United States
Environmental Protection
Agency
Atmospheric Sciences
Research Laboratory
Research Triangle Park, NC 27711
Research and Development
EPA/600/S8-86/013 May 1986
Project Summary
RELMAP: A Regional
Lagrangian Model of
Air Pollution User's Guide
Brian K. Eder, Dale H. Coventry,
Terry L. Clark, and Catherine E. Bellinger
The R Egional Lagrangian Model of Air
Pollution (RELMAP) is a mass-conserving,
Lagrangian model that simulates ambient
concentrations and wet and dry deposi-
tions of SO2, SO4=, and fine and coarse
paniculate matter over the eastern United
States and southeastern Canada (default
domain). Discrete puffs of pollutants,
which are released periodically over the
model's domain, are transported by wind
fields and subjected to linear chemical
transformation and wet and dry deposition
processes. The model, which is generally
run for one month, can operate in two dif-
ferent output modes. The first mode pro-
duces patterns of ambient concentration
and wet and dry deposition over the de-
fined domain, and the second mode pro-
duces interregional exchange matrices
over user-specified source/receptor
regions.
RELMAP was written in FORTRAN IV on
the Sperry UNI VAC 1100/82, and consists
of 19 preprocessor programs that prepare
meteorological and emissions data for use
in the main program, which uses 17 sub-
routines to produce the model simulations.
The procedure necessary for running the
preprocessors and the model is presented
in an example execution, which also allows
the user to verify his results.
A statistical evaluation of the model
reveals that seasonal and annual simula-
tions of sulfur wet depositions for 1980
generally agree within a factor of two with
the observed data. The model, which
generally overpredicts wet deposition in
spring and summer, produced Pearson cor-
relation coefficients that ranged between
0.208 during autumn and 0.689 in spring.
This Project Summary was developed
by EPA's Atmospheric Sciences Research
Laboratory, Research Triangle Park, NC, to
announce key findings of the research pro-
ject that is fully documented in a separate
report of the same title (see Project Report
ordering information at back).
Introduction
During the mid-1970's, SRI International
developed a Lagrangian puff air pollution
model called European Regional Model of
Air Pollution (EURMAP) for the Federal
Environment Office of the Federal Republic
of Germany. This regional model simulated
monthly SO2 and S04= concentration
and wet and dry deposition patterns, and
generated matrices of international ex-
changes of sulfur for 13 countries of
western and central Europe.
By the late 1970's, the U.S. Environmen-
tal Protection Agency (EPA) sponsored SRI
International to adapt and apply EURMAP
to eastern North America. The adapted
version of this model, called the Eastern
North American Model of Air Pollution
(ENAMAP), also calculated monthly SO2
and SO4= concentrations and wet and
dry deposition patterns, and generated
matrices of international exchanges of
sulfur for a user-defined configuration of
regions.
During the early 1980's, EPA modified
and improved the model to increase its
flexibility and scientific credibility. By
1985, simple parameterizations of pro-
cesses involving fine (diameters < 2.5 urn)
and coarse (2.5 ^/m
-------
federal standards for inhalable particulate
matter. This latest version of the model,
called the REgional Lagrangian Model of
Air Pollution (RELMAP), is capable of
simulating concentration and wet and dry
deposition patterns of SO2, S04 = , and
fine and coarse particulate matter. It can
also generate source-receptor matrices of
S02, S04 = , and particulate matter for
user-defined regions.
RELMAP consists of 19 preprocessing
programs that prepare gridded meteor-
ological and emissions data for use in the
main program, which uses 17 subroutines
to generate the simulations. RELMAP was
developed on a Sperry UNIVAC 1100/82,
but can be run on other systems with
minor changes. RELMAP represents state-
of-the-art modeling for the simulation of
the transport, diffusion, transformation,
and deposition of these pollutants.
One of the main features of RELMAP is
its flexibility. The values of more than fif-
ty parameters, which have been defined
with default values, can be redefined by
the user simply by changing the values in
one subroutine. The optimum values of
most of the parameters used in the model
have been investigated by numerical ex-
perimentation, but not all are known.
Therefore, the user has the option of speci-
fying other parameter values. As more
knowledge is gained about the parameter-
izations used by the model, new values
can be substituted, thereby making
RELMAP a constantly evolving model.
Input data required by RELMAP can be
divided into three major categories: (1)
meteorological, which include surface and
850 mb winds, precipitation, and monthly
maximum mixing heights; (2) emissions,
which include point (stack) and area
sources of SO2, S04 = , and fine and
coarse particulate matter; and (3) terrain
data, which are defined by land use types.
