EPA-450/5-83-004
The NAAQS Exposure Model (NEM)
     Applied  to Carbon Monoxide
                  Ted Johnson and Roy A. Paul
                   PEDCo Environmental, Inc.
                505 South Duke Street, Suite 503
                 Durham, North Carolina 27701
                       Prepared for

             U.S. ENVIRONMENTAL PROTECTION AGENCY
                   Office of Air and Radiation
              Office of Air Quality Planning and Standards
                 Research Triangle Park, NC 2771 1

                 Thomas McCurdy, Task Manager
                      December 1983

-------
                           DISCLAIMER
     This report was furnished to the U.S. Environmental Protec-
tion Agency by PEDCo Environmental, Inc./ in partial fulfillment
of Contract No. 68-02-3390, Work Assignments No. 13 and 16.  The
contents of this report are reproduced herein as received from
the contractor.  The opinions, findings, and conclusions expressed
are those of the authors and not necessarily those of the Environ-
mental Protection Agency.

     This draft report is being circulated for review and comment.
Anyone interested in commenting or providing information concern-
ing the material should address their communications to Thomas
Feagans, Ambient Standards Branch, MD-12, U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina 27711.
If you would like to discuss the report, please call him at
(919)541-5655  (FTS 629-5655).
                               111

-------
                            CONTENTS
Figures                                                      vii
Tables                                                        ix
Acknowledgment                                              xvii

1.   Introduction                                           1-1

2.   Overview of the Exposure Model                         2-1

     2.1  The exposure district version of NEM              2-1
     2.2  The neighborhood version of NEM                   2-3
     2.3  References                                        2-5

3.   Simulation of Population Movement                      3-1

     3.1  Composition of cohort files                       3-1
     3.2  Types of neighborhoods                            3-8
     3.3  Estimation of cohort populations                  3-10
     3.4  References                                        3-17

4.   Preparation of Air Quality Data                        4-1

     4.1  Selection of representative data sets             4-1
     4.2  Validation of air quality data                    4-4
     4.3  Simulation of missing values in hourly
            average CO data sets                            4-9
     4.4  References                                        4-18

5.   Simulation of Air Quality Expected at Fixed
       Monitoring Site Under Alternative Carbon
       Monoxide Standards                                   5-1

     5.1  The rollback model                                5-1
     5.2  Air quality indicators                            5-2
     5.3  Background concentrations                         5-15
     5.4  References                                        5-18

-------
                      CONTENTS (continued)
6.    Simulation of Carbon Monoxide Levels in the
       Microenvironment

     6.1  Work-school microenvironment
     6.2  Home-other microenvironment
     6.3  Transportation vehicle microenvironment
     6.4  Roadside microenvironment
     6.5  Other outdoor locations
     6.6  Summary
     6.7  References

7.    Exposure Estimates for Adults with Cardiovascular
       Disease in Four Urban Areas

     7.1  "Best estimate" results
     7.2  Male/female comparisons
     7.3  The significance of indoor sources
     7.4  Uncertainty in NEM exposure estimates
     7.5  References

8.    Nationwide Extrapolations

     8.1  Extrapolation procedure
     8.2  Extrapolation results
     8.3  Uncertainty of the nationwide estimates
Appendix A


Appendix B

Appendix C
Activity Patterns by Age-Occupation
  Subgroup

Cohort Populations by Study Area

Discussion of Air Quality Indicators
  Used in the NEM Analysis and Estimated
  Concentrations Used in the Regulatory
  Analysis
                                              6-1

                                              6-3
                                              6-9
                                              6-18
                                              6-25
                                              6-26
                                              6-26
                                              6-29
                                              7-1

                                              7-1
                                              7-39
                                              7-42
                                              7-42
                                              7-55

                                              8-1

                                              8-2
                                              8-4
                                              8-16
A-l

B-l
                                                             C-l
                              VI

-------
                             FIGURES

Number

 2-1      Flow Diagram of Exposure District Version
            of NEM                                          2-4

 2-2      Flow Diagram of Neighborhood Version of NEM       2-6

 3-1      Simulated Movements Between Exposure Districts
            In The Los Angeles-San Bernardino Study Area    3-2

 3-2      Pattern of Neighborhood Types in a Portion of
            The Philadelphia Study Area                     3-3

 3-3      Representation of People-Movement in the
            Neighborhood Version of NEM                     3-4

 3-4      Contents of Cohort Activity File                  3-11

 4-1      Hourly Average 1978 Carbon Monoxide Data
            Reported by Monitoring Site 261040001G01
            in St. Louis                                    4-15

 4-2      Hourly Average 1978 Carbon Monoxide Data for
            Monitoring Site 261040001G01 in St. Louis
            After Initial Simulation of Missing Values      4-16

 4-3      Hourly Average 1978 Carbon Monoxide Data for
            Monitoring Site 261040001G01 in St. Louis
            After Final Simulation of Missing Values        4-17
                              VII

-------
                             TABLES

Number                                                      Page

 3-1      Description and Apportionment of Activity
            Pattern Subgroups                               3-6

 3-2      Neighborhood Types (NT's) and Codes               3-9

 3-3      Geographic Classification of Exposure Districts   3-12

 3-4      A-0 Group Population By Residential Neighbor-
            hood Classifications in Four Study Areas        3-13

 3-5      Assumptions Concerning Work NT's of A-0 Groups    3-15

 3-6      Collapsed Home-To-Work Trip Tables, Expressed
            As Number Of Trips And Fraction of Trips        3-16

 4-1      Data Sets (mg/m )  Selected For Analysis of
            Population Exposure To Carbon Monoxide in
            Four Cities                                     4-2

 4-2      Anomalies Flagged by Data Screening               4-8

 5-1      Results of Fitting Weibull and Lognormal
            Distributions to 1978 CO Data (ppm) From
            St. Louis, MO by Least Squares Method           5-10

 5-2      Results of Fitting Weibull and Lognormal
            Distributions by Maximum Likelihood Procedure
            to Upper 50 Percent of Daily Maximum 1-Hour
            CO Data                                         5-11

 5-3      Results of Fitting Weibull and Lognormal Dis-
            tributions by Maximum Likelihood Procedure
            to Upper 50 Percent of Daily Maximum 8-Hour
            Running Average CO Data                         5-12

 5-4      Results of Fitting Weibull and Lognormal Dis-
            tributions by Maximum Likelihood Procedure
            to Upper 20 Percent of Daily Maximum 1-Hour
            CO Data                                         5-13
                               IX

-------
                       TABLES (continued)

Number                                                      Page

 5-5      Results of Fitting Weibull and Lognormal Dis-
            tributions by Maximum Likelihood Procedure
            to Upper 20 Percent of Daily Maximum 8-Hour
            Running Average CO Data                         5-14

 5-6      Air Quality Indicators For CO Data                5-16

 5-7      Estimated Annual Average Background Levels        5-17

 6-1      Studies Considered in Developing CO Micro-
            environment Factors                             6-2

 6-2      Estimates of CO Concentrations in
            an Office With Smokers                          6-5

 6-3      Weekday CO Measurements at Street Canyon Site     6-6

 6-4      CO Concentrations  (ppm) at Two Office Sites
            Recorded by Moschandreas, et al.                6-7

 6-5      Indoor-Outdoor CO Ratios Determined for Two
            Office Buildings by Yocum, et al.               6-7

 6-6      Average Indoor/Outdoor CO Ratios Recorded by
            Yocum, et al.                                   6-9

 6-7      Air Exchange Rates Determined by Moschandreas,
            et al.                                          6-12

 6-8      Indoor/Outdoor CO Data Recorded by Cote, et al.   6-15

 6-9      Average Differences Between Kitchen, Living
            Room, and Outside CO Concentrations             6-16

 6-10     Average CO Levels in Various Structures           6-18

 6-11     Ratios of Mean Personal CO Exposures to Mean
            CO Concentrations at Fixed Monitors             6-23

 6-12     Ratios of CO in Motor Vehicles Concurrent to
            CO at Camp Stations                             6-24

 6-13     Ratios of Mean CO Concentrations at Experimental
            Sites and at Fixed Sites                        6-26

 6-14     Estimates of Additive Microenvironmental
            Factors  (am fc)                                  6-27

-------
                       TABLES (continued)

Number                                                      Page

 6-15     Estimates of Multiplicative Microenviron-
            mental Factor (fc>m)                               6-28

 7-1      Estimates of Occurrences for Adults With
            Cardiovascular Disease of 1-Hour CO Exposures
            Above Selected Concentration Values Assuming
            9 PPM/1 EXEX Standard is Attained               7-2

 7-2      Estimates of Adults With Cardiovascular
            Disease Who Have 1-Hour CO Exposures Above
            Selected Concentration Values Assuming
            9 PPM/1 EXEX Standard is Attained               7-3

 7-3      Estimates of Adults With Cardiovascular
            Disease Whose Maximum 1-Hour CO Exposure
            Occurs in Selected Concentration Ranges
            Assuming 9 PPM/1 EXEX Standard is Attained      7-4

 7-4      Estimates of Occurrences for Adults With
            Cardiovascular Disease of 8-Hour CO
            Exposures Above Selected Concentration
            Values Assuming 9 PPM/1 EXEX Standard
            is Attained                                     7-5

 7-5      Estimates of Adults With Cardiovascular
            Disease Who Have 8-Hour CO Exposures Above
            Selected Concentration Values Assuming
            9 PPM/1 EXEX Standard is Attained               7-6

 7-6      Estimates of Adults With Cardiovascular
            Disease Whose Maximum 8-Hour CO Exposure
            Occurs in Selected Concentration Ranges
            Assuming 9 PPM/1 EXEX Standard is Attained      7-7

 7-7      Estimates of Occurrences for Adults With
            Cardiovascular Disease of COHb Levels
            Exceeding Selected Values Assuming
            9 PPM/1 EXEX Standard is Attained               7-8

 7-8      Estimates of Adults With Cardiovascular
            Disease Who Experience COHb Levels Exceeding
            Selected Values Assuming 9 PPM/1 EXEX
            Standard is Attained                            7-9

7-9       Estimates of Adults With Cardiovascular Disease
            Whose Maximum COHb Level Occurs in Selected
            Ranges Assuming 9 PPM/1 EXEX Standard is
            Attained                                        7-10

                               xi

-------
                       TABLES (continued)

Number
 7-10     Estimates of Occurrences for Adults With
            Cardiovascular Disease of 1-Hour CO
            Exposures Above Selected Concentration
            Values Assuming 12 PPM/1 EXEX Standard
            is Attained                                     7-11

 7-11     Estimates of Adults With Cardiovascular
            Disease Who Have 1-Hour CO Exposures Above
            Selected Concentration Values Assuming
            12 PPM/1 EXEX Standard is Attained              7-12

 7-12     Estimates of Adults With Cardiovascular Disease
            Whose Maximum 1-Hour CO Exposure Occurs in
            Selected Concentration Ranges Assuming
            12 PPM/1 EXEX Standard is Attained              7-13

 7-13     Estimates of Occurrences for Adults With
            Cardiovascular Disease of 8-Hour CO
            Exposures Above Selected Concentration
            Values Assuming 12 PPM/1 EXEX Standard is
            Attained                                        7-14

 7-14     Estimates of Adults With Cardiovascular
            Disease Who Have 8-Hour CO Exposures Above
            Selected Concentration Values Assuming
            12 PPM/1 EXEX Standard is Attained              7-15

 7-15     Estimates of Adults With Cardiovascular
            Disease Whose Maximum 8-Hour CO Exposure
            Occurs in Selected Concentration Ranges
            Assuming 12 PPM/1 EXEX Standard is Attained     7-16

 7-16     Estimates of Occurrences for Adults With
            Cardiovascular Disease of COHb Levels
            Exceeding Selected Values Assuming 12 PPM/1
            EXEX Standard is Attained                       7-17

 7-17     Estimates of Adults With Cardiovascular
            Disease Who Experience COHb Levels Exceeding
            Selected Values Assuming 12 PPM/1 EXEX
            Standard is Attained                            7-18

 7-18     Estimates of Adults With Cardiovascular Disease
            Whose Maximum COHb Level Occurs in Selected
            Ranges Assuming 12 PPM/1 EXEX Standard is
            Attained                                        7-19
                               XII

-------
                       TABLES (continued)

Number

 7-19     Estimates of Occurrences for Adults With
            Cardiovascular Disease of 1-Hour CO Exposures
            Above Selected Concentration Values Assuming
            15 PPM/1 EXEX Standard is Attained              7-20

 7-20     Estimates of Adults With Cardiovascular Disease
            Who Have 1-Hour CO Exposures Above Selected
            Concentration Values Assuming 15 PPM/1 EXEX
            Standard is Attained                            7-21

 7-21     Estimates of Adults With Cardiovascular Disease
            Whose Maximum 1-Hour CO Exposure Occurs in
            Selected Concentration Ranges Assuming
            15 PPM/1 EXEX Standard is Attained              7-22

 7-22     Estimates of Occurrences for Adults With
            Cardiovascular Disease of 8-Hour CO Exposures
            Above Selected Concentration Values Assuming
            15 PPM/1 EXEX Standard is Attained              7-23

 7-23     Estimates of Adults With Cardiovascular Disease
            Who Have 8-Hour CO Exposures Above Selected
            Concentration Values Assuming 15 PPM/1 EXEX
            Standard is Attained                            7-24

 7-24     Estimates of Adults With Cardiovascular Disease
            Whose Maximum 8-Hour CO Exposure Occurs in
            Selected Concentration Ranges Assuming
            15 PPM/1 EXEX Standard is Attained              7-25

 7-25     Estimates of Occurrences for Adults With
            Cardiovascular Disease of COHb Levels
            Exceeding Selected Values Assuming
            15 PPM/1 EXEX Standard is Attained              7-26

 7-26     Estimates of Adults With Cardiovascular Disease
            Who Experience COHb Levels Exceeding Selected
            Values Assuming 15 PPM/1 EXEX Standard is
            Attained                                        7-27

 7-27     Estimates of Adults With Cardiovascular Disease
            Whose Maximum COHb Level Occurs in Selected
            Ranges Assuming 15 PPM/1 EXEX Standard is
            Attained                                        7-28

 7-28     Algorithm Used to Calculate Carboxyhemoglobin
            in Blood of Cohorts                             7-31
                               Xlll

-------
                       TABLES (continued)

Number

 7-29     Values Assigned to Variables in Algorithm
            Used to Estimate Carboxyhemoglobin              7-32

 7-30     Percentage of Adults With Cardiovascular
            Disease Experiencing COHb Levels Exceeding
            Selected Values Assuming 9 PPM/1 EXEX
            Standard is Attained         .                   7-35

 7-31     Percentage of Adults With Cardiovascular
            Disease Experiencing COHb Levels Exceeding
            Selected Values Assuming 12 PPM/1 EXEX
            Standard is Attained                            7-37

 7-32     Percentage of Adults With Cardiovascular
            Disease Experiencing COHb Levels Exceeding
            Selected Values Assuming 15 PPM/1 EXEX
            Standard is Attained                            7-38

 7-33     Estimates of Adult Females With Cardiovascular
            Disease Who Experience COHb Levels Exceeding
            Selected Values Assuming 9 PPM/1 EXEX Standard
            is Attained                                     7-40

 7-34     Estimates of Adult Males With Cardiovascular
            Disease Who Experience COHb Levels Exceeding
            Selected Values Assuming 9 PPM/1 EXEX Standard
            is Attained                                     7-41

 7-35     Estimates of Adults With Cardiovascular Disease
            Who Have 1-Hour CO Exposures Above Selected
            Values Under 9 PPM/1 EXEX Standard With In-
            door Sources Omitted                            7-43

 7-36     Estimates of Adults With Cardiovascular Disease
            Who Have 8-Hour CO Exposures Above Selected
            Values Under 9 PPM/1 EXEX Standard With In-
            door Sources Omitted                            7-44

 7-37     Estimates of Occurrences for Adults With
            Cardiovascular Disease of COHb Levels
            Exceeding Selected Values Under 9 PPM/1
            EXEX Standard With Indoor Sources Omitted       7-45

 7-38     Estimates of Adults With Cardiovascular Disease
            Who Experience COHb Levels Exceeding Selected
            Values Under 9 PPM/1 EXEX Standard With In-
            door Sources Omitted                            7-46
                              xiv

-------
                       TABLES (continued)

Number
 7-39     Estimates of Adults with Cardiovascular Disease
            whose Maximum COHb Level Occurs in Selected
            Ranges under 9 PPM/1 EXEX Standard with Indoor
            Sources Omitted                                 7-47

 7-40     Estimates of Adults with Cardiovascular Disease
            Chicago with 1-Hour Carbon Monoxide Exposures
            Above Selected Values Under 12 PPM/1 EXEX Stan-
            dard Using Best,  Lower, and Upper Microenviron-
            ment Factors                                    7-49

 7-41     Estimates of Adults with Cardiovascular Disease
            in Chicago with 8-Hour Carbon Monoxide Exposures
            above Selected Values under 12 PPM/1 EXEX Stan-
            dard using Best,  Lower, and Upper Microenviron-
            ment Factors                                    7-50

 7-42     Estimates of Occurrences for Adults with Cardio-
            vascular Disease  in Chicago of COHb Levels
            Exceeding Selected Values under 12 PPM/1 EXEX
            Standard using Best, Lower, and Upper Micro-
            environment Factors                             7-51

 7-43     Estimates of Adults with Cardiovascular Disease
            in Chicago Experiencing COHb Levels Exceeding
            Selected Values Under 12 PPM/1 EXEX Standard
            Using Best, Lower, and Upper Microenvironment
            Factors                                         7-52

 7-44     Estimates of Adults with Cardiovascular Disease
            in Chicago whose  Maximum COHb Level Occurs in
            Ranges under 12 PPM/1 EXEX Standard Using Best,
            Lower, and Upper  Microenvironment Factors       7-53

 7-45     Sensitivity of COHb Estimates for Chicago to
            Variations in Two Physiological Variables       7-54

 8-1      Urbanized Area Population Data Used to Extra-
            polate Model Results                            8-4

 8-2      Estimates of Occurrences in the Cardiovascular
            Adult Urban U.S.  Population of 1-Hour Average
            CO Exposures above Selected Concentration Values
            under Alternative Air Quality Assumptions       8-5
                                xv

-------
                       TABLES (continued)

Number                                                      Page

 8-3      Estimates of Cardiovascular Adults in Urban U.S.
            with 1-Hour Average CO Exposures above Selected
            Concentration Values under Alternative Air
            Quality Assumptions                             8-6

 8-4      Estimates of Adults in Urban U.S. Whose Maximum
            1-Hour Average CO Exposure Occurs in Selected
            Concentration Ranges Under Alternative Air
            Quality Assumptions                             8-7

 8-5      Estimates of Occurrences in the Cardiovascular
            Adult Urban U.S. Population of 8-Hour Average
            CO Exposures above Selected Concentration
            Values Under Alternative Air Quality Assump-
            tions                                           8-8

 8-6      Estimates of Cardiovascular Adults in Urban U.S.
            with 8-Hour Average CO Exposure Above Selected
            Concentration Values Under Alternative Air
            Quality Assumptions                             8-9

 8-7      Estimates of Cardiovascular Adults in Urban U.S.
            Whose Maximum 8-Hour Average CO Exposure Occurs
            in Selected Concentration Ranges Under Alterna-
            tive Air Quality Assumptions                    8-10

 8-8      Estimates of Occurrences among Cardiovascular
            Adults in Urban U.S. of COHb Levels Exceeding
            Selected Values under Alternative Air Quality
            Assumptions                                     8-11

 8-9      Estimates of Cardiovascular Adults in Urban U.S.
            Experiencing COHb Levels Exceeding Selected
            Values under Alternative Air Quality Assump-
            tions                                           8-12

 8-10     Estimates of Cardiovascular Adults in Urban U.S.
            Whose Maximum COHb Level Occurs in Selected
            Concentration Ranges Under Alternative Air
            Quality Assumptions                             8-13

 8-11     Percentage of Cardiovascular Adult Urban U.S.
            Population Experiencing COHb Levels Exceeding
            Selected Values under Alternative Air Quality
            Assumptions                                     8-14
                               xvi

-------
                       TABLES (continued)

Number

 8-12     Estimates of Cardiovascular Adults in Urban U.S.
            Experiencing COHb Levels Exceeding Selected
            Values a Given Number of Days Assuming 9 PPM/1
            EXEX Standard is Attained                       8-15

 C-l      Estimated Concentrations (EC's) Developed by
            EPA and Corresponding Air Quality Indicators
            (AQI's) from Table 5-6 (concentrations in
            parts per million)                              C-4
                              xvi i

-------
                         ACKNOWLEDGMENTS


     The development of a general model for assessing population
exposures associated with possible National Ambient Air Quality
Standards (NAAQS's) and the application of that model to carbon
monoxide (CO) reported in this document have involved the efforts
of many people.  The following persons associated with PEDCo
Environmental, Inc., played important roles in conducting this
study.  Coauthor Roy Paul developed the population movement
algorithms and created the population data files.  Coauthor Ted
Johnson developed the statistical models used to process air
quality data and managed the development of air quality and
population data bases.  Irene Griffin assisted in the compilation
of population and air quality data.  James Capel created and
debugged the computer programs which processed air quality data.
Barbara Blagun developed estimates of background pollutant con-
centrations.  Dianne Gupton and Dian Dixon typed the report.

     George M. Duggan, Strategies and Air Standards Division  (SASD),
U.S. Environmental Protection Agency  (EPA), designed and imple-
mented the computer programs used to calculate exposure estimates.
Thomas McCurdy, SASD, EPA, facilitated conduct of the study and
co-managed the SASD exposure assessment program with Henry Thomas.
Much of the cohort data used in the model is based upon work done
by SRI International.1

     Dr. William F. Biller, EPA contractor, and Thomas B. Feagans,
SASD, developed the general exposure model and the method used to
extrapolate exposure estimates for individual study areas to the
nation.  The general model makes use of ideas developed by EPA's
Office of Research and Development and others.2  The model was
first applied to carbon monoxide at an early stage of its develop-
ment. 3  A later version of the model was applied to nitrogen
dioxide and particulate matter.l+'5  An exposition of the general
model is available from EPA.6
                           REFERENCES

     Marc F. Roddin, Hazel T. Ellis, and Waheed M. Siddiqee,
     Background Data for Human Activity Patterns, Vols. 1 and 2,
     Draft final report prepared for Strategies and Air Standards
     Division, Office of Air Quality Planning and Standards, U.S.
     Environmental Protection Agency, Research Triangle Park,
     N.C. 27711, August 1979.
                              XVlll

-------
2.    James L.  Repace, Wayne R.  Ott,  and Lance A. Wallace, "Total
     human exposure to air pollution," paper 80-61.6, presented
     at the 73rd Annual Meeting of the Air Pollution Control
     Association, Montreal, June 1980.

3.    William F.  Biller, Thomas  B.  Feagans, Ted R. Johnson,
     George M.  Duggan, and James E.  Capel, Estimated Exposure
     to Ambient Carbon Monoxide Concentrations Under Alternative
     Air Quality Standards,Strategies and Air Standards Division,
     Office of  Air Quality Planning and Standards, U.S.  Environ-
     mental Protection Agency,  Research Triangle Park, N.C.
     27711, January 1981.

4.    Ted Johnson and Roy Paul,  The NAAQS Exposure Model  (NEM) and
     Its Application to Nitrogen Dioxide, prepared by PEDCo Envir-
     onmental,  Inc., for Strategies and Air Standards Division,
     Office of  Air Quality Planning and Standards, Environmental
     Protection Agency, Research Triangle Park, N.C. 27711,
     August 1981.

5.    Ted Johnson and Roy Paul,  The NAAQS Exposure Model  (NEM) and
     Its Application to Particulate Matter, prepared by  PEDCo
     Environmental, Inc.,  for Strategies and Air Standards Divi-
     sion, Office of Air Quality Planning and Standards, U.S.
     Environmental Protection Agency,  Research Triangle  Park,
     N.C. 27711, August 1981.

6.    William F.  Biller, Thomas  B.  Feagans, Ted R. Johnson, George
     M. Duggan,  Roy A. Paul,  Thomas McCurdy, and Henry C. Thomas,
     "A general model for estimating exposure associated with
     alternative NAAQS," paper  81-18.4, presented at the 74th
     Annual Meeting of the Air  Pollution Control Association,
     Philadelphia, June 1981.
                               xix

-------
                            SECTION 1
                          INTRODUCTION

     Under the Clean Air Act, the Environmental Protection Agency
                                                    •
(EPA) is responsible for establishing National Ambient Air Quality
Standards (NAAQS's) and for reviewing them periodically to deter-
mine their adequacies on the basis of recent experience and
research.  In view of these responsibilities, the Strategies and
Air Standards Division (SASD) of the Office of Air Quality Plan-
ning and Standards (OAQPS) is exploring the use of quantitative
methods for assessing health risks associated with proposed air
quality standards.
     An important aspect of health risk assessment is the esti-
mation of population exposure.  For the past few years, SASD has
been engaged in the development of an exposure model suitable for
evaluating alternative ambient air standards.  The model is known
as NEM, an acronym for NAAQS Exposure Model.
     Two versions of NEM have been developed.  The exposure dis-
trict version of NEM simulates the pollutant concentration expected
to occur in selected exposure districts within a study area under
user-specified regulatory scenarios.  It then adjusts these
estimates to account for an exhaustive set of microenvironments
and simulates typical movements of population subgroups, called
cohorts, through the districts and microenvironments.  Outputs of
the simulation program are population exposure estimates at
specified pollutant levels.  Three indices of exposure are used
currently, and more are being investigated.
     The other version of NEM substitutes "neighborhood types"
for exposure districts.  This report describes this alternative
model and its application to four U.S. urban areas (Chicago, IL,
St. Louis, MO-IL, Philadelphia, PA, and Los Angeles,  CA) to
estimate population exposures associated with alternative NAAQS's

                               1-1

-------
proposed for carbon monoxide (CO).   Results of these analyses are
included in this report.  The contribution of indoor CO sources
to total population exposure is also evaluated.
                               1-2

-------
                            SECTION 2
                 OVERVIEW OF THE EXPOSURE MODEL

     Analysis of population exposure under present and proposed
National Ambient Air Quality Standards (NAAQS's) requires that
the significant factors contributing to total human exposure be
taken into account.  Consequently, these factors have been in-
corporated into the NAAQS Exposure Model (NEM), a simulation
model capable of estimating human exposure in selected urbanized
areas under user-specified regulatory scenarios.  The general
model has been designed so that pollutants as diverse as ozone
and CO can be accommodated without making changes to the basic
NEM program.  Instead, input data files are developed to reflect
the assumed spatial characteristics of the specific pollutant
being analyzed.  In one version of the model, air quality data
are selected to represent a small number of discrete exposure
districts.  This version has been applied to N02 and particulate
matter.  Another version incorporating the concept of "neighbor-
hood types" is the basis of the CO exposure analysis described in
this report.  This section briefly discusses the differences
between these two versions of NEM.

2.1  THE EXPOSURE DISTRICT VERSION OF NEM
     In the exposure district version of NEM, land areas within a
selected study area are represented by large, bounded "exposure
districts."  The population within each exposure district is
assigned to a single discrete point, the population centreid.
The air quality level within each exposure district is represented
by the air quality level at the population centroid, which is
estimated for each hour of the year by using monitoring data from
adjacent monitoring sites.  Because pollutants in the air can be

                               2-1

-------
modified considerably when entering a building or vehicle/ these
ambient air quality estimates are adjusted to account for five
different microenvironments:   indoors at work or school, indoors
at home or other locations, inside a transportation vehicle,
outdoors near a roadway, and other outdoor locations.  NEM simu-
lates hour-by-hour movements of representative population groups
through different districts of the city and through different
microenvironments, accumulating the resulting exposure over a
period of one year.
     Because degree of exposure and susceptibility to effects of
pollution vary with age, occupation, and intensity of exercise;
the total population of each study area is divided into age-
occupation (A-0) groups, and each A-O group is further subdivided
into three or more subgroups.  For each subgroup, a typical pat-
tern of activity through the five microenvironments is established,
and an exercise level  (high,  medium, or low) within each is speci-
fied.
     Units of population analyzed by NEM are called cohorts.
Each cohort is identified by exposure district of residence, by
exposure district of employment, by A-O group, and by activity-
pattern subgroup.  During each hour of the year, each cohort is
assigned to a particular exposure district and a particular
microenvironment.  Since NEM simulates hour-by-hour air quality
in each district and each microenvironment, the hourly exposures
of each cohort may be summed over a one year period.  Annual
cohort exposures are summed to provide exposure estimates for
each A-O group, which, in turn, are summed for all groups to
provide an estimate of total population exposure for a particular
study area.  Output of NEM is a series of tables showing fre-
quency distributions of total exposure at different averaging
times  (e.g., 1 hour, 8 hours, 1 year) using different measures of
exposure  (e.g., number of persons with exposures above selected
pollutant levels).
     In developing data bases for this version of NEM, PEDCo had
to establish exposure districts within each study area that would
accommodate the available breakdowns of transportation and census
                                2-2

-------
data; establish air quality data sets for exposure districts; and
set up files listing hourly assignments to an exposure district,
a microenvironment, and an exercise level for each cohort.  In
addition, files had to be established which contained air quality
adjustment factors appropriate to each microenvironment and
rollback factors for adjusting air quality data according to each
air quality standard under consideration.  The methods used in
carrying out these tasks are described in previous reports.1'2
Figure 2-1 shows a flow diagram for the exposure district version
Of NEM.

2.2  THE NEIGHBORHOOD VERSION OF NEM
     Under the exposure district version of NEM, air quality with-
in each microenvironment is a linear function of hourly air quality
values derived from data reported by one or more monitors near
the district centroid.  This representation works best when air
quality within a microenvironment varies slowly with distance
and the sources of pollution are widely distributed.  A good
example of this type of pollutant is ozone (0,), which is formed
relatively slowly in the atmosphere from nitrogen dioxide (NO2)
and hydrocarbon (HC) precursors.  Ozone is transported long
distances with only small changes in concentration, and the
primary sources of pollution (automobiles) are widely distributed.
Under these conditions the use of large exposure districts works
well.
     Large exposure districts are not always appropriate for
determining exposure.  For some pollutants, air quality within
a microenvironment is often more dependent on land use than geo-
graphic location.   In this case, it is more appropriate to
divide a study area into zones which can be classified according
to the types and intensities of emission sources within them.  In
the neighborhood version of NEM, a small number of neighborhood
types  (NT's) are established for this purpose.
                               2-3

-------
                    ANALYST SELECTS

                     (1)  TEST  CITY
                     (2)  NAAQS
                     START PROGRAM
                  SELECT NEXT  COHORT
                   DETERMINE  COHORT
                      POPULATION
             DETERMINE EXPOSURE  DISTRICT
            MICROENVIRON, ACTIVITY  LEVEL
                           I
            DETERMINE AMBIENT AIR QUALITY
              (AQ) IN EXPOSURE DISTRICT
                DETERMINE ADJUSTED AQ
                      PER NAAOS
                           I
                           T
                    DETERMINE AQ IN
                   MICROENVIRONMENT
             ACCUMULATE 1 HOUR OF EXPOSURE
               AT AQ LEVEL IN A REGISTER
             SUMMARIZE  COHORT EXPOSURES FOR
              AGE-OCCUPATION (A-0) GROUPS
                        (OUTPUT)
               FREQUENCY  DISTRIBUTION OF
                 EXPOSURES  BY A-0 GROUP
 COHORT POPULATION
       FILE
 COHORT LOCATION
      FILE
HOURLY AIR QUALITY
      FILE
ROLL-BACK FACTORS
  MICROENVIRONMENT
       FACTORS
Figure  2-1.   Flow  diagram  of  exposure district  version  of  NEM.

                                      2-4

-------
     The neighborhood version of NEM is particularly appropriate
for analyzing CO exposure.  Since suburban neighborhoods generally

have lower population densities and lower traffic densities than
center-city neighborhoods, they generally have lower CO levels.
Likewise, commercial and industrial neighborhoods generally have CO

source patterns different from residential neighborhoods.

     Figure 2-2 shows the flow diagram for the neighborhood version

of NEM.  The principal differences between the two versions lie
in the concept of people-movement and in the structure of data

files used as inputs.  The concept of neighborhood-type and

methods of creating population data bases are described in the
next section.


2.3  REFERENCES

1.   Ted Johnson and Roy Paul, The NAAQS Exposure Model (NEM)
     and Its Application to Nitrogen Dioxide, prepared by PEDCo
     Environmental, Inc., for Strategies and Air Standards Divi-
     sion, Office of Air Quality Planning and Standards, U.S.
     Environmental Protection Agency, Research Triangle Park,
     N. C. 27711, May 1981.

2.   Ted Johnson and Roy Paul, The NAAQS Exposure Model (NEM)
     and Its Application to Particulate Matter, prepared by PEDCo
     Environmental, Inc., for Strategies and Air Standards Division,
     Office of Air Quality Planning and Standards, U.S. Environ-
     mental Protection Agency, Research Triangle Park, N. C.
     27711, August 1981.
                              2-5

-------
                       ANALYST  SELECTS
                        (1)  TEST CITY
                        (2)  NMQS
                        START PROGRAM
                  *  SELECT NEXT COHORT
                      DETERMINE COHORT
                         POPULATION
                DETERMINE  NEIGHBORHOOD TYPE
               MICROENVIRON, ACTIVITY LEVEL
               DETERMINE AMBIENT AIR QUALITY
                 (AQ)  IN NEIGHBORHOOD TYPE
                   DETERMINE ADJUSTED AQ
                         PER NAAQS
                       DETERMINE AQ IN
                      MICROENVIRONMENT
                ACCUMULATE  1  HOUR OF EXPOSURE
                  AT AQ LEVEL IN A REGISTER
               SUMMARIZE COHORT EXPOSURES FOR
                AGE-OCCUPATION  (A-0) GROUPS
                          (OUTPUT)
                 FREQUENCY DISTRIBUTION OF
                   EXPOSURES BY  A-0  GROUP
 COHORT POPULATION
       FILE
 COHORT LOCATION
      FILE
HOURLY AIR QUALITY
      FILE
ROLL-BACK FACTORS
  MICROENVIRONMENT
       FACTORS
Figure  2-2.    Flow  diagram of  neighborhood version of NEM.

