&EPA
United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EPA-450/2-78-016
July 1978
Air
An Approach for
the Preliminary
Assessment of TSP
Concentrations
-------
EPA-450/2-78-016
An Approach for the Preliminary
Assessment of TSP Concentrations
by
Thomas G. Pace
Air Management Technology Branch
Monitoring and Data Analysis Division
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air, Noise, and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
July 1978
-------
11
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current contractors and
grantees, and nonprofit organizations - in limited quantities - from the
Library Services Office (MD-35), Research Triangle Park, North Carolina
27711; or, for a fee, from the National Technical Information Service,
5285 Port Royal Road, Springfield, Virginia 22161.
This document has been reviewed by the Office of Air Quality Planning
and Standards, U.S. Environmental Protection Agency, and approved for
.publication. Approval does not signify that the contents necessarily
reflect the views and policies of the Environmental Protection Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
Publication No. EPA-450/2-78-016
-------
m
TABLE OF CONTENTS
Page
List of Tables v
List of Figures vi
Executive Summary vii
Acknowledgments x
1 Introduction 1
2 Data Base 2
2.1 Site Visits 3
2.2 Tabulation of Data 3
2.3 Local Sources 4
3 Derivation and Discussion of Empirical Relationship 5
3.1 Estimating the Non-Industrial Components of TSP 5
3.1.1 Primary Non-Urban Background (PNB) and
Urban Sulfates, Nitrates (USN) Components 5
3.1.2 Analysis of Local Source (LS) Component 7
3.1.3 Urban Activity (UA) Component 16
3.2 Estimating the Industrial Components of TSP 19
3.2.1 Estimating Values for NI at Each Site 19
3.2.2 Classification of Industrial Influence 20
3.3 Regression Analysis of Non-Industrial and
Industrial Sources 23
3.4 Example Calculation 26
3.4.1 Primary Non-Urban Background and Urban
Sulfate-Nitrate Components 26
3.4.2 Urban Activity Influence 26
3.4.3 Local Source Influence 26
3.4.4 Industrial Influence 29
-------
1 V
TABLE OF CONTENTS (Cont'd)
3.4.5 Calculation in Regression Equation 29
3.4.6 Comparison of Predicted and Observed
Concentrations 30
3.5 Additional Validation of the Empirical Relationships 30
3.6 Discussion 32
4 Potential Applications of the Empirical Equation 34
4.1 Example Application 34
4.2 Data and Site Screening 34
4.3 Preliminary Assessment of TSP Problem 36
4.4 Interpretation of Monitoring Data 39
4.5 Interpreting Diffusion Model Results 39
5 Summary 41
6 References 44
7 Appendix 47
7-1 Data Base 49
7.2 Description of Activity Levels 64
7.3 Listing of Local Activity for Residential
and Commercial Sites 65
7.4 Tabulation of PNB and USN Data 66
7.5 Nomenclature and Abbreviations 67
-------
Table 3.1
Table 3.2
Table 3.3
Table 3.4
Table 3.5
Table 3.6
Table 3.7
Table 4.1
Table 4.2
Table 5.1
LIST OF TABLES
Summary of Regression Coefficients and
Statistics for Local Source Urban Activity
at Sites With High Local Activity
Tabulation of Measured and Predicted
Difference in TSP Concentration For
High-Volume Samplers Located near Each
Other but at Different Heights
Summary of Sites in Data Base
Summary of Regression Analysis of Data
for 142 Urban Sites in 13 Urban Areas
Description of Sites in Charlotte, North
Carolina
Example Calculation of Empirical Prediction
for Charlotte, North Carolina, Sites
Example Calculation of Empirical
Prediction for Allegheny County, Pennsylvania
Example of Application of Empirical Equation
to Data in Philadelphia, Pennsylvania
Page
12
17
24
25
27
28
31
35
Example of Preliminary Source Characterization 38
at a Hypothetical Site
Summary of Empirical Estimate of Annual
TSP Concentration
42
-------
VI
LIST OF FIGURES
Page
Figure 3.1 Constancy of adjusted concentration 9
versus height for sites with "low"
local activity
Figure 3.2 Decay of adjusted concentration versus 10
height for sites with "high" local
activity
Figure 3.3 Plot of relationship between adjusted 15
concentration and height of monitor
-------
VII
EXECUTIVE SUMMARY
This document describes the technical basis, uses and limitations
of an approach for making a preliminary assessment of annual Total
Suspended Particulate (TSP) data. The approach was developed using
a statistical analysis of ambient data. It defines average values
for TSP based on several siting, land use and industrial descriptors.
It is hoped that this document will prove useful to agencies and
others who are interested in understanding the sources of TSP.
Ambient levels of TSP reflect the combined impact of many
sources and source types which collectively contribute to TSP levels.
These levels are above the National Ambient Air Quality Standard for
TSP in many areas of the country. As an aid in identifying these
sources and their relative contributions to annual average TSP
levels, a data base of 142 sites in 13 urban areas was compiled.
Each site was visited and information on monitor placement and the
surrounding neighborhoods was obtained. The sites represented a mix
of undeveloped, residential, commercial and industrial land use.
The 14 urban areas visited represented a variety of industrial and
non-industrial urban centers and spanned the country geographically.
Five components were identified as a result of the site visits
and preliminary analysis as comprising most if not all of the ambient
TSP concentration. Four of the components (primary non-urban background,
urban sulfates and nitrates, local sources and urban activity) are
generally associated with sources other than industrial primary
stack emissions. These four components collectively are referred to
-------
as non-industrial components. They were found to contribute signifi-
cantly to observed TSP levels in all cities and at all site types
(except undeveloped) in varying ratios. The impact of the fifth
component, industrial primary stack emissions (called industrial
component) was found to be restricted primarily to industrial site
types except when major steel making facilities were near residential
or commercial areas. Using statistical analysis, this document
estimates average contributions to observed annual TSP concentrations
attributable to each of these five components.
Of the five components, primary non-urban background and urban
sulfates and nitrates are estimated directly from measurements taken
in non-urban areas and chemical analysis of sulfate and nitrate.
The average contribution of local source and urban activity components
was estimated empirically from the data base gathered in the non-
industrial cities. Using these average values as a guide and referring
to information for each of the 142 sites in the data base, an estimate
was made of the total non-industrial component. A multiple linear
regression technique was next used to estimate the average contribution
of the industrial component to TSP levels.
Thus, this document describes the derivation of average values
for each of the five components comprising TSP annual averages.
These average values were used to compose estimates of total annual
TSP levels for sites in two test cities. These estimates were
compared with actual observations and were found to be a reasonably
accurate approximation of the observed levels. The fraction of
-------
ix
variance in the observed data which is explained by the regression
equation (R2) was .70 and .79 for the two test cities.
The empirically derived relationships between observed annual
TSP and the five previously described components of TSP can be used
to estimate TSP levels. This estimate can be useful in several
ways:
1) The estimate can be compared with the actual concen-
tration at a site to identify those situations which differ substan-
tially from the norm. Thus, such an estimate becomes a screening
technique for identifying abnormal influences. It can also be used
as a screening technique for areas without monitors.
2) The estimate can be further broken down using data
from previous analyses to provide a preliminary estimate of source
categories contributing to TSP levels. This preliminary estimate
can be refined through more extensive analysis or used in those
situations where a more refined estimate (by atmospheric diffusion
models or from measurements such as filter analysis or special
sampling) is precluded by time or resource constraints.
3) The estimate can be useful in interpreting monitoring
data and in identifying possible siting anomalies.
4) Comparing the estimate with dispersion model predictions
may help identify the causes of discrepancies between predictions
obtained with dispersion models and observations, such as certain
improper emission factors or use of a different grid size.
-------
ACKNOWLEDGMENTS
This study was undertaken by the Air Management Technology Branch,
Monitoring and Data Analysis Division. The primary author was
Thompson G. Pace.
Many people were involved in the preparation and review of
this document. Particularly, Warren P. Freas, Dr. Edwin L. Meyer
and Dr. Elsayed Afify provided detailed review and guidance.
Others who participated in the study are Edward J. Lillis, Dallas
W. Safriet, James L. Dicke, Dr. David Marsland, Dr. James Woodburn
and Carole Mask.
-------
1 INTRODUCTION
Measurements of Total Suspended Particulate (TSP) ambient
concentrations have been made routinely since the early 1950's using
the high-volume sampler. This sampler draws air through a glass
fiber filter and the concentration of TSP in the air is expressed as
the ratio of the total weight of particulate collected on the filter,
in micrograms, to the volume of air drawn through the filter, in
cubic meters. Such data, taken for a 24-hour period regularly
throughout the year, are summarized by an annual geometric mean
concentration at each site.
The purpose herein is to describe the derivation and potential
applications of a set of empirical relationships which identify five
major components of ambient TSP concentrations and the relative
contribution of each to TSP levels as measured on high volume samplers.
These components are: 1) primary nonurban background particulates,
2) urban secondary particulates (sulfates and nitrates which are
formed by the atmospheric reaction and transformation of gases, SO
/\
and NO ), 3) particles arising from urban activity, 4) industrial
A
influences and 5) particles arising from area sources in the immediate
vicinity of the monitor (i.e., local sources). The five components
of TSP are reviewed and empirical relationships are developed for
their individual impacts on TSP levels. Potential applications for
these empirical relationships in the preliminary assessment of TSP
problems and explaining variations among the data are suggested.
