EPA-600/4-77-019
March 1977
Environmental Monitoring Series
           DEVELOPMENT OF A METHODOLOGY FOR
                      DESIGNING CARBON MONOXIDE
                              MONITORING NETWORKS
                              Environmental Monitoring and Support Laboratory
                                     Office of Research and Development
                                    U.S. Environmental Protection Agency
                                          Las Vegas, Nevada 89114

-------
                 RESEARCH  REPORTING SERIES

 Research reports of the Office of Research and Development, U.S. Environmental
 Protection Agency, have been grouped into nine series. These nine broad cate-
 gories were established to facilitate further development and application of en-
 vironmental technology. Elimination of  traditional grouping was  consciously
 planned to foster technology transfer and a maximum interface in related fields.
 The nine series are:

       1.  Environmental Health Effects Research
      2.  Environmental Protection Technology
      3.  Ecological Research
      4.  Environmental Monitoring
      5   Socioeconomic Environmental Studies
      6.  Scientific and Technical Assessment Reports (STAR)
      7   Interagency Energy-Environment Research and Development
      8.  "Special" Reports
      9.  Miscellaneous Reports

 This report has been assigned to the ENVIRONMENTAL  MONITORING  series.
 This series describes research conducted to develop new or improved methods
 and instrumentation for the identification and  quantification of environmental
 pollutants at the lowest conceivably significant concentrations. It also includes
 studies to determine the ambient concentrations of pollutants in the environment
 and/or the variance of pollutants as a function of time or meteorological factors.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                              EPA-600/4-77-019
                                              March 1977
         DEVELOPMENT OF A METHODOLOGY FOR
   DESIGNING CARBON MONOXIDE MONITORING NETWORKS
    Mei-Kao Liu, James Meyer, Richard Pollack,
         Phillip M.  Roth,  John H.  Seinfeld
        Systems Applications, Incorporated
                950 Northgate Drive
           San Rafael, California   94903

                        and

Joseph V. Behar, Leslie M. Dunn, James L.  McElroy,
 Pong N. Lem, Ann M.  Pitchford, Nancy T.  Fisher
 Environmental Monitoring  and Support Laboratory
             Las Vegas, Nevada  89114
              Contract No.  63-03-2399
                  Project Officer

                 Edward A. Schuck
  Environmental Monitoring and Support Laboratory
              Las Vegas, Nevada  89114
  ENVIRONMENTAL MONITORING AND SUPPORT 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 and
Support Laboratory-Las Vegas, U.S. Environmental  Protection Agency, and
approved for publication.  Mention of trade names or commercial  products
does not constitute endorsement or recommendation for use.
                                      11

-------
                                   FOREWORD


     Protection of the environment requires effective regulatory actions which
are based on sound technical and scientific information.  This information
must include the quantitative description and linking of pollutant sources,
transport mechanisms, interactions, and resulting effects on man and his
environment.  Because of the complexities involved, assessment of specific
pollutants in the environment requires a total systems approach which trans-
cends the media of air, water, and land.  The Environmental Monitoring and
Support Laboratory-Las Vegas contributes to the formation and enhancement of
a sound integrated monitoring data base through multidisciplinary, multimedia
programs designed to:

          • develop and optimize systems and strategies for moni-
            toring pollutants and their impact on the environment

          • demonstrate new monitoring systems and technologies by
            applying them to fulfill special monitoring needs of
            the Agency's operating programs

     This report discusses the theoretical bases for a method of designing
air quality monitoring networks.  The method was developed for application to
reactive or nonreactive pollutant monitoring.  Specific design steps in the
report, however, are illustrated for a carbon monoxide monitoring network.
•Regional or local agencies may find this method useful in planning or adjust-
ing their air quality monitoring networks.  The Monitoring Systems Design and
Analysis Staff at the EMSL-LV may be contacted for further information on the
subject.
                                        George B.  Morgan
                                        Acting Director
                        Environmental Monitoring and Support Laboratory
                                           Las Vegas
                                      lit

-------
                                   PREFACE
     This document is concerned with the development of a  methodology  for  the
design of a monitoring network for carbon monoxide (CO).   In  actuality,  the
methodology is generally valid for any airborne pollutant.  CO  was  chosen
partially because it is a relatively inert pollutant for which  the  methodology
should be presentable in its basic, simplest form.   In  addition,  CO is a pol-
lutant very susceptible to analysis at both the mesoscale  and microscale
levels.  Finally, the first application of the methodology is for CO and will
appear as a separate report.

-------
                             CONTENTS

Foreword 	,	  i i i
Preface 	   iv
Abbreviations and Symbols 	   vi

     I.   Monitoring of Ambient Air Quality	    1
               Review of Previous Work  	    1
               Objectives of Air Quality Monitoring Programs ..    3
               Factors Pertinent to Monitoring Network Design .    7

    II.   Design of a Network for Air Quality Monitoring 	    9
               Components in the Design of an Air Quality
                Moni tori ng System 	    9
               Selection of Monitoring  Sites 	   11
               Approach Adopted in This Study 	   13

   III.   Analysis at the Mesoscale Level 	   20
               A Description of the Mesoscale Air Quality
                Simulation Model 	   20
               Approximation of Turbulent Transfer by Eddy
                Diffusivities 	   21

    IV.   Analysis at the Microscale Level 	   28
               Overview	   28
               Representativeness of Measurements on the
                Microscale 	   30
               Solution Methodology 	   31

     V.   Field Measurement Program 	   35
               Model ing Region 	   35
               Historical Information 	   37
               Sampling Rationale and Plan 	   37
               Quality Assurance 	   39

    VI.   Application for Selection of Carbon Monoxide
           Monitoring Sites 	   41
               General Methodology 	   41
               Model Input Requi rements 	   41

   VII.   Summary	   48

References	   49

-------
                       LIST OF ABBREVIATIONS AND SYMBOLS
 ABBREVIATIONS

 CAMP
 CAT-P
 CO
 EPA
 g/m3
 g/m/s
 HC
 km
 LDV
 LST
 m
 m/s
 m2/s
 mg/rrr
 NAAQS
 NASN
 NCC
 NEDS
 NOAA
 N02
 NWS
 PHS
 ppm
 SAI

 S°? 3
 yg/m
 UTM
 VMT
Continuous Air Monitoring Program
concentration area time-product
carbon monoxide
U.S. Environmental Protection Agency
grams per cubic meter
grams per meter per second
hydrocarbon
kilometer
light duty vehicles
local standard time
meters
meters per second
square meters per second
milligrams per cubic meter
National Ambient Air Quality Standards
National Air Sampling Network
National Climatic Center
National Emissions Data System
National Oceanographic and Atmospheric Administration
nitrogen dioxide
National Weather Service
Public Health Service
parts per million
Systems Applications, Incorporated
sulfur dioxide
micrograms per cubic meter
Universal Transverse Mercator
vehicle miles traveled
SYMBOLS

A
c(x,y,t)  -


 max

con '
D
pollutant concentration averaging area
pollutant concentration as a function of distance
and time
maximum allowable contribution of any one source to the
concentration
vehicle emission factor
a specified downwind horizontal distance
exposure class
                                      vi

-------
SYMBOLS, continued

dw        --   wind speed class
E         —   roadway types
F(j»k)    --   Figure of Merit
Gmax      ~~   desired maximum concentration gradient away from the roadway
r         —   random variable
H         —   hourly percentage of daily traffic
h*        --   average height of the roughness element within a grid element
k         --   von Karman constant
Kx,Ky,Kz  --   turbulent eddy diffusivities in the x,y and z directions
Kz.       --   vertical diffusivity at height z-j

L         --   Monin-Obukhov length
mon'      ~~   adjusted annual travel distance
Pol       --   temperature correction factor
Q(x,y,t)  --   pollutant emission rate
%Tw'     "~   cold start correction factor
p         --   sensor location at some downwind horizontal distance
s         --   downwind horizontal distance
S         --   stability function which is a digital  version of the Pasquill
               stability category
s*        --   exposure silhouette area
S*        --   lot area
S-j        --   downwind position of i-th particle
az        --   standard deviation of a Gaussian plume in the vertical
t*        —   pollutant concentration time-averaging interval
9         --   angle between wind and line source
u         —   wind speed in the x-direction
U         —   horizontal wind speed
u*        —   friction velocity
Ur        —   reference horizontal wind speed
u'c1      —   pollutant flux in the x-direction
v'c'      --   pollutant flux in the y-direction
w'c1      —   pollutant flux in the z-direction
v         --   wind speed in the y-direction
vgs'      --   speed correction factor
w         —   wind speed in the z-direction
2*        —   height of instrument inlet
Zj        --   height of stable layer capping mixing layer or height above
               which pollutant concentrations level off to background values
Z-;        --   height of particle i
z0        --   surface roughness
Zr        —   reference height
                                      vii

-------
                                  CHAPTER I

                      MONITORING OF AMBIENT AIR QUALITY


     The monitoring of ambient air quality is probably the single  most
important activity in the study and control of air pollution.  "Without
reliable measurements, a legitimate data base cannot be established  for
assessing either the degree of deterioration of ambient air or for enforcing
Federal and local regulations, nor can a firm and valid basis  be obtained  for
examining the cause-effect relationship between the emissions  from the  pollut-
ant  sources and the quality of ambient air.  It is thus not surprising that
prior to the early sixties, when recognition of air pollution  as a national
problem began to emerge, the primary concern of most air pollution studies
focused on the monitoring of ambient air quality.

     In 1953, the U.S. Public Health Service (PHS) set up the  National  Air
Sampling Network (MASM) to measure particulates in air.  In 1962,  an intensive
effort, the Continuous Air Monitoring Program (CAMP), was initiated  by  the PHS
to measure gaseous pollutants in typical urban areas.  Since then, numerous
governmental and private investigations related to air quality measurements
have been carried out.  With the passage of the Clean Air Act  in 1971,  ambient
air monitoring programs became an indispensable part of State  implementation
plans.  Thus, there exists a need to set up guidelines for planning  and siting
of established or prospective air monitoring stations.  As discussed later,
there have been a number of attempts in the past to provide such guidelines.
However, a rigorous guideline with a sound theoretical basis that  is opera-
tionally effective is lacking.  In light of the recent advances in the  theory
of systems design and in the current understanding of the distribution  of  the
various pollutants, an in-depth study that will lead to the development of an
objective methodology for siting monitoring stations is apparently feasible.
The present project is devoted to this goal.

REVIEW OF PREVIOUS WORK

     Guidelines or rules previously used by designers of monitoring  networks
were largely based on accumulated experience derived from practice.   A  brief
review of these efforts is presented below.

     In September 1963, a symposium on environmental measurements  was spon-
sored by the PHS at Cincinnati, Ohio.  Several papers presented there dealt
in part with the objectives of measuring systems and the design of such
systems.   However,  information resulting from that symposium was unsuitable
for specific applications.

-------
     Common practices in the design of air monitoring  networks before 1969
were summarized in a survey made by Yamada (1970).   Based on this study,
Yamada and Charlson (1969) found that differences  in the measurement param-
eters of the air sampling devices between networks of  stations and  between
stations within a single network can be a potential  source  of error.  They
thus pointed out the need for standardization of location and design of air
quality monitoring stations.  Charlson (1969) further  pointed out that  the
siting criteria depend on the particular objective of  interest and  that the
question of "representativeness" of measured air quality was an  important
consideration.