                                 2-6

-------
                           SECTION 3
                SIMULATION OF POPULATION MOVEMENT

     NEM simulates movement of small, homogeneous groups called
cohorts through zones of varying air quality.  This section
describes the development of computer files delineating cohort
movements and the allocation of study area population among the
cohorts.

3.1  COMPOSITION OF COHORT FILES
     A cohort is defined as a group of individuals having a
statistical factor in common in a demographic study.  In the
neighborhood version of NEM, all members of a particular cohort
     (1)  live in the same neighborhood type  (NT),
     (2)  work in the same NT,
     (3)  are members of the same age-occupation group, and
     (4)  are members of a subgroup with**a specified daily
          activity pattern.
Consequently, a computer file describing a cohort is labeled as
to home NT, work NT, age-occupation group, and activity pattern.
     In the exposure district version of NEM, the activity pat-
tern for a cohort contains hourly assignments to predetermined
exposure districts.  Cohort movement can be visualized as trans-
fers between geographically connected districts.  For example,
Figure 3-1 illustrates movement between various districts of the
Los Angeles-San Bernardino study area.  By contrast, the activity
pattern for a cohort in the neighborhood version of NEM contains
hourly assignments to predetermined NT's.  A particular NT may be
scattered over an entire study area and mixed with other NT's, as
shown in Figure 3-2.  Movement from one NT to another may repre-
sent a significant change in air quality level, without involving
a significant change in geographic location.
                               3-1

-------
                                                           CD
                                                           CO
                                                          +J
                                                           O

                                                           si
                                                          4J
                                                           CO
                                                              O)
                                                           OJ i_
                                                           S- «3

                                                           V) >,

                                                           9 ~°
                                                           O. 3
                                                           X -M
                                                             t.
                                                           OJ  0
                                                           >   I
                                                           O  t/>
                                                           E  
-------
                    MORRISTOWN
    BRIDGEPORT
                                          CONSHOCKEN
KEY:

  3 RESIDENTIAL  NEIGHBORHOOD
 ^COMMERCIAL  NEIGHBORHOOD
    INDUSTRIAL  NEIGHBORHOOD
    UNDEVELOPED
             Figure 3-2.  Pattern of neighborhood types  in  a
                 portion of the Philadelphia study area.
                                  3-3

-------
   CM
+J=»t=

CJ  s_
S  3
s:  o
one
•r—
V>  i.
10  O
» o
I—  O
 I
-o  >>
 O  -t->
 O  i-
-c  u
 i_   i
 o  s-
-Q  0)
-C  4J
 O1 C
•i—  cij
 cu  o
 O  +•>
 S-  (0
•i—   I
 >  in
 c  s-
 
Q.-—
^)LT
h"~ "^*
1

(U
• Q.

' 1—
|




c .
o

C
i.

>> "O OJ -O -~ » -i- -M -—
-Q OCJOCM-P 
•^
•I- i.
(U M- (1) 0
Z O S M-

t— 1

CM

IQ
Q.-—
3 CM
O 	 '
«— '

>> Q.
4-> 3
•i- o
> t.
•t- O)
•M jQ
tj 3


                                                                                    £ -a
                                                                                     «3  O
                                                                                    -M  O
                                                                                     C -C
                                                                                     
             CsJ
              O)
              E
    o
 o  c

•^   «
 >  in
 £=  S-
 0)  O
 o  o
 s- -a
 (U


 
-------
     In addition to NT assignments, the activity pattern file
contains hourly assignments of each cohort to a microenvironment
and an exercise level for typical weekdays, Saturdays, and Sundays.
Exposure to CO during a particular hour is determined by adjusting
ambient CO for the assigned NT according to the assigned micro-
environment.  Because high exercise levels may induce a higher
uptake of airborne chemicals into the body, NEM keeps track of
exercise levels so that exposure distributions can be calculated
for each exercise level separately.
     In developing activity patterns appropriate for analyzing
CO exposure, we divided each A-0 group into three to six sub-
groups which could be described by one or more demographic vari-
ables affecting exposure (see Table 3-1).   These variables in-
clude commuting time, work shift, work location (e.g., inside,
outside, in motor vehicle), age, and degree of mobility.  The
population of each age-occupation group was apportioned among
its constituent subgroups according to demographic statistics
obtained from the Bureau of Census and other sources.1  Whenever
possible, the activity patterns developed for the subgroups were
based on actual human activity data.  Because such data are
limited to a small number of studies initiated for other purposes,
many simplifying assumptions were made in constructing the activity
patterns.  For example, retired persons with limited mobility
were assigned to the outdoor microenvironment for fewer hours
than retired persons with full mobility.  Housewives with school-
age children at home were assigned to the transportation vehicle
microenvironment more often than housewives with no children at
home.  In each case, an attempt was made to construct an activity
pattern which was consistent with our intuitive expections of
what members of that subgroup would do on a typical weekday,
Saturday, or Sunday.  The resulting activity patterns are contained
in a separate document.2  Sample activity patterns are presented
in Appendix A.
                               3-5

-------
                TABLE 3-1.  DESCRIPTION AND APPORTIONMENT OF
                        ACTIVITY PATTERN SUBGROUPS
Age-occupation group
Students 18 and over



Managers and professionals





Sales workers




Clerical and kindred workers









Craftsmen and kindred
workers






Operatives and laborers







Subgroup
Code3
Oil
012
013
014
021

022
023

024
031
032
033
034
035
041

042-

043

044

045
046
051

052

053
054
055
056
061

062

063
064
065
066
Description
<30 min commute, 8 a.m. class
<30 min commute, 9 a.m. class
>30 min commute, 8 a.m. class
>30 min commute, 9 a.m. class
<30 min commute, single family
house
<30 min commute, others
>30 min commute, single family
house
>30 min commute, others
Indoor work, <30 min commute
Indoor work, >30 min commute
Outdoor work
Indoor and outdoor work
Traveling
Indoor work, 1st shift, <30 min
commute
Indoor work, 1st shift, >30 min
commute
Indoor work, 2nd shift, <30 min
commute
Indoor work, 2nd shift, >30 min
commute
Outdoor work
Indoor and outdoor work
Indoor work, 1st shift, <30 min
commute
Indoor work, 1st shift, >30 min
commute
Indoor work, 2nd shift
Indoor work, 3rd shift
Outdoor work
Indoor and outdoor work
Indoor work, 1st shift, <30 min
commute
Indoor work, 1st shift, >30 min
commute
Indoor work, 2nd shift
Indoor work, 3rd shift
Outdoor work
Work in motor vehicle
Percent
23
45
11
21

47
21

22
10
43
21
5
9
22

56

26

9

4
1
4

50

24
10
2
4
10

39

18
6
3
18
16
(continued)
                                     3-6

-------
TABLE 3-1 (continued)
Age-occupation group
Service, military, and
private household workers






Housewives


Unemployed and retired





Children less than 5



Children 5 to 17










Subgroup
Code3
081

082

083
084
085
086
091
092
093
101
102
103
104
105
106
111
112-
113
114
121

122

123

124
125

126

Description
Service, day time work, <30 min
commute
Service, day time work, >30 min
commute
Service, night time
Service, in motor vehicle
Military
Private household
No children at home
Some children <13
No children <13, some 13 to 18
Unemployed, job hunting
Unemployed, not job hunting
Disabled
Retired, full mobility
Retired, limited mobility
Retired, confined indoors
0 to 12 months
13 to 24 months
25 to 36 months
37 to 60 months
Elementary school, <30 min
commute
Elementary school, >30 min
commute, walk or bike
Elementary school, >30 min
commute, vehicle
High school, <30 min commute
High school, >30 min commute,
walk or bike
High school, >30 min commute,
vehicle
Percent

36

17
22
3
14
8
42
49
9
20
24
20
30
4
2
21
20
20
39

56

4

7
26

2

5
aFirst two digits indicate age-occupation  group,  third digit  indicates
 subgroup.
                                      3-7

-------
3.2  TYPES OF NEIGHBORHOODS
     Like the exposure district version of NEM, estimation of
human exposure to CO using the neighborhood version involves a
simple two-way trip model.  This model assumes that each cohort
is located either in its home NT or its work NT during each hour
of the year.  To use this model, it is necessary to determine
hourly ambient CO concentration in home NT's and work NT's.
     In reviewing air quality data available from CO monitors,
PEDCo found that information about neighborhood settings of moni-
tors is very limited, except for two National Air Monitoring
Systems (NAMS) monitors in each city where sites have been sur-
veyed in detail.  For most monitors, neighborhood data are limited
to the station-type  (ST) identifiers listed in the Storage and
Retrieval of Aerometric Data (SAROAD) system.  The ST identifiers
classify monitors according to geographic location and land use.
Geographic location categories include
     1.  Center City - core area of the city, not its incorporated
         limits
     2.  Suburban
     3.  Rural
     4.  Remote - far enough from any activity to measure back-
         ground levels.
Land use categories vary with geographic location as follows:
     1.   Center City and Suburban
          a.   Industrial - product-oriented establishments such
               as manufacturing concerns, utilities, mining, and
               graineries.
          b.   Commercial - service-oriented establishments such
               as retail establishments, shopping centers, gas
               stations, and laundromats.
          c.   Residential - because other areas are also used
               residentially, this category is used only in the
               absence of a dominating industrial or commercial
               influence.
          d.   Mobile - sites located in airports, truck or bus
               terminals, or an expressway cloverleaf.  Sites
               placed near parking lots would probably be better
               categorized as industrial or commercial.
                               3-8

-------
     2.
Rural
     3.
a.   Industrial  - same as center city  and suburban
     industrial.
b.   Commercial  - same as center city  and suburban
     industrial.
c.   Near urban  - sites located in  a rural area, yet
     close  enough to a major urban  center to be affected
     by the urban area.
d.   Agricultural - sites located near orchards, crop
     raising,  cattle, and sheep grazing.
Remote
     To simplify  the selection of monitors  to represent NT's, we
established NT's  to conform to the SAROAD ST's.   Since all  study
areas were urbanized,  ST's with rural  and remote categories were
dropped from consideration.  The remaining  eight ST's provide the
basis for eight NT's used for classifying cohorts in a population
exposure analysis (see Table 3-2).  These NT's are coded with
numbers 1 through 9 to facilitate their  use in computer files.
             TABLE 3-2.  NEIGHBORHOOD TYPES (NT's) AND  CODES
                       NT
                                     Computer
                                      code
             Center-city  residential
             Center-city  commercial
             Center-city  industrial
             Center-city  mobile
             Suburban residential
             Suburban commercial
             Suburban industrial
             Suburban mobile
             Not used or  "other"
                                       1
                                       2
                                       3
                                       4
                                       5
                                       6
                                       7
                                       8
                                       9
                                3-9

-------
     Once NT's are established, they can be used like exposure
districts within an activity pattern computer file.  District
codes 1 to 9 can be replaced with NT codes 1 to 9 in both the
definition of cohorts and in hour-by-hour assignments, as shown
in Figure 3-4.  The concept of "home" now refers to home NT
rather than home district, and "work" refers to work NT rather
than work district.  The 2-way trip model is retained; commuters
move from a home NT to a work NT in the morning and return in
the evening.  Consequently, activity patterns developed for the
neighborhood version of NEM use the same format as activity pat-
terns developed for the exposure district version.

3.3  ESTIMATION OF COHORT POPULATIONS
     There were several problems associated with establishing
cohorts and determining cohort populations in the neighborhood
version of NEM.  First, NT's did not provide a convenient
unit for assembling population data.  As shown on a previous map
(Figure 3-2) neighborhoods- occur in irregular patterns scattered
throughout a city.  Boundaries of neighborhoods do not correspond
to the boundaries of census tracts or any other unit used by
the Bureau of Census to organize population data.  Second, even
if it were theoretically possible to collect population data for
each NT, such an effort would require a significant expenditure
of resources which were not available for this purpose.  Conse-
quently, we made certain simplifying assumptions so that popu-
lation data that had already been assembled for the exposure
district version of NEM could be used in CO exposure analysis.
     Population data assembled for previous NEM analyses consisted
of the numbers of people in 12 A-0 groups who resided in each of
the exposure districts comprising each of four study areas.  The
first step in using this data was to determine the number of
people in each A-O group residing in center-city neighborhoods
and the number of people residing in suburban neighborhoods.
                              3-10

-------
Dummy for
city name
        Cohort
          ID
Day-of-week
   code
 AM-PM
 code
        12-hr
location assignments
CITYX.XXXH042W  11  H21H21H21H21H21H21431W11W11W11W11W11
CITYXXXXH042W  12  W11W11W11W11W31H21H21H21H21H21H21H21
Home'
NT
A-0 group-"

Pattern —
(Subgroup
 1-6)

Work NT —
         ^-Activity level  code (2nd hr)

        1—Microenvironment code (2nd hr)

          •Location assignment (2nd hr)
           Work(W) or Home(H)
           Neighborhood Types (NT)
Day-of-week codes:

1 = weekday
2 = Saturday
3 = Sunday
           Activity codes:

           1 = low activity level
           2 = medium activity level
           3 = high activity level
AM-PM codes:

1 = AM,  hrs  1-12
2 = PM,  hrs  1-12
           Microenvironment codes:

           1 = indoors at work
           2 = indoors, other
           3 = transportation vehicle
           4 = transportation, other
           5 = outdoors
 Figure 3-4.  Contents of cohort activity file,
                       3-11

-------
This was possible because each exposure district could be classi-
fied center-city or suburban by its geographic location and pre-
dominant land use pattern.  Classifications of exposure districts
were established in study areas used for previous NEM analyses,
as shown in Table 3-3.

       TABLE 3-3.  GEOGRAPHIC CLASSIFICATION OF EXPOSURE DISTRICTS

Study area
Chicago
Philadelphia
St. Louis
Los Angeles
Exposure districts
Center-city
1,2,3,4,5
1
1,2,3,4
1
Suburban
6,7,8
2,3,4,5,6
5,6,7
2,3,4,5,6,7
Using this breakdown, the A-0 group residence data could  be
classified into suburban or center-city, as listed in Table  3-4.
The next step was to further subdivide these general neighborhood
classifications into specific NT's.  For this breakdown,  it  was
necessary to use transportation data and certain assumptions
about movements of A-0 groups and their place  of residence.
     To simplify the problem of NT assignments, we assumed
that all people live in a residential neighborhood.  This might
be considered obvious by definition, but some neighborhoods  are
better characterized by predominant land uses.  For example,  some
people live in a predominantly industrial neighborhood.   However,
it is not important to use pure definitions of neighborhoods,  but
to differentiate high pollution residential neighborhoods from
presumably low pollution residential neighborhoods.  Using the
NT's listed in Table 3-2, this dichotomy can be approximated by
center-city residential  (high pollution) and suburban residential
(low pollution).  In St. Louis, for example, we approximated
the population living in center-city residential NT's by  the per-
sons living within neighborhoods classified center-city in Table
3-3, and the population living within suburban-residential NT's
                               3-12

-------
a
o
o
O
CD
31
CD
   OO
< LU
i— i ce
O O
1-H p
00 I—
LU CO
02 O
   Lu
_
=) O
a. I-H
o (—
Q. <
   O
a. >-i
=5 U.
O i->
cd oo
O co
   *£.
O -J
 i  (_>
 i
ro
CQ

00


•r~
jr
a
'a!
^5
to
r-
•r*
O.

O
CT
 •!->
C •!-
>
C -i-
OJ O
C_>
C
«3
.a
s-
J3
3
OO
I
i
C i-
CO O
o
c
1C
-0
S-
J3
3
CO
1
rtj ^K
+J 4->
C T-
a> cj
C_)
Age-occupation group
•3- co r-- ui i— i co r^ CM to  CO IT5 CM r— I CO IO
to tr> co o co co CM t-HOroo3r^«
vo»3-cr>  t— i •— i «*
^- r*. vo o co r»»r-» r^^-o coo^ co in coui IOCMCOCOCTN
f— i ^3- CM co •— i coco notor^io
cnco^> o <— i <— ito to^ftocnto
votrjf— i o «j- en tomr-.cnr^
«— i en i— i to CM to ^j-tocO'sr'3'
«— i
*a-coi — f-i rn o •— i CM^-mtoco
crir^io o o cncn t^i-HLnto^a-
U5COLTJ CO CO COCO i-ICOOr»-CM
LO«si-*a- cr> vo «— t «— i COOCM^OCO
r>.r-ii — en i— i to o cr> •— i to o
CM rHi-lr-H <-l CM CVJ I-H U)
OCMCO CM CM COCM UO O O* i-H *f
CMOCTI cn LO tovo cr> •— i r>« co «a-
^roco i — o r>.co toi— icr>«3-^-
ooco^j- «a- cn co p*scntotocn
i— i *a- ,— i to co CT> Lf>cncnr>.o
CM
cntooj i — ir> to cocnLOLOCM
cococo r» cn CM OCOCOLOCT*
toco«a- co r^ to co«tf-co^i-cM
un CTv LD to O lf)l OCMtOCOCM
CM t— i CM to to if> co cn •— i co cn
1— (
CM i— * i— i cn co co i— i cn cn LD cn
CO«TO cn «?r to oocr»cno
cnr^p*. m to co CMCMCocMr^.
cor».CM cn «3- toi i — or»-.oo
cococo to r^. o cor»*tr>ir>CM
CO CM .— i CM t— 1 i— 1 i— 1
(SI
L- -a
 ai m
2; •!- CO •!- 1- S»
\ ^ i. T3 > 3 S_
i^cn o c s- o)&.a;r^
+ i— S. "O M- (O  c -a a» i- -*: "O-cco
(^•r- or— s- coaj >,^: >, c c
•M in S fOO) EV-'i-S- O S-OJT3 O o3(ovicoi— s-i_
&.iOi — i. o. (o-(-  O •— i C\J
1— 1 1— 1 <— 1
CM
t— 1
^J-
CO
r«»
CM
A
to
«3-
CM
cn
•er
LD
T
A
r-H
0
CTV
to
co
o
to
o
CM
CO
cn
cn
IT)
o
r-N
cn
ir>
CM
<— 1
A
CM
CT>
o
o
CM
i-H
CO
ir>
<-H
1— 1
•3-
CM
to
CO
*3-
CO
*3-
o
r-
<-H
cn
<
o
H-
                                                      3-13

-------
by the number of people living in all areas classified suburban.
This allocation procedure allowed the population of each study
area to be divided into residential neighborhood categories which
are also geographic areas.  The resulting breakdown of population
is shown in Table 3-4.
     We next determined work NT's by making simple, intuitive
assumptions about the work environment of A-0 groups.  For example,
we assumed that children under 5 years did not go to work, but
stayed at home.  We assumed that children 5-17 years of age went
to "work" at a school, but that the school was located in the
same NT as their residence.  We assumed that operatives and
laborers worked primarily in industries located within an indus-
trial NT.  The full roster of assumptions is listed in Table 3-5;
these statements allowed the work NT of each A-O group to be
classified residential, commercial, or industrial, but they did
not determine whether NT's are center-city or suburban.  For this
purpose, transportation data were used in the form of home-to-
work trip tables.
     As documented in other reports,3'" home-to-work trip tables
were previously developed for each study area based on data pro-
vided by regional transportation planning agencies.  These trip
tables may be visualized as an n x n array which lists the number
of trips taken during a typical day from each transportation zone
of a city to every other zone.  In previous NEM analyses, arrays
were reduced in size by aggregating transportation zones into
exposure districts.  The resulting smaller arrays showed the
number of trips from each exposure district to every other ex-
posure district.
     In the CO analysis, each study area trip table was further
condensed into a 2 x 2 array, because the only geographic
elements used were suburban and center city NT classifications.
In other words, all exposure districts were aggregated into either
a suburban super-district or a center-city super-district.  The
resulting collapsed trip tables are listed in Table 3-6.  As in
previous NEM analyses, we assumed that the fraction of all trips

                               3-14

-------
         TABLE 3-5.   ASSUMPTIONS CONCERNING WORK NT'S OF A-0 GROUPS
      A-0 Group
            Assumptions
 1.   Students 18+
 2.   Professional and administrative

 3.   Sales workers

 4.   Clerical workers

 5.   Craftsmen

 6.   Operatives  and laborers

 7.   Farmers

 8.   Service, military and household

 9.   Housewives
10.   Unemployed  and retired
11.   Children under 5
12.   Children 5  to 17
Work NT is same as home NT.
All work in a commercial neighbor-
hood; some work in suburban areas,
others in center-city.
Work neighborhood is suburban-
commercial or center-city commercial
Work neighborhood is suburban-
commercial or center-city commercial
Work neighborhood is suburban-
industrial or center-city industrial
Work neighborhood is suburban-
industrial or center-city industrial
There are no farmers in urbanized
area.
Work neighborhood is same as
home.
Work neighborhood is same as home.
Work neighborhood is same as
home.
Work neighborhood is same as home.
Work neighborhood is same as home.
                                   3-15

-------
     TABLE 3-6.  COLLAPSED HOME-TO-WORK TRIP TABLES, EXPRESSED AS
               NUMBER OF TRIPS AND FRACTION OF TRIPS
Study
area
Chicago
Philadelphia
St. Louis
Los Angeles
From home
Center-city
Suburban
Center-city
Suburban
Center-city
Suburban
Center-city
Suburban
To work
Center city
3,182,820(0.935)a
524,260(0.843)
270,517(0.718)
405,867(0.353)
114,627(0.793)
76,977(0.435)
246,970(0.570)
581,480(0.133)
Suburban
221,720(0.065)
97,300(0.157)
106,507(0.283)
743,723(0.647)
29,920(0.207)
99,957(0.565)
1,861,400(0.430)
3,800,720(0.867)
 Number in parenthesis is fraction of trips.

taken from one super-district to  another could be used to  repre-
sent trip distributions  for  all A-0 groups.  These fractions  were
used in a modified  2-way trip model to calculate number of persons
in each cohort, as  follows:
                                Tj __ in i
                                                              (3-1)
        H,A,S,W
                =   P
                     H,A
                                  H
where
       H
       A
       S
       w
       P
        H,A,S,W
        H,A
        "H-W
= home NT
= age-occupation category
= subgroup specified  by  activity pattern
= work NT
= population of a.  cohort which is defined by the
  subscripts
= population of an age-occupation category in a
  home NT  (center-city residential or suburban
  residential)
= fraction of persons in A-0 group allocated to
  subgroup
= number of trips  from a home super-district to
  a work super-district
= total number of  trips  by all A-0 groups from a
  home super-district to both super-districts
   (suburban or center-city).
                                3-16

-------
     Calculation of cohort populations may be illustrated by an
example.  Using the modified trip model, one cohort may be defined
as those persons residing in suburban-residential NT in St. Louis
who are managers or professionals (A-0 Group 2) , who follow typi-
cal activity patterns of subgroup 2, and who are located in center-
city commercial NT during working hours.  This definition is con-
sistent with the list of assumptions in Table 3-5.  This table
shows that 70,853 people live in suburban residential NT's in
St. Louis and belong to A-0 Group 2.  From Table 3-1 we find that
the fraction of A-O Group 2 belonging to subgroup 2 is 21 percent.
From the collapsed trip table, Table 3-6, we find that 110,046
people out of a total of 397,200 people from suburban residential
NT's go to a center-city NT for work.  Thus, we may calculate the
cohort population as follows :
     P5,2,2,2 = (^O, 853) (0.21)           = 4,122 persons.
Populations of all cohorts were calculated similarly; a complete
list of cohorts and estimated cohort populations is provided in
Appendix B.
     Once methods for representing people movement were estab-
lished, and methods were devised for calculating numbers of
people following various movement patterns, there remained the
task of estimating air quality levels to which cohorts were
exposed.  Development of air quality data is the subject of sub-
sequent sections .

3 . 4  REFERENCES
1.   Memorandum from Ted Johnson, PEDCo Environmental, to Thomas
     McCurdy, Strategies and Air Standards Division, U.S. En-
     vironmental Protection Agency, Research Triangle Park,
     North Carolina 27711, April 6, 1982.
2.   Ted Johnson, Activity Patterns for NEM Analysis of Carbon
     Monoxide Exposure, prepared by PEDCo Environmental, Inc.,
     for Strategies and Air Standards Division, Office of Air
     Quality Planning and Standards, U.S. Environmental Protection
     Agency, Research Triangle Park, N. C., October 1982.

                               3-17

-------
3.    Ted Johnson and Roy Paul,  The NAAQS Exposure Model (NEM)
     and Its Application to Nitrogen Dioxide,  prepared by PEDCo
     Environmental, Inc., for Strategies and Air Standards
     Division,  Office of Air Quality Planning and Standards,
     U.S. Environmental Protection Agency,  Research Triangle
     Park, N. C.,  May 1981.

4.    Ted Johnson and Roy Paul,  The NAAQS Exposure Model (NEM)
     and Its Application to Particulate Matter,prepared by PEDCo
     Environmental,Inc.,for Strategies and Air Standards
     Division,  Office of Air* Quality Planning and Standards,
     U.S. Environmental Protection Agency,  Research Triangle
     Park, N. C.,  August 1981.
                               3-18

-------
                            SECTION 4
                 PREPARATION OF AIR QUALITY DATA

     NEM requires representative outside air quality data for
each neighborhood type in the form of a complete year of hourly
average values.  This section describes the procedures used for
selecting appropriate data sets, for validating these data, and
for filling in missing values.

4.1  SELECTION OF REPRESENTATIVE DATA SETS
     To simplify the computer simulation, air quality in all
neighborhoods classified as belonging to a specific NT was assumed
to be a linear function of air quality monitored at a single rep-
resentative monitoring site.  Consequently, for each study area
we had to select one monitoring site per NT.  We originally con-
sidered using the 8 NT's listed in Table 3-1; however, evalua-
tion of data availability indicated that there was an insufficient
number of center-city mobile and suburban mobile sites to include
these NT's in the exposure analysis.  The remaining NT's were:
     o    center-city residential (CR),
     o    center-city commercial (CC),
     o    center-city industrial (CI),
     o    suburban residential  (SR),
     o    suburban commercial (SC),  and
     o    suburban industrial (SI).
     Table 4-1 lists data sets selected to represent these NT's.
These data sets were selected according to the following procedure.
     (1)   Data for the same year were used for all sites in a
          study area.  The year was selected to maximize the
          number of NT's that could be represented by acceptable
                              4-1

-------
       TABLE 4-1.   DATA  SETS  (mg/nT) SELECTED FOR ANALYSIS OF POPULATION
                 EXPOSURE  TO  CARBON MONOXIDE  IN FOUR CITIES

Study area
and year
Chicago (1979)





Los Angeles (1977)





Philadelphia (1978)





St. Louis (1978)







NT
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI


ST
CC
CC
CI
SR
SC
SC
CR
CC
SI
SR
SC
SI
CC
CC
SI
SR
SR
SI
CR
CC
CI
SR
SC
CI


SAROAD code
141220039F01
141220040F01
141220026H01
140780002G01
145680001G01 '
147160005G01
056400003F01
053900001101
050900002101
058720001101
050230001101
050900002101
397140026H01
397140026H01
397140022H01
397140004H01
397140004H01
397140021H01
264280007H01
264280064H01
264280061H01
260200002G01
261040001G01
264280061H01
Probe
height
(ft)
11
11
26
15
15
39
32
20
15
15
15
15
13
13
13
17
17
13
35
?
12
13
14
12

1-hour
obs.
6172
8392
7615
8664
7662
8425
7921
7940
7775
7777
7888
7775
7773
7773
8088
8193
8193
8485
5944
7466
7539
6817
7010
7539

Geo.
mean
1.8
2.9
2.6
1.3
1.4
1.4
1.8
2.6
3.9
3.4
3.0
3.9
2.2
2.2
1.3
0.8
0.8
1.9
?
5.4
2.5
2.6
2.6
2.5
99th
per-
centi le
7.0
15.8
9.8
6.7
7.0
6.4
9.2
18.4
17.3
13.8
13.8
17.3
10.4
10.4
5.8
5.8
5.8
8.1
6.4
10.7
11.8
9.3
9.8
11.9


Notes
b
a



b

a
b
a
a

a,b
a
a9b
a
a,c

a,d
a



b
a:
b:
c:
d:
detailed site location description  available  at MDAD.
site with matching ST not available.
site with matching ST not appropriate.
geometric mean not calculated  because  data  completeness  <75  percent.
                                     4-2

-------
          data sets.  A data set was considered acceptable if it
          contained at least 5839 observations  (66% completeness).
     (2)  When possible, an NT was represented by a data set with
          a corresponding ST (e.g., the center-city commercial NT
          for St. Louis was represented by site 264280064H01,
          which has a center-city commercial ST).
     (3)  If two or more sites with the required ST were available,
          the site judged most representative of the NT was select-
          ed.  In 11 cases, degree of representation was determined
          from photographs of site surroundings and detailed local
          land use descriptions on file in the Monitoring and Data
          Analysis Division (MDAD), EPA.  Summary statistics,
          probe location, and data completeness were also con-
          sidered in selecting sites.
     (4)  If no site with the required ST was available, the site
          with the most similar ST was selected.  Whenever pos-
          sible, the MDAD photographs and land use descriptions
          were used in selecting alternate sites.
     In six cases, NT's and ST's do not match because a site with
the appropriate ST was not available.  For the Chicago center-city
residential NT, we selected a center-city commercial site with
low geometric mean and 99th percentile values.  For the Chicago
suburban industrial NT, we selected a suburban commercial site
located in Skokie, a town near Chicago generally characterized
as industrial.  In Los Angeles, we used a suburban industrial
site (the only industrial site in the study area)  to represent
the center-city industrial NT.   A suburban commercial site in
Afton was used for the St. Louis suburban industrial NT because
Afton is located near an industrial area.  The center-city re-
sidential NT in Philadelphia was represented by a center-city
commercial site which photographs revealed to be surrounded by
apartments.  Similarly, the center-city industrial NT in
Philadelphia was represented by a suburban industrial site because
photographs suggested that the site is actually located in a
heavily urbanized area.
                              4-3

-------
     In one case,  a site with the correct ST was not used because
photographs revealed a site with a different ST was more appropriate,
The only suburban commercial site in Philadelphia  (397140024H01)
is surrounded on all sides by Philadelphia International Airport.
Site 397140004H01, although labeled suburban residential, was used
instead because it had commercial establishments nearby.

4.2  VALIDATION OF AIR QUALITY DATA
     Air quality data sets containing erroneous values can bias
the results of exposure studies, especially if the errors occur
as extreme values.  To ensure good data quality, the data sets
listed in Table 4-1 were screened for anomalous values using
three methods:  (1) the gap test, (2) the patterns test, and  (3)
visual inspection.
4.2.1  The Gap Test
     The Monitoring and Reports Branch (MRB) of MDAD has developed
a standardized data review program called the MRB Validation Re-
port. :  This program uses two different concepts to identify
anomalies  (unexpected data patterns) in hourly data sets.  The
first approach, called the Gap Test, is a statistical analysis
of data over a 1-month period.  This analysis assumes that the
data can be modeled reasonably well by a smooth probability dis-
tribution curve.  Two exponential curves are fit to the data
using the 50th and 95th percentiles of the data for one fit and
the 50th and 99.9th percentiles for the other fit.  Both fits
emphasize the upper tail of the distribution.  All data values
are arranged in order of magnitude and the program examines "gaps"
in the data, i.e., large differences between succeeding ordered
values.  Using the fitted distribution functions, the program
calculates the probability that a gap of magnitude "x" could be
obtained by chance.  A gap that is greater than would be expected
from the underlying assumptions is flagged as a data anomaly.
The strength of the Gap Test is that the criteria for identify-
ing anomalies are based on an analysis of the data set itself.

