EPA-650/4-75-018-b
                 EVALUATION
        OF  THE  MULTIPLE  SOURCE
GAUSSIAN  PLUME  DIFFUSION  MODEL -
                    PHASE II
                        by

              Robert C. Koch and Scott D. Thayer

                     Geomet, Inc.
                    15 Firstfield Road
                Gaithersburg, Maryland 20760
                  Contract No. 68-02-0281
                    ROAP No. 21ADO
                Program Element No. 1AA009
             EPA Project Officer:  D. Bruce Turner

                  Meteorology Laboratory
            National Environmental Research Center
          Research Triangle Park, North Carolina 27711
                     Prepared for

          U.S. ENVIRONMENTAL PROTECTION AGENCY
           OFFICE OF RESEARCH AND DEVELOPMENT
                 WASHINGTON, D.  C. 20460

                      April 1975

-------
                       EPA REVIEW NOTICE

This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication.  Approval does not signify th^t the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
                   RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series.  These broad
categories were established to facilitate further development and applica-
tion of environmental technology.  Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields.  These 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
          9.  MISCELLANEOUS

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 quanti-
fication of environmental pollutants at the lowest conceivably significant
concentrations.  It also includes studies to determine the ambient concen-
trations 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 for sale through the National
Technical Information Service, Springfield, Virginia 22161.

                Publication No. EPA-650/4-75-018-b
                                 11

-------
                     TABLE   OF   CONTENTS
1.0  INTRODUCTION AND SCOPE
2.0  SUMMARY OF THE PHASE I REPORT                                   6

     2.1  Models Compared                            -                6
     2.2  Model Comparisons                                          7
     2.3  Conclusions from the Phase I Report                        9
3.0  PHASE II WORK REPORTED AND ACCOMPLISHED ELSEWHERE              12

     3.1  SCIM Computer Program and User's Manual                   12
     3.2  Training of EPA Personnel                                 14
4.0  CONSIDERATIONS OF EMISSION DATA, EMSU DATA AND
     SHORT-TERM MAXIMUM CONCENTRATIONS                              15

     4.1  Diurnal and Seasonal Variations in Emissions              15
     4.2  Use of EMSU Data as Meteorological Input                  21
     4.3  Procedure for Calculating Annual Short-Term
          Maximum Concentrations                                    35
5.0  REFERENCES                                                     46
Appendix A-l  METHOD OF ESTIMATING THE HEIGHT OF THE
              MIXING LAYER
Appendix A-2  MIXING HEIGHT INTERPOLATION

-------
     Section 1.0
INTRODUCTION AND SCOPE

-------
                              Section 1.0
                         INTRODUCTION AND SCOPE

          This report represents a summation, for record purposes of a
variety of types and phases of work conducted under EPA Contract Number
68-02-0281, "Evaluation of the Multiple-Source Gaussian Plume  Diffusion
Model."  Because the contract work covered an extended period  of time,
and because its products were documented in a number of reports, it is
considered desirable to have this summation document as a matter of com-
plete record of the work; of course, where the work has been formally
reported elsewhere, this report will contain only a summary.
          The scope of this report will  be to briefly cover Phase I  by
reproducing the introduction, model  description,  and conclusions of the
report of that work (Number EF-186).  Phase II will be more definitively
covered by presenting a summary of the report (Number EF-261)  of the com-
puter program and user's manual and describing the training provided EPA
staff, and by accumulating the work reported elsewhere related to special
aspects of the model  (handling variations in emission data input, using
EMSU data as meteorological input, and calculating short-term  maximum
concentrations).   Phase I is covered in  Section 2.0 of this report,  the
computer program, user's manual, and staff training in Section 3.0,  and
the special aspects in Section 4.0.
          For convenience of reference,  and to complete the introductory
remarks, the contract scope of work  is quoted in  the following paragraphs.
                                   -1-

-------
For cross-referencing purposes, the following relationships are given

relating the scope of work's paragraphs 1 and 2 (Phase I), and 3 through

7 (Phase II) with this report's sections:


               Contract Scope of Work                 Report Section

Phase I          1, 2, 3 (Phase II)                        2.1

                      1, 2                                 2.2


Phase II               3                                   3.1

                       4                                   3.2

                       5                                   4.1

                       6                                   4.2

                       7                                   4.3


SCOPE OF WORK
Background:
               This program shall be a continuation of work previously
          supported under Contract No. CPA 70-94, "Validity and Sensi-
          tivity of the Gaussian Plume Urban Diffusion Model."  (Avail-
          able from NTIS as PB 206-951).

               This previous work developed a short-term, steady-state
          Gaussian plume model for urban  diffusion and evaluated this
          model using three-months'  data  for St.  Louis and one-month's
          data for Chicago.  By proper selection  of input data, this
          model can also be used for long-term average concentrations.

               Since this model is expected to replace a currently used
          annual model, it is necessary to make direct comparisons with
          two other models with the same  data base used by these two
          models.

               It is required to completely document the model  so that
          the dispersion modeler can completely understand its steps of
          operation in detail including underlying assumptions and the
                                   -2-

-------
Abstract:
Phase I:
          technical user can properly assemble required input data and
          interpret correctly the concentrations resulting from the model.
               The Contractor shall conduct a research program for further
          evaluation and documentation of a Gaussian Plume Urban Diffusion
          Model.
               For both St. Louis and Chicago data (as considered in the
               previous contract) calculate 1-hour (Chicago) and 2-hour
               (St. Louis) concentration frequency distributions using
               GEOMET (Mean Q) and using sound statistical  techniques
               compare with the results of GEOMET (Variable Q),  calcu-
               lated by the previous contract and with measured  concen-
               trations.

               Calculate concentrations for locations in the New York
               area for 1969 and make comparisons with measurements for
               the averaging times and the models indicated by the sponsor.
               Similar calculations will be made for particulate matter
               for annual averages only.
               Mean annual emission rates for all point and area sources
          and stack characteristics for the point sources for the New York
          region will be furnished by EPA in the format used for IPP.

               Meteorological information for the year 1969 consisting
          of an observation each three hours will also be furnished by
          EPA and will be used as the meteorological data base.

               For each calculation of concentration at a receptor, a
          value for the concentration due to point sources will  be retained
          as well as the calculated total concentration.   All estimates of
          concentrations made will be stored on magnetic tape (hourly and
          24-hour concentrations as time series) and delivered to EPA for
          possible subsequent analysis at the conclusion of the  contract
          period.

               Data on measured S02 concentrations will be furnished the
          contractor by EPA.  Frequency distributions for measured con-
          centrations and for calculated concentrations will be  determined
                                  -3-

-------
Phase II:
          and compared by log-probability plots and by appropriate sta-
          tistical techniques for each station.  (There are approximately
          40 locations with sufficient S02 measurements to obtain fre-
          quency distributions.)  In consultation with the project
          officer, the contractor will select 10 for study of frequency
          distributions.

               Appropriate subsets of calculations will be used to vali-
          date the usefulness of proportionate stratified sampling in
          obtaining frequency distributions.  Resulting frequency distri-
          butions from these subsets need only be compared with the cal-
          culated frequency distribution using all data.

               For each station a linear correlation coefficient, the
          variance (the square of the correlation coefficient), the
          slope and intercept of a least-squares regression line will be
          determined considering the calculated concentration as the inde-
          pendent variable and the measured concentration as the dependent
          variable.  Considering error as the calculated value minus the
          measured value, the mean absolute error, the root-mean-square
          error, and the distribution of errors will be determined.  For
          all but the distribution of errors, there will be a value of
          the above numerated statistics for each station for both 1-hour
          and 24-hour averaging times for each applicable model.  For
          annual averages, the pairs of calculated and measured concentra-
          tions for all stations will be included to calculate one value
          of each of the above statistics for each pollutant for each
          model.
          3.   Considering the results of the sensitivity analysis (pre-
               vious contract) and the results of the evaluation in Tasks  1
               and 2, restructure the computer programs used in  GEOMET
               (Variable Q)  to minimize the digital  computer execution
               times.  One version of these programs must be compatible
               with the IPP.   (Information on the IPP will  be furnished
               by EPA.)  If simplification will  reduce computer  time
               without significant loss of accuracy, a separate  model  shall
               be suggested and documented for annual average concentra-
               tions.  Prepare a user's manual for the use  of these opti-
               mized computer programs for use with  or without the IPP.

          4.   Train 2 to 3 members of the Model  Development Branch,
               Division of Meteorology, on the operation of the  optimized
               models resulting from Task 3, on  a computer  used  by NERC,
               North Carolina.  This shall include a demonstration of  the
                                   -4-

-------
     compatibility with the IPP.   (EPA personnel  shall  be
     responsible for operation of all  phases of the IPP not
     directly connected with the  dispersion model.)

5.   Write procedures to be used  in preparing emission  data so
     that diurnal and seasonal variations in emissions  can be
     used.

6.   Evaluate the use of Environmental Meteorological  Support
     Unit (EMSU) data for determination of meteorological param-
     eter values for input to the optimized model.   Enumerate
     procedures for the use of such data.

7.   Examine the statistical portion (Larsen transform) of the
     AQDM and suggest alternative procedures to be  utilized
     with the optimized dispersion model  to estimate short-term
     (1-hour, 3-hours, and 24-hour) maxima that occur with a
     frequency of once per year.
                         -5-

-------
         Section 2.0
SUMMARY OF THE PHASE I REPORT

-------
                               Section 2.0



                      SUMMARY OF THE PHASE I REPORT





          The report of Phase I (Number EF-186) contains complete docu-



mentation of the work on the basic model, and evalution of its valida-



tion performance against monitored air quality data in comparison to



other models potentially usable for similar purposes.   The work is sum-



marized here by reproducing brief descriptions of the models compared,



analyses performed, and conclusions of that report.



