EPA/600/A-95/077
The Relationship Between Personal Exposures and Ambient
Concentrations of Particulate Matter
David T. Mage
Timothy J. Buckley
US Environmental Protection Agency
Atmospheric Research and Exposure Assessment Laboratory
Human Exposure and Field Research Division
Research Triangle Park, North Carolina 27711
ABSTRACT:
Studies of the correlation of ambient particulate matter
(PM) with mortality implicitly assume that the PM concentration
at an outdoor-ambient monitoring station is a surrogate for the
mean exposure to PM of people in the local community. If the
individuals' probability of mortality from PM exposure is
linearly proportional to their PM exposure, then the expectation
of total PM related mortality in the community is proportional to
the mean personal exposure in the community.
This paper briefly reviews the relationships between ambient
PM concentration and individual personal PM exposure in the
literature and evaluates the validity of the use of ambient PM
data as a surrogate for community PM exposure. The results show
that the surrogate relationship can be quite variable between
communities. In some communities an ambient concentration may be
a good surrogate for mean personal exposures, but in other
communities, ambient concentrations may not have a significant
influence on mean personal exposures.
These findings indicate that in a community, to be studied
prospectively for a relationship between mortality and ambient
PM, it may be useful to monitor personal exposures to PM to
evaluate how well an ambient monitoring station concentration
represents the average personal exposure to PM in the community.
Community monitoring of ambient particulate matter may capture a
substantial portion of the variance in group mean PM exposure in
cross-sectional epidemiological studies. If other sources of PM
exposure remain relatively constant over time, the same could be
true for time-series epidemiological studies.
Disclaimer: The information in this document has been funded
wholly or in part by the United States Environmental Protection
Agency. It has been subjected to Agency review and approved for
publication. Mention of trade names or commercial products does
not constitute endorsement or recommendation for use.
l

-------
A1235
INTRODUCTION:
Human exposure to particulate matter (PM), characterized by
ambient air concentrations of PM, has been associated with excess
morbidity and mortality in both the episodic conditions of the
London Fog of 1952 (Ministry of Health, 1954), and nonepisodic
conditions occurring in the modern USA (Dockery et al., 1993).
This latter occurrence is unexpected, and controversial, because
the ambient air quality, in the US locations that were studied,
meets the current 24-h PM-10 National Ambient Air Quality
Standard (NAAQS) of 150 M9/m3. In all such studies, the
measurements of a PM concentration at one or more stationary
ambient-outdoor monitoring stations (SAM) have been used as a
surrogate for the mean exposure to PM of the people residing in
the community under study.
Ambient air is defined in 40CFR as "that portion of the
atmosphere, external to buildings, available to the general
public". Consequently, while people spend time indoors (e.g. at
home) or on private property (e.g. at work), their personal
exposures to PM may not be regulated by an NAAQS. Measurements
of PM by a personal exposure monitor (PEM) worn by a subject can
include exposures while at home, at work, and in the "ambient
air". A source of PM in a home or at work, such as a plume of
tobacco smoke (ETS), directly influences the PEM of a person at
that specific location. The SAM will not be directly influenced
by that specific smoke plume. Rather, the PM SAM may pick up
traces of the ETS from millions of cigarettes smoked in the
community that escapes into the ambient atmosphere and which
contribute in their totality to the community PM emission
inventory as derived by source apportionment of the SAM PM. As a
result, the measured PM exposure of subjects who do not spend 24-
h per day in the "ambient air", may not be equivalent to the
concurrent SAM PM concentration measured in their community.
This paper discusses the relationships between PM PEM values
and PM SAM values and presents an analysis of how well an ambient
PM concentration can serve as a surrogate for community mean
exposure.
RELATIONSHIP BETWEEN PEM EXPOSURE AND SAM CONCENTRATION:
As defined by Ott (1990), an exposure to PM takes place when
a concentration of PM and a person coexist at a point (x,y,z) in
space at the same moment of time (t). The mass concentration of
PM at that moment is composed of a fraction (w) of materials of
outdoor origin that may be directly related to the SAM
concentration and a fraction (1 - w) of particles with
concentrations unrelated or indirectly related to the SAM
concentration. In general, while outdoors in the ambient air,
w » 1 (some local influences such as leaf burning may not be
picked up by the SAM). However, while indoors (or at work), w is
highly variable (0 < w < 1) depending on the air exchange rate
between indoor and ambient air, and the strength of the immediate
sources of PM indoors. When indoors-at-home, the subject is
2