                               4-4

-------
Its weakness is that the assumed distribution is not always ap-
propriate; consequently some false failures may occur.
     In this study, the output from the gap test was evaluated to
determine whether or not the flagged data were plausible from
other points of view.  It was desired that data not be rejected
unless a strong case for error could be established.  Flagged
data were usually retained if the tested data set contained a
large number of missing values, only one of the two fitted curves
indicated an anomaly, or there were more than five observations
above the gap.  Data flagged by the gap test were more likely to
be rejected if they were flagged by other tests.
4.2.2  The Pattern Test
     The second test in the MRS Validation Program is called the
Pattern Test.  It is composed of 5 subtests carried out for each
24-hour period, as follows:
     (1)  High-value test - the test flags an hourly value that
exceeds a predetermined limit.  For carbon monoxide, the criterion
depends on whether or not the measurement was taken during rush
hours (7-11 a.m. or 4-9 p.m.).  During rush hours the limit is
66 ppm and at other times it is 44 ppm.
     (2)  Adjacent-hour difference test - this test assumes that
data take the form of an auto-correlated time series, i.e., a
large jump or drop in the values within 1 hour is not expected.
If there is a jump or drop greater than 22 ppm, the suspect value
is flagged.
     (3)  Dixon-ratio test - this is a statistical analysis of
the highest and lowest value found during the day.  If A is the
difference between the two highest values and B is the range of
values for the day, then the Dixon Ratio is A/B.  If this ratio
is statistically significant, the suspect values are flagged.
     (4)  Spike-test - the differences between a suspect hourly
value and the preceding and following hourly value is measured.
If either of these two differences is greater than 20 ppm,  or if
                              4-5

-------
the suspect value is 500 percent greater than either of the
adjacent values, the suspect value is flagged.
     (5)  Consecutive high values - it is unusual for a series
of hourly measurements to remain at a high level.  If four con-
secutive hourly values are greater than 40 ppm, the data are
flagged.
     PEDCo evaluated the results of applying MRB Validation Pro-
gram to SO- data in four cities.2  This analysis led to the
development by PEDCo of a revised validation program.  This new
program (denoted here as MRB-2) contains improvements in the
pattern tests and an option for graphical output of flagged data.
The Dixon Ratio Test was enhanced for all pollutants in order to
incorporate the recommendations of a recent EPA report.3  In the
new version, the formulas for testing the high and low hourly
values from each 24-hour period vary according to the number of
hourly measurements recorded during the period.  The spike test
was enhanced so that an individual hourly value is compared to
the two (rather than one) preceeding values and the two succeeding
values.  In addition, MRB-2 incorporates a preliminary screening
test, so that if all the values in a 24-hour period are low,
the pattern tests are skipped.  By passing over data that do
not require testing, the computer program runs more quickly.
     The strength of the pattern test program is that it uses
more than one test to identify possible anomalies.  Its weakness
is that the criteria for identifying anomalies is predetermined
for all tests except the Dixon Ratio Test.  The pattern test
program could be improved by adjusting test criteria according
to the data reported by each individual monitoring site.
4.2.3  Visual Inspection
     George Duggan of SASD has developed a computer program which
will plot all hourly values in a year of data on a single graph.
Such data plots are useful in identifying unusually high values
and long strings of identical values.  All data sets used in the
                               4-6

-------
CO exposure analysis were plotted using this program and reviewed
by PEDCo.  Values which appeared anomalous were flagged for
further investigation.
4.2.4  Results of Data Screening
     CO levels recorded at monitoring sites can be drastically
altered by short-term changes in local emissions.  Exhaust from
a delivery truck idling near a monitor may result in a high read-
ing which appears totally inconsistent with other CO values
recorded that day.  However, this reading should not be excluded
from the exposure analysis since it represents an exposure situ-
ation which may occur frequently throughout the study area.  Con-
sequently, we decided to retain flagged data unless they appeared
to be the obvious result of instrument malfunction or transcription
error.  As discussed below, few of the flagged data fell into these
categories.
     Table 4-2 lists data anomalies flagged by one or more of the
screening methods.  Of the anomalies detected, most were flagged
by the gap test or visual inspection.  No values were flagged by
the high value test, the spike test, or the consecutive hour
test.
     Eight data anomalies were identified by visual inspection.
In two cases, Chicago site 145680001G01 and St. Louis site
261040001G01, long strings of values equal to 0.1 ppm or zero
were found.  We assumed in both cases that these values were
incorrectly entered in place of missing values and removed them
from the data.  The other six anomalies involved unusually high
values.  Analysis showed that the flagged high values occurred
simultaneously at the two Chicago sites and at the two Philadelphia
sites.  We considered this sufficient corroboration to retain
these values.
     We also retained the values flagged by visual inspection at
sites 260200002G01 and 264280064H01 in St. Louis.  In the first
case, four large values (20.8, 26.3, 26.1, and 26.6 ppm) were pre-
ceded and followed by values of 3.1 ppm.  This episode was unusual

                               4-7

-------
00
-
03
O
UJ
uo
UJ
CM
 I
LU
_J
03





CO
3
CO
U
o
&»
Q.
C
c
CO
CO
t.
U
oo




















































IO O
3 CO
to a
•i— tO
> c
-l~
E O
O -i-
X -M
•i- ^
-a 
3 3 V} to
i— i— CO
(O (Q Cn 3
> > E i—
O *O
"~"
en 01
r*^ p***
i i en
CTV CT* I""**
CM CM 1
1 1 CM
CM CM
r— 1 i— 1
on o
00 OO

1— 1 1— 1
o o
CD CD
CM «-H
O 0
0 0
O O
oo oo
P*""* {Q
O Lf>
^" *st*
i-H r— t




0
en
 >

CO CO
en en
&- s-
(O fO

r^ r-N.
p**» r^**
i i
p*** ?**•*
O <-H

CM CM
0
O

r- 1
O

i-H
O
O
o
o
a*
m
urt
o

CO
pMB
CO
en
s:
•^^

to
O
— 1


X X





X










X X

EH
CL CL
CL
O
O •-«
!"•!
A]A
O. to
ia to co
en co 3
3 i —
. CO i — to
CD tO >
C- >
(B CM

oo oo
r**>» P*H,
CO 1 1
f-. IT) ID
1 CM CM
IO 1 1
i— 1 i— 1
O
00 CJ Ctf
o
1—4 I— 4 t— 1
O O 0
3: z z
t-l CM UD
CM CM CM
000
CD O O

1—4 r—< 1— 1
F-* p** r^
cr* cy^ o^
ro co ro
^>
r"
a_


X XX










X





XX X

E E
CL O CL
a. s_ CL
4- CO
O O Nl LO
CM «-H
cn s_
A E (C A
Q. -i— CO
 CO
3 to (/) 3
CO i— CO i—

i. > E i— >
(O O tO

00 00
^^ QQ ^^
co i r-; i
r^1* ^^ i o^
1 CM CM O
<£> 1 i— I 1
(^ «*
Q£ C_> CJ
OO 00 O

t— 1 «— 1 l-H
O 0 0
CD CD Z
CM i— i ^>
O O 
-------
enough to cause the day to be also flagged by the gap test using
50th and 95th percentiles and the adjacent hour test.  However,
the day was not flagged when the gap test was repeated using the
50th and 99.9 percentiles.  In the other case, two large values
(15.2 and 16.5 ppm) were preceded and followed by smaller values.
These data were also flagged by the 50/95 gap test but not by the
50/99.9 gap test.
     The 50/9*5 gap test also flagged data at site 397140021H01 in
Philadelphia and at site 260200002G01 in St. Louis.  Since these
data were not flagged by any other test, including the 50/99.9
gap test, they were accepted.
     Three days of data were flagged by the Dixon Ratio test.
Because analysis of these data indicated that the assumption of
normality was not valid, we repeated the Dixon Ratio test using
the logarithms of the recorded values as recommended by Nelson,
et al.3  This time no values were flagged by the test.  We con-
cluded there was no probable cause under an assumption of log-
normality for rejecting the data.
     In summary, we investigated 12 cases of data flagged by
various screening procedures.  We retained the anomalies in
all but the two cases where visual inspection had identified long
strings of values equal to 0.1 ppm or zero.  These values were
removed from the data sets.  The next section describes the
methodology used to fill in these and other missing values.

4.3  SIMULATION OF MISSING VALUES IN HOURLY AVERAGE CO DATA SETS
     NEM requires air quality data sets with values for every
hour of the year.  Since absolutely complete data sets were not
available, gaps were filled in using a time series model developed
by Johnson and Wijnberg.1*
                               4-9

-------
4.3.1  The Time Series Model
     A complete year of hourly average data takes the form of  a
time series x,, x_ , ..., x  , . .., x  where n = 8760.  We can fit
             .1.   ^        u        n
this series exactly by the model
                        4380
               x.  = x +  Z   R.cos(cu.t + 9.)                 (4-1)
                t       j=l   :     D     3
where x is the arithmetic mean of the series, R. and 9. are
amplitude and phase angle values determined by Fourier analysis,
and CD. = 2iTJ/8760.  Omission of one or more of the 4380 Fourier
cosine terms will yield an approximate fit.  Because Fourier
cosine functions are orthogonal and because the contribution of
each cosine function to the representation of the original time
series is proportional to its amplitude R., we can provide a least
squares fit to the original time series with m cosine terms by
using the cosine terms with the m largest amplitudes.  We denote
                                           A
each term of this estimated time series as x. where
               *         m
               x  = x +  Z R.cos  (oj.t + 9.)                  (4-2)
                r       i=l X
and R., CD., and 9. are the parameters of the Fourier term having
the ith largest amplitude.  For convenience, we will refer to  the
m Fourier terms in Equation 4-2 as the essential cyclical compo-
nent  (ECC).
                                                   A
     The differences between the x  series and the x  series
comprise the d  series, i.e.,
                         A
               dfc = xfc - xt .                                (4-3)
                            A
We can define how well the x.  series represents the x  series
by the goodness of fit statistic
                               4-10

-------
                                                             (4-4)
As m increases, r  increases and the goodness  of  fit improves.
Note that r  = 1 when m = 4380.
     If the x.  series exhibits autocorrelation, the  d,  series is
likely to exhibit autocorrelation.  One means  of  characterizing
a series which exhibits autocorrelation is  to  use an AR(p)  process
(i.e., an autoregressive process of order p) .   In this  case,  each
dfc term can be expressed as
          dt = at + , , <}>-, ..., 4>  can  be obtained by first  esti-
                   -L   ^        jp                       .A.
mating each autocorrelation p,  , using  the relationship  p, = r,
where
               c,
          r,  = -£          k = 1, 2, ..., p                 (4-6)
           K   GO
and
                        8760
          c,  = (1/8760)   Z  (d,  - d) (d. .   - d) .             (4-7)
           k                   t       t-k
From these estimates, the Yule-Walker estimates of  the  autoregres-
sive parameters can be obtained. s
     Autocorrelation of the d. series will decrease as  m increases
since an increasing portion of the x. series autocorrelation is
explained by the cosine functions.  We assumed that most of  the
autocorrelation in the data corresponding to k >_  3  would be  con-
tained in the ECC we selected and that an AR(2) process would
suffice to characterize the d  series.  In this case,
                              t
               r, (1 - r_)
                   - rl
                        .                                    (4-8)
                               4-11

-------
               r  - r 2
                2    1
               —	±,                                     (4-9)
and

          a   = cQ(l - ^iT, - (Jj-r-) .                         (4-10)

This AR(2) process represents a stationary time series if it meets
certain conditions described by Box and Jenkins.5
     A theoretical AR(2) process will have non-zero values of  p,
for k > 2 that decrease gradually according to the relationship

          pk = *lpk-l + *2pk-2     k > °                     (4"11}

until a point is reached where the distribution of r,  is approxi-
mately normal with mean zero and standard error
          a  (rk) ~   E  (I + 2pl  + 2p2)>                    (4-12)

The values of p. and P2 are estimated by r, and r-.  No more  than
5% of the values of r,  for large values of k should deviate from
zero by more than two standard errors.7
     If we can select an ECC such that the autocorrelations in
the d  series corresponding to k > 2 are consistent with Equations
4-11 and 4-12, then an AR(2) process should suffice to character-
ize the d.  series.  To select this ECC, we can start by determining
the d  series that corresponds to m = 1.  We then calculate r,  for
values of k  that are likely to be significant.  These include k =
3, 4, 6, 8,  12, 24, and 168 for typical air quality data.  If the
r, values are not consistent with Equations 4-11 and 4-12, we
determine the d. for m = 2 and repeat the analysis.  We continue
increasing m until the r,  values for k > 2 meet our criteria.   At
this stage,  we should have a combination of ECC and AR(2) process
that will adequately characterize the data.
                               4-12

-------
4.3.2  Initial Treatment of Missing Values
     Fourier analysis cannot be applied to a time series if one
or more values are missing.  If air quality data to be analyzed
are incomplete, some method of estimating missing values must be
used prior to analysis.  Bloomfield8 recommends replacing each
missing observation by a linear combination of its neighbors if
most of the missing values tend to occur in small, isolated
groups.  If a gap containing b-1 missing values occurs between
values x  and x  , , each missing value x  can be estimated by
        cl      3, ~rD                      u
linear interpolation as

          *t = xa + H(t - a)(xa+b - xa}-                    (4"13

However, linear interpolation may not yield reasonable estimates
of missing one-hour values for large gaps, especially if they are
bounded by extreme values.  In these cases, the arithmetic mean
(x) may be a better estimate of each missing value.  Inspection
of data sets to be used in the CO population exposure analysis
suggested that the arithmetic mean should be used to fill in gaps
whenever gap length exceeded 72 hours and/or one of the boundary
values exceeded the arithmetic mean by more than two standard
deviations.  In other cases, linear interpolation produced reason
able results.
4.3.3  Procedure for Simulating Missing Values
     The time series model described above was the basis for the
following procedure for simulating missing values.
     (1)  The mean and standard deviation of each data set were
          calculated.
     (2)  Gaps with lengths exceeding 72 hours and/or with bound-
          ary values exceeding the arithmetic mean by more than
          two standard deviations were identified.  These gaps
          were filled in with the arithmetic mean.
     (3)  Linear interpolation was used to fill in the remaining
          gaps.

                               4-13

-------
     (4)   Fourier analysis was applied to the augmented time series
          created in steps (2) and (3) .
     (5)   An ECC was constructed which contained the smallest
          number of cosine terms required to produce a d. series
          consistent with Equations 4-11 and 4-12.
     (6)   The d,  series was represented by an AR(2) process by
               c                                          /•.
          using Equations 4-8, 4-9, and 4-10 to determine $, ,
          /S       /N                                        -L
          <$ ~ , and a .
           ^       a
     (7)   An a  series was formed by dividing each term in a
              w                   s»
          N(0,l)  random series by CT .  For consistency, the same
                                   31
          random series was used in each case.
     (8)   Missing d. values were simulated using the relationship


               *t = *A-1 + *2St-2 + V                '   (4"14)

     (9)   Missing x. values were filled in using the model
               ~    _    m                   *
               x.  = x +  t R.cos(o).t + 6.) + d.              (4-15)
                t           i     i     i     t
          to create the final augmented data set.
Note that the final simulation  (Equation 4-15) uses information
concerning the cyclical, autoregressive, and stochastic properties
of the time series which are omitted in the initial estimates made
in steps (2) and  (3) .
     Figure 4-1 shows a data set which is missing 1750 values.
Figure 4-2 shows the augmented data set after the initial simu-
lation of missing values [steps  (1) through (3)].  Figure 4-3
shows the final augmented data set with missing terms filled in
by adding an appropriate AR(2) process to the most significant
Fourier cosine functions [steps  (4) through (9)].  This two-step
process simulates missing terms which are consistent with both
the cyclical and the random character of the known values.
                               4-14

-------
OO • OS
                                                                                                                                                  a
                                                                                                                                                 •M
                                                                                                                                                  s_
                                                                                                                                                  O
                                                                                                                                                  *-\
                                                                                                                                                  OJ
                                                                                                                                                  (3   •
                                                                                                                                                 •M cn
                                                                                                                                                  fO i-
                                                                                                                                                 -Q 3
                                                                                                                                                     O
                                                                                                                                                  CO _!
                                                                                                                                                 •o
                                                                                                                                                 •r—   •
                                                                                                                                                  X 4->
                                                                                                                                                  O OO
                                                                                                                                                  c
                                                                                                                                                -2 =
                                                                                                                                                 o o
                                                                                                                                                 -•a o
                                                                                                                                                 ;.. i—*
                                                                                                                                                 "O C
                                                                                                                                                 O O
                                                                                                                                                    o
                                                                                                                                                 00  t-H
                                                                                                                                                 en 
                                                                                                                                                 S» -f-
                                                                                                                                                 QJ  t/i
                                                                                                                                                 s-  o
                                                                                                                                                 3 -t-i
                                                                                                                                                 O •!-
                                                                                                                                                 O)
                                                                                                                                                 3
                                                                                                                                                 CD
                                                               4-15

-------
OQ- 02
DO' 9
                                                                                                  o
                                                                                                  CD
                                                                                                  i—i
                                                                                                  O
                                                                                                  o
                                                                                                  o
                                                                                                 CM


                                                                                                  O
                                                                                                  f "  i y'» '

                                                                                                  O  CO

                                                                                                  £  oo
                                                                                                 re  o
                                                                                                 •O -r-
                                                                                                   ., J

                                                                                                 C'  rj
                                                                                                 X  =

                                                                                                 O -f-
                                                                                                 C  "5

                                                                                                 O
                                                                                                 c •<-
                                                                                                 O 4J
                                                                                                 J2 •!--


                                                                                                 fC -^
                                                                                                 co
                                                                                                 r~
                                                                                                 en
                                                                                                 >— i

                                                                                                 a
                                                                                                 cn
                                                                                                 fO
                                                                                                 i-
                                                                                                 O

                                                                                                 as
                                                                                                CM
                                                                                                 I

                                                                                                •=3-


                                                                                                OJ


                                                                                                3

                                                                                                O1
    00  _
amu A
                                                oo*a
                                          i r  ~idna
                                             i  id
                                                                GO ' O
                                           4-16

-------
CO"  02
                                                                                              O
                                                                                              O
                                                                                              o
                                                                                              CM
                                                                                              to
                                                                                                  CJ
                                                                                              S_  ra
                                                                                              o  ->
                                                                                              4-J
                                                                                              •r-  cn
                                                                                              c  c
                                                                                              o -.-

                                                                                                  I/I
                                                                                              5- T-
                                                                                              o  =
                                                                                              05  O
                                                                                              +J
                                                                                              to  c
                                                                                              T3  O
                                                                                              O  £
                                                                                              C -r-
                                                                                              O  «/i

                                                                                                 P-™
                                                                                              C  (O
                                                                                              o  c
                                                                                             .a •!-
                                                                                                  O)
                                                                                             oo
                                                                                              c; -
                                                                                              C1
                                                                                              
-------
4.4  REFERENCES

1.    Screening Procedures for Ambient Air Quality Data, publication
     no. EPA-450/2-78-037, U.S. Environmental Protection Agency,
     Research Triangle Park,  N.C., July 1978.

2.    Roy A. Paul, R. B. Faoro, W. F. Hunt, "Screening SO- data
     anomalies from four test cities," Technical Conference
     Transactions,  American Society for Quality Control,
     Milwaukee, Wisconsin, 1981.

3.    A. C. Nelson,  D. W. Armentrout, and T. R. Johnson, Validation
     of Air Monitoring Data,  publication no.  EPA-600/4-80-030,
     U.S. Environmental Protection Agency, Research Triangle
     Park, N.C., June 1980.

4.    Ted Johnson and Luke Wijnberg, "Time series analysis-of
     hourly average air quality data," paper no. 81-33.5, pre-
     sented at the 74th Annual Meeting of the Air Pollution
     Control Association, Philadelphia, Pennsylvania, June 21-26,
     1981.

5.    G.E.P. Box and G.M. Jenkins, Time Series Analysis;  Fore-
     casting and Control, Holden-Day, San Francisco, 1976,  p. 55.

6.    Box and Jenkins, p. 58.

7.    Box and Jenkins, pp. 35  and 59.

8.    Peter Bloomfield, Fourier Analysis of Time Series;  An Intro-
     duction, John Wiley and  Sons, New York,  1976, p. 245.
                              4-18

-------
                            SECTION 5
   SIMULATION OF AIR QUALITY EXPECTED AT FIXED MONITORING SITE
            UNDER ALTERNATIVE CARBON MONOXIDE STANDARDS
     The augmented data sets described in Section 4.3 were assumed
to represent the current status of air quality at a representative
monitor in each neighborhood type  (NT) .  To represent air quality
expected under the current NAAQS for CO and under proposed CO
standards/ these data sets were adjusted using a modified form of
the EPA rollback model.

5.1  THE ROLLBACK MODEL
     Each augmented data set is a time series containing 8760
hourly values, i.e.,

          X-,, ^2 ' •••' ^t ' •••' ^8760*
We assumed that y. ,  the difference between each x,  and an assumed
constant background level x., would increase or decrease in propor-
tion to the changes in emissions dictated by a given air quality
standard, as long as x  > x,  .  If x,  < x, , we assumed x  would not
be affected by changes in emissions.   We further assumed that all
emissions would change in proportion to the change in emissions re-
quired to bring the most polluted NT in the study area into
compliance.
     Air quality in each NT was characterized by air quality
indicators (AQI's)  which varied according to the form of the air
quality standard.  We assumed the most polluted NT to be the one
with the largest AQI with respect to the standard being considered.
     To simulate the air quality expected in each NT under a
standard, we created an adjusted data set
                              5-1

-------
where
              = PYt + xfa                                     (5-1)
and p is a rollback factor.  Consistent with the assumptions
above, values of p were calculated according to the formulas

               xs  " xb
          P =  XS  .  °    if Yt > 0                         (5-2)
               xmax  xb        t
and
          p = 1            if yt <_ 0  ,                       (5-3)
where x  is the highest concentration permitted by the standard
       5
for the stated averaging time and x    is the corresponding
AQI for the most polluted NT.  The rollback model assumes  rea-
sonable estimates of x    and x,  are available; Sections 5.2 an
                      max      b
5.3 describe how these estimates were developed.
5.2  AIR QUALITY INDICATORS
     Use of the rollback model to adjust air quality data requires
parameters for characterizing data which are related to the  form
of each standard under consideration.  At the time of the CO pop-
ulation exposure analyses, four types of parameters were consid-
ered for proposed standards:  the daily maximum  1-hour value ex-
pected to be exceeded once per year, the daily maximum 1-hour
value expected to be exceeded five times, the daily maximum  8-
hour running average expected to be exceeded once, and the daily
maximum 8-hour running average expected to be exceeded five
times.  Reasonable estimates of the 1-hour parameters can be made
by fitting a cumulative distribution  [F(x)l to the daily maximum
values of an augmented data set and then calculating the values
/•v          s*
b, ,,.c and bc -£K such that
 1,365      5,365

          F(bVjn) - 1 - £                                    (5-4)
                                5-2

-------
where v is the number of permitted exceedances and n is the number
of possible daily maximum values.  Similarly, reasonable estimates
of the 8-hour parameters can be made by fitting a cumulative
distribution to the daily maximum 8-hour running averages of an
                                         /N          /"^
augmented data set and again calculating b, -6_ and b_ 355 •  In
statistical theory, b,   is known as the characteristic largest
                     x , n
value and b-   is known as the characteristic fifth largest
           D , n
value. l
     Selection of an appropriate cumulative distribution to fit
the data is important in determining a reasonable characteristic
largest value.  Two distributions which often provide close fits
to ambient air quality data are the Weibull and the lognormal.2
The Weibull distribution is defined as
          P(x) = 1 - exp [-(f)kj         '                   (5-5)
where 5 is the scale parameter and k is the shape parameter.  The
lognormal distribution is defined as
                  1   /*w
               "v^iT  J -oo
          F(x) =/^r   /_«  exp (-t2/2) dt                    (5-6)
where
          w = in x - y                                       (5_7)

                                                           2
and In x is distributed normally with mean y and variance a  .
From Equations 5-4 and 5-5, the characteristic vth largest
value of the Weibull distribution can be estimated as
if good estimates of 6 and k are available.  Similarly, the charac
teristic vth largest value of the lognormal distribution can be
estimated as

          bv,n = exP ^ + Szv,n}                            (5'9)
if good estimates of y and a are available.  The value of z
                                                           v, n
is determined from the normal distribution such that the area
under the standard normal curve from z    to °° is v/n.  Approxi-
mate values for z, ~£t. and zc ,,c are 2.7774 and 2.2058.
                 1,365      5,365
                                5-3

-------
     The results of fitting distributions to a large number of
ambient air quality data sets suggest that the characteristic vth
largest value can be better estimated if the upper tail of the
data is emphasized in the fit.  PEDCo Environmental has used two
methods to fit distributions to data censored on the left  (i.e.,
data from which low values have been excluded) :  the method of
least squares and the method of maximum likelihood.
5.2.1  Fitting Distributions by the Method of Least Squares
     The least squares method requires that the equation defining
the distribution under consideration be expressed as a linear
relationship of the form y = az + b.  Equations 5-5 and 5-6
can be rewritten in linear form using the following identities
where x  is the mth ranked value in ascending order.
       m         —
          Distribution     y_     a         z_         b
          lognormal      In x    a        z          u
                             m             m, n
          weibull        lnxm       l

These identities follow Gumbel's recommendation3 that F (x  )  = — T-T-
when fitting distributions to empirical data.  Values of z    for
                                                          m,n
the lognormal distribution are determined such that  the area under
the standard normal curve from -» to z    is m/(n+l).
                                      m, n
     A linear regression analysis of data transformed by these
identities yields a regression line with an equation in the  form
       ^>.    /N
of y = az + b.  Parameters of the corresponding distribution can
                                 /N     /\
be determined from the values of a and b using the following
equations:

          Weibull distribution
          s*.       /\
          & = exp b                                        (5-10)

          k = i                                            (5-11)
              a
          Lognormal distribution

          i = b                                            (5-12)

          a = a                                            (5-13)
                               5-4

-------
5.2.2  Fitting Distributions by  the Method  of Maximum Likelihood
     In an earlier analysis'* of population  exposure  to  N02,  the
least squares method described in Section 5.2.1 was  the sole
method used to fit distributions to air quality data.   The method
of maximum likelihood was not used because  no procedure was  then
available for applying it to the upper tail of a data set.   During
the PM exposure analysis,5 Louis Wijnberg of PEDCo Environmental
(extending the work of Cohen6'7) developed  the following maximum
likelihood procedure which can fit Weibull  and lognormal distribu-
tions to any portion of the upper tail of a data set.
     The n values in an augmented data set  are ranked from smallest
to largest to yield an ordered series

     xl' X2' *'*' xm' '''' xn
where x  indicates the mth ranked value.  We are interested  in  ob-
       m                —
taining maximum likelihood estimates  (MLE's) of the parameters  9,
and 92 of a two-parameter distribution F(x;9,,92)  fitting the
n- = n-c+1 values that equal or exceed x .  Letting f(x;9,,92)
denote the density function of a two-parameter distribution  and
F(x )  be the value of the cumulative distribution at x  ,
   C                                                  C
     L =
            n!
         (n-nf)!
 n
 n
m=c
f(x;9,,09)
   mi/
F(xc)
                                            n-n.
(5-14)
is the likelihood function of interest.  MLE's of 9^ and 92 are
determined by simultaneously solving the likelihood equations
       9
      39.
         •(log L)  = 0
                                          (5-15)
and
         •(log L)  = 0.
                                          (5-16)
In the case of the Weibull distribution, the likelihood equations
are
                              5-5

-------
     1     1        TT,            m       TT,
     1,  1  v ,  , UK        v  / m>  ir, f  mi  - n
     j- + —-  Z ln(;H - =r-   S  (n-)  In (—)  = 0
     *   nf m=c   Xc    nf m=c Xc     Xc
                                                         (5-17)
and
 -,    ,    n   xm
 £ + ±_  Y  / _jn\   ,
 ?   nf m=c  xc
                           n-n,
                                      = 0,
                                                     (5-18)
where
          x
                                                             (5-19)
When fitting the lognorroal distribution,  the likelihood equations
are:
      (y - V) -
                (n-n
                 nf»(zc)
                        = 0
and
+ (y- .)2 - a2
                              (n-n£)a
                                 nf«(zc)
where
     -   1   n
     y = —  Z  y  ,
         nf m=c  m
 2
3  -
             n
                     - 2
                                            = 0
                                                     (5-20)
                                                         (5-21)
                                                         (5-22)
                                                             (5-23)
       = ln
    zc =
         yc -
                                                         (5-24)
                                                         (5-25)
$ denotes the  standard  normal  distribution,  and  is the standard
normal density function.
     The likelihood  equations  were  solved by using the least square
procedure described  in  Section 5.2.1 to make initial estimates of
the parameters and then improving these estimates using an itera-
tive process  (the Secant method)  until an optimal solution was
reached.
                                5-6

-------
     The method of maximum likelihood has several advantages over
the method of least squares.  In particular, maximum likelihood
estimates (MLE's) of parameter values have minimum variance and
they asymptotically approach a normal distribution about the "true"
parameter value as the number of observations increases.  It is
also possible to construct confidence intervals for MLE's.  Param-
eter estimates developed by the least squares method have none of
these properties.  Consequently, we decided to use maximum likeli-
hood to fit Weibull and lognormal distributions to the CO data.
5.2.3  Determining Goodness-of-Fit
     Following the recommendations of Stephens8, and Green and
Hegazy, 9 we investigated the use of the Cramer-von Mises  (W ) and
                   2
Anderson-Darling (A )  statistics to determine goodness of fit.
These statistics are defined by the expressions

     W2 = l/12n + -  Z  [F(x )-(2m-l)/2n]2                 (5-26)
                  n m=l     m
and

                    -                    -   n . , _
                               ill             il *" JL III
     A2 = -   I.  (2m-l) | In F (xj + In  [1-F (xn . , _J ] [ - n.  (5-27)
                                               "      '
             _ i
             m=l
The null hypothesis (HQ) is that the sample comes from a population
with distribution function F(x).  HQ is rejected at a given sig-
nificance level if a goodness-of-f it statistic exceeds a critical
value corresponding to that significance level.
     In our analysis,  we were interested in which of two distribu-
tions (Weibull and lognormal) better fits a specified portion of
                                        2      2
a sample data set.  Unfortunately, the W  and A  statistics of a
Weibull distribution fit to data cannot be directly compared with
     2      2
the W  and A  of a lognormal fit to the same data.  The corres-
ponding significance levels can be compared, but tables which list
significance levels for censored data or for Weibull distributions
                                             2      2
are not currently available.  Consequently, W  and A  can only be
used to characterize the fit of a lognormal distribution to an
uncensored data set.

                               5-7

-------
     We also investigated various statistics based on the absolute
differences between the sample data set and the fitted distribution
The mth absolute difference is
                     s\
          £m =  |x  - x  j     m = c, c+1,  . .., n             (5-28)
                                  s\
where x  is the mth ranked value, x  is the estimate of x
determined by the parameters of the fitted distribution, and  c  is
the rank of the smallest data value used  in the fit.  In the  case
                                                 /\     y\
of a fitted Weibull distribution with parameters 5 and k,
Similarly,

          x  = exp  (u + a 2   )                             (5-30)
           m               m, n

for a fitted lognormal distribution with parameters u and  a.
     The statistics considered were:
              max diff = max e ,                            (5-31)

                           1    n
             mean diff = 	-TT  £  e  ,                      (5-32)
                         n-c+±      m
                               m=c
                           ,    n     9
          mean diff**2 = -—y  z  em  /                     (5-33)
                         n c   m=c  m

           max reldiff = max  [2e /(x   + x  )], and           (5-34)
                                mm    m
                           ,    n
          mean reldiff = ^r^+T  z   t2£ /(x + x  )].         (5-35)
                               m=c

The closer any one  of these statistics is  to zero, the better  the
distribution fits the data.  The bracketed term  in Equations  5-34
and 5-35 is termed  the relative difference; it is the absolute
difference divided  by the mean of the  observed and estimated
values.
                               5-3

-------
     We initially evaluated these statistics as alternatives to
     2
the R  statistic used for characterizing goodness-of-fit for the
least squares method.  Table 5-1 summarizes the results of fitting
Weibull and lognormal distributions to 1978 CO data from St. Louis
site 261040001G01.  Fits were made to the upper 50%, 20%, 5%, and
1% of the data by the least squares method.  In each of the four
cases, all six statistics indicate the same distribution as the
best fit.  The Weibull distribution provides a better fit in the
5% case; the lognormal distribution is superior in the other three
cases.  According to most of the statistics, the two best fits are
lognormal/50% and Weibull/5%.  The estimated characteristic high,
.A.
b , is close to the recorded maximum value of 14.9 ppm for both
fits.  The two worst fits are Weibull/50% and Weibull/20%; in both
           y*.
instances, b  is significantly different from the recorded maximum
value.
     We ultimately selected mean reldiff as our goodness-of-fit
statistic.  This statistic is particularly robust (i.e., not sig-
nificantly affected by outliers) and weights each value used in
the fit equally.  Max diff, mean diff**2, and max reldiff can be
significantly affected by outliers.  Max diff, mean diff, and
mean diff**2 weight higher values more heavily than lower values.
     Weibull and lognormal distributions were first fit to the
upper 50 percent of the daily maximum values in each augmented
data set  (Tables 5-2 and 5-3).  Evaluation of the results indi-
cated that the closer fitting distribution did not always yield
a close fit to the five largest values.  Repeating the analysis
using the upper 20 percent of the daily maximum values in each
augmented data set produced similar reldiff statistics and
superior fits to the five largest values (Tables 5-4 and 5-5).
Consequently, we decided to use the upper 20 percent of the
augmented data set for all fits.  The general procedure used for
determining characteristic largest and fifth largest values is
described below.
     (1)  Maximum daily 1-hour and 8-hour running average values
          in each augmented data set were ranked from lowest to
          highest.
                              5-9

-------
 00  •
 z o
 o o
 =3 UJ
 03 21
 t— i

 O£ m
 I— UJ
  a:
   o*
   00
 a: oo
 0<
 z uj
   CO
=> 00
CO i-t


uj o
3 _l


C3   •
u. a:
   u.
u_
oo  a.
H-  Q.
  • o
1-4 O
 I
ir> co
   r^
uj en
_J •-!
ca














c
o
•r*
3
•r-
J-
+J
in
•a
J3

+J
•f—
•+-

•3
Q
















X
t— 1

S_
OJ
a.





^"^
to

t.
QJ
Q.








