-------
I
I
I
i
i
i
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
at sulfate steady state, it was also considered useful to estimate changes at an interim time, 25 years.
In estimating interim changes, however, it is important to recognize that (1) sulfate concentrations in
n years (25 in this case), expressed as concentration or as percent of steady-state concentration, can
vary widely among sites, and that (2) site-specific time predictions are available for only a few sites.
To overcome these limitations and develop regional population predictions, relationships between
current sulfate and projected levels at 25 years were evaluated for the modeling sites discussed in
section 5.3.4. Based on those data, sulfate concentrations in 25 years were estimated as:
where So, 83$, and SS represent sulfate concentration (or percent steady state) at 0, 25 years, and
steady state; respectively. Using this equation with current and steady-state sulfate concentrations,
concentrations were estimated at 25 years from present for all NE lakes not already at steady state.
A number of analyses of projected steady-state concentrations of ANC were made based upon
the above data sets and various scenarios of percentage reduction in deposition. These analyses
include (1) three levels of deposition (100%, 80%, and 50% of CLD), (2) four values of F (0, 0.2, 0.4, and
0.7), and (3) two times periods (25 years from present and steady state - about 50 years). Analyses
were performed for each of the NE subregions, and estimates were projected up to the population
scale. Estimates were made of the projected additional acidic systems as well as for the expected
changes in ANC at final steady state.
This projected steady-state analysis was subject to a large degree of unavoidable uncertainty.
Uncertainty is associated with the estimates of runoff, precipitation, chemistry of precipitation,
amount of dry deposition, lake sulfate concentration, and lake ANC. Approximations can be made to
the individual uncertainties associated with each of the components of the calculations, and these
approximations can be combined to yield an estimate of uncertainty associated with the final
computations. For amount of precipitation, amount of runoff, and chemistry of precipitation, the
uncertainties were approximated by the standard deviation of the estimates of values for all sites
within Region 1. For dry deposition, an uncertainty was provided by R. Dennis (personal
communication). This approximation of uncertainty was, in fact, more nearly an estimate of biases of
the model predictions of dry deposition (R. Dennis, personal communication). Lake sulfate
concentrations appear to remain relatively consistent over the course of the year, and an
approximation of the uncertainty in this component was made as the standard deviation of biweekly
values of sulfate concentration observed in Woods Lake and Panther Lake as part of the ILWAS.
These individual estimates of uncertainty were combined in a first-order error analysis to yield an
estimation of the uncertainty associated with the final calculations. Approximately two-thirds of this
estimated uncertainty is due to uncertainties in the estimates of wet and dry deposition. Additional
•5-53
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
uncertainty regarding year-to-year variation of lake ANC as a function of relative amounts of
precipitation during the year were not taken into account in this analysis.
5.5.2.2 Results
Tables 5-5 and 5-6 show the results of the analyses for the projected (population estimate)
additional acidic systems. Projections of recovery of currently acidic systems are described in
Section?. It is important to point out that these estimates include estimates of acidification of all
systems having sulfate concentrations below expected steady state. Thus the analysis includes a
number of lakes that, although they have sulfate concentrations below steady state, might actually
currently be at steady state (due to the uncertainty associated with determining the true steady-state
concentration). Therefore, at least for this factor in the analysis, these projections are overestimates.
TABLE 5-5. POPULATION ESTIMATE OF PROJECTED ADDITIONAL ACIDIC LAKES -
25 YEARS: DRAINAGE LAKES AND RESERVIORS ONLY
(ESTIMATED POPULATION NUMBER = 6349)
Deposition (% of current levels)
F
Factor Region
1A
IB
n 1C
n
V
ID
IE
Total
1A
IB
no 1C
0.2
ID
IE
Total
#
%
#
%
#
%
#
%
#
%
#
%
#
%
#
%
#
%
#
%
#
%
#
%
Base
26
2
56
4
52
4
66
6
0
0
200
3
26
2
21
1
52
4
66
6
0
0
165
3
100%
Lowa
8
1
13
0.9
0
0
59
5
0
0
SO
1
0
0
13
0.9
0
0
52
5
0
0
65
1
Highb
65
6
111
8
115
9
105
10
16
1
412
6
48
4
110
8
115
9
98
9
0
0
371
6
Base
8
0.7
13
0.9
0
0
33
3
0
0
54
0.9
8
0.7
13 .
0.9
0
0
14
1
0
0
35
0.6
80%
Low
0
0
6
0.4
0
0
14
1
0
0
20
0.3
0
0
3
0.2
0
0
14
1
0
0
17
0.3
50%
High
26
2
62
4
52
4
59
5
0
0
199
3
26
2
53
4
11
0.9
52
5
0
0
142
2
Base
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Low High
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
6
0.4
0
0
0
0
0
0
6
0.1
0
.0
6
0.4
0
0
0
0
0
0
6
0.1
(continued)
5-54
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 5-5. (Continued)
I
I
i
i
i
I
I
Deposition (% of current levels)
F
Factor
0.4
.Total
0.7
Total
Region
1A
IB
1C
ID
IE
1A
IB
1C
ID
1C
#
%
#
%
#
%
#
%
#
%
#
%
#.
%
#
%
#
%
#
%
#
%
#
%
Base
26
2
13
0.9
11
1
66
6
0
0
116
2
8
1
3
0.2
0
0
7
0.6
0
0
18
0.2
100%
Low8
0
0
6
0.4
0
0
33
3
0
0
39
1
0
0
3
0.2
0
0
7
0.6
0
0
10
0.2
Highb
41
4
70
5
81
6
92
9
0
0
284
4
17
2
21
1
8
1
33
3
0
0
79
1
Base
0
0
6
0.4
0
0
7
0.6
0
0
13
0.2
0
0
3
0.2
0
0
0
0
0
0
3
0.1
80%
Low
0 .
0
3
0.2
0
0
0
0
0
0
3
0.05
0
0
0
0
0
0
0
0
0
0
0
0
High
26
2
24
2
0
0
33
3
0
0
83
1
8
0.7
6
0.4
0
0
7
0.6
0
0
21
0.3
Base
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
50%
Low
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
High
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
a "Low"=low range uncertainty estimate.
b "High"=high range uncertainty estimate.
TABLE 5-6. POPULATION ESTIMATE OF PROJECTED ADDITIONAL ACIDIC LAKES-
50 YEARS: DRAINAGE LAKES AND RESERVIORS ONLY
(ESTIMATED POPULATION NUMBER = 6349)
Deposition (% of current levels)
F
Factor Region
1A #
%
IB #
%
n 1C #
Q
ID #
%
IE #
%
Total #
%
Base
26
2
65
4
63
5
72
7
8
0.7
234
4
100%
Low*
8
0.7
13
0.9
0
0
52
5
0
0
73
1
Highb
160
15
125
9
168
13
131
12
24
2
608
10
Base
18
2
13
0.9
0
0
52
5
0
0
83
1
80%
Low.
0
0
13
0.9
0
0
14
1
0
0
27
0.4
50%
High
35
3
65
4
52
4
66
6
8
0.7
226
4
Base
0
0
3
0.2
0
0
0
0
0
0
3
0.05
Low High
0
0
0
0
0
0
0
0
0
.0
0
0
8
0.7
6
0.4
0
0
7
0.6
0
0
21
0.3
(continued)
5-55
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 5-6. (Continued)
Deposition (% of current levels)
F
Factor Region
0.2
Total
0.4
Total
0.7
Total
1A
IB
1C
ID
IE
1A
IB
1C
ID
IE
1A
IB
1C
ID
IE
#
%
#
%
#
#
%
#
%
#
%
*
%
#
%
#
#
%
#
%
#
%
#
%
#
%
#
#
%
#
%
#
%
Base
26
2
59
4
52
4
66
6
0
0
203
3
26
2
21
1
52
4
66
6
0
0
165
3
8
1
10
0.7
0
0
26
2
0
0
44
1
100%
Low*
0
0
13
0.9
0
0
52
5
0
0
65
1
0
0
13
0.9
0
0
33
3
0
0
46
6.7
0
0
3
0.2
0
0
7
0.6
0
0
10
0.2
High*
128
12
111
8
134
10
125
12
24
2
522
8
92
8
110
8
115
9
105
10
16
1
438
7
41
4 •
29
2
40
3
72
7
0
0
182
3
Base
8
0.7
13
0.9
0
0
33
3
0
0
54
0.9
8
0.7
13
0.9
0
0
14
1
0
0
35
0.6
0
0
3
0.2
0
0
7
0.6
0
0
10
0.2
80%
Low
0
0
6
0.4
0
0
14
1
0
0
20
0.3
0
0
3
0.2
0
0
0
0
0
0
3
0.05
0
0
0
0
0
0
0
0
0
0
0
0
High
26
2
62
4
52
4
66
6
0
0
206
3
26
2
53
4
52
4
66
6
0
0
197
3
8
0.7
13
0.9
0
0
26
2
0
0
47
0.7
Base
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
50%
Low
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
High
8
0.7
6
0.4
0
0
0
0
0
0
14
0.2
0
0
6
0.4
0
0
0
0
0
0
6
0.1
0
0
0
0
0
0
0
0
0
0
0
0
a "Low" = low range uncertainty estimate.
b "High" = high range uncertainty estimate.
, Table 5-7 shows the results of the analyses of projected changes in ANC. Some of the projected
changes are positive for a reason analogous to that stated above. That is, included in this analysis are
lakes having sulfate concentrations above, but within the uncertainty limit of, steady-state sulfate
concentration. Lakes with sulfate concentrations above the 95% confidence interval for sulfate
steady-state concentration were excluded from the analysis inasmuch as they very likely have great
sources of internal sulfur supply that would remain unaffected by decreases in deposition.
5-56
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
i
Ed
i
u-
aas
i
o
S*
HO
i
EsJ
z
i
i
^o
TS
o
a
a
Q
CO
O)
00
en
en
cocn i--i OTfrt^-. eooOrH^.^> 10
— — ~" — — • •— N T-4 | O) »H ^~. I —^
CM I
*? «? c-" "
- - in CM '. '. TjT • '„ '„ •_ O
encoi-ii-i lococncn <^^c-
i^CM |i^ ^HI^ I 00 rni-t |
•I ?
'. Leo1
co t- i
.en
CM co c-
I CM to
co eo eo en co co >—• ^-* ^
C4CM \ -^ 1-11-1 | ^- TT -^ 1 ^^
I I . CO I I '. CM ,-< -H . W
• . 1O . •* . CO . I I t* •*
CM co •** co o o eo >-i . . CM eo
i-i i— 1 co i-ti-i ico t- eo I CM
TV.
*~* .—x CM
it Js
i-T i-T oo «o"
i i i >o
w •* o co eoeococM CM c- c- -I1 TJ- co
I —• CM CM I •—- ^ «•* I ••—»
'. CO | | '- «0 I t '. CO
CO CO I
i"** CD C£9 ^J* O
I I I Tl* I
co o
CO I
co
CO CO CO CO
CM OJ O CO
I I I"H
,-* ^-K CM
Oi ^ O)
^l1 '.
CO 00 CO ~
CM.
90
CO
CO Cft f* **—' C^ ^ IO
CM ^4 I CN 1M 1M I ^—h y«^ «—V 1
I I - r-< I I - OS Ol C- - ^r
'. -CO . . .10 - I I CM .
00 CM CO fO CO CO (M O . . — O
CSJ-»tS CiCJ^Ci 55S*L£J
CO CM » O
1 tO -f
09 CM CO CM
I up co
CO
en
•H O "* i-t «H O CM
•* IN I CM -H
O>
eo •*• t i-«
to >o « eo
T I t- -
. . i—i CM
00 C- •* *•*
»-< N I •-»
00
<-*• <-». 10 ^—v ^— >-< en
epeort^-» oocMi-*^ ^ — ua
^* i xo eo eo i en en co I
1 f ^ I | _, 1-t »-4 1-1
f 1 .
rn CM °\ CO S <^
If
eo
CO -^ 00 i-t
M i-H W C-
1 1 •-<
eo CM Lfn
- - CO -
*"4 iO C4- ^^
*^* ^ i en
r* tN CM W i-t I Cn
I I FN | |
CM
W3 O3 ^^ ^~" CO t*" CD •*"• ^^* -^^
CM CM I U^ ^^ ^ | ^ O) OO , r -
I | '. -M | | . ^H || .CO
co ^o g^ co wi ^^ ^* 1*0 CM co CM ^^
COCO It- CMMliO — -^ ICM
US 00 •
^H JO
7
c c e e
g.S 3 3
S "S S S
as
— rj< coioencoeMeo-*co
en ^ to eo eo ^H
cegg cccc c c c g
2«33 g . 2 33 g.233
ST P 6 ^^SS ^"^PS
ojCC JS 0 ^ c ^j jp c s
^S ^ ^ ^ tf ^ ^S ^ M
2 c • * .5 S * .5 S
ss
csi
d
rr
o
-c .
Sfl^>
'? *
•M
"0-13
•a^
II
tl
3 £
ti »
II
I
«s
c2
•a 2
« $
.S e
IJ
01 3
§§"
•a 5-
> s
35
S.i
J I
> J
^ a;
u >
Ed j~
a
I
5-57
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
5.5.3 Southern Blue Ridge Province
5.5.3.1 Application
Analogous population projections were made for the SBRP. Levels of deposition were set at
100% of OLD and 120% of CLD. Estimates of wet deposition were provided by A. Olsen and estimates
of dry deposition were provided by R. Dennis.
As was the case in the NE, projections of future surface water chemistry in the SBRP focused
on steady-state conditions, but interim projections at 25 and 50 years have also been developed. As in
the NE, relationships between current sulfate and projected levels at 25 and 50 years were evaluated
for the 12 watersheds modeled in Section 5.3.4, with the best relationship based on the ratio of ASCV2
from 0-25 years to ASO*"2 from 0 years-steady state. Data from the 12 SBRP watersheds were fitted
using iinear regression, resulting in the equation
(r=0.88),
— =0.043 + 1.37
SS-SQ (SS)
where So, 825, and SS are sulfate concentrations (or percent steady state) at 0-25 years and steady
state. The same equation is used for predictions at 50 years, with S§Q and 825 substituted for 825 and
Sg, respectively. This equation was rearranged to solve for 825,
S25=2.327S0+0.043SS-1.37 .
and was used with current and steady-state sulfate data to estimate changes in SC>4~2 and ANC at 25
and 50 years for NSWS lakes and Pilot Survey Streams in the SBRP. Two systems were excluded
from the analysis due to the obvious presence of very great internal sources of sulfur.
5.5.3.2 Results
Results of the projections of acidic systems are given in Tables 5-8 through 5-10. Results of the
projections of changes in ANC are given in Tables 5-11 through 5-16. Numerous systems are
projected to lose ANC or become acidic.
5.5.4 Conclusions
(1) Predicted steady-state ANC in the Northeast - base case (100% of CLD); depending
on the F factor, an additional 0.2 to 3% of drainage lakes and reservoirs in Region 1
NSWS (estimated n = 6349) were projected to become acidic.
(2) Predicted steady-state ANC in the Northeast - expected uncertainty of base case
(100% of CLD); depending on the F factor and due to uncertainties in components of
the calculations, the uncertainty associated with the projections in (1) ranges from
0.2 to 6% of the systems projected as likely to become acidic.
5-58
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
I
I
I
(3) Projected steady-state ANC in the Northeast - 20% decrease in deposition (80% of
CLD); depending on the F factor, an additional 0.1 to 0.9% of drainage lakes and
reservoirs in Region 1NSWS (estimated n=6349) were projected to become acidic.
(4) Projected steady-state ANC in the Northeast - 50% decrease in deposition (50% of
CLD); no drainage lakes or reservoirs in Region 1 NSWS (estimated n = 6349) were
projected to become acidic.
(5) Projected steady-state ANC in the Southern Blue Ridge Province - base case (100%
of CLD); depending on the F factor, 2 to 67% of systems were projected to become
acidic.
(6) Projected steady-state ANC in the Southern Blue Ridge Province - expected
uncertainty of base case (100% of CLD); depending on the F factor and due to
uncertainties in the components of the calculations, the uncertainty associated with
the projections in (5) ranges from 1 to 77% of the systems projected as likely to
become acidic; nearly all remaining systems could lose ANC.
(7) Projected steady-state ANC in the Southern Blue Ridge Province - 20% increase in
deposition (120% of CLD); depending on the F factor, 1 to 77% of systems (drainage
lakes, reservoirs, and streams, n=169) were projected to become acidic; nearly all
remaining systems could lose ANC.
(8) Projected steady-state ANC in the Southern Blue Ridge Province - expected
uncertainty of 20% increase in deposition (120% of CLD); depending on the F factor
and due to uncertainties in the components of the calculations, the uncertainty
associated with the projections in (7) ranges from 2 to 84% of the systems projected as
likely to become acidic; nearly all remaining systems could lose ANC.
TABLE 5-8. SBRP - ADDITIONAL ACIDIC SYSTEMS (25 YEARS)
100% CLD
F Factor
0
Lakes
Streams
Total
0.2
Lakes
Streams
Total
0.4
Lakes
Streams
Total
0.7
Lakes
Streams
Total
Number
2
115
117
1
27
28
0
27
27
0
0
0
(0-2)a
(27-115)
(27-117)
(0-2)
(27-115)
(27-115)
(0-1)
(27-27)
(27-28)
(0-0)
(0-0)
(0-0)
Percent
1
4
4
0.4
1
1
0
1
1
0
0
0
(0-1)
(1-4)
(1-4)
(0-1)
(1-4)
(1-4)
(0-0.4)
(1-1)
(1-1)
(0-0)
(0-0)
(0-0)
120% CLD
Number
2
115
117
1
27
28
0
27
27
0
0
0
(0-2)
(27-115)
(27-117)
(0-2)
(27-115)
(27-117)
(0-1)
(27-27)
(27-28)
"(0-0)
(0-0)
(0-0)
Percent
1
4
4
0.4
1
1
0
1
1
0
0
0
(0-1)
(1-4)
(1-4)
(0-1)
(1-4)
(1-4)
(0-0.4)
(1-1)
(1-1)
(0-0)
(0-0)
(0-0)
a Base case (low range uncertainty estimate - high range uncertainty estimate).
5-59
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 5-9. SBRP - ADDITIONAL ACIDIC SYSTEMS (50 YEARS)
100% CLD
F Factor Number Percent
0
Lakes 21 (10-41)* 8 (4-16)
Streams 675 (616-1,185) 24 (22-43)
Total 696 (626-1,226) 23 (21-41)
0.2
Lakes 10 (7-22) 4 (3-8)
Streams 552 (485-644) 20 (18-23)
Total 562 (492-666) 19 (16-22)
0.4
Lakes 7 (1-10) 3 (0.4-4)
Streams 115 (115-481) 4 (4-17)
Total 122 (116-491) 4 (4-16)
0.7
Lakes 1 (0-1) 0.4 (0-0.4)
Streams 27 (27-27) 1 (1-1)
Total 28 (27-28) 1 (1-1)
120% CLD
Number Percent
27 (16-56) 10 (6-21)
1,173 (634-1,479) 42 (23-54)
1,200 (650-1,535) 40 (21-51)
12 (10-28) 5 (4-11)
634 (496-980) 23 (18-35)
646 (506-1,008) 21 (17-33)
10 (7-10) 4 (3-4)
382 (115-552) 14 (4-20)
392 (122-562) 13 (4-19)
1 (0-1) 0.4 (0-0.4)
27 (27-39) 1 (1-1)
28 (27-40) 1 (1-1)
1
1
1
1
a Base case (low range uncertainty estimate - high range uncertainty estimate).
TABLE 5-10. SBRP - ADDITIONAL ACIDIC SYSTEMS (100 YEARS)
100% CLD
F Factor Number Percent
0
Lakes 97 .(20-451)* 37 (7-51)
Streams 1,930 (1,415-2,179) 70 (51-79)
Total 2,027 (1,435-2,330) 67 (47-77)
0.2
Lakes 56 (13-114) 21 (5-43)
Streams 1,423 (1,099-1,930) 52 (40-70)
Total 1,479 (1,112-2,044) 49 (37-68)
0.4
Lakes 18 (10-71) 7 (4-27)
Streams 1,099 (484-1,393) 40 (18-50)
Total 1,117 (494-1,464) 37 (16-48)
0.7
Lakes 1 (0-10) 0.4 (0-4)
Streams 64 (27-92) 2 (1-3)
Total 65 (27-102) 2 (1-3)
120% CLD
Number Percent
143 (51-72) 55 (19-65)
2,179 (1,545-2,359) 79 (56-85)
2,322 (1,596-2,531) 77 (53-84)
92 (20-151) 35 (8-58)
1,930 (1,415-2,179) 70 (51-79)
2,022 (1,435-2,330) 67 (47-77)
39 (13-91) 15 (5-34)
1,393 (672-1,486) 50 (24-54) •
1,432 (685-1,577) 47 (23-52)
6 (0-15) 2 (0-6)
92 (64-560) 3 (2-20)
98 (64-575) 3 (2-19)
1
1
1
1
a Base case (low range uncertainty estimate - high range uncertainty estimate).
5-60 I
-------
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 5-11. PREDICTED CHANGE IN ANC IN STREAMS
(SRBP REGION 3A) (25 YEARS)
Level of Deposition (percent of CLD)»
I
I
I
I
I
I
I
F
o
0 2
\J, £*
0 4
V.**
0 7
V. 1
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
-36
-36
-77
-12
-28
-28
-62
-9
-21
-21
-46
-7
-11
-11
-23
-4
100%
(-24,^7)
(_24, -47)
(-66, -88)
(-1.-23)
(-20, -37)
(-19, -37)
(-53, -71)
t-0,-18)
(-15, -28)
(-15, -28)
(-39, -53)
(-0.-14)
(-7, -14)
(-7, -14)
(-20, -26)
(-0.-7)
-39
-39
-93
-24
-32
-31
-75
-19
-24
-24
-56
-14
-12
-12
-28
-7
120%
(-28, -51)
(-27, -51)
(-81, -106)
(-12, -35)
(-22, -41)
(-22, -41)
(-65, -84)
(-9, -28)
(-17, -31)
(-16, -31)
(-49, -63)
(-7, -21)
(-8, -15)
(-8, -15)
(-24, -32)
(-4, -11)
Each value is followed in parentheses by values obtained by adding or subtracting the estimated
uncertainty. The mean is computed using population weights; the median, minimum, and maximum are
computed using sample data only.
TABLE 5-12. PREDICTED CHANGE IN ANC IN STREAMS
(SBRP REGION 3A) (50 YEARS)
Level of Deposition (percent of CLD)a
F
n
- \J
0.2
0 4
V*TC
0.7
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
-86
-89
-140
-8
-68
-71
-112
-6
-51
-54
-84
-5
-26
-27
-42
-2
100%
(-72, -99)
(-76, -103)
(-127, -154)
(6, -21)
(-57, -79)
(-81, -82)
(-101, -123)
(5, -17)
(.43, _60)
(-45, -62)
(-76, -92)
(4, -13)
(-22, -30)
(-23, -31)
(-38, -46)
(2, -6)
-98
-99
-172
-38
-78
-79
-138
-31
-59
-60
-103
-23
-29
-30
-52
-12
120%
(-82, -114)
(-83, -115)
(-156, -188)
(-22, -54)
(-65, -91)
(-67, -92)
(-125,-151)
(-18, -44)
(-49, -68)
(-50, -69)
(-94, -113)
(-13, -33)
(-25, -34)
(-25, -35)
(-41, _57)
(-7, -16)
Each value is followed in parentheses by values obtained by adding or subtracting the estimated
uncertainty. The mean is computed using population weights; the median, minimum, and maximum are
computed using sample data only.
5-61
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
TABLE 5-13. PREDICTED CHANGE IN ANC IN STREAMS
(SBRP REGION 3A) (100 YEARS)
Level of Deposition (percent of CLD)a
100%
120%
o
02
\Jt£t
0 4
w»^
07
W« I
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
-129
-129
-180
-9
-103
-103
-144
-7
-78
-77
-108
-5
-39
-39
-54
-3
(-99, -160)
(-98, -160)
(-149, -210)
{22, -39)
(-79, -128)
(-79, -128)
(-119, -168)
(17, -32)
(-59, -96)
(-59, -96)
(-89, -126)
(13, -24)
(-30, -48)
(-29, -48)
(-45, -63)
(7, -12)
-162
-160
-221
-41
-129
-128
-177
-32
-97
-96
-133
-24
-48
-48
-66
-12
(-125, -198)
(_123, -196)
(-185, -258)
(-4, -77)
(-100, -158)
(-98, -157)
(-148, -206)
(_3p _62)
(-75, -119)
(-74, -118)
(-111, -155)
(-2, -46)
(-38, -59)
(-37, -59)
(-55, -77)
(-1.-23)
a Each value is followed in parentheses by values obtained by adding or subtracting the estimated
uncertainty. The mean is computed using population weights; the median, minimum, and maximum are
computed using sample data only.
I
I
I
I
TABLE 5-14. PREDICTED CHANGE IN ANC IN LAKES
(SBRP REGION 3A) (25 YEARS)
Level of Deposition (percent of CLD)»
F
Mean
n Median
Minimum
Maximum
Mean
Q 2 ' Median
Minimum
Maximum
Mean
Q 4 Median
Minimum
Maximum
Mean
n „ Median
Minimum
Maximum
-45
^0
-96
17
-36
-32
-77
13
-27
-24
-58
10
-13
-12
-29
5
100%
(-32, -58)
(_27,-53)
(-83, -109)
(30,4)
(-25, -46)
(-22, -43)
(_66, -87)
(24, 3)
(-19, -35)
(-16, -32)
(-50, -65)
(18, 2)
(_10,-17)
(-8, -16)
(-25, -33)
(9,1)
-52
-44
-113
-21
-42
-35
-90
-17
-31
-26
-68
-13
-16
-13
-34
-6
120%
(-32, -58)
(-27, -54)
(-82, -109)
(30,3)
(-25, -47)
(-22, -43)
(-66, -87)
(24,3)
(-19, -35)
(-16, -32)
(.49, -66)
(18, 2)
(-9, -17)
(-8, -16)
(-25, -33)
(9,1)
Each value is followed in parentheses by values obtained by adding or subtracting the estimated
uncertainty. The mean is computed using population weights; the median, minimum, and maximum are
computed using sample data only.
I
I
I
I
5-62
I
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
TABLE 5-IS. PREDICTED CHANGE IN ANC IN LAKES
(SBRP REGION 3A) (50 YEARS)
s
I
i
Level of Deposition (percent of CLD)a
F
Mean
_ Median
Minimum
Maximum
Mean
Q 2 Median
Minimum
Maximum
Mean
Q 4 Median
Minimum
Maximum
Mean
Q 7 Median
Minimum
Maximum
-109
-100
-178
9
-87
-80
-142
7
-65
-60
-107
6
-33
-30
-53
3
100%
(-86, -132)
(-77, -123)
(-155, -201)
(32, -14)
(_69, -105)
(-61, -98)
(-124, -161)
(26, -11)
(-51, -79)
(-4B, -74)
(-93, -121)
(19, -8)
(_26, -40)
(-23, -37)
(_47,-60)
(10, -4)
-126
-113
-214
-38
-101
-90
-171
-30
-76
-68
-128
-23
-38
-34
-64
-11
120%
(-103, -150)
(-89, -136)
(-190, -237)
(-15, -62}
(-82, -120)
(-71, -109)
(-152, -190)
(-12, -49)
(-62, -90)
(-54, -82)
(-114, -142)
(-9, -37)
(-31, -45)
(-27, -41)
(-57, -71)
(-4, -18)
Each value is followed in parentheses by values obtained by adding or subtracting the estimated
uncertainty. The mean is computed using population weights;, the median, minimum, and maximum
are computed using sample data only.
TABLE 5-16. PREDICTED CHANGE IN ANC IN LAKES
(SBRP REGION 3A) (100 YEARS)
Level of Deposition (percent of CLD)a
F
0
0.2
0.4
0.7
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
-184
-154
-361
11
-147
-123
-288
9
-111
-92
-216
7
-55
-46
-108
3
100%
(-112, -257)
(-81, -227)
(-288,^133)
(84, -61)
(-89, -206)
(-65, -181)
(-230, -347)
(67, -49)
{-67, -154)
(-49, -136)
(-713, -260)
(50, -37)
(-33, -77)
(-24, -68)
(-86, -130)
(25, -18)
-231
-190
-437
^13
-185
-152
-349
-34
-139
-114
-262
-26
-69
-57
-131
-13
120%
(-144, -318)
(-103, -277)
(-349, -524)
(45, -130)
(-115, -255)
(-82, -221)
(-280, -419)
(36, -104)
(-86, -191)
(-62, -166)
(-210, -314)
(37, -78)
M3.-96)
(-31, -83)
(-105, -157)
(13, -39)
Each value is followed in parentheses by values obtained by adding or subtracting the estimated
uncertainty. The mean is computed using population weights;, the median, minimum, and maximum
are computed using sample data only.
