-------
      Scavenging of dry deposition by vegetative canopies (especially coniferous canopies) is subject
to a pronounced "edge effect", whereby lower windspeeds and ambient concentrations within interior
canopies result in markedly lower effective dry deposition to those interior regions (Dasch, 1987;
Grennfelt, 1987). We reasoned that this process could be represented by a function of the form of the
well-known Michaelis-Menton equation and used to adjust the "effective" coniferous canopy scaveng-
ing. In this way, scavenging of base cations and chloride within our watersheds would not be com-
puted using total coniferous LAIs, but rather the coniferous LAIs would be adjusted (in effect) so that
as the areal coniferous coverage of a watershed  increased, its effect on scavenging reached an effec-
tive plateau  rather than increasing linearly. We used this approach to adjust total dry deposition until
the chloride budgets approximately balanced in undisturbed NE sites.  The final equation used in the
adjustment was
           % CONe   =    (30 * % CON)/(15 + % CON)
where     % CON    =    mapped percent coniferous coverage
           % CONe   =    effective percent coniferous coverage
(5-1)
These computations of dry base cation deposition leave a great deal to be desired.  The final values,
however, relate well to (1) estimates of dry deposition-to-wet deposition ratios observed by Lindberg et
al. (1986) fora southeastern forested catchment, and (2) previously modelled estimates at Woods and
Panther Lakes in the Adirondack Mountains (R. Munson, pers. comm.).

5.6.3.2.3 Nitrate and ammonium --

      We had no objective or mechanistic approach for estimating dry deposition of nitrate and
ammonium. Instead, we assumed that total dry deposition of nitrate was equal to wet deposition and
that total dry deposition of ammonium was equal to one-half wet deposition.  These ratios approximate
values measured  by Lindberg et al. (1986) in an eastern forested watershed.

5.6.3.2.4 Hydrogen ion --

      We computed dry H+ deposition as the difference between dry anions and other dry cations.
When the sum of other dry cations was greater than the sum of dry anions, we set dry H+ to zero.

5.6.3.2.5 Ion ratios --

      The ratios of dry deposition to wet deposition for all ions for the M-APP study sites for the TY
dataset are shown in Table 5-16 in relation to values obtained for the NE and SBRP.

5.6.3.2.6 Comparisons with direct measurements --

      Although extensive data do not exist with which to compare the DDRP estimates, Church et al.
(1989) used some limited  information (B. Hicks, pers. comm.) for this purpose for the previous esti-
mates made for the NE and SBRP.  We had no comparable information for the M-APP Region.
                                             121

-------
Table 5-16.  Ratios of Dry Deposition to Wet Deposition for DDRP Study Sites
             for the Typical Year (TY) Deposition Datasets
NE
S042"
Ca2+
Mg2+
Na+
K+
cr
aN03-
aNH4+
H+
M-APP
S042'
Ca2+
Mg2+
Na+
K+
cr
aN03-
aNH4+
H+
Median
0.44
1.13
1.92
1.29
1.56
0.38
1.0
0.5
0.47
Median
0.85
1.21
1.49
1.01
1.17
0.28
1.0
0.5
0.92
Mean
0.48
1.12
1.82
1.29
1.66
0.33
1.0
0.5
0.46
Mean
0.87
1.30
1.43
0.94
1.12
0.27
1.0
0.5
0.94
Standard Deviation
0.12
0.42
0.72
0.61
0.71
0.12
-
-
0.23
Standard Deviation
0.18
0.36
0.47
0.32
0.34
0.09
-
- •
0.24
                                                                 (Continued)
 a ratio for nitrate set to 1.0, ammonium ratio set to 0.5
                                              122

-------
Table 5-16. (Continued)
SBRP
S042'
Ca2+
Mg2+
Na+
K+
cr
°N03-
aNH4+
H+
Median
0.62
1.72
1.83
1.14
1.48
0.40
1.0
0.5
0.50
Mean
0.60
1.54
1.69
1.06
1.36
0.36
1.0
0.5
0.52
Standard Deviation
0.12
0.39
0.45
0.35
0.36
0.10
-
-
0.16
0 ratio for nitrate set to 1.0, ammonium ratio set to 0.5
                                                 123

-------
     As reported by Church et al. (1989), preliminary NOAA estimates of wet and dry sulfur deposi-
tion for the NE (sites in central Pennsylvania, Whiteface Mountain, New York, and Rowland, Maine)
and the SBRP (Oak Ridge, Tennessee) were highly comparable to regional averages of the DDRP esti-
mates.  The regional average of DDRP estimates of wet sulfur deposition in the NE was roughly 15
percent greater than the NOAA estimate for the sites examined and the DDRP estimate of dry sulfur
deposition was about 25 percent less than the NOAA estimate. The estimates of total sulfur deposition
in the NE were virtually identical; the total and even the individual component estimates (i.e., wet and
dry) in the SBRP were within 4 percent (B. Hicks, pers. comm.).

     A comparison of regional (i.e., NE and SBRP) dryAotal deposition ratios as obtained from the
DDRP estimates and "as quantified by NOAA for the same region" showed remarkable agreement for
both regions for sulfate, nitrate, and ammonium (Church et al., 1989). No NOAA values were available
for chloride, we could make so no comparisons for that ion.  The DDRP estimates of the dryAotal ratio
for base cations was generally just over twice as high as the NOAA values for the NE and ranged from
4 to 10 times as great in the SBRP. This difference is due at least partly to the fact that DDRP
estimates for base cation deposition included an estimate for coarse-particle dry deposition, whereas
the NOAA values did not. For example, a comparison of the average DDRP estimates of fine-particle
base cation deposition for five watershed sites in  proximity to the NOAA West Point station showed
good agreement (i.e., within 20 percent for calcium, sodium, and magnesium; within 50 percent for
potassium) with the measured NOAA values at that site (B. Hicks , pers. comm.). To account in part
for the uncertainties associated with the DDRP estimates of base cation dry deposition, Church et al.
(1989) performed sensitivity analyses with datasets having much reduced base cation values.  More
formal uncertainty analyses were performed with the integrated ^watershed models. In general, DDRP
analyses and the conclusions drawn from them were not  sensitive to these uncertainties (see Sections
9 and 10, Church et al., 1989).

5.6.3.3  Sulfur Deposition Scenarios

      Typical year total sulfur deposition (as sulfate) for the M-APP is  shown in Plate 5-3.  As
described in Section 5.6.1  (see Figure 5-12), the DDRP was requested to examine the effects of
scenarios  of both current and altered sulfur deposition in  the M-APP.  The sulfur increases and
decreases were performed as sulfate with both dry and wet deposition altered at equal  and constant
percentages (of the total) each year. No good existing data indicate any procedure other than
adjusting wet and dry H+ to coincide with changes in sulfur deposition (A. Olsen, R. Dennis, pers.
comm.). That was the procedure followed. Wet H+ was adjusted  equal to the wet sulfate adjustment
and dry H+ was  recomputed so that the sum of dry cation inputs was not less than the sum of dry
anion inputs (on an equivalent basis) in any month.  Values of typical year sulfate deposition for all
DDRP study sites are shown in Plate 5-4.
                                             124

-------

-------
Plate 5-3.  Pattern of typical year sulfate deposition for the DDRP M-APP study sites.
                                            125

-------

-------
APPALACHIAN  REGION
   Typical  Year
Sulfate Deposition

-------

-------
Plate 5-4. Pattern of typical year sulfate deposition for all DDRP study sites.
                                             126

-------

-------
            DDRP STUDY  REGIONS
       Typical  Year  Sulfate Deposition
SOr2  (g/m2)
©  < 2
©2-4
©4-6
®  6 -  8
•  > 8
 Mid-Appalachian
     Region
Northeast
                                Southern Blue  Ridge
                                     Province

-------

-------
5.7 RUNOFF DATA

     An estimate of average annual runoff for the DDRP study sites was necessary for all levels of
DDRP analyses.  Given the site selection procedures used in the DDRP, it is not surprising that the
DDRP study sites are ungaged and that measured values of annual runoff are not available. Three
options existed for obtaining estimates of runoff.  The first, gaging the systems, would not have been
practical, given the relatively short time frame of the DDRP and the large number of sites.  The second
option was to use an interpolation method, such as kriging,  to estimate runoff at each site. Large
variability in topography across the  regions, and  in other features that influence runoff, limited the
applicability of this method. Thus, we selected a third option for estimating runoff to each DDRP site:
interpolations were made from runoff contour maps developed from existing runoff data by expert
judgment of hydrologists  experienced in runoff mapping.

5.7.1  Data Sources

     Working in cooperation  with the USGS, a runoff contour map of average annual runoff for 1951-
80 (Figure 5-13, Krug et al., 1990) was developed for interpolating runoff at the DDRP sites.  The map
was developed to encompass the NE, M-APP, and SBRP regions of the eastern United States (Figure
5-13).  Average annual runoff  data for the 30-year period were taken  primarily from watersheds of less
than 2,590 km2 having no diversions or regulations. If a gaging station did not have a complete set of
records for the 30-year period, then Krug et al. (1990) calculated a 30-year estimate using the stan-
dard correlation methods described by Matalas and Jacobs (1964).  Runoff contours were plotted at a
scale of 1:500,000 at 5.1-cm (2-in) intervals up to 76.2 cm (30  in) and at 12.7-cm (5-in) contour inter-
vals for runoff greater than 76.2 cm (30 in). Krug et al. (1990)  provide more specific information on
map development and quality assurance.

5.7.2 Runoff Interpolation Methods

      A simple nearest-contour linear interpolation method was used to estimate runoff for each DDRP
site. The Krug et al.  (1990) map  was digitized into a GIS system by the USGS.  By means of the GIS
(Campbell  et al., 1989), the DDRP study sites were overlaid  onto the runoff contour maps and runoff
was interpolated at each  DDRP site to the nearest inch, based on the nearest contour to a site. The
nearest contour was determined  using an engineer's scale to  measure a line from the station location
perpendicular to the contour  (Rochelle et al., 1989b).

5.7.3  Uncertainty Estimates

      Determining a quantifiable  estimate of the uncertainly associated with the runoff interpolations is
important to the effective use  of the runoff data in the Levels I, II, and 111 analyses. Working with the
USGS, we conducted an analysis to estimate the uncertainties in using a runoff contour map to deter-
mine runoff at a specific site.  This  analysis was  incorporated  into the development of the 1951-80
runoff map (Rochelle et al., 1989b;  Krug et al., 1990).  We randomly  selected a subset of the total
USGS sites avaijable for map development and withheld these sites from use by the USGS as they
developed the map.  Then we used the runoff contour map to interpolate runoff at these sites and

                                             127

-------
                              ANNUAL RUNOFF 1951  - 1980
Figure 5-13. Example of average annual runoff map for 1951-80 (Krug et al., 1990).
                                        128

-------
compared the interpolated values to the actual long-term measured values. We determined that runoff
could be estimated, on the average, within approximately 8.9 cm (3.5 in. or 14.9 percent) of the actual
measured runoff. Rochelle et al. (1989b) provide a complete discussion of the uncertainty analysis.

      We conducted a second analysis to test the consistency of interpolating runoff using the hand-
linear interpolation method described in Section 5.7.2.  For the NE  region, we plotted 883 NSWS
watersheds on the 1951-80 runoff contour map, and interpolated runoff to each site.  Of the 883
NSWS sites, a subset of 146 watersheds was plotted onto the runoff contour maps, and runoff was
interpolated at the test sites a second time. We compared the two independent estimates to check for
consistency in using the hand-linear interpolation method.  We found that 11 percent of the sites had a
2.5-cm (1-in.) difference between the two interpolations (5 percent  runoff difference) and 1 percent had
a 5.1-cm (2-in.) difference  between the two runoff interpolations.  The results of a paired t-test indicate
that the hand  interpolation method is reasonably consistent, with no significant differences in runoff
between the two iterations (t = 0.65, p  = 0.51). Rochelle et al. (1989b) provide a description of these
uncertainty analyses.
                                              129

-------

-------
                                        SECTION 6
                           REGIONAL POPULATION ESTIMATION
6.1  INTRODUCTION
     This section describes the procedures used to extrapolate analyses on individual watersheds to
the target populations in the study regions.  This process of extrapolation is called population
estimation. •

6.2 PROCEDURE

6.2.1  Use of Variable Probability Samples

     As was done for the DDRP sample in the Northeast (NE) and Southern Blue Ridge Province
(SBRP)  (Church et al,, 1989), probability samples were selected for stream watersheds in the Mid-
Appalachian region (M-APP) (Section 5.2). Any quantity that can be defined for a sample unit (i.e., for
each watershed) can be extended to a corresponding population quantity through the probabilistic
structure of the sample. The quantity can be a measured variable or a model-based estimate. It can
be a number, a vector, or a function.  In the National Stream Survey (NSS), most quantities were
measured values, and the measurement error tended  to be small relative to the sampling variation.  In
contrast to the NSS, many  of the quantities produced  in the DDRP are model outputs believed to have
significant uncertainty associated with them.  The population estimation techniques provided in this
section  apply to any probability sample with defined inclusion probabilities.  Thus, they are applicable
to any identifiable subset of the DDRP sample. Explicit provision is made for including uncertainty
associated with the quantity that is extended to the regional population.

      In the NSS, and hence the DDRP, the size of the target population is not precisely known.  The
sampling frame for the NSS was an area/point frame based on USGS maps. There was not a popula-
tion of known size from which to draw samples. Thus, the size of the target population  is estimated
from the sample size. This presents no particular difficulty, as long as each unit in the sample has a
known inclusion probability.

      The design of the NSS and the DDRP also permits arbitrary subsetting of the sample.  In some
cases, the subsetting may  correspond to a redefinition of the target population (e.g., the exclusion  of
streams affected by urbanization). In such cases, the inclusion probabilities for the remaining sample
units would  not change, which implies a smaller target population.  In other cases, the subset should
be viewed as a subsample. In these cases, a smaller sample would be used to make an inference
about the same target population, and the inclusion probabilities would change. This might occur  if a
selected stream could not be sampled for some reason.  Inferences could still be made about the
same target populations, but the inclusion probabilities would change.
                                             130

-------
 6.2.2  Estimation Procedures for Population Means

      The structure of the DDRP sample is almost identical to the structure of the NSS sample. The
 estimation procedures parallel those detailed in the Sampling and Analysis Plan for Streams in the
 National Surface Water Survey (Overton, 1987).  Let n be the size of the sample selected from the
 target population, let p, be the probability that sample unit i was included in the sample, and let Py be
 the joint inclusion probability of units i and j. For sample unit i, let y( be the "true" quantity, and let Zj
 be the observed quantity, i.e., the unknown  true value with an associated error, e-t. The error may be
 an observation error or a measurement error; it could also be a prediction error. In each case, we
 assume the characteristics of the error distribution are known, and that the uncertainty in the observed
 values is characterized by that error distribution.  The basic estimation procedures follow the Horvitz-
 Thompson estimator (Cochran, 1977) for variable probability samples; some details, however, depend
 on assumptions made about the observation error. Several distinct error models are treated here.

      In one case, the uncertainty is due to  an additive error term, so that the magnitude of the uncer-
 tainty is constant over the range of the response.  The observation is related to the true value through
 the equation zl = Vj + e j. Two distributions were available to handle this case:  the error term is
 assumed to have either a normal distribution with mean 0 and variance a2 or a  uniform distribution
 over the interval (-a,a). For this uniform distribution, the mean is 0 and o2 = a2/3.

      In a second case, the magnitude of the uncertainty depends on the magnitude of the response.
 This can be modelled with a multiplicative error term, where the uncertainty is proportional to the
 response, so that z, = y (e,. We assume that the uncertainty follows a log-normal distribution with a
 mean value of 1 and a variance a2 = RSD2, where RSD is the relative standard  deviation.

      An implication of the above multiplicative model is that the uncertainty goes to 0 along with the
 response. In some instances, however, there is appreciable uncertainty even when the response is 0.
 For these cases, we assume that the uncertainty is proportional to the sum of the response plus an
 offset (h), so that the observation equation is z, = yl + (y, + h)e s = ys (ej + 1) + hej. The mean value
 of the error term is 0, and the o2 = RSD2. As above, we use a log-normal distribution for this case.

      The error structure affects only the variance of the population total, the variance of the
 population mean, and the estimator of the cumulative distribution function and its associated variance.
The estimator of the target population size and population total take the same form under all of the
above error structures.
                           *            *
Estimator of population total, T:
                                         T = Zz, /p.
(6-1)
Estimator of the size of the target population, M:
                                             131

-------
                                          N = 21/p,


                                               1
                                                                  (6-2)
Estimator of population average, ?:
                                           Y =  TIN.
                                                                                         (6-3)
      Both T and N are random variables, and both are unbiased estimators of the respective popu-

lation quantities.  However, ?, similar to most ratio estimators, is a slightly biased estimator of the

population average.




6.2.3 Estimators of Variance




      For all three error models, the estimator of the variance of T has the form
           n O-A
War (T ) =  2 _
                z    "  "
                 '  + 2  2
                                                              ) zizi\
                                                                ' '
                                    p
                                               />/     Pi Pj Pif
                                                                                         (6-4)
where g(e,z) is a function that depends on the error model and the sample data.  For the additive


model, g(e,z) = o2 N; for the multiplicative model, g(e,z) = o22zi2/pj, and for the multiplicative model


with offset, g(e,z)  = a2 2(z, + h)2/pj, where h is the offset.



               A

The variance of N is estimated by
           77  (1 -Pi)   '"  "
Var (N ) = 2      '   + 2  2
                                       p
                                                />/    PiPjPii
                                                                                         (6-5)
The joint inclusion probabilities, p^, are determined by the structure of the DDRP sample. They are

computed according to the algorithm in the Sampling and Analysis Plan for Streams in the National

Surface Water Survey (Overton, 1987).




      Finally, the variance of the estimator of the population average is obtained from a first-order

variance propagation using Equations 6-4 and 6-5:


                                                                                         (6-6)
                       Var(Y) =
                       Var(A/)/AT - 2TCov(T,N)/N*,
where
                                              132

-------
                                 n n
                    Cov (f,N ) = 2 2  (p, P/ - P// )(1/P/ - 1/Py )(z;- /Pf - zy /P/2)
                                                                                         (6-7)
Confidence intervals (Cl) are derived from the usual normal theory, e.g., a 95 percent Cl on the
population average is given by
                                        Y± 1.96\/Var (Y)
                                                                                         (6-8)
6.2.4 Estimator of Cumulative Distribution Function

      Let N(y) be the total number jn the population with the value of Y < y, so that the
cumulative distribution function of Y is FY (y) = N(y)/N. An estimator of N(y) is
                                A/z(y) =  2  1/P/ =  2 v, (y) /Pi
                                                                                         (6-9)
where

                             v. (y)    =   {i,Zi^  y
                               ' vy;        X0, Zj > y

An estimator of the cumulative distribution function of Y is


                                         Fv(y) = N(y)/N-
                                                                                       (6-10)
The variance of FY has both a sampling component and a component due to measurement uncer-
                      A                     A        A
tainty.  The variance of N(y) and covariance of N(y) and N are needed to calculate the sampling
variance of FY.  These are given by
                        Var(N(Y))=FY(y)(l-FY(y)) 2 1/P;2 + FY2
                                                                                      (6-11)
and
                                   Cov(N,N(y)) = FY(y)Var(N).
                                                                                       (6-12)
                                              133

-------
Then a first-order variance propagation formula gives
     Var(FY) =  Var(N(y))/N 2 + N 2(y)Var(N)/N 4 - 2N(y)Cov(N(y),N)/N2
for the sampling variance.
  (6-13)
      A Monte Carlo procedure is used to calculate the measurement variance.  The sampling vari-
ance and the measurement variance are added to obtain total variance.  The median and quintiles of
                                                          A
the distribution of Y are estimated by the linear interpolation of FY.

6.2.5  Hypothesis Testing for Changes in Population Totals

      In tables such as Tables 10-9 and 10-10, population totals are estimated using the methodology
outlined in Sections 6.2.2 - 6.2.4. It is natural to question whether the apparent changes in population
totals across deposition scenarios, or through time within individual deposition scenarios, are statistic-
ally significant.  In each case, the question is whether the number of discrepant watersheds in the
table  of population estimates is significantly  larger (or smaller) than zero.

      We have specific expectations about the direction of change between scenarios, based on our
understanding of the chemistry of the situation. In moving from a higher deposition scenario to a
lower deposition scenario, we expect to see either minimal change in surface water chemistry or an
increase in surface water ANC.  In this case, the difference of the population totals would be expected
to remain constant or decrease.  For this reason,  we may treat all rejection regions for the hypothesis
tests as one-sided.

      To estimate the number of systems in the target population showing changes between two
deposition scenarios, we can look at the difference between the projected model values for the two
scenarios at a particular site.  Because of the aforementioned scientific considerations, we know in
theory which way to compute the difference so that the difference will be 1 if the model values indicate
a change between scenarios, and 0 otherwise. Then the total number of systems  in the target
population can be estimated by using the Horvitz-Thompson estimator:
                                                 d,/p,
(6-14)
where dj can be written as

                       dj = I{ANC1 < 0} - I{ANC2 < 0},

with l{ }, the characteristic function, taking the value 1 if the quantity inside the brackets is true, and 0
otherwise.
                                              134

-------
      Similarly, to estimate the number of systems in the target population whose chemistry would
 change over time for a particular deposition scenario, we can examine the difference between the
 characteristic functions of the projected model values of ANC for the scenario at two different times.
 Here also, because of the scientific considerations, we know in general which way to compute the
 difference so that the difference will be 1 if the model values indicate a change over time, and 0
 otherwise. Again, the total number of systems in the target population can be estimated by using the
 Horvitz-Thompson estimator.
                                         A
      Now we must estimate the variance of T. This variance should include not only the standard
 terms in the variance estimates for variable probability sampling, using the Horvitz-Thompson
 estimator of the variance expressed previously in this section, but also an estimate of the variance
 component that is due to the variability of the  model inputs and the model sensitivity to these inputs.

      We obtained this second component from the individual model runs performed during the cali-
 bration of the Model of Acidification of Groundwater in Catchments  (MAGIC). For each watershed,
 there were up to 10 acceptably calibrated runs, and a watershed was considered uncalibrated unless
 there were at least 5 such runs. The final watershed value given by the model was the median of the
 values for these runs.

      The program (i.e., MAGIC) used a Latin  Hypercube sampling  algorithm, which is a stratified
 Monte Carlo sampling method, to select the vectors of values for the model runs. Variance estimates
 of the inputs were incorporated in selection of the vectors.  For this reason, we decided that it was
 acceptable to treat the variance of the model outputs as an estimate of the within-watershed variance
 due to input variability and model  sensitivity.

      The analysis was performed on the differences of the characteristic functions rather than on the
 original ANC. Therefore,  we computed this variance on the differences of the characteristic functions
 in the individual model runs. For comparisons between scenarios, we matched the model  runs across
 scenarios such that the input variables other than deposition were matched  (J. Cosby, pers. comm.).
 For comparisons over time, we took the difference between the values of the characteristic function in
 the same model run at the two years  of interest.

      Because we worked with differences of characteristic functions within watersheds, we treated the
 data as  independent observations on a (possibly degenerate) standard trinomial distribution, that is,  a
 discrete distribution taking only the values 1, 0, and -1. The calculated variance is that of one obser-
vation, not the median of 5-10 observations, some of which may have been excluded as  part of unac-
 ceptable runs.  We did not have an estimate of the effect of this censoring, and therefore we have not
calculated the variance of the median of censored observations from a trinomial distribution. Instead,
we used the calculated variance as an overestimate of this component  of the variability.  The effect on
the overall variance appeared to be negligible, except in those cases where the watershed  differences,
dj, were identically zero.   In all these cases, the change in population totals is zero, and hence not a
statistically significant change, regardless of the magnitude of this component of the variance.
                                             135

-------
     We assumed an additive model for the variance estimates. Because we worked with the d|
rather than the ANC, we did not have problems with skewed data, and a multiplicative model was not
necessary. Therefore, for the term representing the total variance due to input variability and model
sensitivity, we used s2 N, where s2 is the pooled variance estimate obtained by taking a weighted
average of these within-watershed variances, with the inverses of the watershed inclusion probabilities
used as the weights.
                                                            A
     We used the Horvitz-Thompson estimator of the variance of T, since we were working with
subsets of the regions. We used the py estimates discussed in Section 6.2.3.

     The sum of the Horvitz-Thompson variance estimate and the model sensitivity term provides an
estimate of the total variance in the estimate of the difference in the target population. The square root
of this sum was treated as the standard deviation, s-}-, of T, and a standard z-test
                                         z = (T - o)/sf.
(6-15)
was used as an approximation.  As stated earlier, we have scientific reasons to expect change in a
particular direction, so a one-sided hypothesis was used in each case.  The usual p-values for one-
sided normal tests were generated in SAS as part of the calculations.

6.3 UNCERTAINTY ESTIMATES

      The quantities displayed in this report are the end result of a sequence of operations, beginning
with the collection of a physical sample in the field and ending with the production of a table or graph.
We conducted chemical analyses, data aggregation, data reduction, and processing of the data
through various mathematical models. The final result contains  an element of uncertainty that
originates, first, in the design, then in the implementation of the  field protocol, and finally in the
precision of the basic measurement  process (e.g., the chemical analytic precision).  The uncertainty
on the final result can be quantified by propagating the uncertainty, or its mathematical analog,
through the same sequence of operations as the data.

      In the DDRP, we have used several techniques to propagate uncertainty through a functional
relationship, which could be a complex simulation model as well as an explicit function.  Let f(x1( x2	
xn) be a function of the variables x1 ,x2	xn, with uncertainties e.,, e2, ..., en, respectively.  The
probability distributions,  or at the least the variances, of the uncertainties are presumed  known. If the
functional relationship is such that partial derivatives can be easily obtained, then the variance of
functional values can be estimated using a first-order linear approximation to the functional
relationship:
                                      Var(f) =
(6-16)
      In the case of a simulation model, the function is the model itself, and the partial derivatives
cannot be calculated explicitly. An approximation to the partials can be obtained  by perturbing the
in turn.  If a suitably small perturbation is chosen, then the ratio of the change in output to the

                                             136

-------
perturbation is an estimate of the partial derivative. These estimates can then be used in a first-order
propagation, as above.

      A disadvantage of both these techniques is that they ignore possible correlations among the
uncertainties. One way to account for such correlations is to propagate not only variances but also
covariance terms. The "first-order, second-moment" technique used in the Enhanced Trickle Down
uncertainty analysis is a means of doing exactly that.  A first-order approximation is made to the
model, and Kalrnan filtering techniques are used to build up an estimate of the state variable variance-
covariance matrix.  A final method that we used in uncertainty assessment was Monte Carlo. The
Monte Carlo method is applied by repeatedly calculating the value of f, each time perturbing  the value
of each Xj by a random quantity drawn from the respective uncertainty distribution.  Monte Carlo is
most easily appSied  when uncertainties are statistically independent, but it can also be applied when
correlations exist. A variant of Monte Carlo, called "fuzzy optimization", was used in the uncertainty
analyses for MAGIC.

6.4 APPLICABILITY

      This section discusses the Level II and III population estimation approaches for DDRP,  including
the statistical formulas used to  estimate population means, variances, and cumulative frequency
distributions. The population estimation procedures are generic and do not depend on the level of
analysis. The specific target populations for inference, however, do depend on the analyses
performed.  Not all DDRP watersheds were used at each level of analysis, so  the target population
varies. The explicit target populations being considered in the analyses are discussed in Sections 9
and 10. The generic uncertainty estimation procedures introduced  in this section also are more
explicitly discussed for each of the individual analyses in Sections 9 and  10.
                                             137

-------
                                         SECTION 7
                              WATERSHED SULFUR RETENTION
7.1  INTRODUCTION
      The fate of sulfur deposited in a watershed is important in determining the response of the
associated surface water because sulfate can act as a mobile anion in the soil matrix (see Section 3).
In systems at steady state with regard to sulfur deposition (i.e., inputs = outputs or zero net retention),
the leaching rate of either basic or acidic cations by the "carriefanion" sulfate has been maximized.
Given no increase in sulfur deposition, future acidification (loss of ANC) of these systems would be
determined principally by cation leaching and the possible depletion of the soil exchange complex. In
systems below sulfur steady state (i.e., inputs > outputs), the acidifying effect of sulfate-driven cation
leaching has not been  maximized.  As sulfate leaching increases in these systems, soil adsorption
sites are filled  on a  net basis, and acidification and the rate of acidification increase over time. A
circumneutral  lake or stream draining a watershed with positive net sulfur retention will continue to
acidify,  possibly becoming acidic (i.e., ANC < 0), as long as rates of sulfur deposition (inputs) exceed
outputs. Thus, even if sulfur deposition decreases, some circumneutral systems will acidify and may
become acidic. Knowing the patterns of watershed sulfur retention, therefore, is important with regard
to understanding and forecasting the potential effects of sulfur deposition on surface water chemistry.
In this section, we examine regional patterns of sulfur retention, as estimated using input/output
budget  analyses.

      The purpose  of the watershed sulfur retention component of the  DDRP Level I analyses is to
estimate the current status of annual sulfur retention in watersheds of the eastern United States, with
primary emphasis  on the NE, M-APP, and SBRP regions.  The M-APP Region provides important  infor-
mation  for the interpretation of sulfur retention patterns for the NE and  SBRP, as well as a more
complete regional perspective for upland areas of the entire eastern United States.  Specific objectives
of these analyses are to:

      •  characterize  current average annual input/output budgets in the NE, M-APP, and SBRP using
          (1) data from intensively studied sites and (2) estimates computed from regionally extensive
          datasets; and
      •  compare annual sulfur retention patterns within and among regions to determine possible
          trends relative to water chemistry, soils,  and atmospheric deposition.

 Many of the analyses presented here were reported with DDRP results for the NE and SBRP region
 (Church et al., 1989), but are repeated here for completeness and to provide a context for discussions
 of sulfur dynamics  in subsequent sections  (9.2 and 10). The previous  DDRP report also provided (1) a
 detailed assessment of sulfate reduction  in wetlands and in-lake sediments as a watershed sink for
 sulfur,  (2) a detailed assessment of internal sulfur sources in the DDRP watersheds, and (3) a region-
 alized uncertainty analysis for sulfur input-output budgets.  These issues are summarized briefly in this
 section; readers are referred to Church et al. (1989) for a more complete discussion of those topics.
                                               138

-------
 7.2  RETENTION IN LAKES AND WETLANDS

       Church et al. (1989, Section 3) identified and discussed several processes that can cause sulfur
 to be retained within watersheds.  One of the processes considered was retention by sulfate reduction
 in wetlands and/or lakes. Retention in these environments occurs principally by dissimilatory reduction,
 with  sulfate used as an electron acceptor and with hydrogen sulfide, organic sulfur, or metal sulfides
 as end products (Rudd et al., 1986a,b; Brezonik et al., 1987).  Analyses in the DDRP (Shaffer et al.,
 1988; Shaffer and Church, 1989; Church et al., 1989) were designed to estimate in-lake sulfur retention
 in the ELS populations of drainage lakes and impoundments; in the NE, SBRP, and Upper Midwest
 (UMW) (Unthurst et al., 1986; Kanciruk et al., 1986),  retention in seepage and closed lake basins was
 not evaluated due to high uncertainty in the hydrologic budgets for those systems. ELS sulfate data,
 along with precipitation and runoff estimates generated within the DDRP (or by ELS for the Upper
 Midwest), were used with an in-lake sulfur retention model developed by Baker et al. (1986b) and Kelly
 et al. (1987) to estimate percent sulfur retention.  Results showed that in all three regions, median in-
 lake sulfur retention was 5 percent or less; in-lake retention was estimated to exceed 10 percent of
 deposition in only 13 percent of NE lakes and in fewer than 1 percent of lakes (mostly impoundments)
 in the Southern  Blue Ridge.

      Regression analyses were used in the DDRP.to evaluate sulfur retention in wetlands and wet
 soils for the NE; results indicated a statistically significant positive correlation between the proportion
 of wetlands on a watershed  and the percent of watershed sulfur retention. For many reasons, quanti-
 tative use of these results or of any similar generalization of wetland area-sulfur budget relationships is
 difficult, and we have not attempted any such use. The importance of wetland retention on watershed
 sulfur budgets depends on the location of the wetland in the watershed and the portion of watershed
 runoff flowing through it. Also, sulfur reactions in wetlands and wet soils can change seasonally or in
 wet/dry years. Wetlands and wet soils can act as sulfur sinks (reduction of sulfur) during wet periods
 when the system is anaerobic, but can become major sulfate sources due to reoxidation of sulfides
 upon drying (Bayley  et al., 1986; Nyborg, 1978). Interpretation of these analyses was further compli-
 cated by high  uncertainties in both the independent (% wetlands) and dependent (% S retention) vari-
 ables in the regression equation.

      Qualitative evaluation of soil and hydrologic characteristics on DDRP M-APP watersheds sug-
 gests that sulfate reduction will not be an important long-term sink on most watersheds.  Histosols
 occur on only five watersheds in  the M-APP  sample, and represent less than 2 percent of watershed
 area on any of those catchments. Poorly drained soils, subject to seasonal flooding, cover more than
 5 percent (maximum  = 19 percent) of the area on  15 watersheds in the M-APP sample. These soils
 could provide reducing environments for sulfate reduction during wet  periods, but any sulfate seques-
 tered  during those times would in all likelihood be quickly reoxidized upon drying and reaeration of the
 soil. Based on these low proportions of permanent anaerobic zones on the watersheds, the potential
for net sulfate reduction in M-APP watersheds appears to be small relative to other sulfur sources/
sinks, and we do not address it further here.
                                             139

-------
7.3 WATERSHED SULFUR RETENTION

      Two approaches have been used in the DDRP to characterize regional patterns of watershed
sulfur retention.  The first approach was to compile and review sulfur inputyoutput budgets at
intensively studied sites (Rochelle et al., 1987). Figure 7-1 summarizes the findings of this review.
Definitive statements about region-wide sulfur retention could not be made on the basis of these data,
however, because of lack of spatial coverage by the intensively studied sites and inconsistencies in
data used for budget calculations.  There are, however, clear trends in sulfur retention from north to
south in the eastern United States, especially relative to the extent of the Wisconsinan glaciation, with
higher retention in the southern areas (Figure 7-1). The DDRP Level I sulfur retention analysis
examines these apparent trends in more detail using a second approach, which has involved gener-
ating regionally consistent sulfur input and output data (Section 5) for the surface water sites sampled
by the Eastern Lake Survey (ELS) and National Stream Survey (NSS), then characterizing the sulfur
budget status for the representative NSWS populations of watersheds on a region-wide basis.

7.3.1  Methods

7.3.1.1  Input-Output Calculation

      Level I sulfur retention analyses use an annualized mass balance approach to estimate percent
retention.  The general equation used to  calculate percent sulfur retention is:

         % Retention = ((Sw +  Sd - R * SS)/(SW + Sd ))*100                               (7-1)

where:   Inputs
             Sw  = wet sulfur deposition (mass/length2/time)
             Sd  = dry sulfur deposition (mass/length2Aime)
         Outputs
             R   = runoff (length/time)
             Ss  = surface water sulfur (mass/length3)

Equation 7-1 relates the total sulfur input (on a mass basis) to total sulfur output for each watershed.
We applied this equation to all ELS lake systems (except seepage and closed lakes) and NSS water-
sheds in the regions of interest.

7.3.1.2  Data Sources

7.3.1.2.1 Inputs-

     Wet sulfur deposition was estimated for each site using chemistry data from the National Trends
Network/National Acid Deposition Program (NTN/NADP) network and precipitation data from the
National Oceanic and Atmospheric Administration (NOAA), National Climatic Data Center (NCDC)
network (Section 5.6).  Briefly, wet sulfate concentrations and precipitation were kriged to each site,
and wet deposition was calculated (see Wampler and Olsen, 1987, for a detailed description of the
                                             140

-------
          \\» 1    '-'"r~"
      SULFUR RETENTION
              RETENTION


                RETENTION


                RETENTION
Figure 7-1.   Percent sulfur retention for intensively studied sites in the United States and

             Canada relative to the southern extent of the Wisconsinan glaciation (adapted

             from Rochelle et al. (1987).
                                          141

-------
calculation). Dry sulfur deposition was estimated based on output from the Regional Acid Deposition
Model (RADM) (see Section 5.6). Several deposition data sets have been developed during the DDRP
to meet a variety of modelling needs; for a thorough description of the nature of each data set and the
differences among them, see Section 5.6 of Church et al. (1989).

7.3.1.2.2 Outputs-

     We used estimates of long-term average annual runoff based on data for the 30-year period of
1951-80 (see Section 5.7 for details). For these analyses we used NSWS sulfate data, which measure
only dissolved inorganic sulfate; it represents the vast majority of sulfate in most surface waters and
does not significantly compromise analyses (David and Mitchell, 1985; Mitchell et al., 1986). Section
5.3 discusses the surface water chemistry  data used in these analyses. For additional information
concerning the ELS and NSS surface water sulfate estimates, see Linthurst et al. (1986), Messer et al.
(1986), and Kaufmann et al.  (1988). Seepage lakes and closed lakes were excluded from the
analyses.

7.3.2 Uncertainty Estimates for Sulfur Input-Output Budgets

     As part of sulfur input-output budget analyses, Church et al. (1989) estimated the uncertainty
associated with calculations  of average annual sulfur retention using  Monte Carlo analysis, similar to
that described in Section 6.3.  Uncertainties for individual input/output variables in Equation 7-1 were
determined, then propagated through the retention equation to determine an estimate of overall uncer-
tainty of the percent retention calculations. A total of 10,000 iterations of the budget calculation were
made for each of 34 randomly selected watersheds in the NE, M-APP, and SBRP.  From those results,
Church  et al. (1989) developed a multiplicative normal relationship between steady-state sulfate and
uncertainly, which  can be used to identify  catchments for which sulfur retention or release is sig-
nificantly different from zero:
                         Est. std. dev.  =  - 4.9 + 0.339
                                                                 2-
                                                                   JSS
(7-2)
Results of this analysis were very useful for identifying watersheds with putative internal sources of
sulfur. We excluded these watersheds from subsequent analyses. An alternate approach for identi-
fying internal sources of sulfur, which focused on bedrock mineralogy, was largely unsuccessful due
to the spatial resolution of state-scale geologic maps. For example, the location of unit boundaries
was uncertain, and small sulfide-bearing bodies either were included in a general map unit or not
delineated  at all.  Watersheds having internal sources of sulfur were ultimately identified on the basis
of (1) the uncertainty estimates described here, (2) identification of outliers/influence points in
regression  analyses, and/or (3) mapped delineation as strip mines, borrow pits, etc. For the two
regions (i.e., M-APP and SBRP) having a majority of watersheds with significant sulfur retention, we
focus subsequent analyses on assessing the importance of adsorption as a retention mechanism
(Sections 9 and 10). For a thorough discussion of uncertainty analyses and of identification  of water-
sheds suspected  of having internal sources of sulfur (including a list of watersheds dropped  from
budget analyses), see Church et al. (1989, Sections 7.3.2 and 7.3.3).
                                              142

-------
7.3.3  Input-Output Budgets - Results and Discussion

      This section summarizes, at a regional scale, the calculated percent sulfur retention for sites
located in the NE, M-APP, SBRP, and several adjacent regions.  We eliminated from the analyses sites
identified as having internal sulfur sources through the steady-state sulfate concentration analysis
(listed in Table 7-6 of Church et al., 1989).  In addition, three sites in the SBRP that were sampled as
part of the Pilot Stream Survey were dropped due to outlier surface water chemistry (i.e., ANC  > 1000
/90 percent  of watershed area) was covered by wetlands or chronically flooded
soils.  Results for the region indicate that most watersheds in the NE are at or near steady state with
respect to atmospheric deposition of sulfur. These analyses also indicate, however, that at a few sites
in the region, sulfate reduction in soils or lake sediments acts to maintain sulfate concentrations at a
level below steady state.

7.3.3.2 Mid-Appalachian Region

      The Mid-Appalachian Region does not present as clear a picture of percent sulfur retention as
the NE (Table 7-1; Figure 7-2B).  Retention for individual watersheds varies widely within subregions,
exceeding 90 percent in several cases, while net release is > 80 percent in catchments in several
subregions. For this study, we have defined the  M-APP Region as a combination of NSS subregions
2Bn and 2Cn (Plate 7-1). Kaufmann et al.  (1988) defined these regions  as the Valley and Ridge and
Northern Appalachians, respectively.  We found that percent retention was widely distributed within
each subregion, with no clear intraregional patterns of low or high percent net retention.  In general,
for both subregions,  net retention is low, averaging less than 30  percent. On average, retention in
                                              143

-------
Table 7-1.    Percent Sulfur Retention - Summary Statistics by Region for Lakes and Stream
              Reaches in the NSWS (ELS) and NSS
Region
NE
ELS Rg. 1b
1A
1B
1C
1D
1E
Mid-App
NSS 2Bn
NSS 2Cn
SBRP
PSS
ELS Rg. 3a
Misc.0
S. App. Pla,
NSS2X
Piedmont
NSS3A
Mid Atlantic Coastal
NSS SB
Poconos/Catskills
NSS 1D
Na
5,828
1,099
1,285
1,190
966
1,288
12,580
6,478

2,031
247

4,329

7,199
Plain
9,535

2,724
Mean
-5
-12
9
-4
-12
-9
28
-4

68
68

43

68

31

-22
Median
-5
-14
8
-7
-9
-12
40
3

75
79

50

78

34

-29
Std. Dev.
28
23
26
27
30
27
43
32

23
33

38

24

38

31
Min.
-70
-64
-66
-66
-70
-62
-83
-83

— *54.
\JT
-64

-64

-10

-60

-71
Max.
73
61
73
63
54
51
91
55

88
93

86

92

93

67
a Estimated target population calculated using NSWS weights (see Linthurst et al., 1986a; Messer et al., 1986a;
  Kaufmann et al., 1988 for information on weights).

b Lakes sampled by the Eastern Lake Survey (ELS). All other sites represented in this table are stream reaches.

c Additional regions sampled as part of NSS Phase I (see Plate 7-1).
                                                  144

-------
                                             NE
                                1.0r
                             O 0.8
                             O
                             0.

                             S 0.6
                             

                            «= 0.4
                            as


                            1

                            o o-2
                                         Upper Bound
                                         Projected
                                         Lower Bound
                               0.0
                                -100     -50     0      50

                                    Percent Suffur Retention
         100
Figure 7-2.    F'opulation-weighted distribution of projected percent sulfur retention, with

               upper and lower bounds for 90 percent confidence intervals, for NSWS target

               population of (A) lakes in the Northeast, and for stream reaches in (B) the

               Mid-Appalachian, and  (C) Southern Blue Ridge Province Regions.
                                              145

-------
Plate 7-1.  Sulfur retention and wet sulfate deposition for National Surface Water Survey
           subregions in the eastern United States.
                                            146

-------

-------
                        NSWS   SUBREGIONS
             yEDIAN  PERCENT   SULFUR  RETENTION
                AND  WET  SULFATE   DEPOSITION
                                                           2.25
MEDIAN  PERCENT
SULFUR  RETENTION
    o  - 20

    20 - 40

    40 - 60

    60 - 80

    80 - 100
Average Annual
Wet Sulfaie       ^   2.75-
Deposliion (g nf2 yr"1)*  3.06-
            3.25
                                                       Eosiern Loke Survey
                                                                 -2.25
                                                       Subregion  X Retention
                                                         1A
                                                         16
                                                         1C
                                                         ID
                                                         IE
                                             -14
                                               8

                                              -a
                                             -12
                                                  2.00
                                                       Notional Stream Survey

                                                                liedian
                                                       Subregion  X Retention
                                                         2Cn
                                                         26(1
                                                         3B
                                                         n
                                                         Us
                                                         3A
                                               3
                                              40
                                              34
                                              50
                                              75
                                              78
                                             *DeP05i-t:ion for 1980 - 1984
                                              (A. Qlsen. Personal Communication)

-------

-------
subregion 2Cn is estimated to be lower than in subregion 2Bn. Subregion 2Cn receives higher sulfur
deposition than does subregion 2Bn (Plate 7-1).  Subregion 2Cn also has a high incidence of systems
with potential acid mine drainage influence, but systems identified by Kaufmann et al. (1988) or
identified by steady state analyses as haying probable internal sulfur sources (Church et al.,  1989,
Section 7.3.3) have been dropped from this analysis and presentation of results, and we do not
believe that they significantly affect our results.

      The Southern Appalachian Plateau and the Mid-Atlantic Coastal Plain have percent retention on
the average of 30 to 40 percent (Table 7-1;  Figure 7-3). Although these regions show a large amount
of scatter in percent sulfur retention, they are characterized as having a majority of systems with
substantial net sulfur retention. Retention in these areas is also consistent with the general trend of
increasing sulfur retention along a north-to-south gradient. The geographic retention gradient is, in
turn, attributable to differences in soil age and physico-chemical characteristics that result in increased
sulfate adsorption along a comparable north-south trend, as  is discussed in Section 9.2.

7.3.3.3 Southern  Blue Ridge Province

      Median net sulfur retention for the SBRP is approximately 75 percent (Table 7-1; Figure 7-2C).
Rochelle and Church (1987), working with sulfur deposition data from Water Year 1984, found similar
results. The average percent sulfur retention for the Piedmont Region (adjacent to the SBRP) is also
high compared to the NE and M-APP  (median = 78.0, Table 7-1 and Figure 7-3).

7.3.4  Regional Patterns of Sulfate Concentration and Steady-State Sulfate

      Comparison of sulfur budgets for the three major DDRP regions in the eastern United States
shows clear regional differences in budget status. Budgets in the NE are, on average, very close
to sulfur steady state; there is high  net retention in the SBRP. Retention is highly variable in the
M-APP Region, ranging from high retention in some systems to high net release in others. Viewed
differently, budget data indicate that the relative potential for increases in surface water sulfate
concentration as watersheds  come to steady state (assuming current deposition)  is small in  the NE,
intermediate (for the region) and highly variable (for individual watersheds) in the M-APP, and high in
the SBRP.

      In considering the potential for future effects of acidic deposition on soils or surface waters, it is
important to consider not only budget status (i.e., the potential for relative changes in sulfate), but also
the potential for absolute change in sulfate (i.e., the -difference between present and steady-state
concentrations). Figure 7-4 and Table 7-2  present data for DDRP sample watersheds for sulfate
concentration, steady-state sulfate concentration, and percent sulfur retention. The sulfur distributions
shown here are somewhat different from those presented earlier  in this section because the previous
data  describe regional ELS or NSS lake/stream populations, whereas these data represent smaller
DDRP target populations, from which we have excluded certain subsets of lakes/streams (e.g.,
seepage lakes, systems with high ANC, watersheds larger than 30 km2).
                                              147

-------
       Southern  Appalachian  Plateau    A
                                                        Mid-Atlantic Coastal Plain
                                    B
    to
  o
    0.8
  o
  0.
  o
 fc
  a
 E
 o
    0.0
     -100
               Upper Bound
               Projected
               Lower Bound
            •so     o      so
         Percent  Sulfur Retention
                                 100
                                                    to
                                                 O 0.8
                                                3=
                                                 O
                                                 a.
                                                  
-------
       LAKE/STREAM SULFATE
                                         STEADY STATE SULFATE
  1.0
              100   150   200   250
                SULFATE ftieq/L)
                                300
                                     350
                                           o
                                             0.0
                                          50  100  150 200  250  300  350  400  450
                                                  SULFATE (neq/L)
    PERCENT SULFUR RETENTION
                                         CHANGE IN SULFATE TO
                                              STEADY STATE
                                             1.0
 o
   0.0
    -60
    -i	r
-40  -20   0   20   40   60   80
   PERCENT SULFUR RETENTION
                                     100
                                           O
                                           I— 0.8-
                                           CC
                                           2
                                           O 0.6 i
                                           CL
                                           111
                                           > 0.4-
                                             0.2-
                                           O
                                             0.0
-50    0    50    100    150   200
         DELTA SULFATE (neq/L)
                                                                               250
Figure 7-4.   Comparisons of regional distributions of (A) measured sulfate concentration,
            (B) steady-state sulfate concentration, (C) percent sulfur retention, and (D)
            change in sulfate concentration to steady state for DDRP target populations of
            lakes or streams in the NE, M-APP, and SBRP.
                                       149
-------
Table 7-2.     Summary Statistics for Distributions of Sulfate Concentration, Steady-State Sulfate
              Concentration, Change in Sulfate Concentration to Steady State, Percent Sulfur
              Retention for DDRP Target Population Watersheds in the NE, M-APP, and SBRP
Variable
SO4 concentration
(Meq/L)

Steady-state SO4
(«eq/L)

Change in sulfate
to steady state
(Keq/L)

Percent sulfur
retention


Region
M-APP
NE
SBRP
M-APP
NE
SBRP
M-APP
NE
SBRP
M-APP
NE
SBRP
Mean
148
110
37
246
118
122
98
8
85
37
4
71
Std.Dev
55
40
26
72
47
23
-
-
24
20
17
Min.
55
34
15
140
50
86
-39
-51
25
-21
-54
17
Median
153
105
24
215
111
120
107
-0
83
44
0
78
Max.
313
214
119
405
239
203
248
28
167
74
69
87
                                          150
-------
      Steady-state sulfate concentrations in the NE and SBRP are similar, with medians of 111 and
120 ^eq/L, respectively. In contrast, due to the higher sulfur deposition to M-APP watersheds, steady-
state sulfate concentrations (median = 215/weq/L, maximum = 405/
-------
lowsulfate adsorption capacities (Section 9.2; see also Church et al., 1989; Rochelle et al., 1987).
Soils of the M-APP Region are predominantly Inceptisols and Ultisols.  Measured sulfate adsorption
characteristics of soils from the three regions, discussed in Section 9.2, clearly demonstrate regional
differences in adsorbed sulfate pools and potential adsorption. Given the current patterns of wet
sulfur deposition  (Plate 7-1) and assuming that the M-APP Region  has received elevated levels of
sulfur deposition for a considerable period of time (Section 9.2), it appears that sulfate breakthrough
from the soils has occurred and that the region is in transition to lowered percent net sulfur retention
and significantly elevated surface water sulfate concentrations. There is little doubt that the increases
in sulfate are a direct consequence of anthropogenically derived increases in atmospheric sulfur
deposition.

     The onset of high rates of sulfur deposition occurred later in  the SBRP than in other regions
(Gschwandtner et al., 1985).  That  late onset, coupled with the high sulfate adsorption capacity of soils
in SBRP watersheds, has allowed the SBRP to experience a "lag" or "delay" phase in  response to
acidic deposition. Model projections for the SBRP and  sulfate trend data for monitored watersheds in
the region indicate, however,  that soils and streams in the SBRP are beginning  a transition to higher
sulfate outputs as sulfate begins to break through the soil column.  The dynamics of  sulfate transitions
in the NE and SBRP were described by Church et al. (1989) (Sections 9 and 10); we present summa-
ries of model projections for those  regions, along with a detailed discussion of projections for the
M-APP Region, in Sections 9 and 10 of this report.
                                             152
-------
                                         SECTION  8
                              LEVEL I STATISTICAL ANALYSES
8.1  INTRODUCTION
     The chemistry of surface waters in natural systems results from the integrated effects and
interactions of inputs from deposition, terrestrial processes, and in-lake or in-stream processes.  In the
design of the DDRP, a sequence of four project objectives and three types of data analyses and model
applications were defined to characterize and quantify linkages  between deposition, watershed
processes, and surface water chemistry (see Section 4). The first type of analysis (Level I) was
designed as a set of statistical analyses, principally multiple regression, that would facilitate
identification and understanding of associations between soil and landscape attributes and current
surface water chemistry. Level I analyses address the first two DDRP objectives:  (1) to provide a
regional description of soil and watershed characteristics, and (ty to characterize  relationships
between watershed attributes (soil chemistry, geology, vegetation, etc., along with deposition) and
surface water chemistry. Because much of our understanding of watershed-surface water chemistry
relationships has been based on site-specific data, which might or might not be representative of
regional conditions, an important objective of Level I statistical analyses in the DDRP was either to
corroborate, at regional scales, those fundamental understandings and assumptions (i.e., associations
between watershed attributes and surface water chemistry) or to identify additional important rela-
tionships not presently incorporated in conceptual or mathematical watershed chemistry models used
for Level  II and 111 analyses.

     In previous DDRP analyses for the NE and SBRP  (Church et al., 1989), we conducted an exten-
sive Level I assessment of relationships between soil chemistry  and/or groups of landscape variables
(geomorphic and hydrologic variables, depth to bedrock, land use and vegetation, bedrock geology,
and areal extent of soil sampling classes) and surface water chemistry. We evaluated landscape
attributes both as individual data sets and as an integrated set of variables.  Results  of those analyses
are summarized in  this section.  We did not conduct an analogous set of analyses for the Mid-
Appalachian (M-APP) Region. This  decision was based on two  principal considerations. First, the
analyses were extremely time consuming. Second and more important,  there were no real surprises
in the results of the Level I analyses for either the NE or the SBRP. The results generally confirmed
the importance, at  regional scale, of previously established  relationships between surface water
chemical constituents and watershed attributes. A detailed description of Level I regression analyses
and results for the  NE and SBRP is presented by Church et al.  (1989), Section 8.

8.2 SUMMARY OF NE AND SBRP APPROACH AND RESULTS
      The general approach used for Level I analyses was multiple regression, in most cases con-
ducted using stepwise procedures (SAS Institute, Inc., 1987, 1988).  Residual analysis allowed us to
identify and remove outliers and leverage points from regression analyses.  Five surface water chemis-
try variables (sulfate concentration, percent sulfur retention, ANC, pH, and sum of base cations [sum
of Ca + Mg in some cases]) were used as dependent variables in the regression analyses.  Five
different groupings of independent variables were used:

                                             153
-------
      •   Deposition to the watershed.

      •   Deposition plus one set of landscape attributes.  Five'sets of landscape attributes were
          developed and used, including: (1) geomorphic/hydrologic variables, (2) land use/vegeta-
          tion, (3) depth to bedrock, (4) bedrock geology, and (5) soil sampling classes.  Several
          approaches were used to quantify individual attributes: (1) areal extent of each  member of a
          class of attributes  (e.g., percent of watershed area with a particular bedrock type,
          percentage in each depth-to-bedrock class) used for bedrock geology, sample  class, and
          depth to bedrock, (2) measured values, average value, or critical value (e.g., minimum or
          maximum) used for geomorphic variables, derived hydrologic variables, bedrock sensitivity
          class, and depth to bedrock, and (3) a set of factors developed from principal component
          analysis (land use/vegetation).

      •   Soil chemical and physical properties, using values aggregated to watershed scale.

      •   Integrated analysis using deposition and all landscape attributes.

      •   Integrated analysis, using deposition, soil chemistry, and  all landscape attributes except
          extent of soil sample classes.

      The results of Level I analyses for the NE and SBRP were generally consistent with the results of
similar regression analyses, and with our conceptual understandings of watershed attributes (and
implicit processes) influencing  surface water chemistry (e.g., Hunsaker et al., 1986; Church and
Turner, 1986; Jenne et al., 1989). Deposition, soil physical and chemical properties, watershed
physical attributes, and anthropogenic factors were all correlated with one or more major surface water
chemistry variables.  Table 8-1  summarizes  results for a small subset of the analyses for the NE and
SBRP, namely those groups of attributes thought to most strongly influence surface water chemistry
and/or that exhibited the highest correlations. At regional scales, other groups of attributes (i.e.,
geomorphic and hydrologic variables, bedrock geology, and depth to bedrock) were less  effective
than the attributes included in Table 8-1 in accounting for variability in surface water chemistry. For
example, the R2 values found in analyses using  bedrock geology were not high, even though bedrock
is the ultimate source of base cations and alkalinity in watersheds, and is widely regarded as one of
the most important factors affecting watershed sensitivity to acidic deposition. This result  is not
surprising in light of the large measurement uncertainties in the bedrock lithology data. The poor
observed relationship with geology is a result, presumably, of several factors:  (1) only surface waters
with low ANC were included in the analyses, so  the range of values for the dependent  variables most
affected by weathering rates (pH, ANC, and sum of base cations) is relatively small, (2) state-scale
geologic maps were used, thus uncertainty  in the position of map unit boundaries was high and
detailed information on spatially heterogeneous  lithologies was not reliable at watershed scale.
Similarly, values of R2 for other analyses using groups of attributes with large measurement error, such
as atmospheric deposition, are also likely to be attenuated.

      Sulfate concentrations  in the NE most clearly show the direct  effects of acidic deposition on
surface water chemistry. Net annual retention of atmospherically deposited sulfur in most NE
watersheds is low and surface water sulfate concentrations for lake  systems in the DDRP sample (i.e.,
with ANC < 400 /teq/L) are highly correlated with rates of sulfur deposition. Sulfur deposition is the
most important predictive variable for surface water sulfate  concentrations in watersheds in the DDRP
                                              154
-------
Table 8-1.        Summary of Level I Regression Analyses for DDRP Watersheds in the NE and
                  SBRP, Listing Values of R2 for Analyses Using Different Groups of Watershed
                  Attributes8
Attributes in
Region Regression
NE Deposition
Depn., land use
Depn., soil sample
class0
Dependent Variable
[S04=]
0.38
0.50
0.57 (141)
Percent S
Retention
nsb
0.19
0.41 (140)
ANC
0.16 (145)
0.37
0.51 (142)
PH
ns
0.30
0.46
SBase Cations
ns
0.42
(142) 0.52
        Depn., all mapped
          attributes'1
        Soil chemistry
        Depn., soil  chemistry,
          mapped attributes
          except sample class6
                                 0.70 (141)    0.44 (129)    0.47 (138)    0.44 (138)   0.57(142)
                                 0.47
                                 0.64
0.34
0.33
0.61
0.50
0.68
 SBRP Deposition               ns            ns            ns            ns          ns
        Depn., land use          ns  (30)       ns (30)        0.21(26)     0.12        0.35(26)
        Depn., soil sample        0.52 (30)      0.41           0.38          0.23 (28)   0.92 (30)
          class
        Depn., all  mapped        0.77          0.52 (30)      0.42 (32)     0.35 (32)   0.89 (25)
          attributes
        Soil chemistry            0.32          -             0.29
        Depn., soil chemistry,     0.73          0.44          0.82          0.65        0.85
          mapped attributes
          except sample class


a   For most analyses 143 (NE) or 31 (SBRP) watersheds were used in regressions; the number of watersheds varies among
    analyses due to deletion of outliers and/or influence points and is indicated if different from 143 or 31.

b   No variables significant in regression equation.

0   Based on sample class  area] extent (percent of watershed area).

d   Independent variables include deposition, geomorphic and derived hydrologic variables, mapped geology (areal extent),
    land use, vegetation, sample class areal extent, and depth to bedrock.

                             :, except sample class extent dropped and soil chemistry (aggregated to watershed   value)
6  Same independent variables as
   added
                                                 155
-------
NE sample. Variables that characterize reducing environments (e.g., % wetlands, extent of Histosols
and wet soils, presence of beaver ponds) are negatively correlated with suifate concentration and are
positively correlated with percent sulfur retention.  Attributes associated with watershed disturbance
(e.g., pits/quarries, urbanization, agricultural activity) are linked to higher suifate concentrations.  In the
SBRP, current suifate concentrations in runoff are controlled principally by adsorption in soils (Church
et al., 1989, Section 9,10), with the result that deposition does not directly control surface water
suifate and is not significantly correlated with stream suifate concentrations. In the SBRP, suifate
concentrations and retention  are most  closely linked to those variables that characterize suifate
adsorption by soils. These include the extent of high- or low-adsorbing sample classes, soil mass,
and soil chemical variables directly or indirectly linked to suifate adsorption.


      More generally, we can summarize the major results of the Level I analyses for the DDRP
sample of watersheds in the NE and SBRP as follows:


      •  Major watershed disturbances, such as quarries and urbanization, are associated not only
         with e!evated suifate concentrations, but also with increases in base cation concentrations
         and ANC.

      •  In both the NE and SBRP, agricultural land use and/or the extent of "open dry" vegetation
         (i.e., nonforested, undeveloped land other than wetlands, a classification that encompasses
         most agricultural land) are associated with high pH, ANC, and base cations and with ele-
         vated suifate concentrations.  The high pH, ANC, and base cations could reflect either the
         preferential use of fertile soils for agricultural activities or anthropogenic amendments, such
         as lime, to those soils.

      •  Surface water chemistry is related to soil depth and particle size in a predictable manner;
         suifate retention in SBRP soils and high  base cation output are associated with deep, rela-
         tively fine-textured soils, whereas high suifate concentration (low retention) and low base
         cations and ANC are linked to predominance of shallow and/or coarse-textured soils on
         watersheds.

      •  Regression analyses using only soil chemistry data (aggregated to watershed values)
         resulted in values of R2 substantially lower than regressions based on combinations of
         deposition and landscape variables and also lower than regressions for soil chemistry data
         used in combination with deposition and with mapped watershed attributes. Soil chemistry
         variables associated with base cation concentration and pool size (pH, CEC, % base satura-
         tion) are positively related  to pH, ANC, and base status of surface waters; suifate adsorption
         variables for the soil (suifate pools, isotherm  coefficients) are positively related to surface
         water suifate concentration.

      •  The soil sampling classes  developed for the  NE and SBRP, which were delineated to. reflect
         variability in  soil chemistry, drainage, texture, thickness, and parent material, are effective for
         describing landscape units. Based on the observation that regression analyses using just
         sample classes and deposition resulted in values of R2 comparable to those developed in
         integrated analyses using  all landscape variables, we conclude that the sample classes
         capture the diversity of both chemical and landscape attributes.  The values of R2 for these
         analyses often were higher than those resulting from analyses based solely on aggregated
         soil chemistry data and (in some cases), were comparable  to the even higher values of R2
        for analyses using deposition, soil chemistry, and all mapped attributes (except sample
         class).
                                             156
-------
Results of a parallel set of regression analyses, conducted using data for riparian buffer
zones, were inconsistent.  In some cases, regressions using attributes within the buffer zone
resulted in values of R2 higher than those for the entire watershed, but in other cases values
of R2 for the buffer were comparable to or lower than those for the entire catchment.  The
inconsistent results for analyses using riparian buffers may be caused by measurement error
(i.e., the spatial resolution and minimum area of map unit delineations).

