EPA-450/4-74-013
SEPTEMBER 1974
(OAQPS NO. 1.2-031)
     GUIDELINES FOR AIR QUALITY
MAINTENANCE PLANNING AND ANALYSIS
        VOLUME 12 : APPLYING
  ATMOSPHERIC SIMULATION MODELS
 TO AIR QUALITY MAINTENANCE AREAS

-------
                                      EPA-450/4-74-013

                                   (OAQPS ISO. 1.2-031)
      GUIDELINES FOR AIR QUALITY

MAINTENANCE  PLANNING AND  ANALYSIS

          VOLUME  12  :  APPLYING

   ATMOSPHERIC  SIMULATION MODELS

 TO  AIR QUALITY MAINTENANCE  AREAS
                ENVIRONMENTAL PROTECTION AGENCY
               Office of Air and Waste Management
            Office of Air Quality Planning and Standards
            Research Triangle Park, North Carolina 27711
                     September 1974

-------
                     OAQPS GUIDELINE SERIES

Tho guideline series of reports is being issued by the Office of Air Quality
I'ir-nning arid Standards (OAQPS) to provide information to state  and local
;»i  nolluiion control agencies; for example, to provide guidance  on the
.-> quisition and processing of air quality data and on  the planning and
analysis requisite for the maintenance of air quality.  Reports published in
•Hs series will  be available - as  supplies permit - rrom the Air Pollution
">. "rhiiifdi Information Center, Research Triangle Park, North Carolina
/:'-!!; or.  for a nominal fee, from the National Technical Information
Service, 5285 Port Royal Road, Springfield, Virginia 22151.
                    Publication No. EPA-"50/4-74- 013
                     (OAQPS Guideline No. 1.2-031)
                                     11

-------
                            FOREWORD

    This document is the twelfth in a series comprising Guidelines for Air
Quality Maintenance Planning and Analysis.  The intent of the series is to
provide State and local agencies with information and guidance for the prepa-
ration of Air Quality  Maintenance Plans required under 40 CFR 51. The volumes
in this series are:

    Volume 1:    Designation of Air Quality Maintenance Areas
    Volume 2j_   Plan Preparation
    Volume 3_:_   Control Strategies
    Volume 4:    Land Use and Transportation Consideration
    Volume 5:    Case Studies in Plan Development
    Volume 6:    Overview of Air Quality Maintenance Area Analysis
    Volume 7j_   Projecting County Emissions
    Volume 8:    Computer-Assisted Area Source Emissions  Cridding
                Procedure
    Volume 9j_   Evaluating Indirect Sources
    Volume 10:   Reviewing New Stationary Sources
    Volume ljj_   Air  Quality Monitoring and  Data~^Analysis
    Volume 12:   Applying Atmospheric Simulation Models to Air Quality
                Maintenance Areas

    Additional volumes may be  issued.

    All references to 40 CFR Part 51 in this document are to  the regulations
as amended through July 1974.
                                 NOTE

      This  guideline  is being released in its present form in order to
 allow its  immediate  use by State and local agencies.  This guideline
 may  be reissued  in the near future in order to incorporate comments
 and  suggested  improvements offered by the EPA Regional Offices and by
 State and  local  agencies and other concerned groups.
                                    ill

-------
       VOLUME 12.  APPLYING ATMOSPHERIC SIMULATION MORELS TO
                  AIR QUALITY MAINTENANCE AREAS

                               PREFACE         .  '

     Volumes 7, 8 and 13 discuss methods v;hich can be  used to
estimate future county-wide emissions and to allocate  these
emission projections within a county.  Volume 11  presents guidance
for locating representative air duality monitoring stations and
for interpreting the resulting data.  The information  contained
in this Volume 12 is intended to provide insight  into  how the  emission
and air quality information thus obtained might be combined to estimate
air quality levels in future years.   The vehicle  for estimating  future
air quality ib cue atmospheric simulation model.
     This guideline is not intended  as a critical  review of atmospheric
simulation models, but only as a guide to the types of models, associated
data requirements, and levels of detail which are available for  use  in
developing control strategies.  Key  simulation model parameters  and  the
data needed to estimate these are first identified. Next, three
tateooHes of models are distinguished in accordance with the  amount
and detail of emissions data which are required to use the model.
Finally, examples of models from each category are identified  and
*r   ioeti in terms of their applicability to different pollutants,
averaging times, emissions and meteorological data requirements, use,
reliability and air quality maintenance analysis.
                                 IV

-------
     The models described in this volume are those which are likely
to be the most available to air pollution control  agencies.   Alternate
models may be used provided available data are consistent with the
model's requirements and the model is appropriate for the pollutant
and sampling time being analyzed.

-------
                             CONTENTS
                                                                Page

   I.   Introduction	   1

  II.   Simulation Model  Parameters	   2

       A.   Pollutant Types and Averaging Times	   2

       B.   Emission Inventory Requirements	   3

       C.   Meteorological  Data	   4a

 III.   Types of Simulation Models .  	   6

  IV.   Applicability of Models	   8

   V.   Background Concentrations	11

  VI.   Model Validation/Calibration Procedures	11

 VII.   Summary	14

VIII.   Caution	14

  IX.   Appendix:  Model  Descriptions	17

       Rollback Model 	  18
       Appendix J HC-0X Relationship	21
       Miller-Holzworth Model 	  23
       Hanna-Gifford Model	25
       HIWAY Model.	28
       Air Quality Display Model	31
       Sampled Chronological Input Model	  34
       APRAC-1A Model 	  36
       SAI Photochemical Model.  ,	38

   X.   References	39

Bibliographical Data Sheet	43
                                 VI

-------
                              TABLES

Table.!.  Summary of Simulation Model Characteristics

Table 2.  Models Applicable to Specific Pollutants and
            Averaging Times
                             FIGURES

Figure 1.  Dispersion Node! Flow Diagram

Figure 2.  Flow Diagram for the Development of a Control Strategy.
                                vii

-------

-------
             APPLYING ATf'OSPHERIC SIMULATION MODELS
               TO AIR DUALITY MAINTENANCE AREAS

I.  INTRODUCTION
    The development of an air quality maintenance plan is  strongly
dependent on en adequate methodology for relating pollutant  emissions
to ambient air quality.  The methodology is required so that con-
clusions can be reached regarding (1) the effect of anticipated growth
in an area on air quality during some specified future period and  (2)
the ability of selected control strategies to maintain air quality
standards throughout this period.  The most commonly used  tool  for
relating emissions and air quality is an atmospheric simulation or
dispersion model.
    An atmospheric simulation model can be defined as a mathematical
description of the transport, dispersion and transformation  processes
that occur in the atmosphere.  In its simplest form, such  a  model
relates pollutant concentrations (x) to pollutant emission rates
(Q) and a background concentration (b),
          x = kQ+b                                           (1)
The constant k is a function of atmospheric conditions and the
spatial relationship between source and receptor.  Atmospheric
simulation models are ultimately concerned with the variability of
k, and of emission rate and their impact on pollutant concentrations.
     This review is intended only as a guide to the types  of models,
associated data requirements, and levels of detail which are available  for
use in developing control strategies.  It is not intended  as a critical
review of atmospheric simulation models.

