j
\
Particulate Matter
Urban-Focused Visibility Assessment
External Review Draft
September 2009
-------
DISCLAIMER
This draft document has been prepared by staff from the Ambient Standards Group,
Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency. Any
opinions, findings, conclusions, or recommendations are those of the authors and do not
necessarily reflect the views of the EPA. This document is being circulated to obtain review and
comment from the Clean Air Scientific Advisory Committee (CASAC) and the general public.
Comments on this draft document should be addressed to Vicki Sandiford, U.S. Environmental
Protection Agency, Office of Air Quality Planning and Standards, C504-06, Research Triangle
Park, North Carolina 27711 (email: sandiford.vicki@epa.gov).
Elements of this report (see Acknowledgements below) have been provided to the U.S.
Environmental Protection Agency (EPA) by Abt Associates Inc. and Stratus Consulting Inc. in
partial fulfillment of Contract No. EP-D-08-100, Work Assignment 0-11. Any opinions,
findings, conclusions, or recommendations are those of the authors and do not necessarily reflect
the views of the EPA or its contractors.
ACKNOWLEDGEMENTS
In addition to EPA staff, personnel from Abt Associates Inc. and Stratus Consulting Inc.
contributed to the writing of this document. The specific chapter where Abt Associates Inc. and
Stratus Consulting Inc. made contributions is chapter 2 (Urban Visibility Preference Studies).
-------
EPA 450/P-09-005
September 2009
Paniculate Matter
Urban-Focused Visibility Assessment
External Review Draft
U.S. Environmental Protection Agency
Office of Air and Radiation
Office of Air Quality Planning and Standards
Health and Environmental Impacts Division
Ambient Standards Group
Research Triangle Park, North Carolina 27711
-------
This page left intentionally blank.
-------
TABLE OF CONTENTS
LIST OF TABLES Hi
LIST OF FIGURES iv
LIST OF ACRONYMS/ABBREVIATIONS vi
1 INTRODUCTION 1-1
1.1 PM NAAQS BACKGROUND 1-3
1.2 SCOPE OF URBAN-FOCUSED VISIBILITY ASSESSMENT 1-5
1.3 VISIBILITY EFFECTS SCIENCE OVERVIEW 1-7
1.4 GOALS AND APPROACH 1-9
1.5 ORGANIZATION OF DOCUMENT 1-11
2 URBAN VISIBILITY PREFERENCE STUDIES 2-1
2.1 METHODS USED IN PREVIOUS STUDIES 2-1
2.2 DENVER, COLORADO 2-3
2.3 VANCOUVER, BRITISH COLUMBIA, CANADA 2-7
2.4 PHOENIX, ARIZONA 2-12
2.5 WASHINGTON, D.C 2-14
2.5.1 Washington, D.C. 2001 2-15
2.5.2 Washington, D.C., 2009 2-17
2.6 SUMMARY OF PREFERENCE STUDIES AND SELECTION OF CANDIDATE
PROTECTION LEVELS 2-23
3 ESTIMATION OF CURRENT PM CONCENTRATIONS AND LIGHT EXTINCTIONS-1
3.1 GENERAL CHARACTERIZATION 3-1
3.1.1 PM2.5 and PMio.2.5 3-1
3.1.2 Light extinction 3-4
3.2 OVERVIEW OF APPROACH AND DATA SOURCES FOR URBAN STUDY ANALYSIS 3-9
3.2.1 Study Period, Study Areas, Monitoring Sites, and Sources of Ambient PM Data 3-10
3.2.2 Use of CMAQ Model Runs for 2004 to Augment Ambient Data 3-15
3.2.3 Use of Original IMPROVE Algorithm to Estimate Light Extinction 3-16
3.3 DETAILED STEPS 3-18
3.3.1 Hourly PM2.5 Component Concentrations 3-18
3.3.2 Hourly PMio-2.5 Concentrations 3-24
3.3.3 Hourly Relative Humidity Data 3-24
3.3.4 Calculation of Hourly and Daily Maximum 1 -Hour Total Light Extinction 3-25
3.4 SUMMARY OF RESULTS FOR CURRENT CONDITIONS 3-26
3.4.1 Levels of Estimated PM2.s, PM2.5 Components, PMi0-2.5, and Relative Humidity 3-26
3.4.2 Levels of Estimated Total Light Extinction 3-26
3.4.3 Patterns of Relative Humidity and Relationship between Relative Humidity and Total
Light Extinction 3-33
-------
3.4.4 Extinction Budgets for High Total Light Extinction Conditions 3-36
3.5 POLICY RELEVANT BACKGROUND 3-37
4 TOTAL LIGHT EXTINCTION UNDER "WHAT IF" CONDITIONS OF JUST
MEETING SPECIFIC ALTERNATIVE SECONDARY NAAQS 4-1
4.1 ALTERNATIVE SECONDARY NAAQS BASED ON MEASURED TOTAL LIGHT
EXTINCTION AS THE INDICATOR 4-1
4.1.1 Indicator and Monitoring Method 4-1
4.1.2 Alternative Secondary NAAQS Scenarios based on Measured Total Light Extinction.. 4-1
4.1.3 Monitoring Site Considerations for Alternative Secondary NAAQS Based on Measured
Total Light Extinction 4-2
4.1.4 Approach to Modeling "What If Conditions for Alternative Secondary NAAQS based
on Measured Total Light Extinction 4-2
4.2 ALTERNATIVE SECONDARY PM2.5 NAAQS BASED ON ANNUAL AND 24-HOUR PM2.5
MASS 4-5
4.2.1 Secondary NAAQS Scenarios Based on Annual and 24-hour PM2.5 Mass 4-5
4.2.2 Approach to Modeling Conditions If Secondary PM2 5 NAAQS Based on Annual and 24-
hour PM2.5 Mass Were Just Met 4-6
4.3 RESULTS FOR "JUST MEETING" ALL ALTERNATIVE SECONDARY NAAQS
SCENARIOS 4-7
5 REFERENCES 5-1
APPENDICES
APPENDIX A - PM2.5 MONITORING SITES AND MONITORS PROVIDING 2005-2007 DATA
FOR THE ANALYSIS OF TOTAL LIGHT EXTINCTION IN THE 15 STUDY AREAS A-l
APPENDIX B - DISTRIBUTIONS OF ESTIMATED PM2.5 COMPONENTS B-l
APPENDIX C - DEVELOPMENT OF PRB ESTIMATES OF PM2.5 COMPONENTS, PMi0.2.5, AND
TOTAL LIGHT EXTINCTION C-l
APPENDIX D RELATIONSHIPS BETWEEN PM MASS CONCENTRATION AND TOTAL LIGHT
EXTINCTION UNDER CURRENT CONDITIONS D-l
APPENDIX E - DIFFERENCES IN DAILY PATTERNS OF RELATIVE HUMIDITY AND TOTAL
LIGHT EXTINCTION BETWEEN AREAS AND SEASONS E-l
APPENDIX F - DISTRIBUTIONS OF MAXIMUM DAILY DAYLIGHT TOTAL LIGHT
EXTINCTION - UNDER "JUST MEET" CONDITIONS F-l
-------
LIST OF TABLES
Table 2-1. VAQ of Denver photos substantively misclassified by majority of participants 2-6
Table 2-2. Summary of photographs used in British Columbia study 2-10
Table 3-1. Annual Mean Reconstructed 24-hour Light Extinction Estimates 3-7
Table 3-2. Urban Visibility Assessment Study Areas 3-10
Table 3-3. PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Tacoma
Study Area 3-13
Table 3-4. Number of days per quarter in each study area 3-14
Table 3-5. Assumed daylight hours by season (Local Standard Time) 3-25
Table 3-6. Percentage of days in which daily maximum daylight 1-hour total light extinction
exceeded three candidate protective levels across the 2005-2007 period, 3-31
Table 4-1. Alternative Secondary NAAQS Scenarios for Light Extinction 4-1
Table 4-2. Current Conditions total light extinction design values for the study areas 4-4
Table 4-3. Percentage reductions in non-PRB light extinction required to "just meet" the
NAAQS scenarios based on measured light extinction 4-5
Table 4-4. Percentage reductions required in non-PRB PM2.5 mass to "just meet" NAAQS
scenarios based on annual and 24-hour PM25 mass 4-7
Table 4-5. Total light extinction design values for "just meeting" secondary NAAQS scenarios
based on measured total light extinction 4-10
Table 4-6. Total light extinction design values for "just meeting" secondary NAAQS scenarios
based on PM2.5 mass 4-10
Table 4-7. Percentage of days across three years (two in the case of Phoenix and Houston) with
maximum 1-hour daylight total light extinction above CPLs when "just meeting"
the NAAQS scenarios 4-11
September 2009 iii DRAFT - Do Not Quote or Cite
-------
LIST OF FIGURES
Figure 1-1. Diagram showing the relationship steps between ambient PM and visibility
impairment 1-9
Figure 2-1. Percent of Denver participants who consider VAQ in each photograph "acceptable."
2-5
Figure 2-2. Photograph time of day information for the percent of participants who consider
VAQ in each photograph "acceptable." 2-7
Figure 2-3. Denver photograph time of day results (9:00 a.m. photographs eliminated), with the
broader range (17.7 dv and 24.6 dv) of the 50% acceptability criteria shown. ...2-7
Figure 2-4. Composite Chilliwack, BC photograph showing VAQ of 14.1 dv and 34 dv 2-9
Figure 2-5. Percent of BC participants who consider VAQ in each photograph "acceptable." .. 2-
11
Figure 2-6. Reproduction of the image with the best VAQ (15 dv) used in the Phoenix study... 2-
13
Figure 2-7. Percent of Phoenix participants who consider VAQ in each image "acceptable." 2-14
Figure 2-8. Reproduction of the image with the best VAQ (8.8 dv) used in the Washington, D.C.
study 2-16
Figure 2-9. Percent of 2001 Washington participants who consider VAQ acceptable in each
image 2-17
Figure 2-10. Percent of 2009 Test 1 study participants who consider VAQ acceptable in each
image, showing the range of the lower and upper bound of 50% acceptability
criteria 2-19
Figure 2-11. Combined results of two Washington preference studies (showing 50%
acceptability criteria from 2009, Test 1) 2-20
Figure 2-12. Comparison of results from Test 1 and Test 2 (Smith and Howell, 2009) 2-20
Figure 2-13. Comparison of results from the Washington, DC (2009) Test 1 and Test 3 2-22
Figure 2-14. Summary of results of urban visibility studies in four North American cities,
showing the identified range of the 50% acceptance criteria 2-24
Figure 3-1. Annual average and 24-hour (98th percentile 24-hour concentrations) PM25
concentrations in ug/m3, 2007 3-2
Figure 3-2. Reconstructed 24-hour light extinction in U.S. urban areas in 2003 3-6
Figure 3-3. Isopleth map of annual total reconstructed particulate extinction based on IMPROVE
data 3-8
Figure 3-4. January and August monthly average diurnal profiles of PM2.5 components derived
from the 2004 CMAQ modeling platform, for the Detroit study area 3-17
Figure 3-5. Sequence of steps used to estimate hourly PM2.5 components and total light
extinction 3-20
Figure 3-6. Example from Detroit study area 3-22
Figure 3-7. Distribution of PM parameters and relative humidity across the 2005-2007 period,
by study area 3-28
Figure 3-8. Distributions of estimated daylight 1-hour total light extinction and maximum daily
daylight 1-hour total light extinction across the 2005-2007 period, by study area.
3-30
Figure 3-9. Distributions of 1-hour total light extinction levels by daylight hour across the 2005-
2007 period, by study area 3-32
September 2009 iv DRAFT - Do Not Quote or Cite
-------
Figure 3-10. Distributions of 1-hour relative humidity levels by daylight hour across the 2005-
2007 period, by study area 3-34
Figure 3-11. Scatter plot of daylight 1-hour relative humidity (percent) vs. reconstructed total
light extinction (Mm"1) across the 2005-2007 period, by study area 3-35
Figure 3-12. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily 1-
hour PM light Extinction for 2005-2007 (Tacoma and Fresno) 3-38
Figure 3-13. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily 1-
hour PM light Extinction for 2005-2007 (Los Angeles and Phoenix) 3-39
Figure 3-14. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily 1-
hour PM light Extinction for 2005-2007 (Salt Lake City and Dallas) 3-40
Figure 3-15. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily 1-
hour PM light Extinction for 2005-2007 (Houston and St. Louis) 3-41
Figure 3-16. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily 1-
hour PM light Extinction for 2005-2007 (Birmingham and Atlanta) 3-42
Figure 3-17. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily 1-
hour PM light Extinction for 2005-2007 (Detroit and Baltimore) 3-43
Figure 3-18. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily 1-
hour PM light Extinction for 2005-2007 (Pittsburgh and Philadelphia) 3-44
Figure 3-19. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily 1-
hour PM light Extinction for 2005-2007 (New York) 3-45
Figure 4-1. Distributions of daily maximum daylight 1-hour total light extinction under two
"just meeting" secondary NAAQS scenarios 4-9
September 2009 v DRAFT - Do Not Quote or Cite
-------
LIST OF ACRONYMS/ABBREVIATIONS
AQS
BAM
BC
CAA
CAIR
CASAC
CBS A
CCN
CDPHE
CMAQ
CONUS
CPL
CRA
CSA
CSN
CTM
CIS
ORE
dv
EPA
FEM
FRM
GEOS
IMPROVE
ISA
Km
LCD
LOESS
Mm
MSA
N
NAAQS
NARSTO
EPA's Air Quality System
Beta Attenuation Mass Monitor
British Columbia
Clean Air Act
Clean Air Interstate Rule
Clean Air Scientific Advisory Committee
Consolidated Business Statistical Area
Cloud Condensation Nuclei
Colorado Department of Public Health and Environment
Community Multiscale Air Quality
CMAQ simulations covering continental US
Candidate Protection Level
Charles River Associates
Consolidated Statistical Area
Chemical Speciation Network
Chemical Transport Model
CMAQ model run
Direct Radiative Effects
deciview
United States Environmental Protection Agency
Federal Equivalent Method
Federal Reference Method
Global Scale Air Circulation Model
Interagency Monitoring of Protected Visual Environment
Integrated Science Assessment
Kilometer
Liquid Crystal Display
Locally weighted Scatter Plot Smoothing
Megameter
Metropolitan Statistical Area
Nitrogen
National Ambient Air Quality Standards
North American Research Strategy for Tropospheric Ozone
September 2009
VI
DRAFT - Do Not Quote or Cite
-------
NCEA
NOAA
NOx
NFS
NRC
NWS
OAQPS
OAR
OMB
ORD
PA
PM
PM2.5
PM10
PMiQ-2.5
PRB
REA
RF
RH
SANDWICH
SEARCH
SMOKE
S
S02
sox
STP
TEOM
UBC
UFVA
VAQ
National Center for Environmental Assessment
National Oceanic and Atmospheric Administration
Nitrogen oxides
National Park Service
National Research Council
National Weather Service
Office of Air Quality Planning and Standards
Office of Air and Radiation
Office of Management and Budget
Office of Research and Development
Policy Assessment
Particulate Matter
Particles with a 50% upper cut-point of 2.5 um aerodynamic
diameter and a penetration curve as specified in the Code of
Federal Regulations.
Particles with a 50% upper cut-point of 10± 0.5 um aerodynamic
diameter and a penetration curve as specified in the Code of
Federal Regulations.
Particles with a 50% upper cut-point of 10 um aerodynamic
diameter and a lower 50% cut-point of 2.5 um aerodynamic
diameter.
Policy Relevant Background
Risk and Exposure Assessment
Radiative Forcing
Relative Humidity
Siulfate, Adjusted Nitrate, Derived Water, Inferred Carbonaceous
mass approach
Southeastern Aerosol Research and Characterization Study
Sparse Matrix Operator Kernal Emissions
Sulfur
Sulfur Dioxide
Sulfur Oxides
Standard Temperature and Pressure
Tapered Element Oscillating Microbalance
University of British Columbia
Urban-Focused Visibility Impact Assessment
Visual Air Quality
September 2009
vn
DRAFT - Do Not Quote or Cite
-------
1 1 INTRODUCTION
2 The U.S. Environmental Protection Agency (EPA) is presently conducting a review of
3 the national ambient air quality standards (NAAQS) for particulate matter (PM). Sections 108
4 and 109 of the Clean Air Act (Act) govern the establishment and periodic review of the NAAQS.
5 The NAAQS are to be based on air quality criteria, which are to accurately reflect the latest
6 scientific knowledge useful in indicating the kind and extent of identifiable effects on public
7 health or welfare that may be expected from the presence of the pollutant in ambient air. The
8 EPA Administrator is to promulgate and periodically review, at no later than five-year intervals,
9 "primary" (health-based) and "secondary" (welfare-based) NAAQS for such pollutants. Based
10 on periodic reviews of the air quality criteria and standards, the Administrator is to make
11 revisions in the air quality criteria and standards, and to promulgate any new standards, as may
12 be appropriate. The Act also requires that an independent scientific review committee advise the
13 Administrator as part of this NAAQS review process, a function performed by the Clean Air
14 Scientific Advisory Committee (CASAC).
15 The current NAAQS for PM include a suite of standards to provide protection from
16 health and welfare effects related to fine and coarse particles, using PM2 5 and PMi0 as
17 indicators, respectively (71 FR 61144, October 17, 2006). With regard to the primary and
18 secondary standards for fine particles, in 2006 EPA revised the level of the 24-hour PM2.5
19 standard to 35 ug/m3 (calculated as a 3-year average of the 98th percentile of 24-hour
20 concentrations at each population-oriented monitor), retained the level of the annual PM2.5 annual
21 standard at 15 ug/m3 (calculated as the 3-year average of the weighted annual mean PM2.5
22 concentrations from single or multiple community-oriented monitors), and revised the form of
23 the annual PM2.5 standard by narrowing the constraints on the optional use of spatial averaging1.
24 With regard to the primary and secondary standards for PMi0, EPA retained the 24-hour PMi0
25 standard at 150 ug/m3 (not to be exceeded more than once per year on average over 3 years) and
26 revoked the annual standard because available evidence generally did not suggest a link between
27 long-term exposure to current ambient levels of coarse particles and health or welfare effects.
28 The 2006 primary standards were based primarily on a large body of epidemiological evidence
29 relating ambient PM concentrations to various adverse health outcomes. The 2006 secondary
30 standards for PM2.5 and PMio were revised to be identical to the primary standards, on the basis
31 that in the Administrator's judgment these standards, in conjunction with the regional haze
1 In the revisions to the PM NAAQS finalized in 2006, EPA tightened the constraints on the spatial averaging option
limiting the conditions under which some areas may average measurements from multiple community-oriented
monitors to determine compliance (see 71 FR 61165-61167, October 17, 2006).
September 2009 1-1 DRAFT - Do Not Quote or Cite
-------
1 program , will provide appropriate protection to address PM-related welfare effects, including
2 visibility impairment, effects on vegetation and ecosystems, materials damage and soiling, and
3 effects on climate change.
4 The next periodic review of the PM NAAQS is now underway.2 In the Integrated
5 Review Plan for the National Ambient Air Quality Standards for Paniculate Matter, March 2008
6 (US EPA, 2008a), EPA outlined the science policy questions that will frame this review, outlined
7 the process and schedule that the review will follow, and provided more complete descriptions of
8 the purpose, contents, and approach for developing the key documents that will be developed in
9 the review.3 EPA is currently completing the process of assessing the latest available policy -
10 relevant scientific information to inform the review of the PM standards. The latest draft of this
11 assessment is contained in the second external review draft of the Integrated Science Assessment
12 for Particulate Matter (ISA, US EPA, 2009a) which was released in July 2009 for review by
13 CAS AC and for public comments. The 2009 second draft PM ISA includes a summary of the
14 scientific evidence for the relationship of PM to visibility effects, remote area and urban haze
15 conditions, the PM components responsible for visibility impacts, and studies of public
16 preference with respect to urban visibility conditions.
17 Building upon the visibility effects evidence presented in the second draft PM ISA, as
18 well as CAS AC advice (Samet, 2009) and public comments on a planning document (US EPA,
19 2009b), EPA's Office of Air Quality Planning and Standards (OAQPS) has developed this draft
20 Urban-Focused Visibility Assessment (UFVA) describing the quantitative assessments being
21 conducted by the Agency to support the review of the secondary PM standards. This draft
22 document is a concise presentation of the methods, key results, observations, and related
23 uncertainties associated with the quantitative analyses performed. Revisions to this draft UFVA
24 will draw upon the final ISA and will reflect consideration of CAS AC and public comments on
25 this draft UFVA.
26 The final ISA and final UFVA will inform the policy assessment and rulemaking steps
27 that will lead to final decisions on the secondary PM NAAQS. A Policy Assessment (PA) is
28 now being prepared by OAQPS staff to provide a transparent staff analysis of the scientific basis
29 for alternative policy options for consideration by senior EPA management prior to rulemaking.
See http://www.epa.gov/ttn/naaqs/standards/pm/sjm index.html for more information on the current and
previous PM NAAQS reviews.
3 On November 30, 2007, EPA held a consultation with the Clean Air Scientific Advisory Committee (CASAC) on
the draft IRP (Henderson, 2008). Public comments were also requested on the draft plan and presented at that
CASAC teleconference. The final IRP incorporated comments received from CASAC and the general public on the
draft plan as well as input from senior Agency managers. CASAC is an independent scientific advisory committee
established to meet the requirements of section 109(d)(2) of the Clean Air Act See
http://vosemite.epa.gov/sab/sabpeople.nsf/WebCommittees/CASAC for more information, and, in particular,
information on the CASAC PM Review Panel activities.
September 2009 1 -2 DRAFT - Do Not Quote or Cite
-------
1 The PA is intended to help "bridge the gap" between the Agency's scientific assessments,
2 presented in the ISA and UFVA, and the judgments required of the Administrator in determining
3 whether it is appropriate to retain or revise the secondary PM standards. The PA will integrate
4 and interpret information from the ISA and the UFVA to frame policy options and to facilitate
5 CASAC's advice to the Agency and recommendations on any new standards or revisions to
6 existing standards as may be appropriate, as provided for in the Clean Air Act. A very
7 preliminary draft PA is planned for release in September 2009 to facilitate discussion on the
8 overall structure, areas of focus, and level of detail to be included in an external review draft of
9 the document, which EPA plans to release for CASAC review and public comment later this
10 year. A discussion of the preliminary draft PA with CASAC will be held in conjunction with
11 CASAC review and public comment of the second draft ISA, this draft UFVA, and a draft
12 assessment document that will inform the review of the primary PM standards - Risk Assessment
13 to Support the Review of the PM Primary National Ambient Air Quality Standards - External
14 Review Draft (US EPA, 2009c).
15 1.1 PM NAAQS BACKGROUND
16 In the review of the secondary PM NAAQS completed in 2006, EPA took into account
17 that the Regional Haze Program4, implemented under sections 169 A and 169B of the CAA, was
18 established to address all human-caused visibility impairment in Class I areas. Recognizing that
19 efforts were underway under that program, EPA focused the 2006 PM NAAQS review on
20 visibility impairment primarily in urban areas. The EPA evaluated the levels of visibility
21 impairment occurring in urban areas and assessed available information on public preferences
22 regarding acceptability of PM-related urban visibility impairment. At that time, EPA's focus
23 continued to remain on particle size and mass and EPA staff determined that PM2.5 size and
24 particle mass, rather than particle composition, remained the most appropriate approach for
25 addressing PM-related urban visibility effects. EPA recognized that PM composition and
26 relative humidity are important factors in the relationship between light extinction (a measure of
27 visibility) and PM mass concentration. The EPA's assessment of PM and meteorological data
28 from 161 cities showed that the least variation in the relationship of light extinction to PM2.5
29 mass concentration was for afternoon periods when low relative humidity conditions generally
30 prevail (EPA, 2005).
31 The EPA proposed to revise the secondary standards by making them identical to the
32 suite of proposed primary standards for fine and coarse particles, providing protection against
33 PM-related public welfare effects including visibility impairment, effects on vegetation and
4 See http://www.epa.gov/air/visibilitv/program.html for more information on EPA's Regional Haze Program.
September 2009 1 -3 DRAFT - Do Not Quote or Cite
-------
1 ecosystems, and materials damage and soiling (71 FR 2620). The EPA also solicited comment on
2 adding a new sub-daily PM2.5 secondary standard to address visibility impairment in urban areas.
3 CASAC provided additional advice to EPA in a letter to the Administrator requesting
4 reconsideration of CASAC's recommendations for both the primary and secondary PM2.5
5 standards as well as standards for thoracic coarse particles (Henderson, 2006). With regard to
6 the secondary standard, CASAC reaffirmed "... the recommendation of Agency staff regarding a
7 separate secondary fine particle standard to protect visibility The CASAC wishes to
8 emphasize that continuing to rely on primary standards to protect against all PM-related adverse
9 environmental and welfare effects assures neglect, and will allow substantial continued
10 degradation, of visual air quality over large areas of the country" (Henderson, 2006).
11 On September 21, 2006, EPA announced its final decisions to revise the secondary
12 NAAQS for PM to provide increased protection of public welfare by making them identical to
13 the revised primary standards (71 FR 61144, October 17, 2006). This was designed to address
14 both visibility and other non-visibility welfare related effects. Specifically, with regard to the
15 secondary standards for fine particles, EPA revised the level of the 24-hour PM2.5 standard to 35
16 ng/m3, retained the level of the annual PM2.5 annual standard at 15 |ig/m3, and revised the form
17 of the annual PM2.5 standard by narrowing the constraints on the optional use of spatial
18 averaging. With regard to the secondary standards for coarse particles, EPA retained PMio as the
19 indicator for purposes of regulating the coarse fraction of PMio (referred to as thoracic coarse
20 particles or coarse-fraction particles; generally including particles with a nominal mean
21 aerodynamic diameter greater than 2.5 |im and less than or equal to 10 jim, or PMio-2.s). EPA
22 retained the 24-hour PMio standard at 150 |ig/m3 and revoked the annual PMio standard.
23 Several parties filed petitions for review following promulgation of the revised PM
24 NAAQS in 2006. These petitions addressed a number of issues, including the decision to set the
25 secondary PM2.5 standards identical to the primary standards. On judicial review the court
26 remanded the secondary PM2.5 NAAQS to EPA because the Agency failed to adequately explain
27 why setting the standards equal to the primary PM2 5 standards provided the required protection
28 from visibility impairment. In particular, the Agency failed to identify a target level of visibility
29 impairment that would be requisite to protect the public welfare, and improperly relied on a
30 comparison of the number of counties which would be in nonattainment for the revised primary
31 NAAQS compared to various alternative secondary standards. Among other things, this
32 equivalence analysis failed to address the issue of regional differences in humidity-related effects
33 on visibility American Farm Bureau Federation v. EPA, 559 F. 3d 512 (D.C. Cir. 2009)
September 2009 1 -4 DRAFT - Do Not Quote or Cite
-------
1 1.2 SCOPE OF URBAN-FOCUSED VISIBILITY ASSESSMENT
2 This chapter provides an overview of the scope and key design elements of the UFVA
3 conducted for this review, including the process that has been followed to design the analyses.
4 Following initiation of the current PM NAAQS review, we began the design of this assessment
5 by reviewing the analyses completed during the previous PM NAAQS review (Abt Associates
6 Inc., 2001; US EPA, 2005, chapter 6) with an emphasis on considering key limitations and
7 sources of uncertainty recognized in that analysis. Furthermore, as an initial step in the overall
8 PM NAAQS review, EPA invited a wide range of external experts as well as EPA staff,
9 representing a variety of areas of expertise to participate in a workshop titled, "Workshop to
10 Discuss Policy-Relevant Science to Inform EPA's Integrated Plan for the Review of the
11 Secondary PM NAAQS" (72 FR 34005, June 20, 2007). This workshop provided an opportunity
12 for the participants to broadly discuss the key policy-relevant issues around which EPA would
13 structure the PM NAAQS review and to discuss the most meaningful new science that would be
14 available to inform our understanding of these issues. One session of this workshop was
15 centered around issues related to visibility impacts associated with ambient PM. Specifically,
16 the discussions focused on the extent to which new research and/or improved methodologies
17 were available to inform how EPA evaluated visibility impairment in this review.
18 Based in part on these workshop discussions, EPA developed a draft IRP outlining the
19 schedule, the process, and the key policy-relevant science issues that would guide the evaluation
20 of the air quality criteria for PM and the review of the primary and secondary PM NAAQS
21 including initial thoughts for conducting quantitative assessments (US EPA, 2007, chapter 6).
22 On November 30, 2007, CAS AC held a teleconference with EPA to provide its comments on the
23 draft IRP (72 FR 63177, November 8, 2007). Public comments were also presented at that
24 teleconference. A final IRP incorporating comments received from CASAC and the general
25 public on the draft plan was issued in March 2008 (US EPA, 2008a).
26 On October 6-8, 2008 the EPA sponsored an urban visibility workshop in Denver,
27 Colorado to identify and discuss methods and materials that could be used in "next step" projects
28 to develop additional information about people's preferences for reducing existing impairment of
29 urban visibility, and about the value of improving urban visibility. Invited individuals came from
30 a broad array of relevant technical and policy backgrounds, including visual air quality (VAQ)
31 science, sociology, psychology, survey research methods, economics, and EPA's process of
32 setting NAAQS. The 23 people who attended the workshop (including one via teleconference
September 2009 1 -5 DRAFT - Do Not Quote or Cite
-------
1 line) came from EPA, the National Oceanic and Atmospheric Administration (NOAA), NFS,
2 academia, regional and state air pollution planning agencies, and consulting firms.5
3 As a next step in the design of the quantitative assessments, EPA developed a planning
4 document outlining the initial design for the PM NAAQS visibility assessment - Particulate
5 Matter National Ambient Air Quality Standards: Scope and Methods Plan for Urban Visibility
6 Impact Assessment, henceforth Scope and Methods Plan (US EPA, 2009b). This planning
7 document was released for CASAC consultation and public review in February 2009. Based on
8 consideration of CASAC and public comments on that Scope and Methods Plan, along with
9 ongoing review of the latest PM-related literature, we made modifications to the scope and
10 design of the visibility assessment and completed our initial analyses. These modifications, as
11 well as the current scope of the UFVA and the rationale supporting it, are described in this
12 section below.
13 The EPA staff continues to believe that a focus on urban area visibility is appropriate. In
14 articulating a rationale for this conclusion, we have reviewed the information contained in the
15 second draft ISA and find the following information compelling: 1) PM levels in urban areas are
16 often in excess of those of the surrounding region since urban haze typically includes both
17 regional and local contributions (US EPA, 2009a; sections 9.2.3.3 and 9.2.3.4), suggesting the
18 potential for higher levels of PM-induced visibility impairment in urban areas; 2) the existence of
19 numerous urban visibility protection programs and goals demonstrates that urban VAQ is noticed
20 and an important value to urban residents (US EPA, 2009a; section 9.2.4), and 3) the existence of
21 large urban populations suggests that potentially more people are routinely affected by poor
22 VAQ than in rural areas. One aspect of the urban visibility conditions assessment, as depicted in
23 Figure 1-1 of section 1.3 of the Scope and Methods document (US EPA, 2009b), has been
24 modified. Taking into account the nature of urban versus more remote area PM composition,
25 and input received at the April 2, 2009 CASAC meeting, EPA staff has concluded that it is
26 unnecessary to develop a new urban-optimized algorithm at this time and that it remains
27 appropriate in the context of this assessment to use the original IMPROVE algorithm to relate
28 urban PM to local haze (PM light extinction).