All input data must be converted from their
original raw forms into the gridded forms
required by the model. This requires spatial
interpolation of the data from reporting
stations to grid cells with a user-defined
(1 ° x 1 ° = default) resolution. Temporal
interpolation is also required to convert the
time increments of the raw data, which
vary from 1 h to 12 h, into the user-defined
time increment of the model (2 h =
default).
Output can be generated in two major
formats. The first format generates user-
defined (45 x 30 = default) arrays of
gridded values of ambient concentrations
and wet and dry deposition of SO2,
S04 = , total sulfur, and fine and coarse
particulate matter. The second output for-
mat, which considers the same param-
eters, produces source-receptor exchange
tables. With both output formats, deposi-
tions are given in kilograms per hectare,
and concentrations are given in micro-
grams per cubic meter. A budget of the
total sulfur and particulate matter
throughout each model simulation pre-
cedes the output arrays. This budget table
contains total pollutant input into the
model, total wet deposition and total dry
deposition of the pollutant, the amount
that was transported off the grid, the
amount of pollutant that remained in puffs
at the completion of each run, and when
applicable, the amount of pollutant
transformed (e.g., S02 into S04=).
Theoretical Basis of the Model
RELMAP is a mass-conserving, regional-
scale Lagrangian model that simulates am-
bient concentrations as well as wet and
dry depositions of SO2, SO4 = , and fine
and coarse particulate matter. The model
performs simulations on a user-defined
latitude-longitude grid with a user-defined
degree of resolution (approximately
10,000 km2 with the default 1 ° x 1 ° grid)
covering the eastern two-thirds of the
United States and southeastern Canada.
The north-south, east-west boundaries of
the model's default domain extend from
25° to 55 °ISI latitude and from 60° to
105 °W longitude, respectively.
RELMAP divides the atmospheric boun-
dary layers into three layers, into which
seasonal emissions are injected. The first
layer is between the surface and 200 m,
and the second is between 201 and 700
m. The depth of the third layer depends
on the maximum mixing height. Default
values vary between 1150 m in winter, to
1300 m in spring and fall and 1450 m dur-
ing the summer.
In the default mode, discrete puffs of
SO2, S04~, and fine and coarse par-
ticulate matter are released every 12 h for
each of the 1350 grid cells that contain
sources. By using vertically integrated and
horizontally and temporally interpolated
wind fields the model transports each puff
in user-specified time increments until the
puff is either transported out of the
model's domain or until the mass of the
pollutant falls below a user-specified
minimum value. The pollutants in each of
the three layers of a puff are transported
in the same direction at the same speed
(i.e., the puff remains an indivisible entity).
RELMAP parameterizes both the
horizontal and vertical diffusion of the
puffs very simply. During the unstable
regimes of midday periods, pollutants from
both point and area sources become well '
mixed below the mixing height long before
the pollutants are transported a distance
comparable to the spatial resolution of the
default grid. For this reason, it is assumed
that instantaneous, complete mixing
within the three layers of the model occurs
during the unstable daylight hours. After
sunset, when mixing is prohibited by
stable conditions, point-source and area-
source emissions are injected into
separate layers and confined to those
layers. All emissions from area sources re-
main in Layer 1, within 200 m of the sur-
face. Emissions from point sources are
allocated into Layer 2, accounting for
typical plume rise, which averages several
hundred meters. Horizontal diffusion of
the puffs occurs at a constant rate, so that
the area of each puff increases at a rate
of 339 km2/h, with the mass of the pollu-
tant remaining homogeneous in the
horizontal at all times. Linear chemical
transformation and wet and dry deposition
processes are simulated by RELMAP as
each puff is transported across the
model's domain. For each time step, the
suspended mass and deposition of each
puff are apportioned into the appropriate
grid cells, based on the percentage of puff
over each grid cell.
RELMAP treats fine and coarse par-
ticulate matter as independent, non-
evolving pollutants; that is, physical and
chemical transformations between fine
and coarse particles are considered to be
negligible. This premise is supported by
particle size distributions obtained from
monitoring data, which exhibit a bimodal
distribution with peaks in the fine and
coarse particle ranges and a deep gap
oscillating between 1 and 5 jjm. RELMAP
does, however, consider the transforma-
tion of SO2 into SO4 = . In the atmos-
phere, this rate varies nonexclusively with
solar insolation (and thus time of day, time
of year, and latitude) and moisture con-
tent. RELMAP simulates the transforma-
tion of SO2 to SO4= through the use of a
heterogeneous component and a homo-
geneous component. The heterogeneous
component accounts for the more rapic
transformation processes that occur ir
saturated environments, while the homo
geneous component simulates transfer
mation that occurs under dry conditions
Both components vary seasonally.
