EP A/600/3-89/061 b
                                           July 1989
          Direct/Delayed  Response Project:
   Future Effects of Long-Term Sulfur Deposition
             on Surface Water Chemistry
in the Northeast and Southern Blue Ridge Province
      Volume II:  Level I and Level II Analyses
                            by

   M. R. Church, K. W. Thornton, P. W. Shaffer, D. L. Stevens, B. P. Rochelle,
      G. R. Holdren, M. G. Johnson, J. J. Lee, R. S. Turner, D. L. Cassell,
      D. A. Lammers, W. G. Campbell, C. I. Liff, C. C. Brandt, L H. Liegel,
       G. D. Bishop, D. C. Mortenson, S. M. Pierson, D. D. Schmoyer
                      A Contribution to the
            National Acid Precipitation Assessment Program
              U.S. Environmental Protection Agency
    Office of Research and Development, Washington, DC 20460
    Environmental Research Laboratory, Corvallis, Oregon 97333

-------
                                        NOTICE

The information in this document has been funded  wholly (or in part) by the U.S. Environmental
Protection Agency.  It has been subjected to the Agency's peer and administrative review, and it has
been approved for publication as an EPA document. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.

-------
                                     CONTENTS

SECTION                                                                       Page

Notice	,	  ii
Tables	,	,	, , , , .   xii
Figures	   xx
Plates		    xxix
Contributors	:	    xxxi
Acknowledgments  	,	   xxxiii

1   ' EXECUTIVE SUMMARY	.....:.;	  1
    ,1,1  INTRODUCTION . .	,	  1
          1.1,1  Project Background  .	  1
          1.1.2  Primary Objectives	  2
          1.1.3  Study Regions	  2
          1.1.4  Time Frames of Concern	  2
     1.2  PROCESSES OF ACIDIFICATION   	;	  4
          1,2.1  Sulfur Retention	  4
          1.2.2  Base Cation Supply	  4
     1.3  GENERAL APPROACH 	  5
          1.3.1  Soil Survey	  5
          1.3.2  Other Regional Datasets	  7
          1,3.3  Scenarios of Atmospheric Deposition	  7
          1.3.4  Data Analysis  	  7
     1.4  RESULTS	  8
          1.4.1  Retention of Atmospherically Deposited Sulfur	  8
              1.4.1.1 Current Retention 	  8
              1.4.1.2 Projected Retention	  8
          1.4.2  Base Cation Supply	10
              1.4.2.1 Current Control	10
              1.4.2.2 Future Effects	10
          1.4.3  Integrated Effects on Surface Water ANC  	12
              1.4.3.1 Northeast Lakes .	12
              1.4.3.2 Southern Blue Ridge Province	15
     1.5  SUMMARY DISCUSSION		.18
     1.6  REFERENCES	 18

2    INTRODUCTION TO THE DIRECT/DELAYED RESPONSE PROJECT	23
     2.1  PROJECT BACKGROUND	23
     2.2  PRIMARY OBJECTIVES	24
     2.3  STUDY REGIONS	24
     2.4  TIME FRAMES OF CONCERN	27
     2.5  PROJECT PARTICIPANTS	27
     2.6  REPORTING	 27

3    PROCESSES  OF ACIDIFICATION	29
     3.1  INTRODUCTION	29
     3.2  FOCUS OF THE DIRECT/DELAYED RESPONSE PROJECT  	•. ;	30
     3.3  SULFUR  RETENTION PROCESSES	30
          3.3,1  Introduction  	30
          3.3,2  Inputs  	31

-------
                                  CONTENTS (Continued)
                                                                                     Page

          3.3.3  Controls on Sulfate Mobility within Forest/Soil Systems	32
               3,3.3,1  Precipitation/Dissolution of Secondary Sulfate Minerals	32
               3.3.3,2  Sulfate Reduction in Soils and Sediments	32
               3.3.3.3  Plant Uptake	34
               3.3.3.4  Retention as Soil Organic Sulfur	 . 34
               3.3.3.5  Suifate Adsorption  by Soils	35
          3.3.4  Models of Sulfur Retention	37
          3.3.5  Summary	38
     3.4 BASE CATION SUPPLY PROCESSES	39
          3.4.1  Introduction  	39
          3.4.2  Factors Affecting Base Cation Availability		 42
               3.4.2.1  Mineral Weathering	42
               3.4.2.2  Cation Exchange Processes	45
          3.4.3  Modelling Cation Supply Processes	47
               3.4.3.1  Modelling Weathering	47
               3.4.3.2  Modelling Cation Exchange Processes	,	48

4    PROJECT APPROACH	,	49
     4.1 INTRODUCTION			49
     4.2 SOIL SURVEY	49
          4.2.1 Watershed Selection	,	49
          4.2.2 Watershed Mapping	49
          4.2.3 Sample Class Definition	51
          4.2.4 Soil Sampling	51
          4.2,5 Sample Analysis	51
          4.2.6 Database  Management  	51
     4,3 OTHER REGIONAL DATASETS	 51
          4,3.1 Atmospheric Deposition	,	52
          4.3.2 Runoff Depth	52
     4,4 DATA ANALYSIS  	52
          4.4.1 Level i Analyses	53
          4.4,2 Level II Analyses	.53
          4.4.3 Level III Analyses 	53
          4.4.4 Integration of Results	 54
          4.4.5 Use of a Geographic Information System	54