          The work accomplished during Phase I concentrated on validation



of a steady-state Gaussian plume urban diffusion model which uses sampled



chronological input data.  The model was developed for EPA by GEOMET in



previous work (Contract Number CPA 70-94).  The model  has been compared



with three other models (using the same data base).





2.1  MODELS COMPARED



          The model  studies include three variations of the multiple-



source, Gaussian plume, meteorological diffusion model.  Two of the models,



the Air Quality Display Model (AQDM) and the Climatological Dispersion



Model (COM)5 are primarily designed to calculate long-term mean concentra-



tions.   The third model, developed by GEOMET under previous EPA sponsorship,



is designed to calculate both the long-term mean concentration and the



frequency distribution of short-term concentrations using selected chrono-



logical data (SCIM).   The frequency distribution is determined by concen-



trations calculated for a statistically selected set of short-term periods.



Representative meteorological characteristics and simulated time-dependent
                                   -6-

-------
emission characteristics are determined for each selected period.  The
other two models use mean emission characteristics and a specified set
of combinations of meteorological characteristics (wind direction, wind
speed and stability).  To determine the long-term mean, the calculations
for each combination of meteorological conditions are weighted by the
relative frequency of occurrence of the combination.
          In addition to the three Gaussian plume models, the simplified
version of the Gifford-Hanna model recently described by Hanna (1971) was
included.  The model is
                                   - r Q
                                 x - C u
where
          x = concentration at a receptor location (yg/m^)
          Q = area source strength surrounding the receptor (ug/m^/sec)
          u = wind speed (m/sec)
          C = dimensionless constant.

2.2  MODEL COMPARISONS
          Calculations using the four models were compared against each
other and against measured values.  Each model was run in its normal  mode.
In addition, certain simplifications were made by averaging the inputs
used for the model.   The model  comparisons include consideration of 10 dif-
ferent variations of model and inputs.
          Two model  comparison tasks were carried out.  The first task was
to compare calculations for SCIM which were obtained in a preceding program
(Contract No. CPA 70-94, "Validity and Sensitivity of the Gaussian Plume
                                   -7-

-------
Urban Diffusion Model") with calculations made using the same model  and

data set, except that mean rather than time-dependent emission rates were

used for all sources.  These calculations involved a 3-month data period

for St. Louis and a 1-month data period for Chicago, for sulfur dioxide

emissions.

          The second task was to compare 10 different combinations of

variations in input and the four models described above with each other

and with measured values using a 1-year data set for New York City.   The

comparisons include 3-hourly, 24-hourly, and annual  concentrations of

sulfur dioxide emissions, and annual  concentrations  of particulate matter

emissions.

          For the SCIM, area source emissions and the meteorological  con-

ditions of atmospheric stability and  height of the mixing layer (grouped

together) were treated either as varying from hour to hour or as being

constant throughout the data period.   Three combinations of input data

conditioning were analyzed, including:


          •    Area source emission rates,  atmospheric stability and
               height of the mixing layer variable

          •    Area source emission rates constant,  but atmospheric
               stability and height of the  mixing layer variable

          e    Area source emission rates,  atmospheric stability and
               height of the mixing layer constant.


          For the simplified Gifford-Hanna  Model  (GHM), area source  emis-

sions and wind speed were treated as  both varying from hour to hour  or  as

being constant throughout the data period.   In addition, the calculated

concentration at a receptor due to point sources  (as estimated by SCIM
                                   -8-

-------
with variable atmospheric stability and mixing layer height) were either

added or not added to the GHM calculations.  This results in four varia-

tions of this model, including:


          •    Constant area source emission rates and wind speed,
               without point sources

          o    Variable area source emission rates and wind speed,
               without point sources

          e    Constant area source emission rates and wind speed,
               with point sources

          «    Variable area source emission rates and wind speed,
               with point sources.


          Calculations for COM, which treat atmospheric stability and

height of the mixing layer as either both variable or both constant,  were

furnished by Mr. D. B. Turner of the Division of Meteorology, EPA/NERC/RTP.

Statistical results of model-to-measurement comparisons for these calcula-

tions were included for comparison with the other models.  Calculations  for

AQDM (no variations) also were furnished by Mr.  Turner and were included

for comparison.


2.3  CONCLUSIONS FROM THE PHASE I REPORT

          Conclusions (1-5)  regarding the use of the Sampled Chronological

Input Model (SCIM), a multiple-source Gaussian plume model, to estimate

short-term SO,, concentrations (e.g., 1-hour and  24-hour concentrations)  are

based on model-to-measurement comparisons for 1  month of Chicago, 3 months

of St.  Louis and 1 year of New York City (NYC) data.  The model  was analyzed

using NYC data for three types of inputs, including:  (1) variable emission
                                   -9-

-------
rates, stability classifications and mixing heights  (variable  Q,  S,  H),

(2) mean emission rates and variable stability classifications and mixing

heights (mean Q, variable S, H), and (3)  mean emission rates,  stability

classifications and mixing heights (mean  Q, S, H).   The model  was analyzed

using St. Louis and Chicago data for the  first two  types of input.  For

comparison purposes, an analysis was also made of the use of the  simplified

Gifford-Hanna Model (6HM).


          1.   Comparing the results for  the three  types of input to SCIM,
it is concluded that:


          0    Use of a mean, rather than a variable, emission rate may
               either increase or decrease the root-mean-square error
               (RMSE)  at a receptor but will decrease the correlation
               with measurements (observed at 10 of 10 St.  Louis  receptors
               for 2-hour concentrations, 5 of 8 Chicago receptors for
               1-hour concentrations, and 10 of 10  NYC receptors  for
               1-hour and 24-hour concentrations).

          e    Based on comparisons using NYC data  and mean emission  rates,
               the use of a neutral stability classification and  a mean mixing
               height will decrease the correlation  with measurements  but  will
               also decrease the RMSE at  a receptor  (observed  at  9 of  10  recep-
               tors for 1-hour concentrations and 8  of 10 receptors  for 24-hour
               concentrations).

          o    Based on NYC comparisons,  the combined use of a mean emis-
               sion rate and mixing height and a neutral stability classi-
               fication will decrease the correlation with  measurements
               but will decrease the RMSE (observed at 10 of 10 receptors
               for 1-hour concentrations  and 7 of 10 receptors for 24-hour
               concentrations).


          2.   In evaluating GHM, it was  concluded  that adding point source
contributions (i.e., calculated using SCIM) to GHM  calculations improved
the results for this model.  The RMSE was smaller at 6 of 10 NYC  receptors,
the correlation coefficient was higher at 6 of 10 receptors, and  the standard
deviation of calculated concentrations was closer to the standard deviation
of measured concentrations at all 10 receptors.
                                   -10-

-------
          3.   Comparing SCIM and GHM using hourly calculations and S02
measurements, NYC data, SCIM produced the least annual mean error at 6
of 10 receptors, the closer agreement, between standard deviations of cal-
culated and measured concentrations at 5 of 10 receptors, the least error
in estimating the maximum measured concentrations at 6 of 10 receptors,
and the highest correlation coefficient at 3 of 10 receptors; GHM produced
the least RMSE at all 10 receptors.  Results for comparison of 24-hour S02
concentrations are similar but slightly more favorable to SCIM.

          4.   There is a need to improve the input data used with the
multiple-source Gaussian plume type model, particularly atmospheric sta-
bility information, since the model is very sensitive to the rather gross
changes in stability which are routinely introduced.  SCIM calculations on
the average, greatly overestimated concentrations associated with Turner-
Pasquill stability classes 2 and 5.

          5.   Calculations based on a NYC emission algorithm developed in
this report, particularly when applied with GHM, generally agree with diur-
nal and temperature dependent trends in measured S02 concentrations.  Fur-
ther improvements in this algorithm are desirable but require more detailed
information.


          Conclusions (6-8) regarding the use of several versions of the

multiple-source Gaussian plume model and GHM to estimate long-term mean

concentrations of S02 and particulates are based on model-to-measurement

comparisons for the same data periods and locations.


          6.   The use of variable emission rates for SCIM and GHM are not
able to demonstrate any conclusive improvement in model  validity over the
use of mean emission rates.  It is inferred that this result is due to the
failure to properly treat other causes of variance, such as those associated
with atmospheric stability.

          7.   Based on results for NYC, the Climatological Dispersion Model
(CMD)  and SCIM versions of the multiple-source Gaussian  plume model  produce
a smaller station-to-station RMSE than the Air Quality Display Model  (AQDM)
version (i.e., RMSE's of 52 and 59, respectively, compared to 92, with an
overall  mean of 135 yg/m3 of S02; RMSE's 22 and 22 compared to 36 with an
overall  mean of 82 yg/m3 of particulates).

          8.   Although the NYC validation statistics for GHM, COM,  and
SCIM are similar for S02, GHM results for particulates have a much higher
station-to-station RMSE than do COM and SCIM (i.e., RMSE of 60 compared to
22, with an overall  mean of 82 yg/m3).
                                  -11-

-------
                   Section 3.0



PHASE II WORK REPORTED AND ACCOMPLISHED ELSEWHERE

-------
                              Section 3.0

           PHASE II WORK REPORTED AND ACCOMPLISHED ELSEWHERE


3.1  SCIM COMPUTER PROGRAM AND USER'S MANUAL

          The treatments called for in the contract scope (Section 1.0)

were performed on the SCIM computer program.  The program itself and its

use were documented in GEOMET Report EF-261, as briefly indicated in the

excerpts from that report which follow.

          The Sampled Chronological Input Model (SCIM) is an urban air

pollution simulation.  It is designed to provide the user with a method

of estimating the air quality characteristics of a particular pollutant

over a specified control area.  Both the mean long-term concentration and

the frequency distribution of short-term concentrations are estimated

using conventional emission inventory and meteorological data.

          The objective of User's Manual is to:


          e    Briefly describe the SCIM computer program and its
               intended applications

          e    Provide guidance and sample programs to process
               conventional data into the input forms required by
               the SCIM program

          e    Describe how to set up and operate the SCIM program.


