THE NAAQS EXPOSURE MODEL (NEM)
AND ITS APPLICATION TO
PARTICULATE MATTER
Ted Johnson and Roy Paul
PEDCo Environmental, Inc.
505 South Duke Street, Suite 503
Durham, North Carolina 27701
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air Quality Planning and Standards
Strategies and Air Standards Division
Research Triangle Park, North Carolina 27711
Henry Thomas, Task Manager
August 31, 1981
-------
CONTENTS
Figures iv
Tables yi
Acknowledgement i-x
1. Introduction 1-1
2. Overview of the Exposure Model 2-1
3. Development of Study Areas 3-1
3.1 Selection of study areas 3-1
3.2 Delineation of exposure districts 3-2
3.3 Reference 3-7
4. Simulation of Population Movement 4-1
4.1 Composition of cohort files 4-1
4.2 Development of population data 4-6
4.3 Minimizing the number of cohorts 4-7
4.4 References 4-14
5. Preparation of Air Quality Data 5-1
5.1 Selection of representative data sets 5-1
5.2 Simulation of missing values in TSP data sets 5-7
5.3 Conversion of 1-hour TSP values to 1-hour
IP,Q values 5-21
5.4 References 5-22
V
6. Simulation of Outdoor Air Quality Expected Under
Alternative Particulate Standards 6-1
6.1 The rollback model 6-1
6.2 Air quality indicators 6-2
6.3 Background concentrations 6-9
6.4 References 6-13
ii
-------
CONTENTS (continued)
7. Simulation of Particulate Levels in the Micro-
environments
7.1 Work-school microenvironment
7.2 Home-other microenvironment
7.3 Transportation vehicle microenvironment
7.4 Roadside and outdoor microenvironments
7.5 Summary
7.6 References
8. Analyses of Population Exposure Using NEM
8.1 Description of computer outputs
8.2 Determinants of exposure estimates
8.3 Uncertainty in NEM exposure estimates
8.4 Effects of different NAAQS's
8.5 Effects of smoking
Appendix A
Appendix B
Appendix C
Appendix D
Computer program used for determining
data adequacy
Weekday and weekend activity pattern data
for each A-0 subgroup
Estimation of cohort location and
population
Computer program used in calculating
population centroids
7-1
7-5
7-8
7-13
7-15
7-16
7-20
8-1
8-1
8-11
8-12
8-16
8-18
A-l
B-l
C-l
D-l
ill
-------
LIST OF FIGURES
Number Page
2-1 NAAQS Exposure Model (NEM) 2-3
3-1 Exposure Districts and Air Quality Monitoring
Sites in the Philadelphia Study Area 3-8
3-2 Exposure Districts and Air Quality Monitoring
Sites in the Los Angeles Study Area 3-9
3-3 Exposure Districts and Air Quality Monitoring
Sites in the Chicago Study Area 3-10
3-4 Exposure Districts and Air Quality Monitoring
Sites in the St. Louis Study Area 3-11
5-1 Geographic Units Used to Organize Population
Data 5-2
5-2 Four Possible Monitor Arrays 5-5
5-3 Population Centroids of Exposure Districts in
Chicago Study Area and Selected TSP Monitoring
Sites 5-9
5-4 Population Centroids of Exposure Districts in
LA-San Bernardino Study Area and Selected TSP
Monitoring Sites 5-10
5-5 Population Centroids of Exposure Districts in
Philadelphia Study Area and Selected TSP
Monitoring Sites 5-11
5-6 Population Centroids of Exposure Districts in
St. Louis Study Area and Selected TSP Monitor-
ing Sites 5-12
5-7 Computer Plot of 24-h Average TSP Data Reported
for 1978 by Monitoring Site 055760004101 5-15
5-8 Computer Plot of 24-h Average TSP Data for Moni-
toring Site 055760004101 after Simulation of
Missing Values 5-16
-------
LIST OF FIGURES (continued)
Number Page
5-9 Scatter Plot of 24-h COH and TSP Values Recorded
During 1977 at Site 14212009F01 in E St. Louis 5-17
5-10 Scatter Plot of 24-h COH and TSP Values Recorded
During 1977 at Site 264280061H01 in St. Louis 5-19
5-11 Scatter Plot of 24-h. Beta Scattering and TSP
Values Recorded During 1977 at Site
26480061H01 in St. Louis 5-20
8-1 Effect of Uncertainty of Microenvironment Factors
on IP 10 Exposure Estimates for Los Angeles
Under Current Air Quality with Smoking Contri-
bution Included 8-15
8-2 Effect of Possible IPio Standards on Exposure
Estimates for Los Angeles 8-17
8-3 Effect of Smoking on IPio Exposure Estimates
for Chicago 8-19
8-4 Effect of Smoking on IPio Exposure Estimates
for Philadelphia 8-20
8-5 Effect of Smoking on IPio Exposure Estimates
for Los Angeles 8-21
8-6 Effect of Smoking on IPio Exposure Estimates
for St. Louis 8-22
v
-------
LIST OF TABLES
Number Page
3-1 Monitoring Sites Reporting Adequate 1978 Data
in 24 Urban Areas 3-3
3-2 Monitoring Sites in Candidate Urban Areas,
1977-78 3-4
3-3 Sample .Output from Program CD (Complete Data) 3-6
4-1 Home-to-Work Trip Tables 4-4
4-2 Assumed Assignments to Home/ Work, and Other
Microenvironments 4-5
4-3 Derivation of Age-Occupation Groups from Census
Data Elements 4-8
4-4 Number of Persons by Age-Occupation Groups for
Exposure Districts in Chicago 4-10..
4-5 Number of Persons by Age-Occupation Groups for
Exposure Districts in Los Angeles 4-11
4-6 Number of Persons by Age-Occupation Groups for
Exposure Districts in Philadelphia 4-12
4-7 Number of Persons by Age-Occupation Groups for
Exposure Districts in St. Louis 4-13
5-1 TSP Sites Used to Represent Air Quality at
District Centroids 5-8
5-2 Peak 1-h and 24-h COS Values Recorded in 1977
at Selected Sites in Chicago and Philadelphia 5-21
6-1 Air Quality Indicators (ug/m ) for TSP Data 6-6
6-2 Comparison of Characteristic Largest Values
Estimated for TSP Data (n=365) by Least
Squares and Maximum Likelihood Fits to Upper
50 Percent 6-10
VI
-------
LIST OF TABLES (continued)
Number page
6-3 Comparison of Characteristic Largest Values
Estimated for TSP Data (n=365) by Least
Squares and Maximum Likelihood Fits to Upper
20 Percent 6-11
7-1 Studies Considered in Developing PM Microen-
vironment.Factors 7-2
7-2 Indoor/Outdoor TSP Ratios for Structures
Corresponding to Work-School Microenvironment 7-7
7-3 Mean Levels of Mass Respirable Particulates
(yg/m3) in Four Homes 7-9
7-4 Indoor-Outdoor TSP Ratios for Structures
Corresponding to Home-Other Microenvironment 7-10
7-5 Air Exchange Rates Determined by Moschandreas
Et Al. 7-12
7-6 Transformation of Ambient TSP Concentrations
from Monitor Height to Breathing Zone 7-17
7-7 Estimates of Microenvironment Factors for TSP 7-18
7-8 Estimates of Microenvironment Factors for IP-in 7-19
8-1 IPIQ Microenvironment Factors for Philadelphia 8-2
8-2 Number of Times Per Year that a Cohort or Person
Occurs in a Microenvironment by Exercise Level
in the Central Philadelphia Exposure District 8-3
8-3 Number of Times Per Year that a Cohort or Person
Occurs in a Microenvironment in Philadelphia
Study Area 8-4
8-4 Number of Pollutant Encounters Per Year in
Philadelphia Under Current Air Quality 8-5
8-5 Number of People Exposed Per Year in Philadel-
phia Under Current Air Quality 8-6
8-6 Number of People at Peak Exposure Per Year in
Philadelphia Under Current Air Quality 8-7
8-7 Number of People Exposed Per Year in Philadel-
phia Under Current Air Quality 8-8
vii
-------
LIST OF TABLES (continued)
Number Page
8-8 Average Number of Hours Spent at Various Exer-
cise Levels in a Typical Week by Age-Occupation
Group 8-13
8-9 People Occurrences at High Exercise Levels and
(Percent of Total) by Study Area 8-13
Vlll
-------
ACKNOWLEDGEMENTS
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 particu-
late matter have involved the efforts of many people. The fol-
lowing persons associated with PEDCo Environmental, Inc., played
important roles in conducting this study. Coauthor Ted Johnson
developed the statistical models used to process air quality
data and managed the development of air quality and population data
and managed the development of air quality and population data
bases. Coauthor Roy Paul wrote the population movement algorithms
and created the transportation files. Dr. Luke Wijnberg developed
procedures used for estimating missing air quality values. Irene
Griffin assisted in the compilation of population and air quality
data and in the selection of study area boundaries. James Capel
created and debugged the computer programs which processed air
quality data. Barbara Blagun developed estimates of background
pollutant concentrations. 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 implemented
computer programs used to calculate the exposure estimates. Henry
Thomas, SASD, facilitated conduct of the study and co-managed the
SASD, exposure assessment program with Thomas McCurdy. Much of the
cohort data used in the model is based upon work done by SRI
International.l
Dr. William F. Biller, EPA contractor, and Thomas B. Feagans,
SASD, developed the general exposure model. The model makes use
of ideas developed by EPA's Office of Research and Development and
others.2 The model was applied at an earlier stage of its
development to carbon monoxide.3
REFERENCES
1. 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..
EPA, Research Triangle Park, N. C. 27711, August 1979.
2. James L. Repace, Wayne R. Ott, and Lance A. Wallace. "Total
human exposure to air pollution," Proceedings of the 73rd
Annual Meeting of the Air Pollution Control Association,
Montreal, Canada, June 1980.
ix
-------
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. EPA,
Research Triangle Park, N.C. 27711, January 1981.
-------
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, ASB has
devoted substantial resources to develop an exposure model suita-
ble for evaluating alternative ambient air standards. The model
is known as NEM, an acronym for NAAQS Exposure Model.
NEM provides ambient pollutant concentration estimates for
selected districts within a study area and adjusts these estimates
to account for five typical microenvironments. It then simulates
typical movements of population subgroups, called cohorts, through
these 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.
This report describes the current version of NEM and its
application in four U.S. urban areas (Chicago, IL, St. Louis,
MO-IL, Philadelphia, PA, and Los Angeles, CA) to estimate popu-
lation exposures associated with alternative NAAQS's proposed for
particulate matter (PM). PM refers to all particles suspended in
the atmosphere. Size-specific categories of PM include fine par-
ticles (FP) with an aerodynamic diameter of 0 to 2.5 ym, coarse
1-1
-------
particles (CP), 2.5 to 15 ym in diameter, and total suspended
particles (TSP), 0 to about 30 ym in diameter. In analyzing
possible PM standards, EPA has been particularly interested in
inhalable particles with diameters of 0 to 10 ym (IP.-) and 0 to
15 ym (IP-ic) • Particles in these size ranges are deposited in
the tracheobronchial and alveolar regions where they may cause
undesirable health effects. At its July 1981 meeting, the Clean
Air Scientific Advisory Committee (CASAC) recommended that EPA
consider a PM standard which would control particles corresponding
to the IP-iQ fraction. Consequently, EPA decided that initial
application of NEM to PM would focus on estimating population
exposure to IP10- Since IPi0 data are not routinely collected,
IP-IO concentrations were estimated by applying proportionality
constants developed by SASD to TSP data. Results of these analy-
ses are included in this report. The contribution of indoor PM
sources to total population exposure is also evaluated.
1-2
-------
SECTION 2
OVERVIEW OF THE EXPOSURE MODEL
To estimate the total exposure of a population under a
specific set of conditions, it would be desirable to monitor the
exposure of representative individuals in the population under
those conditions and to integrate the exposure of each over time.
These individual exposures would then be extrapolated to the
larger population to determine total population exposure. How-
ever, resources were riot available at the time of this project
for such a massive study. Furthermore, some of the conditions to
be studied were hypothetical (e.g., air quality expected under a
proposed NAAQS). Accordingly, simplifying assumptions were made
so that a model could be developed which would produce reasonable
estimates of population exposure and which would be practical to
use.
In the resulting NAAQS Exposure Model (NEM), estimation of
population exposure was greatly simplified by representing the
land area of a study area by large, discrete "exposure districts."
The population within each exposure district was assigned to a
single discrete point, the population centroid. The air quality
level within each exposure district was represented by the air
quality level at the population centroid, which was estimated for
each hour of the year using data collected at nearby monitoring
sites. Because TSP and IP data are recorded as 24-hour averages,
the required hourly average values had to be simulated using 24-hour
monitoring data.
To simulate the change in pollutant levels observed when air
enters a building or vehicle, ambient air quality estimates were
adjusted to account for five different microenvironments: indoors
at work or school, indoors at home or other locations, inside a
2-1
-------
transportation vehicle, outdoors near a roadway, and other outdoor
locations. NEM simulated the hour-by-hour movements of representa-
tive population groups through different districts of the city and
through different microenvironments, accumulating the resulting ex-
posure over a period of one year.
Because degree of exposure and susceptibility to effects of
pollution vary with age, occupation, and level of activity, the
total population of each study area was divided into age-occupa-
tion (A-O) groups, and each A-O group was further subdivided into
three subgroups* For each subgroup, a typical pattern of activity
through the five microenvironments was identified, and the activity
level (high, medium, or low) in each microenvironment was specified.
The basic units of population considered by NEM were called
cohorts. Each cohort was identified by exposure district of resi-
dence, by exposure district of employment, by A-O group, and by
activity-pattern subgroup. During each hour of the year, each co-
hort was assigned to a particular exposure district and a particular
microenvironment. Since NEM simulated hour-by-hour air quality in
each district/microenvironment, exposure of each cohort could be
estimated for a one year period. These exposures were summed over
all A-O groups and activity-pattern subgroups to provide an estimate
of total population exposure. Output of NEM is a series of tables
which depict the frequency distribution of exposure for some or all
of the A^-O groups at different averaging times (e.g., 1 hour, 24
hours, 1 year). Figure 2-1 is a simplified flow diagram of the
model.
In using this model, PEDCo had to establish exposure dis-
tricts within each study area that would accommodate the available
breakdowns of transportation and census data; establish an air
quality data set for each exposure district; and set up files
listing hourly assignments to an exposure district, a microenvi-
ronment, and activity level for each cohort. In addition, files
had to be established which contained air quality adjustment fac-
tors appropriate to each microenvironment and to the height of each
monitor 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 subsequent sections.
2-2
-------
Analyst selects
1) Test city
2) NAAQS
START PROGRAM
SELECT NEXT COHORT
DETERMINE COHORT
POPULATION
COHORT POPULATION
FILE
NOk
USE COHORT?
DETERMINE EXPOSURE DISTRI
MICROENVIRON, ACTIVITY LEV
n
EL
»
DETERMINE AMBIENT AIR QUALITY
(AQ) IN EXPOSURE DISTRICT
T
DETERMINE ADJUSTED AQ
PER NAAQS
»•
. rnnoRT i nrATinw
FILE
HOURLY AIR QUALITY
FILE
ROLL-BACK FACTORS
DETERMINE AQ IN
MICROENVIRONMENT
MICROENVIRONMENT
FACTORS
ACCUMULATE 1 HOUR OF EXPOSURE
AT AQ LEVEL IN A REGISTER
SUMMARIZE COHORT EXPOSURES FOR
AGE-OCCUPATION (A-0) GROUPS
(OUTPUT)
FREQUENCY DISTRIBUTION OF EXPO-
SURES BY A-0 GROUP
Figure 2-1. NAAQS exposure model (NEM).
2-3
-------
SECTION 3
DEVELOPMENT OF STUDY AREAS
NEM has been used to estimate IP10 exP°sure in Chicago, Los
Angeles, Philadelphia, and St. Louis. This section describes how
these cities were selected and how exposure districts were delin-
eated within each city.
3.1 SELECTION OF STUDY AREAS
EPA desired four study areas with large urban populations
which would represent different geographic and climatic regions
of the nation. These study areas were to be used in population
exposure analyses for three pollutants currently undergoing NAAQS
analysis: PM, nitrogen dioxide (NO^), and sulfur dioxide (S02).
Consequently, preference was given to study areas with good air
quality data for two or three of the pollutants rather than study
areas with superior air quality data for only one pollutant.
To simplify the selection process, PEDCo considered only
the 105 urban areas with 1970 populations above 200,000. Initial
inspection of air quality data from these urban areas revealed
that the critical factor in the selection process would be data
adequacy. An urban area was considered to have an adequate data
base for a specific pollutant and year if it had three sites
reporting acceptable data. A data set was acceptable if it had
(1) at least thirty 24-hour values in a year with at least five
values per quarter or (2) at least 5840 hourly values in a year.
Since !PIQ values were to be derived from TSP values, potential
study areas were evaluated according to adequacy of TSP data.
Because of the documented bias in 24-hour S09 bubbler data, only
1-hour S02 data were considered. Although 1-hour NO- data were
preferred, 24-hour data were considered initially. These criteria
3-1
-------
for determining adequacy were considered to be relatively lenient;
however, few of the 105 urban areas had adequate data for all
three pollutants. Table 3-1 lists the 24 urban areas which had
adequate data for at least two of the three pollutants in 1978.
By eliminating urban areas with less desirable data bases, a
shorter list of candidate urban areas was obtained (Table 3-2).
Shown by region of the country are the total number of monitoring
sites within each candidate urban area and the number of sites
with acceptable data in 1977 and 1978. In the Northeast, Phila-
delphia had by far the best data base. In the Southeast, Wash-
ington, D.C., had better data than the other candidate cities,
but was not considered typical of the region. In the Midwest,
Chicago had the best data base for each of the three ^.pollutants;
however, St. Louis had respectable data and was considered more
typical of the region. In the West, Los Angeles was selected for
study, although the N02 monitoring method used at most sites in
1977-78 was not considered acceptable. San Bernardino, an adjacent
city with a respectable data base, was combined with Los Angeles
to form a larger study area (LA-San Bernardino).
3.2 DELINEATION OF EXPOSURE DISTRICTS
Within each urban area, it was necessary to delineate the
boundaries of six or more exposure districts, since transfers
between these exposure districts would represent movement of
cohorts within the urban area. Fortunately, regional transpor-
tation planning agencies had already delineated transportation
districts within each urban area based on census tracts and
enumeration districts and had compiled data on transfers (trips)
between these districts. Since population and other census data
could be obtained for these districts from U.S. census reports or
from planning agencies, the decision was made to combine these
existing transportation districts into exposure districts.
Boundaries of exposure districts were delineated according
to the following criteria:
3-2
-------
TABLE 3-1. MONITORING .SITES REPORTING ADEQUATE 1978 DATA IN
24 URBAN AREAS1
Urban
area
codea
1
2
3
4
5
6
7
8
9
10
11
12
16
19
21
23
24
33
40
47
49
54
56
78
CUyb
New York
Los Angeles
Chicago
Philadelphia
Detroit
San Francisco
Boston
Washington
Cleveland
St. Louis
Pittsburgh
Minneapolis
Milwaukee
San Diego
Cincinnati
Buffalo
Denver
Louisville
San Bernardino
Toledo
Salt Lake City
Bridgeport
Syracuse
Mobile
Number of acceptable sites
S02
(1 hr)c
17
9
14
15
13
7
4
4
4
12
6
6
5
4
6
6
6
0
3
3
4
3
3
3
N02
(1 hr)c
1
0
2
4
1
0
2
8
0
3
2
0
0
4
5
0
2
3
5
0
2
0
2
0
N02
(24 hr)d
3
0
52
1
0
0
0
2
35
0
0
13
4
,; 0
12
0
0
17
0
6
0
3
0
0
TSPd
72
6
91
35
35
9
16
46
46
24
13
20
21
6
45
40
18
18
4
18
6
4
19
7
Rank of urban area by population
largest city in urban area
:Sites with n >. 5840
Sites with 5 samples per quarter and n >_ 30
3-3
-------
TABLE 3-2. MONITORING SITES IN CANDIDATE URBAN AREAS, 1977-78
City
Year
S02 (1 hr)
Total
Acca
N02 (1 hr)
Total
Ace
N02 (24 hr)
Total
Ace
TSP
Total
Ace
Northeast
Philadelphia
77
78
21
21
13
15
12
9
4
4
8
4
1
1
44
41
40
35
Southeast
Washington
Bi rmingharrj
Houston
77
78
77
78
77
78
16
16
1
1
3
2
7
4
0
0
2
0
20
12
0
0
3
3
7
8
0
0
2
0
20
17
3
3
41
43
6
2
3
0
38
37
70
52
17
14
47
47
45
46
12
13
42
39
Midwest
St. Louis
Cincinnati
Chicago
Steubenville
77
78
77
78
77
78
77
78
23
18
9
7
40
36
1
1
4
12
6
6
18
14
1
1
14
18
8
6
12
'5
1
1
1
3
5
5
5
2
1
0
5
0
18
14
73
66
4
3
4
0
15
12
63
52
2
3
45
37
52
50
127
128
5
3
36
24
48
45
95
91
4
3
West
Los Angeles
Phoenix
San Bernardino
77
78
77
78
77
78
14
16
3
1
5
5
12
9
0
0
3
3
14
13
4
3
12
9
1
0
3
1
9
5
4
0
2
1
1
0
3
0
0
0
1
0
14
13
9
9
14
14
9
6
7
0
2
4
a: Sites with acceptable data
3-4
-------
CD Air quality levels at the population centroid of each
exposure district should be represented by monitors
that are close to the centroid;
(2) Exposure districts should be composed of transporta-
tion planning districts, so transportation data may
be used in the study;
(3) Exposure districts should be few in number (no more
than 10) to facilitate programming and to reduce
computer time; and
(4) The geographic area of the study area, as represented
by all districts combined, should cover as much of the
city and surrounding region as possible.
The fourth criterion tends to contradict the first one because of
the difficulty in finding monitoring sites close to district cen-
troids when districts are large. The final choice of district
boundaries represents a trade-off between study area size and ade-
quacy of monitoring data.
A tool that proved useful in selecting district boundaries was
a simple computer program which printed out a table showing data
completeness—an important aspect of data quality. The table was
designed to show which monitors in the study area reported quarterly
air quality data which were at least 50 percent complete and which
monitors reported quarterly data which were at least 75 percent com-
plete. N02 and S02 data were considered 50 percent and 75 percent
complete if a quarter contained 1095 and 1643 hourly average values,
respectively. In the case of TSP, 24-hour average values are usually
recorded at six-day intervals so that a data set which is 100 percent
complete will have 15 or 16 observations per quarter. Consequently,
we required 8 TSP values per quarter for 50 percent completeness and
11 values per quarter for 75 percent completeness.
The source code for this program is listed in Appendix A and
a sample output is shown in Table 3--3. Review of these summaries
indicated that monitors generally had either very complete data or
very incomplete data. Because a modest reduction in the complete-
ness criteria did not substantially increase the number of usable
3-5
-------
TABLE 3-3. SAMPLE OUTPUT FROM PROGRAM CD (COMPLETE DATA)
DATA ADEQUACY
BY QUARTER AT
IN STUDY AREA
TSP MONITORING
LA-SB
SITES
SMSA
4480
6780
6780
6780
4480
6780
6780
4480
6780
4480
4480
4480
4480
4480
44SO
4480
4480
4480
6780
6730
4480
4480
4480
6780
6780
6780
6780
6780
4480
6780
6780
MONITOR SITE
050500002101
G5056C001I01
05C560002I01
050580001101
0509C0002A01
051300001101
C52680001I01
052°40001F01
053420001101
053740001101
053900001101
054100001A01
054100001F01
054180001F01
C54180001I01
054180002101
D542000Q1IC1
G54260001I01
D55380001F01
055640001101
055760002A01
055760002F01
055760004101
056200001101
056400003F01
056535001101
056680001F01
C56680001I01
05822000U01
0584400Q3I01
0585100C1I01
METH
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
91
INT
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
75% COMPLETE QTRS
1976 77 78 79
XXXXXX XXXXXXXX
XX XX
X XXX
XXXX
X
X XX
XX XXX XXX
XX X
XXXXXX XX X X
X XXXX XX
XXXXXXXX X X
XXXXXX XXXX XXX
xxxxxxx
XXX X
XX XXXXXX XX
XX XXX X
XXXXXXXX
XXXXXX XXX X
XX XX
XXXXXX XXXXXX
XX XXX XXX
XXXXXXXXXX XX
XXXXXX XX
X XX XXX
X XXX
X
50Z COMPLETE QTRS
1976 77 7.8 79
XXXXXX XXXXXXXX
X
XXX XXXX
X X XXXX
XXXX
X X
X X «XXX
XX XXX XXXKXXX
xxxxxxx
XXXXXX XXXXXXX
xxxxxxxxx
XXXXXXXX X XX
XXXXXX XXXK XXX
X
XXXXXXXX
X X XK X X X
XXXXXXXXXX XX
XX XXX XXXOCXXX
xxxxxxxxx
XXXXXX XXXOCXXX
XXX *XX
XXXXXX XXXXXXX
XXXXXX X XXXXX
XXXXXXXXXX XX
XXXXXX. XXXXXXX
XXXX XXX
X XXXX
X X 8CX
3-6
-------
sites, no data set that was less than 75 percent complete was cho-
sen, except for isolated cases where a superior monitor location
made a compromise desirable.
