EPA-600/4-81-002
February 1981
METHODOLOGY FOR DESIGNING AN OPTIMUM
AIR QUALITY MONITORING NETWORK
by
Mei-Kao L1u and Joel Avrin
Systems Applications, Incorporated
950\Northgate Drive
San Rafaet, California 94903
James L. McElroy, Project Officer
U.S. Environmental Protection Agency
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada 89114
Contract No. 68-03-2446
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NEVADA 89114
-------
EPA-500/4-81-002
February 1981
METHODOLOGY FOR DESIGNING AN OPTIMUM
AIR QUALITY MONITORING NETWORK
by
Mei-Kao Liu and Joel Avrin
Systems Applications, Incorporated
950 Northgate Drive
San Rafael, California 94903
James L. McElroy, Project Officer
U.S. Environmental Protection Agency
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada 89114
Contract No. 68-03-2446
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NEVADA 89114
-------
DISCLAIMER
.This report has been reviewed by the Environmental Monitoring Systems
Laboratory, U.S. Environmental Protection Agency, and approved for
publication. Approval does not signify that the contents necessarily reflect
the views and policies of the ~.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
ii
-------
CONTENTS
Page
Figures. . . . . . . . . . . . . . . . . . . . .
Tables. . . . . . . . . . . . . . . . . . . . .
Acknow1 edgment . . . . . . . . . . . . . . . . .
. . . . . . . . . . .. iv
. . . . . . . . . . .. v
. . . . . . . . . . . . vi
1. Summary..
. . . . .
. . . . .
. . . .
. . . . .
. . ... . . .
1
2
2.
Introduction
. . . . .
. . . . .
. . . .
. . . . . . . . . . . .
3.
Identification and Ranking of Potential Monitoring Sites. . . .
4
4. Determination of Spheres of Influence and the Optimum
Monitori ng Network. . . . . . . . . . . . . . . . . . . . . .
7
5. Application of the Siting Methodology to the Las Vegas Valley. . 12
The first step--identifying and ranking monitoring sites. . . 14.
The second step--determining the minimum number of
stat ions req ui red. . . . . . . . . . .. . . . . . . . . .. 18
6. Concl udi ng Remarks.
. . . . .
. . . . .
. . . . . . .
. . . . . 25
References
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . .
. . 26
Appendices
A
B
A Heuristic Approach for Optimal Monitoring Network Design. .. 28
Sample Spatial Correlation-Coefficient Isop1eths in
La s Vega s Vall ey . . . . . . . . .. . . . . . . . . .
. . .. . . 31
fif
-------
Number
1
2
3
4
5
6
7
8
9
FIGURES
Page
Generalized correlation coefficient as a function of
di stance. . . . . . . . . . . . ... . . . . . . .
. . . . . .
8
Confidence interval (belts) for the correlation coefficient at
95 ~erJ::_ent CQnfidence level. . . . . . . . . . . . . . . . . . 10
Joint areal coverage for monitoring stations. . . . .
. . . .. 11
Map of the Las Vegas Valley. . . . . . . . . . . . .
. . . . .. 13
Ranked monitoring locations for CO based on one-hour averaging
period near morning traffic peak: Las Vegas Valley. . . . . . 15
Ranked monitoring locations for CO based on one-hour averaging
period near evening traffic peak: Las Vegas Valley. . . . . . 16
Ranked monitoring locations for CO based on eight-hour averaging
period near evening traffic peak: Las Vegas Valley. . . . . . 17
An optimum air quality monitoring network for Las Vegas Valley
based on a cutoff sample spatial correlation coefficient
of 0.8 . . . . . . . . . . . . . . . . . . . . . . e. . . . . . 21
An optimum air quality monitoring network for Las Vegas Valley
based on a cutoff sample spatial correlation coefficient
of' 0.5 . . . . . . . . . . . .. . . . . . . . . . . . . . . . . 23
iv
-------
Number
1
2
3
4
TABLES .4:'
Page
Cutoff Sample Correlation Coefficient to Ensure
Minimum Values for Population Correlation
Coefficient and Variance Explained for 95~ Confidence
with 78 Samples. . . . . . . . . . . . . . . . . . . . .
. . . 19
Identification of Potential Monitoring Locations in the
Las Vegas Valley. . . . . . . . . .. . . . . . . . . . .
. . . 20
Summary Statistics for Selected Monitoring Stations in Las Vegas
Valley Based on a Cutoff Sample Spatial Correlation
Coeffi ci ent of 0.8. . . . . . . . . . . . .. . . . . . . . . 0 . 22
Summary Statistics for Selected Monitoring Stations in Las Vegas
Valley Based on a Cutoff. Sample Spatial Correlation
Coefficient of 0.5. . . . . . . . . . . . . . . . . . . . . . . 24
v
-------
ACKNOWLEDGMENT
We would like to express our sincere thanks to our colleagues, Dr. Martin
J. Hillyer, Dr. Philip M. Roth, and Dr. C. Shepherd Burton for many enlighten-
ing discussions. From these brainstorming sessions, the approach presented in
this report began to take shape. We would also like to thank Dr. George
Flatman of the Environmental Monitoring Systems Laboratory of the U.S.
Environmental Protection Agency for providing critical and timely advice that
helped finalize the statistical aspect of this study.
vi
-------
SECTION 1
SUMMARY
An objective methodology is presenteq for deternrining the optimum number
and disposition of ambient air quality stations in a monitoring network. The
proposed methodology uses climatological information and an air quality simu-
lation model. First, the climatological information is used to generate a
limited number of meteorological scenarios representative of the region of
interest. For each of the scenarios, the air quality simulation model is
employed to produce the corresponding temporal-varying air quality patterns.
The air quality patterns serve as the primary data base i~ a two-step
procedure for determining the monitoring network. In the first step, the air
quality patterns are collapsed into a single pattern through the use of the
figure-of-merit (FOM) concept. For a specific time interval and location, the
FOM is determined as the sum over the meteorological scenarios of the products
of the pollutant concentrations and the associated probabilities of occur-
rence. The identification and ranking of the most desirable monitoring
locations are achieved using the resultant FOM fields. In the second step,
the network configuration is determined- on the basis of the concept of a
sphere of influence (SOl). The 501 are dictated by a cutoff value in the
spatial correlation coefficients between the predicted pollutant concentra-
tions at the monitoring locations identified and the corresponding
concentrations at neighboring locations in the region. This cutoff value is
related to an estimate of concentration variations that can be accounted for
by a given monitoring station. The minimum number of monitoring stations
required is then determined by deleting lower-ranked stations whose 501
overlap the 501 of higher-ranked stations and whose 501 provide non-
overlapping coverage of less than some fixed percentage of the coverage of
the 501 of the higher-ranked stations.
As a demonstration, the siting methodology was applied to the metropolitan
Las Vegas area for carbon monoxide monitoring stations. A 10-station network,
consisting of 7 stations for average peak concentrations and 3 stations for
background concentrations, was selected for a desired minimum detection
capability of 50 percent of the concentration fluctuations {95 percent of the
time} and hence a cutoff spatial correlation coefficient of 0.8 ~etworks with
fewer stations would be selected if smaller minimum detection capabilities of
concentration fluctuations are deemed acceptabl e, and, vice versa.