     At a U.S.  Environmental Protection Agency (EPA) workshop held  in January
1970 (U.S. EPA, 1970), four major objectives of a  regional  air quality  moni-
toring program were cited:

         To measure and document a region's progress toward meeting ambient
         air quality standards.

         To determine ambient air quality in nonurban  areas of the  region.

         To improve the reliability of dispersion  models.

         To provide air quality data during air pollution episodes.

     Four criteria were recommended to ensure that data to  be collected
satisfy all of the above stated objectives:

         Monitoring stations must be pollution-oriented.

         Monitoring stations must be population-oriented.

         Monitoring stations must be located so as to  provide areawide  repre-
         sentation of ambient air quality.

         Monitoring stations must be source-category and/or source-magnitude
         oriented.

     With these qualitative criteria in mind, the following guidelines  were
suggested for the distribution of monitoring stations:

         Heavily polluted or "dirty" areas—in most  cases 3 to 5 stations will
         suffice.

         Nonurban stations--2 to 4, depending upon size of  the area.

         Population-oriented stations —3 to 7.

         Source-oriented stations--3 to 5.

         Reference (center city) station—1.

         Any available stations not accounted for by above  should  be placed

                                      2

-------
         where the concentration gradation is greatest (as predicted by
         dispersion models).

The above recommendations were only speculative in nature and did not define
the criteria for locating monitoring stations.

     More recently, guidelines on siting monitoring stations were provided in
U.S. EPA(1971) and U.S. EPA(1975).  Quantitative rules for designing a mini-
mally adequate surveillance system were suggested in these documents.  For
example, as shown in Figure 1-1, the number of stations that are required to
implement such a system can be determined once the total  population and the
type of measuring systems are known.  Also, formulae were proposed that use
previous air quality records as an aid in determining the number of stations
needed.  Guidelines were provided for locating these stations.  For the most
part, the conclusions drawn were based on the analysis of air quality from
stations sited with little knowledge of the essential ingredients that must
be considered in planning their locations.

     Of more interest, however, are several theoretical studies carried out
(e.g., Morgan et al., 1970; Seinfeld, 1972; Darby et al., 1974) where the con-
cept of systems design was applied.  The design of an air quality monitoring
network was treated in these theoretical studies as an optimization problem
where a set of well-defined constraints related to monitoring objectives were
prescribed.

OBJECTIVES OF AIR QUALITY MONITORING PROGRAMS

     Reflection suggests that consideration of  the objectives of a monitoring
program must precede the actual design of the monitoring  network and the
siting of the stations.  Various objectives of past and present air quality
monitoring programs can be classified into three categories:

General Air Quality Monitoring Programs

     The goal of general monitoring programs is to provide the two-dimensional
air pollutant distribution near the ground in the region  of interest.  Judging
from the specific objectives stated below, it can be concluded that State or
local agencies are likely to be primarily concerned with  this type of moni-
toring program.

Compliance with Air Quality Standards

     As summarized in Table 1-1, the Clean Air Act Amendment of 1971 specified
two types of air quality standards:  the primary standards for the public
health, and the secondary standards for the public welfare (Barth, 1970).  In
addition to these Federal standards, many State and local agencies have set
up their own air quality standards.  Compliance, progress toward compliance,
or lack of compliance with these standards can be determined from the measured
air quality.  When excessive concentration levels are registered by the moni-
toring network, indicating the occurrence of an air pollution episode, the
collected data can be used in activating emergency control measures.

-------
      10.000,000
2,000,000 _


1,000,000 -


  500,000 I

         •M
  200,000 -


  100,000 -

   50,000 I


   20,000 -
         10,000
                    // /   /  /


                   /////
                   I/   / /
                  I'/   I  /  /

                  17   //  /
                  "  I/   /
                  [j  : i  /
                  N
                  II
                      /  /
                          	MECHANICAL-INTEGRATING
                                          (TSP,S02)
                          	AUTOMATIC-CONTINUOUS
                                  (S02,CO,HC,NOx,OXIDANTS)
                     r   i    i^  i    i   i   i    i    i
                 0    5   10  15  20   25   30   35  40  45  50
                           NUMBER  OF STATIONS
Figure 1-1.    Number of stations per air quality control region as
              a function of population (U.S. EPA, 1971).

-------
    TABLE 1-1.   NATIONAL AMBIENT AIR QUALITY STANDARDS
Pollutant

S00
Particulate
matter
CO
Photochemical
oxidants

HC
N02
 Primary
Standards

80yg/m3
(O.OSppm)
       o
365yg/m°
(0.14ppm)
75yg/m-"
Secondary
Standards
10mg/m
(9ppm)
40mg/m3
(35ppm)

160yg/m3
(O.OSppm)

160yg/m3
(0.24ppm)

1OOyg/m3
(O.OSppm)
1300yg/mJ
(0.5ppm)

60yg/m3

150yg/m3

1Omg/m3
(9ppm)
40mg/m3
(35ppm)

160yg/m3
(O.OSppm)

160yg/m3
(0.24ppm)

1OOyg/m3
(O.OBppm)
Averaging
  Times

Annual
arithmetic mean

24 hours
 3 hours


Annual
geometric mean
24 hours

 8 hours

 1 hour


 1 hour


 3 hours
Annual
arithmetic mean
ppm = parts per million

Source:  Code of Federal Regulations, Title 40, Part 50,  pp.  3-28,  July 1,  1975.

-------
 Determination of Long-Term  Air  Pollution Trends

      The implementation  of  realistic air pollution control strategies will
 inevitably take time.  The  effectiveness of these strategies, as reflected by
 the gradual  changes  in air  quality, can be evaluated through painstaking
 comparisons  of historical records of measured air quality data.  In addition,
 the effects  of increases in population and changes in land use on air pollu-
 tion can also be assessed by scrutinizing the long-term trends or patterns.

 Perimeter Monitoring Programs

      In the  general  air  quality monitoring programs described above,
 "areawide" measurements  are being sought.  By contrast, perimeter monitoring
 programs are designed primarily to obtain source-oriented measurements.  The
 eventual goal is the calculation of pollutant fluxes.  The need for this type
 of program can arise out of the following concerns by State or local agencies.

 Enforcement  of Air Pollution Control Regulations

      Federal  or local regulations usually restrict the amount of air
 pollutants emitted by any given industrial installation.  Although in-stack
 monitors can be used to  ensure  that these regulations are being met, enforce-
 ment can also be accomplished by source-oriented monitoring.  The emission
 flux can be  calculated from data obtained from simultaneous monitoring of the
 concentration levels and wind speed and direction along the perimeter of the
 industrial  plant.  This  flux can in turn be used to estimate whether the
 regulations  are being violated.  For example, see Sperling (1975).

 Estimation of Regional Air  Pollutant Fluxes

      "Line-wise" types of measurements can also be used to settle legal dis-
 putes  regarding pollutant fluxes across the boundaries of regions under dif-
 ferent jurisdictions.  For  example, a perimeter monitoring program can provide
 the data necessary to quantify  the transport of airborne pollutants into a
 county from  an adjacent  upwind  county.

 Special  Monitoring Programs

     This  type of  monitoring program is topic- and project-oriented.  The
 measurements  are made over  a short time span, on the order of weeks or months,
 and  are  of interest  only to a special group.  Some of the more important
 goals  of this  type of program are:

 Procurement of a Data Base  for  Regional Model Development

     Regional  models will eventually become an indispensable part of air
 pollution control  programs.  Measured air qualities at a series of monitoring
 stations comprise  an important  part of the data base for validating such
models.  The monitoring  data can also aid in the refinement of regional
models.

-------
Air Quality Impact Studies

     A special monitoring program may be needed to establish the baseline
pollution levels when the construction of a new roadway or shopping center is
contemplated.  The information can be used in transportation control  or land
use planning.

Special Research Studies

     Special measuring programs can be initiated to investigate certain specu-
lated cause-effect relationships.  Examples are studies correlating indoor air
pollution with outdoor air pollution and carbon monoxide concentration on the
roadway with the driver's blood hemoglobin level.

FACTORS PERTINENT TO MONITORING NETWORK DESIGN

     Once the objectives of a monitoring network have been ascertained, an
investigator will be confronted (in the process of planning and designing the
network) with a variety of factors that will affect the final network design.
As it turns out, many of these factors are not necessarily technical.  Most
of the technical aspects will be considered in detail later in this report.
For an overview, it is of interest to briefly summarize both the technical
and nontechnical factors.

Technical Factors

     Included here are factors that can be either evaluated or quantified;
these are:

Emissions

     Because air pollution is a consequence of emissions from sources, this is
probably the most important parameter.

Atmospheric Dispersion and Transformation

     The concentration levels of air pollutants at any receptor point are
determined not only by the emission sources, but also by transport, diffusion,
and chemical and physical transformations that take place in the atmosphere.
These factors can be extremely important in the selection of the monitoring
sites.

Costs
                                                     /
     Both the fixed (capital investment) and variable (recurring) costs for
the construction and maintenance of the monitoring network can also figure
prominently in the decision-making.

Hardware and Software

     A variety of factors concerning the instrumentation (such as precision,
response time, and operating conditions) and data reduction procedures may

-------
also affect the overall design of a monitoring network.
Nontechnical Factors
     A  number of nontechnical factors exist that can also be important in the
deliberation of site selection.  These include:
         Accessibility of the monitoring site.
         Safety of personnel and security of equipment.
         Availability of utilities (electricity, water, etc.).
         Ability to accommodate future modifications or expansions.
         Compatibility of purpose with other surveillance networks.
The important role played by these factors in the final decision of monitor-
ing sites should not be underestimated.  In Yamada's (1970) statistical
survey  of past practices in the design and siting of monitoring stations, the
reasons given for locating stations were as follows:
                 Locating Reason                             Percentage
     Compatibility with station purpose                         50.0
     and availability of site
     Compatibility with station purpose only                    25.0
     Availability of site only                                  12.5
     Others                                                     12.5
The importance of factors other than technical can be clearly seen from
analysis of the information presented in this table.

-------
                                 CHAPTER II

               DESIGN OF A NETWORK FOR AIR QUALITY MONITORING


     Concentration levels of air pollutants are observed not only to fluctuate
with time, but also to vary significantly from one location to another even
within the same proximity.  This variability in observed concentrations is
the consequence of both the complex emission pattern and atmospheric disper-
sion and transformation processes in a region.  Since most conventional
measuring techniques allow only "point" sampling, the measured data at any
arbitrarily chosen site may or may not adequately represent the air quality
in a larger area surrounding the monitoring site.  It is thus clear that a
critical problem in the siting of air quality monitoring stations is to estab-
lish the degree of "representativeness" of the station measurements.  In the
first section of this chapter, major components in the design of an air
quality monitoring network are discussed.  This is followed by a section
delineating the various issues regarding this question of representativeness
of station measurements. Also described is the methodology adopted in this
study for siting monitoring stations.

COMPONENTS IN THE DESIGN OF AN AIR QUALITY MONITORING SYSTEM

     The establishment of an air quality monitoring system, either for
operational use or for special studies, is not a small undertaking.  Usually,
it commits enormous monetary and human resources.  Therefore, the planning  and
design of such a system require the utmost in care.