-------
2 DATA BASE
The data base used in the analysis leading to the development
of empirical relationships was obtained for thirteen urban areas by
visiting monitoring sites and documenting the local sources. All
told, 142 monitors were visited. The visits were made by GCA Technology
Division as a part of a particulate study conducted for the U.S.
Environmental Protection Agency. Many of the thirteen urban areas
visited during the study had monitoring networks extending over a
large area. The entire network for each area could not be visited
due to time constraints. Usually around ten sites from each urban
area were visited. They were about equally divided among residential,
commercial and industrial neighborhoods (in those areas with industry)
and were considered representative of the entire network.[1]
Seven of these areas (Chattanooga, Miami, Oklahoma City, Providence,
San Francisco, Seattle and Washington) were selected because they
had relatively low emissions from industrial sources except in
clearly defined industrial areas. Therefore, these emissions would
not generally contribute greatly to measured concentrations outside
of the industrial neighborhoods in which they are located. This
argument is supported by a recent study of particulates in these
areas.[2] These seven areas provide data at 50 sites which can be
used for estimating the non-industrial components of TSP levels
without the masking effect of major industrial influence.
The other six areas (Baltimore, Birmingham, Cincinnati, Cleveland,
Philadelphia and St. Louis) provide data for 92 additional sites,
-------
which are a mix of residential, commercial, general industrial and
heavy industrial influences. These data provide a perspective on
the industrial influence on TSP levels.[3]
2.1 Site Visits
Photographs of the surrounding area and of the building or
structure on which the monitor was located were taken during the
site visits. Selected localized sources of particulate were often
photographed, as well. At each site the following information was
noted: 1) site classification (residential, commercial, industrial);
2) type and height of support structure for monitor; 3) description
of neighborhood surroundings; and 4) major local sources. In addition,
the typical non-urban levels of TSP and urban sulfate and nitrate
concentrations were noted for each city.
2.2 Tabulation of Data
The data for these 142 sites are summarized in Appendix 7.1.
These data provide insight into the nature and level of influence of
the components of TSP. The 1974 annual geometric mean is used
throughout, because this is the year in which the site visits were
made and local and industrial sources were noted. It would be
advisable to include other years of data in any further analysis if
changes in Ideal environs were known. Several years' worth of data
would enable the consideration of variations in meteorology as a
part of the technique.
-------
2.3 Local Sources
Characterization of the immediate surroundings was made at each
site, and local sources were identified as to type and approximate
distance. As a general rule, only sources within .5 km (about 1/4
mile) were considered. It is recognized that different types of
sources will affect TSP concentrations to varying degrees. Therefore,
the local activity level near each site was classified as "high" or
"low", with the information in Appendix 7.2 used as a guide. It
must be emphasized that some flexibility and judgment was necessary
in specifying the level of local activity, because there are many
situations in the field which do not conform precisely to the cases
in Appendix 7.2.
Appendix 7.3 summarizes the concentrations and activity levels
for the 50 residential and commercial sites visited in the seven
lightly industrialized areas. Industrial sites in these areas were
also visited, but these are not tabulated here because it was suspected
that the industrial influence might mask the impact of the non-
industrial component. The data from these 50 sites are subsequently
used to estimate the impact of the local sources and urban activity
on ambient TSP levels.
-------
3 DERIVATION AND DISCUSSION OF EMPIRICAL RELATIONSHIP
In analyzing the data base described in Section 2, the air
quality concentrations reflect the combined effect of many sources
of particulates. The five components of TSP, previously mentioned,
each were hypothesized (or observed) to vary considerably between
geographic areas, or site types or local siting differences. Each
of these components and the methods used for estimating their "typical"
or average contribution to TSP levels is presented in this section.
Four of the five components (primary non-urban background,
sulfates and nitrates, local sources and urban activity) are generally
associated with sources other than industrial stack emissions.
These four have been grouped together in the following analysis and
are referred to as non-industrial (NI) components. In the derivation
of an empirical relationship, the NI component will be considered
separately from the industrial component.
3.1 Estimating the Non-Industrial Components of TSP
The Primary Non-Urban Background (PNB), the Urban Sulfate,
Nitrate component (USN), the Local Source component (LS) and the
Urban Activity (UA) component are discussed below, along with the
methods used to estimate their average contribution to TSP levels.
3.1.1 Primary Non-Urban Background (PNB) and Urban Sulfates,
Nitrates (USN) Components
This portion of TSP is comprised of non-urban primary particulates
which are homogeneously distributed over a scale of hundreds of
kilometers (a part of the traditional background level) and urban
-------
sulfates and nitrates. It is assumed that sulfates and nitrates
measured in urban areas represent the composite impact of rural and
urban sources. These portions of the TSP may vary considerably from
one urban area to another but are considered in this analysis to be
generally constant and uniformly distributed across a given urban
area.
The magnitude of the primary non-urban background (PNB) portion
of TSP cannot be directly measured in an urban area. It is a generally
accepted (although not precise) practice to use the measurements of
TSP concentration in non-urban areas near the urban area being
studied as an indicator of the total non-urban background. Such
measurements typically range from 15 to 35 pg/m3, depending upon the
region of the country, with highest values in the East. This non-
urban measurement includes both a primary non-urban portion and also
a non-urban sulfate and nitrate portion. In order to estimate the
primary non-urban portion of the non-urban measurement, the non-
urban sulfate and nitrate portions are subtracted from the total
non-urban measurement. It is assumed that all organics which have
been formed by photochemical reaction are included in the primary
non-urban estimate. The urban sulfate-nitrate (USN) portion of the
TSP is estimated by measuring sulfate and nitrates in the urban
area. Thus, USN includes the non-urban sulfate-nitrate as a subpart
of that measured in the urban area. Together, the PNB and the USN
comprise that portion of TSP which is commonly assumed to be (relatively)
constant across a given urban area.
-------
In this analysis, the National Air Surveillance Network (NASN)
sites were used to estimate the primary non-urban and the sulfate-
nitrate levels. The non-urban levels measured by NASN stations can
be divided into sul fate, nitrate and a remainder which is mostly
primary particulate (with a small amount of secondarily-formed
organic particulate). This is commonly assumed to be the PNB value
for the nearby urban area. The urban sulfate and nitrate can be
estimated from the urban NASN data. Of course, if several non-
urban stations and detailed meteorological data were available to
estimate the non-urban TSP influx, it would be preferable to use
such data.
Once the PNB and USN have been determined for an area, these
levels must be subtracted for the measured TSP concentration of each
site within the urban area. The "adjusted" value thus reports the con-
centration at the site due solely to the influence of sources within
the urban area. Consequently, each annual average (reported in Appendix 7.3)
was adjusted by subtracting from it the appropriate PNB and USN
values given in Appendix 7.4. These adjusted values are used in
the further analysis of the Local Source and Urban Activity components.
3.1.2 Analysis of Local Source (LS) Component
Review of the information available from the site visits
provided an indication that the height of the monitor and the
amount of local activity from traffic and parking near the monitors
may be significant factors affecting the concentration measured at
these sites. Residential sites appeared to be only occasionally
-------
8
influenced by this local activity because there was seldom any such
activity, but commercial and industrial sites were nearly always
near such influence. However, industrial sites experience a highly
variable influence from industrial process and fugitive emissions
which would mask the effect of local activity. Thus, it was decided
to base the Local Source and Urban Activity analysis on residential
and commercial sites in non-industrial or only slightly industrial
urban areas to avoid this masking effect. A total of 50 sites in 7
cities are in this subset of the data base (see Appendix 7.3).
The data were plotted, and visual examinations were made to
determine whether a height-concentration relationship was present
for each activity level (see Appendix 7.2). Figures 3.1 and 3.2
show the results of the plot for low and high local activity. It is
clear from Figure 3.1 that the adjusted concentration near "low"
activity sites averages 20 yg/m3 and there is no apparent relationship
between height and concentration. Figure 3.2 for "high" activity,
however, shows a distinct relationship of height and concentration.
The height-concentration relationship is expected, based upon earlier
studies which observed that the number concentration of a colloid in
a gravitational field decreases exponentially with increasing height.[4]
Such an exponential decrease would be more easily observed where
ground-level activity predominates. The lack of an apparent height-
concentration gradient at the "low" activity sites suggests that the
local influence on concentration from this low level of activity
must be very slight. Also, the apparent increase in concentration
-------
90
^ 80
O)
o
oc
ca
a
ta
cc
C/l
oa
-------
10
r>
E
90
80
70
60
09
V)
50
I
oc
UJ
v>
40
o 30
20
o
O
10
3 5
10 15 20
HEIGHT, meters
25
30
Figure 3.2 Decay of adjusted concentration versus
height for sites with "high" local activity
-------
11
between "low" and "high" activity sites suggests that local activity
level is a significant determinant of measured concentrations. The
activity relationship is understandable, if one assumes that TSP
emissions are higher near sites with the high activity level.
Study has shown that the concentration-height relationship can
be satisfactorily described by several mathematical forms. [5] This
same study, however, did suggest the most desirable form. Using the
nomenclature as summarized in Appendix 7.5, this form would be:
y = a e" + c (Equation 3.1)
where: y = predicted cone.,
H = height, meters
a, b, c = empirically derived constants
Thus, an empirical relationship of this form was fit by a least
squares regression procedure. [6] The regression results for the
combined data set are shown in Table 3.1, which indicates that the
significance level is quite high; however, the standard error of
p
10 yg/m3 and the square of the correlation coefficient, R , of .41
indicate that factors other than height are also important, as
mentioned. The resulting equation is:
-.2 H . ~-|
y = 45e + 3I
The distance of the source from the monitor and the variations
in actual emission levels of the sources near the monitor account
for much of the unexplained variance. A rough attempt was made to
estimate the distance of the sources from the monitors and to consider
-------
Table 3.1. Summary of Regression Coefficients and Statistics for Local Source
Urban Activity at Sites With High Local Activity
" hH
For Equation 3.1, y = ae~ +c
Coefficients (Significance Level)
a 45
(.0002)
b .2
N/A*
31
(.0001)
R-square .41
Standard Error 10.1
* Not Available (value of coefficient determined by iteration)
ro
-------
13
this in the analysis. However, a review of the data base indicated
that the distances from each monitor to nearby sources were not
known precisely enough for inclusion in this analysis. The data
base only included a general description of the surrounding area
and not distances to specific sources such as streets.