'-
1
-------
,-
SECTION 2
INTRODUCTION
. The Clean Air Act requires state and local agencies to monitor ambient air
quality, primarily for documenting an area's compliance with the National.
Ambient Air Quality Standards (NAAOS). Additional monitoring may be required
to satisfy secondary objectives such as providing background or baseline con-
centrati ons. Currently, the determi nati on of the number and 1 ocati on of moni-
toring stations required in a network is primarily based .on subjective consid-
erations; semiquantitative rules supported by experience; .or sometimes,
limited use of analytical tools such as simple Gaussian models (Ludwig and
Kealoha 1975). Nontechnical considerations, such as convenience and
accessibility, are usually the dominant factors in selecting a specific
monitoring location within the area of interest. On the other hand, because
of the fluctuations in pollutant emission rates and the turbulent nature of
the atmosphere, pollutant concentration distributions are highly variable,
both in time and space. The concentrations measured at any given site depend
on the emission patterns as well as the atmospheric conditions. The design of.
an optimal monitoring network, therefore, requires an a priori knowledge of
these concentration variabilities. An objective methodology for designing
such a monitoring network is proposed and demonstrated in this 'study.
The methodology uses a data base consisting of a comprehensive set. of
pollutant concentration distributions representative of the region of
interest. The practice of using simulated concentration distributions
generated using an air quality model was adopted for this study; few, if any,
regions have a monitoring network in operation over a sufficient time interval
and of sufficient density to yield the requisite concentration distributions.
The air quality model, of course, produces the distributions by linking the
source emissions with the prevailing meteorological conditions.
The actual siting methodology consists of two steps. me goal of the
first step is simply to ascertain the most favorable locations for making air'
quality measurements. To the goal, a concept called the figure-of-merit,
introduced in an. earlier study (McElroy et al., 1978), is used to facilitate
the identification and ranking of potential monitoring sites. Procedures that
utilize this concept for such a purpose and that develop the data base for use
with the entire. methodology are delineated in section 3. . .
The goal of the second step in the siting methodology is to determine the
minimum number of monitoring stations and, hence, the optimum network
configuration. The spatial correlation coefficient, which is commonly used in
2
-------
statistics and turbulence research, is employed as the parameter with which to
measure the relevance of air quality at one point to that of another point in
its neighborhood. By imposing a minimum value for this coefficient, a sphere
of influence for each potent1al monitoring site can be defined. Subsequently,
by deleting redundancies among.the monitoring stations identified and ranked
in the first step, the optimum monitoring network containing a minimum number
of monitoring stations thus can be determined. The theoretical framework
underlyi~g this approach is described in section 4.
As a demonstration of its utility, the entire siting methodology was
applied to the metropolitan Las Vegas area. A relatively inert pollutant
species, carbon monoxide, was used as an example. The. application of the
siting methodology to Las Vegas is discussed in section 5.
3
-------
SECTION 3
IDENTIFICATION AND RANKING OF POTENTIAL MONITORING SITES
This section discusses the first step of the objective methodology for the
design of an optimum air quality monitoring network -- the procedures for the
identification and ranking of potential monitoring locations. As discussed in
prior studies, the desirability of placing an air quality monitor at a given
location is closely related to specific monitoring objectives (Ludwig and
Kealoha 1975, Ott 1975, Liu et al. 1977, and Ludwig, Berg, and Hoffman 1976).
In general, the primary objective of an air quality monitoring network is to
monitor the highest concentrations in the area of interest to ensure
compliance with air quality standards. These monitoring sites are labeled by
Ott (1975) as the "A"-type stations and by Ludwig, Berg, and Hoffman (1976) as
the "street canyon" and "traffic corridor" stations. In addition to this
primary objective, other secondary objectives for air quality monitoring also
exist. For example, as discussed by Ott (1975) and Ludwig, Berg, and Hoffman
(1976), additional stations may be needed either to meas.ure the population
exposure in a residential area or to provide the background as baseline con- .
centrations typical of the outlying rural areas. The fonner, called the "C"-
type stations by Ott and the "neighborhoodll stations by Ludwig, Berg, and
Hoffman, would require additional information, such as demographic data. The
latter, called the "E"-type stations by Ott and "regional" stations by Ludwig,
Berg, and Hoffman can, however, be incorporated into the present siting algor-
ithm, which is designed prinarily for locating the pollutant concentratio.n
maxima.
In an earlier study by McElroy et al. (1978), the desirability of placing
an air quality rmnitor at a given location in an urban area was accomplished
using a concept called the figure-of-merit (FOM). In its most .general fonn,
the FOM can be defined as the sum over an exhaustive or comprehensive set of
conditions of the products .of an air quality index either observed or
predicted and the associated probability of occurrence:
FOM = L (Ai r Quality Index) . (Probabi lity of Occurrence) .
(1 )
The summation is to be performed over all meteorological scenarios potentially
leading to high air pollutant concentrations. The FOM contains weighting by
the probabilities of occurrence of scenarios to avoid situations related to
extremely rare events or per10ds. These situations. would not necessarily
prQvide the best criteria for deternrining a permanent or senripermanent site
for a. monitoring network.
The air quality index in Equation. (1) can be a composite of several
pollutant concentrati ons, wei ghted again'. by the rel ati ve importance of the
4
-------
individual species, if it is desirable to design a multiple pollutant-species
monitoring network. For example, to locate a site for measuring multiple
pollutant species, the air quality index in Equation (1) can be generalized
using a composJte concentration index proposed by Ott and Thom (1976):
N
I = l wf.c1.
1.=1
,
(2)
where I denotes the overall air quality index, c1. is the concentration of
species f., and Wf is the corresponding weighting factor reflecting the
importance of po lutant species, 1., in the assessment of the overall air
quality. In general, c1. can be either an observed or predicted'
concentration. For the sake of simplicity, only one pollutant species is
considered in the present study. In this case, the FOM at any location can be
defined as the sum of the products of concentrati~n for a specific pollutant
and the associated probability of occurrence of the corresponding
meteorological conditions which are in turn based on available local
climatological data:
M (Concentration at location) (Probability Of)
FOM(x,y) = I (x,y) under meteorological. meteorological.
k=1' pattern k , pattern k
As an alternative, with special emphasis on the detection of maximum
concentrations exceeding the NAAQS, the FOM can be defined as a step function
of the pollutant concentration in a similar manner:
(3)
. FOM(x,y) =
M'
r
k=1
1, if NAAQS or some
fraction thereof is
exceeded at location
(x~y) under meteo-
rological pattern k;
0, i f not.
( Probabi 1 i ty )
of meteorological
. pattern K .
(4)
In Equattons (3) and (4), the pollutant concentration can be either observed
or expected. In the present study, the concentration fields are generated
from an air quality simulation model, which plays a central role in the siting
methodology by linking the known emissions distribution with air quality pat-
terns for a given meteorological scenario. The simulated temporal-varying air
quality patterns, when combined with the corresponding statistics, pe~it the
detennination of the FOM as per Equations (3) and (4). '
A selection of the most favorable air quality monitoring sites can then be
--_..._~~~~mp]J~~,~--~Y _~~~e~~~..~j!1.9 ~~~-~~~~_<:)'!t.!9~.
-------
easily be carried out numerically, essentially completes the first step of the
present siting methodology.