5-63
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
5.6 CLASSIFICATION OF RESPONSE OF NORTHEAST SYSTEMS USING DYNAMIC
WATERSHED MODELS
Previous sections (5.2 - 5.5) have dealt with (1) the principal theories and basic processes of
acidification, (2) the current and predicted status of watersheds regarding sulfate flux, (3) the current
predicted status of supply of base cations from the soil exchange complex to surface waters, and (4) the
application of a steady-state modeling approach to predict ultimate acidification of lake systems in
the NE and stream systems in the SBRP. The purpose of this section is to use integrated, dynamic
watershed models to examine changes in surface water chemistry over the next 50 years. Ten
watersheds in the northeastern United States having lakes of relatively low ANC were examined in
this analysis, using two dynamic watersheds models. Because of the current limited availability of
soils data in the SBRP, this analysis was not applied to this subregion.
5.6.1 Approach
The application of dynamic watershed models is more involved than the steady-state approach
used in Section 5.5 and requires a more thorough description of the individual models and analyses
used. The following description includes
• the dynamic watershed models used,
• forecast assumptions and limitations,
• the watershed selection,
• the watershed-lake data,
* model calibration,
• model sensitivity,
• model forecasts,
• forecast uncertainty, and
• implications for regional changes in surface water chemistry in a region with
stable, deposition sulfate concentrations.
5.6.1.1 Dynamic Watershed Models
Processes that influence the acid-base chemistry of surface water, and that were considered by
the NAS Panel (NAS 1984), are shown schematically in Figure 5-28. Although these processes may
be individually identified, discussed, and empirically represented, they do not occur in isolation and
are not independent. These watershed and lake processes are highly interactive. The observed lake
or stream response to acidic deposition represents the integrated response of numerous watershed
and lake processes controlling surface water chemistry. To predict the future response of a lake or
5-64
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
stream to acidic deposition, therefore, requires dynamic watershed models that incorporate and
integrate the important processes controlling the acid-base chemistry of surface water.
1
I
I
I
I
Watershed Ecosystem Dynamics
CHEMICAL BUDGET
WIT
W»\VNV$V
v^v V\W
OUt DIPOSITfOM
WATER BUDGET
BIOT1C STRUCTURE
PflEClCTATtOH
Miff M It
IPOTRAMSPISAl
JHJUL"
EVAPO
'(RATION
=m
9—^*^
Figure 5-28. Watershed processes thought to be important for modeling surface water
chemistry.
Source: Johnson and Thornton 1986, personal communication.
Both dynamic and steady-state models can be used to forecast changes in surface water
chemistry as a function of changes in acidic deposition. A dynamic watershed model, however,
simulates the time trend of various lake, stream, and watershed constituents, such as ANC, sulfate,
calcium, magnesium, soil base saturation, and sulfate adsorption. A steady-state model can project
conditions at only one time in the future, the time at which steady state is achieved, and does not
5-65
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
provide any indication of the changes that occurred between the initial conditions and steady state. It
is the computation of concentations and processes as a function of time that distinguishes dynamic
models from steady-state models.
Two dynamic watershed models were used to forecast the surface water chemistry for the next
50 years both at current and alternative levels of acidic deposition in the Northeast. These watershed
models follow:
• Integrated Lake-Watershed Acidification Study (ILWAS) (Chen et al. 1983); and
• Model of Acidification of Groundwater in Catchments (MAGIC) (Cosby et al. 1984).
A third model, Enhanced Trickle-Down (ETD), will be used in the DDRP but could not be applied in
the current analysis because it was being expanded based on an earlier version of Trickle-Down
(Schnoor et al. 1986).
These two models were developed by interdisciplinary scientific teams, and each modeling
team had a representative on the NAS Panel (i.e., Gherini - ILWAS; Galloway -MAGIC). Although
each model incorporates the processes considered to control the acid-base chemistry of surface water,
process resolution and detail vary significantly between them. Some of the processes included in the
two models and their spatial/temporal resolution are compared in Table 5-17. The use of multiple
models is important for the following reasons:
• the level of detail with which each process or mechanism is represented varies
between models, reflecting the relative importance of each process in the systems
for which the model was first developed and the philosophy of the scientific team;
• identification of similar key watershed parameters and processes in each model
and their relation to measured watershed characteristics provide greater
confidence in the assumptions of which factors influence the acid-base chemistry of
surface water;
• long-term data sets do not exist for model evaluation, so model accuracy and
precision for long-term forecasts is presently unknown; and
• similar predictions of watershed responses by each model, therefore, provide
greater confidence in the conclusions.
ILWAS and MAGIC were used to forecast changes in surface water chemistry over the next
50 years for 10 low ANC lake-watershed systems in the Northeast. These models are briefly
discussed below.
5-66
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
I
I
I
TABLE 5-17. COMPARISON OF PROCESSES AND RESOLUTION
IN THE MAGIC AND ILWAS MODELS
Modelsa
MAGIC
Atmospheric input
Hydrology
Weathering
Anion retention
SO4~2 adsorption
Nitrification
Denitrification
Base cation buffering
Percent base saturation
Al"1"3 kinetics
Biological
Uptake
Excretion, decomposition
Transformation
Respiration
SCV2 reduction/sediment interactions
Spatial resolution
Temporal resolution
A,LT
A,LT
LT
V
M
ILWAS
E,A
E,A
A,LT
V,H
D
*+ = Process is included in model structure
- = Process is not included in model structure
S = Episodic time scale
A = Annual time scale
LT = Long-term time scale (i.e., > 10 yr)
PT = Point
V = Vertical
H = Horizontal
M = Month
D = Day
5.6.1.2 ILWAS
The ILWAS model was developed to predict both short-term and long-term changes in surface
water chemistry due to acidic deposition. It is the most comprehensive model presently available and
incorporates most of the processes shown in Figure 5-28. The ILWAS model incorporates (1) a canopy
module to simulate forest canopy interactions with both wet and dry deposition; (2) a hydrology and
watershed soil module to route precipitation through the soil horizons and simulate soil-water
physico-chemical processes and biotic transformations; and (3) a lake module to simulate aquatic
biochemical reactions. Surface water constituents predicted by ILWAS are shown in Table 5-18. The
resolution and complexity in model output, however, is tempered by the extensive data requirements
5-67
-------
Chemical
Constituents
Model
MAGIC
ILWAS
ANC
pH
Ca+2
Mg+2
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
for model calibration and aonlication The TABLE 5-18. COMPARISON OF PARAMETERS
tor model calibration and application, l he PREmCTED BY MAGIC AND ILWAS MODELS
model has been applied to about _
25 watersheds in the Adirondack region, as
well as several other watersheds in
Wisconsin, Minnesota, North Carolina,
Wyoming, and California. Regional
assessments are being conducted as part of
the Regional Integrated Lake-Watershed
Acidification Study (RILWAS) of the
Electric Power Research Institute (EPRI)
program and through other independent
applications.
5.6.1.3 MAGIC
MAGIC is an intermediate-
complexity, lumped-parameter model
originally developed to predict the long-
term effects (e.g., decades to centuries) of
acidic deposition on surface water
chemistry. MAGIC assumes that there is a
minimum number of critical processes in a
watershed that influence the long-term
response to acidic deposition. The
watershed model simulates soil solution
chemistry and surface water chemistry and
Total Al+3
ci-
Organic Acid
TIC
Si
Algae
Zooplankton
CO2
predicts a number of water constituents (Table 5-18). Hydrologic flow of water through the soil layers
to the receiving system is simulated using a separate hydrologic model, TOPMOD (Hornberger et al.
1986). The daily flows predicted by the hydrologic model are aggregated to obtain average annual
values, which are used as input for MAGIC. MAGIC does not explicitly incorporate biotic
transformations in either the watershed or lake. The model has been applied to southeastern
streams, Adirondack lakes, and watersheds in England, Scotland, Norway, and Sweden.
5.6.1.4 Model Comparisons
ILWAS emphasizes watershed complexity in its formulations, and MAGIC is a lumped-
parameter model that incorporates a minimum number of processes required to simulate long-term
surface water acidification. Both modeling approaches have advantages and limitations and both
5-68
-------
I
I
I
i
i
i
i
i
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
modeling approaches have been used extensively and appropriately on a variety of engineering and
scientific problems. Differences in forecasts made by MAGIC and ILWAS reflect the uncertainty in
understanding the processes controlling the acid-base chemistry of surface water. Differences
between model forecasts have been used to provide estimates of the uncertainty in the conclusions
and subsequent implications for future surface water acidification.
5.6.1.5 Assumptions and Limitations
The following are primary assumptions that underly the analyses reported in this section:
• long-term acidification is defined in terms of decades (i.e., 10 to 50 years);
• sulfate is the principal acid anion controlling long-term acidification of surface
waters;
• the major soil processes controlling surface water acidification are sulfate
adsorption and base cation supply (ion exchange and mineral weathering);
• the major processes influencing long-term acidification are sufficiently known and
incorporated in the dynamic models to permit realistic predictions of the long-term
surface water chemical response to acidic deposition; and
• the 10 watershed-lake systems selected are typical of the class of northeastern
lakes considered most susceptible to acidic deposition.
These assumptions and their implications are discussed briefly below but are referenced
throughout this section, particularly when discussing and interpreting model forecasts. Each of the
models have their own inherent set of assumptions, which have been identified and documented in
reports or other literature. These assumptions, however, either are subsumed by the assumptions
above or are ancillary to the main purpose of this report and are not discussed unless a specific model
assumption affects output interpretation.
5.6.1.6 Long-term Acidification
Long-term acidification was defined as the change in the average annual lake alkalinity
concentration over .the next 50 years. Considerations of acidic episodes (i.e., hours to days) was
"outside" the scope of these analyses and were not included in this section.
5.6.1.7 Sulfur
The NAS Panel and the DDRP assumed sulfate was the principal acid anion controlling long-
term acidification of surface waters. The corollary to this assumption is that nitrate, chloride, and
organic acids were assumed to have negligible effects on long-term acidification. Based on the
discussion in Sections 3.5.2 and 3.5.4, organic acid and chloride contributions can be neglected. If
nitrate becomes a principal acid anion in the watershed, forecasts of no additional acidic lakes under
5-69
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
current or alternative deposition scenarios, however, may be in error. Nitrate breakthrough, or
nitrate saturation, has been postulated for the Adirondack region (Driscoll, personal
commmunication, 1986). Forecasts based on sulfate only could underestimate the number of
potentially acidic systems. MAGIC does not explicitly consider the nitrogen cycle, and the 1LWAS
nitrogen formulations have not been verified because data are not available.
5.6,1.8 Major Soil Processes
As discussed in Section 5.2, there was general agreement among the NAS Panel members that
sulfate adsorption and base cation supply were the major soil processes controlling surface water
acidification. Water and constituents must come into contact and chemically react with the exchange
sites in the soils, however, for these processes to neutralize acidic inputs. The ILWAS model uses a
classical Darcian approach for simulating vertical and horizontal flow through the soil horizons,
while MAGIC uses a saturation-deficit, variable-contributing-area approach for simulating
subsurface flow, including macropore flow. Both models predict similar stream discharge given the
same precipitation inputs, but might simulate very different flow paths through the soil horizons to
predict stream discharge. There currently is no coherent theory for subsurface flow through forested
watersheds. The flow paths of water through the watershed soils represents one of the greatest
sources of uncertainty in the model forecasts. Alternative flow paths might result in significantly
different forecasts of the acid-base chemistry of surface water.
Lake or stream chemistry decreases some of this uncertainty, however, by integrating the
history of water movement through the watershed. Constituent concentrations in the water entering
the stream or lake reflect the flow path through the watershed and the interaction of the acidic inputs
with soil processes. Different soil horizons have different constituent concentrations so different flow
paths will integrate different proportions of those constituents and result in different stream and lake
constituent concentrations. Similar forecasts of surface water chemistry from ILWAS and MAGIC
would indicate similar interactions of soil processes with acidic inputs and provide greater confidence
in conclusions derived from the forecasts.
5.6.1.9 Surface Water Acidification Models
The specific formulations of the processes controlling acid-base chemistry vary between
models. Regardless of the complexity in the formulations, both models are simplified representations
of the processes and interactions that are thought to occur in the watershed. Some processes may be
implicitly represented without dynamic formulations, such as the nitrogen cycle in MAGIC. Other
processes that are considered important, but for which data are lacking or sparse, might be explicitly
incorporated in the model (e.g., such as the interaction between acidic deposition and tree leaf
exudation in ILWAS). The models reflect the current understanding of the processes controlling the
5-70
-------
I
I
I
i
t
i
i
i
i
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
acid-base chemistry of surface water and, therefore, also reflect the uncertainty in this
understanding.
5.6.2 Watershed Selection
Ten watershed-lake systems were selected out of the 145 DDRP lakes in the northeastern
United States for these analyses. To provide information for policy decisions related to future effects
of surface water acidification, the sample population of 145 NE DDRP watersheds was partitioned
into a subset of watersheds that had associated lake ANC concentrations between 0 and 100 ueq L'l.
Lakes with ANC less than OueqL"1 are, by definition, already acidic, whereas lakes with ANC
greater than 100 ueq L"1 have a high ANC to protect them from becoming acidic in 50 years at current
levels of deposition (Sections 5.4 and 5.5). Lakes with ANC less than 0 or ANC greater than
100 ueq L"1, therefore, were of lower interest and not included in these analyses. Seepage lakes also
were excluded from these analyses because of minimal inputs of soil water and the high level of detail
required to determine groundwater interactions with the lake. Out of the 145 NE DDRP lakes,
66 non-seepage lakes had fall overturn ANC concentrations between 0 and 100 ueq L"1, as measured
by ELS (Linthurst et al. 1986). These 66 lakes are estimated to represent about 1443 lakes in the
restricted target population, or about 33% of the estimated 4302 watersheds in the original NE DDRP
target population.
Statistical cluster analyses were performed on these 66 watersheds to identify groups of
watersheds with similar characteristics. Cluster variables included
• areal percentage of five major soil categories in each watershed including Entisols,
Histosols, Inceptisols, Spodosols, and a category that had impervious surfaces (i;e.,
bedrock outcrops, paved areas) representative of watershed soil characteristics;
• silica concentration in the surface water, a possible indicator of weathering and
base cation supply;
• percent sulfate retention in the watershed, estimated from input/output budgets,
based on wet deposition, an indicator of sulfate steady-state or sulfate as a mobile
anion; and
• ANC as the integrator variable for lake chemistry indicating the present acid-base
status of surface water.
Ten clusters of lakes were identified that had similar characteristics. Some of these clusters,
however, contained only one or two lakes that had a unique combination of watershed characteristics.
Six clusters out of ten contained more than two lakes and the ten watersheds were selected from these
six clusters. General characteristics of lakes and watersheds in these six clusters are listed in
Table 5-19. Because lakes in the 0-50 ueq L"1 ANC category were expected to have a higher likelihood
of becoming acidic in the next 50 years than lakes with greater ANC, seven lakes were selected from
5-71
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
this ANC category and three lakes were selected in the 51 to 100 ANC category. Individual lake-
watershed systems were selected from a cluster if the system was near the centroid or cluster mean
for seven variables:
• watershed area;
• lake area;
• * water shed: lake area ratio;
• silicon concentration;
• percent sulfur retention (based on wet deposition only);
• ANC; and
• percent soil categories on the watershed (i.e., Entisols).
TABLE 5-19. CLUSTER CHARACTERISTICS FOR WATERSHED GROUPS
Cluster
No.
l
• 4
6
7
8
9
No.
Lakes
9
11
17
6
7
8
ANC Mean ANC Silicon
Range ueq L*1 Soil Categories3 Concentration!)
Low
Low
High
Low
Low
Low
29.5
28.3
74.5
30.5
20.5
24.3
Spodosols High
Impervious -
-
Entisols/Inceptisols Low
Histosols -
Inceptisols
Percent
Sulfur
Retention0
_
-
High
_
-
-
* Indicates relatively high fraction of watershed in this soil category, although other soil orders also were present,
b Indicates the mean silicon concentration for this cluster was quite different from the mean concentration.
e Indicates the mean percent Sulfur retention for this cluster was quite different from the mean concentration.
If two lakes were equally likely candidates for selection, the least disturbed watershed (as
indicated by the number of cabins in the watershed) or the watershed with soil depth data also
determined by seismic soundings was selected. Seismic soil depth estimates were made on about 10%
of the NE DDRP watersheds.
The 10 watersheds, selected for these analyses as being typical of watershed types found in the
Northeast with low ANC, represent a target population of 1246 lakes or about 30% of the watersheds
in the original NE DDRP target population. The watershed identification, general characteristics,
ANC ranges, and general location are listed in Tables 5-20 and 5-21. The location of these
10 watersheds is shown in Figure 5-29.
5-72
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
TABLE 5-20. SPECIFIC WATERSHED CHARACTERISTICS
FOR SELECTED WATERSHEDS
I
I
I
I
I
i
i
ELS
Cluster ID No. State
1
1
4
4
6
6
6
7
8
9
1A3-048
1A2-042
1A2-002
1E2-063
1C1-084
1E1-062
1E2-056
1D2-027
1A2-052
1B3-025
NY
NY
NY
ME
NH
ME
ME
MA
NY
NY
WA
(ha)
228
176
148
171
212
111
1518
88
54
122
LA
(ha)
5.1
7.2
12.9
35.6
50.8
105.5
287.2
6.2
6.9
19.4.
WALA
Ratio (
44.7
24.4
11.5
4.8
4.2
7.4
5.3
14.2
7.8
6.3
Silicon
7.04
4.1
1.1
0.4
1.8
1.75
1.4
2.2
3.3
0(?)
Percent*
S-Ret
-99
-91
-124
-67
26
-18
23
-83
-104
5
Available
ANC No. Cabins/ Seismic
(ueq L'l) Watershed Data
14.6
13.6
6.2
38.1
50.7
91.6
67.4
3.7
8.6
35.4
0
0
1
7
10
-
-
6
0
0
No
No
No •
Yes
No
No
Yes
No
Yes
Yes
* Estimates are based on wet deposition only and, therefore, are a minimum estimate of retention.
5.6.3 Watershed-Lake Data Used
Model calibrations and
simulations required data for each
watershed-lake system in the
following categories:
r
• hydroraeteorology;
• morphometry; and
* soil and water chemistry.
5.6.3.1 Hydrometeorology
Hydrometeorological data
included precipitation amount,
other meteorological variables such
as air temperature, barometric
pressure, wind speed, cloud cover,
wet and dry deposition chemistry,
and lake/stream discharge. None of
the 10 watersheds were instru-
mented for hydrometeorological
measurements, so all the hydro-
TABLE 5-21. SUMMARY OF ANC RANGE AND
GENERAL LOCATION OF SPECIAL ANALYSIS
WATERSHEDS
A.
ANC Ranges from 3.7 - 91.6 as:
Number of Lakes ANC Range
3
2
0
2
0
1
1
1
B.
0-10
10-20
20-30
30-40
40-50
50-60
60-70
70-100
General Watershed Location
Four in the Adirondacks
Three in Northeast Maine
One in New Hampshire
One in Pocono/Catskill area
One in Massachusetts Cape area
5-73
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
meteorological data were extrapolated from stations in surrounding watersheds. The lake site and
nearest instrument site, generally within 25 km and always less than 100 km, for the
hydrometeorological data are listed in Table 5-22.
1A2-042
'1A3-048 1A2-002
• 1A2-052
-027
Figure 5-29. General location of the 10 watersheds selected for applications of surface
water acidification models.
TABLE 5-22. SITE FOR HYDROMETEOROLOGICAL DATA USED TO ESTIMATE
CONDITIONS FOR THE 10 STUDY WATERSHEDS
Lake ID
No.
1A2-002
1A2-042
1A2-052
1A3-048
1B3-025
1C1-084
1D2-027
1E1-062
1E2-056
1E2-063
State
NY
NY
NY
NY
NY
NH
MA
ME
ME
ME
Rep. •
Year
1981
1982
1982
1982
1984
1984
1984
1983
1981
1983
Precipitation
Riverbank
Stamford
Stamford
Stamford
Slide Mt.
North Conway
Plymouth
Jonesboro
North Conway
Gardiner
Station
Meteorological
Syracuse
Syracuse
Syracuse
Syracuse
Binghamton
Concord
Boston
Old Town
Portland, ME
Portland
Name
Deposit Chem.
Huntington
Big Moose
Big Moose
Big Moose
Biscuit Brook
Hubbard Brook
251-NACL
Winterport
Bridgeton
Winterport
Discharge
Northwest Bay
Brook
Mine Kill
Mine Kill
Mine Kill
Biscuit Brook
Lucy Branch
Jones River
Pleasant River
Lucy Branch
Togus Stream
5-74
-------
I
I
I
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
. FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Compared with precipitation amount monitoring, the density of deposition chemistry
monitoring stations (NADP, NTN, or other stations) was sparse with a limited period of record at
each station. The availability of deposition chemistry, therefore, was used in the selection of an
average or representative year for the 50-year forecasts. This representative year is listed in
Table 5-22. Precipitation quantity data were obtained from nearby National Weather Service
stations. The precipitation stations were identified based on similar latitude, longitude and elevation
to the watershed-lake site. Further evaluation of the precipitation quantity data indicated that the
representative year selected based on the relatively short period of record at the deposition chemistry
stations was not near the norm based on a 30-yr period of record for most oif the precipitation stations.
The precipitation data, therefore, were normalized based on the 30-yr period of record so that an
average year could be used for the 50-yr forecasts. The normalized precipitation data were used to
volume-weight atmospheric chemistry concentrations for use in the models.
Updated estimates of dry deposition were not received from EPA-Research Triangle Park on
time to include in the analyses. Thus, these deposition data vary from those used in proceeding
sections. Constituent concentrations for dry deposition were based on dry bucket deposition collectors
and canopy enhancement factors estimated for the Adirondack region (Gherini, personal
communication, 1986). Dry deposition was expressed as a ratio of wet deposition. Although these
ratios probably vary across the Northeast, data were not available to verify these relations for other
northeastern subregions. The ratios of dry:wet deposition chemistry computed for the Adirondacks,
therefore, also were used at the other sites.
Meteorological data for air temperature, barometric pressure, wind speed,and cloud cover were
obtained from the stations indicated in Table 5-22. Long-term records (i.e., 30 years) at each station
permitted the development of a normal monthly record for each of the variables by selecting the
month corresponding to the norm and merging these months to obtain an average annual record for
the variables. The data at the beginning and end of each month were smoothed to avoid abrupt
transitions from month to month. Because air temperature played an important role in controlling
snowmelt periods, air temperatures from the monitoring station were adjusted, if necessary, to better
represent air temperatures at each specific lake site.
Discharge records were obtained from nearby USGS gaging stations. Gage locations were
determined based on comparable longitude and latitude to the study site. Watershed size and
elevation were emphasized in selecting representative systems for model calibration. Rather than
normalizing the discharge record and indexing the discharge to the study sites, the hydrologic models
were calibrated on the gaged watersheds and the calibrated hydrologic model was transferred to the
5-75
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
study watershed. This procedure eliminated an additional source of error that would be introduced by
indexing the discharge records to the study site and then calibrating the model on the study site.
5.6.3.2 Morphometry
Basin morphometry such as watershed area, slope, number of stream reaches, kilometers of
streams, lake area, and other basin parameters were measured on photo-enhanced Soil Conservation
Survey (SCS) watershed survey maps. An example of one these maps is shown in Figure 5-30.
Estimated depth to bedrock was obtained from SCS estimates made during the NE DDRP watershed
survey.
Lake volume and area versus volume relationships for the study lakes were obtained from
regression relationships of area versus volume determined for about 50 Adirondack lakes. Stage-
discharge relationships for the study lakes were obtained from regression analyses of stage-discharge
relationships for 15 RILWAS lakes in the Northeast (Gherini, personal communication 1986).
5.6.3.3 Chemistry
Soil chemistry data for each of the 10 study watersheds were collected and analyzed as part of
the DDRP Soil Survey in the Northeast (Section 5.2). The physical and chemical variables measured
in each soil sample are listed in Table 5-23. These data received a cursory QA/QC check but are, as
yet, unverified and unvalidated. Preliminary results indicate minor problems were associated with
the sulfate data, with apparent major problems associated with cation exchange capacity and
exchangeable cation fractions and aluminum. Further analyses are required to evaluate the severity
of these problems on the data sets and the model results.
The lake chemistry data were collected and analyzed as part of the 1984 ELS (Linthurst et al.
1986). The physical-chemical variables measured on each lake sample were discussed previously in
Section 2. The ELS data received a thorough QA/QC analysis.
These two data sources provided the chemistry data used in the model calibration and in the
50-yr forecasts.
5.6.3.4 Other Data
Bedrock geology and vegetative cover also were available for each watershed. The SCS
mapped the watershed vegetation during the NE DDRP watershed survey. An example of one of
these maps is shown in Figure 5-31. Bedrock geology for each watershed was obtained from regional
geology maps (scale, 1:250,000) and was provided by the SCS, as part of the Soil Survey.
-5-76
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
I
85/19
SCAU 1.2*000
HUNOMM&
MB a* mi
tana,
«en «•> » '•"
Figure 5-30. Example of topographic map used in the DDRP Soil Survey and model
calibration.
5-77
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 5-23. SOIL PARAMETERS MEASURED
IN THE NE DDRP WATERSHED SURVEY
Parameter Description (Units)
Watershed ID
Soil permeability (cm hr'l)
Watershed slope (%)
Depth to bedrock (m)
Watershed elevation (m)
Lake residence time (yr)
Lake area (acres)
Watershed area (acres)
Moisture, air dried (%)
Horizon thickness (cm)
Acidity, BaCl2 (meq 100 g'*)
Acidity, KC1 (meq 100 g 1)
Aluminum, CaCl2 (meq 100 g"1)
Aluminum, KC1 (meq 100 g'1)
Aluminum potential
Base Saturation, NH4C1 (fraction)
Bulk density (g cc"1)
Calcium, NH4C1 (meq 100 g'l)
Calcium, CaCl2 (meq 100 g'i)
Cation exchange capacity, NH4C1 (meq 100 g"1)
Clay(%)
Coarse fragments (%)
Potassium, NH4C1 (meq 100 g'l)
Lime potential
Selectivity coefficient, corrected
Selectivity coefficient, uncorrected
Magnesium, NH4CI (meq 100 g'l)
Sodium, NH4C1 (meq 100 g 1)
pH, H2O
pH, 0.002 M CaCl2
pH, 0.01 M CaCl2
Sand(%)
Sulfate,H2O(mgkg-i)
Sulfate,PO4(mgkgi)
5.6.4 Model Calibration
The ILWAS and MAGIC models have different formulations and subroutines and, therefore,
required different calibration procedures. The general approaches for calibrating both models,
however, were similar. First, the hydrologic model or subroutine was calibrated to predict observed
stream and lake discharge. Next, the models were calibrated to predict the observed lake
concentrations of a conservative substance such as chloride. This provided confirmation of mass
balance in the model and also confirmed the hydrologic calibration. If evapotranspiration,
interception, overland flow, or other components of the hydrologic budget were not properly
calibrated, it might be possible to achieve a flow balance, but it would be unlikely for the model to
match the observed conservative constituent concentrations. The final step in calibration was to
correctly predict the observed lake concentrations of other constituents. Calibration of the models to
predict the observed concentrations of multiple constituents provided relatively restrictive
constraints on calibration parameters. Measured variables or variables that could be calculated from
measured soil or lake attributes were incorporated directly into the model without modification.
5-78
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
I
I
I
These variables included constituents such as soil CEC, exchangeable fractions of the base cations,
base saturation, porosity, and hydraulic residence time in the lake. Although there is sampling and
measurement error in these variables, it was assumed that these variables were known and were not
varied during the calibration activities. Calibration error was incorporated entirely in those
parameters that were difficult to measure such as leaf exudation rates or mineral weathering rates.