-------
 II.   SIMULATION MODEL PARAMETERS
      The type of simulation  model v/hich  is  used to examine a specific
 growth situation or maintenance plan  depends on a number of parameters.
 These parameters include  the pollutant and  averaging times under
 consideration,  the  detail  of the emissions  inventory, and available
 meteorological  data.
      A.   Pollutant  Types  and Averaging Times
          The  types  of pollutants that are commonly considered in
 dispersion model  applications are:  (1) S02 and total suspended particu-
 lates  (TSP) as  pollutants  emitted primarily by stationary sources, (2)
 CO and fine particulate matter such as that emitted from mobile sources,
 (3) pollutants  such as HC  and NO  which are emitted from both stationary
                                /\
 and point sources and  react  in the atmosphere to form oxidants.
     For purposes "f  discussion here, CO and particulate matter are
 treated as pollutants which  do not transform or decay in the atmosphere.
 However HC and N0x are involved in numerous chemical  reactions  in the
 atmosphere; these reactions  can be properly accounted for only  by highly
 sophisticated, complex models.  S02 1s known to decay in the atmosphere
 and to form other sulfur compounds.   However such  chemical  changes are
not well documented.  Some models  attempt to account  for these  changes by
ihe use of a half-life term;  usually half-life  is  assumed to be on the
order of a few hours.  Some models  completely ignore  such a  decay.  It
should be noted that N02,  emitted  as a pollutant  from an isolated point
source in a non-reactive atmosphere, may  be treated  in  a manner similar to
S0?;  this should be  done with great  care.

-------
          The averaging times for which dispersion models must be  used  vary
 with the National Ambient Air Quality Standards  applicabl e  to those
 pollutants.  Dispersion models which allow concentration estimates  to  be
 made for the following combinations of pollutants and averaging times  are
 needed:  S02 and particulate matter for 3-hour (S02  only), 24-hour and  annual
 averages; CO for one-hour end eight-hour averages; oxidants  for one-hour
 averages; and N02 for annual  averages.   Dispersion models  estimate  con-
 centrations for a one-hour period or for seasonal  or annual  averages.   If
 an  average concentration  for a period of intermediate length  (e.g., 3-, 8-,
 24-hour is required,  two  options  are available.   One, the  model can be  used
 to  estimate concentrations  hour-by-hour  for  the  period under  consideration
 and an  average  of all  hours taken;  this  is  the  preferred  method.   Two,
 statistical  techniques  suggested  by Larsen   for  urban areas or empirical
                               o
 techniques suggested  by Turner and others for point sources  can  be used
 to  convert a  concentration  for one  averaging time to a  concentration for
 longer  or shorter averaging times.
          Emissions  Inventory  Requirements
          The  detail of  the  emissions  inventory is a  determining factor
of  the  dispersion model used  to estimate concentrations.  Pollutant
sources generally can be  separated  into point sources,  Hne sources and
area sources.  Point sources are defined as those which have emissions
of a pollutant greater  than some limit, such as 100 tons per year.  Any
space heating, process or incineration source which emits this amount
of pollutants or more is considered a point source.  Line sources  are
generally confined to roadways and streets along  which there are
                                  3

-------
well defined straight movements of motor vehicles  and  their  emissions.
Area sources include the multitude of minor sources  with  small  emissions
which are impractical to consider as  individual  point or line  sources.
Such area sources are typically treated as  a grid  network of square
areas, with pollutant emissions distributed uniformly  within each  grid
square.  Commonly, emissions from such sources  are prorated  over areas
of one to IOC square kilometers, depending  on the  available  detail of
emissions and desired refinement of modelling results.  In cases where
there are no data on the spatial distribution of emissions,  the emissions
inventory may be limited to one average emission for the  entire area
of concern.
         For a point source the following minimum data are required  as
input to a dispersion model: pollutants emittedjaverage emission  rates
(some identi-Mcatlr-r, of shorter term variations may be desirable);
physical stack parameters such as stack height, stack diameter, stack
exit velocity and stack exit temperature; and location of the source in
appropriate grid coordinates.  Desirably, a point source emissions
inventory should include these data for all sources with emissions greater
than 100 tons per year.  However a less  precise  estimate can be  made  using
only data from larger sources.
         Typically for line sources, data are required on the width  of the
roadway and its center strip, the types and amounts (grams per second
per meter) of pollutant emissions, the number of lanes and the emissions
from each lane and the height of emissions. The location of the ends of the
straight roadway segments must be specified in appropriate grid coordinates.
                                  4

-------
          Area source information required are  types  and  amounts  of
 pollutant emissions, the physi.cal  size  of the  area over  v/hich  em'ssions
 are prorated, a representative average  stack height  for  the area sources,
 the location of the centroid or the  southwest  corner of  the source area
 in appropriate coordinates.   For circumstances where only  total  area-
 wide emissions ere  available,  supplementary information  on the dimensions
 of the  area  are required.
          It  should  be  noted  that the emissions inventory available
 for use with most dispersion models  is  representative of annual average
 pollutant  emissions.   A  level  of specification greater than this is
 generally  not available, simply  due to  e  lack of data.  The model used
 to estimate  pollutant  concentrations cannot be contingent upon whether
 or not an  hourly emissions inventory is available.   In some cases
 factors which  specify  the diurnal variations of certain source classes
 are  built  into  a model.  However unless otherwise identified it will
 be assumed in this  review that the same inventory is  used to estimate
 hourly as well  as annual average concentrations.
     C.   Meteorological Data
         Meteorological data required to describe transport and
dispersion in the atmosphere are wind direction,  wind speed,  atmospheric
stability and mixing height.   Wind direction determines  the direction
                                 4a

-------

-------
 of movement of the plume.   Wind speed affects  the  initial  dilution
 of the pollutant as it is  emitted from the  stack.   The  atmospheric
 stability determines the rate of dilution as the plume  moves  downwind.
 Mixing height determines the  depth of the atmosphere  through  which
 pollutants can be dispersed in the vertical.   These parameters are
 routinely measured by National Weather Service (NWS)  stations.
 They are available both  as  individual  observations  and  in  summarized
 form from the National Climatic Center in Asheville,  North Carolina.
 Other sources of similar data may be  local  universities, industrial
 companies, pollution control  agencies  and consultants.  However, data
 from these sources are likely to be less comprehensive  in  scope than
 that from the NWS.   Simple  dispersion  models require, as a minimum, data
 on wind  speed.   The  more sophisticated models  require much more detailed
 data on  the  temporal  and spatial  variations of the  above meteorological
 parameters.
          For a dispersion model  to be  useful and valid  the meteorological
 data  input to the  model  must  be  representative of the transport and
 dispersion conditions which the model  is attempting to  simulate.   The
 representativeness of the data  is dependent on (1)   the  proximity of the
meteorological monitoring site  to the area under consideration, (2) the
complexity of the  terrain in the area, and (3)  the  immediate surroundings
of the monitoring  site.  The representativeness of  the data can be
adversely  affected by distance betv/een the source areas and receptors
of interest, valley-mountain terrain,  presence  of large bodies of water
and urban-rural differences.
                                  5

-------
         The r-ost representative meteorological  monitoring  site(s)
should be selected.  For the models to be considered  in  these
guidelines (except for the SAI model)  only one representative,  site  is
required.  Hov;ever, in an area encompassing widely differing meteorological
regimes attention should be given to applying  the  model  with more than
one set of meteorological data.  Meteorological  consultation for the
application of rodels is required in this circumstance,  and is highly
desirable in any case.
         The detail of the emissions and meteorological  data will
determine which dispersion models can be used  to estimate concentrations.
For example, given an urban area for which (1) all ooint/line  sources have
been identified, (2) all other emissions are accounted for as  area
sources and (3) all meteorological parameters  are  available on an hourly
basis, specific detailed dispersion models can be  used.   However, if  it
is not possu'iv. to define the locations and existence of point sources,
it becomes necessary to estimate emissions and distribute these as  area
sources without regard to specific site Ideations.  Thus a dispersion
model requiring less detailed emissions data may be necessary.  In  cases
where only the crudest area-wide emissions are available, very elementary
models must be used.  Obviously, the greater detail with which a model
considers spatial and temporal variations in emissions and meteorological
conditions, the greater its ability to distinguish between the effects
of various air ouality control strategies.
      Atmospheric simulation models can be categorized into three general
groupings.  They are:  (1) models requiring only total area-wide emission
data, (2) models requiring specific information about point/line and area
                                6

