United States
Environmental Protection
Agency
Environmental Research
Laboratory
Duluth MN 55804
Research and Development
EPA-600/S3-84-061 June 1984
&EPA Project Summary
Intermediate-Range Grid Model
and User's Guide for Atmospheric
Sulfur Dioxide and Sulfate
Concentrations and Depositions:
Wisconsin Power Plant
Impact Study
Kenneth W. Ragland and Kenneth E. Wilkening
A three-dimensional time-dependent
grid type model for two chemically reac-
ting species that undergo atmospheric
transport, diffusion, and wet and dry
deposition in the atmosphere boundary
layer is described. The model is useful
for assessing a single source or a group
of sources on a scale of 10s to 100s of
kilometers. The equations are solved by
an implicit finite-difference scheme. The
model is shown to accurately treat
advection, diffusion, time step, and grid
size. The sensitivity of the model to
chemical reaction rate, surface dry
deposition rate, surface roughness,
horizontal diffusivity, pH of the rain and
emissions rate is demonstrated. The
model is applied to sulfur dioxide and
sulfate deposition in the Rainy Lake
Watershed using the available emissions
and meteorological data. Model calcula-
tions are compared to NADP wet depo-
sition data and snow core data. Model
calculations for the Atikokan, Ontario
power plant are also presented.
This Project Summary was developed
by EPA's Environmental Research Labo-
ratory, Duluth, MN, to announce key
findings of the research project that is
fully documented in a separate report of
the same title (see Project Report order-
ing information at back).
Introduction
Much attention is currently being given to
acid impacts of substances carried by and
transformed in the atmosphere. Many
models and methods have been proposed
and tested. The computer model herein
described addresses sulfur dioxide and
sulfate deposition (wet and dry) and the
chemical transformation of sulfur dioxide to
sulfate. The sulfur depositions as they alter
the pH of the soil and water are not a part
of the model. Nitrogen compounds also pro-
duce acidification, but these are not yet in-
corporated into the model. Hence, the sulfur
loadings simulated by the model can only be
taken as indicators of potential acidification.
Atmospheric deposition models may be
divided into two major classes: (1) long-
range transport models, where the region of
interest extends over thousands of kilo-
meters (i.e., the eastern half of the U.S.);
and (2) mesoscale models in which analysis
extends over hundreds of kilometers (i.e.,
the state of Wisconsin). Some advantages
of mesoscale over macroscale models are:
providing higher resolution and more detailed
simulation of meteorological conditions;
focusing on smaller geographical entities
(watersheds, lakes, sensitive biological
regions, etc.); and predicting the relative in-
fluence specific emission sources on nearby
receptors. The main disadvantage of a meso-
scale model is that it is not possible to predict
absolute depositions within the region due
to omission of extra-regional emission
sources and transport.
This report describes a mesoscale
(10s-100s kilometer scale) computer model
for two chemically reacting species which
-------
undergo atmospheric transport, diffusion,
and dry and wet deposition. Application of
the model to sulfur dioxide and sulfate
deposition in the Rainy Lake Watershed in
northern Minnesota and western Ontario is
presented. The model is a three-dimensional,
time-dependent grid which uses an implicit
finite-difference numerical scheme to solve
two coupled species conservation of mass
equations in the atmospheric boundary layer.
The primary purpose for developing the
model was to be able to estimate sulfur diox-
ide and sulfate deposition from the at-
mosphere to the ground over a medium size
(103 to 105 km2) geographical region. One of
the central questions which such a model
can answer is how much in-region sulfur
emissions are deposited locally and how
much leave region boundaries. Other ques-
tions which such a model can answer are:
What is the relative deposition contribution
of single sources to a particular receptor
location? How might one best locate moni-
toring stations given the deposition distribu-
tion pattern predicted by the model? What
is the deposition impact on any given sub-
region (i.e., a watershed) within the larger
region?
Description of UWATM-SOX
Model
In this section the salient features and
operational aspects of the model are
described.
Summary of Model Features
The UWATM-SOX model is a time-
dependent, cell-type model. At its heart is
the numerical solution of coupled S02 and
S04 conservation of mass equations de-
scribed in an Eulerian (fixed) frame of
reference for a dilute species in the at-
mospheric boundary layer. Included are the
processes of wind transport, turbulent dif-
fusion, chemical reactions, source input and
dry and wet deposition. The numerical solu-
tion is a first-order, fully implicit scheme.
