-------

-------
                                                  EPA-600/4-89/026
                                                         June 1989
 RE-EXAMINATION OF INTERIM ESTIMATES OF ANNUAL SULFUR
   DRY DEPOSITION ACROSS THE EASTERN UNITED STATES
                          by

          Terry L. Clark and Robin L. Dennis
        Atmospheric Sciences Modeling Division
Atmospheric Research and Exposure Assessment Laboratory
         U.S. Environmental Protection Agency
     Research Triangle Park, North Carolina 27711

                         and

                   Steven K. Seilkop
           Analytical Sciences, Incorporated
             100 Capitola Drive, Suite 106
     Research Triangle Park, North Carolina 27713
                Contract No. 68-03-3439
                    Project Officers

          Robin L. Dennis and Terry L. Clark
        Atmospheric Sciences Modeling Division
Atmospheric Research and Exposure Assessment Laboratory
           Research Triangle Park, NC 27711
ATMOSPHERIC RESEARCH AND EXPOSURE ASSESSMENT LABORATORY
          OFFICE OF RESEARCH AND DEVELOPMENT
         U.S. ENVIRONMENTAL PROTECTION AGENCY
           RESEARCH TRIANGLE PARK  NC 27711

-------
                                    NOTICE
The information in this document has been funded wholly or in part by the
United States Environmental Protection Agency under Contract No.  68-01-6849
to Kilkelly Environmental Associates, Raleigh, North Carolina, and subcon-
tracted to Analytical Sciences, Incorporated, Research Triangle Park, North
Carolina.   It has been subjected to the Agency's peer and administrative
review, and it has been approved for publication as an EPA document.   Mention
of trade names or commercial products does not constitute endorsement or
recommendation for use.
                                      ii

-------
                                   ABSTRACT
     During the summer of 1987 annual amounts of sulfur dry deposition were
first estimated for more than 7,000 lakes in the eastern United States.  These
estimates, heretofore termed interim estimates since they were expected to be
superceeded in the near future, were derived from predictions of the Regional
Acid Deposition Model (RADM) adjusted using the empirical data from two moni-
toring networks.  Since that time, additional years of empirical data have
become available and a portion of the previously available empirical data has
been superseded.  Consequently, the process of estimating annual amounts of
sulfur dry deposition was repeated to determine whether these interim
estimates should be revised, and if so,  by how much.  This study concludes
that the interim estimates appeared to be too low by 13% and recommends that
the interim estimates be systematically increased by the same amount.

     A comparison of the revised estimates to empirically-derived sulfur dry
deposition amounts suggests that there is some systematic error in the revised
estimates.  Adjusted RADM predictions of dry deposition tend to be biased low
in the most significant source regions (where at least 200 ktonnes S02/yr are
emitted within 80 km of the site).  Conversely, in locations farther removed
from significant sources (81-160 km) there is evidence that the estimates are
biased high.  However, in general, sulfur dry deposition estimates from
adjusted model predictions are within ±60% of the  empirical data.
                                      111

-------
                                CONTENTS


Abstract	iii
Figures	vi
Tables	vii
Acknowledgements 	  viii


   1.   INTRODUCTION 	   1

   2.   REASONS FOR RE-EXAMINING THE INTERIM ESTIMATES 	   3

        2.1  Revision of Empirical Data at U.S.  Sites	3
        2.2  Availability of Additional Data To  Characterize
             Interannual Variability 	   5
        2.3  Possible Improvement of Estimates Using
             Additional Modeling Results 	   7

   3.   SPATIAL DISTRIBUTION OF ANNUAL SULFUR DRY DEPOSITION ...   8

        3.1  Comparison of Model Predictions and Empirical Data.  .   8
        3.2  Uncertainty Assessment of the Revised Estimates ...  20

             3.2.1  Uncertainty of Adjusted RADM Predictions ...  20
             3.2.2  Uncertainty of the Empirical Estimates ....  25
             3.2.3  Interannual Variability of the Empirically-
                    Derived Estimates	27

        3.3  Estimated Seasonal Distribution of  Sulfur Dry
             Deposition	30

   4.   CONCLUSIONS AND RECOMMENDATIONS	31

References	33
                                      v

-------
                                FIGURES
Number
   1      The location of 22 sites of the CORE and APIOS net-
            works from which empirically-derived amounts of
            sulfur dry deposition were available for at least
            one year	10

   2      Empirically-derived annual amounts of sulfur dry
            deposition (kg S/ha/yr)	11

   3      Annual 1980 S02 emissions (10 ktonnes/yr)  from
            anthropogenic sources  (NAPAP, 1987)	12

   4      Scatter plots of empirically-estimated annual amounts
            of sulfur dry deposition (kg S/ha/yr) versus: (a)
            adjusted RADM, (b) unadjusted RELMAP, and (c)
            adjusted ASTRAP predictions at the 22 sites	15

   5      Annually-normalized RADM predictions adjusted for bias
            (a), annual RELMAP predictions (b), and annual
            ASTRAP predictions adjusted for bias (c) of sulfur
            dry deposition (kg S/ha/yr) at the 22 sites	17

   6      Scatter plots of empirically-estimated annual amounts
            of sulfur dry deposition (kg S/ha/yr) versus: (a)
            adjusted RADM, (b) unadjusted RELMAP, and (c)
            adjusted ASTRAP predictions at the four U. S. and
            seven Canadian sites	18

   7      The spatial pattern of annual sulfur dry deposition
            (kg S/ha/yr) derived from adjusted RADM predictions.  . 19

   8      Percent errors in the adjusted RADM predictions of
            annual sulfur dry deposition 	 21

   9      Comparison of the relative error in the adjusted RADM
            predictions and the empirically-derived annual
            estimates of sulfur dry deposition 	 23

  10      The deviations of adjusted RADM predictions from
            empirically-derived annual estimates of sulfur dry
            deposition relative to the four groups of sites in
            Table 4	25
                                      VI

-------
Number
                                FIGURES
  11
  12
  13
Variations of the annual means (1982-1986) of sulfur
  dry deposition along a cross section of southern
  Ontario sites	
Variations of the annual means (1985-1987) of sulfur
  dry deposition at four U.S. sites	
                                                                   27
                                                                   28
The seasonal distribution of annual sulfur dry deposi-
  tion at four U.S. sites (Argonne -- ANL and Oak
  Ridge -- OAK R: 1985-1987; Penn State -- PSU and
  Whiteface Mountain -- WFM: 1986-1987)	
                                                                 .  30
Number
                                TABLES
          The three models used to construct grids of annual
            amounts of sulfur dry deposition across eastern
            North America 	
          Comparison of model predictions and empirically-
            derived estimates of annual sulfur dry deposition
            at the 22 sites	
                                                         13
          Descriptive statistics for adjusted RADM, unadjusted
            RELMAP, adjusted ASTRAP predictions, and empirical
            estimates of annual sulfur dry deposition	14