     The concentration of pollutants in a region and the flux of contaminants
into, within, and out of the region are highly variable quantities in space
and in time.  The location of the limited number of stations permitted by
financial resources requires careful planning so that measurements that are
not typical of the region as a whole can be avoided.  Errors due to local dis-
turbances or instrument malfunctions can be minimized by, an assessment of the
site characteristics, the limitations of the instrumentation, and by estab-
lishment of an adequate instrument calibration and maintenance program.

     The design of a system can be conveniently divided into three parts:

Monitoring Station Selection

     The success of a monitoring program depends heavily on the appropriate
design of the monitoring network.  This design consists of the following
important components:

-------
 Estimation of the Number of Monitoring Stations

     This is one of the factors that must be decided early in the planning
 stage.  Quantitative rules for reaching this decision are extremely useful.
 Ideally, the minimally required number of monitoring stations should be de-
 termined from a cost-effectiveness analysis based upon optimization tech-
 niques.  This was apparently not the past practice.  As we have discussed in
 the  review, Figure 1-1, which was based on statistical data, correlates the
 number of monitoring stations required with the population of the region.
 Despite the fact that people are admitted polluters, their distribution and
 number may not reflect the distribution and strengths of problem pollutants.
 Therefore, any judgment concerning the number of stations should be based on
 more direct factors, such as emissions distribution patterns.

 Determination of the Mode of Monitoring Stations

     In the past, air pollution monitoring has been primarily accomplished by
 fixed-station measurements.  This practice is probably related to the fact
 that instruments used in these measurements were mostly of a wet-chemical
 type, requiring a stable and highly controlled environment.  With the rapid
 advances in instrumentation hardware, mobile monitoring stations have become
 an important monitoring mode.  Although mobile stations usually serve as
 complementary methods in general air pollution monitoring programs, for perim-
 eter monitoring or for special monitoring programs mobile stations may out-
 perform fixed stations, either on economical or operational grounds.  A
 typical description of the design and test of a mobile station mounted on a
 van  can be found in a report by Ingram and Golden (1969).  Other platforms,
 such as airplanes or even satellites, have also emerged recently as good
 candidates for the mobile monitoring of air pollution.

 Selection of Monitoring Sites

     This is, or course, the crucial part of the entire planning process.
 How  close the measurement will fulfill the intended mission of the monitoring
program will  be judged by the "representativeness" of the measured data,
which is in turn critically dependent on the selection of monitoring sites.
This issue will be discussed in detail later.

 Instrumentation System Selection

     Proper selection of the instrumentation system plays an essential role in
assuring a successful monitoring program.  Among the more important elements
are  the following:

         Determination of species to be measured.

         Selection of measuring devices, i.e., choice between manual or
         automated systems.

         Scheduling of sampling frequencies and duration.
                                      10

-------
Data Acquisition and Analysis

     The final  link in the measuring system is the transmission and  transla-
tion of data registered by a sensor.  The following considerations should,
therefore, be included:

         Establishment of data acquisition and processing  system.

         Formulation of procedures for data reduction and  analysis,  including
         quality assurance.

     The foregoing constitutes a brief outline of the three major components
in the design of an air quality monitoring system.  However, only the first
one, i.e., site selection, is addressed in this report.

SELECTION OF MONITORING SITES

     A rational definition of "representativeness" can be  made  in relation to
the objectives of the monitoring program.  Therefore, this chapter first
presents a set of quantitative definitions of the representativeness of
station measurements in terms of some of the objectives  discussed in the  first
chapter.  Most air quality standards state that specified  maximum concentra-
tion levels must not be exceeded more than a certain number of  times during
specific time periods.  Therefore, for air monitoring programs  that  have
compliance with air quality standards as their primary objective, a  measure
of the representativeness of the data collected at a station would be defined
according to the ability of the measured data to reflect true peak concentra-
tions in the region being monitored.  In this case, it is  important  to locate
monitoring stations at maxima in the spatial concentration distributions.  On
the other hand, if the detection of long-term trends is  a  major monitoring
objective, the station measurements can be considered as representative if
they are sensitive to the effect of changes in regional  source  emissions  on
air quality.  It is clear that, to achieve this goal, the  ability to measure
the rate of change of concentration levels with respect  to time—rather than
the concentration maxima themselves—is most critical in assessing representa-
tiveness.  Similarly, other criteria can be established  for different monitor-
ing objectives.  It should be noted, however, that only  the first one—
compliance with, or progress toward attainment of, air quality  standards—is
considered in this study.

     In order to locate monitoring stations to identify  concentrations which
exceed the national ambient air quality standards (NAAQS), one  must  first
make some extremely important interpretations of the standards. First, one
must decide what constitutes a violation.  Clearly, if we  measure close to the
tailpipe of an automobile, ambient air quality standards for several pollut-
ants are  likely to be violated at all  times, even if the  vehicle is operating
well within emission standards.  Therefore, one reaches  the conclusion that
the standards do not apply directly to air spaces which  are close to the  tail-
pipe.   In fact, there is no single definition of where standards should apply.
Similarly, if one were to measure concentrations at curbside along a busy
street,  the chances are good that a large number of measurements exceeding the
standard could also be expected.  Clearly we cannot measure all of them,  nor

                                      11

-------
 is  it necessary.   It  is thus important to clearly define under what situations
 the NAAQS are exceeded.  Obviously this question cannot be resolved without
 other considerations, such as population exposure.

     In the present study, the method used for identifying concentrations
 which exceed the NAAQS can be expressed in terms of a Concentration Area Time-
 Product (CAT-P).
          CAT-P = J ( JJ c(x,y,
t)dxdy I  dt
(II-l)
                  t*
where A  represents an area and t* a time interval for averaging the concen-
trations.  A concept similar to this was proposed by Duckworth (1967) for
describing the severity of air pollution episodes.  The time intervals can be
selected  to match those as specified in the ambient air quality standards
(for example, 1 hour and 8 hours for carbon monoxide).  The choice of aver-
aging area A is dictated by considerations including the spatial  resolution
of  emissions inventory, the size of the region under consideration, and the
storage  capacity of the computer to be used.

     Once a clearly defined interpretation is obtained, the next problem in
the selection of monitoring sites is the determination of pollutant concen-
tration  variations.  Since air pollution is the direct consequence of
emissions from sources, an inventory of emissions, including magnitudes and
temporal  variations, is clearly one of the most important inputs.  Meteorolog-
ical parameters, such as wind speed, wind direction, and atmospheric stability
as  derived from vertical temperature profiles, are also indispensable in
determining the distribution of air pollutants.  With the application of a
mesoscale air quality simulation model, pollutant distributions on a regional
scale can be obtained for varying emission and meteorological conditions.

     Finally, since pollutant concentrations are measured only at particular
points in a region under consideration, the measurements can be expected to be
strongly  affected by environmental factors in the immediate vicinity of moni-
toring sites as selected by the mesoscale model.  For example, the measure-
ments on  a local scale can be affected by localized sources that can cause
local  concentration levels to be significantly higher than the average for the
area.   Local  structures can also influence either the air flow field or
pollutant dispersion in the vicinity of the monitoring site.  At this micro-
scale level,  important parameters that are expected to influence the pollutant
distributions include:

          Relative magnitude of the local sources.

         Distance from local  sources.

         Microscale meteorology.
                                      12

-------
         Natural topography or man-made structures in the neighborhood of the
         monitoring site.

Thus it is clear that the actual siting of a monitoring station requires the
knowledge of pollutant distributions on the microscale.  This problem is dis-
cussed in detail in Chapter IV.

APPROACH ADOPTED IN THIS STUDY

     The primary goal of the monitoring network to be designed is to identify
concentrations which exceed the NAAQS.  Therefore, stations are to be located
in such a fashion as to minimize the probability of not detecting'a violation.
Carrying out this objective requires interpretation of a number of -related
Issues.

     The first issue, as discussed in the previous section, concerns the
interpretation of the NAAQS.  A Concentration Area Time-Product (CAT-P),
given by equation II-l, is used for identifying the concentrations which
exceed the NAAQS.  Assuming that this is appropriate, the use of a mesoscale
model would be ideally suited for computing the CAT-P as follows:


                           CAT-P = c(x,y,t)AxAyAt                 (II-2)


where

     c(x,y,t) = computed concentration

        Ax,Ay = step size in the x- y- directions

           At = step size in time.


Although, in principle, the mesoscale model can be used to obtain annual or
long-term averages, it is not practically feasible.  Because it is implicitly
assumed that the monitoring system is to be designed for long-period opera-
tion, the question then arises as to how to include the long-term fluctuations
in the pollutant concentrations due to not only daily but also seasonal  vari-
ations of the meteorological parameters in the area.  The problem can be
treated by using a frequency-weighted concentration expressed as an index
called the Figure of Merit.

     The Figure of Merit for a particular point can be defined in general as
the sum of the products of the ground-level concentrations and the associated
frequencies of occurrence.   Two types of Figures of Merit are of particular
interest to the present study—one for exceeding ambient air quality standards
and one for general  air quality monitoring.

     For each of the grid points in the modeling region, the Figure of Merit
for exceeding standards can be defined as follows:
                                      13

-------
                 1, if AAQS are exceeded at \    /probability of\
                    grid point j,k under     ] .   (meteorological  I   (II-3)
                    meteorological pattern £/    \ pattern £     /

                 0, if not                  '    \
 Thus,  this Figure of Merit is a measure of the probability that each  grid
 point  is  likely to indicate when an air quality standard is exceeded.

     Similarly, a Figure of Merit for general  air quality monitoring  can be
 defined as
                 /concentration at grid \     /probability of\
   P  ('  b\  -  y  I point j,k under tneteor-1  •   I meteorological  I      (II-4)
   i-  U,KJ  -  /_,  lological pattern £     /     \pattern £      /
             *   \                      /     \              /
 This  index  represents an average pollutant concentration at each grid point
 as weighted by the frequency of occurrence.

      In order to implement the scheme, it is necessary to specify a  set of
 scenarios which completely describe the meteorological conditions in the area
 of concern.  This can be accomplished by examining climatological data in the
 region of interest.

      Once a set of meteorological scenarios is determined, the mesoscale model
 can be exercised to provide the corresponding pollutant distributions.  The
 resultant ground-level concentration distributions, in conjunction with the
 associated  frequencies of occurrence, can then be used to compute the Figure
 of Merit for siting potential monitoring stations.  A computer program has
 been  prepared to perform these calculations.  To illustrate the use of this
 program, values for Figure of Merit defined by equation II-4 were computed
 using simulated surface carbon monoxide (CO) distributions in ppm for a
 1-hour period during peak traffic for two separate days (see Figures II-l and
 II-2) for a hypothetical region with 1-by 1-km grid squares.  Figure of Merit
 values were calculated under the assumption that these situations have an
 equal probability of occurrence.  An isopleth plot of these values is
 presented in Figure II-3.  The computer program also takes the resultant
 values for the Figure of Merit and searches for the highest values which are
 not adjacent to other higher values without an intervening trough.  As shown
 in Figure II-4, locations for the 16 highest Figures of Merit were identified,
 ranked, and plotted.