An important consideration in delineating district boundaries
was the need to minimize the number of districts and thus minimize
computer runtime. Use of a small number of districts was not ex-
pected to significantly affect accuracy of the exposure estimates
because only a relatively small number of monitoring sites were
available for characterizing air quality. SASD decided that the
number of exposure districts in any study area should not exceed
nine. As shown in Figures 3-1 through 3-4, six districts were delin-
eated for Philadelphia, seven for St. Louis, eight for Chicago, and
seven for LA-San Bernardino. Methods used in selecting sites to
represent air quality in each district are discussed in Section 5.
3.3 REFERENCE
1. National Aerometric Data Bank, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina.
3-7
-------
U)
I
00
Pine Valley,
NJ A
Figure 3-1. Exposure districts and air quality monitoring sites
in the Philadelphia study area.
-------
I
\o
Pacific
Ocean
Figure 3-2. Exposure districts and air quality monitoring
sites in the Los Angeles study area.
-------
LAKE
MICHIGAN
7 A I
Figure 3-3. Exposure districts and air quality monitoring sites
in the Chicago study area.
3-10
-------
DA
A O
WOOD RIVER
D
BELLAFONTAINE
NEIGHBORS A
Figure 3-4. Exposure districts and air quality monitoring sites
in the St. Louis study area.
3-11
-------
SECTION 4
SIMULATION OF POPULATION MOVEMENT
An innovative aspect of this population exposure study is
the simulation of population movement through zones of varying
air quality. Previous studies have assumed that the exposure of
each member of a study area population is represented by the air
quality outside his or her place of residence.1 In this study, a
computer was used to simulate the movements of small, homogeneous
groups called cohorts from district to district in each study area
and through microenvironments representing residences, schools,
workplaces, transportation vehicles, roadsides, and the general out-
door environment. By definition, all members of a cohort received
identical exposures. This section describes the development of com-
puter files delineating cohort movements and the allocation of study
area population among the cohorts.
4.1 COMPOSITION OF COHORT FILES
Webster defines a cohort as "a group of individuals having a
statistical factor in common in a demographic study." In this
study, all members of a particular cohort
(1) live in the same district,
(2) work in the same district,
(3) are members of the same A-0 group, and
(4) are members of a subgroup with a specified daily
activity pattern.
Consequently, the file describing a cohort is labeled as to home
district, work district, A-O group, and activity pattern. Each.co-
hort file contains hourly assignments of the cohort to a district,
a microenvironment, and an activity level. Assignments are provided
for both weekdays and weekends. The file also indicates the number
of persons in each cohort.
4-1
-------
The cohort files may be conceived as a hierarchy of identity
codes. For example, a cohort might be identified as A-0 group 4
(clerical workers), activity pattern 2, home district 9, and work
district 8. This cohort is first categorized by A-0 group and
activity pattern (i.e., clerical workers, activity pattern 2).
Sequential assignments of this group to microenvironments and ac-
tivity levels are taken from a typical activity pattern (TAP)
file. The TAP file, developed for EPA by SRI International under
a previous project, is based on detailed surveys of activity
patterns conducted by various researchers.2 Three typical week-
day activity patterns are listed for each of the 12 A-O categories
for a total of 36 patterns. The TAP file also lists percentages
of each A-0 group assigned to each of the three activity patterns.
If census data indicate that a study area contains 100,000 cleri-
cal workers and if the TAP file indicates that 20 percent of
these workers follow activity pattern 2, then 20,000 persons can
be characterized as clerical workers who follow activity pattern
2.
To make the file correspond to home district 9, the cohort
would be assigned to that district during each hour that had a
microenvironment assignment (MA) which suggested that the cohort
members were in their homes. For example, we would assume that a
clerical worker in our sample group would be in his or her home
in district 9 at 5 a.m., since MA and activity level assignments
in the cohort file indicate that cohort members are in an indoor
environment and have a low activity rate at that hour.
To determine how many people can be characterized as clerical
workers, activity pattern 2, home district 9, we consult census
data available for each district. If census data indicate that 15
percent of the study area clerical workers reside in district 9,
then 3,000 persons would be characterized as clerical workers,
activity pattern 2, home district 9.
To determine how many of these 3,000 persons work in dis-
trict 8, we assume that the fraction of trips taken by district 9
clerical workers to work places in district 8 is equal to the
4-2
-------
fraction of total trips beginning in district 9 and ending in
district 8. If that fraction is 25 percent, then we would assign
750 persons to the cohort file labeled clerical workers, activity
pattern 2, home district 9, work district 8. Home-to-work trip
data developed by PEDCo from transportation planning data are
listed in Table 4-1.
Now that the cohort is completely specified, we must deter-
mine its movements through the exposure districts. We have
already assigned the cohort to district 9 during certain hours of
the day when the MA suggested that members of the cohort were
home. We now assign the cohort to district 8 whenever the MA
suggests that cohort members are working or having lunch. If the
MA suggests the cohort is in transit between home and work, it is
assigned to the home district. The cohort is also assigned to
the home district whenever the MA suggests the cohort is shopping
or participating in some other activity that is likely to occur
in the cohort's home district. If the MA suggests the cohort is
in transit between work and home or work and another destination,
it is assigned to the work district. Our assumptions concerning
weekday assignments of A-0 groups to home, work, and "other" are
summarized in Table 4-2. Similar procedures were used to determine
the number of persons in each cohort considered in the exposure
analyses and to make activity level/microenvironment/district
assignments for weekday movements.
Determining weekend assignments was less straightforward.
In addition to weekday patterns, the TAP file contains three
typical Saturday patterns and three typical Sunday patterns for
each A-0 category. Unfortunately, the allocation of a particular
A-0 group to the three Saturday subgroups is different from the
allocation to the three weekday subgroups and from the allocation
to the three Sunday subgroups. In other words, weekend subgroups
are different from weekday subgroups. Therefore, cohorts defined
by weekday activity patterns are not followed into the weekend by
TAP data. Nevertheless, it was necessary to account for weekend
cohort activities because weekends represent a significant portion
4-3
-------
TABLE 4-1. HOME-TO-WORK TRIP TABLES
Home
District
Work District
1
2
3
4
5
6
7
8
Chicago
1
2
3
4
5
6
7
8
29929
74519
30101
12268
53602
19344
11250
10656
4894
29636
7743
684
5333
3810
1885
1020
4493
9591
14766
2625
6288
831
1017
1755
889
592
1350
3397
4017
73
40
353
1445
916
958
1930
16316
242
28
122
614
6540
884
"151
1604
4432
754
189
1013
2416
1229
191
1090
566
1096
606
1031
1754
1720
704
1231
271
739
1077
Philadelphia
1
2
3
4
5
6
270524
34422
65791
70374
61761
23862
34422
59618
16528
4404
2288
2156
65791
16528
52154
20977
7627
3652
70374
4404
20977
100450
30482
3279
61761
2288
- 7627
30482
169931
14911
23862
2156
3652
3279
14911
149034
0
0
- 0
0
0
0
0
0
0
0
0
0
St. Louis
1
2
3
4
5
6
7
14139
26518
10687
25310
11031
18225
12114
16516
20403
9313
8941
3431
10657
3062
10687
9313
15332
12119
6211
16146
2122
25311
8941
12118
64828
10230
13487
3330
11031
3431
6211
10230
61327
19396
1207
18225
10656
16146
13485
19396
141535
3433
.12114-
3062
2123
3330
1208
3433
36219
_.. .0 _.
0
0
0
0
0
0
Los Angeles
1
2
3
4
5
6
7
24697
10607
30662
3159
12534
374
812
;
4959 '
54720
6972
670
12392
111
30
8954
7933
89090
6404
6300
595
115
465
351
3526
19719
3209
2787
198
4205
8748
4106
4653
122655
909
188
--
28
31
177
2107
396
7700
907
•
3
8
27
122
88
1196
10932
0
0
0
0
0
0
0
4-4
-------
TABLE 4-2. ASSUMED ASSIGNMENTS TO HOME, WORK, AND OTHER MICROENVIRONMENTS
ArO
group
1
2
3
^
4
5
6
7
8
9
10.
11
12
Group description
Students, IB and over
Managers, administrators, pro-
fessional, technical, and kin-
dred workers
c&lpc workers
^ G 1 CA « W • *** 1 *
Clerical and kindred workers
Craftsmen, foremen and kindred
workers
Operatives and laborers
Farm workers
Service, military, and private
household workers
Kerried women, not in labor
force
Laid off, unemployed, retired
Children under age 5
Children age 5 through 17
Activity pattern
1
2
3
1
2
3
}
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
' 2
3
1
2
3 .
1
2
3
1
2
3
1
2
3
Weekday
mi croenvi ronmsnt*
changes (24-hr cycle)
M*v* H * M
H-O-W-H
H"*v*n
Mtravel 13 hrs)-^H
U H _JJ
^^T^^^
H-»t "i »r i n i PI i u
Mil.1 5»'J— 0-*-H
H-O-H
H-1MKK>-H
VHr-CWJ
t^t^ij
^^^^^^1
hXWi
U 1 f U
1 I *rl 'It
K-tf-0-«
H-0-fl
U II U
"^^^^^^
H-O-H
U 1 » U
u • n pn
K.
H-0-*
W)
K-W-CWWMHi
K-0-41
ri ' W * n
K.
H-O-W
H-O-H
u « t 'u
1 1 ' n ' j i
B-O-H
H-0-fl
H-0-fl
H-*-0*H
H-0-fl
U < I U
11 T! " 1 J
U . H .. 11
1 1 w It
tJ . M.._U
11 nit
r.icroenvironment abbreviations: H « home ('other1 category interpreted as
home); 0 - other, unspecified (other, interpreted as non-home); W * work-
place.
4-5
-------
of the total hours in a year. One option was to assume that all
persons stay in the home microenvironment during the weekend.
This was viewed as unrealistic. A second option was to use a
Saturday pattern and a Sunday pattern from the TAP file which
were consistent with a given subgroup's weekday pattern. Although
the patterns did not match exactly, this procedure provided a
more realistic simulation than one which assumed all persons
stayed at home.
Weekday and weekend activity pattern data for each A-0
subgroup are listed in Appendix B.
4.2 DEVELOPMENT OF POPULATION DATA
An algorithm was devised to determine the number of persons
in each cohort in a study area, given (1) the number of persons
in each A-0 group who reside in each exposure district, (2) the
fraction of each A-0 group represented by each activity pattern,
and (3) the home-to-work trip table. In Chicago, the transporta-
tion planning agency possessed data on the age and occupational
characteristics of people living in planning districts.5 These
data were used to determine the A-0 group breakdown in each
exposure district. Transportation planning agencies in St.
Louis, Philadelphia, and Los Angeles did not use A-0 data in
their studies, and thus did not have such data available for
transportation planning districts. Consequently, detailed 1970
United States Census A-O data were obtained from the National
Planning Data Corporation (NPDC) under a separate EPA contract.
In the case of LA-San Bernardino, 1970 census data may give an
inaccurate indication of the A-0 distribution of population
across that study area in 1977-78 because rapid growth during the
1970's significantly changed the demographic characteristics of
the population. Nevertheless, no better data were available from
the transportation planning agency or NPDC.
Some preliminary analyses were necessary before NPDC could
extract appropriate census data from its computerized files.
First, each transportation planning district had to be defined in
4-6
-------
terms of census tracts. This task was relatively easy since
transportation planning districts are based on census tracts.
Second, EPA's A-0 categories had to be defined in terms of data
items published by the U.S. Bureau of Census (BOC). These defi-
nitions are summarized in Table 4-3. With these two inputs, NPDC
was able to extract the required data for each census tract,
manipulate the data to yield the desired A-0 groups for each
tract, and summarize the data for all census tracts in each
defined transportation district. Tables 4-4 to 4-7 list the
number of persons in each A-0 category by city and exposure
district.
4.3 MINIMIZING THE NUMBER OF COHORTS
One objective in developing the NEM computer program was to
minimize computer runtime. Since the NEM program accumulates an
annual exposure for each defined cohort, runtime can be reduced
by limiting the number of cohorts.
One way to limit the number of cohorts.is to minimize the
number initially defined. In this study the number was limited
by using only 11 A-0 groups and only 3 activity-pattern subgroups.
Cohort number was further reduced by limiting the number of exposure
districts to no more than nine in each study area.
Another means of reducing the number of cohorts is to select
only those of significant population size. If a cohort had only
200 members, its exposure experience represents only 1/10,000 of
the exposure experience of a city of 2 million people. If the
cohort does not experience extreme air quality levels, it may be
dropped with.no appreciable effect on total exposure estimates.
Preliminary experiments with NEM showed a large number of
cohorts with very small populations. An analysis by George Duggan,
SASD, showed that these cohorts could be excluded without signif-
icantly affecting results. Consequently, we decided that a
minimum cohort size would be adopted for each city such that the
total population of all discarded cohorts would not be more than
4-7
-------
TABLE 4-3. DERIVATION OF AGE-OCCUPATION GROUPS FROM CENSUS DATA ELEMENTS
Age-occupation group
Census data derivation3
i
00
1. Students 18+
2. Managers/professionals
3. Sales workers
4. Clerical workers
5. Craftsmen, foremen .and kindred workers
6. Operatives and laborers
7. Farm workers
8. Service, military, and household workers
9. Married women, not in labor force, hus-
band present
10. Laid off, retired, or unemployed
11. Children under age 5
College school enrollment (Table P-2)
(Professional, technical, and kindred workers (P-3))
.^(Managers and administrators, except farm (P-3))
Sales workers (P-3)
Clerical and kindred workers(P-3)
Craftsmen, foremen and kindred workers (P-3)
(Operatives, except transport)
WTransport equipment operators)
+(Laborers, except farm)
Farm workers
(Total male labor force (P-3)-)
+(Total female labor force (P-3))
- Male civilian labor force (P-3))
- Female civilian labor force (P-3))
-(Males not in labor force (P-3))
'Females not in labor force (P-3))
Service workers (P-3))
Private household workers (P-3))
(Number of husband-wife families (P-3))
-(Married women, husband present, in labor force (P-3))
(Male civilian labor force, unemployed)
'Female civilian labor force, unemployed)
Male, not in labor force, other over 65;
Female, not in labor force)
Male, not in labor force, other over 65)
Male not in labor force)
(Males less than 5)
+(Females less than 5)
(continued)
-------
TABLE 4-3 (continued)
Age-occupation group
12. Children 5-17
Census data derivation
+
+
+
+
+
+
4-
•f
+
Males 5-9)
Males 10-14)
Males 15;
Males 16
Males 17)
Females 5-9)
Females 10-14)
Females 15
Females 16 j
Females 17,
I
VD
-------
TABLE 4-4. NUMBER OF PERSONS BY AGE-OCCUPATION GROUPS FOR EXPOSURE DISTRICTS IN CHICAGO*
Age-occupation group
1. Students 18+
2. Professionals/managers
3. Sales workers
4. Clerical and kindred
workers
5. Craftsmen, foremen,
kindred
6. Operatives and laborers
7. Farm workers
8. Military, service,
household
9. Married housewives
10. Unemployed or retired
11. Children under 5
12. Children 5-17
Total
Percent
Dist 1
36,658
201,789
41,826
155,647
54,301
59,368
-
57,243
43,053
18,549
6,619
15,886
609,939
29.2
Dist 2
20,869
73,074
19,022
44,368
44,624
53,313
-
30,122
50,854
16,439
17,646
42,350
412,681
17.4
Dist 3
9,671
50,033
9,599
31,578
39,649
45,370
-
18,472
30,687
9,052
7,543
18,103
269,757
11.4
Dist 4
5,819
23,888
6,094
16,419
19,941
25,699
-
9.923
23,087
4,040
10,167
24,400
169,477
7.2
Dist 5
10,965
38,957
6,159
21,588
16,134
22,614
-
21,442
22,528
9,318
8,319
19,965
197,989
8.4
Dist 6
9,822
42,698
7,946
23,203
14,897
15,397
-
9,740
33,367
5,690
12,218
29,323
204,301
8.6
Dist 7
9,805
46,454
9,941
28,756
25,556
21,163
-
11,125
32,794
5,856
13,116
31,478
236,044
10.0
Dist 8
6,012
30,183
7,574
14,918
20,343
19,070
-
9.940
26,278
4,841
13,121
31,490
183,770
7.8
Total
109,621
507,076
108,161
336,477
235,445
261,994
-
168,007
262,648
73,785
88,749
212,995
2,364,958
Percent
4.6
21.4
4.6
14.2
10.0
11.1
-
7.1
11.1
3.1
3.8
9.0
100.0
-p.
l-^>
o
-------
TABLE 4-5. NUMBER OF PERSONS BY AGE-OCCUPATION GROUPS FOR EXPOSURE DISTRICTS IN LOS ANGELES"
Age-occupation group
1. Students 18+
2. Professionals/
managers
3. Sales workers
4. Clerical and kindred
workers
5. Craftsmen, foremen,
kindred
6. Operatives and
laborers
7. Farm workers
8. Military, service,
household
9. Married housewives
10. Unemployed or retired
11. Children under 5
12. Children 5-17
Total
Percent
Dist 1
75,046
155,876
46,363
151,099
66,590
134,678
978
71,972
166,345
202,602
115,731
267,644
1,454,924
18.8
Dist 2
100,012
213,582
60,664
200,412
129,717
210,347
2,412
137,441
286,042
227,923
214,646
581,826
2,365,024
30.6
Dist 3
48,615
154,501
46,686
101,891
50,952
66,563
859
49,075
148,959
129,548
76,154
214,017
1,087,820
14.1
Dist 4
25,361
55,571
17,851
39,496
33,404
48,417
1,082
25,865
76,608
39,490
56,306
193,468
612,919
7.9
Dist 5
72,233
178,221
58,343
115,642
88,172
113,097
3,471
74,329
217,943
108,141
140,187
467,283
1,637,062
21.2
Dist 6
7,998
18,649
6,723
13,861
15,811
22,135
3,369
12,676
36,585
21,994
25,295
76,679
261,775
3.4
Dist 7
12,414
25,235
8,581
19,255
15,324
18,383
1,654
22,166
44,149
36,057
26,226
84,368
313,812
4.1
Total
341,679
801,635
254,211
641,656
399,970
613,620
13,825
393,524
976,631
765,755
654,545
1,885,285
7,733,336
Percent
4.4
10.4
3-2
8.3
5.2
7.9
0.2
5.1
12.6
9.9
8.5
24.4
100.0
-------
TABLE 4-6. NUMBER OF PERSONS BY AGE-OCCUPATION GROUP FOR EXPOSURE DISTRICTS IN PHILADELPHIA'
Age-occupation group
1. Students 18+
2. Professionals/managers
3. Sales workers
4. Clerical and kindred workers
5. Craftsmen, foremen, kindred
6. Operatives and laborers
7. Farm workers
8. Military, service, household
9. Married housewives
10. Unemployed or retired
11. Children under 5
12. Children 5-17
Total
Percent
Dist 1
18,429
43,001
14,397
64,791
39,053
98,767
862
57.69C
93,110
96,978
75,482
209,444
812,009
27.6
Dist 2
10,822
32,332
10,577
32,094
20,125
19,976
191
11,943
45,981
29,295
24,482
77,508
315,326
10.7
Dist 3^
22,047
39,197
11,241
30,188
13,584
29,742
524
26,119
39,072
38,268
25,833
82,940
358,755
12.2
Dist 4
19,014
60,326
18,652
51,355
24,512
39,968
402
24,865
68,770
59,796
37,918
117,577
523,155
17.8
Dist 5
13,584
43,477
19,170
50,753
31,826
38,201
238
20,531
75,618
46,350
40,410
114,716
494,874
16.8
Dist 6
10,127
39,546
14,926
35,410
26,254
34,003
536
19,714
61,374
38,346
38,123
115,501
433,860
14.8
Total
94,023
257,879
88,963
264,591
155,354
260,657
2,753
160,867
383,925
309,033
242,248
717,686
2,937,979
Percent
3.2
8.8
3.0
9.0
5.3
8.9
0.1
5.5
13.1
10.5
8.2
24.4
100.0
-------
TABLE 4-7. NUMBER OF PERSONS BY AGE-OCCUPATION GROUPS FOR DISTRICTS IN ST. LOUIS1
Age-occupation group
1. Students 18+
2. Professionals/managers
3. Sales workers
4. Clerical and kindred
workers
5. Craftsmen, foremen,
kindred
6. Operatives and laborers
7. Farm workers
8. Military, service,
household
9. Married housewives
10. Unemployed or retired
11. Children under 5
12. Children 5-17
Total
Percent
Dist 1
875
2,826
538
2,826
1,304
4,862
73
4,061
3,830
8,267
5,647
16,382
51,491
4.3
Dist 2
2,982
7,558
2,337
11,789
5,410
19,049
209
16,180
16,197
21,438
14,888
46,776
164,813
13.6
Dist 3
6,200
8,305
2,144
8,775
3,235
11,019
223
11,915
9,280
15,456
9,834
27,593
113,979
9.4
Dist 4
6,262
17,168
6,711
26,696
14,166
28,259
180
14,480
36,095
42,507
19,604
56,909
269,037
22.3
Dist 5
6,779
26,165
7,842
12,850
6,378
6,889
189
6,313
23,669
13,949
12,243
43,470
166,736
13.8
Dist 6
12,807
36,858
12,256
24,916
14,077
18,261
276
11,727
39,237
24,378
22,145
73,380
290,318
24.0
Dist 7
3,606
7,830
2,966
10,518
7,207
14,869
378
8,115
18,479
15,853
15,640
46,175
151,636
12.6
Total
39,511
106,710
34,794
98,370
51,777
103,208
1,528
72,791
146,787
141,848
100,001
310,685
1,208,010
Percent
3.3
8.8
2.9
8.1
4.3
8.5
0.1
6.0
12.2
11.7
8.3
25.7
100.0
*»
I
M
U)
-------
10 percent of the total population of the study area. Further
analysis suggested that a minimum cohort population of 1,500
persons was appropriate for LA-San Bernardino, Chicago, and
Philadelphia; a minimum cohort population of 500 persons was ap-
propriate for St. Louis.
Initially, 777 cohorts containing a total of 7,719,123
persons were defined for the LA-San Bernardino study area.
Omitting all cohorts with less than 1,500 members resulted in
rejection of 462 cohorts (59.5%) with a population loss of only
128,935 (1.7%). In Chicago, 992 cohorts comprised a study area
population of 2,364,467. Of these, 673 cohorts (67.8%) were
rejected, causing a population loss of 220,598 (9.3%). In St.
Louis, 483 of 777 cohorts were too small to use, representing a
cohort rejection rate of 62.1%. The loss of population was
86,158 out of 1,206,078, a population reduction of 7.1 percent.
In Philadelphia, 588 cohorts comprised a study area population of
2,934,753. Of these, 317 cohorts (53.9%) were rejected, causing
a population loss of only 155,284 (5.3%). Thus the cohort reduc-
tion procedure reduced computer runtime in a cost-effective
manner.
After the small cohorts were set aside, the remaining cohorts
were used as a sample population in NEM simulation. The resulting
exposure estimates were adjusted to reflect the total study area
population. For example, if sample cohorts represented 92.9 per-
cent of the study area population, each sample exposure estimate
was multiplied by Q0 Q or 1.076 to yield a final estimate. This
y £» • y
procedure assumed that cohorts omitted from the simulation have
the same exposure pattern as those included.
4.4 REFERENCES
1. National Air Quality, Monitoring, and Emissions Trends
Report, 1977, Publication No. EPA-450/2-78-052, U.S.
Environmental Protection Agency, Research Triangle Park,
N.C., December 1978, Section 2.
4-14
-------
2. 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.
EPA, Research Triangle Park, N.C. 27711, August 1979.
3. U.S. Bureau of Census, U.S. Census of Population and Housing
1970, Census Tracts, Final reports PHC (1)-17,-43,-159,-181,
and -187.
4. National Planning Data Corporation, Syracuse, New York,
September 1980.
