United States
Environmental Protection
Agency
Atmospheric Sciences
Research Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S8-85/028 Mar. 1986
x°/EPA Project Summary
Methods for Mesoscale
Modeling for Materials
Damage Assessment:
User's Guide
F.W. Lipfert, L.R. Dupius, and J.W. Schaedler
Assessment of acid deposition dam-
age to materials requires, as a mini-
mum, detailed knowledge of SOj (gas)
and wet H* annual deposition fields on
time and space scales consistent with
the mechanisms of damage and the
distribution of materials at risk. Meth-
ods for urban air quality SO, modeling
are reviewed, and a set of simplified
algorithms is presented suitable for
relatively rapid assessment of a large
number of cities. The method is rela-
tively "meteorology free" in that
regional wind roses have been incor-
porated into the parameterization
rather than site-specific wind roses.
Separate algorithms are presented for
point and for area sources, and the
model is designed to produce annual
average concentration estimates over
5-km grids. The Parameterized Air
Quality Model-Annual (PAQMAN) is
designed to mimic the performance of
the Climatological Dispersion Model
(COM). McElroy-Pooler dispersion co-
efficients are used for point sources;
whereas Briggs' plume rise and Pasquill-
Gifford dispersion coefficients a re used
for area sources. PAQMAN results
correlated well with COM calculations
for the New Haven and Pittsburgh
Metropolitan Statistical Areas (MSAs)
with correlation coefficients of 0.90
and 0.80, respectively. Comparisons
between PAQMAN and measured con-
centrations averaged over each MSA
were also favorable over the entire
spectrum of 112 MSAs.
This Project Summary was devel-
oped by EPA's Atmospheric 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 ordering
information at back).
Introduction
One of the components of the National
Acid Precipitation Assessment Program
(NAPAP) deals with damage to materials
in the environment. Such an assessment
requires detailed information on deposi-
tion of corrosive agents, namely S02 and
wet H+, on spatial scales compatible with
materials distributions in urban areas.
The objective is to provide sufficiently
accurate deposition estimates for mate-
rials damage for over 100 cities.
Since extant urban SO2 monitoring
data are inadequate to characterize the
distribution of airquality within a cityfi.e.,
monitors tend to be located only to meet
specific regulatory needs), a modeling
approach is necessary to estimate de-
tailed annual average SO2 concentra-
tions. The basic approach used is a
parameterization in which closed-form
algorithms are developed for point and
area sources; these algorithms mimic the
performance of EPA's Climatological
Dispersion Model (CDM) and allow rapid
assessment of a large number of sources.
The model, called PAQMAN, has been
applied to over 100 MSAs in which
materials distributions have been de-
rived. The intent is that this methodology
will serve as a useful sub-grid feature for
mesoscale and regional scale models.
-------
Objectives
The objectives of this study are to
(1) Develop an acceptable parameter-
ization model for S02 that will allow
the rapid assessment of a large
number of sources using limited
computer resources, and at the
same time mimic the predictions of
a more refined model with reason-
able accuracy.
(2) Apply this parameterized model to
several MS As in order to accurately
predict SO2 distributions for urban
areas.
(3) Develop data bases for regional S02
background (non-urban) concentra-
tions that are consistent with the
urban SOz observations, and for
urban H+ (wet) deposition.
Technical Approach
Since long-term annual average esti-
mates of SO2 are needed for the materials
assessment, a steady state Gaussian
model was considered appropriate as the
baseline model to be used to develop
RAGMAN. COM was selected as the
reference model because of its accept-
ance as a standard EPA regulatory model
and its extensive previous application to
urban areas.
Simplified calculation methods were
developed from parametric studies of
both point and area sources. For point
sources, a range of stack heightsand flue
gas conditions was used; various geo-
metric patterns and emission densities
were used for area sources. Following
the recommendations of EPA for urban
areas,1 the McElroy-Pooler dispersion
coefficients were applied to area sources.