Dry deposition of SO2, SO4 = , and fin
and coarse particulate matter is a high!
variable, complex process that i
parameterized by RELMAP as a functio
of predominant land use, season, an
stability. Twelve land use categories, df
-------
fined by surface characteristics and
vegetation type have been gridded to
RELMAP's 1 ° x 1 ° grid. Dry deposition
velocities, which represent the downward
surface flux divided by the local concen-
tration, were calculated for each land use
type, for six different stability classes and
for each season for SO2, SO4=, and fine
and coarse paniculate matter. The dry
deposition velocities for S02 range bet-
ween 0.05 and 1.15 cm/s, and S04= and
fine particulate matter, velocities range be-
tween 0.05 and 0.50 cm/s. During the
nighttime hours, the dry deposition
velocities for SO2, S04 = , and fine par-
ticulate matter are reduced to 0.07 cm/s,
in order to compensate for the very high
nocturnal resistance, when plant absorp-
tion is minimal.
Because coarse particulate matter con-
sists of a wide range of particle diameters,
two sets of dry deposition velocities are
used by the model. The first set applies to
particulate matter with diameters of 5 ^m,
and the second set applies to particles
with diameters of 10 ^m. The dry deposi-
tion velocities for particles with a 5-^m
diameter range between 0.4 and 5.0 cm/s,
and between 1.0 and 6.0 crn/s for 10- /^rn
diameter particles. Unlike S02, S04=,
and fine particulate matter, the dry deposi-
tion velocities of coarse particulate mat-
ter are much less dependent on the time
of day and the season; therefore, diurnal
and seasonal variation are considered
negligible.
The complex process of wet deposition
of S02, S04 = , and fine and coarse par-
ticulate matter is thought to be a function
of cloud chemistry and cloud type, pollu-
tant concentration, and precipitation type
and rate. RELMAP, however, parameterizes
wet deposition quite simply, treating it as
a function of season and precipitation rate
only. The wet deposition rates are ex-
pressed as percentages per time step,
which depend solely upon precipitation
rate and cloud type. Three cloud types
were considered including Bergeron or
cold-type clouds, warm or maritime-type
clouds, and convective-type clouds.
RELMAP assumes that all winter precipita-
tion results from the Bergeron process,
that spring and autumn precipitation result
from warm cloud formation, and that sum-
mer precipitation is produced by
convective-type clouds.
Model Performance Evaluation
In general, a rigorous evaluation of any
model requires a long-term reliable data
set consisting of measurements of all
parameters simulated by the model across
a network of representative sites similar
to the spatial and temporal resolution of
the model. Unfortunately, such a complete
data set is not available to rigorously
evaluate all aspects of RELMAP. However,
a spatially and temporally consistent sulfur
wet deposition data base is available to
evaluate the model. Such a data set was
acquired from the Acid Deposition System
(ADS), operated for EPA by Pacific
Northwest Laboratory.
Monthly simulations of concentrations
and wet and dry depositions of SO2,
S04 = , and total sulfur were made for all
of 1980. Emissions data used in simula-
tions were from the 1980 National Acid
Precipitation Assessment Program
(NAPAP) Task Group B emissions inven-
tory (Version 2.0) and from the Environ-
ment Canada emissions inventory used in
Phase III of the U.S./Canadian Memoran-
dum of Intent on Transboundary Air Pollu-
tion. Seasonal and annual predictions of
sulfur wet deposition (expressed in
kilograms of SO4= per hectare) were
compared to the amount of seasonal and
annual sulfur wet deposition recorded by
the ADS system. The number of observa-
tions for the seasonal evaluations ranged
from 36 in the winter to 43 during the
spring. The annual evaluation consisted of
34 observations.
Comparison of the means and standard
deviations of the predicted and observed
values, with their corresponding residuals,
provides an indication of the model's
overall performance. RELMAP slightly
overpredicted total sulfur wet deposition
during the winter (0.01 kg SO4 = /ha, or
0.22%). It overpredicted wet deposition
during the spring (2.02 kg S04=/ha, or
25.73%) and summer (3.43 kg S04 = /ha,
or 37.12%). It underpredicted for autumn
(0.37 kg SO4 = /ha, or 9.23%). In the an-
nual simulation, the model over-predicted
total sulfur wet deposition (5.41 kg
S04=/ha, or 20.66%). The percentages
of over/underprediction were calculated by
dividing the residuals by the mean of the
total observed sulfur wet deposition by
season. Examination of the standardized
residuals indicates that there is a consis-
tent tendency within each season for the
model to overpredict total sulfur wet
deposition in the major source regions (i.e.,
Ohio River Valley) and to underpredict wet
deposition in the nonindustrial areas.
