United States
                    Environmental Protection
                    Agency
Atmospheric Sciences
Research Laboratory
Research Triangle Park NC 27711
                   Research and Development
EPA/600/S3-86/024  May 1986
4>EPA          Project  Summary
                    Preliminary  Evaluation  Studies
                    With  the  Regional  Acid
                    Deposition   Model  (RADM)
                     This summary focuses on the find-
                    ings of preliminary evaluation studies
                    of the National Center for Atmospheric
                    Research (NCAR) Regional Acid Deposi-
                    tion Model (RADM). Current efforts in
                    evaluating the gas-phase  chemistry
                    submodel have  met with significant
                    success, and an  understanding of the
                    smog chamber databases has been de-
                    veloped. Due to lack of adequate obser-
                    vation data, evaluation  of aqueous-
                    phase chemistry and dry deposition
                    submodels has not been possible. The
                    cloud process submodel  is  consistent
                    with the available but small data set.
                    Preliminary evaluation of the full RADM
                    system using OSCAR (Oxidation and
                    Scavenging Characteristics of April
                    Rains) meteorology and wet chemical
                    deposition data has  achieved initial
                    success, though  it  is far from conclu-
                    sive. Much larger and extensive data-
                    bases are required  to test the system
                    thoroughly. The meteorology driver for
                    the RADM has shown considerable skill
                    in forecasting the OSCAR IV meteorol-
                    ogy. Simulations of sulfate and nitrate
                    wet deposition  for the first day  of
                    OSCAR IV are quite good.  RADM-
                    simulated three-dimensional chemical
                    species distributions are consistent
                    with preliminary data measured by
                    NCAR under a separate  program. The
                    first test of RADM's capability for ana-
                    lyzing "what if studies with hypotheti-
                    cal reductions in sulfur emissions is
                    demonstrated. The findings confirm the
                    complexity of directly observing poten-
                    tial benefit from emissions reductions.
                     This Project Summary was devel-
                    oped by EPA's Atmospheric Sciences
                    Research Laboratory, Research Triangle
                    Park, NC, to announce key findings of
                    the research project that is fully docu-
mented in a separate report of the same
title (see Project Report ordering infor-
mation at back).

Introduction
  A comprehensive Regional Acid Dep-
osition Model (RADM) is being devel-
oped at the National Center for Atmos-
pheric Research. The RADM is an
evolving  advanced Eulerian computer
model simulating the processes  and
pathways related to acid deposition in
eastern North America. The RADM sys-
tem consists of a mesoscale meteoro-
logical model which drives a transport/
deposition model containing modular
descriptions  of the gas-phase chem-
istry, cloud  processes and aqueous-
phase chemistry, and  dry deposition.
  The objectives for the RADM are
threefold: (1) to result in a state-of-the-
art modeling system  suitable for con-
ducting source-receptor assessment
studies, (2) to be sufficiently flexible to
integrate  current and developing repre-
sentations of the relevant physical and
chemical  processes, and (3) to describe
the spatial and temporal distributions of
pollutants resulting from known source
emissions. Progress toward these ob-
jectives has been very favorable. A first
version of the RADM was completed in
January 1985 and since that time has
undergone numerous sensitivity stud-
ies and preliminary evaluation against
existing field and  laboratory  data. A
summary of the preliminary evaluation
of the RADM and of sensitivity and ap-
plication studies is presented here. The
evaluation of the RADM submodels are
discussed first. Next, 72-h RADM simu-
lations of two OSCAR (Oxidation  and
Scavenging  Characteristics of April
Rains field study, conducted in the

-------
northeastern United States in April
1981) cases are compared to measured
precipitation chemistry data. For one of
these cases, a preliminary evaluation of
the meterological  model is presented.
Emissions sensitivity studies and emis-
sions reduction experiments are also
presented for one of the OSCAR cases.
Finally, databases available for evalua-
tion of the RADM and their limitations
are discussed.