          Deviations of adjusted RADM predictions from
            empirically-derived dry deposition estimates
            relative to annual SO,  point source emissions	22
                                      vii

-------
                             ACKNOWLEDGEMENTS
     Dr. Jack Shannon of the Argonne National Laboratory provided the ASTRAP
predictions, while Dr. Tilden Meyers of NOAA's Atmospheric Turbulence and
Diffusion Laboratory and Ms. Dianne Green of the Ontario Ministry of
Environment provided the empirical estimates of dry deposition.
                                     Vlll

-------
                                SECTION 1
                              INTRODUCTION

     The Aquatic Effects Task Group of the National Acid Precipitation
Assessment Program (NAPAP) required estimates of annual sulfur dry deposi-
tion across more than 7,000 lakes in the eastern United States.  In August
of 1987, in response to this need,  the Atmospheric Sciences Research Labora-
tory (recently renamed the Atmospheric Research and Exposure Assessment
Laboratory) provided estimates of sulfur dry deposition for these locations
in an EPA Internal Report entitled "Interim Estimates of Annual Dry Sulfur
Deposition for the Eastern United States for the Aquatics Research Program"
(Dennis and Seilkop, 1987).  Interim appeared in the title to reflect the
impending improvements in our ability to estimate dry deposition amounts.

     These interim estimates were derived by interpolating the spatial
pattern of annual sulfur dry deposition amounts generated from annually-
normalized predictions of Version I of the Regional Acid Deposition Model
(RADM)  (Chang et al.,  1987) that were adjusted using empirical data at 4
U.S. sites in the COre Research Establishment (CORE) Network (Hales et al.,
1987) and 18 Canadian sites in the Acidic Precipitation in Ontario Study
(APIOS) Network (Ro et al., 1988).   The RADM results were preferred over
those of Version 4D of the Regional Lagrangian Model of Air Pollution
(RELMAP) (Eder et al.,  1986), since at that time the RADM spatial pattern
exhibited less of an oversmoothing problem.

     Since August 1987 several developments have occurred that necessitated
a re-examination of these interim estimates of sulfur dry deposition.
First,  the empirical data at the U.S. sites, as they appeared in the Interim

-------
Report, have been revised as a consequence of improvements to the dry
deposition algorithm of the inferential model used to derive the empirical
estimates.  Secondly, in both the United States and Ontario, data are now
available for additional years.   The expanded data base not only has
provided us with a more statistically representative sample, but also has
enabled network staff to identify outliers and discard or correct erroneous
values.  Thirdly, improvements to the dry deposition module of RELMAP have
greatly reduced its oversmoothing problem, thereby increasing this model's
potential as an appropriate estimator of spatial patterns.  Finally, the
predictions from a third model,  the Advanced Statistical Trajectory
Regional Air Pollution (ASTRAP)  model (Shannon, 1985),  became available.

     In this report we compare the dry deposition predictions from the three
models (RADM, RELMAP and ASTRAP) to the available empirically-derived
estimates to ascertain which of these models most accurately represents the
spatial pattern of sulfur dry deposition.  Using one of these models in
conjunction with the available data, we then recommend systematic revisions
to the interim estimates.  The uncertainty and bias in these revised
estimates are then characterized relative to interannual variability in the
empirically-estimated amounts of sulfur dry deposition, and distance of
sites from the emission source regions.

-------
                                SECTION 2
             REASONS FOR RE-EXAMINING THE INTERIM ESTIMATES

2.1   REVISION OF EMPIRICAL DATA AT U.S. SITES

     Subsequent to the Interim Report, the inferential model developed by
the National Oceanic and Atmospheric Administration's Atmospheric Turbulence
and Diffusion Division (NOAA/ATDD) to estimate dry deposition rates was
upgraded (Hicks et al.,  1987 and Hicks and Matt, 1988).  Two significant
changes resulted.   First, more information about the nature of the surface
has been incorporated into the calculation of the surface uptake resistance.
In the previous version,  only the one dominant plant species at each site
was considered in the calculation of the deposition velocity.  Currently,
the model considers the two dominant plant species at each site.  This was
considered to be a necessary change since many sites are located within
forests that are composed of both deciduous and coniferous species (e.g.,
Whiteface Mountain).

     The second significant change to the inferential model was in its
treatment of deposition to wet surfaces.  The influence of surface wetness
on deposition rates is an important consideration since, for example,
vegetation in the southeastern United States is wet 15% to 20% of the time
as a result of either dew or frost.  In preceeding versions of the model,
wet surfaces caused by rain or dew were considered to be stronger sinks for
sulfur dioxide (S02)  than were dry surfaces.   However,  recent empirical
evidence from research projects conducted in the United States by NOAA/ATDD
and in England (Mike Unsworth, University of Nottingham, personal communica-
tions) suggests that the enhancement of the deposition velocity occurs only
for surfaces that are wet by dew.  It has been suggested that for many

-------
locations, rainwater films enveloping leaves and stems are in equilibrium
with the sulfur in the atmosphere.  Thus, surfaces wet by rainwater consti-
tute a relatively smaller sink compared to the same surfaces enveloped by
dew.  Other factors such as rainfall rates and the accumulation of particles
on the leaves may be significant; however, their influence on uptake rates
is not likely to be understood in the near future.

     Some of the recent estimates of seasonal and annual dry deposition rates
may have changed slightly as a result of minor corrections made to the concen-
trations analyzed from filterpack sampler data.  These corrections compensate
for irregularities in sampler flow rates, which since 1986 are periodically
checked using mass flow controllers.  In some cases, flow rates either have
increased or decreased with time.  In such instances, changes in the flow
rates with time are assumed to be linear; adjustments are performed on a
weekly basis.  The most significant adjustments, a 50% reduction, were
applied to pre-February 1985 concentrations at the Penn State site.  Here,
filter pack S02 concentrations were found to be twice those  simultaneously
measured by a colocated monitor.   Adjustments to concentrations measured at
other locations were much less significant.  Similar discrepancies are also
noted when comparing seasonally averaged S02 concentrations  from 1985  to
1987.