Vegetative Cover Map
1A2-052 + Chub Lake
':
Symbol
OW
5
22
23
25
Hamilton Co., NY
Legend
Map Unit Acreage Percent
Open Wetland 30.0 20.7
Balsam Fir 16.0 11.0
White Pine-Hemlock 11.0 7.6
Eastern Hemlock 9.0 6.2
Sugar Maple-Beech-Yellow Birch 79.0 54.5
Total Land Area inWS 145,0 100.0
Lake Surface Area 1 6.0
•m
^X
^5 I
V— -\
Chub'
Lake
J
^ 25 S\&
^ .^^ JL
5w7 ^^T"22
^Vxxs,7^ Sherman Mt. Quad
7.5 min.
1:24,000
^9
43°15'
74°
74°30'
Figure 5-31. Example of a watershed vegetation map developed during the NE DDRP
survey.
5.6.4.1 ILWAS Calibration
In the ILWAS model, the watershed was partitioned into a series of subcatchments to represent
the horizontal variation in the watershed (Figure 5-32) and vertical layers to represent various soil
horizons (Figure 5-33). Although the ILWAS model incorporates considerable spatial resolution of a
watershed, watershed attributes were aggregated or averaged to obtain representative physical and
chemical parameters for model calibration and simulation. This aggregation or averaging of data
reduced the variance in the parameter values because only the average or weighted average
parameter values were used in the model.
5-79
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Woods Lake
1000 0
i . . i i t
1000 2000 3000 Feet
0 0.5 1 Kilometer
^B^^^^^^A^^^^^
Approximate mean
declination 1979
Figure 5-32. Horizontal segmentation of Woods Lake Basin in ILWAS Model.
Source: Chen etal. (1983)
5-80
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
I
I
I
Prototype
Ocm
Upper
till
(C)
Lower
till
(C)
15 cm
25cm
75cm
Model
Layer 1
Layer 2
Layer 3
Layer 4
Layer 5
Figure 5-33. Representation of vertical layers of Woods Lake Basin in ILWAS Model.
Source: Chen et al. 1983
5-81
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
The ILWAS model, which represents many of the processes shown in Figure 5-28, requires
specification of over 200 parameters, coefficients, and initial conditions for model calibration. These
values can be classified into three groups: constants, measured values, and calibration parameters.
Constant values included thermodynamic constants or other factors that did not vary from watershed
to watershed. Measured values included watershed area, base saturation, lake volume, and other
attributes that were measured or calculated from measured data at a specific site but were not varied
during model calibration. The third set of values were calibration parameters such as mineral
weathering rates, hydraulic conductivity, nitrification rates, and other parameters that were not
well-known and were modified during calibration to match the observed watershed and lake
constituent concentrations.
The number of subcatchments, soil layers, and parameters used in the ILWAS model required
between 2 and 4 weeks of time to calibrate the model for each of the 10 watersheds. Although there
were a significant number of parameters, and, therefore, significant degrees of freedom in selecting
parameter values, only certain combinations of parameter values resulted in predicted constituent
concentrations that matched observed concentrations. The interactions among parameters and
parameter combinations placed limitations on the number of feasible parameter combinations. The
calibration exercise involved identifying the set of parameters that minimized the differences
between the set of predicted versus observed constituent concentrations. The 21 constituents listed in
Table 5-18 were compared with observed lake concentrations during model calibration. A comparison
between predicted and observed ANC concentrations for ILWAS is shown in Table 5-27. ILWAS
predicted average daily constituent concentrations.
5.6.4.2 MAGIC Calibration
The MAGIC model represented the horizontal dimension of the watershed as a homogeneous
unit with no subcatchments or horizontal delineation (Figure 5-34) and the vertical dimension into
two soil layers (Figure 5-35). Watershed data for MAGIC were lumped or aggregated to provide
average or weighted average values for soil layer 1 and soil layer 2. For the 10 study watersheds, soil
layer 1 represented aggregated sample data from the A + B soil horizons and soil layer 2 represented
aggregated sample data from the C soil horizons.
5-82
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Woods Lake
I
I
I
I
I
I
1000 0
I I I I . I
1000 2000 3000 Feet
I L I
9 , , . . °,5 . . . . J Kilometer
Approximate mean
declination 1979
Figure 5-34. Representation of horizontal segmentation of Woods Lake, NY, watershed in
MAGIC Model.
Source: Chen et al. (1983)
5-83
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
B2hir
Upper
till
(C)
Lower
till
(O
Model
Layer 1
Layer 2
Figure 5-35. Representation of vertical layers of Woods Lake, NY, watershed
in the MAGIC Model.
5-84
-------
I
I
I
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
The MAGIC model includes formulations representing the major soil processes controlling soil
solution and stream chemistry. Because MAGIC does not include above-ground terrestrial processes,
there are fewer than 100 parameters, coefficients, and initial conditions that must be specified.
MAGIC required hydrologic input from a separate hydrologic model, TOPMOD (Hornberger et al.
1986). The hydrologic model, TOPMOD, calculated the fraction of runoff that moved through the
watershed as overland flow, macropore flow, shallow, or deep subsurface flow, and calculated the
average storage deficit in the upper soil horizon. TOPMOD used daily meteorological data to predict
daily average stream discharge. This hydrologic information was aggregated to average annual
values and used as input to MAGIC. TOPMOD required about 3 to 4 hr to calibrate on each
watershed. MAGIC required from 2 to 4 hr to calibrate per watershed once the daily hydrologic data
available from TOPMOD was aggregated to average annual values. MAGIC was calibrated first by
forcing the model output to match the observed values of acid anions in the lake. The next step was to
increase or decrease mineral weathering to match the observed lake concentrations of base cations.
Predicted average annual concentrations of and comparison between predicted and observed ANC
concentrations using MAGIC is shown in Table 5-26.
5.6.5 Sensitivity Analysis
One approach used to evaluate the effect of parameter uncertainty on model output is
sensitivity analysis. Sensitivity analysis involves fixing all coefficients and parameters at their
calibrated value, then varying one parameter by some nominal value such as ± 10%, and comparing
the perturbed or altered output with the original calibrated output. A large change in the output
corresponding with a small change in parameter value indicates the model is sensitive to the value of
this particular parameter. Lower confidence in the model output or forecasts might result if the
model were sensitive to parameters that are difficult to measure or that have no physical or chemical
counterparts in the watershed. Sensitivity analysis typically has been conducted on models used for
engineering applications such as wasteload allocation or eutrophication studies.
The sensitivity analyses of MAGIC and ILWAS focused on four major processes: hydrology,
sulfate adsorption, ion exchange, and mineral weathering. Parameters required for calibration of
these major processes were selected for analyses. The purpose of the sensitivity analyses was to
provide an indication of the effect of parameter variation on model output and not an extensive
evaluation of model sensitivity. Not all parameters, therefore, were studied. Parameters were
selected based on previous experience with model simulations and expected effects. In the ILWAS
model, for example, hydraulic conductivity has a greater effect on watershed hydrology and a broader
parameter range than porosity so hydraulic conductivity was selected for study. In MAGIC, the
maximum sulfate adsorption parameter, EMAX, is known to have a greater influence on model
5-85
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
output than the sulfate half-saturation coefficient so the maximum sulfate adsorption coefficient was
selected for study.
Sensitivity analyses were conducted on selected parameters in the ILWAS and MAGIC models
under two scenarios: parameter variation without recalibration, and parameter variation with
recalibration. Recalibration following parameter variation represents an important distinction
between sensitivity analyses conducted on surface water acidification models and on other
engineering models. Surface water acidification models must maintain charge balance for accurate
predictions of surface water chemistry. Arbitrarily varying a parameter by some nominal percentage
can result in other parameter values exceeding realistic ranges with respect to the fixed combination
of parameter values and subsequent charge imbalance in the forecasts. Unfortunately, the range of
feasible parameter values cannot be determined a priori. Without recalibration, parameter values
were varied by about 10 percent. This might be a reasonable range for most parameters but
additional analyses of model behavior are required before this can be determined.
Sensitivity analyses are typically evaluated for steady-state models for which time is not a
variable or for models for which the time frame of interest is a few months or years. The time frame
for surface water acidification models, however, is decades. Small changes in mineral weathering
rates might not become apparent in model output for 10-20 years. Because of time constraints.long-
term model sensitivity was restricted to a limited number of parameters and analyses.
5.6.5.1 MAGIC
Sensitivity analyses of the MAGIC model were evaluated over a 200-yr period, beginning in
1841 and ending in 2031. Selected parameters were varied individually by ±10% with the other
parameters remaining at their nominal calibrated values. MAGIC was run for a 200-yr period to
assess the long-term effects of
these perturbations on model
output. Five parameters or
groups of parameters were
selected for study. Brief descrip-
tions of these parameters are
listed in Table 5-24. The para-
meters selected for sensitivity
analyses, range of variation, and
percent change in model output
are shown in Table 5-25! MAGIC
was sensitive to changes in all WEATH (I)
TABLE 5-24. SENSITIVITY PARAMETERS STUDIED
FOR THE MAGIC MODEL
Parameter
Description
PMAC
D, depth
EMAX
SALCA, SALMG
SALNA, SALK
Proportion of deposition that moves as
macropore flow from the atmosphere to the
lower soil layer without contacting the upper
soil layer
Soil depth of both layers
Maximum sulfate adsorption capacity of
both soil layers
Logarithmic values of selectivity coefficients
for each base cation in both soil layers
five parameters or groups of
Weathering fluxes of the base cations (I) -
Ca, Mg, Na, K - in both soil layers
5-86
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
I
I
I
parameters. The greatest sensitivity was to mineral weathering estimates, followed by soil depth,
maximum sulfate adsorption capacity, and hydrologic routing (i.e., macropore flow), with least
sensitivity to selectivity coefficients.
TABLE 5-25. SELECTED PARAMETER SENSITIVITY RESULTS
FOR THE MAGIC MODEL*
Lake
Constituent
1981
Woods Lake
ANC
H+
Panther Lake
ANC
H+
Clear Pond
ANC
FT
2031
Woods Lake
ANC
H+
Panther Lake
ANC
H+
Clear Pond
ANC
H+
Initial
Values
(ueqL-i)
-15.4
16.0
116.1
0.1
108.8
0.1
-19.2
19.0
120.6
0.1
102.5
0.1
PMAC Depth
-10% +10% -10%
-1.3 * 1.3
1.2 * -1.2
* * 1.6
* * *
* * *
* * *
* * 3.1
* * -3.1
* * 1.7
* * *
* * *
* * *
+ 10%
-2.6
2.5
-2.0
*
*
*
-4.7
4.7
-2.0
*
*
*
E max
-10% +10%
2.0 -3.2
-1.9 3.1
1.8 -2.0
* *
* *
* *
4.2 -4.2
-4.2 4.2
1.7 -1.8
* . *
* *
* • *
Select Weath
> -10% +10% -10%
* -1.3 3.2
* 1.3 -3.2
* * 12.6
* * *
* * . 13.4
* * *
* * 3.6
* * -3.6
* * 14.0
* * *
* * 13.7.
* * *
+ 10%
-3.2
3.2
-13.1
*
-13.9
*
-3.1
3.1
-14.6
-
-13.8
-
a Percent change in ANC and hydrogen ion concentrations following a 10% change in five selected model parameters.
Simulations were initiated in 1841, calibrated on 1981 data, and forecast to 2031.
* No change or > 1.0 percent change.
5.6.5.2 ILWAS
The ILWAS model represents a greater number of watershed processes, including biotic
processes, than the MAGIC model. ILWAS, therefore, has a greater number of parameters and
coefficients than MAGIC. There was insufficient time to perform a formal sensitivity analysis on the
ILWAS model. A qualitative analysis was performed, however, on selected parameters associated
with processes controlling surface water acidification such as hydrology, sulfate adsorption, and base
cation supply. Because the ILWAS model simulates both short- and long-term surface water
chemistry charges, the time scale over which parameter variations are evaluated is important.
5-87
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
For short time frames of 1-2 years, the most sensitive parameters were those related to
watershed hydrology such as permeability and soil depth and ion exchange. The model was
insensitive to mineral weathering in short-term simulations because ion exchange was sufficient to
provide base cations for neutralization of mobile acid anions. Model output was moderately sensitive
to sulfate adsorption in the short-term simulations. An example of short-term sensitivity to soil
depth is illustrated in Figure 5-36. The soil or till depth was reduced to the minimum estimated
depth to bedrock from the measured depth in Panther Lake, resulting in a lake with positive
alkalinity predicted to become acidic.
I
I
Panther Til I at SCS
D Base Case
-25
-50
I
I
I
I
j ft
I
Date
1980
1981
Figure 5-36. Sensitivity of ILWAS Model to soil depth. The lower curve illustrates the
change in ANC predicted when the minimum estimated soil depth was used versus the
measured depth.
I
5-88
I
-------
I
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Long-term forecasts are sensitive primarily to watershed hydrology, mineral weathering, and
sulfate adsorption. Mineral weathering is the source of exchangable base cations so over longer time
frames, the rate at which these base cations are supplied to the exchange sites through mineral
weathering determines the capacity of the watershed to neutralize acid anions.
5.6.5.3 Considerations
The two models have different formulations and different underlying assumptions so a direct
comparison of parameters and parameter sensitivity is not appropriate. Both models, however, were
generally sensitive to parameters that affected major processes considered to control surface water
acidification over longer time scales of decades. These parameters were related to subsurface flow or
the movement of water through the watershed, sulfate adsorption capacity, and mineral weathering.
If these are the primary processes controlling long-term surface water acidification, perturbing
parameters affecting these processes would be expected to change the model output. This general
sensitivity was demonstrated for both models.
5.6.6 Model Forecasts
Model forecasts of changes in surface water chemistry over the next 50 years as a function of
acidic deposition were conducted on the 10 northeastern watersheds with MAGIC and on one of these
watersheds with ILWAS. The annual hydrometeorological sequence for the normal or typical year
was repeated each year for 50 years during the 50-yr forecasts. .
Three deposition scenarios were used in these forecasts: constant acidic deposition at current
levels (100% CLD); 25% increase in acidic deposition (125% CLD); and a 50% reduction in acidic
deposition (50% CLD). The deposition gradient across the Northeast from Maine to the Adirondacks
was preserved by using current deposition occurring at each of the 10 watersheds. Deposition
included both wet deposition chemistry and estimates of dry deposition chemistry. The 25% increase
and 50% reduction in deposition represented an increase or reduction in total deposition (i.e., wet and
dry deposition estimates), respectively. Generation of power in the Northeast is presently below
maximum capacity. The increased deposition reflects the increase in emissions that might result if
power production were increased to maximum capacity.
Deposition was increased 25% or reduced 50% within the first five years of the 50-yr forecast.
This increase/reduction sequence was established solely to maintain model stability and not to
represent any expected or projected increases or reductions in emissions.
5-89
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
5.6.6.1 MAGIC Forecasts
When interpreting the MAGIC forecasts, it is important to remember the forecast values or
constituent concentrations represent average annual estimates and reflect increases or decreases in
average annual estimates.
Constant Deposition (100% of C1D)
MAGIC forecasts for ANC, pH,
and sulfate were made assuming
constant deposition for the next 50 years.
The forecasts are graphically shown in
Figures 5-37 through 5-39 (ANC, pH, and
sulfate, respectively) with constituent
values listed at 10-yr increments in
Table 5-26.
Chub Lake (1A2-052), was forecast
to become acidic (i.e., ANC <0 ueq I/1) in
about 30 years at current levels of
deposition, losing a total of about
16 ueq L"1 ANC over the 50-yr period or
about 0.3 ueq L"1 yr'i ANC on an annual
average basis. The average loss of ANC
over the 50-yr period for all 10 lakes was
about 5 ueq L'l or about 0.1 ueq L'l yr'l.
Five lakes, however, lost more than
5 ueq L'l ANC over the 50-yr period
(Chub Lake, North Branch Lake 1A2-
042, Grass Pond 1A3-048, Upper Beech
Pond 1C1-084, and Long Pond 1E1-062), averaging about 8 ueq L'l of ANC for the 50-yr period.
Two lakes, Chub Lake (1A2-052) and North Branch Lake (1A2-042), that had pH values at or
above 6.0 in 1984, were forecast to have pH values decline below 6.0 by year 2034. The pH in Chub
Lake was forecast to decline from about 6.0 in 1984 to 5.4 in 2034. North Branch Lake had a pH
decline from 6.1 to 5.8 during the 50-yr period. Two lakes, St. John Lake (1A2-002) and Sandy Pond
(1D2-027), had initial pH values around 5.7 and 5.8, respectively, and were forecast to have less than
a 0.05 pH unit decline over 50 years. The other six lakes were forecast to maintain pH values greater
than 6.0.
-10
1984 1994
2004 2014
Year
2024 2034
Figure 5-37. Fifty-year ANC forecasts with the
MAGIC Model for 10 NE watersheds for the
100% CLD.
5-90
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
200
I
I
I
I
I
I
I
4.75
1984
1994
1I2-4SI
1CI-U4
ItZ-CU
UJ-4M
2004 2014
Year
2024
20 4 1984 1994
2004 2014
Year
2024
2034
Figure 5-38. Fifty-year pH forecasts
with the MAGIC Model for 10 NE water-
sheds for the 100% OLD.
Figure 5-39. Fifty-year sulfate forecasts
with the MAGIC Model for 10 NE water-
sheds for the 100% CLD.
Those lakes that had the greatest ANC decreases over the 50-yr period generally also had the
greatest increases in lake sulfate concentration. Those lakes that lost more than 5 ueq L'l ANC over
the 50-yr period had an average increase in lake sulfate concentration of 22 ueq I/1 over 50 yr, and
lakes that lost less than 5 ueq L'l ANC in 50 years had an average increase in lake sulfate
concentration of 10.5 ueq L'l. For example, Chub Lake (1A2-052), the lake that became acidic had the
greatest increase in lake sulfate concentration (i.e., approximately 33 ueq L'l). The average increase
in lake sulfate concentrations for all 10 lakes was about 14 ueq L" 1.
There was a direct relationship between ANC loss and sulfate concentration increase
(AANC = 4.02-0.54.ASO4'2; r2 = 0.90). There was a loss of about 5.4 ueq L'l lake ANC for each
10 ueq L'l increase in lake sulfate over the 50-yr period. This is consistent with the theory presented
in Section 5.2 and the steady-state analyses presented in Section 5.5. The relationship corresponds to
an average F-factor for the 10 watersheds, calculated from A (Ca+2 + Mg*2)/AS
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 5-26. 50-YEAR FORECASTS USING MAGIC WITH 100% CLD
ELS
Lake and Constituent
Constituent Value
1A2- 002 St. John Lake
ANC(p.eqL-l)
pH
6.2
5.5
133.9
Forecast Year
1984
6.09
5.80
133.26
1994
' 5.62
5.78
131.95
2004
5.62
5.78
131.37
2014
5.62
5.78
131.10
2024
6.09
5.80
130.95
2034
6.56
5.82
130.87
1A2-042 North Branch Lake
ANC(yeqL-i)
PH
1A2- 052 Chub Lake
ANC (jieq L"1)
pH
1 A3- 048 Grass Pond
ANC(iieqL-l)
pH
S042
1B3-025 Trout Lake
ANC(p.eqL'l)
pH
13.6
5.7
116.6
8.6
5.4
112.9
14.6
5.5
126.0
35.4
6.6
93.7
f Cf -084 upper Beech Pond
ANC(jieqL-l) 50.7
pH 6.8
S04'2 (peq L-l) 79.4
W2- 027 Sandy Pond
pH
1E1-062 Long Pond
ANC(neqI/l)
pH
1E2-056 Peabody Pond
ANC(peqL'i)
pH
1E2-063 Kalers Pond
ANC(peqL'l)
pH
3.7
5.1
121.6
91.6
. 7.2
64.3
67.4
7.0
70.9
38.1
6.6
71.4
13.65
6.10
116.60
8.64
5.95
112.88
14.61
6.10
126.04
35.47
6.44
93.69
50.98
6.61
79.41
3.64
5.71
121.52
91.48
6.86
64.25
67.64
6.74
70.92
38.07
6.49
71.42
11.69
6.05
122.79
5.27
5.83
120.62
13.11
6.07
130.30
33.85
6.42
99.06
49.83
6.60
83.62
2.76
5.67
116.60
90.46
6.86
67.55
67.65
6.74
72.24
36.33
6.47
75.09
10.08
6.00
128.46
1.68
5.70
127.94
12.31
6.05
134.36
32.30
6.40
103.46
48.71
6.59
87.56
2.29
5.65
114.56
89.44
6.85
70.72
67.66
6.74
73.80
36.34
6.47
78.46
8.46
5.94
133.55
-1.42
5.58
134.71
11.15
6.02
138.12
32.31
6.40
106.99
47.60
6.58
91.24
2.29
5.65
113.69
87.42
6.84
73.79
67.67
6.74
75.51
34.68
6.45
81.52
6.77
5.88
137.99
^4.43
5.48
140.80
10.01
5.99
141.49
30.82
6.38
109.75
46.52
6.57
94.60
2.29
5.65
113.32
87.44
6.84
76.74
66.13
6.73
77.30
34.69
6.45
84.29
5.70
5.84
141.77
-6.94
5.40
146.13
9.27
5.97
144.44
30.82
6.38
111.88
45.45
6.56
97.64
2.29
5.65
113.17
85.46
6.83
79.54
66.15
6.73
79.13
34.70
6.45
86.75
5-92
-------
I
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
5.6.6.2 ILWAS Forecasts
ILWAS forecasts for St. John Lake (1A2-002) assuming 100% CLD are listed in Table 5-27.
Time did not permit forecasts for the other nine systems.
St. John Lake was forecast to decrease from an initial ANC of 6.2 ueq L"l to 4.7 ueq L"1 over the
next 40 years. This represents a decrease of about 1.5 ueq L*1 over 40 years, but the lake was
estimated to lose this ANC over the next 10 years and then approach steady state.
The pH was forecast to decrease from an initial pH of about 5.5 to 5.25 over the next 10 years
and then gradually increase to about 5.4 after 40 years.
St. John Lake was forecast to be in approximate steady state with sulfate, decreasing from an
initial lake sulfate concentration of about 134 ueq L'1 to 126 ueq L"1 after 40 years.
TABLE 5-27. 50-YEAR FORECASTS USING ILWAS WITH 100% CLD
ELS
Lake and Constituent
Constituent Value
1A2-002 St. John Lake
ANC(ueqL'i)
pH
S04-2(ueqL-i)
6.2
5.5
133.9
Forecast Year
1984
6.2
5.54
134.0
1994
3.78
5.25
136.0
2004
3.84
5.33
130.9
2014
4.28
5.37
127.9
2024
4.73
5.42
126.0
2034
-
5.6.6.3 Forecast Comparisons
Although the forecast for only one lake can be compared, the forecast for St. John Lake from
both models were similar. Both models forecast minimal changes in ANC and decreased lake sulfate
concentrations. The ILWAS model forecast a greater initial decrease in pH with a gradual increase
over the 40-yr period. Both models forecast an initial, relatively rapid change in lake chemistry
during the next 10 years, with the lake gradually approaching steady state during the next 30 to
40 years. Model forecasts for St. John Lake using the ILWAS and MAGIC models were considered
comparable.
5.6.7 Regional Estimates
Both the NSWS and DDRP have underlying statistical frames that permit estimates of the
proportion of northeastern lakes in the target population that might become acidic within the next
50 years based on model forecasts. Statistical algorithms were developed to extrapolate from the
10 watersheds in these analyses and provide weighted estimates of the proportion of lakes in the
restricted target population (i.e., lakes with ANC from 0-100 ueq L'l) of 1248 lakes (30%) expected to
show similar changes in surface chemistry over the next 50 years. The cluster number, lake
5-93
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
identifier, and number and proportion of lakes in the target population represented by this lake,
listed in Table 5-28 for each of the 10 watersheds.
TABLE 5-28. NUMBER AND PROPORTION OF LAKES
REPRESENTED BY EACH OF THE 10 STUDY WATERSHEDS
Cluster No.
1
4
6
.7
8
9
Lake ID
1A2-042
1A3-048
1A2-002
1E2-063
1C1-084
IE 1-062
1E2-056
1D2-027
1A2-052
1B3-025
Number of
Representative Lakes in
Target Population
91
' 91
97
97
149
149
149
103
143
179
1,248
Proportion of Target
Population Lakes
0.07
0.07
0.08
0.08
0.12
0.12
0.12
0.08
0.12
0.14
1.00
Uncertainty or confidence limits on the 50-yr cumulative frequency distributions for ANC, pH,
and sulfate values were obtained using a binomial estimate of variance. Although the binomial
formula used assumes simple random samples, the weights for each lake were reasonably unform so
the uncertainty estimation procedure was appropriate. Broad uncertainty estimates reflect the small
sample size used for regional extrapolation.
5.6.7.1 Constant Level of Deposition (100% CLD)
Approximately 143 lakes (12%) of all lakes in the target population having ANC
concentrations between 0 and 100 ueq L"1 were forecast to become acidic (i.e., ANC <0) within the
next 50 years (Figure 5-40a). Uncertainty estimates about the ANC cumulative frequency
distribution after 50 years indicate the number of acidic lakes might range as high as 861 (69%)
(Figure 5-40a). An additional 182 (14%) lakes or a total of about 525 lakes (42%) were estimated to
have annual average ANC concentrations below-10 ueq L"1 after 50 years. The median loss of ANC
was estimated to be about 5 ueq L'l over 50 years.
About 143 lakes (12%) also were forecast to have pH values decrease to less than 5.5 in 50 years
(Figure 5-40b). Uncertainty estimates for the pH cumulative frequency distribution indicate as many
as 349 lakes (28%) might have pH values less than 5.5. The median change in equivalent pH units is
about 0.1 units over 50 years.
Although none of the lakes were estimated to have sulfate concentrations greater than
150 ueq L'l within 50 years, about 711 lakes (57%) were estimated to have sulfate concentrations in
excess of 100 ueq L*1 after 50 years (Figure 5-40c). The number of lakes with sulfate concentrations
5-94
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
I
I
I
> 100 ueq L"i also was estimated to be as low as 387 lakes (31%) (Figure 5-40c). The median increase
in sulfate concentration was estimated to be about 15 ueq L'1 over 50 years.
5.6.8 Watershed Attributes
5.6.8.1 Low (0-50 ueq L'l) ANC
Watersheds
Two clusters had watersheds that
were forecast to become acidic within 50
years assuming either 100% or 125% of
CLD. These watersheds were North
Branch Lake (1A2-042), Chub Lake
(1A2-052), and Grass Pond (1A3-048).
These three watersheds had several
common attributes. First, the three
watersheds were located in the
Adirondacks, a region currently
receiving some of the highest acidic
deposition loading in the Northeast.
Second, the three watersheds were small,
with Chub Lake (1A2-052) having the
smallest area (approximate 60 ha) of any
of the 10 watersheds. Third, the three
watersheds had shallow estimated soil
depths to bedrock, ranging from
aggregated depths of 1.7 to 2.2 m.
Fourth, these three watersheds had lakes
with low current ANC concentrations.
Fifth, the three watersheds had
relatively low base saturations in the A,
B, and C soil horizons. The shallow
watershed depths, in combination with
low base saturation, result in a relatively
small watershed capacity to neutralize
acidic inputs. Sixth, these watersheds
and their respective clusters had
relatively high percentages of soils with
1.00
Si 0.80
§
-g 0.60H
C
- 0.40-
2
3
3
u
0.20
0
1.00.
080>
0.60'
0.40-
0.20
— CDF
—• Uncertainty Limit
0 50
ANCfceqL'1)
100
6
pH
50
100
150
200
Figure 5-40. Cumulative frequency distribu-
tion to 50-yr ANC at 100% CLD (A), 50-yr pH at
100% CLD (B), and 50-yr sulfate at 100% CLD
(C).
5-95
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
low sulfate adsorption capacity (i.e., Hiatosols and Spodosols). These watersheds were near steady
state with respect to sulfate inputs based on annual input/output sulfate budgets. Finally, based on
simulated watershed hydrology using TOPMOD, all three lakes had relatively high' proportions of
subsurface flow moving through the upper soil layer directly into the lake. This upper soil layer had
lower sulfate adsorption capacity, substantially lower base saturation, and lower mineral weathering
rates than the deeper soil layer and, therefore, lower capacity to neutralize acidic inputs.
