United States
                        Environmental Protection
                        Agency
                                  Environmental Sciences Research
                                  Laboratory
                                  Research Triangle Park NC 27711
                        Research and Development
                                EPA-600/S4-81-078/079  July 1983
&EPA
Project Summary
                        Evaluation of  Short-Term
                        N02 Plume  Models for  Point
                        Sources:  Volumes  1   and  2

                        M. A. Yocke, D. A. Stewart,  J. Johnson, and R. J. Frost
                          Four mathematical models for
                        atmospheric pollution transport and
                        reactivity were tested on MISTT data
                        base for their ability to predict patterns
                        of ambient NO? concentrations in the
                        vicinity of large point  sources. The
                        four models tested were: (1) the
                        Reactive Plume Model  (RPM-II), (2)
                        the Total Conversion Method (TCM),
                        (3)  the Oxidant Limiting Method
                        (OLM), and  (4) the Conservation of
                        NO* and Oxidant Method (CNOM). Of
                        these methods, RPM-II was by far the
                        most sophisticated, and yielded more
                        accurate predictions than the other
                        three models.
                          This Project Summary was  devel-
                        oped by EPA's Environmental Sciences
                        Research  Laboratory. Research Tri-
                        angle Park, NC, to announce key find-
                        ings of the research project that is fully
                        documented in a separate report of the
                        same title (see Project Report order in-
                        formation at back).


                        Introduction
                          The burning of fossil fuels by automo-
                        biles,  power plants, and refineries
                        produces appreciable amounts of oxides
                        of nitrogen  (NOX), which are then
                        released into  the atmosphere. These
                        emissions consist primarily of relatively
                        harmless nitric oxide (NO);  however,
                        under suitable conditions, such as in the
                        presence of reactive hydrocarbons and
                        sunlight, NO can be oxidized to form
                        nitrogen dioxide (NOa).
                                    Earlier epidemiological studies indi-
                                  cated that long-term exposure to high
                                  ambient NOa concentrations was asso-
                                  ciated with a higher incidence of respira
                                  tory disease. In addition, there has been
                                  some recent concern about the effects
                                  of short-term peak NOa concentrations
                                  on the frequency and severity of acute
                                  respiratory disease. The results of some
                                  studies show that the magnitude of the
                                  short-term peak NOa concentration had
                                  a significantly greater  influence than
                                  either the  duration or frequency of
                                  exposure. Because of the importance of
                                  these results, the U.S. Environmental
                                  Protection  Agency (EPA)  is giving
                                  serious consideration to promulgating
                                  short-term NOa air quality regulations.
                                  Part of this consideration is the review
                                  of techniques that can be used to predict
                                  short-term (about 1 h) average, ground-
                                  level NOa  concentrations.  Specific
                                  attention is being given to the ability to
                                  predict NOa impacts due to NOX emis-
                                  sions from large point sources  (e.g.,
                                  power plants and other large combustion
                                  processes).
                                    Under the sponsorship of EPA, Sys-
                                  tems Applications, Inc. (SAI) has selected
                                  several models that can be used to
                                  predict ground-level NOa concentration
                                  impacts from point sources  and has
                                  applied the selected models to MISTT
                                  data base  (a large aerometric and
                                  meteorological data base collected near
                                  St. Louis) for the purpose of evaluating
                                  their performance.  The  MISTT data
                                  were selected by EPA because the

-------
MISTT program involved a large urban
and point source monitoring effort that
included a significant number of aircraft
plume measurements.