The model inputs emissions once at the
beginning of the simulation and new mete-
orological data every hour. It performs
calculations every 20 minutes within 1 h.
(The time-step is variable and can be altered
if necessary.) As long as data are available,
any time period for simulation can be
chosen. The model has been run for time
periods up to 1 yr.
The model region is divided into a uniform,
horizontal, two-dimensional grid network.
The rectangular grid size depends on the ex-
tent of the region, the desired cost/efficiency
of the computer runs and the desired ac-
curacy of analysis. Above the grids are
stacked a specified number of vertical cells
from the ground level to an upper limit called
the mixing height. At present the model uses
six vertical cells — three lower fixed-height
(50 m) and three upper variable-height cells.
(This arrangement can be altered.) There-
fore, the space enclosing the region of study
is partitioned completely by a three-di-
mensional cell structure. The single 3D cell
is the basic unit within which the model
calculations are performed.
The model can handle point, line and area
emission sources. The sources are assigned
within one of the grids. All sources are
treated simultaneously, therefore, a specific
source-to-concentration/deposition value
cannot be determined unless the single
source is run separately. The emissions data
are obtained from outside organizations
(usually government control agencies). For
point sources a plume rise is calculated using
one of a set of five equations. The resulting
"effective stack height" is placed within the
appropriate vertical cell. The meteorological
condition known as lofting (effective stack
height above mixing height) is accounted for.
This is an important condition affecting long-
range pollutant transport.
The meteorological data are obtained on
tapes from the National Climatic Center
(NCC) in Asheville, NC. The tapes required
are: hourly surface observations, twice daily
mixing heights and hourly precipitation. An
integrating program (METDATA) converts
the data to the hourly format required by the
model.
The atmospheric boundary layer is char-
acterized by a set of wind speed and eddy
diffusivity profile equations which are
calculated from ground level to the top of
the mixing layer. The profiles describe the
transport and diffusion behavior of the
pollutants. They are vertically variable but
horizontally uniform and constant across the
region of study over a 1 h period. The pro-
files are dependent on the stability of the at-
mosphere.
The boundary layer has been subdivided
into a surface layer and a second layer up
to the mixing height. Therefore, each pro-
file has two forms, one above and one below
the surface layer. The surface layer equations
are based on the Monin-Obukov similarity
theory. In the upper layer the wind profile
is a form of the power-law and the diffusivity
is assumed constant. The profiles are cal-
culated from a knowledge of the geostrophic
wind speed, net heat flux to the ground and
surface roughness. The geostrophic wind is
calculated from a knowledge of the surface
wind speed and height at which it is mea-
sured using a modified Regula Falsi iteration
scheme. The net heat flux is calculated by
first using the NCC surface observations data
and the Pasquill-Turner method of stability
classification to determine the stability class.
Then a heat flux value is assigned to each
class. A single surface roughness value is
chosen to characterize the entire region.
The chemical conversion from S02 to S04
is assumed to be first order. At present two
reaction rates are used — a day value and
a night value — although hourly changes are
possible if the data is available.
The dry deposition is handled by means
of a deposition velocity which is calculated
as a function of terrain and meteorological
conditions. The deposition velocity changes
hourly, but is uniform across the region. It
is determined from an aerodynamic resis-
tance which depends on the meteorological
conditions close to the ground, and a sur-
face resistance which depends on the pollu-
tant species and nature of the surface (snow,
water, vegetation, etc.). The surface re-
sistances are input by the user and aero-
dynamic resistance is internally calculated
from wind profile information.
The wet deposition scheme is based
primarily on a model developed by Scott
(1978). It assumes all precipitation forms ac-
cording to the Bergeron or cold cloud pro-
cess. Ice crystals form in the upper portion
of the cloud and as they sweep downward
through the cloud they capture cloud drop-
lets containing SO4 and dissolved SO2. The
S eventually incorporated into the precipita-
tion is assumed to be drawn in through the
cloud base. A set of three equations for the
wet deposition removal rate are used; one
for rain or snow removal of S04, and one
each for rain and snow removal of S02.
Unlike dry deposition, the wet removal of
S02 and S04 is calculated separately from
the numerical solution of the atmospheric
dispersion equation because it must consider
an entire column of cells from ground level
to mixing height, as opposed to single cells.