5.6.8.2 High (0-50 ueq L i) ANC Watersheds
The cluster containing watersheds with high ANC lakes represents the antithesis of the low
ANC watershed clusters. The three watersheds in this cluster were Upper Beech Lake (1C1-084),
Long Pond (1E1-062), and Peabody Pond (1E2-056). First, these watersheds currently receive
relatively low acidic deposition loading. Second, these were three of the largest watersheds (i.e.,
average area approximately 675 ha) of the 10 study systems. Third, all three watersheds had
estimated soil depths to bedrock that were at least 1 m deeper (i.e., >3 m) than the three low ANC
watersheds. Fourth, these watersheds had lakes with the highest current ANC concentrations.
Fifth, these watersheds generally had higher base saturations in all three soil horizons, A, B, and C.
Sixth, based on the sulfate input/output budgets, watersheds in this cluster are currently retaining
sulfate. Finally, based on simulated watershed hydrology, all three watersheds had a high proportion
of subsurface flow moving through both the upper and lower soil horizons before entering the lake
and, therefore, had a greater capacity to neutralize acidic inputs.
5.6.9 Model Forecast Uncertainty
There are uncertainties associated with any modeling study, but consideration of these
uncertainties becomes particularly relevant when deterministic models are used in forecasting future
effects. A deterministic model, such as MAGIC or ILWAS, provides a single output trajectory
through time with no estimate of error or uncertainty about the predicted concentrations of ANC, pH,
or sulfate. Sampling error, measurement error, and other estimates of uncertainty are standard
QA/QC components of field and laboratory studies, the data used to calibrate these models and
provide the inputs all have inherent error or uncertainty that should be propagated through the
models and associated with the forecasts. Unfortunately, deterministic models do not have the
capability to propagate this uncertainty. Therefore, error or uncertainty estimates must be
implicitly, subjectively, or qualitatively associated with the output. The following section presents
some of the major sources of uncertainty associated with the model forecasts and regional estimates.
This presentation is not intended to reduce the usefulness of the model forecasts but rather to place
these forecasts in proper perspective. Some sources of uncertainty include meteorological and
5-96
-------
I
I
I
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
deposition inputs, initial conditions, watershed and lake data, model processes and parameters, and
regional estimation procedures.
3.6.9.1 Meteorological and Deposition Input Uncertainty
Input uncertainty in these areas is associated with two factors: (1) a constant meteorological
record of one year and (2) dry deposition estimates. One year of normalized meteorological data was
repeated 50 times to generate the 50-yr period of record for model forecasts. Climatic weather
patterns with wet and dry periods, however, are known to influence patterns in surface water
chemistry. ANC concentrations, for example, are typically lower in wet years with higher wet
deposition loading and higher during dry years with lower wet deposition loading. These long-term
climatic patterns were eliminated by repeating the same deposition record for 50 years. The long-
term meteorological patterns associated with the 10 watershed sites are unknown but might affect
whether the watershed becomes acidic in 50 years. If the next 50 years were wetter than normal,
more lakes might become acidic than were forecast by the models.
Dry deposition enhancement factors were based on dry bucket estimates for the Adirondacks.
The NADP network discontinued the dry bucket collection system because of the low reliability and
uncertainty of the dry deposition estimates. Dry sulfate deposition estimates have received the most
attention, ranging from 20-100% of wet sulfate deposition measurements, but dry cation deposition
estimates are virtually unknown! Dry deposition also is expected to have a spatial gradient across
the Northeast similar to wet deposition gradients. The lack of adequate data resulted in the use of a
constant set of dry deposition enhancement factors for all 10 watershed forecasts. Limited analyses
indicated that the 140-yr hindcast was not sensitive to minor changes in the historical dry deposition
factors, but the effects on the 50-yr forecast are unknown. Spatial gradients for dry deposition also
were incorporated strictly as a function of the wet deposition spatial patterns.
5.6.9.2 Initial Conditions for the Forecasts
Although the rate of ANC decrease over the 50-yr forecast varied among sites, the rate of ANC
decrease at each site was linear through time. The ANC concentration in 1984, therefore,
determined, in part, whether the lake became acidic in 50 years. For example, North Branch Lake
and Grass Pond, lakes typical of cluster 1, had an average annual ANC decrease of 0.13 ueq L"1 yr'i.
With initial ANC concentrations of 13.6 and 14.6 ueq L"1, respectively, these two systems were not
forecast to become acidic for over 100 years. Another lake in this cluster, however, had an ANC
concentration of 2.9 ueq L'1. If the average annual ANC depletion rate for this cluster was applicable
for this lake, it would become acidic in about 20 years and the estimated proportion of acidic lakes in
the target population would increase. Estimates of the uncertainty in the 1984 ANC concentrations
for the cluster were calculated using the ANC mean and one standard deviation, assuming the
5-97
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
average annual ANC decrease per year was applicable. Forecasts of the uncertainty in years to
0 ANC for each cluster are listed in Table 5-29.
TABLE 5-29. UNCERTAINTY IN FORECASTED YEARS TO 0 ANC
FOR EACH WATERSHED CLUSTER
Forecast Years to 0 ANC
Cluster Avg. Cluster Std.
No. ANC (ueq L'l) Dev.
1 29.5
4 28.3
6 74.5
7 30.4
8 20.5
9 24.3
20.2
26.0
20.7
35.3
12.5
18.2
Constant Deposition
-1 S.D.
70
80
618
Qa
26
66
Mean
222
976
856
1128
66
262
+ 1S.D
374
1871
1094
2434
106
457
25% Deposition Increase
-1 S.D.
25
22
219
Q*
15
23
Mean
81
270
304
470
38
90
+ 1 S.D.
136
518
388
1014
61
158
a Standard deviation was greater than the mean so the cluster was acidic initially.
5.6.9.3 Watershed and Lake Data
Uncertainty in the watershed and lake data will be illustrated using two examples: estimated
depth to bedrock in the watershed and cation/anion ratios for the lake chemistry.
The maximum depth to which watershed soil pits were excavated for sampling in any
watershed was the standard sample depth of 1.5 m. If bedrock was not encountered before this 1.5m
depth, the depth to bedrock was estimated by SCS. The range in depth classes below 1.5 m, however,
was quite large, i.e., 2-5 m, 5-30 m, and >30 m. This range in depth classes is proportional to the
uncertainty in the soil depth estimate. Sandy Pond (1D2-027), for example, had an aggregated
watershed estimate for soil depth of 17.5 m, which represents the average of the depth class, 5-30 m.
While the soil depth at Sandy Pond is represented in the model as 17.5 m, the actual soil depth might
be as shallow as 5 m or as deep as 30 m and could affect model forecasts. The primary difference .
between Woods Lake being acidic and Panther Lake being alkaline in the Adirondacks was the depth
of the till or estimated soil depth to bedrock (Gherini et al. 1985). Woods Lake had an aggregated
watershed soil depth of 2 m while Panther had an aggregated watershed soil depth of 24 m. Basins
with deep soils are expected to have more exchangeable base cations and weatherable minerals for
neutralization of acidic inputs as the water percolates through the soil (Gherini et al. 1985).
Low ionic strength, dilute concentration water samples, typically collected from lakes
potentially susceptible to acidic deposition, are difficult to measure analytically because constituent
concentrations are generally at, or near, detection or decision limits for the analytical test.
5-98
-------
I
I
I
I
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Cation/anion balances of ± 15% are within acceptable QA/QC criteria for samples reflecting this
imprecision in analytical methods. Two of the lakes, Sandy Pond (1D2-027) and St. John Lake (1A2-
002), had cation deficits of 3% and 12%, respectively, that were within QA/QC criteria and low, but
positive, ANC concentrations (3.7 and 6.2 ueq I/1, respectively). Because MAGIC and ILWAS models
require charge balance, the models included additional hydrogen ions to complete the charge balance,
which results in the calculation of negative ANC concentrations (i.e., ANC = Sum of Base Cations -
Sum of Acid Anions). To compensate for the charge imbalance, chloride was removed from the system
and, if necessary, sodium was added to the system to generate the observed ANC. The effect of these
procedures on the 50-yr forecast is unknown.
5.6.9.4 Model Processes and Parameters
Model calibration and parameterization, obviously, are influenced by the process formulation.
Uncertainty in hydrologic processes has been discussed previously (Section 5.6.3.3). Mineral
weathering rates (i.e., base cation supply) represent both a major source of uncertainty and
parameters to which the models are sensitive. An example of a 10% change in weathering rates is
shown in Figure 5-41, indicating that both the hindcast and forecast MAGIC simulation results are
sensitive to mineral weathering rates. The influence of hydrogen ion concentrations on weathering
rates, the formation and dissolution rates of primary and secondary minerals, the mineralogy of the
watershed, and applicable mineral weathering rates all contribute to the uncertainty in specifying
model weathering parameters. Mineral weathering rates, therefore, were treated as a calibration
parameter within the models. Measured data and parameters calculated from measured data were
used directly in the model. Weathering rates were then selected to achieve a match between observed
watershed and lake constituent concentrations. The hindcast calibration procedure in MAGIC
incorporated the slow weathering rates by simulating the past 100 years so subtle changes in model
outputs dependent on weathering or base cation resupply could be observed. Calibration procedures
for ILWAS used cation/anion ratios, silica concentrations, and other constituent concentrations to
compute mineral weathering rates. Reasonable estimates of mineral weathering were used in the
models, constrained by observed water chemistry data, and the range of currently accepted rates, but
weathering estimates remain one of the primary sources of uncertainty in model applications.
5.6.9.5 Regionaiization Estimates
Ten watersheds represent a small sample size for the current analyses and extrapolation
approach. Theoretical confidence intervals about the forecast ANC, pH, and sulfate values
(Figures 5-40a,b, and c) reflect this uncertainty. Each sample watershed in these analyses represents
about 100 lakes in the target population. Uncertainty in results from the sample watersheds,
therefore, is magnified about 100 times in estimating results for the target population. Future
5-99
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
estimates for the northeast population will be improved significantly in the DDRP because" the
sample size will be 15 times larger.
60
1844
1884
1924
Time (years)
2004
Figure 5-41. MAGIC forecasts of ANC resulting from sensitivity to a 10% change in
weathering rates.
5.6.10 Regional Implications
The primary area of disagreement in the NAS Panel report was with respect to the rate of
acidification in regions where sulfate concentrations were at or near steady state. One hypothesis
was that acidic inputs would continue to titrate bases from soils, and lakes would continue to become
acidic. Another hypothesis was that base cation supply was approximately equivalent to acidic
inputs, and little change in lake alkalinity would occur in the future.
Model forecasts of the effects of acidic deposition on surface water chemistry over the next
50 years under current deposition levels (100% CLD) follow:
• The net change in average annual lake alkalinity is small, with weighted
population estimates for average annual ANC decreases of 0.1 ueq L'l yr~* and an
estimated average ANC loss of 5 ueq L"i for the 50-yr period. This supports the
5-100
-------
I
I
I
I
1
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
hypothesis that the rate of base cation supply might be similar to the rate of acidic
inputs.
* Although the model forecasts indicate the ANC loss rate is small, lakes with ANC
concentrations less than 10 ueq L"1 might become acidic within 50 years because of
± 10 ueq L'l uncertainty in ANC measurements.
• Approximately 143 lakes (12%) in a target population of 1248 northeastern lakes
with ANC between 0 to 100 ueq L"1 were forecast to become acidic within 50 years.
This estimate might range as high as 861 (69%) lakes.
• The watershed attributes associated with the lakes forecast to become acidic
include
- relatively high acidic deposition inputs;
- small watershed areas;
- shallow watershed soils (i.e., aggregated depth <2m);
- soils with low sulfate adsorption capacity;
- low initial ANC concentrations;
- low soil base saturation; and
- shallow subsurface flow paths.
It is the combination of all these attributes and not a specific watershed
characteristic that contributes to an acidic lake forecast.
• Approximately 143 lakes (12%) were forecast to have average annual hydrogen ion
concentrations after 50 years that might be deleterious to aquatic biota (i.e.,
pH < 5.5). This estimate might range as high as 349 (28%) lakes.
5.6.11 Canadian Assessment
The future effects of acidic deposition on surface water chemistry in Canada was assessed using
a regional empirical model. The purpose of the Canadian assessment was to estimate the potential
regional impacts of acidic deposition on surface waters in eastern Canada in the future without
respect to the time frame of these impacts. The Canadian estimates of future effects, therefore, were
based on a steady-state model developed by Jones et al. (1984).
The Jones model consists of both a single-lake, site model and multiple-lake regional model
(Figure 5-42). The site model consists of a set of equations to predict the eventual steady-state
chemical status of a single lake (ANC, pH, cations, and sulfate), based on the watershed's
morphometry and runoff, current chemistry of the lakewater, observed or assumed levels of sulfate
deposition, and assumed values of three other parameters including an F factor. This site model is
embedded in the regional model. The regional model contains frequency distributions for most model
parameters and estimates of the number of lakes within each of 36 secondary watersheds east of the
I
5-101
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Ontario border and south of 52°N. Model output includes regional frequency distributions of the
estimated original and eventual lake ANC, pH, cations, and sulfate values.
^ ,
Regional Model
Regional frequency distributions of model inputs
Regional frequency
distribution of
future condition
pHoo
Figure 5-42. General structure of the overall regional model. The regional model selects
combinations of site characteristics and runs the site model for each combination.
Variables defined in test. Only three of six input frequency distributions are shown. A
frequency distribution of model outputs is generated.
Source: Marmorek et al. (in preparation)
Results of the model application to lakes in three Canadian regions are listed in Table 5-30.
The model results are expressed as the estimated number and percentage of lakes projected to have an
eventual steady-state pH of <5.0 for two assumed values of FW (0 and 0.5), under current levels of
deposition. A pH value less than 5.0 was selected because few Ash species are found in lakes with pH
values less than 5.0.
Under CLD, a total of about 36,000 lakes (6%) were predicted to have pH values less than 5.0.
assuming F = 0, and about 10,000 lakes (1.6%) with pH < 5.0, assuming F=0.5.
5-102
-------
I
I
i
I
a
i
i
DRAFT PREUMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 5-30. PREDICTED AND OBSERVED (ESTIMATED FROM SURVEYS)
NUMBER (%) OF LAKES WITH pH <5, BY REGION, FOR TWO ASSUMED
VALUES OF Fw AND 100% CLD
Number (%) of Lakes
Region
Ontario
Quebec
Mari times
Total
Total
number
of lakes
137,415
229,458
243,245
610,108
WithpH<5
estimated from
surveys*
1,649
3,691
7,165
12,505
(1.2)
(1.6)
(3.0)
(2.1)
Assumed parameter _
values
A
0.67
0.67
0.91
0.91
0.91
0.91
P
100
100
20
20
20
20
F
0
0.5
0
0.5
0
0.5
0
0.5
lumber of L
Steady-Stat
at 100%
15,417
5,629
11,103
915
9,464
3,463
35,984
10,007
akes with
«pH<5
CLD
(11)
(4.1)
(4.8)
(0.4)
(3.9)
(1.4)
(5.9)
(1.6)
a Based on survey data summarized in Jeffries (1986) and scaled up using Counts and Measures information.
Source: Jones et al. 1984, Jones and Cunningham 1985.
Regional forecasts of the percentage of lakes in Eastern Canada were similar in magnitude to
regional estimates of acidic lakes in the Northeast, although the time frame in which lake pH would
decrease to less than 5.0 is unknown for the Canadian lakes. This was expected based on similarities
in deposition, geology, and watershed characteristics between northeast United States and eastern
Canada.
5.7 CONCLUSIONS AND RECOMMENDATIONS
The range in the estimated number of lakes that might became acidic in the next 50 years is
relatively large, from 97 (1.5%) to 2105 (33%) systems in the Northeast and 2 (1%) to 108 (64%) in the
Southern Blue Ridge Province. This range reflects the uncertainty both in the data and processes
underlying the forecasts and uncertainty in the models. Deposition estimates, particularly dry
deposition, are highly uncertain, for both the acid anions such as sulfate and for the base cations.
Improved deposition estimates would permit better watershed budgets for the input and export of
sulfate, nitrate, and base cations and improved estimates of the number of systems susceptible to
acidic deposition. Estimates of the number of systems in the Northeast that are at sulfur and base
cation steady state ranged from 454 (7%) to 2449 (39%) and 445 (7%) to 1932 (30%), respectively.
Soil cation exchange estimates indicated this process, by itself, was not capable of supplying
lake ANC greater than lOOueqL"1. Mineral weathering appeared to be the ANC source for lakes
with ANC > 100 ueq L"i. Lakes with ANC less than 100 ueq L"i appeared to have combined sources
from cation exchange and mineral weathering. The relative contributions from each source is
generally unknown because mineral weathering rates are highly uncertain. Long-term model
1
5-103
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
forecasts (i.e., >5-10 yr), however, are sensitive to the weathering rate estimates, whether it is an
estimate from 0-0.7 for an F factor in steady-state models or an estimate of weathering for the
dynamic models. Because mineral weathering represents the long-term source of base cations to
neutralize acidic inputs, better estimates of weathering rates for watersheds would provide better
estimates of the number of lakes expected to become acidic.
Forecasts using the dynamic surface water acidification models corroborated the estimates
obtained using single-factor approaches. The dynamic model forecasts also had a broad range from
143 (12%) to 861 (69%) lakes that might become acidic in the next 50 years. However, much of this
uncertainty is associated with the small number of watersheds that were modeled for this report.
These dynamic watershed models integrate the processes assumed to control surface water
acidification and provide estimates not only of which systems might become acidic, based on the
interaction among these processes, but also estimates of the time to reach an acidic state. A major
portion of the model uncertainty occurs because these process interactions are poorly understood at
the watershed level. Surface water acidification reflects an integrated system response to. acidic
inputs. To understand and forecast these responses requires integrated watershed-lake studies that
focus on processes and process interactions. The flow paths or water movement through the
watershed illustrates the importance of these interactions. Shallow subsurface flow that moves
through the organic horizons can increase acidic inputs to receiving systems by leaching organic acids
from the soils, while deep subsurface flow might contact highly weathered minerals that neutralize
acidic constituents before entering the receiving system. Knowing the soil characteristics and flow
path through the watershed permits better estimates of which systems might become acidic and
which systems probably were acidic historically. Improved understanding of process interactions at
the watershed level will permit the development and modification of watershed models for more
reliable forecasts of surface water chemistry.
Five recommendations for future studies, therefore, are
(1) improved deposition estimates;
(2) watershed estimates of mineral weathering and base cation supply;
(3) improved understanding of hydrology flow paths with the watershed;
(4) improved estimates of sulfate fluxes and steady state; and
(5) evaluation of model validity for both steady-state and dynamic acid-base surface
water chemistry models.
5-104
-------
I
I
I
I
I
I
I
1
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
5.8 REFERENCES
Beers, Y. 1962. Introduction to the theory of error. Second Edition Reading, MA: Addison-Wesley
Publishing Company, Inc.
Bloom, P.R. and D.F. Grigal. 1985. Modeling soil response to acidic deposition in nonsulfate
adsorbing soils. J. Environ. Qual. 14:489*495.
Chen, C.W., S.A. Gherini, R.J.M. Hudson, and J.D. Dean. 1983. The Integrated Lake-Watershed
Acidification Study, Vol. 1: Model Principles and Application Procedures. Electric Power Research
Institute, Palo Alto, CA. EPRIEA-3221.
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. U.S. Environmental
Protection Agency, Corvallis, OR. NTISPB86-178-118AS.
Cosby, B.J., G.M. Hornberger, R.F. Wright, and J.N. Galloway. 1986. Modeling the effects of
acid deposition: Control of long-term sulfate dynamics by soil sulfate adsorption. Water Resources
Research 22(8): 1283-1291.
Cosby, B.J., G.M. Hornberger, R.F. Wright, and J.N. Galloway. 1985. Modeling the effects of
acid deposition: assessment of a lumped-parameter model of soil water and streamwater chemistry.
Water Resources Research 21(l):51-63.
Cosby, B.J., R.F. Wright, G.M. Hornberger, and J.M. Galloway. 1984. Model of Acidification in
Groundwater in Catchments. University of Virginia. Final Report. Submitted to EPA
Environmental Research Laboratory-Corvallis.
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: 541a-545a.
Gherini, S.A., L. Mok, R.J.M. Hudson, G.F. Davis, C.W. Chen, and R.A. Goldstein. 1985. The
ILWAS Model: Formulation and Application. Water, Air, and Soil Pollut. 26:425-459.
Graczyk, D.J., U.A. Gebert, W.R. Krug, and G.J. Allord. In Press. Runoff for selected time
periods during 1983-85 in the Northeastern Region and Southern Blue Ridge Province of the United
States. U.S. Geological Survey Open File Report.
Gschwandter, G., K.C. Gschwandter, and K. Eldridge. 1985. Historic emissions of sulfur and
nitrogen oxide in the United States from 1900 to 1980. Vol. I Results. EPA Report
EPA-600/7-85-009a.
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 Resources Research 21:1841-1850.
Hornberger, G.M., B.J. Cosby, Jr., 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 Resources Research 22(8):1293-1302.
Jeffries, D.S. 1986. Evaluation of the regional acidification of lakes in eastern Canada using ion
ratios. Proceedings for the ECE Workshop on Acidification of Rivers and Lakes. National Water
Research Institute, Contribution Series #86-79. Burlington, Ontario.
5-105
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Jones, M.J., D.R. Marmorek, and G. Cunningham. 1984. Predicting the extent of damage to
fisheries in inland lakes of eastern Canada due to acidic precipitation. Department of Fisheries and
Oceans Canada.
Jones, M.J. and G.L. Cunningham. 1985. Summary of analyses performed as a follow-up to the
regional acidification impact modelling project. Department of Fisheries and Oceans Canada.
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, 439 pp. EPA/600-4-86/077c. U.S. Environmental Protection
Agency, Washington, DC.
Kelly, J.M. 1984. Sulfur input, output and distribution in two oak forests. In: E.L. Stone, ed.
Proceedings of the Sixth North American Forest Soils Conference, pp. 265-284. Knoxville, TN.
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. In Press. Prediction of biological acid neutralization in acid-sensitive
lakes. Biogeochemistry.
Knox, C.E. and T.J. Nordenson. 1957. U.S. Geological Survey HA-7.
Likens, G.E., F.H. Bormann, R.S. Pierce, J.S. Eaton, and N.M. Johnson. 1977.
Biogeochemistry of a forested ecosystem. New York: Springer- Ver lag.
Linthurst, 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 physio-chemical relationships, 132 pp. EPA/600/4-86/007a. U.S. Environmental
Protection Agency, Washington, DC. \
Lynch, D.S. and N.B. Disc. 1984. Sensitivity of stream basins in Shenandoah National Park to acid
deposition. U.S. Geological Survey Water-Resources Investigations Report, 61 pp. No. 85-4115.
Messer, J.J., C.W. Ariss, J.R. Baker, S.K. Crouse, 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. 1986. National Surface Water Survey: National Stream Survey Phase I - Pilot
Survey EPA-600/4-86-026. U.S. Environmental Protection Agency, Washington, DC.
National Academy of Sciences. 1984. Acid Deposition: Processes of Lake Acidification.
Washington, DC: National Academy Press.
National Academy of Sciences. 1986. Acid Deposition Long-Term Trends. Washington, DC:
National Academy Press.
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.
Rochelle, B.P., M.R. Church, and M.B. David. In Press. Sulfur retention at intensively studied
sites in the U.S. and Canada. Water, Air, and Soil Pollut.
Rochelle, B.P, M.R. Church, D.H. Landers, J.M. Eilers, and J.J. Messer. In Review. Sulfur
retention in watersheds: relationship to effects of acidic deposition on surface water chemistry. In
Proceedings North American Lake Management Society Symposium, Portland, OR.
5-106
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
i
Schnoor, J.L., N.P. Nikolaidis, and G.E. Glass. 1986. Lake resources at risk to acidic deposition
in the upper midwest. J. Wat. Pollut. Cont. Fed. 58:139-148.
Shoemaker, D.P. and C.W. Garland. 1967. Experiments in physical chemistry. Second Edition.
New York: McGraw-Hill Book Company.
Smith, R.A. and R.B. Alexander. 1983. Evidence for Acid-Precipitation-Induced Trends in Stream
Chemistry at Hydrologic Bench-Mark Stations, 12 pp. U.S. Geological Survey, Circular 910.
Swank, W.T. and J.B. Waide. In Press. Characterization of baseline precipitation Land stream
chemistry, and nutrient budgets for control watersheds. In: W.T. Swank and D.A. Crossby, Jr., eds.
Forest Hydrology and Ecology at Coweeta. Springer-Verlag.
U.S. Congress Office of Technology Assessment. 1984. Acid rain and transported air pollutants:
implications for public policy. Assessment Report OTA-0-204, Washington, DC.
i
i
i
i
5-107
-------
1
I
I
I
I
I
I
I
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
• DO NOT CITE OR QUOTE
I
SECTION 6
EFFECTS OF CHANGING SULFATE DEPOSITION ON SURFACE WATER CHEMISTRY
I
I
6.1 SUMMARY
Previously proposed estimates of the S04"2 deposition rates to minimize acidification of low
ANC systems generally fall within the range of 10 to 20 kg SO4~2 ha"1 yr"i. These estimates have
been based on estimated biological thresholds or predicted changes in surface water chemistry.
A steady-state model and a dynamic model, MAGIC, were used to estimate the change in
surface water chemistry for Northeast watersheds with SO4"2 deposition rates of 80% CLD and 50%
CLD. The models indicated between 55 (0.9%) and 83 (1%) lakes were estimated to become acidic in
the next 25 to 50 years, respectively, at 80% CLD. No (0) lakes were estimated to become acidic in the
next 50 years at 50% CLD.
Steady-state models also were used to estimate the change in surface water chemistry for
SBRP lakes and streams at 125% CLD. In the Southeast, at 125% CLD, about 115 lakes and streams
(4%) were estimated to become acidic after 25 years; 1200 lakes and streams (40%) were estimated to
become acidic after 50 years; and 2300 lakes and streams (75%) were estimated to become acidic after
100 years.
Both the steady-state and MAGIC models indicated some currently acidic lakes would recover
(i.e., ANC 3:0 ueq L"*) with an 80% CLD or 50% CLD rate. The relative recovery was greater at the
50% CLD rate.
Steady-state modeling of eastern Canadian lakes indicated a total loading of 18 kg S(V2 ha"1
yr"1 would result in 4000-16,000 acidic lakes (0.6-2.6%), whereas a loading of 13.5 kg SO4"2 ha"* yr'l
would result in about 2400-5900 acidic lakes (0.4-1.0%).
6.2 INTRODUCTION
The term "target loading" has been identified as the SO4"2 deposition rate required to protect
all but the most sensitive systems (MITAP 1983). A similar term, "critical load," has also been used
and represents the greatest SO4~2 deposition rate that will not cause long-term deleterious effects on
the most sensitive ecosystems (Nilsson 1986). Both of these loading concepts, however, assume the
most sensitive systems and deleterious effects can be delineated, which has been extremely difficult.
I
I
6-1
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
In this section, the term target loading will be used in the conventional dose-response sense,
i.e., for a given loading (dose) a corresponding ecosystem effect (response) is elicited. In using these
terms, it is important to be aware of two conditions:
(1) Dose - The chemical species must be specified. In this case, it is necessary to state
whether deposition is expressed as sulfur or sulfate and whether total, wet only,
wet plus dry, etc., components are taken into account.
(2) Response - Not all aquatic systems elicit the same response to a given dose, and the
response may be quite variable even for lakes within the same geographical region.
6.3 METHODS OF ESTIMATING CRITICAL/TARGET LOADINGS
There are three steps required to establish target loadings:
(1) definition of one or more thresholds that provide acceptable levels of protection for
aquatic systems (e.g., mean surface water pH, mean ANC, rate of acid deposition
less than estimated weathering rate);
(2) estimation of the rate of acidic deposition (or concentration of acid in precipitation)
that would cause a lake or stream of a given sensitivity to reach the threshold
condition; and
(3) extention of Steps 2 and 3 to the population of lakes and streams, to evaluate the
extent of damage associated with particular rates of deposition.