5. Chicago Area Transportation Study (CATS), Data provided to
PEDCo Environmental, Inc. and the U.S. Environmental Protec-
tion Agency, August 1980.
4-15
-------
SECTION 5
PREPARATION OF AIR QUALITY DATA
NEM requires representative outside air quality data for
each exposure district in the form of hourly average values with-
out gaps. This section describes the procedures used for select-
ing appropriate data sets and for filling in missing values.
5.1 SELECTION OF REPRESENTATIVE DATA SETS
To simplify the computer simulation, outside air quality
throughout an entire exposure district was assumed to be equal to
the outside air quality at the population centroid of the exposure
district. The population centroid is analogous to a center of
mass for an object; population distribution about the centroid is
balanced in all directions. The coordinates (x., y.) of each
population centroid were estimated from 1970 census data using
the equations
Zx.P.
xi = -rib1 (5-D
and
<5-2>
where, for the jth census division in the ith exposure district,
x. is the longitude of the geographic centroid, y. is the lati-
tude of the geographic centroid, and P. is the population.
As shown in Figure 5-1, exposure districts are composed of
census tracts (CT's), which in turn are composed of block groups
(BG's) or enumeration-districts"(ED's). Since the Bureau of
Census (BOC) has recorded X-Y coordinates and population data for
every BG and ED in the United States, these units provided a con-
venient base for calculating population centroids.
5-1
-------
U1
I
NJ
STUDY AREA
EXPOSURE DISTRICT
CENSUS TRACT
BLOCK GROUP
(Enumeration
District if
area not blocked)
BLOCK
Figure 5-1. Geographic units used to organize population data.
-------
BOG has compiled computer tapes containing ED/BG coordinates
data from the 1970 U.S. census into a master enumeration districts
file (labeled MEDX). Through a previous project, EPA acquired a
compact form of this census file (labeled MEDX-FOR) which contains
only those items of interest to population exposure analyses. l
Because MEDX-FOR is in hierarchical format by state and county,
PEDCo was able to develop a Fortran program (labeled CTDS) which
could search the file and calculate population centroids. The
source code for this program is listed in Appendix D. Target
records in MEDX-FOR were determined using another file which
listed the CT's in each exposure district and the identification
number for the Standard Metropolitan Statistical Areas (SMSA's)
in which the CT's are located. The SMSA number was needed because
certain exposure districts overlap two or more SMSA's which use
identical CT identification numbers.
In executing CTDS, PEDCo discovered duplicate entries for
some BG and ED records. Because these BG's and ED's were counted
twice, the locations of certain population centroids are slightly
biased towards the areas containing duplicate entries. If MEDX-
FOR is to be used frequently, the file should be edited.
Once the location of a population centroid was determined,
there remained the problem of determining outside air quality at
the centroid. Two methods were considered: (1) use of data re-
corded at a single nearby site and (2) interpolation of data re-
corded at three nearby sites. In both cases, only sites with
acceptable data would be used, and missing values would be filled
in by simulation.
The interpolation method assumes that the air quality
throughout a region is a spatial continuum which changes shape
every hour. Monitoring data represent values at discrete points
in this continuum. If there are no discontinuities or steep
slopes in the continuum, the pollutant level at any location
should be reasonably well estimated by interpolating values
recorded at nearby sites.
5-3
-------
Horie and Morrison2 discuss several interpolation schemes
which have the general form
m
ci -
c .w .
(5-3)
where c. is the estimated concentration at the ith centroid, c.
1 ' D
is the observed value at the jth station, and w. is a weighting
coefficient. They recommend the interpolation formula
(5-4)
where r . is the distance from the jth station to the ith centroid.
Note that only the nearest three stations are used in the calcu-
lation.
Another interpolation method determines c. values by fitting
a plane through the points (x-,, y-,, c, ) , (x?' V2' °2^ ' an<^ (X3'
y~, c-) where (x . ,t y. ) are the location coordinates of a station.
The equation of the plane passing through these points is
Ax+By+Cc+l=0.
(5-5)
Constants A, B, and C can be determined from the matrix equation
-1
A
B
C
X
-1
-1
(5-6)
The value c. can be estimated as
where (x., y.) are location coordinates of the ith population
centroid. If more than three stations are desired for the inter-
polation, this method can be expanded to produce a least squares
plane (LSP).
The Horie and LSP methods can yield significantly different
estimates of c. values depending on the relative positions of
the centroid and the nearest three monitors. Figure 5-2 shows
four possible monitor arrays where r, = r2 = r^. Assume c, = 10,
c_ = 20, and c., = 20. Because the centroid is equally distant
5-4
-------
Qci / \
\ ' \
\
' i
c2 c3
Array C
-b
Array A 3
n
M C?
M
f I
f I
c
Ci \ 2
N Array B
\
\
Array D
Figure 5-2. Four possible monitor arrays
-------
from each monitor, the Horie method will yield the same estimate
(16.7) for each of the four arrays. The LSP method yields four
different results.
array c.. (Horie) c.. (LSP)
A
B
C
D
• 16.7
16.7
16.7
16.7
15.8
20.0
15.0
30.0
The LSP estimates differ from the Horie estimates because the
LSP method explicitly assumes a linear pollutant gradient. In a
sense, the LSP method uses more of the available information; it
considers the actual position of each monitor, whereas the Horie
method considers only the distances between monitors and centroids.
Thus, the LSP method should yield more representative estimates
for most monitor arrays.
A serious problem common to both interpolation methods is
the data smoothing which results whenever two or more data sets
are combined. Consider Array A (Figure 5-2) where three monitor-
ing stations are located at vertices of an equilateral triangle,
and the population centroid is located at its exact center. As-
sume that monitoring data at the three sites represent samples
from a single population of independent, identically-distributed
2
normal variates with mean y and variance a . Using either inter-
polation method, the estimate of c. at the population centroid
is the mean of values at the three monitors, and the resulting
time series has a mean approaching y and a variance approaching
2
a /3. Consequently, the variance at the centroid is about 1/3
the variance of any individual monitor. Departures from the above
assumptions may cause less loss of variance, but under all reason-
able assumptions we can expect a significantly smoothed data set.
In LA-San Bernardino, TSP monitors with acceptable data were
so widely separated that spatial interpolation could not be
justified. Because of the scarcity of TSP data in LA-San Bernardino
and because interpolation smoothes data, SASD chose to use data
from a single, nearby site .to._estimate air quality at population
5-6
-------
centroids. In effect, the total population of each district is
assumed to be located at the monitoring site representing the
district. Table 5-1 lists the TSP data sets used to represent
air quality for each district. Figures 5-3 through 5-6 show the
locations of the TSP monitors which provided these data and the
locations of the population centroids.
5.2 SIMULATION OF MISSING VALUES IN TSP DATA SETS
PEDCo Environmental developed a time series model3'4 based on
Fourier analysis which can be used to fill in missing values in
hourly average data sets which are at least 75 percent complete.
This model was not appropriate for TSP data sets which often have
more than 300 missing values out of a total of 365 potential 24-hour
average values. Consequently, an alternative model was developed
for TSP which was less dependent on data completeness.
Preliminary analyses suggested that urban TSP data contain
seasonal and weekday/weekend patterns. In addition, TSP data
tend to be better fit by the lognormal distribution than the
normal distribution. Consequently, we chose to model TSP concen-
trations using the model
,yk = In xk = a + 'B^ + 32D2 + &3D3 + 0^ + e (5-8)
where x, is the 24-hour average TSP concentration for day k, a is
a constant, the 3's are coefficients, e is a random component with
normal distribution N(0,a), and
D, = 1 for weekdays
0 for other days
D2 = 1 for winter
0 for other seasons
D, = 1 for spring
0 for other seasons
D. = 1 for summer
0 for other seasons.
5-7
-------
TABLE 5-1. TSP SITES USED TO REPRESENT AIR QUALITY
AT DISTRICT CENTROIDS
Study area
Chicago
Los Angeles
Philadelphia
St. Louis
Year
1978
1978
1978
1977
District
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
1
2
3
4
5
6
1
2
3
4
5
6
7
Monitoring site
141220018H01
141220029H01
141220036H01
141220015H01
141220032H01
147160007F01
142620001G01
141540016H01
053900001101
054260001101
055760004101
050500002101
050230001101
056400003F01
056680001101
397140026H01
393280108F01
393280108F01
397140004H01
397140024H01
311000001F01
264280006H01
264280015H01
264280062H01
264280025H01
264280010H01
264280015H01
142120009F01
Number of
24-hr values
104
113
95
100
105
52
109
61
72
150
50
193
45
204
46
60
57
57
339
58
61
109
58
61
53
54
58
77
5-8
-------
KEY:
O POPULATION CENTROID
TSP MONITOR
an^ c T + A TCD c?Jtr°lds of exposure districts in Chicago study area
and selected TSP monitoring sites. Numbers in parentheses indicate
districts represented by monitors.
5-9
-------
7
San Bernardino
A(5)
0 ,-
AnaheimK
POPULATION CENTROID
TSP MONITOR '
Pacific
Ocean
Figure 5-4. Population centroids of exposure districts in LA-San Bernardino study area
and selected TSP monitoring sites. Numbers in parentheses indicate
districts represented by monitors.
-------
en
I
KEY:
O POPULATION CENTROID
A TSP MONITOR
Figure 5-5. Population centroids of exposure districts in Philadelphia study area
and selected TSP monitoring sites. Numbers in parentheses indicate
districts represented by monitors.
-------
Bellafontaine
Neighbors
KEY:
O POPULATION CENTROID
A TSP MONITOR
Granite City
Lemay
Figure 5-6. Population centroids of exposure districts in St. Louis
study area and selected TSP monitoring sites. Numbers in
parentheses indicate districts
represented by monitors.
5-12
-------
The model can be stated in matrix form as
Y = A3 + e (5-9)
where Y is a vector random variable of size n, 3 is the unknown
•>*
coefficient vector of size 5, e is a nonobservable random vector
of size n, and A is a known matrix of order n x 5. The rows of A
are
11100 for winter weekdays
10100 for winter weekends
11010 for spring weekdays
10010 for spring weekends
11001 for summer weekdays
10001 for summer weekends
11000 for fall weekdays
10000 for fall weekends .
The expression
3 = (A^)"1 ATy (5-10)
will produce an ordinary least squares (OLS) estimate of 3. The
vector y has n elements corresponding to the logarithms of the n
observed values in the data set.
The minimum variance unbiased estimate (MVUE) of the expected
/^
value of y. is denoted by y, and is given by
" T T> _1 rp
uk = a kCA A) A v ' (5-11)
2
where av is the kth row in matrix A. The MVUE of a is
*s* jt —
a2 = ijCi-Hj (5-12)
where I is an n x n identity matrix and
H - A(ATA)-1 AT. (5-13)
The variance of y, can be expressed as
K.
var y = a. T(ATA)~1 a, a2. (5-14)
5-13
-------
To simulate a value of xfc, we first simulated a value of yfc
by adding a normal random component from a N(0,cr) population to
>N.
the estimate p^. We then simulated the corresponding x, value by
exponentiating the y, value, i.e.,
S\ s^ /N
= exp (u + zcr) (5-15)
where z was the value of a pseudo-random event from a standard nor-
mal population. For example, if a summer weekday value was missing,
it was filled in using the relationship
/N >\ X\ /\ xx
x(weekday, summer) = exp (a + 3j_ + &4 + za) . (5-16)
Figures 5-7 and 5-8 show the recorded TSP data for 1978 at site
055760004101 in the LA-San Bernardino study area and the same
data set after missing values were filled in.
Combining known values with values estimated by the above pro-
cedure yields an "augmented" data set containing 365 values, one
for each day of the year. However, NEM requires an air quality
value for each hour of the year. Since no information was available
on hourly variations in TSP levels, we investigated the possibility
of determining a typical diurnal pattern of hourly values for a
TSP surrogate. Hourly TSP values for a given day would be simu-
lated by multiplying the 24 hourly values in the typical diurnal
pattern by a constant such that the resulting set of 24 values
had a mean equal to the 24-h TSP value for that day in the aug-
mented data set. The use of such a procedure required (1) a TSP
surrogate with 1-h data, (2) a high correlation between surrogate
data and TSP data, and (3) identification of a "typical" diurnal
pattern in the surrogate data.
These three requirements could not be satisfied simultaneously.
The most promising TSP surrogate was coefficient of haze (COH) .
There were at least eight sites in each of the four study areas
which reported 1-h on 2-h COH data. To determine the degree of
correlation between COH and TSP, we computed 24-h COH values from
the 1-h values reported for days that also had 24-h TSP values.
Typical of our results is the COH/TSP scatter plot for site
142120009F01 in East St. Louis (see Figure 5-9) . The equation of
5-14
-------
a
o
I
M
Ui
a
a.
a
o
*
a
w.
m
DO
CM
>~
H-
•-X3
_P
or
n
ICO
o
•
o.
o
JRN
FEB
HRR
RPR NflY
JUN
JUL
DUG
SEP
OCT
NOY
DEC
Figure 5-7. Computer plot of 24-hour average TSP data (yg/m3) reported for 1978 by monitoring
site 055760004101.
-------
o
o
en
I
or—
JUL
RUG
SEP
OCT
NOV
DEC
Figure 5-8. Computer plot of 24-hour-average TSP data (yg/m3) for monitoring site
055760004101 after simulation of missing values.
-------
250
200
E
cn
oT 150
CO
100
•*
CM
50
T
• •
0.5 1.0 1.5
24-HOUR AVERAGE.COM
2.0
Figure 5-9. Scatter plot of 24-hour COH and TSP values recorded during
1977 at site 14212009F01 in East St. Louis.
5-17
-------
2
the regression line is TSP = 84.7 + (11.7)(COH); R is only
0.0087. Regression analysis of COH/TSP data at site 264280061H01
2
(see Figure 5-10) in St. Louis yielded an R value of 0.55—
higher than the East St. Louis site but still poor correlation.
This lack of correlation between 24-h COH and 24-h TSP suggests
an even poorer correlation between 1-h COH and 1-h TSP would be
found if 1-h TSP data were available.
In addition to the poor COH/TSP correlation, we found that
COH data for a given study area could not be realistically char-
acterized by a small number of typical diurnal patterns. Inspec-
tion of selected COH data sets revealed that COH diurnal patterns
varied significantly from site to site on the same day and from
day to day at the same site. The lack of a strong 24-h pattern
was confirmed by Fourier analysis. This technique decomposes a
data set with 8760 values into 4380 sinusoidal functions with
periods of T = 8760/j, j = 1/2,...,4380. Fourier analysis of COH
data from site 143760005F01 indicated that the sinusoidal function
with a period of 24 hours was less important than 9 other functions
in characterizing the data. This function is usually ranked
first or second if a pollutant has a strong diurnal pattern.3
Based on these results, we rejected COH as a possible TSP surro-
gate.
We also investigated using 1-h beta-scattering (3) data as a
TSP surrogate, although St. Louis was the only study area which
reported any 3 data. Regression analysis of 24-h 3 and TSP data
2
at site 264280061H01 in St. Louis yielded an R value of only
0.41. Figure 5-11 is a scatter plot of the data. Because of
poor 3/TSP correlation and lack of 3 data for three of the four
study areas, we rejected 3 as a possible TSP surrogate.
Most of the other potential TSP surrogates had averaging
times exceeding 1 hour or were not reported in sufficient quantity
to allow determination of typical diurnal patterns in the four
study areas. Consequently, we were forced to make the somewhat
simplistic assumption that particulate concentration during each
hour of a day was equal to the 24-hour average TSP value for that
5-18
-------
250
200
en
150
C5
100
I
^1-
csj
50
.. . • ••'
I
I
25 50 75
24-HOUR AVERAGE COM
100
Figure 5-10. Scatter plot of 24-hour COH and TSP values recorded during
1977 at site 264280061H01 in St. Louis.
5-19
-------
250
200
150
Q_
oo
cs
-------
day. The resulting series of 8760 hourly values is termed an
"expanded" data set.
Because an expanded data set contains no diurnal variations
in 1-h values, we expect its peak 1-h value to underestimate the
"true" peak 1-h value. In the absence of actual 1-h TSP data, we
cannot determine this bias directly. However, we can assume that
COS data would have a similar bias. Table 5-2 lists peak 1-h and
24-h values for 1977 for selected COH data sets. Also listed is
the percentage of 1-h values greater than the peak 24-h value.
TABLE 5-2. PEAK 1-H AND 24-H COH VALUES RECORDED IN 1977
AT SELECTED SITES IN CHICAGO AND PHILADELPHIA
Study area
Chicago
Philadelphia
Site
141240001G01
142120009F01
143760005F01
397140004H01
Number of
1-h values
6234
7124
7812
8244
Peak 1-h
value
2.30
5.91
3.37
4.90
Peak 24-h
value
1.41
1.65
1.04
1.60
Percentage of 1-h
values above peak
24-h value
2.9
5.7
1.3
1.6
Peak 1-h COH values are two to three times higher than peak
24-h values. However, the peak 24-h value is exceeded by less
than 3 percent of the 1-h values in three of the four cases. In
addition, the peak 1-h values at sites 143760005F01 and 397140004H01
appear to be abnormally high when compared to other values recorded
on the same days.
These results suggest that 3 to 6 percent of 1-h COH values
will be greater than the peak 24-h value. If typical 1-h TSP
data have a similar distribution, then our procedure for expanding
TSP data sets should yield a similar bias in the simulated 1-h TSP
values.
5.3 CONVERSION OF 1-HOUR TSP VALUES TO 1-HOUR IP1Q VALUES
Analyses5 by SASD indicated that IP1Q concentrations could
be estimated from TSP concentrations by the relationship
5-21
-------
IP1(J = (0.55) (TSP) . (5-15)
This relationship was used to convert expanded-TSP data sets into
expanded IP-,Q data sets. Consequently, the bias described above
is incorporated into the expanded IP-iQ data sets.
5 .4 REFERENCES
1. G.E. Anderson, C.S. Liu, H.Y. Holman, and J.P. Killus,
Human Exposure to Atmospheric Concentrations of Selected
Chemicals, Attachment B, Appendix F, "File Format and Util-
ity Files", SAI No. EF-156R, Systems Applications, Inc.,
San Rafael, California, March 5, 1980.
2. Y. Horie and J. Morrison, Technical Memorandum on Spatial
Interpolation of Air Quality Data, Technology Service Corp-
oration/ Santa Monica, California, December 21, 1977.
3» T. Johnson and L. Wijnberg, "Time series analysis of hourly
air quality data," Paper No. 81-33.5, Presented at the 74th
Annual Meeting of the Air Pollution Control Association,
Philadelphia, Pennsylvania, June 1981.
4. T. Johnson and R. Paul, The NAAQS Exposure Model (NEM) and
Its Application to Nitrogen Dioxide, (draft), PEDCo Environ-
mental, Inc., Durham, North Carolina, May 1981.
5. Memorandum from Thompson Pace, Monitoring and Data Analysis
Division, to Henry Thomas, Strategies and Air Standards
Division, U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina, July 6, 1981.
5-22
-------
SECTION 6
SIMULATION OF OUTDOOR AIR QUALITY EXPECTED
UNDER ALTERNATIVE PARTICULATE STANDARDS
The expanded data sets described in Section 5.2 were assumed
to represent the current status of outdoor air quality in each dis-
trict. To represent outdoor air quality expected under the current
NAAQS for TSP and under proposed TSP and IP10 standards, these
data sets were adjusted using a modified form of the EPA rollback
model.
6.1 THE ROLLBACK MODEL
Each expanded data set can be represented by a series of 8760
hourly values, i.e.,
xl' X2' ' xt' ' X8760'
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 xfc < 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 district in the study area into
compliance.
Air quality in each district was characterized by air qual-
ity indicators (AQI's) which varied according to the form of the
air quality standard. We assumed the most polluted district to
be the one with the largest AQI with respect to the standard being
considered.
To simulate the air quality expected in each district under
a standard, we created an adjusted data set
•* ••" •* ^
' "• ' xt ' " * ' X8760
6-1
-------
where
= pyt + xb (6-1)
and p is a rollback factor. Consistent with the assumptions
above, values of p were calculated according to the formulas
xs " xb
P = XS _ J° if y,. > 0 (6-2)
xmax
and
p = 1 if yt < 0 (6-3)
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 district. The rollback model assumes rea-
sonable estimates of x and x, are available; Sections 6.2 and
6.3 describe how these estimates were developed.
6.2 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 PM pop-
ulation exposure analyses, four types of parameters were consid-
ered for proposed standards: the annual arithmetic mean, the
annual geometric mean, the second highest 24-hour average value,
and the 24-hour average value expected to be exceeded once. Rea-
sonable estimates of the annual arithmetic and geometric means
are the arithmetic mean and the geometric mean of each augmented
data set. These values are easily calculated and are robust (i.e.,
relatively unaffected by anomalous data). Similarly, a reasonable
estimate of the second highest 24-hour average during a year is
the second highest 24-hour average observed in the corresponding
augmented data set. The 24-hour average value expected to be ex-
ceeded once a year can be estimated by fitting a cumulative dis-
tribution [F(x)] to the augmented data and then calculating the
value b such that
n
6-2
-------
F(n) - 1 - <6-4)
for n = 365. In statistical theory, h>n is known as the character-
istic largest value.1
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
F(x) = 1 - exp [-(|)k] (6-5)
where 6 is the scale parameter and k is the shape parameter. The
lognormal distribution is defined as
F(x) =^T /_„ exp (-t2/2) dt (6-6)
where
w = x - u . (6_?)
2
and In x is distributed normally with mean u and variance a .
From Equations (6-4) and (6-5), the characteristic largest value
of the Weibull distribution can be estimated as
/\
bn = fi (In n)1/k (6-8)
if good estimates of & and k are available. Similarly, the charac-
teristic largest value of the lognormal distribution can be esti-
mated as
^\ S\ S\
bn = exp (u + 0z) (6-9)
if good estimates of u and a are available. The value of z is
determined from the normal distribution such that the area under
the standard normal curve from z to °° is 1/n. Values of z for
common data averaging times are listed below.
Averaging time Yearly values(n) z
1 hour 8760 3.6854
3 hours 2920 3.3955
8 hours 1095 3.1171
24 hours 365 2.7775
6-3
-------
The results of fitting distributions to a large number of ambi-
ent air quality data sets suggest that the characteristic highest
value can be better estimated if the upper-tail of the data is em-
phasized 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.
6.2.1 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 rela-
tionship of the form y = az + b. Equations (6-5) and (6-6) can be
rewritten in linear form using the following identities where x is
the mth ranked value.
Distribution 2 £ £ *-L
lognormal In x a z y
Weibull In xm £ In[In( "7 m)] In 6
m K, ri"> i™*m
These identities follow Gumbel's recommendation3 that F (x ) = —rr-
m n+i
when fitting distributions to empirical data. Values of z for
the lognormal distribution are determined such that the area under
the standard normal curve from -» to z_ is m/(n+l).
ni; n
A linear regression analysis of data transformed by these
identities yields a regression line with an equation in the form
of y = az + b. Parameters of the corresponding distribution can
be determined from the values of a and b using the following
equations:
Weibull distribution
6 = exp b (6-10)
k = i (6-11)
a
Lognormal distribution
y = b (6-12)
a = a (6-13)
6-4
-------
After the Weibull and lognormal distributions were fit to
the data, a means was required to determine which curve better
characterized the data. A readily available measure of the good-
ness of fit of each distribution is the coefficient of determi-
2
nation (R ), which is easily determined as part of the linear
2
regression analysis. The closer the R value is to unity, the
better the distribution fits the data.
We also calculated the mean percentage deviation (MPD) be-
tween each distribution and the augmented data in the region of
the fit. Distributions with MPD's less than 2.5 percent were
considered acceptable for determining characteristic largest val-
ues.
Weibull and lognormal distributions were first fit to the
upper 50 percent of the values in each augmented data set. Eval-
uation of the results indicated that, while the closer fitting
distribution had an MPD less than 2.5 percent in all cases, it
often did not yield a close fit to the five largest values. Re-
peating the analysis using the upper 20 percent of the values in
each augmented data set produced an MPD of less than 2.5 percent
in all cases and superior fits to the five largest values.
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 values for 24-hour PM data is
described below.
(1) Each augmented data set was ranked from lowest to high-
est value.
(2) The upper 20 percent of the augmented data values were
fit by Weibull and lognormal distributions using the
least squares method described above.
2
(3) The R values of the two fits were compared and the
parameters of the better fitting distribution (i.e.,
the one with the larger R2 value) were used to deter-
mine the characteristic largest value.
Table 6-1 lists characteristic largest values developed using this
procedure; other AQI's of interest are also listed.