After the identification of the maxima in the FOM field and their ranking
according to the corresponding FOM values, two issues remain to be resolved
before the optimum monitoring network can be developed: The first issue is
related- to the representativeness of air quality data for a selected monitor-
ing station, and the establishment of an area surrounding this station for
which the data can be extrapolated. The second issue is concerned with the
mini mum number of measurement stations needed to obtain sufficient monitoring
coverage, as determined by the capability of detection of concentration
fluctuations by a given monitoring network. These two issues, apparently
interrelated, are addressed in the next section.
-------
SECTION 4
DETERMINATION OF SPHERES OF INFLUENCE
AND THE OPTIMUM MONITORING NETWORK
The determination of the minimum number of monitoring stations required
appears to be the most crucial element in developing an optimum air quality
monitoring network. Intimately related to this element is the determination
of the spatial coverage, or the sphere of influence (501), for each of the
monitori ng 1 ocati ons. In thi s context,. the 501 is defi ned as the surroundi ng
area over which the air quality data for a station can be considered to be
representative. .
Obviously, the specification of an 501 for any selected site is not
unique. Its establishment depends on the method of reconstructing, either
through interpolation, or extrapolation, the concentration field from the data
obtained for a given site. It is ~onceivable that different interpolations or
weighting methods can yield different SOl if the interpolation error is to be
kept to a minimum. For example, a linear interpolation might yield a 501
different from an inverse distance interpolation. In the former case,
gradients are assumed to be constant and can be positive or negative. In the
latter case, the gradients are not spatially constant and are generally
negative; that is, as one progresses outward, the extrapolated values decrease
monotonically. In an earlier study, a heuristic approach based on the
geometry of the computed FOM field was adopted (Liu and Moore 1980). For
comparison purposes, the results of this alternative approach are summarized
in Appendi x. A.
A more rigorous approach 1s adopted in the present study. This approach
is based on the statistical properties of the spatial distributions of the
pollutant concentration distributions used in the first step of the siting
methodology. Analogous to' the study of turbulence in aEu 1 e ri an framewo rk, a
spatial correlation coeffici~nt is' introduced between values of pollutant
concentration at a given site and the corresponding values at its neighboring
points as a function of radial di'stance away from the station:
cov [C(so), C(so + AS)]
/ var [C(so)] . var [C(so + AS)]
r(so,so + AS) =
,
(5)
where C(so) and C(so + AS) can be measured or aredicted values at the
points So and So + AS and AS=[(AX)2 + (Ay)2]1/2. The symbols cov
and var denote covariance and variance, respectively. Statistically, the
correlation coefficient provides a measure of the intensity of association
7
-------
between C(so) and C(so + ~), namely, the concentrations at the monitor-
ing sites and those at its neighboring points. This coefficient, lying
between -1 and +1, thus by itself furnishes an ideal dimensionless tool for
the determination of the SOl. Similar to the characteristics of the correla-
t ion coeffi ci ent common ly used in the study of turbulent vel oci ty, tempera-
ture, and concentration fluctuations, the spatial correlation coefficient is
expected to initially decrease from one as the distance increases. Conse-
quently, a cutoff distance of Sc can be found to determine the SOl for a
predetermined minimum spatial correlation coefficient as is illustrated in
Fi gure l.
Thus, in the second step of this siting methodology, the spatial correla-
tion coefficients surrounding each of the potential monitoring sites are
evaluated. The computation can be carried out along all radial-directions
until the spatial correlation coefficient falls below a predetermined minimum
or cutoff value.. Consequently, the SOl for each of the stations identified in
the first step of the current siting methodology can be determined.
The choice of the cutoff value for the spatial correlation coefficient can
be determined statistically for a given monitoring site. Assume that Cl =
(Cll, C12, . . . Cln) and C2 = (C2~, C22, . . . C2n) denote the pollutant
concentrations at the monitoring s1te and the corresponding pollutant.
concentrations at a neighboring point, respectively.. A computational form of
equation (5) for Cl and C2-with a sample si~e, n, is given by .
(+)
r
rc
Sc
s
o
._-, .
( )
. -- - - ,. .
- -.
Figure 1. Generalized correlation coefficient as a function of distance.
8
-------
n
r = i~l
n
i~l
(Cli - C1)(C2i - C2)
. n
(Cl i - Cl)2 i~1 (C2i - C2)2 .
,
(6 )
where
n
C1 = 1 l C1 i
n i=1
and
n
C2 = 1 l C2i
n i=1
Assuming that these two correlated random variables, C1 and C2, are from a
bivariate normal distribution, a general expression can be derived for the
probability distribution of a correlation coefficient, r, associated with .a
sample size, n, randomly drawn from an infinite population with a true
correl ation coeffi cient p. Thi s probabil ity di stri buti on is gi ven by Davi d
(1938) .
p (r In, p ) =
n-1
(1 - p2)2
,,(n - 3) I
.
n-4 ]
(1 - r2)2 dn-2 r arc cas (-pr)
d(pr)n-2 lJ 1 -(pr)2
(7)
The probability integral, E, given by
r2
E = f
rl
then represents the confidence level of the test hypothesis that r = Po with
r1 < p < r2 as -the confidence interval. .
p(rln,po) dr ,
(8)
The probability integral, Equation (8), can be evaluated using a
quadrature method (David 1938). In Figure 2, the confidence interval for
correlation coefficients at a 95 percent confidence level are reproduced from
TableLQ,f the Ordi~nd Pr~abili~he Distribution of the
. Corre 1 at ion Coeffi c1 ent i n Sma 11 Samp 1 es Da vi d 1938 .
9
-------
-u -u -t.I -tJ -u -u -u -U -U -tJ
+u
+u
-oJ
-II
~ ~
.... ~ I
l.;o ~
10
'II
~ 1/ VIII
1/
,
,/ »
~
1/
1/
, 1/ II
.,
lhr/!.
....
. .OJ +1.2 +U +0.8 .as +0.& +4.7.oJ +u +1.0
+1.0
+u
+u
+u
+o..a
+QJ
+u
+0.1
+0.5
-
1+0.5
~
Au
j
.~
..
==.0.1
B
i 0
=
~
Q,
cf-G.Z
. .
--u
, "
'O-u
.'
. ii-u
.r)i
-OJ
+t.
+0.3
+0.2
+u
o
-QJ
-a.z
-u
-u
-tS
-u
-QJ
-u
-0..1
-OJ
~ -1.0.
~~~~~~~~~~.~~~~~~~~~~
Scale 01 r (-Slmpte CorrNtion Coefficient)
The numtlerson the CUMIS indicate sample size
- _. . . .-. .
Figure 2. Confidence interval (belts) for the correlation coefficient at
95 percent confidence level. (This figure is reproduced with permission
of E. S. Pearson, from David (1938), Cambridge University Press for
the Biometrika Trustees).
If a linear relationship is assumed between the variables, then the square
of the correlation coefficient represents the fraction of the 'variance of
C2 which can be accoun~ed for by the variations in C1 (Ezekiel 1941).