-------
A1235
exposed to PM that entered the home from the outdoor ambient air
through doors, windows, and other miscellaneous openings in the
building envelope. Sources of PM indoors, such as cigarette
smoke, cooking, lint and house dust, make up the remaining (1-w)
fraction of the PM exposure.
The fraction w will be a variable for various PM size
fractions. For simplicity, we can consider that fine-particles,
smaller than 2.5 /im aerodynamic diameter (AD), readily penetrate
a building envelope and tend to persist, and that coarse-
particles (greater than 2.5 nm AD) penetrate it to a lesser
extent and tend to settle out. Consequently, one might expect a
PEM exposure to fine PM to be better correlated to a SAM ambient
fine PM concentration, than for PEM and SAM total PM measures.
RESULTS: RELATION OF INDIVIDUAL PERSONAL PM TO AMBIENT PM.
The literature was surveyed for published studies that
reported simultaneous observations of PM PEM and PM SAM. Table 1
shows the summary of 14 such relationships between PEM and
simultaneous SAM. The studies, at 10 US and 4 foreign locations,
were conducted with varying PEM/SAM sampler cut points (defined
as the AD with 50% penetration to the collection surface), and
varying averaging times. Nine of these data sets show a low
correlation between PEM and SAM, characterized by a value of the
regression R2 not significantly different from zero. In the 9
nonsignificant studies the values of R2 range from less than
0.005.(two	values of r were negative) to 0.15. In the other 5
studies, the significant R2 values range from 0.04 to 0.50, but
in 3 of these 5 studies there were seasons or averaging-times for
which the R2 value was not significantly different from zero
(r was negative in one case). In one study the R2 lost
significance when outliers were removed (Morandi et al., 1986).
This overall pattern implies that variations in ambient PM
may have small influence on the variations of individual PEM PM
exposures, perhaps due to the large amount of time people spend
indoors. In all 7 cases which also reported the R2 between PM
PEM and PM measured by a stationary indoor monitor (SIM), the
values are higher than the PEM/SAM value of R2. One study
(Sexton et al.f 1984) reported PEM/SIM R2 = 0.01, but the other 6
studies reported significant R2 values ranging from 0.25 to 0.90.
Discussion of Results:
The absence of a significant correlation between PEM and SAM
in most studies, has important implications for studies which
seek to develop an "ecological" relation between average measures
of human health and average ambient PM (SAM) concentrations in a
community or between communities (Spengler et al., 1985). The
relatively low correlation between PEM and SAM may result from a
combination of the following factors:
1.	Individual variabilities of activities that generate PM into
the nonambient settings (at home, at work, in traffic) where
people spend most of their time,
2.	Close proximity of the PEM to local sources of PM from
personal activities (within the undiluted plumes of smoking.
3