-------
sources and  (3) models requiring area source emissions allocated on a
sub-county basis.
         The first category encompasses models which are limited to
consideration of only total area-wide emissions.  There is no identification
of individual point sources or specific area sources.   The meteorological
input, if required at all, is in terms of very general parameters such as
average wind speed and average rrixinq height,  Concentrations estimated
with such models are either averages for the whole area or are site
specific and apply only where there are air duality data.  These models
include the Rollback Model;3 the 40 CFR Part 51, Appendix J HC-0
                                                                A
             4                                5
Relationship;  and the Miller-Holzworth Model.   These models have limited
usefulness for evaluating individual air quality control strategies.   They
can only determine the impact of total pollutant emissions on air quality.
They cannot consider in any way how those emissions or the resulting
air quality levels are spatially distributed across an area.
         The second class of models is that which considers in detail
point/line source and area source emissions of pollutants.  These models
have detailed requirements for meteorological inputs and consider complex
atmospheric mechanisms for estimating the downwind transport, dispersion  and
transformation of pollutants.   These models can be used to estimate con-
centrations at any site in an area for which estimates are desired.  This
type of model includes the Air Quality Display Model '  the Sampled
Chronological Input Model7 and the APPAC-1A Model.9'10'11  For cases  where

-------
there is a lack of detailed point/line source data,  the  input  to  these
models can be limited to area source emissions,  wherein  all  point source
emissions are summed in the area sources.   It should be  noted  that such
a summation, should it be necessary, will  considerably detract from the
reliability of the dispersion models.
         The last class of dispersion model  is that  which  considers
only area source emissions.  This includes the SAI Photochemical
Simulation Model12 and the Hanna-Gifford Model.13'14  The  SAI  model
is a sophisticated photochemical dispersion  model which  does not
specifically consider point sources.  The  Hanna-Gifford  model  is  a
simplified area source model for stable pollutants.  While  the
latter model does not consider point/line  sources, the impact  of  point
or line sources can be individually determined by the application of a
point source m r!e1  ''  or a line source model such  as HIWAY.   '
Both the SAI and the Hanna-Gifford models  allow concentrations to be
estimated for a receptor in any of the designated source areas; however
only one representative air quality value  for each area  can  be determined.
IV.   APPLICABILITY OF MODELS
     For those areas with very poor information  on the spatial
distribution of pollutant emissions the application  of the first  type
of simulation model (e.g., Rollback, Appendix J, Miller-Holzworth) is
suggested.  Where possible the Miller-Holzworth  model is preferred for
estimating area-wide average concentrations  of SOp and TSP.  Where oxidant
concentrations must be estimated, either the Appendix J  or rollback
                                   8

-------
 approach may be necessary; those models allow reduction  in  HC  emissions
 to be determined directly from the required reduction  in 0   concentrations.
                                                          /v
 Rollback should be used for estimating CD concentrations in accordance
 with procedures described in Volume 1.
      In those areas where there is detailed  information on pollutant
 emissions (current emissions and projected emissions to  1985),  the AQDM,
 SCIM, APRAC-1A or SAI  models may be used, depending on  the averaging
 times and the pollutants to be considered.   However, it  should  be noted
 that the SCIK and SAI  models are currently in  a developmental and debugging
 phase; they are not now available for  general  distribution  as computer
 programs.   In those cases where the pollutant  emissions  projected to
 1985 can only be  known on an area-source  basis, it  is  recommended that
 either the Hanna-Gifford or the AQDM be used with just area  sources.
          The data  requirements, model  outputs  and general performance
 of these various models  are summarized  in  Table 1.  The  summary is
 structured from the most elementary to  the most sophisticated models.
 In the  table,  "Vs" indicate factors which are the most  general or the
 easiest  to work with.   Increasing  numbers  indicate factors which are more
 detailed or difficult  to work  with.  The Appendix to this guideline (1)
 examines the models  listed  in  Table 1 in greater detail, (2) discusses
 the emissions and meteorological requirements to operate these models, and
 (3) identifies the  availability and reliability of these models.  For
easy reference, Table 2  indicates  the models which are  applicable to specific
pollutants  and averaging  times.  The models are listed  according to level
of detail  and applicability.
                                  9

-------
                                                              Table  I
                                            Sunnary of Simulation  f'odel  Characteristics
Pollutant
Model St"-_.:ifi-
Nai'ic Cdl'C'i
Rollback A:iy
Appendix J QX
Miller- SO?,TSP
Holzv/orth
Hanna- SO,,,TSP
Gifford CO
Hanna-
Gifford S)^,T£P
W. Point SOUIL"
model jJ~,7S?
w. HIWAY CO
AQDM SO?,TSP
SCIM* SJ2,TbP
APRAC-1A CO
5AI* • CO,N02,0X
Averaging
Tine
Snscifi- emission
cation ujta
Any
1 Hour
1 Hour,
Annuul
Annual


1-24 Hour

1-24 Hour
1-24 Hour
Annual
1-24 Hour
1-24 Hour
1-10 Hour
1
1

1

1

2

3
3
3
3
3
2
Keteor-
oloqical
p.it.1
•— ' 	 ' 	 * 	
1
1

3

2

5

5
5
4
5
5
5
Concen-
tration
tstimtcs
3
3

3

3

2

1
1
1
1
1
2
Ease of
Use
1
1

1

1

2

2
2
3
3
3
3 '
Avail-
abillty
1
1

1

1

1

2
2
2
3
2
3
— : 	 . j t _ i i
Reli-
ability
3
3

1

1

1

1
1
1
2
2
2
e 	 — 	
Appl icabili ty
to AQ'1
3
3

3

3

2

1
1
1
1
1
2
_ l ,,.—.,,— . 	
       *1t:ese models, arc currently in a dpveloorental and debugging phase"; ti.ty are not available for general
        distribution as cc-nputer programs.
                                                         Key to Table 1
                                       ,;o,)
                                                   t.   Concentration  Estimates
                                                       1.   Estimates  at  any specified point
                                                       2.   One estimate  for each area source grid
                                                       3.   One estimate  applicable to entire ACTA
                                                   F.   Ease of Use
                                                       1.   Slide-rule
                                                       2."  Small computer effort
                                                       3.   Major computer effort
                                                   G.   Availability
                                                       1.   Open literature
                                                       2.   National Technical Information Service
                                                       3.   EPA, upon request
                                                   H.   Reliability
                                                       1.   Can be verified and  calibrated
                                                       2.   Verification 1s Incomplete, possibility of callbrat
                                                           1s uncertain
                                                       3.   Questionable; acceptable  for crude  estimates  only
var1at.1ors of wind direction, wind speed, stability'-   Applicability to AQM
,;    h, u't                                             ^-   Can distinguish letwcen specific source  anJ  lend  i,
                                                       2.   Can distinguish between land use types  only
                                                       3.   Considers no distinction  between sources  or  land  u1
A.  Pollutant Specification
    Any pollutant
    Specific PolTutap.ts  (V>n.  -
B.  JveTM(j1r.]-t1me Specification
    t.  i  iveraging-tline
 i  a  ?1  hour  Average
!r:!s '1on Data
1   .*rea-wHe Emissions Total
>   t.iial rrrlsslon distributed as finite area sources
i   Metalled point,  line and area sources
: . f;.^ro'njlc#i Data
i   •' ^
    K  t^ra^e ^.f^d speed
      ,'•»!'-' w1''d ','^r-rd and rlxing heiqht
     •f,(j,--r,,^ di'tricutlon of v.ird  direction, wind  speed,
    •t«b;) • ng r.eight
                                                                                                                              ion
                                                                                                                              r<;
                                                            10

-------
                              Table 2
    Models* Applicable to Specific Pollutants  and  Averaging Times
S02 end TSP
Annual Averaqe
AOPM
Hanna-Gifford
Miller-Holzworth
Rollback
S02 and TSP
24-Hour Averaae
SCIfl***
Henna-Gifford***
 with point source
 model
AQDM**
Rollback
S02 and TSP
3-Hour Average

SCIM***
Henna-Gifford***
 with point source
 model
AQDM**
Miller-Holzworth***
Rollback
    CO
1-and 8-Hour Averaae
APRAC-1A***
Hanna-Gifford***
 with HIWAY
Rollback
1-Hour Average,

SAI
Appendix J

Rollback
   N02
Annual Average

Rollback
   *  Listed according to level of detail and applicability.