          The SCIM computer program provides the user with a tool for esti-

mating short-term maxima of pollutant concentrations in addition to  long-

term means.   This is done by calculating concentrations for a sample of

short-term periods selected from a specified long-term period.   The  sample

is then used to estimate the long-term mean concentration, the geometric
                                  -12-

-------
standard deviation and the statistical frequency distribution of short-
term concentrations for all specified locations.  The expected annual
maximum concent!  ;tion may be determined from the frequency distributions
or by means of the geometric standard deviation (e.g., see Larsen 1971).
The calculations are made for specified receptor locations.
          The calculations are made using a multiple-source Gaussian
plume model.  Emissions from large stationary sources are represented by
elevated point sources.  All other emissions are represented by an area
source.  Contributions from the area source to concentrations at a recep-
tor are calculated using a numerical technique to evaluate the integral
equation which must be solved.  The narrow plume concept which implies
that crosswind variations in emission rates may be neglected is an impor-
tant assumption in the numerical technique.  This assumption is valid as
long as the distance between variations in the area source emission rate
is large compared to the crosswind diffusion parameter (a ).  A critical
characteristic of the numerical technique is the spacing of grid points
for which emission rates per unit area are determined.  Model sensitivity
findings show that a spacing of one-quarter mile is important in areas of
high spatial variations of emissions.   More generally, spacings of 1 km
or more are satisfactory.
          A significant feature of this program is that varying patterns
of emissions are linked to a chronology of weather observations so that
related variations in emission rates and in the dispersive capability of
the atmosphere can be taken into account.   In the emission algorithm pre-
sented here emission rates are related to ambient air temperature and to
                                  -13-

-------
hour of the day.  This algorithm is especially applicable to emissions
which are related to space heating requirements.  The program has been
tested and found applicable to estimating sulfur dioxide and particulate
air quality characteristics.
          The program inputs are prepared from conveniently available
data, including.Implementation Planning Program (IPP) or Air Quality
Display Model (AQDM) emission data and standard weather data which is
available on punched cards or magnetic tape from the National Weather
Records Center.  The program analyzes the air quality of a region of inter-
est by calculating a sample of hourly concentrations at specified locations.
The user controls the sample size by specifying the sampling interval
between successive hours for which calculations are made.  The standard
program outputs consist of a data file containing the concentrations cal-
culated for each specified location for each selected hour and a printed
statistical summary of the air quality characteristic of each location
and of all locations combined.   In addition, the user may choose to use a
version of the program which will generate a Source Contribution File in
the correct format to interface with IPP.

3.2  TRAINING OF EPA PERSONNEL
          The final  training of EPA personnel  called for in the  contract
scope was provided at EPA by GEOMET staff in July of 1973.   Ten  to fifteen
staff members of the Office of Air Quality Planning and Standards and of
the Meteorology Laboratory were given  instruction  in  the  use  of  the  pro-
gram.  This instruction  was augmented  by  subsequent  extensive  interaction
by phone and in person between  GEOMET  and  EPA  staff.
                                   -14-

-------
                  Section 4,0

CONSIDERATIONS OF EMISSION DATA, EMSU DATA AND
       SHORT-TERM MAXIMUM CONCENTRATIONS

-------
                               Section 4.0

             CONSIDERATIONS OF EMISSION DATA, EMSU DATA AND
                    SHORT-TERM MAXIMUM CONCENTRATIONS
4.1  DIURNAL AND SEASONAL VARIATIONS IN EMISSIONS

          The SCIM program is primarily designed to analyze air quality

levels associated with emissions of stable pollutants such as sulfur

dioxide, particulates and carbon monoxide.  The emissions from any given

source will vary with hour of the day, day of the week and season of the

year.  Standard emission factors have been developed for most pollutants

which allow estimates of emission rates to be established as a function

of fuel consumption rates or of processing rates for various industrial

activities.  When these fuel consumption rates and processing rates are

described as functions of times, the emission rate of each pollutant is

well defined.

          Unfortunately, information on variations of emissions with time

are not usually available.  However, when emissions result from the con-

sumption of fuel for space heating, the emissions will vary with temper-

ature.  Variations in these emissions with time can be estimated from

local temperature records which are available for almost all locations.

The consumption of fuel for space heating accounts for a certain percent-

age of the emissions of a pollutant from a particular source.  A general

algorithm used in SCIM which describes emissions as a function of
                                   -15-

-------
I

            parameters which can be related to temperature and other activity indexes
•          is the following:

|                    Q(t) = QA K(t)
                      K(t) = (1-F) A(t) + F [H(t) - T(t)] S(t),      T(t) < H(t)
                      T(t) >.H(t)      K(t) = (f-F) A(t).
            where
                      Q(t) = emission rate at time t
                        Q. = average annual emission rate
                      K(t) = time dependent emission factor
                         F = fraction of emissions which result from space heating
                             requirements
                      A(t) = activity factor which defines activity level for time t
                             relative to annual average activity level for activities
                             which control emissions not related to space heating
                             requirements
                      H(t) = temperature threshold for space heating requirements for
                             time t
                      T(t) = temperature at time t
                      S(t) = sensitivity factor which defines rate of emission per
                             degree below temperature threshold at time t relative
                             to annual average rate of emission per degree below
                             temperature threshold.
                      In the above algorithm the parameters A(t), H(t) and S(t) may
            vary with time of day, day of the week and week of the year.  The informa-
            tion required to determine these parameters as functions of time for every
            point source and every square mile of an area source is far too detailed


-------
for what is normally economically feasible to collect and analyze.  However,



it may be useful to derive city-wide parameters which can be applied to



area sources.



          The ideal data for estimating the above parameters would be fuel



consumption records and process operating records for a large number of



sources.  Lacking this, other less desirable data might be used.   In the



Phase I report of this project, a large set of S0? concentration  measure-



ments was used (12 years of almost continuous hourly observations) for



New York City.  Blade and Ferrand (1969) summarized these measurements by



hour of the day, day of the week, and week and month of the year.  The mean



hourly SO^ concentrations for each month of the year were correlated with



mean hourly temperature for each month of the year for the same data period.



Following the methods described in the Phase I report (Appendix A), the



parameter values presented in Table 4.4-1 were developed.  On the basis



of this same data, it was estimated that 29 percent of the emissions are



dependent on temperature variations (i.e., F = 0.29).



          SCIM is programmed to use the above algorithm and the parameter



values in Table 4.1-1 to estimate diurnal variations in area source emis-



sions.  There are some drawbacks to using this data.  The parameter values



for the emission algorithm are specifically applicable to S02 emissions in



New York City.  It is not known how applicable these are to other cities.



Furthermore, since the emission rates were derived from S02 measurements



the parameter values may contain diurnal variations which are associated



with diurnal variations in meteorological conditions.  The diurnal varia-



tion in activity factors shown in Table 4.1-1 does not make much  sense
                                   -17-

-------
Table 4. 1-1. Emission Parameters Developed from New York City SO2 Data
Hour of
the Day
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
A ctivity
Factor, A(A)
1.0272
1.0008
0. 9576
0. 9576
1.00344
1.1784
1.3032
1.3200
1.2624
1.1616
1.0104
0.9336
0. 8760
0. 8232
0.8160
0.8160
0.8232
0. 8424
0.8832
0. 9264
0.9744
1.0104
1.0272
1.0344
Temperature
Threshold, H(t), °F
56
55
55
55
56
58
59
61
63
64
65
65
65
65
65
65
65
65
65
65
65
64
62
60

Sensitivity
Factor, S(t), "^F"1
0.0330
0.0280
0.0280
0. 0293
0.0371
0.0717
0. 1208
0. 1437
0.1214
0. 0977
0. 0893
0.0841
0.0841
0.0852
0. 0856
0. 0886
0. 0983
0. 1060
0.1120
0.1142
0.1114
0. 1024
0. 0696
0.0416
Mean = 0.0826
                                 -18-

-------
when considered in terms of normal variations in business activities.  The
validation analysis reported in the Phase I report showed that, when annual
averages of measured and SCIM calculated concentrations were compared for
different hours of the day, the SCIM calculations overestimate the measured
value by the greatest amount at 7 A.M. and underestimate by the greatest
amount at 1 P.M. and 4 P.M.  These correspond to maximum and minimum values
of A(t), respectively.  These results suggest that a uniform value of A(t) = 1
may be more appropriate than the values in Table 4.1-1.  It is therefore recom-
mended that SCIM be run with A(t)  = 1, rather than the values in Table 4.1-1.
          The sensitivity factors shown in Table 4.1-1 have a logical basis
when considered in terms of people's diurnal activities.  The sensitivity
factors are highest in the early morning with a peak value at 7 A.M.  This
is when a major portion of the population arises.  Residential fuel con-
sumption and probably certain commercial and industrial fuel consumption
is increased greatly relative to other hours of the day.  Therefore, the
sensitivity of fuel consumption relative to the temperature deficit from
the threshold space heating temperature is likely to be very great.  There
is a secondary maximum in the sensitivity factor in the early evening hours.
This corresponds with the time that residential fuel consumption is likely to
be adjusted to temperature changes (i.e., the end of the working day when
people return to apartments and homes).  The low sensitivity factors in the
early morning hours after midnight are times of minimum residential fuel
consumption and low heat replacement in commercial and institutional sources
due to opening doors.  Thus, there is a qualitative basis for using the
sensitivity factors shown in Table 4.1-1.  Although the sensitivity factors
                                    -19-

-------
 and temperature  thresholds  in  Table  4.1-1 were  developed  for  New  York  City,
 they are  probably qualitatively  applicable  to other  cities.   If other
 information  is not available,  they are  a  reasonable  approximation to what
 can be  expected  in other large cities.
           The  fractions  of  emissions which  are  temperature  sensitive is
 likely  to be variable  from  one city  to  another,  depending primarily on
 how cold  the climate is.  One  gross  assumption  which could  be made is
 that the  fraction of SCL emissions which  are temperature  sensitive is
 directly  proportional  to the climatological mean degree days which occur
 at  a given location.   However, it is recommended that estimates of the
 fuel use  and climatological mean degree days be obtained  for  several dif-
 ferent  climates  before attempting to define such a relationship.  One  other
 source  of data is available from a study  by Argonne  National  Laboratory
.(Roberts,  et. al  1970).   From  this report it is  estimated that 72 percent
 of  S0?  emissions  in Chicago are  temperature sensitive.  By way of compari-
 son, it is noted  that  the annual mean degree days are approximately 5000°F
 days and  6200°F  days for New York City  and Chicago,  respectively.  The
 two available estimates  of  29  percent for 5000°F days and 72  percent for
 6200°F  days  are  not very consistent. ' Of course, other factors, such as the
 relative  mix of  industrial, commercial  and residential fuel users in the
 area sources, affect the relationship.  More data on the  amount of fuel use
 which is  related  to temperature  considerations  is needed.   For the time
 being,  one might  reasonably assume that, for large cities with normal  total
 heating degree days (with a base of  65°F) of 5000 to 6000 degree  days, S02
 emissions  from area sources are  50 percent temperature sensitive  and use
 the temperature  thresholds  and sensitivity factors in Table 4.1-1.

