scheraga, J. D. (2000) Mid-Atlantic regional assessment of climate change impacts.
33             Clim. Res. (CR Special 7) 14: 153-269.
34       Yerrapragada, S. S.; Jaynes, J. H.; Chirra, S. R.; Gauri, K. L. (1994) Rate  of weathering of marble due to dry
3 5             deposition of ambient sulfur and nitrogen dioxides. Anal. Chem. 66: 655-659.
36       Yocom, J.  E.; Grappone, N. (1976) Effects of power plant emissions on materials. Palo Alto, CA: Electric Power
37             Research Institute; report no.  EPRI/EC-139. Available from: NTIS, Springfield, VA; PB-257 539.
38       Yocom, J.  E.; Upham, J. B. (1977) Effects of economic  materials and structures. In: Stern, A. C., ed. Air pollution.
39             3rd ed. Volume II. New York, NY: Academic Press; pp. 93-94.
40       Yoshida, L. C.; Allen, E. B. (2001) Response to ammonium and nitrate by a mycorrhizal annual invasive grass and a
41             native shrub in southern California. Am. J. Bot. 88: 1430-1436.
42       Young, P.  (1996) Pollution-fueled "biodeterioration" threatens historic stone: researchers are probing the role
43             microorganisms play in the decay of buildings and artwork. Environ. Sci. Technol.  30: 206A-208A.
44       Young, J. R.;  Ellis, E. C.; Hidy, G. M.  (1988) Deposition of air-borne acidifiers in the western environment.
45             J. Environ. Qual. 17: 1-26.
46       Youngs, W. D.; Rutzke, M.; Gutenmann, W. H.; Lisk, D. J. (1993) Nickel and vanadium in foliage in the vicinity of
47             an oil-fired  power plant. Chemosphere 27: 1269-1272.
48       Zappia, G.; Sabbioni, C.; Gobbi, G. (1993) Non-carbonate carbon content on black and white areas of damaged
49             stone monuments. Atmos. Environ. 27A: 1117-1121.
50       Zappia, G.; Sabbioni, C.; Pauri, M. G.; Gobbi, G. (1994) Mortar damage due to airborne sulfur compounds in a
51             simulation chamber. Mater. Struct. 27: 469-473.
52       Zappia, G.; Sabbioni, C.; Riontino, C.; Gobbi, G.; Favoni, O. (1998) Exposure tests of building materials in urban
53             atmosphere. Sci. Total Environ. 224: 235-244.
54       Zdanovskii, A. B.  (1948) New methods of calculating solubilities of electrolytes in multicomponent systems.
55             Zh.  Fiz. Khim. 22: 1475-1485.
56       Zepp, R. G.; Callaghan, T. V.; Erickson, D. J. (1998) Effects of enhanced solar ultraviolet radiation on
57             biogeochemical cycles. J. Photochem. Photobiol. B 46: 69-82.


         June 2003                                      4-271         DRAFT-DO NOT QUOTE OR CITE

-------
1      Zhang, X. Q.; McMurry, P. H.; Hering, S. V.; Casuccio, G. S. (1993) Mixing characteristics and water content of
2            submicron aerosols measured in Los Angeles and at the Grand Canyon. Atmos. Environ. Part A
3            27: 1593-1607.
4      Zhang, L.; Gong, S.; Padro, I; Barrie, L. (2001) A size-segregated particle dry deposition scheme for an
5            atmospheric aerosol module. Atmos. Environ. 35: 549-560.
6      Zipperer, W. C.; Wu, I; Pouyat, R. V.; Pickett, S. T. A. (2000) The application of ecological principles to urban and
7            urbanizing landscapes. Ecol. Appl. 10: 685-688.
8      Zuburtikudis, L; Triantafyllou, A. (2001) Environmental pollution and its impact on the monuments and works of art
9            in Greece. J. Environ.  Prot. Ecol. 2: 68-72.
       June 2003                                     4-272         DRAFT-DO NOT QUOTE OR CITE

-------
                                Appendix 4A




                      Colloquial and Latin Names
 Alder, hazel



 Alder, red



 Bean, common



 Beech



 Birch, yellow



 Blackhaw



 Brush box



 Ceanothus, hoaryleaf



 Chaparral



 Coachwood



 Corn



 Dogwood, flowering



 Elm



 Fir, balsam



 Fir, Douglas



 Fir, fraser



 Grape



 Grass, red brome



 Grass, purple moor



 Greenbriar



 Gum, sweet



 Haw, black



 Heather, Scottish
Alnus serrulata (Aiton) Willdenow



Alnus rubra Bong.



Phaseolus vulgar is L.



Fagus sylvatica L.



Betula alleghaniensis Britt.



Viburnum prunifolium L.



Lophostemon confertus (R. BR.) P.G. Wilson & Waterhouse



Ceanothus crassifolius Tony



Ceanothus crassifolius



Ceratopetalum apetalum, D.Don



Zea mays L.



Cornus florida L.



Ulmus spp.



Abies balsamea (L) Mill.



Pseudotsuga menziesii (Mirb.) Franco.



Abies fraseri (Pursh.) Poir



Vitis spp.



Bromus rubens L.



Molina caerulea (L.) Moench.



Smilax spp.



Liquidambar styraciflua L.



Viburnum prunifolium 1.



Calluna vulgaris Salisb.
June 2003
             4A-1
DRAFT-DO NOT QUOTE OR CITE

-------
 Hickory



 Hophornbeam



 Ivy, English



 Laurel, mountain



 Lichen, monks hood



 Maize



 Maple, red



 Maple



 Maple, sugar



 Mustard, small podded



 Nettle, stinging



 Oak, bur



 Oak, English



 Oak, chestnut



 Oak, live



 Oak, northern red



 Oak, turkey



 Oak, white



 Oak



 Oats, domestic



 Oats, wild



 Persimmon, common



 Pine, eastern white



 Pine, jack pine



 Pine, loblolly



 Pine, lodgepole



 Pine, Scots (Scotch)



 Pine, slash
Gary a spp.



Ostrya virginiana (Mill.) Koch



Hedera helix L.



Kalmia latifolia L.



Hypogymnia physiodes



Zea mays L.



Acer rubrum L.



Acer spp.



Acer saccharum Marsh.



Brassica geniculata L.



Urtica dioica L.



Quercus macrocarpa Michx.



Quercus rober L.



Quercusprinus = Q. montana Willd.



Quercus virginiana Mill.



Quercus rubra L.



Quercus laevis Walt.



Quercus alba L.



Quercus spp.



Avena saliva L.



Avenafatua L



Diosporos virginiana L.



Pinus strobus L.



Pinus banksiana Lamb.



Pinus taeda L.



Pinus contorta Loud.



Pinus sylvestris L.



Pinus elliotti Englem.
June 2003
              4A-2
DRAFT-DO NOT QUOTE OR CITE

-------
 Pine, lodgepole x jack
 pine
 Poplar, black
 Poplar, white
 Poplar, yellow or tulip
 Privet
 Purple Moor Grass
 Ragweed
 Rhododendron, Catawba
 Rhododendron, rosebay
 Sage, coastal
 Scottish Heather
 Soybean
 Spruce, Norway
 Spruce, red
 Spruce, sitka
 Spruce, white
 Sunflower
 Sweetgum
 Tulip poplar
Pinus contorta (Douglas ex Loud) x P. banksiana Lamb.

Populus nigra L.
Populus alba L.
Liriodendron tulipifera L.
Ligustrum spp.
Molina caerulea (L.) Moench.
Ambrosia spp.
Rhododendron catawbiense Michx.
Rhododendron maximum L.
Artemisia californica Less.
Calluna vulgaris Salisb.
Glycine max (L.) Merr.
Picea abies (L.) Karst.
Picea rubens Sarg.
Picea sitchensis (Bong.) Carr.
Picea glauca (Moench.) Voss.
Helianthus annuus L.
Liquidambar styraciflua L.
Liriodendron tulipifera L.
 EUMYCOTA-FUNGI
 Zygomycota
 Vesicular Arbuscular Mycorrhizae
 Scutellospora
 Gigaspora
 Glomus agrigatum
 Glomus leptototicum
June 2003
              4A-2
DRAFT-DO NOT QUOTE OR CITE

-------
 Ascomycotina



 Chaetomium sp.







 Fungi Imperfect!



 Aureobasidium pullulans



 Cladosporium sp.



 Epicoccum sp.



 Pestalotiopsis



 Phialophora verrucosa



 Pleurophomella =Sirodothis
June 2003                               4A-4       DRAFT-DO NOT QUOTE OR CITE

-------
 i        5.  HUMAN EXPOSURE TO PARTICULATE MATTER
 2                            AND ITS CONSTITUENTS
 3
 4
 5     5.1    INTRODUCTION
 6     5.1.1   Purpose
 7          The U.S. Environmental Protection Agency's (EPA's) regulatory authority for PM applies
 8     primarily to ambient air and those sources that contribute to ambient air PM concentrations.
 9     Most of the epidemiological studies, discussed in Chapter 8, relate measured community levels
10     of airborne pollutants to population-based health statistics.  Of necessity, these studies have to
11     rely on some simplifying assumptions regarding exposures, usually that air pollutant
12     concentrations measured at community (or population-oriented) monitoring sites (or the average
13     concentration of several such sites) can serve as surrogate indices for the average personal
14     exposure to ambient PM for the population.  However, total personal exposure to PM includes
15     both ambient and nonambient components, and both components may contribute to adverse
16     health  effects.  Thus, a major emphasis must be to develop an understanding of exposure to PM
17     from sources that contribute to ambient air pollution.  Ultimately, it will be necessary to account
18     for both ambient and nonambient components of personal exposure in order to fully understand
19     the relationship between PM and health effects. In  addition, an individual's personal  exposure to
20     ambient, nonambient, and total PM would provide useful information for studies where health
21     outcomes are tracked individually.
22          Exposure has many definitions. However, for airborne particulate matter (PM), an
23     individual's exposure is ideally based on measurements of the PM concentrations in the air in the
24     individual's breathing zone as the individual moves through space and  time. However,
25     epidemiological studies frequently use the ambient  concentration as a surrogate for exposure.
26     Therefore, understanding exposure is important because it is the individual who experiences
27     adverse health effects associated with elevated PM  concentrations.  Human exposure  data and
28     models provide the link between ambient concentrations (from monitoring data or estimated
29     with atmospheric transport models) and lung deposition and clearance models to enable
30     estimates of the source-air concentration-exposure-dose relationship for input into dose-response


       June 2003                                 5-1         DRAFT-DO NOT QUOTE OR CITE

-------
 1      assessments for PM from ambient sources.  Personal exposure includes contributions from many
 2      different types of particles, from many sources, and in many different environments.
 3           The goal of this chapter is to provide current information on the development of human
 4      exposure data and models. This includes information on (a) relationships between PM measured
 5      at ambient sites and personal exposures to PM from both ambient and nonambient sources, and
 6      (b) factors that affect these relationships. Human exposure data and models presented in this
 7      chapter provide critical links between ambient monitoring data and PM dosimetry and between
 8      toxicological studies and epidemiologic studies which are presented in other chapters.  Specific
 9      obj ective s of thi s chapter are fourfol d:
10       (1)   To provide an overall conceptual framework of exposure science as applied to PM,
              including the identification and evaluation of factors that determine personal PM
              exposure;
11       (2)   To provide a concise summary and review of recent data (since 1996) and findings from
              pertinent PM exposure studies;
12       (3)   To characterize quantitative relationships between ambient air quality measurements
              (mass, chemical components, number, etc.) as determined by a community monitoring
              site and total personal PM exposure as well as its ambient and nonambient components;
              and
13       (4)   To evaluate the implications of using ambient PM concentrations as a surrogate for
              personal exposure in epidemiologic studies of PM health effects.
14
15      5.1.2   Particulate Matter Mass and Constituents
16           Current EPA PM regulations are based on mass as a function of aerodynamic size.
17      However, EPA also measures the chemical  composition of PM in both monitoring and research
18      studies.  The composition of PM is variable and,  as discussed in Chapters 7 and 8, health effects
19      may be related to PM characteristics other than mass. Since PM from ambient air and other
20      microenvironments may have different physical and chemical  characteristics, PM from such
21      different sources may also have different health effects. Ultimately, to understand and control
22      health effects caused by PM exposures from all sources, it is important to quantify and
23      understand exposure to those chemical constituents from various sources that are responsible for
24      adverse health effects.
        June 2003                                 5-2         DRAFT-DO NOT QUOTE OR CITE

-------
 1           The National Research Council (NRC) recognized the distinction between measuring
 2      exposure to PM mass and to chemical constituents when setting Research Priorities for Airborne
 3      Particulate Matter I: Immediate Priorities and a Long-range Research Portfolio (NRC,  1998).
 4      Specifically, in Research Topic 1, Outdoor Measures versus Actual Human Exposures, NRC
 5      recommends evaluating "the relationships between concentrations of particulate matter and
 6      gaseous co-pollutants measured at stationary outdoor air monitoring sites and the contributions
 7      of these concentrations to actual personal exposures . . ." for PM mass. The NRC Research
 8      Topic 2 recommends evaluating exposures to biologically important constituents and specific
 9      characteristics of PM that cause responses in potentially susceptible subpopulations and the
10      general population. It also was recognized by the NRC that "a more targeted set of studies under
11      this research topic (#2) should await a better understanding of the physical, chemical, and
12      biological properties of airborne particles associated with the reported mortality and morbidity
13      outcomes" (NRC,  1999).  The NRC also stated that the later studies "should be designed  to
14      determine the extent to which members of the population contact these biologically important
15      constituents and size fractions of concern in outdoor air, outdoor air that has penetrated indoors,
16      and air pollutants generated indoors" (NRC, 1999).  Thus, exposure studies should include
17      contributions from all sources. The emphasis in this chapter on PM mass reflects the current
18      state of the science. Where available, data also have been provided on chemical constituents
19      although in most cases the data are limited. As recognized by the NRC, a better understanding
20      of exposures to PM chemical constituents from multiple sources will be required to more fully
21      identify, understand, and control those sources of PM contributing to adverse health effects and
22      to accurately define the relationship between PM exposure and health outcomes due to either
23      short-term or chronic exposures.
24
25      5.1.3   Relationship to Past Documents
26           Early versions of PM criteria documents did not emphasize total human exposure, but
27      rather focused almost exclusively on outdoor air concentrations. For instance, the 1969 Air
28      Quality Criteria for Particulate Matter (National Air Pollution Control Administration, 1969) did
29      not discuss either exposure or indoor concentrations. The 1982 EPA PM Air Quality Criteria
30      Document (1982 PM AQCD), however, provided some discussion of indoor PM concentrations
31      reflecting an increase in microenvironmental and personal exposure studies (U.S. Environmental

        June 2003                                 5-3         DRAFT-DO NOT QUOTE OR CITE

-------
 1     Protection Agency, 1982). The new data indicated that personal activities along with PM
 2     generated by personal and indoor sources (e.g., cigarette smoking) could lead to high indoor
 3     levels and high personal exposures to total PM.  Some studies reported indoor concentrations
 4     that exceeded PM concentrations found in the air outside the monitored microenvironments or at
 5     nearby monitoring sites. Between 1982 and 1996, many more studies of personal and indoor PM
 6     exposure demonstrated that in most inhabited domestic environments indoor PM mass
 7     concentrations and personal PM exposures of the residents were greater than ambient PM mass
 8     concentrations measured simultaneously (e.g., Sexton et al., 1984; Spengler et al., 1985; Clayton
 9     et al., 1993). As a result, the NRC (1991) recognized the potential importance of indoor sources
10     of contaminants (including PM) in causing adverse health outcomes.
11           The 1996 AQCD (U.S. Environmental Protection Agency, 1996) reviewed the human PM
12     exposure literature through early 1996 mainly to evaluate the use of ambient air monitors as
13     surrogates for PM exposure in epidemiological studies. Many of the studies cited showed poor
14     correlations between personal exposure or indoor measurements of PM mass and outdoor or
15     ambient site measurements. Conversely, Janssen et al. (1995) and Tamura et al. (1996a) showed
16     that in the absence of major nonambient sources, total PM exposures to individuals tracked
17     through time were highly correlated with ambient PM concentrations.  Analyses of these latter
18     two studies led to consideration of ambient and nonambient exposures as separate components of
19     total personal exposure. As a result, the 1996 PM AQCD (U.S. Environmental Protection
20     Agency, 1996) distinguished between ambient and nonambient PM personal exposure for the
21     first time. This chapter builds on the work of the 1996 PM AQCD by further evaluating the
22     ambient and nonambient components of PM and by reporting research that evaluates the
23     relationship between ambient concentrations and total, ambient, and nonambient personal
24     exposure.
25
26     5.1.4  Chapter Structure
27           The chapter is organized to provide information on the principles of exposure, review the
28     existing literature, and summarize key findings and limitations in the information; the specific
29     sections are described below.
30       •  Section 5.2 discusses the basic concepts of exposure including definitions, methods for
           estimating exposure, and methods for estimating ambient air components of exposure.

       June 2003                                5-4        DRAFT-DO NOT QUOTE OR CITE

-------
 1       •  Section 5.3 presents PM mass data including a description of the key available studies, the
            relationship of PM exposures with ambient concentrations, and factors that affect the
            relationship.
 2       •  Section 5.4 presents data on PM constituents including a description of the key available
            studies, the relationship with ambient concentrations, and factors that affect the
            relationship.
 3       •  Section 5.5 discusses the implications of using ambient PM concentrations in
            epidemiologic studies of PM health effects.
 4       •  Section 5.6 summarizes key findings and limitations of the information.
 5
 6
 7      5.2   BASIC CONCEPTS OF EXPOSURE
 8      5.2.1   The Concept of Exposure
 9           "There is reasonable agreement that human exposure [to a substance] means contact with
10      the chemical or agent." However, contact can be either with "(a) the visible exterior of the
11      person (skin and openings into the body such as mouth and nostrils), or (b) the so-called
12      exchange boundaries where absorption takes place (skin, lung, gastrointestinal tract)" (Federal
13      Register, 1986). In its 1992 Guidelines for Exposure Assessment (U.S. EPA,  1992), EPA
14      defined exposure as "taking place at the visible external boundary, as in (a) above, [concluding
15      that this definition] is less ambiguous and more consistent with nomenclature  in other scientific
16      fields." This was a change from the 1986 Guidelines (Federal Register, 1986).
17
18             Under this definition, it is helpful to think of the human body as having a hypothetical outer
19             boundary separating inside the body from outside the body. This outer boundary of the body
20             is the skin and the openings into the body such as the mouth, the nostrils, and punctures and
21             lesions in the skin. As used in these Guidelines, exposure to a chemical is the contact of that
22             chemical with the outer boundary.  An exposure assessment is the quantitative or qualitative
23             evaluation of that contact; it describes the intensity, frequency, and duration of contact, and
24             often evaluates the rates at which the chemical crosses the boundary (chemical intake or
25             uptake rates), the route by which it crosses the boundary (exposure route; e.g., dermal, oral,
26             or respiratory), and the resulting amount of the chemical that actually crosses the boundary
27             (a dose) and the amount absorbed (internal dose) (U.S. EPA, 1992).

        June 2003                                   5-5        DRAFT-DO NOT QUOTE OR CITE

-------
 1           When applied to PM exposure by the inhalation route, the concentration of interest is that
 2      of PM in the air which enters the respiratory system, not the average concentration of inspired
 3      and exhaled air which exists at the mouth or nostrils.  Therefore, a measurement of inhalation
 4      exposure to PM is based on measurements of the PM concentration near the breathing zone but
 5      not affected by exhaled air.
 6
 7      5.2.2  Components of Exposure
 8           The total exposure of an individual over a discrete period of time includes exposures to
 9      many different particles from various sources while in different microenvironments.  Duan
10      (1982) defined a microenvironment as "a [portion] of air space with homogeneous pollutant
11      concentration." It also has been defined as a volume in space for a specific time interval during
12      which the variance of concentration within the volume is significantly less than the variance
13      between that microenvironment and surrounding ones  (Mage, 1985). In general, people pass
14      through a series of microenvironments including outdoor, in-vehicle, and indoor
15      microenvironments as they go through time and space. Thus, total daily exposure for a single
16      individual to PM must be expressed as the sum of various exposures for the microenvironments
17      that the person occupies in the day (modified from National Research Council, 1991).
18           In a given microenvironment, particles may originate from a wide variety of sources.
19      For example, in an indoor microenvironment PM may be generated by (1) indoor activities,
20      (2) outdoor PM entering indoors, (3) the chemical interaction of outdoor air pollutants and
21      indoor air or indoor sources, (4) transport from another indoor microenvironment, or (5) personal
22      activities. All of these disparate sources have to be accounted for when estimating total human
23      exposure to PM.
24           An analysis of personal exposure to PM mass (or constituent compounds) requires
25      definition and discussion of several classes of particles and exposure.  In this chapter, PM
26      metrics may be described in terms of exposure or as an air concentration. PM also may be
27      described according to both its source (i.e., ambient, nonambient) and the microenvironment
28      where exposure occurs. Table  5-1 provides a summary of the terms used in this chapter, the
29      notation used for these terms, and their definition. These terms are used throughout this chapter
30      and provide the terminology for evaluating personal exposure to total PM and to PM from
31      ambient and nonambient sources.

        June  2003                                 5-6        DRAFT-DO NOT QUOTE OR CITE

-------
        TABLE 5-1. TYPES OF PARTICIPATE MATTER USED IN EXPOSURE
                            AND CONCENTRATION VARIABLES
 Term
 Notation
                             Definition
 Concentration
 Personal Exposure
                   General Definitions

            General term for the average air concentration over some specified time
            period, used with subscript to indicate concentration of a specific type of
            PM, usually expressed in ug/m3 units.

            General term for the average personal exposure over some specified time
            period, used with subscript to indicate exposure to a specific type of PM,
            quantified as the concentration at the oral/nasal contact boundary.
                                       Types of Particulate Matter
 Term
 Subscript
                             Definition
 Ambient PM
 Ambient-Outdoor
 PM

 Indoor PM

 Ambient-Indoor PM


 Indoor-generated PM

 Indoor-reaction PM


 Personal Cloud PM


 Personal PM
 (of a subject)
    a       PM in the atmosphere measured at a community ambient monitoring site
            either emitted into the atmosphere directly (primary PM) or formed in it
            (secondary PM). Major sources of PM species are industry, motor
            vehicles, commerce, domestic emissions such as wood smoke, and natural
            wind-blown dust or soil. (C without subscript is used for Ca.  Ea is used for
            exposure to ambient PM while outdoors.)

    o       Ambient PM in an outdoor microenvironment. (C0 is frequently considered
            to be equal to, or at least represented by, C.)

    i       All PM found indoors.

    ai       Ambient PM that has infiltrated indoors (i.e., has penetrated indoors and
            remains suspended).  E^ is used for ambient exposure while indoors.

    ig       PM generated indoors.

    ir       PM formed indoors by pollutants from outdoors reacting with indoor-
            generated pollutants.

    pc       PM contributing to personal exposure but not contained in indoor or
            outdoor measurements of PM, usually related to personal activities.

    s       PM at the oral/nasal contact zone as the subject moves through time and
            space.
 Term
Concentration and Exposure Variable Used without Subscripts

 Notation                                Definition
 Ambient
 Concentration

 Total Personal
 Exposure

 Ambient Exposure
 Nonambient
 Exposure
    N
Concentration measured at a community ambient air monitoring site (or the
average of several such sites).

Total personal exposure as measured by a personal exposure monitor
(PEM).

Personal exposure to the ambient component of total personal exposure,
i.e., personal exposure to that PM measured at an ambient air community
monitoring site. Includes exposure to C and Cm but not to resuspended
ambient PM previously deposited indoors.

Personal exposure to nonambient PM.
June 2003
                          5-7
                            DRAFT-DO NOT QUOTE OR CITE

-------
           TABLE 5-1 (cont'd).  TYPES OF PARTICIPATE MATTER USED IN EXPOSURE
                                AND CONCENTRATION VARIABLES
                Relationships among Concentration and Exposure Variables for a Two-Compartment Model
                                           (indoors and outdoors)
                                               T = A + N
                               A = yC + (y-1) 0^, where y = fraction of time outside
 1           Exposures are significant only if they are associated with a biologically relevant duration
 2      of contact with a substance of concern.  Application of this concept to PM exposure is
 3      complicated by a lack of understanding of the biological mechanisms of PM toxicity.  It is not
 4      certain whether the relevant duration is the instantaneous exposure to a peak concentration, or
 5      hourly, daily, or long-term exposure for months or years (or possibly all of the above).
 6      Similarly, it is not certain how PM toxicity depends on particle size or particle composition;
 7      whether number, surface area, or mass is the appropriate metric; or how PM toxicity may be
 8      influenced by conditions that might increase susceptibility (age, pre-existing disease conditions
 9      [COPD, asthma, diabetes,  etc.], exposure to infectious agents, exposure to heat or cold, stress,
10      etc.). It should be clear that a person's exposure is influenced by the distribution of many
11      variables  and parameters.  A measurement at a single point in space or time along each
12      distribution cannot easily describe a person's exposure.  Thus, it is important to think of
13      exposure  as a path function, the instantaneous exposure varying as the PM concentration and
14      composition varies as the person moves through time and space.
15           The 1997 NAAQS were developed largely on the basis of evidence from epidemiologic
16      studies that found relatively consistent associations between outdoor PM mass concentrations
17      and observed health effects. Thus, an emphasis in this chapter is on the relationship between the
18      PM concentrations measured at ambient sites and personal exposures to the PM measured at
19      those ambient sites (NRC, 1998), i.e., ambient PM exposure. Although this is an emphasis,
20      it should be kept in mind that every particle that deposits in the lung becomes part of a dose
21      delivered to the individual. It is likely that the nonambient component of total exposure also has
22      health effects which would not necessarily be detected using community time-series
23      epidemiological studies.
        June 2003                                 5-8        DRAFT-DO NOT QUOTE OR CITE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
5.2.3   Quantification of Exposure
     Quantification of inhalation exposure to PM or any other air pollutant starts with the
concept of the variation in the concentration of the air pollutant in the breathing zone (but
unperturbed by exhaled breath) as measured by a personal exposure monitor as a person moves
through time and space. The relationships among the various types of exposure quantities can
easily be seen in the hypothetical exposure time profile shown in Figure 5-1. The peak exposure,
instantaneous exposure, and average exposure have units of concentration, for PM, usually
|ig/m3. The integrated exposure has units of concentration x time.
                                                    Peak
                                                  Exposure
                                                                 Instantaneous
                                                                        ure
                                                         Time
       Figure 5-1.  Hypothetical exposure time profile: Pollutant exposure as a function of time
                   showing how the averaged exposure, integrated exposure, and peak exposure
                   relate to the instantaneous exposure. (t2 -tl = T.)
       Source:  Adapted from Duan et al. (1989); Zartarian et al. (1997).
       June 2003
                                         5-9
DRAFT-DO NOT QUOTE OR CITE

-------
             The integrated exposure may be written as
                                                                                         (5-1)
 3     where G is the integrated personal exposure during the time period tt to t2 and Cs is the
 4     instantaneous exposure of the subject as the subject moves through space and time (Lioy, 1990;
 5     NRC, 1991; Georgopoulos and Lioy, 1994).  Cs is a function of both time and space so in general
 6     it cannot be correctly approximated by a measurement at a single time or at a single point in
 7     space.
 8           For most of the discussion in this chapter we will be interested in the average exposure
 9     given as
                                                                                         (5-2)
10     where E is the average exposure over the time period t^-tj. Most studies report 24-hour averages
1 1     although some studies measure 12-hour or 2 or 3 day averages.
12           Equations 5-1 and 5-2 apply to a specific individual moving through time and space on a
13     specific path. When treating populations it is necessary to consider the distribution of values of
14     the variables and parameters. Georgopoulos and Lioy (1994) show how these equations can be
15     modified to  consider the probability distributions.
16
17     5.2.4  Methods To Estimate Personal Exposure
18           Personal exposure may be estimated using either direct or indirect approaches. Direct
19     approaches  measure the contact of the person with the chemical concentration in the exposure
20     media over an identified period of time. Direct measurement methods include personal exposure
21     monitors (PEMs) for PM that are worn continuously by individuals as they encounter various
22     microenvironments while performing their daily activities. Indirect approaches use models and
23     available information on concentrations of chemicals in microenvironments, the time individuals
24     spend in those microenvironments, and personal PM generating activities to  estimate personal

       June 2003                                 5-10       DRAFT-DO NOT QUOTE OR CITE

-------
 1      exposure. This section describes the methods to directly measure personal exposures and
 2      microenvironmental concentrations as well as the models used to estimate exposure.  Several
 3      approaches to estimate personal exposure to ambient PM also are described.
 4
 5      5.2.4.1   Direct Measurement Methods
 6      5.2.4.1.1 Personal Exposure Monitoring Methods
 1          In theory, personal exposure to total PM is measured by sampling the concentration of PM
 8      in inhaled air entering the nose or mouth. Practically, it is defined as that PM collected by a
 9      PEM worn by a person and sampling from a point near the breathing zone (but not affected by
10      exhaled breath).  PEMs for PM use measurement techniques similar to those used for ambient
11      PM. Most PEMs rely on filter-based mass measurement of a particle size fraction (PM10 or
12      PM25) usually integrated over either a 24- or 12-h period at flow rates of 2 to 4 L/min using
13      battery-operated pumps. PEMs must be worn by study participants; therefore, they must be
14      quiet, compact, and battery-operated. These requirements limit the type of pumps and the total
15      sample volume that can be collected. Generally, small sample volumes limit personal exposure
16      measurements to PM mass and a few elements detected by XRF. In most studies, PM25 and
17      PM10 have not been collected concurrently; thus, there are very few data available with which to
18      estimate personal exposure to coarse thoracic PM (i.e., PM10_25) exposures.
19          Other methods used for ambient PM also have been adapted for use as a PEM.
20      For example, a personal nephelometer that measures light scattering has been worn by subjects
21      and used in personal exposure studies to obtain real-time measurements of PM (Quintana et  al.,
22      2000; Rea et al., 2001; Magari et al., 2002; Lanki et al., 2002). Light scattering instruments are
23      most sensitive to particles in  the accumulation mode size range. A portable condensation nuclei
24      counter (with a lower  size limit of 20 nm diameter) has also been used in exposure studies
25      (Abraham et al., 2002).
26
27      5.2.4.1.2 Microenvironmental Monitoring Methods
28          Direct measurements of microenvironmental PM concentrations which are used with
29      models to estimate personal exposure to PM also use methods  similar to those for ambient PM.
30      These methods differ from PEMs in that they are stationary with respect to the
31      microenvironment (such as a stationary PEM). Microenvironmental monitoring methods include

        June 2003                                 5-11         DRAFT-DO NOT QUOTE OR CITE

-------
 1      filter-based mass measurements of particle size fractions (PM10, PM25) usually integrated over
 2      either a 24-or 12-h period. Flow rates vary between various devices from 4 to 20 L/min. Larger
 3      sample volumes allow more extensive chemical characterization to be conducted on
 4      microenvironmental samples.  Because more than one pumping system can be used in a
 5      microenvironment, PM2 5 and PM10 can be collected simultaneously. Other continuous ambient
 6      PM measurement methods that have been utilized for microenvironmental monitoring are the
 7      Tapered Element Oscillating Microbalance (TEOM) and nephelometers.  Various continuous
 8      techniques for counting particles by size (Climet, LASX,  SMPS, APS) also have also been used.
 9      Measurement techniques are discussed in Chapter 2.
10
11      5.2.4.2 Indirect or Modeling Methods
12      5.2.4.2.1 Personal Exposure Models
13          Exposure modeling for PM mass (PM2 5 and PM10_2 5) and chemical constituents is a
14      relatively new field facing significant methodological challenges and input data limitations.
15      Exposure models typically use one of two general approaches: (1) a time-series approach that
16      estimates microenvironmental exposures sequentially as individuals go through time or
17      (2) a time-averaged approach that estimates microenvironmental exposures using average
18      microenvironmental concentrations and the total time spent in each microenvironment.
19      Although the time-series approach to modeling personal exposures provides the appropriate
20      structure for accurately estimating personal exposures (Esmen and Hall, 2000; Mihlan et al.,
21      2000), a time-averaged approach typically is used when the input data needed to support a time-
22      series model are not available.  However, the time-varying dose profile of an exposed individual
23      can be modeled only by using the time-series approach (McCurdy, 1997, 2000).
24          Even though the processes that lead to exposure are nonlinear in nature, personal exposure
25      models are often used to combine microenvironmental concentration data with human activity
26      pattern data in order to estimate personal exposures.  Time-averaged models can be used to
27      estimate personal  exposure for an individual or for a defined population.  Total personal
28      exposure models estimate exposures for all of the different microenvironments in which a person
29      spends time, and total average personal exposure is calculated from the sum of these
30      microenvironmental exposures:
        June 2003                                 5-12        DRAFT-DO NOT QUOTE OR CITE

-------
                                   E =
                                                 ?\,
(5-3)
 1
 2     where Ey. is the personal exposure in each microenvironment, y (Duan, 1982).  Example
 3     microenvironments include outdoors, indoors at home, indoors at work, and in transit. Each
 4     microenvironmental exposure, Ep is calculated from the average concentration in
 5     microenvironment y, Cp weighted by the time spent in microenvironment k, t..  T is the sum of t.
 6     over ally. This model has been applied to concentration data in a number of studies (Ott, 1984;
 7     Ott et al., 1988, 1992; Miller et al., 1998; Klepeis et al., 1994; Lachenmyer and Hidy, 2000).
 8           Many exposure studies use 24-hour average measurements of concentration indoors and
 9     outdoors and use these concentrations with the time spent indoors and outdoors in Equation 5-3.
10     It is important to note that although measurement data may be an average concentration over
11     some time period (i.e., 24 h), significant variations in PM concentrations can occur during that
12     time period.  Thus, an error may be introduced if real-time concentrations are highly variable and
13     an average concentration for a microenvironment is used to estimate exposure when the
14     individual is in that microenvironment for only a fraction of the total time.  This may create large
15     errors if the indoor 24-hour average, e.g., in a house, includes significant time periods when
16     there are no people in the house because the indoor concentrations are increased by the activities
17     of people.  In an effort to overcome these errors, the EXPOLIS study (Kousa et al., 2002) turned
18     outdoor samplers off when the subject was indoors  and the indoor sampler off with the subject
19     was outdoors. This provides a better estimate of Ea and E; to compare with the PEM
20     measurement and allows a better calculation of Epc. However, it does not provide data that can
21     be used to regress C or C0 with C;.
22           Microenvironmental concentrations used in the exposure models can be measured directly
23     or estimated from one or more microenvironmental models. Microenvironmental models vary in
24     complexity from a simple indoor/outdoor ratio to a  multi-compartmental mass-balance model.
25     A discussion of microenvironmental models is presented in Section 5.3.4.2.2.
26           On the individual level, the time spent in the various microenvironments is obtained from
27     time/activity  diaries that are completed by the individual. For population-based estimates, the
28     time spent in various microenvironments is obtained from human activity databases.  Many of

       June 2003                                5-13        DRAFT-DO NOT QUOTE OR CITE

-------
 1      the largest human activity databases have been consolidated by EPA's National Exposure
 2      Research Laboratory (NERL) into one comprehensive database called the Consolidated Human
 3      Activity Database (CHAD).  CHAD contains over 22,000 person-days of 24-h activity data from
 4      11 different human activity-pattern studies (McCurdy et al., 2000). Population cohorts with
 5      diverse characteristics can be constructed from the activity data in CHAD and used for exposure
 6      analysis and modeling (McCurdy, 2000). These databases can also be used to estimate
 7      inhalation rates based on activity levels, age, gender, and weight for dosimetry calculations.
 8      A human activity data base may contain information or location, activity, and other information
 9      such as presence of combustion sources (e.g., wood  fireplaces, smokers) and the relative exertion
10      rates. However, in exposure studies "activity" frequently refers to a person's location in space,
11      i.e.,  in what microenvironment at what times. In dosimetry, "activity" is used as an indication of
12      the level of physical exertion and is used to estimate breathing rate and extent of mouth, nose, or
13      combined breathing.  Table 5-2 is a summary of the  human activity studies in CHAD.
14          Methodologically, personal exposure models can be divided into three general types:
15      (1) statistical models based on empirical data obtained from one or more personal monitoring
16      studies, (2) simulation models based upon known or assumed physical relationships, and
17      (3) physical-stochastic models that include Monte Carlo or other techniques to explicitly address
18      variability and uncertainty in model structure and input data (Ryan, 1991; Macintosh et al.,
19      1995).  The attributes, strengths, and weaknesses of these model types are discussed by Ryan
20      (1991), National Research Council (1991), Frey and Rhodes (1996), and Ramachandran and
21      Vincent (1999).  A review of the logic of exposure modeling is found in Klepeis (1999).
22          Personal exposure models that have been developed for PM are summarized in Table 5-3.
23      The regression-based models (Johnson et al., 2000; Janssen et al., 1997; Janssen et al., 1998a)
24      were developed for a specific purpose (i.e., to account for the observed difference between
25      personal exposure and microenvironmental measurements) and are based on data from a single
26      study, which limits their utility for broader purposes. Other types of models in Table 5-3 were
27      limited by a lack of data for the various model inputs.  For example, ambient PM monitoring
28      data is not generally of adequate spatial and temporal resolution for these models.  Lurmann and
29      Korc (1994) assumed a constant relationship between coefficient  of haze (COH) and PM10 and
30      used site-specific COH information to stochastically develop a time series of 1-h PM10 data from
31      every sixth day 24-h PM10 measurements. A mass-balance model typically  was used for indoor

        June 2003                                 5-14        DRAFT-DO NOT QUOTE OR CITE

-------
•— ^
3
to
o
o
OJ

O
H
1

o
H
O

O
H
W
O
?d
O
HH
H
W
TABLE 5-2. ACTIVITY PATTERN STUDIES INCLUDED IN THE CONSOLIDAT
ACTIVITY DATABASE (CHAD)
Diary
Calendar Time Period Documentation or
Study Name of the Study Age1 Days2 Type3 Time4 Rate5 Reference
ED HUMAN

Notes



Baltimore Jan-Feb 1997 65+ 391 Diary; 15-min 24h Standard No Williams et al. (2000a,b) Multiple days, varying from 5-15;
Jul-Aug 1998 blocks part of a PM25 PEM study
CARB: Adolescents Oct 1987-Sept 1988 12-94 1,762 Retrospective 24h Standard No Robinson et al. (1991)
and Adults Wiley et al. (1991a)
CARB: Children Apr 1989-Feb 1990 0-11 1,200 Retrospective 24h Standard No Wiley et al. (1991b)
Cincinnati (EPRI) Mar-Apr and Aug 1985 0-86 2,614 Diary 24h; nominal Yes Johnson (1989) 3 consecutive days; 186 P-D
7 pm-7 am removed7
Denver (EPA) Nov 1982-Feb 1983 18-70 805 Diary 24h; nominal No Akland et al. (1985) Part of CO PEM6 study; 2 consec.
7 pm-7 am Johnson (1984) days; 55 P-D removed7
Los Angeles: Elem. Oct 1989 10-12 51 Diary 24h Standard Yes Spier et al. (1992) 7 P-D removed7
School Children
Los Angeles: High Sept-Oct 1990 13-17 43 Diary 24h Standard Yes Spier et al. (1992) 23 P-D removed7
School Adoles.
National: NHAPS-A8 Sept 1992-Oct 1994 0-93 4,723 Retrospective 24h Standard No9 Klepeis etal. (1995) A national random-probability
Tsang and Klepeis (1996) survey
National: NHAPS-B8 As above 0-93 4,663 Retrospective 24h Standard No9 As above As above
University of Feb-Decl997 0-13 5,616 Retrospective 24h Standard No Institute for Social Research 2 days of data: one is a weekend
Michigan: Children (1997) day
Valdez, AK Nov 1990-Oct 1991 11-71 401 Retrospective Varying 24-h No Goldstein etal. (1992) 4 P-D removed7
period

Washington, DC Nov 1982-Feb 1983 18-98 699 Diary 24h; nominal No Akland et al. (1985)
(EPA) 7 pm-7 am Hartwell et al. (1984)


Part of a CO PEM6 study; 6 P-D
removed7
'All studies included both genders. The age range depicted is for the subjects actually included; in most cases, there was not an upper limit for the adult studies. Ages are inclusive. Age 0 =
< 1 year old.
2The actual number of person-days of data in CHAD after the "flagging" and removal of questionable data. See the text for a discussion of these procedures.



= babies


'Retrospective: a "what did you do yesterday" type of survey; also known as an ex post survey. Diary: a "real-time" paper diary that a subject carried as he or she went through the day.
4 Standard = midnight-to-midnight.
5 Was activity-specific breathing rate data collected?
6PEM = a personal monitoring study. In addition to the diary, a subject carried a small CO or PM2 5 monitor throughout the sampling period.
7P-D removed = The number of person-days of activity pattern data removed from consolidated CHAD because of missing activity and location information;
text.
8National Human Activity Pattern Study; A = the air version; B = the water version. The activity data obtained on the two versions are identical.
9 A question was asked regarding which activities (within each 6-h time block in the day) involved "heavy breathing," lifting heavy objects, and running hard.




completeness criteria are listed







in the





-------
                                    TABLE 5-3.  PERSONAL EXPOSURE MODELS FOR PARTICIPATE MATTER
to
O
o
Model
Study Citation Name

Model Type

Microenvironments or Predictors

Output

Notes


Time- Series Models
Hayes and PMEX
Marshall
(1999)
Johnson et al.
(2000)
Klepeis et al.
(1994)
Lurmann and REHEX-II
Korc (1994)
Deterministic
Regression-
based
Stochastic
Stochastic
Indoors: residential, work, school
Outdoors: near roadway, other
Motor vehicle
Auto travel, roadside, ETS, food prep.
grilling, high ambient PM
ETS, cooking, cleaning, attached garage,
wood burning
12 residential with different sources,
restaurant/bar, nonresidential indoors, in
Inhaled dose of PM10
Hourly for 24 h
By age/gender groups
Source contributions
PM2 5 exposure
24-h average
Respirable particle (PM3 5)
exposure
Distribution of PM10 exposure for
population
Used IAQM
Used human activity data with activity-specific breathing
rate info
Developed from scripted activity study (Chang et al
2000)

Fixed I/O ratio of 0.7 for indoors w/o sources and 1.
in transit


,2 for
          Koontz and      CPIEM
          Niang (1998)
                                                       transit, outdoors
Stochastic        Indoors:  residence, office, industrial
                plant, school, public building,
                restaurant/lounge, other.
                Outdoors, in vehicle
Three averaging times (1 h, 24 h,
season)

Distribution of PM10 exposure for
population
Reduced form mass balance model for indoors with PM
sources

Used California activity pattern and breathing rate data.
Used either a mass balance model or I/O ratio
distribution for indoor microenvironments. Indoor
sources included.

o
H
6
o
0
H
O
O
H
W
O
O
H
W
Time-Averaged Models
Clayton et al. SIM
(1999a)
Janssen et al.
(1997)
Janssen et al.
(1998a)
Ott et al. RCS
(2000)




Stochastic Distribution of annual PM2 5
exposures
Regression- Smoking parent, ETS exposure, outdoor Accounts for difference between
based physical activity personal and microenvironmental
PM10
Regression- Number of cigarettes smoked, hours of Accounts for difference between
based ETS exposure, residence on busy road, personal and microenvironmental
time in vehicle PM10
Statistical Not separated Distribution of PM10 exposure for
population





Based on 3-day ambient measurements
Children only
Adults only


A random-component superposition (RCS) model that
uses distribution of ambient PM10 and estimated
nonambient PM10 concentrations
Results for Ontario, Canada not corrected for 72-h
compared to 24-h averaging time in Riverside, CA and
Phillipsburg, NJ







-------
gr                             TABLE 5-3 (cont'd).  PERSONAL EXPOSURE MODELS FOR PARTICIPATE MATTER
ft>
to                        Model
Q        Study Citation   Name         Model Type      Microenvironments or Predictors        Output                          Notes
OJ      	
           Time-Averaged Models

           Burke et al.       SHEDS-PM    Stochastic        Outdoors, indoors:  residence, office,       PM25 exposure distributions for      A 2-stage Monte-Carlo simulation model for predicting
           (2001)                                         stores, school, in vehicle,                 population, by  age, gender,          population distribution of daily-average personal
                                                         restaurant/lounge,                       smoking and employment status;     exposures to PM.  Model has been applied to
                                                                                               PM2 5 exposure uncertainty           Philadelphia using spatially and temporally interpolated
                                                                                               predictions.  Percent contribution     PM25 ambient measurements from 1992-1993 and 1990
                                                                                               from PM of ambient origin to total    census data. Does not consider PM2 5 exposure from
                                                                                               personal exposures                 active smoking or exposure in subways

           Chao and Tung   None          Mass Balance     Indoors in unoccupied residences in        Predictions of ambient PM in        Model makes corrections for nonideal mixing (residence
           (2001)                         with Empirical    Hong Kong                            indoor microenvironments           with multiple compartments with limited intermixing)
                                         corrections
 H
 6
 o
 o
 H
O
 O
 H
 W
 O
 O
 HH
 H
 W

-------
 1      microenvironments when sufficient data were available, such as for a residence. For most other
 2      microenvironments, indoor/outdoor ratios were used because of the lack of data for a mass
 3      balance model. In addition, only the deterministic model, PMEX, included estimation of inhaled
 4      dose from activity-specific breathing rate information. Data from recent PM personal exposure
 5      and microenvironmental measurement studies should help in the future to facilitate the
 6      development of improved personal exposure models for PM.
 7           An integrated human exposure source-to-dose modeling system that will include exposure
 8      models to predict population exposures to environmental pollutants, such as PM, currently is
 9      being developed by EPA/NERL. A first-generation population exposure model for PM, called
10      the Stochastic Human Exposure and Dose Simulation (SHEDS-PM) model, recently has been
1 1      developed. The SHEDS-PM model uses a 2-stage Monte Carlo sampling technique previously
12      applied by Macintosh et al. (1995) for benzene exposures. This technique allows for separate
13      characterization of variability and uncertainty in the model predictions to predict the distribution
14      of total exposure to PM for the population of an urban/metropolitan area and to estimate the
15      contribution of ambient PM to total PM exposure. Results from a case study using data from
16      Philadelphia have been reported (Burke et al., 2001).  Work is underway to link exposure
17      modeling with dosimetry so as to provide estimates of integrated PM doses for different regions
18      of the lung.  In the future, both exposure and dose metrics generated for various subgroups of
19      concern should aid evaluation of PM health effects.
20
21      5.2.4.2.2 Microenvironmental Models
22           The mass balance model has been used extensively in exposure analysis to estimate PM
23      concentrations in indoor microenvironments (Calder, 1957; Sexton and Ryan,  1988; Duan, 1982,
24      1991; McCurdy, 1995; Johnson, 1995; Klepeis et al.,  1995; Dockery and Spengler, 1981; Ott,
25      1984; Ott et al., 1988, 1992, 2000; Miller et al., 1998; Mage et al., 1999; Wilson et al., 2000).
26      The mass balance model describes the infiltration of particles from outdoors into the indoor
27      microenvironment, the removal of particles in indoor microenvironments, and the generation of
28      particles from indoor sources:
29
30                             VdC;/dt = v,PC- vQ-ytVQ + Q;,                         (5-4)
        June 2003                                 5-18        DRAFT-DO NOT QUOTE OR CITE

-------
 1       where    V   = volume of the well-mixed  indoor air (m3);
 2                C;   = concentration of indoor PM (|ig/m3);
 3                v   = volumetric air exchange rate between indoors and outdoors (m3/h);
 4                P   = penetration ratio, the fraction of ambient (outdoor) PM that is not removed
                         from ambient air during its entry into the indoor volume;
 5                C   = concentration of PM in the ambient air (|ig/m3);
 6                k   = removal rate (h"1); and
 7                Qi   = indoor sources of particles (|ig/h).
 8
 9           Qi contains a variety of indoor, particle-generating sources, including combustion or
10      mechanical processes; condensation of vapors formed by combustion or chemical reaction;
11      suspension from bulk material; and resuspension of previously deposited PM.  The removal rate,
12      k, includes dry deposition to interior surfaces by diffusion, impaction, electrostatic forces, and
13      gravitational fallout. It may include other removal processes, such as filtration by forced air
14      heating, ventilation, or air conditioning (HVAC) or by independent air cleaners.  All parameters
15      except V are functions of time. P and k also are functions  of particle aerodynamic diameter, Da,
16      air exchange rate, v, and house characteristics such as the surface to volume ratio, type of
17      surface, etc.  All variables in Equation 5-3 will have distributions within the population and, in
18      some cases, may vary by a factor of 5 to 10. It is important to determine the distribution of these
19      variables.  Sensitivity and uncertainty analyses are necessary when attempting to explain model
20      results.
21           In addition to the mass balance model, a number of single-source or single-
22      microenvironment models exist.  However, most are used to estimate personal exposures to
23      environmental tobacco smoke (ETS). These models include both empirically based statistical
24      models and physical models based on first principles:  some are time-averaged; whereas others
25      are time series. These models evaluate the contribution of ETS to total PM exposure in an
26      enclosed microenvironment and can be applied as activity-specific components of total personal
27      exposure models. Examples of ETS-oriented personal exposure models are Klepeis (1999),
28      Klepeis et al. (1996, 2000), Mage and Ott (1996), Ott  (1999), Ott et al. (1992, 1995), and
29      Robinson et al. (1994).