6-5
-------
TABLE 6-1. AIR QUALITY INDICATORS (ug/m3)
FOR TSP DATA
Study area
Chicago
Los Angeles
Philadelphia
St. Louis
District
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
1
2
3
4
5
6
1
2
3
4
5
6
7
Arithmetic
mean
92
69
93
76
74
76
74
95
89
123
106
152
105
151
130
54
65
65
74
54
54
87
75
88
76
75
75
96
Geometric
mean
83
61
83
70
60
60
64
87
84
114
96
126
98
137
87
47
60
60
69
48
47
82
70
79
70
69
70
85
Second
high
237
198
275
191
254
317
215
217
219
305
271
453
249
384
415
160
157
157
190
142
160
200
180
238
190
201
180
292
Characteristic
largest value*
251
227
280
200
277
358
246
241
234
338
293
473
266
407
464
172
164
164
187
155
172
218
185
260
205
229
185
315
*Estimated by least squares -fit to upper 20% of data,
6-6
-------
6.2.2 Fitting Distributions by the Method of Maximum Likelihood
In an earlier analysis'* of population exposure to N02, the
least squares method described in Section 6.2.1 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, Louis Wijnberg of PEDCo Environmental
(extending the work of Cohen5'6) 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
X^f ^f 9f
T I J»« , . • . , * , ... A
±z m n
where x indicates the mth ranked value. We are interested in ob-
taining maximum likelihood estimates (MLE's) of the parameters 9,
and 92 of a two-parameter distribution F(x;9lf92) fitting the
nf = 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 , then
c c
n1 f n
L = t* '% i n f(x :9,,9.,
(n-n,;) ! L___ m 1' ^
^ m^**^«
is the likelihood function of interest. MLE's of 9, and 9~ are
determined by simultaneously solving the likelihood equations
L) = 0 (6-15)
and
L) = 0. (6-16)
In the case of the Weibull distribution, the likelihood equations
are
6-7
-------
n
n
i i
1, 1
v + 5- sr- ~ n~ sr~
K nf m=c xc nf m=c xc
(6-17)
and
n-n.
^ , y / .
^ nf m=c Cxc>
_
-1)
= 0,
(6-18)
where
x
(6-19)
When fitting the lognormal distribution, the likelihood equations
are:
Cy - u) -
(n-nf)0<()
and
—
+ (y -
where
y =
n
n
(h-n
_ 2
= 0
(6-20)
(6-21)
(6-22)
(6-23)
- ln
yc -
(6-24)
(6-25)
$ 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 6.2.1 to make initial estimates of
the parameters and then improving these estimates using an intera-
tive process (the Secant method) until the "true" solution was
6-8
-------
reached. Time constraints on the PM exposure analysis did not
permit the evaluation of goodness-of-fit statistics for determining
which distribution better characterized data.
6.2.3 Comparison of Least Squares and Maximum Likelihood Methods
Tables 6-2 and 6-3 list characteristic high values calculated
using Equations (6-8) and (6-9) and parameter values estimated
using the least squares and maximum likelihood methods described
above. In Table 6-2, Weibull and lognormal distributions were
fit to the upper 50 percent of each augmented data set (i.e.,
nf = 183); in Table 6-3, these distributions were fit to the upper
•
20 percent of each augmented data set (i.e., n^ = 73). In general,
the least squares and maximum likelihood estimates are in close
agreement.
Characteristic largest values determined by least squares
were used in the PM exposure analyses described in this report
because goodness-of-fit statistics for the maximum likelihood
procedure were unavailable. PEDCo has since developed methods of
»
determining goodness-of-fit based on the recommendations of
Stephens7 and Green and Hegazy.8 In future exposure analyses,
we will attempt to use maximum likelihood exclusively. MLE's of
distribution parameters have several desirable statistical proper-
ties. In particular, MLE's have minimum variance and they asymp-
totically 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. Parameter estimates
developed by the least squares method have none of these properties.
6.3 BACKGROUND CONCENTRATIONS
The modified rollback model (Section 6.1) assumes a constant
background level (x, ) for each hour of a year. Consistent with
EPA's definition of background, we defined x, as the annual average
pollutant level which would be experienced by a study area if
there were no local man-made sources. Ideally, background concen-
trations should include only pollutant levels arising from natural
6-9
-------
TABLE 6-2. COMPARISON OF CHARACTERISTIC LARGEST VALUES
ESTIMATED FOR TSP DATA (n=365>-B¥' LEAS£=§QSARES
AND MAXIMUM-LIKELIHOOD FITS TO UPPER 50 PERCENT
Study area
Chicago
Los Angeles
Philadelphia
St. Louis
District
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
1
2
3
4
5
6
1
2
3
4
5
6
7
Characteristic largest value (bn), ug/m3
Weibull
LS
234
201
266
1.92
271
320
229
231
204
286
285
447
235
359
485
164
160
160
164
150
164
208
181
252
193
192
181
300
ML
233
202
264
191
268
317
229
231
206
292
283
445
240
362
473
164
159
159
168
149
164
208
181
251
193
197
181
299
loqnormal
LS
266
236
307
219
328
398
271
262
231
325
328
516
265
404
578
194
182
182
184
174
194
237
206
293
222
223
206
356
ML
263
232
304
216
323
390
267
259
228
321
324
510
262
400
569
191
180
180
182
172
191
234
204
289
219
220
204
350
6-10
-------
TABLE 6-3. COMPARISON OF CHARACTERISTIC LARGEST VALUES
ESTIMATED FOR TSP DATA (n=365) BY LEAST SQUARES
AND MAXIMUM LIKELIHOOD FITS TO UPPER 20 PERCENT
Study area
Chicago
Los Angeles
Philadelphia
St. Louis
District
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
1
2
3
4
5
6
1
2
3
4
5
6
7
. ~- *• 3
Characteristic largest value (b ), ug/m
of data n
Wei bull
LS
239
214
265
200
277
330
246
241
221
319
293
445
266
386
464
172
164
164
177
155
172
218
185
260
205
215
185
315
ML
235
210
263
196
271
321
239
237
218
314
287
445
260
381
452
168
161
161
177
152
168
214
182
255
200
211
182
307
lognormal
LS
251
227
280
210
298
358
262
254
234
338
308
473
281
407
490
183
171
171
187
164
183
229
194
276
215
229
194
335
ML
246
222
275
206
290
346
256
249
228
330
303
464
275
398
480
178
168
168
183
161
178
225
190
270
212
224
190
327-
6-11
-------
emissions and from transport of man-made emissions. However, since
local emission sources cannot be removed, background levels must
be estimated indirectly.
Whenever possible, estimates of background levels were based
on data from nonurban rural sites upwind from the urban area. In
some cases, suburban residential sites were considered due to the
lack of the required data at nonurban or background surveillance
stations. In all cases, monitors were screened to identify those
which would not be expected to be impacted by sources within the
urban area. If no monitors meeting this criterion were identified,
average measured concentrations from "regional sites" were used
to estimate background concentration. In cases where background
values had been estimated by previous studies, these values were
used in lieu of the values obtained by the methodology detailed
above.
Background TSP concentrations were determined for all four
study areas included in this analysis. A background surveillance
site in Will County, Illinois indicated a TSP background level of
42 yg/m for the Chicago study area.9 The Air Quality Management
Plan for the South Coast Air Basin10 includes an estimation of 29
yg/m for a TSP background level in the Los Angeles area. This
estimation is for concentrations resulting from geogenic sources
only. Contributions due to transport may result in additional
background pollutant concentrations although supporting data are
not currently available. No nonurban background sites are pres-
ently located in the surrounding Los Angeles area.
In Philadelphia, Aerovironment collected 24-hour hi-vol data
at two nonurban sites—Downington, Pennsylvania and Dover, Delaware-
for an oxidant modeling study conducted by EPA.11 These data
suggest a particulate background level of 34 yg/m .
In a report on the analysis of the St. Louis RAMS ambient
particulate data,12 Technology Service Corporation estimated the
St. Louis TSP background level as 35 yg/m .
6-12
-------
6.4 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, New Orleans, Louisiana, March 1979.
3. Op. cit., Gumbel, p. 34.
4. T. Johnson and R. Paul, The NAAQS Exposure Model (NEM) and
Its Application to N02 Data, PEDCo Environmental, Inc.,
Durham, North Carolina, May 1981.
5. A.C. Cohen, Jr., "Simplified estimators for the normal distri-
bution when samples are singly censored or truncated," Tech-
nometries, Vol. 1, No. 3, August 1959.
6. A.C. Cohen, Jr., "Maximum likelihood estimation in the Weibull
distribution based on complete and on censored samples," Tech-
nometries, Vol. 7, No. 4, November 1965.
7. M.A. Stephens, "EOF statistics for goodness of fit and some
comparisons," Journal of the American Statistical Association,
Vol. 69, No. 347, September 1974.
8. 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.
9. National Aerometric Data Bank, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina.
10. South Coast Air Quality Management District, (draft), Air
Quality Management Plant, October 1978.
11. Personal Communication with Doug Allard, Aerovironment, Pasa-
dena, California, August 1980.
12. J. Trijonis, J. Eldon, J. Gins, and G. Berglund, Analysis of_
the St. Louis RAMS Ambient Particulate Data, (final report),
Vol. 12, EPA-450/4-80-006a, February 1980.
6-13
-------
SECTION 7
SIMULATION OF PARTICULATE LEVELS IN THE MICROENVIRONMENTS
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, and outdoor locations. The exposure model further assumes
that air quality in each microenvironment (x ) of a given district
can be estimated by the expression
xm = am + bmxc (7~I}
where a is the pollutant concentration generated in the microen-
vironment, x is the outdoor air quality estimated for the cen-
c
troid of the district, and b is a coefficient. Estimates of a
* * m m
and b (denoted a and b ) were developed using data reported in
the studies listed in Table 7-1. Unless otherwise indicated, we
assumed that a and b did not vary with city, exposure district,
hour of the day, season, or pollutant level.
A preliminary review by T. Johnson1 of research concerned with
indoor PM sources indicated that cigarette smoke was a signifi-
cant indoor PM source. Typical of the studies which support this
hypothesis is an investigation by Spengler et al.2 in which 12-h
mass respirable particulate (MRP) data were collected from per-
sonal samplers carried by 42 adults and 4 children in Topeka,
Kansas. These monitors, developed jointly by Harvard and
Electric Power Research Institute, collect no particles greater
than 10 ym in diameter, 25 percent of 5 urn particles, 50 percent
of 3.5 urn particles, and 90 percent of particles less than 2 ym
in diameter. Of the 24 households studied, 17 were located on
7-1
-------
TABLE 7-1. STUDIES CONSIDERED IN DEVELOPING
PM MICROENVIRONMENT FACTORS
Microenvironment
Unspecified
Indoors: work or school
Indoors: home or other
Transportation vehicle
Roadside and outdoor
Study
2
Spengler et al .
3
Repace and Lowrey
4
Penkala and OTiveira
5
Thompson et al .
Yocum et al .
Repace et al .
Q
Spengler "et al .
Repace et al .
5
Thompson et al .
Q
Ju and Spengler
3
Repace and Lowrey
Yocum et al .
Moschandreas et al .
Repace et al .
Pace et al.12
Pace13
Pace14
7-2
-------
residential streets with through traffic, four were on dead-end
residential streets, two were on well-used commercial streets,
and one was on a major traffic artery. Nine of the homes were
equipped with gas cooking facilities and 15 had electric facilities,
Only three homes had smokers. Most of the subjects were employed
in office jobs. Four reported routine work exposures to dust or
smoke. Personal sampling data were collected every Tuesday,
Thursday, and Saturday from May 17 through June 28, 1979. Each
home was equipped with stationary indoor and outdoor monitors.
In general, the concentrations measured were close to the detec-
tion limits of the monitors.
The mean for all indoor samples was 23.9 ug/m . The mean
of outdoor samples taken at the homes was 12.7 ug/m . The grand
mean of all the personal samples was 29.6 ug/m . Spengler and
Tosteson concluded that somewhere in their subjects daily activi-
ties they were being exposed to respirable particulate matter at
concentrations higher than what was measured by the stationary
indoor and outdoor monitors. The most likely source of this
additional exposure was cigarette smoke. Time spent in transpor-
tation vehicles was not significant.
The contribution of smoking is important to personal expo-
sures... Exposure to cigarette smoke doubles personal ex-
posures from 20 ug/m3 to 40 ug/m3. Smoking exposed personal
samples had the same overall mean concentration in air con-
ditioned as in nonair conditioned locations. Under the
assumption that personal exposures during automobile travel
may exceed those at other outdoor locations or some indoor
locations, personal samples were grouped by hours of re-
ported transit time. There was only a modest difference
among samples representing less than one hour transit, 1 to
2 hours of transit, and greater than 2 hours. The mean per-
sonal concentrations were approximately 29 ug/m3, 30 ug/m3,
and 31 ug/m3, when not controlled for smoking.2
Although the personal monitoring results indicate the signifi-
cance of cigarette smoke, they do not provide a means of esti-
mating am or bm for a particular microenvironment since the mea-
sured concentrations represent exposure accumulated over 12 hours
in various microenvironments.
7-3
-------
Based on this study and others, SASD decided to estimate
population exposure with and without smoking as an emission
source. As a consequence, different am values were developed for
microenvironments with and without smokers. The smoking a
m
values developed for TSP and IP10 are identical because we assumed
both particulate measurements include 100 percent of the particles
in cagarette smoke.
Occupational exposures to TSP and ^P-in specific to certain
industries such as mining were not considered in estimating
population exposure. Reasonable estimates of occupational ex-
posure would require individual additive factors for each in-
dustry and the creation of industry-specific cohorts. Because
the development of such data bases would be difficult and time
consuming, SASD decided to omit occupational exposures other than
cigarette smoke from the exposure analyses.
The studies listed in Table 7-1 suggested the following
basic relationships between PM measurements at fixed sites and PM
levels in the five microenvironments.
(1) PM concentration is attenuated as air enters buildings
and transportation vehicles.
(2) Small particulates are more likely to penetrate build-
ings and transportation vehicles than large particu-
lates.
(3) PM concentration in the vicinity of automobile traffic
is underestimated by PM measured at a fixed site away
from traffic (i.e., b > 1).
m
(4) PM concentration in the breathing zone is underestimated
by a monitor with a probe above the breathing zone.
Estimates of b developed in the following sections are generally
consistent with these relationships. Because of the scarcity of
data relating indoor and outdoor inhalable particulate concentra-
tions, bm estimates for IP1Q are considered to be less accurate
than b estimates for TSP.
m
7-4
-------
7.1 WORK-SCHOOL MICROENVIRONMENT
As discussed above, smoking is assumed to be the sole source
of indoor generated PM. Two methods are available for estimating
a for indoor microenvironments with smokers: (1) a formula de-
veloped by Repace and Lowrey3 and (2) the application of the equation
/ *r *^ \
3 — V TT f\ V I / ™ ^ )
m indoor m outdoor
to simultaneous data on inside and outside PM levels, given an
appropriate value of b .
m
Repace and Lowrey cite studies which conclude that 11 percent
of adults will be smoking at any given time under most social
conditions. Furthermore, the average number of habitual smokers
in a room is three times the number of active smokers. These
statistics and the results of smoke dispersion studies are the
basis of three equivalent relationships:
am = 25.6(p) (7-3)
m \C/
am = 76.8^) (7-4)
am=23o(|). (7-5)
The parameters are defined as follows: a is the smoker-generated
respirable suspended particulate (RSP) level under equilibrium
conditions, C is the number of air changes per hour, P is the
number of occupants per 1000 square feet, H is the number of
habitual smokers per 1000 square feet, and A is the number of
active smokers per 1000 square feet. All of the expressions
assume rooms with 10-foot ceilings. RSP measurements were made
by a piezo balance which collected 100 percent of particles less
than 3 vtm in diameter. Cigarette smoke particles are generally
less than 1 um in diameter.
The main difficulty in using these relationships is deter-
mining appropriate values for the various parameters. Repace and
Lowrey estimated a practical range for C to be from 1 to 12 air
7-5
-------
changes per hour. They have observed 1 <_ C £ 2 in unventilated
dwellings and 10 £ C £ 12 in well ventilated commercial buildings.
Repace and Lowrey state that the recommended occupancy density
for general office space is 10 persons per 1000 square feet.
Assuming a ventilation rate of 10 air changes per hour, we can
calculate an average smoker-generated RSP level of
am = (25.6)(i£) = 25.6 yg/m3.
/>
The estimates of a developed using the equations of Repace
and Lowrey are supported by other data. Penkala and Oliveira **
studied the generation of suspended particulate matter (SPM) by
a smoking machine in an experimental chamber. Particulate mea-
surements were made using a smoke photometer with unspecified
collection efficiency. Using the results of their study, Penkala
and Oliveira developed the following estimates of average
generated SPM in an office of 400 ft occupied by a smoker who
consumes one cigarette every 48 minutes (a typical smoking rate).
air changes/hour mean SPM, yg/m
0 1980
1 960
2.1 620
7.5 230
Assuming a 10 foot ceiling, one habitual smoker per 400 ft is
2
equivalent to 25 habitual smokers per 1000 ft . Using Equation
~ 3
(7-4), we would estimate a = 256 yg/m for H = 25 and C = 7.5.
The estimates of Penkala and Oliveira are in good agreement with
those of Repace and Lowrey.
Two studies—Thompson5 and Yocum6—give useful PM data for
estimating am by the second method [Equation (7-2)]. Thompson et al.
monitored indoor and outdoor TSP levels in 16 structures in Riverside,
California in the fall of 1971. Of these, 13 can be categorized as
work or school microenvironments. Yocum et al. recorded 12-h TSP
concentrations inside and outside public buildings, office buildings,
and private homes in Hartford, Connecticut from summer 1969 through
7-6
-------
winter 1970. Four building can be categorized as work microen-
vironments. Table 7-2 lists mean indoor/outdoor ratios for each
type of structure and the range of ratios.
TABLE 7-2. INDOOR/OUTDOOR TSP RATIOS FOR STRUCTURES
CORRESPONDING TO WORK-SCHOOL MICROENVIRONMENT
Type of structure
Medical care (4)
School (7)
Store (2)
Library (1)
City hall (1)
Office building (2)
Study
Thompson
Thompson
Thompson
Yocum
Yocum
Yocum
Minimum
ratio
0.08
0.36
0.44
0.46
Mean
ratio
0.19
1.15
0.46
0.38
0.48
0.50
Maximum
ratio
0.29
3.82
0.47
0.54
The numbers in parentheses indicate the number of structures used
in calculating each mean ratio. Note that most of these ratios
are less than unity- Although no information is given concerning
smoking within any of the structures, we can assume that there
were low levels of cigarette smoke in the buildings with small
ratios. Ratios for the office buildings and the city hall range
from 0.46 to 0.54 with a mean of 0.49. Ratios for the stores and
the library range from 0.38 to 0.47 with a mean of 0.43. The
large mean ratio for schools (1.15) probably reflects the dusty
California environment of Thompson's study rather than cigarette
smoke. If we disregard the three schools which had no primary
air filtration, the mean school ratio would be 0.52.
^
These results suggested that 0.33 < bm < 1.00 would be a
reasonable range for the work-school microenvironment given
primary filtration and no smoking. We used bm = 0.50 as our
best estimate. Note that these values are based on TSP data; we
would expect b for IP,,, to equal or exceed these estimates.
c m xu
Few data are available on PM generated by smoking in offices
or schools. Figure 2 in the study by Repace et al.7 records J.L.
Repace's exposure to RSP from 9 p.m., June 13, to 9 p.m., June 14,
1979. From 10 a.m. until 11 a.m., cigarette smoke from an adjacent
office entered Repace's office. The RSP concentration in the office
3
during this time averaged about 50 yg/m . Outside the average
7-7
-------
concentration was about 30 yg/m . Assuming that only the RSP
fraction of outside TSP entered the building and that bm = 0.50,
then
am = 50 - (0.5)(30)
am = 35 yg/m3
by Equation (7-2). If bm = 1.0, am = 20 yg/m3. Figure 4 in the
same study shows hourly RSP concentrations for a typical average
workday based on 5 days of data recorded by Wallace. The average
outdoor concentration at noon was 44 yg/m . The average office
concentration during work hours was 74 yg/m . Assuming that the
elevated level of RSP in the office was caused by smoking and
that b =0.5,
m
a * 74 - "(0.5) (44)
m
am = 52 yg/m .
If no attenuation of outside TSP occurred (i.e., if b = 1.0),
" 3
then a would be 30 yg/m . These values are consistent with
values calculated by the Repace-Lowrey formula using reasonable
parameter estimates.
Although there are no studies which permit the estimation of
a related to smoking for schools, we can reasonably assume that
it is less than the value of a for the office environment. Since
111 S\ M
the Repace-Lowrey formula yields am = 26.5 yg/m for the office
environment, 20 yg/m may be a reasonable value for the combined
work-school microenvironment. However, the scarcity of data and
its variability suggest that the true a value may be anywhere
3 m
from 0 to 50 yg/m . As previously discussed, we assumed *i_>?for
HI
the 'sanievfor TSP, and IP, n.
7.2 HOME-OTHER MICROENVIRONMENT
Table 7-1 lists seven studies which consider indoor particu-
late levels in the home and other indoor areas not characterized
as work or school. To estimate bm for the home microenvironment,
data are needed on Indoor and outdoor particulate levels for
homes with no smokers or else where the smoking contribution has
7-8
-------
been quantified. Unfortunately, only the study by Spengler et al.8
fully meets-these criteria. Spengler's group collected:MRP data
inside and outside of homes in six U.S. cities. Linear regression
was performed on these data after they were stratified according
to whether or not smokers were present. The results of the re-
gression analyses are given below.
2
Group n intercept slope _r
Nonsmoking 42 11 0.78
Smoking
34
32
0.85
0.33
0.10
m
The intercept and the slope provide reasonable estimates of a.
and b, for the grouped data.
m 3 c-
There is fairly good agreement between the slope values of
the two regression lines. The fact that the intercept for the
nonsmoking households is not equal to zero suggests that there are
other significant indoor sources of PM. Studies by Repace et al.7
show that elevated levels of RSP coincide with meal preparation
and persist in residences for several hours afterwards. Thompson
et al.5 found that indoor areas with heavy foot traffic may experi-
ence TSP levels which equal or exceed outdoor levels.
Ju and Spengler9 studied room-to-room variations in levels
in four homes near Boston. Between 25 and 30 24-hour samples
were collected at each house from late November 1979 to early
February 1980. Table 7-3 summarizes the results of the study.
TABLE 7-3. MEAN LEVELS OF MASS RESPIRABLE PARTICULATES (yg/m3) IN FOUR HOMES
Residence
Watertown
Brighton
Waltham
Newton
Outdoor
11.5
10.9
12.5
10.3
Kitchen
17.5
19.4
25.6
16.8
Living
room
' 12.7
19.9
28.8
17.3
Dining
room
16.9
27.9
Bedroom
15.2
23.3
27.6
19.6
Third
floor
13.4
Indoor
average
15.2
20.9
27.. 5
17.9
The Waltham residence, the home with the highest average
indoor particulate concentration, had a woodstove and an occasional
pipe smoker. The mean ratio of indoor averaged concentration to
7-9
-------
outdoor concentration ranged from 1.6 to 2.3 with a four-home
average of 2.0. Since there are smokers in only one of the four
residences, there must be other sources of indoor particulates in
at least three of the homes.
Ju and Spengler found that room-to-room variations were small
but statistically significant. The living room mean approximates
the indoor mean to within 5 percent in all homes except Watertown.
Between-home differences in outdoor concentrations were not sta-
tistically significant; between-home differences in indoor con-
centrations were significant. Ju and Spengler suggest that rooms
with higher particulate levels tend to have lower ventilation rates.
If we assume b =0.80 for MRP as suggested by the above re-
m 2
gression analysis, indoor contributions would be 12 yg/m , 18
.yg/m , and 10 yg/m for the three nonsmoking residences. These
results are consistent with the intercept of 11 yg/m for non-
^ 2
smoking regression line. We ultimately used am = 12 yg/m for
nonsmoking homes during waking hours; our lower bound for a was
3
0 and our upper bound was 20 yg/m .
The studies by Thompson and Yocum described above provide
useful TSP data for hospitals, stores, private residences, a
library, and a city hall. Table 7-4 summarizes the mean indoor/
outdoor ratios for each type of structure.