                                   -20-

-------
4.2  USE OF EMSU DATA AS METEOROLOGICAL INPUT                   ;



          During the late 1960's and early 1970's, Environmental Meteor-



ological Support Units (EMSU) were established by the National Oceanic



and Atmospheric Administration in roughly 20 U.S. cities.  The purpose



of these units was to take observations, prepare forecasts, and provide



advice on the present and future meteorological conditions which 'affect



air pollution levels.  An analysis is made in this study of how data



reported by EMSU's could be used to determine meteorological parameters



for SCIM and how the selected values compared with values determined from



conventional airport weather observations.  Three meteorological param-



eters analyzed were mixing height, atmospheric stability and wind speed



and direction.



          The EMSU data consist of radiosonde observations of temperature,



relative humidity and wind direction and speed from a slow rise balloon



(i.e., about 65 meters per minute).   The soundings are taken from urban



areas and generally provide useful estimates of the temperature, moisture



and wind profiles over large cities.  The soundings are generally taken



at times of expected minimum (near sunrise)  and expected maximum (early



afternoon) dispersion conditions.  Additional soundings may also be



available for intermediate hours.





4.2.1  Mixing Height



          Mixing heights were calculated for EMSU (slow rise)  radiosonde



data and for standard radiosonde data using  the method described in



Appendix A-l.   The data used included all  days  in August and December
                                  -21-

-------
of 1969 for which both EMSU and standard RAOB data were available for
New York City and St. Louis.
          New York City represents a site at which both standard and
EMSU data are available for the same city.  The EMSU data are obtained
from releases at Laguardia Airport which is located well within the
New York Metropolitan area.  The standard data are obtained from releases
at Kennedy Airport which is located on the edge of the metropolitan area.
Mixing heights corresponding to EMSU observation times are determined from
the standard data using an interpolation scheme described in Appendix A-2.
The 12Z mixing height is taken to be representative of 0600 local time
and the OOZ mixing height is taken to be representative of 1400 local time.
Linear interpolation with time is used between 0600 and 1400.  The com-
puted mixing heights, interpolated values and differences between mixing
heights calculated using EMSU and standard RAOB data are presented in
Table 4.2-1.
          St.  Louis represents a site at which standard RAOB data are not
available, but for which EMSU data are available.  The mixing height for
St.  Louis was  estimated using an average of heights calculated for Peoria,
Illinois, and  Columbia, Missouri.   The computed mixing heights, calculated
averages, interpolated values and EMSU less standard RAOB differences are
presented in Table 4.2-2.
          The  question of whether the EMSU data provides information about
mixing heights which is significantly different from that available from
standard radiosondes may be examined using the data in Tables 4.2-1  and
4.2-2.   For each location and each month, the difference between mixing
                                  -22-

-------
            Table 4.2-1.  New York City Mlxjng Height Estimates from Standard RAOB and EMSU Data
Date (1969)
August 1

August 4
August 5

August 6

August 7

August 8

August 11

August 12

August 13

August 14

August 15

August 18

August 19

August 20

Hour
0600
0700
1100
1900
0700
1100
1900
0600
0700
1100
1900
0600
0700
1100
1900
0600
0700
1100
1900
0500
0700
1100
1900
0600
0700
1100
1900
0600
0700
JOOO
1900
0600
0700
1100
1900
0600
0700
1100
1900
0600
0700
1100
1900
0600
0700
1100
1900
0600
0700
1100
1900
0600
0700
1100
1900
Kennedy Airport RAOB
(100)*
107
(2446)
4200
107
(2446)
4200
(4200)
4200
(4200)
4200
(4200)
4200
(1861)
107
(118)
156
(307)
420
(628)
567
(325)
143
(238')
261
(352)
420
(219)
219
(216)
213
(119)
138
(214)
270
(100)
100
(255)
372
(4200)
4200
(4200)
4200
(100)
107
(203) '
275
(1303)
1172
(647)
254
•(100)
423
(2581)
4200
Laguardia Airport EMSU
100
763
100
125
100
187
244
2248
100
791
100
138
100
100
100
881
100
2770
100
172
126
106
100
156
127
930
100
1027
EMSU Minus RAOB
0
-1683
-7
-2321
-4100
-4013
-3956
387
-18
484
-528
-187
-138
-252
-119
665
-19
2556
0
-83
-4074
-4094
0
-47
-1176
282
0
-1554
* Values in parentheses are interpolated (see text).
                                                                                         (Continued)
                                                -23-

-------
        Table 4.2-1.  New York City Mixing Height Estimates from Standard RAOB and EMSU Data (Continued)
Date (1969)
August 21



August 22



August 25



August 26



August 27



August 28



August 29



December 1


December 2


December 3


December 4


December 5


December 8



December 9


December 10


Hour
0600
0700
1200
1900
0600
0700
1100
1900
0600
0700
1100
1900
0600 '
0700
1100
1900
0600
0700
1100
1900
0600
0700
1100
1900
0600
0700
1100
1900
0700
1200
1900
0700
1200
1900
0700
1200
1900
0700
1200
1900
0700
1200
1900
0700
0800
1200
1900
0700
1200
1900
0700
1200
1900
Kennedy Airport RAOB
(292)*
452
(1253)
1573
(100)
168
(93S)
1510
(105)
167
(417)
605
(4200)
4200
(4200)
4200
(642)
643
(649)
653
(162)
180
(255)
310
(155)
149
(125)
107
382
(1666)
2179
797
(1027>
1119
277
(2522)
3420
1057
(1476)
1644
599
743
800
481
(449)
(320)
255
519
(575)
597
285
(153)
100
Laguardia Airport EMSU
100

1281

100

1486

100

671

263

735

190

1194

115

846

192

782


589

561
646

403
1257

1060
1381

913
1274


428
453

544
673

351
352

EMSU Minus RAOB
-192

28

0

551

-5

254

-3937

-3465
«
-451

545

-47

592

37

657


-1077

-236
-381

126
-1265

3
-95

314
530
•

-21
134

26
99

66
199

* Values in parentheses are interpolated (sec text).
                                                                                            (Continued)
                                                 -24-

-------
       Table 4.2-1.  New York City Mixing Height Estimates from Standard RAOR and EMSU Data (Concluded)
Date (1969)
December 11
December 12


December 15


December 16


December 17


December 48


December 19


December 23


December 24


December 29


December 30


Hour
0700
0700
1200
1900
0700
1200
1900
0700
1200
1900
0700
1200
1900
0700
1200
1900
0700
1200
1900
0700
1200
1900
0700
1200
1900
0700
1300
1900
0700
1400
1900
Kennedy Airport RAOB
400
117
(557)*
733
1172
(546)
295
1154
(1135)
1128
878
(1019)
1076
332
(1078)
1377
3342
(26S3J
2377
448
(547)
587
440
(332)
289
460
(5061
514
171
451
451
Laguardia Airport EMS'!
436
221
3638

1103
1240

100
1327

618
1104

219
370

2382
2644

422
743

560
377

531
147

224
327

EMSU Minus RAOB
36
105
3081

-68
693

-1054
192

-260
85

-113 .
-709

-960
-9

-26
196

120
45

71
-359

53
-124

* Values in parentheses are interpolated (see text).
                                           -25-

-------
                  Table 4.2-2. St. Louis Mixing Height Estimates from Standard RAOB and ESMU Data



Date (1969)
August 11


August 12



August 13


August 14


August 18



August 19


August 20


August 21


August 22


August 25




August 26



August 27



August 28



August 29





Hour
0600
1300
1800
0600
1000
1400
1800
0600
1300
1800
0600
1300
1800
0600
1100
1300
1800
0600
1300
1800
0600
1300
1800
0600
1300
1800
0600
1200
1800
0600
1000
1300
1400
1800
0600
1000
1300
1800
0600
1000
1400
1800
0600
1000
1400
1800
0600
1300
1800
RAOB

Peoria
Sounding
227

3269
236


3473
231

725
260

3464
363


3338
255

3163
401

3945
443

1333
250

1336
213



2223
217


2214
228


426
225


3412
233

812
Columbia
Sounding
259

3314
310


2033
294

800
298

3071
316
3093

3230
309

25?5
304

964
777

4056
284

1317
276



3813
280


3574
310


4014
291


3704
284

1565

Average
243
(2910)*
3292
273
(1513)
(2753)
2753
263
(700)
763
279
(2894)
3268
340
3093
(3270)
3284
282
(2523)
2844
352
(2191)
2454
610
(2434)
2694
267
(1062)
1326
245
(16il)
(2671)
(3018)
3018
249
(1571)
(2563)
2894
269
(.1244)
(2220)
2220
258
(1908)
(3558)
3558
258
(1072)
1189