      In conclusion, it should be emphasized that the procedures described
above are by no means definitive.  They are proposed here only to illustrate
an idealized concept for the objective and systematic determination of air
quality monitoring stations as outlined in this chapter.  Modifications of the
basic approach are certainly possible and probably desirable.  For example,


                                      14

-------
Figure II-l.
Simulated surface carbon monoxide distribution across  a
hypothetical region for the hour of peak traffic for a
particular day.  Grid squares are 1 km and isopleth in-
crement is 1 ppm.  (Day 1)
                    15

-------
Figure II-2.
Simulated surface carbon monoxide distribution across  a
hypothetical  region for the hour of peak traffic  for a
particular day.   Grid squares are 1  km and isopleth in-
crement is 1  ppm.  (Day 2)
                                  16

-------
Figure II-3.    Figure of  Merit  distribution for data in Figure II-l and II-2.
               Isopleth increment  is  1  ppm.
                                      17

-------
                                 m
Figure II-4.   Ranking of the potential  monitoring sites by Figure of Merit
               for data in Figure I1-3.
                                     18

-------
normalization of the Figure of Merit as defined by equations  11-3  and  11-4  or
inclusion of demographic information may yield new definitions  which are
intuitively more appealing.  On the other hand, a close examination of the
averaging time is required to clarify certain ambiguities  in  the definitions
of the Figure of Merit.  These questions apparently warrant future considera-
tions.
                                      19

-------
                                 CHAPTER III

                       ANALYSIS AT THE MESOSCALE LEVEL


      The  approach adopted in this study for siting monitoring stations is
 based on  the concept of a CAT-P for identifying the locations of grid areas
 in  a  region whose concentrations exceed the NAAQS.  Since grid models normally
 compute space- and time-averaged pollutant concentrations, it appears that
 this  type of air quality simulation model is ideally suitable for such an
 application.  The first section of this chapter is thus devoted to a discus-
 sion  of the mesoscale or regional model selected for this project.  The
 second section then addresses the very important question concerning the
 incorporation of an improved vertical diffusivity algorithm into the model.

 A DESCRIPTION OF THE MESOSCALE AIR QUALITY SIMULATION MODEL

      The  goal of the mesoscale analysis is the accurate determination of pol-
 lutant distributions on the regional scale under a variety of emission and
 meteorological conditions.  This can be achieved by use of a mesoscale air
 quality simulation model.  The model selected for the present project was
 developed by Systems Applications, Inc., (SAI) (Reynolds et al . , 1973; Roth
 et  al., 1974; Reynolds et al . , 1974).  The model  is based on the following
 simplified form of the equation of mass balance for an inert pollutant
 species like carbon monoxide (CO):
 If *«£* v£* -if ^1),          )(K2f)     (III-l)


where:  c is pollutant concentration; t is time; x,y, and z are the space
coordinates; u,v,w are wind speeds in the x,y, and z directions; and K , Ky,
and KZ are turbulent eddy diffusivities in the x,y, and z directions.

     This partial differential equation in four dimensions (x,y,z,t) is solved
by a finite-difference technique.  The method of fractional steps (Yanenko,
1971) is applied by splitting the original 4-dimensional equations into three,
2-dimensional equations in (x,t), (y,t), and (z,t), respectively.  The solu-
tion is explicit and direct in the line (x,t) and  (y,t) fragments and implicit
and iterative in the (z,t) fragment.  The emissions from areal and point
sources within the region under consideration enter the model  equation as
boundary conditions, via the following expression:
                                      20

-------
                            -K2 ff   =   Q  (x.y,t)                  (III-2)
                                   1 z=0
where:

     Q  (x,y,t) is the pollutant emission rate.


     The top of the modeling region is usually set at the base of an elevated
inversion or stable layer capping the atmospheric mixing layer.  With the
modified diffusivity scheme, which is discussed in the next section, this con-
dition  can be relaxed.  For example, under a stably stratified atmosphere,
the top of the modeling region can be chosen at a height above which pollutant
concentrations level off to the background values.

     The 3-dimensional wind field (u,v,w) as a function of time is considered
as an input that must be prescribed.  In the present model, the 3-dimensional
wind field is calculated based upon surface wind measurements via an inter-
polation algorithm (Liu et al., 1973).  The generation of ground-level  wind
speed and direction at a particular grid cell in time is based on the inter-
polation of the field measurements using a weighted inverse square of the
distance between station locations and the grid cell under consideration.
Once the surface wind is computed, the wind speeds and directions in upper
levels  can be calculated based on the continuity equation of mass.
                    ji

     One of the important assumptions invoked in the derivation of the model
equation discussed above is the gradient transfer approximation, also known as
the K-theory.  This approximation is tantamount to assuming that the turbulent
transfer of pollutants in the atmosphere is proportional to the gradient of
the mean concentration.  For the SAI model, according to a sensitivity study
carried out by Liu, et al., (1975), an order-of-magnitude change in the hori-
zontal  eddy diffusivity will only affect the predicted surface concentration
by less than 3 percent.  Thus, a constant value of 50 m^/s is used which is
compatible with a grid square up to a few kilometers on a side.

     In a previous study, a simple scheme incorporating a height and wind-
speed dependence was used for the prescription of the vertical eddy diffusivi-
ty (Reynolds, et al., 1973).  It was found to be satisfactory for the
Los Angeles basin during daylight hours.  However, the application of the
model  to other regions of interest and for other diurnal periods of the day
requires the use of a more generalized diffusivity algorithm.  The following
section is devoted to a discussion of the basic limitations of the K-theory
and development of such an algorithm.

APPROXIMATION OF TURBULENT TRANSFER BY EDDY DIFFUSIVITIES

     Like many other related studies of the atmosphere, a difficult and also
crucial  part in the simulation of pollutant dispersion is the attainment of a
reasonable scheme to represent the turbulent, processes.  Theoretical studies
based  on higher closure schemes for hierarchies of turbulence moment equations

                                      21

-------
 have recently been  carried out  (e.g., Deardorff, 1970, 1972).  However, these
 have been  restricted  in application to certain special cases and require con-
 siderable  computational effort.  Thus, an alternative based on the simple
 concept of turbulent  eddy diffusivities—the K-theory—is adopted in most
 atmospheric dispersion models of the type used here.  Analogous to the molec-
 ular diffusion,  the K-theory speculates that a pollutant flux in the direc-
 tion of decreasing  concentration is established as a result of turbulent
 fluctuations.   The  magnitude of this flux is proportional to the gradient of
 the average concentration.  Thus:
                                     • - «*  S£                     (III-3)
                                                                    (III-5)
      The  limitations of models based on the K-theory or the gradient transport
 theory are  well  known.  They can be generally grouped into the following two
 categories:

          Length  and time-scale constraints;

          Directional constraints.

 The  first type is related to the spatial and temporal homogeneity of the mean
 concentration field.  Corrsin (1974) summarized the conditions necessary for
 satisfying  such  constraints:

          The transport mechanism length scale must be much smaller than the
          distance over which the curvature of the mean transported field
          gradient changes appreciably.

          The transport mechanism time scale must be much smaller than the time
          during  which the mean transported field gradient changes appreciably.

          The transport mechanism length scale must be essentially constant
          over a  distance for which the mean transported field changes
          appreciably.

     The  second  constraint arises when, for example, the Reynolds stress, a
second-order tensor, is replaced by an inner product of a second rank tensor
and a vector.  The conditions for satisfying this constraint are more
difficult to delineate.  However, qualitative estimates for the validity of


                                      22

-------
the mesoscale model  based  on  the  equation of mass balance for a pollutant
species were obtained  by Lamb and Seinfeld  (1973).  The result seemed to indi-
cate that the gradient-transport  approach is plausible under a variety of
conditions.

     Based on a comprehensive review  of  the literature, an algorithm for pre-
scribing the eddy diffusivity in  the  vertical was adopted.  In the surface
layer, the following general  formula  was used:
                                             (z0() = a functional  relationship to  be  specified.


This formula is the result of the similarity  theory for the constant-flux
surface layer (Businger et al., 1971).  For the neutral case, the -function
equals unity.  For the stable or unstable case, the ^-function is greater or
less than one respectively.   The following empirical expressions for the
<|>-function were proposed by Businger et al.,  based on observational data.

     For the stable case (L > 0)

                             (f) = 1  + 4.7 t^\                     (III-7)



     For the unstable case (L < 0)

                                                  -1
                             (f)   =   (]  - 15 f)
The friction velocity was determined by the  following equation,

                                        kU
                                   u* = T^                          (III-9)


                                      23

-------
where U  denotes a reference horizontal  wind speed measured at a  reference
height, Zf, and
f =  In
or
             f =  In
                   + 2 tan
                          -1
                                   4.7  Zr ' zo
                            *fe
                                1
-  In
 - 2 tan
        '1
                            (111-10)
                            (stable)
                            (III-ll)
                            (unstable)
     Above the surface layer (|L| <_ z <. Z^),  a second-order interpolation
formula first proposed by O'Brien (1970) was  utilized:
                                                           (z -
                                                                    (111-12)
where Z. = height of stable layer capping mixing layer or height above  which
           pollutant concentrations level off to background values  (stable
           conditions).
      K   = vertical diffusivity of height i.
      K(|L|)=dKz
                __
                             = IL.
     The implementation of the proposed diffusivity scheme requires an esti-
mate of the Monin-Obukhov length.  The Monin-Obukhov length can be in general
related to the Richardson numberjwhich can be, in turn, determined experimen-
tally (McElroy, 1969).  The estimate of this length was, however, accomplished
in the present study via the following formula which relates the Monin-Obukhov
                                      24

-------
length to the surface roughness, ZQ, and the stability function, S,
       L =
(a^ + a2S )  z
                                                    -1
                                                              (111-13)
where
al =

a2 =


bl -

bo =


bo =
           0.004349


           0.003724


           0.5034


           0.2310


           0.0325
This formula is a result of the best-fit of observational  data reported by
Colder (1972).  The stability function, S, a digital version of the Pasquill
stability category (see Table III-l), can be calculated as follows:
                       S = \ (3 - dw + |de|) 'Sign (de)
                                                              (111-14)
where
                       sign (d ) =
                                1  de > 0

                                0  d_ = 0
                                     -1  d  < 0
                                          e
and dw and d  are the wind speed class and exposure class.  These are defined
as follows:
                  r     0 <_ U  < 8 m/s
                 2           r
                 4      U  > 8 m/s
                         I
                                      25

-------
                 3
                 2
                 1
                 0
                 -1
                 -2
strong
moderate
slight
heavy overcast
daytime insolation
day or night
                       nighttime cloudiness.
            TABLE  III-l.   PASQUILL CATEGORY AND STABILITY FUNCTION
               Pasquill  Category*
                      A
                      B
                      C
                      D
                      E
                      F
                  Stability Function,  S
                          -3
                          -2
                          -1
                           0
                          +1
                          +2
*Turner  (1969)
     The prescription for the computation of the vertical eddy diffusivity is
now completed.  The primary inputs are the wind speed, the reference height
Z-, the exposure class, and the surface roughness.  The result of a sample
calculation is included in Figure III-l.  Note that an approach for the deter-
mination of vertical eddy diffusivities generally similar to the above was
suggested by Myrup and Ranzieri (1975).
                                      26

-------
ro
                  ZOO
                I  100
                               002       0.04
 0.2        0.4
DIFFUSIVITY (m2/$M)
                                                                                     WIND  SPEED =2 in/sec
                                                                                     STABILITY CLASS =E
                                                                                     REFERENCE HEIGHT =2BOm
0
O
O
O
O
O
O
0
O
1 1
O
O
O
O
O
O
0
O
O
1 1
O
0
O
O
O
0
0
O
0
1 1
2.0        4.0
                        Figure  III-l.    Calculated vertical diffusivity as  a function  of height.