^^
O
CM

cx
Q.
ZID






^4^
O

QJ
CX
CX



1
O
c
co
0
^~
^
,_.
3
•r—
(T3
l_
O
C
CO
o
*~
r-
"^
<"»
•r-
O)

-
g
t-
O
C
en
o
;3
'a
-Q
•r-
QJ
E
s_
o
c
CO
o
£
3
-Q
•r-
QJ

U
•r—

.J2
-M
na

00

o  . uo n  ^^ ^>^
cno«-*ooo«— i»-i r*»
Or-*OOOOCM»-t «q-
r— 1
tOtO^TO O LOOlO
en i-™i co CM crv »~< r*^ CM •""*
CT% ^^ CO C^ CO CO CO to CO
• •••••II •••
OOOOOO OOU3
• f-H

r*»» vo co o% co en ro o
cnopoooo*-" oo
• •'• • • • • •! t •
OrHOOOOCM«-t «:»•
i— <


to 10 ^* co ^o en ^o vo
en to P*. r-t r*, r-t co co 10
cncnoooo en^^ro
• •••••ii •••
OOOOOO OOtO
1— 1
OCOOlO WOCMCM IO
OO *™"^ LjO LjO Os4 ^^4 
o^tnoOr-iO o^^rtn
• •••••ii •••
Of— < O O O O OO^"
•^
CA co OQ ••"< o\ LO en t~~*
r^» eo ro en co CM ••— * CM co
cnio«-iOcoo«— ico «?•
• •••••••ii •
O'O-OOOOCOr-l O
«— t
CM M-
-K S- -r-
4- 4- T- T3
H- 4- >+- T3 (—
i| M ^ 	 •r"* *™^ QJ
•r- "O "O Ol t-
T3 i-
C C C
X (d (O X rC
CM t) QJ QJ (O OJ C
Q£ 21 2Z "Z. S S < tO < -^ < ^L < C* < -Q
 ex
 a.

en
                                                                                             a*
 X
 
-------
TABLE 5-2.   RESULTS OF FITTING WEIBULL  AND  LOGNORMAL  DISTRIBUTIONS  BY
          MAXIMUM LIKELIHOOD PROCEDURE  TO UPPER 50  PERCENT OF
                    DAILY MAXIMUM 1-HOUR  CO DATA




Study area
Chicago





Los Angeles





Philadelphia





St. Louis









NT
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
Wei bull


Mean
reldiff
0.0347
0.0241
0.0197
0.0331
0.0406
0.0179
0.0425
0.0321
0.0315
0.0282
0.0305
0.0315
0.0897
0.0897 •
0.0932
0.0850
0.0850
0.0505
0.0519
0.0185
0.0331
0.1214
0.0433
0.0331
Characteristic
values, ppm
/\
h
Dl,365
10.0
23.5
15.7
11.6
11.7
10.0
14.6
29.0
26.3
20.0
21.8
26.3
16.6
16.6
8.9
11.5
11.5
12.2
8.8
12.7
17.6
16.4
13.0
17.6
/^
h
D5,365
8.4
20.0
13.5
9.2
9.5
8.3
12.1
23.1
21.2
16.3
17.2
21.2
13.4
13.4
7.4
8.6
8.6
10.2
7.3
11.4
14.7
12.9
11.0
14.7
Lognormal


Mean
reldiff
0.0186
0.0121
0.0112
0.0153
0.0153
0.0254
0.0535
0.0351
0.0485
0.0388
0.0435
0.0485
0.0598
0.0598
0.0678
0.0777
0.0777
0.0472
0.0338
0.0118
0.0147
0.0758
0.0214
0.0147
Characteristic
values, ppm
XV
h
Dl,365
11.2
26.4
17.4
13.5
13.1
11.6
17.0
35.0
31.6
23.6
26.9
31.6
17.3
17.3
9.0
13.3
13.3
13.9
10.0
13.4
19.9
17.4
14.3
19.9
/\
h
D5,365
8.7
20.9
14.1
9.6
9.8
8.7
12.8
24.9
22.8
17.4
18.7
22.8
13.0
13.0
7.1
8.9
8.9
10.7
7.6
11.5
15.2
12.6
11.3
15.2
                               5-11

-------
TABLE 5-3.  RESULTS OF FITTING WEIBULL AND LOGNORMAL  DISTRIBUTIONS  BY
        MAXIMUM LIKELIHOOD PROCEDURE TO UPPER 50 PERCENT OF
           DAILY MAXIMUM 8-HOUR RUNNING AVERAGE CO  DATA




Study area
Chicago





Los Angeles





Philadelphia





St. Louis









NT
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
Weibull


Mean
reldiff
0.0383
0.0110
0.0211
0.0588
0.0702
0.0609
0.0208
0.0290
0.0251
0.0278
0.0231
0.0251
0.0954
0.0954
0.0684
0.0509
0.0509
0.0359
0.0512
0.0082
0.0570
0.0901
0.0394
0.0570
Characteristic
values, ppm
s\
L
Dl,365
6.2
14.6
10.4
7.8
8.0
6.8
9.2
21.5
19.5
15.1
16.3
19.5
11.4
11.4
6.4
6.8
6.8
9.1
5.3
10.4
12.0
10.1
9.9
12.0
A.
L
D5,365
5.4
12.7
9.1
6.2
6.4
5.5
7.8
16.7
15.8
12.3
12.9
15.8
9.3
9.3
5.3
5.2
5.2
7.6
4.5
9.5
10.1
8.4
8.5
10.1
Lognormal


Mean
reldiff
0.0178
0.0156
0.0112
0.0247
0.0362
0.0310
0.0393
0.0518
0.0383
0.0219
0.0248
0.0383
0.0521
0.0521
' 0.0340
0.0310
0.0310
0.0172
0.0271
0.0138
0.0314
0.0519
0.0218
0.0314
Characteristic
values, ppm
A
i
Dl,365
6.8
16.3
11.4
8.8
8.7
7.5
10.8
27.0
23.0
17.8
19.8
23.0
11.9
11.9
6.6
8.1
8.1
10.2
5.8
11.2
13.1
10.5
10.9
13.1
yv
K
D5,365
5.5
13.3
9.4
6.3
6.4
5.6
8.3
18.4
16.9
13.1
14.0
16.9
9.0
9.0
5.1
5.5
5.5
7.8
4.6
9.8
10.2
8.2
8.8
10.2
                                5-12

-------
TABLE 5-4.   RESULTS OF FITTING WEIBULL  AND  LOGNORMAL  DISTRIBUTIONS  BY
        MAXIMUM LIKELIHOOD PROCEDURE TO UPPER 20  PERCENT OF
                   DAILY MAXIMUM 1-HOUR CO  DATA




Study area
Chicago





Los Angeles





Philadelphia





St. Louis









NT
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
Wei bull


Mean
reldiff
0.0165
0.0116
0.0139
0.0317
0.0386
0.0231
0.0360
0.0289
0.0252
0.0278
0.0282
0.0252
0.0782
0.0782
0.0843
0.0688
0.0688
0.0439
0.0234
0.0249
0.0353
0.0802
0.0225
0.0353
Characteristic
values, ppm
/s
h
1,365
10.7
24.9
16.3
12.5
12.9
10.2
14.1
29.3
25.0
19.6
21.0
25.0
19.0
19.0
9.9
12.2
12.2
12.8
9.8
13.2
19.0
21.1
14.5
19.0
/\
h
D5,365
8.7
20.7
13.9
9.6
10.1
8.3
11.8
23.3
20.6
16.1
16.8
20.6
14.3
14.3
7.8
8.9
8.9
10.5
7.8
11.6
15.3
14.6
11.7
15.3
Lognormal


Mean
reldiff
0.0196
0.0164
0.0126
0.0210
0.0250
0.0177
0.0432
0.0288
0.0230
0.0300
0.0316
0.0230
0.0607
0.0607
0.0672
0.0718
0.0718
0.0390
0.0327
0.0169
0.0263
0.0608
0.0189
0.0263
Characteristic
values, ppm
>\
K
Dl,365
11.4
26.4
17.2
13.3
13.6
10.8
14.9
31.4
26.2
20.8
22.3
26.2
19.2
19.2
9.7
13.2
13.2
13.4
10.6
13.4
20.1
22.8
15.4
20.1
/\
K
D5,365
8.8
20.9
13.9
9.6
10.0
8.4
12.0
23.5
20.6
16.2
16.9
20.6
13.8
13.8
7.4
8.9
8.9
10.5
7.9
11.5
15.3
14.6
11.8
15.3
                                5-13

-------
TABLE 5-5.  RESULTS OF FITTING WEIBULL AND LOGNORMAL  DISTRIBUTIONS  BY
        MAXIMUM LIKELIHOOD PROCEDURE TO UPPER 20 PERCENT OF
           DAILY MAXIMUM 8-HOUR RUNNING AVERAGE CO  DATA




Study area
Chicago





Los Angeles





Philadelphia





St. Louis









NT
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
Wei bull


Mean
reldiff
0.0165
0.0168
0.0135
0.0467
0.0510
0.0362
0.0170
0.0296
0.0391
0.0253
0.0247
0.0391
0.0944
0.0944
0.0836
0.0467
0.0467
0.0400
0.0406
0.0106
0.0327
0.0821
0.0224
0.0327
Characteristic
values, ppm
/\
h
Dl,365
6.8
15.0
10.8
9.0
9.4
7.8
8.8
20.1
19.4
15.5
16.6
19.4
13.7
13.7
7.2
7.4
7.4
9.9
5.9
10.5
13.8
12.0
10.8
13.8
/\
K
D5,365
5.7
12.9
9.3
6.7
6.9
5.9
7.6
16.1
15.8
12.5
13.1
15.8
10.2
10.2
5.6
5.5
5.5
7.9
4.8
9.5
10.9
9.2
9.0
10.9
Lognormal


Mean
reldiff
0.0111
0.0117
0.0178
0.0345
0.0407
0.0280
0.0222
0.0399
0.0275
0.0330
0.0364
0.0275
0.0757
0.0757
0.0623
0.0437
0.0437
0.0301
0.0297
0.0075
0.0240
0.0665
0.0219
0.0240
Characteristic
values, ppm
s\
K
Dl,365
7.2
15.6
11.4
9.5
10.0
8.3
9.3
21.6
20.3
16.6
18.0
20.3
14.3
14.3
7.2
7.9
7.9
10.4
6.1
10.7
14.7
12.5
11.5
14.7
/\
K
D5,365
5.7
12.9
9.4
6.6
6.9
6.0
7.6
16.3
15.7
12.7
13.3
15.7
9.9
9.9
5.3
5.5
5.5
7.9
4.8
9.5
10.9
9.0
9.0
10.9
                               5-14

-------
     (2)   The upper 20 percent of the daily maximum values were
          fit by Weibull and lognormal distributions using the
          maximum likelihood method described above.
     (3)   The reldiff statistics of the two fits were compared
          and the parameters of the better fitting distribution
          (i.e./ the one with the smaller reldiff value) were
          used to determine the characteristic largest and fifth
          largest values.
Table 5-6 lists characteristic values developed using this pro-
cedure.  Appendix B discusses the relationship between these
values and the expected concentration  (EC) values developed by
EPA for the four study areas.

5.3  BACKGROUND CONCENTRATIONS
     NEM requires a city-specific average background level in
order to calculate the rollback factor applied to ambient pollu-
tant concentrations in each study area.  This background value
should represent the average hourly concentration of a given
pollutant being transported into the urban area, a value unaffected
by any control strategies imposed upon the urban area.  The moni-
toring sites selected to determine CO background should ideally
be located sufficiently upwind from the urban area in a nonlow-
lying location, within no less than five degrees of alignment
with extended straight highway segments.  Also, each site should
be in an area with sufficient ventilation so that air is not
likely to stagnate.  Sites established to monitor regional con-
centrations are preferred to those established to monitor local
concentrations.  PEDCo identified monitoring sites which satisfied
these criteria through an evaluation of (1) regional office and
local agency recommendations, (2)  local wind profiles, and (3)
local land use.  It should be noted that the CO background con-
centration being transported into an urbanized area may in fact
be higher on occasion than some of the reported values within the
area.  This phenomenon is due to dispersion and dilution and is
dependent upon the siting objectives and spatial distribution of
CO monitors across the study area.

                               5-15

-------
TABLE 5-6.   AIR QUALITY  INDICATORS FOR CO DATA



Study area
Chicago
,




Los Angeles





Philadelphia





St. Louis








NT
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
CR
CC
CI
SR
SC
SI
Daily maximum 1-hour
averages (ppm)
Char.
largest
10.7
24.9
17.2
13.3
13.6
10.8
14.1
31.4
26.2
19.6
21.0
26.2
19.2
19.2
9.7
12.2
12.2
13.4
9.8
13.4
20.1
22.8
15.4
20.1
Char.
5th largest
8.7
20.7
13.9
9.6
10.0
8.4
11.8
23.5
20.6
16.1
16.8
20.6
13.8
13.8
7.4
8.9
8.9
10.5
7.8
11.5
15.3
14.6
11.8
15.3
Daily maximum 8-hour
running averages (ppm)
Char.
largest
7.2
15.6
10.8
9.5
10.0
8.3
8.8
20.1
20.3
15.5
16.6
20.3
14.3
14.3
7.2
7.9
7.9
10.4
6.1
10.7
14.7
12.5
11.5
14.7
Char.
5th largest
5.7
12.9
9.3
6.6
6.9
6.0
7.6
16.1
15.7
12.5
13.1
15.7
9.9
9.9
5.3
5.5
5.5
7.9
4.8
9.5
10.9
9.0
9.0
10.9
                    5-16

-------
     Contact with the local EPA Regional Office  resulted  in  iden-
tification of the Chicago Heights site  (SAROAD code:   141240001G01)
as an appropriate background site for the Chicago  study area.   The
site is located at a high school sufficiently far  from areas with
high traffic concentrations.
     A rural site near the urban area of St. Louis (SAROAD code:
264300006G01) was selected as the indicator for  background CO
levels for that study area.  CO levels measured  at this site are
similar to those reported by a site predominantly  upwind  of  the
metropolitan area.
     As a result of diurnal wind cycling caused  by land-sea
breezes, each station in the South Coast Air Basin is  occasionally
upwind and downwind of the center city core.  Consequently,  pre-
dominant wind direction was not considered a valid criterion for
identifying a background site for the Los Angeles  area.   A rural-
agricultural site fairly removed from urban influence  (SAROAD
code:  055160001101) was selected.
     The Philadelphia local agency recommended a site  in  Northwest
Philadelphia (SAROAD code:  397140014H01) as the most  appropriate
indicator for average background concentrations.
     The average hourly concentration was calculated for  a recent
year at each site to estimate annual average background for  the
corresponding study area.  These values are listed in  Table  5-7.

          TABLE 5-7.  ESTIMATED ANNUAL AVERAGE BACKGROUND  LEVELS

Study area
Chicago
Los Angeles
Philadelphia
St. Louis

Year
1979
1977
1978
1978
CO background concentration
mg/m3
1.5
2.0
1.1
2.6
ppm1
1.31
1.75
0.96
2.27
      Converted at STP using 1 ppm = 1145 yg/m2
                               5-17

-------
5.4  REFERENCES

1.    E.  J. Gumbel, Statistics of Extremes, Columbia University
     Press, New York, 1958, p. 82.

2.    T.  Johnson, "A comparison of the two-parameter Weibull and
     lognormal distributions fitted to ambient ozone data," Proc.
     of  Specialty Conference on Quality Assurance in Air Pollution
     Measurement, Air Pollution Control Association, 1979.

3.    Op. cit., Gumbel, p. 34.

4.    T.  Johnson and R. Paul, The NAAQS Exposure Model (NEM) and
     Its Application to Nitrogen Dioxide, prepared by PEDCo Envir-
     onmental, Inc., for Strategies and Air Standards Division,
     Office of Air Quality Planning and Standards, U.S.  Environ-
     mental Protection Agency, Research Triangle Park, N.C.
     27711, August 1981.

5.    Ted Johnson and Roy Paul, The NAAQS Exposure Model (NEM)
     and Its Application to Particulate Matter, prepared by PEDCo
     Environmental, Inc., for Strategies and Air Standards Division,
     Office of Air Quality Planning and Standards, U.S.  Environ-
     mental Protection Agency, Research Triangle Park, N.  C.
     27711, August 1981.

6.    A.  C. Cohen, Jr., "Simplified estimators for the normal dis-
     tribution when samples are singly censored or truncated,"
     Technometries, Vol. 1, No. 3, August 1959.

7.    A.  C. Cohen, Jr., "Maximum likelihood estimation in the
     Weibull distribution based on complete and on censored
     samples," Technometrics, Vol. 7, No. 4, November 1965.

8.    M.  A. Stephens, "EDF statistics for goodness of fit and some
     comparisons," Journal of the American Statistical Association,
     Vol. 69, No. 347, September 1974.

9.    J.  R. Green and Y. A. S. Hegazy, "Powerful modified-EDF
     goodness-of-fit-tests," Journal of the American Statistical
     Association, Vol. 71, No. 353, March 1976.
                              5-18

-------
                            SECTION 6
              SIMULATION OF CARBON MONOXIDE LEVELS
                     IN THE MICROENVIRONMENT

     A basic assumption of NEM is that each member of the study
area population can be assigned during each hour of the day to
one of five microenvironments:  indoors  (work or school), indoors
(home or other),  inside a transportation vehicle, outdoors near a
roadway, or other outdoor locations.  In applying NEM to CO, we
initially assumed that air quality in each microenvironment (x  . )
                                                              m, t
of a given neighborhood type could be estimated by the expression

     xm,t - am,t + bmV                                    (6~l}
where a  .  is the pollutant concentration generated by a particu-
       m f t
lar source in the microenvironment, x." is the monitor-derived
air quality estimated for the neighborhood type, and b  is a multi-
plicative factor.  Consequently, estimates of a    and b  (denoted
/•-        A                                     m, t      m
a  .  and b )  appropriate for CO were needed for each microenviron-
 m,t      m
ment.  We assumed that a  .  will vary with microenvironment, CO
                        m, t
source, and time of day; and that b  will vary only with micro-
environment.   Equation 6-1 was later revised to account for ob-
served lags between indoor and outdoor CO.
     PEDCo reviewed 75 reports with key words or abstracts sug-
gesting they contained information useful in estimating a  ,  and
b—jointly referred to as microenvironment factors (MF's).   The
review indicated that 26 of these reports contained data appli-
cable to our analysis.  These reports are categorized by micro-
environment in Table 6-1.  In the following discussion, results
of these studies are used to estimate MF's and, in some cases, to
develop alternatives to Equation 6-1.
                               6-1

-------
     TABLE 6-1.   STUDIES CONSIDERED  IN  DEVELOPING
              CO MICROENVIRONMENT FACTORS
    Microenvironment
            Study
Indoors:   work or school
Harke1
Penkala and Oliveira2
Moschandreas, et al.3
Yocum, et al ."*
General Electric5
Derham, et al .s
Godin, et al,7
Thompson, et al.8
Indoors:   home or other
Yocum, et al.^
Moschandreas,  et al.9
Cote, et al.ll
Bridge and Corn12
Sterling and Kobayashi13
Penkala and Oliveira2
Repace and Lowrey15
Spengler, et al.17
Sterling and Sterling18
Spengler, et al.20
Godin, et al.7
Elliot and Rowe21
Thompson, et al.8
Transportation vehicle
Ott and Will its22
Ziskind, et al.23
Col will and Hickman21*
Wallace25
Cortese26
Brice and Roesler
Petersen and
Harke, et al
                                            .27
                                         Sabersky
                                                 28
Roadside
Wilson and Schweiss29
Wilson and Schweiss30
Jabara, et al.31
                         6-2

-------
6.1  WORK-SCHOOL MICROENVIRONMENT
     Smoking has been identified by several studies as affecting
CO levels in enclosed working areas.  The contribution of smoking
does not appear to be very significant, however.  CO was monitored
for 18 days by Harke1 in two office buildings, one air-conditioned,
the other not.  Harke found that no significant increase in CO
occurred after employees started to smoke.  In another experiment,
Harke found that CO did not exceed 10 ppm in an unventilated
office room (30 m ) when an occupant smoked at a rate of 2 cigar-
ettes per hour.  Background and outdoor CO levels are not mentioned
in either study.  Using test chamber data, Penkala and Oliveira2
estimate that CO in a 400 ft  room occupied by one smoker consuming
1.25 cigarettes per hour will average 18.6 mg/m  per hour at 0
air changes per hour.  At recommended ventilation rates (2.1 to
7.5 air changes per hour), CO should average 1.2 to 3.6 mg/m .
Moschandreas,  et al.,3 studied CO in two office buildings in
Boston.  They hypothesized that indoor sources of CO are largely
damped by the diffusive effect of the air handling systems.
Elevated CO concentrations related to smoking were not observed.
Yocum, et al.,1* suggest that daytime indoor-outdoor ratios above
1.00 observed in two office buildings in Hartford, CT, are the
result of smoking by occupants and visitors but do not provide
useful data for estimating a  ..
                            m, t
     The report by Penkala and Oliveira is the most useful of
these four studies.  The following excerpt describes their model
and discusses their assumptions.
          Assume a smoker and a nonsmoker occupy the same office
     with a total volume of 400 ft .  Ventilation rates in forced
     ventilations systems are usually between 7 and 25 ft  of
     fresh air per minute per room occupant.  These ventilation
     rates are equivalent to 2.1-7.5 air changes per hour, and
     can be attained by normal leakage around windows and doors.
          A typical smoker consumes one pack of 20 cigarettes
     per day  (16 waking hours).   Each cigarette is smoked in

                               6-3

-------
     about 10 minutes,  creating a high concentration of CO and
     SPM in the room,  and then the ventilation system and other
     removal mechanisms (as measured in this study)  lower those
     concentrations somewhat during a rest period (40 minutes)
     before the next cigarette is lit.  The concentrations can
     be time-averaged by considering the room to be  in a cycle
     consisting of a rapid concentration rise and a  slower ex-
     ponential decay.   The decay rate depends upon the ventilation
     rate and the measured gas removal rate.  Both can be repre-
     sented by the equivalent air changes per hour,  and converted
     to a time constant, T, representing the minutes per equiva-
     lent air change.
          Then C2 = C-^-1/T]
          C- is concentration at time t
          C, is an initial concentration
          T is the number of minutes per equivalent  air change
          t is the average time of one cigarette smoke plus
               following rest period
     Note that (C, - C2) is the concentration added  by smoking a
     cigarette.
          A cycle ends with the room at concentration C2 and is
     raised to a new concentration C, through smoking a cigarette.
     Combining the equations allows computing C, and C2 for any
     equivalent air change rate.  The average concentration C,
     can be found by integration over a smoking period plus rest
     period.2
     The 400 ft  room volume is based on a ASHRAE recommendation
of 200 ft  per office building occupant.  Repace and Lowrey15
estimate that one-third of adults smoke.  They also  state that
the recommended occupancy density for general office space is
10 persons per 1000 square feet.  Assuming an 8-foot ceiling, we
can estimate that there is one smoker per 2400 ft .   Since Penkala
and Oliveira assume there is one smoker per 400 ft , their esti-
mates can be multiplied by 400/2400 to yield the CO levels expected

                                6-4

-------
in an office with one smoker per  2400  ft"
are listed in Table 6-2.
Both sets of estimates
            TABLE 6-2.  ESTIMATES OF CO CONCENTRATIONS IN AN
                         OFFICE WITH SMOKERS


air changes/hour
0
1
2.1
7.5
mean CO (mg/m )
400 ft3 per
smoker
18.6
6.2
3.6
1.2
2400 ft3 per
smoker
3.10
1.03
0.60
0.20
Based on these results, reasonable bounds  for  a
                                               m,t
       during working
                         j                        • ~*
hours would be 0.20 mg/m   (0.17 ppm) and  0.6 mg/mj  (0.52  ppm);  a
reasonable best estimate for a    would be  0.35  mg/m  (0.30 ppm),
                              m,1
the geometric mean of the bounds.
The relationship
     £  =  (x  ,
      m     m, t
                     ~ am,t)/xa,t
                 (6-2)
where, x  .  represents ambient CO levels reported  by  a  fixed monitor,
       a, t
can be used to estimate b  if good data for determining  x  .,  x
    ^                    m3                        3   m, t   a, t
and a    are available.  Two studies—Moschandreas,  et al.,3 and
     in f "c
General Electric5—provide x,,,   and x,   data.  Yocum, et al. , **
                            m, L.      a, t
provide xm ./x, .  values.  None of these studies  list  values
         in, c  a, t
directly relating to a  ..  General Electric measured  CO inside
and outside of two buildings in New York.  One building  was  an
air rights building above the Trans Manhattan Expressway;  the
other was a more conventional high rise structure on one side of
a street canyon in midtown Manhattan.  The following excerpt is
taken from their conclusions.
          Concentrations indoors at the building  base  vary with
     outdoor concentrations.  Indoor concentrations  lag  changes
     in outdoor CO levels.  It is suspected that  this  time delay
     is a variable that is a function of both wind conditions as
     seen at the building and the direction of change  in outdoor
     concentrations.
                               6-5

-------
          Average concentrations inside and outside the buildings
     reduce exponentially with height above ground level.  The
     rate of change with height is essentially constant outdoors
     for both heating and non-heating seasons.  However, indoors
     the decay in average concentrations with height is greater
     during the non-heating season than during the heating
     season.  This variation is the result of changes in the
     roof wind angle from the non-heating to the heating season.
          Indoor concentrations normally are lower than outdoor
     concentrations at all heights above the roadway when outdoor
     concentrations are high.  Conversely, indoor concentrations
     are higher than outdoor concentrations when outdoor concen-
     trations are low.5
     Because the air-rights building is atypical of urban work
places, data for the street canyon building should receive primary
attention.  This building was not air-conditioned; ventilation,
especially during the summer months, was achieved by opening
windows.  Table 6-3 lists average weekday CO concentrations at
9 feet above street level, third floor, fifth floor, llth floor,
and 19th floor.
        TABLE 6-3.  WEEKDAY CO MEASUREMENTS AT STREET CANYON SITE5
season
heating




non-heating




location
9 feet
3rd floor
5th floor
llth floor
19th floor
9 feet
3rd floor
5th floor
llth floor
19th floor
average CO (ppm)
outside
11.2
9.9
7.7
6.6
5.4
11.2
10.3
8.1
4.8
4.2
inside
.
9.5
7.8
6.9
6.8
_
8.2
7.1
4.7
3.8
                               6-6

-------
Inside CO concentrations  are generally the same as  outside CO
concentrations at  the  same building height.  CO decreases with
height so that the ratio  of inside CO to CO 9 feet  above street-
level varies from  0.85 at the third floor to 0.61 at  the 19th
floor during the heating  season.  In the non-heating  season,  the
ratio ranges from  0.73 at the third floor to 0.34 at  the 19th
floor.  The contribution  of indoor sources to indoor  CO is unknown
but is probably small  in  proportion to the ambient  CO levels.
     Moschand"reas,  et  al.,3 measured CO inside and  outside of two
office buildings in Boston.  Their results are listed in Table 6-4,

      TABLE 6-4.  CO CONCENTRATIONS (ppm) AT TWO OFFICE SITES RECORDED
                       BY MOSCHANDREAS, ET AL.3
building
new
old
mean indoor
3.18
2.16
max indoor
11.35
14.36
mean
outdoor/indoor
1.02
0.88
Note they reported  outdoor/indoor ratios rather than  indoor/outdoor
ratios.  Figure  3 in  Moschandreas, et al., shows  indoor CO track-
ing outdoor CO at the new building.
     Table 6-5 lists  indoor-outdoor ratios for two  air-conditioned
office buildings in Hartford,  CT, determined by Yocum,  et al.4

       TABLE 6-5.  INDOOR-OUTDOOR CO RATIOS DETERMINED FOR TWO OFFICE
                      BUILDINGS BY YOCUM, ET AL.1*
Building
100 CP
250 CP
Season
Summer
Fall
Winter
Summer
Fall
Winter
Daytime ratio
1.31
1.32
1.13
1.05
0.96
0.76
Nightime ratio
1.00
1.25
1.21
1.02
1.04
0.96
        CP:  Constitution Plaza
                                6-7

-------
Inside CO was measured on the second floor at 100 CP and the
third floor at 250 CP.  The authors suggest that the start-up of
building ventilation during rush hour is the primary cause of
summer and fall daytime ratios greater than 1.00 at 100 CP.
They further suggest smoking may have elevated ratios in the
winter.
     Derham, et al.,5 monitored CO inside and outside a building
in Los Angeles.  They found that indoor levels of CO reflect
directly the levels outdoors but with a phase lag that can be
explained by means of a simple analytical model which accounts
for ventilation rates but neglects any chemical reactions.  They
do not provide simultaneous indoor/outdoor readings and smoking
is not discussed as a possible CO source.
     Godin, et al., measured CO levels inside and outside a down-
town office in Toronto with the windows closed.  They summarize
their findings as follows:
          At 150 College St., about a mile from the city center,
     outdoor values were 2.7 +1.8 ppm, while the corresponding
     values for the first and third floors were, respectively,
     2.2 + 1.3 ppm and 2.8 + 1.5 ppm.  Values in taller downtown
     buildings apparently depended on the level of air intake for
     the floor in question; at the Toronto Dominion Centre, the
     sidewalk concentration was 6.4 ppm, figures for the first
     and third floors were 4.6 and 4.0 ppm, respectively, but
     the 54th floor  (with a much higher air intake) has a level
     of only 2.4 ppm.7
Godin, et al., conclude that indoor CO concentrations mirror
outdoor concentrations, with a lag of one to two hours.
     These studies suggest that a reasonable model for hourly
average CO in the workplace is

          xm(t) = am,t + IT  [xc(t) + *c(t-1)]  '              (6'3)
The indoor CO at time t is equal to the indoor generated CO at
time t plus b  times the average of the outdoor CO at time t and

                              6-8

-------
at time t-1.  This model assumes  that  building ventilation dampens
variations in indoor CO and causes  a slight  lag between indoor and
outdoor concentrations.  A reasonable  "best" estimate of b, for
                                                           m
buildings of 3 stories or less  is 0.85,  the  ratio of third floor
CO to outside ground floor CO in  the General Electric study.  A
reasonable range for b  is 0.60' (unairconditioned highrise) to
                      m
1.05  (ventilation system of 250 CP).
     The microenvironment under consideration  includes  schools as
well as work places.  Only one  study—Thompson,  et  al.8—measured
indoor and outdoor CO levels at a school.  Accuracy of  their CO
analyzer, +1.0 ppm, prevents a critical comparison of  the  low
values which were measured.   Since NEM treats  indoors work  and
indoors school as the same microenvironment, we  used the model
already developed for indoors at work for the  combined  work-school
microenvironment.
6.2  HOME-OTHER MICROENVIRONMENT
     The value of b  for homes can be estimated by  comparing
indoor and outdoor CO levels of homes with no indoor  CO  sources.
Yocum, et al.,1* measured indoor and outdoor CO at two residences
in Hartford, CT.  Neither home had a gas stove or habitual  smoker
Average indoor/outdoor ratios are listed in Table 6-6.

          TABLE 6-6. AVERAGE INDOOR/OUTDOOR CO RATIOS RECORDED BY
                           YOCUM, ET Al."
Residence
Blinn St.





Carol! St.