Plume rise was calculated according to
Briggs.2 Three wind roses were used for
the parametric studies: JFK Airport (New
York), Pittsburgh, and Cincinnati. These
were then collapsed into generic cate-
gories: "coastal" (JFK) and "interior" (a
blend of Pittsburgh and Cincinnati), in
order to eliminate site-specific wind
roses in favor of regionally representative
meteorology.
The point source calculation algo-
rithms were based on the use of the
average distance (rma«) from the source to
the location of the maximum annual
average ground-level concentration as a
scaling parameter. This distance was
found to be a function of stack height. The
downwind ratio of concentration with
respect to the maximum concentration
was found to be a function of the ratio of
downwind distance to the scale length,
rmax. The maximum ground-level con-
centration lx/Q) was found to be a
function of stack height. These curve fits,
shown schematically in Figure 1, rep-
resent the entire spectrum of point
source dispersion patterns, which were
averaged around the compass to remove
site-specific wind direction effects. The
method implies that all stack heights can
be generalized into one representation,
as demonstrated in Figure 2.
Emission density (Q/A) and scale
length (grid size) were the controlling
parameters for the area source calcula-
tion algorithm. The underlying assump-
tion is that concentration patterns tend to
"flow" from areas of high emission
density into adjacent areas of lower
emission density and that the opposite
effect is negligible. This was deduced
from parametric calculations with COM,
using concentric squares of varying
emission densities.
Calculated area source concentra-
tions are sensitive to the scale length
Point Sources
(X/Q1,
flm.,
H,
1.0
1.0
(emission grid size) applied to the emis-
sions inventory. For example, as the size
of the source diminishes at constant
emission density, it approximates a point
source of effectively zero strength. With
increasing scale, on the other hand,
larger absolute amounts of emissions
become involved, which increase the
spatial impact of the area source. While it
may have been desirable to have the area
scale length compatible with the ma-
terials inventory and land use elements
(census tract level ~1 km2), the parametric
COM runs revealed that a calculation grid
of greater than 1 km was essential in
order to achieve satisfactory computing
times. A 5-km grid provided an accep-
table compromise between spatial reso-
lution and computing resources.
A rectangular grid (5-km spacing) was
overlaid that encompassed the census
tract centroids within each MSA. All
sources inside this grid were included in
Area Sources
XA/Q
Scale Length
X = Ground-Level SOzConcentratii
Q = Emission Rate
H, = Stack Height
R - Distance
ED = Emission Density
Figure 1 . PA QMAN point and area source algorithms.
-------
0.001
Figure 2,
100
Concentration ratio vs. distance
ratio CVG wind rosevarying
H, McElroy-Pooler cr's.
the model calculations. Sources with
stack heights less than or equal to 62 m
were grouped and treated as area
sources; those with stack heights greater
than 60 m were treated as point sources
in the model. In addition, all stack heights
<19 m were set to 19 m, assuming that
they were located atop buildings. Sources
within 50 km outside of the grid bounda-
ries were all treated as point sources, but
were retained only if their concentration
at the nearest grid cell centroid exceeded
0.1 /ug/m3. This procedure was very
effective in eliminating unnecessary
calculations, which reduced computer
time substantially. In addition, the emis-
sions from groups of sources located
within 0.1 km of each other were ag-
gregated, and a weighted average stack
height was used in the calculation
algorithms. While this procedure tends to
overestimate the combined concentra-
tion if the stack heights in the groups
differ greatly, it prevents groups of small
sources from being overlooked because
each is too small.
Emissions data were obtained from
the NAPAP emissions inventory (Version
2.0) for 1980. In addition, emissions due
to residential space heating (oil fuel)
were derived on the census tract level
and summed within each grid square.
These oil-heat emissions were combined
with the NAPAP area source emissions
(stack heights <62 m) for each grid
square. A check on potential problems
with the emissions inventory was im-
plemented by listing all sources that
contributed to a calculated SO2 concen-
tration above the primary air quality
standard of 80/ug/m3. This allowed spot
checks for erroneously calculated low
stacks or high emissions and identified
gross underestimates of ambient con-
centrations (missing sources or er-
roneously high stacks).