-------
A1235
cooking, etc.) that are not well represented by the SAM PM
measure, which is sometimes called a 'personal cloud effect', and
3. Spatial variability of ambient concentration of PM (SAM), as
when people upwind of a point source of PM are unaffected by it,
while those downwind of it are.
Spengler et al. (1985) stated that "Ambient concentrations
provide poor prediction of personal exposures to undifferentiated
respirable size particles. Air pollution epidemiological
investigations must consider the importance of indoor
environments in estimating subject exposures." If we assume that
PM generated indoors (1 - w) on a ng basis is as toxic as the PM
generated outdoors (w), studies that look at a time series of
individual health outcomes (e.g. a symptom diary) should look at
that individual's PEM PM exposure as the independent variable,
rather than a SAM PM concentration. These conclusions relating
to personal exposures of individuals are supported by the results
of the 9 studies that found no significant relation between PEM
PM and SAM PM.
Whereas the variation in any one individual's PM PEM may be
weakly influenced by variation in PM SAM (r or R* not
statistically significant from zero), the variation in the same
PM SAM may have a strong influence on the mean PEM PM exposure of
all the people in the community. Dr. Duncan Thomas (Personal
Communication, 1994) suggested a way to visualize this
relationship between mean personal exposure and ambient
concentration, and the results are presented in the following
section.
RESULTS: RELATION OF MEAN PERSONAL PM TO AMBIENT PM
If all N people in a community carried a PEM, the mean PEM
value is obtained by summing all PM PEM values and dividing by N.
If indoor and outdoor PM are equally toxic on a ng basis, then no
further information would be contained in PM SAM. The mean of a
random sample of PEM PM measurements on subjects in the community
would be an unbiased estimator of the actual community mean PEM
PM, and such a mean may be more appropriate for use than a SAM
measurement.
Figures la, 2a, 3a and 4a show the individual personal PEM
PM and the corresponding ambient SAM PM from four (4) studies
cited in Table 1 for which individual data were available. For
example, Figure la (Lioy et al, 1991) shows a set of PEM samples
obtained from 14 individuals on 14 consecutive days. Because
these 14 subjects were not selected as a probability sample from
the community of Phillipsburg, NJ, we treat their exposures as a
biased sample from the exposure distribution that we would
measure, had every person in the community carried a PM PEM. The
outdoor average is the mean of four (4) ambient SAM PM values
obtained on each day the people carried the PEM for PM-10. The
data plot in a vertical profile corresponding to up-to-14 valid
PEM values obtained on that day. In the regression, each point
is weighted equally, and the R2 value of 0.037 (p = 0.008) would
be significant if the sample were unbiased. In Figure lb, the
4

-------
A1235
mean of the daily PEM values is plotted. The R3 now has
increased from 0.037 to 0.333 (p =0.031). The interpretation is
that, on the average, variation in SAM only explains on the order
of 4% of the variation in an arbitrary individuals PEM, but that
same variation in SAM explains 33% of the variation in their mean
exposure.
Because both these PEM and SAM values are measured with
error, an orthogonal regression may be more appropriate, with
inverse variance weighting for each of the PEM and SAM means.
This would be expected to change the regression slightly but not
the general conclusion that there is much more influence of SAM
variation on the variation in the mean community exposure.
In Beijing (2a-2b), with a nonprobability sample of 20, the
slope remains relatively constant at 0.14 as the R3 value
increases from 0.064 to 0.23 with the usage of mean PM PEM
exposure.
Figure 3a-3b, for Azusa, California, with a nonprobability
sample of 9, the R value between PM PEM and PM SAM is negative
(-0.01) and R3 is 0.0001. In this case, taking the mean of
personal exposures shows no significant improvement in the R3
value. Such a low value of R3, as in several the studies cited
in Table 1, may be caused by several factors, such as sampling
error (too few observations), very strong indoor sources of PM,
or commuting of the people during the day to locations with
significantly different ambient pollution than in their home
community.
In Figure 4a-4b, Riverside, California, with a probability
sample of 178 people, the R3 values are significant, and they
improved from 0.16 to 0.49 while the regression equation remained
essentially the same.
The Riverside CA study and the Phillipsburg NJ study differ
in another important aspect. In Phillipsburg, NJ, 14 people
carried a PEM for 14 days, in Riverside, CA, four different
people each day carried a PEM over the 45-day study period. The
relatively large improvement in the R3 for the Phillipsburg NJ
data may be attributable to a 2-fold more precise estimate of the
mean given by 14 PEM measurements as compared to 4 in Riverside
(l/v/l3 = 0.28, < l/v/3 = 0.57).
DISCUSSION OF RESULTS OF MEAN EXPOSURE ANALYSIS
The improvement in the regression R3 value by talcing the
mean of the PEM PM data is not a discovery or an important
finding of itself. Improvement in the regression coefficient is
predictable from the process of regression to the mean of the
observations - as when random measurement errors are removed.
The higher correlation of categorical exposure assignments has
been noted in epidemiological studies. However, to our
knowledge, the magnitude of the potential improvement involving
individual PM exposure data has not been reported in the
literature. The understanding of the relationship of mean PM PEM
to PM SAM is important in that it begins to define the validity
of using the ambient PM SAM value as a surrogate for the mean PM
5