Description of Models
  The models  evaluated in  this study
were: (1) the  Reactive  Plume Model
(RPM-II); (2)  the  Total Conversion
Method (TCM); (3) the Oxidant Limiting
Method (OLM); and (4)the Conservation
of NOX and Oxidant  Method (CNOM).
The last three are very simple methodo-
logies that have been suggested for use
with standard  Gaussian point  source
dispersion models. RPM-II, on the other
hand, is much more  sophisticated,
allowing treatment  of  time-varying
winds, dispersion rates, reaction rates,
and can accommodate the  use  of very
large photochemical kinetic mechanisms
These models were chosen to provide a
range of complexity in NO2 predictions,
from the very simplest approach (TCM)
to one of the more advanced reactive
plume models (RPM-II).
  In the Conservation of NOx and
oxidant method, the  equilibrium  rela-
tionship among NO, N02 and ozone (Oa)
is used in conjunction with the assump-
tions of oxidant a nd NOx conservation to
simplify the chemical kinetics.  CNOM
suffers the limitation that beyond a few
kilometers from the source its simplified
treatment of chemistry is invalid  in
hydrocarbon-rich  environments.  In
addition, any deficiencies in the disper-
sion  parameter computations—the
Gaussian dispersion relationship, for
example, has many known limitations—
will be reflected  in predicted  N02
concentrations.
  An even simpler model is the oxidant
limiting  method (OLM). The OLM
assumes that N02 concentrations in the
plume are equal to the background Oa
plus some small fraction of the emitted
NO concentration. The dependence on
the background Os is an acknowledg-
ment  that one  NO2 molecule is  formed
for every reaction of Oa and NO
molecules and  that N02 thus may be
limited by the background Oa in  an NO-
rich  plume.  However, it does not
explicitly account for diffusion, entrain-
ment, or the photostationary relationship.
Another simplification,  called total  or
partial conversion  (TCM  or  PCM),
assumes that all or some fraction of the
emitted NOX is ultimately converted to
N02.
  The TCM and OLM suffer the same
deficiencies as CNOM; moreover, these
methods  are further limited by  their
own,  more  restrictive, assumptions.
However, in many cases they are useful
screening  procedures. TCM and OLM
can be usedquicklyand inexpensively to
obtain an estimate of the magnitude of
the potential NOa impacts.
  RPM-II considers  an array of well
mixed reactors  or  cells across  the
plume, each of which moves downwind
and disperses along the mean-wind
trajectory. The model can accommodate
temporally and spatially varying wind
speeds, reaction rates, ambient concen-
tration levels, and dispersion or entrain-
ment  rates.  As  currently  configured,
RPM-II uses a state-of-the-art 68-step
carbon bond photochemical reaction
mechanism developed by Whitten  and
Hogo*; other mechanisms can be easily
substituted for the carbon bond mech-
anism if desired.  The  most  severe
limitation of an involved reactive plume
model like RPM-II is the requirement for
valid  ambient concentration estimates
of reactants along the plume trajectory.
These must by measured, predicted, or
hypothesized. Ambient hydrocarbons
are extremely difficult to characterize
even  by measurement, yet predictions
of plume  NO2 are  very sensitive to
hydrocarbons. Like most other dispersion
models, RPM-II is also limited by the
user's ability to  properly specify wind
speeds and dispersion rates, especially
in complex  terrain  applications.  But
despite such limitations, reactive plume
models  like  RPM-II are clearly more
realistic in  their  approach to NO2
estimation than  are  CNOM, OLM, or
TCM.

Data Base

  The data required for the application
of the models selected are:
  • Atmospheric stability, dispersion
    coefficient,  or  horizontal plume
    spread as a  function of downwind
    distance and time.
  • Point  source emissions rates for
    NO, NO2,  sulfur dioxide (SO2),
    sulfates, and hydrocarbons.
  • Source data:
      - Plume heights.
      - Source parameters, flow rates,
        stack temperatures.
      - Source locations.
  • Ambient concentrations for all
    species as functions of location and
    time.
 'Whitten, G Z  and H  Hogo, Mathematical
 Modeling of Simulated Photochemical Smog. EPA-
 600/3-77-011,  U.S. Environmental Protection
 Agency, Research Triangle Park, IMC, 1977, 307 pp.
  • Sunlight intensity as a function o
    location and time.
  • Inversion heights or vertical plume
    spreads as functions of location anc
    time.
  • Wind field.
These data will also satisfy all require-
ments  of  both  RPM-II and the three
simple NO2 estimation procedures.
  At the start of this study, the useful-
ness of the MISTT data base for applying
and evaluating N02 prediction techni-
ques was uncertain, because  the data
base  was  designed  to characterize
plume sulfur and aerosol transport and
transformations, not to support reactive
plume modeling. However, a significant
amount of  plume  data was collected
during the MISTT study and much of il
for a large point source (Labadie power
plant plume). This study focused on the
Labadie power plant plume data collected
during the MISTT program. Most of the
required data were available as part of
the MISTT data base or from other data
sources. However,  after careful exami-
nation  of the MISTT data, the airborne
hydrocarbon concentration measure-
ments were found  invalid. Plume
dispersion  parameters were  usually
difficult to estimate with much certainty
from the airborne concentration profiles.
  To compensate for  uncertainties in
the dispersion data, the input dispersion
rates were adjusted so that the following
equation was approximately satisfied:

 A(S) = t(E,/U(S) - (C,aA0)]/(C,p(S) - C,a)
where
   A(S) = plume cross-section area at
                downwind distance S
     E, = emissions rate of species i
         (NOx or SO2)
   U(S) = wind speed at distance S
     Cia = measured ambient concen-
         tration of species i
     Ao = exit area  of stacks (121.2 m2)
   C,P(S)= measured average plume
         concentration of species i at
         distance S
Adjustments were made to the total
ambient hydrocarbon levels so that the
plume  Oa  concentrations predicted by
RPM-II approximately matched the
measured  values.  The reactivity of the
hydrocarbon  mass (i.e., the species
distribution  of  hydrocarbons)  was
determined from hydrocarbon  analyses
carried out for  the RAPS program (a
ground-level monitoring program con-
ducted in  the same area during the
MISTT program.)
  These adjustments to the input data
preclude a strict evaluation of model