-------
                                 CHAPTER IV

                      ANALYSIS AT THE MICROSCALE LEVEL


     The primary objective of the microscale analysis  is  to  pinpoint  the
monitor in a microscale environment within the area identified  by the meso-
scale analysis.  The major task is to find a sensor location at which the
sensor's readings are not unduly influenced by local  sources.

OVERVIEW

     The problem of selecting monitoring sites on the  microscale arises
because the mesoscale model predictions only provide an average estimate of
the pollutant concentration over a large geographical  area (typically this
area represents a 1-km by 1-km square), while a conventional sensor measures
only a point value at some arbitrary location within the  grid square.  Depend-
ing on the emissions and meteorology pattern and the relative sensor  location,
the sensor may or may not provide a measurement which  can be considered as a
representative value for the grid square.  Clearly, what  is  needed is an anal-
ysis at the microscale to establish a methodology for  locating  the sensor on
the microscale and a method for relating its readings  to  a mesoscale  average
value.

     The only previous work related to the siting of monitors on a local scale
was that of Ludwig and Kealoha (1975) which forms the  basis  of  the current EPA
siting guidelines for CO (U.S. EPA, 1975).  The essential feature of  their
analysis consisted of defining a variety of locations  at  which  a monitor could
conceivably be placed (street, canyon, corridor, neighborhood,  etc.)  and the
various types of measurement one could make (peak, average,  etc.). As a
criterion for siting for corridor and neighborhood stations, Ludwig and
Kealoha characterized the roadway as a Gaussian line source,
                   c-    -=®	   exp
                               Uozsin  9
where:

     c = concentration (g/nr)

     Q = source strength (g/m/s)

     U = horizontal wind speed  (m/s)
                                      28

-------
     c?z = standard deviation of a Gaussian plume in the vertical  (m)

     9  = angle between wind and line source (degrees)

     z* = height to instrument inlet (m).


The. dispersion coefficient, a , was represented by the equation
                                  a  = as
                             (IV-2)
where a and b are functions of the Pasquill stability class and s is downwind
horizontal distance.  When the conditions that either the contribution from
any one source be small or that the gradients of such contributions be small
were imposed, the following inequalities were derived,

                                 2Q
                              ? Ua sin 6 <  max
                                 1
Ac
                              c sin 6  As
                                           < 6.
                                              max
where Cmax is the maximum allowable contribution of any one source to the con
centration and Gjnax is the desired maximum concentration gradient away from
the roadway.  With the following typical values for the various parameters,

     6     =40 degrees

     U     =  1 m/s

           =  °-001  /m3
     Gmax  =  0.002/m

     Q     .=  0.07 g/m/s

they were able to determine the minimum distance between a large roadway and
a neighborhood monitoring site.  Given the uncertainties in the Gaussian model
and the assumptions involved in its implementation, this approach may have
furnished a qualitative analysis, but it did not provide quantitative informa-
tion on pollutant concentrations at such site locations under a variety of
atmospheric and emission conditions.  In addition, the method failed to pro-
vide information on the representativeness of data obtained from stations
located at places other than their suggested locations.
                                      29

-------
REPRESENTATIVENESS OF MEASUREMENTS ON THE MICROSCALE

     Fundamental to the nricroscale analysis is the concept of the
representativeness of the measurements; i.e., given a point-measurement value,
at what monitor location can the measured value be considered as representa-
tive of an average concentration value over a predetermined length scale?
Schematically, this situation is illustrated in Figure IV-1 for a single
roadway with emission rate, Q, and sensor at distance, p, downwind and height,
z*, above the roadway.
           ROADWAY'
                                          SENSOR
                                            i
it
yX
Q
>**

c

T
z*

      Figure IV-1.   Schematic diagram of roadway and sensor location.
To simplify the analysis, a steady state condition was assumed for an infinite
line source normal to the s-direction.  The wind was assumed to blow in the
s-direction and have a speed varying with height.  For this case,  the equation
of mass balance for a species reduced to:
3?
                                       31
i^   9c
Kz  31
(IV-3)
in which both wind speed, U, and vertical diffusivity profiles must be
specified a priori.  In addition, the emission rate enters the equation as a
boundary condition:
                             K
                             KZ  3Z
       z=0
                                           * -Q.
                                      (IV-4)
                                      30

-------
     Since most sensors are located at a recommended height of 3 meters above
the ground (z* = 3 meters), variations in the vertical  were eliminated in the
final analysis by merely examining those concentration  profiles at the
3-meter level.

     To mathematically quantify the meaning of representativeness, the mean
value theorem of calculus expressed as
          c(p,3) D =
c(s,3)  ds
0 < s < D

D = distance downwind
    of roadway
(IV-5)
was used, from which the mean value of the concentration in the interval
[0,D] can be computed and the location of the sensor p predicting this value
obtained.  Graphically, these two quantities are illustrated in the following
figure:
  Figure IV-2.   Relationship of sensor location to measured concentration,


     For various emissions rates, wind profiles, and Pasquill stability
classes, a series of charts can be constructed relating p to D.

SOLUTION METHODOLOGY

     Two basic difficulties were encountered in trying to solve the governing
equation:
                                      31

-------
                           u  l£-  At r              (IV-8)
                                     >  32

-------
where
     K   = vertical diffusivity at height z.
       r = normally distributed random variable between ± °°
           with mean zero and unit standard deviation
     Wind speed profiles near the ground may be represented by a power law
model of the form
                                U
                                U
in which the exponent n is a function of stability and r refers to parameter
values at a reference height.  The n values calculated as a function of atmos-
pheric stability can be related to Pasquill stability classes.  Table IV-1
lists stability dependent values of mass provided by DeMarrais (1959).  The
n values were assumed applicable to Pasquill classes in the manner shown in
this table.  Since automobiles generate a wake as they move, it could be
anticipated that the wind speed does not smoothly decay to zero as the sur-
face is approached as predicted by the power law model.  According to a study
carried out by the California Department of Transportation (Ranzieri and Ward,
1975), the movement of a vehicle gives rise to a 4-meter-high mixing cell
"above the roadway" in which both dispersion and meteorological parameters
are roughly uniform.  For this reason, the wind speed profile was modified in
such a way that in the range between 0 and 4 meters above the ground the
speed would be set to the 4-meter value.


                  TABLE IV-1.   EXPONENTS FOR WIND PROFILE

                       Pasquill
                    Stability Class          Exponent n

                          A                     0.1

                          8                     0.15

                          C                     0.20

                          D                     0.25

                          E                     0.35

                          F                     0.30

                                      33

-------
      Finally, estimates of the eddy diffusivities were computed from
algorithms discussed earlier in Chapter III as based on the semi-empirical
theory discussed earlier which includes the effects of wind speed, surface
roughness, stability class, and vertical height.

     To facilitate the computations in the microscale analysis, the Monin-
Obukhov lengths for typical values of surface roughness around a roadway were
estimated.  These lengths as a function of prevailing atmospheric stability
class are listed in Table 1V-2.  As was the case with the wind speed profiles,
the diffusivity values were assumed to be uniform in the interval between zero
and 4 meters above the surface and equal to the 4-meter value.


     TABLE IV-2.   MONIN-OBUKHOV LENGTH VERSUS PASQUILL STABILITY CLASS

             Stability Class          Monin-Obukhov Length L (m)

                   A                            -9.09

                   B                           -22.9

                   C                          -133.3

                   D                             «>

                   E                           159.2

                   F                            16.0
     Although the methodology developed here was for the simple case of a
single roadway, this method may be extended to more complex situations in
which the same type of analysis is possible.  For example, if several road-
ways have to be dealt with, the contributions of each roadway must be super-
imposed.
                                      34

-------
                                  CHAPTER V

                          FIELD MEASUREMENT PROGRAM


     Application of mesoscale analysis requires that the validity of the
model used be demonstrated for the region to which it is applied.
Accordingly, an observational program must be developed to support the verifi-
cation effort.  One approach toward establishing such a program proceeds as
follows:

     Various attributes of the area are considered such as the size of the
region, land use, population density, emission patterns, topography, etc.
Examinations of historical information and current data collection activities
are made in order to determine their completeness.  Based on this examination,
any supplemental monitoring needs can be identified.  Thus, pollutants to be
monitored and meteorological conditions to be observed can be specified.  A
decision is made on the number of meteorological and air quality monitoring
sites, the specific locations, and the instrumentation required.  Other
requirements are specified such as resolution, frequency, duration, and
extent of observations and required accuracy, specificity and precision of
the measurements.

MODELING REGION

     To provide a reference system for measurement site locations, a grid
structure should be developed for the modeling region selected.  The bound-
aries of the modeling region must be defined.  A grid size is to be specified
for the area which would be oriented on the Universal Transverse Mercator
(UTM) grid system.  The UTM coordinate system is suggested in order to con-
form to the format of available data.  This grid also provides a reference for
topographic information and emission source locations.

     In a modeling region such as the Las Vegas Valley, the ridgelines of the
surrounding mountains define the boundaries for which a grid structure could
be developed.  A map of this region (Figure V-l) shows the flat, gradually
sloping valley surrounded by the Las Vegas Range to the north, Frenchman and
Sunrise Mountains to the east, the Spring Mountains to the west, and the
McCullough Range to the south.  The urban area consists of vacant desert
scattered among the residential developments.  The population of over 300,000
people is distributed over a larger area than other urban communities of
equivalent population.  Note that a limited access interstate highway trav-
erses the city and major highways crisscross the valley providing access from
Arizona to northwestern Nevada, and from California to Utah.  A large grid of
four-and-six-lane arterial streets and an intermeshed network of secondary
roadways accommodate local traffic in the valley.

                                      35

-------
            -•   *A
           /    "^T^-.
Figure V-l.    Map of Las  Vegas  Valley.
                  36

-------
HISTORICAL INFORMATION

     A survey and preliminary analysis of available aerometric data are
necessary for planning of the field measurement program.  For example, this
will provide guidance for locating field sampling equipment, developing
meteorological scenarios, and determining periods of field-data collection
for both continuous monitoring and intensive measurement efforts.

     Region-wide aerometric data may be available from Federal, State, and
local agencies, particularly with regard to the monitoring and other
provisions of the Clean Air Act.  These agencies and private concerns may
also have such data for specific areas in the region regarding environmental
impact assessments.  Many of the programs may be ongoing and yield quantita-
tive information useful in developing pollutant emission inventories.

     Usually, meteorological data are collected routinely by the National
Weather Service (NWS) at a local airport station or by the military at their
installations.  In some regions, the NWS has data available for locations
in metropolitan areas.  In addition, weather stations are operated in cooper-
ation with the NWS by other Federal agencies often at small  airports and
military establishments.  Data from these and other sources are compiled and
stored at and are available from the National Oceanic and Atmospheric Admini-
stration's (NOAA) National Climatic Center (NCC) in Asheville,  North Carolina.
Inquiries should be made to the NCC regarding such data including  climatic
and other summaries and results of special studies using the data.