In many cases, regional background is
an important portion of the total observed
concentration. Estimates of these back-
ground concentrations were provided by
Dr. Jack Shannon of Argonne National
Laboratory who used the ASTRAP model3
and 1980 NAPAP emissions data (Ver-
sion 2.0). Urban wet H* concentration for
each MSA was interpolated from re-
gional values from the Acid Deposition
Systems (ADS). The rationale for this
assumption was that neutralization of
acidity effects from local urban SO2 and
NOx emissions are neutralized to a cer-
tain extent by alkaline particles in urban
dust.
Results
Comparison of PAQMAN with
COM Calculations
The CDM model was exercised in de-
tail for two case study cities: New Haven,
Connecticut; and Pittsburgh, Pennsyl-
vania. Agreement between the actual
CDM results and those from the simpli-
fied algorithms (PAQMAN) were quite
I
c
o
o
1
a
good, as shown in Figures 3 and 4. Cor-
relation coefficients were 0.90 and 0.80
for CDM and PAQMAN, respectively. For
both cities, the highest concentrations
were due to area sources, for which the
two models agree very well. The scatter
in data at the lower-to-middle concentra-
tion levels, in general, reflect our de-
cision to ignore wind direction effects in
favor of averaging. Since Pittsburgh has
major point sources to the east of the
MSA, PAQMAN S02 concentrations tend
to be over-predicted with respect to CDM
because the prevailing winds are west-
erly. In contrast. New Haven is affected
by major point sources to its southwest;
therefore, PAQMAN under-predicts rela-
tive to CDM for the same reason. It is
hoped that, in general, this effect will
balance and produce no net bias in the
overall assessment of all 112 MSAs.
Comparison with Monitoring
Data
A comparison of both model calcula-
tions with ambient measurements is
shown in Figure 5 for Pittsburgh (se-
lected because of its extensive monitor-
ing network). The two models showed
excellent agreement with ambient mea-
surements, including the background
SOz estimate from ASTRAP.
Figure 6 compares the predicted con-
centrations from PAQMAN averaged
over the entire grid versus the average
observed concentration for the MSAs.
Overall, the model performed very well
(slope is 0.81 ± 0.06) considering that
Symbol = No. of Observations
A = 7
B = 2
etc.
0 / 2 3 4 5 6 78 9 10 11 12 13 14 15
CDM SOz Concentration, fjg/m3
Figure 3. Comparison of PA OMAN and CDM SOz concentration estimates for New Haven.
3
-------
o
Cj
150
140
130
120
no
100
90
80
70
60
SO
40
30
20
10
0
0
COM: M-P a's throughout
PA QMA N: Interior wind rose option
* *
»«
A k
> « < 4
to* aces A» o*
OUtLlCCB»*«
8vnocj»e
CtttOCI
LDIB
Symbol - No. of Observations
A = /
ff = 2
J I
j
40
Figure 4.
50 60 70 80
COM SOt Concentration,
Comparison of PA QMAN and COM SOx concentration estimates for Pittsburgh.
90
100
110
120 130 140 151
monitored data may not necessarily be
representative of the entire urban center.
Many of the higher concentrations (maxi-
mum concentrations, not shown) pre-
dicted by PAQMAN resulted from clus-
ters of sources very close to each other
(industrial complexes in cities such as
Baltimore, Gary-Hammond, Cleveland)
that fall into the same grid square.
Downwash, a frequent feature of such
complexes with short-to-moderate stack
heights, can lead to very high concentra-
tions. In most cases, the maximum con-
centration thus estimated would occur
on plant property, which in a regulatory
sense, is not classified as "ambient air."
Therefore, monitoring data are not re-
quired and are in fact non-existent in
most cases.