RADM Submodel Evaluation
  The RADM version I gas-phase chem-
istry mechanism has been compared
with both smog chamber data and more
complex chemical mechanisms. In gen-
eral, these tests show that the  RADM
mechanism gives  a good representa-
tion of those aspects of tropospheric
chemistry necessary for the modeling
of acid deposition. Initial  tests were
made with data obtained from the
Satewide Air Pollution Research  Center
(SAPRC), and agreement between
RADM simulations and  experimental
values is generally very good for NO2,
NO, and  03. The agreement between
the predicted and experimental  values
for all the reactive species is excellent.
Since reactions with HO radicals is the
dominant loss process for reactive or-
ganics during  photooxidation, the
agreement between the simulation and
measured concentrations of reactive or-
ganic species shows that  the  RADM
mechanism is correctly predicting HO
concentrations within the uncertainty
imposed  by wall sources.
  Comparison of the RADM gas-phase
chemistry predictions with the Univer-
sity of North Carolina smog chamber
data is only in the beginning stages.
Procedures  used to establish the error
bounds of these data, along with com-
parisons  with RADM predictions for a
single case, are reported.  In general,
RADM predictions for O3, NO, and PAM
are in good agreement  with the
chamber data, considering the  uncer-
tainties of the J-values. The simulation
also suggests that for definitive tests of
the chemical  mechanisms for the pro-
duction of H202 in the atmosphere, the
photolytic rate constant must be accu-
rately known.
  The RADM gas-phase chemical mech-
anism was compared to the explicit
mechanism of Leone and Seinfield and
the carbon bond mechanism, version
CBM-X, for conditions ranging from
very clean to highly polluted. Both  of
these latter mechanisms were modified
to include sulfur chemistry. Oxidation
rates for S02 and NOX are in reasonable
agreement by the three mechanisms,
while large differences exist for the total
H202 produced. A sensitivity analysis
was  performed  to determine reasons
for the differences in H2O2.
  The RADM cloud processes and
aqueous-phase chemistry module was
subjectively  evaluated against limited
field  data and more complex models.
Comparisons with OSCAR aircraft ob-
servations of cloudwater composition
suggest that the RADM submodel is ca-
pable of inferring cloudwater pH over
the range of initial conditions encoun-
tered during the observation period.
RADM predictions of aqueous oxidation
within clouds are shown to agree rea-
sonably with the difference in the ratio
of S02 to total sulfur at the base and top
of typical cumulus clouds. Scavenging
ratios predicted  by the RADM are
shown to lie  between the extremes ob-
served in field observations. In addition
to predicting cloud and rainwater com-
position, the RADM cloud processes
model also predicts the  rate at which
pollutants are vertically transported by
sub-grid scale cumulus clouds.  RADM
parameterization for this process was
compared to a  high resolution  cloud
model, and both predicted similar sig-
nificant depletion of a passive tracer
from the boundary layer and "venting"
of this material to the cloud layer.
  The method  used by the RADM to
model dry deposition assumes a series
of three resistances (aerodynamic, sub-
layer, and surface)  and is based on a
highly empirical parameterization, rely-
ing heavily on a relatively sparse data-
base of dry deposition measurements.
Therefore, it  was not possible to evalu-
ate independently the  performance of
the dry deposition model since all avail-
able  measurements of dry deposition
have been  used to tune the current
model parameterizations. Data needs
for  comprehensive dry deposition
model evaluation include heat and mo-
mentum fluxes in  the surface  layer,
wind speed, isolation,  land type (onto
which substances are  deposited), sur-
face  roughness, and a quantification of
surface moisture.

Preliminary Evaluation of
RADM with OSCAR Database
  The overall performance of the mete-
orological model and RADM in predict-
ing the  meteorological processes and
wet  concentration/deposition  during
the OSCAR IV period (April 22-24,1981)
was examined. A number of quantita-
tive measures of the accuracy of the
mesoscale meteorological model are
described. These measures will be usec
to compare the  accuracy of different
versions of the mesoscale model and tc
provide quantitative estimates of the
error in the meteorological data sup
plied to the RADM. Interpretation of the
verification scores and  methods  from
the OSCAR IV case are provided.
  Correlations between  RADM model
results and OSCAR  measurements oi
precipitation  chemistry  variables are
presented for two OSCAR cases for both
hourly  and event totals  of concentra-
tions and deposition. Parameters com
pared were sulfate, nitrate, ammonium
and hydrogen ions in rainwater. In al
cases,  the correlations were much
higher  for the total event comparisor
and for concentrations. Correlations
were also higher for the OSCAR IV case
than OSCAR I.