        June 2003                                 5-19        DRAFT-DO NOT QUOTE OR CITE

-------
 1      5.2.4.3  Methods for Estimating Personal Exposure to Ambient Particulate Matter
 2           In keeping with the various components of PM exposure described in Section 5.3.2,
 3      personal exposure to PM can be expressed as the sum of exposure to particles from different
 4      sources summed over all microenvironments in which exposure occurs. Total personal exposure
 5      may be expressed as
 6
 7                                 T = Ea + Eai + Eig + Eir + Epc                              (5-5)
 8                                            or
 9                                        T = A + N,                                      (5-6)
10
11      where T is the total personal exposure to ambient and nonambient PM, Ea is personal exposure to
12      ambient PM while outdoors, Eai is  personal exposure to ambient PM that has infiltrated indoors
13      while indoors, Eig is personal exposure to indoor-generated PM, Eir is exposure to indoor-
14      reaction PM, and Epc is personal exposure to PM from personal activity (personal cloud). T can
15      also be expressed as A + N where A is ambient PM exposure (Ea + E^) and N is nonambient PM
16      exposure (Eig + Eir + Epc). Although personal exposure to ambient and nonambient PM cannot be
17      measured directly, they can be calculated or estimated from other measurement data.
18      Approaches for estimating these components of PM exposure are described in the following
19      section.
20
21      5.2.4.3.1 Mass Balance Approach
22      Ambient-Indoor Concentrations of Particulate Matter
23           The mass balance model described above (Equation 5-4) has been used to estimate PM
24      concentrations in indoor microenvironments.  This model also may be used to estimate ambient-
25      indoor (Cai) and indoor-generated (Cig) PM concentrations.  The mass balance model can be
26      solved for Cai and Cig assuming equilibrium conditions, i.e., all variables remain constant (Ott
27      et al., 2000; Dockery and Spengler, 1981; Koutrakis et al., 1992) and no indoor reaction PM
28      (Cir). By substituting a = v/V, where a = the number of air exchanges per hour substituting,
29      dCai + dCig for dQ in Equation 5-4, and assuming that dCai and dCig = 0, i.e., ambient-indoor PM
30      (C.J and indoor-generated PM (Cig) are at equilibrium, C^ and Cig are given by Equations 5-7
31      and 5-8.

        June 2003                                 5-20        DRAFT-DO NOT QUOTE OR CITE

-------
                                                    Pa
                                        Cai = C  ——-                                  (5-7)
                                         c" -  F(                                         (5-8>
 i
 2      Equations 5-7 and 5-8 assume equilibrium conditions and, therefore, are valid only when the
 3      parameters P, k, a, C, and Qt are not changing rapidly and when the Cs are averaged over several
 4      hours. It should be understood that equilibrium is a simplification of indoor microenvironments
 5      that are occupied by residents. This assumption of equilibrium may only represent a virtual set
 6      of individuals or populations at risk. Under certain conditions (e.g., air-conditioned homes,
 7      homes with HVAC or air cleaners that cycle on and off, or ambient pollutants with rapidly
 8      varying concentrations), nonequilibrium versions of the mass balance model (Ott et al., 2000;
 9      Freijer and Bloemen, 2000; Isukapalli and Georgopoulos, 2000) are likely to provide a more
10      accurate estimate of Cai and Cig. However, the equilibrium model provides a useful, if
1 1      simplified, example of the basic relationships  (Ott et al., 2000).
12           Equation 5-7 may be rearranged further to give Cai/C, the equilibrium fraction of ambient
13      PM that is found indoors, defined as the infiltration factor (FINF) (Dockery and Spengler, 1981).
14
                                                                                           (5-9)
                                               C   a+ k
15
16      The penetration ratio (P) and the decay rate (K) can be estimated using a variety of techniques.
17      A discussion of these variables and estimation techniques is given in Section 5.4.3.2.2. Both
18      P and k are a function of particle aerodynamic diameter, air exchange rate, and housing
19      characteristics. FINF will also be a function of these parameters and as a result FINF may vary
20      substantially within a population.  Distributions of this parameter should be estimated to
21      understand the uncertainty and variability associated with estimating exposure to PM of ambient
        June 2003                                 5-21         DRAFT-DO NOT QUOTE OR CITE

-------
 1
 2
 3
 4
 5
 6
 7
origin. The distribution of daytime FINF, as estimated from Particle Total Exposure Assessment
Methodology (PTEAM) data, is shown in Figure 5-2a (Wilson et al., 2000).

Personal Exposure to Ambient Particulate Matter
     Personal exposure to ambient PM (A) may be estimated using ambient-indoor PM
concentration (Cai) from the mass balance model, ambient outdoor PM concentrations (C), and
information on the time an individual spent in the various microenvironments. For a two-
compartment model, A may be expressed as
10
11
12
13
14
15
16
17
18
19
20
21
22
23
                                   = yC+(l-y-)C
                                                       Pa
                                                                                 (5-10)
where y is the fraction of time that an individual spent outdoors, and (1 -y) is the fraction of
time spent indoors.
     It is convenient to express personal exposure to ambient PM (A) as the product of the
ambient PM concentration (C) and a personal exposure or attenuation factor.  Following the
usage in several recent papers (Zeger et al., 2000; Dominici et al., 2000; Ott et al., 2000), the
symbol cc will be used for this attenuation factor. Equation 5-10 can be rearranged to obtain an
expression for cc:
                                      T
                                      -
                                                 Pa
                                                a+ k
                                                                                        (5-11)
Substituting equation 5-9 in equation 5-11 gives a relationship for cc in terms of the infiltration
factor FINF and the fraction of time spent in the various microenvironments:
                                              INF-
                                                                                (5-12)
       June 2003
                                         5-22
DRAFT-DO NOT QUOTE OR CITE

-------
25-
20-
•2 15-
-HJ I0
03
0)
.Q
O 10-
O
Number
o en











,— I




I — |















I — |





















































—
























n
0.20 0.25 0.300.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.900.95 1.00
                       Fraction of Ambient PM10 Found Indoors (F,NF= Cai/C)

«2°-
O
1
i15"
o
"5
1 1°"
E
5-

b|






n



^^^



n





i — |






—

















—























—







—







i — |







—



























—






n
                0.20 0.25 0.300.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
                  Fraction of Ambient PM10 Found in Total Personal PM10 Exposure
                                          (a = A / C)
Figure 5-2.  Distribution of FINF (a) and a (b) for daytime as estimated from PTEAM study
            data.

Source: Wilson et al. (2000).
June 2003
5-23
DRAFT-DO NOT QUOTE OR CITE

-------
 1      Thus, personal exposures to ambient PM (A) may be calculated from measurable quantities:
 2
 3                                         A = aC.                                     (5-13)
 4
 5      The factor a, can be measured directly or calculated from measured or estimated values of the
 6      parameters a, k, and P and the time spent in various microenvironments from activity pattern
 7      diaries (Wilson et al., 2000). Because cc depends on housing factors and lifestyle factors, air
 8      exchange rate, and PM deposition rate, it could vary to a certain extent from region to region and
 9      from season to season. Consequently, predicted exposures based on these physical modeling
10      concepts will provide exposure distributions derived conceptually as resulting from housing,
11      lifestyles, and meteorological considerations. For any given population the coefficient a may
12      represent substantial intra- and inter-personal variability based on personal activities, housing
13      characteristics, particle size, and composition. The distribution of daytime a as estimated from
14      PTEAM data is shown in Figure 5-2b. Note that the distribution of a is shifted to higher values
15      compared to FINF  because of the inclusion of time outdoors in a. Distributions of cc should be
16      determined using population studies in order to evaluate the uncertainty and variability
17      associated with model exposures.
18          The mass balance model has been used to separate indoor concentrations into ambient and
19      nonambient components. This approach, based on Equation 5-5 as given in Duan (1982) and
20      called superposition of component concentrations,  has been applied using multiple
21      microenvironments to estimate exposures to carbon monoxide (Ott, 1984; Ott et  al., 1988, 1992),
22      volatile organic compounds (Miller et al., 1998), and particles (Koutrakis et  al., 1992; Klepeis
23      et al., 1994). However, in these studies and in most of the exposure literature, the ambient and
24      nonambient components are added to yield a personal exposure from all sources  of the pollutant.
25      The use of the mass balance model, ambient concentrations,  and exposure parameters to estimate
26      exposure to ambient PM and exposure to indoor-generated PM separately as different classes of
27      exposure has been discussed in Wilson and Suh (1997) and in Wilson et al. (2000).
28
29      5.2.4.3.2   The Sulfate Ratio Technique for Estimating Ambient PM Exposure
30          The ratio of personal exposure to ambient concentration for sulfate has been recommended
31      as a technique to  estimate cc (Wilson et al., 2000). If sulfate has no indoor sources, then As = Ts.

        June 2003                                5-24        DRAFT-DO NOT QUOTE OR CITE

-------
 1      (Superscript S refers to sulfate; superscript 2.5 to PM2 5.) As can be seen in Equation 5-11, P and
 2      k depend on particle size, but a andy do not. Therefore, if sulfate and PM2 5 have the same size
 3      distribution, TS/CS = A2'5/C2'5 = ccs = cc2'5. Sulfate is formed in the ambient air via photochemical
 4      oxidation of gaseous sulfur dioxide arising from the primary emissions from the combustion of
 5      fossil fuels containing sulfur. It also arises from the direct emissions of sulfur-containing
 6      particles from nonanthropogenic sources (e.g., volcanic activity, wind-blown soil).  It has been
 7      known since the mid-1970s that sulfate and accumulation mode mass have similar size
 8      distributions (Whitby, 1978). In the indoor environment, the only common sources of sulfate
 9      may be resuspension by human activity of deposited PM containing ammonium sulfates or soil
10      sulfates that were tracked into the home. However, resuspended PM will be mostly larger than
11      PM2 5. In some homes, an unvented kerosene heater using a high-sulfur fuel may be a major
12      contributor  during winter (Leaderer et al., 1999).  Use of matches to light cigarettes or gas stoves
13      can also be  a source of sulfates.
14           Studies that have used the sulfate ratio technique to estimate a and ambient PM exposures
15      are discussed in Section 5.4.3.1.  When there are no indoor sources of accumulation-mode
16      sulfates, one may deduce that the ambient-to-personal relationship found for sulfates probably
17      would be the same as that for other PM with the same size range and physical/chemical
18      properties.  This assumption has been validated for several homes in Boston (Sarnat et al., 2002).
19      For particle sizes within the accumulation mode size range, the ratio Cai/C was similar for  sulfate
20      and PM2 5 as estimated from SMPS measurements. However, ambient PM with different
21      physical or  chemical characteristics than sulfate will not behave similarly to sulfate. Sulfate has
22      been used as a marker of outdoor air in the indoor microenvironments (Jones et al., 2000; Ebelt
23      et al., 2000). However, the personal exposure of sulfate (Ts = As) should not be taken as an
24      indicator or surrogate for ambient PM2 5 exposure (A2 5) unless it has been previously determined
25      that PM2 5 and sulfate concentrations are highly correlated. This may be the case in some air
26      sheds with high sulfate concentrations but will not be true in general.
27
28      5.2.4.3.3 Source-Apportionment Techniques
29           Source apportionment techniques provide a method for determining personal exposure to
30      PM from specific sources. If a sufficient number of samples are analyzed with sufficient
31      compositional detail, it is possible to use statistical techniques  to derive source category

        June 2003                                 5-25        DRAFT-DO NOT QUOTE OR CITE

-------
 1      signatures, identify indoor and outdoor source categories, and estimate their contribution to
 2      indoor and personal PM.  Daily contributions from sources that have no indoor component can
 3      be used as tracers to generate exposure to ambient PM of similar aerodynamic size or directly as
 4      exposure surrogates in epidemiological analyses. Studies that have used source-apportionment
 5      are discussed in Section 5.4.3.3 (i.e., Ozkaynak and Thurston, 1987; Yakovleva et al., 1999; Mar
 6      et al. 2000; Laden et al., 2000).
 7
 8
 9      5.3  SUMMARY OF PARTICULATE MATTER MASS DATA
10      5.3.1   Types of Particulate Matter Exposure Measurement Studies
11          A variety of field measurement studies have been conducted to quantify personal exposure
12      to PM mass, to measure microenvironmental concentrations of PM, to evaluate relationships
13      between personal exposure to PM and PM air concentrations measured at ambient sites, and to
14      evaluate factors that affect exposure. In general, exposure measurement studies are of two types
15      depending on how the participants are selected for the study. In a,probability study, participants
16      are selected using a probability sampling design where every member of the defined population
17      has a known, positive probability of being included in the sample. Probability study results can
18      be used to make statistical inferences about the target population.  In a purposeful or
19      nonprobability design, any convenient method may be used to enlist participants; and the
20      probability of any individual in the population being included in the sample is unknown.
21      Participants in purposeful samples may not have the same characteristics that would lead to
22      exposure as the rest of the unsampled population.  Thus, results of purposeful studies apply only
23      to the subjects sampled on the days that they were sampled and not to other subjects or other
24      periods of time. Although such studies may report significant differences, confidence intervals,
25      andp values, they do not have inferential validity (Lessler and Kalsbeek,1992). Purposeful
26      studies, however, may have generalizability (external validity).  The extent of generalizability is
27      a matter of judgement based on study participant characteristics.  Purposeful studies of PM
28      personal exposure can provide data with which to develop relationships based on important
29      exposure factors and can provide useful information for developing and evaluating either
30      statistical  or physical/chemical human exposure models.
        June 2003                                 5-26        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Regardless of the sampling design (probability or purposeful), there are three general
 2      categories of study design that can be used to measure personal exposure to PM and evaluate the
 3      relationship between personal PM exposure levels and ambient PM concentrations measured
 4      simultaneously: (1) longitudinal,  in which each subject is measured for many days; (2) pooled,
 5      in which each subject is measured for only one or two days, different days for different subjects;
 6      and (3) daily-average, in which many subjects are measured on the same day.  Only one study, in
 7      which 14 subjects were measured for 14 days, provides sufficient data for a comparison of
 8      longitudinal and daily-average data (Lioy et al., 1990). Longitudinal and pooled studies are
 9      discussed in Section 5.4.3.1.1.
10
11      5.3.2   Available Data
12      5.3.2.1  Personal Exposure Data
13           Table 5-4 gives an overview of the personal exposure studies that have been reported since
14      the 1996 PM AQCD.  In addition, major studies that were reported before that time also have
15      been included to provide a comprehensive evaluation of data in this area. Table 5-4 gives
16      information on the sampling and study designs, the study population, the season, number of
17      participants, PM exposure metric, and the PM size fraction measured.
18           Although there are a number of studies listed in the table, the data available with which to
19      evaluate longitudinal relationships and the factors that influence these are limited. Few studies
20      are based on probability sampling designs that  allow study results to be inferred to the general
21      population and to develop distributional data or exposures and the factors that affect exposure.
22      Unfortunately, none of these probability studies used a longitudinal study design. This limits our
23      ability to provide population estimates and distributional data on the relationship between
24      personal PM exposures and ambient  site measurements. In addition, most of the probability
25      studies of PM exposure were conducted during a single season; thus, variations in ambient
26      concentrations, air exchange rates, and personal activities are not accounted for across seasons.
27      In these cases,  study results are only  applicable to a specific time period. Longitudinal studies,
28      on the other hand, generally have  small sample sizes and use a purposeful sampling design.
29      Some studies did not include ambient site measurements to allow comparisons with the exposure
30      data. Approximately half of these studies monitored PM2 5. Only one or two studies measured
31      both PM10 and  PM2 5 to provide information on PM10_2 5.

        June 2003                                5-27        DRAFT-DO NOT QUOTE OR CITE

-------
•— ^
to
§


Study Design
TABLE
Study Location and
Population
5-4. SU
No. of
Subjects
MMARY OF B

Study Period
LECENT PM PERSONAL EXPOS
Age Days per PM Exposure"
Years Subject Metrics
URE STUDIES
PM Size Co-Pollutant
Measured1" Metrics Reference
Probability Studies














Lft
to

H

6
o
2|
0
H
O
C^j
o
H
W
O
^
o
HH
H
W
Longitudinal


Longitudinal




Longitudinal


Longitudinal


Longitudinal


Longitudinal



Wageningen,
Netherlands, school
children
Amsterdam (Am),
Helsinki (His), elderly
angina or coronary
heart disease

Baltimore, elderly
healthy and COPD

Fresno I
Fresno II (elderly)

Los Angeles, elderly
COPD subjects

Boston, COPD subjects



13


41 (Am)
49 (His)



21


5
16

30


18



1995


Winter 1998
Spring 1999



July-Aug 1998


Feb 1999
Apr-May 1999

Summer/Fall 1996


Winter 1996-7
Summer 1996


10-12 6 P, A, School


50-84 22 (Am) P, I, O
27 (His)



72-93 5-22 P, I, O, A


60+ 24 P, I, 0, A
24 P, I, 0, A

56-83 4 P, 1, 0


12 P, I, O, A



PM25, PM10 Janssen et al.
(1999a)

PM25 Janssen et al.
(2000)



PM2 5, PM10 CO, O3, NO2, Williams et al.
SO2 (2000a,b)

PM2 5, PM10 CO, O3 Evans et al.
PM2 5, PM10 (2000)

PM2 5 Linn et al.
(1999)

PM25, PM10 Rojas-Bracho
et al. (2000)



-------
•— ^

to
O
o
OJ
TABLE 5-4 (cont'd). SUMMARY OF RECENT PM PERSONAL EXPOSURE S


Study Design


Study Location and
Population


Age Days per PM Exposure"
No. of Subjects Study Period Years Subject Metrics

TUDIES (SINCE 1996)

PM Size Co-Pollutant
Measured1" Metrics Reference

Purposeful Studies (cont'd)













U\
i
to
VO




M
N^
^
H
6
o
2"
0
H
O

O
H
W
O
O
H
W
Longitudinal

Longitudinal

Longitudinal


Longitudinal

Longitudinal

Longitudinal



Longitudinal

Longitudinal

Pooled


Probability
Sample,
Pooled

Pooled


Longitudinal


Longitudinal
(iE diary




Nashville, TN,
COPD subjects
Vancouver, British
Columbia, COPD
Amsterdam and
Wageningen, Neth.,
school children
Amsterdam, adults

Baltimore, elderly
subjects
Baltimore, elderly,
COPD, children


Tokyo, Japan,
elderly housewives
Osaka, Japan

Milan, Italy, office
workers

Indianapolis, IN



Banska Bystrica,
Slovakia

Wageningen, NL


Mpala, Kenya





10 Summer 1995 36-88 6 P, I, O

16 April-Sept 1998 54-86 7 P, A

45 1994-1995 10-12 4-8 P, A, School


37 1994 51-70 5-8 P, I, A

15 Summer 1998 75 ±6.8 12 P
Spring 1999
56 Summer 1998 Adults: 75 ± 6.8 12 P, I, O, A
Winter 1999 Children: 9-13
COPD: 65 ±6.6

18 1992 3 P, I, O, A

26 Fall 1990-1995 Multiple days P, I, O

100 Spring/Summer 1 P, Home,
and Winter, year Office,
not stated Commuting
240 1996 16 - ? One 72-h P, I, A, O
sample/subjec
t

49 1997-1998 15-59 1 P, I, O, A


13 Mar- June 1995 12 - 14 5-8 P, A, I at school


252 1996-1998 5-75 2 years I





PM2 5, PM10 Bahadori
etal. (2001)
PM25, PM10 Ebeltetal.
(2000)
PM10 Janssen et al.
(1997)

PM10 Janssen et al.
(1998a)
PM2 5, PM10 O3, NO2, SO2 Sarnat et al.
VOCs (2000)
PM2 5 O3, NO2, SO2, Sarnat et al.
CO, EC,/OC. (2000)
VOC

SPM NO2 Tamura et al.
(1996a)
PM2,PM2.10, Tamura et al.
PM>10 (1996b)
PM10 NO2, CO, Carrer et al.
VOCs (1998)

PM2 5, PM10 Mn, Al, Ca Pellizzari
etal. (2001)


PM10, PM25 SO4~2, Braueretal.
nicotine (2000)

PM2 5, PM10 None Janssen et al.
(1999a)

Undefined CO Ezzati and
Optical MIE Kammen
(2001)




-------
•— -^
£3
to
o
o
OJ














^
i




O
H
1
O
o
0
H
O
o
H
W
O
O
H
W
TABLE 5-4 (cont'd). SUMMARY OF RECENT PM PI

Study Study Location and Age
Design Population No. of Subjects Study Period Years
Longitudinal London, UK 10 1997 9-11

Pooled Zurich, CH 10 1998 Adults

Pooled Minneapolis/St. Paul, 32 Spring, Summer, 24 - 64
MN Volunteers fall 1999
Pooled Birmingham, 1 1 healthy Season and year Adults,
UK - healthy adults, adults, not given Adults > 65
children and 1 8 susceptible Child 10
susceptibles

Pooled Santiago, Chile 8 in 1998 Winters 10-12
children 20 in 1999 1998 & 1999
Pooled Copenhagen, DK 68 subjects Winter 1999, 20 -33
non-smoking students Spring, Summer,
Fall 2000
"All based on gravimetric measurements.
bP = personal, I = indoors, O = outdoors, A = ambient.
"pooled; data from many subjects with only a few days per subject.
longitudinal; one subject measured for many days.















:RSONAL EXPOSURE STUDIES (SINCE 1996)

Days per PM Exposure" PM Size Co-Pollutant
Subject Metrics Measured1" Metrics Reference
5day/season P, I, O PM2 5, PM10 None Wheeler et al.
3 seasons (2000)
12h/dayfor P, I, O Pollen None Riediker et al.
3 days (2000)
2-15 P, I, O PM25 None Adgate et al.
(2002)
adults and P, I, O PM10 CO, NO2 Harrison et al.
children 10. (2002)
susceptibles
5, daytime
only
5 P, I, O PM25 NO2, 03 Rojas-Bracho
PM10 et al. (2002)
2 P, A PM25 None S0rensen,
BS (from PM2 5 et al. (2003)
filter)




















-------
 1           Four large-scale probability studies that quantify personal exposure to PM under normal
 2      ambient source conditions have been reported in the literature. These include the EPA's
 3      PTEAM study (Clayton et al., 1993; Ozkaynak et al.,  1996a,b); the Toronto, Ontario, study
 4      (Clayton et al., 1999a and Pellizzari et al.,  1999); the Air Pollution Exposure Distribution within
 5      Adult Urban Populations in Europe (EXPOLIS) study (Jantunen et al., 1998, 2000; Oglesby,
 6      et al., 2000; Gotshi et al., 2002; Kousa et al., 2002); and a study of a small, highly polluted area
 7      in Mexico City (Santos-Burgoa et al., 1998). A fifth study conducted in Kuwait during the last
 8      days of the oil-well fires (Al-Raheem et al., 2000) is not reported here because the ambient PM
 9      levels were not representative of normal ambient source conditions.
10           Recent longitudinal exposure studies have focused on potentially susceptible
11      subpopulations such as the young and elderly with preexisting respiratory and heart diseases
12      (hypertension, chronic obstructive pulmonary disease, and congestive heart disease).  This is in
13      keeping with epidemiological studies that indicate mortality associated with high levels of
14      ambient PM25 is greatest for elderly people with cardiopulmonary disease (U.S. Environmental
15      Protection Agency, 1996). Longitudinal studies were conducted in the Netherlands by Janssen
16      (1998) and Janssen et al. (1997, 1998a,b, 1999b,c) on purposefully selected samples of adults
17      (50 to 70 years old) and children (10 to 12 years old). School children have also been studied in
18      Chile (Rojas-Bracho et al., 2002).  Several additional  studies have focused on nonsmoking
19      elderly populations in Amsterdam and Helsinki (Janssen et al., 2000), Tokyo (Tamura et al.,
20      1996a), Baltimore, MD  (Liao et al., 1999; Williams et al., 2000a,b,c), and Fresno, CA (Evans
21      et al., 2000).  These cohorts were selected because of the low incidence of indoor sources of PM
22      (such as combustion or cooking). This should allow an examination of the relationship between
23      personal and ambient PM concentrations without the large influences caused by smoking,
24      cooking, and other indoor particle-generating activities. The EPA has a  research program
25      focused on understanding PM exposure characteristics and relationships. Within the program,
26      longitudinal  studies are being conducted on elderly participants with underlying heart and lung
27      disease (COPD, patients with cardiac defibrillator, and myocardial infarction), an elderly
28      environmental justice cohort, and asthmatics. These studies are being conducted in several cities
29      throughout the United States  and over several seasons (Rodes et  al., 2001; Conner et al., 2001;
30      Landis et al., 2001; Rea et al., 2001).
        June 2003                                 5-31         DRAFT-DO NOT QUOTE OR CITE

-------
 1           A series of studies by Phillips et al. (1994, 1996, 1997a,b, 1998a,b, 1999) examined
 2      personal ETS exposure in several European cities. Participants varied by age and occupation.
 3      Respirable particulate matter (RSP) concentrations were reported. These studies are not
 4      included in Table 5-4 because of their focus on ETS exposure (which is not the focus of this
 5      chapter).  A small personal exposure study in Zurich, Switzerland, was reported by Monn et al.,
 6      (1997) for PM10. This study also is not listed in Table 5-4 because indoor and outdoor
 7      measurements were not taken simultaneously with the personal measurements and other details
 8      of the study were not published.
 9
10      5.3.2.2  Microenvironmental Data
11           Usually, personal PM monitoring is conducted using integrated measurements over a
12      12- or 24-h period.  As such, total PM exposure estimates based on PEM measurements do not
13      capture data from individual microenvironments. Recent studies have examined PM
14      concentrations in various microenvironments using a number of different types of instruments
15      ranging from filter-based to continuous particle monitors. Details on the instruments used,
16      measurements collected, and findings of these studies according to microenvironment
17      (residential indoor, nonresidential indoor, and traffic-related) are summarized in Table 5-5.
18      Those studies which collected microenvironmental data as part of a personal exposure
19      monitoring study are summarized in Table 5-4. In general, the studies listed in Table 5-5 are
20      relatively small, purposeful studies designed to provide specific data on the factors that affect
21      microenvironmental concentration of PM from both ambient and nonambient sources.
22           Recently published studies have used various types of continuous monitors to examine
23      particle concentrations in specific microenvironments and resulting from specific activities.
24      Continuous particle monitors such as the scanning mobility particle sizer (SMPS), aerodynamic
25      particle sizer (APS), and Climet have been used to measure particle size distributions in
26      residential microenvironments (Abt et al., 2000a; Long et al., 2000a; Wallace et al., 1997;
27      Wallace, 2000a; McBride et al., 1999; Vette et al., 2001; Wallace and Howard-Reed, 2002).
28      These studies have been able to assess penetration efficiency for ambient particles to indoor
29      microenvironments, as well as penetration factors and deposition rates.  Continuous instruments
30      are also a valuable tool for assessing the impact of particle resuspension caused by human
31      activity. A semi-quantitative estimate of PM exposure can be obtained using personal

        June 2003                                 5-32        DRAFT-DO NOT QUOTE OR CITE

-------
                              TABLE 5-5.  SUMMARY OF RECENT MICROENVIRONMENTAL PM MEASUREMENT STUDIES
 to
 O
 o
           Reference
                 Study Description
Instrument(s)
Size Fraction (urn)
Summary of Measurements
                                                                                                                                                            Notes
           Residential Indoor:  Nonsmoking Homes
 H
 6
 o
 o
 H
O
 O
 H
 W
 O
 O
 HH
 H
 W
Abt et al.         2 homes,
(2000a)          2 seasons,
Boston, MA      6 days
           Long et al.
           (2000a)
           Boston, MA
           Leaderer et al.
           (1999)
           Southwest, VA
Reston, VA

Howard-Reed
et al. (2000)
Fresno, CA
Baltimore, MD
                 9 homes,
                 2 seasons
           Anuszewski      9 homes,
           etal. (1998)      18 days
           Seattle, WA
                                               SMPS

                                               APS

                                               SMPS

                                               APS




                                               Nephelometer (radiance)
                                                                                     0.02-10 urn
                 58 homes, summer
           Wallace et al.      1 home,
           (1997);           4 years
           Wallace (2000b
                             15 participants
                                               SMPS
                                               Climet
                                               PAHs
                                               Black carbon
                                              Nephelometer (personal
                                              MIE)
                                              PEM
                           PM10
                                                                                     PM,
                           6 size bins;
                           100 size channels 0.01-
                           0.4 urn
                           0.1-10 urn

                           PM,,
                                                    Detailed indoor/outdoor traces of PM in various
                                                    size classes for different air exchange rates (<
                                                    Ih'1 to > 2 h'1).

                                                    Continuous PM distributions and size
                                                    distributions obtained for indoor and outdoor air
                                                    using SMPS and APS monitors.
                         Simultaneous indoor and outdoor PM measured
                         continuously; 1-h avgtime, I/O = 0.98; air
                         exchange rate: 0.7-1.7 h"1.

                         24 h mean: Regional air 26.0 ± 11.5 ug/m3 (n =
                         47); Outdoor homes 28.0 ± 17.7 ug/m3 (n = 43);
                         Indoor w/ AC 28.9 ± 18.7 ug/m3 (n = 49);
                         Indoor w/o AC 33.3 ± 14.2 ug/m3 (n = 8).

                         24 h mean: Regional air 20.2 ± 9.9 ug/m3
                         (n = 50); Outdoor homes 21.8 ± 14.8 ug/m3 (n =
                         43); Indoor w/ AC 18.7 ± 13.2 ug/m3 (n = 49);
                         Indoor w/o AC 21.1 ±7.5 ug/m3 (n = 9).

                         Time activity data, whole-house air exchange
                         rates.
                         Continuous carbon monoxide; descriptive data
                         for monitored pollutants; size profiles for six
                         indoor particle sources.

                         Continuous (15-min avg) PM and time activity
                         data; 24-h PM mass; participants from
                         Baltimore and Fresno PM panel studies.
                         Descriptive statistics from each study for five
                         microenvironments.
                                                                                                                                                            Major indoor sources of PM: cooking,
                                                                                                                                                            cleaning, human activity.
                                              Sources of fine particles:  cooking and
                                              outdoor particles; Sources of coarse
                                              particles:  cooking, cleaning, indoor
                                              activities. 50% of particles by volume
                                              generated by indoor events were ultrafine
                                              particles.

                                              Homes contained asthmatic children,
                                              heavy wood burning. Dominant source of
                                              fine particles was outdoor air.

                                              Epidemiologic study of maternal and
                                              infant health effects associated with
                                              indoor air pollution.
                                                                                                                                                 Indoor PM concentrations were lower for
                                                                                                                                                 homes with air conditioning (AC) than
                                                                                                                                                 non-air-conditioned homes.
                                              0.3- to 0.5-um particles linked to outdoor
                                              concentrations, frying, broiling; 0.5- to
                                              2.5-um particles related to cooking
                                              events; > 2.5-um particles influenced by
                                              physical movement.

                                              Time-series plots of personal
                                              nephelometer data showed that each
                                              participant's PM exposure consisted of a
                                              series of short-term peaks, imposed on a
                                              background caused by ambient PM
                                              concentrations.

-------
c
3
to
o
o
OJ
                       TABLE 5-5 (cont'd).  SUMMARY OF RECENT MICROENVIRONMENTAL PM MEASUREMENT STUDIES
           Reference
                           Study Description
Instrument(s)
                                                                        Size Fraction (urn)
                                                   Summary of Measurements
                                                                                                                                                        Notes
           Residential Indoor: Nonsmoking Homes (cont'd)
O
HH
H
W
Rea et al.         15 participants
(2001)
Baltimore, MD
Fresno, CA
           Quintana et al.    Asthmatic children indoor
           (2000)           and outdoor
           San Diego, CA    9 homes
           Chang et al.       1 person performing
           (2000)           predetermined activities
           Baltimore, MD
Nephelometer (personal
MIE)
PEM
                                             Nephelometer (personal
                                             MIE)
                                             Harvard impactors
                                             TEOM

                                             "Roll around" monitor
                                             (RAS)(PM25, CO, VOC,
                                             03N02S02)
                                                                                  0.1-10 urn

                                                                                  PM25 and PM10




                                                                                  0.1-10 urn

                                                                                  PM25 and PM10


                                                                                  PM2,
           Lioy et al.
           (1999)
           NA
                            10 vacuum cleaners
                                                                                  0.3-0.5 urn
                                                                                                           54 ±31% of average daily PM25 exposure
                                                                                                           occurred indoor residences, where participants
                                                                                                           spent 83 ± 10% of their time. A significant
                                                                                                           portion of PM2 5 exposure occurred where
                                                                                                           participants spent 4-13% of their time.
                                                   Nephelometer correlates best with PM2 5:
                                                   vs. Indoor PM25 r = 0.66, vs. indoor PM10 r =
                                                   0.13, vs. outdoor PM25 r = 0.42, vs. outdoor
                                                   PM10r = 0.20.

                                                   1-h personal O3 exposures were significantly
                                                   lower in indoor than outdoor
                                                   microenvironments. 1-h personal CO exposures
                                                   were highest in vehicles. Personal and ambient
                                                   PM2 5 correlations were strongest for outdoor
                                                   microenvironments and those with high air
                                                   exchange rates (i.e., vehicles).

                                                   Vacuum cleaners ranged in collection efficiency
                                                   from 29-99%. Substantial fine particle
                                                   emissions from motors with emission rates from
                                                   0.028 - 128.8 ug/min.
                                                                                                Continuous (15-min avg) PM and time
                                                                                                activity data; 24-h PM mass; modeled PM
                                                                                                mass and time activity data to apportion
                                                                                                time spent in a location. Good
                                                                                                comparison with nephelometer mass
                                                                                                (6-20%).

                                                                                                Indoor and outdoor measurements
                                                                                                collected using passive, active, and active
                                                                                                heated nephelometers for comparison to
                                                                                                PM mass measurements.

                                                                                                1-h personal exposures measured
                                                                                                simultaneously. Personal and ambient
                                                                                                concentrations were compared.



O
>
— ]
1
o
o
2|

H
O
O
H
W
O
Ezzati and
Kammen
(2001)
Mpala, Kenya
Chao and Tung
(2001)
Hong Kong










55 Native huts MiniRam (MIE)
2- years


5 unoccupied homes Dust-Trak (TSI)
measured indoors and
outdoors, along with air
exchange rates









Not specified.
Optical device detects
particles 1-10 um, but it
is not PM10
PM2 5 real time,
calibrated against an
Andersen Mark II










Measured PM surrounding wood fires in
unvented huts. PM measures were up to 8000
ug/m3, but uncalibrated against wood smoke.

In the limit as air exchange goes to zero, there
appears to be a residual source, perhaps from
drafts or thermal effects.
Above air exchange rates of 4. 5/hr penetration
goes to 1, but indoor turbulence resuspends
previously settled PM2 5.







Exposures were related to ARI.



Developed an excellent model for ambient
PM infiltration in the absence
of anthropogenic indoor sources.











-------
                 TABLE 5-5 (cont'd).  SUMMARY OF RECENT MICROENVIRONMENTAL PM MEASUREMENT STUDIES
to
O
o
        Reference
Study Description
Instrument(s)
Size Fraction (urn)
Summary of Measurements
                                                                                                                  Notes
H

6
o


o
H

O

O
H
W

O


O
HH
H
W
Residential Indoor: Nonsmoking Homes (cont'd)
Fischer et al.
(2000)
Amsterdam,
NL


Kingham et al.
(2000)
Huddersfield,
UK
Morawska
etal. (2001)
Brisbane,
Australia

Abraham et al.
(2002)

Geller et al.
(2002)



Gotschi et al.
(2002)


Sarnat et al.
(2002)



Wallace and
Howard-Reed
(2002)


Measured traffic related
differences of PM and VOCs,
indoor/outdoor in 1 8 paired
homes at varying distances
from traffic

Measured PM at ten homes
of non-smokers, < 50 m and
> 300 m from traffic

Measured PM indoors and
outdoors at 16 homes while
residents were absent.
Air exchange rate estimated,
not measured
Measured ultrafine PM in
various home and traffic
microenvironments
Comparison of indoor and
outdoor PM in homes in
Coachella Valley, CA


Measured Black Smoke (BS)
by reflection from PM2 5
filters for 4 European cities,
indoor and outdoor.
Measured PM2 5 for sulfate
analysis in and out of
6 Boston homes


Measured ultrafine, fine and
coarse PM in one Reston, VA
tri-level home for 1 8 months
with air exchange and
meteorologic data
Harvard Impactors





Harvard Impactors



Scanning mobility particle
sizer, aerodynamic particle
sizer, and a TSI dust-trak


TSI 8525


USC Personal PM 5 Lpm




XPOLIS PM monitor and
EEL 43 reflectometer.


Harvard Impactors with
SMPS and APS particle
counters


TSI 3071 SMPS
TSI 3320 APS
Climet 500-1
MIEpDR-1000

PM2 5 and PM10. EC was
measured by reflectance
of the PM2 5 filters. PAH
also measured as
indicator of diesel
traffic.
PM25andPM10and
PAH. EC measured by
filter reflectance.

Submicron PM,
Supramicron PM,
PM25


< 1 [im optical diameter


PM25 and PM10.25




PM25



PM25




10 nmto > 10 (im




Outdoor PM10 and PM25 were approximately
15-20% higher at higher traffic streets than at
the quiet streets on the same days. However,
much larger differences were found for PAH
and EC which are traffic specific.

Median Indoor/outdoor ratio =1 (no indoor
combustion sources).


For supra and sub micron particles, indoor =
outdoor for normal ventilation conditions of > 2
air changes/hour.


1-second readings collected while noting
presence of sources.

12.5 = 4.3 + 0.740
110-2.5 = 3 + 0.30
I EC = 0.840 EC
I OC> O OC.

PM measured I and O for two nights while
subject was home (5 pm - 8 am). 183 sampled
PM2 5 and BS were compared for these samples.

The ratio of I/O PM2 5, sulfate and size
sub-fractions are developed as a function of
season and Air Exchange Rate.


Analysis of particle counts as a function of Air
Exchange Rate and meteorological variables.



"This [study] supports the use [of] traffic
related pollution mapping as an exposure
proxy in large-scale epidemiologic studies
into health effects of motorized traffic
emissions."

Found an absence of a spatial gradient,
perhaps due to wind direction effects (e.g.
sometimes upwind and sometimes
downwind of traffic).
Average outdoor PM concentrations are
good estimates of average indoor
concentrations of PM of ambient origin
for air exchange rates of > 0. 5/h.

The TSI 8525 is a useful instrument for
screening UFPM in microenvironments.

13 volunteers had two to four
23-h measurements I and O analyzed for
elemental carbon (EC) and organic carbon
(OC). Some unoccupied homes
measured.
BS is a supplementary measurement that
can be made on PM2 5 filters as an
indication of EC.

Sulfur is primarily of outdoor origin and
can be used to track ambient PM of
similar (0.06-0.5 [im) AD, but different
relations exist for ultrafine PM and coarse
modePM<2.5 (im AD.
Wind speed has little influence on AER.
The home sometimes acts as
1 compartment and sometimes multiple
compartments.


-------
•— ^
3
to
o
o
OJ















i^f\
i
ON



o

>
H
6
o
2|
0
H
O
O
H
W

O
H
W
TABLE 5-5 (cont'd). SUMMARY OF RECENT MICROE
Reference

Study Description

Instrument(s) Size Fraction (urn)

NVIRONMENTAL PM MEASUR
Summary of Measurements

EMENT STUDIES
Notes

Residential Indoor: Other Home Types
Brauer et al.
(1996)
Mexico

Jenkins et al.
(1996 a,b)
16 U.S. Cities
McBride et al.
(1999)
NA
Vette et al.
(2001)
Fresno, CA

Douce et al.
(2001)

22 rural Mexican homes
(smoking and nonsmoking)


Smoking and nonsmoking
homes

Combustion source (incense)
and walking
(1 room, carpeted)
Detached
semioccupied residence


ETS measured by 4 methods
in smokers homes and offices

Inertial impactor PM10
Radiance
nephelometer PM25

Fluoropore membrane Particle phase ETS
filters markers

Met-One laser particle
counter

SMPS 0.01-2.5 urn
LASX


37 mm filter and XAD-4 PM5
resin backup

Indoor PM25: 132-555 ug/m3 PM10:
282-768 ug/m3. Outdoor PM25: 37 ug/m3 PM10:
68 ug/m3; I/O PM25: 1.8-12.4; PM10: 4.7-10.0.

Mean PM3 5 concentrations were 17-20 ug/m3 in
smoking homes over nonsmoking homes.

Ratios of particle counts a 1.0 and 5.7 m from
the combustion source/activity were obtained.

Temporal relationships between indoor and
outdoor aerosol concentrations evaluated;
penetration factors and deposition rates
estimated. Fresno panel study empty residence.
Samples extracted and analyzed for UV
aborption, fluorescence, solanesol, and
scopoletin.
Variety of cooking fuels used
Nephelometer data were highly correlated
with PM25 and PM10 indoors (r = =0.87-
0.95).



Proximity to source may help explain the
existence of a personal cloud.

Diurnally variable indoor/outdoor aerosol
concentration ratios because of
resuspension from daytime activities.
Penetration factors ranged from 0.5 to 0.9.
Solanesol is best method of 4, but needs
study of solanesol stability on filter.

Nonresidential Microenvironments
Bohadana et al.
(2000)
Donham et al.
(2000)
San Francisco,
CA
Klepeis et al.
(1996)
San Francisco,
CA

Nieuwenhuijse
netal. (1999)
Teschke et al.
(1999)





Manufacturing plant,
woodworkers
34 poultry workers



Airport lounge, ETS




Agricultural activities

Wood production,
wood finishing,
wood construction workers




Not given

NIOSH Method 0600 PM5
monitors
probed respirators

TSI 8510 piezobalance PM35




PM4

PM =50






443 personal time-weighted average
occupations samples of airborne dust.
Total dust sampled indoor respiratory masks.
Personal monitoring: 630 ± 980 ug/m3 (n = 210)
ranging from 10-7,730 ug/m3.

Estimated cigarette emission rate of
1.43 mg/min/cigarette.



Average respirable fraction: 4.5 mg/m3.

1,632 observations from 1979-1997.
Arithmetic mean exposure: 7.93 mg/m3.
Geometric mean exposure: 1.86 mg/m3.






Respirable dust constituted about 10% of
total dust measured.


Personal exposures to ETS can be
modeled in these types of
microenvironments.












-------
             TABLE 5-5 (cont'd). SUMMARY OF RECENT MICROENVIRONMENTAL PM MEASUREMENT STUDIES
to
O
o
H

6
o


o
H

O

O
H
W

O


O
HH
H
W
Reference
Study Description Instrument(s)
Size Fraction (urn) Summary of Measurements
Notes
Nonresidential Microenviromnents (cont'd)
Baek et al.
(1997)
Korea
Ottetal. (1996)
California
Houseman
et al. (2002)
Boston, MA
Brauer and
Mannetje
(1998)
Vancouver, BC
Lee and Chang
(1999)
Hong Kong
Chan (2002)
Levy et al.
(2002)
Indoor and outdoor smoking
restaurants
Bar before and after smoking Piezobalance
prohibited
Indoor and outdoor TSI DusTrak
restaurants,
stores
Indoor restaurants, various
smoking policies
Indoor and outdoor
5 classrooms
A Hong Kong Office TEOM 1400a
monitored indoor and
outdoor
Arbitrary choice of one TSI Dust Irak 8520
library, coffee shop, urban
shopping mall, food court, TSI P-Trak 8525
apartment, hospital, subway,
diesel bus in Boston, MA
PM35 Indoor concentrations: 33-475 ug/m3
Outdoor concentrations: 12-172 ug/m3
I/O: 2.4.
PM3 5 Smoking permitted:
indoor 26.3-182 ug/m3; outdoor < 5-67 ug/m3
Smoking prohibited:
indoor 4-82 ug/m3; outdoor 2-67 ug/m3.
Indoor restaurants: 14-278 ug/m3
Outdoor restaurants: 7-281 ug/m3
Indoor stores: 12-206 ug/m3
Outdoor stores: 7-281 ug/m3.
PM25 Nonsmoking: PM25 7-65 ug/m3;
PM10 PM10 < 10-74 ug/m3
Restricted smoking (> 40% nonsmoking) PM2 5
11-163 ug/m3; PM10 24-89 ug/m3
Unrestricted smoking: PM25 47-253 ug/m3;
PM10 51-268 ug/m3.
PM10 Indoor PM10: 30-470 ug/m3
Outdoor PM10: 20-617 ug/m3
PM2 5 Daily 9 a.m - 6 p.m. weekday for 9 months.
PM counter calibrated to Measured inside and outside the various
PM2 5 microenvironments tested, weekdays morning
and afternoon. At least 3 visits, several 10 min
< 1 (im OD averages.
No significant correlation between indoor
and outdoor measurements.
I/O nonsmoking: 2.2.
I/O smoking: 3.4.
Avg I/O for restaurants: 2.3.
Not known if the restaurants allowed
smoking. In stores, indoor and outdoor
measurements were correlated, avg I/O:
0.83.


Temp, RH and solar intensity influence
In/Out. Wind speed has no effect.
10 min outdoor averages are poor
predictors of indoor values.

-------
                       TABLE  5-5 (cont'd).  SUMMARY OF RECENT MICROENVIRONMENTAL PM MEASUREMENT STUDIES
to
O
o
           Reference
                            Study Description
Instrument(s)
Size Fraction (urn)
                                                                                                             Summary of Measurements
                                                                                                                                                                          Notes
           Traffic-Related Microenvironments (TRM)
 5: greatest concentrations by bicycling
                         and buses.