Precipitation is assumed to occur uniformly
across the region.
A significant feature of the model is a
complete set of regional mass summations
within the main program. Their purpose is
to continually double-check to ensure mass
is conserved at all stages during the internal
calculations and to provide the basis for
deriving a regional atmospheric S budget for
the modeled emissions sources within the
region.
There are several computer techniques
worth noting. First, to handle the numerical
calculations for arbitrary wind directions a
dual grid system is used, one fixed in posi-
tion and one which overlaps the fixed system
and rotates with the wind angle. Transfers
of concentrations are performed back and
forth between the two systems as the wind
direction shifts. Second, the main program
has a "stop-start" capability so that progress
can be stopped anywhere in the stream of
-------
calculation and started exactly where it left
off with no outside intervention other than
re-starting the run. Third, extensive use has
been made of common blocks and the
"FORTRAN Procedure (PROC)" statement
to eliminate unnecessary interprogram
storage locating calls and to make it easier
to make internal changes which affect many
programs at once. Fourth, a very complete
record of the values of all significant variables
is stored on tape hourly for future analysis.
The saved information includes the date,
ground level concentration and deposition
arrays, and certain meteorological, emissions
and regional mass balance data. All this is
compactly stored for each hour as a record
of a tape file. The file covers the entire time
period of simulation.
The model results include 3-h, 24-h,
seasonal and annual average ground level
concentrations, and dry, wet and total
depositions of S02 and SO,,. (These are the
standard averaging periods but the model is
capable of averaging upper level concentra-
tions or using other time periods.) The results
also include pH of the precipitation during
a storm event, and the values and dates of
the 3-h and 24-h maximum concentrations.
All of the above results are printed grid by
grid in an array format and can be isopleth
computer plotted. The results include — a
frequency distribution of concentrations/
deposition values for any selected grid, and
a regional atmospheric S budget which
analyzes, by total mass and percentage of
S input, the flow of S (S02 and S04) through
the various physical and chemical processes
into, out of or within the region.
The model is written in ASCII Fortran and
has been run on the Madison Academic
Computing Center's SPERRY UNIVAC
1100/82 series computer. In an application
of the model to the BWCA wilderness area
in northern Minnesota in which a cell struc-
ture of 11 x 13 x 6 was used with hourly
changes in meteorological data and calcula-
tions carried out every 20 min, the main pro-
gram required 50K core space, and a 1 yr
simulation cost $200 (at the cheapest com-
puter rate) and took 6 hr of CPU time. A 1
day calculation at the "normal" rate (10 x the
cheapest) can be done for $7 and takes 1 min
CPU time.
A complete description of the model and
how to use it is contained in Wilkening and
Ragland (1982) and summarized in Ragland
and Wilkening (1983).
Overview of Mode/ing Operation
This section is an examination of the total
modeling operation: programs, subroutines,
input data and output results, the sequence
of program execution, and storage modes
and formats. A flow chart of the modeling
operation is shown in Figure 1 with the
model features shown in Table 1.
There are five major components to the
modeling effort:
1) Development of emissions inventory
2) Development of meteorological
data-base
3) Simulation of atmospheric transport,
diffusion, chemical reactions and de-
position
4) Processing of hour-by-hour concentra-
tion and deposition output
5) Isopleth computer plotting
In all there is a collection of 11 programs, 11
subroutines, and one FORTRAN PROC
used. They are stored in a program file called
UWATM*ACIDRAIN. Each of the above
tasks are briefly discussed to give the user
an overview of the entire sequence of model
operation.
Emissions Inventory
The model is capable of handling point,
line, and area sources. A detailed meth-
odology for converting agency inventories to
a model form has been worked out for point
sources. Line and area source emissions are
grid apportioned. For each point source the
model requires a set of six variables: loca-
tion, stack height, stack diameter, gas tem-
perature, gas flow rate and emission rate.
Five steps involved in converting outside in-
ventory data to usable form are:
1) Obtain complete inventories from out-
side organizations
2) Select largest sources
3) Combine sources
4) Locate sources within grid network
5) Convert data to model format
Carrying out these steps involves compil-
ing four tables and executing two programs.