In Equation 3.1, the first term, a e" , represents that
portion of the predicted concentration which varies with height. As
"H" (height) becomes higher, this term decreases in value and approaches
zero, and the predicted concentration approaches the constant "c".
Thus, this model implies that monitors located higher than about 40-50
feet are not subject to appreciable height-concentration gradients
attributable to high local activity. The constant "c" is presumed
to represent urban activity influences on a sub-urban scale. As
discussed in the following section on urban activity, this 31 yg/m3
is assumed to represent a component of TSP which is related to
emissions and activity over a larger geographical area than that
which contributes to the local scale component. Thus, the local
scale component includes only the exponential part of the equation.
There were no data points below about 3 meters elevation upon
which to base the empirical relationship in Equation 3.1. Also, it
is possible that there is a limit to the height-concentration relation-
ship below which there is relatively uniform mixing due to ground
level disturbances. Thus, Equation 3.1 should only be applied to
sites higher than 3 meters, until additional data are gathered and
further study made. Meanwhile, it is recommended that the assumption
-------
14
be made that there is a uniform mixing cell 3 meters high for ground
level sources and that there is no concentration gradient in the
first three meters. The relationship between height and concentration
is plotted in Figure 3.3.
There are relatively few (50) data points used in the analysis
and the sites were not selected by a strict random method (although
no bias was intended). Thus, the data set was reviewed to determine
whether there appeared to be a systematic bias which accounted for
the height-concentration relationship. The adjusted concentration
data were against height separately for each city, and the plots
were examined for trends or patterns. It was found that the height-
concentration pattern was evident in Providence, Washington, D.C.,
and Oklahoma City, whereas barely perceivable patterns appeared to
exist in San Francisco and Chattanooga. No pattern is apparent in
Miami. Seattle had only one data point. It was concluded that
there is no systematic bias among cities in the data set which could
account for the pattern in the combined data set. Also, it was
evident that other factors beyond the scope of this investigation
contributed to the weak or apparent lack of pattern in the data in
some cities. For example, it has been shown in previous work by
Record that the data set for Miami can be described by a parameteri-
zation including average daily traffic (ADT) and the slant distance,
/ height^ + distance^, of the site to nearby traffic.[7] Thus, any
height effect among sites in specific cities may be masked by other
variables such as ADT or distance of the monitor from the source.
-------
15
90
80
i
S 60
u
z 50
5
a
aj
£ 40
tn
CO
o
1 30
I-
cc
I-
m 20
u
I.
10
3 5
10
15
HEIGHT, meters
20
25
30
Figure 3.3 Plot of relationship between adjusted
concentration and height of monitor
-------
16
It is very important that further efforts attempt to define more
clearly the relationships among height, distance and an estimate of
source strength such as ADT.
To illustrate the height-concentration effect, a number of
"pairs" of monitors have been located in various urban areas. These
are either monitors located within a block or so of one another and
at different heights (usually part of the local network) or located
on a tower at the same location for a short-term experiment. These
data are summarized in Table 3.2. It can be seen that the overall
correlation between the values predicted by the empirical height
relationship and observed differences in concentration is fairly
2
good (R =.77). The several cases where the model severely underpredicted
differences could be partly because the lower monitor was closer to
the source laterally than the higher one.
3.1.3 Urban Activity (UA) Component
The activity in the area around the high-volume sampler but
outside of the small radius which contributes to the local scale
component, also contributes to TSP levels. This activity and its
contribution to TSP levels is referred to as the Urban Activity (UA)
component. The analysis in Section 3.1.2 estimates that the average
urban activity component at sites with high local activity is about
31 pg/m3. Since at least 90% of the sites which were reported to
have a high activity level in Appendix 7.3 were commercial sites, it
is assumed that most commercial sites would have an average 31 yg/m3
influence from urban activity sources. Moreover, it was observed
-------
Table 3.2 Tabulation of Measured and Predicted Difference in TSP Concentration For
High-Volume Samplers Located Near Each Other But At Different Heights
Site Pair Location
Height, fleters
High Low
Measured
Difference,
ug/m3
Difference Between
Predicted Concentrations
at "High" and "Low"
Activity Sites Using
Equation 3.1
ug/m3
Philadelphia, Broad St.
Philadelphia, Franklin Inst.
Kansas City
Austin, Texas
Austin, Texas
Cincinnati
Chicago
Pi ttsburgh
12 3
20 4
7 2
10 6
6 1
7 2
40 4
26 3
50
25
10
4
11
7
26
47
21
19
13
8
11
1 3
20
23
-------
18
that the non-industrial activity in industrial neighborhoods was
reasonably similar to that in commercial neighborhoods. Thus, until
better data on which to base an estimate become available, it will
be assumed that industrial sites have a similar average 31 yg/m3
urban activity component. The data at sites with low activity
(almost exclusively residential sites) indicate an average residential
urban activity component of about 20 yg/m3, as was shown in Figure 3.1.
It is assumed that sites in undeveloped neighborhoods have no urban
activity component because the primary non-urban background values
would include any urban activity in these areas. Thus, the explicit
"urban activity" term derived here would be zero for these areas.
A recent study of TSP by Record can be used to estimate the
composition of the urban activity component.[8] It appears that
motor vehicle exhaust and tire wear account for 10% to 15% each of
the urban activity influence. Construction and demolition account
for another 5% to 10%, unless major urban renewal projects in the
area increase this component. Space heating, usually from oil-fired
boilers and furnaces, can account for from zero to 40%, and small
industry and power generation can account for from zero to 20+%,
depending upon degree of control, type of fuel burned and climate
conditions. The remainder (around 30-50%) appears to be due to dust
from paved roads and unpaved roads and parking lots, which is suspended
by both the wind and man's activity. These estimates may be used to
provide a rough indication of the sources of TSP comprising the
urban activity component.
-------
19
3.2 Estimating the Industrial Components of TSP
In order to estimate the industrial influence at residential,
commercial and industrial sites, it was necessary to include all the
142 sites described in Appendix 7.1. This gives a data base covering
a variety of industrial and non-industrial areas and sites. A
regression analysis was performed on these data. To do this, the
non-industrial (NI) influence (computed as the sum of PNB, SN, UA,
LS) was estimated for each site. Each site was described mathematically
by a series of binary variables to identify sites near industrial
influence and a multiple linear regression was performed to estimate
the coefficients of the NI and Industrial terms. This procedure is
described in more detail in the following sections.
3.2.1 Estimating Values for NI at Each Site
The variable NI (Non-Industrial) is an estimate of the combined
impact of primary non-urban background, urban sulfates and nitrates,
the urban activity component and the local source component previously
developed. The method used to estimate NI is discussed below.
1) The secondary (urban sulfate, nitrate) and primary
non-urban TSP levels were estimated for each urban area listed in
Appendix A. As mentioned, the values assigned to the sites in each
area are summarized in Appendix 7.4.
2) To this was added the appropriate urban activity
value, based on site type. A value of 0, 20 or 31 yg/m3 was assigned
to each site, depending on its classification as undeveloped,
residential or commercial/industrial.
-------
20
3) Finally, a local scale component was added at each
commercial and industrial site. It was assumed that since the vast
majority of commercial or industrial sites are subject to high local
ground-level activity, they have a local source influence.
3.2.2 Classification of Industrial Influence
It was observed by reviewing the data that industrial sites
generally had higher concentrations if the monitors were near general
industry, a steel mill or related operations. Also, residential-
commercial site values were elevated in concentration near steel
mills, though not as markedly. Thus, a procedure using binary
classification variables (sometimes called "dummy variables") was
developed to identify variations in industrial influence.
At this point, it is appropriate to explain the binary classification
variables and why this method was selected for describing the industrial
influence and incorporating it into an empirical relationship.
First, it was decided that a multi-variable regression procedure
would be used to estimate the industrial influence. Since data was
not readily available to estimate the emissions of the industrial
sources near each monitor, a method was needed to "classify" the
industry as to its type or nature. Binary descriptors were used to
identify whether a specific site met certain conditions of industrial
influence (say, being within 2 km of a steel mill). If the site
were near a steel mill, a "1" would be assigned as the classification
variable for that site. Conversely, if the site were not near a
mill, the "0" classification variable would be used. In the multi-
-------
21
variable regression, a coefficient would then be calculated which
would be the estimate using the regression of the impact of the
steel mill(s) on the site's measured concentration. It should be
noted that these variables were selected based on a preliminary
review of the data so that the regression could be used to evaluate
any differences in concentration for these classifications.
In this analysis, all sites were assigned a series of three
binary classification variables. The first variable (SMIND) was
assigned a "1" if the site was considered industrial and there was a
partially controlled or uncontrolled steel mill or coking operation
within less than 2 km range. A value of "0" was assigned to all
other sites.