Discussion
The underlying assumptions of this
methodology are that all stack heights
can be generalized into a single repre-
sentation and that local meterorology
over a long-term (annual) basis can be
generalized into a regional representa-
tion. These assumptions for the most part
will not introduce any net bias for a large
number of sources scattered "evenly"
around an urban center, but are not
applicable to isolated source assess-
ments. The tendency toward over- or
under-prediction for MS As in which
sources are more concentrated in one
section than another (New Haven and
Pittsburgh) suggests that wind direc-
tional effects may have to be incorpo-
rated in some way within these al-
gorithms. Furthermore, additional al-
gorithms for other meteorological re-
gions (Great Lakes, interior Mid-West,
interior East, etc.) may have to be de-
velped.
The calculations for the 112 MSAs
reveal that despite some high predicted
concentrations, average annual S02
levels within the urban areas are low (20-
30 fig/m3}. This indicates that the urban
contribution, in terms of SOa is com-
parable to that contributed by back
ground levels, on the average. In terms o
damage assessment, these findings im
ply that the sources outside of the city an
just as important as local sources.
Conclusions
The following were concluded frorr
this study.
1. A set of simplified algorithms tha
mimics the COM model and i:
suitable for rapid estimation of SO
distributions for urban areas (MSAs
has been developed. Test-cast
studies have shown that the para
meterization model (PAQMAN
provides reasonable representa
tion of CDM calculations (accurac1
estimates are about ±25%), espe
cially for the higher 862 levels
Both models were shown to com
pare favorably with monitorini
data.
2. PAQMAN predictions for 11:
MSAs indicated that annual aver
-------
age SOz levels that were averaged
over the urban areas were com-
parable to background levels,
which sugests that a considerable
part of the materials damage costs
in urban centers comes from pollu-
tion from outlying areas.
3. The results also reveal some prob-
lems in the model that need to be
examined further, such as sensitiv-
ity of prediction to the assignment
of sources as either point or area
sources, based on stack height; the
incorporation of methodology to
deal with wind direction effects as
well as additional regional mete-
orological regimes; and methods to
estimate urban H+ concentrations.
References
1. U.S. Environmental Protection Agency.
Guideline on air quality models. Draft
(revised). Office of Air and Radiation,
Office of Air Quality Planning and
Standards, U.S. EPA, Research Tri-
angle Park, NC, 1984.
2. Briggs, G.A. Plume Rise Predictions.
In: Lectures on Air Pollution and
Environmental Impact Analysis, Ameri -
can Meteorological Society, Boston,
MA, 1975.
3. Shannon, J.D. A Model of Regional
Long-Term Average Sulfur Atmos-
pheric Pollution, Surface Removal,
and Wet Horizontal Flux, Atmos.
Environ., Vol. 15, No. 5, pp. 689-701,
1981.
150.0
140.0-
130.0-
120.0-
110.0-
100.0
90.0
80.0-
70.0-
60.0-
50.0-
40.0-
30.0
20.0.
10.0.
0.0.
Background
(FromASTRAP
Calculations)
0.0
20.0 40.0 60.0 SO.O 700.0 720.0 740.0
Observed SOt Concentration ffjg/m3)
n COM + PAQMAN
Figure 5. Calculated vs. measured S02 for Pittsburgh.
o
2:
S
60
SO
40
30
20
10-
10 20 30 40 50
Observed SO2 liig/m3)
60
Figure 6. Comparison of average PAQMAN and average observed SOzConcentration for 112
MSA's.
U. S. GOVERNMENT PRINTING OfflCE:1986/646-l 16/20784
-------
F. W. Up/art, L ft. Dupius, and J. W. Schaedler are with Brookhaven National
Laboratory. Upton. NY 11973.
Jack L. Durham is the EPA Project Officer (see below).
The complete report, entitled "Methods for Mesoscale Modeling for Materials
Damage Assessment: User's Guide," (Order No. PB 86-144 862/AS; Cost:
$11.95, 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
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S8-85/028
OOOC329 PS
U S E«„IR PROTECTION ASEHCY
REGION 5 LIBRARY
230 S DEARSORN STREET
CHICAGO II 60604
------- |