                         PM25 background: 21-35 ug/m3; roadway 23-43
                         ug/m3.
                         PM10 background: 13-32 and 29-62 ug/m3;
                         roadway 16-56 and 30-75 ug/m3.
                                                                                                                                                          Tram > circular bus route > radial bus
                                                                                                                                                          route
                                                                                                                                                          Day > night
                                                                                                                                                          No vertical gradient (0-20 m) and
                                                                                                                                                          horizontal gradient (0-80 m) in distance
                                                                                                                                                          from road, each about 13%.
                                                                                                                                                          No significant differences between wet
                                                                                                                                                          and dry periods.

                                                                                                                                                          Air exchange rates measured at various
                                                                                                                                                          ventilation settings and speeds.
                                                                                                                                                          Monitoring vehicle followed a diesel bus
                                                                                                                                                          or truck.
                                                                                                                                                          No concentration gradient with increasing
                                                                                                                                                          distance from the roadways for PM2 5 and
                                                                                                                                                          PM10; concentration gradient did exist for
                                                                                                                                                          black smoke, also found an effect with
                                                                                                                                                          wind direction.

                                                                                                                                                          The average in-vehicle to outdoor ratio
                                                                                                                                                          was 0.99.  Average I/O: 3;  subway values
                                                                                                                                                          were correlated with outdoor
                                                                                                                                                          concentrations.
                                                                                                 Average roadway/background ratio: 3 for
                                                                                                 PM2 5 and PM10. Average increase in
                                                                                                 concentration at the roadway 7.2-12.7
                                                                                                 ug/m3.

-------
•— ^
3
to
o
o
OJ















vi
i
VO




O

p>
H
1
O
o
0
H
O
o
H
W
O
O
H
W
TABLE 5-5 (cont'd). SUMMARY OF RECENT MICROE
Reference

Study Description

Instrument(s) Size Fraction (urn)

NVIRONMENTAL PM MEASUR
Summary of Measurements

EMENT STUDIES
Notes

Traffic-Related Microenvironments (TRM) (cont'd)
Adams et al.
(2001)

Aim et al.
(1999)
Kuopio,
Finland
Chan et al.
(2002)


Hoek et al.
(2001)
Hoek et al.
(2002)
Jinsart et al.
(2002)


Lena et al.
(2002)

Zhu et al.
(2002)















PM by volunteers in TRM
in London, UK

9-km commuter route,
rush hours
1/mo

PM measured in TRM in
Hong Kong


BS in Netherlands
interpolated to outdoor
locations relative to traffic
at subject's homes
PM Exposure of traffic police
at intersection post in
Bangkok, Thailand

PM2 5 and EC in the Bronx,
NY in area of high Diesel
truck traffic
Ultrafme near Interstate
Highway in Los Angeles, CA
with heavy Diesel traffic














16 1pm personal monitor. PM2 5
Porous foam size selector

Climet 6 channels



TSI 8520 DustTrac PM25 and PM10
Calibrated to Partisol
PM25
HiVol PM10
Black Smoke measured Not Recorded
from filters


Sibata personal single PM2 5 and PM10
nozzle
2.5 1pm

PM25 @ 4 1pm PM25
3 Lpm quartz filter for EC
EEL for BS
CPC TSI 3022A 6 nm-220 nm
SMPS TSI 3936
BC aethalometer














Volunteers rode/cycled along fixed routes
repetitively.

Windows closed, vents open
air exchange rate 36-47 h"1.


Repetitive sampling over 8 fixed routes.



GIS used to interpolate background and regional
ambient BS with added increment for distance
to traffic < 50 m and < 100 m.

Sampled 12-h while on duty at post at
intersection.


Sampled 10-h at sidewalk locations while
counting cars and trucks for 3-weeks in summer
of 1999.
200 m upwind and 17, 20, 30, 90, 150 and 300
m downwind.















Personal exposures were generally double
those at fixed-site ambient station.
Subway exposures are maximal.
Morning commutes were generally higher
than afternoon commutes; relationships
determined between PM and wind speed
and vehicle speed.
Highly variable by mode.
Tram exposures are maximal.


May be useful technique. Estimates not
validated with BS measurements at
interpolated loci.

Exposures of both PM25 and PM10 higher
than ambient PM measured at station with
P gauge. Not comparable to U.S.
conditions.
EC is a large component of Diesel PM2 5
and varies with truck traffic.

Ultrafme PM decreased exponentially
from the freeway and was equal to upwind
at 300 m.















-------
 1      nephelometers that measure PM using light-scattering techniques.  Recent PM exposure studies
 2      have used condensation nuclei counters (1 s averaging time) and personal nephelometers (1 min
 3      averaging time) to measure PM continuously (Howard-Reed et al., 2000; Quintana et al., 2000;
 4      Magari et al., 2002; Lanki et al., 2002) in various microenvironments. These data have been
 5      used to identify the most important ambient and nonambient sources of PM, to provide an
 6      estimate of source strength, and to compare modeled time activity data and PEM 24-h mass data
 7      to nephelometer measurements (Rea et al., 2001).
 8
 9      5.3.2.3  Traffic-Related Microenvironments
10           There has been increasing interest in the possible role of traffic-related pollutants.
11      Distance to roadways has been used as a surrogate for exposure to traffic-related pollutants
12      (Hoek et al., 2001), and this exposure indicator was subsequently used in an epidemiological
13      study (Hoek et al., 2002).  A traffic model,  using traffic volume, direct exhaust emissions rate,
14      and a re-entrainment rate has been used to estimate concentrations  of traffic-related emissions at
15      several schools in East Los Angeles (Korenstein and Piazza, 2002). Personal exposure studies
16      have been made in a variety of commuting  situations including vehicle traffic (Adams et al.,
17      2001; Chan et al., 2002).  Other studies have measured various indicators of traffic near
18      roadways (Lena et al., 2002), inside vehicles in traffic (Abraham et al., 2002), and in several
19      types of traffic related microenvironments (Levy et al., 2002). Table 5-5 provides a brief
20      description of these studies, instruments used, measurements made, and key findings.
21
22      5.3.2.4  Reanalyses of Previously-Reported Particulate Matter Exposure Data
23           Papers that  have reanalyzed and interpreted the data collected in previous PM exposure
24      studies are summarized in Table 5-6.  These reanalyses are directed toward understanding the
25      personal cloud, the variability in total PM exposure, and the personal exposure-to-ambient
26      concentration relationships forPM. Brown and Paxton (1998) determined that the high
27      variability in personal exposure to PM makes the personal-to-ambient PM relationship difficult
28      to predict.  Wallace (2000b) used data from a number of studies to test two hypotheses:  elderly
29      COPD patients have (1) smaller personal clouds  and (2) higher correlations between personal
30      exposure and ambient concentrations compared to healthy elderly,  children, and the general
31      population. The analysis by Wallace (2000a) and three subsequent longitudinal  studies

        June 2003                                 5-40        DRAFT-DO NOT QUOTE OR CITE

-------
                     TABLE  5-6.  PAPERS REPORTING REANALYSES OF PARTICIPATE MATTER EXPOSURE STUDIES
to
O
o
           Reference
Study Cited
Objectives/Hypotheses
Findings
           Wallace         PTEAM (Ozkaynak et al., 1990;
           (2000a)         Spengler et al., 1989; Wiener 1988, 1989;
                          Wiener et al., 1990)
                          THEES (Lioy et al., 1990)
                          Nashville COPD (Bahadori et al., 2001)
                          Amsterdam COPD (Janssen et al., 1997,
                          1998a)
                          Boston COPD (Rojas-Bracho et al., 2000)
           Ozkaynak
           and Spengler
           (1996)
Dockery and Spengler, 1981

PTEAM (Ozkaynak et al., 1996a,b)

Netherlands (Janssen et al., 1995)
           Brown and      THEES (Lioy et al., 1990)
           Paxton          PTEAM pilot (Wallace, 1996)
           (1998)          Boston and Nashville COPD (Rojas-Bracho
                          et al., 2000); Bahadori et al., 2001)

           Wilson and      Philadelphia (Burton et al.,  1996;
           Suh (1997)      Suggs and Burton, 1983)

                          EPA AIRS database
           Wilson et al.     New Jersey (Lioy et al., 1990)
           (2000)          Japan (Tamura et al., 1996a)
                          PTEAM (Clayton et al., 1993;
                          Ozkaynak etal., 1996a,b)
                          Netherlands (Janssen, 1998a;
                          Suh etal., 1992)
Examines the differences between pooled and longitudinal
correlations in personal and ambient (or outdoor) data for PM2 5
and PM10.

Discusses the personal cloud for PM2 5 and PM10.

Hypothesizes that COPD patients have (1) smaller personal
clouds (supported) and (2) higher correlations of personal
exposure with outdoor concentrations because of reduced
mobility (not supported).

Uses statistical modeling techniques to examine the
relationship between ambient PM concentrations and personal
exposures. Data analysis involves use of air exchange rates,
penetration factors, and indoor/outdoor ratios, as well as
examining exposure in various microenvironments (traveling,
working, outdoors, indoors) activities (exposure to smoke,
cooking), and source strengths.
                                           Cross-sectional and longitudinal regression analysis on data
                                           sets.
                                           Determines the utility of fine and coarse PM concentrations as
                                           indicators of time-series epidemiology with regard to day-to-
                                           day variability, area uniformity, and indoor/outdoor PM ratios.
                                           Necessary to treat personal exposure to ambient PM  and
                                           personal exposure to nonambient PM as separate components
                                           of total personal PM exposure.

                                           Synoptic review of the "exposure paradox":  Low correlations
                                           between personal exposure and ambient PM concentrations in
                                           spite of the existence of statistical association between ambient
                                           PM and epidemiologic health effects.

                                           Uses personal exposure equation, mass balance, regression
                                           analysis, and deductive logic.
Median longitudinal correlation coefficient is much higher than
the pooled correlation coefficient for the same data sets.
Personal cloud for PM10: 3-67 ug/m3; PM2 5 6-27 ng/m3.
Personal cloud for elderly COPD was much smaller (PM10: 6-
11 ug/m3; PM2 5 = 6 g/m3) than for other healthy populations
(PM10: 27-56 ug/m3; PM25:  11-27 ug/m3) of elderly, children,
and the general population.  However, correlations of personal
exposure with ambient concentrations were not higher for
elderly COPD than for other groups.

The important components of personal exposures are received
during contact with indoor sources, mainly in homes and work
places.

Ambient aerosols contribute about 50% or more to the personal
PM10 exposures of the general population.
The contribution of ambient aerosols to the total toxicity of
inhaled particles is significant.

Individual personal PM exposure is subject to high variability,
which makes the personal-to-ambient PM relationship difficult
to predict.
                                                           Fine and coarse particles should be considered separate classes
                                                           of pollutants.

                                                           Fixed-site ambient fine-particle measurements likely give a
                                                           reasonable indication of the variability in the concentration of
                                                           ambient fine particles across the community. Coarse-particle
                                                           measurements most likely will not.

                                                           Personal PM exposure needs to be divided into different
                                                           classes according to source type:  exposure to ambient PM
                                                           (outdoor and indoors) and exposure to nonambient PM (indoor
                                                           source and personal activity).

                                                           Correlations are higher between personal exposure and ambient
                                                           PM concentrations when PM exposures from nonambient
                                                           sources are removed.

-------
               TABLE 5-6 (cont'd).  PAPERS REPORTING REANALYSES OF PARTICULATE MATTER EXPOSURE  STUDIES
to
O
o
           Reference
Study Cited
                                                           Objectives/Hypotheses
Findings
 75% of ambient PM25 and
> 64% of ambient PM10 measured by the community monitor.
"It is important to note that a personal measurement does not a
priori provide more valid data than a stationary ambient
measurement, i.e. a personal sample in a study investigating
effects from outdoor combustion particles is often influenced
by sources other than outdoor sources and may thus confound
the exposure-effect outcome."

"Despite some lack of correlation between personal (PM10) and
outdoor values, outdoor fine particle concentrations were
strongly associated with mortality and morbidity indicating that
outdoor sources (e.g. vehicular emissions) emit the toxic entity
(Dockerey et al.,  1993; Schwartz et al., 1996)."

Demographic bias exists because women and more-educated
individuals are more likely to respond to survey.

Socioeconomic bias exists in low SES subjects less likely to
participate in diary keeping and exposure monitoring.

Weighting is required for inter-city comparisons.

Selection bias is not a problem for characterizing physical
factors influencing personal exposure.

Distinct male vs female differences: males had higher
exposures to PM2 5, related to ETS, and a larger variance
between sociodemographic groupings.

No sociodemographic differences existed in outdoor PM2 5
concentrations.
Lower occupational status contributed to greater PM2 5
exposures than higher (professional) occupational status.

Mean personal exposure PM2 5  was higher than their apartment
concentrations. Personal cloud of 3 [ig/m3 for PM25 was
negligible but cloud for PM10 was 20 (ig /m3. Indoor PM2 5 data
were less than ambient concentrations.

-------
 1      (Williams 2000a,b,c; Ebelt et al., 2000; Sarnat et al., 2000) supported hypothesis 1 but not
 2      hypothesis 2. Ozkaynak and Spengler (1996) showed that at least 50% of personal PM10
 3      exposure for the general population comes from ambient particles. Wilson and Suh (1997)
 4      concluded that fine and coarse particles should be treated as separate classes of pollutants
 5      because of differences in characteristics and potential health effects. Wilson et al. (2000) gave a
 6      review of what they call the "exposure paradox" and determine that personal PM needs to be
 7      divided into different classes according to source type and that correlations between personal and
 8      ambient PM will be higher when nonambient sources of PM are removed from the personal PM
 9      concentration. Mage (1998) conducted analysis using the PTEAM data and showed that the
10      average person in PTEAM (Riverside, CA in the fall) was exposed to > 75% of ambient PM2 5
11      and > 64% of ambient PM10. Mage et al. (1999) used an algorithm to fill in missing data and
12      outliers to analyzed data sets and show that variation in daily personal exposures for subjects
13      with similar activity patterns and no ETS exposure are driven by variation in ambient PM
14      concentrations.
15
16      5.3.3   Factors Influencing and Key Findings on Particulate Matter
17             Exposures
18      5.3.3.1  Relationship of Personal/Microenvironmental Particulate Matter with Ambient
19             Particulate Matter
20          Understanding the relationship between ambient site measurements and personal exposure
21      to PM is important for several reasons.  First, it allows us to examine the extent to which
22      ambient measurements for PM and various PM constituents can serve as surrogates for exposure
23      to ambient PM or ambient constituents of PM in epidemiological studies.  Second, it provides
24      information that may improve surrogate exposure measurements and, hence, increase  the power
25      of epidemiologic studies.  Finally, because compliance with the NAAQS is based on ambient
26      monitoring, it can be used to understand the effect of regulation  on exposures to PM and its
27      constituents and, hence, can help link the effect of regulations to health outcomes. Many of the
28      studies summarized in Table 5-4 have analyzed this relationship using measurements  of personal
29      PM exposures and ambient PM concentrations.  Of primary interest are the PM concentrations
30      measured in ambient, indoor, and outdoor air; personal exposure measurements; the statistical
31      correlations between measurements; and the attenuation and/or infiltration factors developed for
        June 2003                                 5-43        DRAFT-DO NOT QUOTE OR CITE

-------
 1      personal exposure and indoor microenvironments.  Attenuation and infiltration factors are
 2      discussed in Section 5.3.4.3.1.  Information on correlation analysis is provided below.
 3
 4      5.3.3.1.1  Types of Correlations
 5           The three types of correlation data that will be discussed in this section are longitudinal,
 6      "pooled," and daily-average correlations. Longitudinal correlations are calculated when data
 7      from a study includes measurements over multiple days for each subject (longitudinal study
 8      design). Longitudinal correlations describe the temporal relationship between daily personal PM
 9      exposure or microenvironment concentration and daily ambient PM concentration for each
10      individual subject.  The longitudinal correlation coefficient, r, may differ for each subject.
11      An analysis of the variability in r across subjects can be performed with this type of data.
12      Typically, the median r is reported along with the range across subjects in the study.  Pooled
13      correlations are calculated when  a study involves one or only a few measurements per subject
14      and when different subjects are studied on subsequent days.  Pooled correlations combine
15      individual subject/individual day  data for the correlation calculation. Pooled correlations
16      describe the relationship between daily personal PM exposure and daily ambient PM
17      concentration across all subjects in the study. For some studies, the multiple days of
18      measurements for each subject were assumed to be independent (after autocorrelation and
19      sensitivity analysis) and combined together in the correlation calculation (Ebelt et al., 2000).
20      Daily-average correlations are calculated by averaging exposure across subjects for each day.
21      Daily-average correlations then describe the relationship between the daily average exposure and
22      daily ambient PM concentration.  Cross-sectional is used to refer to both pooled and daily
23      average correlations,  so the meaning of this term must be determined from context.
24           Pooled correlations have been simulated from longitudinal data by using a random-
25      sampling procedure to select a random day from each subject's measurements for use in the
26      correlation.  This procedure was repeated many times, and statistics (such as the mean and
27      standard deviation of the pooled correlation coefficient) were reported (Janssen et al., 1997,
28      1998a, 1999c).
29           The type of correlation analysis can have a substantial effect on the resulting correlation
30      coefficient. Mage et al. (1999) mathematically  demonstrated that very low correlations between
31      personal exposure and ambient concentrations could be obtained when people with very different

        June 2003                                 5-44        DRAFT-DO NOT QUOTE OR CITE

-------
 1      nonambient exposures are pooled, even though their individual longitudinal correlations are
 2      high. The longitudinal studies conducted by Tamura et al. (1996a) and Janssen et al. (1997,
 3      1998a, 1999c) determined that the longitudinal correlations between personal exposure and
 4      ambient PM concentrations were higher than the correlations obtained from a pooled data set.
 5      Wallace (2000a) reviewed a number of longitudinal studies and found that the median
 6      longitudinal correlation coefficient was higher than the pooled correlation coefficient for the
 7      same data (see Tables 1 and 2, Wallace, 2000a).
 8           Mage et al. (1999) examined three longitudinal exposure data sets where several subjects
 9      were measured each day. They showed that by averaging daily  exposures across subjects, daily-
10      average correlations could be obtained. These were all higher than the median longitudinal
11      correlations. Williams et al. (2000a,b) and Evans et al. (2000) have also reported higher
12      correlation coefficients for daily-average correlations compared to longitudinal correlations.  The
13      higher correlations found between daily-average personal exposures and ambient PM
14      concentrations, as  opposed to lower correlations found between individual exposures and
15      ambient PM levels, have been attributed to the statistical  process of averaging (Ott et al., 2000).
16      Personal exposures include contributions from nonambient as well as ambient PM
17      concentrations. When several subjects are measured on the same day, the mean variability due
18      to variations in nonambient exposures is reduced due to averaging. Therefore, the correlation
19      between personal exposure and ambient concentrations increases as the number of subjects
20      measured daily increases.  Ott et al. (2000), using the theory on which their Random Component
21      Superposition (RCS) model  is based, predict expected correlations above 0.9 for the PTEAM
22      study and above 0.70 for the New Jersey study (Lioy et al., 1990) if 25 subjects had been
23      measured daily in  each study.
24
25      5.3.3.1.2  Correlation Data from Personal Exposure Studies
26           Measurement data and correlation coefficients for the personal exposure studies described
27      in Section 5.4.2.1 are summarized in Table 5-7.  All data are based on mass measurements.  The
28      studies are grouped by the type of study design, longitudinal or pooled. For each study in
29      Table 5-7, summary statistics for the total personal PM exposure measurements are presented
30      as well as statistics for residential indoor, residential outdoor, and ambient PM concentrations
31      when available.  The correlation coefficients (r) between  total personal PM exposures and

        June 2003                                 5-45        DRAFT-DO NOT QUOTE OR CITE

-------
•— ^
3
to
o
o
OJ











Y1
ON



o

J>

i
O
0
n
s.**
H
0
o
H
W
O
^
O
H
W




TABLE 5-7. PERSONAL I
MONITORING STUDIES FOR PARTICULATE MATTER: MEASURED CONCENTRATIONS
AND CORRELATION COEFFICIENTS
Personal- Ambient 2 Other Correlation
Measured Concentration Levels (ug/m3)
Size Avg. Sample
Fraction Time Statistic Size1
Longitudinal Studies
Ebelt et al. (2000) - Vancouver, BC
PM25 24 h x±SD 106
Range
Evans et al. (2000) -Fresno, CA
PM25 24 h x 24
Range
PM25 24 h x 12
Range
Jans sen et al. (1997) —Netherlands
PM10 24 h x±SD 301
Range








Jans sen et al. (1998a) —Netherlands
PM10 24 h x±SD 262
Range




Jans sen et al. (1999c) —Netherlands
PM25 24 h x±SD 77
Range


PM25 24 h x±SD 55
Range
PM25 24 h x±SD 22
Range
Residential Residential
Personal Indoor Outdoor Ambient


18.2 ±14.6 11.4±4.1
2-91 4-29

13.3 9.7 20.5 21.7
1-24 4-17 4-52 6-37
11.1 8.0 10.1 8.6
7-16 4-12 5-20 4-16

105.2 ±28.7 38.5 ±5.6
57-195 25-56









61.7 ±18.3 35.0 ±9.4 41. 5 ±4.3
38-113 19-65 32-50





28.3 ±11.3 17.1 ±2.8
19-60 14-22


24.4 ±4.9 17.1 ±2.6
19-33 15-22
37.0 ±17.4 17.1 ±3.7
21-60 14-21

Type3


Median L
P

P

P


Median L
Median L
Median L
MeanP
MeanP
MeanP





Median L
Median P
Median P




Median L
Median P


Median L
Median P


Correlation Coefficients (r) Coefficients (r)

Value (Range)


0.48 (-0.68-0.83)
0.15

0.414

0.844


0.63(0.1-0.9)
0.63
0.59
0.28 (0.12)5
0.45 (0.16)5
0.20 (0.19)5





0.50 (-0.41-0.92)
0.50(0.07-0.83)5
0.34(-0.09-0.67)5




0.86 (-0.1 1-0.99)
0.41 (-0.28-0.93)5


0.92
0.825


Value
Notes Type3 (Range)


n = 16 COPD
subjects

Fresno-1 study Pp.; 0.814
PP-O 0.804
Fresno-2 study Pp.; 0.954
PP-O 0.804

n = 45 school
children
With nonsmoking
parents
With smoking parents
All
With nonsmoking
parents
With smoking parents


n = 37 adults Med. Lp.; 0.72
No ETS exposure Med. L;., (-0.10-
All 0.98)
0.73
(-0.88-
0.95)


n = 13 school
children


With nonsmoking
parents
With smoking parents


-------
•— ^
3

to
O
o
OJ












^
-k


O
!>
'•rj
H
6
o
2;
-^H
0
H


o
H
W
O
O
H
W
TABLE 5-7 (cont.).





Size Avg. Sample
Fraction Time Statistic Size1
Janssen et al. (2000) — Netherlands
PM25 24 h x±SD 338
Range



Janssen et al. (2000) — Finland
PM25 24 h x±SD 336
Range



Linn et al. (1999) -Los Angeles
PM25 24 h x±SD 60
Range
PM10 24 h x±SD 59
Range

Rojas-Bracho et al. (2000) -Boston

PM25 12 h x±SD 224
Range

PM10 12 h x±SD 225
Range

PM10.25 12 h x±SD 222
Range




PERSONAL MONITORING STUDIES FOR PM: MEASURED CONCENTRATIC
CORRELATION COEFFICIENTS


Personal- Ambient 2
Measured Concentration Levels (ug/m3) Correlation Coefficients (r)
Residential Residential
Personal Indoor Outdoor Ambient Type3 Value (Range) Notes

24.3 ±25.7 28.6 ±41. 8 20.6 ± 4.0 Median L 0.79 (-0.41-0.98) n = 36 elderly w/CV
9-134 9-239 13-31 Median L 0.85 disease
No ETS exposures



10.8 ±4.4 11.0 ±4.0 12.6 ±2.0 Median L 0.76 (-0.12-0.97) n = 46 elderly w/CV
4-33 3-27 10-18 disease




23. 8 ±15.1 23. 5 ±15.3 24.8 ±14.5 P 0.26'
4-65 4-92 4-63
34.8 ±14.8 32.6 ±15.6 39.8 ±18.3 33 ±15 P 0.22'
5-85 9-105 7-97 9-??



21.6 ±13.6 17.5±14.1 14.2 ±11.2 Median L 0.61(0.10-0.93)' n = 17 adults
1-128 2-73 1-57

37.2 ±22.8 31.9 ±25.2 22.2 ±18.7 Median L 0.35(0.0-0.72)'
9-211 2-329 3-76

15.6 ±14.6 14.5 ±9.2 8.1 ±6.8 Median L 0.30(0.0-0.97)'
-11-103 -3-255 -2-64




>NS AND



Other Correlation
Coefficients (r)
Value
Type3 (Range)

Med. Lp.; 0.91
Med. L;.,, (-0.28-1.0)
0.84
(-0.00-
0.98)

Med. Lp.; 0.89
Med. L;.a (0.14-1.0)
0.70
(-0.15-
0.94)

P;-, 0.26'
PO-, 0.47'
PiHl 0.32'
po-. O-66'



Med. Lp.; 0.87'
Med. L;.0 0.74'

Med. Lp.; 0.71'
Med. L;.0 0.50'

Med. Lp.; 0.42'
Med. L;.0 0.20'





-------
|r TABLE 5-7 (cont.). P

KJ
o
OJ
Size Avg.
Fraction Time Statistic
Sarnat et al. (2000) -Baltimore
PM2.5 24 h x ± SD
x±SD








PM10 24 h x ± SD
i x±SD
OO
PM10.25 24 h x±SD
x±SD
ERSONAL MONITORING STUDIES FOR PM: MEASURED CONCENTRATIO
CORRELATION COEFFICIENTS
Personal- Ambient 2
Measured Concentration Levels (ug/m3)
Sample
Size1

37
36








37
36

37
36
Residential Residential
Personal Indoor Outdoor Ambient Type3

26.7 ±13.7 25.2 ±11. 5 Median L
18.5 ±11.2 5.6 ±49.0 Median L
P
P
P
P




33.9 ±11.7 34.0 ±12.8 Median L
28.0 ±16.5 7.5 ±73.2 Median L

7.2 ±4.0 8.4 ±2.3 Median L
9.6 ±7.9 -1.3 ±24.2 Median L
Correlation Coefficients (r)

Value (Range)

0.76(-0.21-0.95)7
0.25(-0.38-0.81)7
0.898
0.758
0.508
0.448




0.64(0.08-0.86)7
0.53(-0.79-0.89)7

0.11 (-0.60-0.64)7
0.32(-0.48-0.68)7

Notes

n = 15 adults;
summer
n = 15 adults; winter
High ventilation;
summer
Med. ventilation;
summer
Low ventilation;
summer
WINTER
SUMMER
WINTER

SUMMER
WINTER
NSAND
Other Correlation
Coefficients (r)
Value
Type3 (Range)
















Tamura et al. (1996a) — Tokyo




  PM10      48 h
0.83
              n = 7 elderly adults
H
1
O
0
y^.
0
H
0
crj
o
w
o
^^
-.
HH

W
Williams et al. (2000a,b) -Baltimore

PM25 24 h x 23
Range




PM10 24 h x 28
Range


PM10.25 24 h x 26
Range





13.0 9.4 22.0 22.0 Median L 0.80 (0.38-0.98)6 n = 21 elderly adults
7-25 4-19 7-52 8-59 P 0.894




11.0 30.0 29.9
4-23 13-66 13-74


1.0 8.0 8.0
-3-5 -2-16 1-15





PP.
PP-.
Pi-0

pi!

P;-o

Po-a

P;-o
Pi-a
P




0.90"
0.954
0.944
0.874
0.964

0.824
0.81"
0.944

0.18"
0.084
0.454



-------
•— ^
3

to
O
o
OJ














Y1
VO


O
!>
^
H
6
o
2;
-^H
0
H
O

0
H
W
O
r*"
O
HH
H
W

TABLE 5-7 (cont). PERSONAL MONITORING STUDIES FOR PM: MEASURED CONCENTRATIO
CORRELATION COEFFICIENTS

Personal- Ambient 2
Measured Concentration Levels (ug/m3) Correlation Coefficients (r)
Size Avg. Sample Residential Residential
Fraction Time Statistic Size1 Personal Indoor Outdoor Ambient Type3 Value (Range) Notes
Williams et al. (2000a,b) -Baltimore (cont'd)
Keeleretal. Mean 20 asthmatic 68.4(39.2) 52.2 25.8
(2002) (Std) children (30.6) (11.8)
Detroit, MI
PM1024-h 34.4 15.6
PM2.524-h (21.7) (8.2)
Landisetal. Mean 10 elderly 12.8 10.2 21.0 Pearson r = 0.82
(2001) retirees (0.51-0.92)
Baltimore, MD 4.5 4.0 10.2
PM2524-h r = 0.95
Sulfate24-h (0.74-0.97)
Pooled Studies
Bahadori (1998) -Nashville
PM25 12 h x±SD 30 21.7 ±10.5 15.5 ± 6.6 23.4 ±6.8 P 0.09 n= 10 COPD
Range 10-67 5-40 3-61 subjects; daytime
PM10 12 h x±SD 30 33.0 ±16.9 21.6 ±10.7 32.5 ±8.1 P -0.08 n = 10 COPD
Range 5-88 9-77 7-76 subjects; daytime
Pellizzari et al. (1999) - Toronto
PM25 3d x 922 28.4 21.1 15.1 P 0.23 n = 178; n for indoor,
outdoor lower than
personal

PM10 3d x 141 67.9 29.8 24.3 No correlations
reported
Oglesby et al. (2000) - EXPOLIS Basel

PM25 48 h x±SD 44 23.7 ±17.1 19.0 ±11.7 P 0.07 All
20 17.5 ±13.0 17.7 ±7.1 P 0.21 No ETS exposure

Santos-Burgoa et al. (1998) -Mexico City
PM10 24 h x±SD 66 97 ±44 99 ± 50 P 0.26


Tamura et al. (1996b) - Osaka
PM2 48 h P 0.74
PM10 48 h P 0.67
NSAND


Other Correlation
Coefficients (r)
Value
Type3 (Range)






P vs I r = 0.60
(.41-.85)

r = 0.95
(.73-.97)


PP-, 0.72
Pi-o 0.31
PP-, 0.43
P;-o 0.06

PP-, 0.79
P;-o 0.33










PP-, 0.47
P;-, 0.23





-------
•— ^
3
rt>
to
o
o
OJ












u\
i
o
TABLE 5-7 (cont). I





Size Avg. Sample
Fraction Time Statistic Size1
Pellizzari et al. (2001) — Indianapolis
PM25 72-h Median 250



Brauer et al. (2000) Banska Bystrica
PM25 24-h Mean PM10
PM10 Summer
PM10 Winter
PM25
Summer
PM2 5 Winter
SO4 Winter
'ERSONAL MONITORING STUDIES FOR PM: MEASURED CONCENTRAT
CORRELATION COEFFICIENTS


Personal- Ambient 2
Measured Concentration Levels (ug/m3) Correlation Coefficients (r)
Residential Residential
Personal Indoor Outdoor Ambient Type3 Value (Range) Notes

23 18 18 18 P 0.102 Betweenthe
Logarithms of
concentrations


122 79 35 P PM10 r2<0.17 Multivarwith
120 66 45 nicotine
88 55 22
69 53 32
6.5 4.6 5.7


IONS AND







Other Correlation
Coefficients (r)

Type3

Pvs
outdoor

P vs Indoor

P indoor
PM25
P
Personal
SO4vs
Amb. SO4

Value
(Range)

0.138
0.923



r2 = 0.15
r2 = 0.23





Kousa et al. (2002)
DRAFT-
O
0
2|
0
H
0
r~|
0
H
W


^
O
1 — I
H
W

Helsinki
+ 3 cities
PM2548-h

Abbreviati
Avg. =
Cone. =
CV
d
ETS =

Notes:
1.
2.
3.
4.
5.
6.
7.

ons used:
Averaging (time)
Concentration
Cardiovascular
Day
Environmental tobacco smoke



h
i-a
i-o
L
Med.
o-a



= Hour
= Indoor-ambient correlation
= Indoor-outdoor correlation
= Longitudinal correlation
= Median
= Outdoor-ambient correlation


Pearson r2 = 0.69 P vs I r = 0.60
non-work (0.41-0.85)
pooled
Helsinki r = 0.95
(0.73-0.97)

P = Pooled correlation
p-i = Personal-indoor correlation
p-o = Personal-outdoor correlation
SD = Standard deviation
Stat. = Statistic
x = Mean


Sample size is for personal concentrations; indoor, outdoor and ambient sample sizes may differ.
Correlation coefficient is for personal-residential outdoor if no ambient concentration data reported.
See text for description of types of correlations.



Daily-averaged correlation (values for individual subjects averaged for each day).
Pooled correlations estimated using a Monte Carlo sampling proc
Obtained from a regression equation; r =^/(R2) .
Spearman rank correlations.


edure, n = 1000. If mean P is shown,


then SD given; if median P is shown, then range is given.


8. Calculated, r =V(R ) , from R2 from a mixed model regression.

-------
 1      ambient PM concentrations also are presented and classified as longitudinal or pooled
 2      correlations. When reported, ^-values for the correlation coefficients are included.  Correlation
 3      coefficients between personal, indoor, outdoor, and ambient also are reported when available.
 4
 5      5.3.3.1.3  Correlations Between Personal Exposures, Indoor, Outdoor, and Ambient
 6               Measurements
 1           Longitudinal and pooled correlations between personal exposure and ambient or outdoor
 8      PM concentrations varied considerably between study and study subjects. Most studies report
 9      longitudinal correlation coefficients that range from < 0 to ~ 1, indicating that an individual's
10      activities and residence type may have a significant effect on total personal exposure to PM.
11      General population studies tend to show lower correlations because of the higher variation in the
12      levels of PM generating activities.  In contrast, the absence of indoor sources for the populations
13      in several of the longitudinal studies resulted in high correlations between personal exposure and
14      ambient PM within subjects over time for these populations. But even for these studies,
15      correlations varied by individual depending on their activities and the microenvironments that
16      they occupied.
17
18      Probability Studies
19           In the Toronto study (Pellizzari et al., 1999), pooled correlations were derived for personal,
20      indoor, outdoor, and fixed-site ambient measurements. This study was conducted in Toronto on
21      a probability sample of 732 participants who represented the general population of people
22      16 years and older.  The study included between 185 and 203 monitoring periods with usable
23      PM data for personal, residential indoor, and outdoor measurements.  For PM10 measurements,
24      the mean concentrations were 67.9 |ig/m3 for personal, 29.8 |ig/m3 for indoor air, and 24.3  |ig/m3
25      for outdoor air samples. For PM2 5, the mean concentrations were 28.4 |ig/m3 for personal,
26      21.1 |ig/m3 for indoor air, and  15.1 |ig/m3 for outdoor air samples.  A low but significant
27      correlation (r = 0.23, p < 0.01) was reported between personal exposure and ambient
28      measurements.  The correlations between indoor concentrations and the various outdoor
29      measurements of PM25 ranged from 0.21 to 0.33. The highest correlations were for outdoor
30      measurements at the residences with the ambient measurements made at the roof site (0.88) and
31      the other fixed site (0.82). Pellizzari et al.  (1999) state that much of the difference among the
32      data for personal/indoor/outdoor PM:

        June 2003                                 5-51        DRAFT-DO NOT QUOTE OR CITE

-------
 1            ... can be attributed to tobacco smoking, since all variables reflecting smoking . . . were
 2            found to be highly correlated with the personal (and indoor) particulate matter levels, relative
 3            to other variables that were measured . . . none of the outdoor concentration data types
 4            (residential or otherwise) can adequately predict personal exposures to particulate matter.
 5            (p. 729)
 6
 7           Using a Random Component Superposition (RCS) statistical model, Ott et al. (2000)
 8      calculated an attenuation factor of 0.61 for personal exposure for PM10 for the Toronto study.
 9      The mean nonambient exposure component for PM10 was estimated as 52.62 |ig/m3 with a
10      standard deviation of 84.82 |ig/m3. Although the data were available for PM2 5, similar
11      calculations were not made.
12           PM10 data from the PTEAM study were analyzed using the same approach (Ott et al.,
13      2000). For PTEAM, an attenuation factor of 0.55 was calculated for personal exposure.
14      Infiltration factors were calculated for each residence with an average of 0.56 and a standard
15      deviation of 0.15.  Values  ranged from a minimum of 0.19 to a maximum of 0.87 showing the
16      substantial variability that can be seen between homes depending upon the housing
17      characteristics and operation of the HVAC system.  The mean nonambient exposure component
18      for PM10 was estimated as 59 |ig/m3 with a standard deviation of 46 |ig/m3.
19           Santos-Burgoa et al.  (1998) describe a 1992 study of personal exposures and indoor
20      concentrations to a randomly sampled population near Mexico City. The sample of 66
21      monitored subjects included children, students, office and industrial workers, and housewives.
22      None of the people monitored were more than 65 years old. The mean 24-h personal exposure
23      and indoor concentrations were 97 ± 44 and 99 ± 50 |ig/m3, respectively, with an rPersonal/Ambient
24      = 0.26 (p = 0.099).  Other  correlations of interest were rPersonal/Indoor = 0.47 (p = 0.002) and
25      findoor/Ambient= 0-23 (p = 0.158). A strong statistical association was found between personal
26      exposure and socioeconomic class (p = 0.047) and a composite index of indoor sources at the
27      home (p = 0.039).
28           Correlation analysis  for personal exposure has not yet been reported for EXPOLIS.  Some
29      preliminary results (Jantunen et al., 2000) show that in Basel and Helsinki a single ambient
30      monitoring station was sufficient to characterize the ambient PM2 5 concentration in each city.
31      Using microenvironmental concentration data collected while the subjects were at home, at
32      work, and outdoors,  they calculated the sum of the time-weighted-averages of these data and

        June 2003                                 5-52        DRAFT-DO NOT QUOTE OR CITE

-------
 1      found the results closely match the personal PM2 5 exposure data collected by the monitors
 2      carried by most of the subjects although a few subjects (mostly smokers) were noticeable
 3      exceptions.
 4
 5      Longitudinal Studies
 6           A number of longitudinal studies using a purposeful sampling design have been conducted
 7      and reported in the literature since 1996. A number of these studies (Janssen et al., 1998a,
 8      1999b, 2000; Williams et al., 2000b; Evans et al., 2000) support the previous work by Janssen
 9      et al. (1995) and Tamura et al. (1996a) and demonstrate that for individuals with little exposure
10      to nonambient sources of PM, correlations between total PM exposure and ambient PM
11      measurements are high.  Other studies (Ebelt et al., 2000; Sarnat et al., 2000) show strong
12      correlations for the SO4"2 component of PM25 but poorer correlations for PM25 mass.  Still other
13      studies show only weak correlations (Rojas-Bracho et al., 2000; Linn et al., 1999; Bahadori
14      et al., 2001). Even when strong longitudinal correlations are demonstrated for individuals in a
15      study, the variety of living conditions may lead to variations in attenuating factors or the fraction
16      of ambient PM contributing to personal exposure.  Groups with similar living conditions,
17      especially if measurements are conducted during one season, may have similar a and, therefore,
18      very  high correlations between personal exposure and ambient concentrations, even for pooled
19      correlations. However, when studies contain subjects with homes of very different ventilation
20      characteristics or cover more than one season, variations in a can be high across subjects, thus,
21      showing poor pooled correlations even in the absence of indoor sources.
22
23           Elderly Subjects. Janssen et al. (2000) continued their longitudinal studies with
24      measurements of personal, indoor, and outdoor concentrations of PM25 for elderly subjects with
25      doctor-diagnosed angina pectoris or coronary heart disease. Studies were conducted in
26      Amsterdam, Holland, and Helsinki, Finland, in the winter and spring of 1998 and  1999.  In the
27      Amsterdam study with 338 to 417 observations, mean PM25 concentrations were 24.3, 28.6, and
28      20.6  |ig/m3 for personal, indoor, and outdoor samples, respectively. If the measurements with
29      ETS  in the home were excluded, the mean indoor concentration dropped to 16 |ig/m3, which was
30      lower than outdoor concentrations. In the Helsinki study, the mean PM2 5 concentrations were
31      10.8  |ig/m3 for personal, 11.0 |ig/m3 for indoor air, and 12.6 |ig/m3 outdoor air samples.  The

        June  2003                                5-53        DRAFT-DO NOT QUOTE OR CITE

-------
 1      authors note that for this group of subjects, personal exposure, indoor concentrations, and
 2      ambient concentrations of PM25 were highly correlated within subjects over time. Median
 3      Pearson's correlation coefficients between personal exposure and outdoor concentrations were
 4      0.79 in Amsterdam and 0.76 in Helsinki.  The median Pearson's r for the indoor/outdoor
 5      relationship was 0.85 for the Amsterdam study when homes with ETS were excluded. The
 6      correlation for indoors versus outdoors was  0.70 for all homes.
 7           Results from the correlation analysis can be used to estimate infiltration factors and
 8      penetration factors for these two groups of subjects. In Amsterdam, the attenuation factor was
 9      0.43 and the infiltration factor was 0.47.  Very similar results were seen in Helsinki for the
10      attenuation factor (0.45) and the infiltration  factor (0.51).
11           A series of PM personal monitoring studies involving elderly subjects was conducted in
12      Baltimore County, MD, and Fresno, CA.  The first study was a 17-day pilot (January-February
13      1997) to investigate daily personal and indoor PMj 5 concentrations, and outdoor PM2 5 and
14      PM2 5_10 concentrations experienced by nonsmoking elderly residents of a retirement community
15      located near Baltimore (Liao et al., 1999;  Williams et al., 2000c).  The 26 residents were aged
16      65 to 89 (mean = 81), and 69% of them reported a medical condition such as hypertension or
17      coronary heart disease. In addition, they were quite sedentary: on  average, less than 5 h day"1
18      were spent on ambulatory activities. Because most of the residents ate meals in a communal
19      dining area, the average daily cooking time  in the individual apartments was only 0.5 h
20      (range = 0 to 4.5 h). About 96% of the residents' time was spent indoors (Williams et al.,
21      2000c). Personal monitoring, conducted for five subjects, yielded longitudinal  correlation
22      coefficients between ambient concentrations and personal exposure ranging from 0.00 to 0.90.
23           The Baltimore main study and the Fresno study were conducted using similar monitoring
24      techniques and study design.  Concentrations measured in these studies are summarized in
25      Table 5-8.  For PM2 5, personal exposure and indoor air concentrations are similar for all three
26      studies even though outdoor air concentrations for Fresno in the spring are only half of those
27      measured for Fresno in the winter and for Baltimore. This result is presumably due to high
28      penetration efficiencies in the spring in Fresno when the weather was warm and participants kept
29      the windows and doors of their homes open. These data also show that even when correlations
30      are high, the use of an ambient monitor as a surrogate for exposure in epidemiologic studies can
31      bias the strength of the health effect found due to differing exposure levels.

        June 2003                                  5-54         DRAFT-DO NOT QUOTE OR CITE

-------
           TABLE 5-8.  MEAN CONCENTRATION FOR PM MASS REPORTED FOR THE
                    BALTIMORE (Williams et al., 2000a,b,c) AND THE FRESNO
          	(Evans et al., 2000) STUDIES	
        Study
                            PM2 s Concentration (jig/m3)
                                                         PM10 Concentration (ug/m3)
                 Personal
Indoors
Outdoors
Personal
Indoors    Outdoors
        Baltimore        13.0 ±4.2    10.5 ±4.9   22.0 ±12.0
        Fresno-Winter    13.3 ± 5.9    9.7 ± 5.0    20.5 ± 13.4
        Fresno-Spring    11.1 ±2.8    8.0 ±1.8     10.1 ±3.2
                                                        37.3
                                    13.5 ±6.3   30.0 ±13.7
                                    15.1 ±4.1   28.2±15.9
                                    16.7 ±3.1   28.7 ±6.6
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
     Calculated correlation coefficients are summarized in Tables 5-9 and 5-10. In Table 5-9,
results for Baltimore show excellent daily average correlations for both PM2 5 and PM10. These
results primarily represent the behavior of fine-particle regional sulfate for a group of
participants who have few indoor or personal sources. However, even for this group, there was a
wide range of individual correlation coefficients. The Fresno data, on the other hand, shows
much poorer daily average correlations. Of special note are the poorer correlations for the
ambient to outdoor residential monitors. This could be due to the higher concentrations of nitrate
in the samples. In addition, the residential site may have been influenced by highway traffic.
        TABLE 5-9. DAILY-AVERAGE CORRELATION COEFFICIENTS REPORTED FOR
                  THE BALTIMORE (Williams et al., 2000a,b,c) AND THE FRESNO
                                     (Evans et al., 2000) STUDIES
Study
Baltimore
Fresno-Winter
Fresno-Spring

Ambient/Outdoor
0.92
0.48
0.53
PM2S r2
Personal/Ambient
0.80
(0.14-0.80)a
—
0.70

Personal/Indoors
0.98
(0.20-0.99)a
—
0.77
PM10 r2
Ambient/Outdoor
0.89
0.48
0.61
        aRange for individual participants.
       June 2003
                                         5-55
                        DRAFT-DO NOT QUOTE OR CITE

-------
           TABLE 5-10. REGRESSION ANALYSIS REPORTED FOR INDOOR/OUTDOOR
            RELATIONSHIPS FOR PM25 IN THE BALTIMORE (Williams et al., 2000a,b,c)
                             AND FRESNO (Evans et al., 2000) STUDIES
Daily Average
Study
Baltimore
Fresno-Winter
Fresno-Spring
r2
0.92
0.86
0.56
slope
0.39
nr
nr
Intercept
ftig/m3)
1.5
nr
nr
r2
0.73 ±0.16
0.55 ±0.25
0.39 ±0.21
Individual
slope
0.43 ±0.15
0.25 ±0.17
0.49 ±0.38

Intercept
Oig/m3)
0.9 ±2.6
4.4 ±3.2
3.0 ±3.7
 1          The correlation analysis in Table 5-10 shows correlation coefficients as well as the slope
 2     (infiltration factor) and the intercept (indoor concentration due to nonambient sources) for the
 3     Baltimore and Fresno studies.  These data show strongest correlations for Baltimore where there
 4     are very low indoor concentrations from nonambient sources.  Correlations are not as strong for
 5     Fresno, where there are higher concentrations from nonambient sources. The infiltration factors
 6     for Baltimore and Fresno-Spring are very similar at approximately 0.5.  The infiltration factors
 7     for Fresno-Winter are considerably lower.
 9          Subjects with COPT). Linn et al. (1999) describe a 4-day longitudinal assessment of
10     personal PM25 and PM10 exposures (on alternate days) in 30 COPD subjects aged 56 to 83.
11     Concurrent indoor and outdoor monitoring was conducted at their residences.  This study
12     occurred in the summer and autumn of 1996 in the Los Angeles area. PM10 data from the nearest
13     fixed-site monitoring station to each residence also was obtained. Pooled correlations for
14     personal exposure to outdoor measurements gave R2 values of 0.26 and 0.22 for PM2 5 and PM10,
15     respectively.  Correlations of day-to-day changes in PM25 and PM10 measured outside the homes
16     and correlated with concurrent PM10 measurements at the nearest ambient monitoring location
17     gave R2 values of 0.22 and 0.44,  respectively.  Correlations of day to day changes in PM mass
18     measured indoors correlated with outdoor measurements at the homes gave R2 values of 0.27 and
19     0.19 for PM10 and PM25.
20          Personal, indoor, and outdoor PM2 5, PM10, and PM2 5_10 correlations were reported by
21     Rojas-Bracho et al. (2000) for a study conducted in Boston, MA, on 18 individuals with COPD.
22     Both the mean and median personal exposure concentrations were higher than the indoor
       June 2003
5-56
DRAFT-DO NOT QUOTE OR CITE

-------
 1      concentrations, which were higher than outdoor concentrations for all three PM measurement
 2      parameters. Geometric mean indoor/outdoor ratios were 1.4 ± 1.9 for PM10, 1.3 ± 1.8 for PM2 5,
 3      and 1.5 ± 2.7 for PM25_10. Median longitudinal R2s between personal exposure and ambient PM
 4      measurements were 0.12 for PM10, 0.37 for PM2 5 and 0.07 for PM2 5_10. The relationship between
 5      the indoor and outdoor concentrations was strongest for PM2 5, with a median R2 of 0.55 and
 6      11 homes having significant R2 values. For PM10, the median R2 value was 0.25, with significant
 7      values for eight homes. Only five homes had significant indoor/outdoor associations for PM2 5_10,
 8      with an insignificant median R2 value of 0.04.  The poor correlations for PM10_2 5 are a result of
 9      poorer penetration efficiencies, higher decay rates, and spatial inhomogeneities.
10          Bahadori et al. (2001) report a pilot study of the PM exposure of 10 nonrandomly chosen
11      COPD patients in Nashville, TN, during the summer of 1995. Each  subject alternately carried a
12      personal PM25 or PM10 monitor for a 12-h daytime period (8 am to 8 pm) for 6 consecutive days.
13      These same pollutants were monitored simultaneously indoors and outdoors at their homes.
14      All of the homes were air-conditioned and had low air exchange rates (mean = 0.57 h"1), which
15      may have contributed to the finding that mean indoor PM2 5 was 66% of the mean ambient PM2 5.
16      This can be contrasted with the PTEAM study in Riverside, CA, where no air conditioners were
17      in use and the mean indoor PM2 5 was 98% of the mean ambient PM25 (Clayton et al., 1993).
18      Data sets were pooled for correlation analysis.  Resulting pooled correlations between personal
19      and outdoor concentrations were r = 0.09 for PM25 and r = -0.08 for PM10.
20
21      5.3.3.1.4 Personal Exposure to Sulfate Compared to Personal Exposure to Ambient
22              Particulate Matter
23          A study conducted in Vancouver involving sixteen COPD patients aged 54 to 86 reported
24      low median longitudinal (r = 0.48) and pooled (r = 0.15) correlation  coefficients between
25      personal exposures and ambient concentrations of PM25 (Ebelt et al., 2000). However, the mean
26      correlation between personal exposures to sulfate and ambient concentrations of sulfate was
27      much higher (r = 0.96). Because there are typically minimal indoor  sources of sulfate, the
28      relationship between ambient concentrations and personal exposures to sulfate would not be
29      weakened by variability in an indoor-generated sulfate component as for example in the case of
30      PM2 5 for which there are many primary indoor sources as well as some secondary indoor
31      sources. Correlations of ambient concentrations versus personal exposures for PM2 5 and sulfate
32      are compared in Figure 5-3.