Season
Summer

Fall

Winter

Summer

Fall

Winter

Time of day
Day
Night
Day
Night
Day
Night
Day
Night
Day
Night
Day
Night
Ratio
1.02
1.07
1.03
1.08
1.07
1.08
1.04
1.02
1.03
1.08
0.96
1.08
                               6-9

-------
Note that all ratios are close to unity.  Yocum, et al., do not
provide data useful in determining if indoor CO lags outdoor CO.
Figure 4 from Moschandreas, et al.,9 suggests a lag of one hour
in a conventional residence in Baltimore.  The following is an
excerpt from their study.
     Indoor concentration peaks of CO tend to lag behind
     outdoor CO peaks.  Due to the CO emissions, this behavior
     may be shortened in houses with indoor sources.  The
     observed large fluctuations of the hourly CO concentrations
     display a local structure without a general pattern.  How-
     ever, examination of the CO data base from several weekdays
     leads to identification of a typical pattern with respect
     to 3-h averages.  Typically, the time periods 0800-1000 and
     1900-2100 exhibit the highest observed CO levels.  These
     3-h indoor peaks correspond to outdoor peaks caused by
     automobile traffic during the typical urban rush hours
     (0600-0800 and 1700-1900).  The association of rush-hour
     traffic and typical indoor high level periods reflect the
     time lag monitored earlier.  Figure 4 illustrates the
     indoor and outdoor variation of CO concentrations for a
     typical day, in a dwelling with indoor CO sources.  The
     indoor peak at hours 1400 to 1600 is not a typically observed
     elevation of the indoor concentrations.9
These results suggest that Equation 6-3 is applicable to the home
microenvironment as well as the work microenvironment.  Based
solely on the results of Yocum et al., a preliminary estimate of
b  would be 1.00.  However, analysis by Feagansl° indicates that
1.00 is probably too high.  Feagans suggests 0.85 as a more
appropriate best estimate of b  and 0.70 to 1.10 as a reasonable
          s\                   rn      /N
range for b .  Appropriate values of a  . for different indoor
           m                          m, c
sources are developed below.
     CO sources in the home include smoking, gas stoves, gas
furnaces, coal furnaces, and attached garages.  CO from these
sources combined with CO from outside have resulted in indoor
levels exceeding the CO NAAQS.
                              6-10

-------
     Three studies—Cote, et al.,11 Moschandreas, et al.,9 and
Bridge and Corn12—mention smoking as an indoor CO source in the
home.  Cote, et al., monitored indoor and outdoor CO in four
homes in Hartford, CT.  Unfortunately, the homes with smokers
also had gas appliances so that the contribution of smoking to
indoor CO cannot be determined separately.  Moschandreas, et al.,
monitored CO levels in 15 homes.  Persons living in these houses
were polled as to smoking habits.  Unfortunately, the report by
Moschandreas, et al., provides only a few sample days of CO data
and no smoking data.  Bridge and Corn measured CO at two experi-
mental "parties."  Sterling and Kobayashi provide the following
summary of this study.
     In one 5120 ft  room containing 50 people, 25 people con-
     sumed 50 cigarettes and seven cigars in 1h hours.  With a
     room air exchange rate of seven times per hour, CO averaged
     7 ppm during the course of the party.  During the second
     experiment in a 3750 ft  room containing 73 people, 36
     smokers consumed 63 cigarettes and 10 cigars in Ih hours
     and the average CO content was 9 ppm.13
These results suggest that 7 ppm is a worst case a  value for
smoking that would not be exceeded in the typical home except
during occasional social functions.
     The three studies described above are not useful in deter-
mining a typical a    for smoking.  However, with suitable assump-
            _m. "LJT^.T-L-L,  III / "C
tions we can use the model developed by Penkala and Oliveira to
estimate a  ,  if we have good estimates of air exchange rates-
          m, t
Table 6-7 lists air exchange rates determined by Moschandreas,
et al., for residences of various kinds.
                              6-11

-------
              TABLE 6-7.  AIR EXCHANGE RATES  DETERMINED BY
                        MOSCHANDREAS, ET AL.9
location
Washington
Baltimore
Denver
Chicago
Pitts burg




residence type
experimental
conventional
experimental
conventional
conventional
conventional
experimental
mobile 1
mobile 2
low-rise 1
low-rise 2
low-rise 3
high-rise 1
high-rise 2
high-rise 3
exchanges/h
0.5 - 1.0
0.2 - 0.8
0.5 - 1.2
0.6 - 2.0
0.8 - 1.0
0.6 - 1.0
0.1 - 0.3
0.4 - 1.0
0.3 - 1.1
0.3 - 0.8
0.7 - 1.4
1.6 - 1.7
0.9 - 1.4
0.9 - 1.4
0.9 - 1.2
Air exchange rates  range  from 0.1 to 2.0.  The mean  of the mid-

points of the 15 ranges listed in Table 6-7 is 0.9.   The mean of

the midpoints of the  particular residence types  are  listed below.
                   residence type

                    experimental
                    conventional
                    mobile
                    low-rise
                    high-rise
exchanges/h

   0,6
   0.9
   0.7
   1.1
   1.1
These results  suggest a typical ventilation rate  for a nonexperi-
mental home of  one  exchange per hour.
                           3
     Penkala and Oliveira estimate
                                 3
that one smoker  per 400 ft  in an enclosed  space  will add 6.2 mg/m

(5.4 ppm)  to  indoor CO if there is one air  exchange per hour.  Ac-

cording to U.S.  Census data,11* the average  number of rooms in a

living unit is 5.1.  Assuming the typical five  room house has a

floor area of 1300 square feet and a ceiling  8  feet high, we can

estimate that the typical living unit has a volume of 10,400 ft  .
                                6-12

-------
Housing data indicate that the average living unit has 2.1 adults.15
Repace and Lowrey15 estimate that one third of adults smoke.
Since some teenagers smoke, the average living unit has at least
0.7 smokers per 10,400 ft  or 0.027 smokers per 400 ft .  Smoker-
generated CO would be at least (0.027)(5.4 ppm) = 0.15 ppm.  A
house with 10,400 ft  and two smokers would have a smoker-generated
CO concentration of 0.42 ppm.  These levels are negligible.  In
fact, the number of smokers must be increased to five per 10,400
ft  for the smoker-generated CO concentration to exceed 1.00 ppm.
     In a sample of 69 homes, Spengler,  et al.,17 found 32 percent
had one smoker and 13 percent had two or more smokers.  From
these data we can estimate the average house with smokers has
about 1.3 smokers per 10,400 ft  or 0.05 smoker per 400 ft .
Smoker-generated CO concentration in such a house would be about
0.3 ppm.  Consequently, we used 0.3 ppm as our best estimate of
/\
a  .  for smoking households from 7 a.m.  to 9 a.m. and from 5 p.m.
 m / T.                  s\
to 11 p.m.  A smaller a   , 0.2 ppm, was considered appropriate
                       m 11        />.
from 9 a.m. to 5 p.m.  We assumed a  ,  = 0 from 11 p.m. to 7 a.m.
                                   m, t
     In the CO exposure analysis we are particularly interested
in kitchen and living room CO levels generated by gas stoves.
Peak home CO exposure is expected to occur in the kitchen during
and immediately after meal preparation.   We assume that typical
home CO exposure 'is better represented by CO levels in the living
room.  Data useful in estimating a    for gas stoves are provided
by several studies performed by Research Corporation of New
England.  Yocum, et al. , "* measured CO in two houses with gas
stoves and gas furnaces.  They found that "the heating system had
no measurable effect on the indoor or outdoor CO levels; however,
the gas fired stoves in each house had a significant influence on
indoor CO levels."' Figure 4 in Yocum, et al., shows kitchen levels
in house G-l exceeding outside levels by 3 ppm during meal prepara-
tion; living room levels exceeded outside levels by about 1.5 ppm.
In house G-2, kitchen and family room levels during meal prepara-
tion exceeded outdoor levels by 3.0 to 4.5 ppm and 1.0 to 1.5 ppm,
                              6-13

-------
respectively.  In a later study by Cote, et al.,ll indoor and out-
door CO levels were measured at four homes in Hartford, CT.
House 1 is a 2,000 ft  split-level with well-ventilated kitchen
occupied by a married couple and two children.  The wife smokes a
pack a day.  Yocum, et al., found that an attached garage made a
significant contribution to indoor CO at this house.  House 2 is
         2
a 1500 ft  two-story home with well-ventilated kitchen.  A
single adult lives there who seldom uses the stove.  House 3 is a
1,000 ft  apartment with a small, unventilated kitchen.  A non-
smoking couple and their 2 children live there.  House 4 also has
                                             2
two adults and two children.  It is a 1500 ft  ranch-style house
with kitchen open to other areas of the house.  Table 6-8 lists
the seasonal means of daily average CO concentrations measured in
various areas of the four houses.  Average kitchen and living
room CO values in house 1 exceed outside CO by 1010 ug/m   (0.88
ppm) and 590 yg/m  (0.52 ppm), respectively.  The contribution of
the attached garage is difficult to quantify.  Data from house 2
are probably atypical because of the infrequent stove use.  Houses
3 and 4 are not as well ventilated as house 1 and may be more
appropriate for determining typical a  values.  Average kitchen
CO exceeds average outside CO by 3040 ug/m   (2.66 ppm) and by
4120 yg/m   (3.60 ppm) in house 3 and by 6590 ug/m   (5.76 ppm) in
house 4.  Average living room CO exceeds average outside CO by
980 ug/m   (0.86 ppm)  and 1690 ug/m   (1.48 ppm) in house 3 and by
5780 ug/ra   (5.05 ppm) in house 4.  Closer examination of 2-hour
CO values included in the report reveals that the difference
between inside  (kitchen and living room) and outside CO levels is
usually greatest from 1600 to 1800  (4 p.m. to 6 p.m.) and is
usually smallest from 400 to 600  (4 a.m. to 6 a.m.).  Typical
mealtime CO levels seem to occur between the hours 1200 and 1400
 (noon and 2 p.m.).  Table 6-9 lists average differences between
inside and outside CO levels during these 2-hour periods for
houses 1, 3, and 4.
                                6-14

-------
       TABLE 6-8.  INDOOR/OUTDOOR CO DATA RECORDED BY COTE, ET AL.
                                                         11


House
1


2
3


4


Season
Spring-summer
Fall -winter
Fall -winter
Spring-summer
Spring-summer
Fall -winter
Fall -winter
Fa 11 -winter
Mean daily average CO concentration, ug/m3


Stove
—
4190
4790
3000
4310
7820
7130
9070


Kitchen
4490
3520
4210
-
-
6420
6620
9000

Living
Room
4070
3230
-
3080
3210
5070
-
8190


Bedroom
4170
-
3830
2900
2680
-
5500
-


Outside
3480
1670
2310
2940
2230
3380
2500
2410
Avg.
stove
usage
(min)
198
106
7
43
37
66
115
201
     Sterling and Sterling18  studied  the rate of CO buildup and
dissipation in kitchens, dining  rooms,  and living rooms of nine
homes in Burnaby, British Columbia.   Kitchen levels of CO in
house 1 increased from 6 ppm  to  36 ppm  in 30 minutes,  depending
on the number of burners on.
                    Burners on
                       1
                       2
                       3
                       4
 CO increase,
ppm per minute
    0.2
    0.63
    0.73
    1.20
Rates of increase for the other eight  homes  varied from 0.7 to
3.3 (number of burners on was not  specified).   The average rate
of CO increase for the nine homes  was  about  2  ppm.  Operating a
stove at this rate for 30 minutes  would  yield  an hourly average
of 30 ppm if CO decayed immediately.   Sterling and Sterling
found that CO decayed very slowly  in the test  homes and that it
diffused rapidly throughout the houses.
     An increase in kitchen CO of  30 ppm during meal preparation
is probably atypical since it is based on the  use of three to
four burners continually for 30 minutes.   None of the studies by
Research Corporation of New England suggest  meal-time CO levels
                               6-15

-------
              TABLE  6-9.  AVERAGE DIFFERENCES BETWEEN KITCHEN,
                 LIVING ROOM, AND OUTSIDE CO CONCENTRATIONS
House
1

3

4
Season
Spring/summer
Fall /winter
Spring/summer
Fall /winter
Fall /winter
Time of
day
4-6
12-14
16-18
4-6
12-14
16-18
4-6
12-14
16-18
4-6
12-14
16-18
4-6
12-14
16-18-
3
Difference in CO concentration, ug/m
Kitchen-outside
624 (ll)a
1109 (12)
1742 (12)
1941 (11)
2247 (12)
3380 (14)
-
3284 (21)
3372 (23)
3622 (22)
2704 (9)
7123 (7)
12,424 (9)
Living room-outside
385 (11)
743 (12)
832 (13)
1236 (5)
1154 (6)
2105 (8)
844 (16)
1062 (16)
915 (16)
1757 (9)
2189 (12)
1045 (10)
2256 (9)
6177 (7)
11,328 (9)
Numbers in parentheses  indicate number of days with data.
                                   6-16

-------
this high.  If the data provided by Yocum and Cote are assumed
to be more typical, a reasonable model for kitchen a  ,  in gas
                     ^                              m/ L.
stove homes would be a^ .  = 4.0 ppm during meal-time hours, and
/s                     m, t
a    =2.5 ppm other times.  Reasonable living room estimates
 m, t     /\                                         /v
would be a    = 2.0 ppm during meal-time hours and a  ,  = 1.0 ppm
          nt / n                                       in f t
other times.  Meal-time hours would be defined as the 2-hour
periods 600 to 800, 1100 to 1300, and 1700 to 1900.
     Although the home-other microenvironment includes nonresi-
dential locations such as shopping malls, a single set of a   . and
b  values is used for the combined microenvironment.  Spengler19
cites work by Chapin which suggests that 92 percent of people's
time characterized as spent in home-other microenvironments is
spent in the home.  Consequently, using the indoor home values
for the combined microenvironment should not significantly bias
exposure estimates.  We can assume that a cohort is at home
whenever its activity pattern places them in the home-other micro-
environment during a meal-time hour.  Using the gas stove esti-
mates for a    in these situations is reasonable.  At other times
of the day, home-other could indicate visits to a library,
courthouse, shopping center, sports arena, or doctor's office.
The principal CO source in these enclosed areas is probably
cigarette smoke, although Spengler, et al. ,20 have found  that ice
cleaning machines at hockey rinks can produce one-hour CO levels
exceeding 35 ppm.  Godin,  et al.,7 reported that CO in a theater
foyer where smoking was permitted exceeded CO in the auditorium
by 2 ppm.  Elliot and Rowe20 found an average CO concentration of
25 ppm in a sports arena (not air conditioned) where smoking was
permitted.  CO levels of 9 ppm were recorded in two other arenas
with posted "No Smoking" signs.  Average CO during periods of
nonactivity was 3 ppm in all three arenas.  Thompson, et al.,8
recorded average daytime CO levels in a hospital, YMCA pool,
department store, and shopping mall (see Table 6-10).
                              6-17

-------
          TABLE 6-10.  AVERAGE CO LEVELS IN VARIOUS STRUCTURES'
Kind of structure
community hospital
YMCA pool
department store
shopping mall
CO, ppm
Out
2.1
0.5
6.4
2.7
In
1.7
1.0
3.3
3.1
Thompson, et al., state that the inaccuracy of their analyzer,
+ 1.0 ppm, prevented critical comparison of most of the rather
low values obtained with the possible exception of the CO levels
measured at the department store.  Thompson, et al., suggest  the
following explanation for the relatively low indoor/outdoor
ratio.  Because auto exhaust emissions near the building would be
minimal at night, a mass of air with a minimal level of CO would
accumulate during the night.  If daytime ventilation rates are
low, the inside air would fail to come to equilibrium with out-
side CO.
     None of these studies provide dependable data on typical CO
levels in the "other" microenvironment.  Consequently, we used
the factors determined for "home" as the factors for the combined
home-other microenvironment.

6.3  TRANSPORTATION VEHICLE MICROENVIRONMENT
     The most commonly used transportation vehicle in the four
study areas is the automobile.  The principal internal sources of
CO in automobiles are'probably cigarette smoking and leaky exhausts
In the absence of these sources, available data indicate that
average interior CO is equal to or less than average exterior CO,
although exterior CO shows greater fluctuations.  Ott and Willits22
concluded that the average value of the interior CO concentra-
tion  is approximately equal to the average value of the exterior
CO concentration if the averaging time, T, was much greater  than
the time constant T.  They estimate T = 4.5 minutes for a test
                               6-18

-------
vehicle moving on residential side streets at 20 mph with windows
closed.  Since T decreases as speed increases or windows are open,
we can assume T»T for most moving vehicles.
     Ziskind, et al.,23 studied buses, cabs, and police cars in
Denver and Boston.  They found that interior concentrations "rise
and fall with exterior concentrations, yet are almost always
lower."  They hypothesize that the relatively small difference
between interior and exterior levels provides too small a driving
force for diffusion of CO into the vehicle.  Furthermore,'there
is insufficient time for the two concentrations to equilibriate,
since the external source is constantly changing as long as the
vehicle keeps moving.   Ziskind, et al., found that all vehicles
in their study having interior concentrations in excess of exter-
ior concentrations had both exhaust system leaks and pathways
through to the passenger area.  Since most of the vehicles which
were monitored continuously in their study were selected because
of high interior CO levels, their results cannot be applied to the
general vehicle population.
     Colwill and Hickman2* measured interior and exterior CO
levels of 11 new cars driven around a 35 km route in London.
They report inside/outside ratios of 0.35 to 0.75 with a mean
ratio of 0.55.  Although they did not relate inside CO levels
to stationary monitor readings, Colwill and Hickman state that
occupants of vehicles moving in heavy traffic are exposed to CO
levels higher than those recorded at curbside.
     Several studies provide data which relate interior CO
levels to fixed monitoring data directly.  When Ziskind, et al.,
compared personal sampler data with fixed site data, they found
that total exposures exceeded fixed site concentrations by an
average of 13.9 ppm.   An average ratio was not determined.
Ziskind, et al., also list average interior CO as measured by
continuous monitors in 9 vehicles (8 buses and 1 cab)  and the
average CO levels at corresponding fixed sites.  Interior/fixed
site ratios vary from 1.0 to above 7.0 with a median of 2.7.
Ziskind, et al., are uncertain how much of the difference between
                              6-19

-------
interior and fixed site CO "was due to vehicle self-contamination
and how much was due to the inherent lack of representativeness
of the fixed site monitoring station readings."  However, they
make the following inconsistent statement in their section list-
ing overall study conclusions:
     Typically the CO level measured inside or immediately out-
     side the vehicle significantly exceeded the value recorded
     by the nearest fixed site monitoring station.  Vehicle
     self-contamination does not appear to be the cause of this
     disparity.  Rather, it is postulated that the proximity of
     the vehicle to the emission sources accounts for the dif-
     ference between vehicle and fixed site monitor concentra-
     tions.2 3
     Wallace25 measured CO levels in cars and buses on 37 runs
 (27 by bus, 10 by car) around Washington, D.C.  Mean bus CO was
11.7 ppm, excluding one outlier; mean car CO was 13.8 ppm.  These
values are three to four times higher than mean CO measured simu-
taneously at a stationary monitor at 427 New Jersey Avenue, N.W.
However, Wallace found no significant relationship between ambient
concentrations and interior vehicular concentrations.  His results
suggest that factors associated with particular vehicles—power
source, design, and maintenance—may effect interior CO levels
more than exterior CO levels.
     A doctoral thesis by Cortese26 provides more definitive re-
sults.  In this study, population exposure to CO was measured
by equipping volunteers living and working in the metropolitan
Boston area with portable CO monitors.  The monitored cohort
consisted of 66 nonsmoking volunteers who carried a portable
monitor for 3 to 5 days during commuting and working activities.
Participants' commuting mode and route, residential and occupa-
tional location, exposure to cigarette smoke, and daily activities
were documented.  Volunteers were chosen from populations without
 significant occupational exposures to CO so that measured expos-
ures resulted from ambient air contamination.  Population

                              6-20

-------
exposure data, as measured by personal monitoring, were compared
to CO concentrations measured at 6 fixed location monitoring
stations operated by the Massachusetts Bureau of Air Quality
Control.  Two of the fixed location monitoring stations are
located in downtown Boston.  These urban stations approach
Federal siting criteria for monitoring maximum 1-hour exposure
to CO.  The other four stations are located in suburban areas.
These stations approach federal siting criteria for monitoring
8-hour average CO exposure but are not located close enough to
heavily traveled roadways to monitor maximum 1-hour exposure.
The following conclusions were drawn by Cortese.
     o    Measurements at 6 fixed locations in metropolitan
          Boston underestimated mean 1-hour CO exposure during
          commuting by a factor of 1.8 to 2.0.
     o    Measurements at the two urban monitoring stations,
          whose characteristics approach Federal criteria for
          monitoring maximum 1-hour exposures, underestimated
          the mean 1-hour CO exposure during commuting by a
          factor of 1.4.  Because Boston pedestrians can be
          closer to automobile traffic than the two urban
          stations, measurements from the stations would also
          underestimate pedestrian exposure to CO.
     o    Measurements at the four suburban monitoring stations
          underestimated mean 1-hour CO exposure during commuting
          by a factor of 2.1.  This result is significant because
          a large portion of the average commuting trip in this
          study occurred in suburban areas.
     o    Analysis of the highest 5-7% of the personal exposure
          and fixed location measurements, which are of greatest
          public health importance, indicated that fixed location
          measurements were better estimates of the higher commut-
          ing exposures than of the entire range of commuting
          exposures.  Nevertheless, the mean 1-hour personal
                              6-21

-------
exposure concentration was 1.6 times the mean concen-
tration at all fixed stations and 1.3 times the mean
concentration at urban stations.
10 to 15% of the difference between commuting exposures
and the concentrations measured by fixed location moni-
tors was attributed to an observed reduction in CO
concentrations with increased sampling height between
personal monitors at or near breathing zone (5.5 feet)
and fixed location monitors at a height of 15 feet.
The remainder of the difference was attributed to
commuters being closer to CO emission sources than
fixed location monitors.
No consistent relationship was observed between personal
exposure during commuting and fixed location measure-
ments over the entire range of values encountered.
This result made it impossible to develop a predictive
relationship between personal exposure and fixed loca-
tion measurements".
Mode of travel  (automobile, mass transit, split mode,
i.e., part auto, part transit) and route of travel were
the significant factors influencing personal exposure
to CO during commuting.  Cigarette smoke is the only
other significant source of CO to which a commuter may
be exposed.
Total travel by automobile resulted in a mean CO expo-
sure nearly twice that of rail mass transit commuting
and approximately 1.6 times that of split mode commuting.
Automobile commuting on 4-lane, heavily traveled
arterial roads resulted in a mean CO exposure approxi-
mately 1.4 times the mean exposure during automobile
commuting on other types of roads.
Wind speed, wind direction, season, and automobile age
did not influence commuter population exposure to CO.26
                     6-22

-------
Pertinent data from  the  Cortese study are summarized  in  Table 6-11,
These results suggest  1.4  <_ b  <_ 2.1 for unspecified  Boston trans-
portation vehicles during  commuting hours.  Ziskind's median
ratio of 2.7 may be  the  result of using some vehicles known to
have leaky exhausts  and  not using any rail transit.
         TABLE 6-11.  RATIOS OF MEAN PERSONAL CO EXPOSURES TO MEAN CO
                   CONCENTRATIONS AT FIXED MONITORS25
         mode of travel
         all vehicles
         all vehicles
         all vehicles
   fixed monitors
6 urban sites
2 urban sites
meeting EPA criteria
4 suburban sites
                                             mean personal
                                               exposure
mean monitor CO
  1.8 to 2.0
     1.4
     2.1
     An earlier study  by Brice and Roesler27 compared CO  in  motor
vehicles moving in moderate, to heavy traffic with concurrent con-
centrations measured at  CAMP sites in six cities.  Table  6-12 lists
results of the study.  The  mean of the five ratios is 3.5; the
median is 2.4.  Ratios of vehicles moving in light to moderate
traffic would probably be lower.  Most CAMP sites were located
in downtown areas; probes were usually positioned 15 feet off
the street.  Brice and Roesler state that the low ratio in Chicago
corresponds to a high  average concentration of CO at the  CAMP
site, which is attributed to the close proximity of that  site to
high-density traffic routes.   In-vehicle data for Cincinnati is
heavily weighted toward  downtown street canyons.  The average
ratio for major arteries in Cincinnati is 4.8.
     Petersen and Sabersky28 measured CO inside a car driving a
route in Los Angeles that included a business district, a resi-
dential district, a part of a generally uncrowded freeway, and a
part of a congested freeway.   During a 50-minute drive from  1:52
p.m. to 2:42 during the  summer,  average CO varied from 15 to 20
ppm.  The maximum reading reported by APCD for the day was 8 ppm
                              6-23

-------
and may not have occurred concurrently.   Consequently,  the ratio
of interior CO to fixed site CO  is  at  least 1.9.   These data are
too limited to make any firm estimates of b  for  Los Angeles,
however.
         TABLE 6-12.  RATIOS OF CO  IN MOTOR VEHICLES CONCURRENT
                      TO CO AT CAMP STATIONS27
                      City
                 Chicago
                 Cincinnati
                 Denver
                 St. Louis
                 Washington, D.C.
Interior CO/
  CAMP CO
   1.3
   6.8
   2.4
   2.1
   4.7
     The above studies suggest  that  b   for the transportation
microenvironment should fall between 1.3  and 4.7.   In the NEM
analysis, we used 2.1, the upper  range of Cortese's estimates,
since it incorporates movement  by motor vehicles and trains.  We
assumed reasonable bounds for b  would be 1.4, the smallest ratio
                                m
in Table 6-11, and"~3.5, ' the "mean  of.the ratios in Table 6-12.
     There are few data on typical levels of CO from cigarette
smoke in transportation vehicles.  Ziskind, et al.,23 report that
chi-square analysis of taxicab  data  showed that CO levels were
not significantly higher when drivers  and/or passengers smoked.
However, Harke, et al.,1 measured CO levels of 30 ppm in an un-
ventilated car with an outside  windspeed  of 50 km/hour when 9
cigarettes were smoked intermittently. CO levels averaged 5 to
6 ppm in a well-ventilated car  with  three people smoking contin-
uously.  Unfortunately, Harke does not give outside CO levels.
     Information on the percentage of  automobiles that contain
smokers and the average cigarette-generated CO levels on buses
and trains is unavailable.  Consequently, we let a   ._ = 0 for
                                                   m /1
smoking and assumed that our estimate  b  = 2.1 incorporates some
of the smoker-generated CO to which  commuters in Cortese's study

                                6-24

-------
were exposed.  We made the same assumption concerning CO from
leaky exhausts, since some of Cortese's subjects probably commuted
in cars with leaking exhaust systems.

6.4  ROADSIDE MICROENVIRONMENT
     Persons walking near roadways are usually closer to the
automobiles that produce CO than the nearest fixed CO monitor.
Consequently, fixed monitors usually underestimate roadside CO
levels.  If we assume a  .  = 0, then b  must exceed unity for
                       m/1            m
reasonable estimates of roadside levels.
     Two studies by Wilson and Schweiss29'30 provide data useful
in estimating b .  In 1977, Wilson and Schweiss measured 8-hour
               m
(10 am - 6 pm) CO values at 33 sites in the central business
district and 7 sites in nearby areas of Boise, Idaho, during
November and December, the season when high CO levels frequently
occur.  These values were compared to 8-hour values recorded at
the only continuous CO monitor in Boise.  The fixed site was
located in the center of the downtown business district.  Most of
the 40 study sites were near roadways (but not "hotspot" locations)
Sample probes were mounted 3.5 meters above the ground.  Roadside/
fixed monitor ratios ranged from 0.3 to 1.5.  The mean ratio was
0.92; the median ratio was 0.90.  These results suggest that the
fixed station may have been purposely sited in an area of Boise
with particularly high CO levels.
     Wilson and Schweiss conducted a similar study in Seattle,
collecting data from 36 outside samplers and 4 fixed-site moni-
tors.  Table 6-13 summarizes their results.  In this case, road-
side/fixed monitor ratios range from 0.69 to 2.22 and average
about 1.15.
     Jabara, et al.,31 measured the occupational exposure of
Denver traffic officers to CO during eight hour work shifts and
compared the results to ambient levels at fixed site monitors.
The ratio of mean dosimeter reading to mean fixed site reading
was 21.7/6.4 = 3.39.  Since traffic officers work in areas of

                              6-25

-------
             TABLE 6-13.  RATIOS OF MEAN CO CONCENTRATIONS AT
                 EXPERIMENTAL SITES AND AT FIXED SITES30
Fixed site
Pike St.
University St.
James St.
Fire station
Smejcor St.
mean
Study site CO/fixed site CO
Nearest
study site
0.69
1.07
0.89
2.22
0.98
1.16
2nd nearest
study site
1.03
1.10
1.25
1.26
1.09
1.15
congested traffic, this ratio is probably  high  for  the typical
             A more reasonable estimate of  b  would  be 1.2,  as
                                            m
pedestrian.
suggested by the Seattle data of Wilson  and  Schweiss.
                                                        We assumed
that b  should fall between 0.7 and  2.3,  and  used b  = 1.2 as our
      m                                             m
best estimate.
6.5  OTHER OUTDOOR LOCATIONS
     We assumed that CO levels at outdoor  locations  away from
roads could be approximately represented by  x.",  the monitor-
derived CO concentration,•with no lag  time or  additive factor.
Consequently, we used Equation 6-1  to  estimate CO levels in this
microenvironment.  Following the recommendations  of  Feagans,10
we assumed a  ,  = 0 and that 3. reasonable  range  for  b  would be
            m, t                                       m
0.90 to 1.00.  Feagans' best estimate  for  b  was  0.95.
                                           m
6.6  SUMMARY
     Tables 6-14 and 6-15 summarize  the  estimates of a  .  and b
                                                       m, t      m
for CO according to microenvironment,  room,  CO source, and time
of day.  Equation 6-3 was used to estimate CO  levels in the
work-school and home-other microenvironments.   Equation 6-1 was
used to estimate CO levels in the other  three  microenvironments.
                                6-26

-------
.ABLE 6-14.  ESTIMATES OF ADDITIVE MICROENVIRONMENTAL  FACTORS  (a_  . )
                                                              • n 5 u
Microenvironment
Indoors: work or
school

Indoors: home or
other








Transportation
vehicle
Roadside
Other outdoor
locations
Pollutant
source
none

smoking
none

smoking

gas stove





none
none

none
Room
all


all



kitchen


1 iving
room


NA
NA

NA
Hours ending
all

all
all

8,9,18-23
10-17
7,8,12,13,18,
19
1-6,9-11,
14-17,20-24
7,8,12,13,18,
19
1-6,9-11,
14-17,20-24
all
all

all
Estimated value (ppm)
low
0

0.2
0

0.1
0.1
1.0

0.5
0.7

0.3
0
0

0
best
0

0.3
0

0.3
0.2
4.0

2.0
2.5

1.0
0
0

0
high
0

0.5
0

0.5
0.5
11.0

3.0
10.0

2.0
0
0

0
                              6-27

-------
TABLE 6-15.   ESTIMATES  OF MULTIPLICATIVE MICROENVIRONMENTAL FACTOR (b )
Microenvironment
Indoors: work or school
Indoors: home or other
Transportation vehicle
Roadside
Other outdoor locations
Estimated value
low
0.60
0.70
1.40
0.70
0.90
best
0.85
0.85
2.10
1.20
0.95
high
1.05
1.10
3.50
2.30
1.00
                                6-28

-------
6.7  REFERENCES

 1.  H.  P.  Harke,  "The problem of passive smoking.  I.  The
     influence of  smoking on the CO concentration of office
     rooms," International Archives Arbeitsmedizin, Vol. 33,
     1974,  pp. 199-204.

 2.  S.  J.  Penkala and G. De Oliviera,  "The simultaneous analysis
     of  carbon monoxide and suspended particulate matter produced
     by  cigarette  smoking," Environmental Research, Vol. 9, 1975,
     pp. 99-114.

 3.  D.  J.  Moschandreas, J. Zabransky,  Jr., and D. J. Pelton,
     "Indoor air quality characteristics of the office environ-
     ment," Paper  no.  80-61.2, presented at the 73rd Annual Meeting
     of  the Air Pollution Control Association,  Montreal, Quebec,
     June 22-27,  1980.

 4.  J.  Yocom, et  al., A Study of Indoor-Outdoor Air Pollutant
     Relationships.   Volume I and II, Publication number
     APTD-0592, Research Corporation of New England, Hartford,
     Connecticut,  May 1970.

 5.  General Electric Company, Indoor-Outdoor Carbon Monoxide
     Pollution Study,  Publication number EPA-R4-73-020,  U.S.
     Environmental Protection Agency, Research Triangle Park,
     North Carolina,  December 1972.

 6.  R.  L.  Derham, G.  Peterson, R. H. Sabersky, and F. H.  Shair,
     "On the relation between the indoor and outdoor concentra-
     tions of nitrogen oxides," Journal of the Air Pollution
     Control Association, Vol. 24, No.  2 (February 1974),  pp.
     158-161.

 7.  G.  Godin, G.  Wright, and R. J. Shepard, "Urban exposure to
     carbon monoxide," Archives of Environmental Health, Vol. 25,
     1972,  pp. 305-313.

 8.  C.  R.  Thompson,  E. G. Hensel, and G. Kats, "Outdoor-indoor
     levels of six air pollutants," Journal of the Air Pollution
     Control Association, Vol. 23, No.  10 (October 1973).   pp.
     881-886.

 9.  D.  J.  Moschandreas, J. Stark, J. E. McFadden, and S.  S.
     Morse, Indoor Air Pollution in the Residential Environment,
     Vol. I.  Data Collection, Analysis, and Interpretation,
     Publication number EPA-600/7-78-229a,  U.S. Environmental
     Protection Agency, Research Triangle Park, North Carolina,
     December 1978.
                              6-29

-------
10.  Personal communication to Ted Johnson, PEDCo Environmental,
     from Thomas B. Feagans, Strategies and Air Standards Divi-
     sion, U.S. Environmental Protection Agency, Research Triangle
     Park, North Carolina  27711, December 1981.

11.  W. A. Cote, W. A. Wade III, and J. E. Yocum, A Study of
     Indoor Air Quality, Publication No. EPA-650/4-74-042, U.S.
     Environmental Protection Agency, Research Triangle Park,
     North Carolina, September 1974.

12.  D. P. Bridge and M. Corn, "Contributions to the assessment
     of non-smokers to air pollution from cigarette and cigar
     smoke in occupied spaces," Environmental Research, Vol. 5,
     1972, pp. 215-220.

13.  T. D. Sterling and D. M.  Kobayashi, "Exposure to Pollutants
     in Enclosed Living Spaces," Environmental Research, Vol. 13,
     pp. 1-35.

14.  U.S. Census data.

15.  U.S. Housing data.

16.  J. S. Repace and A. H. Lowrey,  "Indoor air pollution,
     tobacco smoke, and public health," Science, Vol.  208,
     May 2,  1980.

17.  John D. Spengler, et al. , Summary of Air Pollution Measure-
     ments ,  Air Quality Assessment Group, Harvard School of Public
     Health, Boston, Massachusetts.

18.  T. D. Sterling and E. Sterling, "Carbon monoxide levels in
     kitchens and homes with gas cookers," Journal of the Air
     Pollution Control Association,  Vol. 29,  No. 3 (March 1979),
     pp. 238-241.

19.  J. D. Spengler, B. G. Ferris,  Jr., and D. W.  Dockery,
     "Sulfur dioxide and nitrogen dioxide levels inside and
     outside homes and implications  on health effects  research,"
     Environmental Science and Technology, Vol.  13,  No. 10
     (October 1979), pp. 1276-1280.

20.  J. D. Spengler, K. R. Stone, and F. W.  Lilley,  "High carbon
     monoxide levels measured  in enclosed-skating rinks," Journal
     of the Air Pollution Control Association, Vol.  28, No.  8
     (August 1978), pp. 776-779.

21.  L. P. Elliot and D. R. Rowe, "Air quality during  public
     gatherings,"  Journal of the Air Pollution Control Association,
     Vol. 25, No.  6 (June 1975),  pp. 635-636.
                              6-30

-------
22.   W.  Ott and N. Willits, "Modeling the dynamic response of an
     automobile for air pollution exposure studies," Environmetries
     81f  Summaries of Conference Presentations, 1981, pp. 104-105.

23.   R.  A.  Ziskind, M. B. Rogozen, I. Rosner, and T. Carlin,
     Carbon Monoxide Intrusion in Sustained-Use Vehicles,
     Publication number SAI-068-80-535,  U.S. Environmental
     Protection Agency, Research Triangle Park, North Carolina,
     November 15, 1979.

24.   D.  M.  Colwill and A. J. Hickman, "Exposure of drivers to
     carbon monoxide,"" paper no. 79-59.3, 72nd Annual Meeting
     of  the Air Pollution Control Association, Cincinnati, Ohio,
     June 24-29, 1979.

25.   Lance Wallace, "Use of personal monitor to measure commuter
     exposure to carbon monoxide in vehicle passenger compartments,"
     paper no. 79-59.2, 72nd Annual Meeting of the Air Pollution
     Control Association, Cincinnati, Ohio, June 24-29, 1979.

26.   A.  D.  Cortese, Ability of Fixed Monitoring Stations to
     Represent Personal Carbon Monoxide Exposure, thesis sub-
     mitted to the Faculty of the Harvard School of Public Health,
     Boston, Massachusetts, April 1976.

27.   R.  M.  Brice and J. F..Roesler, "The exposure to carbon mon-
     oxide of occupants of vehicles moving in heavy traffic,"
     Journal of the Air Pollution Control Association, Vol. 16,
     No.  11 (November 1966), pp. 597-600.

28.   G.  A.  Peterson and R. H. Sabersky,  "Measurements of pollu-
     tants inside an automobile," Journal of the Air Pollution
     Control Association, Vol. 25, No. 10  (October 1975), pp.
     1028-1032.

29.   C.  B.  Wilson and J. W. Schweiss, Part 1.  Carbon Monoxide
     Study - Boise, Idaho, November 25 - December 22, 1977,
     Publication number EPA-910/9-78-055a, U.S. Environmental
     Protection Agency, Research Triangle Park, North Carolina,
     September 1975.

30.   C.  B.  Wilson and J. W. Schweiss, Carbon Monoxide Study,
     Seattle,  Washington, October 6 - November 2, 1977, publication
     no.  EPA 910/9-78-054, U.S. Environmental Protection Agency,
     Region X, Seattle, Washington, December 1978.