29 With regard to the urban visual air quality preference assessment described in the Scope
30 and Methods document (US EPA, 2009b, section 1.3), more significant modifications have
31 occurred. EPA staff has decided to conduct a reanalysis of the urban visibility preference studies
32 available at the time of the 2006 PM NAAQS review, rather than conduct new public preference
33 studies, as it has become apparent that the results of these studies would be unlikely to be
34 completed in time to inform this review. Recognition that the initial plans described in the Scope
5 To view the complete report from the October 2008 urban visibility workshop, see:
http://vista.cira.colostate.edu/improve/Publications/GrayLit/grav literature.htm
September 2009 1 -6 DRAFT - Do Not Quote or Cite
-------
1 and Methods document were possibly overly ambitious was also shared by members of CASAC
2 (see individual member comments; Samet, 2009). This analysis, therefore, relies on existing,
3 rather than new, urban visibility preference studies and is designed to explore the similarities and
4 differences (comparability) between these studies and assess what information can be drawn
5 from these results to inform the selection of VAQ candidate protection levels (CPLs) to be used
6 in subsequent impact assessments. Further, information presented during the public comment
7 phase of the April 2, 2009 CASAC meeting and later provided to EPA staff, led to the inclusion
8 of a recent study by Smith and Howell (2009) for Washington, D.C. in the reanalysis.
9 As described in the Scope and Methods document (US EPA, 2009b), EPA staff is
10 continuing to focus assessments in this document in terms of an alternative indicator for PM
11 visibility impairment, i.e. PM light extinction, instead of the traditional PM2 5 mass
12 concentration. The 2005 Staff Paper discussed the use of a four-hour afternoon PM2.5 standard,
13 where the underlying rational was that the generally lower afternoon relative humidity tended to
14 produce a more uniform relationship between light extinction and PM2.5 mass concentration
15 throughout the country, therefore providing a more uniform level of visibility protection
16 nationwide. However, this more uniform level of visibility protection was limited to the
17 afternoon hours of the day when relative humidity and visibility impairment are typically the
18 lowest. However, visibility conditions can be the poorest when relative humidity levels are the
19 highest. Thus, from a public welfare perspective, greater protection from visibility impairment is
20 needed during the times when humidity is high. In that regard, morning relative humidity
21 conditions, which are often generally higher in the Eastern US and coastal areas than in the West,
22 causes the same PM concentrations to produce much higher PM-related visibility impairment in
23 those regions than in areas with lower morning relative humidity resulting in unequal visibility
24 impairment at the national scale. Unlike PM mass concentration, which is determined by
25 removing the liquid water from the PM prior to measuring it, PM light extinction can be
26 measured at ambient humidity conditions so that it includes the enhanced light extinction
27 resulting from the liquid water that is associated with the hygroscopic PM components in the
28 atmosphere. PM light extinction, like PM mass concentration, is a measurable physical
29 characteristic of ambient PM. Thus, EPA believes that use of PM light extinction as the indictor
30 for a secondary PM NAAQS is a more appropriate target and more directly related to the welfare
31 effect of interest.
32 1.3 VISIBILITY EFFECTS SCIENCE OVERVIEW
33 Light extinction is the optical characteristic of the atmosphere that best determines the
34 impact potential of PM on perceived visibility. Light extinction is the loss of light per unit of
35 distance and occurs when light is either scattered or absorbed. Particulate matter and gases can
September 2009 1 -7 DRAFT - Do Not Quote or Cite
-------
1 both scatter and absorb light. Light scattering by gases (e.g., nitrogen, oxygen, etc.) that
2 comprise the atmosphere (also known as Rayleigh or clean-air scattering) is related to the density
3 of the air, which is sufficiently constant with elevation that it can be considered a known
4 constant value for any location. NC>2 is the only atmospheric pollutant gas that absorbs light
5 appreciably and its effects are generally small (i.e. less than 5%) compared to PM light
6 extinction, so its contribution to ambient visibility impacts is often ignored (as is done here). By
7 this assumption, light extinction is approximated as the sum of PM light extinction (including
8 both scattering and absorption) plus Rayleigh scattering, where the former characterizes the PM
9 contribution to visibility impacts, and the latter is taken to be a time invariant constant depending
10 only on elevation above sea level. In the same way PM light extinction is a good measure of the
11 degree of visibility impairment.
12 Visual air quality is defined as the visibility effect caused solely by air quality conditions
13 and excluding those associated with meteorological conditions like fog and precipitation. It is
14 commonly measured as either light extinction (in terms of inverse megameters, Mm"1) or the
15 deciview (dv) metric (Pitchford and Malm, 1993), which is a logarithmic function of extinction.
16 Extinction and deciviews are physical measures of the amount of visibility impairment (e.g., the
17 amount of "haze"), with both extinction and deciview increasing as the amount of haze increases.
18 A haziness index measured in deciview units was developed for use in visibility perception
19 studies because it has a more linear relationship to perceived changes in haze compared with
20 light extinction. The haziness index in deciviews (dv) is defined as ten times the natural
21 logarithmic of one tenth of the light extinction in inverse megameter units (Mm"1) (Pitchford and
22 Malm, 1993).
23 There is no simple one-to-one correspondence between PM concentration and PM light
24 extinction. However, as shown in Figure 1-1, the PM light extinction can be estimated from PM
25 composition and relative humidity data, using an algorithm with assumed light extinction
26 efficiencies for each of the major PM species and water growth factors for the hygroscopic
27 species. Though PM light extinction can be accurately determined by direct measurements, there
28 is only limited existing urban PM light extinction data. As a result, the assessment below will
29 principally use monitored and modeled PM mass, species estimates, and relative humidity
30 measurements.
31 The extent to which any amount of light extinction affects a person's ability to view a
32 scene depends on both scene and light characteristics. For example the appearance of a nearby
33 object (i.e. a building) is generally less sensitive to a change in light extinction than the
34 appearance of a similar object at a greater distance. For a scene with known characteristics, the
35 amount of degradation in the scene associated with a change in light extinction can be
September 2009 1 -8 DRAFT - Do Not Quote or Cite
-------
1 determined and the change in appearance can be realistically displayed on a digital photograph of
2 the scene using the WinHaze system.
3 Survey studies have used sets of photographs depicting a range of visibility conditions on
4 urban scenes to assess the public's opinion on the acceptability of conditions. For the specific
5 scenes used in such studies there is a known/predetermined one-to-one correspondence between
6 the percieved haze in the photographs and the amount of PM light extinction. For visibility
7 preference studies, visibility levels are generally characterized using the haze index in units of
8 deciview (similar to the decibel scale for sound).
9
10 Figure 1-1. Diagram showing the relationship steps between ambient PM and visibility
11 impairment.
Progression from PM Characteristics to Visibility Effects
ftwnPWtoVfta «_ .
d« Frafl^a rR^amBC s^«*>
^ TO QJWKMf &UCfe 3
2 n*,|G!emp«Itat — fr ^_JL
/"""" IMHKJVE'X IMniuolluntfihi A
'^^M&sMm^^* rwMwiiumffljy 4.
- WJUghtSeiHHfn|&__fc CpUcdCheraeisnitie -
® PMlbttAbeeipttai PM IgMmUMito) ®
1 " *s|—~J
!^r"!^^ T
taoehwfViwriAlr ft
OH^IlmfS^
1 "
^_____f____^^ PllMOICMI
*'x t,f»iiii.nn«*iiMiti»ii r^^_ ct^n^eittual 9
V W§ttlP8rl«IIWiHl -^^™ ML3n«un«™i w
N- ^ ImAMvme^lbtM
"~— «— . _^.—^"^ WOHIBB8B
1 . •
Vhlweflripwwd „
VWWI^ lfl
«-
«-
PM light extinction can be
approximated by a simple
algorithm using PM composition
and relative humidity data, or it can
be directly measured. There is no
simple one-to-one correspondence
between PM mass concentration
andPM light extinction.
These steps require non-air quality
information. Recent preference
survey studies used WinHaze
generated haze photos to hold the
scene and light conditions constant
so only the effects of light
extinction changes are seen. Survey
studies are tsed to help assess the
relevance of public and scene
contextual factors
12 1.4 GOALS AND APPROACH
13 The principal goal of the current UFVA is to characterize current levels of visibility
14 impairment, with a focus on urban areas, both in terms of the current secondary PM standards, as
September 2009
1-9
DRAFT - Do Not Quote or Cite
-------
1 well as in terms of alternative standards, including indicators and forms that may better reflect
2 the relationship between PM and visibility impairment. In particular, this UFVA focuses on the
3 effectiveness of a light extinction-based indicator for a possible secondary PM NAAQS (see
4 Figure 1-1). This is done by comparing estimates of hourly light extinction in 15 major U.S.
5 urban areas over the three-year period 2005-2007 to the CPLs, which are a range of light
6 extinction values beyond which half of the participants in assessed urban visibility preference
7 studies indicated the haze conditions were unacceptable (see discussion in chapter 2 below and
8 Stratus Consulting Inc., 2009). In addition, the UFVA will include additional characterizations of
9 the effectiveness of a sub-daily PM2.5 mass concentration indicator, which was explored in the
10 2005 PM staff paper and which was considered a viable option by EPA staff and CAS AC in the
11 2006 review. These latter assessments will be summarized in Appendix A.
12 The previous PM NAAQS review used the results of visibility preference survey studies
13 conducted in Denver (1990), Phoenix (2003), and British Columbia (1993) as the basis for
14 suggesting that a standard set to protect visibility conditions to a level within a visual range from
15 between about 40 km to about 60 km (corresponding to light extinction from -100 Mm"1 to -67
16 Mm"1) could represent an appropriate degree of welfare protection from PM. With the exception
17 of a small pilot study conducted in Washington, D.C. in 2001 (9 participants; Abt Associates
18 Inc., 2001), and a replicate study also conducted for Washington, D.C. in 2009 (26 participants;
19 Smith and Howell, 2009), there have been no additional visibility preference survey studies upon
20 which to base the selection of CPLs. The EPA staff, with contractor support, has conducted a
21 more detailed, in-depth assessment of the results from these studies, including the two recent
22 Washington, D.C. studies. This assessment includes an analysis that combines data from across
23 all studies to examine the consistency of the results between the surveys (Stratus Consulting Inc.,
24 2009). Based on the results of this analysis, we have been able to refine the range of visibility
25 conditions that could represent an appropriate degree of public welfare visibility protection that
26 was put forth in the 2006 review, and to determine a central tendency value for the CPLs. These
27 analyses and results are described below in chapter 2.
28 In the previous PM NAAQS review, the characterization of urban visibility conditions
29 were based on IMPROVE algorithm estimates using the 2001 to 2003 PM2.5 mass and speciation
30 data by assuming a constant composition for every hour of the day equal to the 24-hour
31 measured composition and by using either actual or monthly average (10-year mean) hour of the
32 day relative humidity for 161 urban area. Statistical relationships between hourly light extinction
33 estimates and concurrent hourly PM2.5 mass concentrations were used to show that daytime and
34 especially afternoon relationships are relatively strong with a similar linear relationship for both
35 eastern and western urban areas (i.e. R2>0.6, slope -6 m2/g). Relationships that included the
September 2009 1-10 DRAFT - Do Not Quote or Cite
-------
1 non-daylight hours were not as strong and differed more between eastern and western urban
2 areas.
3 The current assessment of urban visibility conditions (as described in chapter 3) is similar
4 in its development of an algorithm to estimate hourly light extinction using PM2.5 mass and
5 speciation data with measured relative humidity. However, it differs in that instead of assuming
6 constant composition for PM2.5, composition is made to vary during the day using urban-specific
7 monthly mean diurnal variations of species concentrations determined from regional air quality
8 model results, while constraining the means of the hourly species concentration for each day to
9 closely match the 24-hour duration measured species concentrations. The current assessment
10 examines 15 urban areas using 2005 to 2007 data sets (i.e. the same cities as used in the current
11 assessment for the primary standard).
12 1.5 ORGANIZATION OF DOCUMENT
13 The remainder of this document is organized as follows: Chapter 2 includes an analysis
14 of the urban visibility preference studies with a discussion of similarities and differences
15 regarding the approaches and methods used and results obtained for each study. This chapter
16 also includes a summary discussion of the results of a composite assessment of the combined
17 four city results and use of these results in the selection of the alternative levels evaluated in the
18 remainder of the assessment. Chapter 3 describes the analytical approach, methods, and data
19 used in conducting the assessment of recent urban visibility conditions, both in terms of PM2.5
20 and light extinction indicators for the set of urban case studies included in this analysis. Selected
21 results are presented in chapter 3, with additional results found in the Appendices. Chapter 4
22 presents estimates of PM2.5 and light extinction conditions generated for the urban case studies
23 for six alternative PM2.5 and light extinction scenarios. Additional information regarding
24 approaches and results for both chapters 3 and 4 are presented in Appendices A-F).
September 2009 1-11 DRAFT - Do Not Quote or Cite
-------
1 2 URBAN VISIBILITY PREFERENCE STUDIES
2 The purpose of this chapter is to present the reanalysis of the methods and results of
3 existing studies of preferences for urban visibility that EPA staff conducted (with contractor
4 support) in order to provide information useful in selecting a range of CPLs in terms of light
5 extinction values for subsequent use in the UFVA assessments of current and alternative VAQ
6 conditions. To date, available urban visibility preference studies have examined individuals'
7 desire for good VAQ by investigating the basic question, "What level of visibility degradation is
8 acceptable?" Preference studies have used a similar group interview type of survey to investigate
9 the level of visibility impairment that participants described as "acceptable." The specific
10 definition of acceptable is largely left to each individual survey participant, allowing each to
11 identify their own preferences.
12 The reanalysis effort included three completed urban visibility preference survey studies
13 plus a pair of smaller focus studies designed to explore and further develop urban visibility
14 survey instruments. The first urban visibility study conducted was in Denver, Colorado (Ely et
15 al., 1991), which developed the basic survey method used in all the subsequent studies. The two
16 other western studies included one in the lower Fraser River valley near Vancouver, British
17 Columbia (BC), Canada (Pryor, 1996), and one in Phoenix, Arizona (BBC Research &
18 Consulting, 2003). A pilot focus group study was also conducted for Washington, DC (Abt
19 Associates Inc., 2001). In response to an EPA request for public comment on the Scope and
20 Methods Plan (74 FR 11580, March 18, 200), Dr. Anne Smith provided comments (Smith, 2009)
21 about the results of a new Washington, D.C., focus group study that had been conducted using
22 methods and approaches similar to the method and approach employed in the EPA pilot study
23 (Smith and Howell, 2009). In total, 852 individuals participated in these studies in four cities,
24 with each individual responding to a series of questions answered while viewing a set of images
25 of various urban VAQ conditions.
26 2.1 METHODS USED IN PREVIOUS STUDIES
27 One direct physical measure of VAQ used in many visibility analyses is light extinction.
28 Light extinction is the loss of light per unit of distance, and measures the ability of particles and
29 gases in the atmosphere to scatter and absorb light traveling between an object and a person (or
30 camera). Extinction and haziness are physical measures of the amount of visibility impairment
31 (e.g., the amount of "haze"), with both extinction and haziness increasing as the amount of haze
32 increases.
September 2009 2-1 DRAFT - Do Not Quote or Cite
-------
1 In all but one6 of the visibility preference studies reviewed in this paper, participants
2 were shown a series of different VAQ conditions projected on a large screen using a slide
3 projector. In the earliest two studies (the Denver and lower Frazer River Valley studies) a range
4 of VAQ conditions were presented by projecting photographs (slides) of actual VAQ conditions.
5 The photographs were taken on different days from the same location, and presented the same
6 scene. Photographs were selected to avoid depicting significant weather events (e.g., rain, snow,
7 or fog), and where measured extinction data were available from the time the photograph was
8 taken.
9 The Phoenix study, as well as the subsequent Washington, D.C. survey instrument,
10 development projects, used photographic-quality images generated by a computer to present
11 different VAQ conditions. The images were developed from an original photograph using the
12 WinHaze software program, which is based on a technique described in Molenar et al. (1994).
13 The Phoenix study and the 2001 Washington, DC project projected slides of digital images
14 prepared by WinHaze. The 2009 Washington, DC project presented images directly from the
15 desktop version of WinHaze using either a liquid crystal display (LCD) projector or a computer
16 monitor.
17 WinHaze analysis synthetically superimposes a uniform haze on a digitized, near-pristine
18 actual photograph. The WinHaze computer algorithm calculates how a given extinction level
19 would impair the appearance of each individual portion of the photograph. A major advantage of
20 presenting WinHaze-generated images is that they provide viewers depictions of alternative
21 VAQ levels, with each image containing exactly the same scene, with identical light angle, time
22 of day properties, weather conditions, and specific scene content details (e.g., the amount of
23 traffic in a intersection). Additional details about WinHaze, and a discussion of the applicability
24 of WinHaze images for regulatory purposes, is in the 2004 PM Criteria Document (U.S. EPA,
25 2004). The desktop version of WinHaze is available online (Air Resources Specialists, 2008).
26 The first urban visibility preference study was conducted in Denver, Colorado (Ely et al.,
27 1991), and developed the basic survey method used in all the subsequent studies. Although there
28 are variations in specific details in each study, all the studies use a similar overall approach (key
29 variations are discussed in the section on each study later in this paper).
30 Visibility preference studies consist of a series of group interview sessions, where the
31 participants are shown a set of photographs or images of alternative VAQ conditions and asked a
32 series of questions. The group interview sessions are conducted multiple times with different
6 Smith and Howell (2009) used digital projection technology not available at the time of the other studies to present
the series of VAQ conditions. Some of the participants in the Smith and Howell study were shown images using a
LCD projector connected to a laptop computer. In other sessions, participants in the Smith and Howell study were
shown images on a computer monitor connected to the computer.
September 2009 2-2 DRAFT - Do Not Quote or Cite
-------
1 participants. Ideally the participants will be a representative sample of the residents of the
2 metropolitan area. While all studies agree that this is the preferred approach, due to the high cost
3 of organizing and conducting a series of in-person group interviews with a large, statistically
4 representative sample, only the Phoenix study was able to fully meet this objective.
5 During a group interview session, the participants were instructed to consider whether the
6 VAQ in each photograph or image would meet an urban visibility standard, according to their
7 own preferences and considering three factors:
8
9 1. The standard would be for their own urban area, not a pristine national park area
10 where the standards might be more strict
11 2. The level of an urban visibility standard violation should be set at a VAQ level
12 considered to be unreasonable, objectionable, and unacceptable visually
13 3. Judgments of standards violations should be based on visibility only, not on
14 health effects.
15 The photographs (images) are not shown in order of ascending or descending VAQ
16 conditions; the VAQ conditions are shown in a randomized order (with the same order used in
17 each group interview session). In order to check on the consistency of each individual's
18 answers, the full set of photographs (images) shown during the group interview included
19 duplicates with the identical VAQ conditions.
20 The participants were initially given a set of "warm up" exercises to familiarize them
21 with how the scene in the photographs or image appears under different VAQ conditions. The
22 participants next were shown 25 randomly ordered photographs (images), and asked to rate each
23 one based on a scale of 1 (poor) to 7 (excellent). They were then shown the same photographs or
24 images again (in the same order), and asked to judge whether each of the photographs (images)
25 would violate what they would consider to be an appropriate urban visibility standard (i.e.,
26 whether the level of impairment was "acceptable" or "unacceptable").
27 2.2 DENVER, COLORADO
28 The Denver urban visibility preference study (Ely et al., 1991) was conducted on behalf
29 of the Colorado Department of Public Health and Environment (CDPHE). The study consisted
30 of a series of focus group sessions conducted in 1989 with participants from 16 civic
31 associations, community groups, and employees of state and local government organizations.7
7 No preference data were collected at a 17th focus group session due to a slide projector malfunction.
September 2009 2-3 DRAFT - Do Not Quote or Cite
-------
1 The participants were not selected to be a fully representative sample of the Denver metropolitan
2 population but were instead selected to take advantage of previously scheduled meetings.
3 During the 16 focus group sessions, a total of 214 individuals were asked to rate
4 photographs of varying visibility conditions in Denver. The photographs were taken November
5 1987 through January 1988 by a camera in Thornton, Colorado. Thornton is suburb of Denver,
6 located approximately six miles north of downtown Denver. The photographs were taken as part
7 of a CDPHE study of Denver's air quality. The scene in the photographs was toward the south
8 from Thornton and included a broad view of downtown Denver and the mountains to the south.
9 Each group was shown one of two sets of 20 randomly ordered unique photographs (13 of the
10 sessions included 5 duplicate slides, for a total of 25 photographs, to evaluate consistency of
11 responses). The two sets of different slides were used to investigate whether the responses
12 between the two sets of photographs were different (no differences were found). Approximately
13 100 participants viewed each photograph. Projected color slides were used to present the
14 photographs to focus group participants, and were projected on a large screen
15 The VAQ conditions in each Denver photograph were recorded when the photograph was
16 taken and measured by a transmissometer yielding hourly average light extinction, bext. The
17 transmissometer was located in downtown Denver, approximately eight miles from the camera
18 and in the middle of the camera's view path. Ely et al. (1991) provide the time of day and
19 measured extinction level for each photograph. The extinction levels presented in the Denver
20 photographs ranged from 30 to 596 Mm"1. This corresponds to 1 Idv to 41dv, approximating the
21 10th to 90th percentile of wintertime visibility conditions in Denver in the late 1980s.
22 The participants first rated the VAQ in each photograph on a 1 to 7 scale, and
23 subsequently were asked if each photograph would violate an urban visibility standard. The
24 individual's rating on the 1 to 7 scale and whether the photograph violated a visibility standard
25 were highly correlated (Pearson correlation coefficient greater than 80%).
26 The percent of participants who found a photograph acceptable to them (i.e., would meet
27 an appropriate urban visibility standard) was calculated for each photograph. Figure 2-1 shows
28 the results of the Denver participants' responses, with VAQ measured in deciviews.
29 Ely et al. (1991) introduce a "50% acceptability" criteria analysis of the Denver
30 preference study results. The 50% acceptability criteria is designed to identify the VAQ level
31 that best divides the photographs into two groups: those with a VAQ rated as acceptable by the
32 majority of the participants, and those rated not acceptable by the majority of participants. While
33 no single VAQ level creates a perfect separation between the two groups, the CDPHE identified
34 a VAQ of 20.3 dv as the point that best separates the Denver study responses into "acceptable"
35 and "not acceptable" groups. Based in part on the findings of the Denver visibility preference
36 study, the CDPHE established a Denver visibility standard at bext = 76 Mm"1 (dv = 20.3).
September 2009 2-4 DRAFT - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Figure 2-1. Percent of Denver participants who consider VAQ in each photograph
"acceptable."
100%
0)
"- 0)
« S
11 50%
'€ re
CO :
Q.
0%
4*
10
15
20
25 30
Deciview
35
40
45
% acceptable - Denver standard
Using 20.3 dv as a 50% acceptability criteria led to six photographs being inconsistently
rated by the majority of the viewers. A photograph was inconsistently rated for two possible
reasons; either the photograph's VAQ was at least 1 dv better than the Denver standard (i.e., dv <
19.3) but was judged to be "unacceptable" by a majority of the participants rating that
photograph, or the VAQ was at least 1 dv worse than the standard (> 21.3 dv) but found to be
acceptable by the majority of the participants. This definition of inconsistent rating helps
evaluate the robustness of the study results to support the selection of the Denver urban visibility
standard at 76 Mm"1 (20.3 dv) by identifying photographs with VAQ a minimum of 1 dv above
or below the standard and ignoring "near misses" involving photographs within 1 dv of the
standard. 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
(U.S. EPA, 2004).
Table 2-1 presents information about the six photographs that were inconsistently rated.
All six of the inconsistently rated photographs were taken at 9:00 a.m. The five inconsistently
rated photographs with a VAQ better than the Denver standard have a VAQ at least 2 dv below
the standard. The VAQ in the only inconsistently rated photograph with air quality worse than
the standard (Photograph #6) is 1.1 dv above the standard. The study used 18 photographs from
9:00 a.m., so a third of the 9:00 a.m. photographs were inconsistently rated. Conversely, none of
the 32 photographs taken at noon or 3:00 p.m. were inconsistently rated.
September 2009
2-5
DRAFT - Do Not Quote or Cite
-------
1
2
Table 2-1. VAQ of Denver photos substantively misclassified by majority of participants
Photograph #
14
18
19
20
24
36
VAQ in photograph
in extinction
(Mm1)
44
54
54
55
60
85
VAQ in
photograph (dv)
13.8
16.9
16.9
17.0
17.9
21.4
% of participants
who rated the photo
"acceptable"
43%
43%
31%
42%
13%
72%
Time of day of
photograph
9:00 a.m.
9:00 a.m.
9:00 a.m.
9:00 a.m.
9:00 a.m.
9:00 a.m.
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Figure 2-2 shows the same data results about percent of participants who rated each
photograph acceptable as in Figure 2-1, but with the time of day of each photograph indicated by
different colors. The time of day colors clearly indicate how inconsistently participants rated
some of the 9:00 a.m. photographs.
Eliminating the 9:00 a.m. photographs creates a "hole" in the range of remaining
photographs; there are no photographs with a VAQ between 17.7 dv and 20.3 dv. As seen in
Figure 2-2, this is a critical range in evaluating the responses. All of the photographs with a VAQ
equal to or better (i.e., a lower dv value) than 17.7 dv are rated acceptable by the majority of the
participants, and all photographs with a VAQ at or above 20.3 dv are rated not acceptable. After
eliminating the 9:00 a.m. photographs, any VAQ level between 17.7dv and 20.3 dv would
completely divide the photographs into two groups with no inconsistent ratings.
A modestly broader range of VAQ conditions provides an even more unambiguous
interpretation of the Denver study results. Every photograph with a VAQ of 17.7 dv or lower was
rated acceptable by 89% or more of the participants, and every photograph with a VAQ of 24.6
or higher was rated not acceptable by 84% or more of the participants. The 17.7 dv to 24.6 dv
range separating the results is shown in Figure 2-3, which also eliminates the 9:00 a.m. results.
September 2009
2-6
DRAFT - Do Not Quote or Cite
-------
1
2
3
4
5
Figure 2-2. Photograph time of day information for the percent of participants who
consider VAQ in each photograph "acceptable."
100%
O)
_c
15 :
« 3
I g-50%
'o *J
t (Q
RJ :
Q.
0%
% •
10
15 20 25 30
Deciview
35
40
45
9:00 AM • 12:OOPM 3:00 PM—Denver standard
Figure 2-3. Denver photograph time of day results (9:00 a.m. photographs eliminated),
with the broader range (17.7 dv and 24.6 dv) of the 50% acceptability criteria shown.
% Participants Rating
"Acceptable"
en c
3 O C
9 sP ~,
x 6s c
. •' 1 • •' .
•
'
, - (
•
> » ••
10 15 20 25 30 35 40 45
Deciview
• 12:OOPM 3:00 PM Denver Standard LowerBound UpperBound
6 2.3 VANCOUVER, BRITISH COLUMBIA, CANADA
7 The BC urban visibility preference study (Pryor, 1996) was conducted on behalf of the
8 BC Ministry of Environment following the methods used in the Denver study. Participants were
9 students at the University of British Columbia, who were in one of four focus group sessions
10 with between 7 and 95 participants. A total of 180 participants completed the surveys (29 did not
11 complete the survey).
September 2009
2-7
DRAFT - Do Not Quote or Cite
-------
1 The BC study used photographs (projected as slides) depicting various VAQ conditions
2 in two cities (Chilliwack and Abbotsford) in the lower Fraser River valley in southwestern BC.
3 Abbotsford is located approximately 75 miles east of Vancouver, BC, and had a 2006 population
4 of 159,000 (Statistics Canada, 2009a). Abbotsford has a diverse and successful economy, with
5 approximately 25% of the labor force working in the Vancouver metropolitan area. Chilliwack is
6 adjacent to Abbotsford to the east. Both cities have experienced rapid population growth,
7 growing faster than the Vancouver metropolitan area, and are considered suburbs (or exurbs) of
8 Vancouver.
9 The survey was conducted at the University of British Columbia (UBC) in 1994. The
10 participants were 206 undergraduate and graduate students enrolled in classes in UBC's
11 Department of Geography. Information about student demographics and where they lived prior
12 to enrolling at UBC (which potentially influences their knowledge of, and preferences for,
13 Vancouver area visibility) is not available.
14 The BC survey showed 20 unique photographs to the participants in random order. Ten
15 photographs were from Chilliwack, and 10 were from Abbotsford. The Chilliwack photographs
16 were taken at the Chilliwack Hospital, and the scene includes a complex foreground with
17 downtown buildings, with mountains in the background up to 40 miles away. Figure 2-4 is a
18 composite of two of the Chilliwack photographs used in the preference study, showing the scene
19 with a good visibility day (14.1 dv) in the middle and a significantly impaired day (34 dv) around
20 the border (Jacques Whitford AXYS, 2007). The Abbotsford photographs were taken at the
21 Abbotsford Airport. The Abbotsford scene includes fewer man-made objects in the foreground
22 and is primarily a more rural scene with the mountains in the background up to 36 miles away.
23 The photographs were taken in July and August 1993 as part of a VAQ and fine
24 parti culate monitoring project sponsored by the BC Ministry of Environment, Lands and Parks
25 (REVEAL, the Regional Visibility Experimental Assessment in the Lower Fraser Valley). All of
26 the photographs were taken at either 12:00 p.m. or 3:00 p.m. VAQ data were available for each
27 photograph from visibility monitors near the location of each camera. The types of VAQ
28 measurement data available from the two locations were not identical. The Chilliwack location
29 used both an open-chamber nephelometer and a long path transmissometer and collected hourly
30 average data on both aerosol light scattering (bsp) and total extinction (bext), respectively. The
31 visibility monitoring at the Abbotsford location had only a nephelometer and collected only bsp
32 data.
September 2009 2-8 DRAFT - Do Not Quote or Cite
-------
1
2
Figure 2-4. Composite Chilliwack, BC photograph showing VAQ of 14.1 dv and 34 dv.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Total light extinction is the sum of scattering by gases (bsg) and particles (bsp) plus light
absorption by gases (bag) and particles (bap). In order to present the preference results from the
BC study in comparable terms, bext for the Abbotsford photographs is estimated by assuming that
the average of the ratios of PM light extinction (i.e., bap+ bsp) to PM light scattering (bsp) for all
ten of the Chilliwack photographs can be multiplied by the Abbotsford nephelometer determined
bsp values corresponding to each of its photographs to estimate its PM light extinction value. By
assuming that absorption by gases (bag) is zero, total light extinction is equal to the PM light
extinction (i.e., bap+ bsp) plus particle scattering by gases (i.e., bsg that is approximately equal to
lOMm"1). Table 2-2 presents the data from the photographs used in the BC study, including the
estimated bext for the Abbotsford photographs.
There are two caveats to be noted about the extinction data for the photographs reported
in Pry or, 1996. First, in Table 2 of the original article, two of the Abbotsford photographs are
listed with the same date and time (12:00 p.m., 7/26/1993). There is no information provided for
a 3:00 p.m., 7/26/1993 Abbotsford photograph, although there is a Chilliwack photograph from
that time. The preference and VAQ data are presumed to be correct for both photographs and
one of the two identical date/time labels is assumed to be a typographic error. The second caveat
is that b^ levels from the same date and time can differ substantially between Abbotsford and
Chilliwack, and the relative levels can change rapidly, even though the two cities are only 25
miles apart. For example, at 12:00 p.m. on 8/19/1993, the bsp level in Chilliwack was about one-
September 2009
2-9
DRAFT - Do Not Quote or Cite
-------
1 Table 2-2. Summary of photographs used in British Columbia study
2
Date
Time
bsp
bext
Ratio
(bexrbsg)/bsp
Estimated
bext
Deciview
Chilliwack
7/26/93
7/26/93
7/27/93
7/27/93
8/2/93
8/2/93
8/5/93
8/5/93
8/19/93
8/19/93
12:00 p.m.