The first two tables are a record of what data
was selected for use from the original inven-
tories and a record of all alterations (and
reasons for) made in the process of selec-
tion. For each of the selected sources a
"potential plume rise" is calculated (using a
program called PLUMEPOT) and this value
is used as the basis of combining sources to
reduce the inventory to manageable size (if
necessary). The second two tables are a
record of the combined sources and a record
of all alterations (and reasons for) made in
the process of combining. After compiling
these tables, the only remaining task is to
grid locate the source. This is done by hand,
using U.S. Geological Survey maps. The
resulting grid ID numbers are recorded on
the combined sources table. Finally, a pro-
gram called MODEL-EMISS takes the com-
bined sources data and changes units and
places the final model emissions inventory
on a mass storage file which is directly ac-
cessed by the model.
Meteorological Data-Base
The model requires a set of seven hourly
meteorological variables: stability, heat flux,
wind speed, wind direction, air temperature,
mixing height, and precipitation. All are
directly or indirectly obtained from a set of
three NCC tapes. There are three steps in-
volved in converting the original NCC tape
data to the model usable form:
1) Purchase hourly surface observations,
twice daily mixing height, and hourly
precipitation tapes from NCC for a
given meteorological station and given
year.
2) Pre-process data with set of three UDH
programs.
3) Execute METDATA program.
Each of the NCC tapes are run through a
separate Unified Data Handler (UDH) pro-
gram (called SFCOBSDATA, MIXINGHT-
DATA, PRECIPDATA), which together
extract the hourly wind speed, wind direc-
tion, air temperature, cloud cover, ceiling
height, precipitation, and afternoon mixing
height data. The data is stored as three files
on another user-supplied tape which is ac-
cessed by the METDATA program. This pro-
gram performs several jobs: 1) uses the
Pasquill-Turner stability class model to deter-
mine the hourly stability, 2) uses the basic
earth-sun angular equations to calculate
sunrise and sunset times and solar altitude
(which are used both in the stability and mix-
ing height determinations), 3) uses a linear
interpolation scheme to estimate hourly mix-
ing height values, 4) determines precipitation
type (rain or snow), 5) assigns hourly heat
flux values on the basis of wind speed and
stability information, 6) converts the units of
wind speed, direction and air temperature,
and finally, 7) stores all the hourly values for
an entire year on a mass storage file which
is directly accessed by the model.
Atmospheric Transport, Diffusion,
Chemical Reaction, and
Deposition Simulation
This component is "the model." It con-
sists of one main program and 10 subrou-
tines. These contain the computer code
equivalent of the mathematical equations ex-
pressing the wind transport, turbulent diffu-
sion, chemical transformation and dry and
wet deposition processes affecting S02 and
S04. The central equation is a pair of cou-
pled continuity equations for dilute pollutant
species in the atmospheric boundary layer
described in an Eulerian (or fixed) reference
system. The simulation technique is a first-
order, fully implicit solution of the finite dif-
ference form of the coupled equations and
are described completely in Ragland and
Wilkening (1983).
-------
Emissions Data
Outside
Agency ~
Inventories
Meteorological Data
3
NCC
Tapes
Plumepot
Model-Emiss
Atmospheric
Simulation
Final Results
UWATM-SOX
Subroutines:
RISE CHOLLU
WINDIF CHOLEQ
DEPVEL DPINDP
WETDEP
QTFM
RFTFM
FRTFM
Other Input Data
User-
Supplied
Values
Figure 1. Schematic of entire modeling operation.
Table 1. Summary of Model Features
Feature
Description
General
Emissions
Meteoro-
logical Data
Cell-type model
Time-dependent (hourly)
Mesoscale analysis 110s to 100s km)
Run in batch mode
Numerically solves coupled SO2 and SO, conservation of mass equations
for a dilute species in the atmospheric boundary layer described in an Eulerian
frame of reference
Includes physical processes of:
wind transport
turbulent diffusion
chemical reactions
source input
dry deposition
wet (rain or snow) deposition
Uses 2D uniform horizontal grid network over region of study,
Uses set of 6 (3 fixed at 50 m and 3 variable) 3D cells in the vertical height
from ground to mixing height
Emission data is input once,
meteorological data is input hourly,
model calculates at 20 min intervals within 1 h
Numerical solution is of first-order fully implicit type
Can handle point, line, area sources
Treats all emission sources simultaneously
For point sources — calculates a plume rise (using 1 of a set of 5 Briggs
or Moses-Carsons equations)
— accounts for lofting condition
— SO, stack emissions X% of S02 (where X is supplied
by user)
Uses data tapes from National Climatic Center (NCC)
Fully computerized'processing of tapes to convert data to hourly format re-
quired by model
For a region which has been partitioned
by a 3D cell structure, the equations are
solved in a downwind-step (x-step) and time-
step (t-step) fashion. They are solved for one
complete block of cells in the yz-plane and
within the block they are solved for each t-
step for the specified number of steps (cur-
rently three, thus calculating at 20-min in-
tervals). This procedure is repeated block by
block from the upwind edge to the down-
wind edge of the region. The matrix solution
for one x-step and one t-step can be sym-
bolized as:
[C] = [A]-1 [D]
where
[C] contains the unknown concentrations,
[A] the known meteorological, chemical
and deposition data, and
[D] the known concentrations from the
previous x-step and t-step.