The second classification variable (SMRC) was assigned a "1" if
a partially controlled or uncontrolled steel mill or coking operation
was within the range of 2-10 km. (These sites were designated as
residential or commercial). Other sites were assigned a "0" value.
The third classification variable (GENIND) was assigned a "1"
if it was near an industrial influence (1-2 km) and not near (less
than 10 km) one or more steel mills or coke ovens. All other sites
were assigned a "0".
The three classification variables are summarized below.
o
,1
SMIND (°) - The presence nearby of partially or uncontrolled
steel mill and/or coking operations on industrial
sites (less than 2 km).
-------
22
SMRC (?) - The presence nearby of partially pr uncontrolled
steel mill and/or coking operations at residential
or commercial sites (within the range of 2-
10 km).
GENIND (?) - The presence nearby of general industry at
industrial sites (less than 2 km away).
The regression model is given by Equation 3.2:
OBS. = Bn NI. + B9 GENIND. + B, SMRC, + BA SMIND. + B,, ug/m3
1 I 1 c. lo If i o
where:
NI = Total estimated non-industrial influence, yg/m3 includes:
'Primary Nonurban Component
'Sulfates, nitrate levels in urban area
'Urban Activity Component (0, 20 or 31 yg/m3 at
undeveloped, residential and commercial/industrial
sites, respectively).
'Local Source Component calculated by:
zero or 45 e"'2 ^i^ (Equation 3.1)
where H- is height of monitor, "i", meters
GENIND d) = The presence of general industry near (less than
2 km) industrial site.
SMIND (?) = The presence of uncontrolled steel mill or coking
operation near (less than 2 km) industrial site.
SMRC (?) = The presence of uncontrolled steel mill or coking
operation near (2-10 km) residential or commercial
site.
-------
23
B-, thru B. = Coefficients to be derived from the Regression Procedure
Be = Regression constant
It is important to note that only one of the above variables
can be assigned a binary classifier of "1" at a particular site.
This would be the one that best describes the site. The other two
must be assigned a "0".
3.3 Regression Analysis of Non-Industrial and Industrial Sources
A multiple regression was performed on the observed concentration
(OBS) compared to the NI, SMIND, SMRC, and GENIND variables.[6]
Table 3.3 summarizes the number of observations for each variable
and Table 3.4 summarizes the results of the regression. It is shown
that the variables predict the observations with a standard error of
16 yg/m3. For an average observation of 79 yg/m3, this represents a
coefficient of variation of +_ 20%. The equation explained 71% of
the variance (R-square) among the observations. As can be seen from
the tabulation, all coefficients are significant at the 99% confidence
level except the regression constant term (unexplained portion),
which is significant at the 95% level. The coefficient of the non-
industrial portion (.88) indicates that it was a reasonable predictor
of the non-industrial component of this data set. The regression
constant represents that portion of the observation that was not
explained by any of the factors considered in the regression.
-------
24
Table 3.3 Summary of Sites in Data Base
SITE TYPE NUMBER
Undeveloped 9
Residential 35
Commercial 52
Industrial (GEN) 21
Industrial (SM) 13
Res/Com (SMRC) 12
Total 142
-------
Table 3.4 Summary of Regression Analysis of Data for 142 Urban Sites in 13 Urban Areas
NI
GENIND
SMRC
SMIND
CONSTANT K
Results of Regression
Coefficients
Significance
Level-
Standard Error
.88
15.0
22.9
52-0
13.3
(.0001)
(.0010)
(.0010)
(.0010)
(.0371)
.10
4.3
4.9
5.1
6.4
R-SQUARE .71
STANDARD ERROR 16.0
ro
-------
26
3.4 Example Calculation
The application of the empirical relationship for estimating TSP
concentrations using the previously derived relationship (Equation 3.2)
is relatively straight-forward. Data were obtained in Charlotte
(Mecklenburg County), North Carolina to estimate TSP levels. Table 3.5
describes the sites, and Table 3.6 summarizes activity levels assigned
to each site and the results of the calculations. As an example,
TSP concentrations arising from each of the previously identified
five major components of TSP are calculated at the Community Hospital
site in Charlotte, North Carolina.
3.4.1 Primary Non-Urban Background and Urban Sulfate-Nitrate
Components
Available NASN data were used to estimate PNB and USN levels.
The non- urban NASN station near Charlotte averaged 30 yg/m3. This
includes approximately 9 yg/m3 of nonurban sulfate (as ammonium
sulfate) and nitrate, leaving a primary component of 21 yg/m3.
Urban sulfate and nitrate levels measured at the urban Charlotte
NASN station totaled 12 yg/m3.
3.4.2 Urban Activity Influence
The neighborhood around the Community Hospital monitor was
primarily commercial for 1-2 km in each direction. Thus a value of
31 yg/m3 is added as urban activity influence.
3.4.3 Local Source Influence
Table 3.5 indicates that there is an unpaved parking lot 100
feet away. There is a major arterial nearby and the surrounding
-------
27
Table 3.5 Description of Sites In Charlotte, North Carolina
Site Name
McAlpine Sewage
Treatment Plant
Mint Hill
Fire Station 14
Mecklenburg
Health
Fire Station 10
Beatties Ford
Carpenter Airport
Community Hospital
Fire Station 11
Davidson Filter
North 29
Patrol Station
Site Height,
Type Feet
Rural 10
Small
Town
Residential ig
Commercial 10
Commercial 16
Commercial 8
Rural 3
Commercial 25
Commercial 15
Small
Town
Light
Commercial
20
8
Description
Very clean looking area. All roads rinrt par».ino o,iv»i.
No bare land. No urbanization nearby. liearcsr thru
road 1s 2 lane linht traffic 1/2 mile away.
Behind telephone Co. Bldq. Unpaved (ireenhou^p ,ir<=a
next door apoearpd to have verv little traffic.
Monitor was 75 feet from 2 lane road, do nafor
activity within 1/4 mile. Not enounh to produce a (jir."
effect.
Residential area surrounding two commercial strips -
Randolph Road and- Sharon Amity Road. Both are ~ lane
and are 125 and 75 feet respectively from monitor.
lio unpaved areas.
Monitor well removed from local traffic. East Blvd.
is 1/8 mile away but is screened from monitor by
trees. No unpaved areas.
Approximately 100 feet from Wilkerson Blvd. (4 lane)
and 50 feet from light traffic side street. Entire
area very commercialized.
Located in dirt parking lot which is used twice a day
by around a hundred cars. Located 15 feet fron Oaklawn
Avenue-moderately traveled 2 lane and 75 feet from
Beatties Ford Road- 4 lane. Expressway 500 feet under
construction in 1973
Very remote area of county.
nearby.
No traffic or activity
Generally active area. Four lane arterial 100 feet
away and very active unpaved parking lot around 100 ft.
Generally quiet area. Arterial within 1/4 mile but
nothing within 100 feet. A fair amount of commercial
activity within 1/4 mile.
Not much activity here-typical small town. Not really
enough urban activity to produce a height effect.
US 29 4 lane 200 feet away. Lightly commercialized
but not in city limits. Rest of area is sparsely
populated.
Davidson Pump
Rural
On lake, no activity, very remote
-------
Table 3.6 Example Calculation of Emoirir.al Prediction fnr Charlotte, North Carolina, Sites
Estimated Concentrations yg/m3-
Sites
Fire Station 10
Fire Station 11
Fire Station 14
Community Hosp.
Davidson Filter
Plant
Mecklenburg
Health Dept.
Beatties Ford
Water
Mint Hill
N. 29 Patrol
McAlpine Sewage
Carpenter
Airport
Statistical Data
Neighborhood
Commercial
Commercial
Residential
Commercial
Rural
Commercial
Commercial
Rural
Commercial
Rural
Rural
: Coefficient
Activity Level
Hi ah
High
High
Htah
Low
Low
High
Low
Low
Low
Low
of Variance: R =
Height, Ft.
16
15
16
25
20
10
8
8
8
10
3
.70
Primary
Nonurban
Background
ug/m3
21
21
21
21
21
21
21
21
21
21
21
Sulfate,
Nitrate
yg/m3
12
12
12
12
12
12
12
12
12
12
12
Urban
Activity
Influence
ug/m3
31
31
20
31 .
0
31
31
0
31
0
0
Local
Influence
ug/m3
15
16
15
9
-
-
23
-
-
-
-
Industrial
Influence
yg/m3
0
0
0
0
0
0
0
0
0
0
0
Emp-jr-ical
Predicted
ug/m3
83
84
73
77
42
65
89
42
70
42
42
1974
Observed
ug/m3
66
62
48
74
46
55
101
39
54
36
30
Residual
ug/m3
+ 17
+ 22
+ 25
+ 2
- 4
+ 7
-12
+ 3
+ ffi
+ 6
+ 12
ro
oo
-------
29
area is commercial. This constitutes a high local activity level.
Thus, the equation:
' YLS = 45 e~'2H (Equation 3.1)
is used where:
YLS = predicted local source concentration,
yg/m3
H = height, 8 meters
therefore:
YLS = 45 e"2
YLS * 9 yg/m3
3.4.4 Industrial Influence
There is no specific industrial influence at the Charlotte,
North Carolina, sites. Some minor industrial influence is probably
reflected in the empirical value of Urban Activity Influence. Also
included are the impact of fuel oil and coal space heating.