-------
2.2   AVAILABILITY OF ADDITIONAL DATA TO CHARACTERIZE INTERANNUAL
      VARIABILITY

     Interannual variability of sulfur dry deposition must be considered in
any estimation procedure that is based on data from a specific year or
other time period.  If there is significant interannual variability,
estimates based on a particular year or other time period may not be
applicable to another period of interest.  This is precisely the situation
that occurs in our estimation of sulfur dry deposition for the 12-month
period chosen for studying aquatic effects (1 October 1983 through 30
September 1984).   Since dry deposition data were not collected until after
this time period, our annual estimates are therefore specific to this later
time period, and could potentially misrepresent the earlier (aquatic
effects) time period.  To characterize the uncertainty in applying estimates
based on data obtained after 1984 to the aquatic effects time period, it is
therefore important to attempt to determine the magnitude of the interannual
variability in sulfur dry deposition.

     In early 1987, empirical data were available for at most two years.  Con-
sequently, at that time the annual variability of sulfur dry deposition could
not be established with confidence.  Now, however, for some sites in Ontario
and in the eastern United States, data are available for as many as five
years.  Although more data are needed to better estimate the actual inter-
annual variability of sulfur dry deposition,  we are now able to derive a
preliminary estimate of the variability with the understanding that addi-
tional data could expand the range of annual amounts.  Therefore, these un-
certainty estimates could be slightly underestimated.

     The expanded empirical data base offers a second benefit in that we now
have for the first time U.S. and Ontario empirical data for the same years
(1985 and 1986).   If interannual variability is indeed significant, then
this is a substantial improvement, since empirical data at the sites are
used to adjust the annual model predictions for bias.  With this approach,

-------
it is therefore desirable to use empirical estimates for the same period so
that differences between sites will not reflect the temporal variability of
sulfur dry deposition.

     Finally, the expanded data base,  in addition to providing a more repre-
sentative sample, enables us to identify outliers and discard or, if possible,
correct erroneous values.  The adjustment of the S02 concentrations  at  the
Penn State site, previously discussed,  is an example of the benefits derived
from an expanded data base.

-------
2.3   POSSIBLE IMPROVEMENT OF ESTIMATES USING ADDITIONAL MODELING RESULTS

     In the August 1987 Interim Report, the annual sulfur dry deposition
predictions derived from RADM's annually normalized results were deemed to
be more representative of the empirical data than the pattern derived from
RELMAP annual results.  At that time, a comparison of empirical data and
RELMAP results indicated that RELMAP underpredicted near the regions of high
emissions and overpredicted across regions farther downwind.  The recent
improvements to RELMAP and the availability of annual predictions from a
third model, the ASTRAP model (Shannon, 1985), warranted a second comparison
of model predictions to the empirical data.

     The RELMAP module for dry deposition was modified to better reflect the
physical processes occurring across the spatial scales of the model grid cells
(i.e., one degree latitude by one degree longitude).  First and foremost,
rather than injecting all of the S02 and sulfate emissions  into the layers
above the surface layer (i.e., 50 m at night and 200 m otherwise), the model
now injects all area source emissions into the surface layer.  This is more
realistic since much of the area source emissions in the United States are
emitted at or near the surface.  Secondly, the order of the calculations of
transport and dry deposition was reversed.  The model now calculates the
amount of dry deposition at the source before transporting pollutant mass
away from it.  The net effect of these changes has increased the amounts at
the source cells and decreased them hundreds of kilometers downwind.

     Finally the last significant modification enabled RELMAP to apply dry
deposition velocities appropriate for each of 11 land-use categories.   These
deposition velocities were also updated by those recommended by Sheih et al.
(1986).  The model predictions are now based on the dry deposition across
each land-use category, whereas before, the cell deposition was based on
only the predominant land-use category.  Thus, in theory, the new approach
yields more representative predictions for the cell as a whole.

-------
                                SECTION 3
            SPATIAL DISTRIBUTION OF ANNUAL SULFUR DRY DEPOSITION

3.1   COMPARISON OF MODEL PREDICTIONS AND EMPIRICAL DATA

     Dry deposition rates are a function of the air concentration near the
surface and the dry deposition velocity.  The latter cannot be measured
directly but is inferred from vertical flux measurements.  In addition,
since dry deposition velocities are a function of atmospheric stability and
surface attributes (e.g., vegetative type, roughness length,  physical con-
ditions, spatial fluctuations of terrain and surface roughness), they can
vary significantly across small areas (e.g., less than 1 km2).   Because of
the potential for small spatial-scale variations in dry deposition velocity,
there is considerable uncertainty in using a dry deposition estimate from an
individual site to represent an average regional value.

     Estimation of a spatial pattern of sulfur dry deposition across the
eastern United States is even more strongly hampered by the fact that there
are only four U.S. sites at which empirical estimates are available.  To
circumvent this paucity of data, the predictions of regional-scale deposi-
tion models are used in conjunction with the available empirical data.  The
following procedure was developed for estimating annual amounts of sulfur
dry deposition across the eastern United States:

         (1)   Construct spatial patterns of annual amounts of S02 and
               sulfate dry deposition from available regional deposition
               models that relate emissions, transport, dispersion and
               transformation to dry deposition using dry deposition velo-
               cities assumed to represent the area of each model grid cell,

-------
         (2)    Adjust model predictions by a constant factor (based on the
               comparison of model predictions to site-specific empirical
               estimates) to correct for model bias,

         (3)    Select the spatial pattern produced by the regional models
               with the smallest mean-square error,

         (4)    Estimate dry deposition amounts at specific locations of
               interest by interpolating adjusted model predictions and

         (5)    Assess the uncertainty of these estimates by examining the
               correspondence between the model predictions and empirical
               estimates and characterizing the spatial and interannual
               variability of the empirical estimates.
     The first step of this approach was executed by constructing grids of
annual sulfur dry deposition from each of three operational deposition models
used by EPA (Table 1).  As was the case described in the Interim Report,  six
three-day episodes of RADM output were averaged and normalized to construct
one annual grid (Cases I,  II and IV of the April 1981 Oxidizing and Scaven-
ging Characteristics of April Rains (OSCAR) Experiment,  the four-dimensional

data assimilation run of OSCAR IV, the August 1979 Northeast Regional Oxi-
dant Study (NEROS) case, and an October 1984 case).  Unlike the RADM grid,

those of RELMAP and ASTRAP were constructed from simulations of the entire
TABLE 1.   MODELS USED TO CONSTRUCT GRIDS OF ANNUAL AMOUNTS OF SULFUR DRY
                   DEPOSITION ACROSS EASTERN NORTH AMERICA
          Model name                         Model genre and approximate
                                             spatial resolution,  km
Regional Acid Deposition Model (RADM)                Eulerian
                                                        80

Regional Lagrangian Model of Air                    Lagrangian
Pollution (RELMAP)                                     100

Advanced Statistical Trajectory Regional        Statistical-Lagrangian
Air Pollution (ASTRAP) Model                           130