This matrix equation is solved for each x-step
and each t-step in the whole region for a 1-h
period, then repeated with new input data.
The whole process is repeated hourly until
the end of the simulation period. All these
calculations are performed in the main pro-
gram (UWATM-SOX) with the help of three
matrix handling subroutines (CHOLLU,
CHOLEQ, DPINDP).
To arrive at the solution, numerous sub-
sidiary calculations (mainly carried out by the
-------
Table 1.
Feature
(Continued)
Atmospheric
Boundary
Layer
Chemical
Reactions
Dry
Deposition
Wet
Deposition
Computer
Techniques
Output
Results
Description
Characterized by set of wind speed and turbulent eddy diffusivity pro-
files, assumed vertically variable but horizontally uniform and constant
within region over 1 h period
Theory uses concepts of stability, geostrophic wind, surface roughness and
mixing height
Layer divided into two regions — surface layer and layer between top of
surface layer and mixing height
Profile equations for surface layer based on the 1954 Monin-Obukov similarity
theory, equations for upper region based on a form of wind speed power-
law and on assumed constant diffusivity
Profiles function of stability (stable, neutral, unstable); stability determined
using the 1961 Pasquill-Turner model
Profiles function of geostrophic wind; geostrophic wind determined solely
from knowledge of surface wind and the height at which it is measured using
a Modified Regula Fa/si iteration scheme
Profiles function of surface roughness; single user-supplied roughness value
assumed to be constant and uniform over region
Profiles function of mixing height; hourly mixing height determined from
NCC Twice Daily Mixing Height tape using linear interpolation scheme
Diffusivity profile based on Mixing-Length Model
Linear and homogeneous
Uses two conversion rates (based on work by Husar) — 2%/hr (day),
0.5%/hr (night)
— Uses deposition velocity of form VD =
1
where r accounts
for atmospheric conditions and rs for terrain
— VD assumed uniform and constant over region
— Based on model by Scott (1978)
— Assumes only Bergeron or cold cloud precipitation development process
— Assumes removal over entire vertical column from ground to mixing height
at once (so calculation is separate from cell-by-cell finite difference
calculations)
— Set of 3 wet removal equations developed:
1. S04 rain or snow
2. SO2 snow
3. SO2 rain
— Precipitation is assumed to occur uniformly across the region and uniformly
over a 1-h period
- Written in ASCII FORTRAN
— Uses complex set of internal total mass summations as basis for
1. continual internal conservation of mass check
2. basis of regional total atmosphere S budget
— Employs dual-grid system to handle arbitrary wind directions
— Employs stop-start capability
— Employs extensive use of COMMON'S and the FORTRAN Procedure (PROC)
— Stores compactly on tape complete set of hourly calculations
— Ground level concentration 3-h
Dry deposition 24-h
Wet deposition SO2 and S0t Seasonal Averages
Total deposition Annual
— Ground level concentration
Dry deposition Maximum 3-h
Dates
S02 and S0t and Values
Wet deposition Maximum 24-h
Total deposition
(All the above are printed on a grid array basis and can be isopleth com-
puter plotted)
— Concentration frequency distribution for any selected grid point
— pH of precipitation
— Total atmospheric SO2/SOt budget for in-region emission sources
subroutines) must be performed. These
include:
1) For point sources, calculate a plume
rise and determine if lofting occurs
(RISE subroutine).
2) Calculate the wind and diffusivity pro-
files (WINDIF subroutine).