        June 2003                                 5-57        DRAFT-DO NOT QUOTE OR CITE

-------
1.00 -

0.75 -
tion Coefficient (r)
o o o
o to bi
o 01 o
i i i
0 -°-25 '
5 -0.50 -

-0.75 -
-i nn
Ebelt etal. ,2000
Pearson's "r"
PM S
•2.b J_


I





Sulfate






Percentile

90th Percentile
75th Percentile
Median
25th Percentile
10th Percentile





Sarnat etal., 2000
Spearman's "r"
PM25
T
'^


T
-
[

^ Summer
| | Winter

J.
f/ -L
I-
I
Sulfate





                             PM2 5   Sulfate
                 PM2 5    Sulfate
       Figure 5-3. Comparison of correlation coefficients for longitudinal analyses of personal
                   exposure versus ambient concentrations for individual subjects for PM2 5 and
                   sulfate.
 1          Another study, conducted in Baltimore, MD, involved 15 nonsmoking adult subjects
 2     (> 64 years old) who were monitored for 12 days during Summer 1998 and Winter 1999 (Sarnat
 3     et al., 2000).  All subjects (nonrandom selection) were retired, physically healthy, and lived in
 4     nonsmoking private residences. Each residence, except one, was equipped with central
 5     air-conditioning; however, not all residences used air-conditioning throughout the summer. The
 6     average age of the subjects was 75 years (± 6.8 years). Sarnat et al. (2000) reported higher
 7     longitudinal and pooled correlations for PM2 5 during summer than winter.  Similar to Ebelt et al.
 8     (2000), Sarnat et al. (2000) reported stronger associations between personal exposure to SO4"2
 9     and ambient concentrations of SO4"2 than for total personal PM2 5 exposure and ambient PM2 5
10     concentrations. The ranges of correlations are shown in Figure 5-1 along with similar data from
11     Ebelt et al. (2000).
       June 2003
5-58
DRAFT-DO NOT QUOTE OR CITE

-------
 1           The higher correlation coefficients and the narrower range of the correlation coefficient for
 2      sulfate suggest that removing indoor-generated and personal activity PM from total personal PM
 3      would result in a higher correlation with ambient concentrations. If, as discussed in 5.3.4.3.1,
 4      there are no indoor sources, a personal exposure measurement for sulfate gives the ambient
 5      exposure of sulfate; the ratio of personal sulfate to ambient sulfate gives the attenuation
 6      coefficient on an individual, daily basis; and the attenuation coefficient times the ambient PM2 5
 7      concentration gives the individual, daily values of ambient PM2 5 exposures (Wilson et al., 2000).
 8      This technique applies only to the nonvolatile components of fine PM, as measured by PM25.
 9      It requires that the sulfate concentration be large enough so that it can be measured with
10      reasonable accuracy.  It does not require that sulfate be correlated with PM2 5 or the non-sulfate
11      components of PM2 5 because the sulfate data is used to estimate the attenuation coefficient not
12      PM2 5. The technique does require that there be minimal indoor sources of sulfate, as indicated
13      by a near-zero intercept for the regression, and that the size distribution of PM2 5 and sulfate be
14      similar.
15           Sarnat et al. (2001) subsequently extended the Baltimore study to include 20 older adults,
16      21 children, and 15 individuals with COPD for a total of 56 subjects. In both studies (Sarnat
17      et al., 2000, 2001), they used their personal and ambient sulfate data to  estimate the ambient
18      PM25 exposure. They used this information in mixed-model analysis (mixed models account for
19      differences among individual subjects), but did not report  correlations between ambient PM2 5
20      exposure and ambient PM25 concentrations based on the pooled data set. However, Sarnat et al.
21      (2001) did report slopes from the mixed model analyses. The t-statistic for the slope of ambient
22      exposure versus ambient concentration as compared to total personal exposure versus ambient
23      concentration increased from 9.96 to 11.12  (total exposure vs. ambient concentration) for the
24      summer period and from 4.36 to 19.88 (ambient exposure  vs. ambient concentration) for the
25      winter period.
26           The study conducted  by Sarnat et al. (2000) also illustrates the importance of ventilation on
27      personal exposure to PM. During the summer, subjects  recorded the ventilation status of every
28      visited indoor location (e.g., windows open, air-conditioning use). As a surrogate for the air
29      exchange rate, personal exposures were classified by the fraction of time the windows were open
30      while a subject was in an indoor environment (Fv).  Sarnat et al. (2000)  report regression
31      analyses of personal exposure on ambient concentration for PM2 5 and for sulfate for each of the

        June 2003                                  5-59        DRAFT-DO NOT QUOTE OR CITE

-------
 1      three ventilation conditions (Figure 5-4). The correlation between personal exposure and
 2      ambient concentration is higher for sulfate than for PM2 5, presumably because PM2 5 has indoor
 3      sources as well as ambient sources but sulfate has only ambient sources. As expected, the
 4      improvement is better for the lower ventilation conditions because under these conditions the
 5      ambient concentration is larger. For the lowest ventilation condition, R2 improves from 0.25 to
 6      0.72.
 7
 8      5.3.3.1.5  Personal Exposure to Ambient and Nonambient Particulate Matter
 9           The utility of treating personal exposure to ambient PM, A, and personal exposure to
10      nonambient PM, N, as separate and distinct components  of total personal exposure to PM, T,
11      was pointed out by Wilson and Suh (1997). The PTEAM study measured, in addition to indoor,
12      outdoor, and personal PM10, the air exchange rate for each home and collected information on
13      the time spent in various indoor and outdoor microenvironments.  This information is available
14      for 147 12-h daytime periods. With this information and statistically estimated values of P and
15      k, it is possible to estimate the daytime A and N as described in Section 5.2.4.3.  Various
16      examples of this information have been reported (Mage et al., 1999; Wilson et al., 2000).
17      Graphs showing the relationships between ambient concentration and the various components of
18      personal exposure (T, A, and N) are shown in Figure 5-5. The correlation coefficient for the
19      pooled data set improves from r = 0.377 for T versus C (Figure 5-5a) to r = 0.856 for A versus C
20      (Figure 5-5b) because of the removal of the N, which, as shown in Figure 5-5c, is highly variable
21      and independent of C. The correlation between A and C is less than 1.0 because of the
22      day-to-day variation in the a of each individual. The regression of T on C gives O~ = 0.711
23      and N = 81.6  |ig/m3. The regression of A on C gives (X  = 0.625.  The regression of N on C
24      gives N = 79.2 |ig/m3.  The finite intercept in the regression with A must be attributed to bias or
25      error in some of the measurements. No reported studies, other than PTEAM, have provided the
26      quantity of data on individual, daily values of T, C, C;, and a that are required to conduct an
27      analysis comparable to that shown in Figure 5-5. It should be noted that the PTEAM study was
28      conducted in southern California in the fall, when houses were open and air exchange rates were
29      high and relatively uniform. These are best case conditions for showing high correlations
30      between ambient site measurements and personal correlations. Such high correlations are not
31      usually found  and would not be expected with lower and more variable air exchange rates.
32
        June 2003                                5-60       DRAFT-DO NOT QUOTE OR CITE

-------
  o
  CL

  .2
  ns
  c
  o
  to

  0)
  CL
  O
  CL
  X
  LU
  O
  Q.

  .2

  (D
  C
  0)
  CL
60'



50'



40'



30'



20-



10'



 0'





60'



50'



40'



30'



20'



10'



 0'





60'



50



40



30



20



10



 0
           = 0.80
                                Well Ventilated Indoor Environment

                                               35
30-



25-


20-


15-


10-


 5-


 0
    R2 = 0.88
         R2 = 0.57
                              Moderately Ventilated Indoor Environment

                                               35
30-



25-


20-


15-


10-


 5-


 0
    R2 = 0.73
         R2 = 0.25
                                       :1 Line
                               Poorly Ventilated Indoor Environment

                                               35
30-



25-



20-



15-



10-



 5-
                                               0
    R2 = 0.72
             10
                   20
                         30


                       PM2.5
                               40
                                     50
                                           60
                                                 0
                                                       10
                                                                15
                       20   25

                      SO4
                                                                                30
                                                                                 40
                           Ambient Concentration (|jg/nn3)
Figure 5-4.  Personal exposure versus ambient concentrations for PM2 5 and sulfate.

            (Slope estimated from mixed models).



Source: Sarnat et al. (2000).
June 2003
                                      5-61
        DRAFT-DO NOT QUOTE OR CITE

-------
                    250
200-
                 0}
                 5  150
                 CO
                 o
                 Q.
                 X
                 LU
                 ra  100-
                 0
                 CL
                 m   50'
                         8 = 40.5+0.7110
                         r= 0.373    *
                         R2 = 0.142
                         #= 147
                    100
                 O •->• N
ft o on c
D O O C
1 1
£§
Z n
o -

N = 39.6+0.086C » (c)
r= 0.051 »
R2 = 0.0026
#=147 . *
* » *
* * *\ * .* *
** * ^ *
* » *^* ** * *
•iv^jj^V ;-:./. • • .
'* ^ "
i i i i
                                50       100       150      200
                                  Ambient Concentration (|jg/m3)
                                                                     250
Figure 5-5.  Regression analyses of aspects of daytime personal exposure to PM10 estimated
            using data from the PTEAM study, (a) Total personal exposure to PM, T,
            regressed on ambient concentration, C. (b) Personal exposure to ambient PM,
            A regressed on C. (c) Personal exposure to nonambient PM, N regressed on C.

Source: Data taken from Clayton et al. (1993). Adapted from Mage et al. (1999) and Wilson et al. (2000).
June 2003
                      5-62
DRAFT-DO NOT QUOTE OR CITE

-------
 1           The RCS model introduced by Ott et al. (2000) presents a modeling framework to
 2      determine the average contribution of ambient PM10 and indoor-generated PM10 to average
 3      personal exposures in large urban metropolitan areas. The model has been tested using personal,
 4      indoor, and outdoor PM10 data from three urban areas (Riverside, CA; Toronto; and Phillipsburg,
 5      NJ).  Results suggest that it is possible to  separate the average ambient and nonambient PM
 6      contributions to personal exposures on a community-wide basis. However, as discussed in the
 7      paper, the authors make some assumptions that require individual consideration in each-city
 8      specific application of the model for exposure or health effects investigations. Primarily,
 9      housing factors, air-conditioning, seasonal differences, and complexities in time-activity profiles
10      specific to the cohort being studied have to be taken into account prior to adopting the model to a
11      given situation.  Finally, this and other available exposure-based analyses presented here do not
12      yet predict the relative contribution of indoor and outdoor PM to particle mass burden to the lung
13      as a function of human activities and different microenvironmental sources and concentrations of
14      PM and its co-pollutants.
15
16      5.3.3.2  Factors That Affect Relationships Between Personal Exposure and Ambient PM
17           A number of factors will affect the relationship between personal exposure and PM
18      measured at ambient-site community monitors. Spatial variability in outdoor
19      microenvironmental concentrations and variations in penetration into indoor  microenvironments
20      will influence the relationship to ambient  PM. Air exchange rates and decay rates in indoor
21      microenvironments will influence the relationship to both ambient and total PM whereas
22      personal activities will influence the relationship to total PM but not ambient PM. Information
23      on these effects is presented in detail in the following sections.
24
25      5.3.3.2.1  Spatial Variability and Correlations Over Time
26           Chapter 3, Section 3.2.3, presents information on the spatial variability  of PM mass and
27      chemical components at fixed-site ambient monitors; for purposes of this chapter, this spatial
28      variability is called an "ambient gradient." The data presented in Section 3.2.3 indicate that
29      ambient gradients of PM and its constituents exist in urban  areas to a greater  or lesser degree.
30      This gradient and any that may exist between a fixed-site monitor and the outdoor
31      microenvironments near where people live, work, and play  obviously affects exposure. The

        June 2003                                 5-63         DRAFT-DO NOT QUOTE OR CITE

-------
 1      purpose of this section is to review the available data on ambient monitor-to-outdoor
 2      microenvironmental concentration gradients or relationships that have been measured by
 3      researchers since 1996. These analyses below are, in general, consistent with the previous
 4      studies covered in the 1996 PM AQCD.  A few outdoor-to-outdoor monitoring studies also are
 5      included to highlight relationships among important microenvironment categories. To assess
 6      spatial variability or gradients, the spatial correlations in the data are usually  analyzed.
 7      However, it should be noted that high temporal correlation between two monitoring locations
 8      does not imply low spatial variability or low ambient gradients. High temporal correlation
 9      between two sites indicates that changes in concentrations at one site can be estimated from data
10      at another site.
11           In a paper on the EXPOLIS-EAS study, Oglesby et al. (2000) conclude that in Basel,
12      Switzerland little spatial variability exists between PM levels measured at fixed site monitors
13      and the participants' outdoor microenvironments. The authors report a high correlation between
14      home outdoor PM25 levels (48-h measurements beginning and ending at  8:00 a.m.) and the
15      corresponding 24-h average PM4 (time-weighted values calculated from midnight to midnight)
16      measured at a fixed monitoring station (n =  38, rsp = 0.96, p < 0.001). They considered each
17      home outdoor monitor as a temporary fixed monitor and concluded that "the PM2 5 level
18      measured at home outdoors . .  . represents the fine particle level prevailing in the city  of Basel
19      during the 48-h measuring period."
20           In a study conducted in Helsinki, Finland, Buzorius et al. (1999) concluded that  a single
21      monitor may be used to adequately describe the temporal variations in concentration across the
22      metropolitan area. Particle size distributions were measured using a differential mobility particle
23      sizer (DMPS; Wintlmayer) coupled with a condensation particle counter (CPC TSI 3010, 3022)
24      at four locations including the  official air monitoring station, which represented a "background"
25      site. The monitoring period varied between 2 weeks and 6 months for the sites, and data were
26      reported for 10-min and 1-, 8-, and 24-h averages.  As expected, temporal variation decreased as
27      the averaging time increased.  The authors report that particle number concentration varied in
28      magnitude with local traffic intensity. Linear correlation coefficients computed for all possible
29      site-pairs and averaging times  showed that the correlation coefficient improved with increasing
30      averaging time. Using wind speed and direction vectors, lagged correlations were calculated and
31      were generally higher than the "raw" data correlations. Weekday correlations were higher than

        June 2003                                 5-64         DRAFT-DO  NOT QUOTE OR CITE

-------
 1      weekend correlations as "traffic provides relatively uniform spatial distribution of particulate
 2      matter" (p. 565).  The authors conclude that, even for time periods of 10 min and 1 h, sampling
 3      at one station can describe temporal variations across relatively large areas of the city with a
 4      correlation coefficient > 0.7.
 5           Dubowsky et al. (1999) point out that, although the variation of PM25 mass concentration
 6      across a community may be small, there may be significant spatial variations of specific
 7      components of the total mass on a local scale.  An example is given of a study of concentrations
 8      of polycyclic aromatic hydrocarbons (PAH) at three indoor locations in a community:
 9      (1) an urban site, (2) a  semi-urban site 1.6 km away, and (3) a suburban site located further
10      away. The authors found the geometric mean PAH concentrations at these three locations varied
11      respectively as 31:19:8 ng/m3 and suggest that the local variations in traffic density were
12      responsible for this gradient. Note that these concentrations are 1,000 times lower than the total
13      PM mass concentration so that such a small gradient would not be detectable for total PM25 mass
14      measurements on the order of 25 jig m"3.
15           The Total Human Environmental Exposure Study (THEES) reported by Waldman et al.
16      (1991) measured indoor, outdoor, and personal benzo(a)pyrene (BaP) levels and found that the
17      outdoor BaP was the same at all outdoor sites across the three sampling periods.  This  study
18      showed the seasonal differences versus BaP levels and exposures due to indoor and outdoor
19      sources and individual activities.
20           Leaderer et al. (1999) monitored 24-h PM10, PM2 5, and sulfates during the summers of
21      1995 and 1996 at a regional site in Vinton, VA (6 km from Roanoke, VA). One similar 24-h
22      measurement was made outdoors at residences in the surrounding area at distances ranging from
23      1 km to > 175 km from the Vinton site,  at an average separation distance of 96 km. The authors
24      reported significant correlations for PM2 5 and sulfates between the residential outdoor values
25      and those measured at Vinton on the same day. In addition, the mean values of the regional site
26      and residential site PM25 and sulfates showed no significant differences in spite of the  large
27      distance separations and mountainous terrain intervening in most directions. However, for the
28      concentrations of PM25.10 estimated as PM10-PM2 5, no significant correlation among these sites
29      was found (n = 30; r = -0.20).
30           Lillquist et al. (1998) found no significant gradient in PM10 concentrations in Salt Lake
31      City, UT, when levels were low, but a gradient existed when levels were high. PM10

        June 2003                                 5-65        DRAFT-DO NOT QUOTE OR CITE

-------
 1      concentrations were measured outdoor at three hospitals using a Minivol 4.01 sampler
 2      (Airmetrics, Inc.) operating at 5 L min"1 and at the Utah Department of Air Quality (DAQ)
 3      ambient monitoring station located between 3 and 13 km from the hospitals for a period of about
 4      5 mo.
 5           Pope et al. (1999) monitored ambient PM10 concentrations in Provo, UT (Utah Valley),
 6      during the same time frame the following year and reported nearly identical concentrations at
 7      three sites separated by 4 to 12 km.  Pearson correlation coefficients for the data were between
 8      0.92 and 0.96.  The greater degree of variability in the Salt Lake City PM10 data relative to the
 9      Provo data may be related to the higher incidence of wind-blown crustal material in Salt Lake
10      City. Pope et al. (1999) reported that increased adverse health effects in the Utah Valley were
11      associated with  stagnation and thermal inversions trapping anthropogenically derived PM10;
12      whereas, no increases in adverse health effects were observed when PM10 levels were increased
13      during events of wind blown crustal material.
14           Vakeva et al.  (1999) found significant vertical gradients in submicron particles existed in
15      an urban street canyon of Lahti, Finland.  Particle number concentrations were measured using a
16      TSI screen diffusion battery and a condensation particle counter at 1.5 and 25 m above the street
17      at rooftop level. The authors found a fivefold decrease in concentration between the two
18      sampling heights and attributed the vertical gradient to dilution and dispersion of pollutants
19      emitted at street level.
20           White (1998) suggests that the higher random measurement error for the coarse PM
21      fraction compared to the error for the fine PM fraction may be responsible for a major portion of
22      the apparent greater spatial variability of coarse ambient PM concentration compared to fine
23      ambient PM concentration in a community (e.g., Burton et al., 1996; Leaderer et al., 1999).
24      When PM2 5 and PM10 are collected independently, and the coarse fraction is obtained by
25      difference (PM2 5_10 = PM10-PM2 5), the expected variance in the coarse fraction is influenced by
26      the variances of the PM10 and PM25 measurements. When a dichotomous sampler collects PM25
27      and PM2 5_10 on two separate filters, the coarse fraction also is expected to have a larger error than
28      the fine fraction. There is a possible error caused by loss of mass below the cut-point size and a
29      gain of mass above the cut-point size that is created by the asymmetry of the product of the
30      penetration times PM concentration about the cut-point size. Because a dichotomous PM
        June 2003                                 5-66        DRAFT-DO NOT QUOTE OR CITE

-------
 1      sampler collects coarse mass using an upper and lower cut-point, it is expected to have a larger
 2      variance than for fine mass collected using only one cut-point.
 3           Wilson and Suh (1997) conclude that PM2 5 and PM10 concentrations are correlated more
 4      highly across Philadelphia than are PM2 5_10 concentrations. Ambient monitoring data from 1992
 5      to 1993 was reviewed for PM2 5, PM2 5_10, and PM10, as well as for PM25 and PM25.10 dichotomous
 6      data for 212 site-years of information contained in the AIRS database (U.S. Environmental
 7      Protection Agency, 2000).  The authors also observed that PM10 frequently was correlated more
 8      highly with PM2 5 than with PM2 5_10.  The authors note that PM2 5 constitutes a large fraction of
 9      PM10 and that this is the likely reason for the strong agreement between PM2 5 and PM10. Similar
10      observations were made by Keywood et al. (1999) in six Australian cities. The authors reported
11      that PM10 was more highly correlated with PM2 5 than with coarse PM (PM2 5_10) suggesting that
12      "variability in PM10 is dominated by  variability in PM25."
13           Lippmann et al. (2000) examined the site-to-site temporal correlations in Philadelphia
14      (1981 to 1994) and found the ranking of median site-to-site correlation was O3 (0.83), PM10
15      (0.78), TSP (0.71),  NO2 (0.70), CO (0.50), and SO2 (0.49). The authors explain that O3 and a
16      fraction of TSP and PM10 (e.g., sulfate) are secondary pollutants that would tend to be distributed
17      more uniformly spatially within the city than primary pollutants such as CO and SO2 which are
18      more likely to be influenced by local emission sources.  Lippman et al. (2000) conclude "Thus,
19      spatial uniformity of pollutants may be due to area-wide sources, or to transport (e.g., advection)
20      of fairly stable pollutants into the urban area from upwind sources. Relative spatial uniformity
21      of pollutants would therefore vary from city to city or region to region."
22           Goswami et al. (2002) used data collected at outdoor monitors of homes in a large
23      exposure study in Seattle, WA to analyze the spatial variability of outdoor PM2 5 concentrations.
24      The day-to-day variability between sites was 10 times higher than the spatial variability between
25      sites. However, differences between sites was sufficient to potentially contribute to
26      measurement error.  An examination of the spatial characteristics of the monitoring sites showed
27      that the most representative monitoring sites were located at elevations of 80-120 m above sea
28      level  and at distances of 100-300 m from the nearest arterial road.
29
        June 2003                                 5-67        DRAFT-DO NOT QUOTE OR CITE

-------
 1      5.3.3.2.2 Physical Factors Affecting Indoor Microenvironmental Particulate Matter
 2               Concentrations
 3           Several physical factors affect ambient particle concentrations in the indoor
 4      microenvironment including air exchange, penetration, and particle deposition. Combined, these
 5      factors are critical variables that describe ambient particle dynamics in the indoor
 6      microenvironments and, to a large degree, significantly affect an individual's personal exposure
 7      to ambient particles while indoors. The relationship between the concentrations of ambient
 8      particles outdoors, C, and ambient particles that have infiltrated indoors, Cai, is given by
 9
10                                     Cai/C=Pa/(a + kl                                (5-14)
11
12      where P is the penetration factor; a is the air exchange rate; and k is the particle deposition rate.
13      (As discussed in Section 5.3.2.3.1, use of this model assumes equilibrium conditions and
14      assumes that all variables remain constant.) Particle penetration is a dimensionless quantity that
15      describes the fraction of ambient particles that effectively penetrates the building shell. "Air
16      exchange" is a term used to describe the rate at which the indoor air in a building or residence is
17      replaced by outdoor air.  The dominant processes governing particle penetration are air exchange
18      and deposition of particles as they traverse through cracks and crevices and other routes of entry
19      into the  building.  Although air exchange rates have been measured in numerous studies, very
20      few field data existed prior to 1996 to determine size-dependent penetration factors and particle
21      deposition rates. All three parameters (P, a, and K) may vary substantially depending on
22      building type, region of the country, and season. In the past several years, researchers have
23      made significant advancements in understanding the relationship between particle size and
24      penetration factors and particle deposition rates.  This section will highlight the studies that have
25      been conducted to better understand physical factors affecting indoor particle dynamics.
26
27      Air Exchange Rates
28           The air exchange rate, a, in a residence varies depending on a variety of factors, including
29      geographical location, age of the building, the extent to which window and doors are open,  and
30      season.  Murray and Burmaster (1995) used measured values of a from households throughout
31      the United States to describe empirical distributions and to estimate univariate parametric

        June 2003                                 5-68        DRAFT-DO NOT QUOTE OR CITE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
probability distributions of air exchange rates. Figure 5-6 shows the results classified by season
and region. In general, a is highest in the warmest region and increases from the coldest to the
warmest region during all seasons. Air exchange rates also are quite variable within and
between seasons, as well as between regions (Figure 5-6).  Data from the warmest region in
summer should be viewed cautiously as many of the measurements were made in Southern
California in July when windows were more likely to be open than in other areas of the country
where air-conditioning is used.  Use of air-conditioning generally results in lowering air
exchange rates. In a separate analysis of these data, Koontz and Rector (1995) suggested that a
conservative estimate for air exchange in residential settings would be 0.18 h"1  (10th percentile)
and a typical air exchange would be 0.45 h-1 (50th percentile).
                         3-
                    a
                    m
                    er
                    ©
                    CD
                    C
                    co
                    o
                    X
                    LU
                  2 -
                                    Coldest Region
                                    Colder Region
                                    Warmer Region
                                    Warmest Region
                                   l
                                                                I
                                   Winter      Spring     Summer
                                                   Season
                                                               Fall
        Figure 5-6.  Air exchange rates measured in homes throughout the United States. Climatic
                    regions are based on heating-degree days: Coldest region ^ 7000, Colder
                    region = 5500 to 6999, Warmer region = 2500 to 4999, and Warmest region
                    ^ 2500 heating-degree days.
        Source: Adapted from Murray and Burmaster (1995).
       June 2003
                                         5-69
DRAFT-DO NOT QUOTE OR CITE

-------
 1           These data provide reasonable experimental evidence that a varies by season in locations
 2     with distinct seasons.  As a result, infiltration of ambient particles may be more efficient during
 3     warmer seasons when windows are likely to be opened more frequently and air exchange rates
 4     are higher. This suggests that the fraction of ambient particles present in the indoor
 5     microenvironment would be greater during warmer seasons than colder seasons. For example, in
 6     a study conducted in Boston, MA, participants living in non-air-conditioned homes kept the
 7     windows closed except during the summer (Long et al., 2000). This resulted in higher and more
 8     variable air exchange rates in summer than during any other season (Figure 5-7).  During
 9     nighttime periods when indoor sources are negligible, the indoor/outdoor concentration ratio or
10     infiltration factor may be used to determine the relative contribution of ambient particles in the
11     indoor microenvironment. Particle data collected during this study were used to determine the
12     indoor/outdoor concentration ratios by particle size (Figure 5-8).  For these nine homes in
13     Boston the fraction of ambient particles penetrating indoors was higher during summer when air
14     exchange rates were higher than during fall when air exchange rates were lower (Long et al.,
15     200 la).
16
                            7 •
in
tu  6 -
D)
c
03
€  5
X
                         (U
                         en
                         c
                         o
                         X
                        UJ
                            3 •
                            2 -
                           0 -
                                                                    »95%
                                                                     90%
                                  Fall
                                            Winter
                                                       Spring
                                                   Season
                                                                 Summer
       Figure 5-7.  Box plots of hourly air exchange rates stratified by season in Boston, MA,
                    during 1998.
       Source: Long et al. (2000).
       June 2003
                         5-70
DRAFT-DO NOT QUOTE OR CITE

-------
                1.1
                1.0 -
                0.9 -
             _O 0.8 -
             "o
             CO 0.7 -
             LJ_
             C 0.6 -
             g
             'ro °-5-
             &_
             1= 0.4 -
             M—
             — 0.3 -
                0.2 -
                0.1 -
                0.0
                                       0.1
                                          Summer  Fall
•<"  LO  CsJ
9  •<-  o
§  °  !*
d  d  d
                                                 m  •* in
                                                 999
                                                 (N  CO -^
                                                 odd
                                                                     co
                                                                 -st- in CD
                                                                 CO -4 LO
                                                          Particle Diameter (|jm)
        Figure 5-8.  Geometric mean infiltration factor (indoor/outdoor ratio) for hourly
                    nighttime, nonsource data for two seasons. Box plots of air exchange rates are
                    shown as inserts for each plot. (Boston, 1998)
        Source: Longetal. (200la).
 1
 2
 3
 4
 5
 9
10
11
12
13
14
     Wallace et al. (2002) conducted a year-long investigation of air rates in an occupied house
to establish the effects of temperature, wind velocity, use of exhaust fans, and window-opening
behavior.  Air exchange rates were calculated by periodically injecting a tracer gas (SF6) into the
return air duct and measuring the concentration in 10 indoor locations sequentially every minute
by a gas chromatograph equipped with an electron capture detector. Temperature and relative
humidity were measured outdoors and in multiple indoor locations. Wind speed and direction in
the horizontal plane were also measured. Use of the thermostat-controlled attic fan was recorded
automatically.  Indoor temperatures increased from 21 °C in winter to 27 °C in summer.
Windows were open only a few percent of the time in winter but more than half the time in
summer.  About 4600 hour-long average air exchange rates were calculated from the measured
tracer gas decay rates.  The mean (SD) rate was 0.65 (0.56) h"1. Tracer gas decay rates in
different rooms were very similar, ranging only from 0.62 to 0.67 h"1,  suggesting that conditions
were well mixed throughout the year. The strongest influence on air change rates was opening
windows, which could increase the rate to as much as 2 h"1 for extended periods, and up to 3 h"1
       June 2003
                                          5-71
                     DRAFT-DO NOT QUOTE OR CITE

-------
 1      for short periods of a few hours. The use of the attic fan also increased air change rates by
 2      amounts up to 1 h"1. Use of the furnace fan had no effect on air change rates (ducts were all
 3      inside the house). A clear effect of indoor-outdoor temperature difference (AT) could be
 4      discerned. However, wind speed and direction were found to have very little influence on air
 5      change rates at the house where the measurements were made.
 6           The air exchange rate, a, is a critical parameter in determining the fraction of ambient PM
 7      found indoors and the extent of build-up of indoor PM due to indoor sources.  Wallace et al.
 8      (2002) provide a brief review of the existing literature on the theory and measurement of air
 9      exchange rates. Open windows and frequent opening of doors lead to higher values of a.
10      However, many homes are kept closed for heating in winter and air-conditioning in summer.
11      Windows may or may not be opened during moderate weather conditions. In some areas,
12      however, heating or air-conditioning may not be required and ventilation by open windows may
13      be more frequent. Thus, a may vary geographically with climate. However, wind speed and
14      direction were found to have very little influence on air exchange rates at the house where
15      measurements were made. The variation of a with AT, as shown in Figure 5-9, is given by a
16      (h'1) = 0.176 + 0.0162 AT (°C). Thus, an increase of 10 °C in AT would lead to an increase in a
17      of 0.164 h"1, almost a doubling of the value of a for no indoor/outdoor temperature difference
18      and no open windows or forced ventilation.
19           The observation of a minimal effect of wind speed on a is an important finding.  If a were
20      strongly  dependent on wind speed, especially at low wind speeds, there might be a correlation
21      between the indoor-generated PM found indoors and the ambient concentration outdoors. Such a
22      correlation could lead to a confounding of the relationship of ambient PM with health outcomes
23      by indoor-generated PM. Wallace et al. (2002) suggest that "the generally tighter construction
24      of homes and the use of vapor barriers may have reduced the effect of wind speed and direction
25      on residential air change rates compared to earlier studies."
26           Wind speed might be expected to have a larger effect on a in a home with open windows.
27      Under conditions of large a, the ambient infiltrated indoors PM concentration will be a large
28      fraction of the outdoor PM and the two concentrations will be highly correlated. However,  the
29      indoor-generated PM concentration will be kept low by the high a so a significant correlation
30      between ambient PM concentrations and indoor-generated PM concentrations would not be
31      likely. The observed lack of a strong wind effect on a in closed houses (Wallace et al., 2002;

        June 2003                                5-72        DRAFT-DO NOT QUOTE OR CITE

-------
            0.85

       f"  0.75 H
        Jc
        I   0.65 H
        oe
        o>
        ra  0.55 H
        c
        re
        |   0.45-
        LLI
        L_
        <   0.35'
            0.25
                           8        12       16       20       24       28
                                  Absolute Temperature Difference (°C)
                                                                           32
                                36
       Figure 5-9. Regression of air exchange rate on absolute indoor-outdoor temperature
                   difference. Overnight values (midnight-8 a.m.) in winter (January-March
                   2000). a (h'1) = 0.176 (0.011 SE) + 0.0164 (0.0005) AJC0) (N= 183, R2 = 0.82).
 1
 2
 3
 4
 5
 9
10
11
12
13
14
Howard-Reed et al., 2002) and the observed lack of correlation between ambient PM
concentrations and indoor-generated PM concentrations indicate that the possibility of
correlation of ambient and indoor-generated PM concentrations can be discounted.

Particle Deposition Rates and Penetration Factors
     Physical factors affecting indoor particle concentrations including particle deposition rates,
k, and penetration factors, P, are possibly the most uncertain and variable quantities. Although k
can be modeled with some success, direct measurements are difficult, and results often vary
from study to study. Particle deposition rates vary considerably depending on particle size
because of the viscous drag  of air on the particles hindering their movement to varying degrees.
The nature and composition of particles also affect deposition rates. Surface properties of
particles, such as their electrostatic properties, can have a significant influence on deposition
rates.  In addition, thermophoresis can also affect k, but probably to a lesser degree in the indoor
microenvironment because temperatures generally vary over a small range.  Combined, these
       June 2003
                                          5-73
DRAFT-DO NOT QUOTE OR CITE

-------
 1      effects can produce order of magnitude variations in k between particles of different size and, in
 2      the case of electrophoresis and thermophoresis, particles of the same size.
 3           Particle penetration efficiency into indoor microenvironments depends on particle size and
 4      air exchange rates.  Penetration varies with particle size because of the size-dependent deposition
 5      of particles caused by impact!on, interception,  and diffusion of particles onto surfaces as they
 6      traverse through cracks and crevices.  Penetration also is affected by air exchange rates.  When
 7      air exchange rates are high, P approaches unity because the majority of ambient particles have
 8      less interaction with the building shell. In contrast, when air exchange rates are low, P is
 9      governed by particle deposition as particles travel through cracks and crevices.
10           Significant advancements have been made in the past few years to better characterize
11      particle deposition rates and penetration factors.  Several new studies, including two in which
12      semi-continuous measurements of size distributions were measured indoors and outdoors, have
13      produced new information on these quantities,  which are key to understanding the contributions
14      of ambient PM to indoor PM concentrations (Equation 5-7).
15           Studies involving semi-continuous measurements  of indoor and outdoor particle size
16      distributions have been used to estimate k and P as a function of particle size (Vette et al., 2001;
17      Long et al., 2001a; Abt et al., 2000b).  These studies each demonstrated that the indoor/outdoor
18      concentration ratios (Cai/C in Equation 5-9) were highest for accumulation mode particles and
19      lowest for ultrafine and coarse-mode particles.  Various approaches were used to estimate size-
20      specific values for k and P. Vette et al. (2001) and Abt et al. (2000b) estimated k by measuring
21      the decay of particles at times when indoor levels were significantly elevated. Vette et al. (2001)
22      estimated P using measured values of k and indoor/outdoor particle measurements during
23      nonsource nighttime periods. Long et al. (2001a) used a physical-statistical model, based on
24      Equation 5-12, to estimate k and P during nonsource nighttime periods. The results for k
25      reported by Long et al. (200la) and Abt et al. (2000b) are compared with other studies in
26      Figure 5-10.  Although not shown in Figure 5-10, the results for k obtained by Vette et al. (2001)
27      were similar to the values of & reported by Abt et al. (2000b) for particle sizes up to 1 jim.
28      Results for P by Long et al. (200la) show that  penetration was highest for accumulation-mode
29      particles and decreased substantially for coarse-mode particles (Figure 5-11). The results for
30      P reported by Vette  et al. (2001) show similar trends, but are lower than those reported by Long
31      et al. (2001a). This likely is because of lower air exchange rates in the one Fresno, CA, test

        June 2003                                 5-74        DRAFT-DO NOT QUOTE OR CITE

-------
          CD
          O
          CD
          Q

;
•
•

10-
\
•
•
1;
-
.
0.1-
.


o Byrne ef al. (1992)
o Foghefa/. (1997)a
v Ligocki ef a/. (1990)b
v Ozkaynak ef al. ( 1 996a)c
n Sinclair ef al. (1988, 1990)b
D Thatcher and Layton (1 995)d
0 Wallace ef a/. (1997)d
T Abt ef a/. (2000b)d
• Longefa/. (2001a)d'e
I
I
* $ f i



1 1 1 1 1 1 1 1 |
0.01 0.1









i


* <


1
n« I-









i


' i


1











j
<



,
j
]









[ { fiH
i q
li ol i
[ '
error bar
includes 0


1 1 1 1 1 1 1










,k
T
X
[








T _
fl
> D? '



f
J
I
I


error bar
includes 0
1
5
1 1 1 1 1 1 1
10
                  a Decay Rates represent Summary
                   Estimates from the four houses examined.
                  b Decay rates are based on sulfate and are
                   presented as < 2.5 um.
                   Estimates were computed using a surface-
                   to-volume ratio of 2 rrr1 (Koutrakis eta/., 1992).
                  c Data represents PWb.g.
                  d Particle sizes are the midpoint of the ranges examined.
                                                       e Decay rates presented are estimates of k for nightly
                                                        average data from all nine study homes.

                                                       f Decay rates are theoretically modeled deposition
                                                        values for smooth indoor surfaces and homogeneous
                                                        and isotropically turbulent air flow.
                                                        Presented curves assume typical room dimensions
                                                        (3 m x 4 m x 5  m) and a friction velocity of 1.0 cm/s.
       Figure 5-10.  Comparison of deposition rates from Long et al., 2001a with literature values
                      (from Abt et al., 2000b).  Error bars represent standard deviations for same-
                      study estimates.

       Source: Adapted from Abt et al. (2000b); Long et al. (2001a).
1

2

3

4

5
home (a ~ 0.5 h"1; Vette et al., 2001) than in the nine test homes in Boston, MA, study (a > 1 h"1;

Long et al., 2001a). These data for P and k illustrate the role that the building shell may provide

in increasing the concentration of particles because of indoor sources and reducing the

concentration of indoor particles from ambient sources especially for homes with low air

exchange rates.
       June 2003
                                               5-75
DRAFT-DO NOT QUOTE OR CITE

-------
U
HI
C
o
"ro
+.»
o>
C
0)
Q.
             1.0
             0.9
             o,
             0.7
             0-6
             0.5
             0.4
             0.3
             0.2
             0.1
             0.0
h
-J
   "-J
                    1K'-i-.-,-,-
                                                          1.0
                                                          0.9
                                                          0.8 ^
                                                          0.7 S
                                                          0.6  «"
                                                              (0
                                                          0.5  *
                                                              O
                                                          0.4
                                                          n o
                                                          0.3  Q.
                                                              0)
                                                          0.2  Q

                                                          0.1

                                                          0.0
                  o  cp
                  g  8
                  o  o
                         CD
                         O
                s
°  00
CD  O
O  o
                              cp
                              in
                         rt
                         9
                         CN|
                         O
                            If)
                            O
                            •4
                            O
                                                            m
                                                            (N
                                                     O
                                                     5
                                            Size Interval (|jm)
       Figure 5-11.  Penetration efficiencies and deposition rates from models of nightly average
                     data.  Error bars represent standard errors.  (Boston, 1998, winter and
                     summer).
       Source: Longetal. (2001a).
 1           Several other studies have investigated particle loss as a function of particle size.  The
 2     penetration of particles across building envelopes has been modeled for several sizes of idealized
 3     rectangular cracks (Liu and Nazaroff, 2001). Particles of 0.1 to 1.0 jim diameter had penetration
 4     efficiencies near 1.0. Supermicron and ultrafine particles were removed to a greater degree by
 5     gravitational settling and Brownian diffusion. Thatcher et al. (2002) conducted an experimental
 6     study of the effects of room furnishings and air speed on particle deposition rates indoors. The
 7     deposition loss rate (K) increased by as much as a factor of 2.6 in going from a bare room (35 m2
 8     surface area) to a fully furnished room (12 m2 additional surface area) with the greatest increase
 9     seen for the smallest particles.  Air speed increases from < 5 to 19 cm/s increased the deposition
10     rates by factors of 1.3 to 2.4 with greater effects on large particles than small particles.  The
11     authors state that, "The significant effect of particle size and room conditions on deposition loss
       June 2003
                                        5-76
                               DRAFT-DO NOT QUOTE OR CITE

-------
 1      rates argues against using a single first-order loss-rate coefficient to represent deposition for
 2      integrated mass measurements (PM25 or PM10)." Riley et al. (2002) have modeled the
 3      infiltration of particles into two building types:  offices and residences.  They developed
 4      representative values ofP, k, and ventilation-system filter efficiencies for particle sizes from
 5      0.001 to 10 |im.  They then used a typical rural and urban outdoor size distribution and
 6      calculated predicted indoor concentrations for number, surface area, and volume distributions.
 7
 8      Compositional Differences Between Indoor-Generated and Ambient Particulate Matter
 9           Wilson et al. (2000) discuss the differences in composition between particles from indoor
10      and outdoor sources.  They note that, because of the difficulty in separating indoor PM into
11      ambient and nonambient PM, there is little direct experimental information on the composition
12      differences between the two.  Although experimental data are limited, Wilson et al. (2000)
13      suggest the following.
14
15             Photochemistry is significantly reduced indoors; therefore, most secondary sulfate [H2S04,
16             NH4HS04, and (NH4)2S04] and nitrate (NH4N03) found indoors come from ambient sources.
17             Primary organic emissions from incomplete combustion may be similar, regardless of the
18             source.  However, atmospheric reactions of polyaromatic hydrocarbons and other organic
19             compounds produce highly oxygenated and nitrated products, so these species are also of
20             ambient origin.  Gasoline, diesel fuel, and vehicle lubricating oil all contain naturally present
21             metals or metal additives.  Coal and heavy fuel oil also  contain more metals  and nonmetals,
22             such as selenium and arsenic, than do materials such as wood or kerosene burned inside
23             homes.  Environmental tobacco smoke (ETS), however, with its many toxic components, is
24             primarily an indoor-generated pollutant.
25
26           Particles generated indoors may have different chemical and physical properties than those
27      generated by anthropogenic ambient sources.  Siegmann et al. (1999) have  demonstrated that
28      elemental carbon in soot particles generated indoors has  different properties than in those
29      generated outdoors by automotive or diesel engines.  In the United States,  combustion-product
30      PM in the ambient/outdoor air generally is produced by burning fossil fuels (e.g., coal, gasoline,
31      fuel oil) and wood; whereas combustion-product PM from indoor sources is produced by
32      biomass burning (e.g., tobacco, wood, foods, etc.). However, some indoor  sources of PM (such
33      as cigarette smoking, meat cooking, and coal burning) occur both indoors and outdoors  and may

        June 2003                                  5-77        DRAFT-DO NOT QUOTE  OR CITE

-------
 1      constitute an identifiable portion of measured ambient PM (Cha et al., 1996; Kleeman and Cass,
 2      1998).
 3
 4      Indoor Air Chemistry
 5           Gas- and aerosol-phase chemical reactions in the indoor microenvironment are responsible
 6      for secondary  particle formation and modification of existing particles. This process could be
 7      complex and may influence the interpretation of exposures to indoor generated particles in
 8      instances when particles are generated by outdoor gases reacting with gases generated or
 9      released indoors to produce fresh particles. For example, homogeneous gas phase reactions
10      involving ozone and terpenes (specifically d-limonene, cc-terpinene, and cc-pinene) have been
11      identified as an important source of submicron particles (Weschler and Shields, 1999). Terpenes
12      are present in  several commonly available household cleaning products, and d-limonene has
13      been identified in more than 50% of the buildings monitored in the Building Assessment and
14      Survey Evaluation (BASE) study (Hadwen et al., 1997).  Long et al. (2000) found that when
15      PineSol (primary ingredient is cc-pinene) was used indoors, indoor PM25 mass concentrations
16      increased from 11 to 32  jig m"3 (indoor ozone concentrations unknown, but ambient ozone
17      concentrations were 44 to 48 ppb).  Similarly, a 10-fold increase in number counts of 0.1 to
18      0.2 |im particles was observed in an experimental office containing supplemented d-limonene
19      and normally encountered  indoor ozone concentrations (< 5  to 45 ppb), resulting in an average
20      increase in particle mass concentration of 2.5 to 5.5 jig m"3 (Weschler and Shields, 1999).  Ozone
21      appears to be the limiting reagent as particle number concentration varied proportionally to
22      ozone concentrations (Weschler and Shields, 1999).  Other studies showed similar results (e.g.,
23      Jang and Kamens, 1999; Wainman et al., 2000).  Such particles, if toxic, would represent an
24      increased health risk due to ambient air pollution. However, the concentration would  depend on
25      the ambient O3 concentration, the O3 infiltration factor, and the indoor generation rate of
26      terpenes. Because the concentration of the resulting particles would not be expected to be
27      correlated with ambient  PM on an individual or population basis, it seems more appropriate to
28      consider indoor-reaction particles as part of nonambient exposure.
29
30
        June 2003                                 5-78        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Indoor Sources of Particles
 2           The major sources of indoor PM in nonsmoking residences and buildings include
 3      suspension of PM from bulk material, cooking, cleaning, and the use of combustion devices such
 4      as stoves and kerosene heaters. Human and pet activities also lead to PM detritus production
 5      (from tracked-in soil, fabrics, skin and hair, home furnishings, etc.) that is found ubiquitously in
 6      house dust deposited on floors and other interior surfaces. House dust and lint particles may be
 7      resuspended indoors by agitation (cleaning) and turbulence (HVAC systems, human activities,
 8      etc.). Ambient particles that have infiltrated into the indoor microenvironments also may be
 9      resuspended after deposition to indoor surfaces. Typically, resuspension of particles from any
10      source involves coarse particles (> 1 jim); particles with smaller diameters are not resuspended
11      efficiently. On the other hand, cooking produces both fine and coarse mode particles whereas
12      combustion sources typically produce ultrafine particles.
13           Environmental tobacco smoke (ETS) is also a major indoor source of PM. It is, however,
14      beyond the scope of this chapter to review the extensive literature on ETS. A number of articles
15      provide source-strength information for cigarette or cigar smoking (e.g., Daisey et al., 1998 and
16      Nelson etal., 1998).
17           A study conducted on two homes in the Boston metropolitan area (Abt et al., 2000a)
18      showed that indoor PM sources predominate when air exchange rates were < 1 h"1, and outdoor
19      sources predominate when air exchange rates were > 2 h"1.  The authors attributed this to the fact
20      that when air exchange rates were low (<  1 h—1), particles released from indoor sources tend to
21      accumulate because particle deposition is the mechanism governing particle decay and not air
22      exchange. Particle deposition rates are generally < 1 h"1, especially for accumulation-mode
23      particles. When air exchange rates were higher (>  2 h"1), infiltration of ambient aerosols and
24      exfiltration of indoor-generated aerosols occur more rapidly, reducing the effect of indoor
25      sources on indoor particle levels.   The study also confirmed previous findings that the major
26      indoor sources of PM are cooking, cleaning, and human activity. They discuss the size
27      characteristics of these ubiquitous sources and report the following.
28
29            The size of the particles generated by these activities reflected their formation processes.
30            Combustion processes (oven cooking, toasting, and barbecuing) produced fine particles and
31            mechanical processes (sauteing, frying, cleaning, and movement of people) generated coarse
32            particles. These activities increased particle concentrations by many orders of magnitude
33            higher than outdoor levels and altered indoor size distributions.  (Abt et al., 2000a; p. 43)

        June 2003                                  5-79        DRAFT-DO NOT QUOTE OR CITE

-------
 1      They also note that variability in indoor PM for all size fractions was greater than for outdoor
 2      PM, especially for short averaging times (2 to 33 times higher).
 3           In a separate study conducted in nine nonsmoking homes in the Boston area, Long et al.
 4      (2000) concluded that the predominant source of indoor fine particles was infiltration of outdoor
 5      particles and that cooking activities were the only other significant source of fine particles.
 6      Coarse particles, however, had several indoor sources, such as cooking, cleaning, and various
 7      indoor activities.  This study also concluded that more than 50% of the particles (by volume)
 8      generated during indoor events were ultrafme particles. Events that elevated indoor particle
 9      levels were found to be brief, intermittent,  and highly variable, thus requiring the use of
10      continuous instrumentation for their characterization. Because the concentration of ultrafme
11      particles will be greater near the source (they will grow in size into the accumulation mode as
12      they age), the personal cloud for ultrafme particles may be higher than for accumulation mode
13      particles if the person is nearer the source than the indoor monitor. Table 5-11 provides
14      information on the mean volume mean diameter (VMD) for various types of indoor particle
15      sources. The differences in mean VMD confirm the clear separation of source types and suggest
16      that there is very little resuspension of accumulation-mode PM. In addition, measurements of
17      organic and elemental carbon indicated that organic carbon had significant indoor sources
18      whereas elemental carbon was primarily of ambient origin.
19           Vette et al. (2001) found that resuspension was a significant indoor source of particles
20      > 1 |im. Concentrations of fine particles were not affected by resuspension. Figure 5-12 shows
21      the diurnal variability in the indoor/outdoor aerosol concentration  ratio from an unoccupied
22      residence in Fresno.  The study was conducted in the absence of common indoor particle sources
23      such as cooking and cleaning. The data in  Figure 5-12 show the mean indoor/outdoor
24      concentration ratio for particles > 1 |im increased dramatically during daytime hours. This
25      pattern was consistent with indoor human activity levels.  In contrast, the mean indoor/outdoor
26      concentration ratio for particles < 1 |im (ultrafme and accumulation-mode particles) remain
27      fairly constant during both day and night.
28           Wallace and Howard-Reed (2002) used three instruments (SMPS, APS, and Climet) to
29      measure ultrafme, fine, and coarse particles in an  inhabited residence for 18 months.  They
30      confirm the observations of Abt et al. (2000a) and Long et al. (2000) that indoor sources
31      primarily generate ultrafme and coarse particles.  Wallace and Howard-Reed report that, "Indoor

        June 2003                                 5-80        DRAFT-DO NOT QUOTE OR CITE

-------
       TABLE 5-11. VOLUME MEAN DIAMETER (VMD) AND MAXIMUM PM2 5
               CONCENTRATIONS OF INDOOR PARTICLE SOURCESa b
Size Statistics
Particle Source
Cooking
Baking (Electric)
Baking (Gas)
Toasting
Broiling
Sauteing
Stir-Frying
Frying
Barbecuing
Cleaning
Dusting
Vacuuming
Cleaning with Pine Sol
General Activities
Walking Vigorously (w/Carpet)
Sampling w/Carpet
Sampling w/o Carpet
Burning Candles
N

8
24
23
4
13
3
20
2

11
10
5

15
52
26
7
Indoor Activity
Mean VMD
(urn)

0.189f
0.107f
0.138f
0.114f
0.184f, 3.48g
0.135f
0.173f
0.159f

5.38g
3.86g
0.097f

3.96g
4.25g
4.28g
0.311f
Background3'6
Mean VMD
(urn)

0.221f
0.224f
0.222f
0.236f
0.223f, 2.93g
0.277f
0.223f
0.205f

3.53g
2.79g
0.238f

3.18g
2.63g
2.93g
0.224f
Maximum
Mean

14.8
101.2
54.9
29.3
65.6
37.2
40.5
14.8

22.6
6.5
11.0

12.0
8.0
4.8
28.0
PM25
Concentrationc>d
SD

7.4
184.9
119.7
43.4
95.4
31.4
43.2
5.2

22.6
3.9
10.2

9.1
6.6
3.0
18.0
 Notes:
 aAll concentration data corrected for background particle levels.
 blncludes only individual particle events that were unique for a given time period and could be detected above
  background particle levels.
 °PM concentrations in ug/m3.
 dMaximum concentrations computed from 5-min data for each activity.
 ^Background data are for time periods immediately prior to the indoor event.
 fSize statistics calculated for PV0 02.0 5 using SMPS data.
 gSize statistics calculated for PV0 7_10 using APS data.