3.4.5 Calculation in Regression Equation
The regression model (Equation 3.2) is as follows:
y = B1 NI + B2 GENIND + B3 SMRC + B4 SMIND + BB yg/m3
therefore:
y = B^PNB t SN t UA t LS)
B0 SMIND + B_ SMRC t B, GENIND + B
£. o 4
-------
30
Substituting,
y = .88 (21 + 12 + 31 + 9) + 0 + 0 + 0 + 13.3
= 77 yg/m3
3.4.6 Comparison of Predicted and Observed Concentrations
Predictions of concentration were similarly performed for all
sites. The predicted and observed values were analyzed by linear
regression to determine the line of best fit. The resulting correlation
p
was R = .70. The equation of the line of best fit had a slope of
.89 (times predicted concentration) and an intercept of -1.0 yg/m3
which indicates excellent agreement between predicted and observed
concentrations.
3.5 Additional Validation of the Empirical Relationship
The Mecklenburg County calculation serves as an independent
check of the regression equation. In addition, a similar calculation
was made for Allegheny County, Pennsylvania. Table 3.7 indicates
the results of that calculation. The R-square was .79, the slope
was .76 and the intercept was 21.6 yg/m3 for a calculation of the
line of best fit between the predicted and observed data. This is
also quite good agreement for a model in a heavily industrialized
area. The estimate for Allegheny County was made from a site description
without actual site visits or prior knowledge the concentration at each site.
-------
Table 3.7 Example Calculation of Empirical Prediction for Allegheny County, Pennsylvania
Sites
Baden
Beaver Falls
Koppel
Brighton Township
Midland
Elco
Downtown
Central Lab
Hazelwood
North Braddock
Ouquesne II
Liberty Boro
Clairton
Airport
South Fayette
Statistical Data:
Neighborhood
Industrial
Commercial
Industrial
Rural
Industrial
Rural
Commercial
Commercial
Industrial
Industrial
Industrial
Industrial
Industrial
Commercial
Rural
: Coefficient of
Activity Level
High
Hiqh
Hiqh
Low
High
Low
High
High
High '
High
High
High
High
Hiqh
Low
o
Variance: R =
Height, Ft.
18
6
30
3
30
-
30
45
45
15
15
30
30
60
30
.79
Primary
Nonurban
Background
ug/m3
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
Sulfate,
Nitrate
ug/m3
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
Urban
Activity
ug/m3
31
31
31.
0
31
0
31
31
31
31
31
31
31
31
0
Local
Influence
ug/m3
13
23
6
0
E
0
6
2
2
16
16
6
6
1
0
Industrial
Influence
ug/m3
52
0
52
23
52
0
23
23
52
52
52
52
52
0
0
Model
Predicted
ug/m3
144
100
137
76
137
53
108
105
133
144
144
137
137
80
53
Average
Observed
1974-76
ug/m3
135
80
105
- 80
140
75
95
110
100
135
150
140
120
84
58
Residual
ug/m1
+ 9
+ 20
+ 32
- 4
- 3
- 22
+ 13
- 5
+ 33
+ 10
- 6
- 3
+ 17
- 4
- 5
-------
32
3.6 Discussion
In the preceding analysis, two empirical relationships were
developed which can be used to help explain variations in particulate
concentrations. In the first relationship, height of the high-
volume sampler above the ground was the independent variable and
the coefficients of the proposed relationship were obtained by
multiple regression. The significance levels of these coefficients
suggest that the probability that the coefficients are not equal to zero
is very high. The coefficient associated with monitor height was
estimated by iterating the linear regression solution of the trans-
formation x1 = e~ x for various values of "b" to maximize the
square of the correlation coefficient and minimize the standard
deviation. Thus, no confidence level for this coefficient could be
readily obtained. In fact, a coefficient of zero (on which a null
hypothesis would be based) would not be very meaningful since ef =
s*.
] and y = constant. A regression analysis was made where "b" = 1
and thus resulted in a very poor fit of the data. Since the regression
statistics were relatively insensitive to small changes in "b" near
the value of "b" = .2, the coefficient was only estimated to one
significant figure. The resulting relationship is meaningful in
that monitor height has been shown to vary greatly among sites.
If, as was shown here, large variations in TSP concentration can
be attributed to monitor height, then much more attention should be
directed to monitor placement in the monitor siting and network
design process.
-------
33
The second regression estimated coefficients of both the non-
industrial and industrial components of TSP. In selecting the
terms for the regression, special attention was given to terms that
would prove meaningful in control strategy development and in
understanding the sources of TSP. The non-industrial component was
estimated for each site using measurements and predictions from the
first empirical relationship. Thus, the coefficient of this term
in the regression would be expected to be close to 1.0 if the
estimates for the non-industrial component were generally accurate.
The resulting coefficient of .88 was considered satisfactory evidence
that the non-industrial component estimates were acceptable. As in
the first regression, the confidence that the null hypothesis can
be rejected is very high. The coefficients of the industrial terms
suggest strongly that industrial sources, particularly steel mills,
are a major influence on TSP levels. The entire equation provides
perspective on the relative impact of non-industrial sources and
indicates that even at industrial sites non-industrial sources are
a significant part of the total measurement. This is very significant
in directing future particulate investigations.
-------
34
4 POTENTIAL APPLICATIONS OF THE EMPIRICAL EQUATION
While it is certainly expected that the use of Equation 3.2
will provide a further understanding of general factors affecting
TSP, there are other more specific uses of the empirical equation
which should be of added benefit in particulate planning and analysis.
Such applications include screening of sites to determine which
sites might not fit the "typical" pattern (the value predicted by
the equation), and thus might be candidates for further, more intensive
analysis. It would serve as a non-data and resource-intensive tool
which could be used to provide a preliminary analysis until data or
resources for a more intensive analysis of TSP data and problems
were available. Interpretation of monitoring data and dispersion
model results are other potential applications.
4.1 Example Application
In this example application, Philadelphia, Pennsylvania (one of
the urban areas in the data base), was analyzed. Using the same
procedure as with Charlotte, Table 4.1 indicates the site characterization
and concentration predictions for Philadelphia. In this instance,
the model estimates compare fairly well with the observed values,
explaining 68% of the variance. It must be pointed out that the
primary use of the model is not to estimate concentrations but to
help explain the observed data. Several examples of this follow.
4.2 Data and Site Screening
An obvious application of the empirical equation is in the
screening of data and sites to determine which sites might not fit
-------
Table 4.1 Example of Application of Empirical Equation to Data 1n Philadelphia, Pennsylvania
Site Name
Belmont Filter
Roxboro Filter
N. E. Airport
AMS Lab
Franklin Inst.
(CAMP)
S. Broad & Spruce
500 S. Broad
Defense Supply
Allegheny River
Int. Airport
Aramingo Fire St.
Activity
Level
Low
Low
Low
Low
High
Hiqh
High
High
High
Hiqh
High
Neighborhood
Residential
Residential
Residential
Residential
Commercial
Commercial
Commercial
Industrial
Industrial
Industrial
Industrial
Monitor
Height, ft.
13
13
13
17
11
13
35
13
13
13
35
Primary
Nonurban
Background
pg/m3
24
24
24
24
24
24
24
24
24
24
24
Sul fates,
Nitrates
ng/m3
18
18
18
18
18
18
18
18
18
18
18
Urban
Activity
Influence
Mg/m3
20
20
20
20
31
31
3i
31
31
31
31
Local
Scale
Influence
pg/m3
0
0
0
0
22
19
4
19
19
19
4
Industry
Influence
pg/m3
0
0
0
0
0
0
0
15
15
15
15
Model
Predicted
pg/m3
68
68
68
68
97
94
8n
109
109
109
95
1974
Observed
pg/m3
72
59
64
78
119
115
76
105
122
94
116
Residual
pg/m3
- 4
+ 9
+ 4
- 10
. 22
. 21
+ 4
+ 4
. 13
t 15
21
Statistical Data: Coefficient of Variance: R = .68
Line of best fit:
y = 1.06X - 1.0 pq/m1
CO
tn
-------
36
the "typical" site as indicated from the empirically predicted
value. Such information would be valuable in highlighting those
sites which would require further study because of concentrations
anomalously higher or lower than the model predicts. In the example
of Philadelphia in Table 4.1, the equation underpredicted the concen-
tration at the Broad and Spruce site, indicating the presence of a
very strong influence not adequately accounted for by the empirical
estimates. Recent inspection of the site indicated that this was
probably due to local traffic with the monitor being extremely close
(35 feet) to the road and low to the ground as well. A more detailed
study was deemed warranted at this site and is now in progress.
Another station which was underpredicted is the CAMP station, which
is within 100 feet of a major unpaved parking area. The close
proximity of the parking lot is probably one reason that the prediction
is low. The underprediction at the Aramingo Fire Station may indicate
the extreme dominance at that site by an industrial source across
the street. As a screening tool, then, this equation appears to
highlight sites which are dominated or influenced by extreme or
unusual conditions. The equation also gives a preliminary estimate
of the influences at all the sites by local, urban activity, Industrial,
PNB and SN components. These examples indicate that distance from
the source is an important consideration, even though it could not
be included in this procedure because of data base constraints.
4.3 Preliminary Assessment of TSP Problem
The empirical equation provides a framework for apportioning
-------
37
the air quality measurements among sources. Table 4.2 illustrates
how such a preliminary apportionment might be accomplished using the
empirical equation. The values suggested for apportioning the urban
activity and local source influences are derived from Record.[8].
It is felt that such a preliminary assessment will give the
control official a framework for further analysis of the problem and
will help to place the various portions of the TSP problem in perspective.