3) Calculate the deposition velocity
(DEPVEL subroutine).
4) Calculate the wet deposition removal
(WETDEP subroutine).
5) Shift emission source coordinates as
wind direction changes (QTFM sub-
routine).
6) Transfer concentrations between the
dual fixed and rotating cell systems as
wind direction changes (FIXROT and
ROTFIX subroutines).
Output Processing
Hourly the main program outputs ground
level concentration arrays, dry and wet
deposition arrays, certain meteorological
data, the wind and diffusivity profiles, cer-
tain emissions data and certain total regional
mass calculations. A program called
CDTAPE (for concentration/deposition tape
processor) does all the summing, averaging
and other manipulations of the hourly out-
put. The program is complex but highly flex-
ible. Besides printing the results, CDTAPE
stores them on tape in a format suitable for
computer isopleth plotting.
Computer Plotting
Our group uses the WISMAP2 computer
graphics package developed by the Univer-
sity of Wisconsin Cartography Laboratory.
To use this package it is first required that
an outline map of the region of study be
"digitized." Once this is done, a wide variety
of plotting techniques are available through
WISMAP2. The plotting process will not be
explained in detail because it is assumed
users will have other similar plotting pack-
ages at their disposal.
Accuracy and Sensitivity
Analysis
In this section accuracy refers to how the
model conforms to expected and known
results when given "ideal," i.e., artificial and
highly restrictive, input data. Sensitivity
refers to the model's relative response to
changes made in important parameters. Ac-
curacy and sensitivity analysis together with
comparison of model predictions to monitor-
ing results provide a basis for judging the
model's reliability and usefulness.
The accuracy of the model was tested in
five simple cases:
1) Pure Advection — The simplest case is
to consider a single steady source under
constant meteorological conditions in
5
-------
Table 1.
Feature
(Continued)
Description
Cost/Time — For 11 x 13 x 6 cells,
of hourly meteorological and emission input,
Application 3 time-steps/h,
50K computer core,
a run on SPERRY UN/VAC 1100/82 series computer,
One year simulation:
Cost = $200 (cheapest rate)
CPU Time = 6 h
One day simulation:
Cost = $7 (normal rate, 10x cheapest!
CPU Time = 1 min
which only a constant steady wind trans-
ports the pollutants.
2) Advection with Diffusion — The next
simplest case to test is that of a single
source under constant meteorological
conditions with a constant steady wind
and constant lateral diffusivity.
3) Time Step and Grid Size — The time step
size was halved (from a 20-min to a
10-min interval). The conclusion is that
increasing the number of calculations
above three does not affect the accuracy
of the results.
4) Grid Subdivision — The model uses a
double grid system to handle arbitrary
wind direction. One grid is assumed fixed
in position and the other rotates with
changing wind directions.
To investigate the sensitivity of the model,
six parameters ranging from doubling reac-
tion and emission rates to defaulting pH
values by increasing one unit were tested.
Results are indicated in Table 2.
Case Study — Rainy
Lake Watershed
To investigate acid rain effects on a sen-
sitive wilderness area the UWATM-SOX
model was applied to the Rainy Lake
Watershed.
Questions to be answered are: How much
of the regional S emissions are deposited
within the region? How much leaves the
Table 2. Summary of Sensitivity Analyses
region? How does this compare to the avail-
able monitoring data in an annual, seasonal
and weekly basis? The years 1976 and 1978
were chosen for the case study because 1976
was the year of the first comprehensive emis-
sions inventory, and 1978 was the most re-
cent meteorological data set available as well
as the start of the National Atmospheric
Deposition Program data at Marcell, Min-
nesota.
Description of Region and
Data Input
The Rainy Lake Watershed is located in
northeastern Minnesota and western On-
tario. It includes the Quetico Provincial Park,
the Boundary Waters Canoe Area and
Voyageurs National Park. This area lies
within two days travel of 50 million people
and yet contains extensive prime wilderness
areas. The modeling area is the 330 by 390
km region of the Rainy Lake Watershed
where the major emission sources are in
Duluth, the Missabe Iron Range, Interna-
tional Falls and Atikokan, Ontario (Ragland
and Wilkening, 1983).
The total emissions of S02 due to point
sources in the region in 1976 were 3,232 g/s
or 112,000 tons/year. In 1978 the S02 emis-
sions were 3,139 gs or 109,000 tons/year.