The pronounced overprediction that oc-
curs during the spring and summer can in
part be attributed to two major factors: (1)
the model does not account for sub-grid-
scale precipitation variability, and (2) the
model does not account for the vertical
transport of pollutants from the mixed
layer into the free troposphere by cumulus
cloud venting. RELMAP assumes that
precipitation occurs everywhere within a
grid cell whenever precipitation occurs at
any site within that cell. Therefore, the fre-
quency of simulated precipitation events
in any grid cell is higher than the actual fre-
quency at one site in that cell, especially
during the months of convective-type
precipitation (spring and summer). Conse-
quently, wet deposition will occur more
frequently in the simulations.
In support of the second factor at-
tributed to model overprediction, it is
estimated that clouds vent 20% of the
subcloud layer air into the troposphere dur-
ing winter and 50% during summer.
Failure to account for such vertical
transport could result in excessive sulfur
wet deposition occurring near the source
regions, especially during the convective
seasons (spring and summer), which is
consistent with the analysis.
Seasonal scatter diagrams, which ex-
hibit the correlation or dependency of the
predicted values on the observed values,
reveal that the model produced higher cor-
relations during the spring (0.689) and
summer (0.562) than it did during autumn
(0.208) and winter (0.479). The annual
simulation produced a Pearson's correla-
tion coefficient of 0.614, indicating that
37.7% of the variance exhibited by the
observed data could be accounted for by
the simulation.
Computer Aspects
The RELMAP model consists of two ma-
jor components, the preprocessors and the
model itself. The software for RELMAP is
written in ASCII FORTRAN and was
developed and tested on the Sperry
UNIVAC 1100/82 at the National Com-
puter Center at the U.S. Environmental
Protection Agency in Research Triangle
Park, North Carolina. Three UNIVAC-
specific routines are used in the software:
ADATE, SORT, and ATAPE. If the user has
a different computer system, comparable
routines must be used.
In its most complex form, the model re-
quires five types of preprocessed input
data (upper-air data, surface data,
precipitation data, emissions data, and dry
deposition velocity data). The preproces-
sors should be run in sequence, because
the output file format of one program is
the input file format for the next sequen-
tial program. Therefore, once the user has
formatted the first file in the sequence, he
does not have to format the remainder of
the files, because their formats are
identical.
-------
The RELMAP model itself consists of
one primary program (MAIN) and 17
subroutines, which are called by MAIN.
Besides calling the subroutines, this pro-
gram performs a number of other func-
tions such as determining when puffs
should be generated, tracking the number
and location of puffs, implementing the
chosen output mode, and defining vertical
profiles of the pollutants. MAIN also con-
tains all the COMMON blocks used by the
model to pass information.
An example execution of the RELMAP
model and its preprocessors, which uses
the Executive Control Language (ECL) of
the Sperry UNIVAC system, is provided. In
this example execution, the preprocessors
and both output versions of the model are
run for the month of January 1980 for a
12° x 13° window located within the
model's default domain. This window,
which extends from 30° to 43 °N and
from 80 ° to 92 °W, contains a total of 156
grid cells.
An annotated precis of the tape that ac-
companies the user's guide is presented
in the full report, which contains a brief
description of the 14 files on the tape. The
first four files are program files that con-
tain the program elements for the model
and the preprocessors; the ten remaining
files contain the data necessary to perform
the example execution.
The EPA authors Brian K. Edor (also the EPA Protect Officer, see below), D. H.
Coventry, and T. L. Clark are with the Atmospheric Sciences Research
Laboratory, Research Triangle Park, NC 27711; and Catherine E. Bellinger is
with Computer Sciences Corporation, Research Triangle Park, NC 27709.
The complete report, entitled "RELMAP: A Regional Lagrangian Model of Air
Pollution User's Guide," (Order No. PB 86-171 394/AS; Cost: $ 16.95, 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/S8-86/013
0000329 PS
U S ENVIR PROTECTION AGENCY
REGION 5 LIBRARY
230 5 DEARBORN STREET
CHICAGO IL 60604
------- |