Overall, the results were consistent with current hypotheses regarding controls on surface
water chemistry and with the processes and relationships incorporated into watershed
chemistry models. Results do not suggest any major omissions or inconsistencies with the
processes incorporated in watershed models used in our Level II and III analyses.
                                     157
-------
-------
                                         SECTION 9
             LEVEL II ANALYSES - SINGLE FACTOR RESPONSE TIME ESTIMATES
9.1 INTRODUCTION

     Although a number of watershed processes influence surface water chemistry (e.g., see
summary reviews in Church and Turner, 1986; Sections 2 and 3 of Church et al., 1989), changes
related to a handful of processes probably represent the major controls on long-term changes in
watershed response to sulfur deposition. The MAS Panel on  Processes of Lake Acidification (NAS,
1984) focused on sulfate adsorption and base cation exchange by soils as critical time-variant
processes that might contribute to a delayed response to acidic deposition.  In its focus on sulfate
adsorption and cation exchange, the NAS panel recognized scientific uncertainties concerning the
present and potential long-term roles of the two processes. Mineral weathering and cation uptake by
vegetation are examples of two rate-limited processes; the magnitudes of which are likely to change
slowly, if at all, over the 50-year period of the DDRP projections. Adsorption and exchange, however,
are capacity-limited processes. As sulfate adsorption sites become occupied or as exchangeable
cations are leached from the soil, the buffering capacities of these two processes decrease monotonic-
ally, increasing the probability of  surface water acidification.  The time frame of such changes can vary
widely as a function of soil physical and chemical properties.  In watersheds with thin or very coarse-
textured soils, buffering of acidic  deposition  by adsorption or exchange can  be very limited and  some
systems can respond almost immediately.  Alternatively, watersheds with deep soils and high adsorp-
tion capacities and/or large exchangeable base cation pools might experience significant changes in
soil leachate chemistry in response to  high acidic deposition loadings only after decades  to centuries.

      This section  presents results of Level II analyses, which involve simulations of the temporal
response of individual watershed processes considered in isolation. We have limited analyses in this
report to an evaluation of the role of sulfate  adsorption and cation exchange as mechanisms contrib-
uting to delays in surface water acidification for watersheds in the M-APP Region and, for comparison,
a summary of results for the NE  and SBRP previously reported in Church et al. (1989). Analyses are
based on models that  consider only adsorption or exchange, and that consider only changes within
the upper regolith  {< 2 m for the M-APP and < 1.5 and 2.0 m in the NE and SBRP, respectively).
These analyses assess the influence of adsorption and exchange on soil and/or surface water chemis-
try at the present time, and project probable future changes in the contributions of adsorption and
 exchange. The analyses use soil chemistry data collected during the DDRP soil surveys  and model-
 based projections of (1) future changes in sulfate mobility controlled by sulfate adsorption, and  (2) two
 independent projections of exchange-mediated changes in base cation leaching and changes in soil
 pH and/or cation exchange pools. By considering cation exchange but not mineral weathering  (i.e.,
 base cation resupply), the exchange models provide a worst-case projection of future changes  in soil
 and solution chemistry.

       The analyses presented in this section evaluate the temporal response of single processes to
 acidic deposition.   Because they assess those processes in  only a portion  of the watershed  (i.e., the
 upper 1.5 - 2.0 m  of watershed soils), the model projections should not be interpreted as integrated

                                              158
-------
 projections of watershed response time. Rather, these analyses represent a set of bounding estimates
 of the relative importance, now and in the future, of the potential role of adsorption and exchange as
 delay mechanisms. Analyses presented by Church et al. (1989, Section 7) showed that processes
 other than adsorption are of limited importance in mediating sulfate mobility in most NE and SBRP
 watersheds, especially in upland catchments such as those considered in this report for the M-APP
 Region. Similarly, although several processes other than exchange and weathering affect cation
 cycling in terrestrial ecosystems, exchange is generally regarded as the rate- or capacity-limited
 process the dynamics of which are most likely to be significantly affected by acidic deposition.  Level
 111 modelling (Section 10) provides projections  of changes in future surface water chemistry based on
 integration of adsorption and exchange with other processes.

 9.2 EFFECTS OF  SULFATE ADSORPTION ON WATERSHED SULFUR RESPONSE TIME

 9.2.1  Introduction

      As noted in Section 9.1 and as discussed in previous DDRP results for the NE and SBRP
 (Church et al., 1989), a major DDRP focus is on sulfate, which is the principal anion in acidic deposi-
 tion and the major  mobile anion contributing to surface water acidification in the eastern United States.
 Because the extent and duration of sulfate retention within watersheds varies widely within and among
 regions, depending on deposition history and on soil physical and chemical properties,  variability in
 sulfate retention by soils represents one of the most important factors influencing the rate of watershed
 chemical response  (i.e., changes in ANC) to acidic deposition (Johnson and Cole, 1980; Galloway et
 al., 1983; NAS, 1984).

      At the start of the DDRP, there was believed to be a strong north-south gradient in sulfate reten-
 tion characteristics  of soils.  Soils in the glaciated northern areas of North America were perceived to
 have low sulfate adsorption capacity, resulting  in negligible sulfate retention by soils, and watershed
 sulfur budgets at or near steady state.  In contrast, watersheds in the southeastern United States were
 characterized by high net sulfur retention, which was attributed to the moderate to high  sulfate adsorp-
 tion capacities of deep, highly weathered soils  in the region (NAS, 1984).  Little was known, however,
 about either adsorption characteristics of soils or watershed sulfur budgets for the M-APP Region.  Soil
 analyses conducted as part of the DDRP in the NE and SBRP demonstrated strong regional differ-
 ences in sulfate adsorption characteristics that  are consistent with this paradigm. Similarly, site-
 specific and regional analyses of watershed sulfur budgets (Rochelle et al., 1987; Rochelle and
 Church, 1987; Church et al., 1989)  have confirmed north-south differences in regional sulfur budgets,
 and in addition have shown a range of retention in M-APP watersheds (Section 7).

      Along with budget data showing a range  of sulfate retention by M-APP watersheds, there are
 scattered data describing sulfate dynamics for watersheds within the region. Monitoring data for
 streams at Fernow,  West Virginia (D. Helvey, pens, comm.), and at White Oak Run and Deep Run,
 Virginia (Ryan et al., 1989), document increases in streamwater sulfate concentrations of about 1-2
/teq/L/yr. Modelling of the White Oak Run catchment projects that sulfate concentration  will increase
 from levels in the mid-1980s of around 80 ,«eq/L to over 200 fieq/L during the next 50-60 years (Cosby
 et al., 1985b,c 1986b). The regional representativeness of these few sites is unknown; the modelling

                                             159
-------
activities described in this section provide the first regional assessment of surface water sulfate
dynamics for the M-APP Region.

9.2.2  Section Objectives

      This section assesses changes in sulfate mobility in DDRP watersheds and regions that are
attributable to sulfate adsorption (and desorption) by soils.  The importance of processes other than
adsorption on sulfate mobility has been previously assessed for watersheds in the NE and SBRP, and
these other processes were found to be of little importance in most DDRP watersheds. The overall
goal of the Level II analysis is to assess the importance of sulfate adsorption as a process contributing
to delays in surface water acidification within study regions.  Specific objectives of Level II sulfate
analyses for the M-APP Region, presented here, are identical to those for the previous NE and SBRP
analyses:

      •  Characterize and compare sulfate pools and  sulfate adsorption capacities of soils within and
         among study regions.

      •  Estimate the response time of soils in DDRP watersheds to changes in  sulfur deposition
         using an adsorption-based model.

      •  For watersheds not presently at steady state, forecast time to steady state under current
         deposition loadings. In addition, for all watersheds, forecast response time to future
         changes in deposition. Results for individual watersheds will be extrapolated to regional
         forecasts.

      •  Summarize the contributions of sulfate adsorption to delays in surface water acidification
         resulting from historic or future changes in deposition.

The results of these analyses will also provide data for evaluating and comparing the relative
importance of sorption and other processes considered by the DDRP models (e.g., cation exchange),
although comparisons are not made within this section.

      It is important to recognize that procedures and models used for this analysis consider only
sorption processes.  Processes affecting watershed chemistry, other than those involving sulfate, are
not considered, and watershed  conditions and fluxes (except sulfur deposition) have been held con-
stant  for the duration of model projections.  It is equally important to recognize that the forecasts and
estimates of time to steady state made here apply only to sulfate.  Although change in sulfate mobility
is one of the principal factors driving change in surface water cations and ANC, processes that don't
involve sulfur also play crucial roles in the dynamics of those ions.  Rates of change in ANC, and
especially potential time to zero ANC, are not necessarily coincident with time to steady state for
sulfur. Systems can reach zero ANC before, concurrently with, or after time to sulfur steady state.
The relationship between changes in sulfate and ANC are characterized and discussed as part of the
Level 111 analyses in Section 10.

      Although  the model projections included in this discussion extend beyond 50 years, attention
should be focused on response over the first 50 years  of simulations.  Projected changes beyond the
                                              160
-------
50-year period have been included for several reasons.  First, in many systems, sulfate response is
projected to occur over time frames exceeding 50 years, in some cases for a period likely to last for
well over a century. Second, including long-term projections indicates the probable magnitude of
changes occurring after the 50-year time frame, until steady state is reached.  Also, for the scenarios
of altered deposition, simulated changes in deposition continue for much  or all of the 50-year
response period. It is difficult, if not impossible, to effectively assess the response to those changes
by considering only a 50-year, rather than a long-term, system response.

9.2.3 Approach

      Level II sulfate analyses are based on model-based projections of future sulfate dynamics in 36
DDRP watersheds within the M-APP Region, with results extrapolated to a regional target population.
Procedures and  data sources are comparable to those used for similar DDRP sulfate analyses for the
NE and SBRP. Soil mapping and chemistry data generated by the DDRP Soil Surveys (Section 5.4,
5.5) were used to make the projections. The principal soil variable for these analyses was sulfate
adsorption isotherms generated for individual soils collected in the surveys, then aggregated to the
watershed level. Forecasts were made using a modified version  of the sulfate subroutine in the Model
of Acidification of Groundwater in Catchments (MAGIC) (Cosby et al., 1985a, 1986b). A detailed
description of the model, data sources, underlying assumptions,  etc., was presented in Church et al.
(1989); key points are repeated below.

9.2.3.1  Model Description

      AH model projections described in this section were made using a modified version of the sulfate
subroutine from the MAGIC model, which was originally developed for stream systems in the Virginia
Blue Ridge (Cosby et al., 1985a,b; 1986b). The model uses a  deterministic, mass-balance approach
that considers only adsorption as a sulfur retention process by soils (Cosby et al.,  1985b,c; 1986b).
Sulfate partitioning between dissolved and  sorbed phases is defined by a hyperbolic (Langmuir) iso-
therm. The original MAGIC subroutine has been modified to accommodate multiple soil horizons (up
to 10, although 3 were used for this study). Soil horizons are treated as a series of continuously
stirred tank reactors (CSTRs); all precipitation and sulfur deposition (wet and dry) inputs are made to
the top mineral soil horizon (organic horizons are not considered in the model, because sorption is
negligible in the  O horizon). Evapotranspiration implicitly occurs  in the top soil horizon.  All flow is
then routed sequentially through each  soil  horizon.  Data are input to the model using annual time
steps. The projected surface water sulfate concentration is defined by (set equal to) the equilibrated
solution sulfate concentration  in the lowest soil horizon.  Because sorption is essentially  an instan-
taneous process, reaction kinetics are not considered and equilibrium between solution  and sorbed
phases is assumed to occur in all cases.

      For these analyses, model simulations were run starting 140 years prior to the base year of 1986
[i.e., the year of the EPA National Stream Survey - Phase I (NSS-I), which included Mid-Appalachian
streams (Kaufmann  et al., 1988)].  Comparison data for NE lakes (Linthurst et al., 1986) and SBRP
streams (Messer et al., 1986)  employ base years of 1984 and  1985,  respectively.  Soil and stream-
water sulfate concentrations were set at the start of simulations by assuming both to be at steady state

                                              161
-------
 with respect to deposition (low historic levels). Simulations were run for a sufficient time into the future
 to allow systems to reach steady state.  For most M-APP watersheds, simulations were run for 140
 years; in a few cases with long response times, simulations were extended to 300 years.

      in calibrating the model for M-APP systems, as described in Section 9.2.3.5, simulated response
 was constrained to compare favorably with two kinds of surface water data.  First, the distribution of
 simulated sulfate concentrations (and of percent sulfur retention) for the base year had to compare
 closely to the observed distribution of streamwater sulfate.  Secondly, because many trajectories of
 simulated response can match current sulfate concentration, we also required simulated rates of
 change for sulfate to be comparable to observed rates for streams monitored within the region.  We
 present and discuss these comparisons in Section 9.2.4.3.

      As in previous analyses for the NE and SBRP (Church et al., 1989), we define response times as
 times for projected sulfate concentrations to reach 95 percent (or for systems with decreasing sulfate
 concentration, 105 percent) of steady state. We use this approach for two reasons. First, model
 projections approach  steady-state concentrations asymptotically, such that true steady state is not
 reached. Second, and more important from a practical perspective, given the measurement uncertain-
 ties and interannual variability in precipitation, runoff, deposition, and surface water sulfate
 concentration, the difference between 95 and 100 percent of steady-state concentration is not
 distinguishable.

 9.2.3.2 Data Sources

       Input requirements for the sulfur model include current sulfur inputs and outputs  (precipitation,
 runoff, total sulfur deposition, and sulfate concentration in runoff), scenarios  of historic and future
 sulfur deposition, and soil variables to describe sulfate partitioning and adsorption capacity (adsorption
 isotherms, soil mass). Data requirements and sources are comparable-to those for similar DDRP
. sulfate modelling in the NE and SBRP.

       The  precipitation and deposition  datasets used here are analogous to the typical year datasets
 generated  for DDRP analyses in the NE and SBRP; generation of these datasets is described in Sec-
 tion 5.6.3.  Runoff estimates, based on interpolation of 30-year average USGS  runoff maps, were
 generated  as described in Section 5.7.1. Data for current streamwater sulfate  concentrations are
 taken from the results of EPA's NSS-I in the Mid-Atlantic and southeastern United States  (Kaufmann et
 al., 1988),  conducted as part of the EPA's National Surface Water Survey.

        Historic sulfate deposition in the region is unknown; initial sulfate inputs  (year -140) were
  estimated  as 5 percent of current deposition. We based estimated sulfur deposition between initial
  and base years (1846-1986) on SO2 emission estimates of Gschwandtner et al. (1985)  for Federal
  Region III  (Delaware, Maryland, Pennsylvania, Virginia, West Virginia); we used  linear interpolation
  between the initial simulation year and 1900. The historic emission pattern served as a scaling  factor
  for each watershed, a procedure that implicitly assumes a constant relationship between regional
  emissions and site-specific deposition  over the last 140 years.
                                                162
-------
       Three scenarios of future sulfur deposition were used for the M-APP Region.  The first scenario
 (A) calls for constant deposition through the 140-year simulation period (Figure 9-1). The second and
 third scenarios are characterized by reduced deposition in the future.  Both start with constant deposi-
 tion at current rates for years 1 to 5, followed by a linear reduction in deposition, to 50 percent of
 current loadings, between years 6 and 15.  For the second scenario (B), future deposition is main-
 tained at 50 percent of current loadings from year 15 through the end of the simulation period.  The
 third scenario (C) allows a linear increase in deposition, from 50 to 80 percent of current loadings
 between years 15 and 50, followed by constant deposition at 80  percent of current input from year 50
 to the end of the simulation period. All comparisons of regional projections among the three DDRP
 regions are based on scenarios of constant future deposition.

       Mapping of soils and quantification of the areal extent of various soils  on DDRP watersheds are
 described in Section 5.4.  Sampling and chemical/physical analyses of soils are described in
 Section 5.5. For each mineral soil horizon, sulfate adsorption data were used to compute adsorption
 isotherms, which were then aggregated, using mass  weighting (computed from horizon thickness, bulk
 density, and coarse fragment content) to obtain sample  class and watershed values. Procedures for
 derivation of adsorption isotherms and for aggregation of adsorption data are identical  to those used
 for previous DDRP analyses in the NE and SBRP, which are described by Shaffer and Stevens (1991)
 and Section 9.2.  of Church et al. (1989).

      During the  process  of sample class development and sampling site selection for the Mid-
 Appalachian Region, we recognized that in those sample classes that occurred across the entire
 region, the potential existed for significant within-class differences in soil chemistry related to
 geography (e.g.,  glaciated parts of the northeastern Pennsylvania Plateau, and Valley and Ridge
 physiographic provinces) and/or to gradients in current and historic rates of acidic deposition. If such
 differences exist,  but were obscured by a forced combination of data during  aggregation, the quality of
 subsequent model simulations might be compromised.  As part of M-APP  analyses, we have analyzed
 spatial trends for soil sulfur data.  Results indicate that, when data for all M-APP pedons are consid-
 ered, there are significant  differences in the spatial distribution of relevant sulfur variables (e.g.,
 extractable sulfate, adsorption isotherm slope), but they result principally from differences in the spatial
 distribution of sampling classes within the region.  With few exceptions, within-class spatial patterns
 are not significant The single exception, sample class END, has  low values for adsorption isotherm
 slope and SO4 -PO4, in the Valley and Ridge province, but covers no more than  20 percent of any
 watershed in that subregion.  If data were subdivided and reaggregated to watershed scale, for
 sample classes that are geographically widespread, there would be only very minor differences in the
 reaggregated data.  Available results indicate that the current sample class structure and aggregation
 are reasonable arid are well-suited to our regional  analyses.
9.2.3.3  Model Assumptions and Limitations

      Several critical assumptions are encompassed by the choice of model and the methods of data
collection. These in turn impose limitations on the scope and interpretation of model projections.  Key
model assumptions and their implications for data interpretation include:

                                              163
-------
      FUTURE SULFUR DEPOSITION SCENARIOS
                DDRP MID-APPALACHIAN REGION

X
D
LL
nr
ID
LL
_l
=)
CO
LU
h-
_J
LU
DC


120-

oo -

80 -
.

RO-

40-

20-


-


\
\
\
* s — — — — —
\
,'
\

*
\ s

	 Scenario A
	 Scenario B
	 Scenario C

I 	 1 	 i • i • i • i • i
        0
25     50     75     100    125
   YEAR OF SIMULATION
150
Figure 9-1.  Deposition scenarios used for modelling future sulfate dynamics in soils of
        DDRP M-APP watersheds.
                          164
-------
      •  Sorption is the only watershed "process affecting sulfate mobility and watershed response -
         time. As noted previously, this limitation provides the most effective means of assessing the
         significance of adsorption by soils as a process delaying surface water acidification.  To the
         extent that other terrestrial processes sequester or generate sulfate on a net basis, model
         projections will under- or overestimate the.time and/or magnitude of the projected response.
         As noted earlier, the net role of other processes in most DDRP watersheds is believed to be
         small, especially in upland systems that lack extensive wetlands or flooded soils.

      •  The  analytical approach used to define sulfate partitioning in the soil (hyperbolic isotherms
         determined by batch equilibrium methods using air-dried soils) adequately describes sulfate
         partitioning by soils under field conditions. A methods comparison  study by Hayden (1987)
         supports the use of hyperbolic isotherms and batch equilibrium methods. Recent analysis of
         soil handling effects (Comfort et al., 1991) indicates that the effects  of air drying on
         adsorption increase in proportion with organic content. Soils with low organic content were
         not significantly affected by drying. Because most adsorption occurs in soils with low
         organic content (organics block anion adsorption), the practical importance of drying effects
         appears to be negligible.

      •  Soil  and watershed conditions influencing adsorption (e.g., soil pH,  Fe, Al, and organic
         content) are assumed to be constant for the duration of model projections. pH is probably
         the most important variable of concern, since adsorption is strongly pH dependent.  Level II
         cation exchange models project moderate decreases in soil pH during the simulation period,
         which would  be expected to lead to modest  increases in sulfate adsorption capacity. The
         quantitative relationship between soil  pH and adsorption is not well  defined (and thus is not
         incorporated  into the model), but large effects on model projections would not be expected.


      •  Hydrologic routing is simple; flow through the soil is represented by a series of continuously
         stirred  tank reactors (CSTRs), and all flow is  routed sequentially through each horizon.  The
        * perfect hydrologic contact represented by a  simplified flowpath used here does not
         realistically reflect lateral flow, macropore flow, etc., that occur in the soil.  Data needed to
         objectively define routing coefficients  for alternate flowpaths are lacking, however, so
         alternate routing was not incorporated into model projections. The effect of modelling flow
         to bypass upper or lower soil horizons under natural conditions would be projections of
         higher initial sulfate leaching (part of the input signal would not be attenuated by sorption on
         the soit), but  a more gradual (in terms of  change in concentration with time) subsequent
         sulfate response. The responses projected here represent an upper bound on initial
         response time, assuming complete contact between the soil and flow through the soil, and a
         lower bound  on time to steady state.

      •  Because the  model  runs on  an annual time step and uses identical  precipitation, runoff, and
         sulfur deposition data from year to year, projections do not capture  the effects of interannual
         variability of natural systems. The lack of such variability in the DDRP projections is
         recognized, but should have little effect on the primary objective of these analyses, i.e. to
         project long-term changes attributable to  chronic sulfur deposition.  If there were any long-
         term trends in precipitation or runoff, they would not, of course, be represented  by model
         projections.


9.2.3.4  Adsorption Data


      Data describing sulfate partitioning by soils, which are used to develop  partitioning functions
(isotherms) of  sulfate adsorption capacity of soils, were generated for individual soils as part of the
DDRP soil survey (Section 5.5). Adsorption isotherms  were developed for each soil, then  aggregated
                                              165
-------
to sample class and to watershed values using mass weighting procedures. Johnson et al (1990)
have described the conceptual objectives of aggregation for DDRP soil chemistry data; actual
aggregation procedures are described by Shaffer and Stevens (1991) and summarized following the
discussion of adsorption isotherms immediately below.

      Adsorption isotherms were generated from data for soil-water slurries equilibrated with six
different amounts of sulfate (0, 2, 4, 8, 16, and 32 mg S/L) designated as SO4_0, SO4_2, etc., in
Section 5.5.4.2.1. For each of the six samples, net sulfate adsorbed by the soil was computed from
the change in dissolved sulfate.  For example, for the 8 mg/L sample:
                      SO4_8n = (SO4_8, - SO4_8f)
                                          L
                                          S
                                   (9-1)
where:  SO4_8j
        SO4_8f
        L
        S
= dissolved sulfate concentration before equilibration (^eq/L)
= final dissolved sulfate concentration after equilibration (aeq/L)
  = volume of liquid (~ 0.050 L)
  = mass of volume of soil (~ 0.010 kg)
Raw data points were then fit to an extended Langmuir isotherm ("extended" by addition of a third
variable to describe the non-zero Y-intercept) (Hayden, 1987), the  partitioning equation taking the
form:
                                  Ec =
                           B1 *  C
                           B  +  C
+ B,
(9-2)
where:  B1 = maximum sulfate adsorption (meq/kg)
        B2 = half saturation constant (aeq/L)
        B3 = Y-intercept (meq/kg)
        C  = dissolved sulfate (weq/L)
        Ec = net adsorbed sulfate at [SO-/'] = C (meq/kg)

The three parameters (B^ B2, and B3) were estimated by a nonlinear least squares procedure
(Fletcher and Powell, 1963; Hayden, 1987) to minimize the sum of squares function. Examples of raw
data and resulting fitted isotherms are shown by Church et al. (1989, their Figure 9-1), and Shaffer and
Stevens (1991, their Figure 1).

      Several approaches were evaluated for aggregating isotherm data from individual soils; the
"aggregate isotherm" approach of Shaffer and Stevens (1991) was selected because it has been
shown to generate effective weighted averages for these nonlinear functions. This procedure involves
three steps as follows: (1) after fitting isotherms for individual soils, compute concentrations of net
adsorbed sulfate corresponding to several concentrations of dissolved sulfate (0, 10, 25, 40, 75, 125,
200, 500, 1000, 2000 ^eq/L) for each soil, (2) for each value of dissolved sulfate, compute a mass
weighted average of the corresponding concentrations of net adsorbed sulfate for the soils in each
aggregation  group, then (3) fit a new isotherm to the set of weighted averages.  The resulting isotherm
                                              166
-------
 was defined as 4he aggregate isotherm and was used to describe sulfate partitioning for that group of
 soils.  For routine applications, isotherm aggregation data were aggregated to the master soil horizon
 (A/E, B, or C), using a three step procedure: (1) individual soil (sub)horizons to master horizon within a
 pedon (mass weighting), (2) pedon (by horizon) to sample class (by horizon), using mass weighting,
 and (3) sample class to watershed (by horizon), using mass and area weighting of each sample class
 occurring on each watershed.  Missing data were assigned  the aggregate value for other subhorizons
 In their respective pedon; for soils having only one subhorizon within a master horizon, the sample
 class average for that horizon was assigned to missing values. For routine uses, data for soil master
 horizons were  used directly in the model and were not further aggregated.

      Because the aggregation approach was not conducive to direct computation of parameter
 uncertainty (i.e., computing a mean and standard deviation  for coefficients), we used an alternate
 approach. Uncertainties for the original isotherm fits were retained, then a Monte Carlo procedure was
 used during each step of aggregation to generate estimates of uncertainly in aggregated coefficients
 at the sample class and watershed level. Uncertainty in sulfate isotherms was propagated through the
 aggregation procedure using the Monte Carlo technique described in Section 6.3, with  steps similar to
 those used for aggregation of other variables.  We repeated the aggregation from individual sub-
 horizons to sample class master horizon 100 times, each time selecting a randomly perturbed set of
 coefficients for each subhorizon isotherm.  The perturbation  of B1 was selected first from a normal
 distribution with a standard deviation obtained from the residual sum of squares and the inverse
 Hessian matrix from the initial nonlinear least squares fit. The perturbed value of B^ along with the
 correlation of B1 and B2 from the inverse Hessian  matrix, was then used to estimate the conditional
 expectation of B? given Br This conditional value was then  perturbed by a value drawn from a normal
 distribution with the conditional standard deviation of B2 given Br  A similar procedure was used to
 perturb B3, except that the  mean  and variance were adjusted for both B1 and B2. The mean values,
 standard deviations, and correlation matrix of the coefficients were summarized at the sample class
 level. These values were then passed to the watershed level aggregation algorithm. The  uncertainty
 calculation for watershed aggregation was conducted using  the same procedure as that described
 immediately above (this paragraph) for sample class uncertainty estimates, except that the correlations
 were derived from the sample class Monte Carlo study rather than  from  an inverse Hessian matrix.

 9.2.3.5 Calibration and Evaluation of Aggregated Data

     Three approaches were used to evaluate aggregated soil sulfate data and to calibrate data for
 model  simulations; all used isotherm data aggregated to watershed level. First, equilibrium soil
 solution sulfate concentrations (the dissolved sulfate concentration  at which net adsorbed sulfate
 equals zero; see Equation 9-2) for B horizon isotherms, aggregated to watershed values, were com-
 pared to streamwater sulfate concentrations. If adsorption by the soil were the sole process influencing
 sulfate mobility, if lab isotherms precisely represented partitioning under field conditions, and if
 aggregation procedures were perfect, a 1:1  correlation between soil and lake/stream sulfate concentra-
tions would be obtained.  Realistically, due to contributions of factors such as hydrologic routing,
 heterogeneity of natural soils,  uncertainties introduced by soil sampling and analysis, and effects of
 regionally focused (rather than watershed by watershed) sampling and aggregation, a high correlation
 between soil solution and surface water sulfate was not expected.   The purpose of this comparison

                                             167
-------
was to evaluate whether the two sets of values had comparable ranges and whether major biases exist
that might invalidate the entire approach.

      The second and third approaches to evaluation of isotherm data use model simulation results,
and compare model-based data to measured stream chemistry. Isotherm and soil data were cali-
brated by comparing projected sulfur retention in the base year to the percent retention computed as
described in Section 7.  Simulations were run starting 140 years before the base year, and model
simulations were not initialized in any way that constrained year 0 projections to match measured
chemistry in the base year.  The comparability (or lack thereof) between measured and observed
sulfate retention for the base year thus provides a check point for simulations; isotherm data were
calibrated by scaling coefficients (B2, or B2 and  B1 together if necessary) such that the mean (for all
watersheds) percent sulfate retention for measured and simulated sulfate were approximately equal.
Calibration was accomplished by scaling the coefficients for all watersheds by the same amount,
resulting in simulations that are unbiased for the population but not matched on a watershed-by-
watershed basis.

      The final check on isotherm data was to compare measured and modelled rates of change in
streamwater sulfate. Rates determined from monitoring studies were compiled and compared to rates
generated from model simulations.  Because field data were available for only a few field sites, this
comparison provides only a rough qualitative check, but it does insure that the trajectory of
simulations (i.e., the rate of increase for sulfate with time) is not unrealistic.

9.2.3.6  Target Populations  for Model  Projections

      Because the modelling watersheds used for DDRP were selected within the probabilistic
framework of EPA's National  Surface Water Survey, results from the sample watersheds can be
extrapolated to an explicitly defined regional target population (Linthurst et al., 1986; Kaufmann et al.,
1988). Except as noted, all results presented for the M-APP Region, and all comparison results from
previous NE and SBRP analyses, represent lake or stream target populations within their respective
regions.  In the M-APP Region, 36 watersheds were modelled within the DDRP, representing a popula-
tion of 5,496 stream reaches within the region (Kaufmann et al., 1988). This population represents a
subsample of the National Stream  Survey watersheds having (1) ANC of <200 fieq/L, (2) sulfur
budgets for which sulfur output was not significantly higher than input (i.e., no major internal
watershed sulfur sources),  (3) catchment areas of 30 km2 or less, and (4)  no evidence of acid mine
drainage (as evidenced by streamwater chemistry, topographic maps, or visible disturbance, or any
combination of these).  Sample and target populations for NE lakes and for SBRP streams were
described in detail in Section 9.2 of Church et al. (1989). For the NE,  simulations were made for 131
lakes  representing  a target population of 3,314 lakes. Thirty-four stream systems were modelled in the
SBRP, representing a population of 1,492 stream reaches.
                                             168
-------
9.2.4 Results
9.2.4.1.  Summary of Soil Adsorption Data

      Sulfate adsorption data for soils in the M-APP are summarized in Table 9-1; for purposes of
comparison, data for the NE and SBRP are also included.  For most of the listed parameters, soil
concentrations or pools for M-APP soils are intermediate between those of the other two regions.
Mass-weighted average concentrations of phosphate-extractable sulfate for M-APP soils range between
25 and 57 mg S/kg; for each horizon, the concentrations are higher than for the equivalent NE soils
but lower than for corresponding SBRP soils. Adsorption isotherm coefficients [adsorption maxima
(B1) and half-saturation constant (B2)] for M-APP soils are also intermediate in value, although the
interdependence of these coefficients makes the relative differences  difficult to interpret.  Figure
5-10 indicates the variability of these data on a sample class and pedon basis.

      If we compare weighted average isotherm slopes (partitioning  coefficients between sorbed and
solution phase) instead of individual coefficients, and consider the B and C soil horizons in which most
adsorption  occurs, the average slopes for M-APP soils are two to four times higher (slopes of 6.8 and
7.35 versus 3.15 and 1.31) than for NE soils, but only about one half the values for soils in the SBRP.
Similar ratios are observed for the amount of sulfate that can be adsorbed per unit mass of soil at
steady state. The differences among the three regions are even more pronounced if we consider the
pools of adsorbed sulfate (at sulfur steady state) per unit watershed  area. Whereas  the average pool
in the M-APP is almost 5 times that of NE soils (17.1  kg/ha vs 3.7 kg/ha) the pool for SBRP soils (55.4
keq/ha) is over 3 times as large as the M-APP pool and about 15 times that of NE soils.  This relation-
ship occurs because average soil mass is greater in the SBRP than  in the M-APP, and the B horizon
mass is greater in the M-APP than in the NE.

      The differences among sulfate retention characteristics for soils in the three  regions reflect differ-
ences in  soil age, parent material, and other soil chemical properties. NE soils are much younger
than those  in the other regions, (around 14,000 years, versus millions of years) and thus are thinner.
They are characterized by lower amounts of inorganic hydrous oxides and clays than soils in the other
regions.  Furthermore, the NE soils have  higher concentrations of organic carbon, which interferes with
or "blocks" sulfate adsorption, than soils in the M-APP or the SBRP.  In comparison to SBRP soils,
adsorption  in the M-APP is lower for several reasons: (1) soils in the  M-APP are shallower; thus (all
other things being equal) they have less soil mass and surface area  than SBRP soils, (2) most soils in
the M-APP have developed from sedimentary bedrock, from which the readily weatherable primary
minerals  were lost during a previous weathering episode, and (3) the more slowly weathering sedi-
mentary rocks in the M-APP experience less dissolution of Fe and Al from aluminosilicate minerals,
providing less substrate for formation of secondary clay mineral phases and hydrous oxide coatings.

9.2.4.2.  Evaluation of Aggregated Data

      As  described in Section 9.2.3, several comparisons between measured  field variables and either
adsorption isotherms or model simulation data were used as checks on the reliability of model  inputs
and outputs.  The first such check compared streamwater sulfate concentrations to equilibrium soil

                                             169
-------
Table 9-1.  Summary Comparison of Adsorbed Sulfate and Adsorption Capacities of Soils on
           Watersheds in the DDRP Target Populations in the M-APP, NE, and SBRPa
Soil
Region Horizon
M-APP A
B
C

NE A
B
C

SBRP A
B
C

Soil
Mass
g/cm2
10.7
78.0
26.1

10.5
55.1
68.9

17.1
90.5
68.2

isotherm Coefficients13 Isotherm
SO4-PO4 B1 B2 slope©
mg/kg meq/kg /*eq/l_ lOO^eq/L
25.0
57.2
33.1

22.0
35.1
13.5

47.7
88.9
42.0

3.23 1788 1.71
5.25 577 6.80
4.82 457 7.35

3.12 1361 2.03
3.57 945 3.15
1.60 1058 1.31

4.38 1093 3.50
7.55 183 17.10
6.25 109 14.70

Adsorption at Sulfur
Steady State
meq/kg keq/ha
0.40
1.58
1.68
pedon 2 =
0.26
0.41
0.16
pedon 2 =
0.45
3.23
3.73
pedon 2 =
0.4
12.3
4.4
17.1
0.3
2.3
1.1
3.7
0.8
29.5
25.4
55.4
a Median Values are Listed.
b Coefficients used in equation 9-2.
                                           170
-------
 solution sulfate (or ESSS, which represents the concentration of dissolved sulfate at which net
 adsorbed sulfate equals zero) for B horizon soils, using soil data aggregated to watershed scale.
 Results of this analysis, which assesses whether measured isotherms predict sulfate concentrations
 comparable to observed streamwater sutfate concentrations, are shown in Figure 9-2.  A scatter plot of
 data shows that in comparison to the wide variability in measured streamwater chemistry (55-313
 fteq/L), values of ESSS fall within a narrow range, with all but two B horizon soils having values of
 ESSS between 131 and 210/teq/L Mean and median sulfate concentrations from the two data sets
 are similar, however.  Mean  and median concentrations for stream water are 152 and  154/teq/L,
 respectively, with corresponding values of ESSS  for B  horizon soils of 164 and 170 fieq/L. The rela-
 tively narrow range of values for ESSS is in all likelihood a result of using soil sampling and data
 aggregation procedures optimized for regional, rather than site-by-site, soil characterization and model
 projections. The regional approach, as discussed by Church et al. (1989), effectively characterizes
 central tendencies and broad regional variability  of watershed characteristics and surface waters, but
 was not intended to (and does not) fully reflect the range of intra-regional variability in  soil and surface
 water chemistry.  The most important result of this comparison is that although the range of values in
 the two data sets is different, mean and median values are similar, suggesting that there are not fun-
 damental problems or major systematic biases with  the isotherm data or with aggregation procedures.
 Recall also that the relative heterogeneity of watershed-scale values of ESSS cannot be attributed to
 aggregation of geographically diverse data for soils within sampling classes. Analysis  of variance for
 data within sample class do not indicate systematic  differences in soil sulfate between geographic/
 physiographic subregions or across deposition gradients.

      A second data check,  comparing measured stream sulfate to modelled stream sulfate for the
 base year of 1986, indicates similar ranges and distributions of sulfate concentrations in the two data
 sets.  Mean measured and modelled sulfate concentrations are 152 and Isy^eq/L respectively; corre-
 sponding mean values of percent sulfur retention are 35 and 34 percent. Site-by-site comparison of
 the data (Figure 9-3, A and B) indicates a great deal of scatter. As already noted, however, this result
 is not especially surprising given that sampling and data aggregation were designed to characterize
 regional, rather than site-specific, characteristics of soils and watersheds. When data are presented to
 compare populations, using  cumulative distributions (Figure 9-3 C and D), the distributions of meas-
 ured and modelled sulfate, for both sulfate concentration and percent sulfur retention,  are very similar.
 In contrast to previous modelling results for the NE and SBRP, for which data were not adjusted
 during the calibration procedure, isotherm data for all M-APP watersheds required significant calibra-
 tion to match distributions of observed and simulated sulfate for the base year.  Two of the isotherm
 coefficients were adjusted, with all values of B1 raised by a factor of three, and all values of B2
 adjusted by a factor of 0.75. Results described here  show that use of the calibrated isotherm data in
 model simulations results in  projections that match not only the mean watershed sulfate condition for
the region, but also very closely match the regional distributions of sulfate concentration and  percent
watershed sulfur retention.
     The final check on the isotherm data used in model projeptions was to compare measured rates
of change for sulfate in stream waters of the M-APP to rates projected by model simulations.  Because
there is very limited surface water monitoring data from which long-term sulfate trends can be deter-
mined, and because the representativeness of those trend data is unknown, we cannot make a
                                              171
-------
        STREAMWATER VS SOIL SOLUTION SULFATE

                  DDRP MID-APPALACHIAN WATERSHEDS
        350
     I  3001
     LLJ
     ID
     CO
     z
     o
     H-
     _j
     O
     CO
        250-
200-
     o  100-
     o
     N
     DC
     O
     X
     CD
150-
   50      100      150      200     250     300

                 STREAMWATER SULFATE (ufiq/L)
                                                             350
Figure 9-2.   Equilibrium soil solution sulfate (B horizon data) and measured stream sulfate
          concentration for the 36 DDRP M-APP watersheds.
                                172
-------
     MID-APPALACHIAN SULFATE
 measured vs modelled - SO4 (u,eq/L)
          100    150    200    250   300
            MEASURED SULFATE (neq/L)
                                   350
    MID-APPALACHIAN SULFATE
36 sites; stream vs model year 0 values
u
KO


0,8


0.6


0.4





0.0
        50   100  150  200  250  300  350  400
               SULFATE (neq/L)
                                                    MID-APPALACHIAN SULFATE
                                              measured vs modelled - % Sulfur Retention
                                                    -40
                                                        B
                                                                       .» «
                                                     -20
                                                         0     20     40     60
                                                             CALCULATED % RTN
                                                                                    80
                                                    MID-APPALACHIAN SULFATE
                                               36 sites; stream vs model year 0 values
                                                    1.0
                                                  o
                                                  C u
                                                  Q_
                                                  UJ
                                                  > 0.4-
                                                    0.2-
                                                  O
                                                       -20   0   20   40   60   80
                                                          PERCENT SULFUR RETENTION
                                                                                    100
  Figure 9-3.   Measured and modelled sulfate concentrations and percent sulfur retention for
              the 36 DDRP M-APP watersheds.
                                         173
-------
rigorous comparison.  Model projections of sulfate change are, on average, somewhat higher than
reported rates of change for surface waters in the region (Table 9-2).  Reported values fall well within
the range of model projections, however, indicating that the model values are not unreasonable esti-
mates of change for sulfate in surface water sulfate.

9.2.4.3. Base Year - Measured Sulfate and Model Projections

     As suggested by results presented in the preceding  section, measured sulfate concentrations
and model projections of the response of M-APP stream systems to atmospheric sulfate deposition are
indicative of a diverse region with a wide range of surface water sulfate concentrations and response
times.  Cumulative distributions of sulfate concentration and of percent sulfur retention for the target
population  (5496 stream reaches), based on measured and modelled base year sulfate for the 36
sample stream reaches in the region, are shown in Figure  9-4 and are summarized in Table 9-3.
Because data presented in this and subsequent plots have been  weighted to reflect the unequal
sampling probabilities of streams in the National Stream Survey (Kaufmann et al., 1988), mean and
median values of sulfate for the population are different from those for only the sample watersheds,
and the plots for the population differ somewhat in appearance from those for the 36 watersheds
shown in Figure 9-3.

      For the target stream population,  measured sulfate concentrations  range from 55 to 312/ieq/L,
with a  mean of 148 and median of 153,ueq/L  The calculated percent sulfur retention  in M-APP
watersheds (computed using NSWS stream sulfate with precipitation, runoff, and sulfur deposition
estimates  described  in Section 5.6) is also highly variable, with a  mean of 37 percent retention and a
range of -21 to 74 percent.  Base year  sulfate concentrations projected by the model range from 40 to
365 /
-------
Table 9-2.  Rates of Increase for Sulfate in Stream Systems In the M-APP Region, Comparing
           Model Projections to Measured Rates of Increase


Site
36 DDRP M-APP
streams


Deep Run, VA
White Oak Run, VA
Fernow, WV, ws #4
Watershed #4


Period of
Record
model-based
estimates


1980- 1987
1980- 1987
1970 - 1985


Stream SO4a
(weq/L)
55 - 313


107
84
85-90

Rate of SO4
Increase
(weq/L/yr)
median = 2.20
range -0.1 - 4.6
Q1 = 1.70
Q3 = 3.53
2.2
2.0
1.0



Reference
this study


Ryan et al., 1989
Ryan etal., 1989
D. Helvey, U.S.
Forest Service
(pers. comm.)
Madison Run, VA
1968, 1982
70
1.3
                                                                        USGS, 1969,
                                                                        1970; Lynch and
                                                                        Dise, 1985
 Concentration In last year of record
                                          175
-------
                          SULFATE CONCENTRATION
          1.0
              0     50    100    150    200    250   300   350    400
                                  SULFATE (neq/L)
          1.0'
       z;
       o
          Hf\ Q ,
       «*-  w«o
       D_
       O

       O  0.6
       Q-
       LU
       >  0.4
       ID
       O
          0.2-
          0.0
                         PERCENT SULFUR RETENTION
     Calculated
	Modelled
             -40     -20      0       20     40      60      80
                           SULFUR (PERCENT RETENTION)
             100
Figure 9-4.   Distributions of measured and modelled sulfate concentration and percent
            sulfur retention for the M-APP target population of 5,496 stream reaches.
            Differences between this figure and plots C and D of Figure 9-3 reflect the
            unequal weighting factors for individual sample watersheds used to generate
            regional population estimates.
                                      176
-------
Table 9-3.  Summary of Sulfur Status in Surface Waters for Watersheds in DDRP Target
           Populations in the NE, M-APP, and SBRP
Variable
SO4 concentration
(«eq/L)




Region
M-APP measure
model
NE measure
model
SBRP measure
model
Mean
148
171
110
128
37
44
Std.Dev
55
74
40
55
26
23
Min.
55
40
34
52
15
13
Median
153
165
105
118
24
41
Max.
313
365
214
281
119
107
Steady state S04    M-APP   measure
concentration
   (weq/L)          NE
                  SBRP
         measure

         measure
246

118

122
72

47

23
140

 50

 86
215

111

120
405

239

203
Change in sulfate  M-APP   measure       98
to steady state
concentration      NE      measure        8
  ("eq/L)
                  SBRP    measure       85
                                          -39

                                          -51

                                           25
                              107

                                0

                               83
                            248

                             28

                            167
Percent sulfur
 retention
M-APP   calculated8
         model
                  NE
         calculated8
         model
                  SBRP    calculated3
                           model
 37
 30

  4
 -8

 71
 64
24
22

20
 4

17
18
-21
-54
-19

 17
 20
 44
 26

  0
 	7

 78
 68
 74
 84

 69
  0

 87
 88
8 Calculated retention, based on measured NSWS sulfate concentrations and on estimates of precipitation, runoff, and sulfur
  deposition as described in Section 5.6.
                                            177
-------
       LAKE/STREAM SULFATE
                                     STEADY STATE SULFATE
  1.0
                                            1.0
  0.0
         50
             100   150  200   250
                SULFATE (neq/L)
                               300
                                    350
                                          O
                                      -i	"-i	r
                                      50  100 150  -200  250  300
                                              SULFATE (neq/L)
                                                                       350 400 . 450
    PERCENT SULFUR RETENTION
                                     CHANGE IN SULFATE TO
                                          STEADY STATE
                                            1.0
    -60
        -40
     —r
 -20   0   20   40  60
PERCENT SULFUR RETENTION
                                    100
                                          O
                                              -50
  50    100   150
DELTA SULFATE Qieq/L)
                                                                         200
                                                                              250
Figure 9-5.   Distributions of measured sulfate concentration, steady-state sulfate concen-
            tration, percent sulfur retention, and change in sulfate concentration to sulfur
            steady state in the NE, M-APP, and SBRP.
                                       178
-------
Table 9-4.  Comparison of Sulfate Concentrations and Sulfur Budget Status for DDRP
            Watersheds in Subregions of the M-APP Region3
Subregion n
1D 7
2B 9
20 20
2Cab 7
2Cbb 5
2Ccb 8
Measured SO4
(aeq/L)
125
55 - 166
124
69 - 209
174
74-313
192
96 - 313
188
154 - 264
149
74 - 223
Steady State SO4
(weq/L)
228
194-313
288
207 - 405
224
140 - 299
268
185 - 299
266
189 - 278
183
140 - 243
Percent Sulfur Retention
44.0
17.9- 71.8
57.2
48.5 - 73.6
21.4
-21 .5 - 64.0
27.5
-8.6 - 65.0
17.3
4.7 - 28.8
18.6
-21.5-61.0
"Values are unweighted mean concentrations and ranges.

b2Ca = First four characters of Watershed ID = 2C28, 2C29, or 2C35
 2Cb
 2Cc<
2C41
2C46, 2C47, or 2C57
                                             179
-------
      The current status of surface water sulfate and of sulfur budgets in the M-APP suggests a region
that is transitional between NE watersheds, most of which are at or near steady state, and those in the
SBRP, all of which are still retaining sulfate.  Although the budget status supports such a "transitional"
state, there is no clear north-to-south gradient of increasing sulfur retention (Table 9-4).  The highest
average sulfur retention in the region (57 percent) occurs in stream survey subregion 2B (Valley and
Ridge Province in Pennsylvania, Maryland, West Virginia and Virginia), and the lowest average
retention (21 percent) occurs in the Allegheny Plateau (stream survey subregion 2C, in Pennsylvania,
Maryland, and West Virginia). Within the plateau region, sulfur retention is highly variable; 3 (of 20)
stream systems in the sample population have net retention higher than 40 percent, but an equal
number are computed as having net sulfur release. Within the same region, a trend of decreased
deposition from north to south is matched by proportional decreases in stream sulfate concentration,
and no spatial trend for retention is apparent.  In the Valley and Ridge, there are similar trends of
decreasing deposition and stream sulfate from north to south, but no trend in percent retention.

9.2.4.4.  Model Results -- Projections for Current Sulfur Deposition

      The principal focus of the Level II analyses is to project the future dynamics of sulfate in the
M-APP watersheds. Figure 9-6 shows temporal trends for both the past and the future (for constant
future deposition) for soils in a typical M-APP watershed, as well as for the fastest and slowest
responding systems among the sample watersheds. As might be expected based on sulfur budget
data,  model projections indicate a wide range of response times to sulfur deposition for soils in the
region.  For many systems in the region, there  is a lag of 25 to 50 years between increases in sulfur
deposition and  corresponding sulfur outputs, resulting in  current sulfur retention of between 30 and 40
percent.  Some systems,  however, are projected to respond with very short lags of only about 10 to
15 years, and are projected to be already at steady state. At the other extreme, a few watersheds with
deep, highly-adsorbing soils  have very long lags and are  projected to have only recently begun to
experience sulfate breakthrough from the soil.  Some of the latter systems are not projected to reach
sulfate steady state for over 150 years.

      Comparison of time traces for the M-APP watersheds to those for the NE and SBRP indicate
major differences  among the three regions, in terms of both historic deposition and response time of
soils to changes in deposition. In the NE, significant increases in sulfur deposition began much earlier
than in the M-APP Region, almost a century ago, and deposition reached  current levels around 1910.
During much of the intervening time, deposition exceeded current levels, at times by as much as 60
percent (Gschwandtner et al., 1985).  Along with very high historic loadings of sulfate, soils in the
region have low adsorption capacities and respond with very short lags to changes in deposition.
Previous DDRP analyses for the region (Section 9.2 of Church et al., 1989) indicate that soils in the NE
respond over time frames of  about a decade.

      Simulations for watersheds  in the SBRP used a historic input sequence (based on Gschwandtner
et al.,  1985) in which sharp increases in deposition occurred later than in the M-APP. Modelled depo-
sition  inputs increased almost exponentially from the late  1800s to 1970 (with a decrease during the
1930s and a temporary increase during the 1940s), but inputs did not exceed 50 percent of current
loadings until 1955 and have never significantly exceeded current levels of deposition. Sulfate

                                             180
-------
                      SULFATE RESPONSE TIME
                      DDRP MID-APPALACHIAN WATERSHEDS
            120
        X
            100-
        I	   ou

        LL
        =!   601
        CO
        LJJ
        >   40"
        ULJ   20-
        CC
             0
                            Deposition Input

                            Stream Output
                                  Typical
                                  Range
                                       T
              1850
1900
1950'    2000     2050     2100     2150
        YEAR
Figure 9-6.   Time trends for modelled inputs and outputs of sulfate for soils on a typical
            watershed in the M-APP Region, and for the fastest and slowest responding
            systems in the region.  Historic inputs are based on Gschwandtner et al.
            (1985). Sulfur flux is expressed on a relative scale, with base year (1986)
            deposition flux equal to 100. The base year is indicated by the arrow.
            Because annual precipitation and runoff are constant during the simulation
            period, input and output curves also simulate relative changes in sulfate
            concentration. Simulated changes  in sulfate concentration in precipitation are
            proportional to changes in deposition flux, and changes in streamwater sulfate
            to output flux (scales for concentration of precipitation and runoff are not
            equal, due to dry deposition and evapoconcentration of sulfate in runoff).
                                      181
-------
response times for SBRP watersheds, as for systems in the M-APP, are characterized by a wide range
of response times. Lags following changes in deposition range from about 20 to more than 150 years,
with an average of roughly 50 years.  The differences in sulfur budget status between M-APP and
SBRP watersheds result both from differences in adsorption capacity and from the earlier onset of high
deposition inputs to M-APP watersheds.

      Responses of M-APP watershed soils to a scenario of constant future deposition are summarized
by Figures 9-7 and 9-8 and in Table 9-5.  Figure 9-7 shows distributions of modelled sulfate concentra-
tion and percent sulfur retention for the base year of 1986, then at 20, 50, and 100 years and at sulfur
steady state (concentration only); also shown are projected changes (increases) in sulfate concentra-
tion between the base year and the  same 20-, 50-, 100-year, and steady-state intervals.  Data for
simulations  exceeding 50 years are included principally to indicate the relative magnitude of change
between 50 years and steady state,  compared to projected changes from the present to year 50.
During the next 50 years, substantial increases in streamwater sulfate are projected, during  which time
soils in a majority of watersheds in the region will approach sulfate steady state.  During the next 20
years, median sulfate concentration  is projected to increase by 25^eq/L (from 165 to 190/feq/L), with
a  mean increase in concentration of 32/^eq/L (maximum increase of 76,weq/L).  During the  same inter-
val, percent sulfur retention is projected to decrease by an average of 15 percent, and the fraction of
M-APP watersheds with less than 5 percent sulfur retention is projected  to increase from 14 to 36
percent.  The projected increase in median sulfate concentration at year 50 is 41 ,«eq/L (165 to 206
/^eq/L) (projected change in median sulfate is not necessarily equal to the median projected change),
with an accompanying decrease in median retention from 26 to 3.1 percent. After 50 years, soils on a
majority (58 percent)  of watersheds  in the region are projected to be within 5 percent of sulfur steady
state, and retention is projected to exceed 10 percent  in only 23 percent of M-APP watersheds.

      For time  periods longer than 50 years, the projected rate of change for sulfate in soils of
M-APP watersheds slows considerably, due to the high percentage of systems that have already
reached or  are near steady state. Between years 50 and 100, median  sulfate is projected to increase
by only 4^eq/L and median percent retention to decrease by less than  3 percent (from 3.1  to 0.4
percent).  At year 100, only 7 percent of watersheds are projected to have more than 5 percent sulfur
retention. Those sites are characterized by soils formed from volcanic parent material that is more
similar to  soils  of the  SBRP than to the soils  of other M-APP watersheds that have formed from sedi-
mentary materials.  Projected changes after  100 years are minimal, since most sites are projected to
be essentially at steady state by that time; increases in sulfate concentration between 100 years and
steady state are projected to exceed 1 yueq/L for only 42 percent of M-APP watersheds and to be
greater than 10 jueq/L in less than 10 percent of watersheds.  Most of the sites projected to have
continued retention after 100 years are located in the Blue Ridge or Piedmont of Virginia.

      Time to sulfur steady state for soils in M-APP watersheds, for a scenario of constant sulfur depo-
sition, is summarized in Figure 9-8.  Consistent with procedures used  in previous analyses for the NE
and SBRP and as discussed earlier, response time (time to sulfur steady state) has  been defined as
the time for sulfate to reach 95 (or 105) percent of steady state concentration. In the M-APP Region,
soils on roughly 15 percent of the watersheds have already reached sulfur steady state.  Median
 projected time to steady state is 35 years; 25th and 75th percentiles of  projected time to steady
                                              182
-------
                         SULFATE CONCENTRATION
                          50  100  150  200  250  300  350  400  450
                     0.0
                                   SULFATE (neq/L)
                  CHANGE IN SULFATE CONCENTRATION
                                TO STEADY STATE
                   o
                                  80    120   160

                                    SULFATE (neq/L)
                                                  200
                                           240
                       PERCENT SULFUR RETENTION
                            T
                            40    80   120    160   200   240

                             PERCENT SULFUR RETENTION
Figure 9-7.
Projected changes in sulfate concentration, delta sulfate (change from base
year concentration),  and percent sulfur retention for M-APP DDRP target
stream reaches.
                                     183
-------
      O

      DC
      O
      Q_
      O
      DC
      D_
      LU
      O
                  TIME TO SULFUR STEADY STATE
         1.0
0.8-
0.6-
          0.4-
          0.2-
          0.0
             0     20    40   60   80   100  120   140   160  180
             TIME TO 95% STEADY STATE (YEARS) AFTER 1986
Figure 9-8.  Distribution of time to steady state for sulfur for soils on DDRP target stream
          reach watersheds in the M-APP Region.
                                184
-------
Table 9-5. Summary Statistics for Modelled Changes in Sulfate Concentration and Percent
           Sulfur Retention, and for Changes in Sulfate Concentrations for Watersheds in
           the M-APP Region8

SO4_conc






%Srtnb





A_S04C




year to 95%
steady state
Year
measured
model-0
20
50
100
140
steady
measured
model-0
20
50
100
140
0-20
0-50
0-100
0-140
0-steady
—

A_SO4 - rate /zeq/LVyr
of change
in year 0


Mean
148
171
204
226
240
243
246
37 •
30
17
8 ,
3
1
32
55
68
72
74
46

2


Std dev
55
74
77
77
74
73
72
24
22
19
14
6
3
21
51
51
55
58
._

1


MIN
55
40
59
94
140
140
140
-21
-1
-0
0
0
0
-2
-3
^3
-3
-3
0

0


P_25
109
121
151
183
187
188
190
20
10 -
2
0
0
0
19
.35
38
38
38
6

. 2


MEDIAN
153 f
165
190
206
211
213
215
44
26
11
3
0
0
33
43
45
45
45
.35 -

2


P_75
175
208
237
268
283
287
293
56
45
25
9
2
1
39
75
99
101
103
62

4


MAX
313,
365
397
404
405
405
405
74
84
75
55
26
11
76
134
171
200
218
160

5


  Model projections for future sulfur deposition at current levels.

  % of systems with >5 % S retention
      Year     0      86%
              20      64
              50      42
             100      7.5

  Change in modelled sulfate concentration from year 0 to n.
                                              185
-------
state are 6 and 62 years, respectively, and the maximum projected response time is 160 years.
Between the base year and year 80, there is an essentially steady increase in the percentage of
systems at steady state, to about 90 percent of systems at steady state by year 80.  Of those systems
with longer response times, most are represented in the 36 modelling watersheds by sites in the
Virginia Blue Ridge or Piedmont and have deep soils with very high adsorption capacities.  Soils and
response times for these sites are similar to those previously characterized by DDRP analyses in the
SBRP.

      The distribution of sulfate response times in the M-APP is°intermediate with respect to the
distribution in the NE and SBRP (Table 9-5, Figure 9-9).  Previous analyses for the other regions,
discussed in Section 9.2 of Church et al. (1989), found that almost  40 percent of NE systems were at
steady state in the base year, and all systems were projected to be at steady state within 9 years
following the base year.  In contrast, no SBRP watersheds are presently at sulfur steady state, and
median projected time to steady state is 61  years, almost twice the 35 year median for the M-APP
Region.  The largest differences in response time between M-APP and SBRP stream systems lie with
those sites responding in the shortest period of time.  In  contrast to the 15 percent of M-APP sites
already at steady state, no SBRP sites are presently at steady  state; the first SBRP watersheds are not
projected to reach steady state for 16 years. Similarly, 25 percent  of M-APP sites are projected to be
at steady state within 6 years, compared to a projected 40-year period before as many as 25 percent
of SBRP sites come to steady state.

9.2.4.5.  Model Results - Projections for Scenarios of Reduced  Future Deposition

      In addition to  projections of future sulfur dynamics  at constant deposition, simulations were run
for the two scenarios of reduced sulfur deposition show in Figure 9-1. Figure 9-10 compares the
response of a typical M-APP watershed  and of fast- and slow-responding  systems in the  region to con-
stant  deposition and to the two scenarios of reduced sulfur deposition.  Probably the most important
result shown by these projections is that under conditions of reduced deposition, systems will even-
tually come to steady state at lower sulfate concentrations than they will if current levels of deposition
are maintained.  The projections reported here are based on an assumption that sulfate adsorption  by
soils is fully reversible. If sorption is only partially reversible, response times to decreases in
deposition will be shorter, because only a portion of sulfate will be  desorbed.  For individual systems,
the nature and timing of the response to decreased deposition are quite variable.  For watersheds with
soils having low adsorption capacities that have responded quickly to historic deposition and are near
steady state for current deposition, decreases in deposition are projected to lead to rapid decreases in
sulfate efflux and to proportional decreases  in sulfate concentration in runoff. Conversely, for the
systems that respond very slowly and that are still at less than 50 percent of steady state, a decrease
in deposition is projected to result in  more gradual increases in sulfate as systems slowly approach
steady state with  the lower level of deposition.  For most watersheds  in the region, a lagged response
comparable to that of the "typical" watershed in Figure 9-10 is  projected, that is a continued increase
in sulfate export for one to three decades, as sulfate currently  adsorbed on upper soil horizons, in
equilibrium with today's high levels of deposition, is desorbed  and  leached from the soil, followed by a
decrease in sulfate as the system comes to steady state with the lower level of deposition. The extent
and duration of such lags (i.e., the time interval between  a decrease in deposition and the onset of

                                             186
-------
                  TIME TO SULFUR STEADY STATE
                          NE, MID-APR, AND SBRP
      Z)
      o
         1.0
      o
      p  0.8
      DC
      O
      O  0.6
      Q_
      LLJ
      >  0.4 H
         0.2-
         0.0
               NE
     M-.
            0
—T—
 40
—r—
 60
—i—
 100
—I—
 120
—i—
 160
20   40   60   80   100  120  140  160  180
  TIME TO 95% STEADY STATE (YEARS)
200
Figure 9-9.  Distributions of time to steady state for sulfur in soils on DDRP target
          watersheds in the NE, M-APP, and SBRP.
                                 187
-------
                                   SCENARIO A
                     120


                   * 100
                   11.
                   L1I
                      80-
                   :D 60-
                   CO
                   ai
                   > 40-
                   LLI  20-
                   cc
                          	1	1	1	1—
                       1850    1900   1950   2000   2050
                                          YEAR

                                   SCENARIO B
                     120
                     100-
                   LU
                      80-
                   r>  60"
                   00
                   LU
                   >  40-


                   UJ  20-
                   GC
                       1850    1900   1950   2000   2050
                                          YEAR

                                   SCENARIO C
                     120
                     100
                   u_
                   UJ
                      80
                   3  60
                   CO
                   HI
                   >  40-
                   Uj  20-
                   CC
	1	
 2100   2150
                                                        2100   2150
                       1850    1900    1950   2000    2050    2100   2150
                                          YEAR

Figure 9-10.  Projected temporal trends for sulfate in a typical M-APP watershed, and for
             the fastest and slowest responding systems in the region, for three scenarios
             of future sulfur deposition.  Sulfur flux is expressed on a relative scale, with
             base year (1986) deposition equal to 100.  The base year is indicated by the
             arrow. Note that because annual precipitation and runoff are constant
             throughout the simulation period, changes in sulfate concentration are
             proportional to changes in sulfur flux.
                                         188
-------
decline in projected streamwater sulfate concentration) in soils of M-APP watersheds is quite variable;
in simulations of the 36 M-APP watersheds (for Scenario B), the lag phase for some watersheds lasted
more than 40 years, during which time projected sulfate concentrations increased by as much as 60
fieq/L (Figure 9-11).

      In addition to qualitative comparisons of sulfate concentrations at steady state, it is important to
compare scenarios of future sulfate to current sulfate concentrations.  As shown in Table 9-6, at
current deposition, mean sulfate concentration is projected to rise sharply during the next 50 years
(171 to 226 fieq/L), with a subsequent increase to 240 ^eq/L in 100 years and to 246 fieq/L at steady
state.  In contrast, under deposition scenario "B"  (50  percent decrease in deposition between years 5
and 15, then constant at 50 percent of current deposition), simulations project a temporary increase in
sulfate, as described above, but then a decrease to 150, 133, and 129 /ueq/L at 50 and 100 years and
at steady  state, respectively.  For scenario "C", as described earlier, projected concentrations initially
rise, but decrease from 171 to les^eq/L by year 50 before increasing by about 20 percent to 191 and
199 /teq/L at 100 years and at steady state.

      These results show that in the long run, sulfate outputs from soils and sulfate concentrations in
solution will be proportional to deposition. In the short term, sulfate concentrations will increase,
sometimes by a substantial amount, even if deposition is sharply  reduced; within 50 years, however, a
reduction  in deposition will result in substantial decreases in sulfate concentration below current
concentrations and especially in comparison to the concentrations that will result if deposition remains
at current levels,  In order to maintain long-term sulfate concentrations at current levels, deposition
would need to be reduced to 63 percent of current input (i,.e., equal to average current output); higher
deposition would result in long-term increases in  deposition.