  **  Statistical conversion of averaging times required.

 ***  Repetitious application of model to each hour under consideration
      is required for averaging times longer than 1-hour.
                                lOa

-------

-------
         It should not be construed that the models discussed here
are the only ones for relating emissions to air Quality.  Other models
which have been summarized and discussed by Lamb, et al,   Calder,
      on          21
Stern,   and Moses   are available from private consultants and other
governmental agencies.  Exclusion of these other models from this volume
is not meant to imply that they cannot or should not be used.  The input
and output of these models should be compared with that of the models
described herein to determine whether they are being used in an appropriate
fashion.  The models discussed in this Volume are (1) those most readily
available to air pollution control agencies and (2) representative of
the state-of-the-art for atmospheric simulation models.
V.  BACKGROUND CONCENTRATIONS
     It must be noted in the application of these simulation models
that background concentrations of pollutants must also be specified.   The
simulation models estimate concentrations only for pollutants which have
Identified sources.  If pollutants occur naturally in the atmosphere  or
are the result of unidentified distant pollutant sources, these pollutant
concentrations must be accounted for and separately added to the dispersion
model estimates.  For example it is commonly assumed that the natural
background concentration of TSP is 30-40 yg/m  over much of the Eastern
              22
United States.    It is usually necessary to add a concentration of this
magnitude to any concentration estimates for TSP.
VI.  MODEL VALIDATION/CALIBRATION PROCEDURES
     To be certain that the dispersion model estimates are as accurate
as possible, validation-calibration is required.  Any model may have
faults v;hich cause estimated concentrations to be in error.  Therefore,
                                11

-------
 it is necessary to validate and calibrate the model  estimates.   The
 availability and accuracy of the input data to these models  will  sig-
 nificantly influence the accuracy of the model estimates.  "The  source
 factors which have the greatest impact on the accuracy  of  the estimates
 are the accuracy and completeness of the emissions data, the accuracy  and
 completeness of data for the physical  plant parameters, and  the  exactness
 with  which the location of the sources is determined.   The accuracy  of the
 concentration estimates are also affected by the  location, exposure, and
 representativeness of the meteorological  monitoring  sites and the overall
 accuracy and completeness of the meteorological data.   Similarly, the
 validation of the  dispersion model  is  affected by the location, exposure, and
 representativeness of the air quality  sampling sites and by  the accuracy
 and completeness of the air quality  data  itself.  These data should  be
 available  for the  same  averaging  times as  the  concentration  estimates.  For
 proper  validation  u,  the  more  sophisticated  dispersion models, air quality
 data  should  describe  the  spatial  variation of  pollutant concentration
 across  the  area.   In  short,  if  the air quality  data are in any way unsuitable
 or  incorrect,  the  validity  of  the dispersion model estimates  cannot be
 determined.   Statistical  methods  available for  validation  of models include
 skill scores,  contingency tables, correlation  analyses and comparisons of
                                   ?1
 timo  series  and spatial variations.
         If  evaluation  by one or more of these statistical  techniques
 indicates that the concentration estimates are not a  satisfactory repre-
 sent,-.* •   , of observed concentrations, even though all source, emissions,
meteorological and air quality data are reliable and  complete,  it must be
                                   12

-------
  assumed that  the  dispersion node!  itself  is  inadequate or inappropriate
  for  the area  under  consideration.  This is likely to occur (1) in areas
  with  complex topographic/meteorological conditions, (2) where model
  inputs  do not allow a complete description of significant source variations,
  (3) where atmospheric reactions or rates of removal of the pollutant from
  the atmosphere are  not adequately accounted for in the model, or (4) for
  pollutants which were not considered in development of the model.   In
  such cases it will be necessary to find a more appropriate model  or make
 appropriate changes to available models.
          Once the dispersion model  estimates  have  been  determined  to be
 acceptable,  they must be calibrated.   The  calibration  should  account for
 systematic errors  in the estimates.   A model  is  calibrated  by comparing the
 calculated concentration estimates  to observed air Quality.   Then  by use of
 a regression equation  or averaged correction  factor, a means  for adjusting
 the  individual model estimates  is developed.   It should be emphasized that
 once a model is calibrated, all future  uses  of the model must employ a  data
 base which is  similar  to  that on which  the calibration was based.  For
 example  if AQDM were calibrated for a 1970 point/area source data base,
 the calibration could then be used to estimate 1985 concentrations with an
 appropriate emission inventory projected to 1985.  However for consistency,
 the projected inventory should identify all point sources in 1985.   If due
to a lack of information, the inventory for 1985  must be generalized to
area source emissions,  then the base year  (1970)  should consider an emissions
inventory which also has  been  generalized  to  area source emissions.
                                  13

-------
VII.  SUMf'ARY
      The use of the above data and dispersion models in air quality
maintenance should proceed as follows.  First, a base year is selected.
The dispersion model is applied to emissions and meteorological  data for
this base year and the dispersion estimates compared to the observed air
quality.  In this way the model can be validated and deficiencies in the
dispersion model can be identified and corrected (Figure 1).  In addition,
any inadequacies in the basic emissions inventory, the meteorological
data, or available air quality data can be identified.  Next, the dispersion
model is applied to the projected 1985 emissions inventory.  For this
application, meteorological data which can be considered representative
of atmospheric conditions in 1985 are required.  This necessitates the
use of climatological means and extremes or the use of data from a year
which exhibited average dispersion conditions.  With these data  the
likelihood of exceeding the air quality standards in 1985 can be ascertained.
This is done with the calibrated model so that some confidence in the
estimates is assured.  Next, if it is determined that  the  air quality is
likely to be greater than National Ambient Air Quality Standards in 1985,
this dispersion model can be used to determine the ability of alternative
strategies for emissions control in  1985  (Figure 2).  Thus, the dual
purpose of air quality maintenance evaluation can be met.
VIII.  CAUTION
     Like any "tool," atmospheric simulation models are useful only if
the user understands how to apply them and is aware of their vagaries.
They are highly specialized tools which require, in many cases,  large
                                  14

-------
amounts of detailed technical data.  As such they should not be used

indiscriminately.  It is strongly urged that the services of competent
                              •           '                *
air pollution meteorologists, engineers and air quality analysts be


solicited in the application of atmospheric simulation models.   This


need is especially critical in the application of the more sophisticated


models and in complex meteorologic/topographic locales.  Without the


availability of such expertise, a model applied improperly or with


incorrect data, can lead to erroneous conclusions about the ability to


maintain air quality.
                                   14a

-------

-------
H. , • f — * , ^.
-Vcr.'i
Gsy
1 D!S?E
5 SIGN
DEL
J
  METEOROLOGICAL
A;.;D TCrGcHAPHICA
        DATA
 3DT~r*t''*Tr-T-\
 r r. c U i v..; c D
A! F; O JI •'" ! 7 v
*->.iu L; J i-. i_ I j j
                EXISTING
              AIS G'JAL.'TY
                                                            VALID
                                                            MODEL
                                                        COnRECTlC.M
            Fig. 1 -  Dispersion mode! flow diagram.
                                  15