This preliminary estimate may provide an adequate level of
analysis in certain areas where the problem seems relatively straight-
forward and the time and resources for more extensive analysis are
not available, nor is the time available to allow a more detailed
study. In most cases, however, further study and refinement of the
estimate are warranted and encouraged. Such study might take the
form of a field experiment to gather new data, examination of filters
to identify source types, analysis of new or existing data by a
variety of statistical methods, and modeling of specific sources or
the entire area using dispersion models, as recommended in EPA
guidance.[9,10,11,12]. In some cases, the resolution of source
contributors may be provided by a single technique, such as dispersion
modeling. However, in many cases, a refined estimate of sources
will involve the synthesis of several analytical methods using good
engineering judgment.
-------
38
Table 4.2 Example of Preliminary Source Characterization
at a Hypothetical Site
Preliminary
Estimate
ng/m3
Primary Non^Urban Background 21
Urban Sulfates 18
Urban Nitrates 4
Urban Activity Influence 31
Tailpipe Exhaust and
Diesel Exhaust 10-15%
Tire Rubber 10-15%
Vehicle and Windblown
Resuspension 30-50%
Space Heating 0-40%
Power Plant and Small
Industry 0-20+%
General Construction 5-10%
Local Source Influence 16
(additional vehicle-related)
Tailpipe Exhaust
Diesel
Tire Rubber
Vehicle Resuspension
Subtotal Nonr-Industrial 90
Industrial Influence 15
Total Predicted Before Using Regression 105
-------
39
4.4 Interpretation of Monitoring Data
One recent study has indicated that there is wide variation in
the placement of monitors in urban areas.[13] Likewise, there is
wide variation in measured concentration. This empirical equation
serves as an estimate of the degree to which monitor placement is a
critical factor in measuring concentrations. Obviously, the equation
places strong emphasis on the urban activity surrounding the monitor,
the height of the monitor and the proximity of high ground level
activity. Identification and tabulation of these factors should
contribute to an understanding of the variations in measured concentrations
among monitors and changes in the site surroundings or height which
resulted in changes in measured levels. Unfortunately, other
important factors such as activity level and distance from the
source to the monitor cannot be considered quantitatively at this
time due to limitations of the data base used in this study.
4.5 Interpreting Dispersion Model Results
There is clearly no direct link between the empirical relationship
and the dispersion model. However, there should be no exclusive
consideration of either approach in the presence of additional
information. Clearly Equation 3.2 is imprecise in the area where it
is potentially most usefulthe identification of industrial impact.
Thus, dispersion models have been and remain important tools in
control strategy development and in assessing TSP problems. In
estimating source contributions through use of dispersion models, it
is possible that discrepancies in the data base, emission factors or
-------
40
dispersion parameters may exist. In such cases, the empirical model
results can be used as an aid in identifying the discrepancies and
for suggesting areas needing refinement, such as certain emission
factors or use of a different grid size. Recent improvements in the
capability of dispersion models, by incorporating particle-size
distributions and natural removal mechanisms are providing better
results for control strategy demonstrations and other uses.
-------
41
5 SUMMARY
A technique is suggested for estimating differences in annual
TSP levels due to monitor siting and location differences. To
accomplish this, a data base of 142 monitoring sites in 13 urban
areas was assembled and analyzed.
The data base was analyzed by using a model to describe TSP
levels at each site as a function of land use, monitor height, etc.,
and then using a multiple regression technique to estimate the
coefficients of each term. The results are summarized in Table 5.1.
In the regression, terms I-IV represent the non-industrial (NI)
portion of urban aerosol, term V represents industrial contribution
and the regression constant (13.3 yg/m3) represents the unexplained
portion.
The resulting empirical model can be used to estimate TSP
levels. This estimate can be useful in several ways.
1) The estimate can be compared with the actual concen-
tration at a site to identify those situations which differ sub-
stantially from the estimate. Thus, it becomes a screening technique
for identifying abnormal influences. It can also be used as a
screening technique for areas without monitors.
2) The estimate can be further broken down using data
from previous analyses to provide a preliminary estimate of sources
contributing to TSP levels. This preliminary estimate can be
refined through more extensive analysis or used in those situations
where a more refined estimate (by atmospheric diffusion models, or
-------
Table 5.1 Summary of Empirical Estimate of Annual TSP Concentration
COMPONENTS METHOD USED TO DESCRIBE COMPONENT
Primary Non-Urban Background
Secondary Particulates
III. Urban Activity Influence
IV. Local Sources
V. Industrial Sources
Urban TSP Minus Non Urban
Sulfates, Nitrates
Urban Sulfates plus Nitrates
Undeveloped Sites
Residential Sites
Commercial/Industrial Sites
H, monitor height, meters
Industrial Sites
'Near General Industry (< 2 km)
'Near Uncontrolled Steel Mill (< 2
km)
EMPIRICALLY DERIVED
OR CALCULATED VALUE
(measured)
(measured)
0 yg/m3
20 yg/m.3
31 yg/m.3
Residential/Commercial Sites
'Near Uncontrolled Steel Mill (2-10 km)
'Other Residential/Commercial Sites
calculated from
equation 45e"'02 H
15 yg/m-3
52 yg/m3
23 yg/m3
0 yg/m3
Total Estimated Concentration = .88 (I+II+III+IV) + V + 13.3 yg/m.3
ro
-------
43
from measurements such as filter analysis or special sampling) is
precluded by time or resource constraints-
3) The estimate can be useful in interpreting monitoring
data by estimating possible siting effects.
4) Comparing the estimate with dispersion model predictions
may help identify the causes of discrepancies between predictions
obtained with dispersion models and observations, such as certain
emission factors or use of a different grid size.
-------
44
6 REFERENCES
1. Frank A. Record, Robert M. Bradway, Gordon L. Deane, Rebecca C.
Galkiewicz and David A. Lynn, National Assessment of the Urban
Particulate Problem, Volume I. EPA-450/3-73-024, U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina, June 1976.
2. Frank A. Record, Robert M. Bradway, Gordon L. Deane, Rebecca C.
Galkiewicz and David A. Lynn, National Assessment of the Urban
Particulate Problem, Volumes VII, VIII, IX, X, XIII, XIV, XVI.
EPA-450/3-73-026e, f, g, h, k, 1, n, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina, July 1976.
3. Frank A. Record, Robert M. Bradway, Gordon L. Deane, Rebecca C.
Galkiewiicz and David A. Lynn, National Assessment of the Urban
Particulate Problem, Volumes IV, V, VI, XI, XII. XV.
EPA-450/3-76-026b, c, d, i, j, m, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina, July 1976.
4. F. W. Sears and M. W. Zemansky, University Physics: Part One
Third edition, Addison Wesley Publishing Company, Reading, Mass.,
1963, page 450.
5. Thompson G. Pace, Warren P. Freas and Elsayed M. Afify,
"Quantifications of Relationship Between Monitor Height and
Measured Particulate Levels in 7 U.S. Urban Areas." APCA
Paper 77-13.4, presented at the APCA Convention, Toronto, Canada,
June 1977.
6. Jolayne Service, A User's Guide to the Statistical Analysis System.
North Carolina State University Student Supply Store, Raleigh, North
Carolina, August 1972.
7. Frank A. Record, Robert M. Bradway, Gordon L. Deane, Rebecca C.
Galkiewicz and David A. Lynn, National Assessment of the Urban
Particulate Problem, Volume KIT. EPA-450/3-76-0261, U.S.
Environmental Protection Agency, Research Triangle Park, North
Carolina, July 1976.
8. Frank A. Record, Robert M. Bradway, Gordon L. Deane, Rebecca C.
Galkiewicz and David A. Lynn, National Assessment of the Urban
Particulate Problem, Volume I.EPA-450/3-73-024, U.S.
Environmental Protection Agency, Research Triangle Park, North
Carolina, June 1976, pp. 213-264.
9. Guideline for Development of Control Strategies in Areas with
Fugitive Dust Problems. OAQPS, No. 1.2-071, U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina,
October 1977.
-------
45
10. Control Strategy Preparation Manual for Participate Matter.
OAQPS 1.2-049, U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina, September 1977.
11. Interim Guideline on Air Quality Models. OAQPS 1.2-080, U.S.
Environmental Protection Agency, Research Triangle Park, North
Carolina, October 1977.
12. Users Guide for the EPA Industrial Source Complex Dispersion
Model (Preliminary Report).H. E. Cramer Company, Inc., Salt
Lake City, Utah, October 1977.
13. Frank A. Record, Robert M. Bradway, Gordon L. Deane, Rebecca C.
Galkiewicz and David A. Lynn, National Assessment of the Urban
Parti oil ate Problem, Volume I. EPA-450/3-73-024, U.S.
Environmental Protection Agency, Research Triangle Park, North
Carolina, June 1976, pp 78-92.
-------
47
7 APPENDIX
-------
49
7.1 DATA BASE
-------
^
SAROAD Code
Site
Mare
Neighborhood
Monitor
Height
(feet)
Neighborhood
Description
1974 Annual
Geometric
Mean
BALTIMORE
21 0120 001
21 0120 006
21 0120 007
21 0120 008
21 0120 009
21 0120 023
21 0120 024
Fire Dept. HQ
NE Police Sta.
NW Police Sta.
SE Police Sta.
SW Police Sta.