EMSU
Sounding
170
2780

168
1301
162

211
567

192
2626

322

3021

336
2467

345
2345

2820
1774

165
154

155
1736
1865
251

157
1926
2005

178
2018
2002

170
3050
2013

173
3080



Minus
Average RAOB
-72
-130

-105
-212
-2591

-51
-133

-87
-268

-18

-249

54
-57

-7
153

2210
-660

-101
-908

-90
105
-806
-2767

-92
355
-559

-91
773
-218

-87
1142
-1545

-89
2008

* Values in parentheses are interpolated (see text).
                                                                                                  (Continued)
                                                   -26-

-------
              Table 4.2-2. St. Louis Mixing Height Estimates from Standard RAOB and EMSU Data (Concluded)



Date (1969)
December 4


' December 5


December 8


December 9


December 11


December 12


December 15


December 16


December 17


December 18


December 19


December 22


December 23


December 24





How
0600
1300
, 1800
0600
1300
1800
0600
1300
1800
0600
1300
1800
0600
1200
1800
0600
1300
1800
0600
1300
1800
OfiOO
1200
1800
0600
1300
1800
0600
1300
1800
0600
1300
1800
0600
1200
1800
0600
1200
1800
0600
1200
1800
RAOB

Peoria
Sounding
299

284
302

557
492

246
301

1414
733

1214
559

805
344

566
231

396
841

712
580

920
483

1238
529

639
902

1552
294

612
Columbia
Sounding
285

1014
544

809
307

341
375

355
821

1196
902

412
988

1070
379

663
799

472
826

444
293

1062
998

1044
1398

1297
912

603

- Average
292
(604)*
649
423
(651)
683
400
(307)
293
338
(816)
885
777
(1098)
1205
'730
(624)
608
666
(799).
818
305
(474)
530
820
(621)
592
703
(648)
682
388
(105S)
1150
764
(822)
841
1150
(1356)
1425
603
(607)
608

St. Louis
EMSU
Sounding
216
928

455
897

942
1313

257
1074

1737
717

1098
1003

921
1109

426
802

318
786

1085
914

345
1794

1175
965

1207
1851

305
753


St. Louis
Minus
Average RAOB
-75
324

32
246

543
1006

-81
258

960
-381

368
380

255
310

121
327

-502
164

381
266

-43
740

412
143

57
495

-297
146

* Values in parentheses are interpolated (see text).
                                                  -27-

-------
heights estimated from EMSU and standard RAOB data is summarized in
Table 4.2-3 for sunrise and for mid-day observation times.  Overall
these comparisons show that the mean differences (-205m) is about
20 percent of mean mixing height (1044m).  However, there is a large
amount of variability for individual comparisons as demonstrated by
the large root mean square difference of 1146m.   These results sug-
gest that for the overall climatological average, the EMSU information
may not be important.  However, for day to day variations, there is
important information available in the EMSU data.  Mixing height is
most important in determining dispersion conditions during the day.
It is less important near sunrise when stable or neutral stability
conditions prevail.  An examination of the data  in Tables 4.2-1  and
4.2-2 shows that in 46 percent (18 comparisons)  for New York City
and 47 percent (16 comparisons) for St.  Louis, the EMSU mixing height
estimate differs from that derived with standard RAOB data by over
50 percent of the standard RAOB estimate.  As a  result, it is con-
cluded that, when EMSU data is available, it should be used in place
of or to supplement the standard data.
          The following procedures are suggested for using the EMSU
data.  Use the interpolation scheme described in Appendix A-l as a model
of diurnal variation in mixing height.   The following steps may  be
followed.
          1.   If a sunrise EMSU sounding is  available,  use  the  mixing
              height from it for the period  from Midnight to 6  a.m.
              If not, use an estiamte from standard RAOB data.
                                  -28-

-------
w


I,
•F-(
 o
X
 M
 a
s
M
o


2

U
 rt

"§
 rt
•M
oo

T3
00

S
«

 a
 a
 0)
 

              00
               U
               o
               c
           0  H   ^


                               S    5:
                               a)    u
                                                -g
U    O

M    X
 ^    u
 O    O
JX    {M


 «:    ^
                                                              .
                                                              -g
                               H
                               0
                               H
                                                                    -29-

-------
          2.  If one or more mid-day EMSU soundings are available,
              obtain mixing heights from each.  Use linear inter-
              polation with time to estimate mixing heights for
              hours between EMSU soundings.

          3.  If more than one EMSU sounding is available these may
              be linearly extrapolated with time to estimate mixing
              heights over the period from 6 a.m. to 2 p.m.

          4.  If only one EMSU sounding is available, compute the
              mixing height for other hours in the period from
              6 a.m. to 2 p.m. by substituting the hourly surface
              temperature for the surface temperature in the sounding
              and computing the mixing for the revised sounding.


4.2.2  Stability Class

          Since EMSU sounding data is obtained from a slow rise bal-

loon, it should be useful in characterizing the temperature and wind

profiles of the lowest layers of the atmosphere, which determine dis-

persion conditions.  Several ways of characterizing the stability of

the atmosphere using EMSU data are compared with the Pasquill stability

classes determined by a method suggested by Turner using surface wea-

ther observations of cloud cover and wind speed.  The extent to which

the EMSU data suggest stability classifications different from the

Turner-Pasquill  categories is discussed.   In conclusion a method of

integrating the two types of data is'proposed.

          Bulk Richardson number, calculated over three different

heights, was used to characterize stability.   The three heights were

from the lowest height in the EMSU sounding with both wind and temper-

ature data to (1) the top of the mixing layer, (2) 140 meters, as used

by McElroy (1969) to classify measurements of a  and a , and (3) the

next lowest height with wind and temperature data.  The'wind speed data
                                  -30-

-------
within the mixing layer were fitted to a power law profile by the method
of least squares.  This too was used to characterize the atmospheric
stability.  These estimates were obtained from St. Louis EMSU soundings
for August and December 1969.  The three bulk Richardson numbers, the
wind speed profile power law and the mean mixing layer wind speed and
direction, for each EMSU sounding with reasonably complete data, are
listed in Table 4.2-4.  For comparison the Pasquill stability class and
surface wind determined from the closest (in time) three-hour surface
weather observation are also listed.
          In order to compare the sounding stability characteristics with
the Pasquill stability classes, the correspondence shown in Table 4.2-5
was assumed.  The correspondence in Table 4.2-5 is hypothetical  and was
selected to be reasonably consistent both with the data shown in
Table 4.2.4 and with information reported by other investigators.  Using
these correspondences, the best agreement between EMSU stability data and
the Pasquill stability classes (listed in Table 4.2-4) is obtained using
the bulk Richardson number over 140 meters.   This gives 19 hours out of
38 in agreement.   The next best was the bulk Richardson number over the
lowest 2 heights, which gives 18 hours of agreement out of 44 comparisons.
          Since 50 percent of the compared hours differ between  EMSU and
surface data stability classifications, there is probably a significant
amount of additional information available in the EMSU data.   However,
it is difficult to see how to use the EMSU data, except to modify the
single hour for which EMSU stability classifications are obtained.   The
stability changes so rapidly from hour to hour during ti.e periods over
                                  -31-

-------
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
I
I
I
I
Table 4.2-4.  Stability and Wind Characteristics for St. Louis
Date
(1969)
Aug. 25
Aug. 25
Aug. 25
Aug. 26
Aug. 26
Aug. 26
Aug. 27
Aug. 27
Aug. 27
Aug. 28
Aug. 28
Aug. 28
Aug. 29
Aug. 29
Dec. 4
Dec. 4
Dec. '-5
Dec. 5
Dec. 8
Dec. 8
Dec. 9
Dec. 9
Dec. 10
Dec. 11
Dec. 12
Dec. 12
Dec. IS
Dec. 15
Dec. 16
Dec. 16
Dec. 17
Dec. 17
Dec. 18
Dec. IS
Dec. 19
Dec. 19
Dec. 22
Dec. 22
Dec. 24
Dec. 24
Dec. 30
Dec. 30
Dec. 31
Dec. 31
Hour
1000
1200
2000
0500
0900
1200
0500
1000
1300
0500
1000
1400
0500
1300
0500
1200
0500
1200
0500
1200
0500
1200
1200
1200
0500
1200
0500
1200
0500
1200
0500
1200
0500
120O
0500
1200
0500
1200
0500
1200
0500
1200
OSOO
1200
Bulk Richardson Number
Over
Mixing
Layer
1.690
-0. 550
0.036
0.146
0.136
-0,632
0.851
4.620
1.125
0.119
7.968
0.200
0.090
0.389
0.027
0.016
0.031
0.051
0.271
0.051
0.026
0.294
0.238
-0.044
0.138
-0.012
0.168
0.039
0.322
-0. 007
0.030
-0. 795
0.020
-0. 035
-0.022
0.094
-0.53S
0.065
0.226
0.069
0.308
-0.544
-1.904
-5.829
Over 140 m
Layer
0.072
-
0.044
0.101
0.131
-0.038
1.028
-0.134
-0.019
0.109
-0.007
0.003
0.097
0.029
0.033
-0. 070
0.009
0.005
0.038
0.010
0.030
0.016
-0.004
-0.007
0.010
-0. 060
0.004
-
0.177
-0. 180
0.034
-0.057
-0. 264
-0. 220
0.091
-
0.330
0.670
-
-
0.233
0.112
0.226
-
Lowest Two
Sounding
Heights
0.009
0.015
0.003
0.149
-O.069
-0.079
0.307
-0. 156
-0. 103
0.073
-0.016
-0.040
O.OS8
-0.043
0.017
-0.060
-0.002
-0.004
0.021
-0.007
0.026
0.007
-0.083
-0.001
0.007
-0. 077
-0.002
-0.038
0.094
-0.138
0.030
-0. 165
0.002
-0.008
-0.012
-0.063
0.170
0.330
0.167
-0.066
0.184
0.035
-0.005
-0.315
Pasquill
Stability
Class
B
C
E
E
C
A
E
B
A
D
D
B
E
D
E
C
D
D
D
C
D
D
D
'D
D
D
E
D
E
C
E
D
D
D
E
D
D
D
D
D
D
D
D
D
Wind Speed
Profile Power
Law
-0.06
0.06
0.28
0.22
0.15
-0.01
0.08
-0.06
0.14
0.34
-0.04
0.06
0.28 '
0.03
0.05
0.04
0.15
0.08
0.17
0.14
0.33
0.17
0.32
0.05
0.08
0.16
O.06
0.07
0.19
0.02
0.38
0.01
0.33
0.08
0.17
0.28
0.10
0.15
0.82
0.26
0.04
0.02
0.08
0.18
Win 1 Lvirection/
Speed (m/sec)
Mixing
Layer
Mean
47/4
65/5
59/6
48/5
70/6
62/5
133/2
169/3
162/7
203/6
228/8
179/9
223/8
210/6
32/12
42/7
157/14
155/11
269/13
254/13
190/11
200/16
69/10
298/14
272/5
301/7
13/19
355/10
85/4
180/8
355/9
175/6
253/15
324/13
318/11
313/21
356/10
158/8
90/3
184/11
7/14
352/13
315/10
271/4
Surface
10/3
10/4
30/3
calm
350/3
70/2
calm
160/5
190/2
calm
160/5
180/2
150/2
200/4
30/3
40/4
130/6
130/6
240/4
230/5
160/4
170/7
30/3
290/7
260/4
310/4
10/3
350/6
60/1
160/3
260/3
130/3
250/5
280/6
230/3
300/4
300/S
100/4
100/4
140/5
320/5
320/S
260/5
280/4
                  -32-