9.2.5. Summary

      The projected sulfate response of soils and surface waters in the M-APP describes  a region
composed principally of delayed response systems. Although a small percentage of watersheds in the
region have soils with very short response times to sulfur deposition, most M-APP watersheds are
mantled with soils that have buffered sulfur inputs for several decades and are continuing, albeit with
diminishing efficiency, to retain sulfate by adsorption. At the present time, median modelled retention
in M-APP watersheds is 26 percent, but modelled sulfate concentrations are increasing as sulfate
breaks through the soil column, and retention is projected to decrease sharply over the next few
decades.  Median modelled retention (assuming  continued sulfur deposition at current levels) is
projected  to decrease from 26 percent at present to 11  percent within 20 years  and to only 3 percent
within 50 years.  Sulfate adsorption has played an important role in minimizing the adverse effects of
acidic deposition on surface waters in many M-APP watersheds, but as sulfate concentrations increase
in the next few decades, there will  be a proportional increase in sulfate-mediated cation leaching, due
to a combination of the salt effect and depletion of exchangeable base cations;  effects of sulfate
leaching on cation removal from M-APP soils are discussed in Section 9.3 and in Section 10.

      The nature of future sulfate dynamics in the soils of M-APP watersheds, as discussed in Sections
9.3. and 9.4, is highly dependent on the nature of sulfur deposition to watersheds of the region. At
                                              189
-------
              MID-APPALACHIAN WATERSHEDS
             PROJECTED LAGS AFTER DECREASE IN DEPOSITION

^
Q)
3;
LLJ
I'-
LL
CO
LLJ
CO
LLJ
DC
O

80 -
70 -
•
60 -
*
50 -
40 -
30 -
20 -

10 -

o-

•
• •
B
B •
*
• " * • "
•••
"" B ."
m*

• •

_-|Q -| 	 1 	 1 	 1 	 1 	 1 	 1 	 1 	 1 	 ' 	 1 ' 1 • r i • i
0 5 10 15 20 25 30 35 40 45 5<
                      DURATION OF LAG (years)
Figure 9-11.  Extent and duration of lags in modelled sulfate response for soils on 36
          M-APP watersheds following a 50 percent decrease in sulfur deposition.
                               190
-------
Table 9-6.  Summary Comparison of Modelled Changes in Sulfate Concentration in M-APP
             Stream Systems Under Three Future Sulfur Deposition Scenarios8
Variable
Year
                                                             Scenario13
 B
Measured SO4
    (aeq/L)
                     148
148
148
Modelled SO4C
(weq/L)



AS04e
(aeq/L)

0
Iagd
50
100
steady
50
100
steady
171
~
226
240
246
55
68
74
171
200
150
133
129
-21
-38
-42
171

168
191
199
-A
19
28
a  Concentrations are weighted means for the DDRP target population.

b  Deposition scenarios as described in Section 9.2.3.2 and shown in Figure 9.1; A is constant future
   deposition at current levels for the entire simulation period; B is constant at current levels for
   5 years, decreasing to 50 percent of current levels by year 15, and constant, thereafter; C is constant at
   current levels for 5 years, decreasing to 50 percent of current loadings by year 15, then increasing to 80 percent
   of current levels si year 50 and constant thereafter.

0  steady «• steady state.

d  Lag is maximum concentration during lag phase after reduction in deposition.

8  A SO4 = change in sulfate from year 0 to n.
                                                   191
-------
constant deposition, average sulfate concentrations are projected to increase by over 40 ^eq/L in the
next 50 years, whereas if deposition is reduced by 50 percent, the average concentration is projected
to decrease by 25 //eq/L. The projected difference in average concentration between the two deposi-
tion scenarios increases to almost 120 /leq/L at sulfur steady state. It is also important to recognize
that despite the lags in sulfate response that are projected following a decrease in deposition, pro-
jected  sulfate concentrations for the scenario of decreased deposition are in all cases, at all times in
the future, lower than those projected if deposition is not reduced.  It is also important to reiterate that
the projections and time frames discussed here refer only to sulfur dynamics as affected by sorption in
soils. Sulfate mobility is considered to be one of the principal factors influencing changes in transport
of cations from the soil, but processes that do not involve sulfur also play crucial roles in the dynamics
of base cations and ANC. Rates of change in ANC,  and especially potential time to zero ANC, are not
necessarily coincident with time to steady state for sulfur. Systems can reach  zero ANC before, con-
currently with, or after time to sulfur steady state. The relationship between changes in sulfate and
ANC will be characterized and discussed as part of the integrated watershed modelling (Level III)
analyses in Section 10.

     The modelling projections described here are  generally consistent with other sulfur analyses for
the M-APP  in this report.  Measured sulfate concentration and percent sulfur retention data, described
in the sulfur budget analyses in Section  7, provided  part of the "true" data used for model calibration
and are therefore similar as a result of fitting during calibration of model inputs. Of greater signifi-
cance, the  projections of future sulfate mobility are very similar to those generated as part of the
integrated modelling described in Section 10. Although the two models are based on the same sulfate
adsorption  algorithm and on the same isotherm and deposition data, several important model proper-
ties (hydrologic routing and consideration of deep (> 2 m) soils) are different,  and calibration
procedures were  different and were conducted independently.  Despite these important differences,
significant similarities exist between the two sets of projections  for sulfate. These similarities contribute
to confidence in both analyses.

      Finally, major results for the Level  II sulfate analyses for M-APP watersheds can be summarized
as follows:

      •  Measured and  modelled  sulfate concentrations in M-APP watersheds are, on  average, higher
         than equivalent values for NE  and SBRP watersheds. Because sulfur deposition to M-APP
         watersheds  is substantially higher than to  most NE or SBRP watersheds, most M-APP
         systems are still retaining sulfur.

      •  Median measured sulfur retention in M-APP watersheds is 44 percent;  median modelled
         retention is somewhat lower at 26 percent; modelled  retention ranges from -1 to 84 percent.

      •  If current levels of sulfur deposition are maintained, sulfate concentrations are projected to
         increase substantially over the next 50 years; the projected median increase is 33 ^aeq/L over
         the next 20 years and 43 /teq/L in the next 50 years.  Maximum projected increases for the
         same intervals  are 76 and 134/ieq/L.  Increases in sulfate concentration between year 50
         and steady state are projected to be much smaller than during years 0 to 50, because most
         systems will be near steady state within 50 years.
                                              192
-------
      •  As sulfate concentrations increase, median percent retention is projected to drop from 26
         percent at present to 11, 3, and 0.4 percent at 20, 50, and 100 years, respectively.  For the
         same time intervals, the projected fraction of the target population at sulfur steady state  (i.e.,
         < 5 percent retention) is projected to increase from 14 percent in the base year to 36, 58,
         and 93 percent.

      *  If sulfur deposition is reduced, there will be proportional decreases in projected steady-state
         sulfate concentration.  Intermediate dynamics are dependent on the amount and timing  of
         decreases in deposition.

      •  For systems with sulfate concentrations significantly lower than steady state, there is likely to
         be a lag phase following any major reduction in deposition, during which sulfate concentra-
         tions in soil leachate continue to increase due to desorption of previously adsorbed sulfate.
         In projections for M-APP systems, such lags last as long as 50 years, during which time
         sulfate concentrations rise as much as 60 ^eq/L  During any such lag phase, sulfate con-
         centrations would be lower than if deposition had not been reduced.

      •  Qualitatively, sulfate adsorption by soils has been, and  continues to be, an important
         process in delaying the effects of acidic deposition in M-APP watersheds.  Most M-APP
         systems  should be regarded as delayed response systems with regard to sulfate. The
         relative importance of adsorption is much higher in M-APP systems than in soils of NE
         watersheds.  It is somewhat lower than in soils of SBRP watersheds, however, due both  to
         the lower average potential for adsorption by typical M-APP watersheds and to the very low
         adsorption (and short response times) for a small percentage of M-APP watershed soils.

9.3   EFFECT OF  CATION EXCHANGE AND WEATHERING ON SYSTEM RESPONSE


9.3.1  Introduction


      Soils provide an important linkage between acidic deposition and surface waters.  Soil can
buffer the effects of acidic deposition by a variety of processes including cation exchange, mobile
anion absorption, and mineral weathering.  The potential of soils to buffer acidic deposition is  often
quite large, especially in the case of calcareous soils.  In the case of M-APP uplands, however, soils
have formed from sedimentary bedrock that consists of the slowly weatherable remnants of previous
weathering cycles. These rocks and associated soils often contain very low quantities of base cations,
and thus have a limited capacity to neutralize acid  inputs.  Consequently, not only are surface waters
in this region at risk from acidic deposition, the  soils in the region are already acidic and could
become even more acidic.


      During the development of the NAS panel report (MAS, 1984), much discussion was devoted to
the role of cation exchange and mineral weathering in protecting  watersheds from acidification. One
group of panel members argued that soil cation exchange in most watersheds provides a large reser-
voir of base cations, sufficient to buffer against potential changes caused by acidic deposition.
Therefore, they argued, even in watersheds where cation exchange (as opposed to mineral weather-
ing) is an important process  in supplying base cations, the future effects of acidic deposition are
probably not a concern.  Another group of panel members argued that the buffering capacity  of soils
was finite, and that continued exposure to current levels of acidic deposition would have long-term,
adverse effects on water quality in some systems.  The conclusion of the committee as a whole was
that the role of cation exchange in buffering against the effects of acidic deposition is an area  of
                                             193
-------
considerable uncertainty, and that these processes need to be considered when attempting to project
future effects of acidic deposition on aquatic ecosystems. Thus, the Level II base cation studies were
designed to determine the role of base cation exchange in controlling future changes in surface water
chemical composition.  The specific objectives were to:

      •  identify the role that base cation exchange has in determining current surface water
         composition;

      •  determine the capacity of base cation exchange processes to buffer against future changes
         in surface water composition as a result of acidic deposition;  and

      •  make projections regarding the magnitude and extent of changes that could occur in region-
         ally representative soils and surface waters as a result of continued exposure to acidic
         deposition.

      We developed several hypotheses to guide the investigations of the role of cation supply
.processes  in regulating surface water chemistry in representative watersheds in the DDRP study
regions.

9.3.1.1  Level II Hypotheses

      Five  hypotheses guide the Level II base cation analyses:

      •  Cation  exchange processes determine surface water composition.
      •  Soils delay surface water acidification.
      •  Increased deposition  induces net cation leaching.
      •  Cation  resupply rate is slow.
      •  Soil chemistry is an indicator of soil response to acidic deposition.

      A brief discussion of each hypothesis follows.

9.3.1.1.1  Cation exchange processes determine surface water composition --

      The first hypothesis is  designed to identify the primary process or processes that regulate surface
water chemistry.  In systems that have attained steady state with respect to sulfate deposition (see
Sections 7 and 9.2), primary mineral weathering and biological uptake are probably the principal
processes  that modify the composition of incident deposition.  Under steady-state conditions, the base
cation exchange  pool should actively reflect the dynamic balance between these two important proc-
esses.  Regardless of their relative importance, however, if soils regulate surface water ANC values,
then this should  be reflected by the composition and chemical properties of the soil exchange complex.

      The  hypothesis is tested by comparing surface water composition projected  using soil cation
exchange models with observed values. A close correspondence between the observed and projected
values suggests that soil exchange processes have a major role in regulating surface water chemistry.
                                              194
-------
9.3.1.1.2  Soils delay surface water acidification --

      The second major hypothesis is that soils will delay, but not prevent, the acidification of surface
waters. The concept behind this hypothesis is that soils have a finite capacity for buffering against
changes in surface water chemistry caused by increased levels of acidic deposition.  In essence, the
chemical and physical characteristics of a soil reflect its response to some given set of environmental
conditions.  Therefore, at a given level of  deposition, vegetative-uptake, mineral weathering, etc., the
cation exchange pool reflects a balance of the various sources and sinks for cations in that area. This
balance is dynamic, changing with the seasons and with the shifting flow of cations among the various
reservoirs.

      When a perturbation such as acidic deposition is imposed on a system, the system (in this case
the soil) evolves toward  a hew state of balance. The rate at which changes take place depends both
on the sizes of  the cation reservoirs in the system and on the flux of material between reservoirs. If
the transfer rates  of material between reservoirs is slow, or if the mass of material in the affected
reservoirs is large relative to the transfer rates, then rates of evolution toward the new balance point
tend to be slow.  Conversely, if reservoirs are small relative to the size of the flux between reservoirs,
then adjustments  to a new system steady state  can occur rapidly.

      This hypothesis contends that the pool of base cations on soil exchange complexes is  large
relative to the rate of cation loss from the  system by leaching.  As a result, the rate of adjustment of
the exchange complex to the new deposition conditions should require decades to centuries  before a
new steady-state, or dynamic balance, condition is attained.

      The hypothesis  is  tested using a model approach.  Measured soil properties serve as inputs to
the various models. The Level II models,  all of which have a mass balance component, track the loss
of base cations from soils at the specified levels of deposition. The models are concerned primarily
with exchange processes and do not explicitly include cation supply via weathering.  Therefore, the
computed mass balances should correspond to the maximum leaching rates that could occur.  The
rate of change of base cation status of the soils included in the study, then, should be related to the
amount of time over which the soils should delay acidification of surface waters.

9.3.1.1.3 Increased deposition induces net cation leaching --

      The third  major  hypothesis is that increased levels  of deposition, specifically, increased  concen-
trations of sulfate  and nitrate in deposition, increase the rate of cation leaching from the soil exchange
complex by way of the mobile anion effect (Johnson et al., 1980; Seip, 1980). Two factors are consid-
ered in evaluating this hypothesis.  First, the average base status of the soil exchange complex repre-
sents a balance among the various supply and demand processes in the ecosystem.  For example,
under steady-state conditions, weathering should supply sufficient base cations to meet the demands
of vegetative uptake while maintaining soil solution concentrations in equilibrium with the soil
exchange complex.  Perturbations to the system, such as changed deposition, will alter this steady-
state condition.
                                             195
-------
      Second, charge balance requirements need to be maintained between the soil exchange com-
plex and soil solutions. Maintaining charge balance, coupled with the increased anion loads provided
by acidic deposition, requires that total (acid plus base) cation concentrations in soil solutions
increase.  The ratio of the base to acid cations will not change dramatically, however, at least during
the initial stages of leaching.  The higher concentrations of base cations in soil solutions lead to a net
depletion of base cations from the exchange complex. If this increased leaching is not matched by an
increased level of supply (e.g., from weathering), then the overall effect will be a net depletion of the
base cations from the exchange complex.

9.3.1.1.4  Cation resupply is slow —

      The fourth hypothesis, that the rates  of cation resupply to the soil exchange complex are slow
relative to the rates of base cation removal, is  not being tested directly in this study. Rather, the
hypothesis is being subsumed in the models as an assumption that exchange reactions provide suffi-
cient buffering for resupply rates  not to be  an  issue for the time scales of concern to the study. As
such, mineral weathering rates have been defined as zero for these Level II simulations. Simulations
that include batch weathering and exchange processes are presented and discussed in Section 10.

      One reason for using this approach is that,  with current technology, no definitive method  exists
for distinguishing the different sources of base cations to surface waters.  Therefore, by assuming that
all base cations are derived from exchange sites,  simulations yield, effectively, a "worst case" scenario
for the depletion of the soil buffering capacity. If, under these circumstances, the results suggest an
extensive capacity of the soil to buffer  against the effects of acidic deposition, then the resupply rate is
not an issue of importance in this study.

9.3.1.1.5  Soil chemistry as an indicator of soil  response to acidic deposition -

      The final hypothesis is intended  to provide the groundwork for using selected soil properties as
qualitative indices of soil "condition" and expected soil response to acidic deposition. Recently,
several attempts have been made to correlate soil properties with the anticipated response of soils to
acidic deposition (VanLoon, 1984; Stuanes, 1984; Lau and Mainwaring, 1985).  Results from these
studies suggest that  soil properties are useful  indicators of how soils will respond to continued
exposure to present  or anticipated levels of acidic deposition.

9.3.1.2  Approach

      As discussed in Section 9.1, the approach used for Level II base cation analyses is model
based.  The primary  processes believed to regulate exchange processes are known, and models have
been developed that describe these processes with internal consistency.  As such, existing model
formulations are used extensively in conducting these studies.

      Data used in running the models were collected specifically for this study.  In collating the data
for use in the models, we made certain decisions regarding how data from individual soils and
watersheds  would be condensed, or aggregated, for use in the models.  Because the primary goal of

                                              196
-------
the DDRP is to make regionally representative projections about future changes in surface water
chemistry as a direct result of acidic deposition, we decided to aggregate the data, first into groups of
soil with similar chemical and physical characteristics, and then to the watershed level.  The models
were run on watershed level data. The results from these runs were then regionally weighted so that
inferences about the DDRP target population of systems could be made.

9.3.1.2.1  Off-the-shelf models --
      In designing the Level II base cation studies, one of the issues considered was the selection of
models.  We decided during the planning stages of DDRP to use only published, peer-reviewed, and
publicly available models.  A primary advantage of this decision was that the data requirements for
these models were known, thus field sampling programs could be developed to collect the appropriate
data. A second advantage was that minor modifications or improvements could be incorporated into
the model codes in a timely manner.  The selected models have been developed and described in
detail by Bloom and Grigal (1985), Reuss (1983), and  Reuss and Johnson (1985).

9.3.1.2.2 Aggregated soil chemistry data —

      Following selection of models for use in the Level II analyses, a major issue was preparation of
data for use in the models. Soil physical and chemical data were gathered for a regionally represen-
tative sampling of soils in the NE and SBRP. These data were obtained from individual pedons and
soil horizons. V/e transformed these data into  a form usable by the models, by aggregating the data
to produce information representative of whole watersheds.

      Details of the aggregation procedures have been presented by Johnson et al: (1990). Briefly,
the steps taken to produce the aggregated data depend on the structure of the model to be applied.
In general,  data are first averaged within the master horizons (i.e., O, A/E, B, or C horizons) of
individual soil sampling classes. Then, if required by the models  (e.g., those that describe the soil as
a single "box"), results from the master horizons are averaged to yield a  single value for each
parameter,  which represents the sampling class as a whole.  The procedures used to average soil
chemical and physical properties at the horizon and sampling class levels varied slightly in accor-
dance with the model for which the data were being developed. For models that use capacity varia-
bles as inputs (e.g., the  Bloom-Grigal model), soil properties were averaged using mass weighting
procedures. For models using intensity variables  as inputs, an intensity weighting scheme (Johnson
et al., 1990) was developed that preferentially weighted the lowest subhorizon in generating values for
master horizons, and then  employed straight numerical averages to produce sample class/pedon data.

      Finally, data from individual sampling classes were weighted, using areal weighting, to produce
soils data representative of the watershed as a whole.  The weighting used in this last aggregation
step was proportional to the relative areal occurrence of the sampling class on a particular watershed.
The weighting effectively assumes that long-term changes in runoff chemistry are controlled by the
extent of soil occurrence on a watershed, not by the location of soils on  that watershed.  For example,
although it  might be argued that riparian  soils have a greater influence on the composition of surface
waters than do ridge-crest  soils, riparian zone soils and those soils immediately adjacent to the lakes

                                             197
-------
are not preferentially weighted relative to upland soils. The decision to use the uniform weighting
approach was based primarily on the difficulty of developing uniform, broadly based algorithms to
apply preferential weighting to specific soils based on geomorphic considerations.

9.3.1.2.3 Scale of temporal forecasts-

      Another decision to be made in implementing the Level II analyses was the time scale over
which the model simulations would be run. In the near term, dramatic, permanent changes to surface
water composition are not expected to occur on annual time scales.  Acidic deposition has affected
parts of eastern North America for at least several decades. As such, changes in soil and runoff have
probably already occurred in many soils and  surface waters; some rapidly changing systems might
have already changed substantially and are near steady state with current high levels of acidic
deposition, whereas soils with relatively  large base cation pools might be in incipient stages of
response, and would be expected to change only slowly.

      For long time-scale projections, the major factor determining the duration of simulations to be
run is the uncertainty associated with the major parts of the modelling efforts. As soil composition and
properties evolve with continued exposure to acidic deposition, the response of these soils is also
expected to change. We anticipate that, for longer time scales, projected changes will become more
dramatic.  The larger changes  are balanced,  however, by increases in the uncertainty of the analyses
for periods exceeding, e.g., 50 years.

      Using these procedures  as guides for bounding the time intervals to be modelled, model
simulations were run for 100 years. The results are presented as projected changes after 20, 50, and
100 years.  The 20- and 50-year projections provide information about relatively near-term changes
that might  be anticipated and are relevant time frames with regard to the potential implementation of
regulatory  controls.

9.3.2 The Bloom-Grigal Mode!

      The Bloom-Grigal soil acidification model (Bloom and Grigal, 1985) is a simple mass action
(nonequilibrium) exchange reaction model. The amount of acidity in deposition displaces an equiva-
lent amount of base cations and/or aluminum from the soil. The model then assumes that these base
cations are completely mobile  and are removed from the system, causing a decrease in the overall
base saturation of the soil. Bloom and  Grigal (1985) use a semi-empirical equation to relate soil base
saturation to soil pH.  Soil pH is then used to determine the activity of AI3+ in soil solution. The
purpose of this analysis is to make a model-based analysis of the effects of the potential long-term
acidic deposition on the soils in the M-APP Region and to extrapolate these results to make regional
projections.

9.3.2.1  Model Formulation

      The  Bloom-Grigal model is formulated  around the assumption that, in steady-state ecosystems,
acidic deposition depletes base cations from the soil ion exchange complex. The simplicity of the

                                              198
-------
 model lies in the fact that soils are treated as a single homogeneous unit or compartment and all
 incoming deposition reacts completely with the soil in that compartment. The simplicity of the Bloom-
 Grigal model makes it a useful tool for assessing the effects of acidic deposition on forested soils for a
 variety of deposition scenarios and for long periods of time.

      The Bloorn-Grigal model assumes that acidity in deposition reacts completely with the soil. In
 other words, the model makes no provision for deposition to be routed around the soil and directly
 into the surface water or into the subsoil strata.  The amount of exchangeable base cations removed
 from the soil compartment is calculated as the difference between the input acidity and the output of
 H+ and A!3+, corrected for the protonation of bicarbonate.  The amount of base cations lost is sub-
 tracted from the pool of exchangeable base cations and a new base saturation is calculated. The
 BIoom-Grigal model then calculates a new soil pH based on an equation that relates soil pH to base
 saturation.  After adjusting parameters, the model then simulates the next year of deposition (Figure
 9-12).

      The Bloorn-Grigal model estimates the loss of base cations on an annual basis using the
 following equation:

                                         S = I - A - C                                    (9-3)

 where S is the sum of base cations, I is the amount of effective acidity in deposition, A is the acid
 leached from the soil, and C is the correction factor for the decrease in acidity due to protonation of
 bicarbonate.

      This model was created to assess the effect of acidic deposition on soils that do  not adsorb
 sulfate. Soils that adsorb sulfate have lower anion flux through the soil, and thus lower base cation
 removal rates (at a given level of deposition) than nonadsorbing soils.  In this regard, the BIoom-Grigal
 model will probably overestimate the rates of base cation removal from the soils  in the M-APP Region,
 because many soils of the region are adsorbing and retaining sulfur.  Retention in watersheds in the
 M-APP ranges from -21 to 74 percent, with median retention of 44 percent  (Section  7.9.2).

     The BIoom-Grigal model incorporates the input of nitrogen in deposition.  Because forested soils
 are generally deficient in available nitrogen, inorganic nitrogen in deposition is  removed by plants and
 organisms in the soil (Bloom and Grigal,  1985). When plants assimilate nitrogen in the form of nitrate
 (NO3"), they release hydroxyls (OH") to the soil, which is a nonacidifying reaction.  When plants assimi-
 late nitrogen as ammonium (NH4+), however, they release protons (H+), thus ammonium uptake  is an
 acidifying  reaction. The biological oxidation of NH4+ to NO3' produces two H+ ions for every mole-
 cule of NH4+ oxidized.  The BIoom-Grigal model incorporates these processes in calculating the net  or
 effective acidity of deposition.

     The original versions of the BIoom-Grigal'model were coded  in FORTRAN and BASIC; the
version used in this analysis is coded in a high speed, compilable form of BASIC. In addition, code
 has been added that allows the input data to be processed  in a batch mode.  The output is formatted
to magnetic media to simplify the data reduction process. Code also has been added to adjust

                                             199
-------
                                    '•• 'i', ss; ;
                                     v-  .  --•;- -   --',--'  -'.
                                     Record watershed
                                      soil pH, base saturation
                                      Calculate protonation
                                         of bicarbonate
                                     Calculate loss of bases
                                        Calculate new base
                                       saturation, soil pH and
                                         soil solution AI3+
                                                                                    Deposition
                                                                                  Precip/runoff
                                                                               Soil chemistry
Modelling
 dataset
Figure 9-12.   Flow diagram for the one-box Bloom-Grigal soil simulation model.
                                                     200
-------
deposition for the nonconstant deposition scenarios as the model is running.  The fundamental
equations in the original model have not been altered, however.

9.3.2.2 Assumptions

      We make a number of assumptions in modelling the effect of acidic deposition on soils with the
Bloom-Grigal model.  Some are implicit to the model, others meet the needs of our current applica-
tion.  The assumptions used in implementation of the Bloom-Grigal model are itemized  below with
explanatory discussion or comments.

9.3.2.2.1  Sulfate adsorption --

      The Bloorn-Grigal model assumes that sulfate is not adsorbed by the soil and  acts as a com-
pletely mobile anion.  As  mentioned previously, in soils that have net sulfate adsorption, this
assumption may lead to an overestimation of the amount of base cations actually leached from the
soil.  If the sulfate adsorption/leaching information from the preceding section were integrated into or
with the Bloom-Grigal model, the effect of sulfate adsorption on base cation leaching could be
assessed.  These two analyses are not coupled in the Level II analyses,  but comparable integration is
used  with the  Level III integrated watershed modelling analyses (Section 10). Consequently, the
analysis presented here represents a worst-case scenario of base cation removal from soils as a result
of acidic deposition.

9.3.2.2.2  Input acidity--

      The total effective acidity (H+tota|) in deposition was defined by Bloom and Grigal, and is
similarly computed in our analyses, as:
                                                                                          (9-4)
Briefly, forested ecosystems are generally nitrogen limited.  Therefore, nitrogen deposition in these
systems can contribute to the total system proton budget (see Section 9.3.2.1 above). Equation 9-4
takes this into account.  Bloom and Grigal (1985) provide a complete discussion of the rationale for
this definition of input acidity.

9.3.2.2.3  Extent of reaction -
      For this analysis, it is assumed that the input acidity, the effective acidity, reacts completely with
the soil.  The acidity is partitioned between solubilizing Al and replacing exchangeable base cations.
This represents a worst-case scenario of base cation removal from soils as a result of acidic deposition.

9.3.2.2.4 Depth of soil --

      The depth of reactive soil material  equals the mean aggregated thickness of the soil sampling
classes represented by the types of soils on the specific watersheds. In their original paper, Bloom

                                             201
-------
and Grigal (1985) considered the effects of deposition only within the rooting zone of soils (the top
25 cm). We consider the effect, however, on the whole soil compartment.  Our soil chemistry input
data are aggregated  to represent the central tendency of the soil chemical characteristics of the whole
soil compartment.  The effect of acidic inputs on data aggregated in this way thus represents a mean
effect.  Actual effects are likely to be greater in some parts of the soil profile than  in others.

9.3.2.2.5 Volume of drainage water -

     The volume of  water moving though the watersheds in each year of simulation is equal to the
long-term annual average runoff. Other than the assumption that water moves vertically in this
analysis, no other assumptions are made regarding watershed hydrologic flowpaths.

9.3.2.2.6 Partial pressure of soil CO2 -

     The partial pressure of ambient CO2 is approximately 0.0003 atm.  Soil air is, however, enriched
with CO2 due to biological respiration and is consequently elevated.  In all  of the  Bloom-Grigal model
runs, the partial pressure of CO2 in the soil air is set at 0.005 atm, a value thought to be reasonable
for forested soils (Fernandez and Kosian, 1987).

9.3.2.2.7 Activity of soil solution AI3+ »

     To calculate the amount of input acidity that is converted to output acidity by aluminum
buffering, the activity of Al   in soil solutions is calculated using the following equation:

                                log(A!3+) = 2.60 - 1.66 * soil pH
(9-5)
This equation in the Bloom-Grigal model represents an empirical relationship based on soil chemical
data for the Upper Midwest. In developing their model, Bloom and Grigal (1985) had a fundamental
                                                                 ,3+
problem in using the solubility of AI(OH)3 to describe the variation in Ar+ with pH. They state that in
very acidic soils, such as forested soils, Al   is undersaturated with respect to the solubility of AI(OH)3.
Therefore, AI(OH)3 solubility was regarded as a poor model for the pH-A!3+ relationship.  To establish
a more realistic relationship between AI3+ and pH, they used laboratory measurements of AI3+ in arti-
ficially acidified soils to define an empirical relationship between pH and Al3+.

9.3.2.2.8  Relating soil solution pH to base saturation --
      The pH of soil solutions is related to base saturation (BS) by the following equation:

                               pH = pKa + n * log [BS/(1 - BS)]
 (9-6)
where pKa is the apparent acidity constant for soil (i.e., aggregate watershed/soil compartment) and n
is an empirical constant. This equation is an extended form of the Henderson-Hasselbach equation.
                                              202
-------
      The Bloom-Grigal model used here calculates pKa and n for each watershed using the initial
aggregated values of soil pH and base saturation. These parameters describe the relationship
between soil pH and base saturation and are unique for each watershed.

9.3.2.2.9  Base cation uptake ~

      The model assumes no net accretion of base cations in biomass. The uptake of base cations
by forest vegetation is an acidifying process by which H+ is exchanged for an equivalent amount of
base cations to maintain charge neutrality.  At the same time, through litterfall and decomposition,
base cations are released to soils.  The Bloom-Grigal model tracks only the flux of base cations
leached from the soil. This no-accretion assumption implies that the uptake of base cations by vege-
tation is exactly equal to the amount recycled to the soil (i.e., steady state).  The model does not take
into account the potential loss of base cations by erosion or chemical complexation followed by leach-
ing reactions.  Because most of the forested systems considered by the DDRP are second-growth
systems still in an aggrading phase, vegetation is probably a significant cation sink in most water-
sheds, and the assumption of no net accretion underestimates potential depletion of soil base cations.

9.3.2.2.10  Mineral weathering —

      Mineral weathering  is the major ultimate source of base cations to soils and runoff. The original
Bloom-Grigal model code had a subroutine that calculated the contribution of base cations to the soil
solution via mineral weathering. The rate of mineral weathering for the simulations presented here is
set to zero for three reasons.  First, by assuming no base cation resupply, a "worst-case" base cation
loss  scenario is evaluated, thereby bounding the projections.  Second, the  relationships between
weathering and soil solution pH are not sufficiently established to provide accurate parameters for the
weathering equations.  One complication, in particular, is that  mineral weathering  rates are a dynamic
function of the chemical weathering environment. Third, the use of this subroutine of the Bloom-Grigal
model requires an  initial estimate of the mineral weathering rate.  These estimates are not available at
this time.

9.3.2.2.11  Cation  exchange  capacity -

      The cation exchange capacity (CEC) of soils is held constant throughout the period of simula-
tion, although this approximation  is not correct in soils if the soil pH changes.  Soil CEC is  derived
from two sources: (1) secondary clay minerals with permanent charge due  to isomorphous substitution
of lower valent cations for cations in the clay crystal lattice, and (2) variable charge sites on organic
matter, para- and noncrystalline hydrous oxides, and edge sites on permanently charged clays.  The
variable charge CEC is a function of pH (i.e., the net soil CEC changes as with changes in pH).  As
pH increases, the variable charge CEC increases, and vice versa.  Because of scientific and data
limitations, we have, however, chosen to hold CEC constant.
                                             203
-------
9.3.2.2.12  Time steps —

      The time step for simulations is annual.  For assessment purposes, yearly time steps are a use-
ful increment.  From a modelling standpoint, any shorter time step (e.g., daily) is data intensive and
computationally demanding. Shorter time steps may provide more accurate short-term projections
(e.g., interannual variability in soil chemistry), but for regional assessments,  longer time steps appear
to be more useful and require less data.

9.3.2.3  Limitations

      Soils are highly complex and no simulation  models exist that accurately depict the fluxes of
energy and matter in soil systems.  Models are based upon simplifying assumptions and are therefore
necessarily incomplete representations of reality.  As with any attempt to project future events, the
Bloom-Grigal soil simulation model is not without limitations. Some of the limitations are due to the
state of the science and others are have been imposed by the DDRP.

      The scientific limitations center around the factors that control aluminum solubility and the
relationship of soil  pH to base saturation.  Bloom and Grigal (1985) developed empirical equations  to
describe this relationship for a selected set of forested  soils in northeastern  Minnesota. As described
in their paper,  the equations appear to be appropriate for forested soils in Minnesota.  For the DDRP,
the relationships described by Bloom and Grigal are assumed to be widely applicable and were not
independently  verified.  It is doubtful, however, that these or any other single equation would be
universally true due to spatial differences in soils and vegetation.

      Soils are dynamic systems. Soil properties  fluctuate on a daily basis,  and daily temperature and
moisture changes affect a broad range of soil processes.  Broader seasonal changes also occur.  The
use of annual time steps assumes that soils are static,  possibly restricting the accuracy of the  pro-
jections.  As mentioned in Section 9.3.2.2.12, however,  shorter time steps are data intensive and
computationally restrictive.

      Individual soil processes are inextricably linked to a number of other processes and any consid-
eration of a single process (e.g., base cation flux) in isolation may distort projections.  In the DDRP
Level II analyses, processes are isolated in order to focus on the principal soil reactions associated
with surface water acidification.  Some of the uncertainly in assessing effects is due to this approach
of isolating facets of the whole ecosystem.

9.3.2.4  Model Inputs

      The Bloom-Grigal model was designed not to be data intensive. The data required to run the
Bloom-Grigal model fall into four categories: (1) deposition data, (2) precipitation data, (3) soil
chemistry data, and (4) fixed parameters.  The sources of the deposition data are described in Section
5. Table 9-7 lists the specific data required to run the Bloom-Grigal model.  The soil chemistry data
used in these simulations have been aggregated to a single compartment representing the watershed
                                             204
-------
Table 9-7. List of Input Data for the BIoom-Grigal Soil Acidification Model
Input Variables
Units
Annual average runoff



Annual H+, NH4+, NO3', and SO42" in wet deposition*1



Annual H+, NH4+, NO3", and SO42" in dry deposition3



Soil pHb



Sum of soil exchangeable base cations (0.1 M NH4CI)



Soil cation exchange capacity (0.1 M NH4CI)
cm



keq/ha



keq/ha



PH(H20)



keq/ha



keq/ha
Fixed Parameters
Value
Length of simulation



Partial pressure of soil CO2



Activity coefficient of AI3+



Activity coefficient of AI(OH)2+
100yr




0.005 atm




0.82




0.92
0 Wet and dry SO42" deposition are used to calculate alternative deposition scenarios.




b Determined using deionized water.
                                              205
-------
level.  These procedures are described in detail by Johnson et al. (1990). The capacity variables, sum
of exchangeable base cations (SOBC), and CEC are capacity weighted.  Soil pH is intensity weighted.

9.3.2.5  Model Outputs

      The Bloom-Grigal model simulates soil processes relevant to the assessment of effects of acidic
deposition on soils. During model simulation runs, soil pH, soil base cation status (i.e., base satura-
tion),  and soil solution AI3+ are tracked.  Each of these is presented in these analyses.

      During the 100-year simulations, soil pH, base saturation, and soil  solution AI3+ are reported for
years 0, 20, 50, and 100.  The projected changes are presented as cumulative distribution functions
(CDF) for graphical comparisons.  The CDFs represent regionally weighted projections for soils on the
target population of watersheds. Summary statistics for the CDFs also are presented for numerical
comparisons.

9.3.2.6  Data Sources

      The data required to run the Bloom-Grigal model include total annual wet and  dry deposition,
total annual runoff, and selected soil chemistry data.  These are listed in Table 9-7. All these data
were  collected as a part of the DDRP and are discussed in detail in  Section 5.

9.3.2.6.1  Deposition data --

      The deposition data in this analysis are  from two sources.  The wet deposition data are
measures of typical year annual loadings. The dry deposition data are estimated, because measures
of dry deposition do not exist for the M-APP.  The details on the acquisition/generation of the DDRP
deposition data sets are given in Section 5.6.  Because there is some uncertainty associated with
these estimates, particularly in the amount of  base  cations, we bracketed the dry base cation deposi-
tion data to look at the effect of including 100, 50, or 0 percent of the base cations in base case
estimates of dry deposition. The reductions in dry  base cations are offset by concomitant increases  in
dry H+ to maintain charge balance. These three deposition datasets are referred to as:  (1) Typical
Year  (TY), (2)  TY Reduced Base Cation (TY-rbc)-TY with 50 percent of the dry base cations, and (3)
TY Zero Base Cation (TY-zbc)-TY with no base cations in dry deposition.  A summary of the regionally
weighted median deposition inputs in the three deposition data sets (TY, TY-rbc, and TY-zbc) used in
the Bloom-Grigal modelling in the M-APP is presented in Table 9-8.

      Median  values, of total acid input (H+tota| = H+ + NH4+ - NO3') in the M-APP are considerably
 higher than in the NE and SBRP (compare Table 9-8 with Table 9-24, Church et. al., 1989).  Median
 acid  loadings are  at least twice those in the other regions.

 9.3.2.6.2 Deposition scenarios -

      Three scenarios of future deposition are used in this analysis.  They are identical to those
 described in Section 9.2.3.2, with the exception of the length of simulation. In lieu of the 140-year

                                              206
-------
Table 9-8.  Regionally-Weighted Median Values of Base Year Annual Acid Inputs for the Bloom-
            Grlgal Model for Watersheds In the M-APP Region, for Three Levels of Dry Base
            Cation Deposition3

TY°
TY-rbc
TY-zbc
H+ NH4+
1.38 0.27
1 .41 0.27
1.55 0.27
NO3" Total Acid lnputb
0.68 0.93
0.68 1 .03
0.68 1.13
a  Values are in keq/ha/yr


b  Total Acid Input = [H+ + NH4+ - NO3" ]


c  TY - typical year wet deposition data for dry deposition; TY-rbc = a 50 percent reduction in loading of base cations in
   dryfaN; TY-zfac = no base cations in dry deposition.
                                              207
-------
simulations described in Section 9.2.3.2,100-year simulations are used here. Constant deposition is
referred to as "Scenario A." A timed 50 percent reduction in wet and dry sulfate deposition is referred
to as "Scenario B." A timed 50 percent reduction in wet and dry sulfate deposition followed by a timed
30 percent increase is referred to as "Scenario C."  Charge  balance in the ramped scenarios is main-
tained by either increasing or decreasing proton deposition.

9.3.2.6.3 Soils data -

     The Bloom-Grigal model uses one value for the following soil chemistry variables to depict the
soil chemistry of a particular watershed: soil pH, CEC, and SOBC. To get results that represent the
central tendency of the DDRP regions, a large number of observations for these variables were aggre-
gated to obtain values for each of the DDRP watersheds.  Combining or aggregating these data can
be accomplished in several ways. It is probably not correct to use a simple average for all variables;
rather, capacity and intensity variables should be weighted  differently.  Of the variables used in the
Bloom-Grigal model simulations, soil pH was aggregated using an intensity variable aggregation
method, whereas CEC and SOBC were aggregated using a capacity variable aggregation method.
The details of these methods are provided in Johnson  et al. (1990). A summary of the regionally
weighted median values of the Bloom-Grigal soil chemistry  input data for the three DDRP regions is
presented in Table 9-9.

      Ranking the regional median values for base saturation, the NE is highest followed by the
M-APP and the SBRP. All other things being equal  (e.g., soil mass, total acid input, runoff), soil base
saturation provides one means of comparing the proton buffering potential of a soil and the potential
duration of delays in response to acidic deposition.  The M-APP, therefore, appears to be intermediate
in buffering potential. We know, however, that this region receives a much larger acid load than the
other two regions. The median soil pH values are low in all three regions, as reflected in the estimates
of AI3+.
9.3.2.7 Results

      This section presents results of the Level II base cation analyses using Bloom and Grigal's
(1985) cation depletion model.  Following discussions about the input data, the projected effects of
acidic deposition on soil base saturation, soil pH, and soil solution AI3+ are presented.  Following
some discussion of these results, comparisons are made to similar projections for the NE and SBRP.
Results for the NE and SBRP are summarized from Church et al. (1989, Section 9.3).

      The DDRP was specifically designed to make projections  of the regional effects of acidic deposi-
tion.  In the DDRP, the basic unit of investigation is the watershed.  Instead of characterizing the
effects of acidic deposition on individual soil pedons, the research focus has been the integrated effect
of acidic deposition on watersheds, using a set of randomly selected watersheds that are regionally
representative. As a result, all of the Bloom-Grigal modelling input data and simulation results are at
the watershed level. Because the DDRP sample of watersheds  serves as the basic link to the target
population of watersheds, watershed level  results are extrapolated to the target population of
watersheds.  The statistical design and methods used in the DDRP are described in Section 6.

                                              208
-------
Table 9-9.  Regionally-Weighted Median Values of Annual Initial Soil Chemical Values of the
            Level II Bloom-Grigal Modelling for the M-APP, NE, and SBRPa
Region
M-APP
NE
SBRP
PH
4.64
4.62
4.85
SOBC
130
40.0
40.4
CEC
958
184
433
BS
13.8
22.0
9.2
AI3+
9.5
10.3
4.3
  pH = intensity weighted soil pH; SOBC = mass weighted sum of exchangeable base cations in keq/ha; CEC = mass
  weighted cation exchange capacity in keq/ha; BS = base saturation [(SOBC/CEC)*100], in percent; AI3+ = calculated based
  upon initial soil conditions in/imol/L
                                               209
-------
9.3.2.7.1  Projections of the effects of acidic deposition on percent base saturation --

      The results of the model projections for the effects of acidic deposition on soil base saturation
are presented graphically in Figure 9-13 and in tabular form in fable 9-10 for the three deposition
scenarios.  Both Figure 9-13 and Table 9-10 present regionally weighted results for years 0, 20, 50,
and 100 in the simulations.  Under current deposition, the regional estimate of percent base saturation
drops from 13.8 percent to 9.4 percent in 100 years.  Most of this change is projected to occur within
the first 50 years. This may be indicative of shift in the soil buffer from cation exchange to the
solubilization of aluminum-oxyhydroxides, as suggested by Ulrich (1983). This  is considered more
fully in Section 9.3.3.2.                                                            -

      The effect of the alternate deposition  scenarios is clearly seen  in Figure 9-13. The middle graph
shows the effect of a ramped 50 percent reduction in sulfate deposition on  soil base saturation.  The
net effect  is seen by the close grouping of the CDF curves.  There is little change in soil base satura-
tion after 100 years with deposition scenario B.  The change in median base saturation is projected to
be less than 0.7 percent.  Deposition scenario C, however, produces a somewhat larger change in
base saturation (the regional median soil  base saturation decreased  by approximately 2 percent after
100 years), but is less than that projected for constant deposition.

      The effects of the two lower concentrations of dry base cations are reported  in Tables 9-11 and
9-12.  The largest change in base saturation is for the constant deposition scenario.  With the TY-rbc
(dry base cations reduced by 50 percent) deposition  dataset, base saturations  are projected to change
from 13.8 percent to 8.6 percent (Table 9-11), a slightly larger decrease than the final median base
saturation of 9.5 percent  for the TY base scenarios. With the TY-zbc (no dry base  cation deposition),
the effects are even more dramatic (Table 9-12). The regional median base saturation  is decreased by
about half (12.75 percent to 6.7 percent). This represents a worst case, because we know that "there
is some component of base cations in  dry deposition. The amount,  however, is uncertain.

9.3.2.7.2  Projections of the effects of acidic deposition on soil pH —

      Soil pH  is a master variable in that it either controls or  strongly influences the solubility of soil
minerals, the composition of soil solutions and the cation exchange complex, and soil biochemistry.
Soils are generally well-buffered systems -  buffered against changes in pH - due primarily to the
complex nature of soils, which are a mixture of inorganic and organic phases,  along with water.

      In assessing the effects of acidic deposition on soils, the acid titration of a mixed buffer is a
good conceptual model of the processes involved. As with any  buffered system, the buffering
capacity of the system can be exceeded, but because soils are  effectively a mixture of several buffers,
as the capacity of one buffer becomes increasingly depleted, the pH of the soil will drop only slightly
as a second buffer takes over. As the acid loading proceeds, the succession of buffer systems
continues. At some point, however, the pH of the system can be reduced to a level at which
dissolution of Al and other metals becomes the primary buffer, and concentrations of these metals
increase to levels that inhibit or are lethal to terrestrial vegetation and aquatic biota.
                                              210
-------
                                 SCENARIO A
                      1.0
                    X
                    VI
                    o
                    fc
                    2
                    g
                    CL
0.8-




0.6-




0.4-




0.2"
                      0.0
  10
                               15     20      25      30

                                PERCENT BASE SATURATION
                                                            35
                      1.0
                                 SCENARIO B
                    x  -
                    VI


                    §

                    cc
                    O 0.4-
                    Q.

                    g
                    ^ 0.2 H
                      0.0
  10
                               15     20      25      30

                                PERCENT BASE SATURATION
                                                            35
                                 SCENARIO C
                               15     20     25      30

                                PERCENT BASE SATURATION
                                                            35
Figure 9-13.  Regional CDFs of BIoom-Griga! model projected base saturations of soils on
            the DDRP target population of M-APP stream watersheds for three deposition
            scenarios (A, B, and C) at 0, 20, 50, and 100 years.
                                       211
-------
Table 9-10. Bloom-Grigal Model Regional Projections of Percent Base Saturation for Soils in the
            M-APP Region as a Function of Time and Deposition Scenario9
YEAR
YEAR
YEAR
                      Deposition Scenario = A

MEAN   STD DEV    MIN      P 25     MEDIAN   P 75
                      Deposition Scenario = B

MEAN    STD DEV    MIN      P 25    MEDIAN  P 75
                      Deposition Scenario = C

MEAN    STD  DEV     MIN     P  25    MEDIAN   P  75
                                                                           MAX
0
20
50
100
16.5
15.5
14.0
12.1
6.0
6.1
6.2
6.5
10.1
9.5
8.2
5.6
12.7
11.9
10.0
8.0
13.8
12.4
10.8
9.4
18.3
18.2
15.4
14.9
34.9
34.4
33.6
32.4
MAX
0
20
50
100
16.5
16.0
15.9
15.8
6.0
6.0
6.0
6.0
10.1
9.9
9.9
9.9
12.7
12.3
12.2
12.2
13.8
13.2
13.2
13.2
18.3
18.3
18.3
18.3
34.9
34.7
34.7
34.7
MAX
0
20
50
100
16.5
16.0
15.6
14.4
6.0
6.0
6.1
6.3
10.1
9.9
9.9
7.7
12.7
12.3
11.8
10.7
13.8
13.2
12.8
11.5
18.3
18.3
18.3
16.3
34.9
34.7
34.6
34.1
    Projections made using typical year deposition, including a full complement of base cations in dry deposition, and three
    deposition scenarios. Results reported for 20-, 50-, and 100-year projections. The data for year 0 represent initial soil
    conditions.
                                              212
-------
Table 9*11.  Bloom-Grigal Model Regional Projections of Percent Base Saturation for Soils in
             the M-APP Region as a Function of Time and Deposition Scenario and for a
             Scenario of a 50 Percent Reduction in Base Cation Dry Deposition8
YEAR
YEAR
YEAR
MEAN
            Deposition Scenario = A

STD DEV   MIN      P 25   MEDIAN   P 75
                      Deposition Scenario = B

MEAN   STD -DEV    MIN      P 25    MEDIAN    P 75
MAX
0
20
50
100
16.5
15.2
13.3
11.1
6.0
6.0
6.1
6.2
10.1
9.3
7.8
5.2
12.7
11.7
9.7
7.4
13.8
12.3
10.4
8.6
18.3
17.5
14.3
13.4
34.9
34.1
33.0
31.1
                                                   MAX
0
20
50
100
16.5
15.8
15.6
15.3
6.0
6.0
6.0
6.0
10.1
9.8
9.8
9.4
12.7
12.2
12.0
11.6
13.8
13.0
12.9
12.4
18.3
18.0
18.0
17.7
34.9
34.5
34.5
34.4
                      Deposition Scenario = C

MEAN    STD DEV    MIN      P 25    MEDIAN   P 75      MAX
0
20
50
100
16.5
15.8
15.2
13.5
6.0
6.0
6.0
6.1
10.1
9.8
9.7
7.1
12.7
12.2
11.3
9.5
13.8
- 13.0
12.1
10.7
18.3
18.0
17.5
14.9
34.9
34.5
34.1
32.9
   Projections made using typical year deposition, with three scenarios of future deposition. Results reported
   for 20-, 50-, and 100-year projections. The data for year 0 represent initial soil conditions.
                                             213
-------
Table 9-12. Bloom-Grigal Model Regional Projections of Percent Base Saturation for Soils in the
            M-APP Region as a Function of Time and Deposition and for a Scenario of No Base
            Cations in Dry Deposition8
YEAR
YEAR
 YEAR
MEAN
            Deposition Scenario = A

STD DEV   MIN      P  25    MEDIAN   P_75
                                                                           MAX
0
20
50
100
16.5
14.9
12.7
10.1
6.0
6.0
6.0
5.9
10.1
9.2
7.5
4.8
12.7
11.4
8.8
6.8
13.8
12.1
10.0
7.7
18.3
16.8
13.3
12.1
34.9
33.9
32.3
29.8
                      Deposition Scenario = B

MEAN    STD DEV    MIN      P 25    MEDIAN   P 75       MAX
0
20
50
100
16.5
15.6
15.2
14.7
6.0
6.0
5.9
5.8
10.1
9.7
9.7
8.8
12.7
12.1
11.7
11.0
13.8
12.7
12.2
12.2
18.3
17.7
17.7
16.2
34.9
34.2
33.8
33.1
MEAN
             Deposition Scenario = C

STD DEV    MIN      P 25    MEDIAN    P 75
                                                                           MAX
0
20
50
100
16.5
15.6
14.2
12.5
6.0
6.0
5.9
5.9
10.1
9.7
9.3
6.6
12.7
12.1
10.9
8.8
13.8
12.7
11.6
9.9
18.3
17.7
16.4 .
14.2
34.9
34.2
33.4
31.6
    Projections made using typical year deposition, with three scenarios of future deposition. Results reported for 20-, 50-, and
    100-year projections. The data for year 0 represent initial soil conditions.
                                              214
-------
      The Bloorn-Grigal modelling results of the effects of acidic deposition on soil solution pH in the
 M-APP uplands are summarized in Figure 9-14 and Tables 9-13, 9-14, and 9-15. The soil pH levels in
 the M-APP are initially low from an agricultural point of view, but are not, however, exceptionally low
 for forested soils. The aggregated soil pH values for the target population of watershed systems in the
 M-APP ranged from about 4.5 to 5.3 (Table 9-13).  In developing the aggregate soil pH values, the
 organic horizons were included. Because organic horizons are often  the most acidic horizons in
 forested soils, their contribution to the aggregate pH  is reflected in these low regional values.

      Under constant deposition (Scenario A) the Bloom-Grigal model projects that the median soil pH
 in the M-APP will decrease by 0.2 pH units over the next 100 years (Table 9-13). The systems with a
 current pH above 4.7 (approximately 10 percent of the systems in the target population; Figure 9-14)
 appear to be strongly buffered against change, and change only slightly if at all. Approximately 90
 percent of the systems (those with initial aggregate soil pH values less than 4.7) are expected to show
 decreases in soil pH under constant deposition at current levels.  Although the projected changes in
 pH are not large, they represent significant increases in H+ concentration in the soil solution, and
 result in significant increases in soil solution  aluminum. The relatively small changes in pH, compared
 to those for exchangeable base cations and  soil solution aluminum, in part reflect the reporting of a
 logarithmic variable, but they result mainly from (1) selectivity coefficients that favor exchange of H+
 from  liquid to solid  phase (with release of base cation from the exchange complex) and (2)  buffering of
 pH changes by dissolution of aluminum. The changes in pH and Al could have significant adverse
 consequences, both for terrestrial vegetation and for biota of receiving surface waters.

      The effect on model projections for pH of using the alternate dry base cation deposition scen-
 arios  is small. For deposition scenario A (constant deposition at current loadings),  use of the TY-rbc
 (Table 9-14) and TY-zbc (Table 9-15) deposition data results in only slightly lower projected soil pH
 values than straight TY data (Table 9-13).  Decreasing the base cations in dry deposition results in
 increased acid  loadings to the soil because H+ is increased to maintain change  balance. As a result,
 the rate of change in base saturation is slightly faster than in the base case,  but there is almost no
 effect of soil pH, because pH is buffered by the soil (Equation  9-6).  Thus the Bloom-Grigal model
 projections in almost all cases include much  larger changes in base cations than in pH, and model
 projections are relatively insensitive to small changes  (such as changes in dry base cation deposition)
 in base cation flux.  Figure 9-15 shows the relationship between-soil pH and base saturation for typical
 values of n and pKa. The flat part of the curve represents conditions under which the soil is buffered
 against changes in  soil pH, such that even large changes in base saturation are associated with minor
 changes in soil pH. The steeper portions of the curve, where the slope is greatest,  are the condition
 of the soil titration curve, where small changes  in  soil  base saturation result in large changes in soil
 pH. This represents a shift from one soil pH  buffering mechanism to another, such as a shift from a
 cation exchange buffer to  an aluminum buffer (Ulrich, 1983).

      As with the soil base saturation projections  reported in the foregoing discussion, the model-
 based projections of change in soil pH under decreased  deposition (Scenario B) result in only slight
changes in soil  pH (Table 9-13 and middle graph on Figure 9-14).  Even though the projected pH
change after 100 years under Scenario B is small, most of it occurs in the first 20 years of the
                                             215
-------
                               SCENARIO A



X
VI

~z.
o

cc
o
CL
O
cc
CL






1.0-

0.8-



0.6-



0.4-



0.2-




o.o -
4.


I
1

^
*
*
•
t
1
1
!
,' /
i i
i / /
' i
I Jf*

'" 1
* — ' /
i i
DO 4.25 4.5(
..^^^^^^f^ff^'
• ft*
/ 1 '"•
1 .'
1 .
1 •
i;

i .'
/ •
i t
•
/ * ----- 0 Year

i* 	 20 Years
50 Year"

	 . "jQQ Years


i i i
) 4.75 5.00 5.25 5.!
                                      SOIL pH

                               SCENARIO B


X
VI

z
o
1-
cc
o
CL
o
cc
CL





i.o-
0.8-



0.6-



0.4-



0.2-



0.0"
4.

/T"
!•'
y •
H "

r
yf
|r
/*
V*
f f
/
J
/
/•


DO 4.25 4.50 4.75 5
.^— -ir^*^











	 20 Years
50 Year"
1 00 Years



00 5.25 5.


















50
                                      SOIL pH

                               SCENARIO C


X
VI
•z.
O
o
0-
o
cc
Q.





1.0-
0.8-


0.6-

0.4-


0.2"



0.0 "
4.
,.^£5***""
jjT
/ // »
/ /i *
/ /i."
;' '•*
|/n
(1 1 	 OYear
IV
> 	 20 Years
' / 50 Year*"
'r 1Afl Yea re
/^ — — — 	 luU I tJafo
,-

DO 4.25 4.50 4.75 5.00 5.25 5.!
                                      SOIL pH
Figure 9-14.  Regional CDFs of Bloom-Grigal model projected pH of soils on the DDRP
            target population of M-APP stream watersheds for three deposition scenarios
            (A, B, and C) at 0, 20, 50, and 100 years, using deposition data for the full
            complement of base cations in dry deposition.
                                     216
-------
Table 9-13. Bloom-Grigal Model Regional Projections of Soil Solution pH for Soils in the M-APP
            Region as a Function of Time and Deposition Scenario for a Scenario of a Full
            Complement of Dry Base cations in Dry Deposition8
YEAR
YEAR
MEAN
            Deposition Scenario = A

STD DEV   MIN     P 25    MEDIAN    P 75
MAX
0
20
50
100
4.7
4.6
4.6
4.5
0.2
0.2
0.2
0.2
4.5
4.5
4.4
4.2
4.6
4.7
4.5
4.4
4.6
4.6
4.5
4.4
4.7
4.6
4.6
4.5
5.3
5.3
5.3
5.2
                                   Deposition Scenario = B

YEAR         MEAN    STD DEV   MIN     P 25    MEDIAN   P 75      MAX
0
20
50
100
4.7
4.6
4.6
4.6
0.2
0.2
0.2
0.2
4.5"
4.5
4.5
4.4
4.6
4.6
4.6
4.6
4.7
4.6
4.6
4.5
4.7
4.7
4.6
4.6
5.3
5.3
5.3
5.3
                     Deposition Scenario = C

MEAN    STD DEV   MIN      P  25    MEDIAN   P 75       MAX
0
20
50
100
4.7
4.6
4.6
4.6
0.2
0.2
0.2
0.2
4.5
4.5
4.5
4.4
4.6
4.6
4.6
4.5
4.6
- 4.6
4.6
4.5
4.7
4.7
4.6
4.6
5.3
5.3
5.3
5.3
0  Projections made using typical year deposition, with three deposition scenarios. Results reported for 20-,
   50% and 100-year projections. The data for year 0 represent initial soil conditions.
                                             217
-------
Table 9-14.  Bloom-Grigal Model Regional Projections of Soil Solution pH for Soils in the M-APP
             Region as a Function of Time and Deposition Scenario and for a Scenario of a 50
             Percent Reduction in Base Cations in Dry Deposition8
YEAR
YEAR
YEAR
MEAN
            Deposition Scenario = A

STD DEV   MIN      P  25    MEDIAN   P 75
MEAN
            Deposition Scenario = C

STD DEV   MIN      P  25    MEDIAN   P  75
MAX
0
20
50
100
4.7
4.6
4.6
4.5
0.2
0.2
0.2
0.2
4.5
4.5
4.4
4.2
4.6
4.6
4.5
4.4
4.6
4.6
4.5
4.4
4.7
4.6
4.5
4.5
5.3
5.3
5.2
5.2
                      Deposition Scenario = B

MEAN   STD DEV    MIN      P 25    MEDIAN    P 75      MAX
0
20
50
100
4.5
4.6
4.6
4.6
0.2
0.2
0.2
0.2
4.5
4.5
4.5
4.5
4.6
4.6
4.6
4.5
4.6
4.6
4.6
4.6
4.7
4.7
4.6
4.6
5.3
5.3
5.3
5.3
MAX
0
20
50
100
4.7
4.6
4.6
4.6
0.2
0.2
0.2
0.2
4.5
4.5
4.5
4.4
4.6
4.6
4.5
4.5
4.6
4.6
4.6
4.5
4.7
4.7
4.6
4.6
5.3
5.3
5.3
5.2
a  Projections made using typical year deposition, with three deposition scenarios. Results reported for 20-,
   50-, and 100-year projections.  The data for year 0 represent initial soil conditions.
                                             218
-------
Table 9-15. BIoom-Grigal Model Regional Projections of Soil Solution pH for Soils in the M-APP
            Region as a Function of Time and Deposition Scenario and for a Scenario of Zero
            Base Cations in Dry Deposition3
YEAR
YEAR
MEAN
            Deposition Scenario = A

STD DEV   MIN      P 25    MEDIAN   P 75
MEAN
            Deposition Scenario = C

STD DEV   MIN     P 25    MEDIAN   P 75
MAX
0
20
50
100
4.7
4.6
4.5
4.4
0.2
0.2
0.2
0.2
4.5
4.4
4.4
4.2
4.6
4.5
4.4
4.3
4.6
4.6
4.5
4.4
4.7
4.6
4.5
4.4
5.3
5.3
5.2
5.2
                                   Deposition Scenario = B

YEAR         MEAN    STD DEV   MIN      P 25   MEDIAN   P 75       MAX
0
20
50
100
4.7
4.6
4.6
4.6
0.2
0.2
0.2
0.2
4.5
4.5
4.5
4.4
4.6
4.6
4.5
4.5
4.6-
4.6
4.6
4.6
4.7
4.7
4.6
4.6
5.3
5.3
5.3
5.2
MAX
0
20
50
100
4.7
4.6
4.6
4.5
0.2
0.2
0.2
0.2
4.5
4.5
4.5
4.3
4.6
4.6
4.5
4.4
4.6
4.6
4.6
4.5
4.7
4.7
4.6
4.5
5.3
5.3
5.3
5.2
    Projections made using typical year deposition, with three deposition scenarios. Results reported for
    20-. 50-, and 100-year projections. The data for year 0 represent initial soil conditions.
                                             219
-------
         8
         2-
         0
         0.0
0.2       0.4       0.6       0.8
   SOIL BASE SATURATION
1.0
               pH = pK  + n log [BS/(1-BS)]
                    pKa = 4.96  n = 0.797
Figure 9-15.  The curvilinear relationship between soil pH and base saturation for typical
          values of n and pKa as defined by Equation 9-6.
                                220
-------
simulations. In all likelihood, this occurs because of the nature of the deposition scenario.  For the
first five years of the simulation, the acid loadings are constant and equal to current levels.  Therefore,
the projected trajectory soil pH is the same for deposition scenarios A and B for the first five years.
For the next 10 years under scenario B, however, the deposition is ramped down by 5 percent per
year, until the initial deposition is halved, then held constant for the duration of the simulation.  Under
scenario B, the rate of decrease in soil pH slows beginning in year 6 and is essentially flat at year 15
and for the remainder of the simulation.

     As expected, the projected changes for soil pH under deposition scenario C (ramped down to
50 percent of current loadings but  ramped back up to 80 percent) are less than those for scenario A
but greater than under scenario  B.  The conclusion to be drawn from this analysis is that reducing
acid loadings diminishes the projected effects of acidic deposition on soil pH.  A corollary can be
drawn  for the concomitant effects on vegetation and surface waters.

9.3.2.7.3  Projections of the effects of acidic deposition on estimates of soil solution AI3+  --

     It has been known for some time in agricultural systems that low soil pH can result in increased
soil solution Al concentrations that  are potentially toxic to plants.  Cronan and Schofield (1979)
proposed that Al could be mobilized by acidic deposition and subsequently transported to surface
waters where aquatic organisms could be affected.  Increasing soil  solution Al concentrations as a
result of acidic deposition has  also been suggested as a factor that may contribute to forest stress and
decline (Ulrich, 1983).  Although Al toxicity levels are not known for very many forest species, we
assume that increases in soil solution Al will have  adverse effects upon forest vegetation,  and will lead
to increased transport of Al to surface waters. The purpose of this analysis is use the Bloom-Grigal
model  projections of soil solution AI3+ concentrations to evaluate the effects acidic deposition on soil
solution Al3* in the M-APP.

     The results of this analysis are presented  in Figure 9-16, and in Tables 9-16, 9-17, and 9-18.
Under  deposition scenario A (Table 9-16), median soil solution AI3+ concentrations are projected to
more than double during the 100-year simulation.  The initial median concentration is 9.5^mol/L; it
increases to 22 .umol/L at the end of the simulation. Approximately  10 percent of the systems are
projected to have AI3+ concentrations in excess of 30 ^mol/L at the end  of 200 years. Reductions in
deposition, represented by scenario B, have a dramatic effect, reducing the projected concentration of
Al3* in solution at 100 years by half. In deposition scenario C, the model projects less AI3+ in
solution compared to scenario A, but slightly more than scenario B.  In all cases,  decreasing the
estimated concentration of base cations in dry deposition resulted in estimates  of lower pH and
greater projected AI3+ concentrations  relative to otherwise identical  scenarios.