Note that the use of r or pZ in this way does not imply a causal rela-
tionship between C1 and C2 but that a linear association exists between
them such that 100 pZ is the percentage of concentration variations
explained by concentration variations at a potential monitoring site. Thus,
variance explained = Pc2
(9)
10
-------
Once the minimum desired value of variance explained is selected, the
minimum acceptable value. of Pc can be obtained from equation (9). This
value for pc can be entered into a nomograph such as Figure 2 as a lower bound
to obtain a cutoff value for the sample coefficient, rc, for a specific
sample size and hence establish the S01 for each of the ranked stations:
Al = A(Xl, Yl)
A2 = A(X2, Y2)
A3 = A(X3, Y3)
. . .' t
AN = A(XN, YN)
,
where Ai, 1 i i iN, is defined by all points (x, y) that lie in the simply
connected region containing station (Xi, Yi) enclosed by a contour
detennined by r > rc. The total areal coverage by the monitoring network
for all N stations, as illustrated in Figure 3, is given by
ANetwork = Al U A2 U A3 U . . . U AN .
(10)
The determination of the minimum number of monitoring stations required can
be then carried out by deleting lower ranking stations whose SOl overlap the
501 of the higher ranking stations and whose SOl provide non-overlapping
coverage of less than some fixed percentage of the coverage of the higher
ranking stations. .
A, u "2 II A3 II A4
-(X4° y 4)
\ .
A4
A3
-(13° "3
Figure 3.
Joint areal coverage for monitoring stations.
11
-------
SECTION 5
APPLICATION OF THE SITING METHODOLOGY
TO THE LAS VEGAS VALLEY
The methodology for designing an optimum network described in the previous
two sections was applied to the metropolitan Las Vegas area. Although the
procedures and the tools developed for this siting methodology are applicable
to chemically inert as well as reactive pollutants, only a relatively inert
species -- carbon monoxide -- was considered in the present study. Las Vegas
is a relatively isolated, urban .community surrounded by desert, with a
population of over 300,000. The Las Vegas Valley, located in southern Nevada,
is bounded by the Sheep Range and Las Vegas Range to the north, Spring
Mountain to the west, Frenchman and Sunrise Mountains to the east, and the
McCullough Range to the south (Figure 4). The floor of the valley slopes
gently from west to east, from about 980 meters (m) mean sea level (MSL) in
the west to about 550 m MSL in the east-southeast. East of the Las Vegas
Wash, the terrain gently rises again. The ~onfiguration of these surrounding
mountains, whose elevations average about a kilometer (km) above the valley
floor, imparts a bowl shape to the valley and. provides relief passes to the
northwest, southwest, and southeast. .
A preliminarY task in the demonstration consisted of an assessment of the
air quality model employed, the SA! Atmospheric Pollution Simulation Model, to
reproduce pollutant concentration distributions under a variety of meteoro-
logical conditions. For this purpose, the Las Vegas Valley was divided into
1 km x 1 km grids over a 48 km x 70 km modeling region. With aerometric data
. gathered in a field measurement program, the SAI model was exercised for six
days during the winter of 1975-1976. The predicted CO concentrations were
compared with field measurements to assess the validity of the model. Good
agreement was shown between predicted and measured values for nearly all the
cases examined. The model predicted diurnal trends well, but sometimes
failed to predict the absolute magnitudes of peak concentrations, especially
in the downtown and Las Vegas Wash areas. The highest value concentrations
resulted from the afternoon traffic peak (downtown). High concentrations also
occurred at the lowest topographic point in the valley (Las Vegas Wash). The
occurrence of microscale phenomena (i.e., those of a smaller scale than the
model can resolve) or uncertainties in input data usually accounted for dis-
crepancies. Linear correlation coefficients between hourly values of measured
and predicted CO concentrations generally ranged between 0.7 and 0.9. These
comparisons are as good as any that have been reported over this time interval
for models finely tuned to a specific region. A detailed presentation of the
field program and the model validation is found in McElroy et al. (1978).
12
-------
. . . . .
SHEEP RAIIGE . /' ~. . \ -
S'~'~<~~~~~t~,~(
\. :' ,'fIf 0,.-:,1
~ . ~~ ~
:t).
. .
Figure 4. Map of the Las Vegas Valley.
13
-------
. With a validated air quality simulation model in hand, the stage is set
for the design of the monitoring network. As indicated earlier, the first
step is the identification and ranking of potential monitoring sites and the
second step is the determination of the optimum configuration for the
monitoring network. These two steps are addressed in the following
subsections.
THE FIRST STEP-~IDENTIFYING AND RANKING MONITORING SITES
For this purpose, six meteorological scenarios were selected on the basis
of historical weather data for the Las Vegas area (McEl roy et ale 1978). The
SAI Air Pollution Simulation Model was exercised for each of the six meteoro-
logical scenarios. In this case, carbon monoxide concentration fields span-
ning 13 hours, at hourly intervals between 7:00 a.m. and 8:00 p.m., LST were
determined. The corresponding FOM fields were subsequently compute~ uSing the
following expression:
6 (Concentration at location) (Probability Of)
FOM(x,y) = l (x,y) under meteorological. meteorological.
k=1 pattern k pattern k
An algorithm developed for identifying potential monitoring sites, as outlined
in section 3, was used for searching for the highest values in the FOM field.
This algorithm eliminates locations having high FOMs that are adjacent to
locations having higher FOMs without an intervening trough. Such locations
are considered to be adequately represented by the adjacent location having
the higher FOM value. The isolated peaks of the FOM thus selected are chosen
as potential candidates for monitoring stations. Because the NAAQS for CO
have been specified as one-hour and eight-hour averages, computations for the
FOM were carried out for each hour from 7:00 a.m. through 8:00 p.m. and for
the two eight-hour periods near the morning and evening traffic peaks. These
periods represent the highest CO concentrations either observed or predicted
in the Las Vegas area because of the dominant contribution of automotive
emissions. The resultant FOMs for the following time periods are shown in
Figures 5, 6, and 7.
(11)
. One-hour period near the morning traffic peak (7:00 a.m.).
. One-hour period near the evening traffic peak (6:00 p.m.).
.' Eight-hour period near the evening traffic peak (12:00 p.m. to
8:00 p.m.).
In these figures, fsopleths for the FOM (in parts per million) overlay the
selected monitoring locations. The locations are ranked alphabetically accord-
ing to the magnitude of the FOM. Consequently, a total of more than 40 moni-
toring stations were identified and ranked. A perusal of these selected moni-
toring stations shows a pattern of proximity to major Las Vegas roadways, a
fact that is not too. surprising since traffic is the major emission source for
. CO. It is, however, interesting to note that all three sets of calculations
14
-------
...
fit
~
MIRTH
10
20
30
.
........
.
.........
...........
.
.
-
en
=
w
. . . .. .. .. .. .. .. .. .. ... .. .. .. .. .. .. .. .. .. .. .. .. . ... .. .. .
. . .
[[[ ...
.
.
30
40
a
SlU1M
Figure 5.
-------
~
en
w
3
HIRTH
10
20
30
.... .
.. . . . . .. .
....
en
a:
IW
. .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
SIUTH
30
40
a
Figure 6.
Ranked monitoring locations for CO based on one-hour averaging
period near evening traffic peak: Las Vegas Valley.
16'
-------
MIRTH
t-
III
....
:II
20
30
.. .. .. .... .. .. .. .. .. .
.. .. .. .. .. .. .. e... ..
. ............~.....,..."..Ii... ...,
"
t-
In
a:
Il.I
.. .. .. .. .. .. .. .. .. .. .. .. .. . .. .. .. .. .. .. .. . .. .. .. .. .. .. .. .. ..