-------

            OF
r
0
i r -\ r •>
.»!;•> >.,.
F sour.
CES
i

                       FO;-".:!JLAT!GM
DATA Orj
AVAIL A'JL
T ;; ^- ;• "^' i
1 L \«»4 . . • I V- w*

E
ES
                       OF CCi.'TfiOL

                   ^   STRATEGIES
                     v
D'JE TO

liRSA,-JiZATIO;\!
                                          STRATEGY 4*
                                                             s:?.:uLAno.\f
 CC;.'TOL7.SOF
[ <\ • •>
                       sr;:A7i cv
                       SELEC: -5
                          Fig.  2  -  Flow diagram for the devslopmcnt of a control strategy.


                                               16

-------
IX.   Model  Descriptions
         17

-------
                               ROLLBACK f-'ODEL
       The rollback rrodel  is based on the  simple  expression  relating
  pollutant concentrations (X)  to  pollutant  emission  rates (0)  and a
  background concentration (b).
       X  = kO +  b.
       The rollback  model  assumes  that the dispersion parameter k
  does  not vary  with time  or with  the source-receptor relationship and
  that  changes in emission rates are  uniform across the area.  Thus the
  relationship of emissions  (Q8g) and air quality in a future year (X85)
  to the emissions (Obase) and air quality (Xbase) in a base  year can be
 expressed by the following proportionality:
      Xbase'b   QBase
 The basic assumption in the model  is  that  a  given  percent  reduction
 or increase in pollutant emissions will  result  in  a  similar  reduction
 or increase in pollutant concentrations.   It is simply  a tool  for
 scaling  concentrations  up or down to  reflect similar changes in the
 gross emission rates.
     The  rollback model  is  applicable to most pollutants and
 averaging times for which  appropriate data are available.  Input to
 tne rollback model requires  total area-wide emissions for the base
year end for 1985 or other years of interest.  A pollutant concentration
    rx.....ative of air quality for the area  and the averaging time
                                 18

-------
 of interest is also necessary.  It should be noted that  since  there  is
 no allowance for specifying the dispersion parameter k or  other
 meteorological parameters, this model cannot be used to  estimate
 concentrations at sites where representative air quality data  do  not
 exist.
      An expansion of the simple rollback procedure is  call ed  Modified
          2
 Rollback.   Modified rollback is a technique for considering a range of
 source categories for the pollutant being considered.  It  can  be
 expressed as
                  N
                    '"1  Gi Fi'85
 where G is  a  growth factor,  e.  g.  the  ratio of population projected
            to 1985 versus  population in  the base year,
       F is  the emission  factor  ratio,  e.g., the ratio of expected
            emissions  per unit of population in 1985 to emissions
            per unit of population  in  the base year,
       N is  the number of source categories,
       i  is  a  particular  source category, e.g. light-duty vehicles,
            stationary sources, etc.
 In cases where  the  area-wide pollutant emissions are contributed by a
variety  of  source  types  with differing emissions,  growth rates and
applicable  controls,  the modified model will allow the situation to
be studied  in more  detail.
     The rollback models are applicable anywhere for which there are
basic data  on area-wide emissions and representative air quality for
a particular base year.  The simple rollback model  can be applied with
                                 19

-------
 simplified hand calculations  and  is  widely  used,   Modified rollback
 has  been computerized and documented.   The  programs and associated
 documentation on the  ."Motor Vehicle  Emission Estimation Program" and
 the  "Modified Rollback  Computer Program", has been  made available to
 all  EPA  Regional  Offices.
     The rollback rrodels  in general  are  valid for  the simplified
 case of  only  one  type of  source uniformly distributed across an area
 affecting  a receptor. . Mobile sources and,associated CO emissions are
 an example of  the type of source which may approximate this situation.
 Accuracy is lost  as the variability of source types  and emission rates
 increase and the  impact of atmospheric processes'on pollutant concentrations
 increase.  Thus due to the importance of point sources for TSP a nd
S02 and the reactive  nature of N02 and px,  this  model  can  provide only
very crude esti^tFc  -f concentrations for  these pollutants.
                                20

-------
                     APPENDIX J  HC-0  RELATIONSHIP

                                     A
      Appendix J of 40 CFR Part 51,  the Requirements  for  Preparation,



 Adoption,  and Subidttal  of Implementation  Plans  (36  P.P..  15486),  is a



 graphical  presentation of the percent  reduction  in hydrocarbon  (HC)



 emissions  required to reduce  an observed peak  hourly average oxidant



 (Ox)  concentration to the national  Ambient Air Quality Standard  (NAAQS)



 for f>x.  The  relationship assumes that the maximum 1-hour 0  concentration
                                                           A


 is  directly affected  by  the quantity of HC emitted during the morning



 hours.   This  assumption  is based on the maximum  observed  relationship



 of  HC and  GX  concentrations for selected cities.  This relationship can



 be  used  to determine  the  effect a change in HC emissions  will have on peak



 oxidant  concentrations.



     To  use the Appendix  graph  it is assumed that HC  emissions are uniform



 across the whole area  and  that  the  peak Ov  concentration  in the area
                                         A


 has been identified.   Inherent  in the  use  of Appendix J is the assumption



 that meteorological conditions  and  the  source-receptor relationship are



 invariant.  Thus there is  no  need for  input of meteorological data.



 Similarly  this  approach cannot  be used  to estimate concentrations at sites



where air  quality data does not  exist.  It  too is simply  a means of



scaling 0^  concentrations  to  variations in  HC  emissions.   Its application



is restricted to peak hourly  average 0  concentrations.    However it is
                                      A


generally  applicable to any area v/here  the formation of 0  is a problem.
                                                          A


     As noted above the Appendix J relationship is presented in the



Federal  Pecistor.  Its application is  elementary.  The reliability of



                                 21

-------
this relationship for determining the effect of HC emissions on GX



concentrations has not been identified.  There is no allowance for



the photochemical reactions which take 'place between HC and NOX and



no consideration of 0  concentrations due to natural sources.  Thus
                     /^


any conclusions concerning the HC reduction required to achieve given



0  levels derived from this relationship should be considered as very
 /\


rough first approximations.
                                   22

-------
                       MILLER-HOLZWORTH f'ODEL

      The Miller-Holzworth dispersion node! is a step in sophistication
 above the basic rollback rrodel for estimating pollutant concentrations.
 In this model  the dispersion parameter k is expressed in terns of mixir-r-
 heiqht, wind speed and the geographical size of the city.  The node! c
-------
to this node! is the uniform average area emission rate (0)  in units
of grains per second per square rreter.  Mixing height (H),  wind speed
through the mixing layer (U) and city size (S) (distance across the
city for a given direction) are also required to estimate  pollutant
concentrations (X).  A discussion of the dispersion model  and appropriate
seasonal average mixing heights and wind speeds is given in  EPA
                   5
publication AP-101.   This publication also provides median, upper
quartile and upper decile (X/Q) values for various city sizes.  Thus  a
range of pollutant concentrations can be  estimated for the  more
restrictive meteorological dispersion conditions.  However,  neither
the emissions inventory used as input nor the output of the  dispersion
model makes it possible to estimate spatial variations in  pollutant
concentrations across the area.  In this sense, any set of emission
control strategic i.r.ich allow similar total  area-wide pollutant
emissions will result in the same air quality for the area.   No matter
how different the strategies are, if the total emissions are the same,
then the air quality estimated will be the same with this  model.
     This is a simple model for which concentration estimates can be
made with a slide rule or from the summaries  in AP-101.  The reliability
of the dispersion model when applied to specific areas to estimate
1-hour concentrations has not been determined.  However, as  noted
earlier, the model has been calibrated.  Such a calibration  is presented
             23
in Anp- l,n A   to Requirements for Preparation, Adoption and Submittal
of Implementation Plans (36 F.R. 15486) for a range of city sizes
so that estimates of annual average concentrations can be made.
                                  24