     It is apparent from this analysis that the soils in the M-APP are likely to have increased AI3+
concentrations at current levels of deposition. This situation increases the likelihood that  surface
waters will have increased AI3+ concentrations that may adversely affect aquatic biota.
                                              221
-------
                     1.0
                   X
                   VI


                   o
                     0.8-
0.6-
                   a:
                   O 0.4
                   CL
                   O
                   cc
                   °- 0.2
                     0.0
                                SCENARIO A
                               10
                	1	1
                 20      30
                AI3+(nmol/L)
                                                      40
                                                             50
                     1.0
                     0.8-
                   X
                   VI

                   •Z. 0.6-
                   O
                   t-
                   CC
                   O 0.4-
                   CL

                   g
                   CL 0.2-
                     0.0
                                 SCENARIO B
                               10
                 20      30
                 A!3+(iamol/L)
                                                      40
                                                             50
                     1.0
                     0.8-
                     :o.6-
                    CC
                    00.4

                    O
                    cc
                      0.0
                                 SCENARIO C
                               10
                                      —r
                                       20
                                              30

                                       AI3+(Hmol/L)
                                40
                                        50
Figure 9-16.  Regional CDFs of Bloom-Grigal model projected soil solution concentrations
             of AI3+ of soils on the DDRP target population of M-APP stream watersheds
             for three deposition scenarios (A, B, and C) at 0, 20, 50, and 100 years.
                                          222
-------
 Table 9-16.   EJIoom-Grigal Model Regional Projections of Soil Solution AI3+ for Soils in the
              M-APP Region as a Function of Time and Deposition Scenario for the Scenario of
              a>. Full Complement of Base Cations in Dry Deposition3
YEAR
YEAR
YEAR
MEAN
             Deposition Scenario = A

STD_DEV    MIN      P 25    MEDIAN   P 75
MEAN
            Deposition Scenario = B

STD DEV   MIN      P  25   MEDIAN   P 75
                                                                          MAX
0
20
50
100
10.3
11.7
14.3
20.0
4.0
4.2
4.9
7.8
0.8
0.9
0.9
1.0
8.0
9.8
12.8
15.0
9.5
11.4
14.6
22.0
11.9
13.0
16.3
25.6
17.3
19.2
23.3
44.5
                                                                          MAX
0
20
50
100
10.3
10.9
11.0
11.2
4.0
4.1
4.0
4.0
0.8
0.9
0.9
0.9
8.0
8.9
9.6
9.6
9.5
10.3
10.3
10.7
11.9
12.2
12.2
12.6
17.3
18.0
18.0
18.0
                      Deposition Scenario = C

MEAN    STD_DEV    MIN      P 25    MEDIAN   P 75      MAX
0
20
50
100
10.3
10.9
11.5
13.6
4.0
4.1
4.0
4.5
0.8
0.9
0.9
0.9
8.0
9.0
9.9
11.0
9.5
10.3
11.1
14.8
11.9
12.2
13.5
16.1
17.3
18.0
18.0
25.9
    Projections made using typical year deposition, with three deposition scenarios. Results reported for 20-,
    50-, and 100-year projections. The data for year 0 represent initial soil conditions. Units are /jmol/L.
                                            223
-------
Table 9-17.   BIoom-Grigal Model Regional Projections of Soil Solution AI3+ for Soils in the
              M-APP Region as a Function of Time and Deposition Scenario for the Scenario of
              a 50 Percent Reduction of Base Cations Concentration in Dry Deposition3
YEAR
YEAR
YEAR
MEAN
            Deposition Scenario = A

STD DEV   MIN      P 25   MEDIAN   P 75
MEAN
            Deposition Scenario = B

STD DEV   MIN      P  25    MEDIAN   P 75
MEAN
            Deposition Scenario = C

STD DEV   MIN      P  25    MEDIAN   P 75
MAX
0
20
50
100
10.3
12.1
15.7
23.7
4.0
4.4
5.4
9.5
0.8
0.9
1.0
1.1
8.0
10.1
14.1
18.1
9.5
11.4
15.4
24.S
11.9
13.3
18.8
28.9
17.3
19.7
25.1
50.2
MAX
0
20
50
100
10.3
11.2
11.4
11.9
4.0
4.1
4.1
4.1
0.8
0.9
0.9
0.9
8.0
9.1
9.8
10.2
9.5
10.6
10.6
11.3
11.9
12.4
12.7
14.1
17.3
18.2
18.2
18.4
MAX
0
20
50
100
10.3
11.5
12.1
15.3
4.0
4.1
4.1
5.2
0.8
0.9
0.9
1.0
8.0
9.1
10.6
12.6
9.5
10.6
11.6
16.1
11.9
12.4
14.2
19.5
17.3
18.2
18.5
29.4
     Projections made using typical year deposition, with three deposition scenarios. Results reported for 20-
     50-, and 100-year projections. The data for year 0 represent initial soil conditions.
                                             224
-------
Table 9-18.   BIoom-Grigal Model Regional Projections of Soil Solution AI3+ for Soils in the
              M-APP Region as a Function of Time and Deposition Scenario for the Scenario of
              Zero Base Cations in Dry Deposition9
                                    Deposition Scenario = A

YEAR         MEAN   STD DEV    MIN      P 25    MEDIAN    P 75       MAX
0
20
50
100
10.3
12.5
17.3
28.2
4.0
4.7
6.1
11.8
0.8
0.9
1.0
1.2
8.0
10.4
14.6
22.0
9.5
12.4
16.8
29.2
11.9
13.7
21.9
34.1
17.3
20.3
26.9
56.4
                                    Deposition Scenario = B

YEAR         MEAN    STD DEV    MIN      P 25    MEDIAN    P 75      MAX
0
20
50
100
10.3
11.5
11.9
12.8
4.0
4.3
4.2
4.4
0.8
0.9
0.9
1.0
8.0
9.3
10.2
10.7
9.5
11.0
11.2
12.4
11.9
12.8
14.1
16.7
17.3
18.5
18.5
20.5
YEAR
                      Deposition Scenario = C

MEAN    STD DEV    MIN      P 25    MEDIAN    P 75      MAX
0
20
50
100
10.3
11.5
12.9
17.8
4.0
4.3
4.4
6.3
0.8
0.9
0.9
1.1
8.0
9.3
11.2
15.0
9.5
11.0
12.2
18.1
11.9
12.8
15.8
21.8
17.3
18.5
19.3
33.3
    Projections made using typical year deposition, with three deposition scenarios.  Results reported for 20-,
    50-, and 100-year projections. The data for year 0 represent initial soil conditions.
                                             225
-------
9.3.2.7.4  Summary --

      Characteristics of soils in the M-APP uplands and model projections of future chemical changes
in soils, using the Bloom-Grigal soil acidification model, can be summarized as follows:

      •   Soils in the M-APP do not presently have extremely low base saturations or phi, or extremely
          high estimated AI3+ concentrations in comparison to other forested temperate soils.

      •   Using current deposition in the Bloom-Grigal model to make projections for a 100-year
          period, the regional median soil base saturations are projected to drop below 10 percent
          (13.8 percent in year 0, 12,4 percent, 10.8  percent and 9.4 percent at years 20, 50, and
          100).

      •   Under a scenario of constant deposition, regional median soil solution AI3+ concentrations
          are projected to more than double, from 9.5/
-------
      Bloom-Grigal model projections were made using the respective regional soils and deposition
data.  The deposition data used in the NE and SBRP were long-term annual average estimates of
current wet and dry deposition (Church et al., 1989, Section 5.6), whereas the M-APP projections are
based on a typical year data set as described in Section 5.6 and Section 9.3.2.1. The median annual
total acid inputs for the NE, M-APP, and SBRP were 0.43, 0.93, and 0.47 Keq/ha, respectively. With
these estimates of current deposition and the regional soils data, the Bloom-Grigal model was used to
project potential! changes in soil and soil-solution chemistry.  The results were regionally weighted and
summarized quinquennially over a 100-year simulation  period.  The median values projected for base
saturation, soil pH, and soil solution AI3+ are  plotted for each region as a function of time in Figures
9-17, 9-18, and 9-19, respectively, and are summarized in Table 9-19.

9.3.2.8.1 Projections of the effects of acidic deposition on soil base saturation -

      The regional median value of base saturation initially is highest for the NE followed by the
M-APP and then the SBRP (Table 9-19).  This north-to-south ranking is expected based upon the age
and degree of v/eathering of the soils in each of the regions. The NE is the youngest landscape of the
three, with soils roughly 15,000 years old. The soils in the NE tend to be Entisols,  Inceptisols, and
Spodosols. The soils in the M-APP are older; most are Inceptisols, Entisols, or Alfisols. The soils in
the SBRP tend to be the highly weathered Ultisols.

      Figure 9-17 shows the projected median base saturations as a function of time.  Median base
saturation in soils in each of the three regions decreases somewhat after 100 years; the greatest
projected change occurs in M-APP soils. This is due largely to the higher loading of acidic deposition
in the M-APP.  The NE appears to be relatively well buffered against changes in soil base saturation.
Soils in the SBRP start at approximately 9 percent base saturation and decrease to slightly less than 7
percent at the end of the simulation. The final projected base saturations in the M-APP and SBRP
soils are low enough that the fraction of AI3+  in soil leachate can increase significantly, and macro-
nutrients may become limiting to terrestrial vegetation.

9.3.2.8.2 Projections of the effects of acidic deposition on soil pH -

      The median pH values of soils in the NE and M-APP initially are similar and slightly lower than in
SBRP soils (Table 9-19).  After 100 years, simulated median pH in the M-APP and SBRP soils
decreases by about 0.2 pH units.  The changes  in pH are relatively small, but represent  large
increases in H+, and are  projected to be accompanied by substantial increases in  soil solution AI3+.
Soils in the NE Region, due to their relatively  high base status, are more strongly buffered, and
experience smaller decreases in soil pH. Figure 9-18 graphs these results.

9.3.2.8.3 Projections of the effects of acidic deposition on estimates of soil solution AI3+ --

      Soil solution AI3+ results were not presented in the earlier DDRP report on the NE and  SBRP
(Church et al., 1989). The Bloom-Grigal model was rerun on these two regions using the methods,
data, and protocols described by Church et al. (1989).  The results of these runs, along with those for
the M-APP are presented here to allow regional comparisons.

                                             227
-------
   25

O

< 20 -j
DC
   CO

   LLJ
   CO

   CD

   r-
   "Z.
   LLI
   O
   or
   LJJ
   Q_
      15-
   10-
        0
                  YEAR 0 = 1985   YEAR 50 = 2035   YEAR 100 = 2085
                                                                  NE
          0
             i
            10
20    30    40    50    60   70    80
     YEARS IN SIMULATION
90    100
Figure 9-17.  Projected changes in median base saturation for 100-year simulations, for
           soils in DDRP target population watersheds in the NE, M-APP, and SBRP
           regions, generated using the Bloom-Grigal model for a scenario of constant
           deposition at current levels.
                                    228
-------
     5.0
     4.8-
  31
  Q.
  O
  CO
     4.6-
     4.4-
     4.2
                 YEAR 0 = 1985   YEAR 50 = 2035   YEAR 100 = 2085
         0    10    20    30    40    50    60   70    80    90   100
                         YEARS IN SIMULATION
Figure 9-18.  Projected changes in median pH, for 100-year simulations, for soils in DDRP
           target population watersheds in the NE, M-APP, and SBRP regions, generated
           using the BIoom-Grigal model for a scenario of constant deposition at current
           levels.
                                    229
-------
     25
                    YEAR 0 = 1985   YEAR 50 = 2035  YEAR 100 = 2085
                           30     40    50     60    70
                            YEARS IN SIMULATION
90    100
                                                 ,3+
Figure 9-19.  Projected changes in median soil solution AT, for 100-year simulations, for
            soils in DDRP target populations in the NE, M-APP, and SBRP regions, gener-
            ated using the Bloom-Grigal model for scenarios of constant deposition at
            current levels.
                                      230
-------
Table 9-19.   Summary Statistics for Regional Medians for Base Saturation (BS), pH, and Soil
              Solution AI3+ Concentration for Soils on DDRP Target Populations of Watersheds
              In the NE, M-APP, and SBRPa
                                  CONSTANT DEPOSITION
Region
                                                 YEAR IN SIMULATION
Parameter
               20
               50
              100
NE
M-APP
SBRP
%BS
%BS
%BS
22.0
13.8
9.2
21.7
12.4
8.1
21.2
10.8
7.3
21.0
9.4
6.6
NE
M-APP
SBRP
  PH
  pH
  PH
 4.6
 4.6
 4.9
 4.6
 4.6
 4.8
 4.6
 4.5
 4.7
 4.6
 4.4
 4.7
NE
M-APP
SBRP
Ald
Al1
   3+
10.3C
 9.5b
 4.3b
10.7
11.4
 5.2
11.2
14.6
 6.3
12.2
22.0
 7.9
0 Projections were made using the Bloom-Grigal model, for scenarios of constant future deposition.
b ;
-------
      Soil solutions in the NE have the highest projected initial soil solution AI3+ concentrations (Table
9-19). After 100 years, however, the median AI3+ levels are projected to be greatest in the M-APP. Al
concentrations increase slightly in NE soil solutions and in the SBRP they increase somewhat more
than in the  NE, as shown in Figure 9-19.  There may be several reasons for these results.  First, the
acid inputs  in the M-APP are approximately twice those in the other two regions, resulting in more
extensive cation leaching from soils and thus larger absolute decreases in base saturation and soil pH
in the region. The second apparent factor is differences in selectivity of exchange reactions for the
two regions, probably resulting from the nature and amount of organic and clay materials in soils of
the three regions. The soils in the NE have low pH, but relatively high base saturations.  The SBRP
soils have low base saturations,  but the highest pH.  The soils in the M-APP are low both in pH and
exchangeable base cations. These characteristics in combination lead to higher projected levels of
Al3+, and consequently indicate that the surface waters in the M-APP are likely to be more susceptible
to acidification than those in the NE and SBRP.

9.3.2.8.4 Summary and conclusions -

      Using the Bloom-Grigal model to make projections of the effects of acidic deposition on soils in
the NE, M-APP, and SBRP, and using the results to make regional comparisons, the following
conclusions can be drawn:

      •  Soils in the NE appear to be buffered against large changes in soil pH, base saturation, and
         soil solution AI3+ concentrations under current levels of deposition over the next 100 years.

      •  The M-APP appears to be susceptible to adverse changes in soil properties due to acidic
         deposition at current levels over the next 100 years. Soil base saturation and soil pH are
         both projected to decrease.  At the same time, soil solution AI3+ concentrations are
         projected to more than double.

      •  The soils in the SBRP are projected to have decreases in base saturation and  soil pH
         comparable to those in the M-APP. Projected  changes in soil solution AI3+ concentrations
         are substantially smaller than those in the M-APP.

Based upon model projections, the M-APP appears to be the most susceptible of the three regions, to
detrimental changes in  soil properties due to acidic deposition.

9.3.3 Reuss Model

9.3.3.1  Model  Description

      The Reuss model was originally developed by Reuss  (1983) and coworkers (Reuss and
Johnson, 1985; Johnson and Reuss, 1984) to simulate changes in soil and soil solution chemistry
associated  with changes in mobile anion leaching, pCO2, etc. In this equilibrium-based,  mass balance
model, the  solubility of a gibbsite-like phase is assumed  to control the concentration of aluminum.
Concurrently, exchange reactions are used to partition the cations AI3+, Ca2+, Mg2+, Na+, and K+
between the solid and solution phases. Figure  9-20 presents schematically the processes considered
                                             232
-------
           Top
        Horizon
             n
         Middle
      Horizon(s)
        Bottom
        Horizon
Figure 9-20.  Row chart showing hydrologic pathways and soil geochemical processes con-
            sidered by the modified Reuss model used for DDRP simulations. Arrows
            indicate major pathways through which ions are interchanged among reser-
            voirs. No attempt was made to distinguish relative fluxes among reservoirs;
            the heavier lines indicate those processes that are the focus of Level II
            modelling efforts discussed here. SEC represents soil exchangeable base
            cations.
                                    233
-------
in the model. The model computes soil pH, soil solution ANC, and base cation and dissolved
aluminum concentrations.  The model then "re-equilibrates" soil solutions with atmospheric carbon
dioxide and computes surface water composition. The development and process representation of the
model, as well as the rationale for its use in the DDRP and a discussion of modifications of the model
for use in the DDRP, are described in detail by Church et al. (1989, Section 9.3).

      Reuss's approach has several  advantages over the use of simple exchange reactions for model-
ling exchange reactions in soil environments.  First, the charge balance requirement of the code
makes the model responsive to ionic strength. Composition of the soil solutions in forested soils
has been shown to depend on ionic strength  (Richter et  al., 1988). Therefore, this aspect of the model
permits a more realistic simulation of natural exchange reactions than do the less involved computa-
tions. Second, the model allows the user to specify the partial pressure of carbon dioxide (pCO2) in
the soil gas.  Although the annual maximum pCO2 in forested  soils is typically about 1  percent
(Fernandez and Kosian, 1987; Solomon and Cerling, 1987), this concentration is about 30 times higher
than atmospheric pCO2 and can significantly affect soil solution composition (Reuss and Johnson,
1985, 1986).  Third,  although use of a gibbsite-like phase to regulate aluminum activities is contro-
versial (c.f., Section  9.3.2.2.7), its  use allows one degree  of freedom in the solution composition
to be effectively constrained. Finally, the mass balance constraints allow the user to track cation
depletion from the exchange complex as a function of time, hydrogen ion loading, and the imposed
physico-chemical environment.

      The Reuss model focuses on soil exchange reactions. The model does not consider other
cation source/sink processes such as mineral weathering, nitrogen transformations, or afforestation,
even though these processes may have equal or greater influence in regulating surface water compo-
sition in certain ecological settings.  Other such cation sources/sinks could be incorporated into the
model, but have been excluded from the present analyses so that the potential role of cation exchange
in the soil as a long-term buffer of acidic deposition can  be assessed more effectively. Models have
been developed that include these processes, and provide integrated projections of system  response
to the imposed deposition (Cosby et al., 1985a,c,  1986a,b; Gherini et al., 1985; Galloway et al., 1983;
Nikolaidis et al., 1988).  These integrated models provide a more complete assessment of ecosystem
response, but cannot be used effectively to understand the contributions of individual processes, such
as soil cation exchange, to the buffering responses of watersheds.

      This section describes Level II analyses of cation exchange buffering for DDRP watersheds in
the Mid-Appalachian Region, conducted using the Reuss model.  As in comparable analyses for DDRP
systems  in the NE and SBRP, our intent is to provide a quantitative assessment of the time frame over
which cation  exchange processes can neutralize acidic deposition, at current or altered rates of
deposition.  Consistent with the objectives of the Level II analyses, these analyses were designed to
consider only exchange processes, and only those occurring in the top two meters of the regolith. No
mineral weathering or other inputs of base cations are considered, except for atmospheric deposition.
As such, these analyses represent a worst-case analysis of exchange-mediated buffering in the soil,
designed to place an upper bound on the potential for the changes  in soil and runoff chemistry that
result from exchange-mediated neutralization  of acidic inputs by exchange  processes within the soils
of DDRP watersheds.
                                             234
-------
9.3.3.2 Model Formulation

      Details of the modification to, and resulting evolution of, Reuss's model as used in this study,
have been described in detail by Church et al., 1989.  The following primary changes were incor-
porated into the model from the original formulation of Reuss (1983):  (1) the model is coded in
FORTRAN instead of BASIC, (2) the model employs the Vanselow exchange formulation (Vanselow,
1932) instead of the Gaines-Thomas formulation (Gaines and Thomas, 1953), (3) nitrogen species are,
to a large extent, presumed to be retained by the ecosystem and are not considered as mobile
species, (4) slightly different aluminum speciation  algorithms are used, (5) multiple horizons, which
permit lateral runoff, are incorporated into the model structure, and (6) the DDRP version of the model
employs time steps in the computations instead of water increment steps. The decision to substitute
the Vanselow exchange expression for the Gaines-Thomas expression used  in the original Reuss
model formulation was based on two criteria. First, although possible problems remain  with the
definition of solid phase activity coefficients, the Vanselow expression is believed by many researchers
to be the most thermodynamically correct exchange formulation.  Second, and more important, in
model simulations conducted as part of DDRP analyses for the NE/SBRP to compare model results
using Gaines-Thomas, Vanselow, and Gapon expressions,  there were not strong differences among
the three formulations.  Residuals generated using the Vanselow formulation were better behaved than
those from the other models.

9.3.3.3 Assumptions

     As with any model,  assumptions are necessary regarding certain processes, the soil environ-
ment, and the characteristics of certain reactions.  The reader is referred to the NE and  SBRP DDRP
report  (Church, et al., 1989) for a detailed assessment of the effect these  assumptions have on model
predictions. The primary  assumptions for current  analyses include:

      •   Gibbsite solubility controls soil AI3+ concentrations.  This is different from the assumption of
         Al control by exchange equilibrium in the Bloom-Grigal model used in Section 9.3.2, and is
         a source of potential differences in model simulation results;

      •   Exchange selectivity coefficients are assumed to  be constant, i.e., not to vary with changes
          in base saturation; and

      •  Soil gas pCO2 is assumed to be a uniform 0.005 atm for all soils for all seasons.

9.3.3.4 Limitations

     The Reuss model focuses on soil exchange reactions. The model does not consider other
processes such as sulfate adsorption, mineral weathering, nitrogen transformations, or afforestation,
even though these processes may have equal or greater influence in regulating surface  water compo-
sition in certain ecological settings (Likens et al., 1977; Johnson et al., 1988). For the present
application of assessing the potential for delays in acidification attributable to soil cation  exchange
processes, these limitations are not only appropriate, but necessary.  Some of the most  significant
omissions in the model are: (1)  no provisions are made to consider organo-cations, and especially
                                             235
-------
organo-aluminum interactions, (2) mineral weathering is not considered, with the result that the pool of
base cations in the soil and the relative abundance of individual cation is controlled solely by
deposition and exchange equilibria, and (3) reactions are modelled as equilibrium processes.  Clearly,
most ongoing  chemical processes in watersheds are subjected to rapidly and constantly changing
chemical environments, but given the rapid nature of exchange reactions and the annual time step
used in simulations, an assumption of exchange equilibrium is not unreasonable.

9.3.3.5  Model Inputs

9.3.3.5.1 Deposition and associated data ~

     The model requires deposition data, including precipitation  quantity (cm/yr) and average annual
concentrations of the major ions (SO42", CI", NO3', NH4+, Ca2+, Mg2+, Na+, and K+)  in precipitation,
along with estimated fluxes of the same ions in dry deposition.  The atmospheric flux of each ion is the
combined wet plus dry average annual deposition. Evapotranspiration (% ET) data required by the
model  are computed as the quantity of precipitation minus runoff. Sources of deposition, precipitation,
and runoff data are described in Section 5.6 and 5.7.

9.3.3.5.2 Soils data -

     The model requires physical and chemical information about each of the horizons included in
the simulations.  Physical parameter data used in the model include horizon thickness,  bulk density,
percent coarse fragments, and a hydrologic runoff parameter.  Chemical parameters  include CEC,
base cation concentrations on the exchange complex, selectivity  coefficients for the Ca/AI, Ca/Mg,
Ca/Na, and Ca/K exchange reactions, soil gas pCO2, the apparent solubility product for AI(OH)3(s),
and the stoichiometric coefficient for H+ to be used in describing the  dissolution of the aluminum solid
phase.  Multiple horizon versions of the model require the above  information for each of the horizons
to be considered. Generation of soil  physical and chemical data  are described  in Section 5.4 and 5.5.
Church et al. (1989, Section 9.3) provide a detailed description of the  procedures used in calculating
exchange selectivity coefficients; procedures used for data aggregation are described by Johnson et
al. (1990).

9.3.3.6 Results - ANC Projections

      For each simulation, the model generates projections for surface water composition and, on a
by-horizon  basis, for soil and soil solution composition for major chemical species. Results are com-
piled for the first and final years of the computation and at user-specified intervals during the
simulation.

      Information on surface water variables retained in the output files include data on pH, ANC,
SO42', NO3",Cr, Ca2+, Mg2+, Na+,  K+, AI3+, sum-(AI)aq, and ionic strength.  For soils, information on
soil pH, base saturation, and exchangeable Ca, Mg, Na, and K are retained for the solid phase, and
ANC, Ca2+, Mg2+, Na+, and K+ data are retained for soil solutions.  For this report, analyses  focus on
surface water ANC, because this parameter provides the most  integrated indicator of system response
to continued exposure to acidic deposition.
                                             236
-------
 9.3.3.6.1  Prediction of current conditions -

      The distribution of current surface water ANC values projected for the M-APP Region using the
 Reuss model is illustrated in Figure 9-21. The ANC values for M-APP streams, from National Stream
 Survey data (Kaufmann et al., 1988) and from initial model calibrations, are listed in Table 9-20. An
 obvious feature of the model projections is an extremely tight clustering of simulated ANC, with a
 mean value of-21 fieq/L (standard deviation  = 8/*eq/L).  Clustered values of predicted ANC values
 have been obseived in all regions for which Reuss model runs have been completed.  The primary
 difference between projections for the M-APP and those for the NE and SBRP regions is the mean
 ANC values observed in each region. For the M-APP, the soils appear to be more highly acidic than
 those in either of the other two regions, resulting in significantly lower predicted surface water ANC
 and pH values.

      As concluded for the other regions, model  results are consistent with the hypothesis that soil
 exchange reactions can buffer soil water and  surface water ANG values, but that in acid soils such as
 those that occur in the DDRP watersheds, ANC values are buffered within a low range of values.  The
 low, tightly clustered ANC values simulated by the Reuss model suggest that exchange is not the
 primary mechanism controlling the compositions of surface waters in DDRP study regions. The
 buffering provided by the soils is effective in an ANC range that is lower than the mean values
 observed.  Figure 9-22 illustrates the relationship between observed and projected ANC values.

       Given the difference between measured and simulated values of ANC, it seems obvious that
 other mechanisms in addition to exchange play a significant role in the control of surface water base
 cations and ANC. Uptake of cations by aggrading vegetation represents a sink for cations, and would
 be expected to deplete soil base cations and  to depress surface water ANC values below those
 computed by the model.  The presumed base cation/ANC source, and  the presumed major process
 altering cation balance and ANC  in the M-APP watersheds, is primary mineral weathering.  Release of
 base cations and ANC through weathering  reactions,  as discussed in Section 3.4 of Church et al.
 (1989), can increase surface water ANC to values well above the limit apparently imposed by soil
 exchange processes.

       According to the knowledge gained from  modeling activities in the three  DDRP  regions,
 mineral weathering is apparently the major ANC generating process in those systems with ANC values
 in excess of about 50 /*eq/L For systems with lower ANC values, both mineral weathering and soil
 exchange processes could be influencing observed surface water chemistry. Available data, however,
 preclude any reliable apportionment of the contribution from each of these two processes to the total
 base cation and ANC levels in surface waters  considered by DDRP.

       The implications of these findings are  significant in terms of projected future changes in
surface water chemistry.  If mineral weathering is, in fact, regulating ANC levels in those systems with
ANC greater than 50 to 100 ^eo/L, then weathering will continue to provide base cations to neutralize
at least a significant fraction of deposition inputs. Because exchange represents a modest or even
negligible net source of base cations in these systems, depletion of the soil reservoir of exchangeable
                                            237
-------
              Reuss Mode
              Year 0
              Current Deposition
  O
      0.0
        -50
-40        -30       -20       -10
       SIMULATED ANC faeq/L)
o
10
Figure 9-21.   Cumulative distribution of year 0 calibrated values of ANC, for the target
            population of DDRP watersheds in the M-APP Region, based on simulations using
            the Reuss model with Typical Year deposition estimates.

                                      238
-------
Table 9-20.    Summary Statistics for Current ANC in DDRP Target Population Stream Reaches
              in the M-APP Region8

Meanb
Standard Deviation
Min
25th percentile
Median
75th percentile
Max
0 National Stream Survey Data from

DDRP Population
Estimates
87.8
58.7
-9.6
48.5
86.8
48.5
199.6
Kaufmann et al., 1988. Units are^eq/L.

Reuss Model
Estimates
-21.0
8.4
-33.7
-28.6
-20.9
-15.7
5.1


   ANC for surface water.
                                          239
-------
 cr
 0)
 =1
    200
    150-
    100-
 Q

 £   50 H
 ID


 c/5
o
     -50
                                                                 1:1
                                                    •  I
                                               «
        -50
              0          50          100         150

                  OBSERVED ANC Qieq/L)
200
Figure 9-22.   Comparison of measured values of ANC for DDRP watersheds in the M-APP

            Region and values simulated using the modified Reuss model with Typical Year

            deposition estimates. Measured values are from Kaufmann et al. (1988).
                                   240
-------
base cations will occur more slowly than in otherwise comparable systems in which exchange is a
major or dominant net source of bases. The potential rate of decline in ANC attributable to depletion
of soil base cations in these systems will tend to vary in direct proportion to the relative importance of
exchange as a net cation source to runoff, and as such will be inversely related to initial ANC.  In the
long term, as long as rates of mineral weathering exceed rates of deposition, ANC will remain above
zero.  In contrast, in systems with ANC values less than about 50 /ieq/L, rates of sulfate deposition
probably exceed rates of primary mineral weathering, such that soil exchange processes are contrib-
uting to the maintenance of ANC.  Some of the low ANC systems are probably experiencing acceler-
ated base cation leaching in. response to ongoing increases in sulfate leaching (Section 9.2). These
systems could experience significant ANC decline in the future, as sulfate mediated leaching continues
to increase and as the exchange  pool is gradually depleted.  Results from the Reuss model are not
sufficient, in and of themselves, to effectively identify those systems at highest risk for such changes.

9.3.3.6.2 Projected future conditions --

      In order to assess the po'tential magnitude of changes in ANC that might occur in the M-APP
watersheds, and to characterize the time frame over which such changes might occur, simulations
were run for a 100-year period, using the Reuss model with a variety of deposition scenarios. The
mass balance component of the model tracks the changes in  the base cation pool on soil exchange
sites through time. For these simulations, precipitation quantity (cm/yr) and the deposition fluxes were
used to specify the total loadings of ions to the soil. Annual time steps were used for these simula-
tions. For the M-APP, model simulations were run for 100 years, with results reported for years 20, 50,
and 100.

      Projected time-dependent changes in ANC values for the population-weighted results are illus-
trated in Figure 9-23 and summary statistics are listed in Table 9-21.  The Reuss model considers only
the effects of the soil cation exchange process in making these projections.  Mineral weathering reac-
tions would supply base cations to the soil system and would, in general, diminish and delay the
response of these systems to the effects of acidic deposition.  At 20 and 50 years, most systems in
the M-APP are projected to experience minimal change in ANC.  The soil buffering capacity in these
systems is sufficient to moderate the effects of acidic deposition over these time scales.  At current
deposition, none of the systems modeled in this region are projected to experience losses of ANC of
more than 20 /^eq/L within the 50-year time frame.

      For longer time frames, however, the changes in base cations and in ANC become much more
pronounced, as soil exchange pools become depleted of base cations (Figure 9-23).  About a third of
the watersheds in the region are projected to experience minimal changes (< -25 jueq/L) over
the 100-year time frame, but others are projected to experience^ much  larger change in ANC (median
change of-73/^eq/L) with a maximum change of -237 ^eq/L  Examination of the results (Table 9-21)
shows that projected changes in the ANC values through time are not linear, but accelerate dramatic-
ally between years 50 and 100, as the buffer capacity of soils is depleted.  Soil response to acidic
deposition is analogous to a buffer being titrated by acidic deposition. When the system is above the
inflection point of the titration curve, the response  of a soil to a unit loading of acidic deposition will be
a gradual, almost linear, decline in projected ANC. Once the system reaches the inflection point,
                                             241
-------
o

DC
O
0.
o
DC
Q_
LLI

P

_J
ID
^
ID
O
      1.0
      0.8 -
YEAR 20

YEAR 50

YEAR 100
                -250     -200     -150     -100      -50       0   -    50

           PROJECTED CHANGE FROM CURRENT ANC (|ieq/L)
Figure 9-23.  Cumulative distributions of projected changes in ANC at 20, 50, and 100
           years, for the target population of DDRP M-APP watersheds, based on Reuss
           model simulations with Typical Year deposition estimates.
                                  242
-------
Table 9-21.    Summary Statistics for Projected Changes in Surface Water ANC Values In
               Streams in the DDRP Target Population in the M-APP Region at 20, 50, and 100
               Years for Three Deposition Scenarios3
                           Scenario A
                            Current
                      Scenario B
                    50% Decrease
                    Scenario C
                  50%/20% Decrease
Current Mean
-21.0
-21.0
0 Units are/ieq/L

  All values, except for current year estimates, are presented as changes from current status.
-21.0
Year 20b
Mean
Std. Dev.
Median
Max (negative)
Year 50b
Mean
Std. Dev.
Median
Max
Year 100b
Mean
Std. Dev.
Median
Max

-0.4
1.0
-0.3
-2.8

-3.4
3.8
-3.3
-19.1

-73.0
61.3
-72.5
-236.9

+9.8
2.6
9.6
5.6,

9.2
3.1
9.3
3.5

4.2
7.5
6.2
-17.3

8.9
2.4
8.7
5.0

2.5
2.5
3.2
-2.7

-14.8
20.2
-7.8
-93.7
                                            243
-------
 however, the rate of decline in ANC dramatically accelerates because the buffer capacity of the system
 has been depleted.

      Results of these simulations have major implications for soils and surface waters in the M-APP
 Region. Small, gradual changes in surface water ANC in the short term do not preclude the possibility
 that more rapid dramatic changes can occur in the future.  This is especially true for systems with low
 current ANC. Rates of change in system response can increase with time, although changes will be
 moderated to some extent by mineral weathering.

 9.3.3.6.3 Ramped deposition scenarios ~

      The last group of simulations addresses the effects that changes in the level of sulfur deposition
 have on projected future changes in ANC.  As described in Section 5.6, three deposition scenarios
 were considered for M-APP simulations. The first scenario is continued deposition at current levels.
 The second scenario (Scenario B) ramps deposition linearly from current levels, starting at year 5, to
 50 percent of current deposition by year 15.  Deposition is then  maintained at this lower level for the
 remainder of the simulation.  The third scenario (C) entails a similar ramping of deposition to 50
 percent of current levels between years 5 and 15, but then gradually increases the fluxes of acidic
 sulfur deposition to 80 percent of current values between years 15 and 50.  After year 50, deposition is
 maintained at 80 percent of current levels.

      Differences between the projections  made using ramped and constant deposition are summa-
 rized in Table 9-21. Differences among the scenarios are substantial at the 20-year point.  Since the
 simulations do not include any source of base cations for the soil exchange complex, except to the
 extent that soils might retain cations from deposition, the projected increases in ANC for the waters
 evolving from these soils is almost entirely a mobile anion effect.  That is to say, the  simple act of
 reducing by half the levels of acidic sulfate deposition and  flux through the soil results in a  decrease in
 acid cation transport  and an increase in projected surface water ANC values.

      The median changes projected for ANC at year 50 using deposition scenario B suggest  that the
 ANC of surface waters in the M-APP would remain comparable to, or slightly higher than present
 conditions.  For the 50 percent reduction in deposition, projected surface water chemistry remains
 almost unchanged during the 30-year interval. For deposition scenario C, which embodies both the
 initial decline and a subsequent increase in deposition, the model projections suggest that the ANC of
 runoff will increase, then gradually decline.  In the absence of a mineral weathering component,
 projected ANC will ultimately decline if sulfur deposition exceeds base cation inputs,  until the
 exchange complex is depleted.  At that time, the soil system will attain a new dynamic equilibrium, in
which the soil solution and runoff will have low pH and ANC because the exchange complex can no
 longer neutralize incoming acidity.

     After 100 years, the differences in the medians among the three deposition scenarios are quite
 marked. Simulations using the constant deposition scenario project major reductions in projected
surface water ANC, with a median decrease of more than 70 ,weq/L Scenario B projects surface water
                                             244
-------
ANC that is little changed from present conditions; scenario C projects small declines in median ANC
(-8 fteq/L compared to -73 /*eq/L at constant deposition).

9.3.3.6.4 Comparison of results from the M-APP, NE, and SBRP -

      Comparison of the results obtained using the Reuss model in the three DDRP regions indicates
both similarities and differences among the regions. Table 9-22 summarizes current surface water
chemistries projected by the model for each of the three regions.  Soils in each region behave as
strong buffers, regulating the surface water ANC values to within narrow ranges; in all cases, median
ANC values are substantially lower than measured ANCs.  These observations again suggest the key
role mineral weathering has in producing ANC for surface waters, even in low ANC systems that are
susceptible to the effects of acidic deposition. The predicted median ANC in the M-APP Region is
considerably  lower than that predicted for systems in the other two regions.  This suggests that the
soils in this region either are naturally more acidic than those in the other regions or that a substantial
fraction of their buffering capacity has been depleted as a result of long-term exposure to acidic
deposition. Other factors might also contribute to the lower predicted equilibrium ANC values for soil
leachate in soils of the M-APP Region, but reasons are unclear.

      The soils In the three regions respond quite differently to continued exposure to acidic depo-
sition (Table 9-23). Soils in the NE Region are projected to undergo the largest changes during the
first 20- and 50-year simulations. At present levels of deposition, soils in the NE appear to be more
susceptible to short-term changes in leachate ANC than are the soils in the other two regions. This
conclusion is, in a sense, counter-intuitive because soils in the NE tend to have higher  levels of base
saturation than the soils in the M-APP or SBRP (see Table 9-9). The soils in the NE, however, are also
younger than those in the M-APP and SBRP, and as a result, tend to have less clay-size material.
Mean changes in the surface water ANC values in the NE  decline almost linearly through time. This is
in marked contrast to the behavior of soils in the other two areas, in which the near-term changes in
projected surface water quality are minimal, but rates of decline then  substantially accelerate as the
buffering capacity is  depleted.  This effect is especially pronounced in the M-APP, and thus is con-
sistent with the hypothesis that long-term exposure to acidic deposition in this region is significantly
depleting the buffering potential of the soils.

9.3.3.7 Conclusion

      Several conclusions can be drawn from the observations made using the Reuss model and the
projected changes to surface water ANC values along the eastern United States.

      •  For watersheds with surface waters currently exhibiting ANC values in excess of 50 to 100
         /teq/L, ANC values projected by an exchange model appear unrelated to observed ANC,
         suggesting that mineral weathering is the dominant watershed process controlling ANC.

      •   For watersheds that are in steady state with regard to sulfate adsorption and exhibit surface
         waters with ANC values of less than 50 ^eq/L, soil exchange processes may be  contributing
         to the observed ANC.  In most systems, the observed levels are probably controlled by a
                                              245
-------
Table 9-22.    Summary Statistics for Projections of Current ANC, Predicted Using the Reuss
              Model with Typical Year Deposition Estimates for the NE, M-APP, and SBRPa
                            NE
                           M-APP
                                                                             SBRP
Mean
Std. Dev.

Min
25th Percentile
Median
75th Percentile
Max
  8.3
 19.0

-46.8
 -1.8
  7.4
 20.6
 67.1
-21.0
  8.4

-33.7
-28.6
-20.9
-15.7
  5.1
  2.2
  6.1

-14.1
 -1.5
  2.3
  4.7
  2.3
    Units
                                             246
-------
Table 9-23.   Projected Changes in ANC at 20, 50, and 100 Years, Based on Reuss Model
              Analyses Using Typical Year Deposition Estimates for the NE, M-APP, and SBRPa
                            NE
 M-APP
  SBRP
Current
       Mean                8.3

Changes at Year 20
       Mean               -6.4
       Std. Dev.            18.0
       Median              -2.0
       Max              -118.3

Changes at Year 50
       Mean             -16.1
       Std. Dev.            26.4
       Median              -6.0
       Max              -138.8

Changes at Year 100
       Mean             -43.1
       Std. Dev.            51.5
       Median            -22.0
       Max              -231.7
 -21.0
  -0.4
   1.0
  -0.3
  -2.8
  -3.4
   7.8
  -3.3
 -19.1
 -73.0
  61.3
 -72.5
-236.9
   2.2
 -1.2
  0.7
 -2.2
 -2.2
 -3.7
  2.0
 -4.0
 -7.4
-23.0
 16.2
-18.5
-58.8
1 Units are /ieq/L
                                          247
-------
          combination of cation exchange and mineral weathering, with a relatively larger role for
          exchange in systems with the lowest ANC.

      •   Modelled reductions in acidic sulfate deposition in the M-APP result in significant projected
          improvements in surface water ANC. The improvement is primarily derived from a mobile
          anion effect, and is not attributable to any increase in soil base cation content.

9.4 SUMMARY

      The Level II analyses indicate that both sulfate adsorption and cation exchange by soils have
been  and will continue to be important processes delaying surface water acidification resulting from
acidic deposition in the M-APP Region.  Sulfate adsorption has maintained  stream water sulfate
concentrations in the region at levels well  below steady state, and model projections indicate that
adsorption will continue, albeit with steadily decreasing efficiency, for a few more decades.  The
sulfate status of streams in the region is in transition; model projections indicate, and monitoring data
confirm, that sulfate concentrations of streams in the region are increasing  by approximately 1-2
,aeq/L/yr (Table 9-2).  Modelled retention in the base year for simulations (1986) was  26 percent,
somewhat lower than the median computed retention  of 44 percent, but is  projected to decrease to a
median of 11 percent within 20 years, and to only 3 percent in the next 50 years. At the  present time,
14 percent of M-APP watersheds are at sulfur steady state; that percentage is projected to increase to
36, 58, and 93 percent in 20, 50, and 100 years, respectively (Tables 9-4, 9-5).  Median time to sulfur
steady state for the M-APP watersheds  is 35 years, compared to 0 years in the NE (i.e., at least 50
percent of NE systems are presently at or above steady state), but is considerably shorter than the 61
year median for streams in the SBRP.

      Because sulfur deposition to M-APP watersheds is much higher than in the NE or SBRP, com-
puted steady-state sulfate concentrations  in the M-APP are, on average, much higher than in the other
regions.  Sulfate concentrations in DDRP target population watersheds in the M-APP are already
higher than in NE or SBRP systems (median  = 153^eq/L in the M-APP, 105,aeq/L in the NE, and 25
fieq/L in the SBRP), but at steady  state, the median sulfate concentration in the M-APP (215y«eq/L) will
be almost twice that of lakes in the NE  (111 ^eq/L) or  SBRP streams (120/leq/L) (Table 7-2).
Although, on average, sulfur retention in M-APP streams is considerably lower than in the SBRP, the
increase  in sulfate concentration as systems come to steady state is larger (median and  maximum
change of 107 and 248 ^eq/L, compared to 83 and 167 ^eq/L in the SBRP).

      If sulfur deposition is reduced in the M-APP,  there will, of course, be proportional decreases in
steady-state sulfate concentration; the rate of response of sulfate concentration depends on the extent
and timing of reductions in deposition and on the reversibility of adsorption. In simulations assuming
complete reversibility,  for most M-APP watersheds (all  but those systems presently at steady state),
there will be an initial lag period if deposition  is reduced,  during which time the stream water sulfate
concentration is projected to continue to increase.  This lag results from desorption of sulfate presently
retained on upper soil horizons. Such lags are projected to last as long as 50 years, with projected
sulfate concentrations increasing as much as 60 fieq/L (Figure 9-11). In all cases, and at any point in
time during such a lag phase, however, the projected  sulfate concentrations are lower than they would
be if deposition were maintained at current levels.
                                             248
-------
      The projections for sulfur relate only to the status of sulfur in soils and runoff; time to steady
state for sulfur is not directly linked to time for systems to reach zero ANC or any other threshold
value.  Although changes in sulfate mobility are a principal control on changes in both base cation
mobility and ANC, other processes also influence the rate and extent of change in cation composition
and ANC of surface waters.  Surface waters can reach an ANC of zero (or any other threshold value)
before, concurrently with, or after sulfur comes to steady state.

      Because of the high sulfur loadings and high sulfate concentrations in M-APP watersheds, the
potential for base cation leaching from the exchange complex, and for depletion  of the labile cation
pool, is high. The Level II base cation exchange analyses represent in most respects a worst-case
analysis (i.e., weathering  rate assumed to be zero, sulfate assumed to be completely mobile), intended
to place an upper bound on the potential rate of cation depletion from the soil and of related changes
in the ANC of runoff.

      Most M-APP soils have formed from sedimentary bedrock, and represent the poorly weatherable
remnants of one or more previous weathering  episodes. Soils in DDRP watersheds in the region,
because of the parent material mineralogy and extensive historic weathering, are naturally base cation
deficient and  usually strongly acidic. Soil analyses confirm the low pH and base status of most soils
on watersheds in the target population. Model simulations for the M-APP Region, as in previous DDRP
analyses for the NE and SBRP, suggest that exchange processes can be a significant source of base
cations to buffer surface water chemistry, but that the role of exchange is probably most important in
those systems with low surface water ANC (<  100 ^eq/L).  Weathering probably  dominates cation
supply to systems with higher ANC.

     Although the two base cation models focus on different aspects of the soil  system (soil pH and
base saturation in the case of the Bloom-Grigal model; runoff chemistry for the Reuss model) results
from the two models are consistent, and  reflect the importance of cation exchange  as a buffer of soil
and runoff pH and ANC.  Under conditions of continued deposition at current levels, the ANC of runoff
(soil leachate) is projected to decrease by only a few^eq/L during the first 50 years of simulations, but
then to decrease rapidly, with a median decrease of 70/^eq/L between years 50 and 100 of simula-
tions.  Simulations using the Bloom-Grigal model project a steady decline in base saturation of soils,
as the exchange pool is depleted during  the 100-year period (from 13.8 to 9.4 percent). Soil pH
changes relatively little  during the first 50 years, but the decrease accelerates between years 50 and
100.

      Under conditions of reduced deposition, projected changes in runoff chemistry reflect both the
depletion of the exchange buffer capacity and  the salt effect. If deposition is reduced by 50 percent,
Bloom-Grigal  projections indicate only  small, slow future decreases in soil pH (<  0.1 units) and base
saturation  (from 13.8 to 13.2 percent) during the 100-year period.  For the same deposition scenario,
Reuss  model  projections  indicate an increase in the ANC of runoff of slightly less than  10/^eq/L,
attributable to a salt effect as sulfate concentrations decline.  For the intermediate deposition  scenario,
soil base cations and pH are projected to initially be almost unchanged as sulfur  deposition decreases,
then to deteriorate as sulfate concentrations increase in the latter part of the simulations. Reuss model
projections of ANC show an initial increase in ANC of almost 10 ^eq/L (salt effect), but ANC then
                                             249
-------
decreases below initial values at years 50 and 100, due to a combination of increased sulfate-
mediated leaching and depletion of the pool of exchangeable base cations. Overall, results of Level II
analyses can be summarized as follows:


      •  Sulfate  adsorption and cation exchange are both important processes that act to delay the
         effects of acidic deposition in soils and surface waters of the M-APP Region.

      •  Both adsorption and exchange are capacity-limited processes that are being depleted.
         Sulfate  concentrations are currently increasing as the adsorption capacity of soils becomes
         saturated, and base cations are projected to be gradually depleted from the soil under
         conditions of continued deposition at current levels.

      •  Under a scenario of continued deposition at current leveis, a median increase in sulfate
         concentrations of more than 100 ^eq/L is estimated as systems come to  steady state. As a
         result of the high rates of sulfate leaching in the M-APP, runoff ANC is projected to have a
         small median  decrease (3 ^eq/L) in the first 50 years  of simulations, but an additional 70
         fieq/L decline between years 50 and 100.  During the same 100-year period, the efficiency
         of the soil base cation buffer systems will be seriously reduced, and final  base saturation is
         projected to be less than 10 percent.

      •  Sulfate  concentrations are projected to decrease, and runoff ANC to increase if deposition is
         reduced by 50 percent. The net depletion of base  cations from the soil exchange pool
         would be virtually stopped  by such a reduction in deposition.
                                              250
-------
-------
                                         SECTION 10
                  LEVEL III ANALYSES - DYNAMIC WATERSHED MODELLING

10.1 INTRODUCTION

     Previous sections discuss (1) the genera! DDRP approach (Section 4), (2) sulfur retention attri-
butes of Mid-Appalachian (M-APP) watersheds (Section 7), and (3) future changes in sulfate adsorp-
tion and base cation exchange (Section 9) that might occur in watersheds  over the next 100 years.
This section discusses the application of a dynamic watershed model to project future changes in
surface water chemistry, referred to as Level III analyses.

     Three terms are used to describe simulations of future change:

     •   Predict - to estimate some current or future condition within specified confidence limits on
          the basis of analytical procedures and historical or current observations.

     •   Forecast - to estimate the probability of some future event or condition as a result of rational
          study and analysis of available data.

     •   Project - to estimate future possibilities based on rational study and current conditions or
          trends.

The model output is intended to  distinguish among these three terms and  definitions. Level III
analyses are defined and intended for use as projections.

     Predictions typically are made to compare different scenarios, controls, or management options.
Predictions can be performed within specified confidence limits because of previous model evalua-
tions, testing, applications, and comparisons with measured data for a variety of system types. Model
predictions of various surface water attributes are legally required  for many proposed management
strategies that range, for example,  from examining potential alterations of hydrologic regimes by land-
use modifications to estimating mixing zones for effluent discharges to estimating phytoplankton
response to nutrient reduction.  Predictions generally are performed for short time periods (e.g., single
events, parts of a season, or a few years) and focus on before-after comparisons such as water quality
before and after wasteload  reductions or plankton biomass before and after nutrient reductions.

      Forecasts  convey some estimate of the likelihood or probability that various conditions or events
will occur in the  future.  Daily weather forecasting, with associated probabilities of showers, thunder-
storms, etc., is a classic example of forecasting. This represents a short-term forecast.  Weather
forecasts also are made for annual or decadal time frames.  Flood forecasts can be short term (daily
or weekly), but also are made for long-term events such as the probability of 100-, 1,000-, and
1,000,000-year events (NRC, 1988).

      Projections, in contrast, are not accompanied by estimates of the probability that any of the
conditions or events might  occur in the future.  Projections can be used as a basis for relative
comparisons among various emission or deposition scenarios. Although the probability that a
scenario will occur cannot be estimated, projections do provide a relative basis for comparing costs

                                              251
-------
and beneficial or deleterious effects associated with different control or management strategies.  This
information generally is relevant to policymakers and decisionmakers for evaluating different control
strategies. The models in the Level III analyses are being used for projecting, not forecasting, the
effects of alternative acidic deposition scenarios on future changes in surface water acid-base
chemistry.

      In Level III analyses, integrated, process-oriented watershed models are used to project long-
term changes (E.e., up to 100 years) in surface water chemistry as a function of current and alternative
levels of sulfur deposition. The watershed models integrate our current understanding of how various
processes and mechanisms interact and respond to acidic deposition.  These mechanisms include
soil-water interactions  (including soil-water contact time), sulfur retention, base cation exchange, and
replacement of base cations through mineral weathering, and other watershed processes (e.g., vege-
tative uptake, in-lake processes, organic interactions).  The present study,  however, does not establish
the adequacy of the formulations that implement these processes, the mode of spatial aggregation of
data, or the calibration approaches used for long-term acidification projections.

      The DDRP is an applied project that has used existing techniques and models for these
analyses (Church et al., 1989).  The watershed model that we have applied to watersheds in the
M-APP Region is the Model of Acidification of Groundwater in Catchments  (MAGIC).  The application
of this model to achieve the objectives of the DDRP  was approved by peer reviewers in accordance
with the Agency's standard competitive funding process and requirement for external review of
environmental data collection programs  (Section 4.4.3).  The MAGIC model has been thoroughly
described in the  literature (Cosby et al.,  1985 a,b,c) and has been reviewed by Jenne et al. (1989) and
by Thornton et al. (1990) in the NAPAP State-of-Science/Technology Report No. 14. The MAGIC appli-
cations for lakes  in  the NE and streams  in the SBRP were extensively reviewed as part of the 1989
DDRP Report (Church  et al.,  1989).

     This section presents:

      •   a brief description of MAGIC,

      •   MAGIC calibration and confirmation on Coweeta watersheds,

      •   MAGIC projection and uncertainty  procedures,

      •   regional population estimation and uncertainty procedures,

      •   regional comparisons and uncertainties, and

     •   discussion and conclusions.        •

     Church et al.  (1989) presented information on:

     •   dynamic watershed models used in the Level III analyses,
                                            252
-------
     •   preparation of the modelling datasets (specifically identifying any differences required for
         Level III analyses compared to Level I and II analyses),

     •   general modelling approach,

     •   model sensitivity analyses, and

     •   regional projection refinements.


The procedures that were used for the NE and SBRP (Church et al., 1989) were also followed in the
MAGIC application in the M-APP Region.

10.2 MODEL OF ACIDIFICATION OF GROUNDWATER IN CATCHMENTS (MAGIC)

     MAGIC is a lumped-parameter model of intermediate complexity, originally developed to project
the long-term effects (i.e., decades to centuries) of acidic deposition on surface water chemistry.  One
of the principal assumptions inherent in the model is that a minimum number of critical processes in a
watershed influence the long-term response to acidic deposition.  The model simulates soil solution
chemistry and surface water chemistry to project the monthly or annual average concentrations of
surface water acid-base chemical constituents (Church et al., 1989; Thornton et al., 1990). Hydrologic
flow of water through soil layers to the receiving system is simulated using a separate hydrologic
model, TOPMODEL (Homberger et al., 1985).  TOPMODEL is a topography-based, variable contribu-
ting area, catchment model adapted from the version of Beven and Kirkby (1979). The model con-
siders overland flow, macropore flow, drainage from the upper zone to the lower zone and to the
stream, and baseflow from the lower zone. TOPMODEL provides flow routing through the watershed
to MAGIC, where both equilibrium and rate-controlled expressions are used to represent geochemical
processes.  Mass balances for the major cations and anions and the effects of aqueous aluminum and
organic acid species on ANC are incorporated into MAGIC.  The components included in the ANC
calculation are:
                                                      2'
'].- [H+] - 3[AI3+]
      ANC = [HC03-j + 2[C032-] + [OK] + [HR"'] + 2[R"']
         - 2[AI(OH)2+] - [AI(OH)2+]

where the terms [HR"~] and [R"2"] indicate organic acid fractions.
      The ANC simulated by MAGIC is similar to the modified Gran ANC, but does not consider
 organic acid contributions as part of the DDRP application.  MAGIC includes explicit watershed geo-
 chemical processes of carbonic acid and aluminum chemistry, mineral weathering, anion retention and
 cation exchange.  It implicitly considers biochemical processes of soil nitrification, nutrient uptake and
 soil respiration (Thornton et al., 1990).

      MAGIC represents the watershed with two soil-layer compartments.  These soil layers can be
 arranged vertically or horizontally to represent the vertical or horizontal movement, respectively, of
 water through the soil.  A vertical configuration was used in the DDRP, and the soil compartments
                                            253
-------
 were assumed to be areally homogeneous.  Annual output is the typical temporal resolution of the
 model, but monthly output also can be obtained.  The model requires standard meteorological inputs
 (i.e., precipitation, air temperature) and deposition inputs for base cations, acid anions, ammonia, and
 fluoride on a monthly or annual time scale.

      MAGIC was originally formulated to be parsimonious in selecting processes for inclusion and
 was intended to be used as a heuristic tool for understanding the influences of the selected processes
 on surface water acidification.  The spatialAemporal formulations in the model reflect the intended use
 for assessment and multiscenario evaluations. It was originally developed and applied to a stream in
 the Mid-Appalachian Region (White Oak Run, Shenandoah National Park, VA;  Cosby et al., 1985a,b,c).
 Subsequently, MAGIC has been applied to numerous individual watersheds (Cosby et al.,  1986c;
 Cosby et al., 1990; Jenkins et al., 1990a; Lepisto et al., 1988; Neal et al., 1986; Whitehead  et al.,
 1988a,b).  It has also been applied in a regional manner to groups of watersheds in the United States
 (Homberger et al., 1986), the United Kingdom (Jenkins et al., 1990c; Musgrove et al., 1990) and in
 Norway (Cosby et al., 1987; Homberger et al., 1989; Wright et al., 1991).  As described previously, it
 was one of three models applied in the DDRP studies in the Northeast and Southern Blue  Ridge
 Province (Church et al., 1989).

      The MAGIC model has been tested more than any other watershed acidification model. Projec-
 tions using MAGIC have been compared to (1) diatom reconstructions of lake chemistry, (2) soil
 column leaching experiments, and (3) whole-watershed acidification manipulations.  Applications of
 MAGIC to diatom reconstructions for four lakes (Big Moose Lake in the Adirondacks, Loch  Grannoch
 in Scotland, Lake Gardsjon in  Sweden, and Lake Howatn in Norway) "indicate that the processes
 linked in MAGIC can account for temporal trends in pH and alkalinity such as those obtained from
 paleolimnological data" (Wright et al., 1986).  Jenkins et al. (1990b) compared  MAGIC hindcasts of
 preindustrial (pre-1850) pH with paleolimnology reconstructions for six Scottish lochs. The two
 approaches yielded comparable historical estimates. Sullivan et al. (1991) compared hindcasts by
 MAGIC and paleoecological methods for reconstruction of pre-industrial (pre-1850) surface water
 chemistry for 33 Adirondack lakes studied as part of both the DDRP (Church et al., 1989) and the
 Paleoecological Investigation of Recent Lake Acidification (PIRLA II) (Charles and Smol, 1990). Both
 methods indicated that acidification had occurred. The MAGIC  hindcasts indicated that acidification
 had occurred across much of the ANC range, whereas the paleolimnological procedures (called
 "Diatom" by Sullivan et al., 1991) indicated acidification primarily for lakes with current ANC less than
 50 |ieq/L.  MAGIC hindcasts indicated higher pre-1850 ANC values than did Diatom.  Agreement
 between the methods for pre-1850 pH was very good above about pH 6, with MAGIC indicating higher
 historical pH values than Diatom below that level.  More recently, Cosby and Ryan (1991) have incor-
 porated a triprotic representation of organic acids using the pK values of Driscoll et al. (1990) for
Adirondack lakes into these MAGIC hindcasts and have found that this yields MAGIC hindcasts of pH
 highly similar to those of Diatom.  Differences remain in the hindcast ANC values, probably because
the Diatom model has a "stronger physiological basis" in pH (Sullivan et al., 1990) than it does in ANC
 (Cosby, pers. comm.).                                                                        ,

      With regard to comparisons to  soil column leaching experiments "(a) formulation of the soil
chemical submodel of MAGIC  predicted closely the results of simple laboratory column experiments

                                            254
-------
when flushing effects associated with drying of soils prior to the experiment were eliminated."
"Despite...reservations the results largely substantiate the validity of the fundamental principles behind
the chemical submodel of MAGIC as applied to the Loch Dee catchment.  It is particularly encouraging
that a lumped model such as this can be applied at a very detailed scale and remain generally valid"
(Grieve, 1989).

     Norton et al. (In press) have used MAGIC to project possible effects of ammonium sulfate appli-
cations to West Bear Brook catchment (Maine, USA) as part of the Bear Brook Watershed Manipula-
tion Project. Early comparisons of predicted and observed effects indicate that "after one year of
treatment, the response of the stream chemistry and the response modelled by MAGIC are similar"
(Norton et al., In press). Wright et al. (1990) used MAGIC to predict the result of whole-catchment
manipulations in the Reversing  Acidification in Norway (RAIN) project. "Application of the MAGIC
model to the whole-catchment experimental manipulations of the RAIN project shows that the
response of water and soil acidification to large and rapid changes in acid deposition can be pre-
dicted.  The aggregation approach used in the MAGIC model to describe a complex system of spati-
ally varying chemical and biological reactions produces predicted trends in runoff chemistry that agree
satisfactorily with observed trends."  'These results reinforce  other evaluations of the model obtained
by  comparison with  paleolimnological reconstructions of lake acidification and changes in regional
lake chemistry.  Together these applications give us confidence in the use of the MAGIC model as a
robust tool for predicting future soil and water acidification following changes in acid deposition"
(Wright et al., 1990).

10.3 OPERATIONAL ASSUMPTIONS

      Several operational assumptions are associated both with DDRP and MAGIC (Table 10-1).
These assumptions  underlie the DDRP Level III analyses in toto.  Additional MAGIC assumptions exist
beyond those made for the DDRP. These specific assumptions, summarized by Jenne et al. (1989),
are described in more detail by the authors and developers  of the model (Cosby et al., 1985 a,b,c).

10.4  MODELLING DATASETS

      The MAGIC model, and its hydrologic model, TOPMODEL, require:

      •  meteorological/deposition data,

      •  runoff estimates,

      •  basin  morphometry,

      •  soils, and

      •  surface water chemistry.
-------
Table 10-1.  Level III Operational Assumptions
 1.    Index sample water chemistry from the NSWS provides an index of chronically acidic systems
      and systems with low ANC that are susceptible to acidic deposition.

 2.    index soil data from the DDRP Soil Survey adequately characterize watershed attributes
      influencing surface water chemistry.

 3.    Projections of future acidification  consider primarily chronic acidification. Episodic acidification is
      considered in the EPA Episodic Response Project (Wigington et al., 1990).

 4.    Surface water acidification is a sulfur-driven process. Sulfur is assumed to be the primary
      acidifying agent in acidic deposition.  Eastern deciduous forests generally are nitrogen-limited
      (Likens et al., 1977; Swank and Crossley, 1988) so there is low export of nitrate.  In addition,
      annual nitrate deposition exceeds annual ammonium deposition in the eastern United States
      (Kulp, 1987) and nitrate  has a slight alkalizing effect in the watershed (Lee and Schnoor, 1988).

 5.    The watershed processes controlling the effects of sulfur deposition on surface waters are
      sulfate adsorption and desorption and base cation depletion and resupply through mineral
      weathering and exchange.

 6.    The effects of organic acids on acid-base chemistry are constant through time and independent
      of sulfate.

 7.    These major processes are known well enough to be incorporated into MAGIC.

 8.    Current watershed attributes  and conditions (e.g., climate, land use, basin characteristics) will
      remain relatively constant over the next 50 years.

9.    Long-term projections using MAGIC are plausible and such projections are a feasible approach
      for evaluating the  regional, long-term effects of sulfur deposition scenarios on surface water
      chemistry.

 10.   "Typical" year projections are not  intended to represent future forecasts of water chemistry but
      rather to provide a common basis for comparisons among deposition scenarios to assess
      potential changes in surface water chemistry.

11.   Acidification is reversible and the  processes in MAGIC are  adequate for describing both
      chemical acidification and deacidification.