-------
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 1
I
I
I
which EMSU data  is  available  that the  surface data is much better that

an extrapolation of  EMSU  data.   The EMSU data has limited use.  The data

would be more useful  if  soundings were available every three hours.



     Table 4.2-5. Proposed Correspondence Between Three Types of Stability Classifications
Time of Day
Day
Day
Day
Day
Night
Night
Pasquill Class
A
B
C
D
D
E
Bulk Richardson Number
< -0. 05
-0.05 to -0.031
-0.03 to -0.011
>-0.01
< 0.01
> 0.01
Wind Speed Profile Power Law
<0.05
0.05 toO. 12
0. 13 to 0. 17
>0. 17
<0.22
>0. 22
          Another uncertainty with  stability data related to vertical

temperature and wind profiles is  that  the  relation of this data to the

commonly used Pasquill dispersion parameters is  not well  established.

In the light of the above  considerations,  it is  recommended that EMSU

data not be used to determine stability  characteristics.


4.2.3  Hind Direction and  Speed

          The data given in Table 4.2-4  shows  comparisons between the

surface wind speed and direction  and the vector  mean wind speed and

direction for the mixing layer.   It is clear from these comparisons that

the EMSU data provides significant  additional  information on the wind

profile.   Of particular interest  is the  frequent occurrence of a notice-

able turning of the wind with height.  This  can  have a  significant

effect on model calculations.  The  need  for  a  detailed  study of how to
                                  -33-

-------
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
 I
I
B
I
I
I
use wind profile data in estimating the wind direction and speed used

in model calculations is clearly indicated by this data.  No attempt

has been made to develop general techniques from the limited data pre-

sented in Table 4.2-4.  However, the data does suggest some possible

ways of using EMSU data to improve the wind direction and speed estimates

used in modeling.  During some periods the vertical wind shear remains

nearly fixed from one EMSU sounding to the next.  This suggests that an

average shear could be derived and applied to all surface wind observations

between the EMSU observation times.  Another possibility is to develop

diurnal patterns of wind shear from other data sources (e.g., St.  Louis

micromet tower data of 1964)  and use the EMSU data to identify and scale

the patterns.   The turning of the wind with height should be taken into

account in dispersion models.

          For the present it is suggested that the following tentative

procedure be used to acco""it for turning of the wind.


          1.  Determine the mean wind direction for the mixing layer
              for each EMSU sounding.   If the sources being modeled
              are mostly elevated sources, use this as the wind dir-
              ection;  if the sources are mostly ground sources, use
              the surface wind direction.

          2.  If the succeeding EMSU sounding is less than 24-hours
              away and the mixing layer mean  wind direction has changed
              by less  than 90°, estimate the  wind direction for inter-
              vening hours by linear interpolation.   Use these wind
              directions  in place of the surface wind direction if
              mostly elevated sources  are being modeled.   If the wind
              direction has shifted by 90° or more or ground sources
              are being modeled use the reported surface wind directions
              for intervening hours.   If the  time period between  sound-
              ings exceeds 24 hours,  use the  reported surface wind
              directions  for  intervening hours.
                                  -34-

-------
4.3  PROCEDURE FOR CALCULATING ANNUAL SHORT-TERM MAXIMUM CONCENTRATIONS





4.3.1  Introduction



          The operational implementation of the Gaussian plume model  of



this study for purposes of evaluating proposed air quality strategies



could impose a severe computational burden.  The model  might be used  to



calculate all hourly concentrations within a long-term  period, e.g.,  one



year, for which both emission and meteorological data are available.   This



is repeated for each of a number of stations in each of many control



regions for each of several  proposed air quality control strategies.   A



statistical sampling procedure was devised to reduce the amount of computa-



tions and was tested on 10 stations in New York City,  the procedure  con-



sists of reducing the number of hourly concentrations calculated; the



number of stations, air quality regions, and control strategies do not



change.  The reduction is achieved by selecting a sample of the variable



hours in a systematic manner and using the sample to calculate the param-



eters required for evaluating air quality strategies (e.g., mean annual



concentration, daily value exceeded only once a year, etc.).



          The test on the 10 stations consisted of choosing samples of



various sizes and determining the loss in accuracy is given by the dif-



ference between a parameter value calculated from all available hours and



the value of this same parameter calculated from a sample.  Six air quality



control parameters were chosen for analyses, and differences were obtained



for each of 29 sample sizes.  Tables and graphs of a function of these



differences, are presented in a fashion to provide information on the



tradeoff between reduction of computations and loss in  accuracy.  These
                                  -35-

-------
I
1
            furnish  guidance  for choosing  a  sample  size, and  thereby reducing compu-
|          tations,  in  any operational  implementation of  the model for evaluating
.          air  quality  strategies.
                      The  test  procedure,  and  the results  of  the  test are described
•          below  including:  method  of  calculating  the six air quality control param-
            eters, the sampling scheme,  and  development of the function of the dif-
|          ferences  which serves  as  an  overall measure of accuracy.
I
4.3.2  Test Procedure

4.3.2.1  Air Quality Parameters
          In current EPA practice, it is generally assumed that air pollu-
tion concentration values follow the log-normal  distribution.   This assump-
tion was adopted in this study.  Although other  distributions  have been
advanced, and may in fact eventually replace the log-normal  assumption,  it
is our opinion that results obtained here would  not be changed substantially.
For any set of hourly concentrations (e.g.,  all  2920 third hours  in a  year
or a sample thereof), the log-normal distribution was fitted by calculating
the mean, 7, and the standard deviation, Sy, of  the logarithms of the
concentrations.  Three air quality values were then derived  as follows:

          Mean = exp (7 + 0.5 sy2)
          Value exceeded once in 1000 hours  = exp (Y + 3.091  SY)
          Value exceeded once in 2920 hours  = exp (Y + 3.396 sv)
                                               -36-

-------
          Three additional values were calculated in a similar manner



from daily mean concentrations, i.e., the arithmetic mean of the eight



hourly concentrations for the day.  The log-normal distribution was fitted



to these means; the parameters are:





          Mean = exp (I + 0.5 sz2)



          Daily value exceeded once in a 100 day = exp (I + 2.33 sz)



          Daily value exceeded once in a year = exp (I + 2.776 s^)





Where Z and s7 are, respectively, the mean and standard deviation of the



logarithms of the daily values.





4.3.2.2  Sampling Procedure



          The sampling procedure for the daily values (average of 8 three-



hourly values) is presented first because it is simpler.   Two terms require



definition:  a sampling interval is the number of values  from one selected



value to the next (e.g., a sampling interval of two means that every other



day is included in the sample); the initial  time indicates the starting



day of the sample.  Twenty-nine sampling intervals were chosen, each value



from 2 through 30.  For each interval, from 2 to 8 different samples were



selected by varying the initial time.  Thus, for a sampling interval of



2, two samples were drawn:  the" first consisted of days 1, 3, 5, ...,



365 and the second of days 2, 4, 6, ..., 364.   The number of samples for



a sampling interval  is given by the maximum of [sampling  interval  or eight].