For the first step, most researchers have used threshold criteria of surface water chemistry,
such as pH levels of 5.3 or 5.8 (MOI1983, Henriksen et al. 1986). Another approach is to calculate the
level of acid anion inputs that would not exceed the base cation supply generally computed from
runoff, lake concentrations of base cations, and an assumed level of cation exchange (described in
Henriksen et al. 1986). A third strategy is to simply use "lack of evidence of biological effects" as the
threshold criterion (Newcombe 1985).
Estimating the level of deposition associated with a given condition of surface water chemistry
(Step 2) is difficult. A considerable number of factors control the effects of acidic deposition on surface
water quality, including soil/sediment contact, weathering replacement, anion retention, base cation
buffering, and instream/inlake processes. Furthermore, the role each of these factors plays in
controlling the acid-base chemistry of surface waters is variable within regions. Much of the current
research on aquatic effects of acidic precipitation is concerned with the development and
improvement of models to simulate these processes (either explicitly or implicitly) and predict
stream- and lakewater chemistry for given levels of deposition. These models, which have been
discussed in Section 5, can be roughly categorized into two groups:
(1) relatively simple statistical or empirical simulation models that treat the
watershed as a "black box," and make predictions on basin or regional scales; and
(2) process-oriented simulation models representing important physical, chemical, and
biological processes for a single basin.
6-2
-------
I
1
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
These models represent two ends of a spectrum rather than mutually exclusive categories. Detailed
process models rely, at some level, on empirical relationships, while statistical models are usually
structured to represent basic processes.
Although empirical steady-state models require less data on individual watersheds and are
easier to apply than process models, they have no temporal resolution. Empirical models make
predictions concerning only the eventual steady-state condition associated with a given level of
deposition, and not the rate at which the aquatic system will move toward that condition. Most, but
not all, empirical models assume that surface water chemistry is currently in equilibrium with
atmospheric deposition, whereas process models explicitly simulate dynamic interactions and
produce the time trend of lake- and streamwater pH, ANC, and other ions. The following sections
describe loadings estimated from biological thresholds and surface water chemistry.
6.3.1 Loadings Estimated from Biological Thresholds
Newcombe (1985) synthesized the results of various intensive studies into a hierarchy of
harmful biological effects at different levels of wet and total sulfate deposition (Figure 6-1). This
figure provides a useful overview of the range of observed responses to sulfate deposition within
relatively sensitive aquatic systems. Newcombe concluded that a limit of between 10 and 20 kg of
i'2 ha'1 yr'l was required for the protection of aquatic systems.
The Minnesota Pollution Control Agency (MPCA) has recently prepared a comprehensive
review of information relevant to the establishment of target loadings (Table 6-1). Included in their
review was a summary of empirical observations adapted from Brydges and Neary (1984; Table 6-2).
This review indicated an increase in damage to aquatic habitat or fisheries for wet deposition above
20 kg of SCV2 ha"1 yr"1. The regions shown in Table 6-2 vary considerably in the level of annual
precipitation received, which affects the estimated target loading.
There are several difficulties associated with directly comparing deposition and biological
responses in different regions. First, an accurate description of biological response within a region
requires response measurements on a large number of systems, so that the regional variability in
aquatic systems is quantified. Measurements of most biotic response variables (with the exception of
presence/absence) require costly intensive studies, which can only realistically be performed on a few
systems. The trade-off, therefore, is between reasonably crude indicators (such as presence/absence)
for a large number of systems versus more sensitive indicators of biological response for a few systems
that may not be regionally representative. Empirical comparisons of deposition and water. chemistry
do not have this problem because it is relatively easy to measure surface water chemistry for a large
number of systems.
6-3
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
.2 o ^ 0.2
•S 5 § t:^
O 3
O D
Vi
09
ca
O (C
fM
'<±
O
c
O
O "
•po '«
O
a.
e
.22
O *
g-ts
c a
.233
!
-o O
o>
«
•oS
« e
3 §
I-
o
UJ c
— o
£ C
£ -o
J5 2
I Q.
D
*->
flj
C
'en
V
-O
si
'
13
o ai
i- -^
O. nj
ii
U. in
c
1}
1
E
o
-
o
vn
*J
w
»*-
0>
"5
<*-
E
.c
o
_
« JSC
. « V
-og
6-4
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
TABLE 6-1. SUMMARY TABLE OF RECOMMENDED PRECIPITATIONjpH LEVELS
AND/OR SULFATE TARGET LOADINGS TO PREVENT LAKE ACIDIFICATION
(Source: Minnesota Pollution Control Agency 1985)
I
I
I
I
I
Source
Sulfate Target
Loading or
pH Standard
Comments
Aimer etal. 1978
Henriksen 1980
Brydges and Neary 1984
Wright 1983
Oppenheimer 1984
Regalia and Brezonik
1985
Schnoor et al. 1986
MPCA1985
15-17 kg/ha/yr Lakes with alkalinities between 100-200 ueq L'1
(total) would lose alkalinity and decline in pH. Very
sensitive lakes would be expected to acidify.
9-12 kg/ha/yr
(total)
20 kg/ha/yr
(wet)
10-12 kg/ha/yr
(wet)
20 kg/ha/yr
(wet)
< 15 kg/ha/yr
(wet)
pH4.7
18 kg/ha/yr
(wet)
< 18 kg/ha/yr
(wet)
pH 4.58-4.78
pH4.6
pH 4.3-4.4
(14-15
kg/ha/yr) (wet)
pH4.7
(11 kg/ha/yr)
(wet)
Very sensitive lakes would be protected from
acidifying; no degradation (loss of alkalinity or pH) in
lakes with aikalinities between 100-200 ueq L"1.
Majority of sensitive lakes would not acidify; this
loading is associated with a precipitation pH of 4.5.
Highly sensitive lakes would most likely acidify.
Highly sensitive lakes with alkalinities less than
50 ueq L'1 would be protected from acidification.
The majority of sensitive lakes will not acidify. Lakes
with alkalinities less than 50 ueq L'1 may acidify.
Protect very sensitive lakes from acidifying. The
majority of sensitive lakes would not show any
significant effects from acid deposition.
Protects lakes with (Ca + Mg) concentrations greater
than 40 ueq L'1 from acidifying.
The majority of sensitive lakes will be protected from
acidifying. Very sensitive lakes may acidify.
Very sensitive lakes will be protected from acidifying.
The majority of sensitive lakes will probably not show
any major effects from acid deposition.
Protects lakes with alkalinities greater than
45 ueq L"1 from acidifying.
Protects the most sensitive lakes (seepage lakes) from
acidifying.
Protects sensitive lakes with alkalinities greater than
60 ueq L"1 from acidifying. Lakes with alkalinities
less than 60 ueq L'1 will most likely acidify.
Protects the very sensitive lakes, alkalinities between
40-60 ueq L"1, from acidifying. Prevents loss of
alkalinity and pH in lakes with alkalinities between
60-100 ueq L'l.
I
6-5
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE ORQUOTE
TABLE 6-2. SUMMARY OF EMPIRICAL OBSERVATIONS OF AQUATIC REGIME
RESPONSE TO SULFATE DEPOSITION IN SPECIFIC STUDY AREAS
(Adapted from Brydges and Neary 1984;
Cited in Minnesota Pollution Control Agency 1985)
Summary
Location
Deposition
(kg sulfate/ha/yr)
Effects
Northern Saskatchewan
Kenora, Ontario
Minnesota
Northern Wisconsin
Algoma, Ontario
Nova Scotia
5 wet (1980)
9-11 bulk (1972)
10-15 wet (1980)
16-17 wet (1981)
24.7 wet (1981)
22 wet(1977-81)
Laurentide Park, Quebec 22-40 wet (1977-80)
Adirondacks, NY
Maine
Hubbard Brook, NH
Muskoka-Haliburton,
Ontario
32-48 wet (1978)
29 wet (1980)
17-28 wet
36 wet (1981)
22 wet (1980)
23-29 wet (1976-78)
31-42 bulk
No chemical effects.
No chemical effects.
No chemical effects.
Some acidification (Nichols and Verry 1985).
pH depression of 2.1 units; elevated excess
sulfate relative to region not receiving acidic
deposition; more lakes of low pH than expected.
Loss of Atlantic salmon species; historic record
of decreased pH in rivers.
Indication of decreased pH in some lakes;
indication of decline in angling success in lower
pH lakes; lower pH in lakes in spring than in
summer.
Evidence of pH declines and loss offish
populations over time.
Evidence of slight pH decrease in lakes (historic
records); no effects on Atlantic salmon; no
evidence of effects on fish in inland lakes.
Spring pH depressions; no long-term change in
stream or lake pH.
pH depressions; fish kill associated with pH
depression in one lake; algal composition in
lakes related to pH.
The second problem with comparisons of deposition and biological response is their lack of
transferability. Intensive studies of biological changes over time, together with extensive studies of
fish presence/absence, have been performed in the La Cloche Mountain and Sudbury regions of
Ontario, the Adirondack region of New York State, Nova Scotia, and southern Norway and Sweden
(reviewed in Harvey et al. 1981 and Baker 1984). These areas provide some of the strongest evidence
for the potential effects of acidic deposition. The difficulty lies in transferring these results to other
regions, with different deposition levels, deposition composition, climates, geology, soils, and aquatic
biota. Simple empirical comparisons of deposition and water chemistry also are difficult to transfer
6-6
-------
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
from one region to another: These comparisons lack the structure provided by quantitative models
that can be developed in one region and formally tested in another.
6.3.2 Loadings Estimated from Predicted Lake Chemistry
Two general categories of models have been used to estimate surface water chemistry based on
various levels of acidic deposition. These general categories are steady-state and dynamic models.
6.3.2.1 Steady-State Models
Empirical steady-state models, such as those of Aimer et al. (1978), Henriksen (1980),
Thompson (1982), Wright (1983), Henriksen et al. (1986), and Jones et al. (1984) have been commonly
used to estimate the level of acidic deposition associated with critical water quality thresholds. Three
of these models are briefly discussed in this section.
The Aimer et al. Model
Aimer et al. (1978) graphed
lake pH versus excess sulfur (S) in
lakewater for lakes in surroundings
of different sensitivity. As Church
(1984) pointed out, the quantity
"excess S in lake water" is not the
same as "total excess S in.
deposition." This limitation can be
partially overcome using measured
S deposition instead of lake S
concentrations on the x-axis. Figure
6-2 shows lake pH graphed against
S deposition for sets of Swedish
lakes with different ranges of lake
concentrations of Ca^ + Mg*2 and
implied differences in weathering
rates (Dickson 1986). The total
SO,*'2 deposition rate that was
estimated to maintain lake pH >5.0
in Sweden was 12 kg SO4 2 ha'l yr'l
for the most sensitive lakes with
Ca+2 + Mg'"2 concentrations of 20 to
90 ueq L'1. It should be stressed
that the total deposition estimates
in Figure 6-2 have high
uncertainties (Dickson 1986).
7.0
5.0
4.0
Lakes with
Ca + Mg(nonmarine)
320-370 p.eq L"i
Increasing Damage
03 9 15 30
Wet Sulfate Deposition (kg SCV2 ha'i yr).
iii i
0 9 15 60
Total Sulfate Deposition (kg $04*2 ha"> yr)
Figure 6-2. The pH values of lakes of different
Ca*"2 + Mg*2 (nonmarine) levels plotted against sulfur
wet deposition and calculated total deposition.
Circled numbers refer to subregions of Sweden.
Source: Dickson (1986), cited in Nilsson (1986)
6-7
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
\
The Henriksen et al. (7986; Model
A method similar to that of Aimer et
al. (1978) was used by Henriksen et al.
(1986) for estimating critical loadings to
streams. The approach is to graph the pH
ranges observed in a set of streams against
stream concentrations of nonmarine
sulfate, noting with each data point
(median pH) the median value for the sum
of nonmarine Ca*2 + Mg+2 (Figure 6-3).
This method addresses the seasonal
variation in stream pH in setting a critical
loading. The "critical concentration" of
sulfate in precipitation is back-calculated
from the critical stream concentration
7.0
6.5-
"5.5
5.0
4.5
Legend
10 20 30 40 50 60 70 80 90 100110
Nonmarine Suifate (peq L'i)
Figure 6-3. Distribution of pH values of 15 rivers
in southernmost and western Norway, plotted
with increasing concentrations of nonmarine
sulfate. 5.50 and 95% percentages are given.
Source: Henriksen etal. (1986)
(that which maintains stream pH above 5.3). This last step is, however, subject to numerous
uncertainties, as outlined by Church (1984).
In the more recent models (i.e., Henriksen et al. 1986), the original ANC is estimated from
current base cation concentrations and an "F factor," which accounts for the change in cation
concentrations associated with a given change in sulfate concentrations (presumably as a result of
cation exchange or mineral weathering processes). An F value of 0 implies no change in cation
concentrations (no neutralization except by lake ANC titration), whereas F = 1 implies that all acidic
deposition is neutralized. The models then predict the steady-state ANC of the system under a given
level of acid loading. Steady-state models do not predict how long it will take the system to reach the
predicted condition. This approach avoids the problem of deciding how F changes with time, but does
not resolve the question of how to estimate F values (see Henriksen [1984] and Church [1984] for a
discussion of this problem).
Assuming F = 0.2, Henriksen et al. (1986) determined a 50% reduction in current levels of
deposition in Norway would reduce the number of Norwegian lakes with ANC <0 from 50% to 30%.
The Jones et al. (1984) Model
Jones et al. (1984) developed a steady-state model that is a solution to a simple dynamic mass
balance model for ANC, base cations, and sulfate. The basic principals of the model are similar to
those of Henriksen (1982) and Wright (1983), with base cation supply affected by sulfate loadings.
According to Wright, the ANC of a watershed is estimated from current surface water cation and
6-8
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
suifate concentrations, but these values are not assumed to be in steady state. Sulfate loading is
estimated directly from deposition data, rather than from surface water suifate concentrations.
The Jones et al. (1984) model consists of both a single-lake and a multiple-lake regional model
(described in Section 5) for use in eastern regions of Canada. The output of the model includes
regional frequency distributions of the estimated original and eventual pH for lakes in eastern
Canada.
Table 6-3 summarizes some of the simplifying assumptions made by the site model and the
consequences for model predictions. Although suifate retention and reduction are not included in the
model, this should not affect the accuracy of steady-state predictions of lake chemistry. Once soils
reach equilibrium with acidic deposition, surface waters will as well. The absence of any suifate
retention then could be responsible for the predicted steady-state pH being lower than that currently
observed at a given site.
TABLE 6-3. MAJOR ASSUMPTIONS IN THE
JONES ETAL. MODEL AND THEIR CONSEQUENCES
Model Assumptions that Might Cause an
Underestimate of the Extent or Magnitude
of Damage to Surface Waters and Fisheries
Model Assumptions that Might Cause an
Overestimate of the Extent or Magnitude of
Damage to Surface Waters and Fisheries
• The original acid neutralizing capacity of
watersheds (eq ANC m"2) is not reduced by
acidic deposition
s
• Episodic pH depressions, which may have
serious consequences for fisheries, are not
simulated
• Model output of the number of lakes pH <5
does not reflect acidification from pH 6 to 5,
where aquatic effects do occur
• ANC generation within the lake is not
included, although the Jones et al. model is
currently being revised to include the model
of ANC generation of Kelly et al. (In Press)
• All suifate in wet deposition is assumed to be
associated with H+
6.3.2.2 Dynamic Models
Dynamic models provide predictions both of the final steady-state lake or stream values at a
given level of deposition and the time necessary to reach this steady-state value. Whether the time to
reach steady state is 40 years or 400 years has important policy implications. A dynamic model,
MAGIC, was used to forecast lake response under three deposition scenarios for northeastern lakes
over the next 50 years. (MAGIC was previously discussed in Section 5.6.)
6.3.3 Previously Proposed Target Loadings
Previous target loading applications generally have employed dose-response information from
laboratory and field experiments to estimate target loadings. Predictive models also can be used to
project the response of an aquatic system to a specified dose. Key references dealing with target
6-9
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
.loadings are only summarized here; the full references should be consulted for a more complete
explanation of how the target loadings were estimated.
It is important to note that a decrease in emissions may not necessarily result in a linear
decrease in deposition (NAS 1983). Furthermore, the relationship between emissions and deposition
is not well known and continues to be an active research topic.
,-
6.3.3.1 Nordic Council of Ministers
At a recent workshop on critical loadings convened by the Nordic Council of Ministers, Nilsson
(1986) and Henriksen et al. (1986) suggested that a concentration limit (either pH or sulfate) may be
preferable to a deposition limit. Their reasons follow: (a) aquatic effects appear to be more strongly
related to concentration than to loading, and (b) a concentration limit has wider application than a
deposition limit because precipitation levels (and therefore total loadings) can be extremely variable
within some regions. The issue of whether to use concentration limits or total loading limits to
protect ecosystems is still unresolved.
The critical loadings recommended by scientists attending the workshop were derived using a
steady-state model (Henriksen 1980) and a dynamic simulation model (Kamari 1986). The
recommended levels are summarized in Table 6-4. Note that the estimated critical loading increases
with both precipitation and lake concentrations of Ca+2+Mg+2 (an index of weathering).
6.3.3.2 Minnesota Pollution Control Agency
The Minnesota Pollution Control Agency (MPCA 1985) used a number of different methods to
estimate target loadings for aquatic systems in Minnesota. The methods included simulation models,
empirical models, and empirical comparisons across regions receiving different levels of deposition.
Table 6-1 summarizes the results of the Minnesota study. After a minimally acceptable pH for
precipitation of 4.7 was determined, MPCA then used a nonlinear regression equation to relate
sulfate and hydrogen ion concentrations in Minnesota precipitation (MPCA 1985). This method
indicated a wet deposition limit of 11 kg SCV2 ha"1 yr'i should be applied to regions containing
sensitive aquatic resources. The important relationship between sulfate and hydrogen ion
concentration in precipitation, however, varies from region to region (Church and Galloway 1984), so
data specific to the region must be used to determine the sulfate loading that corresponds to the
assumed threshold concentration. The model results determined for Minnesota, therefore, are
transferable only to regions with similar climate, geology, soils, and deposition.
6-10
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
i
i
t
i
TABLE 6-4. CRITICAL SULFUR DEPOSITION LOADS SUGGESTED FOR
SENSITIVE AQUATIC ECOSYSTEMS IN SPECIFIC AREAS
(Adapted from Nilsson 1986)
WetSO4'2 Total SO4'2 Concentration
Deposition Deposition of Excess SO^"2
keq km "2 yr"1 keq km"2 yr"1 Rainwater1*
Refo (kg SOY* ha'* yr-*) (kg SO^ ha-* yr'i) (peg L'l)
Empirical data, Sweden
Shallow soils, low ionic strength
waters of 20 to 40 ueq L-1
Ca+Mgc
Glacial till, medium |ieq L-1
Ca+Mg ionic strength 50 to 160
Empirical data, Norway
low ionic strength
Precipitation — 2000 mm
Precipitation —1000 mm
Empirical data, North
Eastern North America*1
Precipitation of 1000 mm
IIASA-model, Sweden6
The most sensitive lake area in
Sweden, 50-year simulation
10 (4.8)
30 (14.4)
15 (7.2)
40(19.2)b
40 (19.2)
20 (9.6)
34-77
(16.3-37)
20(9.6)
10
30
20
20
34.77
• References: 1 - Dickson (1986), 2 - Henriksen et al. (1986), 3 - Henriksen (1980), 4 - Kamfiri (1986).
b 1000-mtn precipitation.
c NonmarineCa+Mg.
d Region 1A of the National Surface Water Survey (Linthurst et al. 1986). Lower value sufficient to protect all lakes in
sample; higher value sufficient to protect 80% of surveyed lakes.
• Hultberg (1985) for Lake Gardsjoen. Dry deposition at this site estimated to be i of the total. Dry deposition is also
important in continental E urope.
I
6.4 MODEL ESTIMATES OF THE EFFECTS OF LOADING CHANGES
The effect of changing S(V2 loading on aquatic chemistry has been simulated by three models:
steady-state models discussed in Section 5 for watersheds in the Northeast and SBRP; the MAGIC
model discussed in Section 5 for DDRP watersheds in the northeastern United States; and the Jones
et al. (1984) model for Canada. The model simulation results are presented in this section. All of
these models allow estimates to be made of changes in aquatic chemistry on a regional basis.
6-11
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
6.4.1 Steady-State Forecasts
The steady-state model and general approach used to forecast the number of acidic lakes at
specific times were described in Section 5.5. The steady-state forecasts were made using F factors
ranging from 0.0 to 0.7. The number and percent of Northeast lakes that might become acidic in 25 or
50 years were estimated at two deposition rates: 80% CLD and 50% CLD. The number and percent of
Southeast lakes and streams that might become acidic at 25, 50, and 100 years were estimated for a
deposition rate of 120% CLD. The time estimates (i.e., 25-100 years) were based on the projected time
to steady state as described in Section 5.5. The estimated change in ANC for systems in each of these
regions also was calculated for each of the deposition rates. Estimates at 100% CLD are included in
the Tables 6-5 through 6-15 for reference.
6.4.1.1 Northeast
The number (percent) of northeastern lakes estimated to become acidic at different deposition
rates, F factors, and times are shown in Tables 6-5 and 6-6. The estimated change in ANC at these
different rates and levels is shown in Table 6-7 .•
The number (percent) of lakes estimated to become acidic in 25 years at 80% CLD, assuming an
F = 0, ranged from 0 (0%) in Central New England (1C) and Maine (IE) to 33 lakes (3%) in Southern
New England (ID) (Table 6-5). A total of 54 acidic lakes (0.9%) was estimated for the Northeast
(Table 6-5). Uncertainty estimates indicated an upper range of 50-60 acidic lakes (4-5%) in the
Pocono/Catskill (IB), Central (1C) and Southern New England (ID) Subregions. Assuming F = 0.7
reduced these estimates to 3 acidic lakes (0.2%) hi the Pocono/Catskill (IB) Subregion. Uncertainty
estimates indicated an upper range of 6 to 8 lakes (0.4-0.7%) might become acidic in 25 years in the
Adirondack (1A), Pocono/Catskill (IB), and Southern New England (ID) Subregions at F = 0.7.
The number of lakes estimated to become acidic in 50 years at 80% CLD, assuming F = 0,
ranged from 0 (0%) in Central New England (1C) and Maine (IE) to 52 lakes (5%) in the Southern
New England (ID) Subregion (Table 6-6). A total of 83 acidic lakes (1%) was estimated in the
Northeast (Table 6-6). Uncertainty estimates indicated an upper range of acidic lakes of about 50 to
60 lakes (4-6%) in the Pocono/Catskill (IB), Central (1C), and Southern New England (ID)
Subregions. Assuming F = 0.7, the number of acidic lakes was estimated to be 3 lakes (0.2%) in the
Poconos/Catskills (IB) and 7 lakes (0.6%) in Southern New England (ID) or a total of 10 lakes (0.2%)
in the Northeast.
6-12.
-------
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 8-5. STEADY-STATE MODEL ESTIMATES FOR NORTHEASTERN LAKES
THAT BECOME ACIDIC AT 25 YEARS
100% CLD
F Factor
0
Region 1A
Region IB
Region 1C
Region ID
Region IE
Total
0.2
Region 1A
Region IB
Region 1C
Region 10
Region IE
Total
0.4
Region 1A
Region IB
Region 1C
Region ID
Region IE
Total
0.7
Region 1A
Region IB
Region 1C
Region ID
Region IE
Total
Number
26
56
52
66
0
200
26
21
52
66
0
165
26
13
11
66
0
116
8
3
0
7
0
18
(8, 65)a
(13,111)
(0, 115)
(59, 105)
(0, 16)
(80,412)
(0,48)
(13, 110)
(0, 115)
(52, 98)
(0,0)
(65,371)
(0,41)
(6,70)
(0,81)
(32, 92)
(0,0)
(38,284)
(0,17)
(3, 21)
(0,8)
(7,33)
(0,0)
(10,79)
Percent
2
4
4
6
0
3
2
1
4
6
0
3
2
0.9
1
6
0
2
1
0.2
0
0.6
0
0.2
(1,6)
(0.9, 8)
(0,9)
(5, 10)
(0,1)
(1,6)
(0,4)
(0.9, 8)
(0,9)
(5,9)
(0,0)
(1,6)
(0,4)
(0.4,5)
(0,6)
(3,9)
(Q,0)
(1,4)
(0,2)
(0.2, 1)
(0,1)
(0.6,3)
(0, 0)
(0.2, 1)
80% CLD
Number
8
13
0
33
0
54
8
13
0
14
0
35
0
6
0
7
0
13
0
3
0
0
0
3
(0, 26)
(6, 62)
(0, 52)
(14, 59)
(0,0)
(20, 199)
(0, 26)
(3,53)
(0, 11)
(14, 52)
(0,0)
(17, 142)
(0, 26)
(3, 24)
(0,0)
(0, 33)
(0,0)
(3,83)
(0,8)
(0, 6)
(0,0)
(0,7)
(0, 0)
(0,21)
Percent
0.7 (0, 2)
0.9 (0.4, 4)
0 (0,4)
3 (1, 5)
0 (0, 0)
0.9 (0.3, 3)
0.7 (0, 2)
0.9(0.2,4)
0 (0, 0.9)
1 (1, 5)
0 (0, 0)
0.6(0.3,2)
0 . (0, 2)
0.4 (0.2, 2)
0 (0, 0)
0.6 (0, 3)
0 (0,0)
0.2 (0.05, 1)
0 (0, 0.7)
' 0.2 (0, 0.4)
0 (0,0)
0 (0, 0.6)
0 (0, 0)
0.05 (0, 0.3)
50% CLD
Number
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(0,0)
(0,6)
(0,0)
(0,0)
(0,0)
(0,6)
(0,0)
(0,6)
(0,0)
(0,0)
(0,0)
(0,6)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
Percent
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(0,0)
(0, 0.4)
(0,0)
(0,0)
(0,0)
(0,0.1)
(0,0)
(0,0.4)
(0,0)
(0,0)
(0,0)
(0,0.1)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
Base case (low estimate, high estimate).
6-13
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 6-6. STEADY-STATE MODEL ESTIMATES FOR NORTHEASTERN LAKES
THAT BECOME ACIDIC AT 50 YEARS
100% CLD
F Factor
0
Region 1A
Region IB
Region 1C
Region ID
Region IE
Total
0.2
Region 1A
Region IB
Region 1C
Region ID
Region IE
Total
0.4
Region 1A
Region IB
Region 1C
Region ID
Region IE
Total
0.7
Region 1A
Region IB
Region 1C
Region ID
Region IE
Total
Number
26
65
63
72
8
234
26
59
52
66
0
203
26
21
52
66
0
165
8
10
0
26
0
44
(8, 160)a
(13, 125)
(0, 168)
(52, 131)
(0,24)
(73, 608)
(0, 128)
(13,111)
(0, 134)
(52, 125)
(0,24)
(65,522)
(0,92)
(13, 110)
(0, 115)
(33, 105)
(0, 16)
(46,438)
(0,41)
(3, 29)
(0, 40)
(7,72)
(0,0)
(10, 182)
Percent
2
4
5
7
0.7
4
2
4
4
6
0
3
2
1
4
6
0
3
1
0.7
0
2
0
1
(0.7, 15)
(0.9,9)
(0, 13}
(5, 12)
(0,2)
(1, 10)
(0, 12)
(0.9, 8)
(0, 10)
(5, 12)
(0,2)
(1.8)
(0,8)
(0.9,8)
(0,9)
(3, 10)
(0,1)
(0.7,7)
(0,4)
(0.2, 2)
(0,3)
(0.6,7)
(0,0)
(0.2,3)
80% CLD
Number
18
13
0
52
0
83
8
13
0
33
0
54
8
13
0
14
0
35
0
3
0
7
0
10
(0,35)
(13,65)
(0,52)
(14, 66)
(0,8)
(27, 226)
(0, 26)
(6,62)
(0, 52)
(14,66)
(0,0)
(20, 206)
(0,26)
(3, 53)
(0,52)
(0,66)
(0,0)
(3, 197)
(0,8)
(0, 13)
(0,0)
(0, 26)
(0,0)
(0,47)
Percent
2 (0, 3)
0.9 (0.9, 4)
0 (0, 4)
5 (1,8)
0 (0, 0.7)
1 (0.4,4)
0.7 (0, 2)
0.9(0.4,4)
0 (0,4)
3 (1, 6)
0 (0, 0)
0.9 (0.3, 3)
0.7 (0, 2)
0.9 (0.2, 4)
0 (0,4)
1 (0, 6)
0 (0, 0)
0.6 (0.05, 3)
0 (0, 0.7)
0.2 (0, 0.9)
0 (0, 0)
0.6 (0, 2)
0 (0, 0)
0.2 (0, 0.7)
50% CLD
Number
0
3
0
0
0
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(0,8)
(0,6)
(0,0)
(0,7)
(0,0)
(0,21)
(0,8)
(0,6)
(0,0)
(0,0)
(0,0)
(0, 14)
(0,0)
(0,6)
(0,0)
(0,0)
(0,0)
(0,6)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
Percent
0
0.2
0
0
0
0.05
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(0,0.7)
(0, 0.4)
(0,0)
(0, 0.6)
(0,0)
(0,0.3)
(0, 0.7)
(0,0.4)
(0,0)
(0,0)
(0,0)
(0,0.2)
(0,0)
(0,0.4)
(0,0)
(0,0)
(0,0)
(0,0.1)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
Base case (low estimate, high estimate).