TABLE 7-4. INDOOR-OUTDOOR TSP RATIOS FOR STRUCTURES CORRESPONDING TO HOME-OTHER
MICROENVIRONMENTS
Type of structure
Medical care (4)
Store (2)
Private home (1)
Library (1)
City hall (1)
Private home (2)
Study
Thompson
Thompson
Thompson
Yocum
Yocum
Yocum
Minimum
ratio
0.08
0.44
-
-
-
0.62
Mean
ratio
0.19
0.46
1.08
0.38
0.48
0.68
Maximum
ratio
0.29
0.47
-
-
-
0.75
The numbers in parentheses indicate the number of structures used
in calculating each mean ratio. Private home ratios range from
0.62 to 1.08 with a mean of 0.82. Ratios for the city hall, the
7-10
-------
stores, and the library range from 0.38 to 0.48 with a mean of
0.44. Note that most of these ratios are less than unity.
Although no information is given concerning smoking within any of
the structures, we assumed that there were low levels of cigarette
smoke in the buildings with ratios less than unity. The data on
smoke-free environments listed in Table 1 of the study by Repace
and Lowrey were not considered appropriate for estimating b ; the
table lists only short-term (< 1 hour) particulate measurements
and notes that food was being prepared near several of the monitors.
A combined TSP b value for home and other is somewhat
m
difficult to estimate since "other" can vary from the highly
filtered air of the typical hospital environment to a private
residence with no filtration. We ultimately decided to use 0.85
for the home b value and 0.44—the mean of the stores, the
m
library and the city hall—for the "other" b value. Since about
8 percent of people's time categorized as home-other is spent in
"other," we used 0.80 = (0.92) (0.85) + (0.08) (0.44) as our best
estimate for the combined home-other TSP b value. This value is
m
identical to the b value suggested by regression analysis for MRP.
Since no other data were available, we used 0.80 as our best esti-
mate of b for both TSP and IP,n; we considered 0.50 < b < 1.00
m 10 — m —
to be a reasonable range for b values.
m
Equation (7-4) can be used to develop initial estimates of
^\
a related to smoking if we have reasonable estimates of the
m
floor area, number of smokers, and ventilations rate in a typical
smoking household. Housing data indicate the average living unit
has 5.1 rooms.10 We assumed a floor area of 1300 ft for a house
of this size. In a sample of 69 homes, Spengleret al.8 -found 32 per-
cent 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 1300 ft ; consequently, H = 1.0. Table 7-5 lists air
exchange rates determined by Moschandreas et al.11 for residences
of various kinds. Air exchange rates range from 0.1 to 1.7. The
mean of the midpoints of the 15 ranges listed in Table 7-5 is 0.9.
The mean of the midpoints of the particular residence types are
listed below.
7-11
-------
Residence type
Exchange/hr
experimental
conventional
mobile
low-rise
high-rise
0.6
0.9
0.7
1.1
1.1
TABLE 7-5. AIR EXCHANGE RATES DETERMINED BY
MOSCHANDREAS ET AL.11
Location
Washington
Baltimore
Denver
Chicago
Pittsburg
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/hr
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
These results suggest a typical ventilation rate for-a non-experimental
home of one exchange per hour. Letting H = 1 and C = 1, Equation
* 3
(7-4) yields a =76.8 ug/m for the typical smoking home during
waking hours when all smokers are present. For a lower bound
estimate of a , we considered the case where one habitual smoker
lives in a large house (2000 square feet) with two air changes
per hour. In this case,
am = 76.8(^~-) = 19.10 ug/m3
for waking hours when the smoker was present. Equation (7-4) esti-
mates are not applicable during working hours in homes where one or
7-12
-------
more members work outside the home. A better overall estimate of am
may be determined from the following stratification of data reported
by Spengler et al.8
Location
Outdoor
Indoor: no smokers
Indoor: 1 smoker
Indoor: 2+ smokers
Homes
74
38
22
9
Mean MRP
(yg/m3)
22.3
24.0
42.8
74.5
MRP levels in homes with one smoker and with two or more smokers
exceed MRP levels in homes with no smokers by 18.8 yg/m and 50.5
yg/m , respectively. The average home with smokers exceeds the
average home without smokers by 28 ug/m . If we assume the above
breakdown represents a random sample, then 45 percent of homes
have one or more smokers.
The lower bound estimate of 19.1 yg/m developed using Equa-
tion (7-4) and the results of the Spengler- study suggested that a
3
for homes with smokers should not be set below about 20 yg/m .
The upper bound was set at 50 yg/m as suggested by the Ferris
results. We used 30 yg/m for our best estimate of a . This
3 m
value is midway between the 28 yg/m value discussed above and
the 32 yg/m value calculated for the intercept of the regression
line for homes with smokers. No adjustment was attempted to
account for "other" activities in the home-other environment.
7.3 TRANSPORTATION VEHICLE MICROENVIRONMENT
None of the studies in Table 7-1 discuss inside/outside par-
ticulate ratios for transporation vehicles. Only Repace et al.7
contains any useful data on PM levels inside transportation
vehicles. Respirable particulate data were collected by Wallace
during five days in August and September 1979. Figure 4 in
Repace et al. shows average daily exposure to RSP in several
common microenvironments. Average concentration values for
outdoors, home, office, bus, and subway are listed below.
7-13
-------
Location
Outdoors
Home
Office
Bus
Subway
Average RSP concentration
yg/m3
43
52
74
87
131
The outdoor measurements were taken at lunchtime at an unspecified
location. Figure 5 of the same study presents average values and
ranges of respirable particulate samples taken in different micro-
environments during spring and summer, 1979. Average values are
listed below.
Location
Outdoors (suburbs)
Home (suburbs)
Autos
Nonsmoking office
Diesel buses
Subway cars
Average RSP concentration
yg/m
3
37
60
78
85
106
134
In neither figure are data presented concerning particulate
concentrations measured by outside monitors concurrently with
particulate concentrations inside autos, buses, and subway cars.
Consequently, these data are not useful in providing best esti-
mates of a and b . They do suggest that PM levels in transpor-
tation vehicles exceed suburban outdoor concentrations and equal
or exceed PM levels in nonsmoking offices. Furthermore, PM
concentrations in excess of 100 yg/m frequently occur on buses
and subway cars in the Washington, D.C. area. According to
Figure 5, the highest RSP concentration recorded in a transportation
vehicle was approximately 180 yg/m . Since most measured outside RSP
concentrations exceeded 30 yg/m , we assumed that a for TSP and IPin
3 m 10
would be less than 150 yg/m for most commuters. We also assumed the
majority of commuters would be in automobiles, buses, or train
cars with no smoking. Thus, we used a = 0 for our low and best
m
7-14
-------
estimates and a =150 ug/m for our high estimate. For h>m, we
used the same values developed for roadside microenvironments (see
Section 7.4). We assumed that PM would not be significantly atten-
uated as it entered a transportation vehicle from the outside.
7.4 ROADSIDE AND OUTDOOR MICROENVIRONMENTS
PM exposures in the roadside and outdoor microenvironments
are assumed to occur at breathing height. Because most TSP
monitors are located at distances other than breathing height,
they yield data which tend to inaccurately represent pedestrian
exposure, especially near roadways.
Pace et al.l2'13 observed the change in TSP concentration
with monitor height at 31 monitors in 7 urban areas. They found
that TSP concentration varies inversely with height, and that the
strongest gradient is found below 9 meters. Pace suggests the
empirical model
y = ae"bh + c (7-6)
where y = predicted concentration in ug/m ; a, b, c are empirically
derived constants; and h is height in meters.
In fitting this model to TSP data from the 31 monitoring
sites, Pace found that a = 45, b = 0.2, and c = 31 ug/m yielded
2
a correlation coefficient (R ) of 0.41. He also found that
factors other than height, such as lateral distance to streets
and traffic density, could significantly affect measured TSP
concentrations.
More recently, Pace11* developed a more complex TSP model
which accounts for a variety of factors which impact on the
monitor, including area sources, point sources, local effects
(monitor height, lateral distance to two streets, and traffic
density on two streets), effects of a visible plume from cars,
and background concentration. However, to use this model it is
necessary to conduct an extensive microinventory of land use and
emission sources in the vicinity of the monitor. Microinventories
7-15
-------
have been conducted for all sites in St. Louis and selected sites
in Philadelphia. However, SASD has decided that a data intensive
model was not desirable at the time of this study. Since height
above ground was known for all monitors in the study, Equation
(7-6) was selected as the method for adjusting PM data.
Equation (7-7) , derived from Equation (7-6), was used to
estimate a b value to transform PM concentrations at each monitor
m
to the corresponding PM concentrations at breathing zone height.
bm = (ae~L'5b + Yh - ae"bh)/yh (7-7)
The variable y, is the annual average TSP concentration at a
monitor located at height h. Using a = 45 and b = 0.2 as recom-
mended by Pace, the numerator of Equation (7-7) yields an estimate
of the annual average value at breathing height. The ratio of
this estimate to y, was assumed to apply at all hours during the
year. Table 7-6 lists each TSP monitor used in the study, its
height, the observed annual arithmetic mean, and an estimate of
b . These b values were applied to the TSP air quality data to
determine TSP levels in the transportation vehicle, roadside, and
outdoor microenvironments.
IP-IO is assumed to vary less with monitor height than TSP.
Consequently, SASD recommended that b =1.00 for all three micro-
environments, regardless of monitor height. SASD also developed
the estimates of a
microenvironments.
the estimates of a listed in Tables 7-7 and 7-8 for these three
m
7.5 SUMMARY
Tables 7-7 and 7-8 summarize our estimates of a and b for
m m
TSP and IP-iQf respectively. Note that a values vary according
to time of day for the work-school and home-other microenvironments.
In most cases, the a values in Table 7-7 were developed from in-
halable particulate data. Research suggests that most TSP generated
indoors falls within the IP range. Consequently, the a values
for indoor microenvironments in Tables 7-7 and 7-8 are identical.
7-16
-------
TABLE 7-6. TRANSFORMATION OF AMBIENT TSP CONCENTRATIONS FROM
MONITOR HEIGHT TO BREATHING ZONE
Study area
Chicago
Los Angeles
Philadelphia
St. Louis
Exposure
district
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
1
2
3
4
5
6
1
2
3
4
5
6
7
Monitor site
141220018H01
141220029H01
141220036H01
141220015H01
141220032H01
147160007F01
142620001G01
141540016H01
053900001101
054260001101
055760004101
050500002101
050230001101
056400003F01
056680001101
397140026H01
393280108F01
393280108F01
397140004H01
397140024H01
311000001F01
264280006H01
264280015H01
264280062H01
264280025H01
264280010H01
264280015H01
142120009F01
Height,
m
14.9
24.7
12.2
16.8
9.1
9.1
8.5
7.6
6.1
6.7
5.5
2.1
3.7
3.7
24.4
4.0
1.2
1.2
5.2
4.0
3.7
13.7
11.6
4.6
10.7
10.7
11.6
4.6
Arith.
mean,
ug/m3
92
69
93
79
74
76
74
95
89
123
106
152
105
151
130
54
65
65
74
54
54
87
75
88
76
75
75
96
/v
b
Dm
1.34
1.48
1.32
1.40
1.35
1.34
1.34
1.25
1.22
1.18
1.17
1.03
1.11
1.08
1.25
1.24
0.97
0.97
1.23
1.24
1.22
1.35
1.44
1.17
1.37
1.38
1.39
1.16
7-17
-------
TABLE 7-7. ESTIMATES OF MICROENVIRONMENT FACTORS FOR TSP
""•"
Microenvironment
Indoors: work or school a
bm
Indoors: home or other am
m
am
am
bm
m
Transportation vehicle a
bm
Roadside am
bm
Outdoors am
m
bm
~ *i
Factor
(smoking)
(nonsmoking)
Hours ending
all
all
1-6, 23, 24
7-22
7-22
all
all
all
all
all
all
all
Estimated value
low
0
0.33
0
20 yg/m3
0
0.50
0
a
0
a
0
a
best
20 yg/m3
0.50
0
30 yg/m3
12 yg/m3
0.80
0
a
20 yg/m3
a
0
a
high
3
50 yg/m
1.00
0
50 yg/m
20 yg/m3
1.00
150 yg/m3
a
50 yg/m
a
0
a
'See Table 7-6.
7-18
-------
TABLE 7-8. ESTIMATES OF MICROENVIRONMENT FACTORS FOR IP
10
Microenvironment
Indoors: work or school a
bm
m
Indoors: home or other a_
am
in
am
bm
m
Transportation vehicle a
bm
Roadside a
"m
Outdoors a
bn
Factor
(smoking)
(nonsmoking)
Hours ending
all
all
1-6, 23, 25
7-22
7-22
all
all
all
all
all
all
all
Estimated value
low
0
0.33
0
20 yg/m3
0
0.50
0
1.00
0
1.00
0
1.00
best
20 yg/m3
0.50
0
30 yg/m
12 yg/m3
0.80
0
1.00
14 yg/m
1.00
0
1.00
high
50 yg/m3
1.00
0
50 yg/m
20 yg/m3
1.00
3
150 yg/m
1.00
35 yg/m
1.00
0
1.00
7-19
-------
The b values in Table 7-7 were developed using TSP data.
Since small particulates infiltrate indoor environments more
readily than large particulates, we assumed that the b values
for IP would equal or exceed the b values for TSP- However, no
data were available for developing b values specific to IP-
Consequently, most of the b values listed in Table 7-8 are iden-
tical to those in Table 7-7.
7.6 REFERENCES
1. Memorandum to Henry Thomas, SASD, EPA, from Ted Johnson,
PEDCo Environmental, Durham, N.C., August 20, 1980.
2. John D. Spengler, et al., "Personal exposures to respirable
particles," paper 80-61.5B, 73rd Annual Meeting of the Air
Pollution Control Association, Montreal, Quebec, June 22-27,
1980.
3. James L. Repace and Alfred H. Lowrey, "Indoor air pollution,
tobacco smoke, and public health," Science, Vol. 208, May 2,
1980.
4. Stanley J. Penkala and Gilberto de Oliveira, "The simulaneous
analysis of carbon monoxide and suspended particulate matter
produced by cigarette smoking," Environmental Research, Vol.
9, pp. 99-114, 1975.
5. C. Ray Thompson, Earl G. Hensel, and Gerrit Kats, "Outdoor-
indoor levels of six air pollutants," Journal of_ the Air
Pollution Control Association, Vol. 23, No. 10 (October 1973),
6. John E. Yocom, William L. Clink, and William A. Cote,
"Indoor/outdoor air quality relationships," Journal of_ the
Air Pollution Control Association, Vol. 21, No. 5 (May 1971).
7. James L. Repace, Wayne R. Ott, and Lance A. Wallace, "Total
human exposure to air pollution," Paper No. 80-61.6, 73rd
Annual Meeting of the Air Pollution Control Association,
Montreal, Quebec, June 22-21, 1980.
8. John D. Spengler, et al. , Summary of_ Air Pollution
Measurements, Air Quality Assessment Group, Harvard School
of Public Health, Boston, Massachusetts.
9. Carole Jo and John D. Spengler, "Room-to-room variations in
concentrations of respirable particles in residences,"
Environmental Science and Technology, Vol. 15, No. 5 (May
1981).
7-20
-------
10. U. S. Census Data, 1970.
11. D. J. Moschandreas, J. Stark, J. E. McFadden, and S. S.
Morse, Indoor Air Pollution in the Residential Environment,
Vol. !_._ Data Collection, Analysis, and Interpretation,
Publication number EPA-600/7-78-229a, U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina,
December 1978.
12. Thompson G. Pace, Warren P. Freas, and Elsayed M. Afify,
"Quantification of relationship between monitor height and
measured particulate levels in 7 U.S. urban areas, paper
no. 77-13.4, 70th Annual Meeting of the Air Pollution Con-
trol Association, Toronto, June 1977.
13. T. G. Pace, An Approach for the Preliminary Assessment of
TSP Concentrations, Publication number EPA-450/2-78-016,
U.S. Environmental Protection Agency, Research Triangle Park,
N.C., July 1978.
14. Thompson G. Pace, An Empirical Approach for Relating Annual
TSP Concentrations to Particulate Microinventory Emissions
Data and Monitor Siting Characteristics, Publication number
EPA-450/4-79-012, U.S. Environmental Protection Agency,
Research Triangle Park, N.C., June 1979.
7-21
-------
SECTION 8
ANALYSES OF POPULATION EXPOSURE USING NEM
This section describes NEM outputs and uses the results of
selected NEM runs to analyze population exposures over a range
of possible IP10 standards. The uncertainty inherent in this
type of analysis is discussed. The contribution of smoking to
IP,Q exposure is also analyzed.
8.1 DESCRIPTION OF COMPUTER OUTPUTS
Tables 8-1 through 8-7 show selected printouts of a computer
run simulating exposure to IP1Q in Philadelphia under current air
quality ("as is") using best-estimate microenvironment factors.
Each printout page has a two-line heading which identifies the
computer run by stating the pollutant under study, the air
quality standard or condition being simulated, and the test city-
A subheading identifies the microenvironment factors being used
(e.g., lower estimate, best estimate, or upper estimate). An
output page number is located in the upper right corner of each
output page.
NEM relates air quality within a microenvironment to air
quality measured at the centroid of an exposure district through
the model x = a + b x , as described in Section 7. Table 8-1
lists the values of am and b used in the computer run. Coeffi-
m m
cient a is called the multiplicative factor (mult) in Table 8-1
and the scalar quantity b is called the additive factor (add).
These factors are listed by exposure district name and number and
by microenvironment (at work or school, at home, inside a vehicle,
along a road, or outdoors) . Analysis of IP-,0 required different-
factors in the work-school and house-other microenvironments for
certain times of day, as indicated under "special case" and
8-1
-------
TABLE 8-1. IP1Q MICROENVIRONMENT FACTORS FOR PHILADELPHIA
(BEST ESTIMATES, SMOKING NOT CONSIDERED)
1P-10/BE ANALYSIS 1 (AS IS) FOR PHILADELPHIA
Best Estimate without Sucking
Page
I I C 1 0
ENV|R}V«E«r
riANSFOR*tTl)«S
._„_____________„:________--_____<
Exposure lone
1 Central Philadelphia
3 South Philadelphia
4 west Philadelphia
6 Canden, New Jersey
"
~
Work/School
lull. Adi
0.5
0.5
0.5
0-5
O.S
. ._
Modi
Hone/Other
.- HulU- . Ail
3.8
a.}
3.1
O.B
0.8
0.1
•
1 I c a t
In vehicle
_
.33
.33
.31
.DO
.00
.00
ion F i
Along Road
JI.U. AitL
1.00 1*
1.00 U
i » DA j 4.
1.00 U
1.00 14
1.00 U
i c t o r
Outdoors
.•.jilt*. JUJLJ
1.00
1.33
1.33
1.03
. 1.0JB_ „
1.00
.____..._--_---- 4
i
L
Special Cases
Add1 AddZ
I
0 12
3- 1ZL
0 12
0 12
0 12
I
r
1
00
I
The nodified air quality values for each exposure category Is equal to the observed cir quality value -ultiplied by the
Multiplicative factor plus_ Jfhe additive Jacioti
_T he_ipeti.a L_c > yf_s_JU_f°r ""orfc /school' tni 'Hoie / Other'.
'Work / Schsal' has additive factor '*J-1* used froa ? l.<
5
'Hpnie /. _Qther_'_ has., addit ive factor ">dd2' used ?ro» 7 A.^. thru 13 P.B.
For this case the year starts on Sunday.
The .(ninua cphoj-t_pjjpulatlpn. used was. 1j5Q3.
the Populil ior>. Reject ed yas 1SS,2B<> fro* 31t Grouos.
The P9pul)tipn used.iias . .. 2«77?t*69 froa ...271 Srpups,..
The Population Total was 2,934,753 froa) 568 Groups.
-------
TABLE 8-2. NUMBER OF TIMES PER YEAR THAT A COHORT OR PERSON OCCURS IN A MICROENVIRONMENT
BY EXERCISE LEVEL IN THE CENTRAL PHILADELPHIA EXPOSURE DISTRICT
1P-10/BE ANALYSIS 1 (AS IS) FOR PHILADELPHIA
Best Estimate without Sioking
N U MIU» . Q LI. J H|S _.tttAJL.» J QHORJL GROUf_pJ_A. PER SI? H OCC URS IN A HI C»0-ENWIHONHENT _AT. _TMf_ _S PECIflEP EKE «e.IlE_.LE.» C L.
Central PhilaJelahfa
00
I
CO
1
1
J
1
Lou Exercise
1
1 Cohort 1 People
nicroenvironc/it. 1 Occurrence* J. Occurrences
I 1
1
Jnrk / Srl,oP( |
1
Hone 1 Other 1
1
1 get dr.. vehicle I
1
Along Road 1
1
_. putdoo_rsL_L.
1
1 1
1 T o t a I |
1 1
1
119. 2s} | i tQi3t 000*000
1
108.120 1 4.7&Q,DJ)O.OM
1
Ji.SDJLl __5»5.JJ1JJ_1D_0
1.664 1 2B.600.POO
1
14.28S 1 2i4.00a.000
1
1
501,844 | 6.6ZDf DDO.DOO
!
i i
1 Ntdiu* Exercise 1
| |
1 1 ' 1
1 Cohort
1
1
1
I 6.665
1..
1 572.
1
JL _ -J.600,
|
1
1 33.215
1
People I
leeurrences | DC
1
1
1
132,300.000 |
1
I
1
M .900.000 |
1
194.113.303 |
1
1
St3.330.OQD |
1
|
High Exercise
.
Cohort I People
currence^ J OJECJI cut i« «.*... _
1
1
1
1
1
1
t
1
364 I 21.200.000
2,241 | 78,533,333
1
1
?,607 | loo.qipa.Qop
.
-------
TABLE 8-3. NUMBER OF TIMES PER YEAR THAT A COHORT OR PERSON OCCURS IN A
MICROENVIRONMENT IN PHILADELPHIA STUDY AREA
IP-1D/3E MlLfSIS 1 US IS) F3R PHILADELPHIA
Best Estlnate without Sucking
NUHBE R OF JIHE5 JHAT A COHORI GROUP OR < PERS3* _Stf UR5 I < •>
-'E«l 1 > Hi *I ._iLTJL_lP E C_! f 1 Ji_E_«_ E«J I S E_ L|J« E L. ___
OVERALL
00
]
I Lou Exercise
|
Microenviron«ent-
Work / School
rt3ie / Dther
lislde \tetiJele__
Along Road
•-*•"--
1 9 1 1 I
Cohort
..Occurrences,
350,204
220,973
6.084
_.. 67,203 ._
People
Occuecence-S__
IZ.llJUDJJjiDJL
94.3DD.QOO
- fiSr.QOOiOOO
22. 4)3. 33). 330
Hedl ui E«f rcl It
Cohort
Occjjrre ntt»_
51,290
3S.3»9
3.432
nj.sn
Occurrences
347,000.000
til. ))3. 303
lOt. ODD. 000
5SZ.OOO.OOO
1, Ji), ))3. 3D3
High Eiercise
|
. Cohort | fttalr
Occurrences J Occurrences
1
2.196
12j094
14,293
72.300,000
253,000,000
>
1
1
325,333,333 1
-------
TABLE 8-4. NUMBER OF POLLUTANT ENCOUNTERS PER YEAR IN PHILADELPHIA UNDER
CURRENT AIR QUALITY (IP1(Ji 24-HOUR AVERAGING TIME)
IP-1D/9E ANALfSIS 1 (AS 15} F3R PHILAJELPHH
3est Estimate without S«oking
POLLUTANT ENCOUNTERS FOR 24 HOUR AVERAGING TIHE
»,--
1
1
1
<
Concentration
Exceeded
700
603
533
400
225
153
110
35
70
55
40
23
0
1 HJM. Concent r » M on
People at NaxinUH
,.__......-_..-. .....4
Low Exercise
7.263.330
62.100.000
6(0.303.333
4.430.000.000
20.100.000.000
23.700.000.000
102
5,280
Aeiiu* Exercise
2.020.000
36.100.03}
309.000.000
1.430.000.000
1.660.000.000
91
9,700
Hi)* Etercile
.
420.000
6.723.133
69.SOO.OOO
304.303.333
343.000.000
95
9,700
.... ,
Any Exercise
7.533,303
64.600.000
6.83.333.303
t.M 0.000 .000
21.800.000.003
25.700.000.000
10?
5,280
^
>
t
'Pollutant Encounters' is the saae as person hours of exposure for 1 hour averaging tin*.
'Any Exercf s e ' »e> s_u.rgA-£ L8J> ».J«r-!t-JLa A e_o_e_fi dejtt
8-5
-------
TABLE 8-5. NUMBER OF PEOPLE EXPOSED PER YEAR IN PHILADELPHIA UNDER CURRENT
AIR QUALITY (IP1Q, 24-HOUR AVERAGING TIME).