 Source:  Long et al. (2000).
June 2003
5-81
DRAFT-DO NOT QUOTE OR CITE

-------
                 .s
                 +3
                  «s
                  o
                  o
                  •O
                  *••
                  a
                  O
                  T£
                  o
                  o
                  •o
                  e
                       0.0
                                            "lime
       Figure 5-12.  Mean hourly indoor/outdoor particle concentration ratio from an
                     unoccupied residence in Fresno, CA, during spring 1999.
       Source: Vette etal. (2001).
 1     sources affecting ultrafine particle concentrations were observed 22% of the time, and sources
 2     affecting fine and coarse particle concentrations were observed 12 and 15% of the time,
 3     respectively . . . Indoor sources, such as cooking with natural gas, and simple physical activities,
 4     such as walking, accounted for a majority (50-90%) of the ultrafine and coarse particle
 5     concentrations, whereas outdoor sources were more important for accumulation-mode particles
 6     between 0.1 and 1 jim in diameter."
 7
 8     5.3.3.2.3 Time/Activity Patterns
 9          Total exposure to PM is the sum of various microenvironmental exposures that an
10     individual encounters during the day and will depend on the microenvironments occupied.
11     As discussed previously, PM exposure in each microenvironment is the sum of exposures from
       June 2003
5-82
DRAFT-DO NOT QUOTE OR CITE

-------
 1      ambient PM (in outdoor and indoor microenvironments), indoor-generated PM and indoor
 2      reaction PM. In addition, there is exposure to PM generated by personal activities that is
 3      determined by the personal activities that the individual conducts while in those
 4      microenvironments.  As mentioned before, PM exposures and their components are variable
 5      across the population; thus, each are distributions rather than point estimates. A thorough
 6      analysis of these distributions would require a comprehensive sensitivity and uncertainty
 7      analysis.
 8           Determining microenvironments and activities that contribute significantly to human
 9      exposure  begins with establishing human activity pattern information for the general population,
10      as well as subpopulations. Personal exposure and time-activity pattern studies have shown that
11      different populations have varying time-activity patterns and, accordingly,  different personal PM
12      exposures.  Both characteristics will vary greatly as a function  of age, health status, ethnic group,
13      socioeconomic status, season, and region of the country. Collecting detailed time activity data
14      can be very burdensome on participants but is clearly valuable in assessing human exposure and
15      microenvironments.  For modeling purposes, human activity data frequently come from  general
16      databases that are discussed below.
17           The gathering of human activity information, often called "time-budget" data, started in the
18      1920s; however, their use for exposure assessment purposes only began to  be emphasized in the
19      1980s.  Many of the  largest U.S. human activity databases have been consolidated by EPA's
20      National Exposure Research Laboratory's (NERL) into one comprehensive database containing
21      more than 22,000 person-days of 24-h activity known as the Consolidated Human Activity
22      Database, or CHAD  (Glen et al., 1997; McCurdy et al., 2000).  The information in CHAD is
23      accessible for constructing population cohorts of people with diverse characteristics that are
24      useful for analysis and modeling (McCurdy, 2000). See Table 5-2 for a summary listing of
25      human activity studies in CHAD. Most of the databases in CHAD are available elsewhere,
26      including the National Human Activity Pattern Survey (NHAPS),  California's Air Resources
27      Board (CARB), and  the University of Michigan's Institute for  Survey Research data  sets.
28           Although CHAD provides a very valuable resource for time and location data, there is little
29      information on PM-generating personal activities.  In addition, very few of the time-activity
30      studies have collected longitudinal data within a season or over multiple seasons.  Such
31      longitudinal data are important in understanding potential variability in activities and how they

        June 2003                                 5-83        DRAFT-DO NOT QUOTE OR CITE

-------
 1      affect correlations between PM exposure and ambient site measurements for both total PM and
 2      PM of ambient origin.
 3
 4      5.3.3.3   Effect of Ambient Sources on Exposures to Particulate Matter
 5           Different sources may generate ambient PM with different aerodynamic and chemical
 6      characteristics, which may, in turn, result in different health responses.  Thus, to fully understand
 7      the relationship between PM exposure and health outcomes, exposure from difference sources
 8      should be identified and quantified. Source apportionment techniques provide a method for
 9      determining personal exposure to PM from specific sources. Daily contributions from sources
10      that have no indoor component can be used as tracers to generate exposure estimates for ambient
11      PM of similar aerodynamic size or directly as exposure surrogates in epidemiological analyses.
12      The most recent EPA PM Research Needs Document (U. S. Environmental Protection Agency,
13      1998) recommended use of source apportionment techniques to determine daily time-series of
14      source categories for use in community, time-series epidemiology.
15           A number of epidemiologic studies (discussed more fully in Chapter 8) have evaluated
16      relationships between health outcomes and sources of particulate matter determined from
17      measurements at a community monitor.  These studies suggest the importance of examining
18      sources and constituents of indoor, outdoor, and personal PM.  For example, Ozkaynak and
19      Thurston (1987) evaluated the relationship between PM sources and mortality in 36 Standard
20      Metropolitan Statistical Areas (SMSAs). Particulate matter samples from EPA's Inhalable
21      Particle (IP) Network were analyzed for SO4"2 and NO3" by automated colorimetry, and elemental
22      composition was determined with X-ray fluorescence (XRF). Mass concentrations from five PM
23      source categories were determined from multiple regression of absolute factor scores on the
24      mass concentration: (1) resuspended soil, (2) auto exhaust, (3) oil combustion, (4) metals, and
25      (5) coal combustion.
26           In another study,  Mar et al. (2000) applied factor analysis to evaluate the relationship
27      between PM  composition (and gaseous pollutants) in Phoenix. In addition to daily averages  of
28      PM2 5 elements from XRF analysis, they included in their analyses organic and elemental carbon
29      in PM2 5  and gaseous species emitted by combustion sources (CO, NO2, and SO2). They
30      identified five factors classified  as (1) motor vehicles, (2) resuspended soil, (3) vegetative
31      burning, (4) local SO2,  and (5) regional sulfate.  Additionally, Laden et al. (2000) applied

        June  2003                                 5-84        DRAFT-DO NOT QUOTE OR CITE

-------
 1      specific rotation factor analysis to participate matter composition (XRF) data from six eastern
 2      cities (Ferris et al., 1979). Fine PM was regressed on the recentered scores to determine the
 3      daily source contributions.  Three main sources were identified (1) resuspended soil (Si);
 4      (2) motor vehicle (Pb); and (3) coal combustion (Se).
 5           Source apportionment or receptor modeling has been applied to the personal exposure data
 6      to understand the relationship between personal and ambient sources of particulate matter.
 7      Application of source apportionment to ambient, indoor, and personal PM composition  data is
 8      especially useful in sorting out the effects of particle size and composition. If a sufficient
 9      number of samples are analyzed with sufficient compositional detail, it is possible to use
10      statistical techniques to derive source category signatures, to identify indoor and outdoor source
11      categories, and to estimate their contribution to indoor and personal PM.
12           Positive Matrix Factorization (PMF) has been applied to the PTEAM database by
13      Yakovleva et al. (1999).  The authors utilize mass and XRF elemental composition data from
14      indoor and outdoor PM2 5 and personal, indoor, and outdoor PM10 samples. PMF is an advance
15      over ordinary factor analysis because it allows measurements below the quantifiable limit to be
16      used by weighting them by their uncertainty. This effectively increases the number of species
17      that can be used in the model.  The factors used by the authors correspond to general source
18      categories of PM, such as outdoor soil, resuspended indoor soil, indoor soil, personal activities,
19      sea-salt, motor vehicles, nonferrous metal smelters, and secondary sulfates. PMF, by identifying
20      the various source factors and apportioning them among the different monitor locations
21      (personal, indoor, and outdoor), was able to estimate the contribution of resuspended indoor dust
22      to the personal cloud (15% from indoor soil and 30% from resuspended indoor soil).  Factor
23      scores for these items then were used in a regression analysis to estimate personal exposures
24      (Yakovleva etal., 1999).
25           The most important contributors to PM10 personal exposure were indoor soil, resuspended
26      indoor soil,  and personal activities; these accounted for approximately 60% of the mass
27      (Yakovleva et al., 1999). Collectively, they include personal cloud PM, smoking, cooking, and
28      vacuuming. For both PM2 5 and PM10, secondary  sulfate and nonferrous metal operations
29      accounted for another 25% of PM mass. Motor vehicle exhausts, especially from vehicles
30      started inside attached garages, accounted for another 10% of PM mass. The PTEAM study was
        June 2003                                 5-85        DRAFT-DO NOT QUOTE OR CITE

-------
 1      conducted in Riverside, CA in the fall of 1990. Yakovleva et al. (1999) caution that their results
 2      may not apply to other geographic areas, seasons of the year, or weather conditions.
 3           Simultaneous measurement of personal (PM10) and outdoor measurements (PM2 5 and
 4      PM10) were evaluated as a three-way problem with PMF, which allowed for differentiation of
 5      source categories based on their variation in time and type of sample, as well as their variation in
 6      composition. By use of this technique, it was possible to identify three sources of coarse-mode,
 7      soil-type PM.  One was associated with ambient soil, one with indoor soil dispersed throughout
 8      the house, and one with soil resulting from the personal activity of the subject.
 9           Two other source apportionment models have been applied to ambient measurement data
10      and can be used for the personal exposure studies.  The effective variance weighted Chemical
11      Mass Balance (CMB) receptor model (Watson et al., 1984, 1990, 1991) solves a set of linear
12      equations that incorporate the uncertainty in the sample and source composition. CMB requires
13      the composition of each potential source of PM and the uncertainty for the sources and ambient
14      measurements. Source apportionment with CMB can be conducted on individual samples;
15      however,  composition of each of the sources of PM must be known.  An additional source
16      apportionment model, UNMIX (Henry et al., 1994) is a multivariate source apportionment
17      model.  UNMIX is similar to PMF, but does not explicitly use the measurement uncertainties.
18      Because measurement uncertainties are not used, only species above the detection limit are
19      evaluated in the model. UNMIX provides the number of sources and source contributions and
20      requires a similar number of observations as PMF.
21           The Yakovleva et al. (1999) study demonstrates that source apportionment techniques also
22      could be very useful in determining parameters needed for exposure models and for determining
23      exposure to ambient PM. Exposure information, similar to that obtained in the PTEAM study,
24      but including other PM components useful for definition of other source categories (e.g.,
25      elemental [EC] and organic carbon [OC]; organic tracers for elemental carbon from diesel
26      vehicle exhaust, gasoline vehicle exhaust, and wood combustion; nitrate; Na, Mg, and other
27      metal tracers; and gas-phase pollutants) would be useful as demonstrated in the use of EC/OC
28      and gas-phase pollutants by Mar et al. (2000).
29
30
        June 2003                                 5-86        DRAFT-DO NOT QUOTE OR CITE

-------
 1      5.3.3.4  Correlations of Particulate Matter with Other Pollutants
 2           Correlations between ambient concentrations and between ambient concentrations and
 3      personal exposures for PM and other pollutants are of importance in understanding possible
 4      confounding in epidemiological studies as will be discussed more fully in Chapter 9.  The
 5      available information from exposure studies is presented in this section.  Several epidemiologic
 6      studies have included the gaseous pollutants CO, NO2, SO2, and O3 along with PM10 or PM2 5 in
 7      the analysis of the statistical association of health responses with pollutants. In a recent study,
 8      the personal exposure to O3 and NO2 was determined, as well as that to PM2 5 and PM2 5_10 for a
 9      cohort 15 elderly subjects in Baltimore, MD, although measured personal exposures to O3, NO2,
10      and SO2 were below their respective level of detection (LOD) for 70% of the samples.
11      Spearman correlations for 14 subjects in summer and 14 subjects in winter are given in
12      Table 5-12 for relationships between personal PM2 5 and ambient concentrations of PM2 5,
13      PM2 5_10, O3, and NO2. In contrast to ambient concentrations, neither personal exposure to total
14      PM2 5 nor PM2 5 ambient origin was correlated significantly with personal exposures to the
15      co-pollutants, PM25.10, nonambient PM25, O3, NO2, and SO2.  Personal-ambient associations for
16      PM2 5_10, O3, NO2, and SO2 were similarly weak and insignificant. Based on these results,  Sarnat
17      et al. (2000) conclude that the potential for confounding of PM2 5 by O3, NO2, or PM 10_25 appears
18      to be limited, because, despite significant correlations observed among ambient pollutant
19      concentrations, the correlations among personal exposures were low.
20           Sarnat et al. (2001) further evaluated the role of gaseous pollutants in particulate matter
21      epidemiology by extending the measurements taken on the earlier adult  cohort of 20 individuals
22      in Baltimore by including additional PM and gaseous pollutant measurements that were
23      collected during the same 1998-1999 period from 15 individuals with COPD and from 21
24      children, 24-h average personal exposures for PM2 5,  O3, SO2 and NO2, and  corresponding
25      ambient concentrations for PM25, O3, SO2, NO2and CO for all 56 subjects were collected  over
26      12 consecutive days.  Results from correlation and regression analysis of the personal and
27      ambient data showed that personal PM2 5 and personal gaseous  pollutant exposures were
28      generally not correlated.  The analysis also showed that ambient PM2 5 concentrations had
29      significant associations with personal PM2 5 exposures in both seasons.  On the other hand,
30      ambient gaseous pollutant concentrations were not correlated with their corresponding personal
31      exposure concentrations. However, ambient gaseous concentrations were found to be strongly

        June 2003                                 5-87        DRAFT-DO NOT QUOTE OR CITE

-------
    TABLE 5-12. CORRELATIONS BETWEEN PERSONAL PM25 AND AMBIENT
                         POLLUTANT CONCENTRATIONS1
Personal PM2 5
vs. Ambient:
SUMMER Subject
SA1
SA2
SA5
SB1
SB2
SB3
SB4
SB5
SB6
SCI
SC2
SC3
SC4
SC5
WINTER WAI
WA2
WA4
WAS
WB1
WB2
WB3
WB4
WC1
WC2
WC3
WC4
WC5
WC6
Median Summer
Median Winter
PM25
0.55
0.85
0.89
0.65
-0.21
0.82
0.73
0.73
0.53
0.95
0.78
0.85
0.78
0.55
0.22
-0.38
-0.18
0.22
0.80
0.62
0.55
-0.12
0.74
0.79
0.28
0.19
0.81
0.01
0.76
0.25
03
0.15
0.31
0.18
0.40
-0.62
0.55
0.62
0.45
0.15
0.78
0.68
0.78
0.66
0.51
-0.18
-0.07
0.67
-0.43
-0.84
-0.32
-0.45
-0.01
-0.62
-0.88
-0.42
-0.84
-0.62
-0.03
0.48
-0.43
NO2
0.38
0.66
0.82
-0.15
0.81
-0.14
-0.34
-0.42
-0.38
0.66
0.36
0.73
0.59
0.32
-0.26
-0.36
-0.22
0.61
0.77
0.59
0.62
0.34
-0.15
0.17
0.03
0.50
0.08
0.65
0.37
0.26
PM2.5.10
-0.12
0.57
0.64
0.38
0.15
-0.04
-0.12
0.23
0.12
0.65
0.51
0.68
0.70
0.43
-0.05
-0.70
-0.29
0.50
0.41
0.09
0.04
-0.10
0.44
0.77
0.57
0.45
0.81
0.37
0.41
0.39
Personal PM2 5 of Ambient Origin
vs. Ambient:
03
0.27
0.21
0.33
0.89
0.26
0.52
0.45
0.36
-0.03
0.83
0.66
0.69
0.50
0.34
-0.78
-0.15
-0.33
-0.72
-0.87
-0.76
-0.77
-0.80
-0.64
-0.87
-0.77
-0.72
-0.76
-0.75
0.41
-0.76
NO2
0.71
0.64
0.81
-0.74
0.08
-0.20
-0.29
-0.48
-0.57
0.63
0.65
0.71
0.50
0.33
-0.04
-0.15
0.20
-0.09
0.53
0.59
0.56
0.68
0.02
0.25
0.30
0.22
0.05
0.19
0.42
0.21
PM,5.10
0.15
0.68
0.79
-0.03
0.33
0.00
-0.14
0.33
0.32
0.57
0.76
0.80
0.51
0.27
-0.24
0.02
0.00
0.40
0.66
0.59
0.60
0.48
0.69
0.71
-0.45
0.67
0.42
-0.45
0.33
0.45
 'Correlations represent Spearman's rvalues; italicized values indicate significance at the a = 0.05 level.

 Source:  Sarnat et al. (2000).
June 2003
5-88
DRAFT-DO NOT QUOTE OR CITE

-------
 1      associated with personal PM2 5 exposures, suggesting that ambient gaseous concentrations for O3,
 2      NO2, SO2 are acting as surrogates, as opposed to confounders of PM2 5 in the estimation of PM
 3      health effects based on multi-pollutant models.  This study did not measure personal CO and did
 4      not find a significant association between summertime ambient CO and personal PM2 5
 5      (a significant winter-time association, however, was found).
 6           Personal EC and SO4"2 were also measured during the winter for the cohort of COPD
 7      patients only.  The analysis of this subset of the data showed that personal SO4"2 was
 8      significantly and negatively associated with ambient O3 and SO2, and personal EC was
 9      significantly associated with ambient O3, NO2, and CO.  The authors interpret these findings as
10      suggesting that O3 is primarily a surrogate for secondary particle exposures whereas ambient CO
11      and NO2 are primarily surrogates for particles from traffic. Sarnat et al. (2001) caution that these
12      findings were from only one location and various physical and personal factors, such as
13      ventilation, time spent outdoors, and household characteristics could affect the strength of the
14      reported associations for certain individuals and cohorts even though the qualitative results
15      found are unlikely to change.
16           A newly developed Roll-Around System (RAS) was used to  evaluate the hourly
17      relationship between gaseous pollutants (CO, O3, NO2, SO2, and VOCs) and PM (Chang et al.,
18      2000).  Exposures were characterized over a 15-day period for the summer and winter in
19      Baltimore, based on scripted activities to simulate activities performed by older adults (65+
20      years of age).  Spearman rank correlations were reported for PM2 5, O3, CO, and toluene for both
21      the summer and winter. The correlations are  given for each microenvironment in Table 5-13:
22      indoor residence, indoor other, outdoor near roadway, outdoor away from road, and in vehicle.
23      No significant relationships (p < 0.05) were found between hourly PM25 and O3. Significant
24      relationships were found between hourly PM2 5 and CO:  indoor residence, winter; indoor other,
25      summer and winter; and outdoor away from roadway, summer. Significant relationships also
26      were found between hourly PM2 5 and toluene: indoor residence, winter; indoor other, winter;
27      and in vehicle, winter. The significant relationships between CO and PM2 5 in the winter may be
28      caused by reduced air exchange rates that could allow them to accumulate (Chang et al., 2000).
29      Although no significant correlation was found between in vehicle PM2 5 and CO, toluene, which
30      is a significant component of vehicle exhaust (Conner et al., 1995), was significantly correlated
31      to PM2 5 in the winter.

        June 2003                                 5-89        DRAFT-DO NOT QUOTE OR CITE

-------
            TABLE 5-13. CORRELATIONS BETWEEN HOURLY PERSONAL PM25 AND
                                      GASEOUS POLLUTANTS
Indoor
Residence

PM2.5vs.
Summer
Winter
PM2.5vs.
Summer
Winter
PM2.5vs.
Summer
Winter
N
03
35
56
CO
41
59
Toluene
46
66
rs

0.29
0.05

0.25
0.43a

0.23
0.38a
Indoor Other
N

16
37

19
39

21
47
rs

-0.14
-0.06

0.59a
0.62a

-0.14
0.44a
Outdoor Near
Roadway
N

10
11

13
13

14
17
rs

0.05
-0.28

0.14
0.37

0.26
0.40
Outdoor Away
from road
N

12
7

12
8

14
8
rs

0.45
0.04

0.62
0.41

0.02
0.48
In Vehicle
N

37
34

46
37

48
42
rs

0.21
-0.10

0.23
0.10

0.12
0.43a
         aCorrelations represent Spearman's rvalues; italicized values indicate significance at the a = 0.05 level.
         Source: Chang et al. (2000).
 1
 2
 3
 4
 5
 9
10
11
12
13
14
emissions, there was no significant correlation between "commuting PM10" and any of the
substances (Carrer et al., 1998).
     Carrer et al. (1998) present data on the correlations among microenvironmental and
personal PM10 exposures and concentrations and selected environmental chemicals monitored
simultaneously (using methods not described). These chemicals were nitrogen oxides (NOX),
carbon monoxide (CO), and total volatile organic compounds (TVOC),  benzene, toluene, xylene,
and formaldehyde. The Kendall T correlation coefficient was used; only results significant at
p < 0.05 are mentioned here. Significant associations were found only between the following
pairs of substances (T shown in parentheses):  personal PM10 (24 h) and NOX (0.34), CO (0.34),
TVOC (0.18), toluene (0.19), and xylene (0.26); office PM10 and NOX (0.31); home PM10 and
NOX (0.24), CO (0.24), toulene (0.17), and xylene (0.25). Even though  most of these chemicals
are associated with motor vehicular emissions, there was no significant  correlation between
commuting PM10 and any of the substances.
       June 2003
                                          5-90
DRAFT-DO NOT QUOTE OR CITE

-------
 1     5.4   SUMMARY OF PARTICULATE MATTER CONSTITUENT DATA
 2     5.4.1   Introduction
 3          Atmospheric PM contains a number of chemical constituents that may be of significance to
 4     human exposure and health effects. These constituents may be either components of the ambient
 5     particles or bound to the surface of particles. They may be elements, inorganic species, or
 6     organic compounds. A limited number of studies have collected data on concentrations of
 7     elements, acidic aerosols, and PAHs in ambient, personal, and microenvironmental  PM samples.
 8     However, there have not been extensive analyses of the constituents of PM in personal or
 9     microenvironmental samples. Data from relevant studies are summarized in this section.  The
10     summary does not address bacteria, bioaerosols, viruses, or fungi (e.g., Owen et al., 1992; Ren
11     etal., 1999).
12
13     5.4.2   Monitoring Studies That Address Particulate Matter  Constituents
14          Relevant studies published in recent years that have measured the constituents of PM in
15     personal or microenvironmental samples are summarized in Tables 5-11 and 5-12 for personal
16     exposure measurements of PM and microenvironmental samples, respectively. Studies that
17     measured both personal and microenvironmental samples are included in Table 5-11.
18          The largest database on personal, microenvironmental, and outdoor measurements of PM
19     elemental concentrations is the PTEAM study (Ozkaynak et al., 1996b). The results are
20     highlighted in the table and discussed below. The table shows that a number of studies have
21     measured concentrations of elements (by XRF), organic carbon (OC), various indicators of
22     elemental carbon (EC), aerosol acidity, sulfate, ammonia, and nitrate.  Additionally, a number of
23     studies have measured PAHs, both indoors and outdoors. Other than the PAHs, there are few
24     data on organic constituents of PM.
25
26     5.4.3   Key Findings
27     5.4.3.1  Correlations of Personal and Indoor Concentrations with Ambient Concentrations
28             of Particulate Matter Constituents
29          The elemental composition of PM10 in personal samples was measured in the PTEAM
30     study, the first probability-based study of personal exposure to particles.  A number of important
31     observations made from the PTEAM data collected in Riverside, CA, are summarized by

       June 2003                                5-91        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Ozkaynak et al. (1996b). Population-weighted daytime personal exposures averaged
 2      150 ± 9 |ig/m3 compared to concurrent indoor and outdoor concentrations of 95 ± 6 |ig/m3.  The
 3      personal exposure measurements suggested that there was a "personal cloud" of particles
 4      associated with personal activities. Daytime personal exposures to 14 of the 15 elements
 5      measured in the samples were considerably greater than concurrent indoor or outdoor
 6      concentrations; sulfur was the only exception.
 7          The PTEAM data also showed good agreement between the concentrations of the elements
 8      measured outdoors in the backyard of the residences with the concentrations measured at the
 9      central site in the community. The agreement was excellent for sulfur. Although the particle
10      and element mass concentrations were higher in personal samples than for indoor or outdoor
11      samples, a nonlinear mass-balance method showed that the penetration factor was nearly 1 for all
12      particles and elements.
13          Similarly to the PTEAM results, recent measurements of element concentrations in
14      NHEXAS showed elevated concentrations of As and Pb in personal samples relative to indoor
15      and outdoor samples (Clayton et al., 1999b).  The elevated concentrations of As and Pb were
16      consistent with elevated levels of PM50 in personal samples (median particle exposure of
17      101 |ig/m3) as compared to indoor concentrations (34.4 |ig/m3). There was a strong association
18      between personal and indoor concentrations and indoor and outdoor concentrations for both As
19      and Pb. However, there were no central site ambient measurements for comparison to the
20      outdoor or indoor measurements at the residences.
21          Manganese (Mn) concentrations were measured in PM2 5 samples collected in Toronto
22      (Crump, 2000). The mean PM2 5 Mn concentrations were higher outdoors than indoors.
23      However, the outdoor concentrations measured at the participants' homes were lower than those
24      measured at two fixed locations. Crump (2000) suggested that the difference in the
25      concentrations may have been because the  fixed locations were likely closer to high-traffic areas
26      than were the participants'  homes.
27          Studies of acidic aerosols and gases typically measure strong acidity  (H+),  SO4"2, NH+4, and
28      NO"3. The relationship between the concentrations of these ions and the relationship between
29      indoor and outdoor concentrations have been addressed in a number of studies during which
30      personal samples, microenvironmental, and outdoor samples have been collected, as shown in
31      Tables 5-14 and 5-15. Key findings from these studies include the following:

        June 2003                                 5-92        DRAFT-DO NOT QUOTE OR CITE

-------
to
O
o
              TABLE 5-14. STUDIES THAT HAVE MEASURED PARTICULATE MATTER CONSTITUENTS IN PERSONAL
                                                                      EXPOSURE SAMPLES
         PM Constituent    Study Name/Reference    Study Location
                                           Population Size/No, of Samples     Summary of Results
vo
H
6
o

o
H
O
O
H
W
O
O
HH
H
W
         Elements
         As and Pb
         Mn
         Acid Aerosol
         Constituents
         PAHs


         Individual
         Particle Analyses
         by CCSEM

         Trace Elemental
         analyses by XRF

         Elemental
         analysis by
         HR-ICP-MS
PTEAM/Ozkaynak
etal. (1996b)
Riverside, CA
178 adults
NHEXAS/Clayton et al.   EPA Region 5
(1999b)

Pellizzari et al. (1998,     Toronto
1999); Clayton etal.
(1999a); Crump (2000)
Samat et al. (2000)
Baltimore, MD
167 samples


925 personal samples



20 adults
                           Brauer et al. (1989)       Boston, MA
Suhetal. (1992)



Suhetal. (1993a,b)

Suh etal. (1994)

Waldman and Liang
(1993); Waldman etal.
(1990)

Zmirou et al. (2000)


Conner etal. (2001)



Landis etal. (2001)


Kinney et al. (2002)
                                                   Uniontown, PA      24 children for 2 days
                                                   State College, PA    47 children
                                                   Georgia and
                                                   New Jersey
                   Hospital, daycares
Grenoble, France    38 adults
Baltimore, MD
Same study
locations for both
New York City,
NY
3 sets of indoor-outdoor-personal
filters, > 2000 particles/filter
19 day s with P, I, A for
10 elderly retirees

46 student volunteers, 1 week in
summer and winter, PM2 5
Outdoor air was the major source for most elements indoors,
providing 70 to 100% of the observed indoor concentrations for
12 of the 15 elements. Correlation coefficients for central
monitoring site versus outdoor at the residences were 0.98 for sulfur
and 0.5 to 0.9 for other elements (except copper).

Personal As and Pb levels higher than indoor or outdoor levels.
No community ambient site for comparison.

Mean PM2 5 Mn higher outdoors than indoors. But PM2 5 Mn
concentrations higher at two fixed locations than at participants'
homes.

High correlations between personal and ambient sulfate
measurements in summer and winter.

Personal exposures to aerosol strong acidity slightly lower than
concentrations measured at stationary site.

Personal exposures to H+ and SO4"2 lower than outdoor levels, but
higher than indoor microenvironmental levels; personal NH4+ and
NO3" higher than indoor or outdoor levels.

Results similar to Uniontown, PA, study.

Results indicate strong neutralization of acidity indoors.

Indoor sulfate levels were 70 to 100% of outdoor levels. Indoor
ammonia levels 5- to 50-times higher than outdoors. Indoors, acid
aerosols were largely neutralized.

Ambient air concentrations close to traffic emissions were 1.1- to
3.5-times higher than personal exposure concentrations.

CCSEM was used to identify individual particles and classify them
by most likely sources.


P and A highly correlated for PM2 5 and sulfates.  P and A not
significantly correlated for soil and trace element oxides.

29 PM constituents were measured.
Most were similar for P, I and O suggesting Outdoor sources
dominant.

-------
vo
H
6
o

o
H
O
O
H
W
O
O
HH
H
W
                     TABLE 5-15.  STUDIES THAT HAVE MEASURED PARTICULATE MATTER CONSTITUENTS IN
                                                         MICROENVIRONMENTAL SAMPLES
o
o
PM Constituent
Acid Aerosol
Constituents
Study
Name/Reference
Jones et al. (2000)
Study Location
Birmingham,
England
Population Size/No, of Samples
12 residences
Summary of Results
Sulfate I/O ratios ranged

from 0.7 to 0.9 for three PM size fractions.
                         Patterson and Eatough   Lindon, UT
                         (2000)
                         Leaderer et al. (1999)    Virginia and
                                               Connecticut
                         Brauer et al. (1990)      Boston, MA
One school
232 homes
11 homes
Ambient sulfate, SO2, nitrate, soot, and total particle number
showed strong correlations with indoor exposure although ambient
PM25 mass was not a good indicator of total PM25 exposure.

The regional ambient air monitoring site provided a reasonable
estimate of indoor and outdoor sulfate at nonsmokers homes.  I/O
sulfate ratio of 0. 74 during summer. Ammonia concentrations were
an order of magnitude higher indoors than outdoors. Nitrous acid
levels higher indoors than outdoors.

Outdoor levels of H+, SO2, HNO3, and SO4'2 exceeded indoor levels
in winter and summer. I/O ratios of H+ lower than I/O ratios of SO4"
2 indicated neutralization of the acidity by ammonia.
PAHs


PAHs and
phthalates
Chuangetal. (1999)
Dubowsky et al.
(1999)
Sheldon et al.
(1993a,b)
PTEAM/Ozkaynak
etal. (1996b),
Sheldon etal. (1993c)
Durham, NC
Boston, MA
Placerville and
Roseville, CA
Riverside, CA
24 homes
3 buildings
280 homes
120 homes
Measurements with continuous monitor; PAH levels generally
higher indoors than outdoors.
PAHs indoors attributable to traffic, cooking, and candle-burning.
Mass balance model used to estimate source strengths for PAH
sources such as smoking, wood-burning and cooking.
12-h I/O ratios for particulate-phase PAHs ranged from 1 . 1 to 1 .4
during the day and 0.64 to 0.85 during night. The concentrations of
phthalates and the number of samples with detectable phthalates
were higher indoors than outdoors.

-------
 1       •  Acid aerosol concentrations measured at the residences in the Uniontown, PA, study were
           significantly different from those measured at a fixed ambient site located 16 km from the
           community. However, Leaderer et al. (1999) reported that the regional ambient air
           monitoring site in Vinton, VA, provided a reasonable estimate of indoor and outdoor
           sulfate measurements during the summer at homes without tobacco combustion.
 2       •  Approximately 75% of the fine aerosol indoors during the summer was associated with
           outdoor sources based on I/O sulfate ratios measured in the Leaderer et al. (1999) study.
 3       •  Personal exposures to strong acidity (H+) were lower than corresponding outdoor levels
           measured in studies by Brauer et al. (1989, 1990) and Suh et al. (1992).  However, the
           personal exposure levels measured by Suh et al. (1992) were higher than the indoor
           microenvironmental levels.
 4       •  Personal exposures to NH4+ and NO3" were reported by Suh et al.  (1992) to be lower than
           either indoor or outdoor levels.
 5           Personal exposures to SO4"2 were also lower than corresponding outdoor levels, but higher
 6      than the indoor microenvironmental levels (Suh et  al., 1992; 1993a,b) as shown in Table 5-16.
 7           The fact that the personal and indoor H+ concentrations were substantially lower than
 8      outdoor concentrations suggests that a large fraction of aerosol strong acidity is neutralized by
 9      ammonia.  Ammonia is emitted in relatively high concentrations in exhaled breath and sweat.
10      The difference between indoor and outdoor H+ concentrations in the Suh et al. (1992, 1993a,b)
11      studies was also much higher than the difference for indoor and outdoor SO4"2, indicative of
12      neutralization of the H+. Results of the Suh  et al. (1992, 1993a,b) studies also showed
13      substantial interpersonal variability of H+ concentrations that could not be explained by variation
14      in outdoor concentrations.
15           Similar results for ammonia were reported by Waldman and Liang (1993).  They reported
16      that levels of ammonia in monitored institutional settings were 10- to 50-times higher than
17      outdoors and that acid aerosols were largely neutralized. Leaderer et  al. (1999) reported that
18      ammonia concentrations during both winter and summer in residences were an order of
19      magnitude higher indoors than outdoors, consistent with results of other studies and the presence
20      of sources of ammonia indoors.
21           Sulfate aerosols appear to penetrate indoors effectively. Waldman et al. (1990) reported
22      I/O ratios of 0.7 to 0.9 in two nursing care facilities and a day-care center. Sulfate I/O ratios

        June 2003                                 5-95         DRAFT-DO NOT QUOTE OR CITE

-------
              TABLE 5-16.  SUMMARY STATISTICS FOR PERSONAL, INDOOR, AND
          OUTDOOR CONCENTRATIONS OF SELECTED AEROSOL COMPONENTS IN
                              TWO PENNSYLVANIA COMMUNITIES
Concentration (nmol m"3)
Aerosol
State College
NO3

so4-2

NH4+
H+


Uniontown
so4-2
NH4+
H+
Home Type

A/C Homes0
Non-A/C
A/C Homes
Non-A/C
All Homes'1
All Homes
A/C Homes
Non-A/C
All Homes6

All Homes6
All Homes6
All Homes6
Sample Site
(In/Out)a

53/71
254/71
56/75
259/75
214/76
314/155
28/74
230/74
163/75

91/46
91/44
91/46
Indoor (12 h)
GM ± GSDb

2.1 ±2.7
3.2 ±2.3
61. 8 ±2.5
96.7 ±2.5
69.1 ±2.6
154.7 ±2.8
4.2 ±4.3
11.2±3.1
9.1 ±3.5

87.8 ±2.1
157.2 ±2.8
13.7 ±2.5
Outdoor (24 h) Personal (12 h)
GM ± GSDb GM ± GSDb

1.4 ±2.1 —
1.4 ±2.1 —
109.4 ±2.4 —
109.4 ±2.4 —
91.0 ±2.5 71.5 ±2.4
104.4 ±2.3 —
82.5 ±2.6
82.5 ±2.6
72.4 ±2.9 18.4 ±3.0

124.9 ±1.9 110.3 ±1.8
139.4 ±2.1 167.0 ±2.0
76.6 ±2.7 42.8 ±2.2
        aIn/Out = Indoor sample site/outdoor sample site.
        bGM ± GSD = Geometric mean ± geometric standard deviation.
        °A/C Homes = Homes that had air-conditioning (A/C); this does not imply that it was on during the entire
        sampling period.
        Non-A/C = Homes without air conditioning.
        dThe sample size (n) for the personal monitoring = 209.
        6n = 174 for personal monitoring.
        Source:  Suhetal. (1992, 1993a,b).
1      were measured for three particle size fractions in 12 residences in Birmingham, England, by
2      Jones et al. (2000). The sulfate I/O ratios were 0.7 to 0.9 for PM< 1.1 |im, 0.6 to 0.8 for PM
3      1.1 to 2.1 jim, and 0.7 to 0.8 for PM 2.1 to 10 |im.  Suhetal. (1993b)  reported that personal and
4      outdoor sulfate concentrations were highly correlated as depicted in Figure 5-13.
5           Indoor/outdoor relationships for a number of PM25 components  and related species in
6      Lindon, UT, during January and February of 1997 by Patterson and Eatough (2000).  Outdoor
7      samples were collected at the Utah State Air Quality monitoring site.  Indoor samples were
8      collected in the adjacent Lindon Elementary School.  The infiltration factors, Cai/C, given by the
       June 2003
5-96
DRAFT-DO NOT QUOTE OR CITE

-------
                      600
                                                                             600
                                    Outdoor Sulfate (nmoles/m )
      Figure 5-13.  Personal versus outdoor SO4 2 in State College, PA. Open circles represent
                    children living in air conditioned homes; the solid line is the 1:1 line.
      Source: Suhetal. (1993b).
1     slope of the regression lines (Table 5-17), were low (0.27 for sulfate and 0.12 for PM2 5) possibly
2     because of removal of particles in the air heating and ventilation system. The authors concluded
3     that indoor PM2 5 mass may not always be a good indicator of exposure to ambient combustion
4     material due to the influence of indoor particle sources. Presumably this occurs because the
5     concentrations of indoor-generated particles are not well correlated with the concentrations of
6     ambient combustion particles.  However, ambient sulfate, SO2, nitrate, soot, and total particulate
7     number displayed strong correlations with indoor exposure, presumably because these species
8     have few indoor sources in the absence of indoor combustion. Ambient PM2 5 mass was not a
      June 2003
5-97
DRAFT-DO NOT QUOTE OR CITE

-------
             TABLE 5-17. REGRESSION ANALYSIS OF INDOOR VERSUS OUTDOOR
          CONCENTRATIONS (units are nmol m3, except for soot and metals, which are ug/m3
                                and absorption units m3, respectively.)"
Species
SO2 All Samples
SO2 Day Samples
SO2 Night Samples
Sulfate All Samples
Sulfate Day Samples
Sulfate Night Samples
Nitrate All Samples
Nitrate Day Samples
Nitrate Night Samples
Soot Day Samples
Soot Night Samples
Total Acidity All Samples
Metals All Samples
Slope = FMF
0.0272 ± 0.0023
0.0233 ±0.0037
0.0297 ± 0.0029
0.267 ± 0.024
0.261 ±0.034
0.282 ±0.035
0.0639 ±0.0096
0.097 ±0.0096
0.047 ±0.011
0.43 ±0.25
0.33±0.13
0.04 ±0.73
0.10 ±0.30
Intercept = C;g
0.34±0.13
0.75 ±0.26
0.099 ±0.075
-0.14 ±0.48
0.40 ±0.66
-0.84 ±0.68
0.9 ±1.5
-0.4 ±1.4
1.5±1.8
3.5±1.7
0.00 ±0.55
0.42 ±0.23
0.0014 ±0.0042
R2
0.73
0.62
0.82
0.70
0.71
0.70
0.54
0.88
0.44
0.43
0.69
0.00
0.01
Average
Outdoors, C0
38
56
20
16
16
16
134
126
139
6
4
0.2
0.0042
        aLindon Elementary School, Lindon, UT, January and February 1997.
        Source: Patterson and Eatough (2000).
 1     good indicator of indoor PM25 mass exposure, presumably due to uncorrelated indoor sources of
 2     PM2 5 mass.
 3          Oglesby et al. (2000) conducted a study to evaluate the validity of fixed-site fine particle
 4     concentration measurements as exposure surrogates for air pollution epidemiology. Using 48-h
 5     EXPOLIS data from Basel, Switzerland, they investigated the personal exposure/outdoor
 6     concentration relationships for four indicator groups: (1) PM25 mass, (2) sulfur and potassium
 7     for regional air pollution, (3) lead and bromine for traffic-related particles, and (4) calcium for
 8     crustal particles. The authors reported that personal exposures to PM2 5 mass were not correlated
 9     to corresponding home outdoor levels (n = 44, r = 0.07). In the study group reporting neither
10     relevant indoor sources nor relevant activities, personal exposures and home outdoor levels of
       June 2003
5-98
DRAFT-DO NOT QUOTE OR CITE

-------
 1      sulfur were highly correlated (n = 40, r = 0.85). These results are consistent with spatially
 2      homogeneous regional pollution and higher spatial variability of traffic and crustal materials.
 3           Indoor, outdoor, and personal concentrations of PM25 mass and a variety of PM
 4      constituents were measured at an 18-story retirement facility in Towson, MD for 28, 24-hour
 5      monitoring periods during July and August 1998 (Landis et al., 2001).  Indoor and outdoor
 6      measurements were made with a Versatile Air Pollutant Sampler (VAPS).  Indoor, outdoor, and
 7      personal samples were made with a Personal Exposure Monitor (PEM). The VAPS
 8      (a dichotomous sampler) collected PM2 5 (15 1 min"1) and PM10_2 5 (2 1 min"1) while the PEM
 9      collected PM25 (2 1 min"1). A comparison of the VAPS and the PEM indicated that the indoor
10      PEM collected much higher mass and more soil components than the indoor VAPS although the
11      differences between outdoor results were smaller and not significant. These differences were
12      attributed to there being a larger coarse particle concentration inside (and perhaps larger
13      diameter particles) and either more particle bounce or a higher 50% cut point for the PEM.
14      In their analysis, Landis et al. (2001) compared indoor and outdoor VAPS data and outdoor and
15      personal PEM data.
16           As shown in Table 5-18 (PEM) and Table 5-19 (VAPS), higher correlations were found for
17      fine-particle components of PM25 and lower correlations for coarse-particle components. Like
18      Patterson and Eatough (2000), Landis et al. (2001) found low infiltration factors for nitrate but a
19      reasonable correlation suggesting that fine-mode ammonium  nitrate may be evaporating after it
20      penetrates indoors. Neither sulfate nor nitrate had indoor sources.
21           Indoor and outdoor PM2 5 and PM10 mass and chemical composition were measured in
22      13 homes (2-4 days for each home) in the Coachella Valley, a unique desert area in southern
23      California during the winter and spring of 2000 (Geller et al., 2002). Maximum infiltration of
24      ambient PM would be expected during this period because the mild climate minimizes the use of
25      heating or air conditioning. Regression analysis was used to estimate FINF and Cig.  Results are
26      shown in Table 5-20.  The Coachella PM is generally considered to be rich in coarse PM and
27      epidemiological studies  have associated PM10 (Ostro et al., 1999)  and estimated PM10_25 (Ostro
28      et al., 2000) with mortality. However, the results of Geller et al. (2002) indicate that even during
29      periods of high air exchange rates indoor exposures would be dominated by PM25.  Geller et al.
30      (2002) also report results for some chemical components of PM25 and PM10_25 and EC and OC in
31      PM25. For EC, ^^ = 0.74, in good agreement with F^ = 0.74. Indoor concentrations of OC

        June 2003                                 5-99        DRAFT-DO NOT QUOTE OR CITE

-------
    TABLE 5-18.  MIXED MODEL ANALYSIS OF PERSONAL VERSUS OUTDOOR
                                   CONCENTRATIONS3

PM25 (ug/m3)
Sulfateb
(ug/m3)
Soil0 (ng/m3)
TEOd (ng/m3)
P?
0.82
0.95
0.03
0.39
a
0.46 ± 0.04
0.40 ± 0.02
0.05 ±0.12
0.43 ±0.11
C
21.0
10.2
458
165
A
9.66
4.08
93
71
N
3.57 ±0.80
0.1 ±0.04
544 ± 95
99 ±23
T
12.8
4.5
553
170
A+N
13.2
4.2
637
170
 a a (slope) and N (intercept) taken from mixed model results. A calculated as a • C; T is measured value.
   Note good agreement between measured T and T estimated from A + N except for soil.
 b as (NH4)2SO4.
 0 Soil = sum of oxides Si, Ca, Fe, and Ti (due to coarse PM in PM25).
 d TEO = trace element oxides (mostly fine PM components).