31.   J.  W.  Jabara, T. J. Keefe, H. J. Beaulieu, and R. M. Buchon,
     "Carbon monoxide:  dosimetry in occupational exposures in
     Denver, Colorado," Archives of Environmental Health, Vol.
     35,  No. 4 (July/August 1980).
                              6-31

-------
                            SECTION 7
                  EXPOSURE ESTIMATES FOR ADULTS
         WITH CARDIOVASCULAR DISEASE IN FOUR URBAN AREAS

     The computer output of NEM provides estimates of population
exposure for various measures of exposure and averaging times.
In the case of CO, NEM also estimates carboxyhemoglobin (COHb)
levels, an important indicator of the physiological effects of
CO on the exposed population.  In this section the results of NEM
analyses of CO exposure in the four study areas under various air
quality assumptions are summarized.  Extrapolations of these re-
sults to the nation are presented in Section 8.
     The exposure estimates presented in this report are for
adults with cardiovascular disease.  Adults are defined to be
those at least 18 years old.  Adults with peripheral vascular
disease are included in the subpopulation considered to have
cardiovascular disease.  Based on the currently available evidence,
this subpopulation is judged to be the most sensitive group of
persons with respect to CO-induced adverse health effects.
     Estimates for three alternative standards are presented in
Section 7.1.  A comparison of male and female estimates is made
in Section 7.2.  The impact on the exposure estimates of omitting
indoor sources from the analyses is discussed in Section 7.3.  A
brief discussion concerning the uncertainty about the accuracy of
the estimates is provided in Section 7.4.
7.1  "BEST ESTIMATE" RESULTS
     Tables 7-1 through 7-27 contain selected printouts of a NEM
analysis of exposure of adults with cardiovascular disease to
CO in the four study areas under various air quality assumptions.
Each table is identified as to CO/COHb indicator and air quality
standard being simulated.  CO exposure estimates are provided for
both 1- and 8-hour average CO concentrations.  In each case, the

                               7-1

-------
TABLE 7-1.  ESTIMATES OF OCCURRENCES FOR ADULTS  WITH  CARDIOVASCULAR
    DISEASE OF 1-HOUR CO EXPOSURES ABOVE SELECTED  CONCENTRATION
         VALUES ASSUMING 9 PPM/1 EXEX STANDARD IS  ATTAINED

1
1 CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20. 0
15.0
12.0
9.0
7.0
0.0
1
MAX. CONCENTRATION
ENCOUNTERS AT MAX.
I


CHICAGO







1,250
25,300
223,000
828,000
3,080,000
10,600,000
1,070,000,000
25.6
523


LOS ANGELES








5,790
145,000
1,310,000
4,930,000
22,100,000
2,670,000,000
21.6
5,790


PHILADELPHIA





1,280
1,280
3,800
22,900
141,000
339,000
1,140,000
4,120,000
1,020,000,000
36.0
1,270


ST LOUIS






709
6,910
13,500
39,200
216,000
878,000
2,830,000
416,000,000
32.0
707

                                 7-2

-------
TABLE 7-2,  ESTIMATES OF ADULTS WITH CARDIOVASCULAR DISEASE  WHO  HAVE
  1-HOUR CO EXPOSURES ABOVE SELECTED CONCENTRATION  VALUES  ASSUMING
                  9 PPM/1 EXEX STANDARD IS ATTAINED

CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM

CHICAGO







1,250
11,400
22,300
54,900-
109,000
121,000
122,000
25.6
523

LOS ANGELES








5,790
62,200
188,000
256,000
304,000
305,000
21.6
5,790

PHILADELPHIA





1,270
1,270
3,800
21,600
36,800
69,300
85,400
110,000
116,000
36.0
1,270

ST LOUIS






707
6,200
7,270
19,500
28,800
36,600
44,800
47,500
32.0
707
                                 7-3

-------
TABLE 7-3.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR DISEASE  WHOSE  MAXIMUM
       1-HOUR CO EXPOSURE OCCURS IN SELECTED CONCENTRATION RANGES
               ASSUMING 9 PPM/1 EXEX STANDARD IS ATTAINED
CONCENTRATION
RANGE
(PPM)
60.0 < C <= 100.0
55.0 < C <= 60.0
50.0 < C <= 55.0
45.0 < C <= 50.0
40.0 < C <= 45.0
35.0 < C <= 40.0
30.0 < C <= 35.0
25. 0 < C <= 30.0
20.0 < C <= 25.0
15.0 < C <= 20.0
12.0 < C <= 15.0
9.0 < C <= 12.0
7.0 < C <= 9.0
0.0 < C <= 7.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
CHICAGO







1,250
10,100
10,900
32,600
54,500
11,700
494
25.6
523
LOS ANGELES








5,790
56,400
126,000
63,000
47,600
1,360
21.6
5,790
PHILADELPHIA





1,280

2,520
17,900
15,100
32,500
16,000
25,100
5,910
36.0
1,270
ST LOUIS






709
5,490
1,070
12,200
9,240
7,860
8,160
2,730
32.0
707
                                   7-4

-------
TABLE 7-4.  ESTIMATES OF OCCURRENCES FOR ADULTS  WITH  CARDIOVASCULAR  DISEASE
     OF 8-HOUR CO EXPOSURES ABOVE SELECTED  CONCENTRATION  VALUES ASSUMING
                      9 PPM/1 EXEX STANDARD IS ATTAINED
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
ENCOUNTERS AT MAX.
CHICAGO










122
107,000
2,070,000
1,070,000,000
12.0
54
LOS ANGELES









/

153,000
2,170,000
2,670,000,000
10.6
4,460
PHILADELPHIA










24,600
322,000
1,030,000
1,020,000,000
14.0
29
ST LOUIS











66,500
429,000
416,000,000
11.5
11
                                   7-5

-------
TABLE 7-5.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR DISEASE WHO HAVE
  8-HOUR CO EXPOSURES ABOVE SELECTED CONCENTRATION VALUES ASSUMING
                  9 PPM/1 EXEX STANDARD IS ATTAINED
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM

CHICAGO










120
5,260
67,200
122,000
12.0
54
1

LOS ANGELES











5,790
147,000
305,000
10.6
4,460

PHILADELPHIA










11,900
34,300
46,300
116,000
14.0
29

ST LOUIS











13,300
24,200
47,500
11.5
11

                                  7-6

-------
TABLE 7-6.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR DISEASE WHOSE  MAXIMUM
   8-HOUR CO EXPOSURE OCCURS IN SELECTED CONCENTRATION  RANGES  ASSUMING
                    9 PPM/1 EXEX STANDARD IS ATTAINED

CONCENTRATION
RANGE
(PPM)
60.0 < C <= 100.0
55.0 < C <= 60.0
50.0 < C <= 55.0
45.0 < C <= 50.0
40.0 < C <= 45.0
35.0 < C <= 40.0
30.0 < C <= 35.0
25.0 < C <= 30.0
20.0 < C <= 25.0
15.0 < C <= 20.0
12.0 < C <= 15.0
9.0 < C <= 12.0
7.0 < C <= 9.0
0.0 < C <= 7.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM

•
CHICAGO










122
5,140
61,900
54,400
12.0
54

LOS ANGELES











5,790
141,000
158,000
10.6
4,460

PHILADELPHIA










11,900
22,400
14,000
68,000
14.0
29

ST LOUIS

'









13,300
10,900
23,300
11.5
11
                                   7-7

-------
TABLE 7-7.   ESTIMATES OF OCCURRENCES FOR ADULTS  WITH  CARDIOVASCULAR  DISEASE
             OF COHb LEVELS EXCEEDING SELECTED VALUES ASSUMING
                      9 PPM/1 EXEX STANDARD IS ATTAINED
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
ENCOUNTERS AT MAX.
CHICAGO











78,300
8,600,000
1,070,000,000
1.92
54
LOS ANGELES






-




115,000
19,800,000
2,670,000,000
1.87
701
PHILADELPHIA








71
1,490
5,150
236,000
2,540,000
1,020,000,000
2.31
15
ST LOUIS










21
53,400
2,000,000
416,000,000
2.02
5
                                    7-8

-------
TABLE 7-8.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE  WHO  EXPERIENCE
              COHb LEVELS EXCEEDING SELECTED  VALUES  ASSUMING
                     9 PPM/1 EXEX STANDARD  IS ATTAINED
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
CHICAGO











9,630
99,000
122,000
1.92
54
LOS ANGELES











5,790
253,000
305,000
1.87
701
PHILADELPHIA








35
685
2,730
34,300
86,200
116,000
2.31
15
ST LOUIS










21
15,700
34,000
47,500
2.02
5
                                    7-9

-------
TABLE 7-9.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE WHOSE MAXIMUM
              COHb LEVEL OCCURS IN SELECTED  RANGES ASSUMING
                    9 PPM/1 EXEX STANDARD  IS ATTAINED
COHB LEVEL
RANGE
(PERCENT)
3.70 < C <= 10.00
3.50 < C <= 3.70
3.30 < C <= 3.50
3.10 < C <= 3.30
3.00 < C <= 3.10
2.90 < C <= 3.00
2.70 < C <= 2.90
2.50 < C <= 2.70
2.30 < C <= 2.50
2.10 < C <= 2.30
2.00 < C <= 2.10
1.50 < C <= 2.00
1.00 < C <= 1.50
0.00 < C <= 1.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
CHICAGO











9,680
89,300
22,600
1.92
54
LOS ANGELES











5,790
248,000
52,000
1.87
701
PHILADELPHIA








36
651
2,040
31,600
51,900
30,200
2.31
15
ST LOUIS










21
15,700
18,300
13,500
2.02
5
                                   7-10

-------
TABLE 7-10.   ESTIMATES OF OCCURRENCES  FOR  ADULTS  WITH  CARDIOVASCULAR DISEASE
    OF 1-HOUR CO EXPOSURES ABOVE  SELECTED  CONCENTRATION  VALUES ASSUMING
                      12 PPM/1  EXEX  STANDARD  IS ATTAINED
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
ENCOUNTERS AT MAX.
CHICAGO






11,200
50,500
223,000
987,000
2,640,000
9,320,000
. 25,700,000
1,070,000,000
34.6
523
LOS ANGELES







17,400
145,000
1,310,000
4,840,000
13,900,000
55,800,000
2,670,000,000
29.1
5,790
PHILADELPHIA



1,280
1,280
1,280
3,800
33,100
141,000
397,000
1,070,000
3,700,000
8,990,000
1,020,000,000
49.0
1,270
ST LOUIS




709
6,910
6,910
24,000
36,100
243,000
623,000
2,000,000
5,430,000
416,000,000
44.0
707
                                   7-11

-------
TABLE 7-11.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE WHO  HAVE
  1-HOUR CO EXPOSURES ABOVE SELECTED CONCENTRATION  VALUES  ASSUMING
                  12 PPM/1 EXEX STANDARD IS  ATTAINED

1
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM

CHICAGO
'






11,200
11,400
22,300
59,200
108,000
119,000
122,000
122,000
34.6
523

LOS ANGELES







5,790
62,200
168,000
256,000
296,000
305,000
305,000
29.1
5,790

PHILADELPHIA



1,270
1,270
1,270
3,800
23,000
36,300
69,300
85,400
109,000
116,000
116,000
49.0
1,270

ST LOUIS




707
6,200
6,200
16,100
19,900
23,800
32,400
43,700
47,500
47,500
44.0
707
                                 7-12

-------
TABLE 7-12.   ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE  WHOSE  MAXIMUM
   1-HOUR CO EXPOSURE OCCURS IN SELECTED CONCENTRATION  RANGES ASSUMING
                    12 PPM/1 EXEX STANDARD  IS  ATTAINED
CONCENTRATION
RANGE
(PPM)
60.0 < C <= 100.0
55.0 < C <= 60.0
50.0 < C <= 55.0
45.0 < C <= 50.0
40.0 < C <= 45.0
35.0 < C <= 40.0
30.0 < C <= 35.0
25.0 < C <= 30.0
20.0 < C <= 25.0
15.0 < C <= 20.0
12.0 < C <= 15.0
9.0 < C <= 12.0
7.0 < C <= 9.0
0.0 < C <= 7.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM

CHICAGO






11.200
182
10,900
36,900
48,700
11,000
2,670
68
34.6
523

LOS ANGELES







5,790
56,400
126,000
68,000
39,500
8,790
714
29.1
5,790

PHILADELPHIA



1,280


2,520
19,200
13,800
32,500
16,000
23,200
7,850

49.0
1,270

ST LOUIS


I

709
5,490

9,900
3,840
8,830
3,590
11,300
3,770
38
44.0
707

                                   7-13

-------
TABLE 7-13.  ESTIMATES OF OCCURRENCES  FOR  ADULTS  WITH  CARDIOVASCULAR  DISEASE
    OF 8-HOUR CO EXPOSURES ABOVE SELECTED  CONCENTRATION  VALUES ASSUMING
                      12 PPM/1 EXEX STANDARD  IS ATTAINED
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
ENCOUNTERS AT MAX.
1


CHICAGO










4,470
61,300
1,880,000
10,900,000
1,070,000,000
16.0
54



LOS ANGELES











72,800
1,480,000
16,400,000
2,670,000,000
13.6
4,460



PHILADELPHIA




'





42,700
267,000
1,020,000
2,640,000
1,020,000,000
18.5
29



ST LOUIS










58
41,700
302,000
1,680,000
416,000,000
15.0
11


                                     7-14

-------
TABLE 7-14.   ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE  WHO  HAVE
  8-HOUR CO EXPOSURES ABOVE SELECTED CONCENTRATION  VALUES  ASSUMING
                  12 PPM/1 EXEX STANDARD IS ATTAINED
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
CHICAGO









1,900
5,260
67,300
93,300
122,000
16.0
54
LOS ANGELES










5,790
95,600
252,000
305,000
13.6
4,460
PHILADELPHIA









17,300
33,500
44,300
104,000
116,000
13.5
29
ST LOUIS









27
3,800
23,700
29,000
47,500
15.0
11
                                 7-15

-------
TABLE 7-15.   ESTIMATES OF ADULTS WITH  CARDIOVASCULAR  DISEASE WHOSE MAXIMUM
   8-HOUR CO EXPOSURE OCCURS IN SELECTED  CONCENTRATION  RANGES ASSUMING
                    12 PPM/1 EXEX STANDARD  IS  ATTAINED
CONCENTRATION
RANGE
(PPM)
60.0 < C <= 100.0
55.0 < C <= 60.0
50.0 < C <= 55.0
45.0 < C <= 50.0
40.0 < C <= 45.0
35.0 < C <= 40.0
30.0 < C <= 35.0
25.0 < C <= 30.0
20.0 < C <= 25.0
15.0 < C <= 20.0
12.0 < C <= 15.0
9.0 < C <= 12.0
7.0 < C <= 9.0
0.0 < C <= 7.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
CHICAGO









1,910
3,360
62,000
26,100
28,300
16.0
54
LOS ANGELES










5,790
89,800
156,000
53,500
13.6
4,460
PHILADELPHIA









17,300
16,200
10,800
59,600
12,400
18.5
29
ST LOUIS









29
8,770
14,900
5,310
18,500
15.0
11
                                   7-16

-------
TABLE 7-16.   ESTIMATES OF OCCURRENCES  FOR  ADULTS WITH  CARDIOVASCULAR DISEASE
             OF COHb LEVELS EXCEEDING  SELECTED  VALUES  ASSUMING
                      12 PPM/1  EXEX  STANDARD  IS ATTAINED
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
'
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
ENCOUNTERS AT MAX.
CHICAGO







134
2,460
16,600
39,700
1,240,000
32,700,000
1,070,000,000
2.52
24
LOS ANGELES








5,190
18,200
35,900
927,000
75,600,000
2,670,000,000
2.38
1,560
PHILADELPHIA




71
329
1,840
8,850
42,300
114,000
163,000
861,000
7,140,000
1,020,000,000
3.03
15
ST LOUIS







187
3,980
16,400
31,500
243,000
5,490,000
416,000,000
2.59
5
                                    7-17

-------
TABLE 7-17.   ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE WHO  EXPERIENCE
              COHb LEVELS EXCEEDING SELECTED  VALUES ASSUMING
                     12 PPM/1  EXEX  STANDARD  IS  ATTAINED
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
CHICAGO







78
1,250
3,770
8,910
60,000
120,000
122,000
2.52
24
LOS ANGELES








2,470
5,090
5,790
65,400
300,000
305,000
2.38
1,560
PHILADELPHIA




35
156
776
4,670
17,100
29,400
32,100
36,300
114,000
116,000
3.03
15
ST LOUIS







186
2,780
7,500
12,600
24,300
43,100
47,500
2.59
5
                                    7-18

-------
TABLE 7-18.
ESTIMATES OF ADULTS WITH CARDIOVASCULAR DISEASE  WHOSE  MAXIMUM
  COHb LEVEL OCCURS IN SELECTED RANGES  ASSUMING
        12 PPM/1 EXEX STANDARD IS ATTAINED

1
1 COHB LEVEL
RANGE
(PERCENT)
3.70 < C <= 10.00
3.50 < C <= 3.70
3.30 < C <= 3.50
3.10 < C <= 3.30
3.00 < C <= 3.10
2.90 < C <= 3.00
2.70 < C <= 2.90
2.50 < C <= 2.70
2.30 < C <= 2.50
2.10 < C <= 2.30
2.00 < C <= 2.10
1.50 < C <= 2.00
1.00 < C <= 1.50
0.00 < C <= 1.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM

CHICAGO







79
1,180
2,510
5,150
51,100
59,600
1,930
2.52
24

LOS ANGELES








2,470
2,620
700
59,600
234,000
5,660
2.38
1,560

PHILADELPHIA




36
121
621
3,890
12,400
12,300
2,730
4,630
77,500
2,050
3.03
15

ST LOUIS







187
2,590
4,720
5,130
11,700
18,800
4,460
2.59
5
                                    7-19

-------
TABLE 7-19.   ESTIMATES OF OCCURRENCES FOR ADULTS  WITH  CARDIOVASCULAR  DISEASE
    OF 1-HOUR CO EXPOSURES ABOVE SELECTED CONCENTRATION  VALUES ASSUMING
                      15 PPM/1 EXEX STANDARD IS ATTAINED
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
ENCOUNTERS AT MAX.

CHICAGO




11.200
21,900
63,300
212,000
667,000
2,520,000
6,190,000
19,300,000
49,300,000
1,070,000,000
43.6
523

LOS ANGELES





5,790
23,200
145,000
849,000
3.440,000
8,030,000
31,500,000
109,000,000
2,670,000,000
37.0
5,790

PHILADELPHIA
1,280
1,280
1,280
1,280
3,800
22,900
55,000
141,000
277,000
1,050,000
2,090,000
5,750,000
13,300,000
1,020,000,000
61.5
1,270
1

ST LOUIS

709
709
1,420
6,910
13,500
24,100
32,300
181,000
513,000
1,130,000
3,270,000
9,220,000
416,000,000
56.1
707

                                    7-20

-------
TABLE 7-20.   ESTIMATES OF ADULTS WITH  CARDIOVASCULAR  DISEASE WHO HAVE
  1-HOUR CO EXPOSURES ABOVE SELECTED CONCENTRATION  VALUES ASSUMING
                  15 PPM/1 EXEX  STANDARD  IS  ATTAINED
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
CHICAGO




11,200
11,200
19,200
22,300
53,200
108,000
118,000
122,000
122,000
122,000
43.6
523
LOS ANGELES





5,790
5,790
62,200
167,000
256,000
287,000
305,000
305,000
305,000
37.0
5,790
PHILADELPHIA
1,270
1,270
1,270
1,270
3,800
21,600
26,500
36,800
50,400
85,400
93,600
116,000
116,000
116,000
61.5
1,270
ST LOUIS

707
707
707
6,200
7,270
16,300
18,600
26,300
32,300
40,000
46,600
47,500
47,500
56.1
707
                                7-21

-------
TABLE 7-21.   ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE WHOSE MAXIMUM
   1-HOUR CO EXPOSURE OCCURS IN SELECTED CONCENTRATION  RANGES  ASSUMING
                    15 PPM/1 EXEX STANDARD IS  ATTAINED
CONCENTRATION
RANGE
(PPM)
60.0 < C <= 100.0
55.0 < C <= 60.0
50.0 < C <= 55.0
45.0 < C <= 50.0
"*0.0 < C <= 45.0
35.0 < C <= 40.0
30.0 < C <= 35.0
25.0 < C <= 30.0
20.0 < C <= 25.0
15.0 < C <= 20.0
12.0 < C <= 15.0
9.0 < C <= 12.0
7.0 < C <= 9.0
0.0 < C <= 7.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
CHICAGO




11,200

8,020
3,050
30,900
54,600
10,400
3,350
68

43.6
523
LOS ANGELES





5,790

56,400
105,000
89,100
30,500
17,700
714

37.0
5,790
PHILADELPHIA
1,280



3,520
17,900
4,880
10,300
13,600
35,000
8,270
22,700


61.5
1,270
ST LOUIS

709


5,490
1,070
8,990
2,380
7,670
6,010
7,640
6,660
880

56.1
707
                                    7-22

-------
TABLE 7-22.   ESTIMATES OF OCCURRENCES  FOR  ADULTS  WITH  CARDIOVASCULAR DISEASE
    OF 8-HOUR CO EXPOSURES ABOVE  SELECTED  CONCENTRATION  VALUES ASSUMING
                      15 PPM/1  EXEX  STANDARD  IS ATTAINED
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
ENCOUNTERS AT MAX.
CHICAGO







4
122
53,500
939,000
7,390,000
29,100,000
1,070,000,000
20.0
120
LOS ANGELES








55,900
647,000
8,150,000
43,200,000
2,670,000,000
17.0
4,460
PHILADELPHIA







12,600
243,000
704,000
2,000,000
5,940,000
1,020,000,000
23.0
29
ST LOUIS








38,900
186,000
1,020,000
3,670,000
416,000,000
18.5
11
                                    7-23

-------
TABLE 7-23.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE  WHO  HAVE
  8-HOUR CO EXPOSURES ABOVE SELECTED CONCENTRATION  VALUES  ASSUMING
                  15 PPM/1 EXEX STANDARD IS ATTAINED

CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM

CHICAGO








120
5,260
54,500
84,200
118,000
122,000
20.0
120

LOS ANGELES








-
5,790
12,200
234,000
298,000
305,000
17.0
4,460

PHILADELPHIA








5,870
33,500
36,600
81,800
116,000
116,000
23.0
29

1
1
ST LOUIS









8,300
23,100
28,800
36,100
47,500
18.5
11
                                  7-24

-------
TABLE 7-24.   ESTIMATES OF ADULTS WITH  CARDIOVASCULAR  DISEASE WHOSE MAXIMUM
   8-HOUR CO EXPOSURE OCCURS IN SELECTED  CONCENTRATION  RANGES ASSUMING
                    15 PPM/1 EXEX STANDARD  IS  ATTAINED
CONCENTRATION
RANGE
(PPM)
60.0 < C <= 100.0
55.0 < C <= 60.0
50.0 < C <= 55.0
45.0 < C <= 50.0
40.0 < C <= 45.0
35.0 < C <= 40.0
30.0 < C <= 35.0
25.0 < C <= 30.0
20.0 < C <= 25.0
15.0 < C <= 20.0
12.0 < C <= 15.0
9.0 < C <= 12.0
7.0 < C <= 9.0
0.0 < C <= 7.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
CHICAGO








122
5,140
49,300
29,700
33,500
3.S90
20.0
120
LOS ANGELES









5,790
6,440
222,000
64,400
7,130
17.0
4,460
PHILADELPHIA








5, 670
27,600
3,050
45,300
34,500

23.0
29
ST LOUIS









8,300
14,800
5,730
7,260
11,500
18.5
11
                                   7-25

-------
TABLE 7-25.
ESTIMATES OF OCCURRENCES  FOR ADULTS  WITH  CARDIOVASCULAR  DISEASE
 OF COHb LEVELS EXCEEDING SELECTED VALUES ASSUMING
         15 PPM/1 EXEX STANDARD  IS ATTAINED
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CCNC.
ENCOUNTERS AT MAX.
CHICAGO



67
240
1,130
6,330
21,300
71,200
180,000
361,000
5,120,000
67,400,000
1,070,000,000
3.10
54
LOS ANGELES





1,560
8,650
21,400
61,000
193,000
319,000
6,320,000
166,000,000
2,670,000,000
2.89
701
PHILADELPHIA
79
833
1,910
6,040
11,900
24,300
65,400
131,000
216,000
371,000
493,000
1,850,000
15,100,000
1,020,000,000
3.75
15
ST LOUIS



120
332
1,640
7,900
20,400
46,300
84,800
108,000
789,000
12,500,000
416,000,000
3.16
5
                                    7-26

-------
TABLE 7-26.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE WHO  EXPERIENCE
              COHb LEVELS EXCEEDING  SELECTED  VALUES ASSUMING
                     15 PPM/1  EXEX  STANDARD  IS  ATTAINED


COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
1




CHICAGO




66
81
800
2,390
4,200
9,170
20,100
32,500
85,300
122,000
122,000
3.10
54





LOS ANGELES






1,560
2,600
5,630
5,790
8,900
16,300
230,000
305,000
305,000
2.89
701





PHILADELPHIA

43
398
822
2,810
6,380
11,300
22,800
30,700
34,300
35,300
35,400
53,400
116,000
116,000
3.75
15


-


ST LOUIS




119
331
1,240
4,160
9,510
15,700
21,600
23,000
28,700
43,900
47,500
3.16
5


                                    7-27

-------
TABLE 7-27.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE WHOSE MAXIMUM
               COHb LEVEL OCCURS IN  SELECTED  RANGES ASSUMING
                     15 PPM/1  EXEX STANDARD  IS  ATTAINED
COHB LEVEL
RANGE
(PERCENT)
3.70 < C <= 10.00
3.50 < C <= 3.70
3.30 < C <= 3.50
3.10 < C <= 3.30
3.00 < C <= 3.10
2.90 < C <= 3.00
2.70 < C <= 2.90
2.50 < C <= 2.70
2.30 < C <= 2.50
2.10 < C <= 2.30
2.00 < C <= 2.10
1.50 < C <= 2.00
1.00 < C <= 1.50
0.00 < C <= 1.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM

CHICAGO



67
15
720
1,590
1,810
4,970
10,900
12,400
52,600
36,200
70
3.10
54

'
LOS ANGELES





1,560
1,040
3,030
161
3,110
7,360
214,000
74,700
639
2.89
701

PHILADELPHIA
44
356
423
1,980
3,570
4,880
11,500
7,870
3,640
981
132
23,000
57,900

3.75
15

ST LOUIS


1
120
212
909
2,920
5,350
6,160
5,930
1,400
5,730
15,200
3,590
3.16
5

                                    7-28

-------
"best-estimate" raicroenvironment factors developed in Section 6
were used to simulate the contribution of gas stoves and smoking
to total CO exposure.
7.1.1  Attainment of 9 ppm/1 ExEx Standard
     NEM estimates in Tables 7-1 through 7-9 were developed by
adjusting the air quality data for each study area using the
roll-back" formula described in Section 5.1 so that the most pol-
luted neighborhood type just meets a "9 ppm/1 ExEx" standard,
i.e., one specifying that the expected number of 8-hour CO values
exceeding 9 ppm shall not be greater than one per year.  Table
7-1 provides estimates of the number of occurrences for adults
with cardiovascular disease of 1-hour exposures to CO concentra-
tions exceeding selected values.  (Exposures exactly equal to
zero are counted as exceeding zero.)  Thus, each column in Table
7-1 presents a cumulative frequency distribution in which the
number of 1-hour exposures increases as CO concentration decreases;
the distribution reaches a maximum at a CO concentration of zero.
This maximum is the number of adults with cardiovascular disease
used in the simulation times the number of possible occurrences
in a year (8760).  Although NEM yields individual frequency distri-
butions for cohorts who are at low,  medium, and high activity lev-
els when a given CO concentration is encountered, only the total
frequency distribution for all activity levels is presented in
Table 7-1.  According to these estimates, none of the four study
areas would have more than 6,910 occurrences of 1-hour CO expo-
sures above 25 ppm if a 9 ppm/1 ExEx standard were just attained.
     Table 7-2 uses an alternative exposure indicator, adults with
cardiovascular disease with 1-hour exposures.  This is the number
of adults with cardiovascular disease in the study area that
experience one or more 1-hour exposures per year to CO concentra-
tions that exceed a specified value.  This exposure indicator is
also expressed as a cumulative frequency distribution.  The number
of adults with cardiovascular disease exposed at zero concentra-
tion (or above) is the total population of the study area.

                              7-29

-------
     Table 7-3 provides estimates of the number of adults with
cardiovascular disease who experience their peak exposure of the
year within selected intervals of 1-hour CO concentrations.  These
estimates are not cumul'ative; each peak exposure falls within a
single interval.
     Tables 7-4 through 7-6 are similar to Tables 7-1 through 7-3
except that exposures are estimated in terms"of 8-hour running
average CO concentrations.  Because the average of any 8 succes-
sive hourly concentrations is less than or equal to the highest
value in the series, pollutant exposures usually occur at lower
concentrations for 8-hour running averages than for 1-hour averages.
For example, the maximum 8-hour running average concentration
experienced in Chicago is 12.0 ppm (Table 7-4), while the maximum
1-hour concentration is 25.6 ppm  (Table 7-1).  Similarly, the
number of 8-hour running average exposures above 9 ppm is 107,000
in Table 7-4, compared with 3,080,000 1-hour average exposures
in Table 7-1.
     Table 7-28 lists the general algorithm used by NEM to esti-
mate COHb levels in the exposed populations.  Specific values
assigned to the variables in the algorithm are listed in Table
7-29.  Full documentation of the rationale for the choice of
these values is provided in an EPA memorandum.:  A brief summary
of the reasons for these choices is given below.  Sensitivity
analysis runs exploring the impact on exposure estimates of using
alternative values for some of these variables are discussed in
Section 7.4.
     The value used for the Haldane constant is 218.  This value
comes from the study by Rodkey, et al.2  Values ranging from 210
to 250 have been reported in the literature.  The Clean Air Scien-
tific Advisory Committee  (CASAC) CO Subcommittee has recommended
218 as a best estimate.
     The value used for the hemoglobin level in the blood is 13.8
g/100 ml for adult females and 15.7 g/100 ml for adult males.
These are the mean values found in HEWs National Health and Nutri-
tion Examination Survey  (NHANES)  for adult males and females aged
18-74 years.3  Since the exposure estimates are for adults with
                               7-30

-------
         TABLE 7-28.   ALGORITHM USED TO CALCULATE  CARBOXYHEMOGLOBIN
                             IN BLOOD OF COHORTS
1.  Given the altitude,  calculate average barometric  pressure  (Pg):
          PB = 760*exp(-0.0000386*Alt)
2.  Calculate capillary  oxygen pressure (P-02):
          PC02 = 0.209*(PB-47) - 46.9
3.  Calculate quantity B:
          B = (1/DL)  + (PB -47)/VA
4.  Let (%02Hb) = 100
5.  Calculate quantity A:
          A = pc02/(M*(%02Hb))
6.  Calculate quantity F:
          F = exp(-t*A*60*104/(1.38*Hb*Vb*B))
7.  Calculate trial  (%COHb) value:
          (%COHb) =  (%COHb)Q*F + (B*VCQ + (Pg-47)*10~6*(CO))*(l-F)/A
8.  Calculate (%02Hb) value for next iteration:
          New (%02Hb) =  100*(%02Hb)/((%02Hb)  + (%COHb))
9.  Starting with the new value of (%02Hb) repeat  Steps  5  through  8.   Com-
    pare the (%COHb)  calculated with that from the previous  iteration.
    Repeat cycle until two successive COHb values  agree  within the desired
    accuracy.
                                    7-31

-------
            TABLE 7-29.   VALUES  ASSIGNED  TO  VARIABLES  IN ALGORITHM
                      USED TO  ESTIMATE  CARBOXYHEMOGLOBIN
     Variable
   Category
     Value
Haldane constant

Hemoglobin concentration


Endogenous CO
  production rate

Alti tude

CO diffusion rate


Blood volume


Ventilation rate
All

Females
Males

Females
Males

All

Females
Males

Females
Males

Low exercise
Medium exercise
High exercise
218.0

13.8 grams/100 ml
15.7 grams/100 ml

0.0062 ml/min
0.0081 ml/min

0

31 ml/min/torr
34 ml/min/torr

4,800 ml
5,800 ml

8,000 ml/min
20,000 ml/min
35,000 ml/min
                                   7-32

-------
cardiovascular disease, values were not developed for the two age/
occupation groups consisting of children.
     The value used for the endogenous CO production rate (VCQ)
is 0.0081 ml/min for adult males and 0.0062 ml/min for adult
females.  These values are simple weighted  (by the number of sub-
jects) averages of the results of six studies for males'*'9 and
four studies for females5"8 reported in the literature.
     The value used for the CO diffusion rate in the lung is 34
ml (min-mm Hg) for adult males and 31 ml (min*mm Kg) for adult
females.  These values are taken from Joumard, et al.10
     The values used for blood volume are 5,800 ml for adult males
and 4,800 ml for adult females.  Each of these values was calcu-
lated by multiplying two other values:  the 74 ml/kg body weight
for average males and 73 ml/kg body weight for average females
reported by Sjostrand1l multiplied by 78 kg average weight for
adult males and 65 kg average weight for females, respectively.
The latter two values are based on data provided by a publication
of the U.S. National Center for Health Statistics.12
     Ventilation rates used for both adult males and adult females
are 8 liters/min for a low exercise level,  20 liters/min for a
medium exercise level, and 35 liters/min for a high exercise level.
The basis for these values is a study by Niinimaa, et al.13  The
low exercise level value represents sleeping and sitting, the
medium exercise level value represents walking and other light
forms of exercise, and the high exercise level value represents
forms of exercise more strenuous than walking.  Obviously these
three categories represent a partitioning of a continuum of exer-
cise levels (see Table A-l of the Office of Air Quality Planning
and Standards Staff Paper on Sulfur Oxides1").
     In essence, the algorithm presented in Table 7-28 estimates
the COHb levels of an individual at the end of every hour of the
year.  Although COHb levels are, strictly speaking, the result
of CO exposure, they can be described using concepts similar to
those used for CO exposure.  For example, Table 7-7 lists the
                              7-33

-------
number of occurrences of COHb levels that exceed selected values.
Table 7-8 lists the number of adults with cardiovascular disease
that experience COHb levels which exceed selected values.  Table
7-9 lists the number of adults with cardiovascular disease who
experience their highest COHb level within selected ranges of COHb
values.  As would be expected, Tables 7-7 and 7-8 present cumula-
tive distributions, while Table 7-9 lists results in discrete
intervals.
     The relative frequencies of high COHb levels among the four
study areas can be compared by normalization, i.e., by converting
the estimates of adults with cardiovascular disease experiencing
different COHb levels to the corresponding percentage of total
adults with cardiovascular disease in the study area population.
Table 7-30 shows that none of the study areas have adults with
cardiovascular disease with COHb levels exceeding 3.0 percent
under the 9 ppm/1 ExEx standard.  Approximately 2.4 percent of
the Philadelphia adults with cardiovascular disease experience
COHb levels exceeding 2.00 percent.  Maximum COHb levels are 1.92
percent for Chicago, 1.87 percent for Los Angeles, 2.31 percent
for Philadelphia, and 2.02 percent for St. Louis.
     As previously noted, the estimates presented in the tables
are for cardiovascular adults.  The values used for the percentage
of adult females with cardiovascular disease was 4.2% and for
adult males 5.8%.  These values are based on U.S. Department of
Health, Education, and Welfare data.15  In this application of
NEM, estimates for the whole population were ratioed down to the
estimates for cardiovascular adults by using these two values in
conjunction with estimates of the percentages of adults who are
male and female in each of the four cities  (52% female and 48%
male).  The fact that married women are all female was accounted
for in the calculation, but the fact that the male/female percent-
age breakdown varies in general from one age/occupation group to
another was not.
     The estimates use 1970 census data for the four cities but
are projected to 1987 by using the multiplicative factor 1.195.