RADM Sensitivity Studies and
Applications
  Ultimately, RADM will be used to
evaluate source-receptor relationships
and the effects of changes in emissions
on downwind receptors. This requires
an understanding of the basis of the pat-
terns of concentration and deposition
forecast by the RADM along with sensi-
tivity of the forecasts to various input
parameters and physical  parameteriza-
tions. Toward this end, wet and dry dep-
osition  patterns for the OSCAR I and IV
cases were analyzed. Results are alsc
presented for RADM simulations of twc
"what  if"  scenarios of 50% and 90°/<
source  reduction in the Ohio Valley re
gion.
  The patterns and magnitudes of dn
deposition of S02 over the northeasterr
United  States and southeastern Canada
integrated over the three-day episode:
were similar for both the OSCAR I anc
IV cases.  In Canada, peaks  showed uf
near Sudbury and  Noranda.  In the
United States, the peaks showed UF
over south-central Pennsylvania; at th(
conjunction of Ohio, Pennsylvania, anc
West Virginia; and in New Jersey, f
maximum dry deposition of 1.14 kg/hi
after three days occurred over south
central Illinois during OSCAR IV. Fo
HNO3,  the accumulated dry depositor
for the two three-day episodes wen
again quite similar, and  mass deposi
tion was considerably less than S02. /
maximum deposition of 0.24 kg/ha oc
curred from  northern New Jersey t<

-------
southern New Hampshire. The accumu-
lated wet deposition differed more be-
tween the two experiments than did dry
deposition. This reflects primarily the
difference in rainfall patterns for the two
episodes. The percentage of total sulfur
deposition that was wet was about 50%
and 60% fo the OSCAR I and  IV cases,
respectively.  The percentage  of nitro-
gen deposition  that was wet was  45%
and 60% for OSCAR I and IV, respec-
tively.
  Several sensitivity tests  were run  in
which the SOX emissions were reduced
by 50% and 90% in the Ohio Valley
(states of Indiana, Ohio, Kentucky,  Ten-
nessee, and West Virginia). In  response
to this regional  reduction in emissions,
the maximum regional reduction for dry
and wet deposition occurred in the re-
duced emissions area, but only weakly
in more distant  receptor states. The de-
crease in dry deposition was greater
than the decrease in wet deposition
only in the four states with decreased
emissions. The decrease in sulfur emis-
sions had very  little effect on nitrogen
chemistry or nitrogen deposition.

Databases for RADM Evalua-
tion Studies
  The evaluation of RADM performance
requires  the assessment of the data-
bases used  for  model execution  and
evaluation. Included in the report are a
description and evaluation of the avail-
able air quality databases for evaluating
regional models, a description of qual-
ity control  procedures employed  in
preparing the  emissions databases
(EPRI  and NAPAP) for model applica-
tions, and the  precipitation database
used in verification of the precipitation
predictions. Also discussed are the  limi-
ations of smog chamber data in evaluat-
ing RADM chemistry.
The National Center for Atmospheric Research is located in Boulder, CO 80307.
John F. Clarke is the EPA Project Officer (see below).
The complete report, entitled "Preliminary Evaluation Studies with the Regional
  A cid Deposition Model (RA DM)," (Order No. PB 86-175 692/A S; Cost: $22.95,
  subject to change) 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:
       Atmospheric Sciences Research Laboratory
       U.S. Environmental Protection Agency
       Research Triangle Park, NC 27711

-------
United States                          Center for Environmental Research                         ,, s*     ""--•,, .,..,yj  jj i
Environmental Protection                Information                                                          .•-'•...   -.«.
Agency                               Cincinnati OH 45268




Official Business
Penalty for Private Use $300


EPA/600/S3-86/024
                0000329    PS


                U  S  ENVIR  PROTECTION  AGENCY
                REGION  S   LIBRARY
                230  S  DEARBORN  STREET
                CHICAGO                 TT
                                            IL    60604

-------