 Source: Landis et al. (2001).
      TABLE 5-19. REGRESSION ANALYSIS OF INDOOR VERSUS OUTDOOR
                                   CONCENTRATIONS3

PM2 5 (ug/m3)
Sulfateb (ug/m3)
OC« 1.4
(ug/m3)
ECC (ug/m3)
NaCl (ng/m3)
Soild (ng/m3)
TEOe (ng/m3)
Nitrate (ng/m3)
R2
0.74
0.98
0.44
0.58
0.32
0.30
0.38
0.83
MNF
0.35
0.41
0.30
0.32
0.06
0.07
0.35
0.09
C
18.9
10.4
5.4
0.50
231
363
94
372
Q
6.6
4.4
1.6
0.16
14
25
33
33
clg
0.32
-0.24
8.0
0.09
37
51
7
12
Q
6.7
4.0
9.7
0.4
48
74
39
68
c* + clg
6.9
4.2
9.6
0.25
53
76
40
45
 a F^p (slope) and Clg (intercept) taken from regression equation. Cm calculated as C • Fmf; Q is measured value.
   Note reasonable agreement between measured Q and Q estimated from C^ + Clg except for EC and nitrate.
 b as (NH4)2S04.
 0 revised regression with outliers omitted.
 d Soil = sum of oxides Si, Ca, Fe, and Ti (due to coarse PM in PM25).
 e TEO = trace element oxides (mostly fine PM components).

 Source: Landis et al. (2001).
June 2003                                  5-100        DRAFT-DO NOT QUOTE OR CITE

-------
             TABLE 5-20. REGRESSION ANALYSIS OF INDOOR VERSUS OUTDOOR
                                        CONCENTRATIONS3
                         R2          FINF          C           Q          Clg           Q
        Fine            0.37          0.74         15.0         11.1          4.3          15.4
        Coarse	035	0.30	8.6	2.6	3.0	5.6
        a FINF (slope) and Clg (intercept) taken from regression equation. Cm calculatedjis C • F^p. Q is measured value.
         Note excellent agreement between measured Q and Q estimated from C^ + Clg.
        Source: Geller et al. (2002).
 1     were much higher than outdoor concentrations; the average indoor to outdoor ratio was 1.77.
 2     PAHs have been measured in studies by EPA and the California Air Resources Board. PAH
 3     results from a probability sample of 125 homes in Riverside are discussed in reports by Sheldon
 4     et al. (1992a,b) and Ozkaynak et al. (1996b). Data for two sequential 12-h samples were
 5     reported for PAHs by ring size (3 to 7) and for individual phthalates. The results scan be
 6     summarized as follows.
 7         •  The particulate-phase 5- to 7-ring species had lower relative concentrations than the more
              volatile 3- to 4-ring species.
 8         •  The 12-h indoor/outdoor ratios for the 5- to 7-ring species ranged from 1.1 to 1.4 during
              the day  and from 0.64 to 0.85 during the night (Sheldon et al., 1993a).
 9         •  An indoor air model used to calculate indoor "source strengths" for the PAHs showed that
              smoking had the strongest effect on indoor concentrations.
10          Results from a larger PAH probability study in 280 homes in Placerville and Roseville, CA
11     (Sheldon et al., 1993a,b) were similar to the 125-home study.  The higher-ring, particle-bound
12     PAHs had lower indoor and outdoor concentrations than the lower-ring species. For most PAHs,
13     the I/O ratio was greater than 1 for smoking and smoking/fireplace homes and less than 1 for
14     fireplace-only, wood stove, wood stove/gas heat, gas heat, and "no source" homes.
15          A study of PAHs in indoor and outdoor air was conducted in 14 inner-city and 10 rural
16     low-income homes near Durham, NC, in two seasons (winter and summer) in 1995 (Chuang
17     et al., 1999). Fine-particle-bound PAH concentrations measured with a real-time monitor were
18     usually higher indoors than outdoors (2.47 ± 1.90 versus 0.53 ± 0.58 |ig/m3). Higher indoor

       June 2003                                5-101        DRAFT-DO NOT QUOTE OR CITE

-------
 1     levels were seen in smokers' homes compared with nonsmokers' homes, and higher outdoor and
 2     indoor PAH levels were seen in urban areas compared with rural areas.
 3          In a study reported by Dubowsky et al. (1999), the weekday indoor PAH concentrations
 4     attributable to traffic (indoor source contributions were removed) were 39 ± 25 ng/m3 in a
 5     dormitory that had a high air exchange rate because of open windows and doors; 26 ± 25 ng/m3
 6     in an apartment; and 9 ± 6 ng/m3 in a suburban home. The study showed that both indoor and
 7     outdoor sources (especially motor vehicular traffic) contributed to indoor PAH concentrations.
 8     BaP concentrations were measured in the THEES study (Waldman et al., 1991).
 9     A comprehensive analysis of the data showed considerable  seasonal variability of indoor and
10     outdoor sources and resultant changes in personal exposures to BaP.
11          The indoor and outdoor concentrations of 30 PAHs were measured in 55 nonsmoking
12     residences in Los Angeles, CA; Houston, TX; and Elizabeth, NJ (Naumova et al., 2002). A
13     comparison of indoor/outdoor ratios of low molecular weight PAHs (3-4 rings) and higher
14     molecular weight PAHs (5-7 rings) indicated that indoor sources had a significant effect on
15     indoor concentrations of 3-ring PAHs and a smaller effect on 4-ring PAHs while outdoor sources
16     dominated the  indoor concentrations of 5-7 ring PAHs.
17
18     5.4.4   Factors Affecting Correlations Between Ambient Measurements
19             and Personal or Microenvironmental Measurements  of
20             Particulate Matter Constituents
21          The primary factors affecting correlations between personal exposure and ambient air PM
22     measurements  have been discussed in Section 4.3.2.  These include air exchange rates, particle
23     penetration factors, decay  rates, removal mechanisms, indoor air chemistry, indoor sources, and
24     freshly-generated particles indoors. The importance of these factors varies for different PM
25     constituents. For acid aerosols, indoor air chemistry is particularly important as indicated by the
26     discussion of the neutralization of the acidity by ammonia which is present at higher
27     concentrations indoors because of the presence of indoor sources. For  SVOCs, including PAHs
28     and phthalates, the presence of indoor sources will substantially affect the correlation between
29     indoor and ambient concentrations (Ozkaynak et al., 1996b; Sheldon et al., 1993b).  Penetration
30     factors for PM will affect correlations between indoors and outdoors for most elements, except
31     Pb, which may have significant indoor sources in older homes. Indoor air chemistry, decay
32     rates, and removal mechanisms may affect soot and organic carbon. Furthermore, reactions

       June 2003                               5-102        DRAFT-DO NOT QUOTE OR CITE

-------
 1      between indoor and outdoor gases and particles may produce freshly generated aerosols indoors.
 2      These factors must be fully evaluated when attempting to correlate concentrations of PM
 3      constituents in ambient, personal, or indoor samplers.
 4
 5      5.4.5   Limitations of Available Data
 6           The previous discussion demonstrates that there is limited data available that can be used
 7      to compare personal, microenvironmental, and ambient air concentrations of PM constituents.
 8      Because of resource limitations, PM constituents have not been measured in many studies of PM
 9      exposure.  There are little data on freshly generated aerosols indoors.  Although there are some
10      data on acid aerosols, the comparisons between the personal and indoor data generally have been
11      with outdoor measurements at the participants' residences not with community ambient air
12      measurement sites. The relationship between personal exposure and indoor levels of acid
13      aerosols is not clear because of the limited database. The exception is sulfate, for which there
14      appears to be a strong correlation between indoor and ambient concentrations.
15           With the exception of PAHs, there are practically no data available with which to relate
16      personal or indoor concentrations with outdoor or ambient site concentrations of SVOCs that
17      may be generated from a variety  of combustion and industrial sources. The relationship between
18      exposure and ambient concentrations of particles from specific sources, such  as diesel engines,
19      has not been determined.
20           Although there is an increasing amount of research being performed to measure PM
21      constituents in different PM size  fractions, the current data are inadequate to adequately assess
22      the relationship between indoor and ambient concentrations of most PM constituents. Additional
23      information is also needed on PM exposures that result from outdoor vapors reacting with indoor
24      vapors.  This is a source that could also vary with outdoor PM, for example, when the outdoor
25      vapor is ozone.
26
27
28
29
        June 2003                                 5-103       DRAFT-DO NOT QUOTE OR CITE

-------
 1     5.5   IMPLICATIONS OF USING AMBIENT PARTICULATE MATTER
 2           CONCENTRATIONS IN TOXICOLOGICAL AND
 3           EPIDEMIOLOGICAL STUDIES OF PARTICULATE MATTER
 4           HEALTH EFFECTS
 5     5.5.1   Toxicology
 6          Most studies of PM toxicity have used either pure chemicals or ambient PM. Indoor-
 7     generated PM differs somewhat from ambient PM in terms of sources, size, and composition.
 8     It is possible, therefore, that indoor-generated PM could have different toxicological properties.
 9     Tobacco smoke has been studied extensively; however, there is little toxicological information
10     on PM from other indoor sources. Long et al. (2001b) have assessed the in vitro toxicity of
11     14 paired indoor and outdoor PM2 5 samples collected in 9 Boston-area homes. Bioasseys were
12     conducted using rat alveolar macrophages (AMs) and tumor necrosis factor (TNF) was measured
13     to assess particle-induced proinflammatory responses.  TNF  production was found to be
14     significantly higher in AMs exposed to indoor PM than to those exposed to outdoor PM. This
15     result held even after normalization for endotoxin concentrations which were higher in indoor
16     samples. The authors feel that their results "suggest that indoor-generated particles may be more
17     bioactive than ambient particles." PM, in its various forms, produces many types of biological
18     effects.  It seems possible that indoor PM could be more active than ambient PM for some
19     effects and less for others.
20
21     5.5.2   Potential Sources of Error Resulting from Using Ambient Particulate
22             Matter Concentrations in Epidemiological Analyses
23          In this section, the exposure issues that relate to the interpretation of the findings from
24     epidemiological studies of PM health effects are examined.  This section examines the errors that
25     may be associated with using ambient PM concentrations in  epidemiological  analyses of PM
26     health effects. First, implications of associations found between personal exposure and ambient
27     PM concentrations are reviewed. This is discussed separately in the context of either community
28     time-series studies or long-term, cross-sectional studies of chronic effects. Next, the role of
29     compositional and spatial differences in PM concentrations are discussed and how these may
30     influence the interpretation of findings from PM epidemiology. Finally, using statistical
31     methods, an evaluation of the influence of exposure measurement errors on PM epidemiological
32     studies is presented.

       June 2003                               5-104       DRAFT-DO NOT QUOTE OR CITE

-------
 1           Measurement studies of personal exposures to PM are still few and limited in spatial,
 2      temporal, and demographic coverage. Consequently, with the exception of a few longitudinal
 3      panel studies, most epidemiological studies of PM health effects rely on ambient community
 4      monitoring data giving 24-h average PM concentration measurements.  Moreover, because of
 5      limited sampling for PM2 5, many of these epidemiological studies had to use available PM10 or in
 6      some instances had to rely on historic data on other PM measures or indicators, such as TSP,
 7      SO4"2, IP15 (inhalable particles with an upper cut of 15 jim), RSP (respirable particles), COH
 8      (coefficient of haze), etc. A critical question often raised in the interpretation of results from
 9      acute or chronic epidemiological community-based studies of PM is whether the use of ambient
10      stationary site PM concentration data influences or biases the findings from these studies.
11           If it is assumed that total personal PM exposure is responsible for observed effects, use of
12      ambient concentrations could lead to misclassification of individual exposures and to errors in
13      the epidemiological analysis  of pollution and health data depending on the pollutant and  on the
14      mobility and lifestyles of the population studied.  Ambient monitoring stations can be some
15      distance away from the individuals and can represent only a fraction of all likely outdoor
16      microenvironments that individuals come in contact with during the course of their daily lives.
17      Furthermore, most individuals are quite mobile and move through multiple microenvironments
18      (e.g., home, school, office,  commuting, shopping, etc.) and engage in diverse  personal activities
19      at home (e.g., cooking, gardening, cleaning,  smoking).  Some of these microenvironments and
20      activities may have different  sources of PM and result in distinctly different concentrations of
21      PM than that monitored by the fixed-site ambient monitors.  Consequently, exposures of some
22      individuals could be classified incorrectly if only ambient monitoring data are used to estimate
23      individual level total personal PM exposures. Thus, improper assessment of exposures using
24      data routinely collected by the neighborhood monitoring stations could conceivably lead to a
25      bias or  increase in the standard error in epidemiological analysis.  Except for extremely unlikely
26      situations, however, the bias would be expected to reduce the estimated health risk coefficient.
27           Because many individuals are typically exposed to particles in a multitude of indoor and
28      outdoor microenvironments during the course of a day, concern about possible error introduced
29      in the estimation of PM risk coefficients using ambient, as opposed to personal, PM
30      measurements has received considerable attention recently from exposure analysts,
31      epidemiologists,  and biostatisticians. Some exposure analysts contend that, for community time-

        June  2003                                5-105        DRAFT-DO NOT QUOTE OR CITE

-------
 1      series epidemiology to yield information on the statistical association of a pollutant with a health
 2      response, there must be an association between personal exposure to a pollutant and the ambient
 3      concentration of that pollutant because people tend to spend around 90% time indoors and are
 4      exposed to both indoor-generated and ambient-infiltrated PM (cf. Wallace, 2000b; Brown and
 5      Paxton, 1998; Ebelt et al., 2000).  Consequently, numerous findings reported in the
 6      epidemiological literature on significant associations between ambient PM concentrations and
 7      various morbidity and mortality health indices, in spite of the low correlations between ambient
 8      PM and concentrations and measures of personal exposure, have been described by some
 9      exposure analysts as an exposure paradox (Lachenmyer and Hidy, 2000; Wilson et al., 2000).
10           To resolve the so-called exposure paradox, several types of analyses need to be considered.
11      The first type of analysis has to examine the correlations between ambient PM concentrations
12      and personal exposures that are relevant to most of the existing PM epidemiological studies
13      using either pooled, daily-average, or longitudinal  exposure data. The second approach has to
14      study the degree of correlations between the two key components of personal PM exposures (i.e.,
15      exposures caused by ambient PM and exposures caused by nonambient PM) with ambient or
16      outdoor PM concentrations, for each of the three types of exposure study designs.  Yet, even
17      with these two approaches, it may still be difficult to examine complex synergisms which, in
18      some situations, may preclude simple decoupling of indoor and outdoor particles either in terms
19      of exposure or total dose delivered to the lung. In  addition, several factors that influence either
20      the exposure or health response characterization of the subjects have to be addressed.  These
21      include such factors as
22        •  spatial variability of PM components,
23        •  health or sensitivity status of subjects,
24        •  variations of PM with other co-pollutants,
25        •  co-generation of fine  and ultrafme particles from outdoor air and indoor gaseous
             pollutants,
26        •  formal evaluation of exposure errors in the analysis of health data, and
27        •  how the results may depend on the variations in the design of the epidemiological study.
28           To facilitate the discussion of these topics, a  brief review of concepts pertinent to exposure
29      analysis issues in epidemiology is presented.

        June 2003                                5-106        DRAFT-DO NOT QUOTE OR CITE

-------
 1      5.5.2.1  Associations Between Personal Exposures and Ambient Particulate Matter
 2              Concentrations
 3           As defined in Sections 5.3 and 5.4, personal exposures to PM result from an individual's
 4      exposures to PM in many different types of microenvironments (e.g., outdoors near home,
 5      outdoors away from home, indoors at home, indoors at office or school, commuting, restaurants,
 6      malls, other public places, etc.). Total personal exposures that occur in these indoor and outdoor
 7      microenvironments can be classified as those resulting from ambient PM (ambient PM exposure
 8      includes exposure to ambient PM while outdoors and exposure while indoors to ambient PM that
 9      has infiltrated indoors) and those primarily generated by indoor sources,  indoor reaction, and
10      personal activities (nonambient PM exposure). The associations between personal exposures
11      and ambient PM concentrations that have been reported from various personal exposure
12      monitoring studies under three broad categories of study design, (1) longitudinal, (2) daily -
13      average, or (3) pooled exposure studies, are summarized below.
14           In Sections 5.4.3.1.2 and 5.4.3.1.3, recent studies mainly conducted in the United States
15      and involving children, the elderly, and subjects with COPD were reviewed, and they indicated
16      that both intra- and inter-individual variability in the relationships between personal exposures
17      and ambient PM concentrations were observed. A variety of different physical, chemical, and
18      personal or behavioral factors were identified by the original investigators that seem to influence
19      the magnitude and the strength of the associations reported.
20           For cohort studies in which individual daily health responses are obtained, individual
21      longitudinal PM personal exposure data (including ambient and nonambient components) may
22      provide the appropriate indicators.  In this case, health responses of each individual can be
23      associated with the  total personal exposure, the ambient exposure, or the nonambient exposure of
24      each individual. Additionally, the relationships of personal exposure indicators with ambient
25      concentration can be investigated.  In the case of community time-series  epidemiology, however,
26      it is not feasible to obtain experimental measurements of personal  exposure for the millions of
27      people over time periods of years that are needed to investigate the relationship between air
28      pollution and infrequent health responses such as deaths or even hospital admissions.  The
29      epidemiologist must work with the aggregate number of health responses occurring each day and
30      a measure of the ambient concentration that is presumed to be representative of the entire
31      community.  The relationship of PM exposures of the potentially susceptible groups to
32      monitored ambient  PM concentrations depends on their activity pattern and level, residential

        June 2003                                5-107       DRAFT-DO  NOT QUOTE OR CITE

-------
 1      building and HVAC factors (which influence the infiltration factor), status of exposure to ETS,
 2      amount of cooking or cleaning indoors, and seasonal factors, among others.  For these special
 3      subgroups, average personal exposures to ambient PM are correlated well with ambient PM
 4      concentrations regardless of individual variation in the absence of major microenvironmental
 5      sources.
 6           Even though both ambient and nonambient PM exposure contribute to daily baseline PM
 7      dose received by the lung, there seem to be clear differences in the relationships  of ambient and
 8      nonambient PM exposure with ambient PM concentration.  Various researchers have shown that
 9      nonambient PM exposure is independent of ambient PM concentration, but that ambient PM
10      exposure is a function of ambient PM concentration. Wilson et al. (2000) explain the difference
11      based on different temporal patterns that affect PM concentrations. "Concentrations of ambient
12      PM are driven by meteorology and by changes in the emission rates and locations of emission
13      sources, while concentrations of nonambient PM are driven by the daily activities of people."
14      Still, although nonambient PM exposure may not correlate with ambient PM concentration or
15      ambient PM exposure, it will nevertheless add to the daily baseline dose received by the lung.
16      An important  concern, for which there is little information, is the relative biological activity of
17      ambient and indoor generated PM both in terms  of the type of toxic effect and the relative
18      potency for a given effect.
19           Ott et al. (2000) also discuss the reasons for assuming that nonambient PM exposure is
20      independent of ambient PM exposure and ambient PM concentration.  They show that the
21      nonambient component of total personal exposure is uncorrelated with the outdoor concentration
22      data. Ott et al. (2000) show the ambient PM exposure is similar for three population-based
23      exposure studies:  two large probability-based studies (the PTEAM study  conducted in Riverside
24      [Clayton et al., 1993;  Thomas et al.,  1993; Ozkaynak et al., 1996a,b] and a study in Toronto
25      [Pelizzarri et al., 1999; Clayton et al., 1999a]), and a nonprobability-based study conducted in
26      Phillipsburg (Lioy et al., 1990). Based on these  three studies, they conclude that the average
27      nonambient PM exposure and the distribution of individual, daily values of nonambient PM
28      exposure can be treated as constant from city to city.
29           Dominici et al. (2000) examined a larger database consisting of five different PM exposure
30      studies and concluded that nonambient PM exposure can be treated as  relatively  constant from
31      city to city although their data show greater variability than the data reported by  Ott (2000).  The

        June 2003                                 5-108        DRAFT-DO NOT QUOTE OR CITE

-------
 1      constancy of nonambient PM exposure remains an open question. If daily values of nonambient
 2      PM exposure were constant, this would imply a zero correlation with ambient PM concentration.
 3      However, this hypothesis of constant individual, daily nonambient PM exposure has not been
 4      established fully because only a few studies have obtained the data needed to estimate daily,
 5      individual values of nonambient PM exposure.  Although nonambient PM exposure is
 6      independent of ambient PM concentration, it may not be independent of the attenuation factor
 7      (ambient PM exposure/ambient PM concentration).  Sarnat et al. (2000) show that nonambient
 8      PM exposure goes up as the ventilation rate (and attenuation factor) goes down. By comparing
 9      winter and summer regression equations Lachenmeyer and Hidy (2000) also show that as the
10      slope, which gives the attenuation factor, decreases, the intercept, which gives the nonambient
11      PM exposure, increases.
12           Mage et al. (1999) assume that the PM10 concentration component from  indoor sources
13      (such as smoking, cooking, cleaning, burning candles, and so on) is not correlated with the
14      outdoor concentration.  They indicate that this lack of correlation is expected because people are
15      unaware of ambient concentrations and do not necessarily change their smoking or cooking
16      activities as outdoor PM10 concentrations vary, an assumption supported by  other empirical
17      analyses of personal exposure data. For the PTEAM data set, Mage et al. (1999) have shown
18      that individual, daily exposures to indoor-generated PM and daily ambient PM concentrations
19      have a correlation coefficient near zero (R2 = 0.005).  Wilson et al. (2000) have shown that
20      individual, daily values of concentration of ambient PM which has infiltrated indoors and
21      indoor-generated PM concentrations also have a near zero correlation  (R2 = 0.03). Figure 5-14
22      shows the relationship of estimated values of nonambient PM exposure with ambient PM
23      concentrations (calculated by EPA, daily,  individual values from PTEAM and daily average
24      values for the  cohort from THEES).
25           Based on these results it is reasonable to assume that ordinarily nonambient PM exposure
26      will have little correlation with ambient PM concentration. A possible exception could be
27      caused by indoor-reaction PM, PM formed when an ambient gas infiltrates indoors and reacts
28      with an indoor-generated gas, i.e., the reaction  of O3 with terpenes from air cleaners. However,
29      ambient O3 does not appear to be highly correlated with ambient or personal PM2 5 (See
30      Table 5-12, Sarnat et al., [2000]). Additionally, not every home will use air fresheners or have
31      the same level of terpene emissions. Hence, indoor-reaction PM concentrations would not be

        June 2003                                5-109        DRAFT-DO NOT QUOTE OR CITE

-------
CD
O
C/5
d)

^ E
c ^^
Q) ?
                       150-
                       100-
                    o  50-
                  CD Q)
                 | §
                 I a
                  c x
                 — LLJ
CD
Q
CD
c
o
(/)
Q) ^^

E ^

o o
                  0) «-
                  O) =*
                  CO g
                  0) Q.
                  > X
                 < LLJ
                  CD
                 Q
                       -50
N = 39.6+0.086C *
r = 0.051
R2 = 0.0026
#=147
                                  * *
                                                                              \Q
                                                        *   *     *
                                               ***
                                     50
                              100
                             150
200
250
                        60
                        50-
                        40-
                        4U
       20'
                             N = 41.5-0.0556C
                             r = 0.291
                             R2 = 0.0846
                             #=14-
                                        50           100           150
                                      Ambient Concentration (pg/m3)
                                                              200
      Figure 5-14.  Plots of nonambient exposure to PM10, (a) daily individual daytime values
                    from PTEAM data and (b) daily-average values from THEES data.

      Source: Data taken from (a) Clayton et al. (1993) and (b) Lioy et al. (1990).
1     expected to correlate with ambient PM concentrations on a community basis.  Therefore, in

2     linear nonthreshold models of PM health effects, nonambient PM exposure is not expected to

3     contribute to the relative risk determined in a regression of health responses on ambient PM
      June 2003
                              5-110
                               DRAFT-DO NOT QUOTE OR CITE

-------
 1      concentration.  Furthermore, in time-series analysis of pooled or daily health data, it is expected
 2      that ambient PM exposure rather than total personal PM exposure will have the stronger
 3      association with ambient PM concentration.
 4
 5      5.5.2.2  Role of Compositional Differences in Exposure Characterization for Epidemiology
 6           The majority of the available data on PM exposures and relationships with ambient PM
 7      have come from a few large-scale studies, such as PTEAM, or longitudinal studies on selected
 8      populations.  Consequently, for most analyses, exposure scientists and statisticians had to rely on
 9      PM10 or PM2 5 mass data, instead of elemental or chemical compositional information on
10      individual or microenvironmental samples.  In a few cases, researchers have examined the
11      factors influencing indoor outdoor ratios or penetration  and deposition coefficients using
12      elemental mass data on personal, indoor, and outdoor PM data (e.g., Ozkaynak et al., 1996a,b;
13      Yakovleva et al., 1999).  These results have been informative in terms of understanding relative
14      infiltration of different classes of particle sizes and sources into residences (e.g., fossil  fuel
15      combustion, mobile  source emissions, soil-derived, etc.). Clearly, in the accumulation-mode,
16      particles associated with stationary or mobile combustion sources have greater potential for
17      penetration into homes and other microenvironments than does crustal material.  There will be
18      variability in the chemical composition of these broad categories of source classes and also
19      probably variations in relative toxicity. Moreover, when particles and reactive gases are present
20      indoors in the presence of other pollutants or household chemicals, they  may react to form
21      additional or different compounds and particles with yet unknown physical, chemical, and toxic
22      composition (Wainman et al., 2000). Thus, if indoor-generated and ambient PM  were
23      responsible for different types of health effects or had significantly different toxicities on a per
24      unit mass basis, it would then be important that ambient and nonambient exposure should be
25      separated and treated as different components, much like the current separation of PM10 into
26      PM2 5 and PM10_2 5. These complexities in personal exposure profiles may introduce
27      nonlinearities and other statistical challenges in the selection and fitting of concentration-
28      response models. Unfortunately, PM health effects models have not yet been able to
29      meaningfully consider such complexities.
30           It is important  also to note that individuals spend time in places other than their homes and
31      outdoors. Many of the interpretations reported in the published literature on factors influencing

        June 2003                                5-111        DRAFT-DO NOT QUOTE OR CITE

-------
 1      personal PM10 exposures, as well as in this chapter, come from the PTEAM study. The PTEAM
 2      study was conducted 10 years ago in one geographic location in California, during one season,
 3      and most residences had very high and relatively uniform air exchange rates. Non-home indoor
 4      microenvironments were not monitored directly during the PTEAM study.  Commuting
 5      exposures from traffic or exposures in a variety of different public places or office buildings
 6      could not be assessed directly. Nonresidential buildings may have lower or higher ambient
 7      infiltration rates depending on the use and type of the mechanical ventilation systems employed.
 8      Because the source and chemical composition of particulate matter affecting personal exposures
 9      in different microenvironments vary by season, day-of-the-week, and time of day, it is
10      conceivable that some degree of misclassification of exposures to PM toxic agents of concern
11      could be introduced when health-effects models use only daily-average mass measures such as
12      PM10 or PM2 5. However, because of the paucity of currently available data on many of these
13      factors, it is not now possible to ascertain the potential magnitude  or severity of any such
14      complex exposure missclassification problems or their potential implications for interpretation of
15      results from PM epidemiology at this point.
16
17      5.5.2.3  Role of Spatial Variability in Exposure Characterization for Epidemiology
18          Chapter 3, Section 3.2.3 and Chapter 5, Section 5.3 present information on the spatial
19      variability  of PM mass and chemical components at fixed-site ambient monitors; for purposes of
20      this chapter, this spatial variability is called an "ambient gradient." Any gradient that may exist
21      between a fixed-site monitor and the outdoor microenvironments near where people live, work,
22      and play obviously affects the concentration profile actually experienced by people as they go
23      about their daily lives.
24          However, the evidence so far indicates that PM concentrations,  especially fine PM (mass
25      and sulfate), generally are distributed uniformly in most metropolitan areas. This reduces the
26      potential for exposure misclassification because of outdoor spatial gradients when a limited
27      number of ambient PM monitors are used to represent population average ambient exposures in
28      time-series or cross-sectional epidemiological studies of PM.  This topic is further discussed in
29      Section 5.6.5. However, as discussed earlier, the  same assumption is not necessarily true for
30      different components of PM such as PM10_25 because source-specific and other spatially
31      nonuniform pollutant emissions could alter the spatial profile of individual PM components in a

        June 2003                                 5-112        DRAFT-DO NOT QUOTE OR CITE

-------
 1      community.  For example, particulate and gaseous pollutants emitted from motor vehicles tend
 2      to be higher near roadways and inside cars.  Likewise, acidic and organic PM species may be
 3      location- and time-dependent.  Furthermore, human activities are complex. If outdoor PM
 4      constituent concentration profiles are either spatially or temporally variable, it is likely that
 5      exposure misclassification errors could be introduced in the analysis of PM air pollution and
 6      health data.
 7
 8      5.5.3   Analysis of Exposure Measurement Error Issues in Particulate Matter
 9              Epidemiology
10           The effects of exposure misclassification on relative risk estimates of disease using
11      classical 2 by 2 contingency design (i.e., exposed/nonexposed versus diseased/nondiseased) have
12      been studied extensively in the epidemiological literature. It has been shown that the magnitude
13      of the exposure-disease association (e.g., relative risk) because of either misclassification of
14      exposure or disease alone (i.e., nondifferential misclassification) biases the effect results toward
15      the null; and differential misclassification (i.e., different magnitudes of disease misclassification
16      in exposed and nonexposed populations) can bias the effect measure toward or away from the
17      null value relative to the true measure of association (Shy et al., 1978; Gladen and Rogan, 1979;
18      Copeland et al., 1977; Ozkaynak et al., 1986). However, the extension of these results from
19      contingency  analysis design to multivariate models (e.g., log-linear regression, Poissson, logit)
20      typically used in recent PM epidemiology has been more complicated.
21
22      5.5.3.1  Time-Series Analyses
23           Researchers have investigated the appropriateness of using ambient PM concentration as
24      an exposure metric and have developed a framework for analyzing measurement errors typically
25      encountered  in the analysis of  time-series mortality and morbidity effects from exposures to
26      ambient PM  (cf. Zeger et al., 2000; Dominici et al., 2000; Samet et al., 2000). Use of this
27      framework, discussed more extensively in Chapters 8 and 9, leads to the following conclusions:
28      the deviation of an individual's personal exposure from the risk-weighted average exposure due
29      to variations  in ambient concentrations, infiltration rates, and indoor-generated PM
30      concentrations is a Berkson error and will not bias the estimated regression coefficient (P, the
31      increase in risk per unit increase in PM) in a time-series analysis of mortality as a function of
32      ambient PM  concentrations. However, the difference between the average personal exposure
        June 2003                                 5-113        DRAFT-DO NOT QUOTE OR CITE

-------
 1      and the true ambient concentration will bias p.  If the daily, individual values of ambient PM
 2      exposure and nonambient PM exposure are poorly correlated and the attenuation coefficient, a,
 3      equal to the ambient PM concentration/ambient PM exposure were constant, the bias would be
 4      given by Pc = «PA where pc is calculated using the ambient PM concentration, C, and PA is
 5      calculated using the ambient PM exposure, A; i.e., the risk determined from an analysis using the
 6      daily ambient PM concentrations would be lower than the risk obtained using ambient PM
 7      exposure by the factor a. However, pc provides the correct information on the change in health
 8      risks that would be produced by a change in ambient concentrations.  However, if the daily,
 9      individual values of ambient PM exposure and nonambient PM exposure are highly correlated
10      and the nonambient PM is toxic for the effect being studied, nonambient PM exposure will act as
11      a confounder and could introduce substantial bias into Pc.  Only one study has reported the
12      association between the daily, individual values of ambient PM exposure and nonambient PM
13      exposure. Wilson et al. (2000), in a further analysis of the PTEAM data set, found a coefficient
14      of determination (R2) of 0.03 suggesting that the daily, individual values of ambient PM
15      exposure and nonambient PM exposure are independent so that nonambient PM exposure will
16      not confound Pc.
17
18      5.5.3.2  Studies of Chronic Effects
19          The Six Cities (Dockery et al., 1993) and American Cancer Society (ACS) (Pope et al.,
20      1995)  studies have played an important role in assessing the health effects from long-term
21      exposures to paniculate pollution. Even though these studies often have been considered as
22      chronic epidemiological studies, it is not easy to differentiate the role of historic exposures from
23      those of recent exposures on chronic disease mortality.  In the Six Cities study, fine particles and
24      sulfates were measured at the community level, and the final analysis of the database used six
25      city-wide average ambient concentration measurements.  This limitation also applies to the ACS
26      study but has less impact because of the larger number of cities considered in that study. In a
27      HEI-sponsored reanalysis of the Six Cities and the ACS data sets,  Krewski et al. (2000)
28      attempted to examine some of the exposure misclassification issues either analytically or through
29      sensitivity  analysis of the aerometric and health data. The HEI reanalysis project also addressed
30      exposure measurement error issues related to the Six Cities study.  For example, the inability to
31      account for exposures prior to the enrollment of the cohort hampered accurate interpretation of

        June 2003                                 5-114        DRAFT-DO NOT QUOTE OR CITE

-------
 1      the relative risk estimates in terms of acute versus chronic causes.  Although the results seem to
 2      suggest that past exposures are more strongly associated with mortality than recent exposures,
 3      the measurement error for long-term averages could be higher and influence these
 4      interpretations. For example, Krewski et al.  (2000), using the individual mobility data available
 5      for the Six Cities cohort, analyzed the mover and nonmover groups separately.  The relative risk
 6      of fine particle effects on all-cause mortality was shown to be higher for the nonmover group
 7      than for the mover group, suggesting the possibility of higher exposure misclassification biases
 8      for the movers. The issue of using selected ambient monitors in the epidemiological analyses
 9      also was investigated by the ACS and Six Cities studies reanalysis team.  Krewski et al. (2000)
10      presented the sensitivity of results to choices made in selecting stationary or mobile-source-
11      oriented monitors.  For the ACS study, reanalysis of the sulfate data using only those monitors
12      designated as residential or urban and excluding sites designated as industrial,  agricultural, or
13      mobile did not change the risk estimates appreciably. On the other hand, application of spatial
14      analytic methods designed to control confounding at larger geographic scales (i.e., between
15      cities) caused changes in the particle and sulfate risk coefficients.  Spatial adjustment may
16      account for differences in pollution mix  or PM composition, but many other cohort-dependent
17      risk factors will vary across regions or cities  in the United States. Therefore, it is difficult to
18      interpret these findings solely in terms of spatial differences in pollution composition or relative
19      PM toxicity until further research is concluded.
20           The influence of measurement errors in air pollution exposure and health effects
21      assessments has also been examined by Navidi et al. (1999).  This study developed techniques to
22      incorporate exposure measurement errors encountered in long-term air pollution health- effects
23      studies and tested them on the data from the  University of Southern California Children's Health
24      Study conducted in 12 communities in California. These investigators developed separate error
25      analysis models for direct (i.e., personal sampling) and indirect (i.e., microenvironmental)
26      personal exposure assessment methods.  These models were generic to most air pollutants, but a
27      specific application was  performed using a simulated data set for studying ozone health effects
28      on lung function decline in children. Because the assumptions made in their
29      microenvironmental simulation modeling framework were similar to those made in estimating
30      personal PM exposures, it is useful to consider the conclusions from Navidi et  al. (1999).
31      According to Navidi et al. (1999), neither the microenvironmental nor the personal sampler

        June 2003                                5-115        DRAFT-DO NOT  QUOTE OR CITE

-------
 1      method produces reliable estimates of the exposure-response slope for O3 when measurement
 2      error is uncorrected. Because of nondifferential measurement error, the bias was toward zero
 3      under the assumptions made in Navidi et al. (1999) but could be away from zero if the
 4      measurement error was correlated with the health response. A simulation analysis indicated that
 5      the standard error of the estimate of a health effect increases as the errors in exposure assessment
 6      increase (Navidi et al., 1999). According to Navidi et al. (1999), when a fraction of the ambient
 7      level in a microenvironment is estimated with a standard error of 30%, the standard error of the
 8      estimate is 50% higher than it would be if the true exposures were known. It appears that errors
 9      in estimating ambient PM indoor/ambient PM outdoor ratios have much more influence on the
10      accuracy of the microenvironmental approach than do errors in estimating time spent in these
11      microenvironments.
12          Epidemiological studies of chronic effects, that use long-term average ambient PM
13      concentrations as the exposure metric, generally do not address the nonambient component of
14      personal exposure (N).  However, if N contributes to the health effect being studied and the
15      average N is different in different cities, the correlation between the average ambient PM
16      concentration and the health effect could be reduced. In an analysis of the effect of nonambient
17      exposure on time-series epidemiology, Dominici et al. examine nonambient exposure data from
18      several cities and conclude that the average nonambient PM exposure varies little among cities
19      in developed countries. Ott et al. (2000) examined estimated average and daily, individual
20      values of nonambient exposure from three studies and concluded that both the average and the
21      distribution of daily, individual values were similar for the three studies.
22
23      5.5.4    Conclusions from Analysis of Exposure Measurement Errors on
24              Particulate Matter Epidemiology
25          Personal exposures to PM are influenced by a number of factors and sources of PM located
26      in both indoor and outdoor microenvironments.  However, PM resulting from ambient sources
27      does penetrate into indoor environments such as residences, offices, public buildings, etc., in
28      which individuals spend a large portion of their daily lives.  The correlations between total
29      personal exposures and ambient or outdoor PM concentrations can vary depending on the
30      relative contributions of indoor PM sources to total personal exposures. Panel studies of both
31      adult and young  subjects have shown that, in fact, individual correlations of personal exposures
32      with ambient PM concentrations could vary person to person, and even day to day, depending on
        June 2003                                5-116        DRAFT-DO NOT QUOTE OR CITE

-------
 1      the specific activities of each person. Separation of PM exposures into two components,
 2      ambient PM and nonambient PM, would reduce uncertainties in the analysis and interpretation
 3      of PM health effects data. Nevertheless, because ambient PM is an integral component of total
 4      personal exposures to PM, statistical analyses of cohort-average exposures are strongly
 5      correlated with ambient PM concentrations when the size of the underlying population studied is
 6      large. Using the PTEAM study data, analysis of exposure measurement errors, in the context of
 7      time-series epidemiology, also has shown that errors or uncertainties introduced by using
 8      surrogate exposure variables, such as ambient PM concentrations, could lead to biases in the
 9      estimation of health risk coefficients. These then would need to be corrected by suitable
10      calibration of the PM health risk coefficients.  Correlations between the PM exposure variables
11      and other covariates (e.g., gaseous co-pollutants, weather variables, etc.) also could influence the
12      degree of bias in the estimated PM regression coefficients. However, most time-series
13      regression models employ seasonal or temporal detrending of the variables, thus reducing the
14      magnitude of this cross-correlation problem (Ozkaynak and Spengler 1996).
15           Ordinarily, exposure measurement errors are not expected to influence the interpretation of
16      findings from either the chronic or time-series epidemiological studies that have used ambient
17      concentration data if they include sufficient adjustments for seasonality and key confounders.
18      Clearly, there is no question that better estimates of exposures to components of PM of health
19      concern are beneficial. Composition of PM may vary in different geographic locations and
20      different exposure microenvironments.  Compositional and spatial variations could lead to
21      further errors in using ambient PM measures as surrogates for exposures to PM. Even though
22      the spatial variability of PM (PM25 in particular) mass concentrations in urban environments
23      seems to be small, the same conclusions drawn above regarding the influence of measurement
24      errors may not necessarily hold for all of the toxic PM components. Again, the expectation
25      based on statistical modeling considerations is that these exposure measurement errors or
26      uncertainties will most likely reduce the statistical power of the PM health effects analysis,
27      making it difficult to detect a true underlying association between the correct exposure metric
28      and the health outcome studied.  However, until more data on exposures to toxic agents of PM
29      become available, existing studies on PM exposure measurement errors must be relied on; thus,
30      at this time the use of ambient PM concentrations as a surrogate for exposures is not expected to
        June 2003                                5-117        DRAFT-DO NOT QUOTE OR CITE

-------
 1     change the principal conclusions from PM epidemiological studies that use community average
 2     health and pollution data.
 3
 4
 5     5.6    SUMMARY OF OBSERVATIONS AND LIMITATIONS
 6     Exposure Definitions and Components
 7      •  Personal exposure to PM mass or its constituents results when individuals come in contact
           with particulate pollutant concentrations in locations or microenvironments that they
           frequent during a specific period of time.  Various PM exposure metrics can be defined
           according to its source (i.e., ambient, nonambient) and the microenvironment where
           exposure occurs.
 8      •  Personal exposure to PM results from an individual's exposure to PM in many different
           types of microenvironments (e.g., outdoors near home, outdoors away from home, indoors at
           home, indoors at office or school, commuting, restaurants, malls,  other public places, etc.).
           Thus,  total daily exposure to PM for a single individual can be expressed as the  sum of
           various microenvironmental exposures that the person encounters during the course of a day.
 9      •  In a given microenvironment, particles may originate from a wide variety of sources. In an
           indoor microenvironment, PM may be generated from within as a result of PM-generating
           activities (e.g., cooking, cleaning, smoking,  resuspending PM from PM resulting from both
           indoor and outdoor sources that had settled out), from outside (outdoor PM entering through
           cracks and openings in the structure), and  from the chemical interaction  of pollutants from
           outdoor air with indoor-generated pollutants.
10      •  The total daily exposure to PM for a single individual also can be expressed as the sum of
           contributions of ambient and nonambient PM. Nonambient PM exposure is due to PM
           generated by indoor sources, personal activities, and chemical reactions  in indoor air.
           Ambient PM exposure includes exposure to ambient PM while outdoors, and ambient PM
           that has infiltrated indoors while indoors.  However, within a large population, there will be
           distributions of total personal exposure and its components due to variations in human
           activities and microenvironmental concentrations and sources each individual encounters.
11      •  Exposure models are useful tools for examining the importance of sources,
           microenvironments, and physical and behavioral factors that influence personal  exposures to

       June 2003                                5-118        DRAFT-DO NOT QUOTE OR CITE

-------
           PM.  However, development and evaluation of population exposure models for PM and its
           components have been limited.  Improved modeling methodologies and new model input
           data are needed.
12
13     Factors Affecting Concentrations and Exposures to Particulate Matter
14       •  Concentrations of PM indoors are affected by several factors and mechanisms: ambient
           concentrations outdoors; air exchange rates; particle penetration factors; particle production
           from indoor sources and indoor air chemistry; and indoor particle decay rates and removal
           mechanisms caused by physical processes or resulting from mechanical filtration, ventilation
           or air-conditioning devices.
15       •  Average personal exposures to PM mass and its constituents are influenced by
           microenvironmental PM concentrations and by how much time is spent by each individual in
           these various indoor and outdoor microenvironments. Nationwide, individuals, on average,
           spend nearly 90% of their time indoors (at home and in other indoor locations) and about 6%
           of their time outdoors.
16       •  Personal exposures are associated with both indoor as well as outdoor sources. The personal
           exposure/outdoor concentration ratios present substantial intra- and inter-personal
           variability. This variability is due to both the presence of personal and microenvironmental
           sources and the varying effect of the outdoor particles on indoor environments.
17       •  Home characteristics may be the most important factor that affects the  relationship between
           the average population exposures and ambient concentrations. Air exchange rate seems to
           be an important home characteristic surrogate that can explain a large fraction of the observe
           inter- and intra-personal variability. One reason why longitudinal studies (many repeated
           measurements per person) provide stronger correlations between personal exposure and
           outdoor concentrations than cross-sectional studies (few repeated measurements per
           individual) may be because home characteristics remain the same.
18       •  Because home characteristics constitute the most important factor affecting personal
           exposures, one would expect that correlations between average population exposures and
           outdoor concentrations will vary by season and geography.
       June 2003                                5-119        DRAFT-DO NOT QUOTE OR CITE

-------
 •  The relative size of personal exposure to ambient PM relative to nonambient PM depends
    on the ambient concentration, the infiltration rate of outdoor PM into indoor
    microenvironments, the amount of PM generated indoors (e.g., ETS, cooking and cleaning
    emissions), and the amount of PM generated by personal activity sources.  Infiltration rates
    primarily depend on air exchange rate, size-dependent particle penetration across the
    building membrane, and size-dependent removal rates.  All of these factors vary over time
    and across subjects and building types.
 •  The relationship between PM exposure, dose, and health outcome could depend on the
    concentration, composition, and toxicity of PM or PM components originating from
    different sources.  Application of source apportionment techniques to indoor and outdoor
    PM2 5 and personal, indoor, and outdoor PM10 composition data have identified the following
    general source categories: outside soil, resuspended indoor soil, indoor soil, personal
    activities, sea-salt, motor vehicles, nonferrous metal smelters, and secondary sulfates.
 •  There have been only a limited number of studies that have measured the physical and
    chemical constituents of PM in personal or microenvironmental samples. Available data on
    PM constituents indicate the following:
    -  personal and indoor sulfate measurements often are correlated highly with outdoor and
       ambient sulfate concentration measurements;
    -  for acid aerosols, indoor air chemistry is particularly important because of the
       neutralization of the acidity by ammonia, which is present at higher concentrations
       indoors because of the presence of indoor sources of ammonia;
    -  for SVOCs, including PAHs and phthalates, the presence of indoor sources will
       substantially affect the relation between indoor and ambient concentrations;
    -  penetration and decay rates are functions of size and will cause variations in the
       attenuation factors as a function of particle size; infiltration rates will be higher for PMX
       and PM2 5 than for PM10, PM10_2 5 or ultrafme particles; and
    -  indoor air chemistry may increase indoor concentrations of organic PM.
 •  Even though there is an increasing amount of research being performed to  measure PM
    constituents in different PM size fractions, with few exceptions (i.e., sulfur or sulfates) the
    current data are inadequate to adequately assess the relationship between personal, indoor,
    and ambient concentrations of most PM constituents.