3:00 p.m.
12:00 p.m.
3:00 p.m.
12:00 p.m.
3:00 p.m.
12:00 p.m.
3:00 p.m.
12:00 p.m.
3:00 p.m.
Average
86
67
63
119
18
20
45
51
46
105
62
128
112
105
185
37
36
70
96
81
170
102
1.372
1.522
1.508
1.471
1.5
1.3
1.333
1.686
1.543
1.524
1.476
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
25.49
24.16
23.51
29.18
13.08
12.81
19.46
22.62
20.92
28.33
21.96
Abbotsford
7/26/93
7/26/93
7/27/93
7/27/93
8/2/93
8/2/93
8/5/93
8/5/93
8/19/93
8/19/93
12:00 p.m.
12:00 p.m.
12:00 p.m.
3:00 p.m.
12:00 p.m.
3:00 p.m.
12:00 p.m.
3:00 p.m.
12:00 p.m.
3:00 p.m.
Average
39
82
104
132
24
25
62
75
67
145
76
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
68
131
205
164
45
47
121
102
224
109
122
19.17
25.73
30.20
27.97
15.04
15.48
24.93
23.22
31.09
23.89
23.67
September 2009
2-10
DRAFT - Do Not Quote or Cite
-------
1 Figure 2-5. Percent of BC participants who consider VAQ in each photograph
2
-3
"acceptable."
100%
"o
.0
3
a.
o
| 50%
0)
c
•5
0£
1
2. o%
1 1
n
Q.
Results of British Columbia Visiblity Study
^ »
'-.•.. *
0 15 20
25 30
Deciview
» Chilliwack • Abbotsford
35 40 45
4
5 third of the Abbotsford bsp level. By 3:00 p.m. the situation was reversed, with the Chilliwack
6 bsp level 50% higher than Abbotsford. In those three hours the Chilliwack bsp level had over
7 doubled (from 46 Mm"1 to 105 Mm"1), and the Abbotsford level had fallen by over half (from
8 145 Mm"1 to 67 Mm"1). Such substantial changes in measured bsp levels occurring across a
9 relatively short period of time and short distance, may reflect an inherent uncertainty introduced
10 by using a single measure of light extinction from a portion of visual scene (where the
11 nephelometer or transmissometer was operating) to assess visibility conditions throughout an
12 actual photographs of a complex scene. Spatial and temporal non-uniformity of visibility
13 conditions within a scene are an atmospheric condition known to occur on some days, and may
14 contribute to the variability in participant responses in preference studies utilizing actual
15 photographs.
16 Figure 2-5 presents the results of the BC study. The division corresponding to the
17 Denver "50% acceptable" criteria occurs between 22.6 dv and 23.2 dv. All of the photographs
18 with a VAQ better than 22.6 dv were rated acceptable by the majority of the participants with
19 one exception (47% of the participants judged the 19.2 dv photograph to be acceptable). All
20 photographs with a VAQ better than 19.2 dv were rated acceptable by over 90% of the
21 participants. All photographs with a VAQ worse than 22.6 dv were rated not acceptable by the
September 2009
2-11
DRAFT - Do Not Quote or Cite
-------
1 majority of the participants, and all photographs with a VAQ worse than 28.3 dv were rated not
2 acceptable by over 90% of the participants.
3 Figure 2-5 also suggests that there may be some difference between the preferences
4 expressed for the Chilliwack scene and those for the Abbotsford scene. All photographs were
5 rated by the same individuals (students at UBC), but the summary of the responses indicate that
6 the participants may have rated as acceptable a worse level of impaired VAQ impairment (e.g.,
7 higher dv levels) in photographs showing more of a downtown area (Chilliwack) than in less
8 congested scenes (Abbotsford). The strongest evidence for this hypothesis, however, is the
9 preference for a single photograph (the 19.0 dv photograph from Abbotsford, rated as acceptable
10 by 47%), previously identified as an outlier observation.
11 The BC Ministry of the Environment is considering the BC urban visibility preference
12 study as part of establishing urban and wilderness visibility goals in BC.
13 2.4 PHOENIX, ARIZONA
14 The Phoenix urban visibility preference study (BBC Research & Consulting, 2002),
15 which was conducted on behalf of the Arizona Department of Environmental Quality, used
16 group interviews based on the methods used in the Denver study, with two major exceptions: (1)
17 the focus group participants were selected as a representative sample of the Phoenix area
18 population, and (2) the pictures presented in the focus groups were computer-generated images
19 to depict specific uniform haze conditions.
20 The Phoenix study included 385 participants in 27 separate focus group sessions.
21 Participants were recruited using random digit dialing to obtain a sample group designed to be
22 demographically representative of the larger Phoenix population. During July 2002, group
23 interview sessions took place at six neighborhood locations throughout the metropolitan area to
24 improve the participation rate. Participants received $50 as an inducement to participate.
25 Three sessions were held in Spanish in one region of the city with a large Hispanic
26 population (25%), although the final overall participation of native Spanish speakers (18%) in
27 the study was below the targeted level. The age distribution of the participants corresponded
28 reasonably well to the overall age distribution in the 2000 U.S. Census for the Phoenix area
29 (BBC Research & Consulting, 2002). Participants slightly over-represented the middle-income
30 range ($50,000 to $74,999), compared with 2000 Census data, and slightly under-represented
31 very low-income ranges (under $24,999). The distribution of participant education levels was
32 fairly consistent with the education distribution in the 2000 Census.
33 Photographic-quality slides of the images were developed using the WinHaze software
34 (Molenar et al., 1994). The scene used in the Phoenix study images was taken at a water
35 treatment plant. The view is toward the southwest, including downtown Phoenix, with the Sierra
September 2009 2-12 DRAFT - Do Not Quote or Cite
-------
1 Estrella Mountains in the background at a distance of 25 miles. Figure 2-6 shows the image with
2 the best VAQ (15 dv).
3
4 Figure 2-6. Reproduction of the image with the best VAQ (15 dv) used in the Phoenix
5 study.
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
The study used a total of 21 unique WinHaze images. Four of the 21 unique images were
randomly selected and used twice to evaluate consistency; participants viewed a total of
25 images. The 25 images were randomly ordered, with all participants viewing the images in
the same order. The WinHaze images used in the Phoenix study do not include layered haze, a
frequent and widely recognized form of visibility impairment in the Phoenix area.
The VAQ levels in the 21 unique images ranged from 15 dv to 35 dv (the extinction
coefficient bext ranged from 45 Mm"1 to 330 Mm"1). As in the Denver study, participants first
individually rated the randomly shown slides on a VAQ scale of 1 (unacceptable) to 7
(excellent). Participants were instructed to rate the photographs solely on visibility and to not
base their decisions on either health concerns or what it would cost to have better visibility. Next,
the participants individually rated the randomly ordered slides as "acceptable" or "not
acceptable," defined as whether the visibility in the slide is unreasonable or objectionable.
Figure 2-7 presents the percent acceptability results from the Phoenix study. The
combination of the use of WinHaze images and the larger number of participants than in the
Denver study may account for the "smoother" backwards S-shaped pattern of preferences.
September 2009
2-13
DRAFT - Do Not Quote or Cite
-------
1
2
Figure 2-7. Percent of Phoenix participants who consider VAQ in each image
"acceptable."
100%
10
15
20
25
30
35
40
45
Deciview
4 90% or more of the participants rated a VAQ of 20 dv or better as acceptable, and 70% rated a
5 VAQ of 22 dv or better as acceptable. The "50% acceptable criteria" was met at approximately
6 24.3 dv (with 51.3% of the participants rating that image as acceptable). The percent
7 acceptability declines rapidly as VAQ worsens; only 27% of the participants rated a 26 DV
8 image as acceptable, and fewer than 10% rated a 29 dv image as acceptable.
9 The Phoenix urban visibility study formed the basis of the decision of the Phoenix
10 Visibility Index Oversight Committee for a visibility index for the Phoenix metropolitan area
11 (Arizona Department of Environmental Quality, 2003). The Phoenix Visibility Index establishes
12 an indexed system with 5 categories of visibility conditions, ranging from "Excellent" (14 dv or
13 less, which was a better VAQ than any of the images used in the Phoenix study) to "Very Poor"
14 (29 dv or greater, which less than 10% of the study participants rated as acceptable). The "Good"
15 range is 15 dv to 20 dv (more than 90% of the participants rated images in this VAQ range as
16 acceptable). The environmental goal of the Phoenix urban visibility program is to achieve
17 continued progress through 2018 by moving the number of days in poorer quality categories into
18 better quality categories.
19 2.5 WASHINGTON, B.C.
20 One of the Washington, D.C. urban visibility pilot studies was conducted on behalf of
21 EPA (Abt Associates Inc., 2001). It was designed to be a pilot focus group study, an initial
22 developmental trial run of a larger study. The intent of the pilot study was to refine both focus
23 group method design and potential survey questions. Due to funding limitations, only a single
September 2009
2-14
DRAFT - Do Not Quote or Cite
-------
1 focus group session took place, consisting of one extended session with nine participants. No
2 further urban visibility focus group sessions were held in Washington, DC, on behalf of EPA.
3 In March 2009, Dr. Anne Smith conducted a separate study of Washington urban
4 visibility, using the same photographs and similar approach as the 2001 study (Smith and
5 Howell, 2009). On behalf of the Utility Air Regulatory Group, Dr. Smith presented comments
6 (Smith, 2009) to the CAS AC at a public meeting held on April 2, 2009 to review EPA's plan
7 (US EPA, 2009b) for conducting further urban visibility studies in support of PM NAAQS
8 reviews. Dr. Smith submitted the Smith and Howell (2009) report to the CASAC as part of the
9 public comment process. The Smith and Howell study conducted three study variations of a
10 Washington, DC, preference study, including one experiment involving 26 participants designed
11 to replicate the EPA 2001 preference study.
12 Both the Abt Associates Inc. (2001) study results and the results of the Smith and Howell
13 (2009) study are discussed below.
14 2.5.1 Washington, B.C. 2001
15 The EPA's Washington, D.C. study (Abt Associates Inc., 2001) adopted the general
16 study methods used in the Denver, BC, and Phoenix studies, modifying them appropriately to be
17 applicable in an eastern urban setting. Washington's (and the entire East's) current visibility
18 conditions are typically substantially worse than western cities and have different characteristics.
19 Washington's visibility impairment is primarily a uniform whitish haze dominated by sulfates,
20 and the relative humidity levels are higher compared with the western study areas. In addition,
21 the relatively low-lying terrain8 in Washington, D.C., provides substantially shorter maximum
22 sight distances. Many residents are not well informed that anthropogenic emissions impair
23 visibility on hazy days.
24 The Washington, D.C. focus group session included questions on valuation, as well as on
25 preferences. The focus group content dealing with preferences for an urban visibility standard
26 was similar to the focus group sessions in the Denver, BC, and Phoenix studies.
27 A single scene of a panoramic photograph taken from Arlington National Cemetery in
28 Virginia was used, and included an iconic view of the Potomac River, the National Mall, and
29 downtown Washington, D.C. All of the distinct buildings in the scene are less than four miles
30 from the camera, and the higher elevations in the background are less than 10 miles from the
31 camera. Figure 2-8 presents the photograph used in the study.
8The maximum elevation in Washington, DC is 409 feet.
September 2009 2-15 DRAFT - Do Not Quote or Cite
-------
1 Figure 2-8. Reproduction of the image with the best VAQ (8.8 dv) used in the Washington,
2 B.C. study.
3
4 The Washington, D.C. study used 20 unique images generated by WinHaze, each
5 prepared from the same original photograph. Humidity and gaseous light scattering was held
6 constant in preparing the WinHaze images, as was the relative chemical mix of aerosol
7 particulates in the photos (i.e., only the aerosol concentrations were increased to create the
8 images with worse VAQ). Five of the images were repeated as a consistency check, so
9 participants viewed a total of 25 slides. The range of VAQ in the images ranged from 8.8 to 38.3
10 dv, which is approximately the 10th to the 90th percentile of the annual distribution of hourly
11 VAQ conditions in Washington.
12 Figure 2-9 presents the percent acceptability results from the 2001 Washington study.
13 Because only nine participants were involved in the study, the possible values of "percent
14 acceptable" are limited to multiples of 1/9. Figure 2-9 also shows an anomalous result involving
15 one of the five repeated images. Three of the repeat images had the same ranking each time they
16 were presented (i.e., all nine participants rated them acceptable or not acceptable both times they
17 rated that slide). One of the images (the image with 8.8 dv, the best VAQ image used in the
18 study) was rated acceptable by all nine participants the first time it was used, but the repeat of
19 that slide was rated not acceptable by one participant. Another image, however, had a
20 substantially different result. The 30.9 dv image was rated acceptable by five of the nine
21 participants the first time it was presented, but the repeat of the slide was only rated acceptable
22 by one of the nine participants. The responses for all five pairs of repeated images are shown in
September 2009
2-16
DRAFT - Do Not Quote or Cite
-------
1 red on Figure 2-9, including the images which were identically rated both times they were
2 presented.
3 Figure 2-9. Percent of 2001 Washington participants who consider VAQ acceptable in each
4 image.
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Results of 2001 Washington DC Preference Study
100%
O)
c
•j=
(0 =
(£ .SJ
^ 1
= •£
Is
£3
(0 =
50%
0%
-•—»-
5 10 15 20 25 30 35 40 45
Deciview
* Unique Images • 5 Repeated Images
In the 2001 Washington, D.C. study, all images with a VAQ below 25.9 dv were rated
acceptable by the majority of the participants, and all images with a VAQ below 29.2 dv were
rated acceptable by at least four of the nine (44%) participants. All images with a VAQ above
30.9 dv were rated not acceptable. The "50% acceptability criteria" division occurs in the range
of 25.9 dv to 30.9 dv, with the anomalous result of the inconsistent responses to the repeated
image with 30.9 dv effectively broadening this range and adding uncertainty to identifying a
clear division.
2.5.2 Washington, D.C., 2009
The Smith and Howell (2009) study conducted additional focus group sessions based on
the methods and materials used in the 2001 Washington, D.C. study. Smith and Howell
recreated the WinHaze images used in the 2001 Washington, D.C. urban visibility preference
study, using the description in the report on the 2001 study (Abt Associates Inc., 2001), and
created images using currently available desktop computer version of WinHaze (Version 2.9.0).
Smith and Howell used a shortened version of the same question protocol as the 2001 study. The
WinHaze images were presented to a total of 64 participants who were all employees of Charles
River Associates (CRA International, Inc). (Smith and Howell also are CRA International
September 2009
2-17
DRAFT - Do Not Quote or Cite
-------
1 employees). The CRA employees were based at the firm's Washington, D.C. and Houston,
2 Texas offices (44 and 20 participants, respectively). The Houston participants were included to
3 explore whether familiarity with Washington, D.C. VAQ conditions developed from currently
4 living in the Washington region noticeably influenced the responses. As noted by Smith and
5 Howell, the participants were not a representative sample of either metropolitan area's
6 population; all participants were employed, and the participant group included a higher
7 proportion of college educated individuals and higher household incomes than the general
8 population.
9 Eight of the Washington-based participants and all of the Houston participants viewed the
10 WinHaze images on a desktop computer monitor. The remaining Washington participants
11 viewed the images projected on a screen.
12 The stated purpose of the Smith and Howell study was to explore the robustness of the
13 2001 results. To investigate this issue, Smith and Howell conducted three different tests
14 concerning urban visibility preferences. Each participant was involved with only one test. The
15 three tests were:
16 * Test 1 - replicated the Abt Associates Inc. (2001) study
17
18 * Test 2 - reduced the upper end of the range of VAQ by eliminating the 11 images
19 used in Test 1 with a VAQ above 27.1 dv
20
21 * Test 3 - increased the upper end of the range of VAQ by including two new images
22 of worse VAQ; the two new images had a VAQ of 42 dv and 45 dv
23
24 Sixteen employees from the Washington, D.C. office and 10 participants from the
25 Houston office took Test 1 (a total of 26 participants). All the participants viewed the same
26 unique 20 Washington, DC WinHaze images as the 2001 study (plus repeated images for a total
27 of 25 images shown to participants). Images were presented in the same random order as in the
28 2001 study. Figure 2-10 presents the results of Test 1. The results for the 16 Washington
29 participants are indicated in blue and results for the 10 Houston participants in red. Although all
30 images used in the study were of Washington, D.C., the results suggest that there is not a
31 significant difference in the preferences of participants based in the two offices. The scene in the
32 images is an immediately recognizable iconic view of the National Mall and downtown
33 Washington, D.C., which may influence the similarity of responses by residents of the two cities.
34 Using the combined Test 1 results from the two CRA offices (26 total participants), the
35 majority of participants in the 2009 study rated all VAQ images with 25.9 dv or less as
36 acceptable and all VAQ images with 29.2 dv or greater as not acceptable. The image of 27.1 dv
37 was rated as acceptable by 50% of the total participants (56% of the Washington-based and 40%
September 2009 2-18 DRAFT - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
of the Houston-based participants). All images with a VAQ less than 22.9 dv were rated
acceptable by at least 90% of the participants, and all images with a VAQ greater than 32.3 dv
were rated not acceptable by 88% of the participants.
Figure 2-10. Percent of 2009 Test 1 study participants who consider VAQ acceptable in
each image, showing the range of the lower and upper bound of 50% acceptability criteria.
100%
Q.
0)
U
u
O)
ra
a.
'o
'•E
0%
_
i
L A A
• A
A _A
•
1
_
"••;*
5 10 15 20 25 30 35 4
Deciview
• Washington A Houston Lower Bound Upper Bound
9 Figure 2-11 presents the 2001 and 2009 study (Test 1) results on a single graph,
10 representing the results of 35 total participants of preferences for urban visibility in Washington,
11 DC. The results from the 2009 study on Figure 2-11 combine the Test 1 responses from the two
12 CRA offices. Figure 2-11 also shows the 50% acceptability criteria range (22.9 dv to 32.3 dv)
13 from the 2009 study, Test 1. In comparison, the 2001 study 50% acceptability range was 25.9 dv
14 to 30.9 dv. Inspection of the points in Figure 2-11 indicate that the results from the 2009 study
15 (Test 1) are not appreciably different than the results of the 2001 Washington study.
16 In Test 2, Smith and Howell reduced the range of VAQ images presented to 26
17 participants to images with a VAQ of 27.1 dv or less. The 26 participants were different people
18 than the Test 1 participants. Test 2 presented only the nine unique clearest WinHaze images
19 from the full Test 1 set of 20 images. This constricted the VAQ levels presented to the range that
20 the majority of participants in the 2001 study rated as acceptable and reduced the upper end of
21
September 2009
2-19
DRAFT - Do Not Quote or Cite
-------
1 Figure 2-11. Combined results of two Washington preference studies (showing 50%
2 acceptability criteria from 2009, Test 1).
.a
as
+j
a.
10 15 20 25 30 35 4
Deciview
B 2001 Study 4 2009, Test 1 Lower Bound Upper Bound
5
6
Figure 2-12. Comparison of results from Test 1 and Test 2 (Smith and Howell, 2009).
-i nno/
B
S
Q.
§
<
C
'•& RCtO/
(3 ou /o
c
&
.0
ns
Q.
ss
noA
i
Results of Smith an
i
•
i
•
5 10 15
d Howell Test
i «
• «
20 25
Deciview
• Test 2 » Test 1
1 and Test 2
«
•
• «
•
•
i
•
30 35 4
0
September 2009
2-20
DRAFT - Do Not Quote or Cite
-------
1 the VAQ range by 11.2 dv. Nine unique WinHaze images were used in Test 2, with three
2 duplicates included, so Test 2 participants were shown 12 images. Figure 2-12 presents the Test
3 1 and Test 2 results. Test 2 found a substantial shift in the responses about which VAQ level is
4 considered acceptable. The smaller number of images used in Test 2 makes identifying the range
5 of the 50% acceptability criteria more difficult than in Test 1. The lower bound of the range
6 occurs between 15.6 and 18.7 dv, and the upper bound occurs between 24.5 and 27.1 dv. Smith
7 and Howell conclude that the shift in the acceptability responses between Test 1 and Test 2
8 suggests that the acceptable responses in an urban visibility preference study conducted using the
9 general approach used in the all the studies may be susceptible to the range of VAQ images
10 presented.
11 One hypothesis (not raised by Smith and Howell) suggested by the Test 2 results is that
12 the 50% acceptability criteria occurs near the middle of the range of images shown to
13 participants. This might be the result of the participants consciously or subconsciously
14 identifying approximately the middle of the VAQ range presented to them. Participants (in all
15 the studies reviewed in this paper) were shown all the images as part of "warm up" exercises and
16 a separate initial rating exercise (ranking the VAQ in each image on a scale of 1 to 7). These
17 initial reviews of the images allow participants to become familiar with the range of VAQ and
18 may consciously or subconsciously calibrate their subsequent responses to the VAQ range they
19 were presented.
20 In Test 3, Smith and Howell expanded the VAQ range of WinHaze images shown to the
21 participants, including two new images with a worse VAQ. The new images had a VAQ of 42
22 dv and 45 dv, raising the upper end of the VAQ range by 6.7 dv. Test 3 reduced the total number
23 of images shown to participants to 19 images by eliminating the use of the five repeat images in
24 Test 1, and also eliminated three additional images in order to reduce the participants' time
25 burden. The three deleted images had a VAQ of 11.1, 15.6, and 24.5 dv. The best VAQ image
26 shown to Test 3 participants was 8.8 dv (same as the best VAQ image in Tests 1 and 2).
27 However, in Test 3 there were no images with VAQ between 8.8 dv and 18.7 dv, creating a
28 significant "hole" in the distribution of VAQ conditions presented to the Test 3 participants.
29 Test 3 was conducted with 12 participants from the CRA Washington office (none of whom
30 participated in Test 1 or Test 2). No Houston participants were involved with Test 3. The results
31 of Test 3 are shown in Figure 2-13, along with the results of Test 1.
32 Increasing the upper end of the VAQ range in Test 3 resulted in an overall increase in the
33 percent of respondents rating as acceptable the VAQ images used in both tests. In Test 3 all
34 images with a VAQ below 22.9 dv were rated acceptable by 100% of the participants (similar to
35 the Test 1 results), implying there was no general change in the acceptability of the images with
36 good VAQ. However, for all VAQ images (that were used in both studies) between 25.9 dv and
September 2009 2-21 DRAFT - Do Not Quote or Cite
-------
1
2
Figure 2-13. Comparison of results from the Smith and Howell (2009) Test 1 and Test 3.
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
100%
50% -
1
I
o
0%
10
15
20
25
Deciview
30
35
40
45
I Test 3 » Test 1
33.6 dv, a noticeably larger percentage of the participants in Test 3 rated the image as acceptable
than in Test 1. At VAQ levels worse than 33.6 dv, the majority of the participants found the
VAQ level not acceptable in both tests.
While not as dramatic as the impact in Test 2 (which substantially reduced the VAQ
range), the impact on the Test 3 results of increasing the VAQ range is consistent with Smith and
Howell's conclusion that changing the range of VAQ presented to the participants affects the
responses about whether a particular VAQ is acceptable. The results of Test 3 also are consistent
with the hypothesis that the "dividing line" for the 50% criteria occurs near the middle of the
range of VAQ presented, and that changing the range of VAQ images changes the 50% criteria
"dividing line," with the "dividing line" remaining in roughly the middle of the VAQ range.
The VAQ ranges that Smith and Howell used in Tests 2 and 3 did not span the range of
actual VAQ conditions that occur in Washington, DC, and Smith and Howell provided no
information about the range of actual conditions in Washington in any of their tests. The images
used in the 2001 Washington, DC study (and Test 1) were deliberately selected to present the
range of VAQ conditions in Washington, DC. In the 2001 study, participants were shown an
image of annual average VAQ in Washington at the time, as well as an image of conditions on a
hazy day (the 20th percentile day in the annual distribution). The Denver, Phoenix, and BC
studies also provided participants with information that the range of VAQ conditions they would
be seeing included the actual annual range of VAQ conditions in their city. It is not known
whether the participants in the Smith and Howell Tests 2 and 3 recognized (based on their own
September 2009
2-22
DRAFT - Do Not Quote or Cite
-------
1 knowledge and experience) that the range of VAQ images presented did not represent the actual
2 annual range, or if they believed the range did depict the annual distribution.
3 2.6 SUMMARY OF PREFERENCE STUDIES AND SELECTION OF
4 CANDIDATE PROTECTION LEVELS
5 Each of the studies reviewed in this assessment investigates the common question, "What
6 level of visibility degradation is acceptable?" The approaches used in the four studies are similar
7 and are all derived from the method first developed for the Denver urban visibility study. The
8 specific materials and methods used in each study vary, however, making direct comparison of
9 the study results challenging. Key differences between the studies include:
10
11 * use of WinHaze (a significant technical advance in the method of presenting VAQ
12 conditions),
13
14 * number of participants in each study,
15
16 * representativeness of participants for the general population of the relevant
17 metropolitan area, and
18
19 * specific wording used to frame the questions used in the group interview process.
20
21 Although the differences between the methods used in the urban visibility preference
22 studies are significant, it is possible to examine the results of the studies to identify overall trends
23 in the study findings. Figure 2-14 present a graphical summary of the results of the studies in the
24 four cities. Figure 2-14 draws on results previously presented in Figures 2-3, 2-5, 2-7 and 2-11.
25 For clarity in Figure 2-14, the Denver results omit the 9:00 a.m. photograph results, the
26 Chilliwack and Abbotsford photographs appear as a single set of data for the BC study, and the
27 results from 2001 and 2009 (Test 1) studies of VAQ preferences in Washington, D.C. are
28 presented as a single combined set of data. The results from the 2009 Washington, D.C. study
29 Tests 2 and 3 are not included on Figure 2-14; those tests are not comparable studies because
30 they did not present the actual range of VAQ conditions in the study city.
31 Figure 2-14 also contains lines at 20dv and 30 dv that effectively and pragmatically
32 identifies a range where the 50% acceptance criteria occurs across all four of the urban
33 preference studies. Out of the 114 data points shown in Figure 2-14, only one photograph (or
34 image) with a VAQ below 20 dv was rated as acceptable by less than 50% of the participants
September 2009 2-23 DRAFT - Do Not Quote or Cite
-------
1 who rated that photograph.9 Similarly, only one image with a VAQ above 30 dv was rated
2 acceptable by more than 50% of the participants who viewed it.10
4
5
6
Figure 2-14. Summary of results of urban visibility studies in four North American cities,
showing the identified range of the 50% acceptance criteria.
iting Acceptable
n c
D C
5 5
% Participants Re
c
D C
o x
x o
u /o
I
- ;f**<*^*v ' ; '
o o **»* * * 1
. i
0 * * I
0 * 1
1.
0* • 1
• ' '
* '**
' 1 *
• * *
»* * 1
°o * • * f
o » • 1 o
4» | »»*•*** *
oo o *» <>t%»*» ° o
) 10 15 20 25 30 35 40 4
Deciview
o Denver • Phoenix • BC
• Washington Upper — — Upper
7
8 Figure 2-14 shows that while there is a high degree of similarity between the preferences
9 found in each study, there may be important differences in VAQ preferences in the four cities as
10 well. For example, the Denver study identified preferences for a relatively good level of VAQ;
11 the 50% criteria occur between 17.7 dv and 20.3 dv. In Washington, D.C., however, the 50%
12 criteria separation occurs at a substantially worse level of VAQ, between 27 dv and 31 dv.
13 There are several major hypotheses that may explain why the results of these studies may
14 be indicating potentially important differences between the preferences for VAQ in different
15 cities. As mentioned, the use of photographs versus WinHaze-generated images may play a
16 significant role in preference studies, perhaps introducing bias (such as suggested by the
9 Only 47% of the BC participants rated a 19.2 dv photograph as acceptable.
10 In the 2001 Washington, D.C. study, a 30.9 dv image was used as a repeated slide. The first time it was shown
56% of the participants rated it as acceptable, and 11% rated it as acceptable the second time it was shown. The
same VAQ level was rated as acceptable by 42% of the participants in the 2009 study (Test 1).
September 2009
2-24
DRAFT - Do Not Quote or Cite
-------
1 responses to the 9:00 a.m. Denver photographs) as well as variability. Use of photographs from
2 different days and times of day that rely on associated ambient measurements of light extinction
3 to characterize their VAQ level introduces two types of uncertainty. The intrinsic appearance of
4 the scene can change due to the changing shadow pattern and cloud conditions, and spatial
5 variations in air quality can result in ambient light extinction measurements not being
6 representative of the sight-path-averaged light extinction. WinHaze has neither of these sources
7 of uncertainty because the same base photograph is used (i.e. no intrinsic change in scene
8 appearance) and the modeled haze that is displayed in the photograph is determined based on
9 uniform light extinction throughout the scene.
10 Second, variation in the degree of representativeness of the participants and the sizes of
11 the participant samples involved may also be important factors. The small sample size and fairly
12 uniform population of respondents is a plausible explanation for the noisiness of the combined
13 Washington, D.C. results (35 participants, including 26 from a single consulting firm and 10 of
14 those from a different city) compared with the larger and more representative population of
15 responders from Phoenix (385 participants, carefully selected to be representative of the Phoenix
16 population).
17 A third hypothesis explored by Smith and Howell (2009) is that the range of VAQ
18 images presented in the survey may influence the results. Though this hypothesis appears to be
19 borne out by Smith and Howell's results for Washington, D.C., it seems an unlikely explanation
20 for the differences in results between the four urban preference studies. For example the Denver
21 study included photographs with the haziest conditions among the four studies, but resulted in
22 the lowest haze condition for the 50th percentile preference ratings among the four, not the
23 highest as might be expected if the range of haze levels were a significant factor influencing the
24 results of preference studies.
25 A fourth major hypothesis is that urban visibility preferences may differ by location, and
26 the differences may arise from inherent differences in the city scape scene used in each city. The
27 key evidence to suggest this hypothesis is that the apparent differences between the Denver
28 results (which found the 50% acceptance criteria occurred in the best VAQ levels among the four
29 cities) and the Washington, D.C. results (which found the 50% acceptance criteria occurred at
30 the worst VAQ levels among the four cities). This hypothesis suggests that these results may
31 occur because the cityscape of Denver includes clearly visible snow-covered mountains in the
32 distance, while the prominent features of the Washington, D.C. cityscape are buildings relatively
33 nearby with only modest changes in elevation.
34 Finally, perhaps of significant importance is that the perceived sensitivity of individual
35 scenes to changes in light extinction can be quite different. As in the fourth hypothesis, this may
36 in part explain why the Denver study scene, with its long distance to the mountain backdrop,
September 2009 2-25 DRAFT - Do Not Quote or Cite
-------
1 resulted a preference for the best VAQ level with a 50% criteria value between 17.7 and 20.3 dv,
2 while in Washington, D.C., the 50% criteria separation occurs at a substantially worse level of
3 VAQ, between 27 and 31 dv from Abt Associates Inc. (2001) and Smith and Howell (2009) Test
4 1. The distinction between the last two hypotheses are that the earlier one speaks to the
5 desirability of seeing distant mountains versus this hypothesis where its ability to perceive haze
6 at lower light extinction levels. Additional studies, including directly comparable studies using
7 similar methods in diverse cities, are necessary to gain further understanding of preferences for
8 urban visibility.
9 Based on the composite results and the effective range of 50th percentile acceptability
10 across the four urban preference studies shown in Figure 2-14, CPLs have been selected in a
11 range from 20 dv to 30 dv (74 Mm"1 to 201 Mm"1) for the purpose of comparing to current and
12 projected conditions in the assessment in chapters 3 and 4 of this document. A midpoint of 25
13 dv (122 Mm"1) was also selected for use in the assessment. These three values provide a low,
14 middle, and high set of light extinction conditions that are used in subsequent sections of the
15 UF VA to provisionally define daylight hours with urban haze conditions that have been judged
16 unacceptable by the participants of these preference studies.