     NWS airport stations providing near-surface weather information are
located about 90 km apart.  They nominally report at 1-hour intervals weather
information including that on cloud cover, visibility, barometric  pressure,
air temperature and moisture, precipitation, and wind speed and direction.
Some NWS, military, and cooperative stations operate on less than  a 24-hour
per day schedule and/or report only certain of the above weather elements.

     NWS stations providing upper air data on temperature, moisture, and wind
speed and direction are situated about 370 km apart.  Vertical  soundings of
these parameters are made nominally at 0000 and 1200Z (international standard
time).  Some military installations also take such soundings on this time
schedule.  Additionally, some NWS and military weather stations, including
many of the above, take vertical soundings of only wind speed and  direction;
the frequency and schedules for these soundings are often dependent upon the
local station responsibilities and, hence, may be irregular.

SAMPLING RATIONALE AND PLAN

     A period of sampling must be chosen both for the continuous monitoring
and the intensive measurement programs.  The examination of historical
aerometric data may allow a determination of both the primary season(s) of
interest and the meteorological situations for which abnormally high levels
of air quality are likely to occur in the region under consideration.

     In a region such as the Las Vegas Valley, the period for full scale
sampling of CO both for the routine and intensive programs would be November


                                      37

-------
through  February,  inclusive.  From a meteorological standpoint the conditions
potentially conducive for high CO concentrations occur most frequently during
this  period.  This might be expected since CO is emitted primarily from near
ground-level sources and the ventilation rate, which is comprised of the
products of mixing depth and the average wind speed through the depth, is
smallest during this period.  The available historical data on CO levels in
the Las Vegas Valley indicate that the CO values are in phase with this mete-
orological pollution potential.

      Las Vegas, which is situated in semi-arid terrain, experiences a desert-
type  climate characterized by nearly clear skies, a large diurnal temperature
change, and a strong nocturnal surface-based inversion.  Periods of minimal
pollutant dispersal, both horizontally and vertically, are associated with
the occurrence of  the nocturnal inversion, especially with the normally
decoupling of air  from upper levels with higher wind speeds from the air
within the inversion layer.  The inversion begins to dissipate and its base
begins to rise shortly after sunrise.  Because of the relatively large tem-
perature increase  in the portion of the inversion nearest the ground, its
base  rises slowest in the first few hours after sunrise.  The inversion
normally reforms at the surface beginning around sunset.  The times of sunrise
and sunset on an annual basis for the local area are shown graphically in
Figure V-2.  As is shown in this figure, sunrise occurs later and sunset
earlier during the November through February period than during the rest of
the year.  Where it is demonstrated that motor vehicles generate the major
portion of the emissions of CO, times of peak emission are critical.  Locally,
these occur between 0630-0830 and 1600-1800 local standard time (LST).  Thus,
the times of peak  emissions coincide most often with diurnal periods of mini-
mal pollutant dispersal in the November through February period.  Hence,
highest short-term (on the order of a few hours) concentrations of CO can
reasonably be expected to occur most frequently during this part of the year.

      Periods for short term intensive sampling can usually be chosen on the
basis of forecasts of synoptic scale weather patterns over the region of
interest.  In the  Las Vegas Valley, days with particularly limited atmospheric
dilution could be  selected.  Such situations generally occur when the valley
is under the direct influence of a slowly moving or stagnant high pressure
area.  A major constraint on intensive sampling in the valley would be the
elimination of weekends for which an adequate CO emission inventory cannot be
established from available data.

      Weather forecasts for intensive sampling can be made using facilities
available at the nearest NWS station.  Here, surface and upper air data
described previously and special aircraft and ship weather reports for at
least the contiguous U.S.  are routinely collected by teletype.  Usually,
charts are available covering this area plus portions of the adjacent oceans.
They contain observed and forecast weather information for ground level and
various constant pressure surfaces above ground.  Additionally, special fore-
cast information is available, generally resulting from simulations with
synoptic-scale meteorological  models.

     Specific forecasts in this manner are developed for periods up  to  72
hours in advance.   In reality, forecasts for periods 48 to 72 hours  in  advance


                                      38

-------
    0400 .

      w
      es
      2
      ao
      55
      m

    0700


    1ST


    1600


      vt
      ts
      V*
      3

    1900
          JAN    FEB   MAR  APR  MAY JUN  JUL  AUG   SEP  OCT NOV  DEC

            Figure V-2.    Sunrise and sunset times for Las Vegas
                           (Local Standard Time).
 normally provide guidance regarding general circulation periods.   Those for
 periods less than 24 hours can usually be expected to provide reasonably
 accurate information on details and timing of specific weather events.   Hence,
 only one day's notice can normally be expected for intensive periods.   Such a
 constraint requires personnel on a standby status based on the longer term
 forecasts.

     Based on predominant synoptic-scale and locally induced wind patterns
 (i.e., upslope and downslope winds), topography, and traffic data, initial
 sampling sites can be selected to provide data for model verification and
 supplementary data for use in the design of a CO monitoring network.  If coop-
 erative arrangement can be made with other monitoring agencies, only a few
 new sites may be required to supplement those existing.

 QUALITY ASSURANCE

     Quality assurance takes on several aspects in the field project.   First,
 if field data are collected by several agencies working cooperatively, it
will be necessary to ensure that measurements within the individual programs
are comparable.   Second, data and data handling procedures must be compatible
and as error-free as possible.
                                      39

-------
     A significant way to ensure uniformity in air  quality measurements  is  to
require that all span gases used in field instrumentation  be  calibrated
against the same National Bureau of Standards standard  reference  gas.  A lead
agency should provide a cylinder of known concentration to be used  as  a  blind
sample for each agency for data correlation purposes.

     Routine and preventive maintenance and frequent instrument calibrations
should also be carried out by each agency.   Standardized check lists and log
books should be kept for all instrumentation.  Routine  maintenance  schedules
should be specified in an operator's manual  developed for  the project.   To
ensure proper identification, strip charts  should be labeled  at each calibra-
tion with the site name, the span value,  time, date, and other information  as
is applicable.

     A complete quality assurance plan should be developed for the  field
program following procedures and format suggested in the EPA  (1976b) publica-
tion, Quality Assurance Handbook for Air Pollution  Measurement Systems.
                                      40

-------
                                 CHAPTER VI

        APPLICATION FOR SELECTION OF CARBON MONOXIDE MONITORING SITES


     An air quality simulation model is an essential component of planning
studies in air pollution control because there commonly exists a need to
establish a quantitative relationship between effluent emissions rates and
the magnitude of ground-level contaminant concentrations that result from
these emissions.   Because each region may have unique topographical  and mete-
orological conditions, and specific pollutant emissions patterns, the model
is a useful tool in the design of a monitoring network to detect concentra-
tions exceeding the NAAQS.  This chapter discusses the development of a
design for an ambient air monitoring network for carbon monoxide using the
techniques developed in earlier chapters.

GENERAL METHODOLOGY

     Since the primary objective of the air monitoring network is to document
compliance with or progress toward meeting of the NAAQS, the CAT-P is used to
identify the locations where concentrations in excess of the ambient CO
standards are most likely to be found.  These are computed with the use of
the mesoscale air  quality simulation model.  A field measurement program is
established to provide air quality and meteorological data for model valida-
tion, and for simulation to yield projected space-time average concentration
distributions of CO.

     Prevailing meteorological patterns as well as the frequency of
occurrence associated with each can be applied, with the projected concentra-
tion distributions, to yield a frequency weighted average of concentrations
called Figure of Merit.  A mapping of these values provides the basis of
selection of the locations and number of sites.

MODEL INPUT REQUIREMENTS

     The SAI model, developed as a nonlinear, second-order partial
differential equation, requires a numerical solution with the aid of a digit-
al computer.  The overall and specific operational characteristics of the
computer programs which make up the model are discussed in a separate docu-
ment as a User's Guide (EPA, 1977).  The program package consists of a data
preparation portion and a simulation portion.  As a general use model, inputs
to the programs are parameters which specify the characteristics of the region
to be modeled.   Because of the number and types of parameters involved, only
a major set of input requirements is discussed.  The User's Guide should be
consulted for further details on specific input parameters.


                                      41

-------
 Emissions Inventory for Carbon  Monoxide

      The emissions inventory can  be  developed in accordance with the EPA
 (1973) publication, Guide for Compiling a Comprehensive Emissions Inventory.
 Emission factors are compiled in  EPA (1976a) AP-42. Compilation of Air
 Pollutant Emission Factors,  second edition, and its five supplements.
 Essentially,  the inventory involves  the summing of emissions from point
 sources, area sources,  and mobile sources, resulting in emissions per hour
 per grid square.

      The emissions factors given  in  AP-42 can be used to compute total
 emissions.   How the emissions might  be spatially distributed throughout the
 day, week or year, must be developed from other data such as traffic patterns
 and densities.

      An emissions inventory  depends  upon quantifiers such as the number of
 kilowatts produced per  hour  or  the number of automobiles traveling at a
 certain speed.   For light duty  vehicles, emissions have also been shown to
 depend on ambient temperature.  Thus, a temperature correction factor and a
 cold-start correction factor which is a function of temperature must be
 applied to their emissions.   For  example, over the temperature range from
 0° to 20°C the  temperature correction changes from 1.57 to 1.06 and the cold-
 start correction factor from 1.3  to  1.1 (for 30% cold starts).  This depend-
 ence on temperature suggests that any one inventory cannot completely charac-
 terize emissions.   A special  inventory is necessary for each different set of
 temperature conditions.   The result  is an inventory of emissions organized
 into separate parts with emissions for which possible temperature dependence
 has not been  quantified,  making up one subtotal and temperature-dependent
 emissions making up the other.  For  this latter part, a computer code can be
 developed to  apply temperature  and cold-start corrections for the given series
 of temperatures that describe each hour of the day chosen.  After the correc-
 tions are applied,  the  various  subtotals can be added, giving an emissions
 inventory specific  to a given day.

      For an emissions inventory in the Las Vegas Valley, the following would
 be considered:   point sources (power plants and industrial plants); area
 sources  (space  heating);  and  mobile  sources (railroads, airplanes, and auto-
 mobiles).   It was  noted  that  total mobile sources in the Las Vegas Valley
 were  responsible for 93  percent of the total emissions, with 84 percent
 attributed  to gasoline-powered vehicles.

      Much of  the point  source data required for most regions is available
 through  the National  Emissions Data  System (NEDS).  For industrial sources,
 the assumption  is  usually made that  emissions are uniform over all operating
 hours.   For power  plants,  emissions  can be estimated from data obtained on
 hourly fuel oil  and  natural gas use  from several chosen consecutive days
during the CO season, the  first of which is chosen randomly.  These values,
averaged for  each hour  and with emission factors applied, yield emissions per
hour.  Natural  gas not  supplied to power plants and industrial users can be
assumed  to be used for space  heating.  Local natural gas suppliers can provide
estimates of  residential,  commercial, and industrial sales.  Using these data,
calculated emissions  can  then be  parceled out to each grid square according to

                                      42

-------
population density.

     Aircraft emissions can be calculated for local  airports with scheduled
commercial flights from data listed in the Official  Airline Guide.  North
American Edition.  This semimonthly publication contains up-to-date airline
schedules, including times, flight numbers, and aircraft used.   Information
on unscheduled charters and light plane traffic may be available through  the
local Federal Aviation Agency.  Application of emission factors by engine
type and number of landing- takeoff cycles per hour determines the actual
emissions.