Thus, for sampling intervals up to eight the number of samples  is  equal  to



the sampling interval; for sampling intervals  beyond eight exactly eight
                                   -37-

-------
I

I
            samples were drawn from the 365 days by varying the initial time from
•          1 through 8.  The value of eight has no significance; it simply reduces
•          the amount of computation.
                      When the sampling procedure is applied to the 3-hourly data
•          some unequal sampling intervals may result, e.g., for interval  two every
 -          other value within one day is selected, but from the last value of one
™          day to the first value of the next, the interval is either 1  or 3.  How-
fl          ever, it was deemed more important to ensure equal  representation of each
            of the eight hours of the day than to maintain a consistent sampling
•          interval.  Again, 29 sampling intervals were used but this time they do
^          not proceed by steps of one but range from 2 to 249.  They are  listed in
"          Table 4.3-1.  As before, for each interval from two through eight the
ft          number of samples equals the sampling interval and  beyond eight exactly
            eight samples were selected.
I
            4.3.2.3  Measure of Accuracy
•                    At each of the 10 stations, three air quality parameters were
            calculated using all 2920 hourly concentrations and an additional three
            using the 365 daily mean concentrations.  For each  sample selected from
f*          the 2920 hourly values, three air quality parameters were calculated and
            differences were taken between them and the corresponding parameter values
            using all 2920 hours.   The same procedure was followed for samples drawn
            from the 365 daily concentrations.   The differences were combined to obtain
            a measure of accuracy for each sampling interval for each air quality
            parameter.
                                              -38-

-------
Table 4.3-1. Measures of Accuracy - Hourly Concentrations
Sampling
Interval
2
• 3
4
5
6
7
9
11
13
IS
17
19
21
23
25
27
29
31
39
47
55
63
71
79
95
119
159
199
249
Average No.
of Cases
in Sample
1460
973
730
584
487
417
325
266
225
195
172
154
139
127
117
108
101
95
75
63
54
47
42
37
31
25
19
15
12
Proportion
of Cases
in Sam pie
0.500
0.333
0.250
0.200
0.167
0.143
0.111
0.091
0.077
0.067
0.059
0.053
0.048
0.043
0.040
0.037
0.034
0.032
0.026
0.021
0.018
0.016
0.014
0.013
0.011
0.008
0.006
0.005
0.004
Measures of Accuracy for
Mean
0.011
0.015
0.032
0.045
0.033
0.056
0.050
0.057
0.056
0.125
0.101
0.077
0.113
0.109
0.080
0.115
0.122
0.106
0.127
0.189
0.232
0.157
0.173
0.260
0.201
0.336
0.234
0.336
0.403
1/1000
0.020
0.024
0.082
0.102
0.077
0.166
0.097
0.144
0.149
0.323
0.287
0.218
0.257
0.380
0.216
0.261
0.258
0.257
0.312
0.406
0.812
0.408
0.535
0.491
0.476
0.810
0.615
0.976
0.842
1/2920
0.023
0.026
0.097
0.113
0.085
0.195
0.109
0.165
0.175
0.376
0.336
0.254
0.290
0.457
0.246
0.291
0.290
0.295
0.362
0.457
0.997
0.481
0.615
0.530
0.524
0.910
0.693
1.105
0.982
                        -39-

-------
I
I
I
i
i
i
i
i
i
t
I
i
i
i
i
i
I
i
                      Let A.,  denote the measure of accuracy for air quality parameter

            j (j = 1, 2, ..., 6) for sampling interval  k (k = 2, 3, ..., 29).   Then,

                                10
                         _    _
                       10 •  N,    -> -4
                                  i   i
                                m=l n=l
                                     x
                                      jlmn
                                                         10
            Where
                      X.,
                      Xilmn
                         N,  =
                  the value of air quality parameter j

                  for sampling interval  k

                  at station m

                  for sample n

                  same as above with sampling interval  of one (i.e.,  using

                  all available data)

                  number of samples for  sampling interval  k.
          The numerator in Equation 4-1  is the root-mean-square of the

differences between air quality parameters calculated from a sample and

the corresponding air quality parameters calculated using all  available

data.  The denominator is the mean over the 10 stations of the parameter

calculated by using all available data.   The measure of accuracy,  A., ,
                                                                   JK

is similar to the coefficient of variation (standard deviation/mean)

except that the numerator is a root-mean-square rather than a  standard

deviation.
                                  -40-

-------
4.3.3  RESULTS



          Table 4.3-1 gives values of A.,  for the hourly concentrations
                                       j K


and Table 4.3-2 contains results for the daily concentrations.   Both



tables indicate considerable savings in computational  effort with reason-



ably small losses in accuracy.  In Table 4.3-1, the loss in accuracy, as



defined by A.,,, is below 20 percent for all  three air quality parameters
            JK


for sampling intervals up to 13 (i.e., sample size only 0.056 as large as



all available 2920 hours).  The loss in accuracy is less for the mean than



it is for the two exceedance values.  This is consistent with statistical



theory which indicates greater accuracy in estimating the mean  of a dis-



tribution than the tails.  In Table 4.3-2, the loss in accuracy for the



daily concentrations is greater than for the hourly concentrations for



the same proportion of cases in the sample.   But even here, the loss is



below 31 percent for sampling intervals up to 10 days.  Again the loss is



less for the mean than for the two exceedance levels.



          To facilitate use of the results,  the measures of accuracy were



plotted against proportion of total cases in the sample, and lines were



fitted by least squares.  Figure 4.3-1 contains the measures from the hourly



concentrations and Figure 4.3-2 the measures calculated from the daily con-



centrations.  Only the first several values  are plotted because, as can be



seen in Tables 4.3-1 and 4.3-2, the measures show large fluctuations for



small proportions of cases.  In an operational problem, these graphs can



be entered with an hypothesized proportion of cases to estimate what loss



in accuracy would occur.  It must be cautioned, however, that the graphs



are based on S02 at 10 stations in New York  City.   It is our subjective
                                  -41-

-------
Table 4.3-2.  Measures of Accuracy - Daily Concentrations
Sampling
Interval
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Average No.
of Cases
in Sample
183
122
91
73
61
52
46
41
37
33
31
28
26
25
23
22
21
20
19
18
17
16
16
15
14
14
13
13
13
Proportions
of Cases
in Sample
0.501
0.334
0.249
0.200
0.167
0.142
0.126
0.112
0.101
0.090
0.085
0.077
0.071
0.068
0.063
0.060
0.058
0.055
0.052
0.049
0.047
0.044
0.044
0.041
0.038
0.038
0.036
0.036
0.036
Measures of Accuracy for
Mean
0.047
0.081
0.094
0.100
0.122
0.090
0.124
0.096
0.149
0.127
0.235
0.245
0.192
' 0.275
0.162
0.125
0.125
0.233
0.208
0.271
0.228
0.313
0.257
0.227
0.229
0.221
0.225
0.283
0.427
1/100
0.088
0.190
0.191
0.181
0.256
0. 202
0.234
0.218
0.270
0.202
0.472
0.515
0.407
0.514
0.283
0.250
0.279
0.488
0.371
0.602
0.366
0.704
0.328
0.516
0.379
0.360
0.376
0.585
0.916
1/365
0.100
0.228
0.230
0.205
0.305
0.244
0.276
0.262
0.309
0.230
0.576
0.629
0.488
0.616
0.333
0.291
0.328
0.596
0.453
0.748
0.411
0.903
0.363
0.673
0.432
0.405
0.422
0.754
1.173
                         -42-

-------
                                                            CQ
                                                            OJ
                                                            
-------
X X
CM CO
cn oo
CO LO
   CM
CO «=1-
                                                                                     a
                                                                                     o
                                                                                     a
                                                                                     o
                                                                                     o
                                                                                     g
                                                                                    u
                                                                                     0)
                                                                                      .
                                                                                     a
                                                                                     rt
                                                                                     o
                                                                                     •43
                                                                                     O
                                                                                    (N
                                                                                     I
                                                                                    ro

                                 jo
                                   -44-

-------
judgement that they are applicable to other pollutants  and other sites,



but this remains to be proven.
                                   -45-

-------
Section 5.0



 REFERENCES

-------
                              Section 5.0

                               REFERENCES
Blade, E. and E.F. Ferrand.  1969.  Sulfur Dioxide Air Pollution on New
  York City:  Statistical Analysis of Twelve Years.  Journal of the
  Air Pollution Control Association. Volume 19, Number 11, pp. 873-878.

Davidson, B.  1967.  "A Summary of the New York Urban Air Pollution
  Dynamics Research Program."  Journal of Air Pollution Control
  Association, 17, pp. 154-158.

Haltiner, G.J. and F.L. Martin.  1957.  Dynamical and Physical Meteorology.
  McGraw-Hill, New York.

Hanna, S.R.   1971.  A Simple Method of Calculating Dispersion from Urba
  Area Sources.  Journal of the Air Pollution Control Association,
  Volume 21,  Number 12, pp. 774-777.

Larsen, R.I.  1971.  A Mathematical Model for Relating Air Quality Measure-
  ments to Air Quality Standards.  Environmental Protection Agency,
  Office of Air Programs, Research Triangle Park, N.C.  1971.

Ludwig, F.L., W.B. Johnson, et al.  1970.  A Practical Multipurpose
  Urban Diffusion Model for Carbon Monoxide.   Contracts CAPA-3-68 and
  CPA 22-69-64.  Menlo Park, California:  Standford Research Institute.

McCaldin, R.O. and R.S. Sholtes.   1970.   Mixing Height Determinations by
  Means of an Instrumented Aircraft.  Contract No.  CPA 22-69-76.  Gaines-
  ville, Florida:  University of Florida.

McElroy, J.L.  1969.   A Comparative Study of Urban  and Rural Dispersion.
  Journal Applied Meteorology 8,  pp. 19-31.

Roberts, J.J. et al.   1970.   Chicago Air Pollution  Systems Analysis
  Program:  A Multiple-Source Urban Atmospheric Dispersion Model,
  ANL/ES-CC-007, Argonne National Laboratory, Argonne, Illinois.

Saucier, W.J.  1955.   Principles  of Meteorological  Analysis.  The  University
  of Chicago Press, Chicago, Illinois.
                                    -46-

-------
                       APPENDICES





                      Appendix A-l



METHOD OF ESTIMATING THE HEIGHT OF  THE MIXING LAYER





                      Appendix A-2



               MIXING HEIGHT INTERPOLATION

-------
                               Appendix A-l

          METHOD OF ESTIMATING THE HEIGHT OF THE MIXING LAYER


          Mixing heights may be estimated using either standard or EMSU

radiosonde data.  The data may be obtained from the NOAA National  Weather

Records Center in Asheville, North Carolina  on magnetic tapes.   The

method outlined here consists of determining the mixing height by a par-

cel displacement method.  The temperature and mositure content of a repre-

sentative parcel are defined for ground level.  The reported surface

temperature may be used, or a more representative temperature from a

nearby urban location, or another time may be selected.  The moisture

content is defined by the maximum mixing ratio in the vertical  profile.