12.   Physical and chemical processes are adequately considered  in MAGIC.

13.   Uncertainty calculations provide estimates of relative error for long-term comparisons among
      deposition scenarios but are  not absolute error estimates.
                                             256
-------
10.4.1  Meteorological/Deposition Data

     Section 5 discusses the meteorological and deposition data, with the exception of daily meteoro-
logical data, which are specific to the Level III analyses. Meteorological data for daily temperature also
are required as model simulation input for TOPMODEL. These data are measured at fewer locations
than daily precipitation. Typical meteorological year (TMY) data have been produced for 238 locations
across the United States.  These locations are usually in major cities. Ten different TMY sites were
selected and matched to each DDRP site, based on geographic location and elevation. Temperatures
were adjusted  to match 30-year normal temperatures for the period 1951-80.  TMY temperature data
also were adjusted to closely match long-term monthly average temperatures at the TMY sites.
Hence, the monthly and daily temporal pattern for each TMY site was representative of the long-term
norm. Because temperature is elevation dependent, the TMY data were adjusted to match the annual
30-year normal temperature at a nearby site with an elevation comparable to the National Climatic
Data Center (NCDC) site assigned to a DDRP lake.  The adjustment was additive  based on the differ-
ence between the annual average TMY temperature and the annual 30-year normal temperature.

     Watershed specific "typical year" meteorological and deposition data were used for each water-
shed in the M-APP Region.  These typical year data were repeated year after year for 100 years
(Scenario A) in performing the watershed projection under current deposition levels  (Figure 10-1).
Two alternative deposition scenarios also were investigated.  The first alternative deposition scenario
(Scenario B) followed the temporal sequence of current deposition levels for the first 5 years of the
projection,  ramped sulfur deposition down for the next 10 years to achieve a 50 percent reduction in
current deposition, and retained sulfur deposition at this reduced level for the next 85 years.  For the
second alternative (Scenario C), sulfur deposition was  retained at current deposition levels for the first
5 years, was reduced  by 50 percent over 10 years,  but then was ramped up to 20 percent below cur-
rent deposition levels over 35 years and retained at 20 percent below current deposition for the next
50 years (Figure 10-1).

10.4.2 DDRP Runoff Estimation

      The DDRP study sites are not gaged, so measured estimates of runoff were unavailable. A
combination of techniques was used, therefore, to obtain estimates of annual and monthly runoff for
the M-APP watersheds.
 10.4.2.1 Annual Runoff

      Annual runoff was estimated for each of the 36 M-APP watersheds, as discussed in Section 5.7.
 Long-term average annual runoff estimates were based on 1951-80 records.  The annual runoff was
 partitioned into average monthly fractions for use in calibrating the hydrologic submodel, TOPMODEL.

 10.4.2.2 Monthly Runoff

      Monthly runoff fractions for each M-APP DDRP site were determined  using the long-term monthly
 mean discharges from USGS gaging station records.  Because none of the DDRP basins had gaging

                                             257
-------
            Change   from   1985  (X)
       .£ o
                                    SOX decrease
                             20     30     40

                            Time,  yrs.
Figure 10-1.  Three deposition scenarios used in the M-APP Region. The constant (current)
         deposition scenario is Scenario A. The 50 percent decrease is Scenario B
         The 50/20 percent decrease is Scenario C.
                            258
-------
stations at the downstream nodes, nearby gaging stations in basins with similar characteristics were
assumed to have similar monthly runoff patterns.

      Initially, the USGS gaging sites and DDRP sites were co-located on the M-APP runoff contour
map (Krug et al., 1990). Any USGS gaging stations outside a 50-mile radius of the cluster of all DDRP
sites were eliminated from further  analysis. Monthly and annual mean discharges for the period of
record for each station were obtained from the USGS (D. Graczyk, pers. comm.) for each of the
remaining gages.  The monthly flow fractions were computed as the ratio of the monthly average dis-
charges to annual discharges for the period of record.  The monthly runoff fractions were plotted to
determine the magnitude of variation in monthly runoff between stations (e.g., Figure 10-2).  The
difference between monthly fraction curves for the USGS stations was no greater than approximately 5
percent per month with the exception of two stations. One station was regulated upstream by mine
pumps and the other station had  only nine years of record.  The two stations were omitted from further
consideration. The monthly flow fractions were assigned to DDRP watersheds based on the following
criteria:

      •  proximity of the basin and its position within runoff contours relative to the DDRP site,

      •  similarity in basin topography, size and elevation to the DDRP site, and

      •  quality of the discharge  record (reported by the USGS).

      Table 10-2 shows the relationship between the DDRP sites and the USGS gaged sites, including
the calculated monthly fractions for each DDRP site and the parameters used to assign the gage to
the DDRP site.

10.4.3 Watershed and Stream Morphometrv Data

      Basin, lake, and stream morphometry and characteristics are discussed in Section 5.4.  These
data, obtained from the DDRP Soil Survey and the NSS for the M-APP, were used in model calibration
for each  watershed  in the M-APP.

10.4.4 Soil/Stream Physical and Chemical  Data

      The soils data, discussed in Section 5.5, were obtained from the DDRP Soil Survey, aggregated
(Section 9), and  used for model calibration.  Surface water chemistry data were obtained from the
NSS and have been described in detail by Kaufmann et al. (1988).

10.4.5 Other Data

       Watershed data such as bedrock geology, land use, vegetative cover, estimated depth to bed-
rock, and other data also were used for calibration of the individual watersheds. These data are
discussed in Sections 5.4 and 5.5.
                                             259
-------
Figure 10-2.  Comparison for consistency among different USGS gages within the M-APP
            Region for monthly flow fractions.
                                       260
-------
Table 10-2.  Monthly Runoff Fractions for Selected USGS Gaging Stations
USGS STA.
01542000 (PA)
ETSPDR"
MEAN ANNUAL
RUNOFF (cfe)
     111
        MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN  FEB   MAR   APR   MAY   JUN   JUL  AUG  SEP  OCT  NOV   DEC
8.24 10.41   17.59  16.40  12.95   8.09   4.09  3.17  2.55  3.42   5.17   7.94

DDRP SITE
2C028069L(PA)
2C028075L(PA)
2C028070L(PA)
USGS STA.
01544500 (PA)
ETPDR"
EQUIVALENT ANNUAL
RUNOFF DEPTH (in)
19
20
19
MEAN ANNUAL
RUNOFF (cfs)
226
EQUIVALENT MONTHLY RUNOFF DEPTH pn)
JAN
1.56
1.65
1.56
FEB
1.98
2.08
1.98
MAR
3.34
3.52
3.34
APR MAY
3.12 2.46
3.28 2.59
3.12 2.46
JUN
1.54
1.62
1.54
JUL
0.78
0.82
0.78
AUG
0.60
0.63
0.60
SEP OCT
0.48 0.65
0.51 0.68
0.48 0.65
NOV
0.98
1.03
0.98
DEC
1.51
1.59
1.51
MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN
7.82
FEB
9.40
MAR
18.51
APR MAY
18.70 12.35
JUN
6.20
JUL
2.98
AUG
1.73
SEP OCT
2.16 3.67
NOV
7.34
DEC
9.15

DDRP SITE
2C029002L(PA)
2C029016L(PA)
2C029020L(PA)
USGS STA.
0155000 (PA)
ETPDR8
EQUIVALENT ANNUAL
RUNOFF DEPTH (in)
21
19
22
MEAN ANNUAL
RUNOFF (cfs)
284
EQUIVALENT MONTHLY RUNOFF DEPTH On)
JAN
1.64
1.49
1.72
FEB
1.97
1.79
2.07
MAR
3.89 "
3.52
4.07
APR
3.93
3.55
4.11
MAY JUN
2.59 1 .30
2.35 1.18
2.72 1.36
JUL
0.62
0.57
0.65
AUG
0.36
0.33
0.38
SEP OCT
0.45 0.77
0.41 0.70
0.48 0.81
NOV
1.54
1.39
1.61
DEC
1.92
1.74
2.01
MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN
7.43
FEB
8.52
MAR
17.62
APR
18.36
MAY JUN
11.66 6.29
JUL
3.35
AUG
2.40
SEP OCT
2.76 4.20
NOV
8.61
DEC
8.81

DDRP SITE
1D029023L(PA)
1 D029031 L(PA)
USGS STA.
01552500 (PA)
ETPDR3
EQUIVALENT ANNUAL
RUNOFF DEPTH (in)
21
21
MEAN ANNUAL
RUNOFF (cfs)
48.6
EQUIVALENT MONTHLY RUNOFF DEPTH (in)
JAN
1.56
1.56
FEB
1.79
1.79
MAR
3.70
3.70
APR
3.85
3.85
MONTHLY
JAN
8.47
FEB
9.44
MAR
15.15
APR
15.51
MAY JUN
2.45 1.32
2.45 1.32
JUL
0.70
0.70
AUG
0.50
0.50
SEP OCT
0.58 0.88
0.58 0.88
NOV
1.81
1.81
DEC
1.85
1.85
FRACTION OF MEAN ANNUAL RUNOFF (%)
MAY JUN
10.69 5.52
JUL
3.14
AUG
2.57
SEP OCT
3.74 5.20
NOV
9.69
DEC
10.88
DDRP SITE
1D029042L(PA)
1D029043L(PA)
USGS STA.
01542000 (PA)
ETSDR8
EQUIVALENT ANNUAL
RUNOFF DEPTH (in) JAN
21
20
MEAN ANNUAL
' RUNOFF (cfs)
111
1.78
1.69
FEB
1.98
1.89
MAR
3.18
3.03
EQUIVALENT MONTHLY RUNOFF DEPTH (in)
APR MAY JUN JUL AUG SEP OCT
3.26
3.10
2.24 1.16
2.14 1.10
0.66
0.63
0.54
0.51
0.79 1.09
0.75 1.04
NOV
2.04
1.94
DEC
2.28
2.18
MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN
8.24
FEB
10.41
MAR
17.59
APR
16.40
MAY JUN
12.95 8.09
JUL
4.09
AUG
3.17
SEP OCT
2.55 3.42
NOV
5.17
DEC
7.94
 DDRP SITE
 2C035027L(PA)
 EQUIVALENT ANNUAL
 RUNOFF DEPTH (in)
      25
                  EQUIVALENT MONTHLY RUNOFF DEPTH (in)
 JAN  FEB  MAR   APR    MAY   JUN  JUL   AUG  SEP   OCT  NOV   DEC
 2.06  2.60  4.40    4.10    3.24   2.02  1.02   0.79  0.64   0.86  1.29    1.98
                                                                                                  (Continued)
   Parameters used in selecting USGS stations:

   E - Elevation
   T - Topography
   S - Size of watershed
   P - Proximity of USGS Station to DDRP Station
   D - Data good
   R - Runoff contours
                                                     261
-------
 Table 10-2  Continued
 USGS STA.
 01538000 (PA)
 ETSOR*
 DDRP SITE
 2B036062L(PA)
 MEAN ANNUAL
 RUNOFF (cfe)
     64.8
               MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
 JAN   FEB   MAR APR  MAY   JUN  JUL  AUG  SEP  OCT  NOV  DEC
 8.85  10.01  15.13 14.82  11.04   6.32  4.56  3.14  3.55  4.87  7.51  9.50
DDRP SITE
1D036011L(PA)
2B038028L(PA)
USGS STA.
01564500 (PA)
ETPDR*
EQUIVALENT ANNUAL
RUNOFF DEPTH (in)
17
16
MEAN ANNUAL
RUNOFF (cfe)
243
JAN
1.50
1.42
JAN
8.01
EQUIVALENT MONTHLY RUNOFF DEPTH On)
FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
1.85 2.57 2.48 1.88 1.07 0.78 0.53 0.60 0.83 1.28 1.62
1.75 2.42 2.34 1.77 1.01 0.73 0.50 0.57 0.78 1.20 1.52
MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
13.63 18.19 16.43 10.80 6.54 2.53 2.12 1.87 3.68 6.27 8.93
 EQUIVALENT ANNUAL
 RUNOFF DEPTH On)      JAN
     17               1.36
                  EQUIVALENT MONTHLY RUNOFF DEPTH (in)
       FEB   MAR  APR  MAY ° JUN  JUL  AUG   SEP   OCT  NOV  DEC
       2.32   326  2.79  1.84  1.11   0.43  0.36   0.32   0.63   1.07  1.52
 USGS STA.
 01448500 (PA)
 ETSOR1
DDRP SITE
2B041008L(PA)
MEAN ANNUAL
RUNOFF (cfe)
     4.86
               MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN    FEB   MAR  APR  MAY  JUN   JUL  AUG   SEP   OCT  NOV  DEC
8.66    8.82  13.36 15.53 10.67   7.29   4.83  3.84   3.89   4.98   7.67 10.36
DDRP SITE
1D037005L(PA)
1D038017L(PA)
2B03S040L(PA)
USGS STA.
01603500 (PA)
ETPDR*
EQUIVALENT ANNUAL
RUNOFF DEPTH On)
27
27
25
MEAN ANNUAL
RUNOFF (cfe)
32.4
JAN
2.34
2.34
2.17
JAN
8.90
FEB
2.38
2.38
2.20
FEB
11.98
MAR
3.61
3.61
3.34
EQUIVALENT MONTHLY RUNOFF DEPTH
APR MAY JUN JUL AUG SEP
4.19 2.88 1.97 1.30 1.06 1.05
4.19 2.88 1.97 1.30 1.06 1.05
3.88 2.67 1.82 i.21 0.98 .097
On)
OCT
1.35
1.35
1.25
NOV
2.07
2.07
1.82
MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
MAR APR MAY JUN JUL AUG SEP OCT NOV
19.71 16.50 11.70 7.06 3.46 2.57 2.55 4.13 4.11
DEC
2.80
2.80
2.59
DEC
7.34
EQUIVALENT ANNUAL
RUNOFF DEPTH On)      JAN
     15               1.34
                  EQUIVALENT MONTHLY RUNOFF DEPTH On)
      FEB   MAR  APR  MAY  JUN  JUL   AUG   SEP   OCT NOV  DEC
      1.80   2.96  2.47  1.76   1.06  0.52   0.38   0.38   0.62  0.62   1.10
USGS STA.
01620500 (VA)
ETSDR1
DDRP SITE
2B047036L(VA)
2B047089L(VA)
2B047076L(VA)
MEAN ANNUAL
RUNOFF (cfe)
     25.8

EQUIVALENT ANNUAL
RUNOFF DEPTH On)
     11
     15
     15
               MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN   FEB   MAR  APR  MAY  JUN   JUL   AUG   SEP   OCT  NOV  DEC
9.17  11.84  17.01  14.79 10.93   7.69   2.02   2.38   2.23   4.63  8.43   8.88

                  EQUIVALENT MONTHLY RUNOFF DEPTH On)
JAN   FEB   MAR  APR  MAY  JUN   JUL   AUG   SEP   OCT  NOV  DEC
1.01   1.30   1.87   1.63  1.20   0.85   0.22   0.26   0.25   0.51  0.93   0.98
1.37   1.78   2.55   2.22  1.64   1.15   0.30   0.36   0.33   0.69  1.26   1.33
1.37   1.78   2.55   2.22  1.64   1.15   0.30   0.36   0.33   0.69  1.26   1.33
USGS STA.
01666500 (VA)
ETSPDR*
DDRP SITE
2B047066L(VA)
MEAN ANNUAL
RUNOFF (cfe)
     220

EQUIVALENT ANNUAL
RUNOFF DEPTH On)
     17
              MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN   FEB   MAR  APR  MAY   JUN  JUL  AUG   SEP   OCT  NOV DEC
9.27  11.20  12.41  11.62  9.58   7.46  4.73  5.49   5.64   6.13  7.91  8.55

                  EQUIVALENT MONTHLY RUNOFF DEPTH On)
JAN   FEB   MAR  APR  MAY   JUN  JUL  AUG   SEP   OCT  NOV DEC
1.58   1.90   2.11   1.98  1.63   1.27  0.80  0.93   0.96   1.04  1.34  1.45
  Poramatets used in selecting USGS stations:

   E - Etavation
   T - Topography
   S - Size of watershed
   P - Proximity of USGS Station to DDRP Station
   D - Daia good
   R - Runoff contours
                                                                                                        (Continued)
                                                      262
-------
Table 10-2  Continued
USGS STA.
03070500 (PA)
ETPDR8
DDRP SITE
2C041002L(PA)
MEAN ANNUAL
RUNOFF (cfe)
     418

EQUIVALENT ANNUAL
RUNOFF DEPTH (in)
     29
              MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN   FEE   MAR APR  MAY  JUN  JUL  AUG   SEP   OCT  NOV DEC
12.59 13.74  15.88 12.77  9.53  5.86  3.43  2.76   2.42   3.41  6.59 11.02

                 EQUIVALENT MONTHLY RUNOFF DEPTH (in)
JAN   FEE   MAR APR  MAY  JUN  JUL  AUG   SEP   OCT  NOV DEC
3.65   3.98   4.61  3.70  2.76  1.70  1.00  0.80   0.70   0.99  1.91  3.20
USGS STA.
0306900 (WV)
ETPDR8
DDRP SITE
2C047007L(WV)
MEAN ANNUAL
RUNOFF (cfe)
     771
              MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN   FEE   MAR APR  MAY  JUN  JUL  AUG   SEP   OCT  NOV DEC
10.91 11.86  15.58 12.71  9.04  6.12  4.82  4.16   3.21   4.59  7.22  9.79

DDRP SITE
2C041039L(WV)
2C041 040L(WV)
2C041045L(WV)
2C041051L(VW)
USGS STA.
03180500 (WV)
ETPDR8
EQUIVALENT ANNUAL
RUNOFF DEPTH (in)
31
33
33
35
MEAN ANNUAL
RUNOFF (cfe)
262
EQUIVALENT MONTHLY RUNOFF DEPTH
JAN
3.38
3.60
3.60
3.82
FEE MAR
3.68 4.83
3.91 5.14
3.91 5.14
4.15 5.45
MONTHLY FRACTION
JAN
11.03
FEE MAR
APR
3.94
4.19
4.19
4.45
MAY
2.80
2.98
2.98
3.16
JUN
1.90
2.02
2.02
2.14
OF MEAN ANNUAL
APR
13.82 18.08 13.89
MAY
9.31
JUN
5.24
JUL
1.49
1.59
1.59
1.69
AUG
1.29
1.37
1.37
1.46
SEP
1.00
1.06
1.06
1.12
(in)
OCT
1.42
1.51
1.51
1.61

NOV
2.24
2.38
2.38
2.53

DEC
3.03
3.23
3.23
3.43
RUNOFF (%)
JUL'
3.08
AUG
2.63
SEP
2.01
OCT
3.50
NOV
7.09
DEC
10.33

DDRP SITE
2C047010L(WV)
2B047032L(WV)
USGS STA.
03050500 (WV)
ETPDR8
EQUIVALENT ANNUAL
RUNOFF DEPTH (in)
27
19
MEAN ANNUAL
RUNOFF (cfe)
516
EQUIVALENT MONTHLY RUNOFF DEPTH (in)
JAN FEE MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
2.98 3.73 4.88 3.75 2.51 1.42 0.83 0.71 0.54 0.94 1.91 2.79
2.09 2.63 3.44 2.64 1.77 1.00 0.59 0.50 0.38 0.66 1.35 1.96
MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN FEE MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
12.02 14.43 16.56 12.64 8.92 5.62 3.61 3.11 2.08 3.17 6.72 11.11
EQUIVALENT ANNUAL
RUNOFF DEPTH (in)      JAN
     30               3.61
                 EQUIVALENT MONTHLY RUNOFF DEPTH (in)
      FEE   MAR APR  MAY   JUN  JUL  AUG   SEP   OCT NOV  DEC
      4.33   4.97 3.79  2.68   1.69  1.08  0.93   0.62   0.95  2.02  3.33
USGS STA.
03186500 (WV)
ETPDR8
MEAN ANNUAL
RUNOFF (cfe)
     262

EQUIVALENT ANNUAL
               MONTHLY FRACTION OF MEAN ANNUAL RUNOFF (%)
JAN   FEB   MAR APR MAY   JUN  JUL  AUG   SEP   OCT  NOV  DEC
11.06 12.98  16.57 12.46 8.57   5.13  4.81  4.19   2.26   4.24  7.47 10.26

                 EQUIVALENT MONTHLY RUNOFF DEPTH (in)
DDRP SITE
2C046033L(WV)
2C046034L(WV)
2C046005L(WV)
2C046041L(WV)
2C046050L(WV)
2C057004L(WV)
RUNOFF DEPTH (in) JAN
34
37
21
30
24
22
3.76
4.09
2.32
3.32
2.65
2.43
FEB
4.41
4.80
2.73
3.89
3.12
2.86
MAR
5.63
6.13
3.48
4.97
3.98
3.65
APR
4.24
4.61
2.62
3.74
2.99
2.74
MAY
2.91
3.17
1.80
2.57
2.06
1.89
JUN JUL AUG
1.74
1.90
1.08
1.54
1.23
1.13
.64 1.42
.78 1.55
.01 0.88
.44 1.26
.15 1.01
.06 0.92
SEP
0.77
0.84
0.47
0.68
0.54
0.50
OCT
1.44
1.57
0.89
1.27
1.02
0.93
NOV DEC
2.54 3.49
2.76 3.80
1.57 2.15
2.24 3.08
1.79 2.46
1.64 2.26
  Parameters used in selecting USGS stations:

   E - Elevation
   T - Topography
   S - Size of watershed
   P - Proximity of USGS Station to DDRP Station
   D - Data good
   R - Runoff contours
                                                      263
-------
10.5  CALIBRATION OF MAGIC
10.5.1 Calibration Procedure
      MAGIC represents the horizontal dimension of the watershed as a homogeneous unit with no
subcatchments and the vertical dimension as two soil compartments.  Watershed data for MAGIC were
lumped or aggregated to provide average or weighted average values for each of the soil layers. The
top soil compartment represented the mass-weighted average conditions  of the A and B master hori-
zons. The lower soil compartment represented the mass-weighted average conditions in the C master
horizon.

      Projecting long-term effects of acidic deposition on surface water chemistry using MAGIC
involves coupling MAGIC with TOPMODEL (Cosby et al.,  1985a,b,c).  Both models were calibrated
using an optimization procedure that selected parameters so that the difference between the observed
and predicted measurements was minimized. The calibration exercise was a three-step  process. The
first step was to specify the model inputs such as precipitation, deposition (both wet and dry), an
estimate of historical inputs for the long-term chemical model, and fixed parameters or parameters
whose values correspond directly to (or can be computed directly from) field measurements, e.g.,
topographic variables such as slope, aspect, area.  This approach, in effect, assigns all the uncertainty
associated with sampling, aggregation, and intrinsic variability to the "adjustable" parameters.  The
adjustable parameters are those that are calibrated or scaled to match observed field measurements.

      The second step  was to select optimal values for the adjustable parameters.  The optimization
method of Rosenbrock  (1960) was employed. Optimal values were determined by minimizing a  loss
function defined by the sum of squared errors between simulated and observed values of system state
variables. Different loss functions were used for the hydrologic and chemical models. The hydrologic
model used  monthly flow estimates.  The chemical model used available stream chemistry records
(i.e., weekly, monthly observations) and observed soil chemistry.

      The final step was to assess the structural adequacy of the model in reproducing the observed
behavior of the criterion variables and parameter identifiability or the uniqueness of the set of optim-
ized parameters. Structural adequacy was assessed by examining the mean error in simulated values
of observed  state variables for those variables used in the calibration procedure, as well as for an
additional state variable that was not used during calibration. Parameter identifiability was assessed
using approximate estimation error variances for the optimized parameters (Bard,  1974).  Cosby et al.
(1989) have  presented additional information on the MAGIC calibration process.

10.5.2 Calibration/Confirmation Studies

10.5.2.1  Study Watersheds

      We used data from Watersheds #34 and #36 of the Coweeta Hydrologic Laboratory for calibra-
tion and confirmation studies.  The USDA Forest Service Coweeta Hydrologic Laboratory, Otto, North
Carolina, encompasses about 2,184 ha (5,400 ac) and contains over 50 experimental watersheds.

                                            264
-------
Topography at Coweeta is steep, with elevations ranging from 686 to 1,600 m and average side
slopes of 50 percent (Swank and Crossley, 1988).  The Coweeta regolith is deeply weathered and
averages about 7 m in depth, which varies dramatically throughout the basin.  Soils generally occur
within two orders - fully developed Ultisols and immature Inceptisols. Meteorological, deposition,
hydrologic, and stream  water chemistry data are monitored at several locations throughout the basin.

      Coweeta Watershed 36 is located at a  higher elevation in the Coweeta Experimental Forest than
Coweeta Watershed 34 and has different hydrologic and chemical characteristics (Table 10-3).  The
Forest Service has compiled a continuous 12-year hydrologic and stream chemistry record on
Coweeta Watershed 36, from June 1973 through May 1985. Data are generally available for the same
period for Coweeta Watershed 34, with the exception of stream chemistry from June 1975 through
May 1980, for which period data are unavailable.  These records for the two watersheds were
partitioned into calibration and confirmation data sets to evaluate longer simulation periods, using the
MAGIC model, than were possible with the data sets available for northeastern lakes (see Church et
al., 1989).  The periods were:
     Watershed
Calibration
Confirmation
      Watershed 36
      Watershed 34
June 73 - May 79
June 82 - May 85
June 79 - May 85
June 73 - May 75
Nov. 80 - May 82
The confirmation period for Watershed 36 represented a forecast, whereas the confirmation period for
Watershed 34 represented a hindcast.

      In the calibration of Coweeta watersheds 34 and 36, TOPMODEL used daily stream flow
volumes.  This was a departure from the standard DDRP modelling protocol (Church et al., 1989) in
which monthly flow fractions were used in the calibration process. This should not introduce any
significant inconsistency, however, inasmuch as Coweeta has no snowpack or overland flow, and thus
we can reliably estimate (on either a daily or monthly basis) that all atmospheric deposition interacts
with one or both soil horizons (J. Cosby, pers. comm.)

10.5.2.2  Results

      Calibration/confirmation results were presented previously for the intensively studied watersheds
in the NE—Woods Lake, Panther Lake, and Clear Pond (Church et al., 1989). MAGIC was developed
for southeastern streams and calibration exercises on these streams have been described elsewhere
(Cosby et al., 1985a,b,c).

      The approach used to calibrate MAGIC on the two Coweeta watersheds was similar to the DDRP
calibration protocol used for the target population projections.  Average annual hydrologic and
chemical conditions were computed for the multiple years of calibration data. The MAGIC model was
then calibrated to these average annual conditions, rather than  the multiple years of calibration data,
to represent the DDRP approach of projecting "typical" year conditions. This approach permitted an

                                             265
-------
Table 10-3.  Inputs, Watershed Characteristics, and Surface Water Chemistries for Coweeta
            Watersheds
SITE CHARACTERISTIC8
         COWEETA
            36
    COWEETA
       34
ATMOSPHERIC
   Avg. Annual Precipitation (cm)
   Volume Weighted pH
   Sulfate Deposition (keq/ha-yr)
   Nitrate Deposition1 (keq/ha-yr)

WATERSHED CHARACTERISTICS
   Basin  Area (km2)
   Relief  (m)
   Depth to Bedrock (m)
   Dominant Soil
   Estimated SO4'2 Sorption
   Vegetative Cover

SURFACE WATER CHEMISTRY
   Average pH
   Average ANC fueq/L)
   Average SO4~Z (aeq/L)
   Average NO3"4 (weq/L)
   Average Annual Flow (cm)
          222
            4.6
            0.7
            0.2
            0.5
          520
            5
        Udults
         high
N Hardwood-Oak-Hickory
            6.5
           55
           19
            1
          168
     201
       4.6
       0.7
       0.2
       0.4
     330
    deep
   Udults
    high
Oak-Hickory
       6.8
      85
       9
       0
     118
0 Swank and Crossley, 1988
                                          266
-------
evaluation of the typical year conditions used in DDRP regional calibrations and made use of the
capability of the MAGIC model for predicting longer term records of stream chemistry.  Although these
projections do not confirm MAGIC'S capability for 100-year projections, they do permit a comparison
of long-term field observations with DDRP "calibrated" values and an assessment of the ability of the
MAGIC model to simulate a multi-year (< 15 yr) period of record.

     The root mean squared error (RMSE) is based on differences between simulated and observed
values over the period of interest. The magnitude of the RMSE is a function of (1) the ability to
correctly simulate the mean value of the observations (i.e., lack of bias in the model), and (2) the
ability to correctly simulate the variation about the mean value (i.e., ability to reproduce seasonal
variation or long-term trends).  For an unbiased model, a comparison  of the RMSE to the standard
deviation of the observed  data can be useful in determining the amount of seasonal or long-term
variation that  is reproduced  (or missed) by the simulations.

     MAGIC  was calibrated to the averages of the observations for the watersheds. Therefore, we
presume that the MAGIC results are unbiased with regard to the means of the observed values.  We
do not expect a priori, however, to capture the variation about the mean values, because the
information on variability was not used in the calibration procedure.  Therefore, a value of the RMSE
approximately equal to the standard deviation of the observations is considered acceptable in terms of
model behavior.  If, in fact, the RMSE is less than the standard deviation, the model simulations are
reproducing some of the dynamic behavior of the system (i.e., the seasonal variations or long-term
trends).  The  calibration/confirmation results, expressed as RMSEs for Watersheds 34 and 36 are
shown in Table 10-4. For most of the variables shown in Table 10-4, the RMSEs are less than, or
approximately equal to, the  standard deviations of the observations.

     There was generally good agreement between the predicted versus the observed values for both
flow and calculated ANC for Watershed 36 over the 12-year period from 1973 through 1985 (Figure
10-3). The RMSE for flow was about 3 cm/mo, whereas the RMSE for ANC was about 8 ^eq/L during
the calibration period and 9 /ueq/L during the confirmation period (Table 10-4). Predicted ANC
concentrations generally were less than observed  values during the calibration period and greater than
observed values during the confirmation period (Figure 10-3).  In the MAGIC model, ANC is a
calculated, not a calibrated, variable.  Calcium, a base cation that is included in the calibration
process, exhibited a pattern similar to that of ANC, with generally good agreement between simulated
and observed values (Figure 10-4).  Simulated calcium concentrations also were slightly greater than
observed values during the confirmation period. Simulated sulfate concentrations were more variable
than calcium  and were not as well correlated with  observed values (Figure 10-4).  The temporal
correspondence between simulated and observed values, including the slight increasing  trend in
observed sulfate concentrations, was represented but there were differences in magnitude between
predicted and observed sulfate  concentrations. An increasing trend in stream sulfate concentration of
about 0.7 fteq/L/yr has been observed in Coweeta Watershed 36 (Waide and Swank, 1987; J. Waide,
pers. comm.). The calibrated MAGIC trend is less than 2/u.eq/L/yr, which is expected because MAGIC
was calibrated on a single composite average  annual constituent value for each calibration variable, as
per the  standard DDRP protocols (Section 10.5; also, Church et al., 1989). The RMSEs for the
individual chemical species  generally were less than 5 /ueq/L throughout both calibration  and
                                             267
-------
Table 10-4. RMSE Values Based on Volume-Weighted Monthly Averages of Simulated and
          Observed Quantities for Coweeta Watersheds over a 12-Year Period of Record
Coweeta Watershed 36
All data


Ca
Mg
Na
K
NH4
SO4
Cl
NO3
SBCd
SAA8
Calkf
Flow8




























Sim.

3.4°
2.7
3.2
1.2
1.2
4.7
4.7
1.7
8.9
8.0
8.4
3.3
Obs

4.4
2.9
3.8
1.0
.2
4.6
1.6
1.0
10.9
5.3
11.4
9.8














Calib.
Sim.

3.7
2.7
3.4
1.2
1.1
4.8
4.7
1.7
9.3
7.9
8.1
3.3
period3
Obs.

4.7
3.3
4.1
1.1
.2
3.8
1.5
1.1
11.7
4.4
10.9
10.6
Valid periodb
Sim.

3.1
2.7
2.9
1.2
1.3
4.7
4.8
1.7
8.5
8.2
8.7
3.2
Obs.

3.7
2.2
3.5
.8
.1
3.9
1.7
.8
9.3
4.8
9.2
8.9




























Coweeta Watershed 34



Ca
Mg
Na
K
NH4
SO4
Cl
NO3
SBC
SAA
Calk
Flow



2
2
3
1

3
2

8
4
7
2
All
Sim

.7
.5
.3
.0
.5
.0
.0
.6
.2
.2
.7
.0
data
. Obs.

3.7
2.8
3.5
1.0
.1
2.8
1.2
.2
9.7
3.5
9.4
5.3















Calib.
Sim.

2.6
2.4
3.0
1.2
.4
2.4
1.7
.6
8.3
3.6
7.1
2.2
period*1
Obs.

3.7
2.5
3.0
1.0
.1
2.1
1.2
.2
8.8
2.7
8.4
5.3
0 Calibr. period = June 73 through May 79 (n=72 months). '
b Valid,
c Units
d Sum
0 Sum
, period = June 79 through May 85
(n=72 months)
« fteq/L (except Flow - cm/mo).
of base cations =
(Ca
of acid anions = (SO4
+ Mg + Na
+ NO3 + O
+ K).
+ F).



h
f
'
Valid
Sim.

2.8
3.1
3.8
.8
.6
2.8
2.9
.6
8.5
4.8
10.9
1.4
SBC-SAA.
period 11
Obs.

3.9
3.1
3.9
1.3
.1
3.7
1.3
.3
10.7
4.2
9.9
6.2

9 RMSE based on total
Calibr. period :
Valid, period 1
Valid, period 2
Valid period 2]















Sim.

2.7
1.9
3.3
1.0
.5
3.9
1.3
.6
7.9
4.5
4.7
1.8

Obs.

4.7
3.4
3.9
1.0
.2
2.8
1.3
.1
12.3
3.4
11.6
3.1

monthly discharge.
= June 82 through May 85 (n=36 months).
= June 73 through Apr 75 (n=
= Nov 80 through May 82 (n=
17 months).
19 months).
                                         268
-------
   120
   100
                   COWEETA WATERSHED 36       (	simulated  	observed)
o
    60
    40
    20
        1974
1976
1978
                                                  1980
                                           1982
                                          1984
        1974
1976
1978
                                                  1980
1982
                                                         1984
                                              Year
 Figure 10-3.  Simulated and observed values for ANC and flow in Coweeta watershed 36
              using the MAGIC model.
                                           269
-------
      50


      40


      30


      20


      10 -


       0
                     COWEETA WATERSHED 36      (	simulated 	observed)
          1974
1976
                                     1978
                                                   1980
                                                                 1982
                                                                               1984
          1974
1976
                                     1978
                           1980
                                               Year
                                                                 1982
                                                       1984
Figure 10-4.  Simulated and observed values for calcium and sulfate concentrations In Coweeta
             Watershed 36 using the MAGIC model.
                                            270
-------
confirmation periods. The individual base cation species typically had better precision than the
individual acid anion species, with an RMSE of about 3 fieq/L for base cations and about 5 fieq/L for
sulfate and chloride. Both potassium and nitrate, which are biologically active chemical species, had
low concentrations in the stream and low standard errors (Table 10-4).  The RMSEs for sum of base
cations, however, was  equivalent to that of the sum of acid anions.

      Predicted versus observed flow and calculated ANC also generally agreed for Watershed 34
(Figure 10-5). The calculated ANC was somewhat lower during the 1973-75 confirmation period, but
was similar to the seasonal trends, which were consistent regardless of the calibration or confirmation
period.  The RMSE for calculated ANC varied  from about 5 ,«eq/L during one confirmation period to
about 10/^eq/L during the second  confirmation period (Table 10-4).  Simulated calcium concentra-
tions, again, exhibited  a  pattern similar to that of ANC (Figure 10-6).' The simulated calcium
concentrations were lower during the confirmation periods and slightly higher during the calibration
period.  Simulated sulfate concentrations also were more variable than calcium and ANC concentra-
tions (Figure 10-6).  Sulfate concentrations were lower than observed concentrations during the first
confirmation period, higher during the second confirmation period, and similar during the calibration
period (Figure 10-6). Simulated RMSEs for individual base cation and acid anion species in Coweeta
Watershed 34 were similar, about 3 /weq/L, for both calibration and confirmation periods, but the sum
of base cations had about twice the RMSEs of the sum of acid anions (Table 10-4).

      In general, the Coweeta simulations indicate that, when calibrated according to the DDRP
protocol, MAGIC is capable of predicting seasonal trends comparable to observed data for longer term
periods of record but might not capture relatively subtle long-term trends in the record. MAGIC was
calibrated to a single composite average annual value for each  calibration  variable, which does not
have a trend. Although  MAGIC will simulate long-term trends driven by trends in the input data, the
projected rate of change is likely to be underestimated because MAGIC did not incorporate these
trends during calibration. The DDRP protocol of calibration using typical year values captures the
general seasonal and  annual pattern and should permit comparisons among deposition scenarios.

10.6  MODEL PROJECTIONS FOR MID-APPALACHIAN REGION

10.6.1  General Approach

      This section discusses the general approach for performing long-term  projections of the effects
of sulfur deposition on surface water chemistry over the next 100 years in the M-APP  Region. The
three simulated deposition scenarios are illustrated in Figure 10-1.  In the first scenario (Scenario A),
current deposition levels throughout the M-APP Region were maintained over the 100-year interval.
Changes in stream water chemistry were projected over this same 100-year period.  In the second
deposition scenario (Scenario B), current deposition levels at each DDRP site were held constant for
the first 5 years, decreased by a total of 50 percent over the next 10 years, and then held constant at
this reduced deposition  rate for the next 85 years.  This phasing corresponded with one possible
scenario of how deposition rates might change if emissions controls were  implemented (U.S. EPA,
Office of Policy, Planning and  Evaluation, pers. comm.).  For the third scenario (Scenario C),
                                             271
-------
   120
   100
    60
    40
    20
                   COWEETA WATERSHED 34       (	simulated
                                                 observed)
        1974
1976
1978
1980
1982
1984
    SO

    40

    30
E
        1974
1976
1978           1980
          Year
              1982
              1984
  Figure 10-5. Simulated and observed values for ANC and flow in Coweeta watershed 34 using
              the MAGIC model.
                                             272
-------
    50


    40


1?  30
3
a
u   20


    10
                   COWEETA WATERSHED 34
                                             (	 _ simulated
                                  observed)
        1974
                  1976
1978
.1980  '
1982
1984
50


40


30


20


10
•a-
O
00
        1974
                  1976
1978
                                                   1980
               1982
              1984
                                               Year
   Figure-10-6.  Simulated and observed values for calcium and sulfate concentrations in
                Coweeta Watershed 34 using the MAGIC model.
                                             273
-------
 deposition rates were held constant for the first 5 years, decreased by a total of 50 percent over the
 next 10 years, increased to 20 percent below current deposition for the next 35 years, and then held
 constant for the remaining 50 years (Figure 10-1). This scenario also corresponds with one possible
 emission control strategy.

      MAGIC was calibrated on individual watersheds using the data sources indicated in Section
 10.5. The projected change in surface water chemistry at the NSS downstream node was simulated
 for the next 100 years using the typical year meteorology and deposition data, discussed in Section
 10.5. The model output represents flow-weighted annual average constituent concentrations.

      Not all watersheds in the M-APP were simulated. There were some watersheds for which the
 optimization and calibration criteria were not satisfied.  Long-term projections for these watersheds,
 therefore, were not performed (Table 10-5). In general, optimization  criteria could not be achieved
 either because cation inputs exceeded outputs or charge balance could not be achieved (i.e., chloride
 imbalances). The results discussed in this section have been obtained by weighting the individual
 watershed estimates by the appropriate inclusion probability, the weight being the inverse of the
 inclusion probability (see Section 5.2, Table 5-1).  Weighting by the appropriate inclusion probability is
 critical for any analyses performed on these data. The emphasis in DDRP is on the relative magnitude
 and direction of changes in attributes of the specified target population.  The individual watershed
 estimates are of interest only as they relate to the distribution of the population  attributes.

     A number of the stream reaches apparently were affected by inputs of roadsalts. We made this
 determination the same way we did for the DDRP analyses for the NE and SBRP (see Church et al.,
 1989, Section 10.5.7). To account for these inputs, we "added" salts (seasalt composition) as a point
 input directly to the stream reaches during the model calibrations, exactly as we did for inland
 watersheds in the NE and SBRP (Church et al., 1989).

 10.6.2  Target Population for MAGIC Protections

     An estimated 4,298 streams, representing 11,246 km of stream length in the target population in
 the M-APP Region, were simulated using MAGIC, as compared to the DDRP total target population of
 5,496 streams  (15,239 stream km in the M-APP). As  presented earlier (Section 5.2.7), the DDRP study
 sites represent target populations that are a subset of the National Stream Survey target population
 (25,715 stream reaches, 69,569 km) for NSS Subregions 1D, 2Cn, and  2Bn (Kaufmann et al., 1988).
 The M-APP DDRP modelling target population reflects an exclusion of stream reaches for which
 MAGIC could not satisfy the calibration criteria for the respective sample watersheds. The modelling
target population includes both disturbed and undisturbed watersheds  based on chloride concentra-
 tions, watersheds that had both positive and negative sulfur retention, and watersheds that had initial
 NSS ANC values ranging from -14 to 198 ^eq/L. MAGIC projections  for this target population are for
a 100-year period.  The 100-year time frame permits an evaluation of the cumulative effects of sulfur
deposition on changes in surface water chemistry.
                                             274
-------
Table 10-5.  Watersheds and Attributes for Which Satisfactory MAGIC Calibration Was Not
            Achieved
Watershed
  ID
State
               Attribute
2B047036
2B047066
2B047076
2C028069
2C029016
2C035027
2C047010
 VA
 VA
 VA
 PA
 PA
 PA
 WV
Na inputs exceeded watershed outputs
S isotherms did not match observed adsorption
Na, Ca inputs exceeded watershed outputs
ANC after calibration = 103, observed ANC = -3,3 fieq/L
No successful calibrations out of 10 tries
No successful calibrations out of 10 tries
No successful calibrations °out of 10 tries
                                           275
-------
10.6.3 Regional Calibration                              "

      A comparison of calibrated values for ANC, pH, sulfate, and calcium plus magnesium with NSS
Phase I values is given in Figure 10-7. There is perfect correspondence between calibrated and NSS
measured values in Figure 10-7 if all the points lie on the 1:1 line. There was general agreement for
the low ANC watersheds but greater scatter between calibrated and observed ANC concentrations for
watersheds with ANC concentrations greater than 50 ^eq/L (Figure 10-7).   Calibrated sulfate concen-
trations generally corresponded with NSS measured sulfate concentrations for concentrations
less than 150 ,aeq/L (Figure 10-7). There was greater scatter about the 1:1 line at the higher sulfate
concentrations.  There is excellent correspondence between calibrated and observed calcium plus
magnesium  concentrations over the range of DDRP watersheds (Figure 10-7). Calcium and magne-
sium are two of the variables considered during calibration optimization. There is considerable scatter
in pH over the entire range of pH values in the DDRP watersheds (Figure 10-7).  Like ANC, pH is a
calculated, not a calibrated, variable.

      Although there is scatter for individual watersheds, there is generally better correspondence for
the target population (Figure 10-8). The same general patterns are evident (e.g., lower calibrated than
observed ANC and pH values; greater calibrated than observed sulfate) but the effects  of individual
watershed deviations are reduced at the target population level.  In part, this occurs because the
population weighting factors and cumulative distribution damp some of the sample variation. The
exception is calcium plus magnesium, where almost perfect correspondence  occurs on the 1:1 plot
(Figure 10-7) but larger deviations occur between calibrated and  observed values for the target
population (Figure 10-8).  Small errors in watersheds  with larger weighting factors result in larger
deviations in the target population. In  addition, once established, any deviation or offset (e.g.,
overestimate) is perpetuated through the cumulative distributions and gives an added visual
impression of difference.

      There  is a general bias in the calibrated DDRP  M-APP target population  (Table  10-6).  The lower
quartile and  median ANC and pH values are lower in the calibrated modelling target population than in
the comparable NSS target population. The lower quartile and median ANC concentrations for the
calibrated target population are 20 and 64 /teq/L, respectively, compared with  NSS ANC concentra-
tions of 48 and 87, respectively. The lower quartile pH value for the calibrated target population is 6.2
compared with NSS values of 6.4. The median pH values are similar between the calibrated and NSS
target population. Sulfate concentrations are generally higher in the calibrated sample than in the
NSS target population.

      This bias in calibrated ANC in the modelled target population results in  a bias in the number and
length of acidic streams in year 0 (initial conditions) (Table 10-7). The initial calibrated percentage
(and percent length) of acidic and low ANC streams (i.e., ANC < 50 fieq/L) was 9 percent (5 percent)
and 38 percent (33 percent), respectively.  For the target population modelled in the M-APP DDRP, the
percentage (and percent length) of acidic and low ANC streams as measured  in the NSS Phase I
survey was 3 percent (4 percent) and 25 percent (22 percent), respectively.  The calibrated  percentage
(and percent length) of acidic and low ANC streams,  therefore, were overestimated by factors of about
1.06 (1.01) and  1.13 (1.11), respectively.
                                             276
-------
   300


   250
 0-200
 CD

 """ 150
 O

 < 100

 O
 ,_ 50
 co
 o>
 >  0
    -50
         Mid-App Stream Reaches
             ANC Comparison
      -50    0   50  100  150 200  250  300
          Phase 1 ANC  (u.eq/L-1)
     Mid-App Stream  Reaches
       [SO42-]  Comparsion
   0  50  100  150  200  250 300  350
    Phase 1 [SO/-] (fieq L-1)
         Mid-App  Stream Reaches
         ([Ca2+]+[Mg2+]) Comparsion
 — 400
        0   50 100  150 200 250 300 350 400
      Phase 1 ([Ca2+HMg2*]) (ueq L-I)
7.5


7.0
                                                  16.5
                                                  Q.
                                                    6.0
                                                  CO
                                                  -5.5
                                                    5.0
                                                    4.5
     Mid-App Stream Reaches
          pH Comparsion
 4.5   5.0   5.5   6.0   6.5   7.0   7.5
           Phase 1 pH
Figure 10-7.  MAGIC calibrated values at year 0 and NSS Phase I values for ANC, sulfate,
             calcium plus magnesium, and pH for the DDRP M-APP sample watersheds.
                                           277
-------
     1.0r
     0.8
    0.6
.1

O
a.
o

IX
  o>

  »3 0.4
  ra
  a
  E

  O
          Mid-App  Stream  Reaches
                    ANC
    0.0
     •TOO
                            Phase 1
                         — Simulated Yr
           0     100

             ANC
                        200

                         L-I)
                              300
                                    400
                                                         Mid-App Stream Reaches
                                                      1-0r
                                                   O 0.8
                                                   O
                                                   o.
                                                   o
                                                      0.6
0}

'••3  0.4
ro
3
E

O
                                                    o.o
                                                                        	 Phase 1
                                                                        	Simulated Yr 0
    "0    100   200   300   400   500   600

        ([Ca2+]+[Mg2*])  (jxeq  !_"i)
         Mid-App Stream Reaches
                     pH
    to
  O 0.8
O


S 0.6
 yz 0.4
E

O °-2
    0.0
                                                   JO
                                                   •^

                                                   o
                                                   a.
                                                   o
                                                      1.0 r
                                                     0.8
                                                     0.6
                                                 o.

                                                 0)

                                                 :P  0.4

—

rnase 1
— Simulation
Year 0
                                                 _

                                                 3

                                                 E

                                                 O
                                                     0.2
     4.0  4.5  5.0  5.5 6.0  6.5  7.0  7.5 8.0

                 ,  pH
                                                     0.0
                                                         Mid-App  Stream  Reaches
                                                                 [S042-]
                           Phase 1
                        — Simulated Yr 0
                                                            100      200     300
                                                             [SO*2']  (jieq L-i)
                                 400
Figure 10-8.  MAGIC calibrated values at year 0 and NSS Phase I values for ANC, calcium

              plus magnesium, pH, and sulfate for the DDRP M-APP target population. Note

              that this does not include all NSS-Phase I stream reaches in the M-APP.
                                            278
-------
Table 10-6.    Descriptive Statistics of Calibrated ANC, pH, Sulfate Concentration, and Calcium
              Plus Magnesium Using MAGIC Compared with NSS Observed Values
                     Descriptive Statistics for MAGIC Calibrated Constituents
Variable
ANC
PH
SO4
Ca + Mg
Mean
77
6.50
172
242
St Dev
71
0.62
59
79
Min
-6
5.02
60
68
P25
20
6.19
120
178
Median
64
6.69
161
246
P75
111
7.27
210
325
Max
252

322
373
     Descriptive Statistics for 1986 NSS Data (29 sites only, corresponding to MAGIC population)
Variable
ANC
pH
SO4
Ca + Mg
Mean
88
6.55
152
239
StDev
59
0.60
47
78
Min
-10
4.91
55
64
P25
48
6.41
110
179
Median
87
6.72
154
245
P75
139
6.96
178
322
Max
200
7.43
264
372
                                            279
-------
Table 10.7.  Population Estimates of Combined Number, Length and Percentages of Stream
            Pleaches Having ANC Less Than Selected Reference Values at Downstream Nodes
            During Spring Baseflow for NSS Phase I; All M-APP DDRP; DDRP Systems
            Simulated (ANC Values Estimated from NSS-I); and DDRP Systems Simulated (ANC
            Values Calculated from MAGIC  Calibrations).
                          NSSa
DDRPb
DDRP Modelled
as Measured0
DDRP Modelled
as Calibrated6
 Sample (n)                164

 Target Population
       (N)                25,715
     Length (km)          69,569

 ANC < 0 u.eq/L
       (N)                  326
       (%)e                 1.3.

     Length (km)           2,324
       (%)f                 3.3

 ANC < 50 |ieq/L
       (N)                 2,199
       (%)                  8.6

     Length (km)           7,430
       (%)                  10.7
  36
 5,496
15,239
  181
  3.3

  597
  3.9
 1,504
 27.4

 3,002
 19.7
       29
      4,298
     11,246
      145
      3.4

      472
      4.2
      1,087
      25.3

      2,505
      22.3
       29


      4,298
     11,246


      390
       9.1

      527
       4.7
      1,656
      38.5

      3,756
      33.4
 a National Stream Survey estimates for Subregions 1D, 2Cn, and 2Bn.
 b DDRP sample and target population estimates for the Mid-Appalachian Region.
 c Summary for NSS Phase I ANC values for stream reaches modelled by MAGIC.
 d Summary for ANC values as calibrated by MAGIC for year 0.
 9 Percentage of estimated target population number.
 ' Percentage of estimated target population length.
                                            280
-------
10.6.4 Mid-Appalachian Projections

      Projected changes in ANC, pH, sulfate, and calcium plus magnesium that might occur over a
100-year period under the three deposition scenarios are presented in this section.

10.6.4.1  Projected Values of ANC, pH, Sulfate, and Calcium Plus Magnesium

      The three deposition scenarios are projected to have different effects on surface water
chemistry.  These projected effects can be presented both as projections of absolute values of
constituents and as changes in  concentrations (or values) of constituents.  Although it is useful to
discuss both absolute values and changes, there are some biases of initial calibrated values of ANC
and sulfate concentration (Table 10-6); thus, for these constituents, the values that are likely to be
most meaningful are not the projected absolute values but rather, the projected changes.  Projected
values represent flow-weighted annual average constituent concentrations.

      Values of ANC are projected by MAGIC to decrease monotonically over 50 years under current
deposition levels (Scenario A), whereas they are projected to increase monotonically over 50 years
with a 50 percent reduction in current deposition (Scenario B) (Figure 10-9, Table 10-8).  With a 50/20
percent change in deposition  (Scenario C), ANC is projected to increase for the first 20 years, return to
approximately initial values (i.e., yr 0) after 50 years,  and decrease after 100 years. The median
projected ANC values after 50 years are 36, 80, and  64 ^eq/L, respectively, for deposition Scenarios A,
B, and C (Table 10-8).  The projected ANC values after 100 years for these three deposition scenarios
are 8, 77, and 45 /weq/L, respectively. The ANC is projected to decrease from yr 50 to yr 100 with  all
three  deposition scenarios (Table 10-8).

      Values of pH are projected to show a pattern similar to that of ANC (Figure 10-10).  The pH
values are projected to decrease monotonically over the 100-year period with current deposition rates
(Table 10-8). The  median pH is projected to increase for the first 20 years with a 50 percent decrease
in deposition but then decrease over the next 80 years. Similar changes in pH are observed with the
50/20 percent deposition decrease, but the changes are greater over the latter 80-year period with this
deposition scenario. The median pH after 50 years with deposition scenarios A, B, and C are 6.4, 6.8,
and 6.7, respectively. After 100 years, the median pH values for these three deposition scenarios are
5.6, 6.8, and 6.5, respectively (Table 10-8). The pH changes are greater for all deposition scenarios at
the lower pH values  (Figure 10-10). All streams with initial calibrated pH values less than about 6.5
exhibited greater changes in pH over time than those streams with initial calibrated pH over 6.5. The
cumulative distributions are similar over time regardless of deposition levels for streams with pH above
6.5.  The lower quartile pH values after 50 years for current, 50 percent, and 50/20 percent deposition
are 5.6, 6.5, and 6.2, respectively; after 100 years, these values are 4.7, 6.4, and 5.5, respectively.  At
the upper quartile, the changes after 50 and  100 years for all three deposition scenarios were to pH
values of 6.8, 7.0, 6.9, and 6.6,  7.0, and 6.8, respectively.
                                              281
-------
                                 1.0
                              O 0.8
                              *3
                              O
                              a.
                              2 0.6
                              a.

                              0>

                              I0'4
                              ZJ


                              O
                                0.0
                                     Mid-App  Stream Reaches
                                       Deposition Scenario A
                                 -100
                                             100
                                                   200
                                          ANC (fieq L-I)
                                                        300
                                                              400
                                 1-Or
                                0.8
                              O
                              g-
                                0.6
                              CP

                              :p 0.4
                              JO
                              =>
                              E

                              O
                                     Mid-App  Stream Reaches
                                       Deposition Scenario B
                                0.0'—
                                 -100

	


Year 20
Year 50

                                        0    100    200   300
                                          ANC (|j,eq L-1)
                                                              400
                                 I.Or
                                     Mid-App  Stream Reaches
                                       Deposition Scenario C
                                        0    100    200   300
                                          ANC ([ieq  L-1)
                                                              400
Figure 10-9. Projections of ANC for the M-APP stream target population under current sulfur
             deposition (Scenario A), 50 percent deposition decrease (Scenario B), and 50/20
             percent deposition decrease (Scenario C) over 100 years.
                                              282
-------
Table 10-8.    Descriptive Statistics of Projected ANC, pH, Sulfate, Percent Sulfur Retention,
              and Calcium and Magnesium Using MAGIC for Current and Decreased
              Deposition
CURRENT DEPOSITION (Scenario A)
Model
ANC
YrO
Yr20
Yr50
Yr 100
YrO
Yr20
Yr50
Yr 100
SO4
YrO
Yr20
Yr50
Yr 100
Percent S
Yr20
Yr50
Yr 100
Ca + Mg
YrO
Yr20
Yr50
Yr 100
MEAN

77
66
51
31
6.50
6.36
6.14
5.69

172
208
232
239
Retention
14
4
1

242
266
273
244
STD DEV

71
69
64
60
0.62
0.71
0.81
0.95

59
66
67
70

12
7
7

79
87
94
96
MIN

-6
-9
-15
-21
5.02
4.92
4.69
4.49

60
100
128
128

-6
-13
-15

68
99
114
100
P25

20
12
4
-15
6.19
6.01
5.65
4.69

120
148
191
192

6
0.7
-0.9

178
207
183
166
MED

64
49
36
8
6.69
6.57
6.42
5.65

161
194
218
226

12
4
2

246
254
267
238
P75

111
102
83
49
6.92
6.90
6.80
6.57

210
236
267
276

24
11
9

325
345
345
298
MAX

252
238
216
183
7.27
7.25
7.20
7.13

322
381
389
389

48
21
12

373
414
467
472
                                                                       (Continued)
                                           283
-------
Table 10-8. Continued
50% DECREASE (Scenario B)
Model
ANC
YrO
Yr20
YrSO
YrlOO
EH
YrO
Yr20
YrSO
YrlOO
SO4
YrO
Yr20
YrSO
YrlOO
Percent S
Yr20
YrSO
YrlOO
Ca + Ma
YrO
Yr20
YrSO
YrlOO
MEAN STD DEV

77
97
93
85

6.50
6.72
6.70
6.63

172
146
130
125
Retention
-20
-8
-3

242
240
221
209

71
71
69
67

0.62
0.42
0.43
0.49

59
45
35
34

14
12
11

79
81
80
78
MIN

-6
5
2
0

5.02
5.71
5.60
5.43

60
80
69
67

-54
-39
-34

68
93
98
91
P25

20
45
40
30

6.19
6.54
6.50
6.37

120
118
109
104

-26
-15
-8

178
170
151
142
MED

64
84
80
77

6.69
6.81
6.79
6.77

161
134
123
120

-16
-3
0.1

246
237
219
210
P75

111
132
129
122

6.92
7.01
7.00
6.98

210
168
143
141

-13
0.3
4

325
310
253
235
MAX

252
260
249
234

7.27
7.29
7.26
7.24

322
255
202
195

15
12
11

373
388
391
376
                                                                       (Continued)
                                           284
-------
Table 10-8. Continued
50%/20% DECREASE (Scenario C)
Model
ANC
YrO
Yr20
Yr50
Yr 100
EH
YrO
Yr20
Yr50
Yr 100
504
YrO
Yr20
Yr50
Yr 100
Percent S
Yr20
Yr50
Yr 100
Ca + Mq
YrO
Yr20
Yr50
Yr 100
MEAN

77
95
72
55

6.50
6.71
6.46
6.19

172
149
172
191
Retention
-13
11
1

242
240
239
240
STDDEV

71
70
67
64

0.62
0.43
0.63
0.81

59
45
47
54

13
9
8

79
81
83
87
MIN

-6
4
-5
-13

5.02
5.67
5.08
4.76

60
83
100
104

-45
-12
-21

68
93
109
102
P25

20
43
19
0.8

6.19
6.53
6.19
5.48

120
120
142
154

-19
8
-2

178
171
164
158
MED

64
82
64
45

6.69
6.80
6.69
6.53

161
138
160
180

-8
12
4

246
238
233
238
P75

111
131
101
78

6.92
7.01
6.89
6.78

210
169
200
219

-5
17
8

325
311
297
298
MAX

252
258
234
209

7.27
7.28
7.24
7.19

322
261
283
311

21
32
13

373
389
412
431
                                           285
-------
                                1.0
                             O  0.8
                             o
                             0.
                             o
                                0.6
                             03

                             I0'4
                             r>


                             O
                               0.0
                                     Mid-App Stream  Reaches
                                      Deposition Scenario A
                                 4.S  5.0  5.5  6.0  6.5  7.0  7.5  8.0
                                               PH
                                    Mid-App  Stream Reaches
                                      Deposition Scenario B
                                4.5   5.0   5.5  6.0  6.5  7.0  7.5
                                                             8.0
                                    Mid-App  Stream Reaches
                                      Deposition Scenario C
                                4.5  5.0   5.5   6.0   6.5  7.0  7.5  8.0
Figure 10-10.    Projections of pH for the M-APP stream target population under current sulfur
                 deposition (Scenario A), 50 percent deposition decrease (Scenario B), and
                 50/20 percent deposition decrease (Scenario C) over 100 years.
                                              286
-------
      Sulfate concentrations are projected to increase over time under current deposition, decrease
under the 50 percent deposition scenario, and decrease for the first 20 years and then increase for the
next 80 years with the 50/20 percent deposition scenario (Figure 10-11).. The median percent sulfur
retention in the watersheds is projected to approach steady state after about 100 years under both
deposition Scenarios A and B (Table 10-8). Under Scenario A, however, the median sulfate concen-
trations are projected to continue to increase in the streams as they approach sulfur steady state,
whereas with Scenario B, the watersheds are projected to desorb sulfate as they approach a new
sulfur steady state and the stream sulfate concentrations decrease (Table 10-8). With Scenario C, the
change in the median percent sulfur retention indicates the watersheds initially desorb sulfate, with
median stream sulfate concentration decreasing over the first 20 years; then the median sulfate
concentrations in the streams are projected to increase as the watersheds approach a new sulfur
steady state (Table 10-8).

      The model projects calcium plus magnesium concentrations to increase for the first 50 years
under deposition Scenario A and then decrease for the next 50 years (Figure 10-12).  Under Scenario
B, calcium plus magnesium concentrations are projected to decrease monotonically over the 100-year
period, whereas for Scenario C, calcium plus magnesium  concentrations are projected to  decrease
slightly for the first 50 years and then increase for the next 50 years (Table 10-8). The changes
projected for Scenario C are relatively small.

      The results of the model  projections can also be represented as (1) the increase or decrease in
the number of stream reaches  with ANC  < 0 ^weq/L or < 50 fieq/L at the lower reach  nodes (Table
10-9), (2) the percentage of the DDRP target population with ANC < 0 ^eq/L or < 50 ^eq/L at the
lower reach nodes (Table 10-10), or  (3) the reach  length in the DDRP target population with ANC < 0
yweq/L or <  50 /^eq/L based on projections at the lower reach nodes (Table 10-11).  Statistical tests of
significance (see Section 6) can be performed to determine if the changes at some target year (e.g.,
year 50) are significantly different from  zero. Using the procedures presented in Section 6, we have
performed such tests for the change in the number of systems with ANC < 0 ^weq/L and ANC < 50
neaJL at the lower reach nodes.  In Section 10.8.5.2, we discuss the important implications of making
projections of reach chemistry  based only on the  chemistry at the lower nodes of each reach.

      For deposition Scenario B, MAGIC projects  a decrease in the number of acidic systems and the
number of systems with  ANC < 50 ^weq/L (in some cases  to zero acidic systems), but with p > .16 for
all cases (Table 10-9), where the null hypothesis is that the projected change is not significantly
different from zero.  For Scenario C,  MAGIC projects first a decrease in the number of acidic and low
ANC systems (p = 16 at 20 years), then  an increase  in the number of such systems (.07 ^ p ^ .09 at
50 and 100 years).  For the current deposition scenario (Scenario A), all cases of projected changes  in
the number of acidic systems at 20,  50, and 100 years have p < .05 (Table 10-9).  For the systems
with ANC < 50 /*eq/L, p = .01  for the projected changes at 50 and 100 years.