6-14
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
1
TABLE 6-7. ESTIMATED CHANGE IN NORTHEASTERN LAKE ANC AT
DIFFERENT DEPOSITION LOADINGS USING A STEADY-STATE MODEL
I
I
I
I
Level of Deposition (percent of CLD)
F Factor
Subregion 1A
n
V
0 2
Wi£l
0 4
u>^
0 7
w. I
Subregion IB
n
U
0 2
\Jm£t
0 4
\Jt^
0 7
\Jt I
Subregion 1C
n
V
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean .
Median
Min
Max
Mean
Median
Min
Max
5
8
-114
55
4
7
-91
44
3
5
-69
33
2
3
-34
16
-23
-14
-158
71
-18
-11
-126
57
-14
-8
-95
43
-7
-4
-47
21
-3
2
-67
40
100%
(41.-31)*
(45, -28)
(-78, -150)
(91, 19)
(33, -25)
(36, -22)
(-62, -120)
(73, 15)
(25, -18)
(27, -17)
(.47, -90)
(55, 11)
(12, -9)
(13, -8)
(-23, 45)
(27, 6)
(18, -63)
(27, -54)
(-117, -198)
(112,31)
(14, -50)
(21, -43)
(-94, -158)
(89,25)
(11, -38)
(16, -32)
(-70, -119)
(67, 19)
(5, -19)
(8, -16)
(-35, -59)
(34, 9)
(22, -29)
(27, -24)
(-41, -92)
(66, 15)
28
31
-73
92
23
24
-58
73
17
18
-44
65
9
9
-22
28
12
15
-101
97
10
12
-81
77
7
9
-60
58
4
5
-30
29
17
19
-40
62
80%
(59, -2)
(61,0) .
(.43, _103)
(122, 62)
(47, -1)
(49, 0)
(-34, -83)
(98, 49)
(35, -1)
(36, 0)
(-26, -62)
(73, 37)
(18, -1)
(18,0)
(-13, -31)
(37, 19)
(45, -21)
(49, -18)
(-67, -134)
(130, 63)
(36, -17)
(39, -14)
(-54, -107)
(104,51)
(27, -13)
(29, -11)
(-40, -81)
(78, 38)
(14, -6)
(15, -5)
(-20, -40)
(39,19) '
(39, -5)
(41, -4)
(-18, -62)
(84,40)
63
64
-18
149
50
51
-14
120
38
38
-11
90
19
19
-5
45
64
57
-30
143
51
48
-24
114
38
36
-18
86
19
18
-9
43
48
47
-4
99
50%
(85,41)
(86,42)
(4, -39)
(171,128)
(68, 32)
(69, 34)
(3, -32)
(137, 102)
(58, 25)
(52, 25)
(3, -24)
(103, 77)
(25, 12)
(26, 13)
(1.-12)
(51,38)
(89,39)
(84, 34)
(-5, -55)
(168, 118)
(71, 31)
(67, 28)
(-4, -44)
(134,94)
(53, 23)
(51,21)
(-3, -33)
(101,71)
(27, 12)
(25, 10)
(1.-16)
(50, 35)
(66, 30)
(65, 29)
(13, -22)
(117,81)
(continued)
6-15
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 6-7. (Continued)
Level of Deposition (percent of CLD)
F Factor
Subregion 1C
(Cont.)
0.2
0.4
n 7
V* I
Subregion ID
n
w
0 2
\ItJU
0 4
w»^
0 7
\J, I
Subregion IE
n
V
0 2
Vf.M
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
Mean
Median
Min
Max
-3
2
-53
32
-2
1
-4Q
24
-1
1
-20
12
-23
-23
-103
33
-18
-18
-83
26
-14
-14
-62
20
-7
-7
-31
10
-6
-5
-63
21
-5
-t
-50
16
100%
(18, -23)
(22, -19)
(-33, _74)
(53, 12)
(13, -17)
(16, -14)
(-25, -55)
(40, 9)
(7, -9)
(8, -7)
(-12, -28)
(20, 4)
(-1.-45)
(-1.-46)
(-81, -125)
(54, 10)
(-1.-36)
(-0.-36)
(-65, -100)
(44,8)
(-0.-27)
(-0, -27)
(-49, -75)
(33, 6)
(-0.-14)
(-0.-14)
(-24, -38)
(16,3)
(12, -23)
(13, -23)
(-45, -81)
(38, 3)
(10, -19)
(10, -18)
(-36, -64)
(31,2)
14
15
-32
49
10
11
-24
37
5
6
-12
19
8
6
-69
66
6
5
-56
53
5
4
-42
40
2
2
-21
20
9
10
-41
40
7
8
-33
32
80%
(31, -4)
(33, -3)
(_U, -49)
(67,32)
(24, -3)
(24, -2)
(-11, -37)
(50, 24)
(12, -2)
(12, -1)
(-5, -19)
(25, 12)
(28, -12)
(26, -14)
(-50, -89)
(86, 46)
(22, -9)
(21, -11)
(40, -71)
(69, 37)
(17, -7)
(15, -8)
(-30, -54)
(51,28)
(8, -4)
(8, -4)
(-15, -27)
(26, 14)
(26, -7)
(26, -7)
(-25, -58)
(57,24)
(21 ,-6)
(21 ,-6)
(-20, -46)
(45, 19)
38
38
-4
79
29
28
-3
59
14
14
-1
30
54
51
-21
117
44
41
-17
94
33
31
-13
70
16
15
-6
35
31
32
-9
70
25
25
-7
56
50%
(53,24)
(52, 23)
(11, -18)
(93, 65)
(39, 18)
(39, 18)
(8, -13)
(70,49)
(20, 9)
(19,9)
(4, -7)
(35,24)
(71,38)
(68, 35)
(-5, -38)
(134, 101)
(57,30)
(54, 28)
(-4, -30)
(107, 81)
(43, 23)
(41,21)
(-3, -23)
(80, 60)
(21,11)
(20, 10)
(-1.-11)
(40, 30)
(46, 16)
(47, 17)
(6, -25)
(85, 55)
(37, 13)
(38, 13)
(5, -20)
(68, 44)
(continued)
6-16
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
TABLE 6-7. (Continued)
I
I
I
I
I
Level of Deposition (percent of CLD)
F.Factor
Subregion IE
(Cont)
Mean
Median
U'4 Min
Max
Mean
_ _ Median
Mm
Max
-3
-3
-38
12
-2
-2
-19
6
100%
(7, -14)
(8, -14)
(_27,^8)
(23, 2)
(4, -7)
(4, -7)
(-14, -24)
(11,1)
5
6
-25
24
3
3
-12
12
80%
(15, -4)
(16, -4)
(-15, -35)
(34, 14)
(8, -2)
(8, -2)
(-7, -17)
(17,7)
19
19
-16
42
9
10
-3
21
50%
(28, 10)
(28, 10)
(4, -15)
(51, 33)
(14, 5)
(14, 5)
(2, -7)
(26, 16)
* Base case (low estimate, high estimate).
With a 50% reduction in deposition (50% CLD), no (0) lakes were estimated to become acidic in
25 years at any assumed F value. At 50 years, 3 acidic lakes (0.2%) were estimated for the
Poconos/Catskills (IB) at F = 0. Confidence estimates indicated a total of 6 lakes (0.1%) might become
acidic in 25 years and 21 lakes (0.05%) in 50 years in the Northeast (Table 6-6).
The average estimated change in ANC was positive or showed an increase in all subregions at
both deposition rates and all F values (Table 6-7). The smaller ANC increases at F = 0.7 resulted
because decreases in SO^2 concentration also had associated decreases in Ca+2 + Mg*2 so there was a
lower net rate of change in ANC. The largest change in ANC at both 80% CLD and 50% CLD was
estimated to occur in the Adirondacks (1A) with the smallest change in ANC estimated to be in
Maine (IE).
6.4.1.2 Southeast
The number (percent) of southeastern lakes and streams estimated to become acidic in the next
25, 50, and 100 years with 120% CLD, assuming different F factors, are shown in Tables 6-8 through
6-10. The estimated change in ANC associated with 120% CLD and different F factors is shown in
Tables 6-11 through 6-13.
6-17
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
The estimated number of acidic lakes and streams at 120% CLD did not differ from the number
estimated at 100% CLD for all F values in 25 years (Table 6-8). There were more lakes and streams
estimated to become acidic at 120% CLD, however, at 50 and 100 years in the future (Tables 6-9
through 6-10). The number of lakes estimated to become acidic at F = 0 in 50 years at 120% CLD
increased to 27 lakes (10%), while the number of streams, under the same conditions, increased to
1173 streams (42%) for a total of 1200 systems (40%) (Table 6-9). Uncertainty in these estimates
indicated an upper range of 56 lakes (21%) and 1479 streams (54%) for a total of 1535 systems (51%).
The number of lakes and streams estimated to become acidic in 50 years at F = 0.7 was similar at
100% CLD and 120% CLD.
TABLE 6-8. STEADY-STATE MODEL ESTIMATES OF THE NUMBER OF
ACIDIC SBRP SYSTEMS IN 25 YEARS
Level of Deposition (percent of CLD)a
F
n
\J
0 2
\f»JU
04.
W*"T
0 7
VI. 1
100%
Lakes #
%
Streams #
%
Total #
%
Lakes #
%
Streams #
%
Total #
%
Lakes #
%
Streams #
%
Total #
%
Lakes #
%
Streams #
%
Total #
%
2
1
115
4
117
4
1
0.4
27
I
28
1
0
0
27
1
27
1
0
0
0
0
0
0
(0, 2)a
(0,1)
(27,115)
(1,4)
(27, 117)
(1,4)
(0,2)
(0,1)
(27,115)
(1,4)
(27, 117)
(1,4)
(0,1)
(0, 0.4)
(27, 27)
(1,1)
(27, 28)
(1,1)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
120%
2
1
115
4
117
4
1
0.4
27
1
28
1
0
0
27
1
27
1
0
0
0
0
0
0
(0,2)
(0,1)
(27, 115)
(1,4)
(27, 117)
(1,4)
(0,2)
(0,1)
(27, 115)
(1,4)
(27,117)
(1,4)
(0,1)
(0,0.4)
(27, 27)
(1,1)
(27,28)
(1,1)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
(0,0)
* Base case (low estimate, high estimate).
6-18
-------
I
i
I
i
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 6-9. STEADY-STATE MODEL ESTIMATES OF THE NUMBER OF
ACIDIC SBRP SYSTEMS IN 50 YEARS
Level of Deposition (percent of CLD)
F
09
,tt
04
W*"*
07
« c
100%
Lakes
Streams
Total
Lakes
Streams
Total
Lakes
Streams
Total
Lakes
Streams
Total
#
%
#
%
#
%
#
%
#
%
#
%
#
%
#
%
#
%
#
%
#
%
#
%
21
8
675
24
696
23
10
4
552
20
562
19
7
3
115
4
122
4
1
0.4
27
1
28
1
(10, 41)a
(4, 16)
(616, 1,185)
(22, 43)
(626, 1,226)
(21,41)
(7, 22)
(3,8)
(485, 644)
(18, 23)
(492, 666)
(16, 22)
(1, 10)
(0.4, 4)
(115,481)
(4, 17)
(116,491)
(4, 16)
(0,1)
(0, 0.4)
(27,27)
(1,1)
(27, 28)
(1,1)
120%
27
10
1,173
42
1,200
40
12
5
634
23
646
21
10
4
382
14
392
13
1
0.4
27
1
28
1
(16, 56)
(6, 21)
(634, 1,479)
(23,54)
(650, 1,535)
(21,51)
(10,28)
(4,11)
(496, 980)
(18, 35)
(506, 1,008)
(17,33)
(7, 10)
(3,4)
(115,552)
(4,20)
(122, 562)
(4, 19)
(0,1)
(0, 0.4) .
(27,39)
(1,1)
(27,40)
(1,1)
• Base case (low estimate, high estimate).
At 100 years, 143 lakes (55%) and 2179 streams (79%) were estimated to become acidic at 120%
CLD and F=0 or a total of 2322 (77%) southeastern systems (Table 6:10). The estimated number of
acidic systems at F = 0.7 was 98 (3%) southeastern systems, consisting of 6 lakes (2%) and 92 streams
(3%). Upper estimates ranged to a total of 2531 (84%) systems.
The estimated change in ANC was negative or decreased for all southeastern systems
regardless of the F value at 120% CLD (Tables 6-11 through 6-13). The change in ANC ranged from
-52 ueq L'l at 25 years to -231 ueq L'l at 100 years (F = 0) in SBRP lakes, and -39 ueq L'l at 25 years
to-162 ueq L'l at 100 years (F = 0) in SBRP streams at F = 0.
6-19
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 6-10. STEADY-STATE MODEL ESTIMATES OF THE NUMBER OF
ACIDIC SBRP SYSTEMS IN 100 YEARS
Level of Deposition (percent of CLD)
F
n
\J
Q 2
\Jt£*
n 4.
\jt**
n 7
VI* 1
100%
Lakes #
%
Streams #
%
Total #
%
Lakes #
%
Streams #
%
Total #
%
Lakes #
%
Streams #
%
Total #
%
Lakes #
%
Streams #
%
Total #
%
97
37
1,930
70
.2,027
67
56
21
1,423
52
1,479
49
18
7
1,099
40
1,117
37
1
0.4
64
2
65
2
(20, 151)a
(7, 51)
(1,415,2,179)
(51,79)
(1,435,2,330)
(47, 77)
(13,114)
(5, 43)
(1,099, 1,930)
(40,70)
(1,112,2,044)
(37,68)
(10,71)
(4, 27)
(484, 1,393)
(18, 50)
(494, 1,464)
(16, 48)
(0, 10)
(0,4)
(27, 92)
(1,3)
(27, 102)
(1,3)
143
55
2,179
79
2,322
77
92
35
1,930
70
2,022
67
39
15
1,393
50
1,432
47
6
2
92
3
98
3
120%
(51, 172)
(19,65)
(1,545,2,359)
(56, 85)
(1,596,2,531)
(53, 84)
(20, 151)
(8, 58)
(1,415,2,179)
(51,79)
(1,435,2,330)
(47,77)
(13,91)
(5,34)
(672, 1,486)
(24, 54)
(685, 1,577)
(23, 52)
(0, 15)
(0,6)
(64, 560)
(2, 20)
(64, 575),
(2, 19)
Base case (low estimate, high estimate).
6-20
-------
1
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
1
TABLE 6-11. STEADY-STATE MODEL ESTIMATES OF THE CHANGE IN
ANC AFTER 25 YEARS IN SBRP LAKES AND STREAMS
I
i
i
i
Level of Deposition (percent of CLD)a
F
Lakes
Mean
_ Median
Minimum
Maximum
Mean
Q 2 Median
Minimum
Maximum
Mean
Q 4 Median
Minimum
Maximum
Mean
Q 7 Median
Minimum
Maximum
Streams
Mean
n Median
Minimum
Maximum
Mean
0 2 Median
Minimum
Maximum
Mean
_ . Median
Minimum
Maximum
Mean
Q ? Median
Minimum
Maximum
-45
-to
-96
17
-36
-32
-77
13
-27
-24
-58
10
-13
-12
-29
5
-36
-36
-77
-12
-28
-28
-62
-9
-21
-21
-46
-7
-11
-11
-23
-*
100%
(-32, -58)
(-27, -53)
(-83, -109)
(30,4)
(-25, -46)
(-22, -43)
(-66, -87)
(24,3)
(-19, -35)
(-16, -32)
(-50, -65
(18, 2)
(-10, -17)
(-8, -16)
(-25, -33)
(9,1)
(-24, -47)
(-24, -47)
(-66, -88)
(-1.-23)
(-20, -37)
(-19, -37)
(-53, -71)
(-0, -18)
(-15, -28)
(-15, -28)
(-39, -53)
(-0, -14)
(-7, -14)
(-7, -14)
(-20, -26)
(-0,-7)
-52
-44
-113
-21
-42
-35
-90
-17
-31
-26
-68
-13
-16
-13
-34
-6
-39
-39
-93
-24
-32
-31
-75
-19
-24
-24
-56
-14
-12
-12
-28
-7
120%
(-32, -58)
(-27, -54)
(-82, -109)
(3.0, 3)
(-25, -47)
(-22, -43)
(_66, -87)
(24, 3)
(-19, -35)
(-16, -32)
(-19, _66)
(18,2)
(-9, -17)
(-8, -16)
(-25, -33)
(9,1)
(-28, -51)
(-27, -51)
(-81, -105)
(-12, -35)
(-22, -41)
(-22, -41)
(-65, -84)
(-9, -28)
(-17, -31)
(-16, -31)
(-49, -63)
(-7, -21)
(-8, -15)
(-8, -15)
(-24, -32)
(-4, -11)
Each value is followed in parentheses by values obtained by adding or subtracting the estimated
uncertainty.
6-21
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 6-12. STEADY-STATE MODEL ESTIMATES OF THE CHANGE IN
ANC AFTER 50 YEARS IN SBRP LAKES AND STREAMS
Level of Deposition (percent of CLD)a
100%
120%
Lakes
0
0.2
0.4
0.7
Streams
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
-109 (-86, -132)
-100 (-77,-123)
-178 (-155, -201)
9 (32, -14)
-87 (-69,-105)
-80 (-61,-98)
-142 (-124,-161)
7 (26,-11)
-65 (-51, -79)
-60 (-46,-74)
-107 (-93,-121)
6 (19,-8)
-33 (-26, -40)
-30 (-23,-37)
-53 (-47,-60)
3 (10,-4)
-126 (-103,-150)
-113 (-89,-136)
-214 (-190,-237)
-38 (-15,-62)
-101 (-82,-120)
-90 (-71,-109)
-171 (-152,-190)
-30 (-12,^19)
-76 (-62,-90)
-68 (-54,-82)
-128 (-114,-142)
-23 (-9,-37)
-38 (-31,^15)
_34 (-27,-41)
_64 (-57,-71)
-11 (-4,-18)
n
V
09
.&
04. -
.^
07
. I
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
-86
-89
-140
-8
-68
-71
-112
-6
-51
-54
-84
-5
-26
-27
-42
-2
(-72, -99)
(-76, -103}
(-127, -154)
(6, -21)
(_57,-79)
(-61, -82)
(_101,-123)
(5, -17)
(_43,_6Q)
(-46, -62)
(-76, -92)
(4, -13)
(-22, -30)
(-23, -31)
(-38, -46)
(2, -6)
-98
-99
-172
-38
-78
-79
-138
-31
-59
-60
-103
-23
-29
-30
-52
-12
(-82, -114)
(-83, -115)
(-156, -188)
(-22, -54)
(-65, -91)
(-67, -92)
(-125, -151)
(_18,^t4)
M9.-68)
(-50, -69)
(-94, -113)
(-13, -33)
(_25,-34)
(-25, -35)
(.47, -57)
(-7, -16)
a Each value is followed in parentheses by values obtained by adding or subtracting the estimated
uncertainty.
6-22
-------
I
1
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 6-13. STEADY-STATE MODEL ESTIMATES OF THE CHANGE IN
ANC IN 100 YEARS IN SBRP LAKES AND STREAMS
Level of Deposition (percent of CLD)<*
F
Lakes
0
,2
0.4
0,
Streams
0
0.2
0.4
0,
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
Mean
Median
Minimum
Maximum
-184
-154
-361
11
-147
-123
-288
9
-111
-92
-216
7
-55
-46
-108
3
-129
-129
-180
-9
-103
-103
-144
-7
-78
-77
-108
-5
-39
-39
-54
-3
100%
(-11 2, -257)
(-81, -227)
(-288, -433)
(84, -61)
(-89, -206)
(-65, -181)
(-230, -347)
(67, -49) .
(-67, -154)
(-49, _136) .
(-173, -260)
(50, -37)
(-33, -77)
(-24, -68)
(-86, -130)
(25, -18)
(-99, -160)
(-98, -160)
(-149, -210)
(22, -39)
(-79, -128)
(-79, -128)
(-119, -168)
(17, -32)
(-59, -96)
(-59, -96)
(-89, -126)
(13, -24)
(_30, -48)
(-29, -48)
(-45, -63)
(7, -12)
-231
-190
-437
-185
-152
-349
-34
-139
-114
-262
-26
-69
-57
-131
-13
-162
-160
-221
' -41
-129
-128
-177
-32
-97
-96
-133
-24
-48
^48
-66
-12
120%
(-144, -318)
(-103, -277)
(-349, -524)
(45, -130)
(-115, -255)
(-82, -221)
(_280,-419)
(36, -104)
(-86, -191)
(-62, -166)
(-210, -314)
(27, -78)
(-43, -96)
(-31, -83)
(-105, -157)
(13, -39)
(-125, -198)
(-123, -196)
(-185, -258)
(-4, -77)
(-100, -158)
(-98, -157)
(-148, -206)
(-3, -62)
(-75, -119)
(-74, -118)
(-111, -155)
(-2, -46)
(-38, -59)
(-37, -59)
(-55, -77)
(-1.-23)
Each value is followed in parentheses by values obtained by adding or subtracting the estimated
uncertainty.
6-23
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
6.4.2 The MAGIC Model
The MAGIC model was previously described in Section 5. MAGIC was used to estimate
changes in the acid-base chemistry of 10 DDRP watershed lake systems in the Northeast'(described
in Section 5.6) for three sulfate deposition scenarios: a 25% increase over CLD (125% CLD), a 20%
decrease in CLD (80% CLD), and a 50% decrease in CLD (50% CLD). Estimates of the change in acid-
base chemistry of the 10 DDRP watersheds using the MAGIC model at 100% CLD were presented in
Section 5.6. Regional estimates of the number of lakes that become acidic at the three levels of
deposition at 25 and 50 years and change in ANC at 25 and 50 years are presented below.
The target population of Northeast lakes with ANC concentrations between 0 and 100 ueq L"1
was 1248 lakes. Estimates of the number of acidic lakes and the change in ANC over 25 and 50 years
are based on this target population and shown in Tables 6-14 through 6-15.
At 125% CLD, 143 lakes (11.5%) were estimated to become acidic within the next 25 years with
upper estimates of 861 lakes (69%) (Table 6-14). The estimated number of acidic lakes increased to
324 lakes (26.1%) within 50 years at 125% CLD (Table 6-14). The upper estimate of acidic lakes was
the same at 50 years as at 25 years.
At both 80% CLD and 50% CLD, no (0) lakes were forecast to become acidic within 25 or
50 years (Table 6-14). Uncertainty estimates indicated that 549 lakes (44%) might become acidic in
25 or 50 years at 80% CLD while an upper range of 449 (36%) might become acidic in 50 years at 50%
CLD.
TABLE 6-14. ESTIMATED NUMBER AND PERCENTAGE OF
NORTHEAST LAKES (ANC 0-100 peq L'l) THAT BECOME ACIDIC
UNDER DIFFERENT LOADINGS USING MAGIC
Level of Deposition (percent of CLD)
100%
25 Years
Number
Percent
50 Years
Number
Percent
0
0
143
11.5
(0,
(0,
(0,
(0,
549)
44)
861)
69)
. 143
11.5
324
26.1
125%
(0,
(0,
(0,
(0,
861)
69)
861)
69)
0
0
0
0
80%
(0,
(0,
(0,
(0,
549)
44)
549)
44)
0
0
0
0
50%
(0, 549)
(0,44)
(0, 449)
(0,36)
The estimated change in ANC was negative for all lakes at 125% CLD and positive for all lakes
at 80% and 50% CLD (Table 6-15). The average estimated ANC decrease ranged from -9.3 to
-12.8 ueq L"1 in 25 and 50 years, respectively, at 125% CLD. The average estimated change in ANC
ranged from +4.5 to +9.4 ueq L'l at 80% CLD and 50% CLD, respectively, at 25 years. At 50 years,
this average estimated change in ANC ranged from +7.5 to +9.5 ueq L'1 at 80% CLD and 50% CLD,
respectively.
6-24
-------
I
i
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 6-15. CHANGE IN ANC ESTIMATED FOR NORTHEAST LAKES
UNDER DIFFERENT LOADINGS USING MAGIC
Level of Deposition (percent of CLD)
A ANC
25 Years
Mean
Median
Minimum
Maximum
50 Years
Mean
Median
Minimum
Maximum
100%
-2.9 (-8.6, -2.8)
-2.7 (-8.4, -2.6)
+ 0.0 (-5.7, +5.7)
-8.5 (-14.2, -2.8)
-5.1 (-10.8, 0.6)
-4,9 (-10.6, 0.8)
0.0 (-5.7, 5.7)
-15.6 (-27.3, -9.9)
125%
-9.3 (-15.0, -3.6)
-9.0 (-14.7, -3.3)
-1.3 (-7.0, -4.4)
-16.0 (-21.7, -10.3)
-12.8 (-18.5, -7.1)
-12.4 (-18.1, -6.7)
0 (-5.7, 5.7)
-26.9 (-32.6, -21. 2)
80%
+ 4.5 (-1.2, 10.2)
+4.3 (-1.4, 10.0)
-1.4 (-7.1, 4.3)
+ 10.4(4.7,16.1)
+ 7.5 (-1.8, +13.2)
. +7.3 ( + 1.6, 13.0)
-0.4 (-6. 1,5.3)
+ 17.2(11.5,22.9)
50%
+ 9.4(3.7,15.1)
+ 9.1 (3.4,14.8)
-0.9 (-6.6, 4.8)
+ 18.8(13.1,24.5)
+ 9.5(3.8,15.2)
+ 9.2(3.5,14.9)
+ 0.8 (-4.9, 6. 5)
+ 18.7(13.0,24.4)
6.4.2.3 Comparisons
Different sulfate deposition rates were used for the steady-state and dynamic model forecasts.
Revised estimates of dry deposition, which incorporate an improved regional gradient, were obtained
and used for the steady-state analyses. There was insufficient time to rerun the MAGIC forecasts
with these revised deposition scenarios.
For the Northeast lakes, both modeling approaches indicated reduced sulfate deposition
resulted in fewer lakes becoming acidic within 50 years with no (0) lakes forecast to become acidic at
50% CLD. Both approaches also indicated a net increase in ANC or positive change in ANC at 80%
CLD and 50% CLD rates.
6.4.2.4 Recovery
The steady-state and MAGIC models were used to forecast the recovery of selected currently
acidic Northeast lakes at 80% CLD and 50% CLD. Recovery was defined as ANC SO ueq I/1. The
steady-state model forecast the potential recovery of 30 acidic (ANC <0), clearwater (Color
<30 Pt-Co units) DDRP drainage lakes in the Northeast. The MAGIC model forecast the potential
recovery of 3 currently acidic, clearwater DDRP drainage lakes, located in the Adirondack and
Pocono/Catskill Subregions. The results of these simulations are shown in Tables 6-16 and 6-17.
These results are not target population estimates. The forecasts are provided for information
only.
The steady-state model estimated the number of lakes recovering at 80% CLD ranged from 23
out of 31 lakes at F = 0 to 9 of 31 lakes at F = 0.7 (Table 6-16). At 50% CLD, 28 of 31 lakes were
forecast to recover with F = 0 in all subregions. At F = 0.7,24 of 31 lakes were forecast to recover in all
subregions. Fewer lakes were forecast to recover because Ca+2+Mg+2 inputs also decreased as
sulfate inputs decreased.