                              7-34

-------
TABLE 7-30.
PERCENTAGE OF ADULTS WITH CARDIOVASCULAR  DISEASE  EXPERIENCING
 COHb LEVELS EXCEEDING SELECTED VALUES  ASSUMING
        9 PPM/1 EXEX STANDARD IS ATTAINED
COHb level
exceeded
(percent)
3.00
2.90
2.80
2.70
2.60
2.50
2.40
2.30
2.10
2.00
1.50
1.00
0.00
Max. COHb
cone.
Percent
at maximum


Chicago










7.93
81.15
100.00

1.92

0.04


Los Angeles










1.90
82.95
100.00

1.87

0.23


Philadelphia







0.03
0.59
2.35
29.57
74.31
100.00

2.31

0.01


St. Louis









0.04
33.05
71.58
100.00

2.02

0.01
                                   7-35

-------
The 1.195 multiplicative factor is the product of 1.115, the
ratio of 1980 total U.S. population to 1970 total U.S. population,
and 1.072, a growth factor corresponding to a projected 1 percent
growth each year from 1980 to 1987.  That is, 1.072 is approxi-
mately equal to (1.01)7 and 1.195 is approximately equal to
1.114 x 1.072.
7.1.2  Attainment of 12 ppm/1 ExEx Standard
     Tables 7-10 through 7-18 provide NEM estimates based on the
assumption that the most polluted neighborhood type in each study
area just meets a standard specifying that the expected number of
8-hour CO values exceeding 12 ppm shall not exceed one per year.
This "12 ppm/1 ExEx" standard is less stringent than the 9 ppm/1
ExEx standard.  Tables 7-10 through 7-12 provide 1-hour exposure
estimates; Tables 7-13 through 7-15 provide 8-hour exposure
estimates; and Tables 7-16 through 7-18 provide COHb estimates.
Table 7-31 provides normalized COHb estimates for the 12 ppm/1
ExEx standard.  Note that all study areas have maximum COHb
levels which equal or exceed 2.38 percent under this assumption;
the maximum COHb level in Philadelphia is 3.03 percent.
7.1.3  Attainment of 15 ppm/1 ExEx Standard
     NEM estimates in Tables 7-19 through 7-27 are based on the
assumption that the most polluted neighborhood type will just
meet a standard specifying that the expected number of 8-hour CO
values exceeding 15 ppm shall not exceed one per year.  The "15
ppm/1 ExEx" standard is the least stringent of the three standards
analyzed.  Normalized COHb estimates for this standard are listed
in Table 7-32.  As expected, COHb levels are higher under the 15
ppm/1 ExEx standard than under the 12 ppm/1 ExEx and 9 ppm/1 ExEx
standards.  All four study areas have COHb levels which equal or
exceed 2.89 percent.  Philadelphia has a maximum COHb level of
3.75 percent, compared with a maximum of 2.31 percent estimated
for the 9 ppm/1 ExEx case.
                              7-36

-------
TABLE 7-31.
PERCENTAGE OF ADULTS WITH CARDIOVASCULAR  DISEASE  EXPERIENCING
 COHb LEVELS EXCEEDING SELECTED VALUES  ASSUMING
        12 PPM/1 EXEX STANDARD IS  ATTAINED
COHb level
exceeded
(percent)
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
Max. COHb
cone.
Percent
at maximum
Chicago



0.06
1.02
3.09
7.30
49.18
98.36
100.00
2.52
0.02
Los Angeles




0.81
1.67
1.90
21.44
98.36
100.00
2.38
0.51
Philadelphia
0.03
0.13
0.67
4.03
14.74
25.34
27.67
31.72
98.28
100.00
3.03
0.01
St. Louis



0.39
5.85
15.79
26.53
51.16
90.74
100.00
2.59
0.01
                                  7-37

-------
TABLE 7-32.
PERCENTAGE OF ADULTS WITH CARDIOVASCULAR  DISEASE  EXPERIENCING
 COHb LEVELS EXCEEDING SELECTED  VALUES  ASSUMING
        15 PPM/1 EXEX STANDARD IS  ATTAINED
COHb level
exceeded
(percent)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
Max. COHb
cone.
Percent
at maximum
Chicago



0.05
0.07
0.66
1.96
3.44
7.52
16.48
26.64
69.92
100.00
100.00
3.10
0.04
Los Angeles





0.51
0.85
1.85
1.90
2.92
5.34
75.41
100.00
100.00
2.89
0.23
Philadelphia
0.04
0.34
0.71
2.42
5.50
9.74
19.66
26.47
29.57
30.43
30.52
50.34
100.00
100.00
3.75
0.01
St. Louis



0.25
0.70
2.61
8.76
20.02
33.05
45.47
48.42
60.42
92.42
100.00
3.16
0.01
                                   7-38

-------
7.2  MALE/FEMALE COMPARISONS
     In this section, a brief comparison is made between selected
COHb estimates for males and for females.  The estimates for females
are in Table 7-33, and the estimates for males are in Table 7-34.
The estimates are based on the assumption that a 9 ppm/1 ExEx
8-hour average standard is met in each of the four study areas.
     The difference in the number of males and females estimated
to exceed selected COHb levels under the same standard is a result
of three factors.  First, different values were assigned to vari-
ous physiological variables for males and females  (see Section
7.1.1).  The differences in these values result in higher esti-
mated COHb levels in the blood of females, assuming the same
pattern of CO exposure.  Second, a slightly greater percentage of
the adult population is female.  Third, the exposure estimates
are for males and females with cardiovascular disease and reflect
the fact that only 4.2 percent of adult females are estimated to
have cardiovascular disease, whereas 5.8 percent of adult males
are estimated to have cardiovascular disease.
     Combining the last two factors, there are approximately 27
percent more adult males with cardiovascular disease in each
study area than adult females.  The fact that there is a slightly
greater percentage of females in the population is outweighed by
the more significant difference between the percentage of adult
males and females who have cardiovascular disease.
     Comparing the estimates in Table 7-33 with the estimates in
Table 7-34, it is apparent that the male/female differences in
physiology have a significant impact on the COHb levels which
result from a given pattern of CO exposure.  More cardiovascular
females reach the highest COHb levels despite there being more
cardiovascular males.  The difference in physiology gives this
result within the model since all cohorts include some females.
The greater number of cardiovascular males begins to dominate as
comparisons between the two move down in COHb level.
                              7-39

-------
TABLE 7-33.  ESTIMATES OF ADULT FEMALES  WITH  CARDIOVASCULAR  DISEASE
   WHO EXPERIENCE COHb LEVELS EXCEEDING  SELECTED  VALUES  ASSUMING
                 9 PPM/1 EXEX STANDARD IS ATTAINED
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
CHICAGO











11,800
85,500
103,000
1.93
103
LOS ANGELES











4,910
219,000
259,000
1.88
1,330
PHILADELPHIA








66
1,040
4,480
29,700
74,500
98,800
2.32
30
ST LOUIS










39
15,300
29,300
40,400
2.03
11
                               7-40

-------
TABLE 7-34.   ESTIMATES OF ADULT MALES WITH CARDIOVASCULAR  DISEASE  WHO
      EXPERIENCE COHb LEVELS EXCEEDING SELECTED  VALUES  ASSUMING
                  9 PPM/1 EXEX STANDARD IS ATTAINED
COHB LEVEL
EXCEEDED
( PERCENT )
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
CHICAGO











7,250
114,000
142,000
1.83
142
LOS ANGELES











6,770
292,000
357,000
1.75
4,070
PHILADELPHIA









289
746
39,500
99,400
136,000
2.19
41
ST LOUIS











16,100
39,500
55,800
1.86
15
                                7-41

-------
7.3  THE SIGNIFICANCE OF INDOOR SOURCES
     The exposure estimates discussed in Section 7.1 assume gas
stoves and smoking contribute to total CO exposure.  To evaluate
the significance of these CO sources, we can repeat the analysis
with the all additive microenvironment factors set equal to zero
(i.e., a    = 0 for all microenvironments).   Tables 7-35 and 7-36
        in f L.
are sample output of such an analysis.  They provide estimates of
the number of people exposed to 1-hour and 8-hour CO concentrations
under the 9 ppm/1 ExEx standard in the absence of indoor sources.
Comparison with Tables 7-2 and 7-5 reveals that indoor sources
have a minor effect on 1-hour and 8-hour exposures.  Maximum 1-hour
CO exposures are less than 1.0 percent higher when indoor sources
are included.  Maximum 8-hour CO exposures are 1.0 to 7.7 percent
higher when indoor sources are included.
     Tables 7-37 through 7-39 provide three indicators of COHb
levels in exposed populations in the absence of indoor sources.
These tables can be compared to Tables 7-7 through 7-9 to deter-
mine the significance of indoor sources on COHb levels.  For the
four study areas analyzed, maximum COHb levels are only 1.0 per-
cent  (St. Louis) to 4.1 percent (Philadelphia) higher when indoor
sources are included.
     These results are not unexpected.  In the NEM model, peak CO
levels are generally experienced in transportation vehicles or
along roadways—microenvironments with "best-estimate" multiplica-
tive factors of 2.10 and 1.20, respectively, and additive factors
equal to zero.  As discussed in section 6.3, the additive factor
corresponding to smoking in transportation vehicles was set equal
to zero because the multiplicative factor was assumed to already
incorporate this CO source.

7.4  UNCERTAINTY IN NEM EXPOSURE ESTIMATES
     Any method used to estimate exposure of large, diverse groups
of people must deal with a myriad of complexities.  The exposure
model can only represent major structural features.  Because the

                              7-42

-------
TABLE 7-35.  ESTIMATES OF ADULTS WITH  CARDIOVASCULAR  DISEASE WHO HAVE
    1-HOUR CO EXPOSURES ABOVE SELECTED VALUES  UNDER 9 PPM/1 EXEX
                STANDARD WITH INDOOR SOURCES OMITTED
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
CHICAGO







1,250
11,400
22,300
53,200
108,000
119,000
122,000
25.7
523
LOS ANGELES








5,790
62,200
188,000
256,000
299,000
305,000
21.7
5,790
PHILADELPHIA





1,270
1,270
3,800
21,600
36,300
69,300
85,400
99,000
116,000
36.0
1,270
ST LOUIS






708
6,200
7,280
19,500
28,800
36,600
43,400
47,500
32.0
708
                                7-43

-------
TABLE 7-36.   ESTIMATES OF ADULTS WITH  CARDIOVASCULAR  DISEASE WHO  HAVE
    8-HOUR CO EXPOSURES ABOVE SELECTED VALUES  UNDER 9 PPM/1 EXEX
                STANDARD WITH INDOOR SOURCES OMITTED
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
CHICAGO











5,260
54,200
122,000
11.6
147
LOS ANGELES







-

"

5,790
13,400
305,000
10.5
5,790
PHILADELPHIA










2,160
32,300
36,300
116,000
13.0
100
ST LOUIS











7,220
22,500
47,500
10.7
45
                                7-44

-------
TABLE 7-37.   ESTIMATES OF OCCURRENCES  FOR ADULTS WITH  CARDIOVASCULAR
   DISEASE OF COHb LEVELS EXCEEDING  SELECTED  VALUES  UNDER  9  PPM/1
              EXEX STANDARD WITH INDOOR  SOURCES OMITTED
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
ENCOUNTERS AT MAX.
CHICAGO











45,200
3,720,000
1,070,000,000
1.89
66
LOS ANGELES











60,100
3,540,000
2,670,000,000
1.82
2,030
PHILADELPHIA









367
1,870
140,000
1,340,000
1,020,000,000
2.22
53
ST LOUIS










21
22,100
775,000
416,000,000
2.00
21
                                7-45

-------
TABLE 7-38.  ESTIMATES OF ADULTS WITH  CARDIOVASCULAR  DISEASE WHO
     EXPERIENCE COHb LEVELS  EXCEEDING  SELECTED VALUES UNDER
        9 PPM/1 EXEX STANDARD WITH  INDOOR  SOURCES OMITTED
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
CHICAGO











9,170
75,800
122.000
1.89
66
LOS ANGELES











5,790
175,000
305,000
1.82
2,030
PHILADELPHIA









156
864
33,000
47,100
116*000
2.22
53
ST LOUIS










21
9,110
26,000
47,500
2.00
21
                              7-46

-------
TABLE 7-39.   ESTIMATES OF ADULTS WITH  CARDIOVASCULAR  DISEASE WHOSE MAXIMUM
      COHb LEVEL OCCURS IN SELECTED  RANGES  UNDER  9  PPM/1 EXEX STANDARD
                        WITH INDOOR  SOURCES OMITTED
COHB LEVEL
RANGE
( PERCENT )
3.70 < C <= 10.00
3.50 < C <= 3.70
3.30 < C <= 3.50
3.10 < C <= 3.30
3.00 < C <= 3.10
2.90 < C <= 3.00
2.70 < C <= 2.90
2.50 < C <= 2.70
2.30 < C <= 2.50
2.10 < C <= 2.30
2.00 < C <= 2.10
1.50 < C <= 2.00
1.00 < C <= 1.50
0.00 < C <~ 1.00
MAX. COHB CONC.
1 PEOPLE AT MAXIMUM
CHICAGO











9,170
66,600
45,800
1.89
66
LOS ANGELES






-




5,790
169,000
130,000
1.82
2,030
PHILADELPHIA









157
709
32,200
14,100
69,300
2.22
53
ST LOUIS










21
9,090
16,900
21,500
2.00
21
                                    7-47

-------
relevant data bases often are incomplete and/or inaccurate,
professional judgment plays a significant role in selecting
monitors to represent neighborhood types, in validating air
quality data, in estimating cohort populations, and in determin-
ing cohort movements.
     Ideally, the uncertainty in each significant factor affect-
ing exposure would be addressed formally within the exposure
model so that a formal representation of the uncertainty in each
exposure estimate would be part of the output of the model.
Formal techniques for characterizing the uncertainty in estimates
generated by applying the NEM model are under development.  Due
to limitations of time and resources, these techniques were not
available for this application.  Instead, several sources of
uncertainty were investigated via a limited sensitivity analysis.
Several values were used for some of the input quantities to see
how sensitive selected exposure estimates are to this variation.
The inputs chosen are the microenvironment factors, the largest
source of uncertainty in estimating exposure, and the physiological
variables used in determining blood COHb levels from exposure
patterns.
     Lower, best, and upper estimates of microenvironmental fac-
tors are presented in Section 6.0.  These differing estimates of
microenvironmental factors were used to calculate exposure esti-
mates for adults with cardiovascular disease in Chicago, assuming
a 9 ppm/1 ExEx standard is just met.  Exposure estimates for
1-hour average and 8-hour average CO concentrations are presented
in Tables 7-40 and 7-41.  The results indicate that the difference
between the lower estimates and the upper estimates is appreciable.
This large variation primarily results from the large differences
between lower and upper estimates of multiplicative microenviron-
ment factors, particularly those for transportation vehicles and
roadsides.
     Tables 7-42 through 7-44 provide COHb level estimates corre-
sponding to the same lower, best, and upper estimates of micro-
environment factors.  The resulting variation in COHb levels is

                              7-48

-------
 TABLE 7-40.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE  IN  CHICAGO
WITH 1-HOUR CARBON MONOXIDE EXPOSURES ABOVE SELECTED VALUES  UNDER 12  PPM/1
   EXEX STANDARD USING BEST, LOWER,  AND UPPER MICROENVIRONMENT FACTORS
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
BEST ESTIMATE






11,200
11,400
22,300
59,200
108,000
119,000
122,000
122,000
34.6
523












*



LOWER ESTIMATE








11,200
22,000
51,100
99,900
120,000
122,000
23.0
523
UPPER ESTIMATE

1,250
11,200
11,400
19,500
22,300
51,100
59,200
99,600
121,000
122,000
122,000
122,000
122,000
58.0
523
                                   7-49

-------
 TABLE 7-41.   ESTIMATES OF ADULTS  WITH  CARDIOVASCULAR DISEASE  IN CHICAGO
WITH 8-HOUR CARBON MONOXIDE EXPOSURES ABOVE  SELECTED VALUES UNDER 12 PPM/1
   EXEX STANDARD USING BEST, LOWER,  AND UPPER MICROENVIRONMENT FACTORS
CONCENTRATION
EXCEEDED
(PPM)
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM

BEST ESTIMATE









1,900
5,260
67,300
93,300
122,000
16.0
54


















LOWER ESTIMATE











4,670
61,600
122,000
10.6
335

UPPER ESTIMATE







776
5,260
31,000
86,400
115,000
121,000
122,000
29.0
1 54
I

                                   7-50

-------
TABLE 7-42.   ESTIMATES OF OCCURRENCES  FOR  ADULTS WITH  CARDIOVASCULAR DISEASE
      IN CHICAGO OF COHb LEVELS  EXCEEDING  SELECTED VALUES UNDER  12 PPM/1
     EXEX STANDARD USING BEST, LOWER,  AND  UPPER MICROENVIRONMENT FACTORS

CQHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
ENCOUNTERS AT MAX.


BEST ESTIMATE







13*
2,460
16,600
39,700
1,240,000
32,700,000
1,070,000,000
2.52
24




















LOWER ESTIMATE











14,700
2,890,000
1,070,000,000
1.78
67


UPPER ESTIMATE
3,980
9,390
24,100
53,000
83,300
120,000
257,000
562,000
1,210,000
2,480,000
3,530,000
31,500,000
409,000,000
1,070,000,000
4.55
24

                                     7-51

-------
TABLE 7-43.  ESTIMATES OF ADULTS WITH CARDIOVASCULAR  DISEASE  IN  CHICAGO
 EXPERIENCING COHb LEVELS EXCEEDING  SELECTED VALUES UNDER  12  PPM/1  EXEX
    STANDARD USING BEST, LOWER,  AND  UPPER MICROENVIRONMENT FACTORS
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
BEST ESTIMATE







78
1,250
3,770
d, 910
60,000
120,000
122,000
2.52
24
















LOWER ESTIMATE











3,760
78,300
122,000
1.78
67
UPPER ESTIMATE
1,310
2,670
3,550
5,090
7,380
8,550
12,100
24,300
55,600
75,100
85,800
117,000
122,000
122,000
4.55
24
                                  7-52

-------
TABLE 7-44.   ESTIMATES OF ADULTS WITH  CARDIOVASCULAR  DISEASE  IN  CHICAGO
 WHOSE MAXIMUM COHb LEVEL OCCURS IN SELECTED  RANGES UNDER  12  PPM/1  EXEX
    STANDARD USING BEST, LOWER,  AND UPPER  MICROENVIRONMENT FACTORS
COHB LEVEL
RANGE
(PERCENT)
3.70 < C <= 10.00
3.50 < C <= 3.70
3.30 < C <= 3.50
3.10 < C <= 3.30
3.00 < C <= 3.10
2.90 < C <= 3.00
2.70 < C <= 2.90
2.50 < C <= 2.70
2.30 < C <= 2.50
2.10 < C <= 2.30
2.00 < C <= 2.10
1.50 < C <= 2.00
1.00 < C <= 1.50
0.00 < C <= 1.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
«*
BEST ESTIMATE







79
1.180
2,510
5,150
51,100
59,600
1,930
2.52
24















'
LOWER ESTIMATE











3,760
74,600
43,300
1.78
67
UPPER ESTIMATE
1,310
1,360
881
1,540
2,290
1,170
3,550
12,200
31,300
19,600
10,600
30,900
4,940

4.55
24
                                 7-53

-------
consistent with the  large variations in 1-hour and  8-hour CO
exposures discussed  above.
     The results of  a  limited sensitivity analysis  on  two of the
physiological variables  which determine COHb levels in the blood
resulting from given patterns of CO exposure are presented in
Table 7-45.  Three different values are used for the Haldane
constant and two different values are used for the  ventilation
rate at low exercise level.   Five different combinations  of
values for these two variables are used for a 12 ppm/1 ExEx
standard.  One combination,  the highest value for each variable,
is used for a 9 ppm/1  ExEx standard.  It is clear that these
variations have a significant effect, but not as large an effect
as the variation in  estimated microenvironment factors.
         TABLE 7-45.   SENSITIVITY OF COHb ESTIMATES FOR CHICAGO
              TO VARIATIONS IN  TWO PHYSIOLOGICAL VARIABLES
               Case
Run
1
2
3
4
5
6
Stand.
(12,1)
(12,1)
(12,1)
(12,1)
(12,1)
(9,1)
Haldane
constant
246
246
218
230
218
246
Venti-
lation
rate,
ml/min
10,000
8,000
10,000
8,000
8,000
10,000
2.0%
COHb
11.5
8.8
8.1
8.1
8.1
1.3
2.5%
COHb
2.9
1.3
0.6
0.8
0.1
-
2.7%
COHb
0.8
0.1
-
-
-
-
  Estimates of the percentage of
adults with cardiovascular disease
  who would experience COHb levels
   exceeding the selected values
     No  sensitivity analysis runs were made in which  microenviron-
ment factors  and physiological values were varied  together.
Obviously,  doing so would result in even more widely  divergent
estimates.  Also, there are other uncertainties which have not
been subjected to analysis in this application.
                                7-54

-------
7.5  REFERENCES

 1.  Memorandum from Harvey Richmond,  Ambient Standards Branch,
     to Mike Jones,  Chief of the Ambient Standards Branch, U.S.
     Environmental Protection Agency,  Research Triangle Park,
     North Carolina 27711.  September  8, 1982.

 2.  F. L. Rodkey, J. D.  O'Neal, and H.  A.  Collison, "Oxygen and
     carbon monoxide equilibria of human adult hemoglobin at
     atmospheric and elevated pressure," Blood, Vol. 33, No. I,
     1960, pp.  57-65.

 3.  U.S.  Department of Health, Education,  and Welfare, Public
     Health Service, Hemoglobin and Selected Iron-Related Find-
     ings  of Persons 1-74 Years of Age:   United States, 1971-1974.
     Advance data number 46, January 26, 1979.

 4.  R. F. Coburn, et al, "Endogenous  carbon monoxide production
     in man," J. of Clin. Invest., Vol.  42, 1963,  pp. 1172-1178.

 5.  S. R. Lynch and A. L. Moede, "Variation in the rate of
     endogenous carbon monoxide production in normal human
     beings," J. Lab. Clin. Med., Vol. 79,  1972, pp. 85-95.

 6.  P. D. Berk, et al, "Comparison of plasma bilirubia turnover
     and carbon monoxide production in man," J. Lab. Clin. Med.,
     Vol.  83, 1974,  pp. 29-37.

 7.  M. Delivoria-Papadopoulos, R. F.  Coburn, and  R. E. Forster,
     "Cyclic variation of rate of carbon monoxide  production in
     normal women,"  J. Appl. Physiol., Vol. 36, 1974, pp. 49-51.

 8.  R. P. Brouillard, M. E. Conrad, and T. A. Bensinger, "Effect
     of blood in the gut on measurements of endogenous carbon
     monoxide production," Blood, Vol. 45,  1975, pp. 67-69.

 9.  C. A. Coltman and G. M. Dudley, "The relationship between
     endogenous carbon monoxide production and total heme mass in
     normal and abnormal subjects," Am.  J.  Med. Sci., Vol. 258,
     1969, pp.  374-385.

10.  R. Joumard, et al, "Mathematical  models of the uptake of
     carbon monoxide on hemoglobin at  low carbon monoxide levels,"
     Env.  Health Persp.,  Vol.  41, 1981,  pp. 277-289.

11.  T. Sjostrand, "Blood volume," Handbook of Physiology, Vol. 1,
     Section 2, Chap. 4,  1962, pp. 51-62.

12.  U. S. National  Center for Health  Statistics,  Advance Data.
     No. 3, November 19,  1976, and No. 14,  November 30, 1977.
                               7-55

-------
13.   V.  Niinimaa,  P.  Cole,  S.  Mintz,  and R.  J.  Shephard,  "Oral
     nasal distribution of  respiratory airflow," Resp.  Physiol,
     Vol. 43,  1981,  pp. 69-75.

14.   Review of the National Ambient Air Quality Standards for
     Sulfur Oxides;   Assessment of Scientific and Technical Infor-
     mation,Strategies and Air Standards Division,Office of Air
     Quality Planning and Standards,  U. S.  Environmental  Protection
     Agency, Research Triangle Park,  N.C. 27711, November 1982.

15.   U.  S. Department of Health, Education,  and Welfare,  Public
     Health Service,  Coronary Heart Disease in Adults.  United
     States:  1960-1962, Vital and Health Statistics Series 11,
     No. 10, December 1975.
                              7-56

-------
                            SECTION 8
                    NATIONWIDE EXTRAPOLATIONS

     The exposure model described in the preceding sections is
applied directly to individual urbanized areas.  To obtain CO
exposure and COHb distributions for all urbanized areas directly
from the model would require that the model be applied to each
urbanized area separately and the distributions obtained be
summed according to the expression
                       n
               E(C) =  I  e. (C) ,                      (8-1)
                      i=l  1
where E(C)  is the total number of exposures to a concentration
above C for all urbanized areas and e.(C) is the exposure dis-
                   •f"H
tribution for the i— area of n urbanized areas.  Analogous
expressions can be written for the number of people with expo-
sures above selected concentrations and the number of people
whose maximum exposure occurs in selected ranges.  To carry out
these computations would require the development of pollutant
concentration and human activity data bases for each urbanized
area in the U.S.  Such an analysis is not feasible at the present
time.  Accordingly, rough estimates of national exposure for
adults with cardiovascular disease were made by extrapolating
the exposure and COHb estimates obtained from modelling the four
study areas discussed in previous sections, namely, Chicago, Los
Angeles, Philadelphia, and St. Louis.
     The extrapolation procedure used is described in Section
8.1.  Results of the extrapolation are presented in Section 8.2.
A discussion of the uncertainty about the accuracy of these
estimates is given in Section 8.3.
                               8-1

-------
8.1  EXTRAPOLATION PROCEDURE
     Equation 8-1 can be rewritten, in terms of exposures per
person in the population, as
                       n
               E(C) =  Z  P.e?(C)                       (8-2)
                      i=l  x x
where e?(C) is the exposure distribution per person in  the popu-
       1                                +-Vi
lation and P. is the population of the i— urbanized area.  As
with Equation 8-1, analogous equations can be written for each
of the exposure and COHb distributions provided by the model.
The effect of factoring out the population is to bring  the e.(C)
values for different areas just meeting a given alternative stan-
dard into closer agreement.  There will continue to be  significant
differences, however, and the basic assumption of the extrapola-
tion is that the e?(C) for the four base study areas are sufficient
to represent these differences exhaustively.  Therefore, the first
step in applying the method was to assign each urbanized area to
one of the four base areas.  The value of n in Equation 8-2 was
set equal to four, and P. became the total population of urban-
                            •f*'H
ized areas assigned to the i— base area.
     The ultimate goal of the extrapolation is to estimate what
the exposure of the sensitive population  (i.e., adults  with
cardiovascular disease) would be in 1987 under each of  three air
quality assumptions.  These assumptions are that the three air
quality standards discussed in Section 7 are just met in all
urban areas.  Since some urban areas are expected to have cleaner
air in 1987 than required by the given standards, NEM estimates
which are based on just meeting the standards are higher than
they would be if they had been based on estimated 1987  quality.
     The CO exposure and COHb distributions for each of the four
base areas are divided by their respective adult population values
to obtain the e.(C) distributions.  To obtain the urban population
estimates to associate with each of the base areas, each of the
urbanized areas with populations greater than 200,000 is assigned

                              8-2

-------
to one of the four base areas based on such considerations as
proximity to the base area, average wind speed, observed peak CO
concentration, climate, and general character of the area.  The
total population for urban areas with population greater than
200,000 associated with each base area (which included that of
the base area) was obtained by summing the associated populations
for each base area.  The population data used at this stage were
based on 1970 census data.
     The total and sensitive populations assigned to each base
area are listed in Table 8-1.  Review of these data reveals that
the 105 urbanized areas with populations greater than 200,000 in
1970 are distributed relatively evenly among the four base areas.
However, although only 22 areas were assigned to Chicago, over 35
percent of the total urban population is associated with this
base area.  This situation occurs because several of the largest
urbanized areas, including the New York urbanized area (pop.
16,200,000), are assigned to Chicago.
     An adjustment is required because the total sensitive popu-
lation of associated urbanized areas with population greater than
200,000 is less than the total population of urbanized areas.
This adjustment is made by using the ratio of the total urbanized
area population in 1970 to the total 1970 population in urban
areas with populations greater than 200,000.  Substitution in
Equation 8-2 of the adjusted population values and the appro-
priate e.(C) values for each exposure and COHb distribution yields
the desired extrapolated distributions.
     Note that although the e?(C) values are based on 1970 urban
area population data they are extrapolated to 1987 (see Section
7.1.1).  The 1970 urban data are not only used to estimate base
populations, but also are used in conjunction with 1980 total
U.S. population data and an estimated growth rate to determine
the factor used for the extrapolation.  The 1970 data were the
best urban population data available for this purpose and for
making the adjustment described in the last paragraph.
                               8-3

-------
      TABLE 8-1.  URBANIZED AREA POPULATION DATA USED TO EXTRAPOLATE
                           MODEL RESULTS
Area
Chicago
Los Angeles
Philadelphia
St. Louis
Totals
1970 1987
Associated
urbanized
areas
22
26
25
32
105
Pop. of associated
urbanized areas
with pop. >200,000
38,894,365
26,339,249
20,553,523
17,350,712
103,137,849
Sensitive pop.
of associated
urbanized areas
1,886,000
1,277,000
997,000
841,000
5,001,000
     By using an expression, which is mathematically  equivalent
to the per person approach described above, the  desired  extra-
polated distributions can be calculated directly from the  exposure
distributions which are calculated for the four  study areas.
That is, E(C) can be calculated from the e.(C) for  the four
study areas by using the expression,

            Total Population  (1970)	    f!
E(C) =
where
Total Population > 200,000  (1970)   ^ J

   f  = Total Pop, of i-type urban areas
    i     Total Pop. of ith urban area
                                           e.(C)
(8-3)
8.2  EXTRAPOLATION RESULTS
     The results of the nationwide extrapolation  are  presented in
Tables 8-2 through 8-12.  The first nine  tables can be  divided
into three sets of three tables.  Tables  8-2,  8-3, and  8-4 present
exposure estimates for a one-hour averaging  time.  Estimates of
occurrences during 1987 among adults with cardiovascular disease
of 1-hour average CO exposures above selected  concentration
values during 1987 under four alternative air  quality assumptions
are presented in Table 8-2.  Estimates  of the  number  of adults
with cardiovascular disease in the urban  U.S.  who would incur 1-
hour average CO exposures above  the same  set of selected
                                8-4

-------
TABLE 8-2.   ESTIMATES OF OCCURRENCES IN  THE  CARDIOVASCULAR  ADULT  URBAN  U.S.
              POPULATION OF 1-HOUR AVERAGE CO  EXPOSURES  ABOVE
              SELECTED CONCENTRATION VALUES  UNDER  ALTERNATIVE
                          AIR QUALITY ASSUMPTIONS
CONCENTRATION
EXCEEDED
(PPM)
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
ENCOUNTERS AT MAX.
9 PPM 8HR 1EXEX



10,200
21,900
163,000
915,000
6,620,000
27,700,000
102,000,000
370,000,000
45,900,000,000
36.0
10,200
12 PPM 8HR 1EXEX

10,200
21,900
124,000
359*000
1,700,000
6,570,000
31,300,000
88,500,000
296,000,000
873,000,000
45,900,000,000
49.0
10,200
15 PPM 8HR 1EXEX
21,900
33,600
359,000
849,000
2,140,000
6,310,000
21,300,000
78,700,000
185,000,000
602,000.000
1,640,000,000
45,900,000,000
61.5
10,200














                                   8-5

-------
TABLE 8-3.  ESTIMATES OF CARDIOVASCULAR ADULTS IN  URBAN  U.S.  WITH  1-HOUR
            AVERAGE CO EXPOSURES ABOVE SELECTED CONCENTRATION
            VALUES UNDER ALTERNATIVE AIR QUALITY ASSUMPTIONS

CONCENTRATION
EXCEEDED
(PPM)
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
IZ.O
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM

9 PPM 8HR 1EXEX



10,200
21,900
157,000
535,000
1,290,000
2,820,000
4,390,000
5,140,000
5,240,000
36.0
10,200

12 PPM SHR 1EXEX

10,200
21,900
113,000
348,000
691,000
1,290,000
2,900,000
4,290,000
5,030,000
5,240,000
5,240,000
49.0
10,200

15 PPM SHR 1EXEX
21,900
21,900
343,000
531,000
372,000
1,270,000
2,510,000
4,290,000
4,800,000
5,220,000
5,240,000
5,240,000
61.5
10,200