-------
A1235
PEM in the community.
There appears to be two distinct categories of exposure
studies that we have examined:
In the first type of study, such as Lioy et al. (1991) [with
and without outliers removed] and Clayton et al. (1990), there is
a significant Ra between individual PM PEM and PH SAM. In this
category, there is an appreciable improvement in correlation
between the mean PM PEM and SAM PM.
It has been suggested that these cases with high correlation
of PEM PM and SAM PM may arise where the fine portion (Xf) of the
ambient PM (PM-2.5) is highly variable from day-to-day, and the
ambient coarse fraction (Xc) is relatively constant. In some
locations, the fine portion (Xf) of the ambient PM (PM-2.5) is
more variable from day-to-day than that of the ambient coarse
fraction (Xc). In an urban area, the fine particle composition
and the fine particle concentration (Xf) are highly correlated
from site-to-site on any given day. This is due, in part, to the
homogeneous gas phase reactions of SOx and NOx to produce
sulfates and nitrates, and aerosol droplet formation with the
condensation nuclei, such as metals, which are emitted from
ubiquitous sources, such as automobiles.
On the other hand, ambient coarse particles (Xc) are
generated locally, and they have higher deposition velocities
than the fine particles (Xf). Their impact may then be limited
by fallout to a locality downwind of their emission point, as
they are not readily transported across an urban area.
Therefore, during an air pollution episode, people living in an
urban area may be exposed to fine PM of similar chemical
composition and concentrations, whereas they will be exposed to
coarse PM with a chemical composition that can depend on the
location of the exposure. Because PM-2.5 could penetrate readily
into a nonambient setting, the correlation between the mean PM
PEM and PM SAM would be high because all the people would have
similar exposure to the ambient fine PM (Xf) plus exposure to
indoor generated PM and ambient coarse PM (Xc) which may have
less fluctuation.
In the second type of study, such as Sexton et al. (1982)
and Spengler et al. (1985), there is negligible correlation
between individual PEM PM and SAM PM, and consequently there will
be little correlation between their mean PM PEM and the SAM PM.
In these cases, if the fine fraction is not an appreciable
portion of the total PM , or there are significant indoor
sources, then the correlations between mean PM PEM and PM SAM may
not be as impressive as for the other case.
IMPLICATIONS FOR PM AND MORTALITY MODELING
PM related mortality may be specific to the most highly
susceptible portion of the population. Such a cohort may be the
elderly people with the most serious chronic obstructive lung
disease (COLD) and cardiac insufficiency. Smithard (1954)
relates the findings of Dr. Arthur Davies (Lewisham coroner) who
autopsied 44 people who died suddenly during the 1952 London Fog.
6

-------
A1235
"The great majority of deaths occurred in people who had pre-
existing heart and lung trouble, that is to say they were
chronic bronchitic and emphysematous people with consequent
commencing myocardial damage. The suddenness of the deaths,
Dr. Davies thought, was due to a combination of anoxia and
myocardial degeneration resulting in acute right ventricular
dilation"
These people with compromised cardio-pulmonary systems may be
relatively inactive, while selecting to live in homes or
institutional settings without sources of indoor pollution. If
their time is spent in clean settings (e.g. where smoking is
prohibited), then they would have little exposure to PM other
than from the ambient pollution (Xf) that intrudes into their
living quarters. The exposure to PM of this cohort would be
highly correlated with PM SAM, and so would be their mortality,
if this PM (Xf) was indeed highly reactive in their pulmonary
tracts. However, the individual PM PEM of people outside this
cohort, who are relatively insensitive to PM, might not be
significantly correlated with PM SAM, as reported in most of the
14 studies cited in Table 1. This suggests a model to relate PM
and mortality as follows:
Let any person on a given day have a probability of mortality,
p(m) = k Z, where
k is the unit probability of mortality per fiq/m3 of PM per day,
Z is the daily average exposure to PM, fig/m3, independent of k.
Let us assume that each individual (i) has their own personal
value of kj that can vary from day-to-day.
The expectation of total mortality (M) in a community of
size N can be shown to be the summation of k Z over all
individuals (i = 1 to N) as follows:
M = E kj Zi
If kj is independent of Zj, then we can define K as (l/N) 23 kj
, and the mean community exposure Z as (l/N) jC Zif and it
follows M - N K Z.
This implies that, given a linear relationship of mortality
with PM PEM exposure (Z), the expected mortality is proportional
to the mean community personal exposure to PM. The individual in
the community, on any given day, with the highest probability of
dying from a PM exposure related condition is that individual
with the highest product Jq Zif not necessarily the highest
exposed individual with the maximum value of Zs.
The Fhillipsburg, NJ, data set is a case in point. In this
study, three subjects had excessively high PEM PM (shown by the
three maxima on Figure la). These values were caused by a hobby
involving welding in a detached garage (900+ ug/m3), a home
remodeling activity (800+ ^g/m3) and usage of an unvented
kerosine heater (400+ jug/m3). Such excessive PM generating
7