17 Though not directly supported by preference or other studies, it is necessary to also
18 identify an averaging time and form to apply along with the CPLs in the assessments described
19 in chapters 3 and 4. For this assessment only daylight hour visibility is being considered. VAQ
20 impacts are instantaneously perceived, suggesting that a short averaging time (e.g. an hour) may
21 be more appropriate than longer time periods (e.g. multiple hours). This is also consistent with
22 the belief that most individuals experience urban VAQ as relatively short-term incidental and
23 intermittent opportunities to be outdoors (e.g. during commutes to work, school, shopping, etc.).
24 Given that some fraction of the public may experience poor VAQ during a relatively small time
25 period and not have the opportunity to see it improve later during the same day, it seems
26 appropriate by EPA staff to consider assessing the current and projected conditions in chapters 3
27 and 4 by comparing the 1-hour daily maximum light extinction to each of the three CPLs
28 supported by the preference studies. Another characteristic that needs to be set for the
29 assessment is the frequency of conditions that should be at or below the CPLs to be considered
30 acceptable. Again, none of the preference studies provided insight into this aspect of
31 acceptability. Because the nature of the public welfare effect is one of aesthetics and/or on
32 feelings of wellbeing and not directly related to a physical health outcome, EPA staff believes
33 that it is not necessary to eliminate all such exposures and that some number of hours/days with
34 poor VAQ can reasonably be tolerated. EPA staff is therefore considering the 90th and 95th
35 percentiles per year averaged over a three year period as a reasonable range of frequencies for
36 meeting the range of PM light extinction CPLs and has incorporated them in this assessment.
September 2009 2-26 DRAFT - Do Not Quote or Cite
-------
1 3 ESTIMATION OF CURRENT PM CONCENTRATIONS AND
2 LIGHT EXTINCTION
3 The goals of the "current conditions" portion of this urban-focused visibility impact
4 assessment are to characterize hourly light extinction conditions in a set of urban study areas in
5 2005-2007, in order (1) to improve understanding of the levels, patterns, and causes of daylight
6 hours light extinction given that essentially no direct measurements are available to inform that
7 understanding, (2) to provide the starting point for projections of light extinction levels under
8 "what if scenarios in each of which it is assumed that each study area complies with a certain
9 secondary NAAQS based either on a measurement-based light extinction indicator or on annual
10 and 24-hour average PM2.5, and (3) to examine the correlation between light extinction and
11 potential alternative indicator(s) based on PM2.5 concentration. This chapter addresses the first
12 goal. Chapter 4 addresses the second goal regarding "what if scenarios. Appendix D addresses
13 the third goal.
14 3.1 GENERAL CHARACTERIZATION
15 3.1.1 PM2.5 and PMio-2.5
16 Chapter 2 of the 2005 Staff Paper from the previous review and chapters 3 (especially
17 section 3.5) and 9 (especially section 9.2.3) and Annex A of the second draft ISA (US EPA,
18 2009a) from the current review present extensive characterizations of the levels, composition,
19 and temporal and spatial patterns of PM2.5 in U.S. urban areas. Both documents present data
20 summaries based on the approximately 1000 PM2.5 monitoring sites in the U.S. The
21 characterizations in the 2005 Staff Paper were based on 2001-2003 data. The characterizations
22 in the ISA are based on 2005-2007 data, which is the same time period used in this visibility
23 assessment. While there generally have been reductions in the concentrations of PM2.5 in many
24 areas as a result of emission reductions of PM2.5 and its precursors, the general patterns, and the
25 diversity of patterns across areas, noted in the 2005 Staff Paper still prevailed in the 2005-2007
26 period.
27 In 2005-2007, 38 urban areas violated the annual PM2.5 NAAQS of 15 |ig/m3, adopted in
28 1997 and retained in the last review completed in 2006. Seventy-six areas violated the revised
29 24-hour NAAQS of 35 |ig/m3. There is considerable but not complete overlap in the areas not
30 meeting the two NAAQS. It should be noted that in many parts of the U.S., PM2.5 concentrations
31 in 2005 were high relative to the next three years. Figure 3-1 illustrates PM2 5 air quality in 2007
32 by representing each monitor by a symbol whose color reflects the annual mean of the
33 concentration at that site or the 98th percentile 24-hour concentration, in both cases in that one
34 year.
September 2009 3-1 DRAFT - Do Not Quote or Cite
-------
th
1 Figure 3-1. Annual average and 24-hour (98 percentile 24-hour concentrations) PM2.s
2
3
concentrations in ug/m , 2007.
4
5
Annual
Concentration Range (|jg/m3}
• 3.4-12.0 (418 Sites)
O 12.1 -15.0 (366 Sites)
O 15.1 -18.0 (86 Sites)
• 18,1 -22.5 (14 Sites)
Puerto Rico
24-hour
Concentration Range (|jg/m3)
• 7-15 (38 Sites)
O 16 - 35 (662 Sites)
O 36-55 (166 Sites)
• 56-73 (18 Sites)
Puerto Rico
6
September 2009
3-2
DRAFT - Do Not Quote or Cite
-------
1 Each urban area exhibits its own detailed patterns of observed concentration levels,
2 temporal and spatial variation, and composition. These differences are due to differences in local
3 and transported emissions and in meteorology. Because of differences in the placement of PM2 5
4 monitoring sites in each urban area, the actual levels and spatial pattern of PM2.5 and PM2.5
5 species concentrations may not be consistently discernable in all areas. This variability and
6 limited monitoring network make it difficult to offer concise generalizations, although some
7 broad similarities can be drawn among areas.
8 Midwestern, southeastern, and eastern urban areas have much higher sulfate levels than
9 do more western areas, attributable to the much higher emissions of SO2 in and upwind of them.
10 Upper midwestern areas and to a lesser extent upper eastern areas have notable nitrate
11 concentrations in winter but not in summer, while southeastern areas generally lack notable
12 nitrate even in winter. Many western urban areas have notable nitrate year round. In all areas,
13 carbonaceous material is an important component of PM2 5 and is attributable to many emission
14 sources of organic material in PM form and of organic PM precursor gases; in some areas with
15 high local use of wood for residential heating carbonaceous material is dominant during the
16 heating season. PM2 5 derived from crustal sources is generally a small faction of total mass,
17 except during local high wind events or due to brief periods of intercontinental transport of dust
18 from Africa or Asia.
19 Comparison of PM2 5 species concentrations within and outside urban areas leads to the
20 conclusion that, in the eastern areas with high sulfate concentrations, the large majority of the
21 sulfate affecting any given urban area originates outside that area. Inward transport and local
22 generation of nitrate and carbonaceous material are more evenly balanced in eastern areas, with
23 some differences among areas. In western areas, local sources dominate for carbonaceous
24 material and nitrate, with the origins of the small sulfate component being more balanced. See
25 Figure 9-24 of the second draft ISA (US EPA, 2009a).
26 Southeastern areas have their highest PM2.5 concentrations in the summer, when
27 conditions are most conducive to sulfate formation. More northern areas, being affected by a
28 more balanced mix of contributors, tend not to have such a strongly seasonal pattern. The
29 seasonal patterns in western areas are individual and varied, related to differences in local
30 sources and formation and dispersion conditions. In all areas, inversion conditions with low
31 wind speeds are conducive to high concentrations due to the trapping of emissions from local
32 sources. Some western areas, especially those with valley or bowl-like topography, are
33 especially affected.
34 There is at present no systematic monitoring network in place for PMi0.2.5, as states have
35 until January 1, 2011, to implement required monitoring sites for PMio-2.5. Consequently,
36 estimates of PMio-2.s must be developed using data from PM2.5 and PMio monitoring sites and
September 2009 3-3 DRAFT - Do Not Quote or Cite
-------
1 equipment, which are not always collocated and consistent. The 2005 Staff Paper presented such
2 estimates in section 2.4.3. The second draft ISA presents such estimates in Figure 3-10 and
3 Table 3-9 of section 3.5.1.1. The 2005 Staff Paper used a data-inclusive approach in which the
4 best available data on PM2.5 and PMio concentrations - in some cases not very robust data -
5 were used to estimate 2001-2003 PMio-2.5 concentrations for 351 metropolitan area counties. For
6 these counties, the annual mean PMio-2.5 concentrations were generally estimated to be below 40
7 ug/m3, with one maximum value as high as 64 ug/m3 and a median of about 10-11 ug/m3. The
8 second draft ISA used a much more data-restrictive approach based only on paired (collocated)
9 low-volume filter-based samplers for both PMio and PM2.5, the most accurate method of
10 measuring PMio-2.5. The second draft ISA reports that only 40 counties have such paired
11 samplers. Using these available co-located PM measurements from 2005-2007, the mean 24-hr
12 PMio-2.5 concentration in these 40 counties was 13 ug/m3. This urban visibility assessment has
13 used a data-inclusive approach to estimating PMio-2.5 concentrations, similar to that used for the
14 2005 Staff Paper, where needed to obtain hourly PMio-2.5 estimates for 15 study areas, which are
15 reported below in section 3.3.2.
16 Additional detail on PM2.5, PMio, and PMio-2.5 concentrations, composition, and patterns
17 appears in section 3.5.1.1 of the second draft ISA. Also, chapter 6 of the 2004 PM Assessment
18 by NARSTO contains more detailed characterizations of PM in different parts of the U.S.
19 3.1.2 Light extinction
20 While light extinction is directly measurable, there are very few regularly operating
21 monitoring sites measuring light extinction in urban areas, and generally those that do operate do
22 not submit data to AQS.u Consequently, any characterization of light extinction conditions
23 based on actual measurements is necessarily less comprehensive than for PM2.s and PMio-2.5.
24 Many monitoring sites that employ nephelometers, which do measure light scattering, operate
25 that equipment in a heated mode for purposes of tracking "dry" PM2.5 mass concentrations, and
26 actual ambient light extinction is not reportable. There are many more filter-based
27 Aethalometers® and similar instruments for measuring light absorption in operation and
28 reporting to AQS, but light absorption is typically a small fraction of total light extinction, so
29 these data alone are not a good indicator of light extinction in urban areas. Also, there are
30 unresolved issues of data corrections and comparability for the light absorption data from these
31 instruments now residing in AQS.
11 There is a large network of "visual range" monitors in operation at U.S. airports, aimed at providing information
to determine landing and takeoff safety. Due to their locations and to the lack of data resolution (values of visual
range above the level needed for unlimited airport operations are not individually reported) the data from these
monitors are not suitable for use in this assessment. The second draft ISA discusses these monitors in section
9.2.2.3.
September 2009 3-4 DRAFT - Do Not Quote or Cite
-------
1 Light extinction can be "reconstructed" from measurements of PM2.5 mass components
2 and PMio-2.s concentrations, along with relative humidity, using the formula known as the
3 IMPROVE algorithm. (Section 9.2.2.2 of the second draft ISA gives an overview of the
4 algorithm and its basis.) PM2.5 component measurements are generally available only on a 24-
5 hour average basis, so it generally is possible to estimate only 24-hour average light extinction,
6 unless additional information on hourly patterns is brought to bear.12 Because EPA's Regional
7 Haze Rule (RHR) currently requires states to address visibility problems in Class I visibility
8 protection areas, which are nearly all rural and remote, there is a large body of literature
9 characterizing light extinction in remote rural areas, based on data from the IMPROVE
10 network's 24-hour samplers and on special studies. Sections 9.2.3.2 and 9.2.3.4 of the second
11 draft ISA summarizes this literature. Section 9.2.3.3 of the ISA contrasts concentrations of PM
12 and PM components between rural and urban areas using data from the rural IMPROVE network
13 and the urban Chemical Speciation Network (CSN), but does not present estimates of light
14 extinction in urban areas.
15 The CSN network provides 24-hour PM2 5 species measurements at about 200 urban
16 sites, from which mass components can be derived. These sites have a mix of daily, one day in
17 three, and one day in six sampling schedules. The 2005 Staff Paper (and its references) may be
18 the only readily available prior assessment to use these urban PM2.5 speciation monitoring data,
19 along with estimates of PMio-2.s concentrations and data on relative humidity, to reconstruct daily
20 24-hour average light extinction in urban areas, for the year 2003. One presentation of the
21 results was in the form of a scatter plot of daily 24-hour reconstructed light extinction versus 24-
22 hour PM2 5 concentration. This graphic appears here as Figure 3-2. (For the immediate purpose
23 of this section, it is the distribution of the data points along the y-axis that is of interest, not the
24 relationship between light extinction and PM2.5 concentrations; the latter subject is addressed in
25 Appendix D.) Generally, most days have light extinction below 200 inverse megameters (Mm"1),
26 but a small percentage of values were as high as about 750 Mm"1.13
27 In addition to this scatter plot, a table developed for the previous PM NAAQS review
28 presented the annual average of estimates of 24-hour reconstructed light extinction values,
29 averaged across 161 urban areas grouped into seven regions (Schmidt, et al., 2005). Table 3-1
30 reproduces these estimates. For regions excluding Southern California, annual average 24-hour
31 light extinction
12 When the IMPROVE algorithm is used to estimate 24-hour light extinction from 24-hour PM2 5 species and
PM10-2.5 concentrations, an assumption is made that every hour has the same PM concentrations but its own relative
humidity value. Hourly estimates of light extinction, including the strongly non-linear effect of relative humidity,
are then averaged to get the 24-hour light extinction estimate.
13 Unfortunately, the file of paired data used to create this scatter plot is no longer available, so the actual
distribution of light extinction values cannot be described more specifically.
September 2009 3-5 DRAFT - Do Not Quote or Cite
-------
1
2
Figure 3-2. Reconstructed 24-hour light extinction in U.S. urban areas in 2003
Source: Schmidt et al., 2005
Output D.3
(Relationship RE & PM2.5; Diurnal RE; Timeframe)
2 of 30
Hemn,
W
70C
COC
50C
40C
IOC
Significant relationship (low p-^slue)
0 0
o o° o
•°#0 ,• V* ,'
°* ,/ •.
0°
0°
East (circles): RE=y-8.1 *PM:,R2=
'
1C
i
,10
, ,
4* 50
ISO
'
70 W
3
4
5
6
7
Relationship benveen recon^rructed light eitinction (RE) and ^4-hour avenge PM,:, 2005. Using atrual/i'Mfl
ranged from 73 to 118 Mm"1. The estimate of the annual average 24-hour light extinction for
Southern California was 168 Mm"1. These estimates were based on 10-year average 1-hour
relative humidity values and 2003 PM monitoring data.
September 2009
3-6
DRAFT - Do Not Quote or Cite
-------
1
2
4
5
6
7
9
10
11
12
13
14
Table 3-1. Annual Mean Reconstructed 24-hour Light Extinction Estimates
by Region (Mm"1)
Region
Northeast
Southeast
Industrial Midwest
Upper Midwest
Southwest
Northwest
Southern California
Reconstructed 24-hour Light
Extinction in 2003
108
98
118
80
73
76
168
Source: Output D.3, Schmidt et al, 2005. We note these regions were used to summarize PM2 5 patterns for the PM
NAAQS review 1997 (US EPA, 1996).
Figure 3-3 is a contour map of annual average reconstructed 24-hour light extinction
based on IMPROVE monitoring sites in 2000-2004, nearly all of which are remote and rural (the
three urban sites in Phoenix, AZ, Washington, D.C., and Puget Sound, WA are indicated by
square symbols). Comparing the mean urban light extinction levels by region listed in Table 3-1,
estimated based on CSN data, with this map of rural light extinction based on IMPROVE data
indicates that remote rural light extinction levels are notably lower than in urban areas in most
parts of the U.S., with the northeast and the southeast regions having the most similarity between
rural and urban light extinction levels. This is consistent with observations of an "urban excess"
September 2009
3-7
DRAFT - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
Figure 3-3. Isopleth map of annual total reconstructed particulate extinction based on
IMPROVE data.
. . •
' •. j • ..
t/.x
jr
Hawaii
I
J V
• IMPROVE Site
• IMPROVE Urban Site
^M o
Puerto Rico /
Virgin Islands
(Source: Spatial and Seasonal Patterns and Temporal Variability of Haze and its Constituents in the
United States Report IV, November 2006.)
of PM2.5 and PMio-2.5 and with the known high regional concentrations of sulfate in these eastern
areas.
One-hour light extinction values of course vary above and below the 24-hour average,
due to diurnal variations in PM2.5 component concentrations, PMio-2.5 concentrations, and
relative humidity. Because low wind speeds, inversion conditions, and lower temperatures are
more prevalent in the night and early morning hours, light extinction generally is higher at those
times, with morning daylight hours being when poor visibility will most often be most
observable. Although light extinction was formally reconstructed on an hourly basis in the 2005
Staff Paper analysis for the last review cited above, the actual full strength of the diurnal pattern
could not be discerned in that analysis because component mix was assumed not to vary from
hour to hour. Under the unverified assumption of constant component mix and using actual
hourly relative humidity data, the daily maximum daylight 1-hour light extinction values were
September 2009
3-8
DRAFT - Do Not Quote or Cite
-------
1 roughly 50 percent higher than the 24-hour average light extinction values.14 The new analysis
2 presented in this document includes a closer look at diurnal patterns, for 15 study areas.
3 3.2 OVERVIEW OF APPROACH AND DATA SOURCES FOR URBAN STUDY
4 ANALYSIS
5 As explained above, there are limited data from direct measurements of light extinction in
6 urban areas. Consequently, this assessment has reconstructed hourly light extinction levels from
7 values of hourly PM2.5 components, PMio-2.5, and relative humidity. Hourly monitoring data for
8 these parameters are also lacking, so the estimates of these parameters necessarily in turn have
9 been developed from a combination of other available ambient monitoring data and air quality
10 modeling results from a chemical transport model (CTM) run. Specifically, the ambient
11 monitoring data starting points are 24-hour PM2.5 mass measured by filter-based Federal
12 Reference Method (FRM) or Federal Equivalent Method (FEM) monitors15, 24-hour PM2.5
13 components measured by the filter-based monitors of the Chemical Speciation Network, and
14 hourly PM2.5 mass measured by continuous instruments (Tapered Element Oscillating
15 Microbalance (TEOM), beta attenuation monitors (BAMs), and nephelometers were used at
16 different sites). The CTS-based diurnal profiles for individual components, in conjunction with
17 hourly PM2 5 measurements, are used to adjust and allocate the 24-hour PM2.5 components
18 measurements to individual hours of each day, as described in detail below. In addition, levels
19 of hourly PMio-2.s mass are calculated from separate measurements of hourly PMio and hourly
20 PM2.s if both are available and by applying PMio-2.s to PM2.s ratios to hourly PM2 5 data if both
21 types of hourly measurements are not available. The ambient data are from 2005-2007 and were
22 all obtained from AQS in the first half of 2009.
23 The CTM run was the "actual emissions" run of the 2004 CMAQ modeling platform with
24 boundary conditions provided by GEOS-Chem global scale CTM.16 The primary use of the CTM
25 modeling is to provide realistic diurnal variations for each of the major PM2.5 components used
26 to estimate light extinction, anchored to site-specific, day-specific measurements of 24-hour
27 concentrations. That is, monthly averaged diurnal profiles for the five major components were
28 generated using the CTM results which were then used to generate realistic hourly concentration
29 variations for each of the 24-hour CSN sample days during the 2005-2007 period.
14 These observations on diurnal patterns come from examination of "Output D.3 (Relationship RE & PM25; Diurnal
RE; Timeframe) 8 of 30" and "Output D.3 (Relationship RE & PM2 5; Diurnal RE; Timeframe) 17 of 30", Analyses
of Paniculate Matter (PM) Data for the PM NAAQS Review, Schmidt et al., 2005.
15 Filter-based Federal Reference Method samplers and filter-based Federal Equivalent Method samplers will both
be referred to as FRM samplers in the remainder of this document.
16 GEOS-Chem is the NASA Goddard Earth Observing System-CHEMistry (global 3-D CTM for atmospheric
composition). This modeling platform, with an appropriately different emissions scenario, is also the basis for the
estimates of policy relevant background concentrations of PM25 presented in section 3.6 of the second draft ISA (US
EPA, 2009a).
September 2009 3-9 DRAFT - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
3.2.1 Study Period, Study Areas, Monitoring Sites, and Sources of Ambient PM
Data
As of the time this assessment began, the ambient monitoring data from 2005-2007, but
not from 2008, had been certified as accurate and complete by the state/local monitoring
agencies that collected them, and the data had been extensively summarized and presented in the
first draft ISA. The EPA staff aimed to develop estimates of daylight hours light extinction for a
reasonably representative number of days in each year of 2005-2007, to allow the application of
statistical forms based on three years of data. However, as explained in more detail below, in
several study areas the limited availability of starting data for these estimates resulted in estimate
sets that do not cover all three years. Also, even in areas with some data in all three years, the
number of days with valid estimates differs by year and is in some cases not large by typical
standards of monitoring data completeness.
For efficiency in the analysis, this visibility assessment uses the same 15 urban study
areas selected for the health risk assessment. These areas are listed in Table 3-2, along with the
area-wide FRM-based 2005-2007 annual and 24-hour PM2.5 design values for each study area.
(See below for an explanation of the "site-specific" columns in Table 3-2.)
Table 3-2. Urban Visibility Assessment Study Areas
Study Area
Tacoma
Fresno
Los Angeles-South
Coast Air Basin
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit- Ann Arbor
Pittsburgh
Baltimore
Philadelphia-
Wilmington
New York-N. New
Jersey-Long Island
Area-wide
2005-2007
Annual
Design Value
10.2
17.4
19.6
12.6
11.6
12.8
15.8
16.5
18.7
16.2
17.2
16.5
15.6
15.0
15.9
Area-wide
2005-2007
24-hour
Design Value
43
63
55
32
55
26
31
39
44
35
43
43
37
38
42
Site-
specific
2005-2007
Annual
Design Value
Same
Same
Same
7.9
10.7
11.5
13.1
14.5
Same
15.7
Same
15.0
14.5
14.7
14.4
Site-
specific
2005-2007
24-hour
Design Value
Same
Same
Same
15
48
25
25
34
Same
33
Same
40
35
37
42
Region
Northwest
Southern
California
Southern
California
Southwest
Northwest
Southeast
Southeast
Midwest
Southeast
Southeast
Midwest
Industrial
Midwest
Northeast
Northeast
Northeast
19
September 2009
3-10
DRAFT - Do Not Quote or Cite
-------
1 For time reasons and because it was anticipated that some study areas would not contain
2 more than one suitable study site, EPA staff sought to identify the single best study site in each
3 area. In identifying the single best study site in each study area consideration was given to the
4 availability of collocated 24-hour data on PM2.5 and its components, because the contribution of
5 PM2.5 components to total light extinction will typically dominate the contribution from PMio-2.5.
6 Ideally, within each study area the three types of PM2.5 data (FRM PM2.5, CSN PM2.5
7 components, continuous PM2 5) would be available at a common site, and that site would be
8 located in a manner consistent with reliance on it to characterize visibility as it would be
9 perceived by a large number of area residents and visitors. Shown in Table 3-2 for convenient
10 comparison are the site-specific FRM-based design values for the monitoring site in each study
11 area from which FRM PM2 5 data were taken for the purposes of this assessment, where not the
12 same as the site providing the area-wide design values.17 As can be seen in Table 3-2, in most of
13 the study areas the site providing FRM data for this assessment is not the area-wide design value
14 site, because the area-wide design value site did not have collocated CSN and/or continuous
15 PM2.5 data.
16 Appendix A provides details on the site(s) identified and used in each study area,
17 including information on the type of monitoring that provided the data and other information that
18 may help interpret the results of the analysis. A portion of this table for a single site - Tacoma -
19 is presented here as Table 3-3 as an example. When viewing this document electronically, the
20 site IDs in these tables are active links and can be used to view the location of the site via
21 GoogleMaps.18
17 2005-2007 PM2 5 design values were taken from the information posted at
http://www.epa.gov/airtrends/values.html. and are consistent with the design values used in the health risk
assessment to "roll back" current concentrations to represent achievement of alternative annual and 24-hour PM2 5
NAAQS. Except in Dallas and Fresno, the area-wide design values are the highest design values of any monitoring
site in the designated (1997 NAAQS) nonattainment area that has sufficiently complete data to allow the calculation
of a design value according to the provisions of 40 CFR 50 appendix N. For Dallas, the design values come from a
site with nearly complete data, and are somewhat higher than the highest values from a site with complete data (see
the draft PM Risk Assessment, US EPA, 2009c, section 3.2.3) For Fresno, the area-wide design value is for the
Fresno-Madera CSA, which is only a portion of the San Joaquin Valley nonattainment area. Also, note that there
are three cases in which the nonattainment area does not include certain areas sometimes thought of as being part of
the area named in Table 2; monitors in these non-included areas were not considered in this assessment. (1) The
design value shown for Pittsburgh is for the Pittsburgh-Beaver nonattainment area; the Liberty-Clairton
nonattainment area is within the Pittsburgh CBSA but is distinct for regulatory purposes, and was not considered in
this assessment. (2) Baltimore was treated separately, although part of a CSA with Washington DC. (3) Berks Co.,
PA is part of the Philadelphia-Camden-Vineland CSA, but not part of the Philadelphia-Wilmington nonattainment
area.
18 Additional meta data on each monitoring site, and access to daily and annual data listings, can be conveniently
obtained using GoogleEarth and the PM2 5, PM10, and CSN monitoring network KML files that can be downloaded
from http://www.epa.gov/airexplorer/monitor kml.htm.
September 2009 3-11 DRAFT - Do Not Quote or Cite
-------
1 In 11 of the study areas, the three types of PM2.5 data were available at a common site. In
2 the remaining four areas, Phoenix, AZ, Pittsburgh, PA, Baltimore, MD, and St. Louis, MO-IL,
3 two types of data were available at one site, but the remaining type of data had to be taken from
4 another site and treated as being representative of the former site.
5 The monitoring agencies described all but one of these sites as neighborhood or urban
6 scale, indicating those agencies' opinion that the sites represent concentrations in an area at least
7 0.5 to 4 km across. An aerial view of the remaining site (in Phoenix) suggests that it may be
8 middle or neighborhood scale. Selected sites are not necessarily the locations of the maximum
9 measured annual or 24-hour PM2.5 levels in their urban area.
10 Hourly PMio-2.5 presented more varied challenges. In four areas (Birmingham, Detroit,
11 Baltimore, and Philadelphia) the site that provides the continuous PM2 5 data also hosts a
12 continuous FEM PMio monitor, and hourly PMio-2.5 could be calculated by difference for most
13 hours. In other areas, this was not the case, and either (1) instruments at two different sites were
14 used in this subtraction (Tacoma, Los Angeles-South Coast Air Basin, Phoenix, St. Louis,
15 Atlanta, and New York-N. New Jersey) or (2) a single regionally applicable PMio-2.5 to PM2.5
16 ratio calculated as part of the last review based on 2001-2003 24-hour FRM/FEM PMio and
17 PM2.5 samples was applied to 2005-2007 hourly PM2.5 data to estimate hourly PMi0-2.5 (Fresno,
18 Salt Lake City, Dallas, Houston, and Pittsburgh). In the case of Los Angeles-South Coast Air
19 Basin, the continuous PMio and PM25 sites were quite distant and separated by a range of hills,
20 so the estimates of PMi0.2.5 and its contribution to total light extinction are more uncertain than if
21 the monitors were clearly within the same air mass. Obviously, for those study areas for which
22 1-hour PMio-2.5 was estimated by application of ratios, PMi0.2.5 estimates can only represent
23 broad trends, not hour-specific conditions at the particular site. More description of the methods
24 used for estimating hourly PMio-2.s appears in section 3.3.2.
25 The sampling schedule for CSN PM2 5 speciation monitoring was one-in-six days for
26 Tacoma, Phoenix, Houston, Detroit, and Philadelphia, and one-in-three days for the other study
27 areas. Not every scheduled CSN site day in 2005-2007 had data for all three types of PM2.5 data,
28 due to missed or invalid samples. Also, for continuous PM2.5, values for a small number of hours
29 of an otherwise data-sufficient day were sometimes missing, due to equipment failure or
30 servicing. EPA staff retained only those days in which 75 percent or more of daylight hours had
31 measurements of PM2.5 (see section 3.3. for more details). If for isolated hours at a site (or site
32 pair) with collocated measurements, PMio-2.s concentrations could not be estimated because of
33 gaps in the same-hour continuous PMio and/or PM2.5 data, EPA staff used the regional ratio
34 approach described above to estimate PMi0.2.5 for those specific hours. Table 3-4 provides more
35 detailed information on the quarterly distribution of the successfully matched and sufficiently
36 complete data available for use. As described later, for some parts of this assessment EPA staff
September 2009 3-12 DRAFT - Do Not Quote or Cite
-------
1
2
Table 3-3. PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Tacoma Study Area
Study Area
First PM2.s Monitoring Site
Second PM2.s Monitoring Site
(if applicable)
data source for PMi0-2.s
Tacoma
AQS ID 530530029
State: Washington
City: Tacoma
MSA: Tacoma, WA
Local Site Name: TACOMA - L STREET
Address: 7802 SOUTH L STREET,
TACOMA
0.5 miles east of 1-5
2005-2007 annual DV = 10.2
2005-2007 24-hr DV = 43
This is the highest 24-hour PM2.5 DV site
in the Seattle-Tacoma-Olympia, WA annual
PM2.5 nonattainment area
Neighborhood Scale
Parameters taken from this site:
* 24-hour FRM PM2.5 mass (AQS
parameter 88101; one-in-three
sampling schedule)
* PM2 5 speciation (one-in-six sampling
schedule)
* l-hourPM25mass (AQS parameter
88502, Acceptable PM2.5 AQI &
Speciation Mass) Correlated Radiance
Research M903 Nephelometry
No continuous PM10 monitoring at this site,
see right hand column.
N/A
AQS ID 530530031
State: Washington
City: Tacoma
MSA: Tacoma, WA
Local Site Name: TACOMA - ALEXANDER AVE
Address: 2301 ALEXANDER AVE, TACOMA, WA
6.4 miles NNE of PM2 5 site
Neighborhood Scale
Parameters taken from this site:
* 1-hour PM10 STP mass (AQS parameter 81102)
* Sample Collection Method: INSTRUMENTAL-R&P
SA246B-INLET
* Sample Analysis Method: TEOM-GRAVIMETRIC
7% of PM10.2 5 values were determined using regional average
PM10.2.5:PM2 5 ratios from 2005 Staff Paper
Additional Explanation
• In this Table, the 1-hour concentration parameter "88502, Acceptable PMisAQI & Speciation Mass" is the same as the IS A refers to as "FRM-like" PM2.5 mass. An entry of "88501, PM2.5
Raw Data" indicates that the monitoring agency makes no representation as to the degree of correlation with FRM PM2.5 mass. The latter type of continuous PM2.5 data were used only when
the former were unavailable.
• Where PM10 was reported in STP, it was converted to LC before PM10.2.5 was calculated.
• For convenient, continuous PM2.5 data was obtained through the AirNow website rather than from AQS, as an initial exploration indicated that not all the desired 1-hour data had been
submitted to AQS.