IP-10/3E »M»LfSIS 1 US IS) FDR PHILADELPHIA
Best Estimate without Sacking
PEOPLE EXPOSED F3R 2» H3lM A»E*A5I«S THE
f
1
1
1
1
1
«
Concentration
Exceeded
(•]/•*»;>
703
609
503
403
^Qf)
225
153
113
85
73
55
43
20
3
Hai. Concentration
People at n»»i«u»
LOK Exercise
365.000
2.030*000
2.930.000
Z.93D.330
2.930.000
2.93Q.aOO
102
5,280
4ediua Exercise
193.000
(B7.009
2.150.003
2.790.033
2.790.000
?. 790.000
92
9,700
Hijh Exercise
-
70.333
137.333
BD7.3DO
1. H3. 333
1.1*0.000
1.I13.33Q
93
9,700
Any Exercise
365.303 I
1
2.383.303 1
-
2.930.003 |
2.J33.303
2.930.000
2.933.303
102
5,280
>
>
'Any Exercise" Measures exposure independent of exercise level.
8-6
-------
TABLE 8-6. NUMBER OF PEOPLE AT PEAK EXPOSURE PER YEAR IN PHILADELPHIA UNDER
CURRENT AIR QUALITY (IP1(J, 24-HOUR AVERAGING TIME)
IP-10/BE ANALYSIS 1 (AS IS) FOR PHILADELPHIA
Best Estimate without Smoking
PEOPLE AT PEAK EXPOSURE FOR 24 HOUR AtfE»ABI<6 THE.
1
1 Concentration
1 Range
1 (•}/•*»*»
1
1
| 733 < C <= 13DD
I
| &3Q < r <= 7nn
1
| 530 < C <= 600
1
| 400 < C <= 500
1
1
| 225 < C
1
Lou Exercise
3SS.330
1,720.000
855. 3DD
102
5,250
leliu* Exercise
193.033
294.000
1.&60.03)
644.000
92
9,700
Hljli Eterclse
71.333
37.200
7)1.113
528.300
8.533
93
9.730
lay Exercise
345,303
1.720.000
353.333
102
5,283
9-
».
1
1
»
"Any Exercise' Measures exposure independent of exercise level.
8-7
-------
TABLE 8-7. NUMBER OF PEOPLE EXPOSED PER YEAR IN PHILADELPHIA UNDER
CURRENT QUALITY (IP1Q, ANNUAL ARITHMETIC MEAN)
IP-10/BE ANALYSIS 1 CAS IS) FOR PHILADELPHIA
Best Estinate without Smoking
PEOPLE EXPOSED FOR ANNUAL AVERA6E
1
1
Concentration
Exceeded
(ng/B**3)
70D
500
4QO
?25
150
11D
85
73
55
43
23
0
Max. Concentration
People at Maxi«u»
Lou Exercise
i
0
NeJiUM Exercise
0
....................
High Exercise
0
>-.-_.......*......<
Any Exercise
130.303
2.933.303
2.930.000
40
14.500
r
1
1
»•
'Any Exercise" Measures exposure independent of exercise level
8-8
-------
explained in the footnotes. The same transformation factors were
used for all exposure districts.
The footnotes contain the statement "for this case the
year starts on Sunday." This information is necessary to call
up the three typical day patterns (weekday, Saturday, and Sunday)
in correct sequence for the year. Another statement says that
the minimum cohort population was 1,500—the criterion used to
accept or reject a cohort in the computer run (see Section 4.3).
A summary statement shows the results of cohort screening using
the 1500 criterion. In Philadelphia, 317 of 588 cohorts were
omitted from the simulation; this resulted in a loss of only
155,284 people, or 5.3 percent from the total study area popu-
lation of 2,934,753.
Tables 8-2 and 8-3 show the number of times that a cohort
occurs within a given microenvironment at each exercise level
(cohort-occurrences) during the year according to the activity
patterns data file. These figures also show the number of per-
sons (people-occurrences) that experienced each microenvironment-
exercise level combination for one or more hours during the year.
These data were obtained by summing populations of all cohorts
that encountered each microenvironment and exercise level. Table
8-2 lists occurrences for Central Philadelphia, one of six expos-
ure districts in the Philadelphia study area. Table 8-3 lists
occurrences for all six exposure districts combined.
Tables 8-4 through 8-7 illustrate the six NEM output tables
which summarize IP-.Q population exposure estimates calculated by
the computer run. Each table uses one of three measures of popu-
lation exposure and one of two averaging times, 24 hours or year.
Table 8-4 summarizes pollutant encounters of exposure for a
24-hour averaging time. The number of pollutant encounters
estimated for a given concentration is the number of times per
year a cohort experiences that concentration (or higher) times
the cohort population summed over the number of cohorts in the
study area population. In other words, it is the number of
times per year that individuals in a study area encounter pollutant
8-9
-------
levels which equal or exceed a concentration. Thus Table 8-4 is
a cumulative frequency distribution in which pollutant encounters
increase as IPi0 concentration decreases; pollutant encounters
reach a maximum when IPi0 equals zero. This maximum is the
population used in the simulation times the number of possible
hourly encounters in a year (8760). The figure shows different
frequency distributions for cohorts who were at low, medium, and
high exercise levels at the end of each 24-hour period when the
given air quality levels were encountered, and a frequency dis-
tribution for occurrences at any of the three exercise levels.
The number of pollutant encounters at all exercise level for the
highest concentration encountered (85 yg/m ) was 7.6 million.
The number of pollutant encounters greater than or equal to zero
was 25.7 billion, the maximum possible for this study area.
At the bottom of each of the three tables with the same
averaging time (Tables 8-4 through 8-6) is listed the highest
concentration encountered by any cohort (max concentration) and
the number of people in the cohorts that encountered that maximum
value (people at maximum). In Table 8-4, the maximum values were
3 3
102 yg/m at low exercise and 93 ug/m at high exercise levels.
In the same figure the number of people at maximum concentration
at low exercise was 5,280, while the number at maximum concentra-
tion at high exercise levels was 9,700,
Table 8-5 uses the second output measure, people exposed,
which is the number of people in the study area that experience a
specific air quality level or higher. Like pollutant encounters,
this measure of exposure is expressed as a cumulative frequency
distribution. However, the number of people exposed at a given
concentration will usually be much smaller than the number of
pollutant encounters since a person can be counted only once (at
most) in determining people exposed at that concentration.
The third measure of exposure is "people at peak exposure."
Table 8-6 lists discrete intervals of pollutant concentrations in
which various cohort populations experience their highest ex-
posure. This distribution is not cumulative; each cohort popula-
tion falls within a single interval.
8-10
-------
The next three tables in a typical computer output for IP10
are similar to Tables 8-4 through 8-6, except that annual average
concentrations are used instead of 24-hour average concentration.
An example of one of these tables is shown in Table 8-7. In this
example, exposure distributions at different exercise levels are
not listed for annual average concentrations because of the dif-
ficulty in determining an exercise level appropriate for an entire
year. Note that population exposures occur at lower concentrations
for annual averages than for 24-hour running averages because the
annual average of hourly concentrations is less than or equal to
the highest hourly value encountered during the year. For ex-
ample, the maximum annual concentration encountered was 40 ug/m
in Table 8-7, while the maximum 24-hour concentration encountered
was 102 yg/m in Table 8-5. Similarly, the number of people
exposed at 40 yg/m and above was 130,000 in Table 8-7, compared
with 2,930,000 in Table 8-5.
Other output tables can be produced by specifying different
study areas, different microenvironment factors, or different air
quality levels. By comparing NEM results for different scenarios,
the factors that determine population exposure may be analyzed.
8.2 DETERMINANTS OF EXPOSURE ESTIMATES
Before a particular scenario is analyzed using NEM, certain
inputs to the model may be identified as determinants which sig-
nificantly affect the esposure estimates. These determinants
include activity pattern data, microenvironment factors, size of
exposure districts, air quality differences between districts,
and age-occupation group distributions across districts. Determinants
vary from one study area to another except for activity patterns
and microenvironment factors. Due to limited data, the same
activity patterns and microenvironment factors were used in each
of the four study areas.
One of the important characteristics of the activities pat-
tern file is the frequency of occurrence of high exercise levels.
8-11
-------
Exercise levels represent useful information because dosage
levels, the amount of pollutant actually entering the body, may
be significantly higher when exercise levels are high. The NEM
model calculates different exposure distributions at different
exercise levels so that output may be better utilized to estimate
health effects. The average number of hours spent at various
exercise levels is shown in Table 8-8. These data represent a
summary of hourly assignments in the activities pattern file.
Analysis of these hourly assignments suggest that the occurrence
of high exercise levels may be underestimated for the total popu-
lation of a study area. Only three age-occupation groups encoun-
ter high exercise levels: operatives and laborers, children
under 5, and children 5-17. We would expect most of the other
age-occupation groups to encounter high exercise levels for at
least one hour during a typical week. Consequently, exposures
at high exercise levels are probably biased low for these groups
and for the study area population as a whole.
Table 8-9 summarizes people-occurrences at high activity
levels for the four study areas as used in NEM. The table shows
that there are a large number of occurrences of hours at high
exercise levels. Note that such occurrences are determined
solely by data in the activity patterns file and by populations
of the respective cities. Although such occurrences represent a
small fraction of occurrences simulated in NEM, adverse health
effects could be realized if such occurrences were to coincide
with high pollutant levels.
8.3 UNCERTAINTY IN NEM EXPOSURE ESTIMATES
NEM, like any method used to estimate exposures of large,
diverse groups of people, must deal with inadcuracies of data and
assumptions about relationships between phenomena. The value of
each of the exposure determinants discussed above is uncertain to
some degree. For example, TSP or IP-iQ can be measured in the
ambient air only within certain limits of accuracy and precision.
8-12
-------
-TABtE- a>a* AVERAGE NUMBER OF HOURS9 SPENT AT VARIOUS EXERCISE LEVELS IN A
TYPICAL WEEK BY AGE-OCCUPATION GROUP
Age-occupation
group
1. Students age 18 +
2. Mgrs. & professionals
3. Sales workers
4. Clerical workers
5. Craftsmen & foremen
6. Operatives & laborers
8. Service & household
9. Housewives
10. Unemployed & retired
11. Children under 5
12. Children 5-17
Exercise level
Lowb
163
165
164
167
157
144
155
161
162
151
146
Medi urn
5
3
4
1
21
20
13
7
6
14
16
High
0
0
0
0
0
4
0
0
0
3
6
All
168
168
168
168
168
168
168
168
168
168
168
Average of hours spent by three activity subgroups weighted by percent in
each subgroup.
\ow exercise includes hours spent sleeping (56 h/week).
PEOPLE OCCURRENCES AT HIGH EXERCISE
LEVELS AND (PERCENT OF TOTAL) BY STUDY AREA
Study area
Philadelphia
Los Angeles
St. Louis
Chicago
Microenvironment
Along road
72,300,000 (0*30%)
190,000,000 (0,29*)
31,200,000 (0.32%)
21,700,000 (0.12%)
Outdoors
253,000,000 (1,04%)
653,000,000 (0.99%)
107,000,000 (1.09%)
109,000,000 (0.58%)
All
325,300,000 (1.34%)
843,000,000 (1,28%)
138,200,000 (1,4.1%-)
130,700,000 (0.70%)
8-13
-------
Likewise, age-occupation distributions are known rather accurately
for a census year, but subsequent shifts in population must be
estimated. The size and shape of exposure districts are determined
by judgement, and such judgements affect the degree to which
people movement in the simulation model correlates with movement
in the real world. It would be desirable to quantify uncertainty
in each of these parameters and to place confidence intervals on
estimated exposures. A detailed sensitivity analysis has not yet
been conducted for the NEM model due to current limitations of
time and resources.
A major source of uncertainty in NEM estimates can be evalu-
ated by examining' microenvironment factors, one of the major
determinants of model results and one of the least certain. As
documented in Chapter 7, a range of factors were derived from a
review of the literature. The lower estimate represented a lower
bound on probable values of microenvironment factors and the
upper estimate represented an upper bound. The best estimate lay
in the middle of the range and represented the best estimate of
the investigator (subject to EPA review) based on current research.
The effects of uncertainty in microenvironment factors may
be analyzed by varying the microenvironment factors and keeping
the other determinants constant. Figure 8-1 shows pollutant en-
counters to IP-.Q (24-h averaging time) for Los Angeles corres-
ponding to low, best estimate, and high microenvironment factors.
In each of the three cases, current air quality levels are
assumed and the exposure measure is pollutant-encounters for 24-
hour averaging time. The results are about the same for very low
pollutant concentrations where there are a large number of
pollutant encounters. At 20 yg/m , the number of pollutant en-
counters was 67.6 billion using the upper estimate and 66.0
billion using the lower estimate. The differences widen for
higher concentrations, where there are relatively smaller numbers
of pollutant encounters. At 150 ug/m , the number of pollutant
encounters was 3.3 billion using the upper estimate and 12.8
million using the lower estimate.
8-14
-------
10 is
QLOHER ESTIMATE
O BEST ESTIHBTE
AUPPER ESTIHflTE
0 50 100 150 200 250 300 350 400 450
IP CONCENTRATION (MG/M**3.)
Figure 8-1. Effect of uncertainty of microenvironment factors on IP1n
exposure estimates for Los Angeles under 10
current air quality with smoking contribution included.
8-15
-------
Best estimate factors yielded as estimate of 203 pollutant
encounters at 150 yg/m . Since microenvironment research is in
an early stage, further work should narrow these differences. At
the present time they represent a major source of uncertainty in
the population exposure estimates.
8.4 EFFECTS OF DIFFERENT NAAQS'S
An important application of NEM population exposure estimates
is the analysis of effects of different NAAQS's on exposure. At
this time it is useful to compare the effects of possible IP1Q
standards on IP10 exposures. This may be accomplished by select-
ing one population measure, one averaging time, and one study
area and comparing the results of different air quality scenarios.
Figure 8-2 shows the number of people exposed to IP10 (24"h
averaging time) in the Los Angeles study area under four different
air quality scenarios. These scenarios include current air qual-
ity and three combined standards specifying annual arithmetic
means and maximum 24-h average values not to be exceeded.' The
"55/150" scenario, for example, assumes that the study area just
meets a combined IPin standard with 55 yg/m as the arithmetic
3
mean not to be exceeded and 150 yg/m as the maximum 24-h value
not to be exceeded.
Since the purpose of the graph was to observe how differences
in ambient air quality affect exposure estimates, the contribution
to exposure from smokers in indoor microenvironments was omitted
from the analysis. All four scenarios used best-estimate micro-
environment factors with indoor smoking excluded. According to
the NEM analysis, exposures estimated for the four scenarios would
not be significantly different at low levels (55 yg/m or lower).
However, the number of people exposed at higher levels would show
substantial variation. At 100 yg/m and above, the number of
people exposed would be 750 thousand for the 55/150 scenario,
4.5 million for the 85/225 scenario, 6.3 million for the "as is"
scenario, and 7.3 million for the 110/300 scenario.
8-16
-------
10
I
-------
8.5 EFFECTS OF SMOKING
A useful application of NEM is to analyze the impact of
indoor-generated pollutants on total exposure. Cigarette smoking
in indoor microenvironments contributes substantially to indoor
concentrations of particulate matter. IPin microenvironment factors
used in the model for home, work, and inside vehicles can be
switched "on" or "off." Switching on or off effectively creates
two scenarios, one with the effects of smoking and one without.
To simulate the effects of smoking, the study area population
is split into two segments. Since research suggests that 45% of
the population live in homes with smokers, one segment containing
45% of the population has an additive factor (a ) of 30 yg/m
applied to the home-other microenvironment between 7 a.m. and 10
p.m. The other segment (55% of the population) is assumed to live
in nonsmoking households, and a lower additive factor of 12 yg/m
is applied to the home-other microenvironment. Both segments are
assumed to work in smoking environments; consequently, an additive
factor of 20 yg/m is applied to this microenvironment for the
total population. To estimate exposures in the absence of smoking,
an additive factor of 12 yg/m is used for all homes and an addi-
tive factor of zero is used for all workplaces.
Figures 8-3 through 8-6 show results of comparing exposure
estimates with and without smoking for the four study areas. All
scenarios are identical except for the smoking contribution. All
scenarios represent current ("as-is") air quality and all use
best estimate microenvironment factors. Each of the figures uses
pollutant encounters for a 24-h averaging time as the measure of
population exposure at any exercise level. In each study area,
the smoking contribution causes the exposure estimates to be
substantially higher than exposure estimates without a smoking
contribution. The magnitude of the differences between smoking
and nonsmoking exposures varied from one study area to another.
For purposes of comparison, observe the differences in estimated
pollutant encounters at 70 yg/m . In Chicago the estimate was
8-18
-------
ion
T
T
o
10 21
QH1THOUT SUCKING
OHITH 46Z SMOKING
0 50 100 150 200 250 300 350 400
IP CQNCENTRflTION (MG/M**3)
Figure 8-3. Effect of smoking on IP1Q exposure estimates for Chicago.
450
8-19
-------
10"
o
CL-
IO 2
QHITHOUT SHORING
OH1TH 462 STOKING
50 100 150 200 250 300 350 400 450
IP CONCENTRflTION (MG/M**3)
Figure 8-4. Effect of smoking on IP1Q exposure estimates for Philadelphia.
8-20
-------
DWITHOUT SMOKING
OMITH 46Z SMOKING
50 100 150 200 250 300 350 400 450
IP CONCENTRflTION (MG/M**3)
Figure 8-5. Effects of smoking on IP1Q exposure estimates for Los Angeles.
8-21
-------
10 2
CJWITHOUT SMOKING
OMITH 46X SMOKING
0 50 100 150 200 250 300 350 400 450
IP CQNCENTRRTION (MG/M**3)
Figure 8-6. Effects of smoking on IP1Q exposure estimates for St. Louis
8-22
-------
0.9 billion without and 1.6 billion with smoking considered. In
Philadelphia the estimate was 64 million without and 169 million
with smoking included. In Los Angeles, pollutant encounters were
12 billion without smoking and 16 billion with smoking. For St.
Louis, the estimate was 331 million without smoking and 559
million with smoking. These results suggest that smoking in
indoor microenvironments can contribute significantly to pollu-
tant levels experienced by the general population. However,
population exposure in the absence of smoking can be substantial;
24-h pollutant encounters to IP10 concentrations of 200 yg/m and
above exceed 100,000 per year in Los Angeles under current air
quality conditions with no smoking.
8-23
-------
APPENDIX A
COMPUTER PROGRAM USED FOR DETERMINING
DATA ADEQUACY
A-l
-------
c
c
...c_ .-_..
c
_c
c
_c
c
c
c
__c
c
c
"c
c
c
c
r
C
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
__PROGRA"I CD (COMPLETE DATA).
FORTRAN V VERSION (CD2)
- _ j
PROGRAM CD PRINTS TABLES SHOWING THE AD
TO TWO ^LTER.MATIVE_CRITER1A__FOR CCKPLETENE
EXPOSURE STUDY WILL BE CONDUCTED USING RAW
SOME OF THE MONITORING SITES LISTED IN THE
EQUACY OF DATA ACCORDING
SS.. A POPULATION
AIR QUALITY DATA FROM
TABLES.
FOUR STUDY AREAS WERE SELECTED, AS FOLLOWS:
AREA SflSA
1 44SO L-A.-LONG BEACH
1 __ _67£Q SAN BERNARDINO;
"RIVERSIDE; ONTARIO
2 16CQ CHICAGO
i 7040 ST. LOUIS
4 6160 PHILADELPHIA
THE INPUT DATA FILE FOR THIS PROGRAM
THE PROGRAK MRB1, WHICH SELECTS 5 DATA PAR
SAROAD S/F FILES. THE FOLLOWING QUARTERLY
SELECTED AND STORED IN FILE STDS*4C3P.:
VARIABLE NA*E
V1CJ) NO. OF OBSERVATIONS
V2CJ) ARITHMETIC MEAN
V3CJ) SOTH PERCENTILE
V4£J) 90fH PFRCFNTTI F
V5(J) 99TH PERCEKTILE
V'HERE INDEX J REPRESENTS QUARTERS 1 TO 16
A-YEAR PERIOD 1976-79.
STATE
05 CA
05 CA
14 IL
26 WO; 14 IL
31 NJ; 39 PA
IS CREATED BY
AHETERS FRO!« THE
DATA HAVE BEEN
WRB1 FIELD
01 IDNC1)
22 IDNC2)
05 IDNC3)
07 TDN(4)
09 IDNC5)
OVER THE SELECTED
OUTPUT ARRAYS ARE USED TO COLLECT THE RESULTS FROM THE PROGRAM.
THEIR INDICES ARE AS FOLLOWS-
IP POl UJTANT fcfo.
IA STUDY AREA NO.
_N=.LSU£,JA> SITE .N0_. FOJJ_ &QLLUT AN7 ANJL.AR.EA
J QUARTER
_._SET_. LiniTS OF THE ARRAYS -BY. KMftX-. LA.RGESJ—Na-
OF SITES EXPECTED FOR ONE AREA.
•• •
PARAMETER NKAX=110
"P»RAKETER._. _\QTRS = 16.
DIrtENSlON IDNC5), V(5,16)
-------
INTEGER ISC3.4>V .OPOLLt . OAREA -
INTEGER C75C3,4,KMAX,16>, C5QC3,4,NKAX,16)
INTEGER ._. NSHSA_C5f4,NKAX> , NMET HC 3, 4,NHAX> , _N INTV C3 ,4 rNMAXX
DOUBLE PRECISION NSAR(3,4,N*AX)
DO 5 IP_=1.,3___ -.- :
DO 4 IA=1 ,4
IS ( IP 11 A)_f Q .-
4 CONTINUE
5 CONTINUE . --
C
C*****_ READ IN _THE SITE DATA AND..THE QUARTERLY DATA.
C
1C CONTINUED _ .... .
READC3TENt> = 40) ISTAT,IAREA,ISITE,IAGY,1PRJrI POLL,IKETH,IINTV,I REG
RfTA£CR,ISKSA,ICNTY,IYRFrIDN,ICFe,ICFA,VQBf VSA,VCRTF,VSYR, ..
RVSQR,VEYR,VElJR,VBYR~,veQR,«V(IfJ),J-1tNQTRS>,r=1,5>
C
C***** SELECT_THE. POLLUTANT AND._AREA.JO-BE LISTED.,._ AND SET POLL. .INDEX.
C IF NONE APPLY, READ IN A NEW RECORD.
..C ....... . :
IF (IPOLL.EQ.'42401'.AND.IINTV.EQ.'Y') IP=1
IF (IPOLL.EQ.'42401'.AND.IINTV.EQ.'Y') GO TO. 15.......
IF (IPOLL.EQ.'42602'.AND.IINTV.EQ.'7') IP=2
IF (IPCLL.EQ.14.260.2'.AND.IINTV.EQ.'7') GO_TO._15 _
IF (IPOLL.EQ.'42602'.AND.IINTV.Eft.'1'> IP=2
. _. I F (I P C L L . E Q . '4 2 6 0 2 ' . A N D . 11N T V . EQ . '.1' ) 6 0_T_Q_15.
IF (IPOLL.EQ.'11101'.AND.IINTV.EQ.'7') IP=3
IF (IPOLL.EQ.'11101'.AND.IINTV.EQ.'7") 60 TO_J 5
GO TO 10 .
C ...... -• l •
C***** SELECT/STUDY ARE AS-6 ^~S«SA. THE FOLLOWING ORDER REFLECTS THE
_C ORDER IN WHICH SITES ARE LISTED IN THE. FILE.. (BY SAROAD... CO &£.)..„
C SET AREA IKDEX. IF NONE APPLY, READ IN A NEW RECORD.
C . „.
15 CONTINUE
IF CISKSA.EC.'448C') GO TO 20
IF (ISMSA.EC..'6780'J_IA = 1_._ _ -
IF CISKSA.EQ.'67SO') GO TO 20
1F__(ISKSA...EP. '16C.O') IA = 2 . •• .
•* If (ISKSA.EQ.'1600'> GO TO 20 *
IF _CISKSA.EQ.*Z040.1>._ IA=3
'IF USKSA.EQ.'7040') GO TO 20 ,
1F_ (ISKSA .EQ._'6.160±>__ IA=.4__ ::
IF (ISKSA.EQ.'6160') GO TO 20
GO TO 10
C
_C *.**** ADVANCE JTHE SJJE_NO_INDEX , _THEN SEE .1F_ ARR AY_SIZE_ LU?IT
C IS EXCEEDED. IF SO, PRINT ERROR MESSAGE.
C ASSIGN. (X)__WJHEN_.DATA_COMPLE.tENES.S_CRUE81A_rS_jaEJ '.
C FOR EITHER HOURLY OR 24-HOUR DATA.
C
20 CONTINUE
1S(IP.,IA> * ISCIP,IA)
N=IS(IP,IA)
-------
IF (N.GT.NP«AX> GO TO 21 " .._ ..
IF (N.LE.NPUX) GO TC 22
21 CONTINUE _. _. _ _. . _-
LIKIT=Nf*AX
WRITE (6,415.>_A.IKIT,N,.IP, IA
GO TO 10
.__2Z CONTINUE.