-------
                         HANNA-GIFFORD MODEL
      The Hanna-Gifford dispersion model  is the next step  upward  in
 dispersion model sophistication.  The Hanna-Gifford model  is .most
 readily applicable to stable pollutants  such  as SOp,  particulates,
 and CO.  It has been used to estimate 1-hour  and annual average  con-
 centrations of such pollutants.   The model can be used  to  estimate
 an average concentration for any defined area.  This  model  differs
 from the previous models in the  form of  the dispersion  constant  (k).
 In the basic Hanna-Oifford model the constant is a  function of
 stability, wind speed and the size and number of area sources.   The
 equation relating concentrations to emissions is
                                    N
                               Q  + [ Q   [(21  + I)1'13-  (21 -  I)1"15
                                °  1=1 i
X =
*x
where
a
f2]
H

(Ax /2)1-b
a(l-b) U

        a,  b  are  empirically  determined  constants used to specify
           dispersion,  they  are  functions of the atmospheric stability
       AX is the size  (width)  of the area sources,
        N is  the  number  of  upwind  sources,
        i is  a specific  upwind source,
       QQ is  the  pollutant  emissions for the area in which the
         receptor site  is  located,
       ^i are the emissions for upwind areas.
The model is applied to each area which encompasses a receptor site.
The application  is made for  the wind direction, wind speed and stability
class  for each meteorological situation under consideration.  All sources
upwind of the receptor area are included in determining the pollutant impact,
This approach is used to estimate hourly average concentrations for
                                 25

-------
 all  situations of interest.   If data  on  diurnal variations  in emission
 rates are available,  they can be used with  the model.  Concentrations for
 other averaging times  can be  obtained by estimating concentrations for
 each hour of the period  and averaging the hourly concentrations.
      Annual  average concentrations can be obtained in a manner similar
 to that  described for  AQDM.   However  Hanna-Gifford have simplified
 this procedure by substituting  a more elementary form of the dispersion
 equation.   For annual  average concentrations the model takes the
                         r
 simplified  form of x*  =  - 0,  where C  is  a constant dependent on the
           _            u
 pollutant,  u  is  the annual average wind  speed, and 0 is the average
                          o
 emission  density  (g/sec/m ) for  the whole area of interest.  However
 this  form of  the  model can only  be used  to estimate a representative
 average concentration  for the whole area.
      In examinations of  the reliability of this model, it has been shown
 that  the more  detailed formulation provides an accuracy similar to
 that  of much more  sophisticated models.   Correlations generally
 varying between  .60_ and  .95_ depending on the pollutant, averaging time
                          14
 and  area have  been found.    For the simplified model, from an application
 to a  large number  of areas, values of the constant C equal  to 225 and
 50 respectively for particulates and S02 were found.   However the
 standard deviation of the error in concentration estimates  obtained
when the mean value for the constant is  used, has  been found to be
about 50 vg/m  for TSP.  Further analysis has shown that  the constant
 is overly simplistic and that  a relationship of x  = 52 +  91.7 Q/u
provides a better estimate for particulate matter.   The use of this
relationship in place  of an average for  C reduces  the standard
deviation of the error to 10 ug/m .  For SOp a  similar
                                  26

-------
relationship could not be derived because the  variation  of the
constant C was too random.  This reflects the  importance of a
largo point source emissions in determining SCL concentrations  and
indicates that caution should be exercised in  the use of a value
of 50 for C when considering S0?.
     The model is basically applicable for areas where there is
no point source information available so that  all emissions are grouped
into area source emissions.  However, if specific point sources
or line sources are known the impact of these  sources can be
estimated independently through the use of an  appropriate model and
concentrations added to those caused by area sources.  Diurnal  variations,
where available^in emission rates should be considered in the application
of such specific models to point and line sources.   Appropriate
point source models are discussed in guidelines for Reviewing New
                   1 n
Stationary Sources.    These models are generally accepted to be
accurate within a factor of two.  The most basic point source
dispersion models are available through the National Technical
Information Service.    An appropriate line source model is discussed
in this Guideline under the HIWAY model.
     Discussions of the Hanna-Gifford dispersion model are available
from various literature sources.  '    The model is simple enough
to estimate concentrations with a slide rule and hand tabulations.
                                  27
                                   a

-------
                             HIWAY KODEL

      The HIWAY model   is a basic line source dispersion rnodel .
 It is applicable to mobile source pollutant emissions along
 roadways such as highways and streets and is applicable  only  to  stable
 pollutant emissions (CO and fine particulate matter)  from
 automotive sources.  The model  is applicable to  estimate one-hour
 average  concentrations of these pollutants.   By  repetitively
 simulating  each  hour in an 8-hour or  24-hour period,  8-hour and
 24-hour  concentrations of the pollutants  of  interest  can  be estimated.
 In this  way concentrations at specified distances downwind from the
 roadway  or  line  source can be obtained.
     The  basic formulation for  this line  source model is
         X - il           1                  y \2
                   i
                  J
                    A
where x is the concentration at the receptor
      u is wind speed
      qe is the constant line source emission rate
     ty oz are the standard deviations of the concentration
            distribution in the crosswind and vertical  directions;
            they are dependent on atmospheric stability and on
            downwind distance from sources to the  receptor
     A,  B  are  the endpoints of the line source
                                 28

-------
     L is the length of the line source from point A to the  ith
       element of the line source
     y is crosswind distance from source element to the receptor.
The model is applied to each lane of traffic which makes up  the  line
source to estimate the contribution to concentrations at specified
downwind distances.  Concentrations can be calculated for any
specified combination of meteorological conditions.
     Basic inputs into this dispersion model are emissions in terms
of grams per meter per second for each lane making up the roadway
for each hour being considered.  Due to the diurnal variations in
emission rates of automotive sources, emissions should be appropriately
varied with time of day. In addition, the length of roadway, its width,
and other physical parameters of the roadway need to be specified.
Meteorological inputs into the model are stability class, wind direction
and wind speed for the hour of interest.  The output of the  model  is
one-hour average concentrations at specified points downwind from  the
roadway.  Concentrations estimated for each hour in a given  sequence  of
hours make it possible to obtain 8-hour and 24-hour concentrations  of
pollutants.  The model is not readily applicable to estimating annual
average concentrations.
     To determine the impact of line sources, in addition to more
generalized area source emissions, the HIWAY model can be used in
conjunction with the Hanna-Gifford rrodel.  The Hanna-Gifford model
is used to determine the impact of all other sources in an area while
the HIWAY model is used to determine the impact of specific  nearby
roadways.
                                 29

-------
     The model is available from the National Technical Information
Service.    HIK'AY has been compared with preliminary data which
indicate that this model has an accuracy comparable to that of point
source dispersion models.
                                   30