-------
A1235
activities are not expected of elderly people who may have
compromised pulmonary systems. In fact, the elderly and infirm
husband of the remodeler had a personal exposure of 45 M9/n3 °n
the day of the remodeling activity. The indoor monitors in the
homes of the welder and remodeler only recorded 55 jxg/m3 and 19
Mg/m3, respectively, during those events, indicating the
specificity of the high exposure to only the individual involved.
If we remove these three 'outliers', as being unrepresentative of
the exposures of those people most at risk from high values of
kj, as defined by Smithard (1953), then as shown on Figures 5a
and 5b, the correlation R2 improves markedly, from 0.250 to
0.914.
It is this relation of the PM PEM exposure to PM SAM
concentration, as shown in Figures 5a-5b, that may be a better
representation of the true situation underlying the PM vs
mortality relationships because of the "healthy worker" effect.
Chronically ill people who are sensitive to PM might move from a
polluted community if able to do so. If unable to move, they
might avoid those activities and situations they are sensitive to
(e.g. by avoiding smokers or replacing carpeting with smooth
flooring). Consequently, healthy people with high PEM PM
measures in occupations and indoor settings can cause the
correlation R2 between PEM and SAM to be low, but they may not be
the individuals at highest risk of the acute effects of PM
exposure.
CONCLUSIONS:
An individual's exposure to PM measured by a PEM might not
be significantly correlated to the local ambient PM SAM value in
the community because of personal PM generating activities and
sources of PM indoors. In terms of time-series morbidity
studies, where one looks at an individual's symptom diary or day-
to-day pulmonary function tests, it may be inappropriate to use
the daily PM SAM concentration as a surrogate for that
individual's exposure without adding additional covariates. In
such a study, the subjects might need to carry a PM PEM to allow
the investigator to test whether the fluctuations in symptoms are
related to the SAM PM or the PEM PM. The increased regression
R2 between community mean PM exposure and the community SAM PM
concentration, in some locales, supports the assumption that the
community SAM PM concentration may be a valid surrogate for mean
community PEM PM concentrations.
References:
Binder , R. E., Mitchell, C. A., Hosein, H. R., Bouhys, A.
Importance of the indoor environment in air pollution exposure.
Archives of Environmental Health 31: 277-270 (1976).
Clayton, C.A., Perritt, R. L., Pellizzari, E. D., Thomas, K. W.,
Whitmore, R. W., Wallace, L. A. Ozkaynak, H. O., Spengler, J. 0.
Particle Total Exposure Assessment Methodology (PTEAM)study:
Distribution of aerosol and elemental concentrations in
8