.. .. .. .. .. .. .. .. .. .. .. .. .. .... .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ..
.. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ..
.. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . .. .. .. . .. .. .. .. .. .. .. .. .. .. ".8 .. .. .. .. .. .. .. .
30
SIUTH
Figure 1.
Ranked monitoring locations for CO based on eight-hour averaging
period near evening traffic peak: Las Vegas Valley.
'11
-------
identify a location in the vicinity of an existing station on East Charleston
(see Figure 4), a location that usually reports the highest CO concentrations
in Las Vegas. The one-hour morning maximum (Figure 5) identified locations at
Henderson and downtown Las Vegas as the second- and third-ranked locations,
whereas the one-hour evening maximums (Figure 6) identified the same two
locations, but in reverse order. This finding is probably due to the fact
that Las Vegas morning traffic is job-related, whereas the evening traffic is
primarily caused by visitors in the downtown area. This result seems to
indicate that the FOM methodology developed under the present project can
indeed detect subtle diurnal variations in an emissions pattern that may be
unique to the Las Vegas area. In the next subsection, an optimum monitoring
network for the metropolitan Las Vegas is established from among these
stations by statistically determining the minimum number of monitoring
stations required.
THE SECOND STEP--DETERMINING THE MINIMUM NUMBER OF STATIONS REQUIRED
As described in section 4, the determination of the optimum network con-
figuration -- the minimum number of monitoring stations required -- is made by
using an SOl. for each of the ranked monitoring locations. The SOls are, in
turn, determined by the spatial correlation coefficients and the associated
cutoff values for a prescribed confidence level.
Concentration fields for the six meteorological scenarios as determined
by the model simulation provide the data base for evaluating the spatial cor-
relation coefficients for each of the monitoring locations identified. Prior
to the calculation of the spatial correlation coefficients, a smoothing of
these concentration fields was accomplished to remove small-scale fluctua-
tions. Similar operations are commonly used in turbulence research and
numerical weather prediction (Shuman 1957, and Haltiner 1971). Smoothing in
this study was accomplished by the following operation: Assuming that
Cg is the gth time-smoothed concentration at grid point (x,y), then the
xy
(g + :1)-th time-smoothed concentration field is obtained by
g+1
Cxy , =
Cg + bC9
9 xy
1 + b
(12)
where
~ = t (ci-l,y + Ci+l,y + ci,y-l + cL+1) ,
and b isa weighting factor to be empirically determined. A value of 2 was
chosen for b, which is comparable to the well-known two-dimensional Shuman
filter (Nelson and Weible 1980).
18
-------
In the present application, three smoothing operations were sufficient to
facilitate further analysis without altering the essential features of the
original concentration fields. The data base thus consists 13 hourly smoothed
concentration fields for each of the 6 meteorological scenarios for a total of
78 samples. Values of the cutoff sample correlation coefficient to ensure
specific minimum values of the population coefficient and variance explained
as determined from a numerical version of Figure 2 for a sample size of 78 and
95 percent confidence level are shown in Table 1.
TABLE 1. CUT-OFF SAMPLE CORRELATION COEFFICIENT TO ENSURE MINIMUM
VALUES FOR POPULATION CORRELATION COEFFICIENT AND VARIANCE
EXPLAINED FOR 95% CONFIDENCE WITH 78 SAMPLES
Sample Correlation
Cut-Off Value
rc
Population Correlation
Minimum Value
Pc
Variation Explained
Mi nimum Value.
pc2
. .- -.. u_. -. .,. ".
0.4
0.5
0.6
0.7
0.8
0.9
0.18
0.1)
0.44
0.56
0.70
0.85
0.03
0.1
0.2
0.3
0.5
0.7
To demonstrate the utility of the siting methodology, a total of 19
monitoring locations was selected from the highest ranking monitoring
locations determined. The 13 highest ranking locations, determined by the
eight-hour FOM for the evening traffic peak, were augmented by three locations
each from the highest ranking one-hour FOM for the morning and evening traffic
peaks, which were not adjacent to the locations already selected. The 19
stat10ns were selected to cover maximum or peak concentrations. In addition,
as a further demonstration of the methodology for secondary monitoring
objectives discussed earlier, three stations located in the northern, western,
and southeastern outskirts of the city were arbitrarily added to measure
either the background or baseline air quality in the Las Vegas area. The
.1 ocati ons and character1 zati ons of these 22 stati ons are 1 isted in Table 2.
A specific selection of sample correlation-coefficient isopleths for each
of the 22 stations is shown in Appendix B. As described in section 3, the 501
is dictated by the OJtoff correlation coefficient. Assuming that the
criterion for an optimum network design is its capability for catching at
least 50 percent of the concentration variations 95 percent of the time, then
Equation (9) yields a minimum value for Pc of 0.7.
-.- .. -
, - --
~. ~'. .- ~ " ..
- -'.-.-._...,...~.....- -.-'---".--- .-.-......--. '- .
.--.-- ~,--- - --
. ..
. . . 19
""'- --- --.- --- -..- ..-........-'.
. -
. .
-------
TABLE 2. IDENTIFICATION OF POTENTIAL MONITORING LOCATIONS IN THE
LAS VEGAS VALL EY
Station
Identification x-Coordinate y-Coordi nate Comments
A 28 40
B 25 37
C 27 37
0 39 26
E 25 34 Locations determined
F 26 32 by eight-hour figure-of
G 34 48 merit near the evening
H 21 45 traffi c peak.
I - 37 30
J 21 39
K 20 41
L 23 22
M 39 54
N 41 32} Locations determined by
O. 18 40 one-hour figure-of-merit at
P 19 46 the evening traffic peak'
Q 23 48 } Locations determined by
R 46 20 one-hour figure-of-merit at
S 22 15 the morning traffic peak
T 4 37 J Arbitrarily chosen rural
U 28 60 locations
V 43 15
With a sample size of 78, Table 1 shows that this value of Pc
corresponds to cutoff sample correlation coefficient of 0.8. Therefore, this
value was used to determine the spheres of influence for each of the 22
stations listed in Table 2. Stations were deleted from the list if the
individual areal coverage, after eliminating overlapping regions already
covered by higher ranking stations, was less than 10 percent of the coverage
of the highest ranked station. As a result, a. total of 10 air quality
monitoring stations was selected, among which 3 are rural background stations.
The locations of these 10 stations and their jOint areal coverage (shaded
areas) are shown in Figure 8. For each of these stations, the following
statistics were compiled to measure the effectiveness of individual stations
as well as that of the. overall network: .'
20
-------
20.
~ 30.
" qO
O. ....... .,... . .
...... ......... ....:. .......... ..! ! !
~' 81." , ~. 30.
; ,~! ...................;.. r#
; ~; ! 8 20.
. . . . ........] 0
: 21. i ,. .
. . 30.
Figure 8 A qo. o.
bas~d n optimum air .
on a cutoff quality monit
sample spatial oring network f
correlation coef~~c~:stVegas Valley
21 n of 0.8. .
On.
I
\
I
1
I
I
I
I
I
I
I
I
I
-------
. The fraction of area covered by the individual station based on the
total area considered (48 km x 70 km = 3~360 km2)~ as determined by
SOl.
e. The cumulative areal coverage~ expressed as the fraction of the
total area considered~ beginning with the highest-ranked station.