The area source emissions of S02 due to
space heating and small industrial and com-
mercial sources were not available, but are
Average
Concentration
3-h Maximum
Concentration
believed to be small in this region. Direct S04
emissions were assumed to be 3% of the
S02 emissions by weight, which is a gener-
ally observed result from stack sampling. The
regional S02 emissions of 112,000 tons/year
may be compared to the Minnesota state-
wide total of 521,000 tons/year, Wisconsin
937,000 tons/year and Illnois 2,344,000
tons/year.
The meteorological data were obtained
from the NCC for International Falls, Min-
nesota, for the years 1976 and 1978. The
other data input parameters used were the
same as the base case in the previous
section.
The annual average S02 concentrations
away from any sources were calculated to
be about 0.5 u.g/m3, while the ambient S04
concentrations were about 10 times smaller.
The 24-h worst-case S02 concentrations
away from any sources were 5 /^g/m3 and
the 24-h worst-case SO, concentrations also
were 5 pig/m3. These levels are reasonable
in view of the relatively few sources in the
region. Monitoring at the remote Fernburg
site in the region showed essentially no
readings above the 5 u.g/m3 threshold of the
instrument.
Annual and Seasonal Deposition
The (1976) annual wet plus dry deposition
isopleths resulting from 35 emission sources
within the region are shown in Figure 2. The
areas of largest annual deposition center
around the Clay Boswell power plant,
Calland and Steep Rock mines, Taconite
Harbor and Duluth.
The regional mass balance, presented in
Table 3 for 1976, shows that 17% of the S
emissions were deposited in the region due
to dry deposition and 0.3% was retained due
to wet deposition. Whereas 82% of the S
emissions were transported out of the region
on an annual basis. Of the S deposited in the
region 8.5 x 109 g out of 9.4 x 109 g were
deposited by dry deposition SO2. The re-
gional mass balance for 1978 was similar.
Dry Deposition
Wet Deposition
Cases
SO,
S04
SO,
S04
SO,
S04
S02
S04
Double reaction rate 0.93 1.68 0.89 1.74 0.96 1.64 0.94 1.66
Double sulfate surface 1.00 1.07 1.00 1.09 1.00 0.55 1.00 1.02
resistance
Reduce surface roughness by 1/2 1.06 1.16 1.18 1.23 1.25 1.18 1.25 1.24
Reduce horizontal diffusivity 0.99 1.00 1.00 1.01 1.01 1.01 1.01 1.00
by 1/2
Set default pH to 5.8 0.93 0.97 0.98 0.98 0.93 0.98 2.30 0.96
Double emissions 1.99 1.97 1.99 1.98 1.97 1.94 J.98 2.01
-------
Rainy Lake Watershed'-®
49 45
4900
WCA
—4800
13
9400
9300
9200
91 00
— 4700
4630
9000
37.28 Miles
0 30 60 Km
Figure 2. Map of the study region where 35 emission sources result in the computed annual
(1976) wet plus dry deposition ofSO~i as kg SOVha. The SOidepositions have been
multiplied by 1.5 (the ratio of molecular weights) and combined with the S04 values
to represent the total loading from regional sources.
Table 3. Regional Mass Balance for 1976
Emissions
Chemical reaction
Dry deposition
Wet deposition
Transport out of region
SO2
(10* g/yr)
102
6
17
1
78
SO,
(10* g/yr)
3.1
9.8
0.6
0.5
11.8
S
(10* g/yr)
52.3
8.7
0.7
42.9
It is of interest to compare the deposition
during the snow season (November-April)
with the non-snow season (May-October).
The calculations are summarized for grid (2,
9) the Marcell, Minnesota NADP monitoring
site, in Table 4. The greatest percentage of
S deposition occurs by dry deposition of S02
during the summer. Wet deposition of S04
is comparable in winter and summer, al-
though total wet deposition is greater in the
summer. It is believed that the reason wet
deposition of S04 is similar in winter and
summer is that there are more hours of
precipitation in winter than in summer
although the total amount of precipitation is
greater in the summer. The wet deposition
in 1978 is nearly twice as great as 1976
because of increased precipitation.