                                   8-6

-------
TABLE 8-4.   ESTIMATES OF CARDIOVASCULAR ADULTS IN URBAN  U.S.  WHOSE  MAXIMUM
        1-HOUR AVERAGE CO EXPOSURE OCCURS  IN  SELECTED  CONCENTRATION
             RANGES UNDER ALTERNATIVE AIR  QUALITY ASSUMPTIONS
CONCENTRATION
RANGE
(PPM)
50.0 < C <= 55.0
45.0 < C < = 50.0
40.0 < C <= 45.0
35.0 < C <= 40.0
30.0 < C <= 35.0
25.0 < C <= 30.0
20.0 < C <= 25.0
15.0 < C <= 20.0
12.0 < C <= 15.0
9.0 < C <= 12.0
7.0 < C <= 9.0
0.0 < C <= 7.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
9 PPM 8HR 1EXEX



10,200
11,700
135,000
378,000
753,000
1,530,000
1,570,000
747,000
107,000
36.0
10,200
12 PPM 8HR 1EXEX

10,200
11,700
90,600
235,000
344,000
603,000
1,610,000
1,390,000
738,000
211,000
4,730
49.0
10,200
15 PPM SHR 1EXEX


326,000
134,000
341,000
401,000
1,240,000
1,770,000
510,000
426,000
18,600

61.5
10,200














                                   8-7

-------
TABLE 8-5.  ESTIMATES OF OCCURRENCES IN THE CARDIOVASCULAR ADULT  URBAN  U.S.
              POPULATION OF 8-HOUR AVERAGE CO EXPOSURES  ABOVE
              SELECTED CONCENTRATION VALUES UNDER ALTERNATIVE
                          AIR QUALITY ASSUMPTIONS
CONCENTRATION
EXCEEDED
(PPM)
50,0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
ENCOUNTERS AT MAX.
9 PPM 8HR 1EXEX








200,000
6,340,000
63,500,000
45,900,000,000
14.0
232
12 PPM 8HR 1EXEX







429,000
4,290,000
54,900,000
322,000,000
45,900,000,000
18.5
232
15 PPM 8HR 1EXEX






104,000
3,330,000
29,200,000
206,000,000
835,000,000
45,900,000,000
23.0
232














                                    8-8

-------
TABLE 8-6.   ESTIMATES OF CARDIOVASCULAR ADULTS  IN  URBAN  U.S.  WITH  8-HOUR
            AVERAGE CO EXPOSURES ABOVE SELECTED CONCENTRATION
            VALUES UNDER ALTERNATIVE AIR QUALITY ASSUMPTIONS
CONCENTRATION
EXCEEDED
(PPM)
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
12.0
9.0
7.0
0.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
9 PPM 8HR 1EXEX








98,000
618.000
2,650,000
5,240,000
14.0
232
12 PPM 8HR 1EXEX







176,000
_
538,000
2,410,000
4,090,000
5,240,000
13.5
232
IS PPM 8HR 1EXEX






49,400
529,000
1,770,000
3,660,000
4,950,000
5,240,000
23.0
232














                                   8-9

-------
TABLE 8-7.   ESTIMATES OF CARDIOVASCULAR ADULTS  IN  URBAN U.S. WHOSE MAXIMUM
        8-HOUR AVERAGE CO EXPOSURE OCCURS  IN  SELECTED  CONCENTRATION
             RANGES UNDER ALTERNATIVE  AIR  QUALITY  ASSUMPTIONS
CONCENTRATION
RANGE
(PPM)
50.0 < C <= 55.0
45.0 < C <= 50.0
40.0 < C <= 45.0
35.0 < C <= 40.0
30.0 < C <= 35.0
25.0 < C <= 30.0
20.0 < C <= 25.0
15.0 < C <= 20.0
12.0 < C <= 15.0
9.0 < C <= 12.0
7.0 < C <= 9.0
0,0 < C <= 7.0
MAX. CONCENTRATION
PEOPLE AT MAXIMUM
9 PPM 8HR 1EXEX








98,100
520,000
2,030,000
2,590,000
14.0
232
12 PPM 8HR 1EXEX







176,000
362,000
1,870,000
1,680,000
1,160,000
18.5
232
15 PPM 8HR 1EXEX






49,400
480,000
1,240,000
1,890,000
1,290,000
291,000
23.0
232














                                    8-10

-------
TABLE 8-8.   ESTIMATES OF OCCURRENCES  AMONG  CARDIOVASCULAR ADULTS  IN  URBAN U.S.
         OF COHb LEVELS EXCEEDING SELECTED  VALUES  UNDER  ALTERNATIVE
                           AIR QUALITY  ASSUMPTIONS
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
ENCOUNTERS AT MAX.
9 PPM 8HR 1EXEX








570
12.000
41,700
4,730,000
296,000,000
45,900,000,000
2.30
120
12 PPM 8HR 1EXEX




570
2,650
14,300
76,700
473,000
1,580,000
2,730,000
33,300,000
1,070,000,000
45,900,000,000
3.02
120
15 PPM 8HR 1EXEX
637
6,690
15,300
51,800
106,000
250,000
810,000
1,880,000
4,100,000
8,580,000
13,900,000
151,000,000
2,270,000,000
45,900,000,000
3.75
120


,













                                    8-11

-------
TABLE 8-9.  ESTIMATES OF CARDIOVASCULAR ADULTS  IN  URBAN  U.S.  EXPERIENCING
         COHb LEVELS EXCEEDING SELECTED VALUES  UNDER ALTERNATIVE
                         AIR QUALITY ASSUMPTIONS
COHB LEVEL
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
9 PPM 8HR 1EXEX








281
5,500
22,200
742,000
4,140,000
5,240,000
2.30
120
12 PPM 8HR 1EXEX




281
1,250
6,230
42,000
217,000
452,000
660,000
2,100,000
5,090,000
5,240,000
3.02
120
15 PPM 8HR 1EXEX
345
3,200
6,600
25,800
58,200
132,000
308,000
506,000
732,000
1,060.000
1,350,000
3,480,000
5,180,000
5,240,000
3.75
120
















                                   8-12

-------
TABLE 8-10.
ESTIMATES OF CARDIOVASCULAR ADULTS IN URBAN U.S.  WHOSE  MAXIMUM
COHb LEVEL OCCURS IN SELECTED CONCENTRATION RANGES
     UNDER ALTERNATIVE AIR QUALITY ASSUMPTIONS
COHB LEVEL
RANGE
( PERCENT )
3.70 < C <= 10.00
3.50 < C <= 3.70
3.30 < C <= 3.50
3.10 < C <= 3.30
3.00 < C <= 3.10
2.90 < C <= 3.00
2.70 < C <= 2.90
2.50 < C <= 2.70
2.30 < C <= 2.50
2.10 < C <= 2.30
2.00 < C <= 2.10
1.50 < C <= 2.00
1.00 < C <= 1.50
0.00 < C <= 1.00
MAX. COHB CONC.
PEOPLE AT MAXIMUM
9 PPM 8HR 1EXEX








285
5,230
16,700
720,000
3,400,000
1,100,000
2.30
120
12 PPM 8HR 1EXEX




285
972
4,990
35,900
175,000
235,000
208,000
1,440,000
2,990,000
149,000
3.02
120
15 PPM 8HR 1EXEX
351
2,860
3,400
19,200
32,400
74,100
175,000
198,000
227,000
327,000
291,000
2,130,000
1,700,000
63,000
3.75
120
















                                   8-13

-------
TABLE 8-11.  PERCENTAGE OF CARDIOVASCULAR  ADULT  URBAN  U.S. POPULATION
         EXPERIENCING COHb LEVELS  EXCEEDING  SELECTED VALUES
              UNDER ALTERNATIVE AIR QUALITY  ASSUMPTIONS
COHb level
exceeded
(percent)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00
9 ppm 8 hr
1 ExEx








0.01
0.10
0.42
14.16
79.01
100.00
12 ppm 8 hr
1 ExEx




0.01
0.02
0.12
0.80
4.14
8.63
12.60
40.08
97.14
100.00
15 ppm 8 hr
1 ExEx
0.01
0.06
0.13
0.49
1.11
2.52
5.88
9.66
13.97
20.23
25.76
66.41
98.85
100.00
                                 8-14

-------
TABLE 8-12.  ESTIMATES OF CARDIOVASCULAR ADULTS IN URBAN U.S.
 EXPERIENCING COHb LEVELS EXCEEDING SELECTED VALUES A GIVEN
  NUMBER OF DAYS ASSUMING 9 PPM/1 EXEX STANDARD IS ATTAINED
f*nMCFkTT"D ATTrtM
UUrlwtLri 1 KA I Awrt
EXCEEDED
(PERCENT)
3.70
3.50
3.30
3.10
3.00
2.90
2.70
2.50
2.30
2.10
2.00
1.50
1.00
0.00


1 DAY








281
5.500
21,800
372,000
673,000


NUMBER OF
2-4 DAYS










1,120
656,000
2,570,000


TIMES
5-25 DAYS




'






909.000
25,300,000



> 25 DAYS












39,200,000
1,910,000,000

                             8-15

-------
concentrations under the same assumptions are presented in Table
8-3.  Estimates of the number of urban U.S. adults whose maximum
1-hour average CO exposure would occur in various concentration
ranges are presented in Table 8-4.
     Analogous estimates for 8-hour average CO exposures are pre-
sented in Tables 8-5, 8-6, and 8-7 respectively.  Similar esti-
mates for COHb levels resulting from CO exposure are presented
in Tables 8-8, 8-9, and 8-10.  The absolute numbers presented in
Table 8-9 are presented in percentage form in Table 8-11.
     Estimates of the number of adults with cardiovascular disease
who would have their blood COHb levels elevated above selected
concentrations for various numbers of days if an 8-hour average
9 ppm/1 ExEx standard were just met in all urban areas are
presented in Table 8-12.  The table indicates the frequency of
repeated peak COHb levels.  The table indicates, for example,
that of the 5,500 adults with cardiovascular disease who are
estimated to have their blood COHb level exceed 2.1 percent under
the 9 ppm/1 ExEx standard, none would have it occur more than one
day.

8.3  UNCERTAINTY OF THE NATIONWIDE ESTIMATES
     The uncertainty of the CO exposure and COHb estimates made
for the four base cities was discussed in Section 7.4.  The nation-
wide estimates are even more uncertain because of the additional
uncertainty introduced by the extrapolation of exposure estimates
for these four cities to all urban areas in the U.S.
     Formal means of dealing with the uncertainty of nationwide
estimates are under development, but were not available for this
analysis.  Hence, no attempt was made to formally represent the
uncertainty of the estimates presented in Section 8.2.  The analy-
ses discussed in Section 7.4 indicate that uncertainty is already
great at the city level.  That even greater uncertainty exists
in the nationwide estimates should be recognized when considering
the estimates presented in Tables 8-2 through 8-12.
                              8-16

-------
                           APPENDIX A

     Section 3.1 describes the development of activity patterns
for the 56 population subgroups used in the NEM analysis.  Ref-
erence 2 of Section 3.4 contains these 56 activity patterns.  This
appendix contains three examples of these activity patterns.  At
the top of each table is a label indicating the age-occupation
group, the subgroup, and the percentage of the age-occupation
group falling into the subgroup.  In the body of the table are
hourly assignments to locations, microenvironments, and activity
levels for weekdays, Saturdays, and Sundays.  Note that the hour
designated "1 AM" is the hour which ends at 1 AM.
                              A-l

-------
           ACTIVITY PATTERNS BY ASE-OCCUPATIOh SUBGROUP
A-0 GROUPS 4—Clerical -orkers      SUBGROUPlZ    PCT IN SU86ROUP=26
SAY OF TIME LOCATION7MICR
WEEK OF DAY 1
WEEKDAYS AM H
2
1
P« W
2
1
SATURDAY An H
2
t
^}-tt u
• ™ n
2
1
SUNDAY AM H
2
1
PW H
2
1
LOCATION CODES; H=ho«e
niCROENVIRONMENT CODES:
1 = uork or school 2
4 = roadside 5
2
H
2
1
w
1
1
H
2
1
H
2
1
H
2
1
H
2
1


3
H
2
1
V
1
t
H
2
1
H
2
1
H
2
1
H
2
1
W=work

QENVIflONKENT/ACTIVITY-LEVEL BY
4
H
2
1
w
1
1
H
2
1
H
2
2
H
2
1
H
2
1


= bo «e or
= outdoo
rs
5
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
5
3


other

6
H
2 .
1
U
3
1
H
2
1
H
2
1
H
2
1
H
4
2




7
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1


3 =
6 -
8
M
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1


9 1
y
1
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1


0
u
1
1
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1


transport
ki
tchen

11
y
1
1
H
2
1
H
5
2
H
2
1
H
3
1
H
2
1


vehic

HOUR
12
u
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1


le

 ACTIVITY  LEVELSs  1=low   2=«ediu«  3=high
                                    A-2

-------
           ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
A-0 GROUP: 6—Operatives ^Laborers SUb6«OUP:6
PCT IN SUBGROUP:!6
DAY OF TIME
y£EK Of DAY
WEEKDAYS AM
*

PN


SATURDAY A*


PH


SUNDAY AH


f»H


LOCATION/MICR
1
H
2
1
y
3
1
H
2
1
H
2
1
H
2
1
n
2
1
2
H
2
1
y
3
1
H
2
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=hoae
HICROENVIRONMENT
1 = work or schoo
4 = roadside
COOES
I

-
2 =
5 =
3
H
2
1
y
2
1
H
2
1
H
5
2
H
2
1
H
2
1
w=*ork

OENVIRONMENT/ACTIVITY-LEVEL BY HOUR
4
H
2
1
tt
3
1
H
2
1
H
2
1
H
2
1
H
2
1
5
H
2
1
y
3
1
H
2
1
H
2
1
H
2
1
H
5
2
6
H
2
1
JJ
2
1
H
2
1
„
2
1
H
2
1
H
2
1
7
y
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
3
y
3
1
H
2
1
H
2
1
H
4
2
H
2
1
H
2
1
9 1
y
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
0 11
y y
3 4
1 2
H H
2 2
1 1
H H
2 2
1 2
H H
2 2
1 1
H H
2 3
1 1
H H
2 2
2 1
12
y
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1


ho»e or
outdoo
rs

other


3
6

s

tr

an sport

vehi

cle
= kitchen
 ACTIVITY LEVELSr 1=low   2=«ediu«  3=high
                                   A-3

-------
           ACTIVITY PATTERNS BY A6E-OCCuPATION  SUBGROUP
A-0 GROUP: 9—Housewives           SUBGROUPS'!     PCT IN SUBGROUP:42
DAT OF TIME LOCATION/HI
VEEK OF DAT 1
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AR H
2
1
PW R
6
1
SUNDAY AM H
2
1
PH R
6
1
2
H
2
1
H
2
t
H
2
1
H
2
1
H
2
1
H
2
1
3
H
2
1
H
3
1
H
2
1
H
5
2
H
2
1
H
2
1
CROENVIRGNMEHT/ACTIVITY-LEVEL BY
4
H
2
1
H
2
1
H
2
1
H
2
1
H
Z
1
H
2
2
5
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
2
6
H
2
1
H
6
1
H
2
1
N
2
1
H
2
1
H
2
1
7
• H
6
1
H
2
1
H
2
1
H
2
1
H
2
1
H
6
1
a
H
2
2
H
2
1
H
6
1
H
4
2
H
6
1
H
2
1
9
H
2
1
H
2
1
H
2
2
H
3
1
H
2
1
H
2
1
10
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
11
H
2
2
H
2
1
h
2
1
H
2
1
H
2
1
H
2
1
HOUR
12
H
5
1
K
2
1
H
2
1
H
2
1
H
3
1
H
2
1

 LOCATION COOES:  H=ho«e    U=work

 MICRQENVIRONMENT  COOES:
 1 = work or school    2  = ho»e  or other
 4 = roadside          5  - outdoors

 ACTIVITY LEVELS:  1-tou    2=nediu«  3-high
3 - transport vehicle
6 - kitchen
                                    A-4

-------
                                 APPENDIX  B



                       COHORT POPULATIONS  BY  STUDY AREA
Cohort description

A-0 group
Students 18+
01







Professionals
02














Sales workers
03








Home
NTa
CR
1



SR
5


CR
1


CR
1


SR
5


SR
5


CR
1



CR
1



Work
NTa
CR
1



SR
5


CC
2


SC
6


CC
2


SC
6


CC
2



SC
6



Sub-
group
1
- 2
3
Cohort population

Chicago
19,316
37,792
9,238
4 17,636
i
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
5
1
2
3
4
5
5,897
11,538
2,820
5,384
170,393
76,133
79,758
36,254
11,845
5,293
5,545
2,520
47,282
21,126
22,132
10,060
8,806
3,934
4,122
1,874
33,250
16,238
3,866
6,959
17,011
2,312
1,129
269
484
1,183
Phila-
delphia
4,239
8,293
2,027
3,870

17,387
34,017
8,315
15,875
14,501
6,479
6,788
3,085
5,710
2,551
2,673
1,215
35,661
15,933
16,692
7,587
65,332
29,191
30,581
13,901
4,442
2,169
517
930
2,273
1,749
854
203
366
895

St. Louis
3,753
7,344
1,794
3,427

5,334
10,436
2,551
4,870
13,364
5,971
6,256
2,834
3,489
1,559
1,633
742
14,490
6,474
6,782
3,083
18,812
8,405
8,805
4,002
4,000
1,953
465
837
2,046
1,044
510
121
219
534
Los
Angeles
17,261
33,771
8,255
15,760

61,326
119,985
29,330
55,993
41,759
18,658
19,547
8,885
31,503
14,076
14,746
6,702
40,366
18,036
18,895
8,589
263,140
117,573
123,172
55,987
11,364
5,550
1,321
2,378
5,814
8,573
4,187
997
1,794
4,386
[ continued)
                                     B-l

-------
Cohort description

A-0 group
Sales workers
03 (cont.)








Clerical
workers 04






















Home
NTa
SR
5



SR
5



CR
1




CR
1




SR
5



1
SR
5




Work
NTa
CC
2



SC
6



CC
2




SC
6




CC
2




SC
6




Sub-
group
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
Cohort population

Chicago
9,244
4,514
1,073
1,932
4,722
1,719
839
200
360
879
141,163
65,540
22,687
10,083
2,520
10,083
9,813
4,556
1,577
701
175
701
31,571
14,658
5,074
2,255
564
2,255
5,880
2,730
945
420
105
420
Phila-
delphia
11,322
5,529
1,317
2,370
5,792
20,742
10,130
2,412
4,341
10,612
26,033
12,087
4,184
1,860
465
1,860
10,250
4,759
1,647
732
183
732
39,508
18,343
6,349
2,822
706
2,822
72,380
33,605
11,633
5,170
1,293
5,170

St. Louis
4,315
2,107
502
903
2,208
5,602
2,736
651
1,173
2,866
22,242
10,327
3,575
1,589
397
1,589
5,806
2,696
933
415
104
415
11,765
5,462
1,891
840
210
840
15,274
7,092
2,455
1,091
273
1,091
Los
Angeles
11,372
5,554
1,322
2,380
5,818
74,133
36,204
8,620
15,516
37,928
48,231
22,393
7,751
3,445
861
3,445
36,385
16,893
5,848
2,599
650
2,599
36,537
16,963
5,872
2,609
652
2,609
238,175
110,175
38,278
17,013
4,253
17,013
(continued)
                                     B-2

-------
Cohort description

A-0 group
Craftsmen 05























Laborers 06












Home
NTa
CR
1




CR
1




SR
5




SR
5




CR
1





CR
1




Work
NTa
CI
3




SI
7




CI
3




SI
7




CI
3





SI
7




Sub-
group
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6

1
2
3
4
5
6
Cohort population

Chicago
81,648
39,191
16,330
3,266
6,532
16,329
5,676
2,725
1,135
227
454
1,135
25,626
12,300
5,125
1,025
2,050
5,125
4,772
2,291
954
191
382
954
75,251
34,731
11,577
5,789
34,731
30,872

5,231
2,414
805
402
2,414
2,146
Phila-
delphia
14,010
6,725
2,802
560
1,121
2,802
5,516
2,648
1,103
221
441
1,103
20,533
9,856
4,107
821
1,643
4,107
37,618
18,056
7,524
1,505
3,009
7,524
27,638
12,756
4,252
2,126
12,756
11,339

10,882
5,022
1,674
837
5,022
4,464

St. Louis
9,562
4,590
1,912
382
765
1,912
2,496
1,198
499
100
200
499
6,018
2,889
1,204
241
481
1,204
7,813
3,750
1,563
313
625
1,563
19,542
9,020
3,007
150
9,020
8,017

19,542
2,354
785
392
2,354
2,093
Los
Angeles
18,978
9,109
3,796
759
1,518
3,796
14,317
6,872
2,863
573
1,145
2,863
22,170
10,641
4,434
887
1,774
4,434
144,520
69,370
28,904
5,781
11,562
28,904
29,939
13,818
4,606
2,303
13,818
12,283

22,586
10,424
3,475
1,737
10,424
9,266
(continued)
                                     B-3

-------
Cohort description

A-0 group
Laborers 06
(cont.)










Service
workers 08










Housewives
09




Retired 10





Home
NTa
SR
5




SR
5




CR
1




SR
5




CR
1

SR
5

CR
1




Work
NT*
CI
3




SI
7




CR
1




SR
5




CR
1

SR
5

CR
1




Sub-
group
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
1
2
3
1
2
3
4
5
6
Cohort population

Chicago
18,289
8,441
2,813
1,406
8,441
7,503
3,406
1,572
524
262
1,572
1,397
49,393
23,324
30,184
4,116
19,208
10,976
11,090
5,236
6,777
924
4,312
2,464
71,488
83,402
15,319
38,824
45,295
8,320
11,480
13,776
11,480
17,219
2,296
1,148
Phila-
delphia
22,294
10,289
3,430
1,715
10,289
9,146
40,843
18,851
6,284
3,142
18,851
16,756
20,770
9,808
12,693
1,731
8,077
4,616
37,142
17,539
22,698
3,095
14,444
8,254
39,106
45,624
8,380
122,142
142,499
26,173
19,396
23,275
19,396
29,093
3,879
1,940

St. Louis
6,791
3,134
1.045
522
3,134
2,786
8,817
4,069
1,356
678
4,069
3,617
16,789
7,928
10,260
1,399
6,529
3,731
9,416
4,446
5,754
785
3,662
2,092
27,469
32,047
5,886
34,182
39,879
7,325
17,534
21,040
17,534
26,300
3,507
1,753
Los
Angeles
24,843
11,466
3,822
1,911
11,466
10,192
161,945
74,744
24,914
12,457
74,744
66,439
25,910
12,235
15,834
2,159
10,076
5,758
115,759
54,664
70,741
9,647
45,017
25,724
69,865
81,509
14,971
340,320
397,040
72,926
40,520
48,624
40,520
60,781
8,104
4,052
(continued)
                                     B-4

-------
Cohort description

A-0 group
Retired 10
(cont.)




Children <5
11






Children 5-17
12










Home
NT*
SR
5




CR
1


SR
5


CR
1




SR
5




Work
NTa
SR
5




CR
1


SR
5


CR
1




SR
5




Sub-
group
1
2
3
4
5
6
1
2
3
4
1
2
3
4
1
2
3
4
5
6
1
2
3
4
5
6
Cohort population

Chicago
3,277
3,933
3,277
4,916
655
327
10,562
10,059
10,059
19,615
8,076
7,691
7,691
14,997
67,594
4,828
8,449
31,383
2,414
6,035
51,683
3,692
6,460
23,996
1,846
4,615
Phila-
delphia
42,411
50,893
42,411
63,617
8,482
4,241
15,851
15,096
15,096
29,438
35,021
33,353
33,353
65,039
117,289
8,378
14,661
54,455
4,189
10,472
284,616
20,330
35,577
132,143
10,165
25,412

St. Louis
10,836
13,003
10,836
16,254
2,167
1,084
10,494
9,995
9,995
19,489
10,506
10,006
10,006
19,511
82,690
5,906
10,336
38,391
2,953
7,383
91,294
6,521
11,412
42,387
3,261
8,151
Los
Angeles
112,631
135,157
112,631
168,946
22,526
11,263
24,304
23,146
23,146
45,135
113,151
107,763
107,763
210,137
149,881
10,706
18,735
69,587
5,353
13,382
905,879
64,706
113,235
420,587
32,353
80,882
B-5

-------
                            APPENDIX C
               DISCUSSION OF AIR QUALITY INDICATORS
              USED IN THE NEM ANALYSIS AND ESTIMATED
          CONCENTRATIONS USED IN THE REGULATORY ANALYSIS

     A number of reviewers of early drafts of this report have
asked how air quality indicators (AQI's) used in the NEM analyses
of CO compare with estimated concentrations  (EC's) used in the
regulatory impact analysis of alternative CO NAAQS's.  This appen-
dix discusses how EC's are determined and why they differ from
AQI's.
     For regulatory impact analysis purposes, EPA characterizes
air quality levels in urbanized areas by a single value.  This
value, the EC, is determined from existing air quality data
according to the same criteria that states would use to determine
whether or not an area attains a proposed NAAQS.  These criteria
vary according to the "form" of the standard being analyzed and
the allowed violation rate.  In the case of CO, forms under con-
sideration include one-hour and eight-hour daily maximum standards
with allowed violation rates of one and five expected exceedances
per year over a three year period.
     The EC for a given urbanized area is usually based on air
quality data from the monitor which reported the highest air
quality values over a two or three year period.  According to
current EPA guidance, the EC may be determined by applying the
simple formula
                              / number \ / allowed  \
descending rank of EC value = I of-years I (exceedance)  + 1   (B-l)
                              \analyzed7 y   rate   /

to a multi-year data set from this monitor.  Thus if the permitted
exceedance rate is five and three years of data are considered,
                               C-l

-------
the EC value would be the 16th highest concentration in the data
set.  For two years of data and one allowed exceedance, the third
highest concentration would be used.  Note that each EC corresponds
to an actual observed value.
     AQI values used in the NEM analysis are determined by fitting
distributions to single-year data sets which have had missing
values filled in by time series analysis (see Section 5).  Values
with expected exceedance rates of one and five are represented by
the characteristic largest and fifth largest values, respectively.
These values correspond to quantiles in the fitted distributions
rather than particular observed values.
     Table C-l lists the EC's for the four study areas which have
been developed for alternative CO NAAQS's which consider  (1) the
daily maximum one-hour concentration with one expected exceedance,
(2) the daily maximum eight-hour running average concentration
with one expected exceedance, and (3) the daily maximum eight-
hour running average concentration with five expected exceedances.
Also, listed is the value of the largest corresponding AQI from
Table 5-6.  In over half the cases, EC and AQI values differ by
more than 10 percent.
     There are a number of reasons for such large differences.
EC's are based on observed values from incomplete data sets.
AQI's are quantiles on curves fit to the upper tails of filled-in
data.  In addition, EC and AQI values are determined from data
representing different time periods.  EC's represent average air
quality over three years (1977-79), while AQI's represent air
quality for a single year  (1977, 1978, or 1979).  Air quality
during a single year may differ significantly from the three year
average.  A third reason is that an EC and the corresponding AQI
may represent different monitors.  The monitor used to determine
the EC for a city is determined by analyzing data from all moni-
tors in an urbanized area and identifying the monitor which re-
corded the highest CO levels.  The selection of monitors for
determining the corresponding AQI is limited to the sites used in
                               C-2

-------
the NEM analysis.  Because no more than six sites  (one per neigh-
borhood type)  are used to represent CO levels across a NEM study
area and because the boundary of the study area is smaller than
the corresponding urbanized area, the monitor used for determining
the AQI is often different from that used to determine the EC.
                              C-3

-------
TABLE C-l.  ESTIMATED CONCENTRATIONS (EC'S) DEVELOPED BY EPA AND
   CORRESPONDING AIR QUALITY INDICATORS (AQI'S) FROM TABLE 5-6
              (concentrations in parts per million)
Study area
Chicago
Los Angeles
Philadelphia
St. Louis
1-h average value,
1 expected exceed.
EC
30.9
37.8
32.9
27.9
AQI
24.9
31.4
19.2
22.8
8-h running average value
1 expected exceed.
EC
18.0
24.4
14.7
17.0
AQI
15.6
20.3
14.3
14.7
5 expected exceed.
EC
14.0
17.0
11.0
10.0
AQI
12.9
16.1
9.9
10.9
                              C-4

-------
                                    TECHNICAL REPORT DATA
                             (Please read Instructions on the reverse before completing)
 1 REPORT NO.
  EPA-450/5 84  003
                               2.
                                                             3. RECIPIENT'S ACCESSION NO.
 4. TITLE AND SUBTITLE
  The NAAQS  Exposure Model  (NEM)
  Applied to Carbon Monoxide
              5. REPORT DATE
               December 1983
             6. PERFORMING ORGANIZATION CODE
 7. AUTHOR(S)

  Ted Johnson  and  Roy A. Paul
                                                             8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
  PEDC.O Environmenta, Inc.
  505 South Duke  Street  Suite  503
  Durham. North Carolina  27701
                                                             10. PROGRAM ELEMENT NO.
              11. CONTRACT/GRANT NO.
 12. SPONSORING AGENCY NAME AND ADDRESS.
  U.S. tnvironmental  Protection Agency
  Office of  Air  and Radiation
  Office of  Air  Quality Planning  and Standards
  Research Triangle Park, North Carolina  27711
              13. jyPEOF RE PORTAND PERIOD CO VERED
                Final
              14. SPONSORING AGENCY CODE
 15. SUPPLEMENTARY NOTES
 16. ABSTRACT
      This  report presents a version of the National  Ambient Air  Quality Standard
 (.NAAQS) Exposure Model (NEM)  suitable for assessing  carbon monoxide (CO) exposure  and
 presents the  results of applying  it to CO.  NEM  is  a simulation  model  that simulates
 the intersection of a population  with pollutant  concentrations over space and
 time to estimate exposures that would obtain  if  various alternative NAAQS were
 just met.   Estimates are presented for adults with  cardiovascular  disease in four
 urban study areas and for a nationwide extrapolation.
17.
                                 KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                               b.IDENTIFIERS/OPEN ENDED TERMS
                           c.  COSATI field/Group
  Carbon Monojxi.de
  Air Pollution
  Exposure Assessment
  Air Quality  Standards
jlS. DISTRIBUTION STATEMEN1
'  Release to Public
19. SECURITY CLASS (This Report,
  Unclassified
21. NO. OF PAGES
  197
                                               20 SECURIT1- CLASS 'This page)
                                                                           22. PRICE
 EPA Form 2220-1 (Rev. 4-77)   PREVIOUS EDITION is OBSOLETE

-------
                                                         INSTRUCTIONS

   1.   REPORT NUMBER
       Insert the LPA report number as it appears on the cover of the publication.

   2.   LEAVE BLANK

   3.   RECIPIENTS ACCESSION NUMBER
       Reserved for use by each report recipient.

       TITLE AND SUBTITLE
       'itle should indicate clearly and briefly the subject coverage of the report, and be displayed prominently.  Set subtitle, if used, in smaller
         •x or otherwise subordinate it to mam title. When a report is prepared in more than one volume, repeat the primary title, add volume
         mber and include subtitle for the specific title.

   5.   REPORT DATE
       Each report shall carry a date indicating at least month and year.  Indicate the basis on which it was selected (e.g., date of issue, date of
        pf.  oval, date of preparation, etc.}.

   6.   PERFORMING ORGANIZATION CODE
       Leave blank.

   7.   AUTHOR(S)
       Give name(s) in conventional order (John R. Doe, J. Robert Doe, etc.).  List author's affiliation if it differs from the performing organi
       zation.

   8.   PERFORMING ORGANIZATION REPORT NUMBER
       Insert if performing organization wishes to assign this number.

   9.   PERFORMING ORGANIZATION NAME AND ADDRESS
       Give name, street, city, state, and ZIP code. List no more than two levels of an organizational hirearchy.

   10.  PROGRAM ELEMENT NUMBER
       Use the program element number under which the report was prepared.  Subordinate numbers may be included in parentheses.

   11.  CONTRACT/GRANT NUMBER
       Insert contract or grant number under which report was prepared.

   12.  SPONSORING AGENCY NAME AND ADDRESS
       Include ZIP code.

   13.  TYPE OF  REPORT AND PERIOD COVERED
       Indicate interim final, etc., and if applicable, dates covered.

   14.  SPONSORING AGENCY CODE
       Insert appropriate code.

   15.  SUPPLEMENTARY NOTES
       Enter information not included elsewhere but useful, such as:  Prepared  in cooperation with, Translation of, Presented at conference of,
       To be published in,  Supersedes, Supplements, etc.

   16.  ABSTRACT
       Include a brief (200 words or less) factual summary of the most significant information contained in the report. If the report Contains a
       significant bibliography or literature survey, mention it here.

   17.  KEY WORDS AND DOCUMENT ANALYSIS
       (a) DESCRIPTORS - Select from the Thesaurus of Engineering and Scientific Terms the proper authorized terms that identify the major
       concept of the research and are sufficiently specific  and precise to be used as index entries for cataloging.

       (b) IDENTIFIERS AND OPEN-ENDED TERMS - Use identifiers for project names, code names, equipment designators, etc. Use open-
       ended terms written in descriptor form for those subjects for which no descriptor exists.

       (c) COSATI FIELD GROUP - Field and group assignments are to be taken from the 1965 COSATI Subject Category List. Since the ma-
       jority of documents are multidisciplinary in nature,  the Primary Field/Group assignment(s) will be specific discipline, area of human
       endeavor, or type of physical object. The apphcation(s) will be cross-referenced with secondary Field/Group assignments that will follow
       the  primary posting(s).

   18.  DISTRIBUTION STATEMENT
       Denote releasability to the public or limitation for reasons other than security for example "Release Unlimited."  Cite any availability to
       the  public, with address and price.

   19. &20. SECURITY CLASSIFICATION
       DO NOT submit classified reports to the National Technical Information service.

   21.  NUMBER OF PAGES
       Insert the total number of pages, including this one and unnumbered pages, but exclude distribution  list, if any.

   22.  PRICE
       Insert the price set by the National Technical Information Service or the Government Printing Office, if known.
EPA Form 2220-1 (Rev. 4-771 (Reverse)

-------