-------
A1235
personal, indoor and outdoor air samples in a Southern
California community. Journal of Exposure Analysis and
Environmental Epidemiology 3:227-250 (1993).
Dockery, D. W., Schwartz, J, and Spengler, J. D. Air pollution
and daily mortality: Associations with particulates and acid
aerosols. Environmental Research 59: 362-373 (1992).
Dockery, D. W., Spengler, J. D., Personal exposuresto respirable
particulates and sulfates. Journal of the Air Pollution Control
Association 31: 153-159 (1981).
Lioy, P. J., Waldman, J. M., Buckley, T., Butler, J., Pietarinen,
C. The personal, indoor and outdoor concentration of PM-10
measured in an industrial community during Winter. Atmospheric
Environment 24B, 57 - 66 (1990).
Ministry of Public Health. Mortality and Morbidityduring the
London Fog of December 1952. Report No. 95 on Public Health and
Medical Subjects. Her Majesty's Stationary Office, London
(1954).
Morandi,. M. T., Stock, T. H., Contant, C. F. A comparative
study of respirable particulate microenvironmental
concentrations and personal exposures. Environmental Monitoring
and Assessment 10: 105-112 (1988).
Ott, W. R. Concepts of Human Exposure to Air Pollution.
Environment International, 7: 179 - 186 (1982).
Perritt, R. L., Clayton, C. A., Pellizzari, E. D., Thomas, K.
W., Wallace, L. A., Spengler, J. D., Ozkaynak, H. O. Particle
Total Exposure Assessment Methodology Study: personal, indoor
and outdoor particulate concentration distributionsfor Southern
California Fall 1990 - preliminary results. In: Measurement of
Toxic and Related Air Pollutants: Proceedings of the 1991
USEPA/A&WMA International Symposium, v 2; May; Durham, NC.
Pittsburgh PA: Air & Waste Management Association; pp. 665-671.
(A&WMA publication VIP Network-21) 1991.
Sexton, K., Spengler, J. D., Treitman, R. D. Personal exosures to
respirable particles: A case study in Waterbury, Vermont.
Atmospheric Environment 18: 1385-1398 (1984).
Smithard, E. H. R. The 1952 Fog ina Metropolitan Borough.
Monthly Bulletin of the Ministry of Health and Public Health
Laboratory Services (London) 13: 26-35 (1954)
Spengler, J. D., Treitman, R. D., Tosten, T. D., Mage, D. T.,
Soczek, M. L. Personal exposures to respirable particulates and
implications for air pollution epidemiology. Environmental
Science & Technology 19: 700 -707 (1985).
WHO/UNEP Human Exposure to Carbon Monoxide and suspended
particulate matter, in Zagreb, Yugoslavia. Internal Document
EFP/82.33, WHO, Geneva (1982a).
WHO/UNEP Human Exposure to S02, N02 and suspended particulate
matter, in Toronto, Canada. Internal Document EFP/82.38, WHO,
Geneva (1982b).
WHO/UNEP Human Exposure to suspended particulate matter and
sulfate in Bombay, India. Internal Document EFP/84.66, WHO,
Geneva (1984).
9

-------
WHO/UNEP Human Exposure to Carbon Monoxide and suspended
particulate matter, in Beijing, People's Republic of China.
Internal Document EFP/82.33, WHO, Geneva (1985).
10

-------
A1235
Table l. Comparison of PM PEM exposure of individuals to the simultaneous
ambient (SAM) PM concentration in 10 US Cities and 4 foreign cities.
Reference
Year
Location
PM nm
n
time
Mean
PEM
Mean
SAM
R* PEM
vs SAM
P
Binder
et al.
1973
Ansonia
5
20
24—h
115
59
NS
NS
Dockery &
Spengler
1975
Watertown
3.5
18
24-h
35
17
0.00
NS
Dockery &
Spengler
1976
steubenvil
3.5
19
12-h
57
64
0.19
NR
Spengler
et al.
1979
Topeka
3.5
46
12-h
30
13
0.04
NS
WHO/UNEP
1981
Win
Win
Sum
Sum
Toronto
n-asthma
n-asthma
asthma
asthma
25
13
13
13
13
8-h
8-h
8-h
8-h
122
124
91
124
68
78
54
80
0.15
0.10
0.00
0.07
NS
NS
NS
NS
Spengler
et al.
1981
Kingston
Harriman
3.5
97
24-h
44
18
0.00
NS
WHO/UNEP
(a)
1982
Sum
Win
Zagreb
5
12
l-wk
114
187
55
193
0.00
0.50
NS
NR
Sexton &
Spengler
1982
Waterbury
3.5
48
24-h
36
17
0.00
NS
WHO/UNEP
1982
Win
Sum
Bombay
3.5
15
24-h
127
67
58
117
65
51
0.26
0.20
0.02
NR
NR
NS
WHO/UNEP
(b)
1985
Win
Slim
Beijing
3.5
20
24-h
l-wk
177
66
421
192
0.07
0.03
.09
NS
Morandi
et al.
1986
Houston
3.5
30
12-h
27
16
0.34
NR
Perritt
et al.
1989
Azusa
2.5
10
9
9
24-h
24-h
79
115
43
62
0.01
0.01
NS
NS
Clayton
et al.
1993
Riverside
10
141
24-h
113
84
0.23
NR
Lioy
et al
1991
Phillipsbg
10
14#
14*
24-h
24-h
86
60
0.04
0.25
.008
.0001
NS:	Not statistically significant from 0.
NR:	p Value not reported, but mentioned as significant.
#:	14 subjects carried PEMS for 14 days for 191 valid measurements.
*:	Three outliers are removed and regression is for 188 measurements.
11