-------
                    AIP OU/UITY DISPLAY f'ODFL

      The Air Ouality Display Model (AODr) is a long-term average
 concentration urban dispersion model which is best used to determine
 the impact of a wide variety of stationary source classes on annual
 averaqe concentrations of S02 and TSP.  The model has been widely
 applied to areas with numerous point and area sources; it has been
 validated and calibrated for these areas.
      The model  is based on the standard long-term concentration
 equation
                        2 Q f


 where  f(e,  S, N)  is  frequency during  the  period of  interest  that
                  the  wind is  from  direction  e, for  the  stability
                  condition  S,  and  the  wind -speed  class  N.
              oz  is the vertical dispersion  parameter  at  distance x
                  for  stability condition S
               u  is the representative wind  speed for class N
               H  is the effective  stack height for wind speed u.
The model is used to determine the impact of all  sources at a oiven
receptor, for a ciiven set of nieteoroloaical conditions.  It then
weights this concentration by the frequency with  which that particular
set of meteorological conditions occurs and then  sums over all
                                 31

-------
meteorological  conditions, thus producing a long-term average
concentration.  Basic  inputs to the model are a comprehensive emissions
inventory  including both noint'sources and area sources.  Meteorological
input  is a joint frequency distribution of wind speed (6 classes), wind
direction  (16 cardinal points), and stability class (Pasquill classes
A-F) along with an annual average mixing height.  The dispersion model can
be used to estimate concentrations at any point downwind that is specified.
     When adequate point source information is not available a properly
calibrated model could be used with just area source information,
that is, all point sources included in area source emissions.  However,
it must be recognized that the reliability of this model or any model
will be adversely affected by such an assumption, namely that all
point sources can be considered in the area source emissions.
     This model is available from National  Technical Information Service6
and as a computer system.  A similar model, called Climatological Dispersion
Model, is also available.8'    The two models have differences in calculation
techniques, however they require basically the same inputs  and have the
same type of outputs.   A significant difference betv/een  the two is
that there is a source contribution file produced by A(W which allows
the impact of each individual  source on air quality to be obtained.
The same sort of output cannot readily be obtained from  CDM.   It should
be noted that the  AODM was  originally developed with the Holland plume
rise equation for  calculating  effective stack heights.   A. change to the
computer program is now available  for the inclusion of the  Briggs1  plurr.e
                                 32

-------
rise equation.   Ihis latter equation is  preferred,  especially  for
large sources,  e.q. power plants.
     AOPr can he calibrated using  regression  analysis  techniques.
In nurerous applications of the model  a  correlation coefficient  of
C.8 or greater has been found to be typical.
                                  33

-------
                 SAMPLED CHRONOLOGICAL INPUT MODEL

      The Sampled Chronological Input Model (SCIM) is one of several
                  20
 dispersion models   which are applicable to both point and area source
 emissions in urban areas.  These models  as a class  are basically
 used to estimate hourly average concentrations  of SC>2 and particulate
 matter which are emitted from stationary sources.   SCIM specifically
 determines e frequency distribution of one-hour concentrations.   If
 this model or any other model is  to be used to  estimate 24-hour
 concentrations,  a series of hourly  observations  must be used  to
 determine 24 1-hour concentrations.   While this  model  does  not
 sppr.ifically pstimate  24-hour concentrations, it provides  a summary
 of all  estimated hourly concentrations.   Thus estimates  of  con-
 centrations  for  averaging  times  longer than  1 hour  can  be  tabulated.
 Through  the  use  of  standard  statistical  techniques  SCIM  summarizes the
 1-hour  estimates  as  a  distribution of  concentrations.  This model
 can  be  used  to estimate concentrations at  any desired site.  The model
 is based  on  the  standard Gaussian dispersion equation;
    -V        Q         F1/'V\2]F1/'H\-1
     y^	 c\-p	„-  — -    tvp    	--A —
            » «r. 
-------
 no  allcvvnce for diurnal variations in emissions.  Limited  tests  with
 variations  in emissions based on time of day did not show a  consistent
 improvement in concentration estimates.  The meteorolooical  data
 required  e.re specific hourly data on wind speed, wind direction,
 stability class and mixing height.  The model does provide  the option
 of  estii..£'ting concentrations for selected intervals, e.p. every 3
 hours  to  lessen the computational burden.  /* judiciously selected
 interval, which insures that estimates for all times of day  for a
 sufficiently long  period of time, will not adversely influence the
 frequency distribution for hourly concentrations.  Documentation  of
 this model  is  in progress.
     The  reliability of this model has not been widely determined;
 howpvpr.  annl lV;>t-j npc  tn thrpo  riti'oc  i prH r:\-t-rNc -i-ha-t- +-Kn •Fv^^i.r.M,-,/
        >  ~r[~- -'-'•—  -~ «.-.---,  ^-.w.^i  i I, v- i w v. w i. ^ 1.11 w> w LIIV. iii.t^wC'ii^^
distributions  of 1-hour concentrations estimated by the dispersion
model approximately reproduce the observed frequency distributions.
                                 35

-------
                        APRAC-1A  KODEL

      APRAC-1A is  a  dispersion  model which  is applicable only to line
 and area  sources  of stable  automotive pollutants  (CO and participate
 matter).   The model  can be  used  to determine 1-hour averages of these
 pollutant  concentrations  or a  frequency distribution of such concentra-
 tions.  It too can  be used  to  estimate concentrations for other
 averaging  times by  averaging a series of hourly estimates.  It can
 be  used to estimate  concentrations at any desired site.
     APRAC-1A is  based  on a modification of the Gaussian dispersion
 equation.   The  basic input  to  this dispersion model is in terms of
 individual  street segments  or  line sources.  The model takes these
 segments and  determines area source emissions from these line sources;
 these emissions are combined with other specified area source emissions
 to  indicate total emissions. A method for determining diurnal  variations
 in  emissions  is included in this model.   These area sources are oriented
 in  the upwind direction.  A logarithmic spacing of the area sources
 allows the nearby sources to be considered in greater detail than  those
 farther sources, whose individual contributions tend to be merged
 during longer travel.  Meteorological  input to the model  is the same
 :s with other dispersion models;  wind direction,  wind speed, stability
class and  mixing height on an hourly  basis.
     "uus  model also incorporates a street effects submodel.  This
submodel  allows the  impact of localized  emissions  and wind circulations
                                 36

-------
on pollutant concentrations to be determined.   It basically  takes  the
form of a box model with slightly different methods  for calculating
concentrations on opposite sides of the road.   This  difference is  an
attempt to reflect the impact of the complex v/ind circulation which
has been identified in street canyons.
     The model is available from National Technical  Information
Service/'    '     This dispersion model has been applied to St.
Louis and San Jose and estimates compared to observed data;  it can
be expected to have a reliability similar to that of other dispersion
models.
                                 37

-------
                    SAI PHOTOCHEMICAL f'ODEL

      The SAI model is a ohotochemical dispersion model.   It not
 only considers the transport and dispersion of pollutants but also
 the transformation of HC and NOX into photochemical  oxidant pollutants.
 It estimates the hourly concentration variations in  these pollutants;
 CO, HC, NO, NOp and 0 .
               ^-      J\
      The mathematical formulation for this model is  considerably
 different from that discussed for the other dispersion models.  This
 model  uses  finite difference techniques  over a qrid  of area sources
 to solve the classical  equations of  conservation of  mass  v/hich
 include local  change, advection, diffusion,  photochemical  reaction
 and emission.
     The emissions  input  to  this dispersion  model is specified  as
 uniform gridded area  sources;  hourly  emission  rates  and their diurnal
 variations  are  specified.  Meteorological  input  are  hourly  data
 on  the  spatial  variation  of  wind direction,  wind  speed and mixing
 height.   The model  is capable  of estimating  for  each hour an
 average  concentration for each area identified as an area source.
 'fr.is model  is most  applicable when large amounts  of data are available
 :,.,* a sophisticated analysis of  a region.  To this date it has only
 reen applied to areas in California.
     This model is one of the better sophisticated ohotochemical
models currently available.  The reliability of this  model has not
been thoroughly examined.
                                  38