-------
year of 1980.  The RELMAP results presented here were derived from the
improved model version and differed from those presented in the Interim
Report.
     The second step was accomplished by first comparing model predictions
with the annual means of the empirical data at each of 22 sites of the CORE
and APIOS networks (Figures 1 and 2).   Empirical data at the CORE sites were
based on weekly mean dry deposition velocities and air concentrations for
the years 1985 to 1987,  inclusive.  Dry deposition data from two CORE sites
were not used here.  A significant bias, as high as +50%, was suspected at
West Point, New York due to "edge" effects (i.e., significant spatial
gradients in terrain or surface roughness).  In addition, the revised data
from Bondville, Illinois were not available.   In contrast, Ontario empirical
data were based on a cruder, less accurate method -- since only annual mean
dry deposition velocities were available, sulfur dry deposition amounts at
the APIOS sites were determined from these deposition velocities and annual
mean concentrations for the years 1982 to 1986, inclusive.
Figure 1.
The location of 22 sites of the CORE and APIOS networks from which
empirically-derived amounts of sulfur dry deposition were
available for at least one year.
                                     10

-------
Figure 2.  Empirically-derived annual amounts of sulfur dry deposition
           (kg S/ha/yr).

     A comparison of the annual mean amounts of sulfur dry deposition at the
CORE and APIOS sites (Figure 2) and the spatial distribution of 1980 anthro-
pogenic S02 emissions  (Figure 3)  gives  evidence of a strong correlation
between the two.   Specifically, sulfur dry deposition tends to be greatest
across regions of high S02 emissions.   Consequently,  model predictions of
spatial gradients of sulfur dry deposition are expected to generally reflect
those of emissions.

     From the comparisons of empirical estimates and model predictions,
mean-square errors were calculated and used as a measure of concurrence.  To
reduce the bias of the annually-normalized RADM and the ASTRAP predictions,
the ensembles of predictions were adjusted systematically by factors of 0.43
and 0.57, respectively.  These factors are the regression coefficients of a
linear regression of empirical estimates on model predictions forced through
the origin.  Similar adjustments to RELMAP predictions were unnecessary, as
the estimated regression coefficient was almost identical to one.  Thus,
                                     11

-------
                 0 0 1
                  014 000000 10 701 ^0—- 000100  002
                  000000022 a—'"'! 200 12 20  1020
                  ooo    9 1 i  i ~" o a^e—<—a—s    o o o o  4.
Figure 3.   Annual 1980 S02 emissions (10 ktonnes/yr) from anthropogenic
            sources (NAPAP, 1987).

unlike the other models, RELMAP gives no evidence of bias.  Compared  to  the
results of the earlier version of RELMAP (presented  in the Interim Report)
these results represent a significant improvement.

     For each of the 22 sites, Table 2 compares the  empirically-derived
amounts of annual sulfur dry deposition with the adjusted RADM predictions,
unadjusted RELMAP predictions and adjusted ASTRAP predictions.  As the
descriptive statistics in Table 3 indicate, the adjusted RADM predictions
best replicate the empirical data; its bias, root-mean-square error,  and
average error are lower than those of the other two models.  This is  also
supported by the scatter plots of Figure 4.  The scatter plots also illus-
trate the tendency of the models to overpredict the  sulfur dry deposition  in
the empirical data range of 3 to 6 kg S/ha/yr.  This is indicative of the
models' slower rate of decreasing dry deposition away from the high
emissions regions, or in other terms, their smoothing tendency near steep
                                     12

-------




UJ
, t
CO
CM
CM
UJ
3:
h-

<

O


CO
o
UJ



0
QS
ID

ID
CO

_J






0
CO
LU
1 —
<
t—
CO

Q
LU


QC
LU
O

>—
_J
^
O
a:
a!
UJ
a
<£

00
z
o

t—
u

o
LU
ae
a.

	 i
LU
8
x
u.
O


CO

a:

a.
Z
u





UJ

ca

t—
it
u
u
u
u
u
ii >»
II -*x
II flJ
u -g
ti "*-.
II CO
II
ii en
u
u •»
II C
II O

11 *-• 3*2
II —
II CO *
II O %
ii a. c
II 0) O
II "O i|—
1 (- '—
1 T3 >
II (U
ii c_ -a
ii ID
II M- 0)
II — ' >
II D •-
1 CO

i ra o


1 C
II CD
u
II *4-
1 0
II CO
11 c
It O
II "-
II *-»
II U
1 (_
1
1
1
1
1
1
1
1
1

1
1
1
1
1
1
II
II
11
It
II
II
II
II
II
II
II
II
II
It
It
II
II
II
II
1
It
II
It
II
II
II
II
11
u
11
II
II
II
II
II
II
11
11
II
1
It
II
II
II




<
!-^
00
<
u
+•>

3
mJ
•a"
<





a.
s
UJ
a:

"D
^j
Ul
3
-Q
ID
C







Z
<
IX

T>
%
•^
<









^
"•s.
CO ^ CO
•-- oo ra
— 01 >.
t" -s
aT

CD to o

• *








ro ro
00 O

+ «-
+

rvj co

O^ in









S 0
O*> N*



8 S
jsj fyj













f^J OJ

ro ro
u-I (NJ













a. >- o
* i_
4-J OJ 4J
OJ O C
CO H- ^>


Ci "— A)

a. 3 —
CO

CO
o

CL
<
* *

O 0





II
II
(1
II
II
>OOOsLn«->ocOKifMrvJroorooroOi^pJ n
N.iACOcoooN-u"»(NJ'O''Oininrorvjf^co>o "

i

o*O'Om^ocoorii**OLniAcooO'~*~fMPs*'4' i
O^cocomrooin-^rororororo-^rvfMOo i

i
i
i
i
i
i
i
i
i
^rocoor^rjorofMco^»-in^fOLnOf\j 1
%±^>rofvjr**iio«— «— «— OtOJcvirooooN-O11 i

1 + 1^- + + +,-<\j++ +. , i
+ + + t
i
>o o CNJ *"~ ^O ^o o «"• ^ m «~ m m *~ ^~ ro r^- o i
OOO«-S.OJ^-N-CO{NJsO'Of*-Ofs-CO'O'OPO 1
•OmcOLnfM«— roro>J"u^(Nioj(\ifM»— ^-00 i

u
u
n
n
ii
i
i
i
rocoocooco-Ovor^^->or-.N.fMLnrocON. i
<\iOi»^^f*"D»ioo*— oo**f^-r\j-*O'— r*-*or*- !