June 2003                                5-120       DRAFT-DO NOT  QUOTE OR CITE

-------
 1     Correlations Between Personal Exposures, Indoor, Outdoor, and Ambient Measurements
 2       •  Most of the available personal data on PM measurements and information on the
           relationships between personal and ambient PM come from a few large-scale studies (e.g.,
           the PTEAM study) or the longitudinal panel studies that have been conducted on selected
           populations, such as the elderly.
 3       •  Panel and cohort studies that have measured PM exposures and concentrations typically have
           reported their results in terms of three types of correlations: (1) longitudinal, (2) pooled, and
           (3) daily-average correlations between personal and ambient or outdoor PM.
 4       •  The type of correlation analysis performed can have a substantial effect on the resulting
           correlation coefficient.  Low correlations with ambient concentrations could result when
           people with very different nonambient exposures are pooled even though their individual
           personal exposures may temporally be correlated highly with ambient concentrations.
 5       •  Longitudinal and pooled correlations between personal  exposure and ambient or outdoor PM
           concentrations reported by various investigators varied considerably among the different
           studies and in each study between the study subjects.  Most studies report longitudinal
           correlation coefficients that range from close to zero to near one, indicating that individual's
           activities and residence type may have a significant effect on total personal exposures to PM.
 6       •  Longitudinal studies that measured sulfate found high correlations between personal and
           ambient sulfate.
 7       •  In general, probability-based population studies tend to show low pooled correlations
           because of the high differences in levels of nonambient PM-generating activities from one
           subject to another.  In contrast, the absence of indoor sources for the populations in several
           of the longitudinal panel studies resulted in high  correlations between personal exposure and
           ambient PM within subjects over time for these populations. However, even for these
           studies, correlations varied by individual depending on their activities and the
           microenvironments that they occupied.
 9
10
       June 2003                                 5-121       DRAFT-DO NOT QUOTE OR CITE

-------
 1     Potential Sources of Error Resulting from Using Ambient Particulate Matter
 2     Concentrations in Epidemiological Analyses
 3          As yet, there is no clear consensus among exposure analysts as to how well community
 4     monitor measurements of ambient air PM concentrations represent a surrogate for personal
 5     exposure to total PM or to ambient PM.
 6      •  Measurement studies of personal exposures to PM are still few and limited in spatial,
           temporal, and demographic coverage. Consequently, with the exception of a few
           longitudinal panel studies, most epidemiological studies on PM health effects have relied on
           daily-average PM concentration measurements obtained from ambient community
           monitoring data as a surrogate for the exposure variable.
 7      •  Because individuals are exposed to particles in a multitude of indoor and outdoor
           microenvironments during the course of a day, concerns about error introduced in the
           estimation of PM risk coefficients using ambient, as opposed to personal PM measurements,
           have been raised.
 8      •  Total personal exposures to PM could vary from person to person, and even  day to day,
           depending on the specific activities of each person. Separation of PM exposures into two
           components, ambient and nonambient PM, would reduce potential uncertainties in the
           analysis and interpretation of PM health effects data.
 9      •  Available data indicate that PM mass concentrations, especially fine PM, typically are
           distributed relatively uniformly in most metropolitan areas, thus reducing the potential for
           exposure misclassification because of spatial variability when a limited number of ambient
           PM monitors are used to represent population  average ambient exposures in  community
           time-series or long-term,  cross-sectional epidemiological studies of PM.
10      •  Even though the spatial variability of PM (in particular, PM25) mass concentrations in urban
           environments seems to be small, the same conclusions drawn above regarding the influence
           of measurement errors may not necessarily hold for all PM components.
11      •  There are important differences in the relationship of ambient PM concentrations with
           ambient PM exposures and with nonambient PM exposures. Various researchers have
           shown that ambient PM exposure is a function of ambient PM concentration and that
           concentrations of ambient PM are driven by meteorology, by changes in source emission
           rates, and in locations of emission sources relative to the measurement site. However,

       June 2003                                5-122        DRAFT-DO NOT QUOTE OR CITE

-------
           nonambient PM exposure is independent of ambient PM concentration because
           concentrations of nonambient PM are driven by the daily activities of people.
12      •  Because personal exposures also include a contribution from ambient concentrations, the
           correlation between daily-average personal exposure and the daily-average ambient
           concentration increases as the number of subjects measured daily increases. An application
           of a Random Component Superposition model has shown that the contributions of ambient
           PM10 and indoor-generated PM10 to community mean exposure can be decoupled in
           modeling urban population exposure distributions.
13      •  If linear nonthreshold models are assumed in time-series analysis of daily-average ambient
           PM concentrations and community health data, nonambient PM exposure is not expected to
           contribute to the relative risk estimates determined by regression of health responses on
           ambient PM concentration.
14      •  Using the PTEAM study data, analysis of exposure measurement errors in the context of
           time-series epidemiology has shown that the error introduced by using ambient PM
           concentrations as a surrogate for ambient PM exposure biases the estimation of health risk
           coefficients low by the ratio of ambient PM exposure to ambient PM concentration (called
           the attenuation factor). However, the health risk coefficient determined using ambient PM
           concentrations provides the correct information on the change in health risks that would be
           produced by a change in ambient concentrations.
15      •  Because sources and chemical composition of particulate matter affecting personal exposures
           in different microenvironments vary by season, day-of-the-week, and time of day, it is likely
           that some degree of misclassification of exposures to PM toxic agents of concern will be
           introduced when health-effects models use only daily-average mass measures such as PM10
           or PM2 5. Because of the paucity of currently available  data on many of these factors, at this
           point it is impossible to ascertain the significance of these more complex exposure
           misclassification problems in the interpretation of results from PM epidemiology.
16      •  Exposure measurement errors may depend on particle size and composition. PM25 better
           reflects personal exposure to PM of outdoor origin than PM10.  Various indicators of ultrafme
           particle concentrations or other components of PM may also be useful exposure indicators
           for epidemiological studies.


       June 2003                                 5-123        DRAFT-DO NOT QUOTE OR CITE

-------
    Seasonal or temporal variations in the measurement errors and their correlations between
    different PM concentration measures and co-pollutants (e.g., SO2, CO, NO2, O3) could
    influence the error analysis results but not likely the interpretation of current findings.
    Multipollutant personal exposure studies have suggested that ambient concentrations of
    gaseous co-pollutants serve as surrogates of personal exposures to particles rather than as
    confounders.  The implications for epidemiology are discussed in Chapters 8 and 9.
    Ordinarily, PM exposure measurement errors are not expected to influence the interpretation
    of findings from either the community time-series or long-term epidemiological studies that
    have used ambient concentration data if they include sufficient adjustments for seasonality
    and key personal and geographic confounders.
    In the context of long-term epidemiological studies, it appears that the errors introduced in
    estimating ambient PM indoor/ambient PM outdoor ratios have much more influence on the
    accuracy of the microenvironmental exposure estimation approach than do errors in
    estimating time spent in these microenvironments.
    To reduce exposure misclassification errors in PM epidemiology, conducting new cohort
    studies  of sensitive populations with better real-time techniques for exposure monitoring and
    further  speciation of indoor-generated, ambient, and personal PM mass are essential.
    Based on statistical modeling considerations, it is expected that existing PM exposure
    measurement errors or uncertainties most likely will reduce the statistical power of the PM
    health effects analysis, thus making it difficult to detect a true underlying association
    between the correct exposure metric and the health outcome studied.
    Although exposure measurement errors for fine particles are not expected to influence the
    interpretation of findings from either the community time-series or the long-term,  cross-
    sectional epidemiological studies that have used ambient concentration data, they  may
    underestimate the strength of the effect.  Sufficient data are not available to evaluate the
    effect of exposure measurement error for other PM species or size fractions.
June 2003                                 5-124       DRAFT-DO NOT QUOTE OR CITE

-------
1      Key Findings
2      • Most people spend most of their time indoors where they are exposed to indoor-generated
         PM and ambient PM that has infiltrated indoors.
3      • Indoor-generated and ambient PM differ in sources, sizes, chemical composition, and
         toxicity.
4      • The ambient PM concentration and the indoor PM concentration can be measured by outdoor
         and indoor monitors. The total personal exposure can be measured by a personal exposure
         monitor carried by the person.  However, the concentrations of indoor-generated PM and
         ambient PM that has infiltrated indoors and the related values of ambient and nonambient PM
         exposures must be estimated.
5      • From a regression of individual, daily values of total personal exposure  on daily PM
         concentrations, the intercept gives the average nonambient PM exposure and the slope gives
         the average attenuation factor (the ratio of ambient PM exposure to ambient PM
         concentration).
6      • Similarly, from a regression  of individual, daily values of indoor PM concentration on daily
         ambient concentrations, the intercept gives the average concentration of indoor-generated PM
         and the slope gives the average infiltration factor (concentrations of ambient PM that has
         infiltrated indoors/ambient PM concentration).
7      • The attenuation factor and the infiltration factor depend on the penetration coefficient, the
         fraction of ambient PM that penetrates through the walls, doors, windows, etc.; the deposition
         or removal rate, a measure of how rapidly PM within the indoor microenvironment is
         removed by deposition to surfaces or by filtration in a heating/cooling system; and the air
         exchange rate, a measure of how rapidly indoor air is replaced by outdoor air.  The
         attenuation factor also depends on the fraction of time spent outdoors.
8      • The air exchange rate is an important variable  for determining the concentration of ambient
         PM found indoors. It can be measured by release and measurement of an inert tracer gas
         indoors.  The air exchange rate increases with  opening of windows or doors or operation of
         window or attic fans. It also increases as the indoor/outdoor temperature difference
         increases.  For closed homes, i.e., no open windows or doors,  the air exchange rate does not
         appear to be a function of wind speed or direction.


       June 2003                               5-125       DRAFT-DO NOT  QUOTE OR CITE

-------
   The penetration coefficient and the deposition rate can be estimated from measurements of
   outdoor and indoor concentrations under conditions when there are no indoor sources (night
   time or unoccupied home).  These parameters are functions of particle size. The penetration
   coefficient is high and the deposition rate is low for accumulation mode particles (0.1 to
   1.0 |im). The penetration coefficient is lower and the deposition rate is higher for ultrafme
   particles (< 0.1 jim) and coarse mode particles (> 1.0 jim).  The attenuation factor and the
   infiltration factor are higher for particles in the accumulation mode than for ultrafme or
   coarse particles.
   The attenuation factor and the infiltration factor will vary as the air exchange rate and
   therefore will vary with season and housing characteristics.  These factors will increase with
   increased opening of windows and doors. For closed homes, these factors will increase with
   an increase of the indoor/outdoor temperature difference but they do not appear to be affected
   by wind speed or direction.
   The regression technique is useful for finding average values of the attenuation factor and the
   nonambient exposure and possibly for estimating the distribution of individual, daily values
   of the nonambient PM exposure.
   However, individual, daily values of the ambient PM exposure, the nonambient PM exposure,
   and the  attenuation factor are needed.  These may be determined from individual, daily values
   of the total PM personal exposure and  daily ambient PM concentrations by several
   techniques:
    (1)  Mass balance technique.  Direct measurement of the air exchange rate, measurement of
        the fraction of time spent outdoors by a diary of the subject's activity pattern,  and use of
        the equilibrium mass balance equation for the attenuation factor with estimated values
        of the penetration coefficient and the deposition rate.
    (2)  Sulfate ratio technique. Individual, daily values of the attenuation factor (for PM2 5) will
        be given by individual, daily values of personal exposure to sulfate / the daily ambient
        sulfate concentration provided there are no indoor  sources of sulfate and sulfate and
        PM2 5 have similar particle size distributions.
June 2003                                 5-126        DRAFT-DO NOT QUOTE OR CITE

-------
    (3)  Recursive technique.  Indoor-generated emissions, which tend to be episodic, can be
        removed from a continuous record of indoor PM concentration, allowing separation of
        indoor-generated PM and ambient PM that has infiltrated indoors.
   In pooled studies (different subjects measured on different days), individual, daily values of
   the total PM exposure are usually not well correlated with the daily ambient PM
   concentrations. In longitudinal studies (each subject measured for multiple days), individual,
   daily values of the total PM personal exposure and the daily ambient PM concentrations are
   found to be highly correlated for some, but not all subjects.
   Only one study has reported estimated individual, daily values of ambient and nonambient
   PM exposure.  Individual, daily values of the total PM personal exposure and the daily
   ambient PM concentrations were poorly correlated.  However, individual, daily values of
   ambient PM exposure and the daily ambient PM concentrations were highly correlated.
   Individual daily values of ambient and nonambient PM exposure were not well correlated.
   Individual daily values of nonambient PM exposure and daily ambient PM concentrations
   were also not well correlated.
   These key findings will be used in examining the implications of using the ambient PM
   concentration instead of the total PM personal exposure (or the ambient PM exposure) in
   epidemiological studies.
June 2003                                5-127       DRAFT-DO NOT QUOTE OR CITE

-------
  1       REFERENCES

  2       Abraham, J. L.; Siwinski, G.; Hunt, A. (2002) Ultrafme paniculate exposures in indoor, outdoor, personal and
  3             mobile environments: effects of diesel, traffic, pottery kiln, cooking and HEP A filtration on
  4             micro-environmental particle number concentration. Ann. Occup. Hyg.  46(suppl. 1): 406-411.
  5       Abt, E.; Suh, H. H.; Allen, G.; Koutrakis, P. (2000a) Characterization of indoor particle sources: a study conducted
  6             in the metropolitan Boston area. Environ. Health Perspect. 108: 35-44.
  7       Abt, E.; Suh, H. H.; Catalano, P.; Koutrakis, P. (2000b) The relative contribution of outdoor and indoor particle
  8             sources to indoor concentrations. Environ. Sci.  Technol. 34: 3579-3587.
  9       Adams, H. S.; Nieuwenhuijsen, M. J.; Colvile, R. N.; McMullen, M. A.  S.; Khandelwal, P. (2001) Fine particle
10             (PM2 5) personal exposure levels in transport microenvironments, London, UK. Sci. Total Environ.
11             279: 29-44.
12       Adgate, J. L.;  Ramachandran, G.; Pratt, G. C.; Waller, L. A.; Sexton, K. (2002) Spatial and temporal variability in
13             outdoor, indoor, and personal PM25 exposure. Atmos. Environ. 36: 3255-3265.
14       Akland, G. G.; Hartwell, T. D.; Johnson, T. R.; Whitmore, R. W. (1985) Measuring human exposure to carbon
15             monoxide in Washington, D.C., and Denver, Colorado, during the winter of 1982-1983. Environ. Sci.
16             Technol. 19: 911-918.
17       Al-Raheem, M. Y. A.; El-Desouky, M.; Mage, D. T.; Kollander, M.; Wallace, L. A.; Spengler, J. D. (2000) Personal
18             exposures to PM and gaseous air pollutants at the end of the 1991 Kuwait oil fires. Presented at: PM2000:
19             paniculate matter and health—the scientific basis for regulatory decision-making, specialty conference &
20             exhibition; January; Charleston, SC. Pittsburgh, PA: Air & Waste Management Association.
21       Aim, S.; Jantunen, M. J.; Vartiainen, M. (1999) Urban commuter exposure to particle matter and carbon monoxide
22             inside an automobile. J. Exposure Anal. Environ. Epidemiol. 9: 237-244.
23       Anuszewski, J.; Larson, T. V.; Koenig, J. Q. (1998) Simultaneous indoor and outdoor particle light-scattering
24             measurements at nine homes using a portable nephelometer. J. Exposure Anal. Environ. Epidemiol.
25             8:483-493.
26       Baek, S.-O.; Kim, Y.-S.; Perry, R. (1997) Indoor air quality in homes, offices and restaurants in Korean urban
27             areas-indoor/outdoor relationships. Atmos. Environ. 31: 529-544.
28       Bahadori, T. (1998) Human paniculate exposure assessment: relationship between outdoor, indoor, and personal
29             measurements (air pollution) [dissertation]. Boston, MA: Harvard University.
30       Bahadori, T.; Suh, H. H.; Rojas, L.; Koutrakis, P. (1999) Characterization of personal PM10 and PM2.5 exposures
31             for individuals with COPD: results from a pilot study in Nashville, TN. J. Exposure Anal. Environ.
32             Epidemiol.: in press.
33       Bahadori, T.; Suh, H. H.; Rojas-Bracho, L.; Koutrakis, P. (2001) Personal exposure to paniculate matter of
34             individuals with chronic obstructive pulmonary disease (COPD).  J. Exposure Anal. Environ. Epidemiol.:
3 5             accepted.
36       Bohadana, A.  B.; Massin, N.; Wild, P.; Toamain, J.-P.; Engel, S.; Goutet, P. (2000) Symptoms, airway
37             responsiveness, and exposure to dust in beech and oak wood workers. Occup. Environ. Med. 57: 268-273.
38       Brauer, M.; Mannetje, A. (1998) Restaurant smoking restrictions and environmental tobacco smoke exposure. Am.
39             J. Public Health 88: 1834-1836.
40       Brauer, M.; Koutrakis, P.; Spengler, J. D. (1989) Personal exposures to acidic aerosols and gases. Environ. Sci.
41             Technol. 23: 1408-1412.
42       Brauer, M.; Koutrakis, P.; Keeler, G. J.; Spengler, J. D. (1990) Indoor and outdoor concentrations of acidic aerosols
43             and gases. In: Indoor air '90: precedings of the 5th international conference on indoor air quality and climate,
44             volume 2, characteristics of indoor air; July-August; Toronto, ON, Canada. Ottawa, ON, Canada:
45             International Conference on Indoor Air Quality and Climate, Inc.; pp. 447-452.
46       Brauer, M.; Bartlett, K.; Regalado-Pineda, J.; Perez-Padilla, R. (1996) Assessment of paniculate concentrations from
47             domestic biomass combustion in rural Mexico. Environ. Sci. Technol. 30: 104-109.
48       Brauer, M.; Hirtle, R. D.; Hall, A. C.;  Yip, T. R. (1999) Monitoring personal fine particle exposure with a particle
49             counter. J. Exposure Anal. Environ. Epidemiol. 9: 228-236.
50       Brauer, M.; Hruba, F.; Mihalikova, E.; Fabianova, E.; Miskovic, P.; Plzikova, A.; Lendacka, M.; Vanderberg, J.;
51             Cullen, A. (2000) Personal exposure to particles in Banska Bystrica, Slovakia. J. Exposure Anal. Environ.
52             Epidemiol. 10: 478-487.
53
54
         June 2003                                      5-128         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Brown, K. G.; Paxton, M. B. (1998) Predictability of personal exposure to airborne paniculate matter from ambient
  2             concentrations in four communities. In: Vostal, J. I, ed. Health effects of paniculate matter in ambient air.
  3             Proceedings of an international conference; 1997; Prague, Czech Republic. Pittsburgh, PA:  Air & Waste
  4             Management Association; pp. 333-350. (A&WMA publication VIP-80).
  5       Burke, J.; Zufall, M.; Ozkaynak, H. (2001) A population exposure model for paniculate matter: case study results
  6             for PM2 5 in Philadelphia, PA. J. Exposure Anal. Environ. Epidemiol. 11: 470-489.
  7       Burton, R. M.; Suh, H. H.; Koutrakis, P. (1996) Spatial variation in paniculate concentrations within metropolitan
  8             Philadelphia. Environ. Sci. Technol. 30: 400-407.
  9       Buzorius, G.; Hameri, K.; Pekkanen, J.; Kulmala, M. (1999) Spatial variation of aerosol number concentration in
10             Helsinki city.  Atmos. Environ. 33: 553-565.
11       Byrne, M. A.; Lange, C.; Goddard, A. J. H.; Roed, J. (1992) Indoor aerosol deposition studies using
12             neutron-activatable tracers. J. Aerosol Sci. 23(suppl. 1): S543-S546.
13       Calder, K. L. (1957) A numerical analysis of the protection afforded by buildings against BW aerosol attack.
14             Fort Detrick, MD: Office of the Deputy Commander for Scientific Activities; BWL tech. study no. 2.
15       Carrer, P.; Alcini, D.; Cavallo, D. (1997) Daily personal exposure to air pollutants of office workers inMilano. In:
16             Woods, et al. (eds.), Proceedings of Healthy Buildings '97, v.2, pp. 249-254; Washington, DC.
17       Carrer, P.; Alcini, D.; Cavallo, D.; Bollini, D.; Ghalandar, R.; Lovato, L.; Ploia, E.; Vercelli, F.; Visigalli, F.;
18             Maroni, M. (1998) Personal exposure to PM10 of subjects living in Milan. In: Vostal, J. J., ed. Health effects
19             of paniculate matter in ambient air. Proceedings of an international conference; 1997; Prague, Czech
20             Republic. Pittsburgh, PA: Air & Waste Management Association; pp. 359-365. (A&WMA publication
21             VIP-80).
22       Cha, Q.; Chen, Y.-Z.; Du, Y.-X. (1996) Influence of indoor air pollution on quality of outdoor air in city of
23             Guangzhou, China. In: Indoor air '96: the 7th international conference on indoor air quality, volume I; July;
24             Nagoya, Japan. Tokyo, Japan: Institute of Public Health; pp. 1131-1135.
25       Chan, A. T. (2002) Indoor-outdoor relationships of paniculate matter and nitrogen oxides under different outdoor
26             meteorological conditions. Atmos. Environ. 36: 1543-1551.
27       Chan, L. Y.; Lau, W. L.; Lee, S. C.; Chan, C. Y. (2002) Commuter exposure to paniculate matter in public
28             transportation modes in Hong Kong. Atmos. Environ. 36: 3363-3373.
29       Chang, L.-T.; Koutrakis, P.; Catalano, P. J.; Suh, H.  H. (2000) Hourly personal exposures to fine particles and
30             gaseous pollutants-results from Baltimore, Maryland. J. Air Waste Manage. Assoc. 50: 1223-1235.
31       Chao, C. Y. H.; Tung, T. C. (2001) An empirical model for outdoor contaminant transmission into residential
32             buildings and experimental verification. Atmos. Environ. 35:  1585-1596.
33       Chuang, J. C.; Callahan, P. J.; Lyu, C. W.; Wilson, N. K. (1999) Fob/cyclic aromatic hydrocarbon exposures of
34             children in low-income families.  J. Exposure  Anal. Environ. Epidemiol. 2: 85-98.
35       Clayton, C. A.; Perritt, R. L.; Pellizzari, E. D.; Thomas, K. W.; Whitmore, R. W.; Wallace, L. A.; Ozkaynak, H.;
36             Spengler, J. D. (1993) Particle total exposure  assessment methodology (PTEAM) study: distributions of
37             aerosol and elemental concentrations in personal, indoor, and outdoor air samples in a southern California
38             community. J. Exposure Anal. Environ. Epidemiol. 3: 227-250.
39       Clayton, C. A.; Pellizzari, E. D.; Rodes, C. E.; Mason, R. E.; Piper, L. L. (1999a) Estimating distributions of
40             long-term paniculate matter and manganese exposures for residents of Toronto, Canada. Atmos. Environ.
41             33:2515-2526.
42       Clayton, C. A.; Pellizzari, E. D.; Whitmore, R. W.; Perritt, R. L., Quackenboss, J. J. (1999b) National human
43             exposure assessment survey (NHEXAS): distributions and associations of lead, arsenic, and volatile organic
44             compounds in EPA Region 5. J. Exposure Anal. Environ. Epidemiol. 9: 381-392.
45       Conner, T. L.; Lonneman, W. A.; Seila, R. L. (1995) Transportation-related volatile hydrocarbon source profiles
46             measured in Atlanta. J. Air Waste Manage. Assoc. 45: 383-394.
47       Conner, T. L.; Norris, G. A.; Landis, M. S.; Williams, R. W. (2001) Chemical and physical characterization of
48             indoor, outdoor, and personal samples collected in and around a retirement facility. In: Measurement of Toxic
49             and Related Air Pollutants: proceedings of a specialty conference; September, 2000; Research Triangle Park,
50             NC. Pittsburgh, PA: (A&WMA publication VTP-100/CD; pp. 678-700).
51       Copeland, K. T.; Checkoway, H.; McMichael, A. J.; Holbrook, R. H. (1977) Bias due to misclassification in the
52             estimation of relative risk. Am. J. Epidemiol.  105: 488-495.
53       Crump, K. S. (2000) Manganese exposures in Toronto during use of the gasoline additive, methylcyclopentadienyl
54             manganese tricarbonyl. J. Exposure Anal. Environ. Epidemiol. 10: 227-239.
55       Daisey, J. M.; Mahanama, K. R. R.; Hodgson, A. T.  (1998) Toxic volatile organic compounds in simulated
56             enviromental tobacco smoke: emission factors for exposure assessment. J. Exposure Anal. Environ.
57             Epidemiol. 8:313-334.


         June 2003                                      5-129        DRAFT-DO NOT QUOTE OR CITE

-------
  1       Dockery, D. W.; Spengler, J. D. (1981) Indoor-outdoor relationships of respirable sulfates and particles.
  2             Atmos. Environ. 15: 335-343.
  3       Dockery, D. W.; Pope, C. A., Ill; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. E.; Ferris, B. G., Jr.; Speizer, F. E.
  4             (1993) An association between air pollution and mortality in six U.S. cities. N. Engl. J. Med. 329: 1753-1759.
  5       Dominici, F.; Zeger, S. L.; Samet, J. (2000) A measurement error model for time-series studies of air pollution and
  6             mortality. Biostatistics 1: 157-175.
  7       Donham, K. J.; Cumro, D.; Reynolds, S. J.; Merchant, J. A. (2000) Dose-response relationships between
  8             occupational aerosol exposures and cross-shift declines of lung function in poultry workers: recommendations
  9             for exposure limits. J.  Occup. Environ. Med. 42: 260-269.
10       Douce, D. S.; Clench, M. R.; Frost, B. (2001) Variations in the estimation of the contribution of environmental
11             tobacco smoke (ETS) to respirable (< 5\im) indoor air particulates obtained by the use of different analytical
12             methods. J. Environ. Monit. 3: 295-301.
13       Duan, N. (1982) Models for human exposure to air pollution. Environ. Int. 8: 305-309.
14       Duan, N. (1991) Stochastic microenvironment models for air pollution exposure. J. Exposure Anal. Environ.
15             Epidemiol. 1:235-257.
16       Duan, N.; Dobbs, A.; Ott, W. (1990) Comprehensive definitions of exposure and dose to environmental pollution.
17             In: Proceedings of the EPA/A&WMA specialty conference on total exposure assessment methodology;
18             November; Las Vegas, NV. Pittsburgh, PA: Air & Waste Management Association; pp. 166-195
19             (A&WMA publication VIP-16).
20       Dubowsky, S. D.; Wallace, L. A.; Buckley, T. J. (1999) The contribution of traffic to indoor concentrations of
21             polycyclic aromatic hydrocarbons. J. Exposure Anal. Environ. Epidemiol. 9: 312-321.
22       Ebelt, S. T.; Petkau, A. J.; Vedal, S.; Fisher, T. V.; Brauer, M. (2000) Exposure of chronic obstructive pulmonary
23             disease patients to paniculate matter: relationships between personal and ambient air concentrations. J. Air
24             Waste Manage. Assoc. 50: 1081-1094.
25       Esmen, N. A.; Hall, T. A. (2000) Theoretical investigation of the interrelationship between stationary and personal
26             sampling in exposure estimation. Appl. Occup. Environ. Hyg. 15:  114-119.
27       Evans, G. F.; Highsmith, R. V; Sheldon, L. S.; Suggs, J. C.; Williams, R. W.; Zweidinger, R. B.; Creason, J. P.;
28             Walsh, D.; Rodes, C. E.; Lawless, P. A.  (2000) The 1999 Fresno paniculate matter exposure studies:
29             comparison of community, outdoor, and residential PM mass measurements. J. Air Waste Manage. Assoc.
30             50: 1887-1896.
31       Ezzati, M.; Kammen, D. M. (2001) Quantifying the effects of exposure to indoor air pollution from biomass
32             combustion on acute respiratory infections in developing countries. Environ. Health Perspect.  109: 481-488.
33       Federal Register. (1986) Guidelines for estimating exposures. F. R. (September 24) 51: 34,042-34,054.
34       Ferris, B. G., Jr.; Speizer, F. E.; Spengler, J. D.; Dockery, D.; Bishop, Y. M. M.; Wolfson, M.; Humble, C. (1979)
3 5             Effects of sulfur oxides and respirable particles on human health: methodology and demography of
36             populations in study. Am. Rev. Respir. Dis. 120: 767-779.
37       Fischer, P. H.; Hoek, G.; VanReeuwijk, H.; Briggs, D. J.; Lebret,  E.; Van Wijnen, J. H.; Kingham, S.; Elliott, P.  E.
3 8             (2000) Traffic-related differences in outdoor and indoor concentrations of particles and volatile organic
39             compounds in Amsterdam. Atmos. Environ. 34: 3713-3722.
40       Fogh, C. L.; Byrne, M. A.; Roed, J.; Goddard, A. J. H. (1997) Size specific indoor aerosol deposition measurements
41             and derived I/O concentrations ratios. Atmos. Environ. 31: 2193-2203.
42       Freijer, J. I.; Bloemen, H. J. T.  (2000) Modeling relationships between indoor and outdoor air quality. J. Air Waste
43             Manage. Assoc. 50: 292-300.
44       Frey, H. C.; Rhodes, D. S. (1996) Characterizing, simulating and analyzing variability and uncertainty:
45             an illustration of methods using an air toxic emissions example. Hum. Ecol. Risk Assess. 2: 762-797.
46       Geller, M. D.; Chang, M.; Sioutas, C.; Ostro, B. D.; Lipsett, M. J.  (2002) Indoor/outdoor relationship and chemical
47             composition of fine and  coarse particles in the southern California deserts. Atmos. Environ. 36: 1099-1110.
48       Georgopoulos, P. G.; Lioy, P. J. (1994) Conceptual and theoretical aspects of human exposure and dose assessment.
49             J. Exposure Anal. Environ. Epidemiol. 4: 253-285.
50       Gladen, B.; Rogan, W. J. (1979) Misclassification and the design of environmental studies.  Am. J. Epidemiol.
51             109:607-616.
52       Glen,  G.; Lakkadi, Y.; Tippett, J. A.; del Valle-Torres, M. (1997) Development of NERL/CHAD: the National
53             Exposure Research Laboratory consolidated human activity database. Research Triangle Park, NC: U.S.
54             Environmental Protection Agency, Office of Research and Development; contract no. 68-D5-0049.
55       Goldstein, B. R.; Tardiff, G.; Hoffnagle, G.; Kester, R. (1992) Valdez health study: summary report. Ankorage, AK:
56             Alyeska Pipeline  Service Company.
         June 2003                                      5-130         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Goswami, E.; Larson, T.; Lumley, T.; Sally, Liu, L.-J. (2002) Spatial characteristics of fine paniculate matter:
  2             identifying representative monitoring locations in Seattle, Washington. J. Air Waste Manage. Assoc. 52: 324-
  3             333.
  4       Gotschi, T.; Oglesby, L.; Mathys, P.; Monn, C.; Manalis, N.; Koistinen, K.; Jantunen, M.; Hanninen, O.;
  5             Polanska, L.; Kunzli, N. (2002) Comparison of black smoke and PM2.5 levels in indoor and outdoor
  6             environments of four European cities. Environ. Sci.  Technol. 36: 1191-1197.
  7       Hadwen, G. E.; McCarthy, J. F.; Womble, S. E.; Girman, J. R.; Brightman, H. S. (1997) Volatile organic compound
  8             concentrations in 41 office buildings in the continental United States. In: Woods, J. E.; Grimsrud, D. T.;
  9             Boschi, N., eds. Proceedings of healthy buildings/indoor air quality international conference; September;
10             Bethesda, MD. Washington, DC: Healthy Buildings/IAQ. [Healthy Buildings - International Conference; v. 2,
11             pp. 465-470.]
12       Harrison, R. M.; Thornton, C. A.; Lawrence, R. G.; Mark,  D.; Kinnersley, R. P.; Ayres, J. G. (2002) Personal
13             exposure monitoring of paniculate matter, nitrogen  dioxide, and carbon monoxide, including susceptible
14             groups. Occup. Environ. Med. 59: 671-679.
15       Hartwell, T. D.; Clayton, C. A.; Michie, R. M., Jr.; Whitmore, R. W.; Zelon, H. S.; Whitehurst, D. A.; Akland, G. G.
16             (1984) Study of carbon monoxide exposures of residents of Washington, D.C. Research Triangle Park, NC:
17             U.S. Environmental Protection Agency, Environmental Monitoring Systems Laboratory; report no.
18             EPA-600/D-84-177. Available from: NTIS, Springfield,  VA; PB84-230051.
19       Hayes, S. R.; Marshall, J. R. (1999) Designing optimal strategies to attain the new U.S. paniculate matter standards:
20             some initial concepts. J. Air Waste Manage. Assoc.  49: PM192-PM198.
21       Henry, R. C.; Lewis, C. W.; Collins, J. F. (1994) Vehicle-related source compositions from ambient data:
22             the GRACE/SAFER method. Environ. Sci. Technol. 28: 823-832.
23       Hoek, G.; Fischer, P.; Van Den Brandt, P.; Goldbohm, S.;  Brunekreef, B. (2001) Estimation of long-term average
24             exposure to outdoor air pollution for a cohort study  on mortality. J. Exposure Anal. Environ. Epidemiol.
25             11:459-469.
26       Hoek, G.; Brunekreef, B.; Goldbohm, S.; Fischer, P.; Van  den Brandt, P.  A. (2002) Association between mortality
27             and indicators of traffic-related air pollution in the Netherlands: a cohort study. Lancet 360: 1203-1209.
28       Houseman, E. A.; Ryan, L.; Levy, J.; Richardson, D.; Spenger,  J. D. (2002) Autocorrelation in real-time continuous
29             monitoring of microenvironments. J. Appl. Stat. 29: 855-872.
30       Howard-Reed, C.; Rea, A. W.; Zufall, M. J.; Burke, J. M.;  Williams, R. W.; Suggs, J. C.; Sheldon, L. S.; Walsh, D.;
31             Kwock, R. (2000) Use of a continuous nephelometer to measure personal exposure to  particles during the
32             U.S. Environmental Protection Agency Baltimore and Fresno panel studies. J. Air Waste Manage. Assoc.
33             50: 1125-1132.
34       Howard-Reed, C.; Wallace, L. A.; Ott, W. R. (2002) The effect of opening windows on air change rates in two
35             homes. J. Air Waste Manage Assoc. 52: 147-159.
36       Institute for Social Research. (1997) Child development supplement to the panel study of income dynamics.
37             Ann Arbor, MI: University of Michigan. Available:  www.isr.umich.edu/src/child-development/home.html
38             [2001, January 31].
39       Isukapalli, S. S.; Georgopoulos, P. G. (2000) Modeling of  outdoor/indoor relationships of gas phase pollutants and
40             paniculate matter. Presented at: PM2000: paniculate matter and health—the scientific basis for regulatory
41             decision-making, specialty conference & exhibition; January; Charleston, SC. Pittsburgh, PA: Air & Waste
42             Management Association.
43       Jang, M.; Kamens, R. M. (1999) Newly characterized products and composition of secondary aerosols from the
44             reaction of a-pinene with ozone. Atmos. Environ. 33: 459-474.
45       Janssen, N. (1998) Personal exposure to airborne particles: validity of outdoor concentrations as a measure of
46             exposure in time series studies [dissertation]. Wageningen, Netherlands: Landbouwuniversiteit Wageningen.
47       Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1995) A relation between personal and ambient.
48             Epidemiology 6(suppl.): S45.
49       Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1997) Childhood exposure to PM10: relation between
50             personal, classroom, and outdoor concentrations. Occup. Environ.  Med. 54: 888-894.
51       Janssen, N. A. H.; Hoek, G.; Brunekreef, B.; Harssema, H.; Mensink, L; Zuidhof, A. (1998a) Personal sampling of
52             particles in adults: relation among personal, indoor,  and  outdoor air concentrations. Am. J. Epidemiol.
53             147: 537-547.
54       Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1998b) Personal sampling of airborne particles:
55             method performance and data quality. J. Exposure Anal. Environ. Epidemiol. 8: 37-49.
56       Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1999a) Personal exposure to fine particles in children
57             correlates closely with ambient fine particles. Arch.  Environ. Health  54: 95-101.


         June 2003                                      5-131         DRAFT-DO NOT QUOTE  OR CITE

-------
  1       Janssen, N. A. H.; Hoek, G.; Brunekreef, B.; Harssema, H. (1999b) Mass concentration and elemental composition
  2             of PM10 in classrooms. Occup. Environ. Med. 56: 482-487.
  3       Janssen, L. H. J. M; Buringh, E.; van der Meulen, A.; van den Hout, K. D. (1999c) A method to estimate the
  4             distribution of various fractions of PM10 in ambient air in the Netherlands. Atmos. Environ. 33: 3325-3334.
  5       Janssen, N. A. H.; De Hartog, J. J.; Hoek, G.; Brunekreef, B.; Lanki, T.; Timonen, K. L.; Pekkanen, J. (2000)
  6             Personal exposure to fine paniculate matter in elderly subjects: relation between personal, indoor, and outdoor
  7             concentrations. J. Air Waste Manage. Assoc. 50: 1133-1143.
  8       Jantunen, M. J.; Hanninen, O.; Katsouyanni, K.; Knoppel, H.; Kuenzli, N.; Lebret, E.; Maroni, M.; Saarela, K.;
  9             Sram, R.; Zmirou, D.  (1998) Air pollution exposure in European cities: the EXPOLIS study. J. Exposure
10             Anal. Environ.  Epidemiol. 8: 495-518.
11       Jantunen, M. J.; Kousa, A.; Hanninen, O.; Koistinen, K.;  Sram, R.; Lanki, T.; Lioy, P. J. (2000) Long term
12             population PM exposures correlate well with outdoor air levels - short term individual exposures do not.
13             Presented at: PM2000: paniculate matter and health—the scientific basis for regulatory decision-making,
14             specialty conference & exhibition; January; Charleston, SC. Pittsburgh, PA: Air & Waste Management
15             Association.
16       Jenkins, R. A.; Palausky, M. A.; Counts, R. W.; Guerin, M. R.; Dindal, A. B.; Bayne, C. K. (1996a) Determination
17             of personal exposure of non-smokers to environmental tobacco smoke in the United States. Lung Cancer
18             14(suppl. 1): S195-S213.
19       Jenkins, R. A.; Palausky, A.; Counts, R. W.; Bayne,  C. K.; Dindal, A. B.; Guerin, M. R. (1996b) Exposure to
20             environmental tobacco smoke in sixteen cities in the United States as determined by personal breathing zone
21             air sampling. J. Exposure Anal. Environ. Epidemiol. 6: 473-502.
22       Jinsart, W.; Tamura, K.; Loetkamonwit, S.; Thepanondh, S.; Karita, K.; Yano, E. (2002) Roadside paniculate air
23             pollution in Bangkok. J. Air Waste Manage. Assoc. 52: 1102-1110.
24       Johnson, T. (1984) A study of personal exposure to carbon monoxide in Denver, Colorado. Research Triangle Park,
25             NC: U.S. Environmental Protection Agency, Environmental Monitoring Systems Laboratory; report no.
26             EPA-600/4-84-014.
27       Johnson, T. (1989) Human activity patterns in Cincinnati, Ohio [final report]. Palo Alto, CA: Electric Power
28             Research Institute; project no. RP940-06; report no. EN-6204.
29       Johnson, T. R. (1995) Recent advances in the estimation of population exposure to mobile source pollutants.
30             J. Exposure Anal. Environ. Epidemiol. 5: 551-571.
31       Johnson, T.; Long, T.; Ollison, W. (2000) Prediction of hourly microenvironmental concentrations of fine  particles
32             based on measurements obtained from the Baltimore scripted activity  study. J. Exposure Anal. Environ.
33             Epidemiol. 10:403-411.
34       Jones, N. C.; Thornton, C. A.; Mark, D.; Harrison, R. M.  (2000) Indoor/outdoor relationships of paniculate matter in
35             domestic homes with roadside, urban and rural locations. Atmos. Environ. 34: 2603-2612.
36       Keeler, G. J.; Dvonch, J. T.; Yip, F. Y.; Parker, E. A.; Israel, B. A.; Marsik, F. J.; Morishita, M.; Barres, J. A.;
37             Robins, T. G.; Brakefield-Caldwell, W.; Sam, M. (2002) Assessment of personal and community-Level
3 8             exposures to  paniculate matter among children with asthma in Detroit, Michigan, as part of Community
3 9             Action Against Asthma (CAAA). Environ. Health Perspect. 110(suppl. 2): 173 -181
40       Keywood, M. D.; Ayers, G. P.; Gras, J. L.; Gillett, R. W.; Cohen, D. D. (1999) Relationships between size
41             segregated mass concentration data and ultrafine particle number concentrations in urban areas.
42             Atmos. Environ. 33: 2907-2913.
43       Kingham,  S.; Briggs, D.; Elliott, P.; Fischer, P.; Lebret, E. (2000) Spatial variations in the concentrations of
44             traffic-related pollutants in indoor and outdoor air  in Huddersfield, England. Atmos. Environ. 34: 905-916.
45       Kinney, P. L.; Chillrud, S. N.; Ramstrom, S.; Ross, J.; Spengler, J. D. (2002) Exposures to multiple air toxics in New
46             York City. Environ. Health Perspect. 110(suppl. 4): 539-546.
47       Kleeman, M. J.; Cass, G. R. (1998) Source contributions  to the size and composition distribution of urban  paniculate
48             air pollution. Atmos. Environ. 32: 2803-2816.
49       Klepeis, N. E. (1999) An introduction to the indirect exposure assessment approach: modeling human exposure
50             using microenvironmental measurements and  the recent national human activity pattern survey. Environ.
51             Health Perspect. Suppl. 107(2): 365-374.
52       Klepeis, N. E.; Ott,  W. R.; Switzer, P. (1994) A Total Human Exposure Model (THEM) for Respirable Suspended
53             Particles (RSP). Presented at:  87th annual meeting & exhibition of the Air & Waste Management Association;
54             June; Cincinnati, OH. Pittsburgh, PA: Air & Waste Management Association; paper no. 94-WA75A.03.
55             Available from: NTIS, Springfield, VA; PB94-197415.
56       Klepeis, N. E.; Ott,  W. R.; Switzer, P. (1995) Modeling the time series of respirable suspended particles and carbon
57             monoxide from multiple smokers: validation in two public smoking lounges. Presented at: 88th annual