     Data on actual vehicle miles traveled (VMT) at different speeds,  on  each
type of roadway per grid square, and on diurnal traffic variations  can often
be supplied by the State Department of Highways.  Heavy duty vehicle emis-
sions are not considered to be temperature-dependent and are handled accord-
ing to procedures outlined in AP-42.

     For light duty vehicles, the equation (from EPA, 1976a, AP-42) for CO
emissions is
1000
               [Hl
               TOO
                                       1975
                                                                    (VI-1)
(VMTE LDVE    ^   c;n,  yjs, m*nl
                                      0=1963


where

      Q. DV = total carbon monoxide emissions (kg/hour/grid square)


         H = hourly percentage of daily traffic

         E = roadway types

      VMTr = vehicle miles traveled for each roadway type E


      LDVE = percent light duty vehicles for roadway type E


      c* ,  = Federal test procedure average emission factor for model  year o,
             calendar year n

      v* ,  = speed correction factor for speed s', model  year o


      m* ,  = adjusted annual travel for model year o, calendar year n1


      p T  = temperature correction factor for model year o, temperature T
                                      43

-------
     q T  , = percent cold start correction factor for model  year o,
             temperature T and percent cold starts w1.

     There are  two sets of temperature and cold-start correction factors for
 the  above equation.  One set applies to model  years 1963 through 1974 which
 do not use catalytic converters.  The other set applies to model years 1975
 and  later.  The equation then becomes
   Q  =
  1
1000
               100
(pq)
                       (p'q')
                              ^
                                                 1974
(VMT£ LDVj
                                               0=1963
                                                             (VI-2)
vos
o      '
 where

     p,q       are the correction factors for the years 1963 through 1974

     p',q'     are the correction factors for 1975

     With this equation, emissions from light duty vehicles can be calculated
 for a  reference temperature.  Adjustments for specific temperatures can be
 made during simulations, using data collected at the nearest NWS or military
 weather observing station.

 Topography

     Ground elevation information can be derived from standard U.S. Geological
 Survey topographic maps (1:62,500 scale).  A map mosaic can be prepared using
 a clear plastic grid overlay of some specified grid size oriented over the
 area using UTM coordinate designations.  Grid square elevations can be visual-
 ly estimated taking into account features of the terrain.

 Initial and Boundary Conditions

     The model program takes measured and specified air quality data at a
 number of sites throughout the modeling region and, via an interpolation/
 extrapolation procedure, computes both initial and boundary concentration dis-
 tributions.   The input data requirements for the program are:

     a.   The grid coordinates for each measurement site.

     b.   Hourly air quality data for each pollutant of interest at each
     station.

 Initial conditions are given for 1  hour prior to the hour of simulation while
boundary conditions are input for each hour of simulation.
                                      44

-------
Atmospheric Thermal Structure

     Data on the diurnal character of reference height Zj, i.e., the
atmospheric mixing height, are required at hourly intervals.   This informa-
tion is generally determined through analysis of vertical profiles of air
temperature.  Such profiles can be obtained from soundings using free or
tethered radiosondes, or from vertical spirals using fixed wing, helicopter,
or drone aircraft as measurement platforms.  Data collected using remote
sounding devices such as a LIDAR or monostatic acoustic sounder can often be
used for this purpose.  Vertical profiles of air quality, dewpoint
temperature, visual range (with an integrating nephelometer), and wind speed
and direction can often be useful as supplementary information.

     Special measurement programs may be required to ascertain the necessity
of allowing for spatial variability in mixing height across the modeling
region.  This is especially true for areas which contain large urbanized
sections or severe topographical features, or are under the direct influence
of large bodies of water.

Surface Wind Speed and Direction

     Information on the 3-dimensional wind for each of the grid cells in the
model region is required at hourly intervals.  The interpolation algorithm
described in Chapter  III provides this data from values of horizontal wind
speed and direction reported for the grid locations of each of the surface
wind stations in the  field network.

Vertical Eddy Diffusivity Requirements

     Profiles of vertical eddy diffusivity for each grid column are computed
by model algorithms (see Chapter III) from values of near surface horizontal
wind speed, roughness parameter, height of reference level (i.e., mixing
height), cloud cover  and exposure class.  The information on cloud cover may
be obtained from hourly weather observations made at the nearest NWS or mili-
tary station.  Exposure class may be determined in daytime from the sun angle
which is in turn obtained by an algorithm using latitude, longitude, time of
day, and Julian date  as input parameters.

     Values of the surface roughness parameter, z0, for inclusion into the
diffusivity scheme can be estimated using the aerodynamic technique developed
by Lettau (1969).  Basically, the technique considers the height, surface
area, and distribution of roughness elements exposed to the wind.  This tech-
nique was developed from field measurements of the aerodynamic properties and
wind profiles for various geometric configurations of roughness elements
placed on a frozen lake surface.  The explicit formulation is


                                                                     (VI-3)
                                      45

-------
 where h* is  the average  height of the roughness elements within a grid
 element; s*  is  the  silhouette area of the elements exposed to the wind; and
 S* is the lot area.   The lot area is the quotient of the total surface area
 over which the  elements  are being considered divided by the number of ele-
 ments within the area.

      Average values  of z0 can be computed for each of the horizontal area!
 grid elements of the model area within the region of interest.  Percentages
 of the land-use types within each of the grid elements can be established
 using street maps,  real  estate maps, aerial photographs, and results of
 visual  ground surveys, etc.  A value of z0 for each of the grid elements can
 then be computed by appropriate weighting of the percent coverage of land-use
 type existing within the grid elements.

      Variations in  silhouette area and hence of z0 as a function of season of
 the year and of wind direction may need to be considered.  The former is
 especially true for  areas with significant deciduous tree cover and/or agri-
 cultural activity.   The  latter is particularly important for areas with rela-
 tively heterogeneous land-use tracts and the above-mentioned tree cover.

 Scenario Selection

      Selection  of optimal locations for CO monitoring stations by the Figure
 of Merit technique  is accomplished using simulations with the mesoscale air
 quality simulation model  as a data base.  The simulations can be run on mete-
 orological scenarios developed from historical weather data supplemented by
 historical air  quality data and current meteorological information collected
 during the field sampling program.

      Analysis of historical air quality data in relation to the prevailing
 meteorological  situation or pattern should provide definitive information on
 the types of such situations likely to produce high pollutant concentrations.
 This is particularly important since the function of the CO monitoring net-
 work is for  monitoring to detect concentrations exceeding the NAAQS.

      Classification ,of meteorological situations into types may be
 accomplished through  statistical analysis of sea level pressure charts and
 upper air constant pressure charts applicable for the area as available from
 the NCC of the  NWS.   Objective classification may be accomplished in the
 manner  outlined  by such  investigators as Lund (1963) and Roach and MacDonald
 (1975).   Basically,  this  is done for a large sample size, on the order of 5
 to  10 years  of data,  by  linear correlation of the heights on a constant pres-
 sure surface (or  pressures on a constant height surface) at regular grid
 intervals.   The maps  correlating highest with each other are grouped together
 and  form classes.  The percentage frequency of each class establishes its
 probability  of occurrence.  Alternatively, visual examination of such charts
 may  allow a  subjective classification of meteorological situations with the
 frequency of occurrence of each establishing its probability of occurrence.

      Meteorological  information on winds, surface air temperature, mixing
 depths,  and  atmospheric  stability for input into the model for each such class
may  be  obtained from  historical surface and upper air data for reporting

                                      46

-------
stations within the region.  Where the appropriate stations are not available
locally, selective use may be made of data for stations normally within the
same air mass.  Diurnal mixing depths, for instance, may be estimated from
the 1200Z radiosonde profile and hourly surface air temperature values in the
manner of Holzworth (1972).

     Statistical analysis of the resultant meteorological data for the
particular dates corresponding to the various classes can lead to the selec-
tion of the scenarios to be utilized.  For example, averages and/or percen-
tiles of frequency distributions of the various parameters may be selected.
Alternatively, typical cases for subsets within the basic classes may be
chosen.

     In regions containing significant topographic features, under the direct
influence of  large bodies of water or containing large urbanized areas, it
may well be necessary to use supplementary weather data to assist in estab-
lishing meteorological input for the model.  Here, large spatial variations
in certain of the parameters may persist across the region.  Objective analy-
sis or other  statistical models may be used to develop fields of such varia-
bles based upon the historical and supplementary data.  In areas where the
meteorology for the relevant scenarios is determined almost exclusively by
local effects due to large bodies of water or complex topography, it may even
be necessary  to use meteorological simulation models to develop the relevant
inputs.
                                       47

-------
                                 CHAPTER VII

                                   SUMMARY


     This document has presented a methodology for designing an air
monitoring network which has as its objective the identification of pollutant
concentrations that exceed the NAAQS.  Emphasis has been placed on the theo-
retical development of mathematical expressions which describe the behavior
of pollutants that have been emitted in an urban environment.  Computer
solution of these equations under specified meteorological conditions form
the data base from which predicted pollutant concentrations can be mapped,
which in turn provide the bases for the selection of monitoring sites for a
network.

     A procedure has been described for applying this methodology to a design
for selecting carbon monoxide monitoring sites.  Application of the methodol-
ogy requires the establishment of a field measurement program, the develop-
ment of an appropriate historical data base, the assembly of a complete emis-
sions inventory, and the development of other pertinent data required as
input to the solution of the mathematical model.

     This document contains only the rationale behind the air monitoring
network design methodology and a procedure for applying this methodology.
The instructions for implementing and executing a computer solution of the
mathematical model is published separately as a User's Guide.  A complete
example of the application of the methodology including details of an executed
field program, emissions inventory development, meteorological scenarios
development, simulation results, and recommended network design for the
Las Vegas Valley are presented in a separate publication.
                                      48

-------
                                 REFERENCES


Barth, D. (1970), "Federal Motor Vehicle Emission Goals for CO, HC, and NO ,
     Based on Desired Air Quality Levels," Jour. Air Poll. Cont. Assoc.,  x
     Vol. 20, p. 519.

Businger, J. A., J. C. Wyngaard, Y. Izumi, and E. F. Bradley (1971), "Flux-
     Profile Relationships in the Atmospheric Surface Layer," J. Atmos. Sci.,
     Vol. 28, pp. 181-189.                                         "~~

Charlson, R. J.  (1969), "Note on the Design and Location of Air Sampling
     Devices," Jour. Air Poll. Cont. Assoc., Vol. 19 (10), p. 802.

Corrsin, S.  (1974), "Limitations of Gradient Transport Models in Random Walks
     and in Turbulence," Adv. in Geophys., Vol. 18A, pp. 25-60.

Darby, W. P., P. J. Ossenbruggen, and C. J. Gregory (1974), "Optimization of
     Urban Air Monitoring Network," J. Environ. Energy. Vol. EE 3, pp. 557-591.

Deardorff, J. W. (1970), "A Three-Dimensional Numerical Investigation of the
     Idealized Planetary Boundary Layer," Geophys. Fluid Dyn., Vol. 1,
     pp. 377-410.

Deardorff, J. W. (1972), "Numerical Investigation of Neutral  and Unstable
     Planetary Boundary Layers," J. Atmos. Sci., Vol.  29, pp. 91-115.

DeMarrais, G. A. (1959), "Wind Speed Profiles at Brookhaven National Labora-
     tory," J. Applied Meteor., Vol. 16, p. 181.