These summary statistics are listed in Table 3.
TABLE 3. SUMMARY STATISTICS FOR SELECTED MONITORING STATIONS IN
LAS VEGAS VALLEY BASED ON A CUTOFF SAMPLE SPATIAL CORRELATION
COEFFICIENT OF 0.8
Individual Station Cumul ative
Fractional Areal Fractional
Station x-Coordi nate y-Coordi nat e Coverage Areal Coverage
A 28 40 0.0723 0.0723
B 25 37 0.0253 0.0976
D 39 26 0.0455 0.1432
H 21 45 0.0155 0.1586
I 37 30 0.0122 0.1708
L 23 22 0.0211 0.1920
R 46 20 0.0104 0.2024
T 4 37 0.0182 0.2205
U 28 60 0.0336 . 0.2542
V 43 15 0.0119 0.2661
As a sensitivity test of the siting methodology~ identical computations
were made using a cutoff sample correlation coefficient of 0.5. This network
configuration and its correspondi ng joint areal coverage (shaded areas) are
presented in Figure 9. A total of 7air quality monitoring stations were
selected by the siting methodology. Among these stations~ 2 are rural .
background stations. Summary statistics for the 7-station network are given
f n Tab 1 e 4. It is interest i ng to note that cumu 1 ati ve a rea 1 coverage
increases from 26.6 percent of the total region for the 10-station network to
62.3 percent for the 7-station network.. However, it should be noted as 1s
shown in Table 1, that the 7-station network can only detect a minimum of 10
percent of the concentration variations, whereas the 10-station network can
detect a minimum of 50 percent of the concentration variations for the area
within the combined spheres of. influence (~ee Figures 8 and 9). .
22
-------
I
1
,
I
I
I
I
I
I
I
I
I
I
I
,
-------
TABLE 4. SUMMARY STATISTICS FOR SELECTED MONITORING STATIONS IN
LAS VEGAS VALLEY BASED ON A CUTOFF SAMPLE SPATIAL CORRELATION
. COEFFICIENT of 0.5
Indivfdual Station Cumulative
Fractional Areal Fracti ona 1
Station x-Coordinate y-Coordinate Coverage' Areal Coverag~
A 28 40 0.2604 0.2604
H 21 45 0.0664 0.3268
I 37 30 0.0327 0.3595
o 18 40 0.0429 0.4024
R 46 20 0.0354 0.4378
T 4 37 0.1232 0.5610
U 28 60 0.0622 0.6232
Further observations can be made concerning these monitoring networks as
determined by the siting methodology:
. As shown in Figures 8 and 9, all urban stations chosen are located,
as expected, along the main transportation corridors that constitute
the bulk of carbon monoxide emissions sources (about 80 percent) in the
Las Vegas area. Such stations corresond to the "street canyon" and
"traffic corridor" stations referred to by Ludwig, Berg, and Hoffman
(1976) and the "A "-type stations referred to by Ott (1975).
.' The joint areal coverage of these stations (as shown in Figures 8
and 9) tends to shift toward the east, southeast, and south of the
major emissions sources, apparently reflecting prevalent local
wind directions in Las Vegas for the meteorological scenarios used.
.. The western and northern rural background stations designated as
stations T and U, were selected in both networks presumably because air
quality in the neighborhood of these sites is not significantly
affected by the major emissions sources in the metropolitan Las Vegas
area.
24
-------
SECTION 6
CONCLUDING REMARKS
An objective methodology was presented for determining an optimum air
quality monitoring network for an urban area. This methodology, in totality,
yields both the number and disposition of the monitoring stations. Although
the methodology is applicable to all pollutants, it was applied to the Las
Vegas Valley using a relatively inert species, carbon monoxide, as an example.
The exercise of the methodology for Las, Vegas suggests that the design of an
air quality monitoring network can be accomplished in a technically rigorous
manner.
25
-------
~--- --
REFERENCES
David, F. N. (1938), Tables of the Ordinates and Probabilit tnte ral of the
Distribution of the Correlation Coefficient in Small Samples the Bio-
metrika Office, Cambridge University Press, Cambridge, England).
Ezekiel, M. (1941), Methods of Correlation Analysis (John Wiley and Sons,
Inc., London, Eng~.
Haltiner, G. J. (1971), Numerical Weather Prediction (John Wiley and Sons,
Inc., New York, New Y~.
Liu, M. K. and G. E. Moore (1980), "Development and Application of a
Methodology For Air Quality Monitoring Network Design," Proc. of Second
Joint Conference on Applications of Air Pollution Meteorology and the
Second Conference on Industrial Meteorology, March 1980, New Or,leans,
Louisiana, pp. 727-733.
Liu, M. K. et al. (1977), "Development of a Methodology for the Design of a
Carbon Monoxide Monitoring Network," EPA-600/4-77-019, U.S. Environmental.
Protection Agency, Las Vegas, Nevada.
Ludwig, F. L. and J. H. S. Kealoha (1975), "Selecting Sites for Carbon
Monoxide Monitoring,"EPA 450/3-75-077. U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina.
Ludwig, F. L., N. J. Berg, and A. J. Hoffman (1976), liThe Selection of Sites
for Air Pollutant Monitoring," Paper presented at the 69th Annual Meeting
of the Air Pollute Control Assoc., 27 June - 1 July 1976, Portland,
Oregon.
McElroy, J. L. et al. (1978), "Carbon Monoxide Monitorlng Network Design
Methodology," EPA-600/4-78-053, U.S. Environmental Protection Agency, Las
Vegas, Nevada.
Nelson, S. P. and M. L. Weibel (1980), "Three-Dimensional Shuman Filter," i.
Appl. Meteorol., Vol. 19, pp. 464-469.
Ott, w. R. (1975), "Development of Criteria for Siting Monitoring Stations,"
Paper presented at 68th Annual Meeting of the Air Pollute Control Assoc.,
Boston, Massachusetts.
Ott, W. R. and G. C. Thorn (1976), itA Critical Review of Air Pollution Index
Systems in the United States and Canada, II i. Ai r Poll ute Control Assoc.,
Vol. 26, pp. 460-470.
26 '
-------
Shuman, F. G. (1957), "Numerical Methods in Weather Prediction: II.
Smoothing the Filtering, II 1iQn. Wea. Review, Vol. 85, pp. 357-361.
u.s. Environmental Protection Agency (1971),"Guidelines:
Surveillance Network,1I Public Health Service, AP-98.
Air Quality
27
-------
APPENDIX A
A HEURISTIC APPROACH FOR OPTIMAL MONITORING NETWORK DESIGN
,
In this appendix, a summary is presented of a heuristic approach used in a
prelinrlnary study by Liu and Moore (1980) for determining the optimum moni-
toring network and its application to metropolitan Las Vegas.
Similar to the statistical approach described in the text, the heuristic
approach also consists of two steps. Furthermore, the first step in both
approaches is identical. Maxima or peak values of figure-of-merit (FOM) are
used to identify and rank potential monitoring stations. In the second step,
the two approaches differ principally in the definition of the sphere. of
influence (501).