      The projections at current deposition of acidic systems at 50 and 100 years correspond to 11
percent and 24 percent, respectively, in the number of stream reaches, based on lower node
chemistry (Table 10-10) and to 1,298 km (at 50 years) and 1,886 km of stream reach length (at 100
years) (Table 10-11).  The projections for stream reaches  with lower node ANC < 50^eq/L at 50 and
                                             287
-------
                              1.0r
                                   Mid-App Stream Reaches
                                    Deposition Scenario  A
                                      100
                                             200     300
                                      [SO,2-] (jieq L-1)
                                                           400
                              0.0
                                   Mid-App  Stream Reaches
                                    Deposition Scenario  B
                                                          400
                              1.0r
                                  Mid-App  Stream Reaches
                                    Deposition  Scenario C
                                                          400
Ftgure 10-11.     Projections of sulfate concentration for the M-APP stream target population
                 under current sulfur deposition (Scenario A), 50 percent deposition decrease
                 (Scenario B), and 50/20 percent deposition decrease (Scenario C) over 100
                 years.
                                             288
-------
                                 Mid-App Stream Reaches
                                   Deposition  Scenario  A
                              0   100   200   300   400  500  600
                                  ([Ca2+]+[Mg2D (neq  1_-1)
                                  Mid-App Stream Reaches
                                   Deposition Scenario B
                                   100  200  300  400   500
                                   ([Ca2+]+[Mg2*]) (jieq  L"0
                                                           600
                             1.0
                           O 0.8
                           O

                           &0.6
                           Q.
                           O
                           V D.4


                           i
                           O 0-2



                             0.0
                                  Mid-App Stream  Reaches
                                   Deposition  Scenario  C
                                   100   200   300  400  500
                                   ([Ca2+]+tMg2*]) (jieq L-1)
                                                           600
Figure 10-12.    Projections of calcium plus magnesium for the M-APP stream target population
                 under current sulfur deposition (Scenario A), 50 percent deposition decrease
                 (Scenario B), and 50/20 percent deposition decrease (Scenario C) over 100
                 years.
                                              289
-------
Table 10-9.   Change Projected by the MAGIC Model In the Number of M-APP DDRP
               Lower Reach Nodes Having (A) ANC < 0 or (B) ANC < 50 fiec\/L
ANC < 0 tiey/L
                                        Depos'rtion Scenario8
Year
20

50

100

Current
Scenario A
50
p=.02
347
p=.04
906
p=.02
50% Decrease
Scenario B
-145
p=.16
-145
p=.16
-130
p=.19
50%/20% Decrease
Scenario C
-145
p=.16
30
p=.07
192
p=.09
B
ANC < 50 uea/L
Year
  Current
Scenario A
Deposition Scenario8

   50% Decrease
     Scenario B
50%/20% Decrease
    Scenario C
20
50
100
  229
  p=.16

  1240
  p=.01

  1300
  p=.01
       -145
       p=.16

       -145
       p=.16

       83
       NS
      -145
      p=.16

      229
      p=.07

      546
      p=.08
  The number given is the estimated difference in number of population lower reach nodes between the listed year
  and year 0 for tha same MAGIC model runs. Each ANC value is calculated as the NSS observed value plus the change
  in model ANC since year zero.

  in each case, the difference is compared against a statistical estimate of the standard deviation of the difference,
  and a p-value is given underneath the difference. The test is a one-sided hypothesis test where the null hypothesis
  Is that the difference is not significantly different from zero. The estimate of the variability is designed to
  capture the variability of the inputs into the model, the sensitivity of the model to the input variability, the
  variability of the model between scenarios, and the variability involved in extrapolating to the population estimates.
                                                 290
-------
Table 10-10.  Percentage of M-APP Target Population Stream Reaches Projected by the MAGIC
              Model to Have (A) ANC < 0 or (B) ANC < 50 /*eq/L at the Lower Reach Nodes
A
ANC < 0 weq/L
Year Current
Scenario A
Oa 3
20b 5
50b 11
100b 24

Deposition Scenario
50% Decrease
Scenario B
3
0
0
0

50%/20% Decrease
Scenario C
3
0
4
8

B
ANC < 50 weq/L
Year Current
Scenario A
Oa 25
20b 31
50b 54
100b 56

Deposition Scenario
50% Decrease
Scenario B
25
22
22
27

50%/20% Decrease
Scenario C
25
22
31
38
a  Estimates from National Stream Survey for the target population stream reaches corresponding to the watersheds modelled
   (see Table 10-7).

b  Estimates at 20, 50 and 100 years are computed by adding change in ANC projected by MAGIC to ANC observed by
   Kaufmann et al. (1988) for the National Stream Survey.
                                                291
-------
Table 10-11. Length (km) of M-APP Target Population Stream Reaches Projected by the MAGIC
              Model to Have (A) ANC < 0 or (B) ANC < 50 fieq/L at the Lower Reach Nodes
A
ANC < 0 /tea/L
Year
Oa
20b
50b
100b

Current
Scenario A
472
773
1298
1886

Deposition Scenario
50% Decrease 50%/20% Decrease
Scenario B Scenario C
472 472
0 0
0 645
80 1016 -

B
ANC < 50 «ea/L
Year
Oa
20b
50b
100b

Current
Scenario A
2505
2777
5831
6178

Deposition Scenario
50% Decrease 50%/20% Decrease
Scenario B Scenario C
2505 - 2505
2221 2221
2221 2777
2493 4694
a  Estimates from the National Stream Survey for the target population stream lengths corresponding to the watersheds
   modelled (see Table 10-7).

b  Estimates at 20, 50. and 100 years are computed by adding change in ANC projected by MAGIC to ANC observed by
   Kaufmann et al. (1988) for the National Stream Survey.
                                                292
-------
100 years are over 50 percent of the number of stream reaches (54 percent and 56 percent, respec-
tively). These projections are equivalent to 5,831 km and 6j178 km of stream reaches, respectively, at
50 and 100 years (Table 10-11). Over half the DDRP target population of stream reaches (at the lower
nodes, see Section 10.8.5.2) are projected to have ANC < 50 ^eq/L and thus, presumably, to be
susceptible to episodic acidification to ANC < O^eq/L (Eshleman,  1988).

10.6.4.2  Projected Rates of Change

      The projected changes in ANC and pH values over the next 100 years can be displayed as box
and whisker plots (Figures 10-13 through 10-18), which illustrate how both the target population
constituent medians and interquartile ranges vary through time.

      The change in median ANC concentration projected for current deposition (Scenario A) was -28
,weq/L (64 ^eq/L to 36 ^ueq/L) after 50 years and -56 ^eq/L after 100 years (Table 10-8). The projected
rate of ANC change was relatively constant over time  (Figure 10-13). For the 50 percent decreased
deposition scenario (Scenario B), the projected change in median ANC was +16 ^weq/L (64 to 80
^weq/L) at year 50, and +13lweq/L (64 to 77/^eq/L) at year 100 (Figure 10-14). With the 50/20 percent
deposition scenario (Scenario C), the change in median ANC was 0 ^ueq/L or equivalent to the initial
calibrated median ANC at 50 years and -19^eq/L (64 to 45,aeq/L) at  100 years (Figure 10-15).  The
projected change in median ANC varied among ANC  groups (Table 10-12). For deposition Scenario
A, the projected changes in median ANC for streams with an initial ANC < 25 /*eq/L were about -7
fieq/L and -16 fieq/L after 50 and 100 years, respectively. With an initial ANC concentration of 25 to
100 fteq/L, the projected changes were -33 and -61 /ieq/L, respectively.  The projected change in
ANC was intermediate for streams with an initial calibrated ANC between  100 and 40O/*eq/L (Table
10-12).  These changes are shown  graphically in Figure 10-19.  For Scenario B, the greatest projected
ANC increase occurred in streams with an initial calibrated ANC between 100 and 400^eq/L (Figure
10-19).  For Scenario C, almost no change was projected for streams with an initial calibrated ANC
< 25 jMeq/L, whereas for streams with an initial calibrated ANC between 25 and 100 ,weq/L the
projected change was -7 fieq/L (Table 10-12). After 100 years, the greatest ANC change (-27yweq/L)
was  projected for streams with initial calibrated ANC between 100 and 400
      For deposition Scenario A, the projected change in median pH was -0.25 units at 50 years and
-1.0 units during the next 100-years (Table 10-8, Figure 10-16).  Median pH for Scenario B remains
nearly constant over the 100-year period (Table 10-8, Figure 10-17), increasing by 0.10 units between
years 0 and 50, but then declining by .02 units between years 50 and 100.  For Scenario C, median
pH is projected to change little during the initial 50 years of simulation, but to decline by about 0.20
units between years 50 and 100 (Table 10-8, Figure  10-18).

      The change in median ANC is a population attribute, not an individual watershed attribute. The
change in ANC for each individual watershed was relatively independent of the initial, calibrated ANC
values.  When watershed groups (e.g., < 25, 25-100, 1 00-200 fieq/L ANC) are considered, however,
different watershed  characteristics and differential population weights result in different changes in
median ANC, not as a function of the initial ANC, but rather as a function of the initial ANC group to
which the watershed was assigned.
                                             293
-------
          ANC   vs.  TIME
         Constant  Deposition
             Model * MAGIC
          -
                            o
                            CM
                            a:
                           o
                           LO
                                      Mid-
                                   Appalachian
                                     Region
O
o
                                         >-
Figure 10-13.   Box and whisker plots of projected ANC for M-APP streams under current
          deposition (Scenario A).
                           294
-------
          ANC   vs.  TIME
            SOX  Decrease
             Model -  MAGIC
           D"
           Q)
           D
          O
250
200
 150
 100
 50
   0
-50
       Upper
       Q3
       Mean
       Median
       Qi
       Lower
o
>-
                            O
                            CM
                      O
                      m
                      a:
                                 Mid-
                             Appalachian
                                Region
o
o
Figure 10-14.   Box and whisker plots of projected ANC for M-APP streams under a 50 percent
          deposition decrease (Scenario B).
                           295
-------
          ANC   vs.  TIME
         50X/20X Decrease

             Model - MAGIC
          cr
          •
                            O
                            CN
                     O
                     m
                                Mid-'
                             Appalachian
                               Region
O
O
Figure 10-15.   Box and whisker plots of projected ANC for M-APP streams under a 50/20
          percent deposition decrease (Scenario C).
                           296
-------
            pH  vs. TIME
         Constant Deposition

             Model - MAGIC
                 8r
                 7
             a_
                 4-
       Upper

       Q3

       Mean

       Median

       Qi

       Lower
o
CM
O
in
                  Mid-
               Appalachian
                 Region

o
o
Figure 10-16.  Box and whisker plots of projected pH for M-APP streams under current

          deposition (Scenario A).
                            297
-------
            pH   vs.  TIME
             SOX  Decrease


              Model -  MAGIC
Upper



Mean

Median

Qi

Lower
            Mid-
        Appalachian
          Region
                 8
                 7
                              o
                              CM
o
LO
O
o


a:
Figure 10-17.  Box and whisker plots of projected pH for M-APP streams under a 50 percent

          deposition decrease (Scenario B).
                            298
-------
            pH  vs. TIME
         50X/20X  Decrease

             Model - MAGIC
                         Mid-
                     Appalachian
                       Region
                 8
                 7
             i.
                 4
       Upper

       Q3

       Mean

       Median

       Qi

       Lower
>-
o
CM

Of
             o
             lO
O
o
Figure 10-18.  Box and whisker plots of projected phi for M-APP streams under a 50/20
          percent deposition decrease (Scenario C).
                            299
-------
Table 10-12. Changes in Median ANC, pl-i, Sulfate Concentration, and Calcium plus Magnesium
           Concentration over 50 and 100-Year Periods as a Function of the Initial NSS ANC
           Groups and Deposition Scenarios
                                       NSS ANC Group
Mode! Output
Variable
< 25
25 - 100
100 - 400
Scenario

ANC (Meq/L)
50 yr
100 yr
EhL
50 yr
100 yr
SO4 (Meq/L)
50 yr
100 yr
Aa

-7
-16

-0.4
-0.7

28
31
Bb

15
10

0.7
0.5

-39
-41
cc

0.5
-7

0.1
-0.4

-8
2
A

-33
-61

-0.3
-0.9

35
43
Scenario
B

9
6

0.1
0.0

-6
-63
C

-7
-26

-0.1
-0.2

-23
-2
Scenario
A

-23
^6

-0.1
-0.3

106
116
B

24
17

0.1
0.1

-27
-34
C

-5
-27

-0.0
-0.1

17
56
(Ca + Ma) (uea/L)
50 yr
100 yr
-4
-30
* Deposition Scenario A =
Deposition Scenario B =
c Deposition Scenario C =
-32
-37
-22
-24
14
5
-22
-32
-10
-6
28
-37
-82
-96
-38
-37
current deposition.
50 percent decrease.
50/20 percent decrease.
                                         300
-------
Change

in Median ANC

Year 0 to Year 50
Model - MAGIC

Mid-
Appalachian
Region rv

w
Constant 50X 50//20X
Deposition Decrease Decrease
\ 30 r
cr
CD 9O
^ Z.U
~ 10-
CD
f~~D f~\
C U &g$&::::$.:&:
5 -10 -
O >>:::^:iS
, -20-
O svvSX;
< ' ' ' '•'
30r 30
9 0 ::;::;;;:-':;;; ? 0
z.vj n J-.-.--.-.-.-.-J ^.vj
-If) i^>%; :W:W: 1Q


H :-:-::-:-::-:-;: •:••:•:•-.-••.-:-- :•:••:•:-•:•:•-: pi
U U
>^ -10 - -10

li -20 - -20
-10 - -"50

r-
_


$-:;::::$£M
-

-


moo LOOO LOOO
CMOO CNOO ^|Op
x/ i
OJ '
•NT v , ^sT " i ^J
§ S g Sg
ANC Group ANC Group ANC Group
(ueq/L) (ueq/L) (ueq/L)
Figure 10-19.  Projected change in median ANC concentration over 50 years for all three
             deposition scenarios as a function of the NSS Phase i ANC used to classify
             DDRP streams in the target population.
                                          301
-------
      Projected changes in pH are closely related to year 0 pH and ANC of the streams.  The greatest
 changes in pH are projected to occur in streams with low ANC (Figure 10-20), although as shown by
 Figure 10-21, the projected decreases in streams with an initial pH  between 6.0 and about 6.75 are
 greater than  for streams with initial pH < 6.0.  This pattern can be attributed to two factors: (1) as pH
 decreases below 6.0, further decreases in pH are partially buffered  by dissolution of aluminum, and
 (2) pH is a logarithmic, rather than a linear scale. Thus, although increases in H+ are greatest in
 those surface waters with the lowest pH, corresponding changes in pH are smaller.  Under current
 deposition, pH is projected to decrease by 0.1 to 1.9 units by year  100. For the 50 percent decreased
 deposition scenario, pH is projected to increase in most systems that currently have pH > 6.5;
 streams that  presently have a high pH are projected to change very little over the next 100 years
 (Figure 10-21). For the 50/20 percent scenario, pH is projected to change very little between years 0
 and 50, but to decline in most systems between years 50 and 100; the largest declines are projected
 to occur in those streams with year 0 pH between 6.0 and 6.5 (Figure 10-21).

      Projected changes in sulfate reflect the continuation of an ongoing trend of soils and runoff to
 come to steady state with atmospheric sulfur deposition. Currently some soils in DDRP watersheds in
 the M-APP Region are retaining a significant portion of incident sulfur  by adsorption on soils; as soils
 continue to adsorb sulfate and approach a dynamic equilibrium with current levels of deposition,
 sulfate concentrations in drainage waters will increase to steady-state.concentrations.  Conversely, if
 deposition is  reduced, soils will desorb sulfate and come to steady  state with the lower level of inputs.
 For the M-APP scenario  of continued deposition at current loadings (Scenario A), the projected
 change in sulfate during the next 50 years is +57/*eq/L (from 161 to 218 /*eq/L) and +65 ^eq/L
 during the 100-year simulation period (Table 10-8).  For the 50 percent deposition scenario (Scenario
 B), median sulfate concentration is projected to decrease by SS^eq/L during the next 50 years and by
 41 /
-------


Change in Median pH
Year 0 to Y
Model- MAC
Constant
Deposition
0.8r 0.8
0^™* ^\ /*^
.6 - 0.6
c? 0.4 - 0.4
g 0.2- 0.2
-^ &£#ffi^W3 °
^-0.2 Ililil -0.2
OA ££i££ n A
. "r 	 U . "f
LO 0 0
Csl O O
v s I
ANC Group
(ueq/L)
ear 50
3IC
SOX
Decrease
r 0.8
-TxX'*-v

$m
_
0.6
0.4
0.2
h 	 1 	 Tt 0
-0.2
n /
V^f • 1^

Mid-
Appalachian
Regionrx
50X/20X
Decrease
-
in o o in o o
CM O O (N O O
"^ 0 ^0
OJ (-) CN] (-)
00 0 °-
ANC Group ANC Group
(ueq/L) (ueq/L)
Figure 10-20.  Change in median pH concentration over 50 years for all three deposition
             scenarios as a function of the NSS Phase I ANC used to classify DDRP streams
             in the target populations.
                                         303
-------
            Mid-App Stream Reaches
              Deposition  Scenario A
                   Year =  50
       1.0


       0.5


       o.o


     ••0.5


      -to


      -15
      -2.0
                                8>
        4.0  4.5   5.0  5.5   6.0  6.5  7.0   7.5
               Year 0 Model  pH
                                                              Mid-App  Stream  Reaches
                                                               Deposition Scenario  A
                                                                    Year =  100
                                                              tOr
                                                        0.5
                                                        0.0
                                                      CL
                                                        -0.5
                                                        -1.0
                                                             -1.5
                                                             -2.0
                                                         4.0   4.5  5.0   5.5  6.0   6.5  7.0   7.5
                                                                 Year 0  Model  pH
   i.
 to


 0.5


 0.0


-0.5


 •to


 •ts
     -2.0
             Deposition  Scenario B
                   Year  =  50
                 O  00
       4.0  4.5  5,0  5.5  6.0   6.5  7.0   7.5
               Year 0  Model  pH
  to


  0.5


  0.0


L-0.5
I

 -1.0


 -1.5
                                                               Deposition Scenario  B
                                                                    Year -  100
                                                                                   *  °D°
                                                                                    o  8 e>
                                                         4.0   4.5  5.0   5.5  8.0   6.5  7.0   7.5
                                                                 Year 0  Model  pH
      to


      0.5


      0.0


     L-0.5J
     I

      •to


      •ts
     -2.0
             Deposition Scenario C
                   Year  - 50
       4.0   4.5  5.0   5.5  6.0   6.5  7.0   7.5
               Year 0  Model  pH
                                                              tOr
                                                        0.5
                                                        0.0
                                                       •"0.5
                                                        -1.0
                                                             -1.5
                                                             -2.0
                                                               Deposition Scenario  C
                                                                    Year -  100
                                                         4.0   4.5  5.0   5.5  6.0   6.5  7.0  7.5
                                                                 Year 0  Model pH
Figure 10-21.  Change in pH for sampled streams in the DDRP M-APP Region as a function of
                the initial calibrated pH at year 0.
                                                  304
-------
10.6.4.3  Comparison With ILWAS Projections for Two Watersheds

     Staff of Tetra-Tech, Inc., used the Integrated Lake-Watershed Acidification (ILWAS) Model to
perform projections of potential future effects of sulfur deposition for two of the DDRP M-APP
watersheds (R. Munson, pers. comm.).  They performed simulations for the DDRP watersheds of
Johnson Run, West Virginia {ID 2C046033; observed ANC = 6 jieq/L) and North Branch Rock Run,
Pennsylvania (ID 1D029023; observed ANC = 32fieq/L). They used data for surface water chemistry,
watershed characteristics, soil attributes and chemistry, annual runoff, and current atmospheric
deposition chemistry and volume provided by DDRP staff.  They performed simulations using three
deposition scenarios provided by the Electric Power Research Institute (EPRI) (R. Munson, pers.
comm.).  Of the three scenarios, only one corresponded directly to a scenario used by DDRP. This
was the deposition scenario of continuing deposition at current levels (Scenario A). Tetra-Tech, Inc.,
performed projections for periods of up to 50 years.

     Here, we briefly examine comparisons between the ILWAS and MAGIC projections for the two
watersheds using flow-weighted average annual values for ANC, the primary variable of interest in the
DDRP. For ANC, it is best to compare projected changes (rather than absolute values) because
complications may arise from calibration offsets for current values. As noted earlier, in our reporting of
final projected ANC values  we combine projected changes with current values as observed by the
National Stream Survey (Kaufmann et al., 1988). This comparison of model  projections parallels that
approach.

     For Johnson Run, West Virginia (ID 2C046033), both MAGIC and ILWAS projected very slight
acidification (loss of ANC) in the future under current deposition. Projections with ILWAS indicated
decreases in ANC of 1 and 2 neq/L at 20 and 50 years, respectively.  Projections with  MAGIC
indicated ANC decreases of 2 and 5 [j.eq/L at 20 and 50 years, respectively.

     For North Branch Rock Run, Pennsylvania (ID 1D029023), projections from MAGIC and ILWAS
both indicated acidification slightly greater than that for Johnson Run under  current deposition
loadings. At 20 years, MAGIC projected a decrease of 8 neq/L, and ILWAS  a decrease of 6 neq/L.  At
50 years, MAGIC and ILWAS projected decreases  of 15 and 11  neq/L, respectively.

     Two primary observations can be made from these comparisons. First, both models project
future acidification for the watersheds (very slight for one, more pronounced for the other) under
current loads of atmospheric deposition. Second, the changes in ANC projected by the models for
the two watersheds are in close correspondence, with differences in projections ranging from  1 to 4
|ieq/L

10.7 DISCUSSION

10.7.1   Projections of Future Changes in Acid-Base Surface Water  Chemistry

      One of the objectives of the DDRP Level III analyses is to estimate the general direction and
relative magnitude of possible future changes in surface water acid-base chemistry as a function of
                                             305
-------
alternative sulfur deposition scenarios.  Making relative comparisons among different management or
control alternatives represents one of the most effective uses of models. These comparisons are
based on relative changes in the population attributes of M-APP stream reaches as a function of the
three deposition scenarios. Annual typical year meteorological and deposition data were repeated for
100 years to provide a consistent base for assessing the relative magnitude and direction of the
change in stream chemistry of M-APP watersheds and to permit relative comparisons among deposi-
tion scenarios.  These projections are not and were not intended to be forecasts of future conditions.
Forecasts relate to the actual probability that an event or condition will occur. The relative magnitude
and direction of change can be placed in perspective, however, by comparing the projections with
other information on  M-APP watersheds, such as monitoring and survey data.

      Dynamic watershed models represent an integration of our knowledge and working hypotheses
of how watershed processes control changes in surface water acid-base chemistry (Jenne et al., 1989;
Thornton et al., 1990).  The observed differences in process inclusion/representation  among models
reflect the different philosophies among model developers regarding both the quantitative importance
of individual processes and the different descriptions of the nature of those processes.  A discussion
and evaluation of these alternative hypotheses and underlying assumptions, as well as other uncertain-
ties, provides information on our level of understanding and assists in identifying important remaining
questions.  Tabfe 10-13 lists the implications of a variety of factors on estimates of future  change in
ANC and pH. Section 10.8 discusses some of these factors and their implications on conclusions that
can be drawn from comparisons among model projections of different deposition  scenarios.

10.7.2 Systems of Interest

      The Mid-Atlantic highlands consists of the Mid-Atlantic parts  of the Blue Ridge,  Ridge and Valley,
and Appalachian Plateaus, an area bounded  by the Virginia/North  Carolina border on the south and
the Adirondack Mountains of northern New York on the north (Kaufmann et al., 1988). This area
includes the DDRP Mid-Appalachian Region.  In the Mid-Atlantic highlands, Kaufmann et al. (1988)
found acidic streams (excluding streams acidic due to acid mine drainage, which were removed from
this analysis) only in forested watersheds with areas less than 30 km2. These forested watersheds
typically occur at relatively high elevations within individual physiographic regions, and are located on
ridges in the Ridge and Valley Province and in upland areas of the Appalachian Plateau.  Kaufmann et
al. (1988) concluded  that the most probable source of acidity in the acidic streams was acidic
deposition.  In the acidic systems, sulfate.was the dominant anion  and stream sulfate concentrations
could be accounted for based on  evapoconcentration of sulfate deposition, making substantial
watershed contributions unlikely. Nitrate and organic anion concentrations were very low and
contributed little to chronic surface water acidity (Kaufmann et al., 1988).

      The major factor controlling  surface water ANC concentrations  in the Mid-Atlantic highlands,
based on current chemistry, is apparently neutralization of acidic inputs  through mineral weathering or
cation exchange with the concomitant release of base cations (predominantly calcium) (Kaufmann et
al., 1988).  The acidic streams had mean sulfate concentrations of about 150 ^eq/L and mean base
cation concentrations of about 135^eq/L.
                                             306
-------
Table 10-13.  Effects of Critical Assumptions and Uncertainties on Projected Rates of Change
       Assumptions Resulting in Under-
       Estimates of ANC and pH Changes
Assumptions Resulting in Over-
Estimates of ANC and pH Changes
1.      Mineral weathering overestimated             1.

2.      Nitrate assimilation overestimated             2.


3.      Total sulfur deposition underestimated         3.

4.      Calibrated ANC greater than observed         4.

5.      Watershed land use changes                 5.

6.      Desorption is not the reverse of              6.
       adsorption-sulfate is irreversibly
       bound

7.      Biotic uptake/assimilation reducing            7.
       available base cation pool

8.      Effects of distribution extremes over-
       smoothed through aggregation
Mineral weathering underestimated

Organic acids buffer surface water
chemistry

Total sulfur deposition overestimated

Calibrated ANC less than observed

Watershed land use changes

Desorption is not the reverse of
adsorption-hysteresis-related delays in
change

Weathering and sulfate adsorption
increased by decreased soil pH
                                             307
-------
      The DDRP M-APP target population simulated by MAGIC "represents streams with measured  -
ANC < 200 /teq/L, and watersheds having areas < 30 km2 and showing no evidence of acid mine
drainage or other significant internal sources of sulfur. The DDRP stream systems are predominantly
higher elevation, forested watersheds.  The initial calibrated median ANC for the stream target
population was 64/^eq/L, with median sulfate and base cation concentrations of 161 and 246 ^eq/L,
respectively.  The lower quartile and median pH values were 6.2 and 6.7, respectively. This target
population of watersheds represents the streams that are likely to respond to acidic deposition inputs
over the next several decades.
10.7.3 Relative.' Magnitude and Direction of Change

      Under current levels of deposition, the direction of change projected by model simulations was a
decrease in ANC. Simulations result in a statistically significant increase in the percent of acidic
systems at 20, 50, and 100 years. With decreased deposition, there was a general projected trend of
increasing ANC, and a decrease in the percentage of currently acidic systems.  For the 50/20 percent
deposition scenario, projected changes in sulfate paralleled changes in sulfur deposition (i.e., a
decrease, then a substantial increase in sulfate concentration), and resulted in changes in pH and
ANC that were inversely related to the changes in sulfate. The general direction of the change in
surface water acid-base chemistry projected by MAGIC for DDRP M-APP watersheds, in relation to
changes in sulfur deposition inputs, is consistent with current theories of surface water acidification,
and with laboratory and field observations (Church and Turner, 1986; Baker et al., 1991; Galloway et
al., 1983; NAS, 1984; Schindler,  1986, 1988; Turner et al., 1990; Wright et al.,  1990).

      The rates and magnitudes of chemical changes projected by MAGIC for surface waters in the
M-APP Region are, in general, consistent with those observed at monitoring sites within the region.
For the current deposition scenario, median projected changes for ANC and sulfate between years 0
and 50 are -0.56 /*eq/L/yr and +1.1 /teq/L/yr, respectively. In comparison, for two stream systems
developed over sedimentary bedrock in the Blue Ridge Province of Virginia, that have soil and other
watershed characteristics comparable to those of the modeled DDRP target systems, sulfate concen-
trations have increased about 2 /*eq/L7yr and ANC has decreased by 0.5 to 0.7 ^ueq/Uyr over the past
10 years (Ryan  et al., 1989).  Sulfate increases of similar magnitude have been  reported for Coweeta
streams; sulfate concentrations have increased and ANC has decreased to the point that  sulfate is
now the dominant anion in some of these streams (Waide and Swank, 1987).  Analysis of data for
USGS Hydrologic Bench Mark stations in the Mid-Atlantic region indicates that sulfate increases were
statistically significant in low ANC streams in the region (Smith and  Alexander, 1983). There was no
statistically significant trend in sulfate concentration, however, for higher ANC USGS stream systems in
the southern Mid-Atlantic highlands.  The rates and magnitude of change projected by MAGIC are
within the range of measured rates observed under current deposition rates.  Because the DDRP
target population is limited to sites with ANC < 200 /
-------
the surveyed streams were acidic (ANC < 0 ^eq/L).  Watersheds in these regions of Virginia are, on
average, currently retaining sulfur (about 68 percent  net annual retention), with an estimated steady-
state suifate concentration of 220 fieq/L based on current deposition levels for the region. The median
net annual sulfur, retention for Mid-Appalachian watersheds, based on the DDRP analyses, is currently
40 percent but is projected to be near steady state in 50 years (Sections 7 and 9 of this report).

     Webb  et al. (1989) used empirical steady-state models to investigate future changes in the
Virginia brook trout streams.  Assuming F factors of 0.4 and 0.8, they projected median ANC values for
sulfur steady state of -38 and 22 ^weq/L, respectively. The percentage of acidic systems for these two
F factors (0.4 and 0.8) were 88 percent and 32 percent acidic  streams, respectively, when stream
suifate concentrations reach  steady state (Webb et al., 1989).  Projected median ANC values after 50
and 100 years, using MAGIC for current deposition, were 36 and 8 peq/L, respectively.  The percen-
tages of acidic systems after 50 and 100 years were 11 percent and 24 percent (Table 10-9), or an
increase of 8 percent and 21 percent,  respectively.  The  projected percentages of low ANC  streams
(ANC < 50 ,aeq/l_) after 50 and 100 years were 54 percent and 56 percent, or an increase of 29 per-
cent and 31  percent, respectively.  The projected, relative changes are consistent among these two
approaches.

10.7.4  Relationships Among Model  Projections and Watershed Processes

10.7.4.1  Sulfur Retention Processes

10.7.4.1.1 Adsorption-

     The MAGIC subroutine for suifate retention is based on two assumptions; the first is that suifate
adsorption is the only terrestrial process influencing  watershed sulfur retention/release.  The second
assumption, which affects projections  of .watershed response when deposition is reduced, is that
adsorption is completely reversible. The review of sulfur retention processes in soils included  with
DDRP  analyses for the NE and SBRP  (Church et al., 1989, Section 3)  noted that a variety of processes
can retain or release suifate  in terrestrial systems. That review concluded, however, that adsorption
was the principal net retention mechanism in  most upland terrestrial systems.  The adsorption capacity
of soils varies widely on both local and regional scales, and is related to soil pH, Fe, and Al content,
texture, and organic matter.  Similarly, the reported reversibility of adsorption varies widely;  suifate is
readily and completely desorbed from some soils (Weaver et al., 1985; Sanders and Tinker, 1975),
whereas less than 10 percent of suifate can be displaced from some highly weathered tropical soils
(Bornemisza and Llanos, 1967).  Reversibility of sorption has not been well characterized for soils in
the M-APP Region and contributes to  uncertainty in  model projections; to the extent that suifate is
irreversibly sorbed to soils, recovery time in response to reductions in sulfur deposition will be
overestimated.

10.7.4.1.2 Retention of organic sulfur —

      Along with adsorption, sulfur can be retained in soils by incorporation into soil organic matter.  If
there is significant net retention or release of sulfur from organics, model estimates will over- or under-
                                              309
-------
estimate actual watershed response time.. Any assessment of apparent reversibility of sorption and/or
hysteresis in desorption can similarly be confounded by changes in the rates of sulfur immobilization
and/or mineralization in soil organic matter. Although soil sulfur occurs predominantly in organic
forms in most soils, and although there is substantial cycling of sulfur between organic and inorganic
fractions, several lines of evidence suggest that there is little net retention of sulfur in the organic
fraction of most forest soils in the eastern United States.  Two  studies of sulfur dynamics by Fuller and
coworkers (at the Hubbard Brook Experimental Forest, New Hampshire, and the Huntington Forest,
New York), the first based on a model of kinetic data for conversion among soil sulfur pools and the
second assessing  distribution of sulfur isotopes in soil sulfur pools, concluded that overall rates of
immobilization and mineralization of organic sulfur in soils were about equal and that net retention in
organic pools was near zero (Fuller et al., 1986a,b). Sulfur input-output budgets developed for the
northeastern United States as part of the DDRP (Rochelle and  Church,  1987; Church et al., 1989)
show that budgets are, on average, very close to steady state, indicating that there is not significant
net retention in organics or other watershed sulfur pools on a broad regional basis. Finally, for soils in
each of the three DDRP regions, median carbon-to-sulfur ratios in mineral horizon soils were less  than
160:1, and were less than 100:1 in B  and C horizons. These ratios are lower than the 200:1 ratio
below which soils would be expected to have net mineralization of sulfur, and far below the 400:1 ratio
above which soil organic matter is expected to be accumulating sulfur (Barrow, 1961; T. Strickland,
pers. comm., 1989).  Based on these data, it seems unlikely that soils in DDRP watersheds are
accumulating sulfur in organic matter; they might even be a net sulfur source.

10.7.4.2 Base Cation Supply

10.7.4.2.1  Cation exchange --

     Baker et al.,  (1990) determined  that mineral weathering and cation exchange were major factors
influencing surface water acid-base chemistry. The MAGIC model is based on a conceptual model
having sulfate  adsorption as the principal control on acid anion mobility and neutralization of acidic
cations  by cation exchange and/or mineral weathering (Galloway et al.,  1983; Cosby et al., 1985a,b,c).
In M-APP simulations, the projected increase in median stream sulfate over 50 years was +57 ^eq/L,
with an  associated increase in base cation (Ca + Mg) concentrations of +21 fieq/L.  In the MAGIC
model simulations, mineral weathering rates are fixed during calibration, such that long-term changes
in base  cations and ANC are controlled by exchange equilibrium. Thus, although ANC formulations in
MAGIC  do consider aluminum dynamics, ANC can also be estimated as Z base cations - 2 acid
anions.  For the constant deposition scenario, the decrease in  the median ANC for M-APP projections
was about 30 fteq/L, which is consistent with the greater increase in sulfate than in base cation
concentrations.

     Watershed exchange pools are tightly coupled with deposition inputs.  This tight coupling of
declining base cation concentrations with declining surface water sulfate concentrations was recently
reported for the Hubbard Brook Experimental  Forest (Driscoll et al., 1989b). Two mechanisms can
contribute to this coupling: (1) decreasing atmospheric deposition of base cations and sulfate and
(2) decreasing release of base cations due to reduced sulfate efflux and depletion of the exchangeable
base pool in soils (Driscoll et al., 1989b; Holdren and Church, 1989).
                                             310
-------
     A primary assumption of the DDRP is that decreased sulfur deposition rates occur in conjunction
with decreased emission rates only for sulfate and hydrogen concentrations (i.e., that concentrations
of other acid anions and base cations are exchanged). For our model simulations, electroneutrality of
deposition was maintained by decreasing hydrogen ion concentrations on a 1:1 equivalent basis with
sulfate. There currently is no  good evidence that decreases in anthropogenic emissions of sulfur are
or should be necessarily accompanied by changes in base cation deposition rates (A. Olsen, pers.
comm.; R. Dennis, pers. comm.) and it remains uncertain whether base cations would decrease with
emission controls.  If the major source of base cations is local entrainment and resuspension,
emission controls will have little influence on base cation concentrations in deposition.  If local sources
of dust-borne base cations are reduced (i.e., by paving roads and employing dust control at construc-
tion sites) long-range transport of base cations from emissions would become an increasingly impor-
tant component of the cation/anion budget of deposition, although total cation deposition would
decrease. By  ignoring changes in base cations, therefore, an additional source of uncertainty has
been included in model projections.

     Process-oriented interpretation of experimental and monitoring data for watersheds in the M-APP
Region are consistent with the processes embodied in MAGIC  simulations (i.e., that mineral weather-
ing, cation exchange, and sulfate sorption are the predominant controls on surface water chemistry in
the M-APP Region). Questions remain, however,  as to whether there are other important watershed
processes affecting surface water acidification that should be incorporated into watershed models.  It
also remains quantitatively unclear how much the model formulations, operational assumptions,
parameter selection, and data aggregation affect  the results  of long-term projections.

10.7.4.2.2  Mineral weathering -

     Mineral weathering is critical for all long-term projections, but it is a process about which little
information at  the watershed scale can be obtained.  Mineral weathering parameters are calibrated as
part of the watershed-by-watershed calibration procedure used by MAGIC, but weathering rates are
not completely unconstrained. Rather, there are limits on the range over which these parameters can
vary while maintaining reasonable ranges for other, related parameters (e.g., historical base saturation)
and still matching observed surface water chemistry constituent concentrations (e.g., silica, calcium,
sodium). Acidic deposition can result in  changes in soil pH  (Falkengren-Grerup et al., 1987; Tamm
and Hallbacken, 1988) and changes in soil pH may influence mineral weathering rates and sulfate
absorption  capacities. In the  M-APP Region, however, median projected changes in 50 years for soil
solution pH were less than 0.2 units (Section 9.3). Although mineral weathering rates cannot be
unequivocally  estimated, the model formulations and mass balance approaches used for the DDRP
simulations are analogous to  the mass balance approaches that have been  used to estimate weather-
ing in watershed studies (Velbel, 1986a; Paces, 1973).

10.7.4.3  Nitrogen Dynamics

      MAGIC focuses on the effects of sulfur deposition on surface water acidification.  MAGIC
considers total deposition acidity,  including nitrate, but the dynamic nitrate formulations are
rudimentary.   Because most eastern forested watersheds are nitrogen limited (Likens et al., 1977;

                                              311
-------
 Swank and Crossley, 1988), ft is likely that nitrogen inputs in most systems are effectively retained in
 vegetation and soil, and nitrate is not an important mobile anion. Deposition inputs of nitrate are
 about twice the inputs of ammonium for the eastern United States (Kulp, 1987). Although nitrification
 has an acidifying effect, nitrate assimilation has an alkalizing effect (Lee and Schnoor, 1988). Nitrate
 concentrations typically are low in M-APP streams and do not play a large role in surface water acid-
 base status at present (Baker et al., 1990). It is possible that nitrogen might become an important
 contributor to chronic surface water acidification in the future for some forested systems. Evaluation of
 this possible effect was outside of the scope of this study, however.  It is well known that nitrate can
 be an important component of episodic acidification (Wigington et al., 1990) in some situations.

 10.7.4.4  Land Use Changes

      Land use changes are  inevitable over a 50- to 100-year period and can have a major influence
 on water quality. Aggrading forests can have an acidifying effect on watershed soils due to  depletion
 of soil base cations and production of organic acids in litter and root exudate (Troedsson, 1980) and
 also due to increased capture of dry deposition. Agricultural land uses can  also significantly affect soil
 chemistry through soil amendments (lime, fertilizer), alteration of soil carbon pools, etc.  The objective
 of the DDRP long-term simulations is not to forecast the future  but rather to  provide a basis for
 comparing the relative effects of alternative acidic deposition scenarios through model projections.
 Omitting considerations of land use change does not seriously compromise this objective. Estimations
 of land use changes are important only if absolute estimates of chemical change are required.  This is
 not the objective of DDRP simulations.

 10.7.4.5  Organic Acidification

      One of the operational assumptions of MAGIC was that the status of organic acids would remain
 constant. Krug and Frink (1983) hypothesized that reversing surface water acidification by reducing
 the input of strong mineral acids could result in increased dissociation of humic acids and mobility of
 organic acids.  This, in turn, could return the acid-base  status of systems that are currently acidic (due
 to mineral acidity) to a previous acidic state (due to organic acidity).  Organic acid concentrations in
 streams throughout the Mid-Atlantic highlands  currently are low (Kaufmann et al., 1988).  Because of
 the nature of the watersheds  and  soils of the region, we believe that a reduction in acidic deposition
 inputs would not lead to significantly increased leaching of organic acids to the point of causing acidic
 conditions.

 10.7.5 Other Considerations
10.7.5.1  Data Aggregation

      In generating data at watershed scale for use in DDRP models, physical and chemical attributes
of soils were averaged (weighted) to master horizons, master horizons were aggregated to sampling
classes, and sampling class attributes were aggregated to the watershed values, which were then
used for calibration and simulations. This averaging or aggregation process preserves the central
tendency in watershed attributes and subsequent projected effects, but  reduces the variability (i.e.,

                                              312
-------
extreme values in the distribution of soil attributes at the horizon-to-watershed scale).  Data
aggregation thus results in underestimates of change in the tails or extremes of the distributions.
Although these extremes represent a small proportion of the target population, the changes in these
watersheds might be underestimated, so the changes in ANC or pH for a fraction of regional surface
waters is likely to be larger than projected.

10.7.5.2 Effects on Upstream Versus Downstream Nodes

      The  DDRP target population represents the lower stream nodes sampled during the NSS. Very
important differences in water chemistry exist between the upstream and downstream nodes of these
reaches (Kaufmann et al., 1988). For example, the median ANC for the target population of upstream
nodes of the reaches modelled in DDRP is over 20 ^eq/L  lower than that for the downstream nodes
(Table 10-14).  Also, about 40 percent more upstream nodes than downstream nodes have ANC
values < 50 fieq/L.  Most strikingly, over five times as many upstream nodes as downstream nodes
are acidic.

      When Kaufmann et al. (1988) presented target population estimates of the lengths of stream
reaches that were acidic or had ANC < 50 ^eq/L, they  used linear interpolation along the  reaches.
The DDRP estimates of stream reach length projected by MAGIC to be acidic or to have ANC < 50
jueq/L are  based solely on chemistry data and simulations for the lower nodes.  If the DDRP estimated
stream reach length in the manner used by the NSS, the  estimates of stream reach length with ANC
< 0 or 50 fteq/L would surely be much greater.  For example, for the stream reaches modelled by
MAGIC, the estimate of current reach length with ANC <  50 ^eq/L based on measured  NSS values for
the lower nodes is 2,505 km  (Table 1O-11).  Using the NSS interpolating procedure, the corresponding
estimate is 4,222 km (A. Herlihy, pers. comm.). Similarly, the estimate of acidic stream reach length is
472 km (Table 10-11), based on measured ANC at lower  nodes, whereas the corresponding estimate
using the  NSS interpolation is 1,557 km (A. Herlihy, pers.  comm.), over three times as great.

      From conducting transects of stream reach chemistry, Kaufmann et al. (1988) found that for
upper stream reaches with low ANC (like the DDRP subsample from NSS), there was a  fairly smooth
and continuous decline in pH upstream from the lower node. This observation supports a hypothesis
that as watershed soils grow more shallow in the upper reaches of catchments, cumulative buffering of
soils  on receiving water decreases and so, presumably, does the ability of the catchment  to buffer
against the effects of acidic deposition.  Thus, estimation of stream reach length < 0 or 50 /*eq/L
based solely on lower node chemistry is surely an underestimate of the length affected, as is the
estimation of future  conditions based upon modelling of conditions only at the lower node. The
estimates of stream reach length projected to have ANC  values < 50 /*eq/L or projected to become
acidic at some future time, as presented  in this report,  are probably large underestimates  of
acidification or recovery as compared to projections that  could be generated  using interpolation
 procedures analogous to those used by the National Stream Survey.
                                             313
-------
Table 10-14. Comparison of ANC Values at Upstream and Downstream Reach Nodes for the
            M-APP DDRP Modelling Target Population
                                                       Node
Median ANC («eq/L)

# sites with current ANC <

# sites with current ANC < 0 fieq/L
Downstream

     87

   1504

    145
Upstream

    64

  2370

   772
                                        314
-------
10.8 REGIONAL COMPARISON: NE, M-APP, SBRP

     Church et al. (1989) found that in the NE, lakes would respond relatively rapidly to changes in
sulfate deposition because the NE was, on the average, near sulfate steady state and because sulfate
retention capacity of NE soils is small. In contrast, for either current or increased sulfur deposition, it
might be 150-200 years before the SBRP watersheds are, on the average, at or near sulfate steady
state (Church et al., 1989). Most M-APP watersheds, on the average, were projected to reach sulfur
steady state in about 50 years.  The M-APP Region has watersheds with characteristics of both NE
and SBRP watersheds and may represent a transition region where some systems can be regarded as
responding relatively  rapidly to sulfate deposition, whereas other systems may have longer response
times. In contrast to  the SBRP, which has no currently acidic streams, about 4 percent of the M-APP
stream reaches are currently acidic at their lower nodes, with acidic deposition the most likely
contributor to stream  acidity (Kaufmann et al., 1988).

     The MAGIC model was used to project changes in surface water chemistry in all three DDRP
regions, so relative comparisons among the three regions can be made using projections based on
simulations from a  common model. In the NE, MAGIC projections indicated that less than an
additional 1  percent of the lakes might become acidic (ANC < 0 ,weq/L) in 50 years under current
deposition.  This estimate represents the difference between the number of nonacidic lakes projected
to become  acidic and currently acidic lakes that might become nonacidic under current  deposition
levels.  In the SBRP,  MAGIC projections indicated 1  percent of the currently nonacidic streams might
become acidic within 50 years under current deposition and the ANC of an  additional 9 percent of
streams was projected to decrease to <  50 /weq/L  By contrast, in the M-APP, MAGIC projections
indicate much larger  changes, with an additional 8 percent of the stream population projected to
become acidic under current deposition after 50 years, and an additional 29 percent of the stream
population with ANC  decreasing to < 50 ^
      With a 30 percent decrease in deposition, there might be about a 2 percent increase in the
target population of nonacidic NE lakes (i.e., currently acidic lakes might become nonacidic) and
about a 2 percent increase in the target population of NE lakes with ANC > 50 ^eq/L in 50 years. In
the M-APP, a 50 percent deposition decrease might result in no (0) acidic streams after 20-25 years
and a 3 percent increase in the target population of M-APP streams with ANC > 50 ^eq/L after 50
years. For the currently acidic  streams, this represents an increase in the median ANC of about 15
^eq/L after 50 years. With an initial 50 percent decrease in deposition followed by a relative increase
to a final 20 percent deposition decrease after 50 years, there might be an increase in ANC for the first
20 years, with no acidic streams after 20 years.  Following this initial ANC increase, however, there
might be increased surface water acidification with about 4 percent of the streams estimated to be
acidic after 50 years. This percent of acidic streams is comparable to the initial calibrated percent of
streams  in the M-APP.

      Comparing these three regions and projections with the MAGIC model for different deposition
scenarios indicates there might not be significant changes in surface water chemistry under current
deposition in the NE but that at least a 30 percent decrease in deposition might be necessary to see
an increase in surface water ANC and minimize continued surface water acidification.  In the SBRP,
                                             315
-------
some streams might become chronically acidic within 50 years under current deposition levels.
Scenarios of decreased deposition were not simulated for the SBRP. In the M-APP, a significant
percent of stream reaches might become acidic within 50 years under current deposition. A 50
percent deposition decrease might be necessary in this region to see an increase in ANC and to
minimize continued surface water acidification.


     We are well aware that long-term modelling is a difficult and imperfect activity. Particularly
vexing is the problem of confirmation, both of models themselves and of their projections. Twelve-year
calibration and confirmation studies on Coweeta Watersheds 34 and 36 indicate that the RMSEs
among the models and the observed standard errors of the data are similar. Subtle long-term trends
were not effectively simulated, however, because of calibration procedures. Regardless, long-term
projections can be confirmed only with long-term periods of record (Simons and Lam, 1980), which do
not exist.  Although our model projections cannot be confirmed, they are consistent with theory and
limited field experiments and observations.


10.9 CONCLUSIONS FROM LEVEL III ANALYSES


     Conclusions from the Level III analyses follow:


     • The M-APP Region is a transition region between NE watersheds with short response times
        and SBRP watersheds with long response times.'

     • Current deposition levels might result in an increase in acidic streams in the M-APP of about
        threefold over 50 years, and about eightfold over 100 years.

     • A 50 percent deposition decrease might result in no (0) acidic streams (acidic  due  to
        atmospheric sulfur deposition)  after 50 or  100 years.

     • A 50/20 percent deposition  decrease might result in an initial increase in ANC after 20-25
        years followed by surface water acidification.  Surface water conditions after 50 years might
        be similar to initial calibrated surface water chemistry for the target population  at year 0.
        Some additional acidification is projected at 100 years.

     • Current deposition levels might result in about a doubling of stream reaches with ANC
        < 50 fieq/L (and thus susceptible to acidic episodes) at  both 50 and 100 years.

     • The rates of change in projected surface water acid-base chemistry using MAGIC are
        comparable to observed rates of change in surface water chemistry ascribed to effects of
        acidic deposition.

     • A quantitative evaluation of the absolute accuracy of long-term model projections is not
        available at present.  Comparisons of relative projected effects among deposition scenarios is
        perhaps more reliable than absolute projections.
                                            316
-------
                                        SECTION 11
                                   SUMMARY OF RESULTS

     This section presents an overview of the results of the DDRP analyses.  Church et al. (1989)
performed statistical analyses of the interrelationships among deposition, edaphic factors, and surface
water chemistry for lake/watersheds in the NE and stream/watersheds in the SBRP.  Those analyses
generally supported the postulated relationships incorporated into both the single factor models (for
sulfate adsorption and cation supply)  and the integrated watershed models.  For example, soil depth,
soil chemical characteristics, and watershed hydrology factors all appear as important explanatory
variables in the regressions performed.  Additionally, wetlands in northeastern watersheds appear to
have an important role in influencing sulfur dynamics (see Sections 7 and 8 of Church et al., 1989).
The watersheds we studied in the M-APP, however, contain few wetland areas (see Sections 5 and 7);
thus, such effects are relatively unimportant in this region.  As discussed in the Preface and in Section
8, we did not perform the lengthy statistical analyses for the watersheds of the M-APP Region.

11.1 RETENTION OF ATMOSPHERICALLY DEPOSITED SULFUR
11.1.1 Current Retention

       We computed net annual retention of atmospherically deposited sulfur for watersheds not
having apparent significant sulfur internal sources (see Section 7). At present, watersheds in the NE
appear to be approximately at steady state, whereas median net retention is about 75 percent in the
SBRP. These observations are qualitatively consistent with theory (Galloway et al., 1983).

      The M-APP Region is a transition zone between the NE and SBRP in terms of observed current
sulfur retention.  Because of the similarities between soils in the M-APP Region and the SBRP, it is
possible that the M-APP Region at one time retained as much of the elevated sulfur deposition as is
now evident in the SBRP (i.e., 70 - 80 percent). It is also possible that sulfur deposition has
decreased this retention, perhaps very dramatically in the westernmost area (subregion 2Cn of the
National Stream Survey), which  now has median percent sulfur retention of only three percent (Plate
11-1). High sulfur deposition and decreased retention, in turn, may have led to the low ANC and
acidic stream reaches (excluding stream reaches affected by acid mine drainage) identified there by
the National Stream Survey (Kaufmann et al., 1988).

11.1.2  Projected Retention

      As indicated in Section 10.1.1  current sulfur retention in M-APP watersheds is highly variable,
ranging from systems already at steady state to sites with retention exceeding 70 percent. The
projected dynamic response of  sulfur in M-APP watersheds is similarly variable; some watersheds have
already reached steady state while others are  not projected to come to steady state under current
deposition for 150 years. The sulfate dynamics described in the remainder of this  section are based
on projections using the integrated MAGIC model, but projections from the Level II sulfate modelling
 (Section 9.2) are almost identical. For a scenario of constant deposition, model  projections under-
 estimate current sulfur retention; median simulated retention in year 0 was 26 percent (compared

                                             317
-------
-------
Plate 11-1. Sulfur retention and wet sulfate deposition for National Surface Water Survey
subregions in the eastern United States.
                                            318
-------
-------
                         NSWS  SUBREGiONS
             MEDIAN  PERCENT  SULFUR  RETENTION
                 AND  WET  SULFATE  DEPOSITION
                                                           2.25
yEDIAN  PERCENT
SULFUR  RETENTION



g  o  - 20

    20 - 40

    40 - 60

    60 - 80

    80 - 100
Average Annual
Wet Sulfaie       ^   2.75-

Deposiiion (g m"2 yr~')*  3.00^
            3.25
                                                       Eastern lake Survey
                                             -2.25
                                                                lledion
                                                       Subregion  X Retention
                                                         1A
                                                         IB
                                                         1C
                                                         ID
                                                         IE
                                             -H
                                               8

                                              -8
                                             -12
                                                  X2.00
 .00
                                   Notional Sirsorn Survey

                                            lledion
                                   Subregion I Retention
                                                         2Cn
                                                         2Br,
                                                         3B
                                                         n
                                                         2As
                                                         3A
                                               3
                                              40
                                              34
                                              50
                                              75
                                              78
                                              Deposition for 1980 - 1984
                                              (A. Olseni Personal Communicat ion)
-------
-------
to 44 percent for measured retention). Retention is projected to decrease sharply during the simula-
tion period, to a median of 12 percent in 20 years, and to only 4 percent and 2 percent by years 50
and 100, respectively (Plate 11-2). Most of the projected decreases in retention occur during the first
50 years of simulations; the subsequent  rate of decrease in retention (increase in sulfate concentra-
tion) slows substantially because most systems are near steady state. Median time to sulfur steady
state in the M-APP, based on Level II analyses, is considerably shorter (35 years) than in the SBRP
watersheds (61 years) characterized by Church et al. (1989).

     The changes in sulfur retention represent large projected changes in sulfate concentration
during  the simulation period.  Levels of sulfur deposition in the M-APP are much higher than in the NE
or SBRP (Section 5.6).  Steady-state sulfate concentrations in the M-APP, under a scenario of constant
deposition, are also much higher (median = 215/teq/L compared to 111 ^eq/L in the NE  and 120
,aeq/L in the SBRP), and changes in concentration between current conditions and steady state are
large.  Although current retention in the M-APP is lower than in the SBRP, the median change in
sulfate as systems come to steady state is larger, 107 ,weq/L versus 83 ^eq/L in the SBRP. The
maximum increase in the M-APP is almost 250 ^eq/L.

     Under scenarios  of reduced sulfur deposition, projected changes in sulfate are much smaller.
For the scenario with a 50 percent decrease in deposition (Scenario B), current sulfate concentrations
in most systems would be above the steady-state concentration. Concentrations are thus projected to
decrease as systems come to steady state with the lower level of deposition, from a simulated median
of 161  ^eq/L in year 0,  to 134, 123, and  120 /
-------
-------
Plate 11-2.  Changes in sulfur retention up to 100 years for watersheds of the M-APP Region as
projected by the MAGIC model for its specified DDRP target population (see Section 10.7.1).
                                           320
-------
-------
  Mid-Appalachian
       Streams
       PERCENT
SULFUR  RETENTION
     Model = MAGIC
         3rd Quartile +
         (1.5 x Interquartile Range)"
         3rd Quartile
         Mean
         Median
         1st Quartile
         1st Quartile -
         (1.5 x Interquartile Range)"
    " Not to exceed extreme value.
                                    Current Deposition
ct:

ce:
   50 —
  -50
 '-100
                            -  100 —|
                            •—
                            S   50 —
                            Ul_J
                            "    0 -
                            O£L
                            Z3
                            •-1-  -50 -
  -100
                                      50% Decrease
              CSI   IO   O
                                    50%/20%  Decrease
                                50 -
                            CE:

                            CE:
     0 —
                             i  -50

                             '-100
                                      C3   O
                                      >-   ct:
-------
-------
essentially an equilibrium process in soils, but the pool of exchangeable bases is limited. Because
most soils in watersheds considered by the DDRP are already strongly acidic, projected effects of
acidic deposition on soil pH are small (Section 9.3), and weathering rates would not be expected to
rise substantially.  As long as weathering rates exceed the rate of acid anion flux from a watershed,
base cation resupply will exceed acid anion output, and  systems will not become acidic. If, however,
acid anion output exceeds weathering rates, exchange can provide a temporary source of bases to
neutralize acidity;  with time, however, the finite exchange reservoir will be depleted and runoff will
eventually become acidic.

      Our analyses suggest that the importance of cation exchange in the control of cation concentra-
tions and ANC is greatest in  systems with low surface water ANC (< 100 ^eq/L), and that in systems
with higher ANC, weathering resupply will maintain positive surface water ANC.  In low ANC systems,
the relative contributions of weathering and exchange cannot be accurately determined.

11.2.2 Future Effects

      Using two soil chemistry models,  developed  by Reuss and Johnson (1986) and by Bloom and
Grigal (1985), we  assessed the potential importance of soil cation  exchange in delaying surface water
acidification.  These analyses represent a worst-case assessment that considers only exchange
processes in the upper two meters of soil (i.e., weathering is not considered). Projections were
generated assuming sulfate to be completely mobile. Model projections reflect two processes  occur-
ring within the soil:  (1) a mobile anion effect (i.e., to maintain charge balance, concentrations of both
basic and acidic cations increase or decrease with changes in sulfate) and (2) depletion of the soil
base  cation exchange pool (i.e., as the  proportion  of base cations on the solid phase is depleted, the
proportion of bases in solution will also  decrease and the proportion of acidic cations in solution will
increase).  Results of simulations from the two cation models are generally consistent.  For a
deposition Scenario A, of constant deposition, ANC is projected to decrease  only slightly during the
first 50 years of simulations,  then to drop  sharply between years 50 and 100  as the exchange pool
becomes depleted (median projected decreases 0.3, 3,  and ys^weq/L at 20, 50, and 100 years).  For
the same period,  soil base saturation is projected to decline steadily, from 14 to 9 percent; soil pH
drops slightly during the first 50 years of simulations, but the decline then accelerates as the
exchange buffer system is depleted.  Soil solution  aluminum concentrations increase in a corre-
sponding nonlinear pattern, from 9 ^mol/L at the start of simulations to 22 jumol/L at year 100.   For
the scenarios of reduced sulfur deposition, changes are much smaller. ANC is projected to increase
(median increase of 6 ^aeq/L after 100 years) as a consequence of deposition Scenario B,  due to a
mobile anion effect.  Soil pH, base saturation, and soil solution aluminum remain almost constant
under the same scenario. For Scenario C, ANC is projected to initially increase, then to decrease by a
median of 8 ^eq/L after 100  years.  For the same scenario, soil pH and base saturation decrease over
the 100-year simulation period, but by much smaller amounts than if current  rates of deposition are
maintained.
                                              321
-------
 11.3  INTEGRATED EFFECTS OF SURFACE WATER ANC
      As noted in the Preface of this report, as discussed in Section 1.3.4, and as fully described in
Section 10, we applied the MAGIC watershed model to study the stream watersheds in the M-APP
Region.  At present, it is not possible to quantify the long-term (i.e., 50-100 years) accuracy of future
projections of any watershed acidification model, including the MAGIC model.  Calibration and short-
term (10 years) verification tests, as well as comparisons to other field and laboratory studies and
paleoecological reconstructions of past lake chemistries, however,  are reasonably encouraging (see
Sections 10 and 1.5).

      The MAGIC model was successfully calibrated for 29 of the 36 DDRP stream reaches (at the
lower nodes;  see Section 10) in the M-APP. The 36 DDRP watersheds represent 5,496 stream
reaches with  a combined reach length of 15,239 km. The 29 watersheds simulated by MAGIC
represent 4,298 stream reaches totalling 11,246 km of stream reach length. Modelling estimates of
effects on stream chemistry were performed for time periods up to 100 years. As discussed in Section
10, the modelling of the stream reaches, based on calibrations to observed chemical concentrations at
the lower nodes of each stream reach, resulted in projections of conditions for the most buffered part
(i.e., the lower end) of the stream reaches under study (see also Kaufmann et al., 1988).  Therefore,
projections of changes in chemical concentrations for stream reach lengths are conservative (i.e.,
underestimates).

      The fact that calibrated sulfate concentrations exceed observed concentrations also has an
effect (see Table 10-6). For projections of increased sulfate concentration, the change from the
calibrated value to the steady-state value is an underestimate, leading to an underestimate of acidi-
fication (i.e., loss of ANC). For projections of decreased sulfate concentrations, the effect is an
overestimate  of ANC recovery.

      As discussed in Section 10, the watershed modelling analyses make use of watershed soil
representations as aggregated from the DDRP Soil Survey. Because the focus of the DDRP is on
regional characteristics and responses, soils data were gathered and aggregated to capture the most
important central tendencies of the systems under study.  As a result, extremes of individual watershed
responses are underestimated  in the analyses presented in this report.  Those systems that MAGIC
projects to respond to the greatest extent or most quickly to current or altered levels of sulfur
deposition might,  in fact, respond even more extensively or more quickly than indicated here. This
should be kept in mind when evaluating the results of the MAGIC simulations.

      The projected changes in the acid-base status of M-APP streams can be presented in  a variety
of ways.  Plate 11-3 illustrates the projected changes in the distributions of ANC for each of the three
deposition scenarios for up to 100 years. Plate 11-4 shows the associated projected changes in pH.
More detail on the projected changes in ANC and pH at 50 years is shown in Plates 11-5 and 11-6,
respectively.  Plates 11-3 and 11-4 indicate patterns of (1) a monotonic decrease in both ANC and pH
for the constant deposition scenario, and (2) an increase in ANC and pH in the short term (20 years),
with longer term decreases for  both decreased deposition scenarios. For the 50 percent decreased
deposition scenario, there are slight decreases in ANC and pH over the long term, presumably due to

                                             322
-------
Plate 11-3.  Changes In ANC up to 100 years for stream reaches of the M-APP Region as
projected by the MAGIC model for its specified DDRP target population (see Section 10.7.1).
                                          323
-------
-------
Mid-Appalachian
      Streams

  ANC  vs.   TIME
    Model  =  MAGIC
                       300


                       200


                       100


                         0


                       •100
                                      Current Deposition
                                   Cvl   UO



                                   CtZ   C£.
     r.
3rd Quoriiie +
(1.5 x Interquartile Range)"

3rd Quortile

Mean

Uedian

1st Quorlile

1st Quartile -
(1.5 x Interquartile Range)"
  " Not to exceed extreme value.
                               300 —i


                             < 200 -
                             CT

                             .£100-


                             =   0 H


                              -100 -
                                        50%  Decrease
                                       o


                                       az.
Cvl


oz
                       300


                     ^L 200
                     cr
                     a>
                     => 100
                                      50%/20%  Decrease
                             <-3
                              -too
                                           CvJ   LO



                                           CtZ   &Z
                                                    o
                                                    <=s
-------
-------
Plate 11-4.  Changes in pH up to 100 years for stream reaches of the M-APP Region as
projected by the MAGIC model for its specified DDRP target population (see Section 10.7.1).
                                         324
-------
-------
Mid-Appalachian
      Streams
   pH  vs.   TIME
    Model  =  MAGIC
8.0

7.0

6.0

5.0

4.0
                                      Current Deposition
            CV1   LO   O
       Ct±            •«—
       >—   Ct:   CrZ
            >-   >—   ct:
        3rd Quortile +
        (1.5 x Interquartile Range)"
        3rd Quartile
        Mean
        Median
        1st Quartile
        1st Quartile -
        (1.5 x Interquartile Range)"
  " Not to exceed extreme value.
8.0-

7.0 -

5,0 —

5.0 -

4.0-
                                        50%  Decrease
            c-j   to
8.0

7.0

6.0

5. 0

4.0
                                      50%/20%  Decrease
                                       O   C3

                                       O£
-------
-------
Plate 11-5.  Changes in median ANC at 50 years for stream reaches of the M-APP Region as
projected by the MAGIC model for its specified DDRP target population (see Section 10.7.1).
                                         325
-------
-------
CHANGE  IN  MEDIAN   ANC
     Year 0 to Year 50
        Model = MAGIC
                                 l-APP Study Area
 C3T>
 c=
 C3
30
20
10H
        Constant
       Deposition
    20-
    30-
                     50%
                   Decrease
30n
20
10-I
                       rr~
               1 0-
               20-
               30-
                        is

                        :» j
                   50%/20%
                   Decrease
30n
20-
10-
 0
10
20
30
        10
                        UO
        V
            i  i

           tO  CD
                        V
                          to
                                          to  cz>
                                          OJ  CZ>
                 ANC Group (ueq/L)
-------
-------
Plate 11-6.  Changes in median pH at 50 years for stream reaches of the M-APP Region as
projected by the MAGIC model for its specified DDRP target population (see Section 10.7.1).
                                         326
-------
-------
 CHANGE  IN  MEDIAN  pH
      Year 0  to  Year  50
        Model =  MAGIC
                                    I-APP Study Area
er-
as
CTl
c=
1 ,0

0 ,5H

  0
 Constant
Deposition
                           50%
                         Decrease
   o ,5
     -0,40
             -0.11
          -0,29
        IO  CD  CD

        OM  CD  CD
        \S
           i   i

           LO  CD
 1  .0

 0,5

    0

-0.5H
                        0.67
                           0.05
                              °-09
                        Cvj


                        \s
                         CD  CD

                         CD  CD
                         I   I

                         IO  CD
                         Osl  CD
                  ANC Group (ueq/L)
                                       50%/20%
                                       Decrease
                            1.0-1

                            0,5-

                              0

                            0.5H
0.05
                                         -0.05 -0.02
                                 to  o
                                 CM  C3>
                                            OJ  CD
-------
-------
modelled depletion of the soil cation exchange complex. For the 50/20 percent decrease deposition
scenario, the decreases are due principally to the ramped increase of deposition back up to the 20
percent level at year 50.  Plates 11-5 and 11-6 show that, although the greatest ANC changes are not
projected for the lowest ANC group (i.e., < 25 fieq/L), the largest pH changes are projected for this
group. This reflects, of course, the logarithmic relationship between these two variables.