Duckworth, S., and E. Kupchanko (1967), "Air Analysis:  The Standard Dosage
     Area Product," Jour.  Air Poll. Cont. Assoc.. Vol.  17, p. 379.

Golder, D. (1972), "Relations Among Stability Parameters in the Surface
     Layer," Boundary-Layer Meteor., Vol. 3, pp. 47-58.

Harlow, F. H., and J. E. Welch (1965), "Numerical Calculation of Time
     Dependent Viscous  Incompressible Flow of Fluids with Free Surfaces,"
     Physics of Fluids. Vol.  3, pp. 2182-2159.

Holzworth, G. C. (1972), "Mixing Depths, Wind Speeds, and Potential for Urban
     Air Pollution Throughout the Contiguous United States," U.S.  Environmen-
     tal  Protection Agency, AP-101.
                                      49

-------
 Hotchkiss, R. S., and 0. S. Hirt (1972), "Particulate Transport in Highly
     Disturbed Time Dimensional Flow Fields," LP-DC-72-364, Los Alamos
     Laboratories, Los Alamos, New Mexico.

 Ingram, W. T., and J. Golden  (1969), "Mobilized Monitoring," Interim Report,
     School of Engineering and Science, New York University, University
     Heights, Bronx, New York.

 Lamb, R. G., and J. H. Seinfeld (1973), "Mathematical Modeling of Urban Air
     Pollution-Generalized Theory," Environmental Science and Technology,
     Vol. 7, #3, pp. 253-261.

 Lettau, H. H. (1969), "Note on Aero-Dynamic Roughness-Parameter Estimation on
     the Basis of Roughness-Element Description," J. Appl. Meteor., Vol. 8,
     pp. 828-832.

 Liu, M. K., D. C. Whitney, S. D. Reynolds, and P. M. Roth (1973), "Automation
     of Meteorological and Air Quality Data for the SAI Urban Airshed Model,"
     Report R73-32, Systems Applications, Inc., San Rafael, California.

 Liu, M. K., D. C. Whitney, J. H. Seinfeld, and P. M. Roth (1975), "Continued
     Research Mesoscale Air Pollution Simulation Modeling: Analysis of Model
     Validity and Sensitivity and Assessment of Prior Evaluation Studies,"
     prepared for U.S Environmental Protection Agency (Contract 68-02-1237),
     Systems Applications, Inc., San Rafael, California.

 Ludwig, F. L., and J. H. S. Kealoha (1975), "Selecting Sites for Carbon Mon-
     oxide Monitoring," Final Report on EPA Contract 68-02-1471, Stanford
     Research Institute, Menlo Park, California.

 Lund, I. A. (1963), "Map-Pattern Classification by Statistical Techniques,"
     J. Appl. Meteor., Vol. 2, pp. 56-65.

McElroy, J. L. (1969), "A Comparative Study of Urban and Rural Dispersion,",,
     J. Appl. Meteor., Vol. 8, p. 19.

Morgan, G. B., G. Ozolins, and E. C. Tabor (1970), "Air Pollution Surveil-
     lance Systems," Science, Vol. 170, p. 289.

Myrup, L.  0., and A. J. Ranzieri (1975), "A Unified Approach to the Estimation
     of Diffusivity in the Planetary Boundary Layer," California Department
     of Transportation Report.

O'Brien, J.  J.  (1970), "On the Vertical Structure of the Eddy Exchange
     Coefficient in the Planetary Boundary Layer," J. Atmos. Sci., Vol 27,
     pp. 1213-1215.

Ranzieri,A.  J.,  and C. E.  Ward (1975), "Caline Z—An Improved Microscale Model
     for the Diffusion of Pollutants from a Line Source," California Depart-
     ment  of Transportation,  Transportation Laboratory, Sacramento,
     California,  presented at the Air Quality Workshop in Washington, D.C.
                                      50

-------
Reynolds, S. D., P. M. Roth and J. H. Seinfeld  (1973), "Mathematical Modeling
     of Photochemical Air Pollution, Part I-Formulation of the Model,"
     Atmos.  Environ., Vol.  7,  pp.  1033-1061.

Reynolds, S. D., M. K. Liu, T. A.  Hecht, P. M.  Roth, and J. H. Seinfeld
     (1974), "Mathematical Modeling of Photochemical Air Pollution—III.
     Evaluation of the Model," Atmos. Environ.. Vol. 8, pp. 563-596.

Roach, G. E., and A. E. MacDonald  (1975), "Map-Type Precipitation Probabili-
     ties for the Western Region," U.S. Department of Commerce, NOAA, NWS,
     COM-75-10428.

Roth, P. M., P. J. W. Roberts, M.  K. Liu, S. D. Reynolds, and J. H. Seinfeld
     (1974), "Mathematical Modeling of Photochemical Air Pollution—II.  A
     Model and  Inventory of Pollutant Emissions," Atmos. Environ., Vol. 8,
     pp. 97-130.

Seinfeld, J. H. (1972), "Optimal Location of Pollutant Monitoring Stations in
     an Airshed," Atmos. Environ., Vol. 6, p. 847.

Sklarew, R.  C., A. J. Fabrick, and J. E. Rager  (1971), "A Particle in Cell
     Method  for Numerical Solution of the Atmosphere Diffusion Equations and
     Applications to Air Pollution Problems," Final Report No. 3SR-844,
     EPA Contract 68-02-0006,  Systems, Science  and Software, La Jolla,
     California.

Sperling, R. B. (1975), "Evaluation of the Correlation Spectrometer as an
     Area S02 Monitor," Environmental Measurements, Inc., EPA-600/2-76-077.

Turner, D. B. (1969), "Workbook of Atmospheric  Dispersion Estimates," Public
     Health  Service Publication No. 999-AP-26.

U.S. Environmental Protection  Agency (1970), "Air Quality Surveillance,"
     Working papers distributed to participants in the Regional Implemen-
     tation  Workshop.

U.S. Environmental Protection  Agency (1971), "Guidelines: Air Quality
     Surveillance Networks," AP-98.

U.S. Environmental Protection  Agency (1973), "Guide for Compiling a Compre-
     hensive Emissions Inventory," APTD-1135.

U.S. Environmental Protection  Agency (1975), "Guidance for Air Quality
     Monitoring Network Design and Instrument Siting," Office of Air Quality
     Planning and Standards, Number 1.2-012, Research Triangle Park,
     North Carolina.

U.S. Environmental Protection  Agency (1976a), "Compilation of Air Pollutant
     Emissions Factors," AP-42, and Supplements 1 through 5, Second Edition.

U.S. Environmental Protection Agency (1976b), "Quality Assurance Handbook  for
     Air Pollution Measurement Systems," EPA-600/9-76-005.


                                      51

-------
U.S. Environmental  Protection Agency (1977), "User's Guide  to  the  SAI Urban
     Airshed Model,"  in  preparation.

Yamada, V. M.  (1970),  "Current Practices in Siting and Physical  Design of
     Continuous Air Monitoring Stations," Jour. Air Poll. Cont.  Assoc^,
     Vol. 20,  p. 209.                      '

Yamada, V. M., and  R.  J.  Charlson (1969), "Proper Siting of the  Sampling
     Inlet Line for a  Continuous Air Monitoring Station,"   Environmental  Sci-
     ence and  Technology, Vol. 3, p. 483.

Yanenko, N. N.  (1971), The Method of Fractional Steps, Springer-Verlag,
     New York, New  York.
                                       52         *U.S. Government Printing Off ice: 1977-784-677/84 Region No. 9-1

-------
                                   TECHNICAL REPORT DATA
                            (Please read Instructions on the reverse before completing)
 . REPORT NO.
  EPA-600/4-77-019
                             2.
             3. RECIPIENT'S ACCESSION-NO.
 4. TITLE AND SUBTITLE
   DEVELOPMENT OF A METHODOLOGY FOR DESIGNING  CARBON
   MONOXIDE  MONITORING NETWORKS
                                                          5. REPORT DATE
                                                           March 1977
             6. PERFORMING ORGANIZATION CODE
 7.AUTHOR(S)M>  K>  Liu> j^ „     ^ R>         j   i   >      j
 J.  H.  Seinfeld,  J.  V. Behar, L. M. Dunn, J.  L. McElroy,
 P.  N.  Lem, A.M.  Pitchford. N. T. Fisher
             8. PERFORMING ORGANIZATION REPORT NO.
 9. PERFORMING ORGANIZATION NAME AND ADDRESS

 Systems  Applications, Incorporated
 950  Northgate Drive
 San  Rafael,  California  94903
                                                           10. PROGRAM ELEMENT NO.
              1HD620
             11. CONTRACT/GRANT NO.


              68-03-2399
 12. SPONSORING AGENCY NAME AND ADDRESS
 U.S. Environmental Protection Agency-Las Vegas, NV
 Office  of Research and Development
 Environmental  Monitoring and Support  Laboratory
 Las Vegas,  NV  89114
             13. TYPE OF REPORT AND PERIOD COVERED
             	Interim Report	
             14. SPONSORING AGENCY CODE
              EPA/600/07
 15. SUPPLEMENTARY NOTES
 This report was jointly authored by  personnel  of the Systems Applications, Incorporated
 and the Environmental Monitoring and  Support  Laboratory,  Las Vegas, Nevada.
 16. ABSTRACT
      A  methodology is presented for  designing  a  carbon monoxide monitoring network
 based on  the objective of identifying concentrations that exceed the national ambient
 air quality standards (NAAQS).  The  basis  for  identifying concentrations in excess of
 the NAAQS is the Concentration Area  Time-Product,  where the concentrations are inte-
 grated  over an area (i.e., a grid square in  a  gridded system) and integrated over a
 time interval  for averaging the concentrations.   These are computed with a mesoscale
 air quality simulation model formulated as a 4-dimensional (x,y,z,t), partial differ-
 ential  equation of mass balance for  the pollutant  species which yields space-time
 average concentration  distributions.  A frequency-weighted average of concentrations
 called  Figure of Merit is determined from  these  projected concentration distributions,
 prevailing  meteorological patterns,  and the  frequency of occurrence associated with
 each of the meteorological patterns.  A mapping  of these Figure of Merit values
 provides  the basis of selection of the locations and number of sites in the network.
      The  methodology was applied in  a design of  an ambient air monitoring network for
 carbon  monoxide.   The establishment  of a field measurement program is described which
 would provide air quality and meteorological data  for model validation and simulation
 as required in development of the specifications for the number and location of sites
 in the  network design.  Discussions  are limited  to the design methodology.  Actual
 field data,  simulation exercises, pollution  concentration isopleths, and mappings are
           in  d  sapdrdlti
                                   WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                              b.IDENTIFIERS/OPEN ENDED TERMS
                          c.  COS AT I Field/Group
Air Pollution
 Mathematical Models
 Systems Analysis
 Air Monitoring Networks
 Air Quality Monitoring
 Monitoring Sites
  05H
  13B
  14A
 3. DISTRIBUTION STATEMENT


         RELEASE  TO PUBLIC
19. SECURITY CLASS (ThisReport)

    UNCLASSIFIED
21. NO. OF PAGES

     64
20. SECURITY CLASS (Thispage)

    UNCLASSIFIED
2,2. PRICE
EPA Form 2220-1 (9-73)

-------