-------
A1235
Philiipsberg, NJ (Winter 1988)
(all data included, n=191)
Figure la
Figure lb.
CO
I
3
O

o
w
&
m
1000
800
600
400
200
RA2 = 0.037 (p=0.008)
m = 0.46 (std err = 0.17)
b = 57.8 (std err = 96.5)
n = 191
0 50 100 150 200
MEAN OUTDOOR PM-10, ug/mA3 (4 SITES)
CO
2
O
C/D
W
Ph
<
200
"Bi)
A 150
RA2 = 0.333 (p=0.031)
m = 0 478 (std err = 0 20)
b = 55.0 (std err = 29.3)
n = 14
0
50 100 150 200
MEAN OUTDOOR, ug/mA3 (4 Sites)

-------
A1235
Beijing, PRC (Winter 1985)
Figure 2a.
Figure 2b.
~-j
<
£
o
C/3
w
Ou
400
m
<
6
^3)
3
300
200
100
0
RA2 = 0.064 (p=0.094)
m = 0.14 (std err = 0.08)
b = 116 (std err = 37)
n = 45
0 100 200 300 400 500 600 700
OUTDOOR PM-10 (ug/mA3)
m
400
I
A 300
^ 200
£
O
cn
&
W
Ph
£
<
100
0
RA2 = 0.27 (p=0.084)
m = 0.15 (std err = 0.08)
b = 111 (std err = 35)
n = 12
0 100 200 300 400 500 600 700
OUTDOOR PM-10 (ug/mA3)

-------
A1235
AZUSA, CA (Spring 1989)
Figure 3 a.
Figure 3b.
300
250
A 200
150
100
50
0
RA2 = 0.00
n = 52
¦ ¦
«#
.¦ ¦¦
m
0 20 40 60 80 100 120 140
MEAN OUTDOOR PM-10 (ug/mA3)
Outdoor mean represents 2 to 4 sites
m
<
300
~a> 250
Oh
RA2 = 0.00
n = 11
z
o
cn
Qdi
W
Pu
5
0 20 40 60 80 100 120 140
MEAN OUTDOOR PM-10 (ug/mA3)
Outdoor mean represents 2 to 4 sites
Personal Mean represents 4 to 6 persons

-------
Personal exposures (Mg/m3)
500
Riverside, California
A1235
400
300
200
100
0
pERS = 0.54*OUT + 62

o ~
Q nQi
100	200	300	400	500
Backyard concentrations (ug/m3)
600
FigureHrx. Personal exposures vs. residential (back ysrd) outdoor PM)S concentrations.
16*.
158
40
PERS = 63.8 ~ a.SZ'OUT H'Z « 4?v.
Riverside, California
e
0	«a	m	128 isb	sm
Outdoor PH-10 (ugsn3)
Figure 4b. Daily averase persona I e*po»ura» to 1*111®
v* dally average backward eoncantratIons

-------
A1235
Phillipsberg, NJ (Winter 1988)
Three 'outliers' removed (971,809, & 453)
Figure 5a.
Figure 5b.
RA2 = 0.250 (p=0 007)
m = 0 54 (std err = 0.07)
b = 42 5 (std err = 5.0 )
n =188
0 50 100 150 200 250 300
MEAN OUTDOOR PM-10, ug/mA3 (4 Sites)
m

300
6 250
OJQ
3
200
-------