      Another way to present the results of the modelling is as changes in the percent or length of
stream reaches projected to  have ANC < Q /teqfL (Figures 11-1 and  11-2) or < 50,weq/l_ (Figures 11-3
and 11-4) in the future.  Because MAGIC has some bias in its  calibration to year 0 (see Section 10),
these values are computed by adding the projected changes in ANC to the values observed by
Kaufmann et al. (1988) for the National Stream Survey.  The changes in ANC projected by MAGIC are
unbiased with respect to year-0 ANC (J. Cosby, pers. comm.). A one-sided hypothesis test of
significance for the changes  in numbers of systems  (i.e., different from zero; see Section 6) indicates
changes (p  < .05) at years 20, 50, and 100 for stream reaches with ANC < 0 fieq/L for the current
deposition scenario (see Section  10). Changes at 50 and 100 years for the projections of systems
with ANC < 50 /ieq/L are also evident (p = .01). For the 50/20 percent decreased deposition
scenario, changes (increases) in the numbers of stream reaches with ANC < 0 or ANC < 50 yeq/L are
indicated (.07 - p ^ .09) at both  50 and 100 years.  All other projected changes are less well
supported (p > .16) (see Table 10-9).

      Considered together, the results shown in Plates  11-3 through  11-6 and Figures 11-1 through
11-4 indicate a modelled projection of marked decreases in the ANC and pH of M-APP stream reaches
(as represented by chemical  conditions at the lower reach nodes) under continued levels of current
deposition.  These projected  changes in ANC translate  also to marked increases in (1) the number of
acidic stream reaches (ANC  < 0 ^weq/L), and (2) the number of stream reaches with ANC < 50 jueq/L.
Stream reaches with ANC  <  50 /ueq/L are probably highly susceptible to episodic acidification to
acidic conditions during periods of heavy snowmelt or rainfall (Eshleman, 1988).

      Projections for the 50 percent decreased deposition scenario indicate somewhat improved (rela-
tive to toxic conditions for fish) chemical conditions at 50 years (especially pH for low ANC systems),
with perhaps a slight decrease in water quality after that time.  Projections for the 50/20 percent
decreased deposition scenario indicate chemical conditions at 50 years very close to those observed
by Kaufmann et al. (1988)  in  the National Stream Survey. Decreases in water quality are projected at
100 years under this deposition scenario.

11.4  SUMMARY DISCUSSION

      Our analyses of (1) current sulfur mobility, (2) projected sulfur mobility, (3) cation exchange
buffering, and (4) integrated watershed modelling present an internally consistent picture of the
potential long-term effects  of sulfur deposition on the chemistry of our target population of stream
reaches in the M-APP Region.  Although soils of our study watersheds have greater ability to retain
future atmospheric inputs of sulfate than do those of the NE, this ability is quite limited relative to soils
of the SBRP (Church et al., 1989). Soils of the M-APP Region  are projected to reach sulfur input-
output steady state faster than those of the SBRP (median time to steady state  is 35 years for the
                                             327
-------
                 d>
                     30
                     20-
                 c
                 0>
                 o
                 o>  10
                 D.
                               Current  Deposition     ^
                            eo
                         Y////A
                                     20
                               50
         100
    30

O
D)
«   20
C
                o
                    10-
                           50% Decreased  Deposition
                            CO
                         Y////A
                           0
                    20
50
100
                    30
                 0
                 O)
                 «  20-
                 c
                 a)
                 o
                 o
                DL
50
%/20% Decreased Deposition
CO
CO *
"7 S~7 S~7 7
X/////A ° '/////,
•
                           0        20       50      100
                                   Time  (yr)
Figure 11-1.   F3ercentage of stream reaches of the DDRP M-APP modelling target population
            (see Section 10.7.1) projected by the MAGIC model to have ANC < 0 fieq/L at the
            lower reach nodes. Estimates at year 0 are from the National Stream Survey
            (Kaufmann et al., 1988) for the same target population.  Estimates at 20, 50, and
            100 years are computed by adding the change in ANC projected by the MAGIC
            model to ANC observed by the National Stream Survey.
                                      328
-------
               ,
              c
              
-------
                  100
               §,
               CO
                   50-
               
-------
              D)
              C
              -
r-
CM
                                      20
                           50
                                                          co
                          100
              5
              o
              (0
              o
              CC
10000 -
7500-
5000-
2500-
rt .
50%

m
o
in
CM
HI
Decreased Deposition

t~
CM
CM
CM
^

CM 2
S S
Wk HI
                                      20
                           50
                          100
                    10000
              D)
              C
              0>


              JC
              o
              C3
              o
              CC
 7500-



 5000-



 2500-



    0
                             50%/20%  Decreased Deposition
                            O)
                            CD
in
o
m
CM
CM
CM
CM
                              0       20      50      100


                                     Time  (yr)



Figure 11-4.  Length (km) of stream reaches of the DDRP M-APP modelling target popula-

            tion (see Section 10.7.1) projected by the MAGIC model to have ANC < 50
            fieq/L at the lower reach nodes.  Estimates at year 0 are from the National
            Stream Survey (Kaufmann et al.,  1988) for the same target population. Esti-

            mates at 20, 50, and 100 years are computed  by adding the change in ANC
            projected by the MAGIC model to ANC observed  by the National Stream
            Survey.
                                       331
-------
M-APP, 61 years for the SBRP).  Calculated steady-state sulfate concentrations in the M-APP Region
(214
-------
                                         SECTION 12
                                        REFERENCES
Abrahamsen, G., J. Hoviand, and S. Hagver. 1980. Effects of artificial acid rain and liming on soil
      organisms and the decomposition of organic matter, pp. 341-362. In: T.C. Hutchinson and M.
      Havas, eds. Effects of Acid Precipitation on Terrestrial Ecosystems. Plenum Press, New York.

Adams, F., and Z. Rawajfih. 1977. Basaluminite and alunite: A possible cause of sulfate retention by
      acid soils. Soil Sci. Soc. Am. J. 41:686-692.

Altshuller, A.P., and R.A. Linthurst. 1984. The Acidic Deposition Phenomenon and Its Effects: Critical
      Assessment Review Papers. EPA/600/8-83/016bf. U.S. Environmental  Protection Agency,
      Washington, DC.

Asbury, C.E., F.A. Vertucci, M.D. Mattson, and G.E. Likens. 1989. Acidification of Adirondack lakes.
      Environ. Sci. Technol. 23:362-365.

Aylmore, L.A., G.M. Karim, and J.P. Quirk. 1967. Adsorption and desorption of sulfate ions by soil
      constituents. Soil Sci. 103:10-15

Baker, L.A., and P.L. Brezonik. 1988. Dynamic model of in-lake alkalinity generation. Water Resour.
      Res. 24:65-74.

Baker, LA., P.L. Brezonik, and E.S. Edgerton. 1986a. Sources and sinks of ions in a softwater acidic
      lake in Florida. Water Resour. Res. 22:715-722.

Baker, L.A., P.L. Brezonik, and C.D. Pollman. 1986b. Model of internal alkalinity generation: Sulfate
      retention component. Water Air Soil Pollut. 31:89-94.

Baker, LA., P.R. Kaufmann, A.T. Herlihy, and J.M. Eilers.  1990.  Current Status of Surface Water Acid-
      Base Chemistry. NAPAP SOS/T Report 9, In: Acid Deposition: State of Science and Technology.
      National Acid Precipitation Assessment Program, Volume II, Washington, DC.

Baker, L.A., P.R. Herlihy, P.R. Kaufmann, and J.M. Eilers. 1991. Acidic lakes and streams in the United
      States: The role of acidic deposition. Science. 252:1151-1154.

Bard, Y. 1974. Nonlinear Parameter Estimation.  Academic Press, New York. 341pp.

Barrow, N.J.  1961. Studies on mineralization of sulfur from soil organic matter. Aust. J. Agr.  Res.
      12:306-319.

Barrow, N.J.  1967. Studies on the adsorption of sulfate by soils. Soil Sci. 104:342-349.
                                             333
-------
Bartling, M.H., M.L. Papp,, and R.D. Van Remortel. 1988. Direct/Delayed Response Project:
     Preparation Laboratory Standard Operating Procedures. In: M.L Papp and R.D. Van Remortel. in
     prep. Direct/Delayed Response Project: Preparation Laboratory Operations and Quality
     Assurance Report for the Mid-Appalachian Soil Survey.

Bayley, S.E., R.S. Behr, and C.A. Kelly. 1986. Retention and release of S from a freshwater wetland.
     Water Air Soil Pollut. 31:101-114.

Bettany, J.R., J.W.B. Stewart, and E.H. Halstead. 1973. Sulfur fractions and carbon, nitrogen and sulfur
     relationships  in grasslands, forest and associated transitional soils. Soil Sci. Soc. Am. J.
     37:915-918.

Beven, K.J., and J.M. Kirkby. 1979. A physically based, variable constant contributing area model of
     basin hydrology.  Hydrol. Sci. Bull.  24:43-69.

Bloom, P.R., and DP. Grigal. 1985. Modeling soil response to acidic deposition in nonsulfate
     adsorbing soils. J. Environ. Qual. 14:489-495.

Bornemisza, E., and R. Llanos. 1967. Sulfate movement, adsorption, and desorption in three Costa
     Rican soils. Soil Sci. Soc. Am. Proc. 31:356-360.

Brezonik, P.L., L.A.  Baker, and T.E. Perry. 1987. Mechanisms of alkalinity generation in acid-sensitive
     softwater lakes, pp. 229-262. In: R.  Hites and S.J. Eisenreich, eds. Sources and Fates of Aquatic
     Pollutants. Advances in Chemistry,  Series 216. American Chemical Society, Washington, DC.

Byers, G.E., R.D. Van Remortel, M.J. Miah, J.E. Teberg, M.L. Papp, B.A. Schumacher, B.L. Conkling,
     D.L. Cassell, and P.W. Shaffer. 1990. Direct/Delayed Response Project: Quality Assurance
     Report for Physical and Chemical Analyses of Soils from the Mid-Appalachian Region of the
     United States. EPA/600/4-90/001. U.S. Environmental Protection Agency, Washington, DC.
     337 pp.

Campbell, W.G., and M.R. Church. 1989.  EPA uses GIS to study lake and stream acidification. Federal
     Digital Cartography Newsletter 9:1-2.

Campbell, W.G., M.R. Church, G.D. Bishop, D.C. Mortenson, and S.M. Pierson. 1989. The role for a
     Geographic Information  System in a large environmental project. Internal. J.  GIS.  3:349-362.

Chan, W.H., D.B. Orr, and R.J. Vet. 1982a. The Acid Precipitation in Ontario Study  (APIOS). An
     Overview: The Cumulative Wet/Dry  Deposition Network. Report #ARB-15-82-ARSP, Ontario
     Ministry of the Environment.

Chan, W.H., D.B. Orr, and R.J. Vet. 1982b. The Acid Precipitation in Ontario Study  (APIOS). An
     Overview: The Event Wet/Dry Deposition Network. Report  #ARB-11-82-ARSP, Ontario Ministry of
     the Environment.
                                             334
-------
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. 92:14681-14700.

Chao, T.T., M.E. Harward, and S.C. Fang. 1962a. Movement of ^S tagged sulfate through soil
      columns. Soil Sci. Soc. Am. Proc. 26:27-32.

Chao, T.T., M.E. Harward, and S.C. Fang. 1962b. Adsorption and desorption phenomena of sulfate
      ions in soils. Soil Sci. Soc. Am. Proc. 26:234-237.

Chao, T.T., M.E. Harward, and S.C. Fang. 1964a. Anionic effects on sulfate adsorption by soils. Soil
      Sci. Soc. Am. Proc. 28:581-583.

Chao, T.T., M.E. Harward, and S.C. Fang. 1964b. Iron or aluminum coatings in relation to sulfate
      adsorption characteristics of soils. Soil Sci. Soc. Am. Proc. 28:632-635.

Charles, D.F., and J.P. Smol. 1990. The PIRLA II Project:  Regional assessment of lake acidification
      trends. Verh. Internat. Verein. Limnol. 24:474-480.

Chen, C.W., S. Gherini, R.J.M. Hudson, and J.D. Dean. 1983. The Integrated Lake/Watershed
      Acidification Study, Volume 1: Model and Procedures. Rep. No. EA-3221. Electric Power
      Research Institute, Palo Alto, CA.

Chen, C.W., S.A. Gherini,  R.J.M. Hudson, and J.D. Dean. 1984. The Integrated Lake-Watershed
      Acidification Study, Volume 2: Hydrologic Analysis. C.W. Chen, ed. Rep. No. EA-3221. Electric
      Power Research Institute, Palo Alto, CA.

Chou, L, and R. Wollast. 1985. Steady-state kinetics and dissolution mechanisms of albite. Am. J. Sci.
      285:963-993.

Chrisman, N.R. 1987. Efficient digitizing through the combination of appropriate hardware and software
      for error detection and editing. Intl. J. Geog. Info. Systems. 1:265-277.

Church, M.R. 1989. Predicting the future long-term effects of acidic deposition on surface water
      chemistry: The Direct/Delayed Response Project. Eos, Transactions, American Geophysical Union
      70:801-802,812-813.

Church, M.R., and R.S. Turner. 1986. Factors Affecting the Long-term Response of Surface Waters to
      Acidic Deposition: State of the Science. EPA/600/3-86/025. NTIS PB86 178 188/AS. U.S.
      Environmental Protection Agency, Corvallis, OR. 274 pp.
                                             335
-------
Church, M.R., K.W. Thornton, P.W. Shaffer, D.L Stevens, B.P. Rochelle, G.R. Holdren, M.G. Johnson,
      JJ. Lee, R.S. Turner, D.L Cassell, DA Lammers, W.G. Campbell, C.I.  Liff, C.C. Brandt, LH. Liegel,
      G.D. Bishop, D.C. Mortenson, S.M. Pierson and D.D. Schmoyer.  1989. Direct/Delayed Response
      Project:  Future Effects of Long-term Sulfur Deposition on Surface Water Chemistry in the
      Northeast and Southern Blue Ridge Province. EPA/600/3-89/061 a-d,  U.S. Environmental
      Protection Agency, Washington, DC, 887 pp.

Clark, T.L, R.L. Dennis, and S.K. Seilkop. 1989. Re-examination of Interim Estimates of Annual Sulfur Dry
      Deposition Across the  Eastern United States. EPA/600/4-89/026. U.S.  Environmental Protection
      Agency, Atmospheric Research and Exposure Assessment Laboratory, Research Triangle Park,
      NC, 34 pp.

Cochran, W.G. 1977. Sampling Techniques, pp. 259-261. John Wiley & Sons, Inc., New York,  NY.

Comfort, S.D., R.P. Dick, and J. Baham.  1991.  Air-drying and pretreatment effects on soil sulfate
      sorption.  Soil Sci. Soc. Am. J. 55:968-973.

Cook, R.B., and D.W. Schindler. 1983. The biogeochemistry of sulfur in an experimentally acidified lake.
      Ecol. Bulletin (Stockholm). 35:115-127.

Cosby, J., University of Virginia, Charlottesville, Virginia, personal communication.

Cosby, B.J., G.M. Hornberger, J.N. Galloway, and R.F. Wright. 1985a.  Modeling the effects of acid
      deposition: Assessment of a lumped  parameter model of soil water and streamwater chemistry.
      Water Resour. Res. 21:51-63.

Cosby, B.J., G.M. Hornberger, J.N. Galloway, and R.F. Wright. 1985b.  Time scales of catchment
      acidification: A quantitative model for estimating freshwater acidification. Environ. Sci. Technol.
      19:1144-1149.

Cosby, B.J., G.M. Wright, G.M.  Hornberger, and J.N. Galloway. 1985c. Modeling the effects of acid
      deposition: Estimation  of long-term water quality responses in a small forested catchment.  Water
      Resour. Res. 21:1591-1601.

Cosby, B.J., G.M. Hornberger, E.B. Rastetter, J.N. Galloway, and R.F. Wright. 1986a. Estimating
      catchment water quality response to acid deposition using mathematical models of soil ion
      exchange processes. Geoderma 38:77-95.

Cosby, B.J., G.M. Hornberger, R.F. Wright, and J.N. Galloway. 1986b.  Modeling the effects of acid
      deposition: Control of long-term sulfate dynamics by soil sulfate adsorption. Water  Resour. Res.
      22:1283-1292.
Cosby, B.J., P.G. Whitehead, and R. Neale. 1986c. A preliminary model of long-term change in stream
     acidity in southwest Scotland.  J. Hydro!. 84:381-401.

                                             336
-------
Cosby, B.J., G.M. Hornberger, and R.F. Wright. 1987. A regional model of surface water acidification in
      southern Norway: calibration and validation using survey data. In: J.  Kamari, ed. Proceedings of
      the IISAS-IMGW Task Force Meeting on Environmental Impact Models to Assess Regional
      Acidification. Reidel Publishing.

Cosby, B.J., G.M. Hornberger, P.P. Ryan, and D.M. Wolock. 1989. MAGIC/DDRP Final Report. Volume
      1.  Models, Calibration, Results Uncertainty Analysis,  QA/QC. U.S. EPA Environmental Research
      Laboratory, Corvallis, OR.  286 pp.

Cosby, B.J., A. Jenkins, R.C. Ferrier, J.D. Miller, and  TAB.  Walker.  1990.  Modelling stream acidification
      in afforested catchments:  Long-term reconstructions  at two sites in Central Scotland.  J. Hydrol.
      120:143-162.

Cosby, B.J. and P.P. Ryan.  1991.  Effects of inorganic and organic constituents on the long-term
      variation of the acid-base status of small lakes. Presented °at the Symposium on Environmental
      Chemistry of Lakes and Reservoirs, American Chemical Society National Meeting, April 16-17,
      Atlanta, GA.

Couto, W., D.J. Lathwell, and D.R.  Bouldin. 1979. Sulfate sorption by oxisols and an alfisol of the tropics.
      Soil Sci. 127:108-116.

Cozzarelli, I.M. 1986. Processes of aluminum mobilization and immobilization  in a natural soil system,
      White Oak Run, Virginia.  M.S. Thesis, Department of Environmental Sciences,  University of
      Virginia, Charlottesville, VA.

Cronan, C.S., and C.L Schofield. 1979. Aluminum leaching response to acid precipitation: Effects on
      high-elevation watersheds in the Northeast. Science  204:304-306.

Gumming, B.F.,  J.P. Smol, J.C.  Kingston, D.F. Charles, H.J.B. Birks, K.E. Cambun, S.S. Dixit,
      A.J. Uutala, and A.R. Selle.  How much acidification  has occurred in Adirondack region lakes
      (New York, USA) since pre-industrial times? Can. J.  Fish. Aquat. Sci. (In press).

Dahlgren, R.A., D.C. McAvoy, and  C.T. Driscoll. 1990. Acidification and Recovery of a Spodosol Bs
      Horizon from Acidic Deposition.  Environmental Science and Technology.  24:531-537.

Dasch, J.M. 1987. Measurement of dry deposition to surfaces in deciduous and pine canopies. Environ.
      Pollut. 44:261-277.

David, M.B., and M.J. Mitchell.  1985. Sulfur constituents and cycling in waters, seston, and sediments of
      an  oligotrophic lake. Limnol. Oceanogr. 30:1196-1207.

David, M.B., M.J. Mitchell, and  J.P. Nakas. 1982. Organic and inorganic sulfur constituents of a forest
      soil and their relationship to microbial activity. Soil Sci. Soc. Am. J. 46:847-852.
                                               337
-------
 David, M.B., M J. Mitchell, and S.C. Schindler. 1984. Dynamics of organic and inorganic sulfur
      constituents in hardwood forest soils, pp. 221-246. In: E.L Stone, ed. Forest Soils and Treatment
      Impacts: Proceedings of the Sixth North American Forest Soils Conference, June 1983. Dept.
      Forestry, Fisheries, and Wildlife, University of Tennessee, Knoxville.

 David, M.B., M.J. Mitchell, and T.J. Scott. 1988. Importance of biological processes in the sulfur budget
      of northern hardwood ecosystem. Biol. Fert. Soils. 5:258-264.

 Dennis, R., National Oceanic and Atmospheric Administration, Research Triangle Park, North Carolina,
      personal communication.

 Dillon, P.J., R.A. Reid, and R. Girard.  1986. Changes in the chemistry of lakes near Sudbury, Ontario,
      following reductions of SC^  emissions. Water Air and Soil Pollut. 31:59-65.

 Driscoll, C., Syracuse University, Syracuse, New York, personal communication.

 Driscoll, C.T., and R.M. Newton. 1985. Chemical characteristics of Adirondack lakes. Environ. Sci.
      Technol. 19:1018-1024.

 Driscoll, C.T., C.P. Yatsko, and F.J. Unangst. 1987a. Longitudinal and temporal  trends in the water
      chemistry of the North Branch of the Moose River. Biogeochemistry 3:37-61.

 Driscoll, C.T., B.J. Wyskowski, C.C. Consentini, and M.E. Smith. 1987b. Processes regulating temporal
      and longitudinal variations in the chemistry of a low-order woodland stream in the Adirondack
      Region of Mew York. Biogeochemistry 3:225-241.

 Driscoll, C.T., N.M. Johnson, G.E.  Likens, and M.C.  Feller. 1988. Effects of acidic deposition on the
      chemistry of headwater streams: A comparison between Hubbard Brook,  New Hampshire, and
      Jamieson Creek, British Columbia. Water Resour. Res. 24:195-200.

 Driscoli, C.T., R.D. Fuller, and W.P. Schecher. 1989a. The role of organic acids in the acidification of
      surface waters in the eastern U.S. Water Air Soil Pollut. 43:21-40.

 Driscoll, C.T., G.I-.. Likens, LO. Hedin, J.S. Easton, and F.H.  Bormann. 1989b. Changes in the chemistry
      of surface waters: 25-year results at the Hubbard Brook Experimental Forest, NH. Environ. Sci.
      Technol. 23:137-143.

Driscoll, C.T., M. Lehtinen,  R. Munson, and S.A. Gherini.  1990. The contribution of strong inorganic
      acids and organic acids in regulating hf and Al in Adirondack, NY (USA)  Lakes. Presented at the
      International Conference on Acidic Deposition. Its Nature and Impacts.  Glasgow, Scotland.
      September 16-21.
                                              338
-------
Eder, B.K., and R.L. Dennis. 1990. On the use of scavenging ratios for the inference of surface-level
     concentrations and subsequent dry deposition of Ca2*, Na+, and K1". Water Air Soil Pollut.
     52:197-216.

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 PB86  171 394/AS. U.S. Environmental
     Protection  Agency, Research Triangle Park, NC.

Eshleman, K.N. 1988. Predicting regional episodic acidification of surface waters using empirical models.
     Water Resour. Res. 24:1118-1126.

ESRI. 1987. ARC/INFO Users Guide. Environmental Systems Research Institute, Redlands, CA.

Eyre, F.H. 1980.  Forest cover types of the United States and Canada. Society of American Foresters,
     Washington, DC.

Falkengren-Grerup, U. 1987. Long-term changes in pH of forest soils in southern Sweden. Environ.
     Pollut. 43:79-90.

Fernandez, I.J., and P.A.  Kosian. 1987. Soil air carbon dioxide concentrations in a New England
     spruce-fir forest. Soil Sci. Soc. Am. J. 51:261-263.

Fisher, R.A.  1935. The logic of inductive inference (with discussion). J. Royal Stat. Soc. 98:39-54.

Fitzgerald, J.W.,  and D.W. Johnson. 1982. Transformations of sulphate in forested and agricultural lands,
     pp. 411-426. In: A.I. Moore, ed. Sulphur, Volume  1. British Sulphur Corp., London.

Fitzgerald, J.W.,  and M.E. Watwood. 1987. Sulfur retention in hardwood forest litter and soil: Mechanisms
     and environmental  parameters, pp. 119-126. In: NAPAP Aquatic Effects Task Group VI Peer Review
     Project Summaries, Volume I, May 17-23. New Orleans, LA.

Fitzgerald, J.W.,  T.C. Strickland, and W.T. Swank. 1982. Metabolic fate of inorganic sulphate in soil
      samples from undisturbed  and managed forest ecosystems. Soil Biol. Biochem. 14:529-536.

Fletcher, R., and M.J.D. Powell.  1963. A rapid convergent descent method for minimization. Computer J.
      6:163-168.

Fuller,  R.D., M.B. David, and C.T. Driscoll. 1985.  Sulfate adsorption  relationships in forested Spodosols of
     the northeastern USA. Soil Sci. Soc. Am. J. 49:1034-1040.

Fuller,  R.D., C.T. Driscoll, S.C. Schindler, and M.J. Mitchell.  1986a. A simulation model of sulfur
     transformations in forested Spodosols. Biogeochemistry 2:313-328.
                                              339
-------
Fuller, R.D., MJ. Mitchell, H.R. Krouse, B.J. Wyskowski, and C.T. Driscoll. 1986b. Stable sulfur isotope
      ratios as a tool for interpreting ecosystem sulfur dynamics. Water Air Soil Pollut. 28:163-171.

Gaines, Jr.,  G.J., and H.C. Thomas. 1953. Adsorption studies on clay minerals. II. A formulation of the
      thermodynamics of exchange adsorption. J. Chem. Phys. 21:714-718.

Galloway, J.N., S.A. Norton, and M.R. Church. 1983. Freshwater acidification from atmospheric
      deposition of sulfuric acid: A conceptual model. Environ. Sci. Technol. 17:541-545.

Galloway, J.N., G.R. Hendrey, C.L. Schofield, N.E. Peters, and A.H. Johannes. 1987. Processes and
      causes of lake acidification during spring snowmelt in the west-central Adirondack Mountains, New
      York.  Can. J. Fish. Aquat. Sci. 44:1595-1602.

Gherini, S.A., L Mok, R.J. Hudson, G.F. Davis, C.W. Chen, and  R.A. Goldstein. 1985. The ILWAS model:
      Formulation and application. Water Air Soil  Pollut. 26:425-459.

Gobran, G.R., and  S.I. Nilsson. 1988. Effects of forest floor leachate on sulfate retention  in a Spodosol
      soil.  J. Envir. Qual. 17:235-239.

Goldstein, R.A., and S.A. Gherini, eds. 1984. The Integrated Lake-Watershed Acidification Study. Volume
      4: Summary  of Major Results. EA-3221, Research Project  1109-5.  Electric Power Research
      Institute, Palo Alto, CA.

Graczyk, D., U.S. Geological Survey, Madison, Wisconsin, personal communication.

Grennfelt, P. 1987.  Deposition processes for acidifying compounds. Environ. Technol. Letters 8:515-527.

Grieve,  I.C.  1989.  A Laboratory test of the soil chemical submodels of two  models of catchment
      acidification.  Hydrological Processes. 3:339-346.

Gschwandtner, G.,  K.C.  Gschwandtner, and K. Eldridge. 1985. Historic emissions of sulfur and nitrogen
      oxides in the United States from 1900 to 1980. EPA/600/7-85/009a. U.S. Environmental Protection
      Agency, Research Triangle Park, NC. 105 pp.

Haines,  T.A. 1981. Acidic precipitation and  its consequences for aquatic ecosystems: A review. Trans.
      Am. Fish. Soc. 110:669-707.

Haque,  I., and D. Walmsley. 1973. Adsorption and desorption of sulfate in some soils of  the West Indies.
      Geoderma 9:269-278.

Hayden, N.J. 1987. Sulfate Retention and Release in Six Eastern Soils. M.S.  Thesis.  Dept. Civil and
      Environ. Engineering, Michigan State University, East  Lansing, Ml. 165 pp.
                                              340
-------
Hedin, L.O., G.E. Likens, K.M. Postek, and C.T. Driscoll.  1990. A field experiment to test whether
      organic acids buffer acid deposition.  Nature.  345:798-800*

Helgeson, H.C., W.M. Murphy, and P. Aagaard. 1984. Thermodynamic and kinetic constraints on
      reaction rates among minerals and aqueous solutions: II. Rate constants, effective surface area
      and the hydrolysis of feldspar. Geochim. Cosmochim. Acta 48:2405-2432.

Helvey, D., USDA Forest Service, Parsons, West Virginia, personal communication.

Henriksen, A., and D.F. Brakke. 1988. Increasing contributions of nitrogen to the acidity of surface waters
      in Norway.  Water Air Soil Pollut. 42:183.

Herlihy, A., Oregon State  University,  Corvallis,  Oregon, personal communication.

Hicks, B., National Oceanic and Atmospheric Administration, Washington, D.C., personal  communication.

Hicks, B.B., R.P. Hosker,  Jr., and J.D. Womack. 1986. Comparisons of wet and dry deposition derived
      from the first year of trial dry deposition  monitoring, pp. 486-490. In: Symposium on Acid Rain: I.
      Sources and Atmospheric Processes. Presented before the Division of Petroleum Chemistry, Inc.
      American Chemical Society, April 13-18, New York, NY.

Hingston, F.L, R.J. Atkinson, A.M. Posner, and J.P. Quirk. 1967. Specific adsorption of anions. Nature
      215:1459-1461.

Hingston, F.J., A.M. Posner, and J.P. Quirk. 1972. Anion adsorption by geothite and gibbsite: I. The role
      of the proton in determining adsorption envelopes. Soil Sci.  23:177-192.

Holdren,  G.R., and P.M. Speyer. 1985. pH dependent changes in the rates and stoichiometry of
      dissolution of an  alkali feldspar at room temperature. Am. J. Sci. 285:994-1026.

Holdren,  G.R., and M.R. Church. 1989. Discussion of Driscoll, C.T., G.E. Likens, L.O.  Hedlin, J.S. Eaton,
      and F.H. Bormann. "Changes in the chemistry of surface waters." Environ. Sci. Technol. 23:137-
      143.

Hornberger, G.M., K.J.  Beven, B.J. Cosby, and D.E. Sappington. 1985. Shenandoah Watershed Study:
      Calibration of a topography-based, variable contributing area hydrological model to a small
      forested catchment. Water Resour. Res. 21:1841-1850.

Hornberger, G.M., B.J.  Cosby, and J.N. Galloway. 1986. Modeling the effects of acid deposition:
      Uncertainty and spatial variability in estimation of long-term sulfate dynamics in a region.  Water
      Resour. Res. 22:1293-1302.

Hornberger, G.J., B.J. Cosby, and R.F. Wright. 1989. Historical reconstructions and future forecasts of
      regional surface water acidification in southernmost Norway. Water Resour. Res. 25:2009-2018.

                                              341
-------
Hosker, R.P., and J.D. Womack. 1986. Simple meteorological and chemical filterpack monitoring system
      for estimating dry deposition of gaseous pollutants, pp. 23-29. In: Proceedings of the Fifth Annual
      National Symposium on Recent Advances in Pollutant Monitoring of Ambient Air and Stationary
      Sources, January. EPA/600/9-85/029. Atmospheric Turbulence and Diffusion Division,
      NOAA/ERL/Air Resources Laboratory, Oak Ridge, TN.

Huang, P.M. 1977. Feldspars, olivines, pyroxenes and amphiboles, pp. 553-602. In: J.B. Dixon and S.B.
      Weed, eds., Minerals in Soil Environments.  Soil Science Society of America, Madison, Wl.

Hunsaker, C.T., S.W. Christensen, JJ.  Beauchamp, R.J. Olson, R.S. Turner, and J.L. Malanchuk. 1986.
      Empirical Relationships Between Watershed Attributes and Headwater Lake Chemistry in the
      Adirondack Region. Environ. Sci. Div. Pub. No. 2884. ORNL/TM-9838. Oak Ridge National
      Laboratory, Oak Ridge, TN.  123  pp.

Hutchlnson, T.C., and M. Havas. 1986. Recovery of previously acidified lakes near Coniston, Canada,
      following reductions in atmospheric sulphur and metal emissions. Water Air Soil Pollut.  23:319-333.

Hynes, H.B.N. 1975. Edgardo Baldi memorial lecture: The stream and its valley. Verh. Internal. Verein.
      Umnol. 19:1.

Jenkins, A., B.J.  Cosby, R.C. Ferrier, TAB. Walker, and J.D. Miller. 1990a. Modelling stream acidification
      in afforested catchments: An assessment of the relative effects of acid deposition and
      afforestation. J. Hydrol. 120:163-181.

Jenkins, A., P.G. Whitehead,  B.J. Cosby, and H.J.B. Birks. 1990b. Modelling long term acidification: A
      comparison with diatom reconstructions and the implications for reversibility. Phil. Trans. R. Soc.
      Lond. B. 327:435-440.

Jenkins, A., P.G. Whitehead,  T.J. Musgrove, and B.J. Cosby. 1990c. A regional model of acidification in
      Wales. J. Hydrol. 116:403-416.

Jenne, E.A., LE. Eary, LW. Vail, D.C. Gin/in, A.M. Liebetrau, LF. Hibler, T.B. Miley, and M.J. Monsour.
      1989. An Evaluation and Analysis of Three Dynamic Watershed Acidification  Codes (MAGIC, ETD,
      and ILWAS). PNL-6687/UC-11.  Richland, WA.

Johnson, D.W., and D.W. Cole. 1980. Anion mobility in soils: Relevance to nutrient transport from forest
      ecosystems. Environ. Internal. 3:79-90.

Johnson, D.W., and J.O. Reuss. 1984.  Soil mediated effects of atmospherically deposited sulphur and
      nitrogen. Phil. Trans. R. Soc. Lond. 3056:383-392.

Johnson, D.W., and D.E. Todd. 1983. Relationships among iron, aluminum, carbon, and sulfate in a
      variety of forest soils. Soil Sci. Soc. Am. J. 47:792-800.
                                              342
-------
Johnson, D.W., and D.E. Todd. 1987. Nutrient export by leaching and whole-tree harvesting in a loblolly
      pine and mixed oak forest. Plant and Soil 102:99-109.

Johnson, D.W., J.W. Hornbeck, J.M. Kelly, W.T. Swank, and D.E. Todd. 1980. Regional patterns of soil
      sulfate accumulation: Relevance to ecosystem sulfur budgets, pp. 507-519.  In: D.S. Shriner, C.R.
      Richmond and S.E. Lindberg, eds. Atmospheric Sulfur Deposition: Environmental Impact and
      Health Effects. Ann Arbor Science, Ann Arbor, Ml.

Johnson, D.W., J. Turner, and J.M. Kelly. 1982. The effects of acid rain on forest nutrient status. Water
      Resour.  Res. 18:449-461.

Johnson, D.W., G.S. Henderson,  and D.E. Todd. 1988. Changes in nutrient distribution in forests and
      soils of Walker Branch watershed, Tennessee, over an eleven-year period. Biogeochemistry
      5:275-293.

Johnson, M.G., P.W. Shaffer, D.L Stevens, K.W. Thornton, and R.S.  Turner. 1990. Predicting and
      Forecasting Surface Water Acidification: A Plan to Assess Data Aggregation Effects.  U.S.
      Environmental Protection Agency, Environ. Res. Lab., Corvallis, OR. 93 pp.

Johnson, R.A., and D.W. Wichern.  1982.  Applied Multivariate Statistical Analysis. Prentice-Hall,  Inc.,
      Englewood Cliffs, NJ.  594 pp.

Kanciruk, P., J.M. Eilers, R.A. McCord, D.H. Landers, D.F. Brakke, and R.A. Linthurst. 1986.
      Characteristics of Lakes in the Eastern United States, Volume  III: Data Compendium of Site
      Characteristics and Chemical Variables. EPA/600/4-86/007c.  U.S. Environmental Protection
      Agency, Washington, DC. 439 pp.

Kaufmann, P.R., AT. Herlihy, J.W. Elwood, M.E. Mitch, W.S. Overton, M.J. Sale, J.J. Messer, K.A.
      Cougar, D.V. Peck, K.H. Reckhow, A.J. Kinney, S.J. Christie, D.D. Brown, C.A. Hagley, and H.I.
      Jager. 1988. Chemical Characteristics of Streams in the Mid-Atlantic and  Southeastern United
      States, Volume I: Population Descriptions and Physico-Chemical Relationships.
      EPA/600/3-88/021 a. U.S. Environmental Protection Agency, Washington, DC. 397 pp.

Keller, W., and J.R. Pitblado. 1986. Water quality changes in Sudbury area lakes: A  comparison of
      synoptic surveys in 1974-1976 and 1981-1983. Water Air Soil  Pollut. 29:285-296.

Kelly, C.A., J.W.M. Rudd, R.H. Hesslein, D.W. Schindler, P.J. Dillon,  C.T. Driscoll, S.A. Gherini, and R.E.
      Hecky. 1987.  Prediction of biological acid neutralization in acid-sensitive lakes. Biogeochemistry
      3:129-140.

Kern, J., ManTech Environmental Technology, Inc.,  Corvallis,  Oregon, personal communication.
                                              343
-------
Kern, J.S., and J.J. Lee. 1990. DDRP: Field Operations and Quality Assurance Report for Soil Sampling
      in the Mid-Appalachian Region of the US. EPA/600/3-90/045. US Environmental Protection
     Agency, Washington, DC. 56 pp

Kern, J.S., D.A. Lammers, D.L Cassell, and W.G. Campbell. 1990. DDRP: Field Operations and Quality
     Assurance Report for Watershed Mapping in the Mid-Appalachian Region of the US. EPA/600/3-
     90/036. U.S. Environmental Protection Agency, Washington, DC. 169 pp.

Khanna, P.K., and F. Beese. 1978. The behavior of sulfate salt input in podzolic  brown earth. Soil Sci.
      125:16-22.

Khanna, P.K., J. Prenzel, K.J. Meiwes, B. Ulrich, and E. Matzner. 1987. Dynamics of sulfate retention by
     acid forest soils in an acidic deposition environment. Soil Sci. Soc. Am. J. 51:446-452.

Kittrick, J.A. 1977.  Mineral equilibria and the soil  system, pp. 1-26. In: J.B. Dixon  and S.B. Weed, eds.,
     Minerals in Soil Environments. Soil Science Society of America, Madison, Wl.

Krug, E.G. 1989. Assessment of the theory and hypotheses of the acidification of watersheds. Illinois
     State Water Survey Division, SWS Contract Report 457. 25? pp.

Krug, E.C.,  and C.R. Frink. 1983. Acid rain on acid soil: A new perspective. Science 221:520-525.

Krug, E.C.,  and P.J. Isaacson. 1984. Comparison of water and dilute  acid treatment on organic and
     inorganic chemistry of leachate from organic-rich horizons of an acid forest soil. Soil Sci.
     137:370-378.

Krug, E.C.,  P.J. Isaacson, and C.R. Frink.  1985. Appraisal of some current hypotheses describing
     acidification of watersheds. J. Air Pollut. Contr. Assoc. 35:109-114.

Krug, W.R., W.A. Gebert, D.J. Graczyk, D.L. Stevens, B.P. Rochelle, and M.R. Church. 1990. Map of
     Mean Annual Runoff for the Northeastern, Southeastern, and Mid-Atlantic  United States, Water
     Years 1951-80. U.S. Geological Survey Water Resour. Invest. Report No. 88-4094, Madison, Wl.
     11pp.

Kulp, L 1987. Interim assessment, the causes and effects of acidic deposition.  Volume I:  Executive
     Summary.  National Acid Precipitation Assessment Program, Washington, DC.

Lammers, DA, D.L Stevens, and M.G. Johnson. 1987. Soil mapping  data management system. In:
     Agronomy Abstracts, American Society of Agronomy, Atlanta, November 29 - December 4, 1987.

Lammers, D.A., J.J. Lee, and J.A. Ferwerda. 1988. Mapping Protocols for Direct/Delayed Response
     Project Soil Mapping Activities in the Mid-Appalachian Region. U.S. Environmental Protection
     Agency, Corvallis, OR.  39 pp.
                                             344
-------
Landers, D.H., J.M. Eilers, D.F. Brakke, W.S. Overton, P.E. Kellar, M.E. Silverstein, R.D. Schonbrod, R.E.
      Crowe, R.A. Linthurst, J.M. Omernik, S.A. league, and E.P. Meier. 1987. Characteristics of Lakes in
      the Western United States, Volume I: Population Descriptions and Physico-Chemical Relationships.
      EPA/600/3-86/054a. U.S. Environmental  Protection Agency, Washington, DC. 176 pp.

Lasaga, A.C.  1981. Rate laws of chemical  reactions, pp. 1-68. In: A.C. Lasaga and R.J. Kirkpatrick, eds.,
      Kinetics of Geochemical  Processes. Reviews in Mineralogy Series, Volume 8.  Mineralogical
      Society of America, Washington, D.C.

Lau, W.M., and S.J. Mainwaring.  1985. The determination of soil sensitivity to acid deposition. Water, Air,
      Soil Pollut. 25:451-464.

LaZerte, B.D., and P.J. Dillon. 1984. Relative importance of anthropogenic versus natural sources of
      acidity in lakes and streams of central Ontario. Can. J. Fish. Aquat. Sci. 41:1664-1677.

Lee, J.J., and D.E. Weber. 1983.  Effects of sulfuric acid rain on decomposition rate and chemical
      element content of hardwood leaf litter. Can. J. Bot. 61:872-879.

Lee, J.J., D.A. Lammers, M.G. Johnson, M.R. Church, D.L Stevens, D.S. Coffey, R.S. Turner, LJ. Blume,
      LH. Liegel, and G.R. Holdren. 1989a. Watershed surveys to support an assessment of the regional
      effect of acidic deposition on surface water chemistry. Environ. Manage. 13:95-108.

Lee, J.J., D.A. Lammers, D.L. Stevens, K.W. Thornton, and K.A. Wheeler. 1989b. Classifying soils for
      acidic deposition aquatic effects: A  scheme for the northeastern U.S. Soil Sci. Soc. Am. J.
      53:1153-1163.

Lee, J.J., D.L. Stevens, J.S. Kern, and D.A. Lammers. 1990. Direct/Delayed Response Project: Mid-
      Appalachian Soil Correlation and Sampling Class Development Workshop. EPA/600/3-90/037. US
      Environmental Protection Agency, Washington, DC. 43 pp.

Lee, S. 1987. Uncertainty Analysis for Long-Term Acidification of Lakes in Northeastern USA. Ph.D.
      Thesis. University of Iowa, Iowa City.

Lee, S., and  J.L. Schnoor. 1988. Reactions that modify chemistry in lakes of the NSWS. Environ. Sci.
      Technol. 22:190-197.

Lepisto, A.,  P.G. Whitehead, C. Neal, and B.J. Cosby.  1988.  Modeling effects of acid deposition:
      Estimation of long-term water quality responses in forested catchments in Finland. Nordic
      Hydrology.  19:99-120.

Liegel,  L.H.,  D. Cassell, R. Church, D. Stevens, and P. Shaffer. 1991.  Regional characteristics of land
      use in the Northeast and Southern  Blue Ridge Province: Associations with acid rain effects on
      surface-water chemistry. Environ. Manage. 15:269-279.
                                               345
-------
 Likens, G.E., F.H. Bormann, R.S. Pierce, J.S. Eaton, and N.M. Johnson. 1977. Biogeochemistry of a
      Forested Ecosystem. Springer-Verlag, Inc., New York, NY. 146 pp.

 Llndberg, S.E., G.M. Lovett, D.D. Richter, and D.W. Johnson. 1986. Atmospheric deposition and canopy
      interactions of major ions in a forest. Science 231:141-145.

 Llnthurst, R.A., D.H. Landers, J.M. Eilers, D.F. Brakke, W.S. Overton, E.P.  Meier, and R.E. Crowe. 1986.
      Characteristics of Lakes in the Eastern United States, Volume I: Population  Descriptions and
      Physico-Chemical Relationships. EPA/600/4-86/007a. U.S. Environmental Protection Agency,
      Washington, DC: 136 pp.

 Lynch, D.D., and N.B. Dise. 1985. Sensitivity of Stream Basins in Shenandoah National Park to Acid
      Deposition. USGS Water Resources Investigation Report 85-4115. U.S. Geological Survey,
      Richmond, VA. 61 pp.

 Mast, M.A., and J.I. Drever. 1987. The effect of oxalate on the dissolution  rate of oligoclase and tremolite.
      Geochim. Cosmochim. Acta 51:2559-2568.

 Matalas,  N.C., and B. Jacobs. 1964. A Correlation Procedure for Augmenting Hydrology Data. U.S.
      Geological Survey Professional Paper 434, Washington, DC. 7 pp.

 McColl, J.G., and A.A. Pohlman.  1986.  Soluble organic acids and their chelating influence on Al and
      other metal dissolution from forest soils.  Water Air Soil Pollut.  31:917-927.

 Messer, J.J., C.W. Ariss, J.R.  Baker, S.K. Drous6, K.N. Eshleman, P.R. Kaufmann,  R.A. Linthurst, J.M.
      Omernik, W.S. Overton, M.J. Sale, R.D. Schonbrod, S.M. Stambaugh,  and J.R. Tuschall, Jr. 1986.
      National Stream Survey Phase I,  Pilot Survey. EPA/600/4-86/026. U.S. Environmental Protection
      Agency, Washington, DC. 179 pp.

 Mitchell,  M.J., M.B. David, D.G. Maynard, and S.A. Telang. 1986. Sulfur constituents in soils and streams
      of a watershed in the Rocky Mountains of Alberta. Can. J. For. Res. 16:315-320.

 Mohnen,  V.A. 1988. The challenge of acid rain. Scientific American 259:30-38.

 Morgan,  M.D. 1990. Streams in the New Jersey Pinelands directly reflect changes in atmospheric
      deposition chemistry. J. Environ. Qual. 19:296-302.

 Mortenson, D.C. 1989a. Geographic Information System Documentation of Watershed Data for
      Direct/Delayed Response Project - Northeast Database. EPA/600/3-89/001 and Project Summary.
      U.S. Environmental Protection Agency, Corvallis, OR. 35 pp.

Mortenson, D.C. 1989b. Geographic Information System Documentation of Watershed Data for
      Direct/Delayed Response Project - Southern Blue Ridge Province Database. EPA/600/3-89/002
      and Project Summary. U.S. Environmental Protection Agency, Corvallis,  OR. 25 pp.

                                             346
-------
Munson, R., Tetra-Tech, Inc., Hadley, Massachusetts, personal communication.

Musgrove, T.J., P.G. Whitehead, B.J. Cosby, G.M. Hornberger, and R.F. Wright. 1990. Regional modeling
     catchments in the Galloway region in southwest Scotland. In: J. Kamari, ed. Impact Models to
     Assess Regional Acidification. Kluwer Academic Publishers, Amsterdam.

NAS (National Academy of Sciences). 1984. Acid deposition: Processes of lake acidification. Summary of
     a Discussion. National Research Council Commission on Physical Sciences, Mathematics, and
     Resources. Environmental Studies Board, Panel on Processes of Lake Acidification. National
     Academy Press, Washington, DC. 1 1  pp.

NAS (National Academy of Sciences). 1986. Acid Deposition: Long Term Trends.  National Research
     Council Commission on Physical Sciences, Mathematics, and Resources. Environmental Studies
     Board. National Academy Press. Washington, D.C.

Neal, C., P.G. Whitehead, R. Neale, and B.J. Cosby. 1986. Modeling the effect of  acid deposition and
     conifer afforestation on stream acidity in the British uplands. J. Hydrol. 86:15-26.

Neary,  B.P., and  P.J. Dillon. 1988. Effects of sulphur deposition on lake-water chemistry in Ontario,
     Canada. Nature 333:340-343.

Neller,  J.R. 1959. Extractable sulfate - sulfur in soils of Florida in relation to amount of clay in the profile.
     Soil Sci. Soc. Am. J. 23:346-348.

Nevell, N., and M. Wainwright. 1986. Microbial growth on polluted leaf and litter leachates: Effects on
     leachate pH and sulphate concentration. Plant and Soil. 92:303-308.

Nikolaidis, N.P., H. Rajaram, J.L Schnoor, and K.P. Georgakakos. 1988. A generalized soft water
     acidification model. Water Resour. Res. 24:1983-1996.

Nodvin, S.C., C.T. Driscoll, and G.E. Likens. 1986. The effect of pH on sulfate adsorption by a forest soil.
     Soil Sci. 142:69-75.
Nordstrom, O.K. 1982. The effects of sulfate on aluminum concentrations in natural waters: Some
      stability relations in the system AI^  - SQj - h^O at 298 K. Geochim. Cosmochim. Acta
      46:681-692.
Norton, S.A., R.F. Wright, J.S. Kahl, and J.P. Scofield. The MAGIC simulation of surface water
      acidification at, and first results from, the Bear Brook Watershed Manipulation, Maine, USA.
      Environ. Poll. In press.

NRC (National Research Council). 1988. Estimating probabilities of extreme floods. Methods and
      Recommended Research. National Academy Press. Washington, DC.
                                              347
-------
Nyborg, M. 1978. Sulfur pollution and soil, pp. 359-390. In: J.O. Nriagu, ed. Sulfur in the Environment,
      Part II: Ecological Impacts. Wiley-lnterscience, New York, NY.

Nye, P.H., and D.J. Greenland. 1960. The Soil Under Shifting Cultivation. Commonwealth Bureau of Soils
      Tech. Comm. No. 51. Commonwealth Agricultural Bureaux, Farnham Royal, Bucks.

Olsen, A., U.S. EPA, Corvallis, Oregon, personal communication.

Olsen, A.R., E.G. Voldner, D.S. Bigelow, W.H. Chan, T.L. Clark, M.A. Lusis, P.K. Misra, and R.J. Vet.
      1990.  Unified wet deposition data summaries for North America: Data summary procedures and
      results for 1980-1986. Atmos. Environ.  24A:661-672.

Overton, W.S.  1987.   A Sampling and Analysis Plan for Streams in the National Surface Water Survey.
      Technical Report 117.  Department of Statistics, Oregon State University, Corvallis, Oregon, 97331.
      50 pp.

Paces, T. 1973. Steady-state kinetics and equilibrium between ground water and granitic rock. Geochim.
      Cosmochim. Acta 37:2641-2663.

Papp, M.L, and R.D. Van Remortel. 1990. Direct/Delayed Response Project: Preparation Laboratory
      Operations and Quality Assurance Report for the Mid-Appalachian Soil Survey. EPA/600. US
      Environmental Protection Agency, Washington, DC.  226 pp.

Papp, M.L, R.D. van Remortel, CJ. Palmer, G.E. Byers, B.A. Schumacher, R.L Slagle, J.V. Burton, K.C.
      Shines, J.E. Teberg and M.J. Miah. 1989. Direct/Delayed Response Project: Quality Assurance
      Plan for Preparation and Analysis of Soils from the Mid-Appalachian Region of the United States.
      EPA/600/4-89/031. U.S. Environmental Protection Agency, Washington, D.C. 226 pp.

Parfitt, R.L, and R.S.C. Smart. 1978. The mechanism of sulfate adsorption on iron oxides. Soil Sci. Soc.
      Am. Proa 42:48-50.

Peters, N.E., and C.T. Driscoll. 1987. Hydrogeologic controls of surface-water chemistry in the
      Adirondack Region of New York State. Biogeochemistry 3:163-180.

Rajan, S.S.S. 1978. Sulfate adsorbed on hydrous alumina, ligands displaced, and changes in surface
      charge. Soil Sci. Soc. Am. J. 42:39-44.

Rajan, S.S.S. 1979. Adsorption and desorption of sulfate and charge relationships in allophanic clays.
      Soil Sci.  Soc. Am. J. 43:65-69.

Reuss, J.O. 1983. Implications of the calcium-aluminum exchange system for the effect of acid
      precipitation in soils. J. Environ. Qual. 14:26-31.
                                             348
-------
Reuss, J.O., and D.W. Johnson. 1985. Effect of soil processes on the acidification of water by acid
      deposition. J. Environ. Qual. 14:26-31.

Reuss, J.O., and D.W. Johnson. 1986. Acid Deposition and the Acidification of Soils and Waters.
      Ecological Studies Volume 59. Springer-Verlag, Inc., New York, NY.

Richter, D.D., P.J. Comer, K.S. King, H.S. Sawin, and D.S. Wright. 1988. Effects of low ionic strength
      solutions on pH of acid forested soils. Soil Sci. Soc. Am. J. 52:261-264.

Rochelle, B.R., and M.R. Church. 1987. Regional patterns of sulfur retention in watersheds of the eastern
      U.S. Water Air Soil Pollut. 36:61-73.

Rochelle, B.P., M.R. Church, and M.B. David. 1987. Sulfur retention  at intensively studied sites in the U.S.
      and Canada. Water Air Soil Pollut. 33:73-84.

Rochelle, B.P., C.I. Lift, W.G. Campbell, D.L Cassell, M.R. Church, and R.A. Nusz. 1989a. Relating map
      geomorphic/hydrologic parameters to surface water chemistry regionally: Relative to acidic
      deposition. J. Hydro!.  112:103-120.

Rochelle, B.P., D.L. Stevens, and M.R. Church. 1989b. Uncertainty analysis of runoff estimates from a
      runoff contour map. Water Resour. Bull. 25:491 -498.

Rosenbrock, H.H. 1960. An automatic method for finding the greatest or least value of a function.
      Comput. J. 3:175-184.

Rudd, J.W.M., C.A. Kelly,  and A. Furutani. 1986a. The role of sulfate reduction  in long term accumulation
      of organic and  inorganic sulfur in lake sediments. Limnol. Oceanogr. 36:1281-1291.

Rudd, J.W.M., C.A. Kelly,  V. St.  Louis, R.H.  Hesslein, A. Furutani, and M.H. Holoka.  1986b. Microbial
      consumption of nitric and sulfuric acids in north temperate-lakes. Limnol. Oceanogr. 31:1267-1280.

Ryan, P.F., G.M. Hornberger, B.J. Cosby, J.N Galloway, J.R. Webb, and E.B. Rastetter.  1989.  Changes
      in the chemical composition of stream water in two catchments in the Shenandoah National Park,
      Virginia, in response to atmospheric deposition of sulfur. Wat. Resour. Res.  25:2091 -2099.

Sale,  M.J.,  P.R. Kaufmann, H.I. Jager, J.M.  Coe,  K.A. Cougan, A.J. Kinney, M.E. Mitch, and W.S.
      Overton. 1988.  Chemical Characteristics of Streams in the Mid-Atlantic and Southeastern United
      States, Volume II: Streams Sampled,  Descriptive Statistics, and Compendium of Physical and
      Chemical Data. EPA/600/3-88/021 b. U.S.  Environmental Protection Agency, Washington, DC. 597
      pp.

Sanders, F.E., and P.B.H. Tinker. 1975. Adsorption of sulphate by a sandy loam soil (calcic cambisol).
      Geoderma 13:317-324.
                                              349
-------
SAS Institute, Inc. 1987. SAS/STAT Guide for Personal Computers, Version 6 Edition. SAS Institute, Inc.,
      Gary, NC. 1028 pp.

SAS Institute, Inc. 1988. SAS/STAT User's Guide, Release to 6.03 Edition. SAS Institute, Inc., Gary, NC.

Schindler, D.W. 1986. The significance of in-lake production of alkalinity. Water Air Soil  Pollut.
      30:931-946.

Schindler, D.W. 1988. The effects of acid rain on freshwater ecosystems. Science 239:149-157.

Schindler, S.C., M.J. Mitchell, T.J. Scott, R.D. Fuller, and C.T. Driscoll. 1986a. Incorporation of 35S-sulfate
      into inorganic and organic constituents of two forest soils. Soil Sci. Soc. Am. J. 40:457-462.

Schindler, D.W., M.A. Turner, M.P. Stainton, and G.A. Linsey. 1986b.  Natural sources of acid neutralizing
      capacity fn low-alkalinity lakes of the Precambrian Shield. Science 232:844-847.

Schnoor, J.L, 8. Lee, and N.P. Nikolaidis. 1986. Lake resources at risk to acidic deposition in the
      eastern United States. Water Air Soil Pollut. 31:1091-1101.

Schofield, C.L 1976. Acidification of Adirondack Lakes by Atmospheric Precipitation: Extent and
      Magnitude of the Problem. Final report, Project F-28-R. New York State Dept. Environ. Conserv.,
      Albany, NY.

Sellkop, S., AnaJytical Sciences, Inc., Durham, North Carolina, personal communication.

Selp, H.M. 1980. Acidification of freshwaters  - sources and mechanisms, pp. 358-366. In: D. Drabl0s and
      A. Tollan, eds. Ecological Impact of Acid Precipitation: Proceedings of an International Conference,
      Sandefjord, March 11-14. SNSF Project, Oslo-As, Norway.

Shaffer, P.W. 1989.  Organic blocking of sulfate adsorption by a Virginia forest soil. Agronomy Abstracts.
      p. 311.

Shaffer, P.W., and M.R. Church. 1989. Terrestrial and in-lake contributions to the alkalinity budgets of
      drainage lakes: An assessment of regional differences. Can. J.  Fish. Aquat. Sci. 46:509-515.

Shaffer, P.W., and D.L Stevens, Jr.  1991. Procedures for aggregation of nonlinear data: an assessment
      using soil sulfate adsorption isotherms. Soil Science Society of America Journal.  55:684-692.

Shaffer, P.W., R.P. Hooper, K.N. Eshleman, and M.R. Church. 1988. Watershed vs. in-lake alkalinity
      generation: A comparison of rates using input-output studies. Water, Air, Soil Pollut. 39:263-273.

Simons, T.J., and D.C.L Lam. 1980. Some limitations of water quality models for large lakes: A case
      study of lake Ontario. Water Resour. Res. 16:105-116.
                                              350
-------
Smith, R.A., and R.B. Alexander. 1983. Evidence for Acid-Precipitation Induced Trends in Stream
     Chemistry at Hydrologic Bench-Mark Stations. Geological Survey Circular 910. U.S. Geological
     Survey, Alexandria Virginia, 12 pp.

Soil Survey Staff. 1951. Soil Survey Manual. USDA Soil Conservation Service, Agricultural Handbook No.
     18. U.S. Gov. Print. Office, Washington, DC.

Soil Survey Staff. 1975. Soil Taxonomy: A Basic System of Soil Classification for Making and Interpreting
     Soil Surveys. USDA Soil Conservation Service, Agricultural Handbook No. 436. U.S. Gov. Print.
     Office, Washington, DC. 754 pp.

Soil Survey Staff. 1983. National Soils Handbook.  USDA Soil Conservation Service, Agricultural
     Handbook No. 430. U.S. Gov. Print. Office, Washington, DC.

Soil Survey Staff. 1984. Procedures for Collecting  Soil  Samples and Methods of Analysis for Soil Survey.
     USDA Soil Conservation Service, SSIR 1. U.S. Gov. Print. Office, Washington, DC. 68 pp.

Solomon, K.P., and T.E. Cerling. 1987. The annual carbon dioxide cycle in a montane soil: Observations,
     modeling, and  implications for weathering. Water Resour. Res. 25:2257-2265.

Stensland, G., Illinois State Water Survey, Champaign, Illinois, personal communication.

Stevenson, F.J. 1982. Humus Chemistry: Genesis, Composition, Reactions. Wiley-lnterscience, New
     York. NY. 443 pp.

Strickland, T., ManTech Environmental Technology, Inc., Corvallis, Oregon, personal communication.

Strickland, T.C., and J.W. Fitzgerald. 1984. Formation and mineralization of organic  sulfur in forest soils.
     Biogeochemistry 1:79-95.

Strickland, T.C., J.W. Fitzgerald, J.T. Ash, and  W.T. Swank.  1987. Organic sulfur transformations and
     sulfur pool sizes in  soil and litter from a  Southern Appalachian hardwood forest. Soil Sci.
      143:453-458.

Stuanes, A.O. 1984. A simple extraction as an  indicator of a soil's sensitivity to acid precipitation. Acta
     Agric. Scand. 34:113-127.

Sullivan, T.J., J.M.  Eilers,  M.R. Church, D.J. Blick,  K.N. Eshleman, D.H. Landers, and M.S. DeHaan. 1988.
     Atmospheric wet sulphate deposition and lakewater chemistry. Nature 331:607-609.

Sullivan, T.J., J.A. Bernert, E.A. Jenne, J.M. Eilers, B.J. Cosby, D.F. Charles, and A.R. Selle. 1991.
      Comparison of MAGIC and DIATOM Paleolimnological Model Hindcasts of Lakewater Acidification
      in the Adirondack Region of New York. PNL-7487, Pacific Northwest Laboratory, Richland,
     Washington.

                                              351
-------
Swank, W.T., and D.A. Crossley, Jr., eds. 1988. Ecological Studies, Volume 66: Forest Hydrology and
      Ecology at Coweeta. Springer-Verlag, New York, NY. 469 pp.

Swank, W.T., and J.B. Waide. 1988. Characterization of baseline precipitation and stream chemistry and
      nutrient budgets for control watersheds, pp. 57-79. In: W.T.  Swank and D.A. Crossley, Jr., eds.
      Ecological Studies, Volume 66: Forest Hydrology and Ecology at Coweeta. Springer-Verlag, New
      York, NY.

Swank, W.T., J.W. Fitzgerald, and J.T. Ash. 1984. Microbial transformations of sulfate in forest soils.
      Science 223:182-184.