On the basis of a heuristic approach, the 501 was defined by Liu and Moore
(1980) as the monitoring site itself plus those grid locations along a
straight line in any direction from the site showing a continually decreasing
FOM. An al gorithm was developed that employs both:a forward mode and a back-
ward mode to screen the most suitable candidates for the monitoring network
from the N monitoring locations identifled in the first step. The forward
mode starts with the highest FOM grid and defines the 501 for each of the N
monitoring. stations in descending rank of FOM.
The backward mode is used to eliminate lower-ranked stations that do not
contribute to the total spatial or FOM coverage of the monitoring network. In
the backward mode, the SOls of the lower-ranked stations are overlayed by the
SOls of the higher-ran~ed stations. At the end of this operation, the
stations that do not appear in the listing are those that are completely
contained in the SOls of the higher-ranked stations. These stations,
considered as redundancies, can be thus deleted from the network without
affecting the adequacy of coverage. Consequently, an optimum monitoring
network is determi ned. .
This heuristic approach was applied to metropolitan Las Vegas using the
data base described earlier. Summary statistics generated during the forward
and backward modes of the site selecting algorithm are shown in Tables A-I,
A-2, and A-3. The station locations are identified in Figures 5, 6, and 7,
respectively, in the text of this report. It can be seen from the tables
that, though the inclusion of all five stations is necessary on the basis of
. the one-hour morni ng maxima, only three stations overl appi ng the five stations
for the 'morning maxima are required on the basis of the one-hour evening and
the eight-hour evening maxima. These five stations, denoted as A to E in
Figure Sof the text, then represent the optimal network for monitoring carbon
monoxide in the Las Vegas area, as determined using the heuristic approach.
28
-------
A comparison of the 5-station network developed using the heuristic
approach with the 10-station or 7-station networks developed using the
statistical approach shows that both approaches place the majority of the
monitoring stations, with a varying degree of emphasis, in downtown Las Vegas
and Henderson in the southeast (Figure 4). The principal difference between
the two approaches lies in the areal coverage statistics. The areal coverage
1n the heuristic approach may be grossly inflated because the SOl based on a
line of sight can be extended indefinitely. The areal coverage determined by
the statistical approach is apparently more realistic. Thus, the statistical
approach is favored over the heuristic approach in the determination of the
optimum monitoring network.
TABLE A-I. CUMULATIVE COVERAGE STATISTICS FOR THE MONITORING NETWORK
F~R CO BASED ON ONE-HOUR AVERAGING PERIOD NEAR MORNING TRAFFIC PEAK--
. LAS VEGAS VALLEY (7 a.m.)
.., . - Peak . '" - Cumulative Cumulative Percent
Value Percent of of Fi gure-of-
Station (ppm) Areal Coverage Merit Coverage
A 13.763 70.710 67.710
B 10.200 87.0 80.5
C 9.492 88.5 83.2
D 8.249 88.6 83.5
E 7.390 . 98.2 93.7
TABLE A-2. CUMULATIVE COVERAGE STATISTICS FOR THE MONITORING NETWORK
FOR CO BASED ON ONE-HOUR AVERAGING PERIOD NEAR EVENING TRAFFIC PEAK--
LAS VEGAS VALLEY (6 p.m.)
Peak Cumulative Cumulative Percent
Value Percent of of Figure-of
Station (ppm) Areal Coverage Merit Coverage
A 15.472 89.610 91.410
B 6.233 96.8 96.6
C 4.870 99.8 99.9
D 2.796 99.8 99.9
. E 2.445 99.9 99.9
29
-------
TABLE A-3. CUMULATIVE COVERAGE STATISTICS FOR THE MONITORING NETWORK FOR CO
BASED ON EIGHT-HOUR AVERAGING PERIOD NEAR EVENING TRAFFIC PEAK--LAS VEGAS
VALLEY (12 p.m. - 8 p.m.).
Peak Cumul ative Cumulative Percent
Value Percent of of Figure-of-
Station (ppm) Areal Coverage Meri t Coverage
A 7.461 90.0% 91.5%
B 5.209 98.4 98.2
C 4.593 98.9 98.7
D 3.410 100.0 100.0
E 2.672 100.0 100.0
~ ..- _. ._. ....
. h- ..- . - .
.. --.. "
- .... -.. '.' -.-. .
-"'h ..- -, ~~,- ,'- '-"'-' "_'8 ,-,~""',"""'-'-"-'" _..~.,...........-_._-. ..-.,.....
. - 30
-------
APPENDIX B
SAMPLE SPATIAL CORRELATION-COEFFICIENT ISOPLETHS IN LAS VEGAS VALLEY
31
-------
to.
20.
30.
70.
50.
so.
40.
30.
20.
:
I
f
i
i
i
20.
30.
40.
o.
(a)
Sfte A
Figure 8-1.
Sample spatial correlation-coefficient
in Las Vegas Valley.
isopleths
32
-------
70.0. 10. 20. 30. 40. 70.
.
f .
60.
~o.
so.
10.
!
!
I
i
I
i
I
!
!
i
I
I
f
i .
i
i
!
I
20.
30.
(b)
Site 8
Figure B-1.
(continued)
33
-------
so.
20.
'0.
50.
30.
20.
10.
t
j
i
i
i
i
20.
30.
(c)
S1t8 C
F1 gure 8..1.
(continued)
34
-------
.
-.-----
I
150.
i
I
i- ..
!
i
!
SQ.
40.
30..
20.
10.
lD.
30.
I
20.
.
I
i
20.
(d)
3D.
Site D
Figure 8-1.
(continued)
35
-------
so.
50.
40.
30.
20.
10.
10.
20.
30.
.t ~ ...- ---.-
20.
Ce)
30.
SIte E
F1 gure 8..1.
(continued)
36
-------
. .
. .
. .
. .
: : :
, ",;..-.........-.-....)1 ,
-------
ISO.
so.
40.
30.
20.
10.
0'0.
LO.
20.
30.
i
I.
i
I'
i
I
20.
30.
(0)
Sfta Q
Figure 8-1.
(continued)
38
-------
30.
20.
30.
40.
70.
50.
zo.
10.
-11
i ---
10.
20.
(It)
30.
Stta H
F1 gure 8-1.
(continued)
39
-------
60.
so.
40.
30.
20.
JO.
20.
30.
40.
70.
.
.
.-,, .
'""..._."
..
I
i
i
.
I
i
,.
20.
o.
30.
40.
(1)
SUa 1
Fi gure 8-1.
(continued)
40
-------
80.
20.
30.
so.
40.
!D.
20.
JO.
20.
30.
(j)
Sfte J
Figure 8-1.
(continued)
41
-------
20.
30.
.
60.
1&4..
50.
40.
'"1""___-"
so.
I
°13.
10.
20.
(t)
Site It
Figure B-1.
(continued)
42
-------
30.
20.
30.
40,
70.
10.
so.
20.
10.
20.
30.
(1)
Site L
Figure 8-1.
(continued)
43
-------
70!1'
10.
zo.
30.
60.
so.
40.
30.
20.
i
. f
~
i
I
i
20.
30.
(a)
Site "
Figure B-1.
(conti nued)
44
-------
7Q!J. 10. 20. 30.
80.
50. -
40.
so.
20.
10.
i
I
. :
i
!
I
Figure B-1.
.
!
i
!- .
I
i
:
i
i
!
i
i
:
.1
i
i
1
. .
i .
i
I
20.
30.
(n)
Stte N
(continued)
45
-------
70!1.
10.
20.
30.
40.
70.