t
SgSSS^^K^^^SgK^S^Si ! o


1 X
1

I ^
1 
1 *->
I ca
1 E

1 CO
1 LU
1 —
rou^ruro^-<~rOLnrOLnLnin^LnLOLn-jf\j t o
*^ t t_
**j f\j *o CM m O* rvj o «"" * — ro ^ -o O* co m (\i o* i -^~
O»^Of\J(Min(\jc\JOjT-»— r-ro«— ro«— oo ILU


1 ^s.
i



i ra
1 .5

Ii cn
CO II LU

en — • ti — '
>- O "D -^ — u (0
t_ — • •*-» c nj o (D j»; ti •—
*-»too*j  £ «^ —• ^ t*. v^ 4-j _* 3 O C— i i

U CL 3 Q- CO Z O O CO O O O Z Z ^ (J 21 Q (C
1 O

1 U

1 "D
I O
***** * * lt_

r- ^- ^_ ,__ y_ ^_ p^J f^J fyj (>J f^J fyj j^ | ^^j


1
1
( %


























































t
M
QJ

XJ
CO
l_



0)
1

tfl

a>

CO

CU



c



(U
tfl
D


QJ

• •—




*
13

-------
   TABLE 3.  DESCRIPTIVE STATISTICS FOR ADJUSTED RADM,  UNADJUSTED RELMAP,
  ADJUSTED ASTRAP PREDICTIONS,  AND EMPIRICAL ESTIMATES  OF ANNUAL SULFUR DRY
                                 DEPOSITION
Empirical RADM RELMAP
Estimates
AT ALL 22 U.S. AND ONTARIO SITES:
Mean 4.12 4.42 4.83
Bias 7.3% 17.2%
RMSE* 1.42 (34%) 2.41 (58%)
Average Error 1.11 (25%) 1.90 (46%)
AT 11 SITES (4 UNITED STATES AND 7 ONTARIO SITES):
Mean 6.18 6.49 7.11
Bias 5.0% 15.0%
RMSE* 1.69 (27%) 3.08 (50%)
Average Error 1.42 (23%) 1.90 (31%)
ASTRAP

5.76
39.8%
3.04 (74%)
2.42 (59%)

7.62
23.3%
3.38 (55%)
2.99 (48%)
*   Root-mean-square error
                                     14

-------
                Q.
                    2 -
                    0 -i
                              (a)  Adjusted RADM
                                   Adjusted RADM

                            (b)  Unadjusted RELMAP
                               4    6    8    10
                                 Unadjusted RELMAP

                             (c)  Adjusted ASTRAP
                                   6    8    10
                                  Adjusted ASTRAP
                                                 12
                                        8    10   12    14    16
                                                          16
Figure 4.    Scatter plots of  empirically-estimated annual amounts  of sulfur
             dry deposition  (kg  S/ha/yr) versus:  (a)  adjusted RADM,  (b)
             unadjusted RELMAP,  and (c) adjusted  ASTRAP predictions  at the 22
             sites.
                                       15

-------
gradients (Figures 2 and 5).  Moreover, the degree of smoothing appears to
be a function of the spatial resolution of the models (Table 1). That is,
the coarser the spatial resolution, the more smoothing is evident.

     Descriptive statistics were also computed using only the empirical esti-
mates at four of the U.S. sites and seven Ontario sites (Table 3) to provide
a more equal balance between the available U.S. and Ontario data, thereby
reducing the possibility of overweighting discrepancies in Ontario relative
to those found at United States sites.  The Ontario sites, identified in
Table 2, were selected on the basis of their position relative to the steep
gradient both in terms of sulfur emissions and sulfur dry deposition from
Lake Huron to eastern Ontario.  By computing descriptive statistics from
this subset, one can assess: (1) the performance of the models in replica-
ting steep gradients, and (2) the degree of smoothing intrinsic to each
model.

     As was the case for the entire data set, adjusted RADM results best
replicate the empirical data of the 11-site subset.  The scatter plots of
Figure 6 also support this conclusion.  Therefore, as was the case in the
Interim Report, the adjusted RADM predictions appear to provide better
estimates of the annual sulfur dry deposition amounts than those obtained
from the simpler linear models.  Figure 7 illustrates the spatial pattern of
annual sulfur dry deposition derived from the adjusted RADM predictions.
                                     16

-------
 3


•D
<
-a
 a
 c
 o
 <
 an
                                                                                                                             u

                                                                                                                            •5
                                                                                                                             Q.


                                                                                                                            a.
                                                                                                                                .c
                                                                                                                                -•*.
                                                                                                                                in
                                                                                                                              _   _

                                                                                                                             g  £


                                                                                                                             * "S
                                                          17

-------
                              (a)  Adjusted RADM
                   16

                   14

                   12

                   10
                                         ~1	1	1	1	1	1—
                                            10    12    14
                                  Adjusted RADM
                           (b)  Unadjusted RELMAP
                   16	

                   14 -

                   12 -

                   10 -

                    8 -
                          2    4    6    8    10    12
                                 Unadjusted RELMAP

                             (c)  Adjusted ASTRAP

                   12 -
                    8 -
                    2 -

                    0 -
                              4    6     8    10
                                  Adjusted ASTRAP
Figure 6.   Scatter plots of  empirically-estimated annual amounts  of sulfur dry
            deposition (kg S/ha/yr)  versus:  (a)  adjusted RADM,  (b)  unadjusted
            RELMAP, and (c) adjusted ASTRAP predictions at the  four U.S.
            sites and seven Ontario  sites.
                                       18

-------
Figure 7.    The spatial pattern of annual sulfur dry deposition (kg S/ha/yr)
            derived from adjusted RADM predictions.
                                     19

-------
3.2   UNCERTAINTY ASSESSMENT OF THE REVISED ESTIMATES

     The uncertainty in sulfur dry deposition amounts obtained from the
estimation procedure described above is primarily related to three main
factors:  (1) the accuracy with which the RADM captures the underlying
spatial pattern of dry deposition, (2) the accuracy of the empirical dry
deposition estimates that are used to adjust the predictions,  and (3) the
potential systematic differences between the empirical dry deposition
estimates from recent years and the period of interest.  This  section
discusses these factors and their effects on the uncertainty in estimated
regional patterns of sulfur dry deposition.

3.2.1.   Uncertainty of Adjusted RADM Predictions

     To some extent, we have already investigated the uncertainty in the
adjusted RADM predictions by comparing them to the empirical estimates
(Tables 2 and 3 and Figures 4 and 5).   Some random differences between the
adjusted RADM and the empirical estimates are expected because of
differences in spatial scales.  For example, the RADM is a regional model
that is not expected to simulate spatial gradients on scales below its grid
cell resolution of approximately 80 km.  Similarly, empirical  estimates from
an individual site imperfectly represent a regional average.  Because we
have so little data from the United States against which to compare the
adjusted RADM predictions, it is important to attempt to determine whether
any of the observed difference is related to model error.  In  particular,
examination of model performance relative to emission patterns can provide
insight into the model's ability to accurately represent physical processes,
and give a sense of the degree of confidence that can be placed in its
predictions.