         June 2003                                       5-132         DRAFT-DO NOT QUOTE OR CITE

-------
  1             meeting and exhibition of the Air & Waste Management Association; June; San Antonio, TX. Pittsburgh, PA:
  2             Air & Waste Management Association; paper no. 95-MPH.03.
  3       Klepeis, N. E.; Ott, W. R.; Switzer, P. (1996) A multiple-smoker model for predicting indoor air quality in public
  4             lounges. Environ. Sci. Technol. 30: 2813-2820.
  5       Klepeis, N. E.; Switzer, P.; Ott, W. R. (2000) Estimating exposure to environmental tobacco smoke particles using
  6             24-hour time-activity recall diaries. Presented at: PM2000: paniculate matter and health — the scientific basis
  7             for regulatory decision-making, specialty conference & exhibition; January; Charleston, SC. Pittsburgh, PA:
  8             Air & Waste Management Association.
  9       Koistinen, K. J.; Kousa, A.; Tenhola, V.; Hanninen, O.; Jantunen, M. J.; Oglesby, L.; Kuenzli, N.; Georgoulis, L.
10             (1999) Fine particle (PM25) measurement methodology, quality assurance procedures and pilot results of the
11             EXPOLIS study. J. Air Waste Manage. Assoc. 49: 1212-1220.
12       Koontz, M. D.; Rector, H. E. (1995) Estimation of distributions for residential air exchange rates: final report.
13             Washington, DC: U.S. Environmental Protection Agency, Office of Pollution Prevention and Toxics; contract
14             nos. 68-D9-0166 and 68-D3-0013.
15       Koontz, M. D.; Niang, L. L. (1998) California population indoor exposure model (CPIEM): version 1.4F. Final
16             report A933-157: user's guide. Sacramento, CA: California Air Resources Board; GEOMET report no.
17             IE-2632.
18       Korenstein, S.; Piazza, B. (2002) An exposure assessment of PM10 from a major highway interchange: are children
19             in nearby schools at risk? J. Environ. Health 65: 9-17.
20       Kousa, A.; Oglesby, L.; Koistinen, K.; Ktinzli, N. Jantunen, M. (2002) Exposure chain of urban air
21             PM2 ^-associations between ambient fixed site, residential outdoor, indoor, workplace and personal exposures
22             in four European cities in the EXPOLIS-study. Atmos. Environ. 36: 3031-3039.
23       Koutrakis, P.; Briggs, S. L. K.; Leaderer, B. P. (1992) Source apportionment of indoor aerosols in Suffolk and
24             Onondaga Counties, New York. Environ. Sci. Technol. 26: 521-527.
25       Krewski, D.; Burnett, R. T.; Goldberg, M. S.; Hoover, K.; Siemiatycki, J.; Jerrett, M.; Abrahamowicz, M.;
26             White, W. H.  (2000) Reanalysis of the Harvard Six Cities study and the American Cancer Society study of
27             paniculate air pollution and mortality: a special report of the Institute's Particle Epidemiology Reanalysis
28             Project.  Cambridge, MA: Health Effects Institute.
29       Lachenmyer, C.; Hidy, G. M. (2000) Urban measurements of outdoor-indoor PM2 5 concentrations and personal
30             exposure in the deep south. Part I. Pilot study of mass concentrations for nonsmoking subjects. Aerosol Sci.
31             Technol. 32: 34-51.
32       Laden, F.; Neas, L. M.; Dockery, D. W.; Schwartz, J. (2000) Association of fine paniculate matter from different
33             sources with daily mortality in six U.S. cities. Environ. Health Perspect. 108: 941-947.
34       Landis, M. S.; Norris, G. A.; Williams, R. W.; Weinstein, J. P. (2001) Personal exposures to PM2 5 mass and trace
35             elements in Baltimore, MD, USA. Atmos. Environ. 35: 6511-6524.
36       Lanki, T.; Aim, S.; Ruuskanen, J.; Janssen, N. A. H.; Jantunen, M.; Pekkanen, J. (2002) Photometrically measured
37             continuous personal PM2 5 exposure: levels and correlation to a gravimetric method. J. Exposure Anal.
38             Environ. Epidemiol. 12: 172-178.
39       Leaderer, B. P.; Naeher, L.; Jankun, T.; Balenger, K.; Holford, T. R.; Toth, C.; Sullivan, J.; Wolfson, J. M.;
40             Koutrakis, P. (1999) Indoor, outdoor,  and regional summer and winter concentrations of PM10, PM2 5, SO42",
41             H+, NH4+, NO3", NH3, and nitrous acid in homes with and without kerosene space heaters. Environ. Health
42             Perspect. 107:223-231.
43       Lee, S. C.; Chang, M. (1999) Indoor air quality investigations at five classrooms. Indoor Air 9: 134-138.
44       Lena, T. S.; Ochieng, V.; Carter, M.; Holguin-Veras, J.; Kinney,  P. L. (2002) Elemental carbon and PM25 levels in
45             an urban community heavily impacted by truck traffic. Environ. Health Perspect. 110:  1009-1015.
46       Lessler, J. T.; Kalsbeek, W. D. (1992) Nonsampling error in surveys. New York, NY: John Wiley & Sons, Inc.; p. 7.
47       Levy, J. I.; Dumyahn, T.; Spengler, J. D. (2002) Paniculate matter and polycyclic aromatic hydrocarbon
48             concentrations in indoor and outdoor microenvironments in Boston, Massachusetts.  J. Exposure Anal.
49             Environ. Epidemiol. 12: 104-114.
50       Liao, D.; Creason, J.; Shy, C.; Williams,  R.;  Watts, R.; Zweidinger, R. (1999) Daily variation of paniculate air
51             pollution and poor cardiac autonomic control in the elderly. Environ. Health Perspect. 107: 521-525.
52       Ligocki, M. P.; Liu, H. I. H.; Cass, G. R.; John, W. (1990) Measurements of particle deposition rates inside Southern
53             California museums. Aerosol Sci.  Technol. 13: 85-101.
54       Lillquist, D. R.; Lee, J. S.; Ramsay, J. R.; Boucher, K. M.; Walton, Z. L.; Lyon, J. L. (1998) A comparison of
5 5             indoor/outdoor PM10 concentrations measured at three hospitals and a centrally located monitor in Utah. Appl.
56             Occup. Environ. Hyg. 13: 409-415.
         June 2003                                      5-133         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Linn, W. S.; Gong, H., Jr.; Clark, K. W.; Anderson, K. R. (1999) Day-to-day paniculate exposures and health
  2             changes in Los Angeles area residents with severe lung disease. J. Air Waste Manage. Assoc.
  3             49:PM108-PM115.
  4       Lioy, P. J. (1990) Assessing total human exposure to contaminants: a multidisciplinary approach. Environ. Sci.
  5             Technol. 24: 938-945.
  6       Lioy, P. J.; Waldman, J. M; Buckley, T.; Butler, J.; Pietarinen, C. (1990) The personal, indoor and outdoor
  7             concentrations of PM-10 measured in an industrial community during the winter. Atmos. Environ. Part B
  8             24:57-66.
  9       Lioy, P. J.; Wainman, T.; Zhang, J.; Goldsmith, S. (1999) Typical household vacuum cleaners: the collection
10             efficiency and emissions characteristics for fine particles. J. Air Waste Manage. Assoc. 49: 200-206.
11       Lippmann, M.; Ito, K.; Nadas, A.; Burnett, R. T. (2000) Association of paniculate matter components with daily
12             mortality and morbidity in urban populations. Cambridge, MA: Health Effects Institute; research report
13             no. 95.
14       Liu, D.-L.; Nazaroff, W. W. (2001) Modeling pollutant penetration across building envelopes. Atmos. Environ.
15             35:4451-4462.
16       Long, C. M.; Suh, H. H.; Koutrakis, P. (2000) Characterization of indoor particle sources using continuous mass and
17             size monitors. J. Air Waste Manage. Assoc. 50: 1236-1250.
18       Long, C. M.; Suh, H. H.; Catalano, P. J.; Koutrakis, P. (2001a) Using time- and size-resolved paniculate data to
19             quantify indoor penetration and deposition behavior. Environ. Sci. Technol. 35:  2089-2099.
20       Long, C. M.; Suh, H. H.; Kobzik, L.; Catalano, P. J.; Ning, Y. Y.; Koutrakis, P. (200Ib) A pilot investigation of the
21             relative toxicity of indoor and outdoor fine particles: in-vitro effects of endotoxin and other paniculate
22             properties. Environ. Health Perspect. 109: 1019-1026.
23       Lurmann, F. W.; Korc, M. E. (1994) User's guide to the regional human exposure (REHEX) model. San Francisco,
24             CA: Bay Area Air Quality Management District; draft report no. STI-93150-1414-DR.
25       Macintosh, D. L.; Xue, J.; Ozkaynak, H.; Spengler, J. D.; Ryan, P. B. (1995) A population-based exposure model for
26             benzene. J. Exposure Anal. Environ. Epidemiol. 5: 375-403.
27       Magari, S. R.; Schwartz, J.; Williams, P. L.; Hauser, R.; Smith, T. J.; Christiani, D. C. (2002) The association
28             between personal measurements of environmental exposure to particulates and heart rate vanability.
29             Epidemiology 13: 305-310.
30       Mage, D. T. (1985) Concepts of human exposure assessment for airborne paniculate matter. Environ. Int.
31             11:407-412.
32       Mage, D. T. (1998) How much PM of ambient origin are people exposed to? In: Measurement of toxic and related
33             air pollutants. Volume II: proceedings of a specialty conference; September; Gary, NC. Pittsburgh, PA: Air&
34             Waste Management Association; pp. 666-681. (A&WMA publication VTP-85).
3 5       Mage, D. T.; Ott, W. R. (1996) Accounting for nonuniform mixing and human exposure in indoor environments.
36             In: Tichenor, B. A., ed. Characterizing sources of indoor air pollution and related sink effects. Philadelphia,
37             PA: American Society for Testing and Materials; pp. 263-278.
3 8       Mage, D.; Wilson, W.; Hasselblad, V.; Grant, L. (1999) Assessment of human exposure to ambient paniculate
39             matter. J. Air Waste Manage. Assoc. 49: 174-185.
40       Mar, T. F.; Norris, G. A.; Koenig, J. Q.; Larson, T. V. (2000) Associations between air pollution and mortality in
41             Phoenix, 1995-1997. Environ. Health Perspect. 108:  347-353.
42       McBride, S. J.; Ferro, A. R.; Ott, W. R; Switzer, P.; Hildemann, L. M. (1999) Investigations of the proximity effect
43             for pollutants in the indoor environment. J. Exposure Anal. Environ.  Epidemiol. 9: 602-621.
44       McCurdy, T. (1995) Estimating human exposure to selected motor vehicle pollutants using the NEM series of
45             models:  lessons to be learned. J. Exposure Anal. Environ. Epidemiol. 5: 533-550.
46       McCurdy, T. (1997) Modeling the dose profile in human exposure assessments: ozone  as an example. Rev. Toxicol.
47             1:3-23.
48       McCurdy, T. (2000) Conceptual basis for multi-route intake dose modeling using an energy expenditure approach.
49             J. Exposure Anal. Environ. Epidemiol.  10: 86-97.
50       McCurdy, T.; Glen, G.; Smith, L.; Lakkadi, Y. (2000) The National Exposure Research Laboratory's Consolidated
51             Human Activity Database. J. Exposure Anal. Environ. Epidemiol. 10: 566-578.
52       Mihlan, G. J.; Todd, L. A.; Truong, K. N. (2000) Assessment of occupational exposure patterns by
53             frequency-domain analysis of time sertes data. Appl. Occup. Environ. Hyg. 15:  120-130.
54       Miller, S. L.; Branoff, S.; Nazaroff, W. W. (1998) Exposure to toxic air contaminants in environmental tobacco
55             smoke: an assessment for California based on personal monitoring data. J. Exposure Anal. Environ.
56             Epidemiol. 8:287-311.
         June 2003                                      5-134         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Monn, C. (2001) Exposure assessment of air pollutants: a review on spatial heterogeneity and
  2             indoor/outdoor/personal exposure to suspended paniculate matter, nitrogen dioxide and ozone.
  3             Atmos. Environ. 35:  1-32.
  4       Monn, Ch.; Carabias, V.; Junker, M; Waeber, R.; Karrer, M; Wanner, H.  U. (1997) Small-scale spatial variability
  5             of paniculate matter <10 \im (PM10) and nitrogen dioxide. Atmos. Environ. 31: 2243-2247.
  6       Morawska, L.; He, C.; Hitchins, I; Gilbert, D.; Parappukkaran, S. (2001) The relationship between indoor and
  7             outdoor airborne particles in the residential environment. Atmos. Environ. 35: 3463-3473.
  8       Murray, D. M.; Burmaster, D. E.  (1995) Residential air exchange rates in the United States: empirical and estimated
  9             parametric distributions by season and climatic region. Risk Anal. 15: 459-465.
10       National Air Pollution Control Administration. (1969) Air quality criteria for paniculate matter. Washington, DC:
11             U.S. Department of Health, Education, and Welfare, Public Health Service; NAPCA publication no. AP-49.
12             Available from: NTIS, Springfield,  VA; PB-190251/BA.
13       National Research Council. (1991) Human exposure assessment for airborne pollutants: advances and opportunities.
14             Washington, DC: National Academy of Sciences.
15       National Research Council. (1998) Research priorities for airborne paniculate matter. I. Immediate priorities and a
16             long-range research portfolio. Washington, DC: National Academy  Press.
17       National Research Council. (1999) Research priorities for airborne paniculate matter. II. Evaluating research
18             progress and updating the portfolio. Washington, DC: National Academy Press.
19       Naumova, Y. Y.; Eisenreich, S. I; Turpin, B. I; Weisel, C. P.; Morandi, M. T.; Colome, S. D.; Totten, L. A.;  Stock,
20             T. H.; Winer, A. M.; Alimokhtari, S.; Kwon, I; Shendell, D.; Jones, J.; Maberti, S.; Wall, S. J. (2002)
21             Polycyclic aromatic hydrocarbons in the indoor and outdoor air of three cities in the U.S. Environ. Sci.
22             Technol.  36: 2552-2559.
23       Navidi, W.; Thomas, D.; Langholz, B.; Stram, D. (1999) Statistical methods for epidemiologic studies of the health
24             effects of air pollution. Cambridge, MA: Health Effects Institute; research report no.  86.
25       Nelson,  P. R.; Conrad, F. W.; Kelly, S. P.; Maiolo, K. C.; Richardson, J. D.; Ogden, M. W.  (1998) Composition of
26             environmental tobacco smoke (ETS) from international cigarettes part II: nine country follow-up. Environ.
27             Int. 24:251-257.
28       Nieuwenhuijsen, M. J.; Noderer, K. S.; Schenker, M. B.; Vallyathan, V.; Olenchock, S. (1999) Personal exposure to
29             dust, endotoxin and crystalline silica in California agriculture. Ann.  Occup. Hyg. 43: 35-42.
30       Oglesby, L.; Ktinzli, N.; Roosli, M.; Braun-Fahrlander, C.; Mathys, P.; Stern, W.; Jantunen, M.; Kousa, A. (2000)
31             Validity of ambient levels  of fine particles as surrogate for personal exposure to outdoor air pollution—results
32             of the European EXPOLIS-EAS study (Swiss Center Basel). J. Air Waste Manage. Assoc. 50: 1251-1261.
33       Ostro, B. D.; Hurley, S.; Lipsett, M. J. (1999) Air pollution and daily mortality in the Coachella Valley, California:
34             a study of PM10 dominated by coarse particles. Environ. Res. 81: 231-238.
35       Ostro, B. D.; Broadwin, R.; Lipsett, M. J. (2000) Coarse and fine particles  and daily mortality in the Coachella
36             Valley, California: a follow-up study.  J. Exposure Anal. Environ. Epidemiol. 10: 412-419.
37       Ott, W. R. (1984) Exposure estimates based on computer generated activity patterns. J. Toxicol. Clin. Toxicol.
38             21:97-128.
39       Ott, W. R. (1999) Mathematical models for predicting indoor air quality from smoking activity. Environ. Health
40             Perspect. 107(suppl. 2): 375-381.
41       Ott, W.; Thomas, J.; Mage, D.; Wallace, L. (1988) Validation of the simulation of human activity and pollutant
42             exposure (SHAPE) model using paired days from the Denver, CO, carbon monoxide field study. Atmos.
43             Environ. 22:2101-2113.
44       Ott, W. R.; Mage, D. T.; Thomas, J. (1992) Comparison of microenvironmental CO concentrations in two cities for
45             human exposure modeling. J. Exposure Anal. Environ. Epidemiol. 2: 249-267.
46       Ott, W. R.; Klepeis, N. E.; Switzer, P. (1995) Modeling environmental tobacco smoke in the home using transfer
47             functions. Presented at: 88th annual meeting and exhibition of the Air & Waste Management Association;
48             June; San Antonio, TX; paper no. 95-WP84B.03. Research Triangle Park, NC: U.S. Environmental Protection
49             Agency, Atmospheric Research And Exposure Assessment Laboratory; report no. EPA/600/A-95/079.
50             Available from: NTIS, Springfield,  VA; PB95-225488.
51       Ott, W.; Switzer, P.; Robinson, J. (1996) Particle concentrations inside a tavern before and after prohibition of
52             smoking: evaluating the performance of an indoor air quality model. J. Air Waste Manage. Assoc.
53             46:  1120-1134.
54       Ott, W.; Wallace, L.; Mage, D. (2000) Predicting paniculate  (PM10) personal exposure distributions using a random
55             component superposition (RCS) statistical model. J. Air Waste Manage. Assoc. 50: 1390-1406.
56       Owen, M.  K.; Ensor, D. S.;  Sparks, L. E. (1992) Airborne particle sizes and sources found in indoor air.
57             Atmos. Environ. Part A 26: 2149-2162.


         June 2003                                       5-135         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Ozkaynak, H.; Spengler, J. (1996) The role of outdoor paniculate matter in assessing total human exposure.
  2             In: Wilson, R.; Spengler, J., eds. Particles in our air: concentrations and health effects. Cambridge, MA:
  3             Harvard University Press.
  4       Ozkaynak, H.; Thurston, G. D. (1987) Associations between 1980 U.S. mortality rates and alternative measures of
  5             airborne particle concentration. Risk Anal. 7: 449-461.
  6       Ozkaynak, H.; Ryan, P.  B.; Spengler, J. D.; Laird, N. M. (1986) Bias due to misclassification of personal exposures
  7             in epidemiologic  studies of indoor and outdoor air pollution. In: Berglund, B.; Berglund, U.; Lindvall, T.;
  8             Spengler, J.; Sundell, J., eds. Indoor air quality: papers from the third international conference on indoor air
  9             quality and climate; August 1984; Stockholm, Sweden. Environ. Int. 12: 389-393.
10       Ozkaynak, H.; Spengler, J. D.; Ludwig, J. F.; Butler, D. A.; Clayton, C. A.; Pellizzari, E.; Wiener, R. W. (1990)
11             Personal exposure to paniculate matter: findings from the Particle Total Exposure Assessment Methodology
12             (PTEAM) prepilot study. In: Indoor air '90: precedings of the 5th international conference on indoor air
13             quality and climate, volume 2, characteristics of indoor air; July-August; Toronto, ON, Canada. Ottawa, ON,
14             Canada: International Conference on Indoor Air Quality and Climate, Inc.; pp. 571-576.
15       Ozkaynak, H.; Spengler, J. D.; Xue, J.; Koutrakis, P.; Pellizzari, E. D.; Wallace, L. (1993) Sources and factors
16             influencing personal and indoor exposures to particles, elements and nicotine:  findings from the particle
17             TEAM pilot study.  In: Jantunen, M.; Kalliokoski, P.; Kukkonen, E.; Saarela, K.; Seppanen, O.; Vuorelma, H.,
18             eds. Indoor air '93: proceedings of the 6th international conference on indoor air quality and climate, v. 3,
19             combustion products, risk assessment, policies; July; Helsinki, Finland. Helsinki, Finland: Indoor Air '93;
20             pp. 457-462.
21       Ozkaynak, H.; Xue, J.; Spengler, J.; Wallace, L.; Pellizzari, E.; Jenkins, P. (1996a) Personal exposures to airborne
22             particles and metals: results from the particle  TEAM study in Riverside, California. J. Exposure Anal.
23             Environ. Epidemiol. 6: 57-78.
24       Ozkaynak, H.; Xue, J.; Weker, R.; Butler, D.; Koutrakis, P.; Spengler, J. (1996b) The particle TEAM (PTEAM)
25             study: analysis of the data: final report, volume III. Research Triangle Park, NC: U.S. Environmental
26             Protection Agency, Atmospheric Research and Exposure Assessment Laboratory; report no.
27             EPA/600/R-95/098. Available from: NTIS, Springfield, VA; PB97-102495.
28       Patterson, E.; Eatough, D.  J. (2000) Indoor/outdoor relationships for ambient PM25 and associated pollutants:
29             epidemiological implications in Lindon, Utah. J. Air Waste Manage. Assoc. 50: 103-110.
30       Pellizzari, E. D., Mason, R. E.; Clayton, C. A.; Thomas, K. W.; Cooper, S.; Piper. L.; Rodes, C.; Goldberg, M.;
31             Roberds, J.; Michael, L. (1998) Manganese exposure study (Toronto). Final report. Research Triangle Park,
32             NC: Research Triangle Institute, Analytical and Chemical Sciences; report no. RTI/6312/02-01 F.
33       Pellizzari, E. D.; Clayton,  C. A.;  Rodes, C. E.; Mason, R. E.; Piper, L. L.; Fort, B.; Pfeifer, G.; Lynam, D. (1999)
34             Paniculate matter and manganese exposures in Toronto, Canada. Atmos. Environ. 33: 721-734.
35       Pellizzari, E. D.; Clayton,  C. A.;  Rodes, C. E.; Mason, R. E.; Piper, L. L.; Fort, B.; Pfeifer, G.; Lynam, D. (2001)
36             Paniculate matter and manganese exposures in Indianapolis, Indiana. J. Exposure Anal. Environ. Epidemiol.
37             11:423-440.
38       Phillips, K.; Howard, D. A.; Browne, D.; Lewsley, J. M. (1994) Assessment of personal exposures to environmental
39             tobacco smoke in British nonsmokers. Environ. Int. 20: 693-712.
40       Phillips, K.; Bentley, M. C.; Howard, D. A.; Alvan,  G. (1996) Assessment of air quality in Stockholm by personal
41             monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke. Scand. J.
42             Work Environ. Health 22(suppl. 1): 1-24.
43       Phillips, K.; Bentley, M. C.; Howard, D. A.; Alvan,  G.; Huici, A. (1997a) Assessment of air quality in Barcelona by
44             personal monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke.
45             Environ. Int. 23:  173-196.
46       Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan,  G. (1997b) Assessment of air quality in Turin by personal
47             monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke. Environ. Int.
48             23:851-871.
49       Phillips, K.; Bentley, M. C.; Howard, D. A.; Alvan,  G. (1998a) Assessment of air quality in Paris by personal
50             monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke. Environ. Int.
51             24:405-425.
52       Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan,  G. (1998b) Assessment of environmental tobacco smoke and
53             respirable suspended particle exposures for nonsmokers in Lisbon by personal monitoring. Environ. Int.
54             24:301-324.
55       Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan,  G. (1999) Assessment of environmental tobacco smoke and
56             respirable suspended particle exposures for nonsmokers in Basel by personal monitoring. Atmos. Environ.
57             33: 1889-1904.


         June 2003                                      5-136          DRAFT-DO NOT QUOTE OR CITE

-------
  1       Pope, C. A., Ill; Thun, M. I; Namboodiri, M. M; Dockery, D. W.; Evans, J. S.; Speizer, F. E.; Heath, C. W., Jr.
  2             (1995) Paniculate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am. J. Respir.
  3             Crit. Care Med. 151: 669-674.
  4       Pope, C. A., Ill; Hill, R. W.; Villegas, G. M. (1999) Paniculate air pollution and daily mortality on Utah's Wasatch
  5             Front. Environ. Health Perspect: 107: 567-573.
  6       Praml, G.; Schierl, R. (2000) Dust exposure in Munich public transportation: a comprehensive 4-year survey in
  7             buses and trams. Int. Arch. Occup. Environ. Health 73: 209-214.
  8       Quintana, P. J. E.; Samimi, B. S.; Kleinman, M. T.; Liu, L.-J.; Soto, K.; Warner, G. Y.; Bufalino, C.; Valencia, J.;
  9             Francis, D.; Hovell, M. H.; Delfino, R. J. (2000) Evaluation of a real-time passive personal particle monitor in
10             fixed site residential indoor and ambient measurements. J. Exposure Anal. Environ. Epidemiol. 10: 437-445.
11       Ramachandran,  G.; Vincent, J. H. (1999) A Bayesian approach to retrospective exposure assessment. Appl. Occup.
12             Environ. Hyg. 14: 547-557.
13       Rea, A. W.; Zufall, M. J.; Williams, R. W.; Sheldon, L. S.; Howard-Reed, C. (2001) The influence of human activity
14             patterns on personal PM exposure: a comparative analysis of filter-based and continuous particle
15             measurements. J. Air Waste Manage. Assoc. 51: 1271-1279.
16       Ren, P.; Jankun, T. M.; Leaderer, B. P. (1999) Comparisons of seasonal fungal prevalence in indoor and outdoor air
17             and in house dusts of dwellings in one northeast American county. J. Exposure Anal. Environ. Epidemiol.
18             9:560-568.
19       Riediker, MM.; Keller, S.; Wuthrich, B.; Koller, T.; Monn, C. (2000) Personal pollen exposure compared to
20             stationary measurements. J. Invest. Allergol. Clin. Immunol. 10: 200-203.
21       Riley, W. J.; McKone, T. E.; Lai, A. C.  K.; Nazaroff, W. W. (2002) Indoor paniculate matter of outdoor origin:
22             importance of size-dependent removal mechanisms. Environ. Sci. Technol. 36: 200-207.
23       Robinson, J. P.;  Wiley, J. A.; Piazza, T.; Garrett, K.; Cirksena, K. (1991) Activity patterns of California residents
24             and their implications for potential exposure to pollution  (final report). Sacramento,  CA: California Air
25             Resources Board report no. CARB-A6-177-33.
26       Robinson, J. P.;  Switzer, P.; Ott, W. (1994) Microenvironmental factors related to Californians' potential exposure to
27             environmental tobacco smoke (ETS). Research Triangle Park, NC: U.S. Environmental Protection Agency,
28             Atmospheric Research and Exposure Assessment Laboratory; report no. EPA/600/R-94/116. Available from:
29             NTIS, Springfield, VA;  PB94-214814.
30       Rodes, C.; Sheldon, L.; Whitaker, D.; Clayton, A.; Fitzgerald, K.; Flanagan, J.; DiGenova, F.; Hering, S.; Frazier, C.
31             (1998) Measuring concentrations of selected air pollutants inside California vehicles [final report].
32             Sacramento, CA: California Environmental Protection Agency, Air Resources Board; contract no. 95-339.
33       Rodes, C. E.; Lawless, P. A.; Evans, G.  F.; Sheldon, L. S.; Williams, R. W.; Vette, A. F.; Creason, J. P.;  Walsh, D.
34             (2001) The relationships between personal PM exposures for elderly populations and indoor and outdoor
35             concentrations for three retirement center scenarios. J. Exposure Anal. Environ. Epidemiol. 11: 103-115.
36       Rojas-Bracho, L.; Suh, H. H.; Koutrakis, P. (2000) Relationships among personal, indoor, and outdoor fine and
37             coarse particle concentrations for individuals with COPD. J. Exposure Anal. Environ. Epidemiol.
38             10:294-306.
39       Rojas-Bracho, L.; Suh, H. H.; Oyola, P.; Koutrakis, P. (2002) Measurements of children's exposures to particles and
40             nitrogen dioxide in Santiago, Chile. Sci. Total Environ. 287: 249-264.
41       Roorda-Knape, M. C.; Janssen, N. A. H.; De Hartog, J. J.; Van Vliet, P. H. N.; Harssema, H.; Brunekreef, B. (1998)
42             Air pollution from traffic in city districts near major motorways. Atmos. Environ. 32: 1921-1930.
43       Rotko, T.; Oglesby, L.; Ktinzli, N.; Jantunen, M. J. (2000a) Population sampling in European air pollution exposure
44             study, EXPOLIS: comparisons between the cities and representativeness of the samples. J. Exposure Anal.
45             Environ. Epidemiol. 10: 355-364.
46       Rotko, T.; Koistinen, K.; Hanninen, O.;  Jantunen, M. (2000b) Sociodemographic descriptors  of personal exposure to
47             fine particles (PM25) in EXPOLIS Helsinki. J. Exposure Anal. Environ. Epidemiol. 10: 385-393.
48       Ryan, P. B. (1991) An overview of human exposure modeling. J. Exposure Anal. Environ. Epidemiol. 1: 453-474.
49       Samet, J. M.; Dominici, F.; Zeger, S. L.; Schwartz, J.; Dockery, D. W. (2000) National morbidity, mortality, and air
50             pollution study. Part I: methods and methodologic issues. Cambridge, MA: Health Effects Institute;  research
51             report no. 94.
52       Santos-Burgoa, C.; Rojas-Bracho, L.; Rosas-Perez, I.; Ramirez-Sanchez, A.; Sanchez-Rico, G.; Mejia-Hernandez, S.
53             (1998) Modelaje de exposicion a particulas en poblacion general y riesgo de enfermedad respiratoria [Particle
54             exposure modeling in the general population and risk of respiratory disease]. Gac. Med. Mex.  134: 407-417.
55       Sarnat, J. A.; Koutrakis, P.; Suh, H. H. (2000) Assessing the relationship between personal paniculate and gaseous
56             exposures of senior citizens living in Baltimore, MD. J. Air Waste Manage. Assoc. 50: 1184-1198.
         June 2003                                      5-137        DRAFT-DO NOT QUOTE OR CITE

-------
  1       Sarnat, J. A.; Schwartz, I; Catalano, P. I; Suh, H. H. (2001) Gaseous pollutants in paniculate matter epidemiology:
  2             confounders or surrogates? Environ. Health Perspect.  109: 1053-1061.
  3       Sarnat, J. A.; Long, C. M; Koutrakis, P.; Coull, B. A.; Schwartz, J.; Suh, H. H. (2002) Using sulfur as a tracer of
  4             outdoor fine particulate matter. Environ. Sci. Technol. 36: 5305-5314.
  5       Schwartz, J.; Dockery, D. W.; Neas, L. M. (1996) Is daily mortality associated specifically with fine particles? J. Air
  6             Waste Manage. Assoc. 46: 927-939.
  7       Sexton, K.; Ryan, P. B. (1988) Assessment of human exposure to air pollution: methods, measurements, and models.
  8             In: Watson, A. Y.; Bates, R. R.; Kennedy, D., eds. Air pollution, the automobile,  and public health.
  9             Washington, DC: National Academy Press; pp. 207-238.
10       Sexton, K.; Spengler, J. D.; Treitman, R. D. (1984) Personal exposure to respirable particles: a case study in
11             Waterbury, Vermont. Atmos. Environ. 18:  1385-1398.
12       Sheldon, L.; Clayton, A.; Jones, B.; Keever, J.; Perritt, R.; Smith, D.; Whitaker, D.; Whitmore, R. (1992a) Indoor
13             pollutant concentrations and exposures. Sacramento, CA: California Air Resources Board; report no.
14             ARB/R-92/486.
15       Sheldon, L.; Clayton, A.; Keever, J. Perritt, R.; Whitaker, D.  (1992b) PTEAM: monitoring of phthalates and PAHs
16             in indoor and outdoor air samples in Riverside, California, volume II [final report]. Sacramento, CA:
17             California Environmental Protection Agency, State Air Resources Board, Research Division; contract no.
18             A933-144. Available from: NTIS, Springfield, VA; PB93-205649/XAB.
19       Sheldon, L.; Whitaker, D.; Keever, J.; Clayton, A.; Perritt, R. (1993a) Phthalates and PAHs in indoor and outdoor air
20             in a southern California community. In: Jantunen, M.; Kalliokoski, P.; Kukkonen, E.;  Saarela, K.;  Seppanen,
21             O.; Vuorelma, H., eds. Indoor air '93: proceedings of the 6th international conference on indoor air quality
22             and climate. Volume 3: combustion products, risk assessment, policies; July; Helsinki, Finland.  Helsinki,
23             Finland: Indoor Air '93; pp. 109-114.
24       Sheldon, L.; Clayton, A.; Keever, J.; Perritt, R.; Whitaker, D. (1993b) Indoor concentrations  of polycyclic aromatic
25             hydrocarbons in California residences. Sacramento, CA: Air Resources Board; ARE contract no. A033-132.
26       Sheldon, L.; Clayton, A.; Perritt, R.; Whitaker, D. A.; Keever, J. (1993c) Indoor concentrations of polycyclic
27             aromatic hydrocarbons in California residences and their relationship to combustion source use. In: Jantunen,
28             M.; Kalliokoski, P.; Kukkonen, E.; Saarela, K.; Seppanen, O.; Vuorelma, H., eds. Indoor air '93: proceedings
29             of the 6th international conference on indoor air quality and climate. Volume 3: combustion products, risk
30             assessment, policies; July; Helsinki, Finland. Helsinki, Finland: Indoor Air '93; pp. 29-34.
31       Shy, C. M.; Kleinbaum, D. G.; Morgenstern, H. (1978) The effect of misclassification of exposure status in
32             epidemiological studies of air pollution health effects.  Bull. N. Y. Acad. Med. 54: 1155-1165.
33       Siegmann, K.;  Schemer, L.;  Siegmann, H. C. (1999) Physical and chemical properties of airborne nanoscale particles
34             and how to measure the impact on human health. J. Mol. Struct. (Theochem) 458: 191-201.
35       Sinclair, J. D.; Psota-Kelty, L. A.; Weschler, C. J. (1988) Indoor/outdoor ratios and indoor surface accumulations of
36             ionic substances at Newark, New Jersey. Atmos. Environ. 22: 461-469.
37       Sinclair, J. D.; Psota-Kelty, L. A.; Weschler, C. J.; Shields, H. C. (1990) Measurement and modeling of airborne
3 8             concentrations and indoor surface accumulation rates of ionic substances at Neenah, Wisconsin. Atmos.
39             Environ. Part A 24: 627-638.
40       Serensen, M.; Daneshvar, B.; Hansen, M.; Dragsted, L. O.; Hertel, O.; Knudsen, L.; Loft, S.  (2003) Personal PM25
41             exposure and markers of oxidative stress in blood. Environ. Health Perspect. Ill: 161-165.
42       Spengler, J. D.; Treitman, R. D.; Tosteson,  T. D.; Mage, D. T.; Soczek, M. L. (1985) Personal exposures to
43             respirable particulates and implications for air pollution epidemiology. Environ. Sci. Technol. 19: 700-707.
44       Spengler, J. D.; Ozkaynak, H.; Ludwig, J.; Allen,  G.; Pellizzari,  E. D.; Wiener, R.  (1989) Personal exposures to
45             particulate matter: instruments and methodologies P-TEAM. In: Measurement of toxic and related air
46             pollutants: proceedings of the 1989 U.S. EPA/A&WMA international symposium; May; Raleigh, NC.
47             Pittsburgh, PA: Air & Waste Management Association; pp. 449-463. (A&WMA publication VTP-13).
48       Spier, C. E.; Little, D. E.; Trim, S. C.; Johnson, T. R.; Linn, W. S.; Hackney, J. D. (1992) Activity patterns in
49             elementary and high school students exposed to oxidant pollution. J. Exposure Anal. Environ. Epidemiol.
50             2:277-293.
51       Suggs, J. C.; Burton, R. M. (1983) Spatial characteristics of inhalable particles in the Philadelphia metropolitan area.
52             J. AirPollut. Control Assoc. 33: 688-691.
53       Suh, H. H.; Spengler, J. D.; Koutrakis, P. (1992) Personal exposures to acid aerosols and ammonia. Environ. Sci.
54             Technol. 26:2507-2517.
55       Suh, H. H.; Koutrakis, P.; Spengler, J. D. (1993a) Indoor and outdoor acid aerosol and gas concentrations.
56             In: Jantunen, M.; Kalliokoski, P.; Kukkonen, E.; Saarela,  K.; Seppanen, O.; Vuorelma, H., eds. Indoor air
         June 2003                                      5-138         DRAFT-DO NOT QUOTE OR CITE

-------
  1             '93: proceedings of the 6th international conference on indoor air quality and climate. Volume 3: combustion
  2             products, risk assessment, policies; July; Helsinki, Finland. Helsinki, Finland: Indoor Air '93; pp. 257-262.
  3       Suh, H. H.; Koutrakis, P.; Spengler, J. D. (1993b) Validation of personal exposure models for sulfate and aerosol
  4             strong acidity. J. Air Waste Manage. Assoc. 43: 845-850.
  5       Suh, H. H.; Koutrakis, P.; Spengler, J. D. (1994) The relationship between airborne acidity and ammonia in indoor
  6             environments. J. Exposure Anal. Environ. Epidemiol. 4: 1-23.
  7       Suh, H. H.; Allen, G. A.; Koutrakis, P.; Burton, R. M. (1995) Spatial variation in acidic sulfate and ammonia
  8             concentrations within metropolitan Philadelphia. J. Air Waste Manage. Assoc. 45: 442-452.
  9       Tamura, K.; Ando, M.; Sagai, M.; Matsumoto, Y. (1996a) Estimation of levels of personal exposure to suspended
10             paniculate matter and nitrogen dioxide in Tokyo. Environ. Sci. (Tokyo) 4: 37-51.
11       Tamura, K.; Ando, M.; Miyazaki,  T.; Funasaka, K.; Kuroda, K. (1996b) Indoor concentrations and personal
12             exposure to paniculate matter at roadside houses. In: Indoor air '96: the 7th international conference on indoor
13             air quality, volume I; July; Nagoya, Japan. Tokyo, Japan: Institute of Public Health; pp. 599-602.
14       Teschke, K.; Marion, S. A.; Vaughan, T. L.; Morgan, M. S.; Camp, J. (1999) Exposures to wood dust in U.S.
15             industries and occupations, 1979 to 1997. Am. J. Ind. Med. 35: 581-589.
16       Thatcher, T. L.; Layton, D.  W. (1995) Deposition, resuspension, and penetration of particles within a residence.
17             Atmos. Environ. 29:  1487-1497.
18       Thatcher, T. L.; Lai, A. C. K.; Moreno-Jackson, R.; Sextro, R.  G.; Nazaroff, W. W. (2002) Effects of room
19             furnishings and air speed on particle deposition rates indoors. Atmos. Environ. 36: 1811-1819.
20       Thomas, K. W.; Pellizzari, E. D.; Clayton, C.  A.; Whitaker, D. A.; Shores, R. C.; Spengler, J.; Ozkaynak, H.;
21             Froehlich, S. E.; Wallace, L. A. (1993) Particle total exposure assessment methodology (PTEAM) 1990
22             study: method performance and data quality for personal, indoor, and outdoor monitoring. J. Exposure Anal.
23             Environ. Epidemiol.  3: 203-226.
24       Tsang, A. M.; Klepeis, N. E. (1996) Descriptive statistics tables from a detailed analysis of the national human
25             activity pattern survey (NHAPS) data. Las Vegas, NV: U.S. Environmental Protection Agency, Office of
26             Research and Development; report no.  EPA/600/R-96/148.
27       U.S. Environmental Protection Agency. (1982) Air quality criteria for paniculate matter and sulfur oxides. Research
28             Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
29             Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield, VA; PB84-156777.
30       U.S. Environmental Protection Agency. (1992) Guidelines for exposure assessment. Washington, DC: Risk
31             Assessment Forum, USEPA 600Z-92/001.
32       U.S. Environmental Protection Agency. (1996) Air quality criteria for particulate matter. Research Triangle Park,
33             NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
34       U.S. Environmental Protection Agency. (1998) Particulate matter research needs for human health risk assessment to
3 5             support future reviews of the national ambient air quality standards for particulate matter. Research Triangle
36             Park, NC: National Center for Environmental Assessment; report no. EPA/600/R-97/132F.
37       U.S. Environmental Protection Agency. (2000) Aerometric Information Retrieval System (AIRS). Research Triangle
38             Park, NC: Office of Air Quality Planning and Standards. Available: www.epa.gov/ttn/airs/ [2000, October 6].
39       Vakeva, M.; Hameri, K.; Kulmala, M.; Lahdes, R.; Ruuskanen, J.; Laitinen, T. (1999) Street level versus rooftop
40             concentrations of submicron aerosol particles and gaseous pollutants in an urban street canyon.
41             Atmos. Environ. 33:  1385-1397.
42       Van Vliet, P.; Knape, M.; De Hartog, J.; Janssen, N.; Harssema, H.; Brunekreef, B. (1997) Motor vehicle exhaust
43             and chronic respiratory symptoms in children living near freeways. Environ. Res. 74: 122-132.
44       Vette, A. F.; Rea, A. W.; Lawless, P. A.; Rodes, C. E.; Evans, G.; Highsmith, V R.; Sheldon, L. (2001)
45             Characterization of indoor-outdoor aerosol concentration relationships during the Fresno PM exposure
46             studies. Aerosol Sci.  Technol. 34:  118-126.
47       Vyas, V M.; Sun, Q.; Chandrasekar, A.; Wang, S.; Georgopoulos, P. G.; Burke, J.; Xue, J.; Ozkaynak, H. (2002)
48             Source-to-dose analysis of population exposures to PM. Presented at: Eleventh annual meeting of the
49             International Society of Exposure Analysis; November 2001; Charleston, SC.
50       Wainman, T.; Zhang, J.; Weschler, C. J.; Lioy, P. J. (2000) Ozone and limonene in indoor air: a source of submicron
51             particle exposure. Environ.  Health Perspect. 108: 1139-1145.
52       Waldman, J. M.; Liang, C. S. K. (1993) Penetration and neutralization of acid aerosols at institutional settings.
53             In: Jantunen, M.; Kalliokoski, P.; Kukkonen, E.; Saarela, K.; Seppanen, O.; Vuorelma, H., eds. Indoor air
54             '93: proceedings of the 6th international conference on indoor air quality and climate. Volume 3: combustion
55             products, risk assessment, policies; July; Helsinki, Finland. Helsinki, Finland: Indoor Air '93; pp. 263-268.
56       Waldman, J. M.; Liang, C. S. K.; Menon, P. (1990) Indoor exposures to acidic aerosols at child and elderly care
57             facilities. In: Indoor air '90: precedings of the 5th international conference on indoor air quality and climate,


         June 2003                                       5-139         DRAFT-DO NOT QUOTE OR CITE

-------
  1             volume 2, characteristics of indoor air; July-August; Toronto, ON, Canada. Ottawa, ON, Canada:
  2             International Conference on Indoor Air Quality and Climate, Inc.; pp. 607-612.
  3       Waldman, J. M; Lioy, P. J.; Greenberg, A.; Butler, J. P. (1991) Analysis of human exposure to benzo(a)pyrene via
  4             inhalation and food ingestion in the Total Human Environmental Exposure Study (THEES). J. Exposure Anal.
  5             Environ. Epidemiol. 1: 193-225.
  6       Wallace, L. (1996) Indoor particles: a review. J. Air Waste Manage. Assoc. 46: 98-126.
  7       Wallace, L. (2000a) Correlations of personal exposure to particles with outdoor air measurements: a review of recent
  8             studies. Aerosol Sci. Technol. 32: 15-25.
  9       Wallace, L. (2000b) Real-time monitoring of particles, PAH, and CO in an occupied townhouse. Appl. Occup.
10             Environ. Hyg. 15: 39-47.
11       Wallace, L.; Howard-Reed, C. (2002) Continuous monitoring of ultrafine, fine, and coarse particles in a residence
12             for 18 months in 1999-2000. J. Air Waste Manage. Assoc. 52: 828-844.
13       Wallace, L.; Quackenboss, J.; Rodes, C. (1997) Continuous measurements of particles, PAH, and CO in an occupied
14             townhouse in Reston, VA. In: Measurement of toxic and related air pollutants. Volume II: proceedings of a
15             specialty conference; April-May; Research Triangle Park, NC. Pittsburgh, PA: Air & Waste Management
16             Association, pp. 860-871. (A&WMA publication VIP-74).
17       Wallace, L. A.; Emmerich, S. J.; Howard-Reed, C.  (2002) Continuous measurements of air change rates in an
18             occupied house for 1 year: the effect of temperature, wind, fans, and windows. J. Exposure Anal. Environ.
19             Epidemiol. 12: 296-306.
20       Watson, J. G.; Cooper, J. A.; Huntzicker, J. J. (1984) The effective variance weighting for least squares calculations
21             applied to the mass balance receptor model. Atmos. Environ.  18: 1347-1355.
22       Watson, J. G.; Robinson, N. F.; Chow, J. C.; Henry, R. C.; Kim, B. M.; Pace, T. G.; Meyer, E. L.; Nguyen, Q.
23             (1990) The USEPA/DRI chemical mass balance receptor model, CMB 7.0. Environ. Software 5: 38-49.
24       Watson, J. G.; Chow, J. C.; Pace, T. G. (1991) Chemical mass balance. In: Hopke, P. K., ed. Receptor modeling for
25             air quality management. Amsterdam, The Netherlands: Elsevier Science Publishers B.V.; pp. 83-116.
26             (Vandeginste, B. G. M.; Kvalheim, O. M., eds. Data handling in science and technology, v. 7).
27       Weschler, C. J.; Shields, H. C. (1999) Indoor ozone/terpene reactions as a source of indoor particles. Atmos.
28             Environ. 33:2301-2312.
29       Wheeler,  A. J.; Williams, L; Beaumont, R. A.; Hamilton, R. S. (2000) Characterisation of paniculate matter sampled
3 0             during a study of children's personal exposure to airborne paniculate matter in a UK urban environment.
31             Environ. Monit. Assess. 65: 69-77.
32       Whitby, K.  T.  (1978) The physical characteristics of sulfur aerosols. Atmos. Environ. 12: 135-159.
33       White, W. H. (1998) Statistical considerations in the interpretation of size-resolved paniculate mass data. J. Air
34             Waste Manage. Assoc. 48: 454-458.
35       Wiener, R. W. (1988) Measurement and evaluation of personal  exposure to aerosols. In: Measurement of toxic and
36             related air pollutants: proceedings of the 1988 EPA/APCA international symposium; May; Research Triangle
37             Park, NC. Pittsburgh, PA: Air Pollution Control Association; pp. 84-88. (APCA publication VTP-10).
3 8       Wiener, R. W. (1989) Particle total exposure assessment methodology-an overview of planning and
39             accomplishments. In: Measurement of toxic and related air pollutants: proceedings of the 1989 U.S.
40             EPA/A&WMA international symposium; May; Raleigh,  NC. Pittsburgh, PA: Air & Waste Management
41             Association; pp. 442-448. (A&WMA publication VIP-13).
42       Wiener, R. W.; Wallace, L.; Pahl, D.; Pellizzari, E.; Whittaker,  D.; Spengler, J.; Ozkaynak, H. (1990) Review of the
43             particle TEAM 9 home field study. In: Measurement of toxic and related air pollutants: proceedings of the
44             1990 EPA/A&WMA international symposium, v. 2; May; Raleigh, NC. Pittsburgh, PA: Air & Waste
45             Management Association; pp. 452-460. (A&WMA publication VTP-17).
46       Wiley, J. A.; Robinson, J. P.; Piazza, T.; Garrett, K.; Cirksena, K.; Cheng, Y.-T.; Martin, G. (199la) Activity
47             patterns of California residents. Final report. Sacramento, CA: California Air Resources Board; report no.
48             ARB/R93/487. Available from: NTIS, Springfield, VA.;  PB94-108719.
49       Wiley, J. A.; Robinson, J. P.; Cheng, Y.-T.; Piazza, T.; Stork, L.; Pladsen, K. (1991b) Study of children's activity
50             patterns: final report.  Sacramento, CA: California Air Resources Board; report no. ARB-R-93/489.
51       Williams, R.; Suggs, J.; Zweidinger, R.; Evans, G.; Creason, J.; Kwok, R.; Rodes, C.; Lawless, P.; Sheldon, L.
52             (2000a) The 1998 Baltimore Paniculate Matter Epidemiology-Exposure Study: part 1. Comparison of
53             ambient, residential outdoor, indoor and apartment paniculate matter monitoring. J. Exposure Anal. Environ.
54             Epidemiol. 10: 518-532.
55       Williams, R.; Suggs, J.; Creason, J.; Rodes, C.; Lawless, P.; Kwok, R.; Zweidinger, R.; Sheldon, L. (2000b) The
56             1998 Baltimore paniculate matter epidemiology-exposure study: part 2.  Personal exposure assessment
57             associated with an elderly study population. J. Exposure  Anal. Environ.  Epidemiol.: 10: 533-543.


         June 2003                                     5-140        DRAFT-DO NOT  QUOTE OR CITE

-------
  1       Williams, R.; Creason, J.; Zweidinger, R.; Watts, R.; Sheldon, L.; Shy, C. (2000c) Indoor, outdoor, and personal
  2             exposure monitoring of paniculate air pollution: the Baltimore elderly epidemiology-exposure pilot study.
  3             Atmos. Environ. 34: 4193-4204.
  4       Wilson, W. E.; Suh, H. H. (1997) Fine particles and coarse particles: concentration relationships relevant to
  5             epidemiologic studies. J. Air Waste Manage. Assoc. 47: 1238-1249.
  6       Wilson, W. E.; Mage, D. T.; Grant, L. D. (2000) Estimating separately personal exposure to ambient and
  7             nonambient paniculate matter for epidemiology and risk assessment: why and how. J. Air Waste Manage.
  8             Assoc. 50: 1167-1183.
  9       Yakovleva, E.; Hopke, P. K.; Wallace, L. (1999) Receptor modeling assessment of particle total exposure
10             assessment methodology data. Environ. Sci. Technol. 33: 3645-3652.
11       Zartarian, V. G.; Ott, W. R.; Duan, N. (1997) A quantitative definition of exposure and related concepts. J. Exposure
12             Anal. Environ.  Epidemiol. 7: 411-437.
13       Zeger, S. L.; Thomas, D.; Dominici, F.; Samet, J. M.; Schwartz, J.; Dockery, D.; Cohen, A. (2000) Exposure
14             measurement error in time-series studies of air pollution: concepts and consequences. Environ. Health
15             Perspect. 108: 419-426.
16       Zhu, Y.; Hinds, W. C.; Kim, S.; Shen, S.; Sioutas, C. (2002) Study of ultrafine particles near a major highway with
17             heavy-duty diesel traffic. Atmos. Environ.  36: 4323-4335.
18       Zmirou, D.; Masclet, P.; Boudet, C.; Dor, F.; Dechenaux, J. (2000) Personal exposure to atmospheric polycyclic
19             aromatic hydrocarbons in a general adult population and lung cancer risk assessment. J. Occup. Environ.
20             Med. 42: 121-126.
21
         June 2003                                      5-141         DRAFT-DO NOT QUOTE OR CITE

-------