-------
VIII.  REFERENCES







(1)  l.ersen, R. I.; "A Mathematical  Model  for Relating Air Quality



Measurements to Air Quality Standards";  Office of  Air Programs



Publication I'o. AP-£9 (NTIS PB 205277);  Office of  Technical



Information and Publications; U.  S.   F.PA;  Research Triangle  Park,



N. C.  27711; (November 1971).



(2)  Turner, P. B.; "Workbook of Atmospheric  Dispersion  Estimates";



PUS Publicetion No. 999-AP-26 (NTIS  PB 191482); Office of Technical



Information and Publications, U.  S.   EPA;  Research Triangle  Park,



N. C.  27711; (1969).



(3) de Nevers, f!. and J. P. Morris,  "Rollback Modeling—Basic  and



Modified"; Paper Number 73-139; Presented  at  1973  Annual Air



Pollution Control Association Meeting; Chicago, 111.;  (June  1973).



(4)  U. S.  EPA; "Requirements for Preparation, Adoption,  and



Submittal of Implementation Plans—Appendix J"; Federal  Register 36;



No. 158; p. 15502; (August 14, 1971).



(5)  Holzworth, G. C.; "Mixing Heights,  Wind  Speeds,  and Potential



for Urban Air Pollution Throughout the Contiguous  United States";



Office of Air Programs Publication Ho. AP-101 (NTIS PB 207103);



Office of Technical Information and Publications;  U.  S.   EPA;  Research



Triangle Park, N. C.  27711;  (January 1972).



(6)  TRW Systems Group; "Air Ouality Display Model";  Prepared  for



the rational Air Pollution Control Administration  under  Contract Ho.



PH-22-68-60 (NTIS PB 189194), DHEW,  U. S.  Public Health  Service,



Washington, 0. C.  (November  1949).



                                 39

-------
(7)  Koch, R. C. and G. H. Stadsklev; "A User's  Manual  for the
Sampled Chronological Input Model (SCIM)";  GEOMET Report  No.  E-261
prepared for U. S.  FPA Under Contract Number 68-02-0281; U.  S.   EPA,
OAQPS, Research Triangle Park, N. C.  27711;  (August 1973)  (available
in draft form only) .
(8)  Busse, A. D., and J. R. Zimmerman; "User's  Guide for the
Climatological Dispersion flodel"; Environmental  Monitoring Series
EPA-R4-73-024 (NTIS PB 227346AS) NERC, EPA, Research Triangle Park,
N. C.  27711 (December 1973).
(9)  Ludwig, F. L. and R. L. Mancuso; "User's Manual for  the
APRAC-1A Urban Diffusion Model Computer Program," Prepared for
U. S.  EPA Division of Meteorology Under Contract CAPA-3-68 (1-69)
(NTIS PB 213091); U. S.  EPA, Research Triangle  Park, N.  C.
27711; (September 1972).
(10)  Ludwia, F. L. and W. F. Dabberdt; "Evaluation  of the APRAC-1A
Urban Diffusion Model for Carbon Monoxide"; Prepared for  U. S.
EPA, Division of Meteorology Under Contract CAPA-3-68 (1-69)  (NTIS
PB 210819); U. S.  EPA, Research Triangle Park,  N.  C. 27711
(February 1972).
(11)  U. S.  EPA; User's Network for Applied  Modeling of  Air  Pollution
fUNAMAP); (Computer Programs on Tape for Point Source Models, HIWAY,
Clinatological Dispersion Model and APRAC-1A) NTIS PB 229771
National Technical Information Service, Springfield, Virginia  22151
                                 40

-------
(12)   Systorrs Applications  Inc.;  Urban  Air  Shed Photochemical
Sinuletion Model  Study—Volumes  I-I II"; Prepared  for U. S.
EPA,  CPD Under Contract Nurrber 68-02-0339;  EPA-P4-73-030g;
Washington, D. C.  20460;  (July  1973).  (Available from APTIC).
(13)   Hanna, S. P.; "A Simple f-'ethod of Calculating Dispersion
from Urban Area Sources";  J.  Air  Pollution  Control Ass'n.,
Vol.  21, No. 12,  pp. 774-777; (December 1971).
(14)   Gifford, F. A., and  S.  R.  Hanna;  "Modeling  Urban Air  Pollution";
Atnospheric Environment, Vol. 7,  pp. 131-136;  (1973).
(15)   U.S.  EPA, OAQPS; "Reviewing New Stationary Sources";
Guidelines for Air Quality Maintenance  Planning  and Analysis, Volume  10;
OAQPS No. 1.2-029; (September 1974).
(16)   Zimmerman,  J. R. end R. S.  Thompson;  "User's Guide  for
HIWAY:  A Highway Air Pollution Model"; Environmental Monitoring
Series EPA-650/4-008, NERC, EPA,  Research Triangle Park,  N. C.
27711 (in preparation).
(17)   U. S.  EPA, OAQPS, CPDD; "Guidelines  for Designation  of Air
Quality Maintenance Areas"; OAQPS Publication  Mo. 1.2-016;  U. S.
EPA,  OAQPS, Research Triangle Park, N.  C.  27711; (January  1974).
(18)   Lamb, D. V., et al;  "A Critical Review of  Mathematical
Diffusion Modeling Techniques for Air Quality  with Relation to  Motor
Vehicle Transportation"; A Study Prepared for  the Washington  State
Highway Commission, Department of Highways; University  of
Washington, Seattle, Washington  (June 1973).
                                 41

-------
(19)  Calder, K. L.; "Mathematical  Modeling of Air  Quality
Through Calculation of Atmospheric  Transport and Diffusion"  in
Proceedings of the Third Meeting of the Expert Panel  on  Air
Pollution Modeling, CCMS/NATO, Paris,  France (October 2-3, 1972).
(20)  Stern, A. C., "Proceedings of Symposium on Multiple-Source
Urban Diffusion Models"; Air Pollution Control Office Publication
No. AP-86 (NTIS PB 198400), Office  of  Technical  Information  and
Publications; U.S.  EPA; Research Triangle Park, N.  C.   27711;
(1970).
(21)  Moses, H.; "Mathematical Urban Air Pollution  Model";
Argonne National Laboratory Peport  ANL/ES-PPY-001;  Argonrie
National Laboratory, Argonne, Illinois, 60439 (April  1969)
(Limited Distribution Only).
(22)  McCormick, R. A ; "Air Pollution Climatology  in Ai_r
Pollution Volume 1, Edited by A. C. Stern, Academic Press,
New York, flew York  10003 (1968).
(23)  U.S.  EPA; "Requirements for  Preparation,  Adoption
and Submittal of Implementation Plans—Appendix A";  Federal
Register, 36. No. 158; pp. 15494-15495; (August 14,  1971).
                                  42

-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing!
' -. " ; NO 2.
PA .1^0/4-74-013
j -'.;-. fl\D SUBTITLE
•auin-'i jnes for Air Quality Maintenance Planning and
. Vo^o'fz: Applying Atmospheric Simulation Models to
A"'-"1 'Tniity Mai nienance Areas
. -< ' 3 '
... ,T pt OP MING ORGANIZATION NAME AND ADDRESS
'. -i--° Receptor Analysis Branch
/•V,-v; ';.v(-ing and Data Analysis Division, OAQPS, EPA
; R"-,
-------

-------