     There is some evidence to suggest that the RADM exhibits  biases
relative to the emissions pattern.  Although it appears to represent an
improvement over Lagrangian models relative to oversmoothing in areas within
                                     20

-------
approximately 200 km downwind of significant emission source regions,  there
is still some estimation bias in these areas.   As illustrated by Figure 8,
the largest positive errors occur in southwestern Ontario and to the
northeast of Toronto; sites in this region are 100 to 200 km northeast of
major sources.  In addition,  the adjusted model estimate at the Penn State
site, downwind of a major source region (Figures 3 and 4),  is 50% greater
than the empirical estimate.

     It appears that the pattern of over/underestimation is linked to  the
proximity of major source regions,  as indicated in Table 4 and Figure  9.
In regions within 80 km of major sources (i.e., those emitting more than 200
ktonnes of S02 per year),  RADM tends  to  underestimate dry deposition.
Figure 8.  Percent errors in the adjusted RADM predictions of annual sulfur
           dry deposition.
                                     21

-------
TABLE 4.   DEVIATIONS OF ADJUSTED RADM PREDICTIONS FROM EMPIRICALLY-
            DERIVED DRY DEPOSITION ESTIMATES RELATIVE TO ANNUAL S02
                           POINT SOURCE EMISSIONS
S02 emissions, ktonnes/yr
Site
ID
GROUP
1
101
10
4
102
Within
80 km
A SITES (>200
421
409
304
297
209
Within
81-160 km
KTONNES WITHIN
746
345
205
803
169
Sulfur dry deposition,
kg S/ha/yr
RADM
80 KM)
7
15
5
9
8

.2
.7
.1
.4
.0
Empirical
estimates

9.2
13.7
5.7
10.7
10.2
Relative
deviation*, %


-22
14
-10
-12
-22
GROUP B SITES (<125 KTONNES WITHIN 80 KM BUT >125 KTONNES WITHIN 81-160 KM)
103
3
8
11
13
9
95
<1
3
102
<1
2
904
957
620
382
146
137
8
6
4
5
2
3
.0
.9
.3
.5
.6
.8
5.
4.
2.
2.
2.
2.
3
3
8
8
9
4
50
60
51
94
-11
57
GROUP C SITES (<125 KTONNES WITHIN 160 KM)
16
104
15
31
27
17
20
27
64
4
14
8
<1
0
69
60
33
16
13
2
<1
2.4
2.4
2.6
0.9
1.4
1.6
1.8
1.7
2.3
2.0
0.6
0.7
1.4
1.7
40
4
32
57
97
10
8
GROUP D SITES (WITHIN 160 KM OF SUDBURY, ONTARIO)
 23
 25
 21
 22
896
  0
  0
  9
  135
1,633
1,040
1,031
2.2
1.5
2.1
1.8
3.7
1.7
3.1
1.7
-41
 -7
-33
  6
    [(Prediction - Empirical Estimate) / (Empirical Estimate)] x 100
                                     22

-------
                   DEVIATIONS OF ADJUSTED RADM PREDICTIONS
                    100
                    80-
                    60 -
                    40 -
                    20-
                 V
                 Q
                   -20 -
                   -40 -
                   -60 -
                   -80 -
                   -too
                             A      B      C      D
                                    Group ID

Figure 9.    The deviations of adjusted RADM predictions from empirically-
            derived annual estimates of sulfur dry deposition relative to
            the four groups of sites in Table 4.

The one exception is the Argonne site (101), which, unlike the other Group A
sites, is  normally situated on the upwind side of the source region.

     The model predictions are greater than the empirical estimates at sites
located within 81 to 160 km of major source regions.  This suggests that the
model tends to oversmooths dry deposition gradients near source regions.
The pattern of overprediction does not, however, emerge at sites within 81
to 160 km of the major sources in Sudbury, Ontario.  In fact, in this region
the model performance appears to parallel that in regions within 80 km of
major source regions, with underpredictions and a single modest overpredic-
tion of 6%.  One cause of this difference in model behavior might be related
to the Sudbury stacks, which are much taller than those elsewhere in North
America.  These tall stacks might deposit sulfur compounds farther  downwind
than typical sources, with much of the deposition occuring in the 81-160-km
range rather than within 80 km.  Consequently, it would not be unreasonable
to expect that the model's behavior within  81-160 km of the Sudbury source
region might be similar to that observed  in regions within 80 km of typical
large source regions.
                                      23

-------
     Oversmoothing of the spatial gradients is also evident in the groups of
sites near less significant source regions (Group C).   Within 160 km of
source regions emitting between 20 and 125 ktonnes of S02,  RADM estimates
are considerably greater than the empirical estimates.  With one exception,
the relative error at these sites is of the same order as that at sites
within 81 to 160 km of much larger sources.  In contrast, the model predic-
tions at the two sites removed from major source regions (i.e., <3 ktonnes/
yr within 160 km of the site) are within 10% of the empirical estimates.

     In summary, a comparison of the revised estimates to empirically-
derived amounts of sulfur dry deposition suggests that there is a systematic
error in the revised estimates.  Adjusted RADM predictions of dry deposition
tend to be biased low in the most significant source regions (where at least
200 ktonnes S02/yr are emitted within 80 km of the site).   Conversely,  in
locations farther removed from significant sources (81-160 km) there is
evidence that the estimates are biased high.

     It is noteworthy that the adjusted model predictions are generally
within ±60% of the empirical estimates (Figure 10).  Relative errors are
generally less than 40% (in absolute value) at high deposition sites (where
empirical estimates exceed 9 kg S/ha/yr).  At sites with moderate deposition
(where empirical estimates are between 2 and 6 kg S/ha/yr), which are fairly
close to significant source regions (e.g., Penn State, south-central
Ontario, and northeast of Toronto), estimation errors are generally between
45% and 60%, with one error approaching 100%.  For other sites with moderate
deposition, Figure 8 illustrates the consistency of model behavior with
regard to over and underestimation, with errors ranging ±40%.   However,  as
noted previously, some of the greater deviations  (in absolute value) in this
group of sites might also be explained by the relationship of the sites and
their distance from the sources.
                                     24

-------
                     100
                               RELATIVE DEVIATIONS
                            Adj. RADM Predictions Ic Emp. Estimotes
                     80 -
                     60-
                     40 -
                     20-
                      0
                    -20-
                    -40
                    -60-
                    -80 -
                    -100
                            2    4    6    8    10
                             Empirical Estimates, kg S/ha/yr
                                                     12
Figure 10.  Comparison of the relative  error  in the adjusted RADM pre-
            dictions and the empirically-derived annual estimates of sulfur
            dry deposition  (kg  S/ha/yr).