September 2009
3-13
DRAFT - Do Not Quote or Cite
-------
1
2
Table 3-4. Number of days per quarter in each study area
Study Area
Tacoma
Fresno
Los Angeles-
South Coast
Air Basin
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann
Arbor
Pittsburgh
Baltimore
Philadelphia-
Wilmington
New York-
N.New Jersey-
Long Island
Total
Number
of Days
110
324
302
86
306
274
149
294
350
295
141
284
187
145
228
2005
Qi
0
19
28
0
27
22
21
27
30
22
12
26
19
15
22
Q2
0
24
28
13
28
24
20
27
30
25
12
23
17
11
23
Q3
0
27
22
11
30
26
10
24
29
25
10
25
15
13
13
Q4
0
27
28
14
26
22
14
27
30
24
11
23
11
10
15
2006
Qi
13
30
26
12
20
23
14
28
29
28
12
22
15
9
23
Q2
15
29
26
13
28
23
12
19
29
27
13
25
16
13
19
Q3
15
29
27
11
31
24
8
27
30
26
11
24
19
10
18
Q4
14
27
22
12
20
24
12
29
30
27
15
26
18
13
21
2007
Qi
13
26
21
0
23
18
15
29
30
25
11
22
12
13
19
Q2
13
28
26
0
25
23
14
25
30
19
11
22
12
14
15
Q3
14
30
24
0
19
24
9
22
27
26
12
23
17
12
19
Q4
13
28
24
0
29
21
0
10
26
21
11
23
16
12
21
3 Note: Only days with matched and sufficiently complete data were retained in the assessment.
September 2009
3-14
DRAFT - Do Not Quote or Cite
-------
1 substituted data for the single missing quarters of data in Phoenix and Houston, to achieve
2 seasonal balance.
3 3.2.2 Use of CMAQ Model Runs for 2004 to Augment Ambient Data
4 Because systematic monitoring data are not available on hourly PM2 5 component
5 concentrations, EPA staff extracted and applied certain information from the modeling platform
6 for calendar year 2004 described in section 3.7.1.2 of the second draft ISA, in which the global-
7 scale circulation model GEOS-Chem was paired with the regional scale air quality model
8 CMAQ.19 The main use of this platform in the ISA is to estimate policy-relevant background
9 concentrations of PM2.5. For the urban-focused visibility assessment described here, however, we
10 used results from the validation run of the platform, in which emissions for all emission source
11 types and countries are included, to develop realistic diurnal variations of the major PM2.5
12 components.
13 The EPA staff identified the 36 km-by-36 km CMAQ grid cell corresponding to the
14 location of the CSN monitoring site used in each study area. We then extracted from the detailed
15 model output for this grid cell the day/hour-specific concentrations of sulfate, nitrate, elemental
16 carbon, organic carbon, and "crustal/unspeciated" PM2.s during 2004, and then we averaged
17 across days within the month for each individual hour of the day.20 Thus, for each species, EPA
18 staff obtained 24 values for a month, for each of the 12 calendar months. We then averaged the
19 24 hourly values in each monthly set for each component to obtain the 24-hour average
20 concentration for the month. We then divided each hourly value by the 24-hour value, to obtain
21 a normalized diurnal profile for the pollutant, which was taken to represent all days in that month
22 for 2005, 2006, and 2007. In total, this resulted in 5 (components) x 12 (months) x 15 (study
23 areas) = 900 profiles. Visual examination of a number of these showed them to be reasonably
24 smooth and generally to show morning (and sometimes also late afternoon/evening) peaks which
25 are the anticipated effect of higher vehicle traffic and lower mixing heights. The peaks were
26 generally moderate, as would be expected in light of the large grid cells and generally moderate
27 diurnal profiles for SMOKE pre-processing of emissions in the CMAQ modeling platform.
28 Sulfate, as would be expected for a regionally transported pollutant, generally had a flatter
29 diurnal profile than for other components. Hourly nitrate concentrations were low when
30 expected: during warmer months and in warmer areas. Figure 3-4 shows example diurnal profiles
19 Similar modeling was not available for 2005, 2006, or 2007.
20 For several of the listed components that are not direct CMAQ outputs, concentrations were estimated by post-
processing to aggregate the appropriate CMAQ outputs. The "crustal/unspeciated" CMAQ output results from non-
reactive dispersion of that portion of the PM25 emission inputs not assigned during SMOKE processing to a more
specific CMAQ species, and is considered in most EPA analyses to represent the same material as the "soil"
component reported for IMPROVE sampling.
September 2009 3-15 DRAFT - Do Not Quote or Cite
-------
1 for the five PM2.5 components, for the Detroit study area for the months of January and August.
2 Diurnal profiles like these were applied to 24-hour CSN measurements of component
3 concentrations, as explained in detail below.
4 3.2.3 Use of Original IMPROVE Algorithm to Estimate Light Extinction
5 The EPA staff used the original IMPROVE light extinction algorithm, rather than the
6 more recent revised version, because the original version is considered more representative of
7 urban situations, when emissions are still fresh rather than aged as at remote IMPROVE sites.
8 To maintain consistency with the form of the candidate protective levels (CPLs) identified in
9 chapter 2, we staff included a value of 10 Mm"1 to represent clear air Rayleigh scattering, and we
10 use the term "total light extinction" to indicate this.21 No presumption is intended regarding
11 whether a possible secondary NAAQS using measured light extinction as the indicator should
12 include or exclude light extinction due to gases. The formula for total light extinction using the
13 traditional IMPROVE algorithm is shown below.
14
+ 3 x f(RH)
+ 4 x [Organic Mass\
+ 10 x [Elemental Carbon]
+ 1 x [Fine Soil]
— 0.6 x [Coarse MxssJ
-rlO
15
16 Total light extinction (bext) is in units of Mm"1, the mass concentrations of the
17 components indicated in brackets are in ug/m3, and f(RH) is the unitless water growth term that
18 depends on relative humidity. We refer to the first five terms in this algorithm as the five PM2.5
19 components. In this algorithm, the sulfate and nitrate components are to be expressed as fully
20 neutralized, without associated water since the water absorption effect is reflected in the f(RH)
21 term. The organic mass component is to include the mass of associated elements other than
22 carbon. As described below, we included steps in our development of estimates of hourly
23 component concentration to ensure consistency with these aspects of the IMPROVE algorithm.
21 We did not include a term for light absorption by NO2 or other gases.
September 2009 3-16 DRAFT - Do Not Quote or Cite
-------
1 Figure 3-4. January and August monthly average diurnal profiles of PM2.s components
2 derived from the 2004 CMAQ modeling platform, for the Detroit study area.
Normalized Diurnal Profile for PM2.5 Components
(January, Detroit)
o
o
o
o
(N
O
O
O
O
(N
O
O
O
O
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
rsi
o
o
o
o
rsi
o
o
o
o
rsi
o
o
o
o
rsi
Normalized Diurnal Profile for PM2.5 Components
(August, Detroit)
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
September 2009
3-17
DRAFT - Do Not Quote or Cite
-------
1 3.3 DETAILED STEPS
2 3.3.1 Hourly PM2.s Component Concentrations
3 The task of estimating hourly PM2 5 component concentrations is in a sense over-
4 determined, given the four types of available information: 24-hour PM2.5 mass by filter-based
5 FRM, 24-hour component concentrations by CSN, hourly PM2.5 mass by continuous instrument,
6 and diurnal profiles of components from the 2004 CMAQ run. There are multiple ways in which
7 two or three of these four data sources could be used to estimate hourly PM2.5 component
8 concentrations, and the result generally can be expected to be at least somewhat inconsistent with
9 the information in the remaining data sources. For example, each 24-hour PM2.5 component
10 mass from CSN sampling can be apportioned to hours based on the diurnal profile developed
11 from the 2004 CMAQ run, but then in general the hourly values of PM2 5 mass determined by
12 summing the components in an hour would not exactly match the data from the continuous PM2.5
13 instrument. EPA staff therefore used a sequence of steps which achieves a prioritized
14 compromise among the data sources. In this sequence, we have given greater weight to the 24-
15 hour FRM, CSN, and continuous PM2.5 mass data because these are instrument-based and
16 location- and day-specific, than to the CMAQ-based profiles which are CTM-based, averaged to
17 the month, and extrapolated from 2004 to each of 2005, 2006, and 2007.
18 Because of differences in filter materials, laboratory analysis, and data reporting, there
19 are differences between the contribution of some PM components to PM2 5 mass as, reported by
20 filter-based 24-hour FRM sampler or by continuous instruments, and the mass of the same
21 component as reported by CSN (or IMPROVE) sampling. The following summary of these
22 differences may be helpful in understanding the steps used to develop estimates of hourly PM25
23 components in this analysis. In the IMPROVE algorithm for reconstructing total light extinction,
24 the light extinction contribution multipliers per unit of mass concentration of components are
25 different for each of the five principle components. Consequently, care is required to estimate
26 these components as consistently as possible with the IMPROVE sampling and analytical
27 methods.
28 • Nitrate: CSN (and IMPROVE) sampling uses a Nylon filter for purposes of nitrate
29 quantification, while FRM sampling uses a Teflon filter for PM2 5 as a whole. The
30 Nylon filter limits the loss of nitrate in the form of nitric acid vapor, compared to the
31 Teflon filter. Hence, the nitrate mass reported by CSN (and IMPROVE) sampling
32 typically will be higher than the nitrate contribution to FRM PM2.5 mass, particularly
33 under warm ambient conditions. In addition, the IMPROVE program does not measure
34 ammonium ion and hence must make an assumption that nitrate ion is fully neutralized
35 by ammonium ion. In contrast, in FRM sampling ammonium ion is measured and it is
36 possible for nitrate to be found not to be fully neutralized. These two factors tend to
37 make nitrate mass as reported for a CSN or IMPROVE site higher than the nitrate
September 2009 3-18 DRAFT - Do Not Quote or Cite
-------
1 contribution to PM2.5 mass reported for a FRM site. On the other hand, FRM sampling
2 results in some water that is associated with nitrate being included in the reported PM2 5
3 mass, while the nitrate mass reported by CSN (or IMPROVE) sampling excludes all
4 water. Continuous PM2 5 samplers employ a variety of methods for measuring PM2 5
5 mass, with correspondingly different behavior regarding retention/loss of nitrate.
6 However, it is generally accepted that the continuous PM2 5 sampling methods used at the
7 study sites have less nitrate retention than a CSN sampler, and are more like the FRM
8 method in that regard.
9 • Sulfate: Unlike nitrate, sulfate is not subject to loss once collected by a filter, but like
10 nitrate the issues of neutralization and water retention apply. Also, as for nitrate, as a first
11 order approximation, continuous PM2 5 instruments can be assumed to be more like FRM
12 samplers in reporting the mass effect of sulfate than like CSN samplers.
13 • Elemental and Organic Carbon: Only the mass of carbon atoms is included in the
14 reported elemental carbon and organic carbon for a CSN (or IMPROVE) sampler. In
15 addition, the distinction between elemental and organic carbon atoms is dependent on the
16 specifics of the two different thermo-optical analytical methods used in the CSN vs. the
17 IMPROVE network.22 Also, the quartz filter used to quantify carbonaceous material in
18 CSN and IMPROVE sampling both absorbs and loses organic vapors during sampling,
19 while the Teflon filter in a FRM sampler does not absorb organic vapors (although PM
20 on the filter may do so). Therefore, some method other than direct measurement must be
21 used to estimate the total mass concentration of organic carbonaceous material in ambient
22 air. The IMPROVE program adjusts for absorption of vapors by subtracting a monthly
23 average backup filter value, and then applies a standard adjustment factor (1.4 in the
24 "old" IMPROVE method) to the remaining organic carbon measurement to estimate
25 organic carbonaceous material. In contrast, the standard reports from CSN sampling
26 submitted to AQS do not include these two adjustments, but it is routine for EPA staff to
27 apply similar adjustments for the same purpose, after reporting of CSN data to AQS. For
28 this assessment the SANDWICH approach to such adjustments (Frank, 2006) is used to
29 estimate the organic mass through a mass balance of component measured on the CSN
30 and FRM samplers.
31 • Hourly PM^: The continuous instruments used for measuring hourly PM2 5 mass were
32 different among sites (as listed in Appendix A), and none of the instrument types that
33 provided data for this assessment perfectly measures "true" ambient concentrations.
34 None of them, when averaged over 24 hours, exactly matches either the measurements of
35 PM2 5 mass from a FRM sampler or the sum-of-components reportable from CSN
36 sampling. Differences can arise because of differences in water capture and retention,
37 inconsistent absorption and loss of organic vapors and nitric acid vapor, etc. In 2006,
38 EPA developed and promulgated criteria for approval of continuous PM2.s samplers as
39 "federal equivalent methods". These criteria assure a minimum level of correlation
40 between approved continuous instruments and the FRM method, when data from both are
22 While CSN carbon sampling and analysis methods have recently been harmonized with IMPROVE methods at
many CSN sites, it was not until mid-2007 that the first 57 sites were using the harmonized methods. Consequently,
most of the elemental and organic carbon data used in this assessment were obtained with the original CSN methods.
September 2009 3-19 DRAFT - Do Not Quote or Cite
-------
1 expressed as 24-hour average concentrations. However, in 2005-2007 no commercially
2 available instruments were yet approved under those criteria. Consequently, the
3 continuous instruments providing data to this assessment can be assumed to have a range
4 of correlation performance versus the FRM. In light of these consistency issues, the
5 hourly data from the continuous instruments were taken to be most indicative of the
6 relative concentrations of PM2 5 from hour-to-hour, with less reliance on the absolute
7 accuracy of the continuous instruments.
8 Taking into consideration the above information, EPA staff combined the four types of
9 available PM2.5 data in each study area using the following steps. Figure 3-5 provides a flow
10 chart that may assist in understanding these steps.
11
12 Figure 3-5. Sequence of steps used to estimate hourly PM2.s components and total light
13 extinction
Consistent with FRM
FRM Data:
CMAQ: Diurnal
Profiles of PM25
Components
i 24-hour PM
! Components by SANDWICH
2.5
Steps 2 and 3
Tentative 1-hour PM2.5 Components
and Sum-of-Components
Steps 5 and 6
1-hour PM2.5
Components Adjusted to be
Consistent With measured
1-hour PM2.5
i 1-hour Relative
: Humidity Data
Steps 7 and 8
Estimates of 1-hour
PM,
'10-2.5
IMPROVE Light Extinction Algorithm
September 2009
3-20
DRAFT - Do Not Quote or Cite
-------
1 1. The SANDWICH method (Frank, 2006) was used to subdivide the 24-hour PM2.5 mass
2 reported by the FRM for each day and site into sulfate (including associated ammonium
3 and residual water during filter weighing), nitrate (including associated ammonium and
4 residual water during filter weighing), elemental carbon, organic carbonaceous mass, and
5 fine soil/crustal mass. This is done using information from the CSN measurements,
6 physical models, and day-specific temperatures. Significantly, in the SANDWICH
7 method, the component referred to as organic carbonaceous mass is actually a residual
8 whose value is determined as the difference between the PM2 5 mass determined from
9 weighing the FRM filter and the sum of the estimated masses of the other four mass
10 components as listed above.
11
12 2. The CMAQ-derived monthly diurnal profiles for sulfate, nitrate, elemental carbon,
13 organic carbon and fine soil/crustal, like the examples for Detroit in Figure 3-4, were
14 multiplied by the day-specific SANDWICH-based estimates of the 24-hour average
15 concentrations of these five PM2.5 components, to get day-specific hourly estimates of
16 these five components (including ammonium and water associated with sulfate and
17 nitrate ion).
18 3. The hourly concentrations of these five components (including ammonium and water
19 associated with sulfate and nitrate ion when the filter is weighed) were added together, to
20 get a sum-of-components estimate of hourly PM2.5 mass for the day of the FRM
21 sampling.
22 4. The hourly data from the continuous PM2.5 instrument on the day of the FRM sampling
23 were normalized by their 24-hour average, to get a diurnal profile. (Recall that days were
24 not used in this assessment if hourly PM2.5 mass data were missing for more than 25
25 percent of daylight hours.) This profile was applied to the 24-hour PM2.5 mass reported
26 by the FRM sampler, to get a second, FRM-consistent estimate of hourly PM2.5 mass for
27 the day of the FRM sampling. This is straightforward when all 24 values of 1-hour PM2.5
28 mass were available for the day. However, for some (but not many) days, some values
29 for continuously measured hourly PM2.5 mass were missing. In such cases, EPA staff
30 used only the hours with valid 1-hour PM2.5 mass values to develop the diurnal profile
31 and then applied the profile to the FRM value as just described. This keeps the average
32 of the valid 1-hour PM2.5 values equal to the 24-hour value from the FRM sampler.
33 5. The two estimates of hourly PM2.5 mass from steps 3 and 4 were compared, hour-by-
34 hour. By virtue of the way they were derived, the averages of these estimates across all
35 24 hours of the day will necessarily be the same (and will be equal to the 24-hour FRM
36 measurement). However, while the diurnal pattern of these two estimates of the same
37 physical parameter should also be generally similar, it can be expected (and it is
38 observed) that the hourly measurements from the continuous PM2.s instruments (after
39 adjustment to be consistent with the FRM data) have more hour-to-hour variability.
40 Figure 3-6 gives an example of this comparison, for one day for the Detroit study area.
41
42
September 2009 3-21 DRAFT - Do Not Quote or Cite
-------
Figure 3-6. Example from Detroit study area.
35
•5(1
„ 25 -
E
B>
3 on
c 20 -
o
I 15
c ' J
01
o
c
0 10 -
c;
0 -
^•^^ ft
~" "V \ /V AT^.X
» jr~^ IV / ^v * * /^« — ™ ^"
^^-V' ¥X/ ^^"^ ~^^ »
!_• — • • ^—*^'^
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
—•—FRM adjusted Hourly PM2. 5
— • — Sum of diurnalized SANDWICHED species data using CMAQ profiles
Example comparison from the Detroit study area of hourly PM2.5 mass on March 24, 2006 as
estimated by applying CMAQ-based diurnal profiles to SANDWICH estimates of 24-hour
component concentrations versus applying a diurnal profile derived from continuous PM2.5
measurements to FRM PM2 5 mass.
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Given that the continuous instrument is reacting to hour-specific local conditions that can
vary from hour-to-hour due to real variations in local emissions and dispersion/transport
conditions, while the CMAQ-based estimates contain much less specific information, the
diurnal pattern of PM2.5 mass observed by the continuous instrument (adjusted to be
consistent with the FRM value for 24-hour average PM2.s) was taken as more reliable.
Within each hour, the estimates of all five components from step 2 were increased or
decreased by a common percentage (referred to below as A; where the subscript i
indicates the hour) so that the sum of the five components after this adjustment was equal
to the estimate of the hourly PM2.5 mass from step 4. The adjustment percentage varied
from hour-to-hour. Necessarily, in some hours the adjustment is an increase in the
concentrations of all components, and in other hours it is a decrease. While this
adjustment preserves the consistency between the 24 values of hourly PM2.5 mass and the
24-hour FRM mass, it can disturb the consistency between the hourly estimates of PM2.5
components and the SANDWICH-based estimates of 24-hour average component
concentrations. This disturbance was generally small, because the adjustments
necessarily go in one direction for some hours and the other direction for other hours. For
example, for the particular day in Detroit used for illustration purposes in Figure 6, the
effect of this step was to cause a discrepancy of 3 percent between the SANDWICH-
based values of 24-hour sulfate concentration and the average of the 24 estimates of 1-
September 2009
3-22
DRAFT - Do Not Quote or Cite
-------
1 hour sulfate concentrations (the positive percent indicates a higher concentration in the
2 result of this step than the SANDWICH-based value). The discrepancies were 1, 1,2,
3 and 2 percent for nitrate, elemental carbon, organic carbon, and fine soil/crustal,
4 respectively.
5 7. Each hourly estimate of sulfate concentration from step 6 (which includes estimates of
6 associated ammonium and particle bound water) was adjusted so that it excludes water
7 and reflects full neutralization and therefore is consistent with the reporting practices of
8 the IMPROVE program and the IMPROVE algorithm. This was done via these sub-
9 steps:
10 a. The 24-hour CSN value for the dry mass of sulfate ion (not SANDWICHed, no
11 ammonium) was multiplied by 1.375 to reflect an assumption of full
12 neutralization.
13 b. The ratio of this fully neutralized 24-hour sulfate mass to the SANDWICH-based
14 24-hour sulfate value was calculated.
15 c. This ratio was applied to each individual hour's sulfate concentration from step 6.
16 As in Step 6, it is possible for the 24 final hourly sulfate estimates to no longer be
17 exactly consistent with the 24-hour CSN sulfate measurement.
18
19 8. A similar adjustment as in step 7 (for sulfate) was made to each hour's nitrate
20 concentration from step 6, so that the estimate of hourly nitrate would reflect actual
21 atmospheric conditions and be consistent with the IMPROVE algorithm. However, the
22 ratio approach used in step 7(b) for sulfate could not be applied for nitrate, so this
23 adjustment had to be more complicated. Because in warm weather the FRM Teflon filter
24 does not retain nitrate, the initial FRM-consistent nitrate estimate derived by applying the
25 SANDWICH method to the FRM and CSN data can be zero. Such a zero value makes it
26 impossible to use the ratio approach in 7(b). Instead, the adjustment was made as
27 follows:
28 a. The 24-hour CSN value for nitrate ion (not SANDWICHed, no ammonium) was
29 multiplied by 1.29 to reflect an assumption of full neutralization.
30 b. This 24-hour value was then diurnalized using the CMAQ-based profile, similar
31 to step 2.
32 c. Each resulting hourly value of nitrate was further multiplied by the Ai factor from
33 step 6.
34 d. This new estimate of hourly nitrate was used to replace the initial nitrate value
35 that had resulted from step 6.
36 For cooler areas and days in which the 24-hour SANDWICH results include some nitrate,
37 the effect of these steps for nitrate are exactly the same as the effects of step 7 for sulfate (except
38 for the 1.29 vs. 1.37 neutralization factor). For warmer areas and days in which the 24-hour
September 2009 3-23 DRAFT - Do Not Quote or Cite
-------
1 SANDWICH results did not include any nitrate even though nitrate was measured on the CSN
2 Nylon filter, the effect of these steps is to assign the CSN nitrate to each hour using a
3 combination of the information in the CMAQ-based profiles and the information provided by the
4 continuous PM2.5 sampler. As in Step 6, it is possible for the 24 final hourly nitrate estimates to
5 no longer be exactly consistent with the 24-hour CSN nitrate measurement.
6 3.3.2 Hourly PMi0-2.s Concentrations
7 Three different paths were used to estimate hourly PMi0-2.5 concentrations, in the
8 following order of preference:
9
10 1. When hourly data from a collocated PMio instruments were available at the continuous
11 PM2 5 site in a study area, PM2 5 was subtracted hour-by-hour from PMi0. Negative values
12 were reset to zero.
13
14 2. When collocated continuous PMio data were not available at the continuous PM2.5 site in
15 a study area, but continuous PMio data were available at another site in the same study
16 area, PMio-2.5 was estimated by subtraction, implicitly assuming that the latter site was
17 also representative of PMio at the former site.
18
19 3. If neither of the first two methods was possible, a regional average ratio of PMi0-2.5 to
20 PM2.5 determined from an analysis of 24-hour data for the 2005 Staff Paper was applied
21 to hourly PM2.5 from the continuous instrument associated with the study area.
22
23 The estimation of PMi0-2.5 was further complicated because some types of data were
24 missing for isolated hours in the 2005-2007 period. As result, even for a single study area more
25 than one method sometimes had to be used to estimate hourly PMio-2.5. Appendix A gives more
26 specifics about the estimation of hourly PMio-2.5 in each study area.
27 The three-path approach described here is similar to that used for the visibility analysis
28 reported in the 2005 Staff Paper. While the second and third paths involve the use of data and
29 assumptions that are not robust compared to the use of paired, collocated, same-method
30 continuous instruments or to the use of paired low-volume filter-based samplers, in most areas
31 and periods the contribution to total light extinction from the resulting PMio-2.5 concentrations
32 was not large compared to the light extinction due to PM2.5 components.
33 3.3.3 Hourly Relative Humidity Data
34 Hourly relative humidity (RH) data for each study area's primary monitoring site were
35 obtained hour-by-hour from the closest available non-missing relative humidity measurement, as
36 reported by either an air monitoring station reporting such data to AQS or a National Weather
37 Service (NWS) station. For the AQS RH data, parameter 62201 values were utilized. RH data
38 from both sources are expressed as percentages.
September 2009 3-24 DRAFT - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
3.3.4 Calculation of Hourly and Daily Maximum 1-Hour Total Light Extinction
The original IMPROVE algorithm was applied hour-by-hour to estimate total light
extinction in each study area.
Because the interest in this analysis is on visibility during daylight hours, EPA staff
applied an hour-of-day filter to identify those hours that were daylight hours. The actual times of
local sunset and sunrise for each day and area were taken from tables of sunrise and sunset
available from the US Naval Observatory for each of the 15 urban areas. Daylight hours were
defined as any hour (e.g., 8:00 AM to 8:59 AM) containing no minutes before sunrise or after
sunset. For simplicity, these were generalized, so that all the days within each "season" in all
study areas were considered to have the same daylight hours.23 Table 3-5 shows the resulting
definition of daylight hours for the study areas. Unless otherwise stated, all subsequent
discussion of the results refers only to the values of parameters during these daylight hours.
Table 3-5. Assumed daylight hours by season (Local Standard Time)
First hour that is
entirely daylight
Last hour that is
entirely daylight
Number of
daylight hours
November-
January
8:00-9:00 AM
3: 00-4: 00PM
8
February-April
7:00-8:00 AM
5:00-6:00 PM
11
May-July
5:00-6:00 AM
6:00-7:00 PM
14
August-October
6:00-7:00 AM
5:00-6:00 PM
12
16
17
18
19
20
21
22
23
24
25
26
27
28
Daily maximum 1-hour total light extinction is the statistic of most interest in this
assessment, as briefly discussed in section 2.6. Days were set aside and not used to determine
this statistic if the hourly PM2.5 mass value was missing (or reported to be less than zero) for
more than 25 percent of daylight hours. Two or three missing daylight hours were allowed,
depending on season.
In this assessment, we capped the value of the humidity adjustment factor in the
IMPROVE algorithm ("f(RH)") at the value it has for a relative humidity of 95 percent. The
effect of measurement errors in relative humidity at values above 95 percent on the value of
f(RH) and thus on reconstructed total light extinction is considerable because of the highly
nonlinear form of the function in that range. This creates uncertainty as to the representativeness
of the extinction values calculated with high values of relative humidity. In addition, very high
values of relative humidity may be due to ongoing or very recent precipitation or fog. Persons
23 This simplification may be eliminated for the final version of this assessment.
September 2009
3-25
DRAFT - Do Not Quote or Cite
-------
1 may not expect or value clear visibility during such conditions. Later, consideration will be
2 given to the interpretation of these results and the appropriate treatment of days with very high
3 relative humidity in the statistical form of possible secondary PM NAAQS aimed at protection of
4 visibility in urban areas.
5 3.4 SUMMARY OF RESULTS FOR CURRENT CONDITIONS
6 3.4.1 Levels of Estimated PM2.s, PM2.s Components, PMi0-2.s, and Relative Humidity
7 Figure 3-7 presents box-and-whisker plots to illustrate the distributions in each study area
8 of the estimates of 1-hour PM2.5 (the diurnalized FRM value, resulting from step 4 in section
9 3.4.1), PMio-2.5, and relative humidity over the entire 2005-2007 study period. In the plot for
10 each parameter, areas are ordered by longitude, to make it easier to see east-versus-west regional
11 differences. For these three parameters, the distributions are given for all the daylight 1-hour
12 estimates. Similar plots of the daily maximum daylight 1-hour values of PM2.5, PMio-2.5, and
13 relative humidity concentration are available in Appendix B, as are plots of all daylight 1-hour
14 values for each of the PM2.5 component species.
15 From these plots we see that the distributions of PM2.5 generally trend toward higher
16 concentrations from west to east except for the two California urban locations which have PM2 5
17 concentrations more typical of eastern areas. The lowest median PM2.5 concentrations are in
18 Tacoma, WA, and Phoenix, AZ. Median PMio-2.s concentrations are highest in St. Louis, MO,
19 and Phoenix, AZ, and lower elsewhere. The highest outlier PMi0.2.5 concentrations are in St.
20 Louis, MO, and Los Angeles, CA. Relative humidity is lowest for the western urban areas
21 except for Tacoma, WA, which is similar to the northeastern urban locations with respect to
22 humidity. These hourly daylight PM concentration and relative humidity box and whisker plots
23 are consistent with our expectations based on regional 24-hour PM concentration values and
24 humidity climatology.
25 3.4.2 Levels of Estimated Total Light Extinction
26 Figure 3-8 presents box-and-whisker plots to illustrate the distributions of the estimates
27 of daylight 1-hour reconstructed total light extinction levels in each area in each year. The
28 distribution of the individual 1-hour values (8a) and the daily maximum 1-hour values (8b) are
29 both shown. The horizontal dashed lines in the plots represent the low, middle, and high
30 candidate protective levels (CPL) as discussed in section 2.6. These benchmarks for total light
31 extinction are 74, 122, and 201 Mm"1, corresponding to the benchmark VAQ values of 20 dv, 25
32 dv and 30 dv. Table 3-6 provides the percentages of days (across all of 2005-2007, unweighted)
33 in which the daily maximum daylight 1-hour total light extinction level was greater than each of
34 the three candidate protective levels.
September 2009 3-26 DRAFT - Do Not Quote or Cite
-------
1 As was also seen in the comparable PM2.5 box and whisker plots in Figure 3-7, the hourly
2 total light extinction values in Figure 3-8 tend to be higher in the eastern urban areas and lower
3 in the non-California western urban areas. The distributions of maximum daily total light
4 extinction values are higher (Figure 3-8b), as expected, than for all hours (Figure 3-8a). Both
5 Figure 3-8 and Table 3-6 indicate that all 15 urban areas have daily maximum hourly total light
6 extinctions that exceed even the highest of the CPL some of the time. Again, the non-California
7 western urban locations have the lowest frequency of maximum hourly total light extinction with
8 values in excess of the high CPL less than 10% of the time. Except for the two Texas and the
9 non-California western urban areas, all of the other urban areas exceed that high CPL about one-
10 quarter to one-half the time. Based on these estimated maximum hourly total light extinction
11 estimates, all 15 of the urban areas exceed the low CPL for about 60% to 100% of the days. As
12 noted in section 3.2.1, in most of the study areas the study site used in this assessment is not the
13 site in the study area with the highest concentrations of PM2.5.
14 In the last review of the secondary PM NAAQS, the pattern of light extinction during the
15 day was of particular interest. To illustrate the distributions of 1-hour total light extinction levels
16 in specific daylight hours, Figure 3-9 shows the distributions of 1-hour total light extinction
17 across the entire three-year study period, individually for the study areas. (Appendix E provides
18 additional graphics related to temporal/spatial patterns of light extinction.) These plots show that
19 high total light extinction can occur during any of the daylight hours, though for most of these
20 urban areas the early morning hours have the highest total light extinction. Urban areas without
21 a prominent preference for early morning high total light extinction include Phoenix, AZ; Salt
22 Lake City, UT; Tacoma, WA; Fresno, CA; and Philadelphia, PA.
September 2009 3-27 DRAFT - Do Not Quote or Cite
-------
1
2
3
4
5
6
1
Figure 3-7. Distribution of PM parameters and relative humidity across the 2005-2007
period, by study area
(a) Estimates of 1-hour PM2.s mass, based on applying continuous instrument-based
diurnal profiles to 24-hour FRM PMi.s mass
PM 2.5 hourly (Daylight Hours)
i
T r
(b) Estimates of 1-hour PMi0-2.s
PM Coarse hoyrty (DayKght Hoyrs)
« -
I.I
—
i
i
i i
September 2009
3-28
DRAFT - Do Not Quote or Cite
-------
1
2
3
4
Figure 3-7. (cont.). Distribution of PM parameters and relative humidity across the 2005-
2007 period, by study area
(c) 1-hour relative humidity
Relative Humidity hourly (Daylight Hours)
September 2009
3-29
DRAFT - Do Not Quote or Cite
-------
1
2
3
4
Figure 3-8. Distributions of estimated daylight 1-hour total light extinction and maximum
daily daylight 1-hour total light extinction across the 2005-2007 period, by study area.