50.
60.
40.
30.
20.
10.
(0)
Sfta 0
Figure 8-1.
(continued)
46
-------
60.
10.
20.
30.
40.
70.
so.
.. .. .. ..
. . . ..
: : : :
. . . .
. . . .
. . . .
: : : :
. . .. .
. . . .
. . .. .
. . . .
. . . .
.. . . .
. . .. ..
.. .. .. ..
.. .. .. ..
.. .. . ..
.. .. .. ..
--~. :1 ....- ~.._-t-_.---.-I.--.--l.__.__...
o. i'.j... [0 j j 1
i ~ I I' . . I
: : ~ :
I 81 ' ~; I
! ..! : ~ ;
: ~~: : 0" :
_._-l-..~-.T .-...;; r.-..-.---.r............
- .I~.~...!-.__. L..-..
I Lf ~: !
i i i i
---1- .._-~-_..._.. t- .................._1_._.....-....
: '-r I I
~tJ\! i . !
. ~. . ---~kfr: -.... --1.-..--..--
!! . !
i «11; .-1/1 . 1
i . .. : :
f i I I
: : : i
! !..! !.
0'0.
10.
20.
(p).
30.
40.
30.
20.
10.
Site P
F1 gure B-1.
(continued)
47
-------
70!3'
10.
20.
3D.
40.
70.
60.
so.
40.
3D.
20.
10.
20.
30.
40.
o.
(q)
Site Q
Figure 8-1.
(continued)
48
-------
70~'
10.
zo.
30.
20.
SQ.
.. .. .. .
. . . .
. . . .
. . . ..
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . .. ..
.. .. . .
.. .. .. ..
.. .. .. .
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. .
.. .. .. ..
.. .. .. ..
.. . .. .
----...-....---..........--..-........--..................-..-...-....-..-............................
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. . .. .
.. .. . ..
.. .. .. ..
. .. .. ..
.. .. .. .
.. .. .. .
.. .. .. ..
. .. .. ..
.. .. .. .
.. .. .. .
.. .. .. .
.. . .. ..
.. . .. ..
.. .. .. .
.. .. .. .
.. .. . .
.. .. .. ..
.. .. .. ..
.. .. . ..
.. . .. .
.. .. .. ..
.. .. .. ..
. . . .. .
.--.--.....--.-.............---..-..--....-....--......--....-.....-......-.........................
.. .. . .
.. . . ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. . .. .
.. .. . .
.. .. .. .
.. .. .. .
. .. .. ..
.. . .. .
.. .. .. ..
.. .. .. .
.. . .. .
.. . .. .
. .. .. .
.. .. . .
.. . .. ..
.. .. .. .
.. . .. ..
.. .. .. .
.. . . .
.. .. .. ..
. . .. ..
.. .. .. ..
. . .. .
. .. .. .
.. . . .
. . .. .
.----...-....-.-.-............................-............-.....-..-.......--........--............-..............
. . . .
.. . .. .
. . .. .
.. . . ..
.. .. . ..
.. . . ..
.. . . .
. .. . ..
. . .. ..
. . . .
. .. .. .
. .. . .
. . . .
. . . .
. .. . .
. . .. .
.. .. . ..
. . .. .
. . . ..
. .. . .
.. .. . .
.. .. . .
. .. . .
.. . . .
.. .. .. ..
.. . .. .
.. . .. ..
. .. . ..
-----r------ t----..---r------r--.---....'..
I f.-I..J
so.
40.
30.
20.
(r)
30.
Site R
Figure B-1.
(continued)
49
-------
70~.
LO.
20.
30.
40.
70.
so.
so.
40.
30.
20.
10.
10.
20.
(5)
30.
Site S
Figure 8-1.
(continued) .
50
-------
20.
20.
30.
40.
70.
60.
. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
('\: : : :
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
: ! :. :
..-zt-l----..t-._-t............
..- : : :
r I I I
'-r---ri--m...
+-4--.--1----..
i ! !
! I i
-i '''-'--1-1---''
i : :
! i :
so.
40.
30.
-
u...
10.
20.
(t)
30.
40.
o.
Site T
Figure 8-1.
(continued)
51
-------
70~.
so.
10.
20.
30.
so.
40.
so.
20.
10.
(u)
Sfte U
F1 gure 8-1.
(conti nued)
52
-------
. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
.. .. .. ..
. .. . ..
- L_.J____I_-1-___.
i ~;;I- .
i
7OP'
so.
so.
40.
3D.
20.
10.
LO.
20.
30.
.
i
i
I
i
20.
(y)
30.
Site V
Figure 8-1.
(continued)
53
40.
70.
-------
. -...J
'. . --..- J
J
-1
J
.]
.j
,
,
i
't TECHNICAL REPORT DATA
(Plet1Se rettd /TUtrUCtiOTU on the revene before completing)
',. REPORT NO. 12. 3. RECIPIENT'S ACCeSSION NO.
EPA-600/4-8l-002.
, 4. TITLE AND SUBTITLE 5. REPORT DATE
'METHODOLOGY FOR THE DESIGN OF AN OPTIMUM AIR QUALITY - 'v 1 qRl
MONITORING NETWORK 6. PERFORMING ORGANIZATION CODE
7. AUTHORIS) 8. PERFORMING ORGANIZATION REPORT NO.
M. K. Liu, J. Avrin 158-EF79-146R4
9. PERFORMING ORGANIZATION NAME ANO ADORESS 10. PROGRAM ELEMENT NO.
Systems Applications, Incorporated
950 Northgate Drive 1'. CONTRACi'TGRANT NO.
San Rafael, California 94903 68-03-2446
12. SPONSORING AGENCY NAME AND ADDRESS 13. TYPE OF REPORT AND PERIOD COVERED
U.S. Environmental Protection Agency--Las Vegas, NV
Office of Research and Development 14. SPONSORING AGENCY CODE
Environmental Monitoring Systems Laborato~
Las Vegas, Nevada 89114 EPA/600/07
15. SUPPLEMENTARY NOTES
J. L. ~cE 1 roy, EMSL-LV, Project Officer
16. ABSTRACT
A two-step objective method is presented for determining the optimum number and
disposition of ambient air quality stations in a monitoring network. The. method uses a
data base consisting of a comprehensive set of simulated or measured air quality
patterns representative of the region of interest. In the first step, the most -
desirable monitoring locations are identified and ranked. The minimum number of
required locations is determined in the second step through eliminating redundancies
among the locations identified in the first step with regard to spatial coverage over
the region of interest. As a demonstration, the method is applied to the Las Vegas
Valley of Nevada for the pollutant species carbon monoxide.
17. -
KEY WORDS AND DOCUMENT ANAL YSIS
a. DESCRIPTORS b.IDENTIFIERS/OPEN ENDEO TERMS C. COSA TI Field/Group
Meteorology Design Methodology S5C
Air Quality Monitoring net~rork 68A
Air pollution 91A
1B. DISTRIBUTION STATEMENT 19. SECURITY CLASS (Thir Report) 21. NgOOF PAGES
UNCLASSIFIED
RELEASE TO PUBLIC 20. SECURITY CLASS (Thir page) 122. PRICE
UNCLASSIFIED
EPA Form 2220-1 (Rev. 4-77)
PREVIOUS EDITION IS OBSOL.e:TE
. .s..
------- |