IF(IINTV.EQ.T) GO TO 25
D0.__24 _J=1 ,_16 _.
C75 (IP,IA,N,J)=' '
1 E JtYJCUJ ) .JS.E..JJ... OL3L C 7 5 (IP , LA , K, I )_=rx:
IF (V(1,J).EQ.99999.9) C75(IP,I A,N,J)='
C 5 0 (1P , 1A , N , J ) = *_ '
IF (V(1rJ).GE.8.0> C50(IP,IA,N,J)-'X
IF (V (1 , J ) . E Q .99999. .9 ) _. C 5 Q (.1 P , I A , K, J )_= '
' 24 CONTINUE
GO TO 3.C ._._... . -..
25 CONTINUE
DO 29~^M~,T6 - —
IF (V(1,J).GE.1643.0) C75(IP,1A,N,J)='X'
IF CVC1 ,J)_.EQ.99999.9) C75 (IP, IA, N , J) ='__^ .
C50 nPt!A,N,J)=' '
IF, (V <1, J.) . G E . 1 0_9 5 - P )_._. C 5 0 (I P_, I A , N , J ) -'*'_
IF (rf(1,J).EQ.99999.9) C50(IP,IA,N,J)=' '
_ 29 CONTINU!
C
C**_*** .ASSIGN. J.HE...I.NRUT DATA FROM EACH RECOR D__.TO._THE _APPPROPR1ATE _____
£ OUTPUT ARRAY.
c __________ : INPUTS_FPR_SAROA&. SITE_.CODE_.ARJ=_.COMBINEP_.TP_. REDUCE ____
C NO. OF ARRAYS.
.C ____________________________________________ ., _____________ ......... ____________ . _____________ ......
30 CONTINUE
NSKSA (IP\IA,N)=ISM.SA ______ ___________________________ ._ _ ______________ _____
ENCODE (12,420,NSAR(IP,IA,N) ,IDUM*Y) ISTAT , I AR EA , IS ITE , IA6Y,IP
NINTVC1P,IA,N)=IINTV
GjO_. T0...1.Q
C
Jt***.**._P R I W T_ TH E _ ?. E.S.UL.T S _ F R OM._T.KE _ 0 UTPUJ. _AR R A.Y S_.
C . ASSIGN CHARACTER NAMES TO OPOLL AND OAREA.
OK THIS_
C AREA OR THIS POLLUTANT, PRINT THE HEADIN AND 1 DATA LINE.
C ______ IF KOT_jr_H.E_1.ST._LJINEt J UST_P.RINT_ T_,DAIA_LI_NE. _________
C
______ 40 CONTINUE ____________________________________________________ _____ _____ __ _
DO 5C IP = 1 ,3
IF (1P.EQ.1) OPOLL='S02'
I F_ .(l.P_..EQ . 21_OP_OLL='NOZ' _
IF (IP.EQ.3) OPOLL='TSP'
. I F. . (IA - I_C .1) _. ,0 A R E A = ' L A - S B *
-------
IF (1A.EG.2) OAREA='CH1C '
I F__{lA_.Ej8_.J.)_OAJR£ArlSI..UO.U*
IF fIA.EQ.4) OAREA='PH1LA '
DO 4fc N=1 ,NS
IF (N.GT. 1) .GO.JTO 41
WRITE (6,403) OPOLL,OAREA
41 CONTINUE
WRITE (6,410) NSKSA(IP,IAfN), NSAR(IP,1A,N),
NINIV(IP,1A,N) ,
S(C75(IP,IA,NfJ),J=1,16), CC5Q(IP«IAvN,J)fJ-1ff16)
.4.8 _C0NT 1NUE :
49 CONTINUE
J_0 CONJ1NUE
GO TO 999
C***** PRINTING FORMATS FOLLOW.
C
403 FCR"AT(1H1 ,16X,"DATA ADEQUACY BY QUARTER AT ',A3,
___ R" MO{ilTORIN6_SIT.ES"/ _ _____
&31X,'IN STUDY AREA ',A6/
SI OX ,6 5 ('=')/
&3°X,'75X COMPLETE GTRS ','50% COMPLETE GTRS'/
810X,'SMSA MONITOR SITE METH INT 'T2('1976 77 78 79
S.10X ,65('___')J
'
410 FORKATdH ,9X ,A4 , ZX ,'A 1 2 , 2X , A2 , 3Xf A1 ,3X r
415 FORf,AT(1H ,"***OUTPUT ARRAY SIZE LKTIT',' EXCEEDED*** LIl*riT='t
._S.I.4.., '__N-'_, 14*.^.. P.OLLUTAN t=r*I.1., -1 __AREA =' »II )
420 FOR«AT(A2TA4rA3,A1,A2)
999_ CONTINUE
END
-------
APPENDIX B
WEEKDAY AND WEEKEND ACTIVITY PATTERN DATA
FOR EACH A-O SUBGROUP
B-l
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 1—STUDENTS AGE 18+ SUBGROUP:1 PCT IN SUBGROUP:25
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL
WEEK OF DAY 12345678910
WEEKDAYS AM H
2
1
PM H
1
1
SATURDAY AM H
2
1
PM H
1
1
SUNDAY AM H
2
1
PM H
5
1
H
2
1
H
1
1
H
2
1
H
1
1
H
2
1
H
5
1
H
2
1
H
1
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
H
1
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
H
1
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
3
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
2
H
2
1
H
2
1
H
3
1
H
5
2
H
2
1
H
2
1
H
2
1
H
2
1
BY
11
H
1
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
HOUR
12
H
1
1
H
2
1
H
1
1
H
2
1
H
2
1
H
3
1
LOCATION CODES: H=HOME
MICROENVIRONMENT CODES:
3=TRANSPORT VEHICLE
W=WORK
1=INDOORS AT WORK
4=TRANSPORT OTHER
2=INDOORS - OTHER
5=OUTDOORS
ACTIVITY LEVELS: 1=LOW
2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT' ALL
A-0 GROUP: 1—STUDENTS AGE 18+ SUBGROUPS PCT IN SUBGROUP:17
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
: 1
PM H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
3
1
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1
H
1
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
1
2
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
1
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
H
2
1
HOUR
12
H
2
1
H
3
1
H
5
2
H
2
1
H
2
t
H
2
1
LOCATION CODES'- H = HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
SUBGROUPS PCT IN SUBGROUP:58
A-0 GROUP: 1 — STUDENTS AGE 18+
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 9 10 11
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
3
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
«• ^ •»
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
2
1
• •• • • •
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
5 ™ ^ ™
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
HOUR
12
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME
MICROENVIRONMENT CODES:
3 = TRANSPORT VEHICLE
W=WORK
1=INDOORS AT WORK
«» = TRANSPORT OTHER
2=INDOORS - OTHER
5 = OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 2—MGRS £ PROFESSIONALS SUBGROUP:1 PCT IN SUBGROUP:19
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM W
3
1
SATURDAY AM H
2
1
PM H
3
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
W
3
1
H
2
1
H
2
2
H
2
1
H
3
1
H
2
1
W
3
1
H
2
1
H
5
2
H
2
1
H
5
1
H
2
1
W
3
1
H
2
1
H
5
2
H
2
1
H
5
1
H
2
1
W
3
1
H
2
1
H
5
2
H
2
1
H
5
1
H
2
1
W
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
3
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
W
3
1
H
2
1
H
2
1
H
2
1
If
2
1
H
2
2
W
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
3
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
W
3
1
H
2
1
H
2
1
H
2
1
H
3
1
HOUR
12
H
3
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 2—MGRS £ PROFESSIONALS SUBGROUPS
PCT IN SUBGROUP:51
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM W
1
1
SATURDAY AM H
2
1
PM H
2
2
SUNDAY AM H
2
1
PM H
2
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
3
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
2
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
= = 8 = S
HOUR
12
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME
MICROENVIRONMENT CODES:
3=TRANSPORT VEHICLE
W=WORK
1=INDOORS AT WORK
4=TRANSPORT OTHER
2=INDOORS - OTHER
5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 2—MGRS £ PROFESSIONALS SUBGROUPS PCT IN SUBGROUP-'SO
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 1 2 3 4 5 6 7 8 9101112
WEEKDAYS AM H
2
1
PM W
1
1
SATURDAY AM H
2
1
PM M
1
1
SUNDAY AM H
2
1
PM H
5
1
H
2
1
W
4
1
H
2
1
W
1
1
H
2
1
H
5
1
H
2
1
W
1
1
H
2
1
W
1
1
H
2
1
H
5
1
H
2
1
M
1
1
H
2
1
W
1
1
H
2
1
H
5
1
H
2
1
W
1
1
H
2
1
W
3
1
H"
2
1
H
3
1
H
2
1
M
3
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
W
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
M
2
1
H
3
1
H
2
1
H
2
1
W
3
1
W
1
1
14
3
1
W
1
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
W
1
1
H
3
1
W
1
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
W
1
1
H
2
1
H
3
1
H
2
1
LOCATION CODES: H=HOME M=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 1=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOM 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 1 2 3 H 5 6 7 8 9101112
WEEKDAYS AM H
2
1
PM W
5
1
SATURDAY AM H
2
1
PM H
3
1
'SUNDAY AM H
2
1
PM H
2
1
H
2
1
W
2
1
H
2
1
H
2
1
"H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
W
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
W
2
1
H
2
1
H
2
1
"H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
3
1
H
2
1
H
3
1
H
2
1
W
1
1
H
2
1
H
2
1
H
3
1
H
3
1
H
2
1
LOCATION CODES: H=HOME
MICROENVIRONMENT CODES:
3=TRANSPORT VEHICLE
W=MORK
1=INDOORS AT WORK
4=TRANSPORT OTHER
2=INDOORS - OTHER
5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 3—SALES WORKERS SUBGROUPS PCT IN SUBGROUP:24
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 12 3456789 10 11 12
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
5
2
SUNDAY AM H
2
1
PM H
5
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
3
1
H
2
1
H
2
1
H
2
1
H
5
2
H
3
1
H
2
1
H
2
1
H
5
2
H
2
1
H
5
2
H
3
1
H
2
1
H
2
1
H
5
2
H
2
1
H
5
2
H
3
1
LOCATION CODES: H-HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE H=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 3—SALES WORKERS SUBGROUPS PCT IN SUBGROUP:31
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 1 2 3 4 5 6 7 8 9 10 11 12
WEEKDAYS AM H
2
1
PM W
1
1
SATURDAY AM H
2
1
PM H
3
1
SUNDAY AM H
2
1
PM W
1
1
•• • •• ^ — •<—•••••• — • — — ^j. — — — ••
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
W
1
1
•K « V « W •
H
2
1
W
2
1
H
2
1
H
2
1
H
2
1
W
1
1
• •• ™» ™
H
2
1
W
3
1
H
2
1
H
2
2
H
2
1
W
3
1
H
2
1
W
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
W
3
1
H
2
1
H
2
1
H
2
1
W
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
W
1
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
2
W
1
1
H
2
1
H
3
1
H
2
1
W
1
1
H
2
1
W
1
1
H
2
1
W
1
1
H
2
1
W
1
1
H
2
1
W
1
1
H
2
1
W
1
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: H—CLERICAL WORKERS SUBGROUP:1 PCT IN SUBGROUP:17
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 1 2 3 4 5 6 7 8 9101112
WEEKDAYS AM W
1
1
PM H
2
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
pri H
3
1
W
1
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 1=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 4—CLERICAL WORKERS SUBGROUPS PCT IN SUBGROUP:70
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM W
1
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
PM W
2
1
H
2
1
W
1
1
H
2
1
H
2
2
H
2
1
W
1
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
W
3
1
H
2
1
H
2
1
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
W
3
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
3
1
H
2
1
H
3
1
H
2
1
HOUR
12
W
1
1
H
2
1
H
2
1
H
2
1
W
3
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: l» — CLERICAL WORKERS SUBGROUPS PCT IN SUBGROUP: 13
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
HOUR
12
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 1=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 5—CRAFTSMEN G FOREMEN SUBGROUP:! PCT IN SUBGROUP:33
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM W
1
2
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
PM W
1
1
H
2
1
U
1
1
H
2
1
H
3
1
H
2
1
U
1
1
H
2
1
U
1
1
H
2
1
H
5
1
H
2
1
U
1
1
H
2
1
U
1
2
H
2
1
H
5
1
H
2
1
U
1
2
H
2
1
U
3
1
H
2
1
H
2
1
H
2
1
U
1
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
2
U
1
1
H
2
1
U
1
2
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
HOUR
12
U
1
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
LOCATION CODES: H=HOME
MICROENVIRONMENT CODES:
3=TRANSPORT VEHICLE
ACTIVITY LEVELS: 1=LOW
W=WORK
1=INDOORS AT WORK
«4 = TRANSPORT OTHER
2=MEDIUM 3=HIGH
2=INDOORS - OTHER
S=OUTDOORS
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 5—CRAFTSMEN £ FOREMEN SUBGROUPS PCT IN SUBGROUP'27
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
MEEK OF DAY 1 2 3 1 5 6 1 8 91011
WEEKDAYS AM H
2
1
PM U
1
2
SATURDAY AM H
2
1
PM H
3
1
SUNDAY AM H
2
1
PM H
3
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
U
1
2
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
U
5
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
U
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
U
3
1
H
2
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
1
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
; = = = =
HOUR
12
W
1
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
LOCATION CODES: H=HOME M=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE ^'TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOM 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT? ALL
A-0 GROUP: 5--CRAFTSMEN £ FOREMEN SUBGROUPS PCT IN SUBGROUP:f0
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
5
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
H
5
2
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
3
1
H
2
1
H
3
1
H
5
2
H
2
1
H
2
1
H
5
1
H
2
1
H
5
1
H
5
1
H
2
1
H
2
1
H
5
2
H
2
1
H
5
1
H
5
2
H
2
1
H
2
1
H
5
2
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
HOUR
12
H
5
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE t=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 6—OPERATIVES -LABORERS SUBGROUP:1 PCT IN SUBGROUP:12
===============================================================.
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM W
1
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
W
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
2
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
HOUR
12
W
1
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 6—OPERATIVES -LABORERS SUBGROUPS PCT IN SUBGROUP:23
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM H
5
1
SATURDAY AM H
2
1
PM H
1
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
H
5
1
H
2
1
H
1
2
H
2
1
H
2
1
H
2
1
H
5
3
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
5
3
H
3
1
H
2
1
H
2
1
H
5
1
H
2
1
H
3
1
H
1
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
1
2
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
1
1
H
2
2
H
2
1
H
2
1
H
5
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
1
2
H
2
1
H
2
1
H
2
1
HOUR
12
H
5
2
H
2
1
H
1
1
H
3
1
H
2
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 6—OPERATIVES -LABORERS SUBGROUPS PCT IN SUBGROUP:65
================================================================
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
MEEK OF DAY 1 2 3 H 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM W
1
1
SATURDAY AM H
2
1
PM H
5
1
SUNDAY AM H
2
1
PM W
1
1
H
2
1
W
1
2
H
2
1
H
5
1
H
2
1
U
1
1
H
2
1
U
1
1
H
2
1
H
5
2
H
2
1
U
1
2
H
2
1
U
1
2
H
2
1
H
5
3
H
2
1
W
1
2
H
2
1
U
3
1
H
2
1
H
3
1
H
2
1
U
1
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
U
3
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
H
5
2
H
2
2
H
3
1
H
2
1
U
1
2
H
2
1
H
5
3
H
2
2
U
1
1
H
2
1
U
1
1
H
2
1
H
5
1
H
2
2
U
1
1
H
2
1
HOUR
12
U
1
1
H
2
1
H
5
1
H
3
1
W
1
1
H
2
1
LOCATION CODES: H=HOME M=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 8—SERVICE £ HOUSEHOLD SUBGROUPM PCT IN SUBGROUP:13
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL
MEEK OF DAY 12315678910
WEEKDAYS AM H
2
1
PM W
3
1
SATURDAY AM H
2
1
PM H
5
2
SUNDAY AM H
2
1
PM H
2
1
H
2
1
«
U
5
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
H
1
2
H
2
1
H
2
1
H
2
1
W
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
U
3
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
1
2
H
2
1
H
3
1
U
5
1
H
2
2
H
2
1
H
3
1
H
2
1
H
2
1
U
1
1
H
2
1
H
3
1
H
5
1
H
2
1
H
2
1
U
1
2
H
2
1
H
1
1
H
5
2
H
2
1
H
2
1
• = — —
BY
1 1
U
5
2
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
V « « W
HOUR
12
U
5
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 8—SERVICE £ HOUSEHOLD SUBGROUPS PCT IN SUBGROUP:29
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 23456789 10 11
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM U
1
1
SUNDAY AM H
2
1
PM U
1
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
U
1
1
H
2
1
H
2
2
H
2
1
U
3
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
U
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
2
U
1
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
U
1
1
H
2
1
U
1
1
H
2
1
HOUR
12
H
3
1
H
2
1
U
1
2
H
2
1
U
1
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 8—SERVICE £ HOUSEHOLD SUBGROUPS PCT IN SUBGROUP:58
=================================================::==================
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 1 2 3 4 5 6 7 8 9 10 11 12
WEEKDAYS AM H
2
1
PM W
1
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
3
1
H
2
1
H
3
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
t
H
3
1
H
2
1
U
1
2
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
U
1
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
U
1
1
H
2
1
H
2
2
H
2
1
H
3
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
H
2
1
U
3
1
W
3
1
H
3
1
H
2
1
H
2
1
H
3
1
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
2
U
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
U
1
2
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 9—HOUSEWIVES SUBGROUPS PCT IN SUBGROUP:20
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
3
1
SUNDAY AM H
2
1
PM H
2
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
HOUR
12
H
2
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICTS ALL
A-0 GROUP: 9—HOUSEWIVES SUBGROUPS PCT IN SUBGROUP:2H
====================================================================
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
MEEK OF DAY 1 2 3 4 5 6 7 8 9101112
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME
MICROENVIRONMENT CODES:
3=TRANSPORT VEHICLE
U-UORK
1=INDOORS AT WORK
4=TRANSPORT OTHER
2=INDOORS - OTHER
5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 9—HOUSEWIVES SUBGROUPS PCT IN SUBGROUP:56
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 1 2 3 4 5 6 7 8 9101112
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME W=UORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP:10—UNEMPLOYED £ RETIRED SUBGROUP:! PCT IN SUBGROUP: 8
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 1 2 3 4 5 6 7 8 9101112
WEEKDAYS AM H
2
1
PM W
1
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
W
3
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
W
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME
MICROENVIRONMENT CODES:
3=TRANSPORT VEHICLE
W=WORK
1=INDOORS AT WORK
4=TRANSPORT OTHER
2=INDOORS - OTHER
5=OUTDOORS
ACTIVITY LEVELS: 1=LOW
2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP: 10—UNEMPLOYED £ .RETIRED SUBGROUPS PCT IN SUBGROUP:«*2
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 1 2 3 4 5 6 7 8 9101112
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
PM H
3
1
H
2
1
u
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
3
1
H
2
1
H
3
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
3
1
H
2
1
H
3
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
S
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
5
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP:10—UNEMPLOYED £ RETIRED SUBGROUPsS PCT IN SUBGROUP'50
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL
WEEK OF DAY 1 2 3 4 5 6 7 8 9101
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
2
1
SUNDAY AM H
2
1
PM H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
BY
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
HOUR
12
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME U-UORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE H=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP:11—CHILDREN UNDER 5 SUBGROUP:1 PCT IN SUBGROUP:41
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
MEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM H
5
1
SATURDAY AM H
2
1
PM H
4
1
SUNDAY AM H
2
1
PM H
5
1
H
2
1
H
3
1
H
2
1
H
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP:11—CHILDREN UNDER 5 SUBGROUP'2 PCT IN SUBGROUP:39
============================================_==================:
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 234567891011
WEEKDAYS AM H
2
1
PM H
3
1
SATURDAY AM H
2
1
PM H
5
2
SUNDAY AM H
2
1
PM H
2
1
H
2
1
H
5
1
H
2
1
H
5
2
H
2
1
H
5
3
H
2
1
H
4
2
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
3
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1
HOUR
12
H
1
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP:11—CHILDREN UNDER 5 SUBGROUPS PCT IN SUBGROUP: 20
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 1 2 3 H 5 6 7 8 9101112
WEEKDAYS AM H
2
1
PM H
2
1
SATURDAY AM H
2
1
PM H
2
2
SUNDAY AM H
2
1
PM H
H
3
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
H
3
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
5
2
H
2
1
H
5
1
H
2
1
H
5
2
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
5
2
H
2
1
H
5
3
H
2
1
LOCATION CODES: H=HOME U=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE? 1 RESIDENT DISTRICT: ALL
A-0 GROUP:12—CHILDREN 5 TO 17 SUBGROUP:! PCT IN SUBGROUP:36
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
WEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM H
n
1
SATURDAY AM H
2
1
PM H
5
2
SUNDAY AM H
2
1
PN H
5
2
H
2
1
H
4
3
H
2
1
H
5
1
H
2
1
H
5
3
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
5
3
H
2
1
H
5
1
H
2
1
H
5
1
• •" ™ •••"•"•
H
2
1
H
5
2
H
2
1
H
5
1
H
2
1
H
2
1
w w «••
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
•» «• • •
H
2
1
H
2
1
H
2
1
H
5
2
H
2
1
H
2
1
H
1
1
H
4
2
H
2
1
H
5
2
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
1
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
HOUR
12
H
1
1
H
2
1
„
5
*
H
2
1
H
2
1
H
2
1
LOCATION CODES: H=HOME
MICROENVIRONMENT CODES:
3=TRANSPORT VEHICLE
W=WORK
1=INDOORS AT WORK
4=TRANSPORT OTHER
2=INDOORS - OTHER
5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT: ALL
A-0 GROUP:12—CHILDREN 5 TO 17 SUBGROUPS PCT IN SUBGROUP:30
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY
MEEK OF DAY 1 2 3 4 5 6 7 8 91011
WEEKDAYS AM H
2
1
PM H
5
2
SATURDAY AM H
2
1
PM H
5
1
SUNDAY AM H
2
1
PM H
5
2
H
2
1
H
1
1
H
2
1
H
5
2
H
2
1
H
5
3
H
2
1
H
1
1
H
2
1
H
5
2
H
2
1
H
5
3
H
2
1
H
3
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
1
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
5
3
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
= = = = =
HOUR
12
H
1
1
H
2
1
H
5
1
H
2
1
H
5
2
H
2
1
LOCATION CODES: H=HOME M=WORK
MICROENVIRONMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE 4=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
ACTIVITY PATTERNS BY AGE-OCCUPATION SUBGROUP
CLIMATE ZONE: 1 RESIDENT DISTRICT? ALL
A-0 GROUP:12—CHILDREN 5 TO 17 SUBGROUPS PCT IN SUBGROUP'S^
DAY OF TIME LOCATION/MICROENVIRONMENT/ACTIVITY-LEVEL BY HOUR
WEEK OF DAY 1 2 3 4 5 6 7 8 9101112
WEEKDAYS AM H
2
1
PM H
1
1
SATURDAY AM H
2
1
PM H
4
2
SUNDAY AM H
2
1
PM H
5
1
H
2
1
H
1
1
H
2
1
H
5
2
H
2
1
H
5
1
H
2
1
H
1
1
H
2
1
H
5
1
H
2
1
H
5
1
H
2
1
H
3
1
H
2
1
H
5
3
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
3
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
2
1
H
1
1
H
2
1
H
5
2
H
2
1
H
3
1
H
2
1
H
1
1
H
2
1
H
5
2
H
2
1
H
2
1
H
2
1
H
1
2
H
2
1
H
2
1
H
2
1
H
5
1
H
2
1
LOCATION CODES: H=HOME W=WORK
MICROENVIROHMENT CODES: 1=INDOORS AT WORK 2=INDOORS - OTHER
3=TRANSPORT VEHICLE H=TRANSPORT OTHER 5=OUTDOORS
ACTIVITY LEVELS: 1=LOW 2=MEDIUM 3=HIGH
-------
APPENDIX C
ESTIMATION OF COHORT LOCATION
AND POPULATION
C-l
-------
ESTIMATION OF COHORT LOCATION
The cohort location algorithm is an editing program which
expands a small dummy file into a large cohort location file
(COHORT-CITY). A sample of the dummy file is exhibited in
Figure .C-l. The partial cohort identification includes home
district/ age-occupation (A-0) group,, and activity pattern sub-
group. The work district is not presented, but a dummy variable
is used to indicate whether the subgroup stays in the home (H)
district or whether it moves to a work (W) district during a
12-hour period. These conditions were treated differently by
the cohort location algorithm. Six basic lines were required
to represent the location assignments for each of two 12-hour
periods in a typical weekday, a typical Saturday, and a typical
Sunday.