(a) Individual 1-hour values
Hourly Extinction (DayjighS Hours)
123S 3643 33S3
34S7 3106 1652 3273 3930 3262
3O9 2095 161S 2515
5
6
cT
tf*
(b) Maximum daily values
Daily Maxsimim Extinction (DaylighS Hours)
110 324 302 86 306 27$ 149 294 350 295 141 284 18? 14S 228
s -• •
"T—r ^'
X'
-------
1
2
3
4
Table 3-6. Percentage of days in which daily maximum daylight 1-hour total light
extinction exceeded three candidate protective levels across the 2005-2007 period,
by study area
Study Area
Tacoma
Fresno
Los Angeles-South
Coast Air Basin
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann Arbor
Pittsburgh
Baltimore
Philadelphia-
Wilmington
New York-
N. New Jersey-Long
Island
Average
Number of Days with
Estimates
110
324
302
86
306
274
149
294
350
295
141
284
187
145
228
232
Candidate Protective Level
74 Mm -1
122 Mm -1
201 Mm -1
Percentage of days
68
80
92
59
61
86
89
100
96
95
91
93
88
95
91
86
36
51
80
13
24
53
58
86
80
80
79
70
65
76
70
61
9
24
49
3
9
14
21
55
52
34
57
43
38
46
46
33
September 2009
3-31
DRAFT - Do Not Quote or Cite
-------
1 Figure 3-9. Distributions of 1-hour total light extinction levels by daylight hour across the 2005-2007 period, by study area
2
1000 -
500 -
0-
E
1000-
500 -
o -
Phoenix, A2
Fresno, CA
o 3 i g
o f g §
"iff!
..»•**«,...
Atlanta, GA
Pittsburgh, PA
8 a §
Houston, TX
JJ|L|I 2« I $ J*
I *t* *li* fi*
Baltimore, MD
Salt Lake City, UT
o ooo
88 ° o o ° °
0 - o ° a
° a S 8 I °
**********
Los Angeles, CA
Birminqham, AL
St. Louis, IL
New York, NY
Dallas, TX
**uiii §°
iii!!*Uii
i ° °
I A t I
S E ****** K
Tacoma, WA
Philadelphia. PA
o 8
Detroit,
0 . n O
1000
- 500
-0
05 06 07 08 09 10 11 12 13 14 15 16 17 18 05 06 07 08 09 10 11 12 13 14 15 16 17 18 05 06 07 08 09 10 11 12 13 14 15 16 17 18 05 06 07 08 09 10 11 12 13 14 15 16 17 18 05 06 07 08 09 10 11 12 13 14 15 16 17 18
September 2009
3-32
DRAFT - Do Not Quote or Cite
-------
1 3.4.3 Patterns of Relative Humidity and Relationship between Relative Humidity
2 and Total Light Extinction
3 Figure 3-10 shows the distribution of relative humidity values at each daylight hour, for
4 each study area across 2005-2007 (Seasonal patterns are shown in Appendix E). As expected, in
5 every area relative humidity is lowest in the early afternoon, typically the warmest part of the
6 day. Relative humidity is most similar across areas in this time period, as observed in the 2005
7 Staff Paper. However, even in this period there are notable differences among areas. This
8 variation was not as evident in the information presented in the 2005 Staff Paper because only
9 regionally averaged information was presented. In all areas, there is considerable variation in
10 hour-specific relative humidity during the three-year period.
11 To allow closer inspection of estimated total light extinction values that have been
12 calculated using high relative humidity values, Figure 3-11 is a scatter plot of actual (uncapped)
13 1-hour relative humidity and 1-hour reconstructed total light extinction. Horizontal lines are
14 included in each of the individual plots corresponding to the three benchmarks for total light
15 extinction and a vertical line in each for 90 percent relative humidity. As stated above, f(RH) was
16 capped at its value when relative humidity is 95 percent. While some of the highest values of
17 total light extinction occur when relative humidity is above 90 percent, the majority of the
18 instances with total light extinction greater than the candidate protective levels occur when
19 relative humidity is 90 percent or lower. Notice that in Figure 3-11 there are plenty of high
20 humidity conditions for each urban area that correspond to low total light extinction values. This
21 is because humid air does not by itself contribute to light extinction. Particles composed of
22 material that absorbs water in high relative humidity conditions (e.g., sulfate and nitrate PM)
23 swell to larger solution droplets that scatter more light than their smaller dry particle counterparts
24 in a less humid environment. The magnitude of the relative humidity effect on light extinction
25 depends directly on the concentration of these hygroscopic PM components. (Figure 3-11
26 reveals skips in reported relative humidity values for some but not all the study areas. This is a
27 result of calculations of relative humidity from dry and wet bulb temperatures reported to the
28 nearest whole Celsius degree.)
September 2009 3-33 DRAFT - Do Not Quote or Cite
-------
1 Figure 3-10. Distributions of 1-hour relative humidity levels by daylight hour across the 2005-2007 period, by study area
2
60
40
20
0
;S
'E
100
80
60
40
20
0
Phoenix, AZ
o o
OOo
Fresno
CA
Atlanta
GA
Pittsburgh. PA
A
o ..
o o
Houston. TX
Baltimore, MD
Salt Lake City, UT
Los Angeles
CA
Birmingham
AL
St. Louis, IL
New York. NY
Dallas, TX
o o o o o o
-'-^§00000
) 8 B
Tacoma, WA
Philadelphia. PA
Detroit. Ml
100
- 80
60
-40
- 20
0
05 06 07 08 09 10 11 12 13 14 15 16 17 18 05 06 07 08 09 10 11 12 13 14 15 16 17 18 05 06 07 08 09 10 11 12 13 14 15 16 17 18 05 06 07 08 09 10 11 12 13 14 15 16 17 18 05 06 07 08 09 10 11 12 13 14 15 16 17 18
September 2009
3-34
DRAFT - Do Not Quote or Cite
-------
1 Figure 3-11. Scatter plot of daylight 1-hour relative humidity (percent) vs. reconstructed total light extinction (Mm *) across
2 the 2005-2007 period, by study area
UJ
0 20 40 60
0 20 40 60 80 100
0 20 40 60 80 100
Percent Relative Humidity
0 20 40 60 80 100
September 2009
3-35
DRAFT - Do Not Quote or Cite
-------
1 3.4.4 Extinction Budgets for High Total Light Extinction Conditions
2 An extinction budget for a single period shows the contribution that each PM component
3 makes to total light extinction via the additive terms of the IMPROVE algorithm. It can be
4 expected that the pattern in the extinction budgets will vary by time of year and by study area.
5 Examination of extinction budgets allows initial insights into what emission reduction approach
6 may be most effective in reducing total light extinction.
7 Figures 3-12 through 3-19 present day-specific maximum daylight 1-hour light extinction
8 budgets for the 10% of the days in each study area that have the highest daily maximum 1-hour
9 light extinction levels. (These figures show PM light extinction, not total light extinction. The
10 Rayleigh scattering term of 10 Mm"1 is not shown in Figures 3-12 through 3-19.) The pattern of
11 results shown in Figures 3-12 through 3-19 is generally as expected in light of emissions and
12 climate differences among study areas. Except for the PM2 5 soil component, each of the
13 components of total light extinction is a major contributor to extreme light extinction events at
14 some time and location. In the West, carbonaceous PM2.5 (i.e., organic mass and elemental
15 carbon), nitrate, and/or coarse mass (especially in Phoenix) tend to be most responsible for these
16 high haze hours. In the East it tends to be sulfate, nitrate, and the carbonaceous PM2 5
17 components that are the large contributors to total light extinction. From the sample period dates
18 we can determine the seasonal variations in major components. Nitrate and carbonaceous PM2.s
19 contribute more to the extreme light extinction periods during winter, while sulfate contributes
20 more in the summer. In many of the more northerly eastern urban areas, a combination of sulfate
21 and nitrate contributes to high light extinction year-round.
22 Looking at individual urban areas, Tacoma has its highest light extinction hours in the
23 colder months and primarily due to carbonaceous PM2.5 components. Extreme haze hours in the
24 two California urban areas are primarily caused by high nitrate PM2 5, though Los Angeles has
25 several extreme hours associated with coarse PM. Phoenix is unique among the 15 urban areas
26 in having most of its extreme light extinction caused by coarse PM, though there are a few top-
27 10-percent days where the maximum hourly haze is dominated by carbonaceous, sulfate, and
28 nitrate PM2.5. Salt Lake City has extreme haze hours caused mostly by nitrate in the winter with
29 some periods with carbonaceous PM2 5 being the major contributor. Dallas and Houston have
30 high contributions to total light extinction by sulfate PM2.5, but Dallas includes winter nitrate and
31 carbonaceous-caused extreme hours in the winter while Houston seems to have less contribution
32 by nitrate. Sulfate in the summer and nitrate in the fall and winter are responsible for most of the
33 extreme light extinction at St. Louis, though there are several maximum hourly periods where
34 coarse PM is a major component. Birmingham and Atlanta are similar in having sulfate year-
35 round and winter carbonaceous PM2 5 as major contributors to their extreme light extinction
36 periods. Detroit has frequent large light extinction contributions from nitrate PM2.5, mostly in
September 2009 3-36 DRAFT - Do Not Quote or Cite
-------
1 the winter, as well as some contributions from sulfate PM2.5 year-round and one winter period
2 with high contributions from carbonaceous PM2.5. The remaining four urban locations (i.e.,
3 Pittsburgh, Baltimore, Philadelphia, and New York) are similar in that most of their extreme
4 light extinction is from year-round combinations of sulfate and nitrate. New York also has some
5 winter carbonaceous contributions to its extreme light extinction.
6 3.5 POLICY RELEVANT BACKGROUND
7 Policy relevant background levels of total light extinction have been estimated for this
8 assessment by relying on outputs for the 2004 CMAQ run in which anthropogenic emissions in
9 the U.S., Canada, and Mexico were omitted, as described in the second draft ISA. Estimates of
10 PRB for total light extinction were calculated from modeled concentrations of PM2 5 components
11 using the IMPROVE algorithm. The necessary component concentrations were extracted from
12 the CMAQ output files, as they were not summarized in the second draft ISA. More detail is
13 provided in Appendix C.
14 It is also necessary to have estimates of PRB for PMio-2.5, as input to the IMPROVE
15 algorithm. The second draft ISA for this review does not present any new information on this
16 subject. The approach used in the two previous reviews was to present the historical range of
17 annual means of PMio-2.5 concentrations from IMPROVE monitoring sites selected as being least
18 influenced by anthropogenic emissions (US EPA, 2004, Table 3E-1). For this assessment, EPA
19 staff estimated PRB for PMio-2.5 using a contour map based on average 2000-2004 PMio-2.5
20 concentrations from all IMPROVE monitoring sites, found in a recent report from the
21 IMPROVE program (DeBell, 2006).
September 2009 3-37 DRAFT - Do Not Quote or Cite
-------
1 Figure 3-12. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily
2 1-hour PM light Extinction for 2005-2007 (Tacoma and Fresno)
Top 10% Daylight Hour Daily Maxima Tacoma, WA
soil • EC • NO3 n SO4 O OCM D PMc
Top 10% Daylight Hour Daily Maxima Fresno. CA
4
5
soil • EC • NO3 n so* a OCM a PMC
September 2009
3-38
DRAFT - Do Not Quote or Cite
-------
1
2
3
4
Figure 3-13. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily
1-hour PM light Extinction for 2005-2007 (Los Angeles and Phoenix)
Top 10% Daylight Hour Daily Maxima Los Angeles. CA
Top 10% Daylight Hour Daily Maxima: Phoenw, AZ
soil • EC • N03 D S04 D OCM O PMc
"•%•
September 2009
3-39
DRAFT - Do Not Quote or Cite
-------
1 Figure 3-14. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily
2 1-hour PM light Extinction for 2005-2007 (Salt Lake City and Dallas)
Top 10% Daylight Hour Daily Maxima: Salt Lake Cty, UT
o,, "o,,
~ „, <%, ^
•or •»„ ">, ">, V, ~~o, ''o, '"o, ^o, ~~o, la- V, s/s V
Vu ^ ? » ;i V> "is 'iJ a. \a ^ ^> j ? jo, (i. ^
• soil • EC • NO3 D SO4 n OCM D PMc
Top 10% Daylight Hour Daily Maxima Dallas, TX
September 2009
3-40
DRAFT - Do Not Quote or Cite
-------
1 Figure 3-15. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily
2 1-hour PM light Extinction for 2005-2007 (Houston and St. Louis)
Top 10% Daylight Hour Daily Maxima- Houston. TX
soil • EC • HO? n soi n OCM a PMC
Top 10% Daylight Hour Daily Maxima: SI. Louis. IL
4
5
vvv"
EC • NO3 n SO4 n OCM a PMC
September 2009
3-41
DRAFT - Do Not Quote or Cite
-------
1 Figure 3-16. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily
2 1-hour PM light Extinction for 2005-2007 (Birmingham and Atlanta)
Top 10% Daylight Hour Daily Maxima: Birmingham, AL
• soil • EC • NO3 D SO4 D OCM D PMc
Top 10% Daylight Hour Daily Maxima Atlanta. GA
oil • EC • NO3 a $04 a OCM a PMC
September 2009
3-42
DRAFT - Do Not Quote or Cite
-------
1 Figure 3-17. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily
2 1-hour PM light Extinction for 2005-2007 (Detroit and Baltimore)
Top 10% Daylight Hour Dally Maxima. Detroit Ml
• soil • EC • NO3 D SO4 n OCM D PMc
Top 10% Daylight Hour Daily Maxima Baltimore MD
• • i
> <%> % %> % %>
\
en • EC • NO3 a $04 a OCM a PMC
September 2009
3-43
DRAFT - Do Not Quote or Cite
-------
1 Figure 3-18. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily
2 1-hour PM light Extinction for 2005-2007 (Pittsburgh and Philadelphia)
Top 10% Daylight Hour Daily Maxima- Pittsburgh. PA
• soil • EC • NO3 D SO-4 0 OCM D PMc
Top 10% Daylight Hour Daily Maxima: Philadelphia, PA
lll.
^ "%
en • EC • NO3 a $04 a OCM a PMC
September 2009
3-44
DRAFT - Do Not Quote or Cite
-------
1 Figure 3-19. Light Extinction Budgets for the Top 10 Percent of Days for Maximum Daily
2 1-hour PM light Extinction for 2005-2007 (New York)
Top 10% Daylight Hour Daily Maxima: New York. NY
soil • EC • NO3 D SO4 D OCM O PMc
September 2009
3-45
DRAFT - Do Not Quote or Cite
-------
1
2
4 TOTAL LIGHT EXTINCTION UNDER "WHAT IF"
CONDITIONS OF JUST MEETING SPECIFIC ALTERNATIVE
SECONDARY NAAQS
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
4.1 ALTERNATIVE SECONDARY NAAQS BASED ON MEASURED TOTAL
LIGHT EXTINCTION AS THE INDICATOR
4.1.1 Indicator and Monitoring Method
As proposed in the Scope and Methods plan, the indicator considered in this section is
total light extinction, assumed to be measured by a continuous instrument, or instrument pair,
capable of reporting both light scattering and light absorption. For example, the measurement
method could be an Aethalometer® or similar instrument for measuring light absorption paired
with a nephelometer, with both instruments using a PMio inlet so that total light extinction due to
PM2.5 and PMio-2.s combined (and gases) would be measured. The measurement would include
the effect of Rayleigh scattering by gases, while the alternative NAAQS would be intended to
provide protection from the loss in visual air quality due to PM. Therefore, it would be
necessary to account for the contribution to measured total light extinction from Rayleigh
scattering either when setting the level of the NAAQS (by adding an increment of about 10 Mm"1
to the intended permitted level of light extinction caused by PM) or in the data interpretation rule
for comparing instrument readings to the NAAQS (by subtracting about 10 Mm"1 before
comparison to the level of the NAAQS).
4.1.2 Alternative Secondary NAAQS Scenarios based on Measured Total Light
Extinction
Six alternative NAAQS scenarios presented in Table 4-1 are analyzed in this section,
each based on daily maximum daylight 1-hour total light extinction. The scenarios are ordered
from least to most stringent.
Table 4-1. Alternative Secondary NAAQS Scenarios for Light Extinction
Level (including Rayleigh
scattering)
201 Mm-1
201 Mm-1
122 Mm-1
122 Mm-1
74 Mm-1
74 Mm-1
Annual
Percentile
90
95
90
95
90
95
Form
3-year average of percentile value
3-year average of percentile value
3-year average of percentile value
3-year average of percentile value
3-year average of percentile value
3-year average of percentile value
27
September 2009
4-1
DRAFT - Do Note Quote or Cite
-------
1 It is useful to think ahead to monitor siting aspects of NAAQS implementation, so that
2 the suitability of the monitoring sites used in this assessment for the purpose of this section can
3 be considered.
4 4.1.3 Monitoring Site Considerations for Alternative Secondary NAAQS Based on
5 Measured Total Light Extinction
6 It is most likely that instruments that would be used to implement a secondary NAAQS
7 with a measured light extinction indicator will be "closed path" instruments that react only to air
8 quality in their immediate vicinity. However, light paths that matter to perceived visual air
9 quality are likely to be several kilometers long. Therefore, a monitoring site should be at least
10 neighborhood in scale, i.e., its relationship to emission sources and transport should be such that
11 measurements made at the site reasonably reflect concentrations in an area surrounding the site
12 of at least about 0.5 to 4 kilometers in diameter.
13 It would be logical to require that in any urban area for which light extinction monitoring
14 is deemed a necessary requirement, at least one monitoring site would be placed in an area
15 expected to have the maximum total light extinction conditions, subject to the above scale of
16 representation consideration and possibly also subject to the condition that the site be in an area
17 (or reasonably represent such an area) where scenic vistas are able to be perceived by people.
18 i.e., that the site is "population oriented." Given that site paths of concern will typically be
19 several kilometers long, it is difficult to imagine a neighborhood scale monitoring location within
20 the census-defined urbanized area of an urban area which would not be "population oriented" for
21 purposes of visual air quality, as "neighborhood" size land areas typically would have residents,
22 workers, etc. somewhere within them during daylight hours.
23 With regard to the monitoring sites used in this assessment, all are reported to be, or
24 appear to be, neighborhood or larger scale, and all are in areas where people are present during
25 daylight hours. The sites in Detroit (Dearborn) and New York-N.New Jersey are, however, rather
26 close to an industrial source and a major interstate highway interchange/turnpike exit,
27 respectively. Significantly, most of the study sites are not the highest PM2.5 concentration site
28 in their urban area, so a "what if scenario that manipulates the "current conditions" at these sites
29 to "just meet" an alternative secondary NAAQS might implicitly leave other parts of their urban
30 areas with total light extinction above the NAAQS.
31 4.1.4 Approach to Modeling "What If Conditions for Alternative Secondary
32 NAAQS based on Measured Total Light Extinction
33 Before modeling "what if conditions, EPA staff augmented the data set described in
34 Table 4 so that the sets of study days for Houston and Phoenix were seasonally balanced despite
35 the lack of actual monitoring data for one quarter in each city. For the first quarter of 2005 in
September 2009 4-2 DRAFT - Do Note Quote or Cite
-------
1 Phoenix, we substituted the available 12 days from the first quarter of 2006. For the fourth
2 quarter of 2007 in Houston, we substituted 13 randomly drawn days from the fourth quarters of
3 2005 and 2006.
4 Also, Tacoma (originally) and Phoenix (after this augmentation) each have only two
5 calendar years of suitable data, while the form of the alternative NAAQS scenarios requires the
6 averaging of the 90th or 95th percentile values from three years. In Tacoma and Phoenix, for
7 every step in the analysis at which a design value is used as an input or reported as an output, we
8 averaged the percentile values from the only two available years.
9 We modeled daily maximum daylight 1-hour total light extinction under each of the
10 "what if scenarios (in which each study area "just meets" one of the alternative secondary
11 NAAQS listed in section 4.1.2) via the following steps. These steps are essentially the same as
12 the "proportional rollback" steps that have been used in the health risk assessment modeling of
13 "what if conditions in several previous NAAQS reviews for PM and other criteria pollutants.
14
15 1. Identify the appropriate percentile (90th or 95th) daily light extinction value in each
16 year, noting the day and hour each occurred, and average these values across years to
17 calculate the light extinction design value for each site consistent with the percentile
18 form of the NAAQS scenario. The two resulting design values for each area (for the
19 90th and 95th percentile forms) are shown in Table 4-2. (Note that in a few cases,
20 which are identified by a footnote, the study area meets one or more of the NAAQS
21 scenario under current conditions. In these cases, the "current conditions" total light
22 extinction values are not adjusted, i.e., total light extinction values are never "rolled
23 up.")
24
25 2. Using the same days and hours, find the three (or two, in the case of Phoenix and
26 Houston for which there were only two years of suitable data available)
27 corresponding values of PRB total light extinction, and average these values across
28 years to calculate the PRB portion of the design value.
29
30 3. Subtract the value from step 2 from the value from step 1, to determine the non-PRB
31 portion of the design value.
32
33 4. Calculate the percentage reduction required in non-PRB total light extinction in order
34 to reduce the design value to the total light extinction level that defines the NAAQS
35 scenario, using the following equation:
36
37 Percent reduction required = 1 - (NAAQS level - PRB portion of the design value)/(non-PRB
38 portion of the design value)
39
40 The percentage reductions determined in this step are shown in Table 4-3.
411.
September 2009 4-3 DRAFT - Do Note Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
5. Turning to the entire set of day/hour-specific actual and PRB daylight total light
extinction values for the three (or two) year period, determine the non-PRB portion of
total light extinction, reduce it by the percentage determined in step 4, and add back
in the PRB total light extinction. The result is the "just meets" total light extinction
value for that day and hour.
Note that in these steps, it is not necessary to make any explicit or implicit assumption
about what PM components would be reduced to allow the area to meet the NAAQS scenario, as
the NAAQS scenario's target design value is itself in units of light extinction.
Table 4-2. Current Conditions total light extinction design values for the study areas.
Study Area
Tacoma
Fresno
Los Angeles-South Coast Air
Basin
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann Arbor
Pittsburgh
Baltimore
Philadelphia- Wilmington
New York-N.New Jersey-
Long Island
Design Value for 90th
Percentile Form (Mm *)
228
308
323
117*
184
213*
235
420
436
269
444
425
441
409
449
Design Value for 95th
Percentile Form (Mm *)
278
403
436
154*
256
262
275
512
565
291
636
481
484
436
538
* This design value meets one or more of the NAAQS scenarios.
12
September 2009
4-4
DRAFT - Do Note Quote or Cite
-------
1
2
Table 4-3. Percentage reductions in non-PRB light extinction required to "just meet" the
NAAQS scenarios based on measured light extinction.
Total Light Extinction
Level (Mm ')
Percentile Form
Area
Tacoma
Fresno
Los Angeles
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit
Pittsburgh
Baltimore
Philadelphia
New York
NAAQS Scenarios Based on Maximum Daily 1-hour Daylight Total
Light Extinction, Average of Percentile Value Over Three Years
201
90th
201
95th
122
90th
122
95th
74
90th
74
95th
Percentage Reduction Required in
Non-PRB Total Light Extinction
13
38
40
0
0
7
16
55
56
28
58
55
58
54
59
29
53
56
0
23
26
29
64
67
33
71
61
61
57
65
51
65
66
0
36
49
53
75
75
60
77
75
76
74
77
59
73
75
23
56
59
60
80
82
63
84
78
79
76
80
74
82
82
43
65
75
76
87
87
79
88
87
88
86
88
77
86
87
59
76
79
79
90
90
80
91
88
89
88
90
3
4
5
6
7
8
9
10
11
12
13
14
15
16
4.2 ALTERNATIVE SECONDARY PM2.5 NAAQS BASED ON ANNUAL AND
24-HOUR PM2.5 MASS
4.2.1 Secondary NAAQS Scenarios Based on Annual and 24-hour PM2.s Mass
In this draft version of the assessment, EPA staff have modeled two "what if scenarios
based on the same indicators and averaging periods as define the current secondary PM2.5
NAAQS:
• 15 |ig/m3 weighted annual average PM2.5 concentration and 35 |ig/m3 24-hour average
PM2.5 concentration with a 98th percentile form, both averaged over three years. These are
the current secondary NAAQS for PM2.5.
• 12 |ig/m3 weighted annual average PM2.5 concentration and 25 |ig/m3 24-hour average
PM2.5 concentration with a 98th percentile form, both averaged over three years.
These are the highest and lowest alternative NAAQS scenarios considered in the health
risk assessment, and therefore encompass the full range of alternative primary PM2.5 NAAQS
being analyzed by EPA staff.
September 2009
4-5
DRAFT - Do Note Quote or Cite
-------
1 4.2.2 Approach to Modeling Conditions If Secondary PM2.s NAAQS Based on
2 Annual and 24-hour PM2.s Mass Were Just Met
3 Because these NAAQS scenarios are based on PM2 5 mass as the indicator, rather than
4 light extinction, the steps needed to model "what if conditions are somewhat different, and
5 involve explicit consideration of changes in PM2.5 components.
6
7 1. Apply proportional rollback to all the PM2.5 monitoring sites in each study area,
8 taking into account PRB PM2.5 mass, to "just meet" the NAAQS scenario for the area
9 as a whole, not just at the visibility assessment study site. The health risk assessment
10 document describes this procedure in detail. The degree of rollback is controlled by
11 the highest annual or 24-hour design value, which in most study areas is from a site
12 other than the site used in this visibility assessment. The relevant result from this
13 analysis is the percentage reduction in non-PRB PM2 5 mass need to "just meet" the
14 NAAQS scenario, for each study area. These percentage reductions are shown in
15 Table 4-4. Note that Phoenix and Salt Lake City meet the 15/35 NAAQS scenario
16 under current conditions, and require no reduction. PM2.5 levels in these two cities
17 were not "rolled up."
18
19 2. For each day and hour for each PM2.5 component, subtract the PRB concentration
20 from the current conditions concentration, to determine the non-PRB portion of the
21 current conditions concentration.
22
23 3. Apply the percentage reduction from step 1 to the non-PRB portion of each of the
24 five PM2.s components. Add back the PRB portion of the component.
25
26 4. Re-apply the IMPROVE algorithm, using the reduced PM2 5 component
27 concentrations, the current conditions PMi0-2.5 concentration for the day and hour, and
28 relative humidity for the day and hour. Include the term for Rayleigh scattering.
September 2009 4-6 DRAFT - Do Note Quote or Cite
-------
1
2
Table 4-4. Percentage reductions required in non-PRB PM2.5inass to "just meet" NAAQS
scenarios based on annual and 24-hour PM2.s mass
Study Area
Tacoma
Fresno
Los Angeles-
South Coast
Air Basin
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann
Arbor
Pittsburgh
Baltimore
Philadelphia-
Wilmington
New York-
N.New Jersey-
Long Island
Percentage Reduction Required
Annual PM2.5 NAAQS = 15 ug/m3
24-hour PM2.5 NAAQS = 35 ug/m3
19
45
37
0*
37
0*
6
10
22
8
19
Being recalculated
6
8
17
Annual PM25 NAAQS = 12 ug/m3
24-hour PM2.5 NAAQS = 25 ug/m3
43
61
55
22
56
7
27
37
45
30
43
Being recalculated
33
35
41
* These areas meet this NAAQS scenario under current conditions.
3
4
5
6
7
8
9
10
11
12
13
14
15
4.3 RESULTS FOR "JUST MEETING" ALL ALTERNATIVE SECONDARY
NAAQS SCENARIOS
The modeling described in sections 4.1 and 4.2 resulted in estimates of total light
extinction for each day and hour in each study area. Four summaries of these conditions are
presented here.
Figure 4-1 shows two box-and-whisker plots of daily maximum daylight 1-hour total
light extinction. The top panel (a) is for the single illustrative scenario of a NAAQS based on
measured light extinction with a level of 122 Mm"1 and a 90th percentile form, which was chosen
for this illustration because it is approximately mid-way among the six such scenarios in terms of
stringency. Plots for all six scenarios of NAAQS based on measured total light extinction are
provided in Appendix F. The bottom panel (b) is for the scenario of meeting the current
secondary PM2.5 NAAQS of 15 |ig/m3 for the annual average and 35 |ig/m3 for the 98th percentile
24-hour average. A notable feature of this comparison is that in the top panel, all the study areas
September 2009
4-7
DRAFT - Do Note Quote or Cite
-------
1 have a similar distribution of the daily maximum daylight 1-hour total light extinction, while in
2 the bottom panel this is not the case. This is expected, since a NAAQS based on a measured
3 total light extinction indicator will of course result in areas achieving similar total light extinction
4 patterns once each area reaches a "just meets" condition; in areas with generally higher relative
5 humidity conditions, concentrations of PM2.5 components and/or PMio-2.5 would need to be lower
6 to achieve the "just meet" condition. In contrast, in the NAAQS scenario represented by the
7 bottom panel, concentrations of PM2.5 mass will be similar across areas, but concentrations of
8 PM2.5 components may not be, and levels of total light extinction will not be similar in areas with
9 dissimilar levels of relative humidity. The specific differences among areas in the bottom panel
10 are generally as expected, with the drier study areas having lower levels of total light extinction.
11 Tables 4-5 and 4-6 summarize the "just meets" conditions in the eight NAAQS scenarios
12 in terms of the total light extinction design value. Table 4-5 addresses the six scenarios of
13 NAAQS based on measured total light extinction, and the form of the design value given in the
14 Table corresponds to the assumed percentile form of the NAAQS. Table 4-6 addresses the two
15 scenarios of NAAQS based on PM2.5 mass, and total light extinction design values in both
16 percentile forms are shown. Note that the design values in Table 4-5 resulting from the rollback
17 steps described in section 4.1.4, in some cases do not exactly equal the assumed level of the
18 NAAQS, although all are quite close. This is a result of hours switching their ranking in the
19 rollback process. This can happen because the level of PRB total light extinction varies with
20 each hour, so a uniform percentage reduction in non-PRB light extinction (step 5) can result in
21 non-uniform percentage reductions in actual total light extinction. In principle, rollback could be
22 iterated to exactly achieve a design value equal to the level of the NAAQS for each scenario.
23 However, the discrepancies indicated in Table 4-5 were judged too small to justify iterative
24 rollback, given other uncertainties in the analysis.
25 Finally, Table 4-7 summarizes all eight scenarios in terms of the percentage of days
26 (across 2005 to 2007, but after rollback) in which the daily maximum daylight 1-hour total light
27 extinction under "just meeting" conditions exceeds each of the CPLs. Also shown at the bottom
28 of the table in each column representing a NAAQS scenario is the average of these percentages
29 across the 15 study areas. Comparisons of these percentages allows a rough indication of how the
30 two scenarios of a NAAQS based on PM2 5 mass compare to the other six scenario in terms of
31 protecting visual air quality. Notice that even the most restrictive of the two NAAQS scenarios
32 based on PM2.5 mass would permit projected 1-hour maximum daily light extinction above the
33 least restrictive CPL (201Mm"1) more that 10% of the time for most of the Eastern urban areas
34 (Dallas, Houston and Atlanta have values near 10%), while the percent of maximum hourly days
35 for the Western urban areas are all less than 10%.