Tamm, C.O., and L Hallbacken. 1988. Changes in soil acidity in two forest areas with different acid
      deposition: 1920s to 1980s. Ambio 17:56-61.

Tang, A.J.S., W.H. Chan,  D.H.S. Chung, and M.A. Lusis. 1986. The Acid Precipitation in Ontario Study:
      Precipitation and Air Concentration and Wet and Dry Deposition Fields of Pollutants in Ontario,
      1983. Report #ARB-008-86-AQM, Ontario Ministry of the Environment.

Thornton, K., D. Marmorek, P. Ryan, K. Heltcher, and D. Robinson. 1990. Methods for Projecting Future
      Changes in Surface Water Acid-base Chemistry. SOS/T Report 14. National Acid Precipitation
      Assessment Program

Troedsson, T. 1980. Long-term changes of forest soils. Ann. Agr. Fenniae.  19:81-84.

Turner, R.S., R.B. Cook, H. Van Miegroet, D.W. Johnson, J.W. Elwood, O.P. Bricker, S.E. Lindberg, and
      G.M. Hornberger.  1990. Watershed and Lake Processes Affecting Chronic Surface Water Acid-
      Base Chemistry, NAPAP SOS/T Report 10, In: Acidic Deposition: State of Science and"
      Technology. National Acid Precipitation Assessment Program, Volume II, Washington, DC.

Ulrich, B. 1983. Soil acidity and its relation to acid deposition. In: B. Ulrich, and J. Pankrath, eds. Effects
      of Accumulation of Air Pollutants in Forest  Ecosystems. D. Reidel Publ. Co., Boston, MA.

U.S. Environmental Protection Agency, Office of Policy, Planning and Evaluation, Washington, D.C.,
      personal communication.

U.S. Environmental Protection Agency. 1984.  National Surface Water Survey: National Stream Survey;
      Draft 1985 Research Plan. Environmental Research Laboratory, Corvallis, Oregon.  134 pp.

USEPA (U.S. Environmental Protection Agency). 1988. Statements of Work for Soil Analysis in the
      Direct/Delayed Response Project Mid-Appalachian Soil Survey. Attachment A in Invitations for Bid,
      Solicitations W802498D1 and W802621D1.  U.S. Environmental Protection Agency, Las Vegas, NV.
      313 pp. and 36 pp.
                                              352
-------
U.S. Geological Survey. 1969. Water Resources Data, Virginia, Water Year 1968. U.S. Geological Survey
     Water Report VA-68-1.

U.S. Geological Survey. 1970. Water Resources Data, Virginia, Water Year 1969. U.S. Geological Survey
     Water Report VA-69-1.

VanLoon, G.W.  1984. Acid rain and soil. Can. J. Physiol. Pharmacol. 62:991-997.

Vanselow, A.P. 1932. Equilibria of the base-exchange reactions of bentonites, permutites, soil colloids,
     and zeolites. Soil sci. 33:95-113.

Velbel, M.A. 1985. Geochemical mass balances and weathering rates in forested watersheds of the
     Southern Blue Ridge. Am.J. Sci. 285:904-930.

Velbel, M.A. 1986a. The mathematical  basis for determining rates of geochemical and geomorphic
      processes in small forested watersheds by mass balance: Examples and implications, pp. 439-452.
      In: S.M. Colman and D.P. Dethier, eds. Rates of Chemical Weathering of Rocks and Minerals.
     Academic Press, Orlando, FL.

Velbel, M.A. 1986b. Influence of surface area, surface characteristics, and solution composition of
     feldspar weathering rates,  pp. 615-634. In: J.A.  Davis and K.F. Hayes, eds. Geochemical Processes
     at Mineral Surfaces. American Chemical Society Symposium Series No. 323, Washington, DC.

Vong, R., S. Cline, G. Reams, J.  Bernert, D. Charles, J. Gibson, T. Haas, J. Moore, R. Husar, A. Olsen, J.
      Simpson, and S. Seilkop. 1989.  Regional Analysis of Wet Deposition for Effects Research.
      EPA/600/3-89/030. U.S. Environmental Protection Agency, Environmental Research Laboratory,
      Corvallis, OR. 57 pp.

Waide, J., FTN  & Associates, Little Rock, Arkansas, personal communication.

Waide, J.B., and W.T. Swank. 1987. Patterns and Tends in Precipitation and Stream Chemistry at the
      Coweeta Hydrologic Laboratory. Aquatic Effects Task Group VI Peer Review. Volume II. May 17-
      23, New Orleans, LA.

Wampler, S.J., and A.R. Olsen. 1987. Spatial estimation of annual wet acid deposition using
      supplemental  precipitation data, pp. 248-251. In: Tenth Conference on Probability and Statistics,
      Oct. 4-5,  Edmonton, Atla., Canada. American Meteorological Society, Boston, MA.

Weaver, G.T., P.K. Khanna, and F. Beese.  1985. Retention and transport of sulfate in slightly acid forest
      soil.  Soil Sci. Soc. Am. J. 49:746-750.

Webb, J.R., B.J. Cosby, J.N. Galloway, and G.M. Hornberger. 1989. Acidification of native brook trout
      streams in Virginia. Wat. Resour. Res.  25:1367-1377.
                                              353
-------
Welder, R.K., and G.E. Lang. 1988. Cycling of inorganic and organic sulfur in peat from Big Run Bog,
      West Virginia. Biogeochemistry 5:221-242.

Whitehead, P.G., S. Bird, M. Hornung, J. Cosby, C. Neal, and P. Paricos. 1988a. Stream acidification
      trends in the Welsh uplands - A modelling study of the Uyn Brianne catchments.  J. Hydrol.
      101:191-212.

Whitehead, P.G., B. Reynolds, M. Hornung, C. Neal, J. Cosby, and P. Paricos.  1988b. Modelling long
      term stream acidification trends in Upland Wales at Plynlimon. Hydrol. Proc. 2:357-368.

Wigington, P.J., Jr., T.D. Davies, M. Tranter, and K.N. Eshleman. 1990. Episodic Acidification of Surface
      Waters Due to Acidic Deposition.  NAPAP SOS/T Report 12. In: Acidic Deposition: State of
      Science and Technology. National Acid Precipitation Assessment Program, Volume II, Washington,
      DC.

Wollast, R. 1967. Kinetics of the alteration of K-feldspar in buffered solutions at low temperature.
      Geochim. Cosmochim. Acta 31:635-648.

Wright, R.F., B.J. Cosby,  G.M. Hornberger, and J.N. Galloway.  1986. Comparison of paleolimnological
      with MAGIC model reconstructions of water acidification. Water Air Soil Pollut. 30:367-380.

Wright, R.F., B.J. Cosby,  M.B. Platen, and J.O. Reuss. 1990.  Evaluation of an acidification model with
      data from manipulated catchments in Norway. Nature. 343:53-55.

Wright, R.F., B.J. Cosby, and G.M. Hornberger. 1991. A regional model of lake acidification in
      southernmost Norway. Ambio. 20:222-225.

Wright, R.F., E. Lotse, and A. Semb. 1988. Reversibility of acidification shown by whole-catchment
      experiments. Nature 334:670-675.
                                              354
-------
                                       SECTION 13
                                       GLOSSARY
13.1 ABBREVIATIONS AND SYMBOLS
13.1.1 Abbreviations
ADS
AERP
ANC
APIOS

AREAL-RTP
BMD
BMK
END
BNS
BXP
BXW

CDF
CFP
Cl
CMS

DDRP
DOC
DQO

ELS-I
EMSL-LV
EPA
EPRI
ERL-C
ERP
ETD

FLW

GIS
Acid Deposition System
Aquatic Effects Research Program
Acid neutralizing capacity
Acid Precipitation in Ontario Study conducted by Ontario Ministry of the
Environment
USEPA Atmospheric Research and Exposure Assessment Laboratory -
Research Triangle Park

A DDRP soil sampling class in the M-APP Region
A DDRP soil sampling class in the M-APP Region
A DDRP soil sampling class in the M-APP Region
A DDRP soil sampling class in the M-APP Region
A DDRP soil sampling class in the M-APP Region
A DDRP soil sampling class in the M-APP Region

Cumulative distribution function
A DDRP soil sampling class in the M-APP Region
Confidence  interval
A DDRP soil sampling class in the M-APP Region

Direct/Delayed Response Project
Dissolved organic carbon
Data quality objective

Eastern Lake Survey-Phase I
USEPA Environmental Monitoring and Systems Laboratory - Las Vegas
U.S. Environmental Protection Agency
Electric Power Research Institute .
USEPA Environmental Research Laboratory - Corvallis
Episodic Response Project
Enhanced Trickle Down Model

A DDRP soil sampling class in the M-APP Region

Geographic Information System

                      355
-------
HST
   A DDRP soil sampling class in the M-APP Region
ILWAS
IQR

LA!
LEVIS
LTA

MAGIC
MLRA
MQO

NAS
NADP/NTN
NAPAP
NCDC
NE
NOAA
NRC
NSS-I
NSWS

ORNL

PE
PNL
-  Integrated Lake/Watershed Acidification Study
-  Interquartile range

—  Leaf area index
-  Laboratory Entry and Verification System
-  Long-term annual average deposition

-  Model for Acidification of Groundwater in Catchments
—  Major land resource areas
-  Measurement quality objective

—  National Academy of Sciences
-  National Acid Deposition Program/National Trends Network
—  National Acid Precipitation Assessment Program
-  National Climatic Data Center
—  Northeast Region
-  National Oceanographic and Atmospheric Administration
-  National Research Council
—  National Stream Survey-Phase I
-  National Surface Water Survey

-  Oak Ridge National Laboratory, Tennessee

-  Performance evaluation
-  Battelle-Pacific Northwest Laboratories
QA
QC

RADM
RELMAP
RCC
RMSE
RSD
   Quality assurance
   Quality control

   Regional Acid Deposition Model
   Regional Lagrangian Model of Air Pollution
   Regional Coordinator/Correlator
   Root mean square error
   Relative standard deviation
SAS
SBRP
SCS
SE
SOBC
   Statistical Analysis System
   Southern Blue Ridge Province
   Soil Conservation Service
   Standard error
   Sum of base cations
                                            356
-------
TMY
TPD
TWO
TWM
TXP
TXW
TY

UMW
USDA
USFS
USGS
UTM
Typical meteorological year
A DDRP soil sampling class in the M-APP Region
A DDRP soil sampling class in the M-APP Region
A DDRP soil sampling class in the M-APP Region
A DDRP soil sampling class in the M-APP Region
A DDRP soil sampling class in the M-APP Region
Typical year

-Upper Midwest
U.S. Department of Agriculture
U.S. Forest Service
U.S. Geological Survey
Universal Transverse Mercator
13.1.2 Symbols
2As
2Bn
2Cn
2X
3A
3B

AC_BACL
AI3+
AL_AO
AL_CD
AL_CL
AL_CL2
AL_PYP
AW

C_TOT
Ca2+
CaCI2
CA_CL2
CAJDAC
CEC_CL
CECJDAC
cr
CLAY
C02
CON
Southern Blue Ridge subregion (NSS Pilot Survey)
Valley and Ridge subregion (NSS Pilot Survey)
Northern Appalachians subregion (NSS Pilot Survey)
Southern Appalachians subregion (NSS Pilot Survey)
Piedmont subregion (NSS Pilot Survey)
Mid-Atlantic Plain subregion (NSS Pilot Survey)

see Table 5-7
aluminum ion
see Table 5-7
see Table 5-7
see Table 5-7
see Table 5-7
see Table 5-7
total watershed area, in kilometers squared

carbon total
calcium ion
calcium chloride
see Table 5-7
see Table 5-7
see Table 5-7
see Table 5-7
chloride ion
see Table 5-7
carbon dioxide
mapped percent coniferous coverage.
                                           357
-------
CONe
COS
COSI
effective percent coniferous coverage
see Table 5-7
see Table 5-7
FE_AO
FEJDD
FE_CL2
FE_PYP
FRAG
FS
FSI

H+
u-f
" total
H20
H2S04
ha
HC03-
H-DRY
H-WET

K
K+
K_CL
K_CL2
keq/ha
kg
km
KDAC
LTA-rbc
LTA-zbc
see Table 5-7
see Table 5-7
see Table 5-7
see Table 5-7
fragments > 2 mm diameter
see Table 5-7
see Table 5-7

hydrogen ion
total effective acidity (H+ + NH4+ -
water
sulfuric acid
hectare (2.47 acres or ten thousand square meters)
bicarbonate ion
annual hydrogen ion loading in dry deposition
annual hydrogen ion loading in wet deposition

hydraulic conductivity
potassium ion
see Table 5-7
see Table 5-7
kilooquivalent per hectare
kilogram
kilometer
see Table 5-7
sulfate mass transfer coefficient (m/yr)

long-term annual average, reduced dry base cation
long-term annual average, zero dry base cation
MA_MP_CM
MA_MP_UN
MA_SHEDS
MACMPNT
MAGIS01L
MATRAN
MATRANRC
/
-------
Mg2+
MG_CL
MG_CL2
MG_OAC
MOIST
MS
   -magnesium ion
   see Table 5-7
   see Table 5-7
   see Table 5-7
   see Table 5-7
   see Table 5-7
NJTOT
Na+
NA_CL
NA_CL2
NAOAC
NH
OH'

pC02
PH_002M
PH_01M
PH H20
-  see Table 5-7
-  sodium ion
—  see Table 5-7
-  see Table 5-7
-  see Table 5-7
   -ammonium ion
-  nitrate ion

-  hydroxide ion

   -partial pressure of carbon dioxide
-  see Table 5-7
-  see Table 5-7
—  see Table 5-7
R
r
R2

RTR

S_TOT
SAND
SBC_CI
sd
SiO2
SI_AO
SILT
SO4_0
SO4_2
SO4_4
SO4_8
SO4J6
S04_32
SO4_B2
SO4_EMX
S04 H2O
    runoff (length/time)
    correlation coefficient
    coefficient of determination, the proportion of variability explained by a
    regression model
    lake retention time, in years

    see Table 5-7
    see Table 5-7
    sum of base cations as measured in unbuffered 1N ammonium chloride
    dry sulfur deposition (mass/length2/time)
    silicon dioxide
    see Table 5-7
    see Table 5-7
    see Table 5-7
    see Table 5-7
    see Table 5-7
    see Table 5-7
    see Table 5-7
    see Table 5-7
    half saturation constant
    adsorption asymptote
    sulfate, water extractable
                                           359
-------
SO4_PO4
SO4_SLP
SO4_XIN
isotherms
S042'
—  sulfate, phosphate extractable
-  slope of sulfate adsorption isotherm at zero net adsorption
—  zero net adsorption concentration for sulfate, determined from adsorption

   -sulfate
-  steady state sulfate concentration
—  surface water sulfur (mass/length3)
-  wet sulfur deposition  (mass/length2/time)

-  hydraulic residence time, in years
-  hydrologic retention time, in years
VCOS
VFS
   see Table 5-7
   see Table 5-7
13.2 DEFINITIONS

ACCURACY - the difference between the approximate solution obtained using a numerical model and
the exact solution of the governing equations (or a known standard concentration), divided by the
exact solution (or known standard concentration).

ACID ANION - negatively charged ion that combines with hydrogen ion to form an acid.

ACID CATION - hydrogen ion or other metal that can hydrolyze water to produce hydrogen ions, e.g.
Al, Mn, Fe.

ACID CRYSTALLINE - in the Southern Blue Ridge Province, rocks or bedrock which, upon
weathering, form secondary phases including HIV clays.

ACID DEPOSITION SYSTEM (ADS) - a  national database of precipitation amount and chemistry
maintained at Battelle-Pacific Northwest Laboratories.

ACID MINE DRAINAGE - runoff with high concentration of metals, sulfate, and acidity resulting from
the OXIDATION of sulfide minerals that have been exposed to air and water (usually from mining
activities).

ACID NEUTRALIZING CAPACITY - the total acid-combining capacity of a water sample determined
by titration with a strong acid to a preselected equivalence point pH: an integrated measure of the
ability of an aqueous solution to neutralize strong acid inputs. Acid neutralizing capacity includes
strong bases (e.g., hydroxide) as well as weak bases (e.g., borates, carbonates, dissociated organic
acids, alumino-hydroxyl complexes).
                                            360
-------
ACIDIC DEPOSITION - rain, snow, or dry fallout containing high concentrations of sulfuric acid, nitric
acid, or hydrochloric acid, usually produced by atmospheric transformation of the by-products of fossil
fuel combustion (power plants, smelters, autos, etc.). Precipitation with a pH of less than 5.0 is
generally considered to be unnaturally acidic, i.e., altered by ANTHROPOGENIC activities.

ACIDIC EPISODE - an episode in a water body in which ACIDIFICATION of SURFACE WATER to an
ACID NEUTRALIZING CAPACITY less than or equal to 0 |ieq/L occurs.

ACIDIC LAKE OR STREAM - an  aquatic system with an ACID NEUTRALIZING CAPACITY less than or
equal to 0 ^eq/L.

ACIDIFICATION - any temporary or permanent loss of ACID NEUTRALIZING CAPACITY in water or
BASE SATURATION in soil by natural or ANTHROPOGENIC processes.

ACIDIFIED - a natural water that  has experienced any temporary or permanent loss of ACID
NEUTRALIZING CAPACITY or a soil that has experienced a reduction in BASE  SATURATION.

ACTIVITY COEFFICIENTS - empirically derived coefficients used to transform  concentration data to
salt or ion activities.

ADJUSTED R2 - the standard R2 of regression analysis,  modified to balance increasing the R2 against
increasing the number of explanatory variables.

AFFORESTATION - the natural process through which non-forested lands become forested.

AGGRADING FORESTS - forests in which there is a net annual accumulation  of biomass.

AGGREGATION - a method for statistically reducing a set of data to a single calculated or index value
for each parameter (e.g., a weighted average).

ALFISOLS - in Soil Taxonomy, the ORDER of mineral soils with an argillic horizon with at least 35
percent base saturation.

ALKALINITY - the titratable base of a sample containing hydroxide, carbonate, and bicarbonate ions,
i.e., the equivalents of acid  required to  neutralize the basic carbonate components.

ALKALINITY MAP CLASS - a geographic area defined by the expected ALKALINITY of SURFACE
WATERS (does not necessarily reflect measured alkalinity); used as a STRATIFICATION FACTOR in
ELS-I design.

ALUMINUM BUFFERING - a chemical process in which hydrogen ion activities are buffered by the
precipitation/dissolution of aluminum hydroxides.

ALUMINUM BUFFER RANGE - pH 4.2 - 2.8
                                           361
-------
ANAEROBIC - without free oxygen (e.g., hypolirnnetic lake waters, sediments, or poorly drained soils).

ANALYTE - a chemical species that is measured in a water soil, or tissue sample.

ANALYTICAL CHARACTERIZATION - physical and chemical properties of water soil, or samples
measured in the laboratory.

ANALYTICAL DUPLICATE - a QUALITY CONTROL sample made by splitting a sample.

ANION - a negatively charged ion.

ANION CATION BALANCE - a method of assessing whether all CATIONS and ANIONS have been
accounted for and measured accurately; in an electrically neutral solution, such as water, the total
charge of positive ions (cations) equals the total charge of negative ions (anions).

ANION EXCHANGE/ADSORPTION - a reversible process occurring in soil in which ANIONS are
adsorbed and released.

ANTHROPOGENIC - of, relating to, derived from, or caused by human activities or actions.

APPARENT SOLUBILITY PRODUCT - an approximate form of an equilibrium constant calculated using
solution concentration data instead of activities.

AQUEOUS SPECIES - any dissolved ionic or nonionic chemical entity.

AQUIC - a moisture regime of soils in which a water table and reducing conditions occur near the
surface.

AQUIFERS - below-ground stratum capable of producing water as from wells or springs.

AQUO LIGAND - a water molecule held to Fe or Al in a clay edge or hydrous oxide by ligand exchange.

ARC - represents line features and borders of area features.  One line  feature may be made up of many
arcs.  The arc is the line between two nodes.

ARC/INFO - a commercial geographic information system (GIS) software used to automate, manipulate,
analyze, and display geographic data in digital form.

ATTRIBUTE  - the class,  characteristics or other properties associated  with a specific feature, area on a
map, lake or stream.

AVAILABLE TRANSECT -  a transect identified to represent a map unit and listed for random selection.
                                           362
-------
BASE CATION - a nonprotolytic CATION that does not affect ACID NEUTRALIZING CAPACITY;
consists principally of calcium, magnesium, sodium, and potassium.

BASE CATION EXCHANGE - the process by which BASE CATIONS (Ca2+, Mg2+, Na+, K+) are
adsorbed or released from negatively charged sites on soil particles from or to, respectively, soil
solutions. Such exchange processes are instrumental in determining pH of soil solutions.

BASE CATION SUPPLY - the pool of BASE CATIONS (Ca2+, Mg2+, Na+, K+) in a soil available for
exchange with hydrogen ions (H+). The base cation pool is determined by the CATION EXCHANGE
CAPACITY of the soil and the percentage of exchange sites occupied by BASE CATIONS.

BASE SATURATION - the percentage of total soil CATION EXCHANGE CAPACITY that is occupied by
exchangeable cations other than hydrogen and aluminum, i.e., the base cations Ca2+, Mg2+, Na+,
and K+.

BEDROCK - solid rock exposed at the surface of the earth or overlain by unconsolidated material.

BEDROCK GEOLOGY - the physical and chemical nature and composition of solid rock at or near the
earth's surface.

BEDROCK LITHOLOGY - see L1THOLOGY.

BEDROCK SENSITIVITY SCORES - a six point scale, developed for DDRP, designed to distinguish
the relative reactivities of different lithologies.

BEDROCK UNITS - the smallest homogenous entity depicted on a bedrock map.

BIAS - a systematic error in a method caused by artifacts or idiosyncracy of the measurement system.

BIOMASS - the quantity of particulate organic matter in units of weight or mass.

BIOMASS ACCRETION - net accumulation of plant mass in a growing, or aggrading, ecosystem; also
refers  to net accumulation of an individual  nutrient associated with accumulation of biomass.

BLOOM-GRIGAL MODEL - a numerical model used to investigate the evolution of soil exchange
characteristics under various H+ ion deposition SCENARIOS. The code is based on  mass balance
consideration with empirical functions used to describe the pH-base saturation relationships.

BONFERRONI INEQUALITY - an inequality from probability theory that is used to carry out  multiple
simultaneous statistical comparisons.

BOXPLOT - a graph of data with a box drawn from the 25th percentile to the 75th percentile; lines
extending from the box as far as the data extend to a distance of at most 1.5 times the
INTERQUARTILE RANGE, and more extreme observations marked individually.
                                          363
-------
BUFFERING CAPACITY - the quantity of acid or base that can be added to a water sample with little
change in pH.

BULK DENSITY - the integrated density of a volume of soil, including solid matter, soil solutions,
voids, roots, etc.

Ca/AI EXCHANGE REACTION - the reaction describing the distribution of Ca and Al between the soil
exchange complex and the soil solution.

CALC1TE - a mineral with the formula CaCO3. A carbonate mineral.

CALIBRATION - process  of checking, adjusting, or standardizing operating characteristics of
instruments and model appurtenances on a physical model or coefficients in a mathematical model
with empirical data of known quality.  The process of evaluating the scale readings of an instrument
with a known standard in  terms of the physical quantity to be  measured.

CALIBRATION BLANKS - a zero-concentration QUALITY CONTROL standard  that contains only the
matrix of the CALIBRATION standard.

CAPACITY  FACTOR - a chemical property of a system defined as a function of the quantity or size of
that system.

CAPACITY-LIMITED PROCESS - A mechanism (e.g., sulfate adsorption or cation exchange) for which
the long-term ability to supply  or consume cations or anions is constrained by the size of a watershed
pool or capacity (e.g., pool of exchangeable bases and sulfate adsorption capacity) rather than  by
reaction kinetics.

CARBON-BONDED SULFUR - a reduced form of organic sulfur, characterized by C-S bonds.

CARBONIC ACID - a weak acid, H2CO3, formed  by dissolution of carbon dioxide in water.
Dissociation of carbonic acid (to H+ and  HCO3") and subsequent consumption of H+ by exchange or
weathering reactions generates ANC in the form of bicarbonate ions.

CATCHMENT - see WATERSHED.

CATION - a positively charged ion.

CATION DEPLETION - a  process through which  base cations on a soil exchange site are progres-
sively replaced by ACID CATIONS at rates higher than those expected during normal pedogenesis.

CATION EXCHANGE - a  reversible process occurring in soil and/or sediment  in which ACIDIC
CATIONS (e.g., hydrogen  ions) are adsorbed and BASE CATIONS are released.

CATION EXCHANGE CAPACITY - the sum total  of exchangeable cations that a  soil can absorb.
                                           364
-------
CATION (OR ANION) LEACHING - movement of cations (or anions) out of soil, in conjunction with
mobile anions in soil solution.

CATION RETENTION - the physical, biological, and geochemical processes by which cations in
watersheds are held, retained, or prevented from reaching receiving SURFACE WATERS.

CHRONIC ACIDIFICATION - see LONG-TERM ACIDIFICATION.

CIRCUMNEUTRAL - close to neutrality with respect to pH (pH = 7); in natural waters, pH 6 - 8.

CLAY - a soil separate consisting of particles with an equivalent diameter less than 0.002 mm; also a
soil textural class containing > 40 percent clay-sized material, < 45 percent sand and < 40 percent
silt.

CLAY MINERALS - any of a series of sheet silicate minerals formed in a soil or low-temperature
diagenetic environment.

CLOSED LAKES - a lake with a surface water inlet  but no surface water outlet.

CLUSTER ANALYSIS - a multivariate classification  technique for identifying similar (or dissimilar)
groups of observations.

COARSE PARTICLE DRY  DEPOSITION - atmospheric DRY DEPOSITION of particles greater than 2
microns in effective diameter.

COLLINEAR - see MULTICOLLINEARITY.

COMBINATION BUFFER - land area surrounding a lake including area within a 40-foot elevation
contour, area within a linear buffer adjacent to perennial streams, and area around contiguous
wetlands.

COMPLEX - a map unit consisting of two or more dissimilar soil components or miscellaneous areas
occurring in a regularly repeating pattern.

COMPONENTS - see MAJOR COMPONENTS, MINOR COMPONENTS, and MAP UNIT
COMPOSITION.

CONSOCIATION - a map unit dominated by a single soil taxon (or miscellaneous area) and similar
soils.

CONTOUR LINE - a line connecting the  points on the land surface that  have the same elevation.

CONVERGENCE - state of tending to a unique solution. A given scheme is convergent if an
increasingly finer computational grid leads to a more accurate approximation of the unique solution.
Note that a numerical method may sometimes converge on a wrong solution.
                                           365
-------
COVERAGE - a digital analog of a single map sheet; forms the basic unit of data storage in
ARC/INFO.

CUMULATIVE DISTRIBUTIVE FUNCTION - a function, F(x), such that for any reference value X, F(x)
is the estimated proportion of individuals (lakes, streams, estuaries, coastal waters) in the population
having a value x :< X.

DATABASE FILE - a collection of records that share the same format.

DEPOSITIONAL FLUXES - the mass transfer rate to the earth's surface of any of a number of
chemical species.

DEPTH TO BEDROCK - depth to solid, fixed, unweathered rock underlying soils.

DEPTH TO A SLOWLY PERMEABLE OR IMPERMEABLE LAYER - depth to a layer in soils or
underlying soils that restricts the downward flow of water (e.g., bedrock, dense till or fragipan).

DETECTION LIMIT QC CHECK SAMPLE - a QUALITY CONTROL sample that contains the ANALYTE
of interest at two to three times the contract required detection limit.

DIGITIZATION - the process of entering lines or points into a GEOGRAPHIC INFORMATION SYSTEM.

DIGITIZED COORDINATES - lines or points that have been entered into a GEOGRAPHIC
INFORMATION SYSTEM.

DISSIMILATORY REDUCTION - a process in which an oxidized chemical species (e.g., SO4  - S) is
utilized by an organism as an electron acceptor in the absence of free oxygen and released in a
reduced form (e.g., S2") rather than assimilated.

DISSOCIATION - separation of an acid into free H+ and the conjugate base of that acid (e.g.,
H2CO3 -> H+ + HCO3"), or separation of a base into a free hydroxyl and the conjugate acid of the
base (e.g., NH4OH -> NH4+ + OH").

DISSOLUTION RATES - the rate at which a mineral is transformed to aqueous species or secondary
minerals in an aqueous environment.

DISSOLVED ORGANIC CARBON - a measure of organic (nonorganic) fraction of carbon in a water
sample that is dissolved or unfilterable.

DOWNSTREAM REACH NODE - see LOWER NODE.

DRAINAGE - the frequency and duration of periods when the soil is free of saturation or partial
saturation and the depth to which saturation commonly occurs.

DRAINAGE BASIN - see WATERSHED.
                                         366
-------
DRAINAGE CLASS - any of the seven classes that characterize the frequency and duration of soil
saturation.

DRAINAGE LAKE - a lake with SURFACE WATER outlet(s) or with both inlets and outlets.

DRY DEPOSITION - for the purposes of DDRP analysis, atmospheric deposition of materials to
watersheds in any form other than rain or snow.

DRY DEPOSITION VELOCITY - an effective velocity used with airborne concentrations to compute dry
depositional flux of materials to surfaces or watersheds.

ELECTRON ACCEPTOR - an oxidized (or at least partially oxidized) chemical species capable of
undergoing a reduction reaction by addition of an electron.

ELS PHASE I LAKES - the population of lakes sampled during phase I of the Eastern Lake Survey of
the EPA's National Surface Water Survey.

EMPIRICAL MODEL - representation of a real system by a mathematical description based on
experimental data rather than on general physical laws.

ENTISOLS - in Soil Taxonomy, the ORDER of mineral soils with no or very poorly developed genetic
horizons.

EPISODE - a short-term change in stream pH and ACID NEUTRALIZING CAPACITY during  storm flows
or snowmelt runoff.

EQUIVALENT - unit of ionic concentration; the quantity of a substance that either gains or loses one
mole of protons or electrons.

ESTER SULFATE - an oxidized form of sulfur in soil organic matter, characterized  by C-O-SO3 or N-O-
SO3 linkages.

EVAPOTRANSPIRATION (%ET) - the amount or proportion of precipitation that is returned to the air
through direct evaporation or by transpiration of vegetation.

EXCHANGE POOL - the reservoir of BASE CATIONS in soils available to participate in exchange
reactions.

EXTENSIVE PARAMETERS - variables that depend on the size (extent) of the system.

EXCHANGE REACTIONS - any of a number of reactions that describe the partitioning of two chemical
species between a solution and soil exchange complex.

FELDSPARS - a group of tectosilicate  minerals that are the most abundant group in the earth's crust.
                                          367
-------
FIELD REVIEW - a review of soil surveys made in the field by supervisory soil scientists to help field
soil scientists maintain standards that are both adequate for the objectives of the survey and
consistent with those of other surveys. Samples of the fieldwork are examined for soil identification,
placement of boundaries, and map detail in relation to survey objectives.

FINE PARTICLE: DRY DEPOSITION - atmospheric DRY DEPOSITION of particles of size less than 2
microns in effective diameter.

FIRST-ORDER REACTION - a chemical reaction, the rate of which is proportional to the concentration
of the limiting reactant.

FOREST COVER TYPE - a descriptive classification of forest land based on present occupancy of an
area by tree species.  (The term 'Vegetation" implies total forest community, whereas the focus here is
on trees defining type.  Whenever the term "vegetation" is used in this report it should be construed as
FOREST COVER TYPE.)

FREUNDLICH ISOTHERM - an exponential adsorption isotherm of the form E0 = aCb, where: Ec =
concentration of adsorbed species (per unit mass adsorbent), C = dissolved concentration of species
being adsorbed, and a and b are derived coefficients.

FULVIC ACID - a family of naturally-occurring weak organic acids found in soils and surface waters;
fulvic acids are operationally defined as the acid-soluble (pH = 1.0) fraction of an alkali-soluble soil
extract; pK is roughly 3.5.

GAINES THOMAS FORMULATION - a formulation used to describe exchange processes.

GAPON - a formulation used to describe exchange processes.

GENERIC BEDROCK TYPE -  see GENERIC ROCK TYPE.

GENERIC ROCK TYPE - a general classification of different BEDROCK UNITS into groups according
to the primary LITHOLOGY.

GEOGRAPHIC INFORMATION SYSTEM (GIS) - a computerized system designed to store, process,
and analyze data.

GEOLOGY - see BEDROCK GEOLOGY.

GEOMORPHIC POSITION - the relative location in the landscape  described by hillslope elements
(cross section view) and slope components (plane view), e.g., sideslope footslope.

GIBBSITE - a mineral with the chemical formula AI(OH)3.
                                           368
-------
GIS BUFFERS - land area surrounding a lake, stream, or wetland, delineated using a GIS.  See
COMBINATION BUFFER and LINEAR BUFFER.

GLACIAL TILL - see TILL

GLACIOFLUV1AL - a material that has been deposited by glaciers and sorted by meltwater.

GLEY SOIL - a soil developed under conditions of poor drainage, characterized by oxygen depletion
and reduction of iron and other metals (Mn), resulting in  gray colors and mottles.

GRAN ANALYSIS - a mathematical procedure used to determine the equivalence points of a
TITRATION CURVE for acid and base neutralizing capacity.

GROUNDWATER - water in the part of the ground that is completely saturated.

HEURISTIC MODEL - representation of a real system by a mathematical description based on
reasoned, but unproven argument, intended for use as an aid to studying and exploratory analysis of
the system being modelled.

HINDCAST - to estimate some prior event or condition as a result of a rational study and analysis of
available pertinent current and historical data.

HISTOSOLS - in Soil Taxonomy, the ORDER of soils formed from organic PARENT MATERIAL.

HORIZON - a horizontal layer of soil with distinct physical and/or chemical characteristics. Genetic
horizons are the result of soil-forming process.

HYDRAULIC HEAD - hydrostatic pressure created by a difference in height of water columns in
different portions of a connected aquifer.

HYDRAULIC RESIDENCE TIME - a measure of the average amount of time water is retained in a lake
basin. It can be defined on the basis of inflow/lake volume, represented as RT, or on the basis of
outflow/lake volume and represented as Tw. The two definitions yield similar values for fast flushing
lakes, but diverge substantially for long residence time SEEPAGE LAKES.

HYDROLOGIC FLOW PATHS - the distribution and circulation of water deposited by precipitation on
the surface of the land, in the soil, and underlying rocks  within a WATERSHED.

HYDROLOGIC RETENTION TIME - see HYDRAULIC RESIDENCE TIME.

HYDROUS OXIDE - a collective term referring to any of a group of amorphous or crystalline species
of iron or aluminum that are  partially or fully hydrated (e.g., MO(OH), M(OH)3).
                                           369
-------
 IMMOBILIZATION REACTION - conversion of an inorganic form of a nutrient (especially S or N) to
 organic matter.

 IMPOUNDMENT - a man-made lake created by construction of a dam; also applied to natural lakes
 whose level is controlled by a human-made spillway.

 INCEPTISOLS - in Soil Taxonomy, the ORDER of soils with at least one diagnostic horizon, but with
 no horizon strongly enough developed to place them in another ORDER.

 INCLUSIONS - see MINOR COMPONENTS.

 INDEX OF CONTACT TIME - the theoretical maximum potential of contact between runoff and the
 soil matrix.  The index is calculated by dividing the soil water flow rate (obtained using Darcy's law) by
 average annual runoff.

 INDEX SAMPLE - in NE lakes, one sample per lake, used to represent chemical conditions on that
 lake.  In streams, any sample (or the average of one to three samples)  collected at a stream NODE
 during the SPRING BASEFLOW INDEX PERIOD, used to represent chemical conditions in the stream.

 INFO - a database management system that stores, maintains, manipulates, and reports information
 associated with geographic features in ARC/INFO.

 INITIAL CONDITIONS - given values of DEPENDENT VARIABLES or relationship between dependent
 and independent variables at the time of start-up of the computation in a mathematical model.

 IN-LAKE SULFUR RETENTION -  net retention of sulfur within a lake, occurring principally by
 reduction within sediments.

 INTENSITY FACTOR - a variable with properties defined by concentration or a surface and/or in
 solution, and therefore independent of the quantity or size of the system.

 INTENSIVE PARAMETERS - variables whose values are independent of the size or extent of the
 system, e.g., temperature and pH.

 INTERQUARTILE RANGE  - the difference between the 75th and the 25th percentiles.

 IONIC STRENGTH -  a measure of the interionic effect resulting from the electrical attraction and
 repulsion between various ions. In very dilute solutions, ions behave independently of each other and
the ionic strength can be calculated from the measured concentrations  of ANIONS and CATIONS
present in the solution.
                                          370
-------
ISOTHERM - a linear or nonlinear function describing partitioning of an absorbent between solid and
sorbed phases.  Such isotherms were originally used to characterize (nearly) ideal processes (e.g., the
Langmuir equation was developed to describe adsorption of a gas by a solid), but are often
empirically defined for adsorption of anions or organic compounds on soils because they provide a
convenient shorthand to describe partitioning.

KINETIC MODELS - any of a family of numerical models that use kinetic considerations as the
unifying principle in describing  natural processes.

KRIGING - a technique for spatial interpolation.

LABEL - represents point features or is used to assign identification numbers to POLYGONS.

LAKE TYPE  - a classification of lakes based on the presence or absence of inlets, outlets, and dams
as represented on LARGE-SCALE MAPS.

LAND COVER - see FOREST COVER TYPE.

LAND USE - the dominant use of an area of land (e.g., crop land).

LANDFORM SEGMENT - a small part of the local landform that is uniquely related to landscape
processes.

LANGMUIR  ISOTHERM  - a hyperbolic adsorption isotherm (used in this project for sulfate) of the
form Ec = (B1 * C)/(B2 + C), where: Ec = net adsorbed sulfate, C = dissolved sulfate, and B1 and B2
are empirically derived coefficients.  When appropriate, the isotherm can be "extended" by addition of
a third coefficient to describe a non-zero Y-intercept.

LARGE-SCALE MAPS - 1:24,000, 1:25,000, or 1:62,500 scale U.S. Geological Survey topographical
maps.

LEACHING - the transport of a solute from  the soil in the soil solution.

LEVERAGE POINT - a data point that strongly influences the parameter estimates in a regression.

LIGAND EXCHANGE - a mechanism of bond formation between an oxyanion and a soil mineral
bearing  hydroxyl groups. The exchange involves formation of inner sphere complexes of anions to
Lewis acid sites, following replacement of water from the Lewis acid site by the oxyanion.

LIMESTONE - a rock type consisting primarily of CALCITE.

LINEAR BUFFER - land area within a set distance of a lake or stream.
                                           371
-------
LITHOLOGY - the physical characteristics of a rock or mapped BEDROCK UNIT.  Generally relates to
mode of formation, mineralogy, and texture.

LITTERFALL - fresh organic detritus, usually leaves, needles, twigs, etc., that compose the bulk of the
forest floor.

LOCAL LANDFORM - a subdivision of the regional landform that is the result of localized landscape
processes.

LONG-TERM ACIDIFICATION - a long-term partial or complete loss of ACID NEUTRALIZING CAPACITY
from a lake or stream.

LONG-TERM ANNUAL AVERAGE DEPOSITION (LTA)  - a dataset of atmospheric deposition
representing atmospheric deposition during the early-to-mid 1980s for the purposes of the DDRP.

LOWER NODE - the downstream NODE of a STREAM REACH.

MAJOR LAND RESOURCE AREA - a geographic area characterized by a particular pattern  of soils,
climate, water resources, and LAND USE.

MAJOR COMPONENTS - soil components or miscellaneous areas that are identified in the name of a
map unit.

MAP COMPILATION  - the process of checking and measuring soil map unit data.

MAPPING PROTOCOLS - instructions that guide the field mapping and provide for quality control.

MAP SYMBOL - a symbol used on a map to identify map units.

MAP'PING TASK LEADER - the person responsible for field mapping activities.

MAP UNIT - see SOIL MAP UNIT.

MAP UNIT COMPOSITION - the relative proportion (expressed in percent) of all soil components and
miscellaneous areas in a map unit.

MAP UNIT COMPOSITION FILE - a DATABASE FILE that contains ail components and their relative
proportion for each map unit in the survey area (components are identified by an assigned code, i.e.,
SCODE).

MAP UNIT CORRELATION - see SOIL CORRELATION.

MAP UNIT DELINEATION - an area on a map uniquely identified with a symbol.  A delineation of a soil
map has the same major components as identified and named in the map unit.

                                          372
-------
MAP UNIT NAME - the title of a map unit identified by the major soil components or miscellaneous
areas followed by appropriate phase terms.

MASS ACTION MODELS - any of a family of numerical models that use equilibrium-based principles as
the central unifying theme.

MASS BALANCE MODELS - any of a family of numerical models that use conservation-of-mass
principles as the central unifying theme.

MASS TRANSFER COEFFICIENTS - a removal or rate constant used in models of in-lake alkalinity
generation (and elsewhere) to quantify the average removal rate of a reactant from solution.  Specific
reference in this project is transfer from solution to sediment by all processes, including sedimentation
and diffusion.  In many systems, the mass transfer coefficient for sulfur is essentially a diffusion constant
for sulfate across the water-sediment interface; for nitrate a biological uptake/sedimentation rate.

MASTER HORIZONS - the most coarsely based delineations within a pedon. Usually,  A/E horizons
denote zones of net mass depletion, B horizons are zones of net accumulation, and C  horizons indicate
minimal pedogenic evolution.

MATRIX SPIKE - a QUALITY CONTROL sample made by adding known quantity of an ANALYTE to a
sample aliquot.

MAX - the maximum sensitivity code observed on a WATERSHED.

MEAN - the weighted average of sensitivity codes for a WATERSHED.

MEDIAN (M) - the value of x such that the cumulative distribution function F(x) = 0.5; the 50th
percentile.

MID-APPALACHIAN REGION - one of the three geographic regions considered  by the, DDRP,
consisting of upland areas (subregions 2Bn and 2Cn) of the Mid-Atlantic region (MD, PA, VA, WV)
defined by the National Stream Survey.

MINERAL WEATHERING - dissolution of rocks and minerals by erosive forces.

MINERALIZATION - microbially-mediated conversion of nutrients from an  organically bound  (especially
N and S) to an inorganic form.

MINOR COMPONENTS - soil components or miscellaneous areas that are not identified in the name of
the map unit.  Many areas of these  components are too small to be delineated separately.

MISCELLANEOUS AREA  - land areas that have no soil and thus support  little or no vegetation without
major reclamation.  Rock outcrop is an example.
                                            373
-------
MISCELLANEOUS LAND AREAS - see MISCELLANEOUS AREA.

MOBILE ANION - an anion that remains in solution and passes through a soil without significant delays
due to biological or chemical processes; also a model or paradigm for cation leaching from soils, based
on the premise that the rate of cation leaching from a soil is controlled by the sum of mobile anions
(which are regulated by a suite of more-or-!ess independent processes).

MONTE CARLO METHOD - technique of STOCHASTIC sampling or selection of random numbers to
generate synthetic data.

MOTTLING - spots or blotches of different color in a soil,  including gray to black blotches in poorly
drained soils due to presence of reduced iron and other metals.

NITROGEN TRANSFORMATION - biochemical processes through which nitrogen deposited in an
environment is converted to other forms.

NODE - the points identifying either an upstream or downstream end of a REACH.

NONPARAMETRIC - referring to a statistical procedure that does not make the classical distributional
assumptions.

NON-SILICATE IRON AND ALUMINUM - soil iron and/or aluminum occurring in the soil as an
amorphous or a (hydrous) oxide phase rather than as an ion incorporated within a silicate mineral
lattice.

OFFICIAL SOIL SERIES  DESCRIPTION - a record of the definitions of a soil series and other relevant
Information about each series. These definitions are the framework within which most of the detailed
information about soils of the United States  is identified with soils at  specific places.  These definitions
also provide the principal medium through which detailed information about the soil and its behavior at
one place is projected to similar soils at other places.

ORDER - in Soil Taxonomy, the highest level of classification, e.g., SPODOSOLS.

ORGANIC ACID - organic compound  possessing an acidic functional group; includes fulvic and humic
acids.

ORGANIC ANION - an organic molecule with a negative net ionic charge.

ORGANIC "BLOCKING" - a reduction  in the sulfate (or other anion)  adsorption capacity of a soil
resulting from  preferential sorption of organic acids by the soil.

ORGANIC HORIZONS - any identifiable soil horizon containing in excess of 20 percent organic matter
by weight.
                                            374
-------
OUTLIER - observation not typical of the population from which the sample is drawn.

OXIDATION - loss of electrons by a chemical species, changing it from a lower to a higher oxidation
state (e.g., Fe2+ to Fe3+ or S'2 to S+6, with intermediates).

PARAMETER - (1) a characteristic factor that remains at a constant value during the analysis, or (2) a
quantity that describes a statistical population attribute.

PARENT MATERIAL - the material from which soils were formed.

PARTIAL PRESSURE - the pressure of a gaseous sample that is attributable to one particular
component.

PEDON - the smallest block of soil that contains all the characteristics of a soil (usually about 1 m2); a
soil individual.

PERCENT COARSE FRAGMENTS - the percentage of soil, by volume, that is composed of rock
fragments unable to pass through a 2-mm sieve.

PERMEABILITY - the  ease with which gases, liquids, or plant roots penetrate or pass through a bulk
mass of soil or a layer of soil.

pH - the negative logarithm of the hydrogen ion activity. The pH scale runs from 0 (most acidic) to 14
(most alkaline); a difference of 1 pH unit indicates a tenfold change in hydrogen activity.

POLYGON - represents area features.

PRECISION - a measure of the capacity of a method to provide reproducible measurements of a
particular ANALYTE (often represented by variance).

PRIMARY MINERAL WEATHERING - the natural  process by which thermodynamically unstable
minerals are converted to more stable phases under earth surface conditions.

PRINCIPAL COMPONENT ANALYSIS - a statistical analysis concerned with explaining the variance-
covariance structure through the use of PRINCIPAL COMPONENTS.

PRINCIPAL COMPONENTS - particular linear combinations of the original data, which geometrically
represent a new coordinate system with axes in the directions of maximum variability.

PROBABILITY SAMPLE - a sample in which each unit has a known probability of being selected.

QC CHECK SAMPLE  - a QUALITY CONTROL sample that contains the ANALYTE of interest at a
concentration in the mid-calibration range.
                                           375
-------
QUALITY ASSURANCE - steps taken to ensure that a study is adequately planned and implemented
to provide data of known quality, and that adequate information is provided to determine the quality of
the database resulting from the study.

QUALITY CONTROL - steps taken during a study to ensure that data quality meets the minimum
standards established by the quality assurance plan.

QUARTILE - any of three values (Q1, Q2, Q3) that divide a population into four equal classes, each
containing one-fourth of the population.

QUINTILE - any of the four values (Q1 , Q2 , Q3 , Q4) that divide a population into five equal classes,
each representing 20 percent of the.population; used to provide additional values to compare
characteristics among populations of lakes and streams.

RATE-LIMITED REACTION - a process  (e.g., mineral weathering) for which the long-term ability to
supply reaction products (e.g., base cations) is constrained by reaction or transport kinetics.

RCC TRANSECTS - transects conducted by the Regional Coordinator/Correlator (RCC).

REACH - segments of the stream network represented as blue  lines on 1:250,000-scale U.S.
Geological Survey maps. Each reach (segment) is defined as the length of stream between two blue-
line confluences. In the NSS-I, stream reaches were the sampling unit.

REACTION ORDER - the relationship between the rate of a chemical reaction and the concentration
of a reaction substrate, defined by the value of the exponent of that substrate.

REAGENT BLANK - a QUALITY CONTROL sample that contains all the reagents used and in the
same quantities used  in preparing a soil sample for analysis.

REDUCTION/OXIDATION - chemical reaction in which substances gain or lose electrons.

REGION  - a major area of the conterminous  United States where a substantial number of streams with
ALKALINITY less than  400  jieq/L can be found.

REGIONAL LANDFORM - physiographic areas that reflect a major land-shaping  process over a long
period of time.

REGIONAL SOILS LEGEND - a correlated and controlled legend for an entire region (see SOIL
IDENTIFICATION LEGEND).

RELMAP - a source-receptor model designed to estimate dry deposition of sulfur; not used directly in
the DDRP.
                                           376
-------
REPORTS - relative to GIS activities, a format designed by the user for printing out information
containing the data files.

RESERVOIR - a body of water collected and stored for future use in a natural or artificial lake.

RESIDUAL - in regressions, the difference between the observed dependent variable and the value
predicted from the regression fit.

REUSS MODEL - a numerical model used to describe exchange processes in a soil environment.

RIPARIAN - a zone bounding and directly influenced by SURFACE WATERS.

ROBUST - a statistical procedure that is insensitive to the effect of OUTLIERS.

ROUTINE TRANSECTS - transects conducted by field soil scientists responsible for the mapping.

SALT EFFECT - the process by which hydrogen ions are displaced for the soil exchange complex by
BASE CATIONS (from neutral salts). The result is a short-term increase in the acidity of associated
water.

SAMPLING CLASS - see SOIL SAMPLING CLASS.

SAMPLING CLASS CODE - a three-character code assigned to each SOIL SAMPLING CLASS.

SAMPLING CLASS COMPOSITION - the relative proportion of sampling classes in a map unit.

SAND - a soil separate between 0.05 and 2.0 mm in diameter; also a soil texture class containing at
least 85 percent sand, and whose percentage of silt, plus  1.5 times the percent clay, does not exceed
15.

SATURATION INDEX - the ratio of the ion activity product (of dissolved ions) to the solubility product
for a solid phase; if the saturation index (SI) exceeds 1.0,  the solution is supersaturated with respect to
that phase; if SI = 1.0, the solution  is at equilibrium, if SI < 1.0, the solution is undersaturated with
respect to that solid phase.

SCENARIO - one possible deposition  sequence following implementation of a control or mitigation
strategy and the subsequent effects associated with this deposition sequence.

SECONDARY MINERALS - any inorganic mineral phase formed by transformation of another mineral or
by precipitation from an aqueous phase.

SEEPAGE LAKE - a lake with no permanent SURFACE WATER inlets or outlets.
                                            377
-------
SELECTIVITY COEFFICIENT - the apparent constant used to describe the partitioning of species in
an exchange reaction.

SENSITIVITY ANALYSIS - test of a model in which the value of a single variable or parameter is
changed, and the impact of this change on the DEPENDENT VARIABLE is observed.

SENSITIVITY CODES - see BEDROCK SENSITIVITY SCORES.

SIGNIFICANCE LEVEL - the conditional probability that a statistical test will lead to rejection of the
null hypothesis, given that the  null hypothesis is true.

SILICA - the dissolved form of silicon dioxide (SiO2).

SILT - a soil separate consisting of particles between 0.05 and 0.002 mm in equivalent diameter; also
a soil texture class containing at least 80 percent silt and < 12 percent clay.

SILVICULTURAL PRACTICES - forest management  practices to increase wood yields:  thinning,
pruning, fertilization, spraying with herbicides/insecticides, and irrigating.

SIMULATION - replication of the prototype using a model.

SKELETAL SOILS - soils with  at least 35 percent rock fragments in the control section.

SLOPE PHASE - the slope gradient of a soil map unit or taxonomic unit expressed in percent.

SLOPE SHAPE ACROSS - shape of the surface parallel to the contours of the landscape (e.g.
concave, convex, plane).

SLOPE SHAPE DOWN - shape of the land surface at right angles to the contours of the landscape.

SMALL-SCALE MAP - 1:250,000-scale U.S. Geological Survey map.

SOIL - unconsolidated material on the surface of the earth that serves as a natural medium for the
growth of plants.

SOIL ACIDIFICATION - a process  through which BASE CATIONS are removed from the soil and are
replaced by ACID CATIONS.

SOIL BUFFERING CAPACITY - the capacity of a soil to resist changes in pH with the addition of
acids to the sysrtem.

SOIL COMPONENT CODE - four-character code assigned to each  soil or miscellaneous area
component of map units in a survey area. Codes were used to link data files.
                                           378
-------
SOIL COMPONENTS FILE - a computer data file that contains all the soil and miscellaneous area
components in a survey area and identified with a code (i.e., SCODE).

SOIL CORRELATION - the process of maintaining consistency in naming, classifying, and interpreting
kinds of soils and of the units delineated on maps.

SOIL EXCHANGE COMPLEX - all components of a soil that contribute to the absence of exchange
properties of that soil.

SOIL FAMILY - next to the lowest category in Soil Taxonomy in which classes are separated mainly on
particle size, temperature, and mineralogy.

SOIL IDENTIFICATION LEGEND - a map legend that lists the symbols used to identify SOIL MAP
UNITS and the names of the  map units.

SOIL LEGEND - see SOIL IDENTIFICATION LEGEND.

SOIL MAP UNIT - a collection of areas defined and named in terms of their soil components or
miscellaneous areas or both.  Each map unit differs in some respect from all others in a survey area and
is uniquely identified on a soil map.

SOIL SAMPLING CLASS - an arbitrary grouping of soils either known or expected to have similar
physical and/or chemical effects on drainage waters with respect to effects of acidic deposition.

SOIL SERIES - the most homogenous category in the taxonomy used in the United States.  A group of
soils that  have horizons similar in arrangement and in differentiating characteristics.

SOIL SOLUTIONS - those aqueous  solutions in contact with soils.

SOIL TAXONOMIC CLASS - the soil members within limits of ranges set by Soil Taxonomy.  Taxonomic
units are members of the taxonomic class.
SOIL TAXONOMIC UNIT - a kind of soil described in terms Of ranges in soil properties of the
polypedons referenced by the taxonomic unit in the survey area.

SOIL TEXTURE - the relative proportion by weight, of the several soil particle size classes finer than 2
mm in equivalent diameter (e.g., sandy loam).

SOIL TEXTURE MODIFIER - suitable adjectives added to soil texture classes when rock fragments
exceed about 15 percent by volume, for example, gravelly loam.  The terms 'Very" and "extremely" are
used when rock fragments exceed about 35 and 60 percent by volume, respectively.

SOIL TRANSECT - a distance on the surface of the earth represented by a line on a map.  Transects
can be straight, dogleg, or zigzag.

                                            379
-------
SOLID PHASE IEXCHANGERS - those components of soils, primarily organic matter, clay minerals
and mineral oxides, that serve as the sites for exchange reactions.

SOLUM - soil layers that are affected by soil formation.

SPECIATION MODEL - a numerical model used to describe the distribution of aqueous species
among various possible complexes and ion pairs; usually for the purpose of estimating single ion
activities.

SPECIFIC ADSORPTION - adsorption of sulfate by  ligand exchange, often involving exchange of two
ligands and formation of a bridged (M-O-SO2-O-M) structure.

SPODIC HORIZONS -  a soil hon'zon in which iron oxides, aluminum oxides, and organic matter have
accumulated from higher horizons.

SPODOSOLS - in Soil Taxonomy, the ORDER of mineral soils with well-developed SPODIC
HORIZONS.

SPRING BASEF:LOW INDEX PERIOD - a period of the year when streams are expected to exhibit
chemical characteristics most closely linked to ACIDIC DEPOSITION.  The time period between
snowmelt and leafout (March 15 to May 15 in the NSS-I) when NSS-I  stream reaches were visited
coinciding with expected periods of highest geochemical and assessment interest (i.e., low seasonal
pH and sensitive  life stages of biota).

STABILITY  (NUMERICAL OR COMPUTATIONAL) - ability of a scheme to control the propagation or
growth of small perturbations introduced in the calculations. A scheme is unstable if it allows the
growth of error so that  it eventually obliterates the solution.

STANDARD DEVIATION - the square root of the variance of a given statistic.

STEADY-STATE - the condition that occurs when a  property (e.g., mass, volume, concentration) of a
system does not  change with time. This condition requires that sources and sinks of the property are
in balance (e.g., inputs equal outputs; production equals consumption).

STEPWISE  REGRESSION - a statistical model-selection technique in which variables are added to the
model one at a time using F statistics as criteria, and then previously included variables are examined
for removal from the model.

STOCHASTIC  - an  occurrence that is probabilistic in nature.

STRATIFICATION FACTOR - factor used to  define strata prior to lake selection; the factors used in
the ELS-I were region, subregion, and alkalinity map class.

STRATUM - in the NSS-I, sampling strata were designated as REGION and expected alkalinity map
class  (greater than or less than 50 fieq/L).

                                           380
-------
STREAM REACH - see REACH.

STREAM SAMPLING NODE - the points identifying either an upstream or downstream end of a
REACH. In the NSS-I, the points above or below the confluence of two streams.

SUBGROUP - in Soil Taxonomy, the level of classification just below GREAT GROUP, e.g., Typic
Haplorthods.

SUBORDER - in Soil Taxonomy, the level of classification just below ORDER, e.g., Orthods.

SUBREGIONS - areas within each region that are similar in water quality, physiography, vegetation,
climate, and  soil; used as a STRATIFICATION FACTOR in ELS I design.

SUBTENDING - of or pertaining to subjacent or downslope landscape positions.

SULFATE ADSORPTION - reversible retention of sulfate on the surface of positively-charged  surfaces
of soils, by either electrostatic attraction or by ligand exchange.

SULFATE ISOTHERM - a mathematical relationship between the amounts of sulfate absorbed on soil
and the concentrations of sulfate in solution equilibrated with that soil.

SULFATE MINERALS - a mineral containing sulfur in the +6 oxidation state (SO4); sulfate minerals
are typically formed as evaporites; gypsum  and  anhydrite are common sulfate minerals.

SULFATE MOBILITY - unrestricted movement (in the soil solution) of sulfate through the soil, with
subsequent release in soil leachate; also the extent to which  sulfur supplied to a watershed from
external sources (i.e., acidic deposition) passes  though watershed soils rather than being retained by
adsorption or some other process.

SULFATE RETENTION - the physical, biological, and geochemical processes by which sulfate in
WATERSHEDS is held, retained, or prevented from  reaching  receiving SURFACE WATERS.

SULFIDE - an ion  consisting of reduced sulfur (S2") or a compound containing sulfide, e.g., hydrogen
sulfide (H2S) or the iron sulfide pyrite (FeS2).

SULFIDE OXIDATION - chemical reaction in which a sulfide  loses electrons and assumes a higher
oxidation state; sulfate is the completely oxidized end product.

SULFIT1C - containing sulfide minerals, usually pyrite.

SULFUR INPUT/OUTPUT BUDGET - an approach to describing sulfur mobility in a watershed by
comparing fluxes of sulfur to and from the watershed (as the difference between input and output or
as a ratio).
                                            381
-------
 SURFACE WATER - streams and lakes.

 SURFACE WATER RUNOFF - precipitation that flows overland to reach SURFACE WATERS.

 SURFiCIAL GEOLOGY - characteristics of the earth's surface, especially consisting of unconsolidated
 residual, colluvial, or glacial deposits lying on the BEDROCK.

 SYNOPTIC - relating to or displaying conditions as they exist at a point in time over a broad area.

 SYSTEMATIC ERROR - a consistent error introduced in the measuring process.  Such error commonly
 results in biased estimations.

 TARGET POPULATION  - a subset of a population explicitly defined by a given set of exclusion criteria
 to which inferences are to be drawn from the sample attributes.

 THERMODYNAMIC CONSTANTS - an empirically derived constant used to describe the relative
 distribution of chemical species in a specified reaction when at equilibrium.

 THROUGHFALL - precipitation that has interacted with a forest canopy, the chemistry of which is thus
 modified from that of incident precipitation due to wash off of dry-deposited material and leaf exudates
 as well as by Ion exchange and uptake by leaf surfaces.

 TILL - unstratified material deposited by glaciers.

 TITRATION CURVES - a loci of points describing some solution property, usually pH, as a function of
 the sequential addition of a strong acid (or base) to the system.

 TOPMODEL - topographically based, variable source area hydrologic model.

 TOPOGRAPHIC MAP - a map showing contours of surface elevation.

 TRANSECT - see SOIL TRANSECT.

 TRANSECTING - a field  activity involving the collection of data at points along a designated line (see
 TRANSECT POINTS).

 TRANSECT POINTS - locations along  a TRANSECT where data are collected.

 TRANSECT SEGMENT UNION - all transect stops in the same map unit on a WATERSHED.

TRANSECT STOPS - see TRANSECT POINTS.

TRANSFORMATION ERROR - calculates the residual mean square error of the digitized TIC locations
and the existing TICs.
                                           382
-------
TRAVERSING - a field activity that involves observation at uncontrolled representative locations in the
landscape.

TYPICAL YEAR (TY) DEPOSITION DATA - a dataset of atmospheric deposition developed within the
DDRP for specific use with the integrated watershed models.

UNCERTAINTY ANALYSIS - the process of partitioning modelling error or uncertainty to four sources:
intrinsic natural variability, prior assumptions/knowledge, model identification, and prediction error.

UNIVERSAL TRANSVERSE MERCATOR (UTM) PROJECTION - a standard map projection used by
the U.S. Geological Survey.

UPPER NODE - the upstream NODE of a STREAM REACH.

UPSTREAM REACH NODE - see UPPER NODE.

UTM COORDINATES - lines or points as represented  in a UNIVERSAL TRANSVERSE MERCATOR
PROJECTION.

VALIDATION - comparison of model results with a set of prototype data not used for verification.
Comparison includes the following: (1) using a dataset very similar to the verification data to
determine the validity of the model under conditions for which it was designed; (2) using a dataset
quite different from the verification data to determine the validity of the model under conditions for
which it was not designed but could possibly be used; and (3) using post-construction prototype data
to determine the validity of the  predictions based on model results.

VANSELOW EXCHANGE FORMULATION - a formulation used to describe soil exchange reactions.

VARIABLE - a quantity that may assume any one of a set of values during the analysis.

VARIABLE SOURCE AREA - A topographically convergent, low transmissivity area within a watershed
that tends to produce saturation excess overland flow during storm runoff periods.

VEGETATION - see FOREST COVER TYPE.

VERIFICATION - check of the behavior of an adjusted model against a set of prototype conditions.

WATERSHED - the geographic area from which SURFACE WATER drains into a particular lake or
point along a stream.

WATERSHED STEADY STATE - a condition in which inputs of a constituent to a  WATERSHED equal
outputs.
                                          383
-------
WATERSHED SULFUR RETENTION - retention of sulfur by any of a number of mechanisms within a
WATERSHED.

WEATHERED BEDROCK - soft or partly consolidated BEDROCK that can be dug with a spade.

WEATHERING - physical and chemical changes produced in rocks at or near the earth's surface by
atmospheric agents with essentially no transport of the altered materials.

WEIGHT - the inverse of a sample's inclusion probability; each sample site represents this number of
sites in the TARGET POPULATION.

WET DEPOSITION - for the purposes of the DDRP, atmospheric deposition of materials via rain or
snow.

WETLAND - an area, generally with hydric soils, that is saturated, flooded, or ponded long enough
during the growing season to develop anaerobic conditions in the upper soil horizons and that is
capable of supporting the growth of hydrophitic vegetation.

ZERO-ORDER REACTION  - a chemical reaction, the rate of which is independent of reactant
concentration.
                                          384
-------