3.2.2   Uncertainty of the  Empirical  Estimates

     Since appropriate techniques  are not yet developed to directly measure
dry deposition, we must rely on either  inferential techniques that estimate
dry deposition from measurable  parameters [such as vertical fluxes of pollu-
tants near the surface -- the method  used at  the U.S.  sites (Hicks and Matt,
1988)], or the less desirable method, used at the Ontario sites, based on
long-term mean concentration measurements and estimates of dry deposition
velocities.  Specifically,  the  Ontario  estimates of annual dry deposition
were computed from annual mean  measurements of S02 and sulfate concentra-
tions and annual mean dry deposition  velocities.  The dry deposition veloci-
ties were estimated via the method first reported by Masse and Voldner
(1983) and updated by Voldner et al.  (1986).   In contrast, the U.S. amounts
were computed from weekly mean  measurements of S02 and sulfate concentra-
tions and mean dry deposition velocities inferred from flux measurements for
the same week.  Regardless  of which of  these  two approaches are used, errors
arise and contribute to the deviations  illustrated in Figure 8.
                                      25

-------
     Since dry deposition inferential methods rely directly on vertical flux
measurements of both S02 and sulfate,  the accuracy of these measurements
directly determines the accuracy of the U.S. empirical estimates.  For
individual measurements this could translate into significant deviations
between model predictions and empirical estimates.  Fortunately, however,
the empirical estimates at the U.S. sites are actually annual averages based
on enough weekly measurements to virtually eliminate the effects of random
measurement errors.  Therefore, only a bias in the measurement technique
would affect the accuracy of the U.S.  estimates.   If there are biases in the
empirical deposition estimates, it is unlikely that they would be related to
the emissions pattern, as the deviations of adjusted model predictions from
empirical estimates appear to be.

     The key sources of uncertainty of the Ontario estimates are related to
the uncertainties arising from the estimation of the climatological dry
deposition velocities and the assumption that the product of the long-term
mean concentrations and deposition velocities is representative of the sum of
the products of the short-term concentrations and deposition velocities.
These uncertainties have yet to be assessed.

     With the very limited available data, it is impossible to separate and
quantify the three sources of error (i.e., model bias, subgrid-scale varia-
bility, and empirical estimation errors).  Therefore, the best that we can
do in characterizing the uncertainty in our dry deposition estimates is to
consider the aggregate of all these errors, as reflected in the distribution
of RADM deviations from empirical estimates.  These deviations suggest that
the adjusted RADM predictions of dry deposition generally are expected to
lie within ±60% of the actual values.   Although there is some evidence that
the magnitude and direction of the errors in model-predicted dry deposition
might be related to distance from significant source regions, we do not feel
that the available data allow us to further refine the ±60% estimate of
uncertainty.
                                     26

-------
3.2.3   Interannual Variability of the Empirically-Derived  Estimates

     Since the Aquatics Effects Program of NAPAP  focuses  on the period
I October 1983 to 30 September 1984,  it then follows  that our  estimates  of
annual sulfur dry deposition should be confined to  that same period.  How-
ever, since empirical data in the United States exist only  after  this
period,  we must base our estimates on the years subsequent  to  the period of
interest.  Therefore, one component of the uncertainty of our  best  estimates
for this 12-month period relates to the interannual variability of  sulfur
dry deposition as determined from the empirical data.

     To assess the magnitude of this  uncertainty, the interannual variabili-
ty was estimated at the Ontario APIOS sites using all available data passing
the screening criteria.  Figure 11 illustrates the  annual variations in
sulfur dry deposition for five years  (1982-1986)  along a  cross section of
the APIOS network from Site 1 near Detroit to Site  16 in  eastern  Ontario.

               DRY  DEPOSITION  VARIABILITY IN S.ONTARIO


cc
i^
3
in
|
i—
8
O.
a
I
5
U.

-------
This cross section was selected to relate annual variability to the pattern
of sulfur emissions, which along this line exhibits an emissions decrease of
two orders of magnitude from sites 1 and 4 to 3.   Except for Site 4, where
annual amounts varied as much as 5 kg S/ha/yr, a variation of 1 to 2 kg
S/ha/yr is apparent.  This translates to an annual variation of 10% to 30%
of the average annual amount of sulfur dry deposition.

     Since data for only two or three years are available at the U.S. sites,
our ability to quantify interannual variability at the U.S. sites is
limited.  However, as Figure 12 illustrates, from the available data, only
small interannual variations, on the order of 1 kg S/ha/yr, are observed
during the period 1985 to 1987.  At Argonne and Oak Ridge this amounts to
only a 10% interannual variation, considerably less than that at the APIOS
sites.  Less than a 10% variation occurs in the two years of data at Penn
State and Whiteface Mountain.
                 ANNUAL VARIABILITY OF  S DRY  DEPOSITION
            cr
            I
            CO
            CO
            en
            o
            D.
            tr
            o
                              1985 - 1987  U.S. DATA
                      ANL
OAK R
                            1985
                                             PSU
             ZZZ2 1987
                        WFM
Figure 12.  Variations of the annual means  (1985-1987) of sulfur dry
            deposition at four U.S. sites.
                                     28

-------
     The apparent differences in the interannual variabilities at the Ontario
and U.S. sites can possibly be explained by the significant differences in
the approaches of estimating sulfur dry deposition,  as discussed in the
preceeding section.  Since the latter approach is theoretically more pre-
cise, the interannual variability apparent in the Ontario annual estimates
might be related to the uncertainty imposed by the estimation method and not
an indication of the true annual variability.  Therefore, until additional
U.S. data are available, a ±10%  interannual variability is  assumed.   Since
the magnitude of this variation is very small compared to the ±60%  estimate
of the uncertainty discussed previously, our uncertainty estimate remains at
±60%.
                                     29