An editing routine is used to create the cohort location
file (Figures C-2 and £-3) . The Univac Editor reads each
activity pattern listed in the dummy file. If the work district
dummy variable shows that the subgroup stays in their home dis-
trict, the subgroup becomes a cohort by exchanging the home
district number for the 'H1. However, if the dummy variable
shows that the subgroup has a work district which may be dif-
ferent from the home district, then the subgroup is split among n
districts where n is the number of districts comprising the study
area. The six basic lines of the dummy file are duplicated for
each home district and work district combination. The H and W
codes are replaced by the appropriate district number. This
process creates a complete file that locates every cohort in a
district and in a microenvironment at an activity level for every
hour of the year.
-------
Dummy for
city name
Day-of-week
code
Cohort
ID
CITYXXXXH
(H042W
AM-PM 12-hr
code location assignments
'W - -T - * --
LI H21H21H21H21H21H21431W11W11W11W11W11
CITYXXXXH042W 12 W11W1]
Home
district
A-0 group-
Pattern—
(Subgroup
1-3)
Work district
>, > „,
W11W11W31H21H21H21H21H21H21H21
-Activity level code (2nd hr)
1—Microenvironment code (2nd hr)
Location assignment (2nd hr)
Work(W) or Home(H)
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 £-1- Sample contents of cohort dummy file.
-------
w*JVtR/RN9 RUN-lDiACCOUKT-MO. ,STDS
*ST^ PRINT*, ,PR
£ = !>,U COHORT-CITYX.
04 BRIEF
LUIT L 15 15
LMIT C 9 55
1?3
L3D? 11
DITTO 1 1=5
-1?7
LPSU3 Li S
C /H/J/ 193 G
-1 >7
L33P 13!
L «
M 5
LDD* 1! 2
DITTO X4 X3
-5
L^S'JS L2»J
C frf/S/ 6 S
L>r»
-------
1:CITYXXXXHD11H
2:CITYXXXXH311H
3:CITYXXXXHD11H
;:C!TYXXXXH01lH
5:CITYXXXXHD11H
i:CITYXXXXHD1lK
7:CITYXXXXHD12H
3:CITYXXXXH012H
?: CITYXXXXtOI 2H
1D:CITYXXXXH^12H
11:CITYXXXXKD12H
12:CITYXXXXH012H
13:CITYXXXXH313H
14:CITYXXXXHD13H
15:CITYXXXXH!D13H
15:CITYXXXXH313H
17:CITYXXXXHD13H
15:C1TYXXXXH313H
19:CITYXXXXHC21W
21:CI
22:CI
23:CI
24:CI
25:CI
25:CI
27:CI
2S:C!
2?:CI
33:CI
32:CITYXXXXH023tf
33:CITYXXXXH023W
34:CITYXXXXH023tf
35:CITYXXXXH023tf
35:CITYXXXXHD2IM
37: CITYXXXXH031W
33:CITYXXXXHD31«
39:CITYXXXXHD31W
tD:CITYXXXXHD31tf
41:CITYXXXXH331H
A2:CITYXXXXHD31W
43:CITYXXXXH032H
A4:CITYXXXXH332H
45:C-ITYXXXXHQ32H
A5:CITYXXXXH332H
47:CITYXXXXHD32H
43:CITYXXXXH032H
TYXXXXH021U
TYXXXXHD21rf
TYXXXXH221W
TYXXXXHD21W
TYXXXXH022W
TYXXXXhD22tf
TYXXXXHD22tf
TYXXXXHD22tf
TYXXXXHD22W
TYXXXXHD22tf
5D:CITYXXXXH033tf
51 :CITYXXXXH033W
52:CITYXXXXHD33ii
12
21
12
31
32
11
12
21
22
31
3?
11
12
21
22
31
32
11
12
21
22
31
32
11
12
21
22
31
32
11
12
21
22
31
32
11
12
21
22
31
32
11
12
21
22
31
32
11
12
21
22
H21H21
H11H11
H21H21
H11H11
H21H21
H51H51
H21H21
H21H21
H21H21
H21H21
H21H21
H31H21
K21H21
H21H31
H21H21
H31H21
H21K21
H21H21
H21H21
H21H21H21H21H31H21H21
H11H11H11H11H31H21H52
H2lH2lH21H2lH2lH21H2l
HHHllH3lH2lH5lH21H22
H21H21H21H21H21H21H21
H21H21H21H21H21H21H21
H21H21H21H21H21H21H21
H21H21H21H31H11H11HT2
K31H21H21H21H21H21H21
H31H21H21H22H31H21H21
H21H21H21H21H21H21H21
H21H21H31H21H21H21H21
H21H21H21H21H21H?1H31
H21H21H21H21H21H21H21
H21H21H21H21H21H21H21
H2lH22H31H21H21H21H21
H21H21H21H21H21H21H21
H21H21H21H21H21H21H21
H21H21H21H21H21H21H22
H31H11H11
H52H21H21
H2lH11ri11
H21H2iH21
H21H21H21
H21H21H31
H3.1HE1H21
H12H11H31
H21H21H52
K21H21H21
H2THZ1H21
H21H21H21
H21H21H21
H21H21H21
H21H21H21
H2iH2lH21
H21H21H21
H31H21H21
H21H21H31
H21H21H21
H52H52H52
H21H21H21
H51H51H51
H2lH2lH2l
Ul1tf11Ul1
H21H21H21
H21H21H21
H21H21H21
H21H21H21
H2lH2lH2l
H21H21H2
H21H31H2
H21H21H2
H21H21H2
H?lH2lH3
u3lH3lH2
H21H21H2
H31H21H2
H21H21H2
H21H22H2
H2tH2lH2
H21H21
H31H22
H21H21
H21H31
K2lH2l
Ul1Ul1
HZ1H21
H22H21
H21H21
'H21H21
H2lH2l
H21H21H21H21H21H21H21H31W11W11H11W11
Wl1tf11Wl1Wl1tf3lH2tH2lH21H2lH2lH2lH21
HZ1H21H21H21H21H21H21H21H21H21H21H31
H5lH5lH5lH5tH31Wl1Wl1y3lH2lH2lH2lH21
1H21
1H21
1H21
1H21
lWl1
lH2l
1H21
1H31
1H21
1H21
lUl1
H21H21H31
H31H21H21
H21H21H21
H21H31H21
tf11tf11Ul1
H2lH2lH21
H21H21H21
H21H21H21
H21H21H21
H21H21H21
Wl1Wl.1y11
H21H21
H31H21
H21H?1
H21H21
K21H21
H21H21
H21H21
H52H52
H21H21
H51H21
H21H21
Wl1Wll
H2lH21
W31H21
H21H2
H21H2
H21H2
H31H3
H21H2
H21H2
H21H2
H52H5
H21H2
H31H3
H21H2
W21tf3
H2lH2
H21H2
1H21H21
1H21H21
1H21H21
1HZ1H31
1M21H21
1H21H21
1H21H21
2H21H21
1H21H21
1H31H31
1B21H21
1W21w3l
1H2lH2l
2H21H21
H21H21H21
H21H21H21
K21H21H21
H21H21H21
H21H21H21
H21H21H21
H21H21H21
H21H21H21
H21H21H21
H21H31H31
H21H21H21
H2lH2lH3l
H2lH31tf11
H21H21H21
H21H31H21
H21H21H31
H21H31H31
H21H21H21
K31H21H21
H21H21H21
H21H52'H52
H21K21H21
H52H52H52
H31H31H31
Wl1w11d11
H22H2lH21
WllWl1*11
H21H21H21
Figure C-3. Cohort dummy file
(partial).
-------
ESTIMATION OF COHORT POPULATION
The cohort population algorithm requires as inputs the pop-
ulation of each age-occupation (A-O) group in each exposure dis-
trict and the fraction of each A-O group that may be assigned to
each of three alternative activity patterns. This information
is available from the TAP file, previously described in Section 4.
These fractions, together with the A-O group populations, were
assembled into an input file, called SUBPOP (not included in
this appendix because of its length). The SUBPOP file is simi-
lar in appearance to the dummy file shown in Figure C-l, except
that location assignments are replaced by a number representing
the population of the age-occupation group., In the place of the
day and time codes is the percent allocation of population to
subgroup.
Home-to-work trip tables were developed (see Table C-l) which
listed the number of work-related trips taken from a district of
residence to a district of employment, and back. The trip data
were expressed in person-trips (rather than vehicle trips) and
in origin-destination (0-D) format. Non-home based trips and
non-work trips could not be accommodated in a simple two-way
model and were not used. A separate computer file was estab-
lished for the trip-table for each city.
A simple Fortran routine (see Figures x>4 and C-5) produces
a cohort population file from the two input files. The program
first reads in the trip table and then reads the A-O group data
listed in the input file SUBPOP-city. If the work district dummy
variable shows that the subgroup stays in their home district,
the cohort population is simply the A-O group multiplied by the
fraction of the group in the subgroup. However, if the dummy
-------
TABLE C-l. HOME-TO-WORK TRIP TABLES
Home
District
Work District
1
2
3
4
5
6
7
8
Chicago
1
2
3
4
5
6
7
8
29929
74519
30101
12268
53602
19344
11250
10656
4894
29636
7743
684
5333
3810
1885
1020
4493
9591
14766
2625
6288
831
1017
1755
889
592
1350
3397
4017
73
40
353
1445
- 916
958
1930
16316
242
28
122
614
6540
884
151
1604
4432
754
189
1013
2416
1229
191
1090
566
1096
606
1031
1754
1720
704
1231
271
739
1077
Philadelphia
1
2
3
4
5
6
270524
34422
65791
70374
61761
23862
34422
59618
16528
4404
2288
2156
65791
16528
52154
20977
7627
3652
70374
4404
20977
100450
30482
3279
61761
2288
7627
30482
169931
14911
23862
2156
3652
3279
14911
149034
0
0
1 - o
0
0
0
0
0
0
0
0
0
St. Louis
1
2
3
4
5
6
7
14139
26518
10687
25310
11031
18225
12114
16516
20403
9313
8941
3431
10657
3062
10687
9313
15332
12119
6211
16146
2122
25311
8941
12118
64828
10230
13487
3330
11031
3431
6211
10230
61327
19396
1207
18225
10656 ..
16146
13485
19396
141535
3433
12114
3062
2123
3330
1208
3433
36219
. 0
0
0
0
0
0
0
Los Angeles
1
2
3
4
5
6
7
24697
10607
30662
3159
12534
374
812
-
4959
54720
6972
' 670
12392
111
30
8954
7933
89090
6404
6300
595
115
_L
465
351
3526
19719
3209
2787
198
4205.
8748
4106
4653
122655
909
188
28
31
177
2107
396
7700
907
3
8
27
122
88
1196
10932
0
0
0
0
•o
0
0
-------
variable shows that the subgroup has a work district which may
be different from the home district, then the subgroup -is split
among n work districts where n is the number of districts com-
prising the study area. The trip table was used to allocate the
subgroup population among the districts. Since trip table data
were not available for different A-O groups, it was assumed that
each A-O subgroup makes the same number of trips to any district
and has the same distribtuion of trips over all districts as
every other subgroup. Therefore, the ratio of trips to a certain
work district to the total trips to all districts (t /T) is the
same for all cohorts who live in the same home districts and
commute to the same work district. Thus, we can summarize the
calculation for estimating each cohort population as follows:
PHASW
•w
where H = home district,
A = age-occupation category
S = subgroup specified by activity pattern
W = work district
P = population of a cohort which is defined by the
HASW subscripts
P _ = population of an age-occupation category in a
given home district
F = fraction of persons allocated to subgroup s
S
tH_w = number of trips by all A-O groups from a home
district to one work district
TH = total number of trips by all A-O groups from
. a home district to all work districts.
The program writes the resulting cohort data into a cohort popu-
lation file in the format requested by ASB. When completed, the
cohort location and the cohort population files are combined into
one file and sorted for use in the NEM program.
-------
iSJ^P/RMB RUN-ID,ACCOUNT-NO,STDS,5
2SY* PRINTS,,PR
2AS5,A RAP*PR.
SFDR.N RAP*PR.COHPOP,TPFJ.COHPOP
IE:F
2
-------
£ ***** PROGRAM COHPOP (COHORT POPULATION) ******************
C
C EVTER NTD---NO.OF TRANSPORTATION DISTRICTS ACTUALLY USED.
C EVTER NSUB --- NO. OF LINES IN THE SUBPOP-XXX. FILE.
C ENTER CITYA I CITYB --- CITY S'AME IN EISHT CHARACTERS.
C
(,*****»*****************************************«***********•*
PARAMETER NTD=7
PARAMETER NSUB=2S1
1VTE3ER TDC9), ROW, COL, CITY A, CITYS
CITYA='L. A.'
CITY3='
INTEGER HTD,AOGRP,AOSU3,KyORK,SU8PCT,AOPOP
INTESER HOME, WORK, COHPOP
DIMENSION TABL£<9,9), ROWTOT(NTD)
30 2 HTD=1,NTD
DO 2 ICTD=1,9
I TABLE(HTD, KTD) =0
^
C,
C***** ^EAD IN THE TRIP TA3LE FOR THE CITT. ***
C
DO ID J=1,NTD
SEAD(5,5) (TD(K),<=1 ,9)
5 FDR*AT<9(I6,1 X),17X)
ROWTDT(ROW) = FLOAT+ TD (6
DO 10 1=1,9
COL^I
TA5LE(ROW, COL) = TD(I)
1D CONTINUE
*** READ IV THE A-0 SUBGROUP DATA, PERFORfl CALCULATIONS,
AND PRINT THE RESULTS IN A1 OUTPUT FILE.
11 READ(5,12,E^D=999)HTD,AOSRP, AOSUB,KVOR<,SUBPCT, AOPOP
1Z FOR*.AT(I1,12,I1,A1,1X,I2,1X,I6,55X)
IF (iCWORK.EB.'H') GO TO 25
,RDiJ = HTD
DD 1? KTD=1,STD
COL = KTD
TRIPS = TABLE (ROW, COL)
TTRIPS= ROKTOT(ROW)
SU£TPOP= (FLOAT(SUBPCT)/1DO)*FLOAT(AOPOP)
REAL? = (TRIPS/TTRIPS)*SUBPOP
COHPOP= IFIXCR.EALP)
=
HOME = HTD
fc'ORK = KTD
.'RITE(10,14)CITYA,CITY3,HOHE,AOSRP, A OSUB, WORK., COHPOP
1^ FDRMATdH ,T1 ,A4,A4-,J1, J2,J1,J1,'OOD3*,1X, J6)
19 CONTINUE
30 TO 11
Figure G-5. Source code for population program.
-------
25 5J3DOP= (FLDAT(SU3PCT)/1DO)*FLOAT(AOPOP)
CDHP3P= IFIX(SUBPOP)
rtOSE = H^O
JRITE(1j!l<»)CITY4|ClTY3thOKEt*06RP«AOSU3iHOMEfCOHPOP
53 TO 11
?99 BSD
Figure C-5 (continued)
-------
APPENDIX D
COMPUTER PROGRAM USED IN CALCULATING
POPULATION CENTROIDS
D-l
-------
.************** PROGRAM CTDS ICENTROIDS) **- ____ ............ --*****
C
C PROGRAM CTDS READS A FILE CONTAINING A LIST OF CENSUS TRACTSvCT'S)
C rfHICH COMPOSE THE TRANSPORTATION D ISTRI CTS (TD *S ) WITHIN AN SwSA.
C THEN IT READS A FILE CONTAINING A LIST OF CT'S, THEIK ASSOCIATED
C POPULATIONS, AND THE X-Y COORDINATES OF THE POPULATION CENTRUIDS . ._
C OF ENUMERATION D ISTRICTS (ED 'S > WITHIN THE CT.
DATA, THE COORDINATES OF THE POPULATION CENTROIDS _ ____
_ __ _
C OF THE TRANSPORTATION DISTRICTS ARE CALCULATED AUD LISTED
C __ _JfHIS VERSION OF THE PROGRAM ACCESSES FILE SASD*MEDX FROM 6* OUGGAN _
C -WHICH IS A COMPACT FILE CREATED BY S.A.I. INC. THc OTHER FIuES
C __ DEFINE THE CT^S COMPOSING EACH TO (TD -CT-CHIC . , TD-CT-PHI LA ., __
C TD-CT-LA.tTD-CT-STLOU.)
C ____ THE FOLLOWING PARAMETERS MUST BE PUT INTO THE PROGRAM: __
c _ _ _... .. .. ._
C CITYA THE CITY NAME (6 CHARACTERS IN PARENS.) FOR TABLE HEADING
C ~ CITYB '" SECOND HALF OF CITY NAME C 6 CHARACTERS IN PARENS.-).
C __ _J!A_XTJ>^ ___ THE NUMBER OF TRANSPORTATION DISTRICTS DEFINED F0« THE
C CITY AND WHICH ARE USED IN THE FILE TO -CT -XXXXX .
C MAXCT THE LARGEST NUMBER OF CENSUS TRACT FOR A SINGLE Tu .
C STATE1 A STATE IN WHICH THE CITY IS LOCATED (FIPS C ODE—1 N TEG Ert ) .
C STATE2 A SECOND STATE OF INTEREST. _
C~ CfcfYl A COUNTY IN WHICH~1 OR MORE TD IS LOCATED (Fl'PS COa E --INIE GER) ,
C CNTY2 A SECOND COUNTY OF INTEREST.
C BLAT THE MINIMUM LATITUDE FOUND IN THE CITY ,
c BLONG THE MINIMUM LONGITUDE FOUND IN THE CITY
C A RUNSTREAM FOR THIS PROGRAM MAY BE FOUND IN STDS*UTIL.CTDS.
C
C INITIAL ASSIGNMENTS FOLLOW:
c ****************** ******* *************************************************
PARAMETER CITYA ='PriILAD'
PARAMETER CITY& ='ELPHIA'
PARAMETER STATE1 =42
PARAMETER' STATEZ =34
PARAMETER CNTY1A =091
PARAMETER CNTY1B =1u1
PARAMETER CUTY1C =045
PARAMETER CNTY1D =017
PARAMETER CNTY2A =007
PARAMETER CNTY2B =005
PARAMETER fAXTD =7
PARAMETER MAXCT =157
BLAT = 39.0000
BLONt>= 74.000C
INTEGER TD, CT, RECORD, RTYPE
INTEGER NCTCMAXTD), POP(MAXTD), IDNCT(KAXTD,MAXCT3
DOUBLE PRECISION SUMLAT (MAXTDJ, SUMLNG(MAXTD )
DO 5 TD=1,MAXTD
POP(TD)=0
NCT(TD)=C
SUMLAT(TD)=0
SUMLM6(TD)=C
DO 5 CT=1,MAXCT
IDNCT(TD,CT)=0
5 CONTINUE
-------
DATA JNUM/0/
C***** REAO IN THE CT'S WHICH DEFINE THE TD'S.
C
10 KEADC5,11,END=20) JTD.JCT
11 FORMAT(II,IX,14)
_TD=JTD
NCT(TD)= *NCT(TD)+1
CT =_NCT(TD) _
IfiNCT(TDtCT) = /CT
60_To_ic___
c "" ~ " ""
C***** READ IN A RECORD TO LOOK FOK THE STATE OF INTEREST.
C
20 READC10,21,ERR=51,END=8C)RTYPE,1 STATE
"21 FOR MAT(A 1,1X ,12 , 2 4 X , 5 2 X )
22 IF (ftTYPE.NE.'T) 60 TO 2C
IF (CISTATE.NE.STATE1>.AND.(ISTATE.NE.STATE2J) GO TO 2C
C
C***** IN STATES OF INTEREST, LOOK FOR THE COUNTY (S) OF INTEREST
C _..
30 READC10,31,ERR=51,ENC=80)RTYPE,ICNTY
31 FORMAT (A1,I3,24X>52X)
32 IF
IF
IF
IF
IF
IF
IF
60
(RTYPE. NE.'U') 60 TO 30
( (I STATE. EQ. STATED. AND.
( CI STATE. E Q. S TATE 1). AND.
(( I STATE .EQ. STATED. AND.
(CI STATE. EG. STATE D.AND.
(( I STATE. EQ. STATE 2). AND.
((ISTATE.EQ.STATEZ).AND.
TO 30
(I
(I
(I
(I
(I
(I
CNTY
CNTY
CNTY
CNTY
CNTY
CNTY
.EQ
.EQ
.EQ
.EQ
.EQ
.EQ
•
•
•
•
•
•
CNTY1A
CNTY1B
CNTY1C
CNTY1D
CNTY2A
CNTY26
)
>
5
)
)
)
)
)
)
)
)
)
60
60
60
60
60
60
TO
TO
TO
TO
TO
TO
33
33
33
33
33
33
33 JNUM=0
C
C
C
C
* -.
** FOR DATA RECORDS THAT
DETERMINE WHETHER THE
OF THE DEFINED TD'S.
ARE
CT
FR
IN
OM
THI
A
S
STATE
RECOR
AND C
D MATC
OUNTY
HES A
OF
CT
INTEREST
IN ANY
i
40 READC10,41,ERR=51fEND=80)RTYPE,1CNTY,1CT,IEDPOP,I LONG,1LAT
41 FORMAT (A1,13,13X,l4,1eX,15 ,1_7 ,i6 ,23X
IF CftTYPE.NE." '> 60 TO 51 "
142 R£CORD=*I-DATA*
JNUM=JNUM+1
IF (JNUM.6T.1) 60 TO 43
WRITE (6, 42")"
42 FQRMATC1H ,'FIRST LINE OF DESIRED *,*DATA
43 TD=C
44 TO=Tu+1
IF (TD.GT.MAXTD) 60 TO 4G
___ CT_=p ______
"45 CT = CI-H
__.-_ _
IF (CT.6T.NUMCT) "60 TO 44
KC_T»I DN C^T C T D , C T )
IF (ICT.E'Q.KCT) t.0 TO 60
60 TO 45
C
J***** FORHAT ERROR IN DATA RECORD (END OF COUNTY OR STATE)
C
-------
51 R£ADCQ,21,tRR = 52,END=iiQ)RTYP£ ,1 STATE
J»,UM=0
IF CRTYPt.EG.'1'j 60 TO 22
52 RcAD(Ot31fERR=53,END=oQ)RTYPE,:iCNTY
JNUM=0
IF (RTYPE.EQ.'O'i GO TO 32
53 R£AD<0,41fERR=57fEND=60)RTYPE ,1CNTY,ICT ,1EDPOP,ILOfcG,ILAT
IF_(ICNTY.EQ.CNTir1> GO TO 142 _
IF (ICNTY.EQ.CNTY2) GOTO 142
57 «RITt(6.58)
58 FORMATC1H ,'FJLE FORMAT ERROR" ~*V*ENCOUNTfcRED "*",*—S TOP PROGRAM,
SO T0_999^ ___
C
C***** FOR RECORDS IN A CT OF INTEREST, ACCUMULATE SUflS OF PRODUCT^
C" FOR LATITUDE AND LO NGITUDE , At = SU«LAT (
RLONS = FLOAT(ILONG)
XLONG = RLON&/10COO. - BLONG
IF (KLON6.LT.BLOKG) XLONG=0.0
PROD2 = XLONG*IEOPOP
SUHLMG(TD) = SUHLNG(TD) •> PRQQl
POP(TD> = POPCTO) + 1EDPOP
D ,1CT,ILAT,YLAT,S U M L A T(TD>
65 FORMATUH 111 V3X .14, 3X(I6 V3X , F6.4 V3X , F2 C.4>
60 TO 40 _ _
' c ~ " "
C***** CALCULATE LATITUDE S, LONGITUDE FOR EACH TO AND PRINT A TABLt
80 KCITY=CITYA
JCITY=CITYB
URIT£(6,200)KCIT YtJCITY
00 90 TD=1,MAXTD
ZLAT = SUKLAT(TD)/POPCTD) + SLAT
ZLONG = SUMLNG(TO)/POP(TD) + BLONG
WRITE(6,201)TD,ZLAT,ZLONG
90 CONTINUE
C***** PRINTING FORMATS FOLLOW:
C _ __ _
200 FORMATC1H1,'COORDINATES OF *,"POPULATION CENTROIDS ",
8'FOR TRANSPORTATION "/7X t "01 STRI CTS IN THE CITY_0_F_ *,2(Ao)//_
S55('=*>/* TRANSPORT. 6Ts TRICJ*,11X,'LATITUDE',7X,'LONGITUDc"
855C-'))
2C1 FORM^TdH ,8X ,11 ,19X , Fl 0.4 ,6x ,F1 0.4)
202 FORHATC1H ,57X)
999 "~
------- |