Fort McHenry
Fire Co #22
COMM
RES
RES
IND
RES
IND
IND
30
20
20
20
20
50
30
Near Expressway
Construction
105
53
68
105
85
102
95
en
o
-------
SAROAD Code
BIRMINGHAM
01 0380 005
01 2140 003
01 3200 001
01 0380 019
01 0340 001
01 0380 003
01 0380 012
01 1880 002
01 1300 003
01 0380 Oil
01 0380 010
01 0570 001
01 2540 001
Moni
Site - Heig
Name Neighborhood (fee
N. Birmingham IND 6
Leeds IND 6
Tarrant City IND 6
E. Thomas IND
6
Dessemer COMM 6
NASN COMM 45
Downtown COMM 10
Irondale COMM 6
Fairfield IND 10
West End RES 6
Wood lawn RES . 6
Huffman RES 6
Mountain Brook UNDEV 6
tor 1974 Annual
ht Neighborhood Geometric
t) Description Mean
Foundry Across
St. w/ Controls
By Passed 144
Small Rural Town
W/ Major Cement PI. 143
Gravel Quarry,
Coke Ovens, Cupola 130
On Road Downwind From
Steel Mill 1/4 Mile 128
General CBD And
Some Unpaved Pdg. 98
96
Near Well Swept
Street 94
Small Town CBD
Railroad Yards 90
Major Steel Works
6-10 Blocks 90
88
84
51
47
en
-------
SARD
AD Code
Monitor 1974 Annual
Site - Height Neighborhood Geometric
Name Neiahborhood (feet) Description Mean
CHATTANOOGA
44 0380
44 0380
44 1280
44 0380
44 0380
44 0380
44 0380
44 0380
020
006
003
017
019
021
015
024
City Hall COMM 50 --- 80
WDEF, W. Broad IND 30 Foundry With
Building Emissions 86
Brainerd RES 7 --- 60
Lookout Mtn. UNDEV 3 --- 38
E. Chattanooga COMM 25 Railway
Street Traffic 81
Silver Dale UNDEV 30 --- 38
Shall owford Rd. IND Quarry and Unpaved
10 Roads 101
APC Bureau COMM 12 Street Traffic 84
on
INJ
-------
1
1
1
CI
36
36
36
36
36
36
36
36
36
36
36
36
.
SAROAD Code
Monitor 1974 Annual
Site Height Neighborhood Geometric
Name Neighborhood (feet) Description Mean
NCINNATI
1220
1220
1220
3540
1220
1220
7700
5880
1220
1220
1220
5880
001
002
on
001
014
013
001
001
015
016
020
002
Public Library COMM 80 75
College Hill Fire Hse RES 30 57
Oakley Fire Hse COMM 25 Train yard and
Unpaved Parking 70
Lockland IND 20 100
Carthage Fire Hse IND 25 95
Price Hill Fire Hse RES 25 65
Wyoming RES 12 56
St. Dernaro IND 16 Truck Term.,
Train Yd., Material 130
Transfer
Corryville RES 35 72
Fairmont IND 25 Foundry, Paved Rds.
Railway 92
Drake Hospital RES 15 71
St. B. St Clements RES 45 76
tn
CO
-------
SARO
AD Code
Monitor I974 Annual
Site " - Height Neighborhood Geometric
Name Neighborhood (feet) Description Mean
CLEVELAND
36 1300
36 1300
36 1300
36 1300
36 1300
36 1300
36 1300
36 1300
36 1300
36 1300
36 1300
013
024
001
005
008
012
033
026
006
027
029
ARC Lab IND Unpaved Industrial
20 Steel Mills, Trains 175
Brooklyn YMCA RES 50 --- 76
Health Museum RES 24 --- 90
Pneumatic Tool IND 59 - 88
Fire St. #13 IND 23 147
Fire St. #19 COMM 25 Unpaved Pkg.
Gen. Commercial 124
St. Vincents Hosp. IND 4 - 149
Harvard Yards IND 60 Steel Mills, Etc.
Unpaved Rds. , Etc. 168
JFK School RES 60 - 50
P. L. Dunbar School RES 20 93
Suppl. Ed. Center COMM 65 Downtown
Unpaved Parking 112
in
-------
j SAROAD Code
MIAMI
10 3220 001
10 0860 013
10 2700 006
10 2700 003
10 0480 001
10 2760 001
10 2700 002
10 3040 001
10 4740 001
10 0860 003
10 2720 007
10 1766 001
10 0220 002
Site
Name Neighborhood
OPA Locka COMM
10001 NW 87th Avenue COMM
6400 NW 27th Avenue COMM
3700 NW 7th Avenue COMM
16770 NW 37th Avenue COMM
Miami Springs COMM
864 NW 23rd Street COMM
19th Avenue Miami Beach SPECIAL**
West Miami COMM
600 SW 87th Avenue RES
Was'h. Ave. Miami Beach SPECIAL**
Hialeah RES
Bay Harbor Island RES
Monitor
Height Neighborhood
(feet) Description
1974 Annual
Geometric
Mean
18 Bare Area
Extremely Dusty Area 86
13 Cement Trucks
Spillage Resuspended 79
29 Unpaved Berms
Sewer Construction 75
Near Major
1 4 Expressways
73
Suburban Shopping
20 Center Near Road 70
27
12 Near Unpaved
Parking Lot
21
16 Near Unpaved
Parking Lot
24
Hotel District
14 Near Street
35
12
70
69
42
56
44
42
54
50
**Near Ocean
in
in
-------
SAROA.D Cede
OKLAHOMA CITY
37 2200 015
Site
Name Neighborhood
428 W. Calif. COMM
Monitor
Height Neighborhood
(feet) Description
On Fire Station
1974 Annual
Geometric
Mean
15 Street, Urban Ren. 92
37 2200 001
200 N. Walker COMM
70 On Courthouse
Near Urb. Renewal 55
37 2200 002
37 2180 005
37 1940 006
37 1940 010
37 0260 014
37 2200 018
37 2200 019
37 2200 020
37 2200 021
37 3300 022
37 0940 016
37 2200 017
SW 66th & Denning UNDEV
SW First & Main UNDEV
300 Mid America RES
NE 10th & Douglas RES
3919 N. Rockwell RES
2045 NW Tenth COMM
NW Hi way & Meridian UNDEV
Ranger Station UNDEV
SE 74 & Hiway UNDEV
SW Second & Robinson COMM
Edmund RES
NE 13th & Phillips COMM
14
14
41
39
14 Unpaved Parking
Near Fire Station 59
14
4 On Ground W/
Apt. House Const
Near Corner
8 Traffic
15
15
15
Near On Ramp
12 Traffic
15
15 On Office Near
49
80
98
62
43
54
101
53
CJI
«\
Const, of Hospital 89
-------
SARO/V) Code
PHILADELPHIA
39 7140 020
39 7140 001
39 7140 024
39 1400 004
39 7140 008
39 7140 026
39 1400 003
39 7140 022
39 7140 019
S"i*°
'.3,~2
Neighborhood
Monitor
Height
(feet)
i
1974 Ann
Fieighborhcod Gecrat-.ri
DescriptiOi: Msari
Belmont Filter RES 13 72
Roxboro Filter RES 13 59
N.E. Airport RES 13 64
AMS Lab RES 17 78
Franklin Inst (CAMP) COMM Street Traffic
11 Unpaved Parking 119
S. Broad & Spruce COMM 13 Street Traffic 115
500 S. Broad COMM 35 76
Defense Supply IND 13 Railway 105
Allegheny River IND 13 Grain Handling 122
39 7140 021 Int. Airport
Aramingo Fire St.
IND
IND
Coal Storage
13 Refineries
General Dusty
Area
35 Paint factory
50 yds.
94
116
en
-------
SAP.GAD Code
Site
Name
Neighborhood
Monitor
Height
(feet)
Neighborhood
Description
1974 Annual
Geometric
Mean
PROVIDENCE
41 0300 005
41 0300 006
41 0300 007
41 0120 003
41 0100 002
41 0300 008
41 0100 001
41 0120 001
St. Office Bldg. COMM
Westminster St. CCMM
Dyer Street COMM
Tristam Burges Sch. RES
General Hospital RES
St. Josephs Hospital RES
Police Station COMM
Jr. High School COMM
50
100
15
30
50
110
45
48
Traffic Island
Near Expressway
63
68
61
43
46
53
49
en
00
-------
SAPOPD Code
ST. LOUIS AREA
26 0200 001
26 0200 002
26 0260 001
26 1040 002
26 2630 002
26 2630 003
26 4120 001
26 4300 003
26 4280 061
26 4280 066
26 4280 006
26 4280 007
14 0160 004
14 2120 006
14 2120 008
14 2960 005
Site ' -
Name Ne
Monitor
Height
ighborhood (feet)
Neighborhood
Description
907 Chambers Road COMM 25
Rt 67 & 1270 UNDEV 10
Holman School RES 4
Clayton Health COMM 55
Lemay Mt. St. Rose RES 4
Lemay AC 1C I NO
4
St. Ann RES 9
Coking and
Ti02 Production
Street Traffic
Old Jamestown UNDEV 4
Shreve Rd. & 170 COMM
15
River & Sulfur Ave. IND
10
Street Traffic
Railway
Unpaved Parking
And Quarry
Munic Courthouse COMM 60
8227 SDBWY IND 35
Alton COMM 32
City Hall COMM 50
Cahokie COMM 15
Unpaved Parking
Truck Terminals
Granite City #1 COMM 15
1974 Annual
Geometric
Mean
78
50
59
64
71
128
65
40
112
93
80
126
69
89
109
117
en
to
-------
SAROAD Code
Site
Name
Neighborhood
Monitor
Height
(feet)
Neighborhood
Description
1974 Annual
Geometric
Mean
ST. LOUIS AREA (Continued)
14 8520 007 Wood River
14 2960 008 Granite City #06
14 2960 009 Granite City #07
14 2960 010
26 4280 010
26 4280 012
26 4280 032
Granite City #08
Donovan Avenue
Munic. Arts
St. Louis Univ.