.6-25
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
TABLE 6-16. NUMBER OF INDIVIDUAL ACIDIC (ANC £0) CLEARWATER
(I.E., PT-CO COLOR < 30 UNITS) DRAINAGE DDRP LAKES/RESERVOIRS
IN THE NORTHEAST ESTIMATED TO RECOVER AT 80% CLD
AND 50% CLD USING A STEADY-STATE MODEL
Deposition
Subregion3
1A
IB
1C
ID
TOTAL
F Factor
CIO
0.2
0.4
0.7
0.0
0.2
0.4
0.7
0.0
0.2
0.4
0.7
0.0
0.2
0.4
0.7
0.0
0.2
0.4
0.7
Number
Estimated Recovery Frequency0
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
, Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number"
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
Number
Estimated Recovery Frequency
80%
14
14/14
14
14/14
12
12/14
4
4/14
6
6/11
5
5/11
5
5/11
3
3/11
1
1/1
1
1/1
1
1/1
1
1/1
2
2/5
2
2/5
2
2/5
1
1/5
23
23/31
22
22/31
20
20/31
9
9/31
50%
14
14/14
14
14/14
14
14/14
12
12/14
8
8/11
8
8/11
8
8/11
6
6/11
1
1/1
1
1/1
1
1/1
1
1/1
5
5/5
5
5/5
5
5/5
5
5/5
28
28/31
28
28/31
28
28/31
24
24/31
a There were no <0) acidic lakes in Subregion IE.
b These are NOT population estimates and percentages should not be calculated.
6-26
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
f
I
I
With the MAGIC model, 2 out of 3 lakes were forecast to recover within 25 years at both 80%
CLD and 50% CLD (Table 6-17). The change in ANC was quite variable among the 3 lakes,
indicating the influence of site-specific watershed characteristics on lake recovery.
TABLE 6-17. ESTIMATED CHANGE IN ANC FOR
THREE ACIDIC LAKES UNDER DIFFERENT
DEPOSITION RATES USING MAGIC
Duck Lake 1A2-004
1984
25 yr
50 yr
Trout Lake 1A2-054
1984
25 yr
50 yr
Island Pond 1B3-059
1984
25 yr
50 yr
80% CLD
-17.3
-14.4
-14.4
-6.0
1.0
0.5
-9.2
4.6
4.0
50% CLD
-17.3
-7.0
-6.9
-6.0
11.1
11.1
-9.2
23.9
23.9
6.4.3 Model Application to Aquatic Systems in Canada
The Jones et al. (1984) model was applied to lakes in eastern Canada. Results of this
application are presented in Table 6-18 and Figure 6-4. Table 6-18 includes predictions of the number
and percentages of lakes in eastern Canada projected to have an eventual steady-state pH <5 for six
acidic deposition scenarios. The number and percentages of lakes in eastern Canada currently
observed to have a pH < 5 are also included in the table. Figure 6-4 presents the highest and lowest
estimates of the number of lakes in eastern Canada with a predicted steady-state pH < 5 for two
alternative values of background sulfate (20 and 100 ueq L"1) and for each of the two Fw assumptions.
The geometric mean of the four predictions is also shown. Because two extra parameter combinations
were used in computing values in Figure 6-4, the range of prediction values was larger than values in
Table 6-18. Model predictions differed from currently observed conditions whenever the lakewater
sulfate concentration was not in equilibrium with the level of sulfate deposition.
Under current deposition, the number of lakes in eastern Canada estimated to eventually have
steady-state pH <5 ranged from 10,000 to 60,000 lakes (Figure 6-4), as compared with the 12,000
lakes estimated based on extensive surveys of present conditions in eastern Canada (Table 6-18;
Kelso et al. 1986).
6-27
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
V
s
05 OS
OZ0
>•=.
1<
£X
A
5«fc
w«w
S0^
o*s
§gQ
«3<
a«o
S5J
aP ^ ^ S 5 *? «.
S«3 33 « -• "3
CO — O
O —
— to oon
^- ^ >ot~
•a «e t- o
'
n **
<§«
§ S
3 31
«. ce «.
"
g
Si m
E- •*
n- C
«
O 10
o o
§ §
d d
oia
od
o
w
C4
Ii
5 «
o m
d. d
do
o
10
to
S3
on
oo
3
O
d
S
s
"8
00
3
<3
60
I
Q.
a
( 1 986 1 and
ham 1985.
ff
in J eil'r
d Cunn
a »
'c s
a =
6-28
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
i
Upper bound
~ Current lakes
withpH<6
Geometric mean
Lower bound
Current lakes
with pH< 4,7
i ri i
1 2 T 3
Total Sulfate Deposition (10*9, kg yr'1)
20 40 60 80 100 120 140
Proportional Deposition Control, Percent of Current Rate
9 - 18 2736
Threshold Deposition Control (kg SOa"2. ha'i yr'i)
Figure 6-4. Predicted relationship between the number of lakes with pH < 5 and total
suifate deposition in eastern Canada, south of 52°N. The lower and upper bounds
represent the minimum and maximum damage predictions from four model runs, using a
wide range of model parameters. The proportional deposition control axis assumes
proportional reductions in deposition in all secondary watersheds. The threshold
deposition control axis lowers deposition to the threshold or to the current deposition,
whichever is lower, in all secondary watersheds.
Source: Minns and Kelso (1986)
6-29
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
The model predictions indicated decreases in the extent of acidification were possible with
reduced deposition. Table 6-18 indicated that a 40% reduction in deposition would produce an 80 to
86% decrease in the predicted number of lakes in eastern Canada with a pH < 5 (9000 to 29,000
lakes), while a 20% increase in acidic deposition was predicted to cause 72 to 100% more lakes to
acidify (10,000 to 26,000 lakes).
Changing the target loading in Canada by 33%, from 27 to 18 kg SOi"2 ha"1 yr"1, decreased by
41 to 51% the predicted number of acid lakes in eastern Canada (3000-16,000 lakes). A further 25%
reduction in the target loading from 18 to 13.5 kg SCV2 ha"1 yr"1 (Table 6-18) lowered the predicted
number of acid lakes in Canada by 65 to 67% (400 to 2400 lakes). Note that the model predicted no
acid lakes under a zero deposition scenario. This probably was an underestimate, because of the large
number of acidic, highly colored lakes in eastern Canada.
The model predictions included both wet and dry deposition in the input loading level, so target
loadings expressed as wet deposition alone would be lower.
6.5 CONCLUSIONS AND RECOMMENDATIONS
6.5.1 Conclusions
Target loadings that were previously proposed to protect low, ANC systems generally fell in
the range of 10 to 20 kg SO4~2 ha*1 yr"1 (wet). The methods for estimating these loadings were (1) to
define a threshold (biological, chemical, or both) that provided acceptable levels of protection for
aquatic systems; (2) to estimate, generally using simple, empirical steady-state models, the loading
rate that would cause a lake or stream of a given sensitivity to reach the threshold; and (3) to
extrapolate the loading to the entire population of lakes and streams in a region. Recent advances in
providing estimates (2) and (3) using dynamic watershed models, and statistically based estimates of
the number of lakes of a given sensitivity have provided another approach for estimating the regional
effects of decreased loading.
Steady-state models estimated about 55 lakes (0.9%) in the Northeast might become acidic in
25 years at 80% CLD with an estimated 83 acidic lakes (1%) after 50 years at 80% CLD. No (0) lakes
were forecast to become acidic after 25 years at 50% CLD with 3 lakes (0.05%) forecast to become
acidic after 50 years at 50% CLD. In the Southeast, at 125% CLD, about 115 lakes and streams (4%)
were estimated to become acidic after 25 years; 1200 systems (40%) were estimated to become acidic
after 50 years; and 2300 systems (75%) were estimated to become acidic after 100 years.
The MAGIC model forecast about 140 (11%) Northeast lakes were estimated to become acidic
after 25 years at 125% CLD with about 325 acidic lakes (25%) after 50 years at 125% CLD. No (0)
lakes were forecast to become acidic in either 25 or 50 years at 80% CLD or 50% CLD.
6-30
-------
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Both the steady-state and MAGIC models indicated some currently acidic lakes would recover
(i.e., ANC >0 ueq L"1) with an 80% CLD or 50% CLD rate. The relative recovery was greater at the
50% OLD rate because of lower SC<4"2 inputs.
Modeling results from the Jones et al. (1984) model, which was applied to lakes in eastern
Canada, indicated that a total loading of 18 kg SO4"2 ha"1 yr"* would result in 4000 to 16,000 acidic
lakes or 0.6 to 2.6% of the total lakes in eastern Canada. At 100% CLD, the model estimated that
10,000 to 30,000 (1.6-5.9%) lakes would acidify. About 2400 to 5900 acidic lakes (0.4% to 1.0%) were
estimated at a target loading of 13.5 kg SCV2 ha"1 yr'1 (total deposition).
A reduction of current loading of 20% and 50% would result in a wet deposition of about 15 to
30 and 10 to 20 kg SO4"2 ha"1 yr"1 in the northeastern United States.
6.5.2 Recommendations
Because it is difficult to predict aquatic chemistry under current deposition, it is clear that
attempting the reverse procedure (estimating the loading that yields a desired chemical condition)
can only be approximate, with considerable uncertainty. Improvements in model predictions can be
brought about by improving estimates of the key model parameters: sulfate sorption equilibria,
cation exchange equilibria, mineral weathering rates, and hydrologic flow.
The greatest amount of information regarding aquatic system response comes from situations
in which deposition has been altered. Therefore, if target loadings and associated emissions are
imposed, it will be essential to carefully design a monitoring program to assess surface water
responses and provide appropriate information for policy considerations.
6.6 REFERENCES
Baker, L.A. 1984. Mineral and Nutrient Cycles and Their Effect on the Proton Balance of a
Softwater, Acidic Lake. Ph.D. Dissertation. Gainesville, FL: University of Florida.
Aimer, B., W. Dickson, C. Ekstrom, and E. Hornstrom. 1978. Sulfur pollution and the aquatic
ecosystem. In: J.O. Nriagu, ed. Sulfur in the Environment. Part II: Ecological Impacts, pp.273-311.
New York, NY: John Wiley and Sons.
Brydges, T.G. and B.P. Neary. 1984. Target loadings to protect surface waters. Speech to Air
Pollution Control Association. Annual Meeting, Edmonton, Alberta.
i
Church, M.R. 1984. Predictive modeling of the effects of acidic deposition. In: The Acidic
Deposition Phenomenon and Its Effects. Critical Assessment Review Papers, Vol. II. Effects Sciences,
pp. 4-113-4*127.
Church, M.R. and J.N. Galloway. 1984. Application of Henriksen's "acidification indicator" and
"predictor nomograph" to two Adirondack lakes. Air, Water, and Soil Pollut. 22:111-120.
6-31
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Dickson, W. 1986 Some data on critical loads for sulphur on surface waters. In: I. Nilsson, ed.
Critical Loads for Sulphur and Nitrogen, pp. 143-158. Report from a Nordic Working Group. Nordic
Council of Ministers. Stockholm.
Harvey, H.H., R.C. Pierce, P.J. Dillon, J.P. Kramer, and D.M. Whelpdale. 1981. Acidification
in the Canadian Aquatic Environment: Scientific criterion for assessment of the effects of acidic
deposition on aquatic ecosystems. NRC Canada Report No^ 18475, Ottawa, Ontario.
Henriksen, A. 1980. Acidification of freshwaters - a large scale titration. In: D. Drablos and
A. Tollan, eds. Proceedings of the International Conference on Ecological Imapct of Acid Deposition,
pp. 68-74. Oslo, Norway.
Henriksen, A. 1982. Changes in base cation concentrations due to freshwater acidification. Acid
Rain Research Report 1/1982. NIVA, Oslo, Norway.
Henriksen, A. 1984. Changes in base cation concentrations due to freshwater acidification. Verh.
Internal. Verein. Limnol. 22:692-698.
Henriksen, A., W. Dickson, and D.F. Brakke. 1986. Estimates of critical loads for sulphur to
surface waters. In: I. Nilsson, ed. Critical Loads for Sulphur and Nitrogen, pp.S7-l2Q. Report from a
Nordic Working Group. Nordic Council of Ministers. Stockholm.
Hultberg, H. 1985. Budgets of base cations, chloride, nitrogen and sulphur in the acid Lake
Gardsjon,SW Sweden. Ecol.Bull. 37:133-157.
Jeffries, D.S. 1986. Evaluation of the regional acidification of lakes in eastern Canada using ion
ratios. Proceedings for the ECE Workshop on Acidification of Rivers and Lakes. National Water
Research Institute, Contribution Series #86-79. Burlington, Ontario.
Jones, M.J. and G. Cunningham. 1985. Summary of analyses performed as a follow-up to the
regional acidification impact modelling project. Department of Fisheries and Oceans Canada.
Jones, .M.J., D.R. Marmorek, and G. Cunningham. 1984. Predicting the extent of damage to
fisheries in inland lakes of eastern Canada due to acidic precipitation. Department of Fisheries and
Oceans Canada.
KamarL J. 1986. Critical deposition limits for surface waters assessed by a process-oriented model.
In: I. Nilsson, ed. Critical Loads for Sulphur and Nitrogen. Report from a Nordic Working Group.
Nordic Council of Ministers. Stockholm.
Kelly, C.A., J.W.M. Rudd, R.H. Hesslein, D.W. Schindler, P.J. Dillon, C.T. Driscoll, S.A.
Gherini, R.E. Hecky. In Press. Prediction of biological acid neutralization in acid-sensitive lakes.
B iogeochemistry.
Keiso, J.R.M., C.K. Minns, J.H. Lipsit, and D.S. Jeffries. 1986. Headwater lake chemistry
during the spring freshet in north-central Ontario. Water, Air, and Soil Pollut. 29:245-259.
Minnesota Pollution Control Agency. 1985. Proposed Acid Deposition Standard and Control
Plan. Statement of Need and Reasonableness.
National Academy of Sciences. 1983. Acid Deposition: Atmospheric Processes in Eastern North
America. Washington, DC: National Academy Press.
6-32
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
Newcombe, C.P. 1985. Acid deposition in aquatic ecosystems: setting limits empirically. Env.
Management 9(4):277-288.
Nichols, D.S. and E.S. Verry. 1985. Evidence for the cultural acidification of lakes in the northern
lake states. Proceedings for the Conference on Air Pollutant Effects on Forest Ecosystems, pp. 253-265.
The Acid Rain Foundation. St. Paul, MN.
Nilsson, I. (ed.) 1986. Critical Loads for Sulphur and Nitrogen. Report from a Nordic Working
Group. Nordic Council of Ministers. Stockholm.
Oppenheimer, M. 1984. Reducing acid rain: The scientific basis for an acid rain control policy.
Environmental Defense Fund. MPCA Exhibit 181.
Rogalla, J. and P. Brezonick. 1985. Empirical modeling to predict acidification of Minnesota
lakes. Report to the Minnesota Pollution Control Agency. MPCA Exhibit 182.
Schnoor, J.L., N.P. Nikolaidis, and G.E. Glass. 1986. Lake resources at risk to acidic deposition
in the Upper Midwest. J. Water Pollut. Cont. Fed. 58:139-148.
Thompson, M.E. 1982. The cation denudation rate as a quantitative index of sensitivity of eastern
Canadian rivers to acidic atmospheric precipitation. Water, Air, and Soil Pollut. 18:215-226.
United States/Canada. 1983. Memorandum of Intent on Transboundary Air Pollution. Impact
Assessment, Work Group I.
Wright, R.F. 1983. Predicting acidification of North American lakes. NIVA Report 0-81036.
Norwegian Institute for Water Research, Oslo, Norway.
I
6-33
-------
I
I
I
I
I
I
I
I
-------
I
I
I
I
I
1
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
- DO NOT CITE OR QUOTE
SECTION 7
RATES OF RECOVERY
7.1 SUMMARY
If acidic inputs to an aquatic system are reduced, the natural processes of ANC generation will
eventually increase the ANC and pH of the system. Theoretical considerations and empirical
evidence indicate that the temporal trends of ANC, pH, and SO^2 concentrations in surface water
during the recovery phase will not be the reverse of the acidification phase (i.e., hysteresis occurs
during recovery). Furthermore, simulation models indicate that the rate of recovery for an acidic
lake might be slower than the rate at which it became acidic. These dynamic models indicate that the
response of lakes to reduction of deposition depends most strongly on the hydraulic residence.time
and the nature and relative magnitudes of the various hydrologic flow paths.
The dynamics of acidification and recovery reflect different geochemical processes.
Acidification rates are controlled by SCV2 adsorption characteristics of the soil and the capacity of the
soils to replace H+ with base cations. During recovery, the asymptotic approach of ANC and pH to
pre-acidification levels is regulated by the rate of supply of base cations to the soil exchange complex
and to the soil solution from mineral weathering.
The observed recovery of lakes in Sudbury and western Sweden and of rivers in Nova Scotia
has been an immediate response to decreases in sulfur deposition. In contrast, reductions in nitrogen
deposition or reductions in H+ deposition are not necessary to induce recovery. The critical factor
controlling the recovery of once-acidic systems appears to be a decrease in sulfate deposition.
7.2 INTRODUCTION
Aquatic systems will recover from the effects of acidic deposition after emissions of SOa and
NOx have been reduced; in question are the rate and extent of recovery, and whether recovery will
result in the same biological community as existed prior, to acidification. Recovery may be the result
of either of two conditions: (1) the reduction of acidic inputs to the system, or (2) the introduction of
substances to the system that neutralize acidity or increase the rate of ANC production. In the first
condition, the major issues are the rate and the extent of both chemical and biological recovery. In
the second condition, the major issues are the rate and the extent of biological recovery, following the
return of pH to circumneutral values, and the continuing costs of maintaining circumneutural pHs.
If acidic inputs to the system are reduced, the natural processes of ANC generation, such as soil
and rock weathering and inlake bacterial sulfate reduction, will eventually increase the ANC and pH
7-1
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
of the system. The rate and extent of this increase depend on many factors, including the magnitude
of the deposition reduction and numerous lake and watershed characteristics.
If a mitigation program such as lake or watershed liming is undertaken, the system will
increase in ANC to the extent and at the rate that the ANC-generating substance (such as lime) is
applied, or that it reacts with the water. Mitigative measures will result in a long-term increase in
ANC only if they are applied continually and at a rate exceeding the rate of acidic deposition,
7.3 MODELING APPROACHES TO ESTIMATING RATES OF RECOVERY
Simulation models, developed to project future acidification of aquatic systems, can also be
used to forecast the time trend and chemical extent of recovery for acidified systems, following
reductions in acidic deposition. The dynamic watershed models used to forecast recovery from
acidification are described in Section 6.4.2.4.
Cosby et al. (1985) used the Model for Acidification of Groundwaters in Catchments (MAGIC)
to estimate the time trend of acidification and recovery of a hypothetical catchment subsequent to
changes in deposition rates (Figure 7-1). Recovery, simulated by a stepped reduction in deposition to
estimated background sulfate levels, followed a 120-year period of acidic deposition. Several phases
can be identified in the recovery or post-acidification period. Initially, following the cessation of
acidic deposition, there was an immediate reduction of SO4"2 and H+ concentrations. Streamwater
concentrations of SO4~2 declined as a function of the S04"2 sorption capacity of the soils and the degree
of reversibility of sorption. Streamwater base cation concentrations were depressed below pre-
acidification levels, and ANC increased as fewer H* and Al+3 were needed to balance the flux of
strong acid anions. However, Streamwater concentrations of H* and ANC cannot completely recover
until soil base saturation has returned to pre-acidification levels; the resupply of base cations to
exchange sites in turn is controlled by mineral weathering. The simulation results from MAGIC for
this hypothetical watershed imply that recovery takes twice as long as the acidification process.
MAGIC forecasts assuming a 80% and 50% reduction from the current level of deposition
(CLD) for three Northeast watersheds were discussed in Section 6 and shown in Table 6-17. In these
simulations, all three systems were forecast to respond relatively rapidly during the first 10 years
following decreased deposition and then to slowly approach a new steady state during the next
40 years. Two of the three lakes recovered within 25 years at both 80% and 50% CLD. These results,
however, indicated recovery rates were quite variable. Hysteresis effects during the recovery phase
can be investigated using MAGIC, but were not evaluated for these forecasts.
7-2
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
I
Stage 1 Stage 2
200-
io deposition
120 160
Years
I MttmiMMMt*
200
Postacidic
deposition
240
280
Figure 7-1. Response of hypothetical catchment having moderate SCV2 adsorption to
square wave of deposition: A, changes in sum of strong base cations (SBC), sum of
strong acid anions (SAA), and alkalinity (ALK) of streamwater; B, changes in pH of soil
water, streamwater, and bulk deposition.
Source: Cosby et al. (1985)
I
7-3
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
The Integrated Lake/Watershed Acidification Study (ILWAS) model was used to examine the
effects of reduced sulfur loads on the chemistry of two lakes located in the Adirondacks, Woods and
Panther (Chen et al. 1985; Gherini et al. 1985). Current annual sulfur loading rates of about
1200 eq ha'i were reduced by 50% to 600 eq ha"i. Input data from 1979 to 1981 were run sequentially
for 12 years to provide a synthesized base period for evaluating the effects of the reduction in sulfur
loading. Panther Lake, which lies in a watershed with deeper soil than Woods Lake, showed a
predicted increase in mean annual (air-equilibrated) pH from 7.22 to 7.34 in the first three years. The
rate of recovery was predicted to decrease with time, however, and after an additional two years of
reduced sulfur loadings the pH had increased to only 7.35. The predicted response of the more poorly
buffered Woods Lake was more substantial and rapid.
7.4 EVIDENCE FOR CHEMICAL AND BIOLOGICAL RECOVERY
7.4.1 Evidence of Chemical Recovery from Deposition Reductions
Investigations of the effects of reductions in deposition are difficult because there are few long-
term records of surface water chemistry and biota. Nevertheless, studies conducted in areas where
deposition has decreased are invaluable, because they provide the strongest empirical evidence either
for or against current acidification models and hypotheses. Of particular interest are the changes in
cation export rates and changes in surface water concentrations of SO/t"2, pH, and organic acids that
result from changes in acidic deposition. Cation export rates are expected to decrease as SO,*"2
deposition declines. The F-factor (the ratio of the change in Ca+Mg to the change in lakewater
SO4~2) is generally positive and has an important effect on the predictions of most steady-state models
(Henriksen 1982; Wright 1983; Jones et al. 1984). According to Krug and Frink (1983), humic acids
will be released from the soil following a reduction in strong acidic deposition, with "little or no
measurable change in pH."
The areas for which long-term records exist and where deposition reductions have occurred are
Nova Scotia, the Sudbury area of Ontario, and western Sweden; the chemical recovery of aquatic
systems in these three areas will be discussed in Sections 7.4.1.1 and 7.4.1.2, and the biological
recovery will be presented in Section 7.4.2.
7.4.1.1 Chemical Recovery in Canada
In Nova Scotia, a decrease in deposition over a 5- to 10-year period has coincided with
decreased concentrations of sulfate in surface waters and increased ANC and pH (Thompson 1986).
Runoff-corrected cation export rates have remained constant.
In the Sudbury area, both large-scale surveys (Keller and Pitblado 1986) and site-specific,
intensive studies (LaZerte and Dillon 1984; Dillon et al. 1986; Hutchinson and Havas 1986) provide a
7-4
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
I
I
I
I
detailed picture of surface water responses to reductions in deposition. Emissions of SQ% from
Sudbury smelters have substantially declined since the 1960s: 4240-7034 tonne/day during the
period 1960 to 1969; 3663-6383 tonne/day from 1970 to 1977; and 1065-2562 tonne/day from 1978 to
1983 (Keller and Pitblado 1986). These declines in emission rates have resulted in a nearly fourfold
decrease in average SO^2 deposition during the summer months from 254 eq ha"1 month"1 in 1970 to
65 eq ha"1 month'i in 1977 (Hutchinson and Havas 1986). Significant declines in trace metal
emission have accompanied the SOj emission reductions.
During the summers of 1982 and 1983, Keller and Pitblado (1986) resampled 209 Sudbury area
lakes originally sampled in 1974 to 1976. Between the two study periods, smelter emissions of S(>2
declined by about 50%. Though sampling methods differed somewhat between the two periods,
comparative tests indicated that these differences in methods did not cause significant differences in
measured parameters. A plot of average lake pH in the 1974 to 1976 period against average pH
recorded from 1981 to 1983 demonstrates that the more acidic lakes (most of which are located
relatively close to Sudbury) have experienced pH increases (Figure 7-2). Circumneutral lakes, most
of which are relatively far from Sudbury, showed no consistent pattern of change. Reductions in
average concentrations of sulfate, nickel, and copper were also greatest in lakes closest to Sudbury.
8.0'
7.0
6.0
X
a
IX
5.0'
4.0'
4.0
5.0
6.0
xpH 1974-1976
7.0
8.0
Figure 7-2. Average pH in 1974-1976 plotted against average pH in 1981-1983 for the
study lakes. The line represents a 1:1 relationship. Numbers indicate coincident points.
Source: Keller and Pitblado (1986)
7-5
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE ORQUOTE
Keller and Pitblado (1986) performed more detailed analyses for a subset of 21 lakes, each of
which had an average pH <5.5 during the 1974 to 1976 period. They found statistically significant
declines in sulfate and copper concentrations (though no declines in nickel) and statistically
significant increases in pH. Significant correlations were found between the decrease in sulfate,
nickel, and copper concentrations in the 21 lakes and the distance of these lakes from the smelters.
On the other hand, watershed-to-lake-area ratios did not correlate significantly with the changes in
pH, nickel, or copper, and showed only a weak positive correlation with decreases in SO4~2
concentrations. These findings suggest that deposition levels were more important than watershed
factors in determining the rate and extent of surface water chemical responses.
In addition to the surveys of Keller and Pitblado (1986), water chemistry data from five
intensively studied lakes in the Sudbury area (Clearwater, Lohi, Middle, Hannah, and Swan) have
been analyzed by Dillon et al. (1986); the data from Clearwater and Swan are discussed in LaZerte
and Dillon (1984). These data show substantial declines in SCV2 concentrations in all five lakes
during the period from 1978 to 1984, ranging from 25 to 62%. Differences among lakes in the
proportional declines were probably due to differences in three factors: distance from Sudbury; lake
order (headwater versus second-order lakes); and the extent of anaerobic hypolimnia (bottom waters
devoid of dissolved oxygen), which affects bacterial sulfate reduction (LaZerte and Dillon 1984; Dillon
et al. 1986). LaZerte and Dillon point out that the changes in lake sulfate concentrations should not
be directly proportional to the changes in local emissions, .because some portion of the area's
deposition originates outside the region.
Three of the five lakes were limed; therefore, the changes in pH for these three lakes cannot be
attributed to changes in atmospheric deposition. In the other two lakes, mean pH levels increased
from 4.23 (1973-77 mean) to 4.61 (1984) in Clearwater and from 3.96 (1977) to 4.80 (1982) in Swan.
The H* concentrations in Clearwater declined by over 50%, decreasing from an average of 59 ueq L"1
in 1973-1977 to 25 ueq L"1 in 1983. During this period, sulfate concentrations decreased from an
average of 545 ueq L'1 in 1973-1977 to 379 ueq L'1 in 1983, and essentially paralleled declines in lake
water H+ and decreases in emissions of SC*2 from smelting activity in the Sudbury region ("see
Figure 7-3).
Swan Lake, which was studied less intensively during the same period, showed even more
dramatic shifts in H+ and SC*4"2 concentrations; H* activity declined 85% (from pH 3.96 to 4.80)
between 1977 and 1982, and SOi'2 concentrations decreased by 62% from 580 to 220 ueq L'l. LaZerte
and Dillon (1984) suggest that the different recovery rates shown by the two lakes may reflect
differences in hydraulic flushing rates. The hydraulic residence time of Swan Lake is only about
1 year compared to approximately 3 to 4 years for Clearwater Lake.
7-6
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
1000-1 100-1
I
I
I
I
I
I
I
or
v
3500-
o
{ft
o-1
tr
jj5
+
o-1
Year
Figure 7-3. Sulfur emissions at Sudbury and average H + and SO4"2 concentrations in
Clearwater Lake, 1973 to 1983.