36
September 2009 4-8 DRAFT - Do Note Quote or Cite
-------
Figure 4-1. Distributions of daily maximum daylight 1-hour total light extinction under
two "just meeting" secondary NAAQS scenarios
(a) Secondary NAAQS based on measured total light extinction with a level of 122 Mm"
1 and a 90th percentile form
ExtRollbackDailyMaxNAAQS122Pctl90
-I X
(b) Secondary NAAQS of 15 ug/m3 for the annual average and 35 ug/m3 for the 98th
percentile 24-hour average
PMRollbackDailyMaxCase! NAAQS
T T
-------
Table 4-5. Total light extinction design values for "just meeting" secondary NAAQS
scenarios based on measured total light extinction
Level (Mm ')
Percentile Form
Tacoma, WA
Fresno, CA
Los Angeles, CA
Phoenix, AZ
Salt Lake City, UT
Dallas, TX
Houston, TX
St. Louis, IL
Birmingham, AL
Atlanta, GA
Detroit, MI
Pittsburgh, PA
Baltimore, MD
Philadelphia, PA
New York, NY
Secondary NAAQS Scenario
74
90th
74
95th
122
90th
122
95th
201
90th
201
95th
Total Light Extinction Design Value
(based on same percentile form as the NAAQS scenario)
74
72
74
74
75
71
78
75
77
73
71
74
73
74
72
90
74
74
74
73
73
80
74
76
80
75
76
75
76
77
122
121
122
122
122
120
124
123
123
121
121
122
122
122
121
126
122
122
122
122
122
124
122
122
127
122
124
122
122
124
201
201
201
201
201
201
201
202
201
200
201
200
202
201
201
201
201
201
201
201
201
201
201
201
203
201
203
201
201
202
Table 4-6. Total light extinction design values for "just meeting" secondary NAAQS
scenarios based on PMi.s mass
Annual/l-hour PM2.5
NAAQS
City Name
Tacoma, WA
Fresno, CA
Los Angeles, CA
Phoenix, AZ
Salt Lake City, UT
Dallas, TX
Houston, TX
St. Louis, IL
Birmingham, AL
Atlanta, GA
Detroit, MI
Pittsburgh, PA
Baltimore, MD
Philadelphia, PA
New York, NY
15ng/m3/35ng/m3
90th %tile
Design Value
(Mm1)
188
183
221
117*
126
213*
224
384
355
249
364
Recalculating
419
377
377
95th %tile
Design Value
(Mm1)
228
238
311
154*
174
262*
261
477
476
271
520
Recalculating
459
403
450
12ng/m3/25ng/m3
90th %tile Design
Value (Mm1)
139
139
175
107
98
200
182
311
268
197
264
Recalculating
308
273
274
95th %tile Design
Value (Mm1)
165
179
261
145
133
245
211
383
369
218
376
Recalculating
335
296
325
* Phoenix and Dallas meet 15 ug/m 735 ug/m under current conditions, so these entries are the same as for current
conditions.
September 2009
4-10
DRAFT - Do Note Quote or Cite
-------
Table 4-7. Percentage of days across three years (two in the case of Phoenix and Houston) with maximum 1-hour daylight
total light extinction above CPLs when "just meeting" the NAAQS scenarios
Mm-1
Level
Percentile
Form
Annual/24-
hour
Area
Tacoma
Fresno
Los Angeles
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit
Pittsburgh
Baltimore
Philadelphia
New York
Average
Days above 74 Mm"1 (Percent)
201 201 122 122 74 74
90 95 90 95 90 95
157 127
35 25
Percentage of days
56 46 26 23 11 9 55 39
54 42 29 21 9 3 48 33
81 74 53 24 8 5 82 76
88 66 50 30 9 5 46 42
54 32 25 14 11 4 26 19
77 66 43 30 9 4 80 77
75 68 45 32 14 9 81 69
75 62 40 28 10 6 98 94
67 56 42 26 13 6 90 82
86 84 57 50 9 7 91 85
66 50 40 18 11 5 82 79
55 52 30 25 10 6 *
55 50 25 22 10 8 82 71
67 65 33 28 8 6 90 81
57 50 27 23 10 6 80 70
68 58 38 26 10 6 73 65
Days above 122 Mm"1 (Percent)
201 201 122 122 74 74
90 95 90 95 90 95
127
15/35 25
Percentage of days
25 18 8 6 2 1 23 11
27 19 10 4 1 1 23 14
49 20 8 5 2 1 59 28
46 27 9 5 2 1 76
24 13 10 4 3 2 11 6
42 26 11 421 46 42
40 30 11 711 48 31
36 25 10 621 79 66
36 22 11 521 66 52
50 42 10 700 68 49
36 16 11 510 70 59
27 22 10 6 0 0
23 22 10 600 60 46
29 26 8 6 0 0 70 54
26 21 11 611 60 45
34 23 10 511 49 36
Days above 201 Mm"1 (Percent)
201 201 122 122 74 74
90 95 90 95 90 95
15/35 12/25
Percentage of days
862000 6 3
10 4 1 1 0 0 8 2
8 5 3 1 0 0 12 7
953210 2 2
10 4 3 2 1 0 3 2
11 5 2 1 0 0 13 11
11 6 1 1 0 0 14 7
10 6 2 1 0 0 44 32
11 5 2 1 0 0 37 21
10 5 0 0 0 0 22 10
11 5 1 0 0 0 45 22
10 5 0 0 0 0
960000 34 22
860000 37 21
12 5 2 1 0 0 30 19
10 5 2 1 0 0 22 13
* EPA is currently recalucating these values.
September 2009
4-11
DRAFT - Do Note Quote or Cite
-------
5 REFERENCES
Abt Associates Inc., 2001. Assessing Public Opinions on Visibility Impairment due to Air Pollution: Summary
Report. Prepared for EPA Office of Air Quality Planning and Standards; funded under EPA Contract No.
68-D-98-001. Abt Associates Inc., Bethesda, MD. January. Available:
http://www.epa.gov/ttncaaal/tl/reports/vis rptfinal.pdf. Accessed 9/16/2008.
Air Resource Specialists. 2003. 2003 WinHaze Air Quality Modeler, Version 2.9. Air Resource Specialists, Inc.
Available: http://www.air-resource.com/downloads.php. Accessed 9/16/2008.
Arizona Department of Environmental Quality. 2003. Recommendation for a Phoenix Area Visibility Index.
Visibility Index Oversight Committee. Available: http://www.phoenixvis.net/PDF/vis 031403final.pdf.
Accessed 9/16/2008. BBC Research & Consulting. 2002. Phoenix Area Visibility Survey. Draft Report.
Available: http://www.azdeq.gov/environ/air/download/vis_021903f.pdf. Accessed 9/16/2008.
BBC Research & Consulting. 2003. Phoenix Area Visibility Survey. Draft Report. Available:
http://www.azdeq.gov/environ/air/download/vis 021903f.pdf. Accessed 9/16/2008.
DeBell L. (2006). Spatial and seasonal patterns and temporal variaibility of haze and its constituents in the United States: Report
IV. 1 Cooperative Institute for Research in the Atmosphere. Colorado State University Fort Collins, CO. 156388
Ely, D.W., J.T. Leary, T.R. Stewart, and D.M. Ross. 1991. The Establishment of the Denver Visibility Standard.
Presented at the 84th Annual Meeting & Exhibition of the Air and Waste Management Association. June
16-21.
Frank NH. (2006). Retained nitrate, hydrated sulfates, and carbonaceous mass in federal reference method fine
paniculate matter for six eastern U.S. cities. J Air Waste Manag Assoc, 56: 500-11
Henderson, R. (2006) Letter from Dr. Rogene Henderson, Chair, Clean Air Scientific Advisory Committee to the
Honorable Stephen L. Johnson, Administrator, US EPA. Clean Air Scientific Advisory Committee
Recommendations Concerning the Proposed National Ambient Air Quality Standards for Paniculate
Matter. March 21, 2006. Available: http://www.epa.gov/sab/pdf/casac-ltr-06-002.pdf
Henderson, R. (2008) Letter from Dr. Rogene Henderson, Chair, Clean Air Scientific Advisory Committee to the
Honorable Stephen L. Johnson, Administrator, US EPA. Clean Air Scientific Advisory Committee
Consultation on EPA's Draft Integrated Review Plan for the National Ambient Air Quality Standards for
Paniculate Matter January 3, 2008.
Jacques Whitford AXYS. 2007. The View Ahead; Identifying Options for a Visibility Management Framework for
British Columbia. Report for the British Columbia Ministry of Environment. Available:
http://www.env.gov.bc.ca/air/airqualitv/pdfs/view ahead.pdf. Accessed 8/5/2009.
Molenar, J.V., W.C. Malm, and C.E. Johnson. 1994. Visual air quality simulation techniques. Atmospheric
Environment 28(5): 1055-1063.
NARSTO (2004) Paniculate Matter Science for Policy Makers: A NARSTO Assessment. P. McMurry, M.
Shepherd, and J. Vickery, eds. Cambridge University Press, Cambridge, England. ISBN 0 52 184287 5
Pitchford, M. and W. Malm. 1993. Development and applications of a standard visual index. Atmospheric
Environment 28(5): 1049-1054.
Pryor, S.C. 1996. Assessing public perception of visibility for standard setting exercises. Atmospheric Environment
30(15):2705-2716.
September 2009 5-1 DRAFT - Do Note Quote or Cite
-------
RWDI AIR. 2008. Final Report: Establishing a Visibility Goal for Wilderness and Urban areas in British Columbia
and Canada. Report to the British Columbia Ministry of Environment. Available:
http://www.env.gov.bc.ca/air/airqualitv/pdfs/visibilitv goal report final.pdf. Accessed 8/5/2009.
Samet, J. (2009) Letter from Dr. Jonathan M. Samet, Chair, Clean Air Scientific Advisory Committee (CASAC) to
The Honorable Lisa P. Jackson, Administrator, U.S. EPA. Consultation on EPA's Particulate Matter
National Ambient Air Quality Standards: Scope and Methods Plan for Urban Visibility Impact Assessment.
May 21, 2009. Available: http://yosemite.epa.gov/sab/sabpeople.nsf/WebCommittees/CASAC.
Schmidt. M; Frank, N.; Mintz, D.; Rao, T.; McCluney, L. (2005). Analyses of paniculate matter (PM) data for the
PM NAAQS review. Memorandum to PM NAAQS review docket EPA-HQ-OAR-2001-0017. June 30,
2005. Available: http://www.epa.gOv/ttn/naaqs/standards/pm/sjm_cr_td.html
Smith, A. 2009. Comments on the First External Review Draft of EPA's "Integrated Science Assessment for
Particulate Matter." CRA International, Washington, DC. March 30. Prepared for the Utility Air
Regulatory Group. Submitted as public comment to the public meeting EPA Clean Air Science Advisory
Council. April 2.
Smith, A.E. and S. Howell. 2009. An Assessment of the Robustness of Visual Air Quality Preference Study Results.
CRA International, Washington, DC. March 30. Prepared for the Utility Air Regulatory Group. Submitted
as supplemental material to presentation by Anne Smith to the public meeting of the EPA Clean Air
Science Advisory Council. April 2.
Statistics Canada. 2009a. Population and Dwelling Count Highlight Tables, 2006 Census. Available:
http://wwwl2.statcan.gc.ca/census-recensement/2006/dp-pd/hlt/97-550/Index.cfm7Page = INDX&LANG
= Eng. Accessed 7/13/2009.
Statistics Canada. 2009b. Population: Chilliwack, British Columbia (Census Agglomeration). Available:
http://wwwl2.statcan.ca/census-recensement/2006/dp-pd/prof/92-591/details/page.cfm7Lang = E&Geol =
CMA&Codel = 930_&Geo2 = PR&Code2 = 59&Data = Count&SearchText = Chilliwack&SearchType
= Begins&SearchPR = 01&B1 = Population&Custom. Accessed 7/13/2009.
Stratus Consulting Inc., 2009. Review of Urban Visibility Public Preference Studies: Final Report. Prepared for EPA
Office of Air Quality Planning and Standards; funded under EPA Contract No. EP-D-08-100 with Abt
Associates Inc., Bethesda, MD. September. Available:
US EPA (1996a). Air Quality Criteria for Particulate Matter. Research Triangle Park, NC: National Center for
Environmental Assessment-RTF Office; report no. EPA/600/P-95/001aF-cF. 3v. Available:
http: //www. epa. gov/ttn/naaqs/standards/pm/s_pm_pr_cd. html.
US EPA (1996b). Review of the National Ambient Air Quality Standards for Particulate Matter: Policy Assessment
of Scientific and Technical Information, OAQPS Staff Paper. Research Triangle Park, NC 27711: Office
of Air Quality Planning and Standards; report no. EPA-452/R-96-013. Available:
http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_pr sp.html.
US EPA (2004). Air Quality Criteria for Particulate Matter. National Center for Environmental Assessment, Office
of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711;
report no. EPA/600/P-99/002aF and EPA/600/P-99/002bF. October 2004. Available:
http://www.epa.gov/ttn/naaqs/standards/pm/s_pm cr cd.html
US EPA (2005). Review of the National Ambient Air Quality Standards for Particulate Matter: Policy Assessment
of Scientific and Technical Information, OAQPS Staff Paper. Research Triangle Park, NC 27711: Office
of Air Quality Planning and Standards; report no. EPA EPA-452/R-05-005a. December 2005. Available:
http://www.epa.gOv/ttn/naaas/standards/pm/s pm cr sp.html
September 2009 5-2 DRAFT - Do Note Quote or Cite
-------
U.S. Environmental Protection Agency. (2007). Draft Integrated Review Plan for the National Ambient Air Quality
Standards for Paniculate Matter. October 2007. U.S. Environmental Protection Agency, Research
Triangle Park, NC, EPA 452/P-08-006. Available at:
http://www.epa.gov/ttn/naaqs/standards/pm/s_pm 2007_pd.html
US EPA (2008a). Integrated Review Plan for the National Ambient Air Quality Standards for Paniculate Matter.
National Center for Environmental Assessment and Office of Air Quality Planning and Standards, U.S.
Environmental Protection Agency, Research Triangle Park, NC. Report No. EPA 452/R-08-004. March
2008. Available at:
http://www.epa. gov/ttn/naaqs/standards/pm/data/2008_03_final_inte grated_review_plan.pdf
U.S. EPA (2008b). Integrated Science Assessment for Paniculate Matter: First External Review Draft. National
Center for Environmental Assessment-RTF Division, Office of Air Quality Planning and Standards,
Research Triangle Park, NC. EPA/600/R-08/139 and 139A. December 2008. Available:
http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_isa.html.
US EPA (2009a). Integrated Science Assessment for Paniculate Matter: Second External Review Draft. National
Center for Environmental Assessment-RTF Division, Office of Air Quality Planning and Standards,
Research Triangle Park, NC. EPA/600/R-08/139B. July 2009. Available:
http://www.epa.gOv/ttn/naaqs/standards/pm/s jm_2007_isa. html.
US EPA (2009b). Paniculate Matter National Ambient Air Quality Standards: Scope and Methods Plan for Urban
Visibility Assessment. Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency, Research Triangle Park, NC. EPA-452/P-09-001. February 2009. Available:
http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_pd.html.
September 2009 5-3 DRAFT - Do Note Quote or Cite
-------
APPENDICES
A. PM2.s Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of
Total Light Extinction in the 15 Study Areas
B. Distributions of Estimated PMi.s Components under Current Conditions
C. Development of PRB Estimates of PM2.s components, PMi0-2.s, and Total Light
Extinction
D. Relationships between PM Mass Concentration and PM Light Extinction under Current
Conditions
E. Differences in Daily Patterns of Relative Humidity and Light Extinction between Areas
and Seasons
F. Distributions of Maximum Daily Daylight Total Light Extinction under "Just Meets"
Conditions
-------
APPENDIX A - PM2 5 MONITORING SITES AND MONITORS
PROVIDING 2005-2007 DATA FOR THE ANALYSIS OF TOTAL
LIGHT EXTINCTION IN THE 15 STUDY AREAS
September 2009 A-l
DRAFT Do Not Quote or Cite
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 1 5 Study Areas
Study Area
Tacoma
First PM2.5 Monitoring Site
AQS ID 530530029
State: Washington
City: Tacoma
MSA: Tacoma, WA
Local Site Name: TACOMA - L STREET
Address: 7802 SOUTH L STREET,
TACOMA
0.5 miles east of 1-5
2005-2007 annual DV = 10.2
2005-2007 24-hr DV = 43
This is the highest 24-hour PM2.s DV site in
the Seattle-Tacoma-Olympia, WA annual
PM2.5 nonattainment area
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 881 01 ; one-in-three sampling
schedule)
• PM2.5 speciation (one-in-six
sampling schedule)
• 1-hour PM2.5 mass (AQS
parameter 88502, Acceptable PM2.5 AQI
& Speciation Mass) Correlated Radiance
Research M903 Nephelometry
No continuous PM1 0 monitoring at this
site, see right hand column..
Second PM2.5 Monitoring Site (if
applicable)
NA
PM10 data source for PM10.2.s
AOS ID 530530031
State: Washington
City: Tacoma
MSA: Tacoma, WA
Local Site Name: TACOMA - ALEXANDER AVE
Address: 2301 ALEXANDER AVE, TACOMA, WA
6.4 miles NNE of PM2.5 site
Neighborhood Scale
Parameters taken from this site:
. 1-hour PM10 STP mass (AQS parameter 81 102)
o Sample Collection Method:
INSTRUMENTAL-R&P SA246B-INLET
o Sample Analysis Method: TEOM-
GRAVIMETRIC
7% of PM10-2.5 values were determined using
regional average PM10-2.5:PM2.5 ratios from 2005
Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-2
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Fresno
AQS ID 060190008
State: California
City: Fresno
MSA: Fresno, CA
Local Site Name: None given
Address: 3425 N FIRST ST, FRESNO
2.5 miles west of the airport, 3 miles NNE
of central Fresno
2005-2007 annual DV = 17.4
2005-2007 24-hr DV = 63
This is not the highest annual or 24-hr
PM2.5 DV site in the San Joaquin
nonattainment area.
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; every day sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
. 1-hour PM2.5 mass (AQS
parameter 88501, PM2.5 Raw Data) Met-
One BAM
No continuous PM10 monitoring at this
site, see right hand column..
NA
PM10-2.5 values were determined using regional
average PM10-2.5:PM2.5 ratios from 2005 Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-2
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 1 5 Study Areas
Study Area
Los Angeles-
South Coast Air
Basin
First PM2.s Monitoring Site
AQS ID 060658001
State: California
City: Rubidoux (West Riverside)
MSA: Riverside-San Bernardino, CA
Local Site Name: None given
Address: 5888 MISSION BLVD.,
RUBIDOUX
Eastern SCAB, 0.4 miles from Pomona
Freeway.
2005-2007 annual DV = 19.6
2005-2007 24-hr DV = 55
This site is not the highest DV site in the
LA-South Coast nonattainment area.
Neighborhood scale.
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 881 01; every day sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
. 1 -hour PM2.5 (AQS parameter
88502, Acceptable PM2.5 AQI &
Speciation Mass) [still investigating
instrument type]
No continuous PM1 0 monitoring at this
site, see right hand column..
Second PM2.S Monitoring Site (if
applicable)
NA
PM10 data source for PM10.2.s
AQS ID 060710306
State: California
City: Victorville
MSA: Riverside-San Bernardino, CA
Local Site Name: MOVED FROM 060710014
Address: 14306 PARKAVE., VICTORVILLE, CA
36 miles north of PM2.5 site, on the other side of a range of
hills. 0.4 miles from 1-15
Measurement Scale not given in AQS, but appears
Neighborhood by aerial image.
Parameters taken from this site:
. 1-hour PM10 STP mass (AQS parameter 81 102)
o Sample Collection Method:
INSTRUMENTAL-R&P SA246B-INLET
o Sample Analysis Method: TEOM-
GRAVIMETRIC
6% of PM10-2.5 values were determined using
regional average PM10-2.5:PM2.5 ratios from 2005
rij. pp T-\
Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-4
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Phoenix
AQS ID 040137020 (FRM & CSN)
State: Arizona
City: Scottsdale
MSA: Phoenix-Mesa, AZ
Local Site Name:
Address: 10844 EAST OSBORN ROAD
SCOTTSDALE' AZ
Reporting Agency: Salt River Pima-
Maricopa Indian Community of Salt River
Reservation
Eastern edge of the metro area, largely
surrounded by agricultural fields.
2005-2007 annual DV = 7.9
2005-2007 24-hr DV = 15
This site is not the highest DV site in the
Phoenix-Mesa CBSA.
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; one-in-six sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
No continuous PM10 monitoring at this
site, see right hand column.
AQS ID 040139998 (Continuous)
State: Arizona
City: Phoenix
MSA: Phoenix-Mesa, AZ
Local Site Name: Vehicle Emissions
Laboratory
Address: 600 N 40th St & Fillmore St
Measurement Scale not available; 0.75
miles from intersection of two freeways, 1
mile from Phoenix airport.
Parameters taken from this site:
• 1-hour PM2.5 mass.
Nephelometer.
AQS ID 040133002
State: Arizona
City: Phoenix
MSA: Phoenix-Mesa, AZ
Local Site Name: CENTRAL PHOENIX
Address: 1645 E ROOSEVELT ST-CENTRAL PHOENIX STN
1.8 miles NE of central Phoenix
Neighborhood Scale
Parameters taken from this site:
. 1-hour PM10 STP mass (AQS parameter 81102)
o Sample Collection Method:
INSTRUMENTAL-R&P SA246B-INLET
o Sample Analysis Method: TEOM-
GRAVIMETRIC
2% of PM10-2.5 values were using regional average
PM10-2.5:PM2.5 ratios from 2005 Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-5
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Salt Lake City
AQS ID490353006
State: Utah
City: Salt Lake City
MSA: Salt Lake City-Ogden, UT
Local Site Name: UTM COORDINATES =
PROBE LOCATION
Address: 1675 SOUTH 600 EAST, SALT
LAKE CITY
2.5 miles SSE of central Salt Lake City
2005-2007 annual DV = 10.7
2005-2007 24-hr DV = 48
This is not the highest DV site in the Salt
Lake City CSA.
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; every day sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
. 1-hour PM2.5 mass (AQS
parameter 88501, PM2.5 Raw Data)
FDMS-Gravimetric
No continuous PM10 monitoring at this
site, see right hand column.
NA
PM10-2.5 values were determined using regional
average PM10-2.5:PM2.5 ratios from 2005 Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-6
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Dallas
AQS ID 481130069
State: Texas
City: Dallas
MSA: Dallas, TX
Local Site Name: DALLAS HINTON
Address: 1415 HINTON STREET
4.5 miles NE of central Dallas
2005-2007 annual DV = 11.5
2005-2007 24-hr DV = 25
This is not the highest DV site in the
Dallas-Ft. Worth CSA.
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; every day sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
• 1-hour PM2.5 mass (AQS
parameter 88502, Acceptable PM2.5 AQI
& Speciation Mass) TEOM Gravimetric 50
deg C
No continuous PM10 monitoring at this
site, see right hand column..
NA
PM10-2.5 values were determined using regional
average PM10-2.5:PM2.5 ratios from 2005 Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-7
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Houston
AQS ID 482010024
State: Texas
City: Not in a city
MSA: Houston, TX
Local Site Name: HOUSTON ALDINE
Address: 4510 1/2 ALDINE MAIL RD
10 miles NNE of central Houston
2005-2007 annual DV = 13.1
2005-2007 24-hr DV = 25
This is not the highest DV site in the
'Houston-Baytown-Huntsville, TX CSA.
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; one-in-six day sampling
schedule)
• PM2.5 speciation (one-in-six
sampling schedule)
• 1-hour PM2.5 mass (AQS
parameter 88502, Acceptable PM2.5 AQI
& Speciation Mass) TEOM Gravimetric 50
deg C
No continuous PM10 monitoring at this
site, see right hand column.
NA
PM10-2.5 values were determined using regional
average PM10-2.5:PM2.5 ratios from 2005 Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-8
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 1 5 Study Areas
Study Area
St. Louis
First PM2.s Monitoring Site
AQS ID 2951 00085
State: Missouri
City: St. Louis
MSA: St, Louis, MO-IL
Local Site Name: BLAIR STREET
CATEGORY A CORE SLAM PM2.5.
Address: BLAIR S
2 miles north of central St. Louis
2005-2007 annual DV = 14.5
2005-2007 24-hr DV = 34
This is not the highest DV site in the St.
Louis nonattainment area.
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 881 01; every day sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
. 1 -hour PM2.5 mass (AQS
parameter 88502, Acceptable PM2.5 AQI
& Speciation Mass) TEOM Gravimetric 30
deg C
No continuous PM1 0 monitoring at this
site, see right hand column.
Second PM2.S Monitoring Site (if
applicable)
NA
PM10 data source for PM10.2.s
AQS ID 295100092 (2005 and 2006 data)
State: Missouri
City: St. Louis
MSA: St, Louis, MO-IL
Local Site Name:
Address: 3 NORTH MARKET
0.7 miles ESE of PM2.5 site, across the street from the
eastern edge of what appears to be a recycling/municipal
works yard.
Middle Scale
Parameters taken from this site:
. 1-hour PM10 STP mass (AQS parameter 81 102)
o Sample Collection Method:
INSTRUMENTAL-R&P SA246B-INLET
o Sample Analysis Method: TEOM-
GRAVIMETRIC Site was on the other (western) side of the
recycling/municipal works yard as site 295100093, below.
295100093 (2007 data)
State: Missouri
City: St. Louis
MSA: St, Louis, MO-IL
Local Site Name: None given
Address: Branch Street
0.6 miles ESE of PM2.5 site, across the street from the
western edge of what appears to be a recycling/municipal
works yard.
Middle Scale
Parameters taken from this site:
. 1-hour PM10 STP mass (AQS parameter 81 102)
o Sample Collection Method:
INSTRUMENTAL-R&P SA246B-INLET
o Sample Analysis Method: TEOM-
GRAVIMETRIC
4% of PM10-2.5 values were determined using
regional average PM10-2.5:PM2.5 ratios from 2005
Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-9
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Birmingham
AQS ID 010730023
State: Alabama
City: Birmingham
MSA: Birmingham, AL
Local Site Name:
Address: NO. B'HAM.SOU R.R., 3009
28TH ST. NO
2.3 miles north of central Birmingham
2005-2007 annual DV = 18.7
2005-2007 24-hr DV = 44
This is the highest DV site in the
Birmingham nonattainment area
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; every day sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
. 1-hour PM2.5 mass (AQS
parameter 88502, Acceptable PM2.5 AQI
& Speciation Mass) TEOM Gravimetric 50
deg C
. 1-hour PM10STP mass (AQS
parameter 81102)
o Sample Collection
Method: INSTRUMENTAL-R&P SA246B-
INLET
o Sample Analysis
Method: TEOM-GRAVIMETRIC
NA
Same as PM2.5 site.
0.3% of PM10-2.5 values were determined using
regional average PM10-2.5:PM2.5 ratios from 2005
Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-10
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Atlanta
AQS ID 130890002
State: Georgia
City: Decatur
MSA: Atlanta, GA
Local Site Name: 2390-B WILDCAT
ROAD, DECATUR, GA
Address: SOUTH DEKALB
About 7 miles SE of central Atlanta
2005-2007 annual DV = 15.7
2005-2007 24-hr DV = 33
This is not the highest DV site in the
Atlanta nonattainment area.
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; every day sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
. 1-hour PM2.5 mass (AQS
parameter 88502, Acceptable PM2.5 AQI
& Speciation Mass) TEOM Gravimetric 30
deg C
No continuous PM10 monitoring at this
site, see right hand column.
NA
AQS ID 131210048
State: Georgia
City: Atlanta
MSA: Atlanta, GA
Local Site Name: Georgia Tech, Ford Environmental Science
and Technology Bldg, roof
Address: GA. TECH., Ford ES&T Bldg, 311 Ferst St NW,
Atlanta GA
8.6 miles NW of PM2.5 site
Neighborhood Scale
Parameters taken from this site:
. 1-hour PM10STP mass (AQS parameter 81102)
o Sample Collection Method:
INSTRUMENT MET ONE 4 MODELS
o Sample Analysis Method: BETA
ATTENUATION
8% of PM10-2.5 values were determined using
regional average PM10-2.5:PM2.5 ratios from 2005
Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-ll
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Detroit-Ann
Arbor
AQS ID 261630033
State: Michigan
City: Dearborn
MSA: Detroit, Ml
Local Site Name: PROPERTY OWNED
BY DEARBORN PUBLIC SCHOOLS
Address: 2842 WYOMING
About 0.2 miles from Ford River Rouge
auto plant
2005-2007 annual DV = 17.2
2005-2007 24-hr DV = 43
This is the highest annual and 24-hr DV
site in the Detroit nonattainment area
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; every day sampling
schedule)
• PM2.5 speciation (one-in-six
sampling schedule)
. 1-hour PM2.5 mass (AQS
parameter 88501, PM2.5 Raw Data)
TEOM Gravimetric 50 deg C
. 1-hour PM10STP mass (AQS
parameter 81102)
o Sample Collection
Method: INSTRUMENTAL-R&P SA246B-
INLET
O Sample Analysis
Method: TEOM-GRAVIMETRIC
NA
Same as PM2.5 site.
2% of PM10-2.5 values were determined using
regional average PM10-2.5:PM2.5 ratios from 2005
Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-12
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Pittsburgh
AQS ID 420030008
State: Pennsylvania
City: Pittsburgh
MSA: Pittsburgh, PA
Local Site Name: None given
Address: BAPC 301 39TH STREET BLDG
#7
3 miles NE of central Pittsburgh, 0.5 miles
from Allegheny River
2005-2007 annual DV = 15.0
2005-2007 24-hr DV = 40
This site is not the highest DV site in the
Pittsburgh nonattainment area.
Urban Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; every day sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
. 1-hour PM2.5 mass (AQS
parameter 88502, Acceptable PM2.5 AQI
& Speciation Mass) TEOM Gravimetric 50
deg C
No continuous PM10 monitoring at this
site, see right hand column.
NA
PM10-2.5 values were determined using regional
average PM10-2.5:PM2.5 ratios from 2005 Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-13
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Baltimore
AQS ID 240053001 (FRM & CSN)
State: Maryland
City: Essex
MSA: Baltimore, MD
Local Site Name: Essex
Address: 600 Dorsey Avenue
7 miles east of central Baltimore
2005-2007 annual DV = 14.5
2005-2007 24-hr DV = 35
This is not the highest DV site in the
Baltimore nonattainment area.
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; every day sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
• 1-hour PM10 LC mass (AQS
parameter 85101
AQS ID 245100040 (Continuous)
State: Maryland
City: Baltimore
MSA: Baltimore, MD
Local Site Name: Oldtown
Address: Oldtown Fire Station, 1100 Hillen
Street
1 mile NNE of Inner Harbor area
Middle Scale
Parameters taken from this site:
. 1-hourPM2.5 mass (AQS
parameter 88502, Acceptable PM2.5 AQI
& Speciation Mass) TEOM Gravimetric 50
deg C
Same as PM2.5 site.
5% of PM10-2.5 values were determined using
regional average PM10-2.5:PM2.5 ratios from 2005
Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-14
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
Philadelphia-
Wilmington
AQS ID100032004 (DE)
State: Delaware
City: Wilmington
MSA: Wilmington-Newark, DE-MD
Local Site Name: CORNER OF MLK
BLVDANDJUSTISONST
2.5 miles NE of central Wilimington, 0.25
miles from the Delaware River, 22 miles
SWfrom central Philadelphia
2005-2007 annual DV = 14.7
2005-2007 24-hr DV = 37
This is not the highest DV site in the
Philadelphia nonattainment area
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; every day sampling
schedule)
• PM2.5 speciation (one-in-six
sampling schedule)
. 1-hour PM2.5 mass (AQS
parameter 88501, PM2.5 Raw Data) Beta
Attenuation
. 1-hour PM10STP mass (AQS
parameter 81102)
NA
Same as PM2.5 site.