IND
IND
IND
IND
RES
RES
COMM
15
25
25
15
30
35
38
Unpaved Parking
Near Steel Mill
72
85
158
118
61
59
70
-------
SAP.QAD Cede
SAN FRANCISCO
05 5880 001
05 5300 004
05 6860 001 .
05 6860 003
05 8080 001
05 6300 003
05 6240 001
05 4020 002
05 0740 001
05 6980 004
Site
Neine Neighborhood
Pittsburg COMM
Oakland NASN COMM
San Francisco NASN COMM
San Francisco COMM
Sunnyvale RES
Richmond COMM
Redwood City COMM
Livermore COMM
Berkeley NASN RES
San Jose COMM
Monitor
Height Neighborhood
(feet) Description
20 Gravel Parking
Lots
62 Expressway 25 yds
truck terminals
32
15
33
17 Generally dusty
area, dirt lot
adjoining
11 Street traffic
truck terminals
1974 Ar.nua
Geometric
Mean
50
52
51
53
41
50
50
18 Gravel pits unpaved 74
parking, hghwy const.
90
15 heavy traffic
59
58
gravel parking lot
CT>
-------
SAROAD Code
Site
Name
Neiahborhocd
Monitor
Height
(feet)
Neighborhood
Description
1974 Annual
Geometric
Mean
SEATTLE
49 1840 058
49 1840 001
49 1840 013
49 1840 057
49 1840 066
49 1840 059
McMicken Hts.
Duwamish Fire Sta.
Public Safety
Food Circus
4500 Marginal Way
Harbor Island
RES
RES
COMM
COMM
IND
IND
D.O.E. 6700 Marginal Way IND
15
25
80
70
20
15
15
35
48
63
45
Street, Railway,
Gen Industry 68
Grain Mill, Battery,
Cement, Steel 77
Street Traffic
105
-------
SARQAD Code
WASHINGTON, DC
09 0020 003
09 0020 001
09 0020 015
09 0020 009
09 0020 005
09 0020 012
09 0020 Oil
09 0020 007
09 0020 008
Site
Name
427 N. J. Avenue
Municipal Center
Catholic Univ.
Amer. Chem. Soc.
Brightwood Police
National Arboretum
Cleve Pk. Library
Nevel Thomas Sen.
General Hospital
"1 ' Monitor
Height Neigh
Neighborhood (feet) Descr
1974 Annual
Dorhood Geometric
iption Mean
COMM 15 Bldg. Const.
Street Canyon 102
COMM 80
92
RES 50 University Near
Cut and Cover Subway 70
COMM 100
67
COMM 30 Commercial < 1 mi . to
Cut and Cover Subway 63
UNDEV 3
COMM 25
RES 3
RES 40
59
54
54
52
en
GO
-------
64
7.2 Description of Activity Levels
HIGH ACTIVITY
Presence of a network of several two and/or four lane roads which
collectively contribute to a high level of urban activity. Examples
of this would be commercial or industrial areas. Areas which are
primarily residential except for perhaps a single moderately traveled
street would not usually qualify. Likewise, the commercial activity
in a small rural community would usually not be high activity. How-
ever, the presence of one or more of the following in a primarily
residential or small community, and within 1/4 mile of the site,
would suggest high activity:
'Expressways or major arterials carrying more than 25,000 ADT.
'Roads with moderate traffic (over 3,000 cars per day) and a
noticeable accumulation of dirt in the traffic lanes.
'Construction activity with demolition, grading or mud carry-
out of several months duration. Shorter or less intensive
activities should not be counted.
'Active unpaved roads or parking areas. Activity of over 20
to 30 vehicles per day would be a general guideline but the
length of travel and speed would be appropriate considerations
that would modify this guide.
'Industrial plants with large and active unpaved areas within
or around the plant.
LOH ACTIVITY
Low activity would be suggested by the following:
'Primarily residential, undeveloped or seldom used areas.
"Small rural communities.
'Sites with nearby activity screened by trees or buildings
from the monitor.
-------
65
7.3 Listing of Local Activity for Residential and
Commercial Sites
City
Chattanooga
11
11
11
Seattle
it
11
11
Providence
"
11
ii
11
H
11
Wash. D.C.
"
II
"
"
II
Okla. City
"
II
II
"
"
"
II
San Francisco
11
11
"
"
n
n
it
11
Miami, Fla.
"
11
II
"
11
II
II
II
11
n
Site
City Hall
E. Chattanooga
APC
Bra i nerd
Pub Safety
Food Circus
McMicken Hts.
Duwamish Fire Station
State Office
Westminister
Police Station
Dyer St.
Tristam Burgess School
General Hospital
St. Joseph Hospital
ACS
Cleveland Park
Brighton Police
Camp
Nevel Thomas School
General Hospital
West California
North Walker
NW 10th
2nd Robinson
13 and Phillip
Mid America
10th and Douglas
Edmund
SFR
SU
PT
OK
RI
RC
SJ
LI
BK
12-Westwood
20-NW 87
10-NW 27
8-NW 7
1-NW 23
16-SW 62
19-Ali Baba
600-SW 87
Hialeah
Bay Harbor Island
28-NW 37
Height, Ft.
50
25
12
7
80
70
15
25
50
100
45
15
30
50
110
100
25
30
15
3
40
15
70
8
12
15
14
14
15
32
33
20
62
17
11
15
18
90
27
13
29
14
12
16
18
24
14
12
20
Annual Geometric
Mean ug/m"'
80
81
84
60
63
45
35
48
63
68
66
88
61
43
46
67
54
63
102
54
52
92
55
98
107
89
59
49
53
51
41
50
52
50
50
58
74
59
70
79
75
73
69
56
86
44
54
50
70
Activity
Level
High
High
High
Low
High
Low
Low
Low
High
High
High
High
Low
Low
Low
High
Low
High
High
Low
Low
High
High
High
High
High
High
Low
Low
High
Low
High
High
High
Low
High
High
Low
High
High
High
High
High
High
High
Low
Low
Low
High
-------
66
7.4 Tabulation of PNB and USN Data
Urban
Area
Chattanooga
Miami
Okla. City
Providence
San Francisco
Seattle
Washington, DC
Cleveland
Birmingham
Philadelphia
Baltimore
St. Louis
Cincinnati
Total
Non-urban3
35
25
25
30
15
15
30
30
30
35
35
25
35
Non-urban
Sb
8
7
4
9
3
4
11
13
9
13
10
8
14
Non-urban
Nb
2
1
1
1
1
0
2
1
1
1
1
1
1
PNB
25
17
20
20
11
11
18
16
20
21
24
16
20
Urban
USC
15
7
4
12
7
9
16
13
18
18
13
15
15
Urban
UNC
2
1
3
3
4
3
4
3
3
4
3
3
3
(SUM)
USN
17
8
7
15
11
12
20
16
21
22
16
18
18
a From non-urban NASN network (PNB Total non-urban Non-urban Sulfates, Nitrates)
Calculated as ammonium sulfate, nitrate
From urban NASN network
-------
67
7.5 Nomenclature and Abbreviations
TSP Total Suspended Participate
PNB Primary Non-Urban Background
USN Urban Sulfates-Nitrates
LS Local Source
UA Urban Activity
A
y Predicted concentration, ug/m3
H Height of monitor, meters
a Empirically derived constant
b Empirically derived constant
c Empirically derived constant
e Base of natural logarithm, 2.71828
R-square The square of the correlation coefficient
ADT Average Daily Traffic
B-j thru B. Regression coefficients
Be Regression constant
x Slope of Linear Regression Line
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-450/2-78-016
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
"An Approach For The Preliminary Assessment
of TSP Concentrations"
5. REPORT DATE
July 1978
6. PERF'O'R'MING ORGANIZATION CODE
7. AUTHOR(S)
Thompson G. Pace
8. PERFORMING ORGANIZATION REPORT NO,
9. PERFORMING ORGANIZATION NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Monitoring and Data Analysis Division
Research Triangle Park, N.C. 27711
10. PROGRAM ELEMENT NO.
2AA635
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Air quality data for Total Suspended Particulate (TSP) in 13 U.S. urban areas was
examined. The data from 142 monitoring sites were grouped so that residential and
commercial sites in non- or light-industrial urban areas could be examined. A relation
ship between height and concentration was noted at the sites with nearby ground-level
activity due to traffic, parking, etc., such that the concentration decreased expon-
entially with increasing height of the monitor above ground. No such relationship was
found at sites with no ground-level activity. Commercial and industrial sites were
found to be near ground-level activity in 90 percent of the cases examined while resi-
dential sites were virtually never located near such activity.
The entire data base was then examined using a multiple regression procedure to
estimate the relative impacts of non-industrial, general industrial and steel mill in-
fluences on TSP levels. Non-industrial influences were found to account for over half
of the total concentration estimate in all cases.
Several potential applications of the linear regression technique are suggested.
It can be used as a screening technique for examining TSP data to identify sites with
unusual concentrations or to provide a preliminary estimate of sources of TSP. It can
be used to interpret the variations in TSP data by estimating siting effects and it can
help to identify the causes of discrepancies between predictions obtained with disper-
sion models and observations.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COS AT I Field/Group
Measurement
Particulates
Model
Monitor Siting
Hi-volume Sampler
Land Use
Steel
Urban Activity
Height
Fugitive Dust
Resuspension
18. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
77
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDI TICN i s OBSOLETE
68
------- |