Source: Lazerte and Dillon (1984)
The data from Clearwater and Swan can be used to examine the two hypotheses mentioned at
the beginning of this section. Swan Lake is the only one of the five lakes not to have received
substantial quantities of road salt, therefore permitting an estimate of the F-factor, the ratio of the
change in Ca+Mg to the change in SO4-2 (Henriksen 1982). Dillon et al. (1986) estimated F = 0.64 for
Swan Lake, a value higher than that reported by Henriksen (1982) for other locations and indicative
of substantial changes in cation export (or substantial transport of cations between the sediment and
water) associated with changes in deposition. LaZerte and Dillon (1984) used the results from
Clearwater Lake to refute the hypothesis of Krug and Frink (1983), who contend that a reduction in
deposition of strong acid will induce the release of equivalent amounts of organic acids with
subsequently little change in pH. According to the Krug and Frink hypothesis, reductions in H* and
SC>4~2 deposition in Sudbury should have been accompanied by increases'in organic acid
concentrations, rather than increases in pH. No changes in the organic acid content of either
Clearwater Lake or Swan Lake were observed, although both lakes experienced substantial increases
inpH.
In Baby Lake, a fairly small lake (11.7 ha) in a watershed dominated by exposed granitic and
gneissic bedrock, SC>4"2 concentrations have declined by 50% and pH has increased from 4.05 in 1972
to 5.8 in 1984 as a result of decreased emissions from smelting activities near Sudbury (Figure 7-4;
7-7
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Hutchinson and Havas 1986). Conductivity levels also have declined correspondingly. Similar
results have been observed for nearby Alice Lake (Figure 7-4; Hutchinson and Havas 1986), although
the increase in pH for this lake is not as striking as that of Baby Lake. The slower rate of pH recovery
reflects the comparatively better buffering of Alice Lake. This is because the Alice Lake watershed is
dominated by granitic till, whereas Baby Lake lies in a watershed of exposed bedrock. Also, it should
be noted that the pH recovery of Alice Lake may reflect high ANC inputs from an upstream slag pond.
6-
4_
r = 0.947**
1970
1975
1980
Baby Lake
-6
-5
-4
60-
'40-
20-
Sulfate
r
r =-0.928**
1970
1975
1980
mM
-6
-4
••*?
B
7.0-
X
0.
6.0-
Alice Lake
-7.0
-6.0
300-
J200
01
E100
1970
1975
1980
Sulfate
r = -0.83**
T
1970
1975
1980
mM
h3
Figure 7-4. The pH and sulfate concentrations in (A) Baby Lake and (B) Alice Lake
from 1968 to 1984. The straight arrow represents the closure of the Coniston smelter
and corresponds to a significant improvement in local air quality. The open circles
are data from International Nickel Company. The asterisks (**) represent a
significant correlation at the 1% level of probability.
Source: Hutchinson and Havas (1986)
7.4.1.2 Western Sweden
In recent years, SC*4"2 and H+ concentrations in a number of lakes in western Sweden have
progressively declined after reaching maximum levels in the mid-1970s. Forsberg et ai. (1985)
documented reductions of 40 to 100 ueq L'l for SO/i"2 in several lakes as well as the river Atran in
western Sweden (Figure 7-5). These declines followed reductions in measured regional SC<4~2
deposition of 15 to 21% that occurred between the periods 1962-1973 and 1974-1983 (Figure 7-6).
78
-------
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
I
I
I
I
I
I
I
500-
^300-
100-
1
1965
67
1
69
71
1 1 1
73
Year
75 77
79
1
81
Figure 7-5. Sulfate concentrations in surface waters in
western Sweden. A, River Atran, annual (n=12)
average values (data from the Swedish National
Environment Protection Board, MK-LAB, Uppsala); B,
Lake Rishagerodvatten; C, Average values, vegetative
season, seven lakes based on 45 to 60 samples per lake;
D, Lake Horsikan; E, Lake Stora tssjon.
Source: Forsbergetal. (1985)
Corresponding increases in pH
of 0.3 to 0.4 units accompanied
the reductions in lakewater
SO*"2. Decreases in surface
water H* correlate closely
with changes in SO4~2; similar
correlations were observed
during the acidification phase
as well (Figure 7-7). The
decreases in surface water H+
observed in western Sweden
occurred despite an increase in
H+ deposition relative to rates
observed during the acidifi-
cation phase of the mid-1970s.
Total H* deposition during the
recovery period (1977-1983)
was 2440 eq ha'1 compared to
1140 eq ha"1 during the
acidification phase (1970-
1976), apparently as a result of
increased NOx emissions.
These empirical studies
illustrate several phenomena
that must be considered in the
context of lake recovery from
acidification. Recovery of
lakewater pH levels has
proceeded in Sudbury, Nova
Scotia, and western Sweden in
the absence of corresponding declines in' NOs" deposition. In these areas, the driving variable
appears to be reductions in atmospheric loadings of SO^2. Indeed, recovery of the lakes in western
Sweden has occurred despite a nearly twofold increase in H* deposition. This phenomenon
underscores the importance of controlling SO2 emissions rather than NOx emissions. Both of the
emitted gases are Lewis bases subject to biogeochemical reduction or consumption in natural waters;
»- -
_l
or -
a -
O -
Sohus Malm on
(Lon. -11.37, Lat
o
O
Forshult
Lon. -13.78, Lat.
r • i • i •
1962 64 66
0
* 58.20) A
A
V~"\A ^V
°-° \ /— ° \/
\ / °
4-60.08
68 70 72 74 76 78 80 82
Year
1 21 5%
1 1 5%
84
Figure 7-6. Concentrations of SO4~2 (neq L'1) at two
stations in western Sweden, 1962-1983. Each value is a
weighted annual mean (n=12). The 1972 value for
Bohus Malmon was omitted when calculating the
concentration level for the period 1962-1973.
Source: Forsberg et al. (1985)
7-9
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
I
20
1977-1983
Lake Ramsjon
OMOH
a
0 -
1968-1976
B
30-
Lake St. Halevatten
= 0.85
.1976-1981
the net result of such reactions is, in general, the production of an equivalent amount of ANC (Cook et
al. 1986; Rudd et al. 1986). Atmospheric deposition of NOx to surface waters is almost completely
consumed (>95%) in all but the most extreme examples of NOa" loading (Driscoll and Newton
1985); SO4"2 retention in watersheds
and SCV2 losses due to bacterial
reduction typically are less than
NOs* uptake and in some cases are
minimal (Rudd et al. 1986; Kelly
et al., in press; Cook et al., in press).
The Swedish experience also
illustrates another important facet
concerning the reversibility of lake
acidification: hysteresis. Although
reductions in SO4*2 loadings have
resulted in an almost immediate and
corresponding reduction in lakewater
SO4~2 concentrations, the associated
increases in pH indicate that only
partial recovery from acidification
has occurred for a given return to pre-
vious SO4~2 levels. This phenomenon
is illustrated in Figure 7-7 for Lakes
Ramsjon and St. Halevatten in
western Sweden. Comparison of H+
activities as a function of acidifica-
tion and post-acidification phase
SO4~2 levels shows that H"1" activities
during the recovery phase are higher
than corresponding H+ activities
10-
.0 -
1968-1975
150 200
I
250
I
300
I
I
I
I
I
I
during the acidification phase.
Forsberg et al. (1985) suggest that
the hysteresis may reflect increased
H* loadings during the recovery
phase. Although there is some merit
to this explanation, the thermody-
I
Figure 7-7. Increases and subsequent decreases
in H + and sulfate concentrations in Lake
Ramsjon and Lake Stora Halevatten, western
Sweden, 1968 -1983.
Source: Forsberg et al. (1985)
I
7-10
I
-------
I
I
I
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERN ALUSE ONLY
DO NOT CITE OR QUOTE
namics of ion exchange dictates that surface water ANC will not return to pre-acidifieation levels
until base saturation is restored to pre-acidification levels (Cosby et al. 1985). More specifically,
infiltration of ground water during the recovery phase of reduced acid loadings results in an exchange
of H* adsorbed on soil ion-exchange sites for Ca+2 and other base cations in solution until an
equilibrium is reached; this equilibrium is established between the concentrations of H+ and base
cations being deposited and the concentrations of these constituents adsorbed to soil particles. If, as
in the case of the watersheds in western Sweden, increases in NOg" loading have not exceeded the
capacity of the watershed for assimilation, ion-exchange effects may indeed account for the
hysteresis. As time progresses and soil base saturation returns to pre-acidification levels, lakewater
pH and ANC levels will asymptotically approach their original values.
7.4.2 Evidence for Biological Recovery from Deposition Reductions
Although water quality has improved in Sudbury area lakes, many lakes still have a water
quality that is unsuitable for fisheries (e.g., 10% have a pH <5.0; 16% have a pH <5.5 [Keller and
Pitblado 1986J). Information on fish communities in the affected lakes is scarce, though Keller and
Pitblado report preliminary evidence of improvements in two lakes.
Yan (1985a) examined the zooplankton community of Clearwater Lake, based on intensive
sampling during the period from 1973 to 1984. He computed 11 zooplankton community parameters
that have been empirically related to lakewater pH or metal levels in published surveys and assessed
whether these parameters had changed significantly over time. There were no significant changes in
any of the 11 parameters, indicating that the community is not yet recovering. Analysis of possible
reasons for the lack of recovery showed that the only feasible explanation is continuing toxic
concentrations of H+, copper, and nickel.
7.4.3 Biological Recovery following Chemical Restoration: Results from Canada
Lakes and streams in the Sudbury area, in other areas of Ontario, and in Nova Scotia have
been limed. The results of these mitigation efforts are described in a report by the Federal/Provincial
Research and Monitoring Coordinating Committee (1986), from which most of the following text is
drawn.
7.4.3.1 The Sudbury Lakes Environmental Study
Between 1973 and 1975, three highly acidic, metal-contaminated lakes near Sudbury, Ontario,
were limed by the Ontario Government using a combination of calcium hydroxide and calcite in an
attempt to restore fish (Yan and Dillon 1984). Copper and nickel levels in lakewater were extremely
high prior to liming because of metal deposition from nearby smelting operations. Although liming
resulted in major increases in pH (topH>7) and reductions in metals levels, the waters remained
7-11
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
toxic as a result of residual metals (Yan and Dillon 1984). Introductions of smallmouth bass, Iowa
darter, brook stickleback, brook trout, and rainbow trout were all unsuccessful because of this metal
toxicity. As inputs of acid and metals from atmospheric deposition continued, the lakes began to
reacidify and metals concentrations increased to their original levels.
Two of the lakes were fertilized with several annual additions of phosphate and ammonium
nitrate after liming (Yan and Lafrance 1984). Although fertilization of limed lakes showed some
promise as a tool for delaying reacidiiication in the Sudbury study, several problems were apparent:
(1) concentrations of labile forms of metals were unaffected by nutrient additions;
(2) additions of Nr^NOs led to pH decreases because of the preferential uptake of
by algae; and
(3) phytoplankton communities had higher rates of primary productivity in response
to fertilization than did circumneutral waters of similar trophic status, perhaps
because higher trophic levels were not present to limit algal biomass in the limed
lakes (Yan and Lafrance 1984; Marmorek 1984).
Yan (1985b) compared the zooplankton communities of two limed lakes with 15 nonacidic
reference lakes from the Muskoka-Haliburton Region. In spite of having circumneutral pH for close
to a decade, fundamental qualities of the zooplankton communities in the limed lakes (including total
zooplankton biomass, species richness, mean organism size, relative daphnid biomass, relative
cyclopoid biomass, relative predator biomass, and the coefficient of variation in total biomass) differ
significantly from those of the reference lakes. Possible reasons for the levels of recovery are the
absence offish and invertebrate predators.
7.4.3.2 Experimental Lake Neutralization in Ontario
In Ontario, the Ministries of Natural Resources and the Environment have undertaken (as of
1981) an Experimental Neutralization Program, to assess the feasibility of using lake liming to
rehabilitate acidified lakes, and to protect the biological communities in acid-stressed lakes. The
program is still under way, but some preliminary findings have been published (Booth et al., in
press).
In 1983, the neutralization of Bowland Lake, an acidic lake near Sudbury, Ontario, that
formerly supported a viable lake trout population, raised the whole-lake pH from 5.0 to 6.8 and
resulted in reduced concentrations of total aluminum (from 130 to 65 ug L"1). Young lake trout (aged
0+ and 1+) that were stocked after liming survived and exhibited good growth. Transferred adult
lake trout spawned in both years after liming. The resident yellow perch population showed both
increased growth (older cohorts) and decreased growth (younger cohorts) and a fourfold increase in
number. These responses of the yellow perch populations have not yet been attributed to the liming.
7-12
-------
I
I
I
I
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
The long-term response of Rowland Lake fish populations has yet to be determined, but chemical
models suggest that, if it is not relimed, Bowland Lake will reacidify to pre-neutraiization water
quality by 1990. The lake trout population may be in jeopardy much earlier.
Trout Lake, a low ANC lake near Parry Sound, Ontario, was limed in 1984. No immediate
response of the fish community was observed; however, the community did not appear stressed prior
to lake neutralization. After the neutralization began, Booth et al. (in press) observed 100%
mortality of lake trout fingerlings held in cages at one near-shore site during spring snowmelt,
suggesting whole-lake liming was not entirely effective in mitigating episodic pulses of acidic
snowmelt at all littoral (near-shore) sites.
Because whole-lake neutralization may not protect critical spawning habitat, Ontario is
experimenting with shoal liming (the addition of limestone gravel to spawning shoals) as an
alternative of whole-lake liming.
Ontario's current approach to neutralization is an experimental one. The province maintains
that abatement of emissions at the source is the only measure to restore water quality that has been
affected by acidic deposition. Until the long-term response of fish communities to neutralization has
been assessed, the use of lake liming will not be widespread.
7.4.3.3 Salmon River-laming in Nova Scotia
Long-range transport of sulfuric acid has caused the extinction of Atlantic salmon (Salmo
solar) stocks in 13 Nova Scotian rivers and severe declines in an additional 18 rivers (Watt et al.
1983; Watt, in press). Fearing further losses, the Department of Fisheries and Oceans (Canada) has
undertaken experiments (Watt et al. 1984; White et al. 1984) to test the feasibility of mitigating the
acidification of Atlantic salmon rivers in Nova Scotia by adding limestone or other substances to
lakes and streams. Two liming methods have been tested extensively, instream limestone gravel
additions (at six river treatment sites) and headwater lake liming (in five lakes). Estimates have
been made of the relative costs and effectiveness of these mitigation techniques.
Three years of observations at six sites where limestone gravel had been placed in the
streambed produced disappointing results (Watt et al. 1984) and led to the conclusion that this
approach is impractical (Watt, in press). Low water temperatures during winter and early spring
reduced the rate of limestone dissolution, and, at high flows, the amount of limestone gravel that
would theoretically be required to increase pH by as little as 0.3 units was so large as to be
prohibitive.
The second approach has been the liming of headwater lakes to create a reservoir of treated
water, which naturally discharges from the lakes.to protect downstream salmon habitat. The
7-13
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
limestone was usually added to the lakes as a slurry sprayed evenly over the entire surface. Watt
et al. (1984) showed in a four-lake study that, even in lakes with mean residence times as short as five
months, satisfactory lake pH levels can be maintained for up to one year if the limestone dose exceeds
three times the whole-lake acidity. The overdosing provides an excess of lime, which settles to the
bottom, thus effectively sealing off acid demand from the sediments and dissolving slowly to provide
additional ANC at each fall and spring turnover. The treatment must be repeated annually in lakes
having mean residence times of less than one year. Calcitic limestone was found to give better
dissolution efficiencies than the dolomitic variety.
A major problem with the headwater lake liming approach, as noted by White et al. (1984), is
that during episodes of heavy autumn and winter rains the lakes can become covered by a thin
surface layer of highly acidic rainwater which then flows out of the lake resulting in a downstream
acidic shock. This phenomenon is most common under ice cover, when an inverse thermal
stratification prevails (Watt et al. 1984). Watt (in press) reports that this problem has been overcome
by spreading a layer of dry limestone powder evenly over the ice surface. This approach is effective in
preventing the buildup of an acidic surface layer, because the first water entering the lake from a
winter rainstorm is drainage from the ice cover.
Positive responses in Atlantic salmon were obtained with both the instream-gravel and
headwater-lake liming methods (Watt et al. 1984). Significantly higher densities of juvenile salmon
were found in the immediate vicinity of instream limestone gravel deposits, but this beneficial effect
did not persist far downstream. In the lake liming studies, salmon parr introduced into the
previously barren and toxic outlet stream of Sandy Lake were still surviving one year after liming. In
addition, wild native salmon adults migrated (apparently attracted by the higher pHs) into the
previously unused outlet stream and spawned successfully. The resulting native salmon fry were
showing good survival up to one year after liming.
A large area of Nova Scotia has already been rendered barren of Atlantic salmon (Watt et al.
1983). After the acidic deposition problem is brought under control, it will be necessary to initiate a
salmon restoration program for the barren habitats. Restoration would be easier and have a higher
probability of success if a number of nuclei of native wild local stocks were available to provide a
genetically diverse selection of potential donor stocks. To achieve such a gene bank, Watt (in press)
recommends that a number (approximately six) of the 18 river stocks currently threatened with
extinction be preserved by using the headwater lake liming technique to create deacidified refuges in
tributaries.
Estimating the costs associated with liming lakes in eastern Canada and rivers in Nova Scotia
to mitigate the effects of acidic deposition is beyond the scope of this assessment. Because of the large
7*14
-------
I
I
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
numbers of lakes in eastern Canada that are acidic (about 10,000 have ANC <0 ueq L"1; Kelso et al.
1986), liming lakes would be expensive. Recent estimates indicate that the costs would be many tens
of millions of dollars (Canadian; Minns and Kelso, in press).
7.4.3.4 Canadian Mitigation Studies: Conclusions
Mitigation studies in Ontario and Nova Scotia suggest that, in the absence of metals
contamination, biological recovery can occur, though the rate and extent are still uncertain. There is
a need for continued monitoring of both chemical and biological recovery following liming. Such data
will be particularly valuable for evaluating the extent to which lakes and streams exhibit a
hysteresis in recovery. . ,. .
With respect to mitigation, the lake and river liming experience in Ontario and Nova Scotia
highlight the problems inherent in using full-scale liming operations as a way to rehabilitate and
protect aquatic ecosystems:
• in highly metal-contaminated systems, liming may not reduce metal concentrations
to nontoxic levels;
• without concomitant reductions in acidic deposition, liming becomes a continuous
process;
* liming may not protect the habitat for critical developmental periods of all fish
species;
• the long-term response of resident fish populations to liming has not yet been
assessed; and
• the full recovery of aquatic systems may require the introduction offish.
7.5 CONCLUSIONS AND RECOMMENDATIONS
7.5.1 Conclusions
It is clear from the Sudbury, Nova Scotia, and western Sweden studies that declines in
emissions and deposition lead to chemical recovery. Levels of ANC and pH increase; sulfate
concentrations (and metals concentrations in the case of Sudbury) decline. Data are insufficient to
evaluate the regional extent of biological recovery, although chemical concentrations and site-specific
studies near Sudbury suggest that biological recovery is probably either absent or is only very slight
in many lakes, perhaps because of high levels of trace metals remaining after decreased deposition.
The conclusions from modeling studies and from field studies of changes in lake chemistry
resulting from decreases in atmospheric deposition are as follows.
(1) Consistent with the mobile anion hypothesis, lakes in Sudbury and western Sweden
and rivers in Nova Scotia have responded immediately to changes in sulfur
7-15
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
deposition. Corresponding reductions in nitrogen deposition or, as has been the case
for the lakes in western Sweden, reductions in H* deposition are not necessary to
induce recovery; rather, the critical variable appears to be sulfur.
(2) Theoretical considerations and empirical evidence both indicate that an initial
hysteresis effect will develop for pH and ANC as lakewater SC^"2 concentrations
return to pre-acidification levels. Simulation studies using MAGIC show that
complete recovery of pH and ANC will not occur until base saturation is restored to
pre-acidification levels.
(3) Simulation models indicate that the response of lakes to reductions in acidic
deposition will be dictated by the hydraulic residence time and the nature and •
relative magnitudes of the various water flow paths. Systems dominated by direct
precipitation inputs and with short hydraulic residence times will respond both more
rapidly and with greater relative changes in concentrations of SC>4"2 and H+ than
watersheds with long SCV2 residence times. Catchments having low cation
exchange capacity and little ability to adsorb SO^"2 will have response times
approximating a year. Catchments with more moderate ability to adsorb SC>4~2 will
both become acidified and recover from acidification at a substantially slower rate.
7.5.2 Recommendations
The key processes regulating the rate and extent of recovery of aquatic systems from
acidification are (1) base cation production from cation exchange and from mineral weathering and
(2) sulfate sorption. Because knowledge of the mechanisms and rates of base cation production and
sulfate sorption is incomplete, future research is needed to examine these processes during recovery.
Estimates need to be made of rates of cation exchange, of mineral weathering, and of replenishment
of the soil cation exchange complex from mineral weathering. For example, if the cation exchange
complex is depleted during acidification and the rate of base cation replenishment from weathering is
slow, then the rate of recovery will likely be slower than acidification and will be highly dependent on
the rate of mineral weathering. The slower recovery rate than acidification rate in western Sweden
has been interpreted as a slow return to pre-acidification levels of base saturation.
Additional work is also required to determine if processes controlling the desorption of sulfate
from soils are the same as those controlling adsorption. The sulfate recovery would be faster if sulfate
sorption were irreversible than if it were reversible.
Detailed, process-level studies of watershed recovery from acidification, similar to the
Reversing Acidification in Norway Study, are required to determine the mechanisms and rates of
recovery. These types of studies could address the issues of base cation production and sulfate
adsorption. Chosen watershed sites should represent a range of sulfate adsorption and base cation
exchange characteristics to allow estimates to be made of regional differences in recovery.
7-16
-------
I
I
I
I
I
I
I
I
I
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
7.6 REFERENCES
Booth, G.M., J.G. Hamilton, and L.A. Malat In Press. Whole-lake and shoal lining in Ontario. In
situ bioassays and short-term biological and chemical changes. Water, Air, Sand oil Pollut.
Chen, C.W., S.A. Gherini, J.D. Dean, R.J.M. Hudson, and R. Goldstein. 1985. Development and
calibration of the Integrated Lake/Watershed Acidification Study Mode!. In: J.L. Schnoor, ed.
Modeling of Total Acid Precipitation Impacts, pp. 175-209. Ann Arbor Science.
Cook, R.B., C.A. Kelley, J.C. Kingston, and R. Kreis. In Press. Chemical limnology of soft water
lakes in the Upper Midwest. Biogeockemistry.
Cook, R.B., C.A. Kelly, D.W. Schindler, and M.A. Turner. 1986. Mechanisms of hydrogen ion
neutralization in an experimentally acidified lake. Limnol. Oceanogr. 31:134-148.
Cosby, B.J., G.M. Hornberger, J.N. Galloway, and R.F. Wright 1985. Time scales of catchment
acidification: a quantitative model for estimating freshwater acidification. Environ. Sci. Technol.
19:1144-1149.
Dillon, P.J., R.A. Reid, and R. Girard. 1986. Changes in the chemistry of lakes near Sudbury,
Ontario, following reduction of SO2 emission. Water, Air, and Soil Pollut. 31(l-2):59-65.
Driscoll, C.T. and R.M. Newton. 1985. Chemical characteristics of acid-sensitive lakes in the
Adirondack region of New York. Environ. Sci. Technol. 19:1018-1024.
Federal/Provincial Research and Monitoring Coordinating Committee. 1986. Assessment of
the State of Knowledge on the Long-Range Transport of Air Pollutants and Acid Deposition.
Environment Canada, Downsview, Ontario.
Forsberg, C., G. Morling, and R.G. Wetzel. 1985. Indications of the capacity for rapid reversibility
of lake acidification. Am&io. 14:164-166.
Gherini, S.A., L. Mok, R.J.M. Hudson, G.F. Davies, C.W. Chen, and R.A. Goldstein. 1985. The
ILWAS model: formulation and application. Water, Air, and Soil Pollut. 26:425-459.
Henriksen, A. 1982. Susceptibility of surface-waters to acidification. In: T.A. Haines and
R.E. Johnson, eds. Acid Rain/Fisheries, pp. 103-121. Proceedings for the International Symposium
on Acid Precipitation and Fishery Impacts in North America. Northeastern Division of the American
Fisheries Society.
Hutchinson, T.C. and M. Havas. 1986. Recovery of previously acidified lakes near Coniston,
Canada, following reductions in atmospheric sulphur and metal emissions. Water, Air, and Soil
Pollut. 28:319-333.
Jones, M.J., D.R. Marmorek, and G. Cunningham. 1984. Predicting the extent of damage to
fisheries in inland lakes of eastern Canada due to acidic precipitation. A report to the Steering
Committee of a project sponsored by the Department of Fisheries and Oceans.
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, and Soil Pollut. 29:285-296.
7-17
-------
DRAFT PRELIMINARY INTERPRETIVE REPORT
FOR INTERNAL USE ONLY
DO NOT CITE OR QUOTE
Keller, W., J.R. Pitblado, and N.I. Conroy. In Press. Water quality changes in Sudbury area lakes
related to reduced smelter emissions. Water, Air, and Soil Pollut.
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. In Press. Prediction of biological acid neutralization in acid-
sensitive lakes. Biogeochemistry.
Kelso, J.R.M., C.K. Minns, J.E. Gray, and M.L. Jones. 1986. Acidification of surface waters in
eastern Canada and its relationships to aquatic biota. Canadian Special Publication of Fisheries and
Aquatic Sciences 87. Department of Fisheries and Oceans, Ottawa, Canada.
Krug, B.C. and C.R. Frink. 1983. Acid rain on acid soil: A new perspective. Science 221:520-525.
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.
Marmorek, D. 1984. Change in the temporal behavior and size structure of plankton systems in
acid lakes. In: G. Hendrey, ed. Early Biotic Responses to Advancing Lake Acidification, pp. 23-42.
Boston, MA: Butterworth Publishers.
Minns, C.K. and J.R.M. Kelso. In Press. Estimates of existing and potential impact of acidification
on the freshwater resources of eastern Canada. Water, Air, and Soil Pollut.
Rudd, J.W.M., C.A. Kelly, V. St Louis, R.H. Hesslein, A. Furitani, and M.H. Holoka. 1986.
Microbial consumption of nitric and sulfuric acids in acidified north temperate lakes. Limnol.
Oceanogr. 31:1267-1280.
Thompson, M.E. 1986. The cation denudation rate model - its continued validity. Water, Air, and
Soil Pollut. 31:17-26.
Watt, W.D. In Press. The case for liming some Nova Scotian salmon rivers. Water, Air, and Soil
Pollut.
Watt, W.D., G.J. Farmer, and W.J. White. 1984. Studies on the use of limestone to restore Atlantic
salmon habitat in acidified rivers. Lake and Reservoir Management, Proceedings of the Third
Annual Conference of the North American Lake Management Society, pp. 374-379.
Watt, W.D., C.D. Scott, and W.J. White. 1983. Evidence of acidification of some Nova Scotian
rivers and its impact on Atlantic salmon, Salmo salav. Can. J. Fish. Aquat. Sci. 40:462-473.
White, W.J., W.D. Watt, and C.D. Scott 1984. An experiment on the feasibility of rehabilitating
acidified Atlantic salmon habitat in Nova Scotia by addition of lime. Fisheries 9:25-30.
Wright, R.F. 1983. Predicting acidification of North American lakes.
Norwegian Institute for Water Research, Oslo, Norway.
NIVA Report 0-81036.
,_Yan, N.D. 1985a. Biological effects of acidification. III. Long-term changes in the plankton of
Clearwater Lake near Sudbury, Ontario: Have the communities responded to reduced acid inputs?
Presented at the International Symposium of Acidic Precipitation, Muskoka, Ontario.
U.S. Environmental Protection Agsaoy
Library. Rocra 2404 PM-211-A
401 M Street, S.W.
Washington, DC 20460
7-18
-------
A-'
I
i
1
n
i
Bale. Due I
*
'•f&tt'3b
wov- n.
. . :
'..'-•-.: ' " '"
1 • ' ' * '
P
•
• .... •? - ' -*; -><
-iyyi . : -
• ^ * ."*
.
'•* ' * r'- .*•• "
•' " -
- . , .; -.
-• • .. ; •
, ••. • "'
- •
,>• • ' -
• • -;' - ;
«".;•' ~ - •/ *
"
^ •
; !
.
OftODART, INC
-------
-------