Comparison With Monitoring
Data — NADP and Snow Cores
The National Atmospheric Deposition Pro-
gram located a site in the region at Marcell,
Minnesota, starting in July 1978. The NADP
site measures wet S04 deposition on a
weekly basis from inter- and intraregional
transport. Total wet S04 deposition moni-
tored at the site was considerably more than
that calculated for the site, mainly due to
transport of SO, from outside the region. Dry
deposition is not measured reliably at the
NADP site and hence cannot be compared.
Sixty-five snow cores taken in April 1979
in the BWCA provide valuable information
on wet plus dry deposition. They yielded S04
depositions mostly in the range of 1 to 2
kg/ha, whereas the regional model predicted
0.5 kg/ha. The NADP site at Marcell mea-
sured 1.6 kg/ha S04 wet deposition during
the 1978 to 1979 snow season.
Extreme-Case Weeks
The calculated wet and dry deposition
fluxes also may be compared on a weekly
basis for several extreme case weeks. The
weekly S04 wet deposition data at the
Marcell NADP site show that a few big
events account for much of the annual SO4
wet deposition. The wet S04 deposition is
clearly dominated by long-distance transport
of S04 during the worst-case weekly events.
Single Source
The model was also run for a new single
source to be located at Atikokan, Ontario
(Ragland and Wilkening, 1983). The emis-
sions were 70.5 x 109 g/yr of S02 and 2.1
x 109 g/yr of SO4. The stack height was 200
m and the plume rise was typical of a coal-
fired power plant. The model also was run
with the stack height reduced to 100 m.
Overall, 18.6% of the S remained in the
region compared to 15% at full stack height.
Table 5 gives the results of the predicted
mass balance for sulfur emissions.
References
Ragland, K.W. and K.E. Wilkening. 1983.
Intermediate-Range Grid Model for At-
mospheric Sulfur Dioxide and Sulfate
Concentrations and Depositions. Atmos.
Environ. 17: 935-947.
Scott, B.C. 1978. Parameterization of
Sulfate Removal by Precipitation. J. Appl.
Meteorol. 17: 1375-1389.
Wilkening, K.E. and K.W. Ragland. 1982. A
User's Guide to the University of Wiscon-
-------
Table 4. Calculated Regional Deposition at the Marcell, Minnesota Site
Dry
S Deposition (kg/ha)
Winter S02 deposition
Summer S02 deposition
Winter SO, deposition
Summer SO, deposition
Winter deposition as SO,
Summer deposition as SO,
Annual deposition as SO,
1976
0.24
1.23
0.02
0.02
0.36
1.86
2.22
1978
0.21
1.13
0.02
0.02
0.34
1.72
2.06
Wet
1976
0.01
0.10
0.03
0.03
0.06
0.18
0.23
1978
0.02
0.19
0.03
0.06
0.05
0.35
0.40
Wet and Dry
1976
0.25
1.33
0.05
0.05
0.41
2.04
2.45
1978
0.23
1.32
0.05
0.08
0.39
2.07
2.46
Table 5. Mass Balance Due to Single Source at Atikokan3
S02 (10" g/yr) SO, {JO3 g/yrt
S 110* g/yr)
Emissions
Chemical reactions
Dry deposition
Wet deposition
Transport out of region
70.5
4.7
10.0 (12.4)
0.4 ( 0.4)
55.4 (52.9)
2.1
7.1
0.3 (0.5)
0.4 (0.4)
8.4 (8.4)
35.9
—
5. 1 ( 6.4)
0.3 ( 0.3)
30.5 (29.2)
"Note: Numbers in parentheses represent a stack height of 100 m instead of 200 m.
sin Atmospheric S02/S04 Air Pollution
Computer Model (UWATM-SOX). U.S.
Environmental Protection Agency, Cincin-
nati, Ohio. (Available from NTIS.)
Kenneth W. Ragland is with Department of Mechanical Engineering, The
University of Wisconsin, Madison, Wl 53706.
Gary E. Glass is the EPA Project Officer (see below).
The complete report, entitled "Intermediate-Range Grid Model and User's Guide
for Atmospheric Sulfur Dioxide and Sulfate Concentrations and Depositions:
Wisconsin Po wer Plant Impact Study." (Order No. PB84-189 257; Cost: $ 13.00.
subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, v'A 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Research Laboratory
U.S. Environmental Protection Agency
Duluth. MN 55804
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
U.S. GOVERNMENT PRINTING OFFICE: 1984-759-102/980
------- |