3% of PM10-2.5 values were determined using
regional average PM10-2.5:PM2.5 ratios from 2005
Staff Paper
September 2009
DRAFT Do Not Quote or Cite
A-15
-------
PM2.5 Monitoring Sites and Monitors Providing 2005-2007 Data for the Analysis of Total Light
Extinction in the 15 Study Areas
Study Area
First PM2.s Monitoring Site
Second PM2.S Monitoring Site (if
applicable)
PM10 data source for PM10.2.s
New York-
N.New Jersey-
Long Island
AQS ID 340390004 (NJ)
State: New Jersey
City: Elizabeth
MSA: Newark, NJ
Local Site Name: ELIZABETH LAB
Address: NEW JERSEY TURNPIKE
INTERCHANGE 13
1.75 miles south of Elizabeth, at the I-95
interchange with I-278
2005-2007 annual DV = 14.4
2005-2007 24-hr DV = 42
This is not the highest DV site in the New
York nonattainment area
Neighborhood Scale
Parameters taken from this site:
. 24-hour FRM PM2.5 mass (AQS
parameter 88101; every day sampling
schedule)
• PM2.5 speciation (one-in-three
sampling schedule)
. 1-hour PM2.5 mass (AQS
parameter 88502, Acceptable PM2.5 AQI
& Speciation Mass) TEOM Gravimetric 30
deg C
No continuous PM10 monitoring at this
site, see right hand column.
NA
AQS ID 360610125
State: New York
City: New York
MSA: New York, NY
Local Site Name: PARK ROW
Address: 1 PACE PLAZA
Near the on-ramp to the Brooklyn Bridge, Manhattan end
Measurement scale not stated.
Parameters taken from this site:
. 1-hour PM10 STP mass (AQS parameter 81102)
o Sample Collection Method:
INSTRUMENTAL-R&P SA246B-INLET
o Sample Analysis Method: TEOM-
GRAVIMETRIC
2% of PM10-2.5 values were determined using
regional average PM10-2.5:PM2.5 ratios from 2005
Staff Paper
Notes:
• In this Table, the 1-hour concentration parameter "88502, Acceptable PM2 5 AQI & Speciation Mass" is the same as the ISA refers to as "FRM-like"
PM25mass. An entry of "88501, PM25 Raw Data" indicates that the monitoring agency makes no representation as to the degree of correlation with
FRM PM2 5 mass. The latter type of continuous PM2 5 data were used only when the former were unavailable.
• Where PM10 was reported in STP, it was converted to LC before PM10.2.5 was calculated.
• For convenience, continuous PM2 5 data were obtained through the AirNow website rather than from AQS, as an initial exploration indicated that not all
the desired 1-hour data had been submitted to AQS.
September 2009
DRAFT Do Not Quote or Cite
A-16
-------
APPENDIX B - DISTRIBUTIONS OF ESTIMATED PM2 5
COMPONENTS
September 2009 B-l
DRAFT Do Not Quote or Cite
-------
Figure B-l - Distribution of daily maximum PMi.s, PMio-i.s, and relative humidity across the 2005-2007 period, by study area
(a) Daily maximum daylight PM2.5
Daily Maximum PM2.5 (Daylight Hours)
109 304 288
300 263 143 274 346 276 131 268 187 142 213
-r T T
-r
September 2009
DRAFT Do Not Quote or Cite
B-2
-------
(b) Daily maximum daylight PMio-2.5
Daily Maximum Coarse (Daylight Hours)
o
O
o
o
rj-
109 304 288 86 300 263 143 274 346 276 131 268
142 218
September 2009
DRAFT Do Not Quote or Cite
B-3
-------
(c) Daily maximum daylight relative humidity
Daily Maximum Rel Hum (Daylight Hours)
o
o
s
0>
a.
^T o
£• CD
2
D
I
0)
o
324
O
302
O
306
O
274
O
149
O
295
O
141
O
284
O
v
<«P ti«
,0'
^ ^
-------
Figure B-2 - Distribution of hourly PM2.s components across the 2005-2007 period, by study area
(a) 1-hour daylight sulfate (dry, fully neutralized)
Sulfate hourly (Daylight Hours)
CO
o
O
1238 3643 3383
3457 3106 1652 3273 3930 3262 1567 3179 2095 1618 2515
T
September 2009
DRAFT Do Not Quote or Cite
B-5
-------
Figure B-2 - Distribution of PM2.s components across the 2005-2007 period, by study area, continued
(b) 1-hour daylight nitrate (dry, fully neutralized, CSN method consistent)
Nitrate hourly (Daylight Hours)
CO
~
o
a
1
3383
O
,
-------
Figure B-2 - Distribution of PM2.s components across the 2005-2007 period, by study area, continued
(c) 1-hour daylight elemental carbon
Elemental Carbon hourly (Daylight Hours)
o
O
1238 3643 3383
3457 3106 1652 3273 3930 3262 1567 3179 2095 1618 2515
\\
#~ -«PN
-------
Figure B-2 - Distribution of PM2.s components across the 2005-2007 period, by study area, continued
(d) 1-hour daylight organic carbonaceous material
Organic Carbon hourly (Daylight Hours)
CO
o
1238 3643
3457 3106 1652 3273 3930 3262 1567 3179
1618 2515
,
-------
Figure B-2 - Distribution of PM2.s components across the 2005-2007 period, by study area, continued
(e) 1-hour daylight fine soil
Soil hourly (Daylight Hours)
1238 3643 3383 988 3457 3106 1652 3273 3930 3262 1567 3179 2095 1618 2515
S °
O O
1
i
8 1
il
\\
#~ .*&•
&? ,,v'
September 2009
DRAFT Do Not Quote or Cite
B-9
-------
APPENDIX C - DEVELOPMENT OF PRB ESTIMATES OF PM2 5
COMPONENTS, PM10_2 5, AND TOTAL LIGHT EXTINCTION
Policy relevant background levels of total light extinction have been estimated for this
assessment by relying on outputs for the 2004 CMAQ run in which anthropogenic emission
in the U.S., Canada, and Mexico were omitted, as described in the second draft ISA.
Estimates of PRB for total light extinction were calculated from modeled concentrations of
PM2.5 components using the IMPROVE algorithm. The necessary component concentrations
were extracted from the CMAQ output files, as they were not summarized in the second draft
ISA.
More specifically, for each study area, EPA staff overlaid CMAQ grid cells over
shapes representing the Census-defined urbanized area for each study area, and visually
identified the CMAQ grid cells that had a substantial portion of their area coincident with the
urbanized area. For each such grid cell, for each of the 12 months of the year, we obtained
the 24 values of the hour-specific average concentrations of the five PM2.5 components. We
then averaged these across the selected grid cells. Thus, a given hour of the day has the same
PRB estimate for a component on all days within a month, but months and study areas differ.
Table C-l summarizes these PRB estimates for the PM2.s components (including the specific
form assumed for sulfate, nitrate, and organic carbon). The most notable observed feature of
the PRB estimates is relatively high values for elemental and organic carbon PRB for the
Tacoma study area. This area is often affected by wildfires for extended periods in the
autumn months, and such fires were included in the 2004 emissions scenario for the PRB
CMAQ run. A cursory review of information on fire events in 2005-2007 confirmed that the
fire situation in this part of the country in 2004 was not an anomaly.
September 2009 C-l
DRAFT - Do Not Quote or Cite
-------
Table C-l. Summary of PRB estimates for the five PMi.s components: average 1-hour
values across 2005-2007
Study Area
Tacoma
Fresno
Los Angeles-South
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann Arbor
Pittsburgh
Baltimore
Philadelphia-
New York-N.New
Jersey-Long Island
Average 1-Hour PRB Concentration Across 2005-2007 (ug/m3)
Sulfate
(dry, no
ammonium)
0.45
0.4
0.36
0.31
0.25
0.27
0.3
0.31
0.29
0.3
0.34
0.3
0.34
0.34
0.36
Nitrate
(dry, no
ammonium)
0.026
0.00062
0.0037
0.000052
0.00028
0.0022
0.0055
0.0027
0.007
0.016
0.00062
0.00052
0.0016
0.00097
0.0038
Elemental Carbon
0.15
0.08
0.028
0.02
0.025
0.055
0.091
0.047
0.099
0.1
0.024
0.029
0.039
0.03
0.026
Organic
Carbonaceous
Material
1.3
0.74
0.3
0.26
0.26
0.59
0.86
0.53
1.1
1.1
0.32
0.36
0.44
0.36
0.31
Fine
Soil/Crustal
0.31
0.19
0.036
0.015
0.034
0.092
0.17
0.07
0.19
0.19
0.018
0.034
0.054
0.032
0.022
September 2009
DRAFT - Do Not Quote or Cite
C-2
-------
It is also necessary to have estimates of PRB for PMio-2.5, to feed into the IMPROVE
algorithm. The second draft ISA for this review does not present any new information on
this subject. The approach used in the previous two Criteria Documents was to present the
historical range of annual means of PMio-2.5 concentrations from IMPROVE monitoring sites
selected as being least influenced by anthropogenic emissions. See Table 3E-1 of the 2004
Criteria Document. For sites in the lower 48 states, these annual means range from a low of
1.8 ug/m3 to a high of 10.8 ug/m3. For this assessment, EPA staff estimated PRB for PMio-2.5
using a contour map based on average 2000-2004 PMio-2.5 concentrations from all
IMPROVE monitoring sites, found in a recent report from the IMPROVE program. (Spatial
and Seasonal Patterns and Temporal Variability of Haze and its Constituents in the United
States: Report IV, November 2006). We located each study area's position on this map, and
assigned it the mid-point of the range of concentrations indicated by the contour band for that
location. The contour map is reproduced here as Figure C-l. Stars show locations of the 15
study areas. In this reproduction, the midpoints of the contour ranges have been added to the
legend.
The results are shown in Table C-2. Lacking any other information, these PRB
values are taken to apply to every hour of the year. While the contour map and thus these
values are influenced by data from IMPROVE sites that were not considered in the 2004
Criteria Document to be the sites most isolated from the influence of anthropogenic
emissions, including three IMPROVE sites in urban areas, these values are generally within
the range of values presented in the Criteria Document for such isolated sites. Further, these
PRB values are low enough that their exact values will have little effect on the results of
"what if estimation of total light extinction levels under possible secondary PM NAAQS.
Table C-3 presents the resulting 2005-2007 average PRB daylight total light
extinction by study area, determined by using each daylight hour's f(RH), the hour-specific
PRB PM2.s component estimates summarized as annual averages in Table C-l, the PRB
PMio-2.5 estimates in Table C-2, and the IMPROVE algorithm. The sulfate and nitrate
component values in Table C-l are multiplied by 1.375 and 1.29 to reflect full neutralization,
before being used in the IMPROVE algorithm.
September 2009 C-2
DRAFT - Do Not Quote or Cite
-------
Figure C-l. Selection of PRB values for PM10.2.5 based on contoured IMPROVE
monitoring data
9.5
8.5
7.5
6.5
5.5
4.5
3.5
2.5
1.5
0.5
22.1
8.92
7.93
6.94
5.95
4.96
3.96
2.97
1.98
0.99
0.00
(jg/m3
I
IMPROVE Site
IMPROVE Urban Site
5>*
Puerto Rico /
Virgin Islands
September 2009
DRAFT - Do Not Quote or Cite
C-4
-------
Table C-2. Policy Relevant Background Concentrations of PMi0-2.s Used in This
Assessment, Based on Measurements at IMPROVE Sites
Study Area
Tacoma
Fresno
Los Angeles-South Coast Air Basin
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann Arbor
Pittsburgh
Baltimore
Philadelphia- Wilmington
New York-N.New Jersey -Long Island
PRB PMio-2.sMass (ug/m3)
4.5
5.5
4.5
5.5
4.5
8.5
5.5
7.5
5.5
5.5
9.5
3.5
3.5
6.5
3.5
Table C-3. 2005-2007 Average Policy Relevant Background Daylight Total light
Extinction
Study Area
Tacoma
Fresno
Los Angeles-South Coast Air Basin
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann Arbor
Pittsburgh
Baltimore
Philadelphia- Wilmington
New York-N.New Jersey -Long Island
2005-2007 Average Policy Relevant
Background Daylight Total Light Extinction,
Mm1
22
21
18
18
15
18
20
19
19
19
17
17
19
18
18
September 2009
DRAFT - Do Not Quote or Cite
C-5
-------
APPENDIX D RELATIONSHIPS BETWEEN PM MASS
CONCENTRATION AND TOTAL LIGHT EXTINCTION UNDER
CURRENT CONDITIONS
In the last review, the 2005 Staff Paper examined the correlation between total light
extinction and PM2.5 mass concentrations, each defined for various consistent time periods.
The 2005 Staff Paper analysis assumed that the percentage mix of PM2.5 components was the
same in all 24 hours of each day, equal to that indicated by 24-hour CSN sampling. The
modeling of 1-hour total light extinction in this new assessment allows these correlations to
be re-examined, with the more realistic treatment in which the mix of PM2.5 components is
modeled to vary during the day, based in part of diurnal profiles from CMAQ modeling (see
section 3.2.4).
Four figures are presented here, using different time periods for PM2.5 mass
concentrations and total light extinction, not always matching. In each figure, the solid red
line represents a LOESS fit (a form of locally weighted polynomial regression, see
http://support.sas.com/rnd/app/papers/loesssugi.pdf) to the data. Table D-l presents squared
correlation coefficients between observed and LOESS model-predicted values for all four
figures.
Figure D-l compares hourly PM2.5 mass (as actually measured by the continuous
instruments) vs. same-hour daylight total light extinction. As the 2005 Staff Paper explained,
the scatter is due to variations in the mix of PM2.5 components and in relative humidity across
hours. In addition, continuous PM2.5 mass instruments do not register the mass of each
component consistently with CSN samplers and analysis, which affects the scatter in this
figure because the estimates of light extinction are ground truthed to the CSN measurements
more strongly than to the continuous PM2.5 measurements.
Figure D-2 compares 12-4 pm average PM2 5 mass vs. 12-4 pm average total light
extinction. Because this time period is generally the time of lowest relative humidity, in most
study areas and on average as indicated by the squared correlation coefficients, the scatter in
Figure D-2 is less than in Figure D-l.
Figure D-3 compares 12-4 pm average PM2 5 mass vs. daily maximum daylight 1-
hour total light extinction. The scatter in Figure D-3 is typically more than in Figure D-2,
because daily maximum daylight 1-hour total light extinction often occurs earlier in the day
than the 12-4 pm period used to average PM2.5 mass, when relative humidity is higher.
September 2009 D-l
DRAFT - Do Not Quote or Cite
-------
Figure D-4 compares 8 am-12 pm average PM2.5 mass vs. daily maximum daylight 1-
hour total light extinction. The scatter in Figure D-4 is typically less than in Figure D-3 and the
squared correlation coefficients larger, because this earlier averaging period for PM2 5 mass more
often encompasses the period of maximum total light extinction.
Table D-l. Squared correlation coefficients between observed and LOESS
model-predicted values of total light extinction
Area
Tacoma
Fresno
Los Angeles-
South Coast Air
Basin
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann
Arbor
Pittsburgh
Baltimore
Philadelphia-
Wilmington
New York-
N.New Jersey-
Long Island
AVERAGE
Figure D-l
1-hour PM2.s
mass vs. same-
hour total light
extinction
0.82
0.79
0.58
0.63
0.86
0.51
0.52
0.44
0.63
0.62
0.49
0.47
0.43
0.43
0.68
0.59
Figure D-2
12-4 pm average
PMi.s mass vs.
12-4 pm average
total light
extinction
0.80
0.89
0.65
0.66
0.94
0.50
0.57
0.27
0.60
0.74
0.58
0.51
0.48
0.36
0.82
0.62
Figure D-3
12-4 pm average
PMi.s mass vs.
daily maximum
daylight 1-hour
total light
extinction
0.36
0.60
0.35
0.16
0.70
0.15
0.20
0.25
0.25
0.35
0.11
0.39
0.47
0.16
0.49
0.33
Figure D-4
8 am-12pm
average PM2.5
mass vs. daily
maximum
daylight 1-hour
total light
extinction
0.73
0.69
0.46
0.17
0.80
0.28
0.31
0.41
0.33
0.58
0.31
0.41
0.49
0.25
0.52
0.45
September 2009
DRAFT - Do Not Quote or Cite
D-2
-------
Figure D-l. - Relationship between 1-hour PM2.s mass vs. same-hour total light extinction.
0 50 100 150 200
_J I | | I I
0 50 100 150 200
Phoenix. AZ
Pittsburgh. PA
Salt Lake City, UT
St. Louis, II
Tacoma, WA
1000 -
500 -
0 -
Fresno, CA
Houston, TX
Los Angeles, CA
New York. NY
Philadelphia. PA
LU
£
-1000
- 500
- 0
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit,
1000 -
500 -
0 -
\ I T
0 50 100 150 200
0 50 100 150 200
PM2.5 (ug/m3)
50 100 150 200
September 2009
DRAFT - Do Not Quote or Cite
-------
Figure D-2. Relationship between 12-4 pm average PM2.s mass vs. 12-4 pm average total light extinction.
50 100
_] | I
50 100
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, IL
Tacoma, WA
1000 -
400 -
200 -
0 -
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
re
^c
"x
LLJ
£
•2> 1000
800
600
400
200
0
-1000
- 800
- 600
- 400
- 200
- 0
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, I
50 100
50 100
Pm2.5 12-4 average
i
50
100
September 2009
DRAFT - Do Not Quote or Cite
D-4
-------
Figure D-3. Relationship between 12-4 pm average PM2.s mass vs. daily maximum daylight 1-hour total light
extinction.
50
100
I
50
100
I
Phoenix, AZ
Pittsburgh, PA
Salt Lake City. UT
St. Louis, IL
Tacoma, WA
1200 -
1000
800
600
400
200 -
••..V
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
c
o
SI
Ul
- 1200
- 1000
- 800
- 600
- 400
- 200
- 0
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas. TX
Detroit,
1200
1000
800
600
400
200
0
..•*•'•
.'•. .s ••:
"A ~ 4/
50
100
0 50 100
Pm2.5 12-4 average
50
100
September 2009
DRAFT - Do Not Quote or Cite
D-5
-------
Figure D-4. Relationship between 8 am-12pm average PM2.s mass vs. daily maximum daylight 1-hour total light extinction
50 100
50 100
Phoenix, AZ
Pittsburgh. PA
Salt Lake City. UT
St. Louis. II
Tacoma, WA
1200
1000
800
600
400
200
0
•*•*.
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia. PA
c.
o
LU
£
- 1200
- 1000
- 800
- 600
- 400
- 200
- 0
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas. TX
Detroit, I
1200
1000
800
600 -
400 -
200 -
0 -
50 100
50 100
Pm2.5 8-12 average
50 100
September 2009
DRAFT - Do Not Quote or Cite
D-6
-------
APPENDIX E - DIFFERENCES IN DAILY PATTERNS OF
RELATIVE HUMIDITY AND TOTAL LIGHT EXTINCTION
BETWEEN AREAS AND SEASONS
In the last review of the secondary PM NAAQS, the pattern of total light extinction
during the day was of particular interest. It was noted, using estimates of hourly total light
extinction based on a simpler approach than described for this analysis, that both (1) mid-day
total light extinction and (2) the slope of the relationship between total light extinction and
PM2.5 concentration varied less among regions of the country that at other times of the day.
This was attributed to greater homogeneity of relative humidity across regions in the mid-day
period. This is in contrast to the situation in the morning and later afternoon hours, when
more eastern areas typically experience higher relative humidity levels than the more arid
western and southwestern areas. The current analysis allows these patterns to be re-
examined.
Figures E-l through E-4 show the diurnal pattern of season-average, hour-specific
total light extinction and relative humidity for the four "daylight seasons." Light extinction
and relative humidity for a given clock hour are averaged across the days in the season,
across all three years. Daylight hours (per the simplified schedule of Table 3-5) are indicated
by solid circles. Average 1-hour total light extinction generally is highest in the morning,
corresponding to higher relatively humidity (mostly due to lower temperature), higher
vehicle traffic, and less dispersive conditions than later in the day. As was observed in the
last review, there is more variation in average 1-hour total light extinction among areas in the
morning than at mid-day.
September 2009 E-l
DRAFT - Do Not Quote or Cite
-------
Figure E-l. Diurnal and seasonal patterns of relative humidity (percent) and total light extinction (Mm"1) for 2005-2007
(a) November-January
i-
y 3 -
8 ->.
(
I-
m S -
8 -:
(
1-1
aJ
2 8 -
S -
<
p^n.
ivr-rf^i (^T^
B^^p-p*1
6 12 17 2
Hour
Fresno, CA
H >
V
6 12 17 2
Hour
Atlanta, GA
tftfCCO» _^
• cP^
W
V^p*
6 12 17 2
Hour
- R Si
- s a -
r S S -:
3 <
-8 a _
1—
r R a -
- S 8-
r 8 S! -
3 <
- 1 8-1
L R 8 J
- S § H,
-8 S -
3 1
CflHI«^HZOH>
^^^i^^
-ff»fc.
St^cP*0*'
) 6 12 17 2
Hour
Houston, TX
aafgt>\ ^
\.o
Qii*^**,^*0^
V-^
) 6 12 17 2
Hour
Baltimore, MD
tf*1*00* -ff**03
CAWI\L^w
^*
) 6 12 17 2
Hour
r R a-
- S I -
' UJ g
3 C
L
L R a-
- s l-i
_ tn "2 —
3 C
- i §n
- s gj
- S 8 -
|
~ SO SG ~
3 C
^ ^
b. ^ /"
6 12 17 2
Hour
.os Angeles, C/>
• rf^
^%£^
6 12 17 2
Hour
Birmingham, AL
"""Xx
*• J^30
m^
B 12 17 2
Hour
-R ^
- S 8 -
r 8 8 -
3 (
L
- § 8-.
-a 8-
3 I
~ S ~
"- R a -
- S 8 -
-8 S J
3 (
>«iaillX.
\*o°
6 12 17 2
Hour
-2 8 -
L R a -
- s 1-
10 8
3 (
-! l~*
t S 8-
3 (
- § I'
i- |£> S J
- S 8 -,
u S S -
3 (
raoxt»Vlw.tf(ww
a
• (J'
) 6 12 17 2
Hour
Philadelphia, Pfi
^-^^
^^^ rA?
^W'Ul
) 6 12 17 2
Hour
Detroit, Ml
•"^N^^cftfro
^^V^rfW
) 6 12 17 2
Hour
- R
- S £
8
3
i
1
-si • RH
r 8
• Ext
3
r §
^ R
r S i
8
3
September 2009
DRAFT - Do Not Quote or Cite
E-2
-------
Figure E-2. Diurnal and seasonal patterns of relative humidity (percent) and total light extinction (Mm"1) for 2005-2007,
continued
(b) February-April
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, IL
Tacoma, WA
1-
a g-
s -<
<
8 -
It
a § i
8 -:
(
1-
IH
a 8-
8 -.
(
tff01*^ <<£>
°roX"*' "*"
) 6 12 17 2
Hour
Fresno, CA
•^ ^
V-V-' ,<•
\^C)
) 6 12 17 2
Hour
Atlanta, GA
• j£
>w>0
V xP
^•5S^
) 6 12 17 2
How
- P 8-
- B S -
!
- s a -
h SB SB -
3 <
- § in
- R a -
- S |-
r 8 S r
3 <
- 1 in
4
- S 8 -
to in _
3 (
c(Sjfl^.
cfto ^i|^*a»;
) 6 12 17 2
HC.II
Houston, TX
""% ^
_
QJ5rf«L^_f(SO
T^«^
) 6 12 17 2
Hour
Baltimore, MD
^. x^
rtf?*-*
*+&^fS^
) 6 12 17 2
Hour
- B a-
- S 8-
r SB S3 -
3 C
I
- 1 in
- R a-
- S 8-
r !B S -
3 C
- 1 in
r R a^
- S 8-
^ SQ 10 _
3 C
vx
•*^^^
S 12 17 2
Hour
.os Angeles, Cf
• _rfp
P^**1^.^ rft
_S£?;
S 12 17 2
Hour
Birmingham, AL
• ^^
n 'nl AH
^ ". V-F!'
V-^f0^
6 12 17 2
Hour
-5 B-
- R a-
- S 8-
r S S -
3 (
- 1 in
- R a-
- S 8-
r S S -i
3 1
- S S-,
i- PI
_ jrt S -
-a B-
_ un SQ —
3 1
"""N^^
Vc^
6 12 17 2
Hour
New York, NY
^^V ^
dk^
a^ '^^^^n
6 12 17 2
Hour
Dallas, TX
r*^ * j-P
_'X^.»
%"
^•fCO
N—^
6 12 17 2
Hour
-8 H -
-a 1-
•• /Q S ••
3 (
- 1 1 -
r e 8-;
- S 8-
r 8 8 -
3 (
- i 8-
- is a i
- S 8-
k- S3 8 n
3 (
^x^^,
trrrV — ^ AU"
^^^i^^^
) 6 12 17 2
Hour
Philadelphia, PA
^Xx*
'°*atv*. <
j "^^••^jr>k*J
) 8 12 17 2
Hour
Detroit, Ml
t**^ jt
P A
•M»9
3 6 12 17 2
Hour
*•*
- s?
*\
3
i
r -
r ^
o
in
- S
3
r §
r IS
- S
- S
3
•RH
•Ext
September 2009
DRAFT - Do Not Quote or Cite
E-3
-------
Figure E-3. Diurnal and seasonal patterns of relative humidity (percent) and total light extinction (Mm"1) for 2005-2007,
continued
(c) May-July
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, IL
Tacoma, WA
1 -
1-
a 8-
s -
<
§ -
1-
^ a -<
s -1
(
1-
1^
uu g _
j
S -
(
^ V. ^
>bc ""Tbiitr^
) 6 12 17 2
Hour
Fresno, CA
**% -*
' ^^-IIT^
"a"*" "— ^
) 6 12 17 2
Hour
Atlanta, GA
^ 0^
a^w*>-''r
**T_»itr^
) 6 12 17 2
How
- 1 §H
- R S -
- s a j
-a SB -
3 <
- 8 § -,
- S aj
- S 8 -
r 8 S3 -
3 <
"* 2 SS ""
L R a-
- s a -,
r 8 S3 -
3 (
-\^-
VrU^*
T/*rr
) 6 12 17 2
Hc.ii
Houston, TX
OB\ /^
_^V
) 6 12 17 2
Hour
Baltimore, MD
3&*
•"^^y
*1_|ll~*
) 6 12 17 2
Hour
- 1 8-1
- R a-
L S 8 -,
_ tn Jfl —
'
3 C
I
-8 | -,
- R a-
- s a-
*- 8 S -
3 C
- 1 8-1
L R aj
- s s -s
^ SQ 10 _
3 C
0
•• J*'
GCZOMIMHHgModb
S 12 17 2
Hour
.os Angeles, Ct
(rf**V
\_jy^^
S 12 17 2
Hour
Birmingham, AL
^•. o^
DCP *,^^X
^^*X^
6 12 17 2
Hour
r § gi
- R a-
r s §,
-a s -
3 (
i
r § 8 -|
-s a-1
-a 1-
- a s -
3 (
r i §n
"~ jG S -j
r S I-
-8 8 -i
3 1
*7 v -/"
^v*. ^
_^s
6 12 17 2
Hour
New York, NY
"Xx
«»pc V
^***p^
6 12 17 2
Hour
Dallas, TX
». <&
%x^.*
co^\_ fot
V •<'
6 12 17 2
Hour
- i 8 -
- R a J
r S 8 -
_ i/j 12 —•
3 (
-8 | -
L fi g -
L S 8-
_ in to _
3 (
- i 8-
r R 3J
- S | 7
k- 8 S n
3 (
-------
Figure E-4. Diurnal and seasonal patterns of relative humidity (percent) and total light extinction (Mm"1) for 2005-2007,
continued
(d) August-October
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, LIT
St. Louis, IL
Tacoma, WA
s -
a-
3 -
i
(
a-
8"
3 -t
aj
Sl
IS -
(
****•.
^- -er
EOOXIW^^/VP3*'
) 6 12 17 2
Hour
Fresno, CA
^X5
^S2S
^•^
) 6 12 17 2
Hour
Atlanta, GA
*""•. 0^
. 0
«*JJNV^",«
) 6 12 17 2
Hour
I
- R i -
- S? 8 J
- S 3 -
3 (
r S 8-
t- S3 3 -i
3 (
L R a-
- S 8 -
_ tn S ~
3 <
a*«V^ ^0XH
octf*-*^* o**^*
v^
) 6 12 17 Z
Hour
Houston, TX
• cT^
v/
itrf^ ^tf3D
-•M^1
) 6 12 17 2
Hour
Baltimore, MD
(5**1\ /*
^\*ffi
} 6 12 17 2
Hour
- s •-
L S 8 -!
r {3 3 -
I
3 (
I
- 1 In
L R a-
- S 8-
- S 3 -
3 (
- R aj
r S S J
_ jrj g _
3 (
^-o«
•"\>1
N.X
coox^^^a ^PO01
6 12 17 2
Hour
.os Angeles, Cf
•2^1$^
6 12 17 2
Hour
Birmingham, AL
o\«. d^
0 - V j°
_ '-^V
*^ut*
6 12 17 2
Hour
-« a-
" S 8 -
_ in to _
3 (
1
r S 8-
- S S3 r
a t
- £ 8 -
L a 1 -
- R S J
3 1
>*CCO\ rf^
*- jtf°
*%^
6 12 17 2
Hour
New York, NY
rTY
-------
1
2
3
APPENDIX F - DISTRIBUTIONS OF MAXIMUM DAILY
DAYLIGHT TOTAL LIGHT EXTINCTION - UNDER "JUST
MEET" CONDITIONS
5
6
(a) 201 Mm"1, 90th percentile
ExtRollbackDailyMaxNAAQS201Pctl90
(b) 201 Mm'1, 95th percentile
ExtRollbackDailyMaxNAAQS201Pctl95
o
-8-
*v d>-
^ "
•«,* <<• ,
6*' <*-
September 2009
DRAFT - Do Not Quote or Cite
F-l
-------
2
3
(c) 122 Mm"1, 90th percentile
ExtRollbackDailyMaxNAAQS122Pctl90
!il 8 -
110 324 302 98 306 274 159 294 350 295
284 137
o d
I X
(d) 122 Mm'1, 95th percentile
ExtRollbackDailyMaxNAAQS122Pctl95
110 324 302 98 306 274 159 294 350 295 141 284 187 145 228
^
September 2009
DRAFT - Do Not Quote or Cite
F-2
-------
2
3
(e) 74 Mm"1, 90th percentile
ExtRollbackDailyMaxNAAQS74Pctl90
!il 8 -
110 324 302 98 306 274 159 294 350 295
2S4 1S7
X
-§--§--§-
^* ^ ^r
Xs ^
(f) 74 Mm'1, 95th percentile
ExtRollbackDailyMaxNAAQS74Pctl95
~ S -
110 324 302 98 306 274 159 294 350 295
2S4 1S7
-§--§--§.
September 2009
DRAFT - Do Not Quote or Cite
F-3
-------
(g) 15 ug/m3 annual, 35 ug/m3 24-hour
PMRollbackDailyMaxCasel NAAQS
2
3
4
110 324 302 98 306 274 159 294 350 295
O O
-*' ov
284 187 145 22S
T T r ^? T t t
-------
-------
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-452/P-09-005
Environmental Protection Health and Environmental Impacts Division September 2009
Agency Research Triangle Park, NC
------- |