The temperature change which will occur if the parcel is displaced upward

is traced until the parcel temperature is 1°C less than the sounding

temperature.  The temperature change is assumed to be adiabatic between

the ground surface and the mixing condensation level, and pseudo-adiabatic

above the mixing condensation level.  The following seven steps are used

to determine the mixing layer height for each observation time.


          1.   Read and store the height, pressure, temperature, and
relative humidity of each data level.

          2.   Convert all relative humidities to mixing ratios using
the following equations (Saucier 1955):
                              M = 0.01 U S
                                                                        (A-l)
                       r  _ 0.62197 f E _n goo E                        ,.   .
                            P - f E	~0.62Z pr                        (A-2)
                                   A-l

-------
where
                          E = 6.11  (10)
                                         7.5 T
                                       T + 237.3
(A-3)
          M = mixing  ratio

          U = reported  relative humidity  (percent)
                •
          S = saturation mixing ratio

      f w 1 = correction factor for  departure  from  ideal  gas  laws

          E = saturation vapor pressure of water  (mb)

          P = reported  pressure (mb)

          T = reported  temperature (°C)«
          3.  Find the maximum mixing ratio  for the observation  time  (M  ).
                                                                       A

          4.  Find the mixing condensation level by the  following equa-
tions (Saucier 1955):
                          7  _ 1000  ,,    n ,
                          Zc - "O1  (To ' V
(A-4)
where
         Z  = mixing condensation level (m)
          v»

         T  = ground level temperature (°C)

         D  = ground level dewpoint (°C).


In order to account for evaporation of dew during the early morning, it
is assumed that the rrr'xed atmosphere will contain moisture equal to that
indicated by Mx.   D0  is  determined from Mx by means of "Equations A-2 and
A-3 using S = Mx  and  T = D0:
D
0
237.3
7.5 -
log10
Iog10
" M P
X 0
6.11 (0.622)
f MxPo 1
6.11 (0.622)_
                                                                       (A-5)
                                  A-2

-------
where
         P  = ground level pressure.
              •

          5.  Using the reported data levels to define layers, find the
layer (Z-j_] to Zj) containing the top of the mixing layer.  The top of
the mixing layer is identified by the parcel method.  When the reported
vertical temperature profile exceeds the temperature of a parcel lifted
from the surface by 1°C, this is assumed to be the top of the mixing
layer (Zm).  The layer containing the mixing layer height is identified
by testing if
                              Ti i Ti + !                              (A-6)
where
          T. = temperature of ith level (°C)

          Tr = temperature of parcel lifted to ith level (°C)


The parcel temperature is calculated as follows:
            y- =  -0.0098, Z. < Z,.
                            I —  C

                           T ,     L.$J:I                              (A-?)

                  -0.0098
                                   i   o .
                                   L  o.  -t
                              cp Rv ^i
where
          ' = temperature lapse rate (°C/m,  Haltiner and Martin  1957)

          L = 2500 = latent heat of vaporization  (joules/g)
                                   A-3

-------
       $il] ~ saturation mixing ratio of parcel lifted to (i-l)th level,
              estimated from Equations A-2 and A-3 using P.  ,  and T.',

         Rd = 0.287 = gas constant for dry air (joule/g/°C)

         c  = -1.003 = specific heat of dry air at constant pressure
          P   (joule/g/°C)

         R  = 0.461 = gas constant for water vapor (joule/g/°C).


          6.  Estimate the height of the mixing layer by linear interpo-
lation as follows:




            Zn, = Zi-l  *	(V\r) -'(Vl  " Vl'1    "  HS           
-------
                              Appendix A-2                       j



                       MIXING HEIGHT INTERPOLATION





          The vertical mixing ceiling is defined as that height above



ground level at which there is a marked reduction in vertical diffusion.



Such barriers are observed as a sharp drop in the concentration observed
                                                                  i


in a vertical sounding (e.g., Davidson, 1967).  It may be observed as a



delineation between the smoke-filled layer and cleaner air aloft over



many cities in the early morning.  Much higher ceilings typical of



afternoon hours are clearly visible to air travelers climbing to or



descending from cruising altitudes.  The ceiling may vary from 100 meters



at night to over 1500 meters during the day.  Hourly estimates of the



ceiling are required for use in the model.



          Unfortunately, this mixing ceiling is not always visibly dis-



cernible and no routine systems for taking vertical soundings of pollu-



tant materials are in operation.  Therefore, the ceiling is generally



inferred from temperature soundings which are routinely observed twice



daily at certain airports by the National Oceanic and Atmospheric Admin-



istration (NOAA).  These observing locations are separated by about



200 km on the average and are usually located outside the urban area.



The mixing layer is generally characterized by a near adiabatic lapse



rate extending from the ground to some latitude at which a deep, (several



kilometers) more stable lapse rate exists.  However, the vertical temper-



ature structure of the atmosphere is frequently not this well defined.



As a result, considerable judgement may be required to define where,  in



a vertical temperature profile, an effective mixing ceiling exists.
                                   A-5

-------
Unfortunately very little data have been collected on the relationship
between vertical pollution and temperature profiles which could be used
to develop and substantiate rules for defining the mixing ceiling over
an urban region.
          The procedure which is generally used to define the mixing
ceiling is the following:  Determine the general rural vertical temper-
ature profile from the nearest appropriate (same air mass) radiosonde,
or by interpolation of two or more nearby radiosondes.  Estimate mini-
mum morning and maximum afternoon air temperatures which are representa-
tive of the urban area.  The afternoon temperature may be obtained directly
from airport observations or other available data.  In most cases the
morning urban temperature will exceed the rural temperature.  The follow-
ing equation (Ludwig, et al . , 1970) may be used to estimate morning
urban temperatures (T ) from rural temperatures (T ) using the urban
population $ and the radiosonde temperature lapse rate (dT/dp) as param-
eters:

                 Tu  = Tr  +  4>°'25(0.0633 - 0.298 ^1)                    (A-8)
Construct adiabatic temperature profiles from the urban temperatures
which intersect the rural temperature profile.  The height of these inter-
sections are assumed to be the minimum and maximum mixing ceilings.
                                  A-6

-------
  A method  of interpolating between these  values  to  give hourly estimates

  is to:                                                                 :
                                                                         \

             1.    Use  the morning minimum from midnight to 6  a.m.

             2.    Linearly interpolate between the minimum and the maximum
  between 6 a.m. and  2  p.m.                                            \
                                                                         i

             3.    U.se  the afternoon maximum between 2 p.m. and midnight.


  This pattern  of diurnal variations is illustrated  in Figure A.2-1.
          f
•1,   ooz*
*1-1
wJ Sounding
          I
to  Ceiling  |
         4
 9999999®

-------
          Limited simultaneous observations of temperature and S0? or



particle concentration profiles reported by Davidson (1967),  Roberts,



et. al (1970) and McCaldin and Sholtes (1970)  attest to the general



validity of this approach.
                                   A-8

-------
                                   TECHNICAL REPORT DATA
                            (Please read Instructions on tlie reverse before completing)
 1. REPORT NO.
 EPA-650/4-75-018-b
                              2.
                                                           3. RECIPIENT'S ACCESSION-NO.
 4. TITLE AND SUBTITLE
 Evaluation of  the  Multiple Source Gaussian Plume
 Diffusion Model  -  Phase II
                                 5. REPORT DATE
                                      April 1975
                                 6. PERFORMING ORGANIZATION CODE
 7. AUTHOR(S)
 Robert C. Koch
 Scott D. Thayer
                                                           8. PERFORMING ORGANIZATION REPORT NO.
                                  GEOMET Report  No.  EF-467
 9. PERFORMING ORGANIZATION NAME AND ADDRESS
 GEOMET, Incorporated
 15 Firstfield Road
 Gaithersburg, Maryland
                                 10. PROGRAM ELEMENT NO.
                                    1AA009
                                 11. CONTRACT/GRANT NO.
20760
                                                              68-02-0281
 12. SPONSORING AGENCY NAME AND ADDRESS

 Meteorology Laboratory
 National Environmental  Research Center
 Research Triangle  Park, North Carolina  27711
                                 13. TYPE OF REPORT AND PERIOD COVERED
                                     Final
                                 14. SPONSORING AGENCY CODE
 15. SUPPLEMENTARY NOTES
 Phase I report was  published April 1973 as GEOMET  Report Number EF-186,
 16. ABSTRACT

      This report  summarizes work done to compare  a  computer model for estimating  air
 pollution concentrations from multiple sources with measured S02 and particulate
 concentrations  and  with other model calculations.   The model is capable of  estimating
 short-term and  long-term concentrations, and  produces results which are equivalent in
 validity to results produced with other models.   Since the model represents hourly
 variations in both  emissions and meteorological conditions, this report considers
 available sources of data and how these can best  be used to estimate parameters for
 the model.  Use of  temperature and industrial and commercial activity indexes  to
 estimate seasonal and diurnal variations in emissions is discussed.  Use  of slow-
 rise balloon soundings taken in urban areas is discussed as a possible supplement
 to conventional weather data.  Finally, the applicability of using sampled  calcula-
 tions when estimating short-term maximum concentrations is evaluated.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                              b.IDENTIFIERS/OPEN ENDED TERMS
                                               c.  COS AT I  Field/Group
  Air Pollution
  Urban Areas
  Atmospheric  Diffusion
  Diurnal Variation
  Emission
  Sequential Sampling
                     Air  Quality Model
1302
0401
 3. DISTRIBUTION STATEMENT
  Release Unlimited
                                              19. SECURITY CLASS (This Report)

                                                 UNCLASSIFIED
                                               21. NO. OF PAGES

                                                    64
                                              20. SECURITY CLASS (Thispage)
                                                 UNCLASSIFIED
                                                                        22. PRICE
EPA Form 2220-1 (9-73)

-------