-------
performances or verification of  model
predictions. However, it is possible to
judge whether the models are capable
of acceptable performance when reason-
able input values are selected.
  In addition, a "restricted" data base
was assembled for input to the models.
The purpose of using this data set was to
evaluate model  performance for the
limited  data  normally available to a
prospective user. Model input values for
all parameters in the "restricted" data
set  were identical  to  those  in the
"unrestricted" set,  except  for  wind
speeds,  plume  widths, and  plume
depths. These exceptions were instead
derived  from local National Weather
Service  station data and standard
Pasquill-Gifford-type dispersion curves.
  A compendium of data collected by
Meteorology  Research, Inc. and the
University of Washington during the
MISTT program is presented  in plotted
form in Volume 2 of the Project Report.
This includes maps showing aircraft
flight paths and  graphs of  the data
collected along the paths.

Results
  Ten case-study days  were selected
from the MISTT study data base, and 28
observations of average plume concen-
trations at various downwind distances
were derived from the data on those 10
days. The results of the model runs with
the "unrestricted" and "restricted" data
bases were studied with the  aid of
several  statistical and  graphic tech-
niques. The RPM-II performed better than
the  simple NO2 models in  both the
"restricted" and "unrestricted"  cases.
This result might be expected because
RPM-II is intended to account in greater
detail for more processes than are the
simpler  models.
  Inspection of scatter diagrams reveals
that RPM-II is capable of matching the
observations more closely than the
simple models across a wide range of
NO2 concentrations. The least-squares
best fit line has a slope of 0.887 and an
intercept of 1.6 ppb, compared with an
ideal slope of 1.0. The  RPM-II clearly
performed within the expected accuracy
of the monitoring instruments (~ ± 15
percent of the true va lue). The computed
statistics are also good, but they do not
constitute validation of RPM-II  perfor-
mance  because of the input  data
uncertainties. Rather,  the following
conclusions  about RPM-II are  war-
ranted:
  • The model is capable of acceptable
     performance  when reasonable
    ambient hydrocarbon and plume
    dispersion input values are selected.
  • No  apparent defect that  affects
    performance is  revealed  in  the
    model's formulation, although the
    formulation  does contain  several
    simplifying assumptions.
  Because the simple models contain a
relatively simplistic treatment of chem-
istry, they may  not be appropriate for
use in  situations where hydrocarbon
oxidation significantly influences NOa
concentrations. It is therefore desirable
to define concisely the character of the
situations requiring more sophisticated
models. On the basis of the simulations
carried out  in this study, however, it is
clear that several important parameters
interact to  determine NOa  production
rates. Thus, a  concise definition of
conditions  requiring  a  sophisticated
model is not possible on the basis of
these results. NOa production appears
generally to depend on the existence of
a sufficient amount of ambient ozone
and hydrocarbons, solar insolation, and
plume dispersion in combination. The
precise  combinations  depend  on the
following variables:
  • Amount  of  NOX  present  in the
    plume and in the ambient air.
  • Time of release from the stack.
  • Ratios of NO to NOs in the plume
    and in the ambient air.
  • Wind speed and variability.
  • Split of the reactive hydrocarbons.


Conclusions
  The four models, RPM-II, TCM, OLM,
and CNOM performed reasonably well
by comparison with MISTT data both
"unrestricted" and "restricted"  input
data sets. Of the four, RPM-II was the
superior performer using both data sets
over a  wide range of concentration
levels. TCM, OLM, and CNOM tended to
underpredict  but mostly  at the high
concentration ranges.  It must be stres-
sed that model  performances  in this
study cannot be construed as a verifica-
tion of model performances because of
input data deficiencies and the  input
data adjustments that were necessary.
  M. A.  Yocke, D. A. Stewart, J. Johnson,  and R.  J. Frost are with  Systems
    Applications. Inc., Sam Rafael, CA 94903
  K. L. Schere is the EPA Project Officer (see below).
  The complete report consists of two volumes entitled "Evaluation of Short- Term
    NOz Plume Models for Point Sources:"
    "Volume I.  Technical Discussion," (Order No. PB 82-234 329; Cost: $17.50,
    subject to change).
    "Volume II. Data," (Order No. PB 83-217 232; Cost: $4.50, microfiche only,
    subject to change).
  The above reports will be available only from:
          National Technical Information Service
          5285 Port Royal Road
          Springfield, VA 22161
          Telephone: 703-487-4650
  The EPA Project Officer can be contacted at:
          Environmental Sciences Research Laboratory
          U.S. Environmental Protection Agency
          Research Triangle Park, NC 27711

-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
Official Business
Penalty for Private Use $300

-------