-------
3.3   ESTIMATED SEASONAL DISTRIBUTION OF SULFUR DRY DEPOSITION

     In addition to estimates of total annual sulfur dry deposition, some
aquatic effects researchers might need to know the distribution of dry
deposition amounts across seasons.  This distribution was determined using
1985 to 1987 empirical estimates at the U.S. sites.  As Figure 13 shows,
approximately 30% of the annual sulfur dry deposition occurs in each the
spring and summer and approximately 20% in each the winter and autumn at
Argonne,  Oak Ridge and Penn State.  The seasonal variation of dry deposition
was less at Whiteface Mountain; this site differed from the other three U.S.
sites in that it was located in a mixed conifer/deciduous forest and,
therefore, had a higher wintertime dry deposition velocity for S02.
            U.
            O
            LU
            CD
            CC
            LU
            a.
                  SEASONALITY OF  SULFUR  DRY DEPOSITION
                               1985 - 1987  U.S. DATA
                 10
                     WINTER
                    §• ANL
     SEASON
OAK R     V7Z PSU
Figure 13.  The seasonal distribution of annual sulfur dry deposition at
            four U.S. sites (Argonne -- ANL and Oak Ridge -- OAK R: 1985-
            1987; Penn State -- PSU and Whiteface Mountain -- WFM: 1986-
            1987) .
                                     30

-------
                                  SECTION 4
                       CONCLUSIONS AND RECOMMENDATIONS

     In August 1987 annual sulfur dry deposition to more than 7000 lakes in
the eastern United States were estimated from interpolated model calcula-
tions anchored to the empirical data.  Since that time, several developments
have occurred to warrant a second examination of these best estimates.  The
procedure described in the Interim Report was repeated; however this time,
the predictions of an additional model and an expanded data base of empiri-
cal estimates were available.

     Three regional models,  RADM, RELMAP and ASTRAP, were applied to con-
struct grids of annual sulfur dry deposition.  Comparison of RELMAP predic-
tions, adjusted RADM and ASTRAP predictions with empirical data at 22 sites
indicated that RADM best replicated the steep gradient downwind of a signi-
ficant emissions source region.  Although each model exhibited a tendency to
smooth the gradient, the degree of smoothing appeared to be a function of
the spatial resolution of the model.   Based on the model comparisons with
available empirical data, the adjusted RADM predictions appear to be the
best estimates to date of the spatial distribution of annual sulfur dry
deposition in the eastern United States.

     Since the RADM adjustment factor here was 13% greater than that used in
the Interim Report, an identical systematic increase in the interpolations
appearing in the Interim Report is recommended.  The difference in adjustment
factors was a result of using as many as five years of Canadian data and two
to three years of U.S. data, as opposed to only one to two years of data
that were available at the time of the Interim Report.
                                     31

-------
        The comparison of the revised estimates to empirically-derived
amounts of sulfur dry deposition suggests that there is a systematic error
in the revised estimates.  Although adjusted RADM predictions of dry
deposition are generally within ±60%  of the  empirical estimates,  they tend
to be biased low in the most significant source regions (where at least 200
ktonnes S02/yr are  emitted within 80  km of the site).   Conversely,  in
locations farther removed from significant sources (81-160 km) there is
evidence that the estimates are biased high.

     Because of the anticipation of data from additional sites and periods
and impending improvements to the algorithms that calculate dry deposition
velocities, it is recommended that this procedure be repeated at a later
time.  Therefore, these estimates of the annual sulfur dry deposition across
the eastern United States could be revised in the future.
                                     32

-------
                               REFERENCES
Chang, J.S., R.A.  Brost, I.S.A.  Isaksen,  S.  Modronich,  P.  Middleton,  W.R.
     Stockwell,  and C.J. Walcek.   1987.   A three-dimensional Eulerian acid
     deposition model:  physical  concepts  and formulation.   J.  Geophys.  Res.,
     92(012):14681-14700.

Dennis, R.L. and S.K.  Seilkop.   1987.   Interim estimates of annual dry sulfur
     deposition for the eastern  United States for the Aquatics Research Pro-
     gram.  Internal Report,  U.S.  Environmental Protection Agency, Research
     Triangle Park, NC,  16 p.

Eder, B.K.,  D.H. Coventry, T.L.  Clark, and C.E. Bellinger.   1986.   RELMAP:  a
     Regional Lagrangian Model  of Air Pollution user's guide.   EPA/600/8-86/
     013 (NTIS:  PB 86-171 394/AS),  U.S.  Environmental Protection Agency,
     Research Triangle Park,  NC.

Hales, J.M., B.B Hicks,  and J.M.  Miller.   1987.  The role of research
     measurement networks as contributors to federal assessments of acid
     deposition.  Bull,  of Amer.  Meteor.  Soc.,  68:216-225.

Hicks, B.B., D.D.  Baldocchi,  R.P.  Hosker, and D.R.  Matt.  1987.  A prelim-
     inary multiple resistance  routine for deriving dry deposition veloci-
     ties from measured quantities.  Water,  Air and Soil Pollution, 36:311-
     330.

Hicks, B.B.  and D.R. Matt.  1988.   Combining biology, chemistry, and meteoro-
     logy in modeling and measuring dry deposition.  J. of Atmos.  Chem.,
     6:117-131.

Masse, C. and E.G. Voldner.  1983.   Estimation of dry deposition velocities
     of sulfur over Canada and the United States east of the Rocky
     Mountains.   Proceedings of  the Fourth International Conference on Pre-
     cipitation Scavenging, Dry  Deposition,  and Resuspension,  Santa Monica,
     CA, November 29 - December  3,  Pruppacher et al. Eds.,  Elsevier,  NY.
     p. 991-1001.

NAPAP.  1987.  Interim assessment - the causes and effects of acidic deposi-
     tion, Volume II:  emissions  and control.  The National Acid Precipitation
     Assessment Program, 722 Jackson Place,  N.W., Washington,  DC 20503.
                                     33

-------
Ro, C.U., A.J.S. Tang, W.H. Chan, R.W.  Kirk,  N.W.  Reid, and M. Lusis.  1988.
     Wet and dry deposition of sulfur and nitrogen compounds in Ontario.
     Atmos. Environ., 22:2763-2772.

Shannon, J.D.  1985.  User's guide for the Advanced Statistical Trajectory
     Regional Air Pollution (ASTRAP) Model.  EPA/600/8-85/016 (NTIS: PB 85-
     236 784/AS),  U.S. Environmental Protection Agency, Research Triangle
     Park, NC, 83 p.

Sheih, C.M.,  M.L.  Wesely, and C.J. Walcek.  1986.   A dry deposition module
     for regional acid deposition.  EPA/600/3-86/037 (NTIS: PB 86-218104),
     U.S. Environmental Protection Agency, Research Triangle Park,  NC, 63 p.

Volnder, E.G., L.A. Barrie, and A. Sirois.  1986.   A literature review of
     dry deposition of oxides of sulphur and nitrogen with emphasis on long-
     range transport modelling in North America.   Atmos.  Environ.,  20:2101-
     2123.
                                     34

-------

-------

-------