5    DATA SOURCES AND DESCRIPTIONS	 . 55
     5.1 INTRODUCTION	55
     5.2 STUDY SITE SELECTION		, ,  ,	55
          5.2.1  Site Selection Procedures	55
          5.2.2  Eastern Lake Survey Phase I Design	55
          5.2.3  Pilot Stream Survey Design	58
          5.2.4  DDRP Target Population	58
               5.2.4.1  Northeast Lake Selection  	58
               5.2,4.2  Southern Blue Ridge Province Stream Selection	 60
               5.2.4.3  Final DDRP Target Populations   	  82
     5.3 NSWS LAKE AND STREAM DATA  	  82
          5.3.1  Lakes in the Northeast Region	  82
               5.3.1,1  Lake Hydrologic Type		  82
               5,3.1.2  Fail  Index Sampling	 .  82
               5.3.1.3  Chemistry of DDRP Lakes	  89
                                            IV

-------
                             CONTENTS (Continued)


     5.3.2  Streams in the Southern Blue Ridge Province Region	  91
          5.3.2.1  Spring Baseflow Index Sampling 	  91
          5.3.2.2  Chemistry of DDRP Stream Reaches  	  93
5.4 MAPPING PROCEDURES AND DATABASES   	  93
     5.4.1 Northeast Mapping	  95
          5.4.1.1  Soils   	  95
          5.4.1.2  Depth to Bedrock 	  99
          5.4.1.3  Forest Cover Type   	101
          5.4.1.4  Bedrock Geology    	101
          5.4.1.5  Quality Assurance   	101
          5.4.1.6  Land Use/Wetlands   	105
          5.4.1.7  Geographic Information  Systems Data Entry  	118
     5.4.2 Southern Blue Ridge Province Mapping	132
          5.4.2.1  Soils   	134
          5.4.2.2  Depth to Bedrock 	137
          5.4.2.3  Forest Cover Type/Land Use   	137
          5.4.2.4  Bedrock Geology          	137
          5.4.2.5  Drainage  	139
          5.4.2.6  Quality Assurance    	139
          5.4.2.7  Land Use/Wetlands	142
          5.4.2.8  Geographic Information  Systems Data Entry  	143
5.5 SOIL SAMPLING PROCEDURES AND DATABASES	146
     5.5.1 Development/Description of Sampling Classes	147
          5.5.1.1  Rationale/Need for Sampling Classes   	147
          5.5.1.2  Approach Used for Sampling Class Development  	147
          5.5.1.3  Description  of Sampling Classes   	148
     5.5.2 Selection of Sampling Sites  	150
          5.5.2.1  Routine Samples  	150
          5.5.2.2  Samples on Special Interest Watersheds  	155
     5.5.3  Soil Sampling  	155
          5.5.3.1  Soil Sampling Procedures  	156
          5.5.3.2  Quality Assurance/Quality Control of Sampling   	156
     5.5.4  Physical and Chemical Analyses  	157
          5.5.4.1  Preparation Laboratories	157
          5.5.4.2  Analytical Laboratories	159
     5.5.5  Database Management	167
          5.5.5.1  Database Structure  	172
          5.5.5.2  Database Operations	174
     5.5.6  Data Summary	 178
          5.5.6.1  Summary of Sampling Class Data	178
          5.5.6.2  Cumulative  Distribution  Functions  	178
5.6 DEPOSITION DATA	178
     5.6.1  Time Horizons of Interest 	190
          5.6.1.1  Current Deposition	190
          5.6.1.2  Future Deposition 	190
     5.6.2  Temporal Resolution  	190
          5.6.2.1  Level I Analyses  	190
          5.6.2.2  Level II Analyses	190
          5.6.2.3  Level III Analyses	190

-------
                                   CONTENTS (Continued)
                                                                                      Page

          5.6.3 Data Acquisition/Generation	192
                5.6.3.1  Level III Analyses - Typical Year Deposition Dataset	192
                5.6.3.2  Level I and II Analyses - Long-Term Annual Average
                        Deposition Dataset	208
          5.6.4 Deposition Datasets Used in DDRP Analyses	224
     5.7 HYDROLOGIC DATA	224
          5.7.1  Runoff	224
                5.7.1.1  Data Sources	224
                5.7.1.2  Runoff Interpolation Methods	224
                5.7.1.3  Uncertainty Estimates	227
          5.7.2 Derived Hydrologic Parameters  	227
                5.7.2.1  TOPMODEL .  . .	228
                5.7.2.2  Soil Contact (Darcy's Law)    	231
                5.7.2.3  Mapped Hydrologic Indices      	234

6    REGIONAL POPULATION ESTIMATION  	242
     6.1 INTRODUCTION	242
     6.2 PROCEDURE	242
          6.2.1  Use of Variable Probability Samples	242
          6.2.2 Estimation Procedures for Population Means	243
          6.2.3 Estimators of Variance	244
          6.2.4 Estimator of Cumulative Distribution Function	245
     6.3 UNCERTAINTY ESTIMATES  	245
     6.4 APPLICABILITY  	246

7    WATERSHED SULFUR RETENTION	247
     7.1 INTRODUCTION	247
     7.2 RETENTION IN LAKES AND WETLANDS	248
     7.2.1  Introduction  	248
          7.2.2 Approach .,	249
          7.2.3 Results .  .	251
     7.3 WATERSHED SULFUR RETENTION  	253
          7.3.1  Methods	253
                7.3.1.1  Input/Output Calculation  	253
                7.3.1.2  Data Sources   	255
          7.3.2 Uncertainty Estimates	255
                7.3.2.1  Introduction 	255
                7.3.2.2  Individual Variable Uncertainties  	255
                7.3.2.3  Uncertainty Calculation  - Monte Carlo Analysis  	260
          7.3.3 Internal Sources of Sulfur	262
                7.3.3.1  Introduction/Approach   	262
                7.3.3.2  Bedrock Geology  	.662
                7.3.3.3  Upper Limit Steady-State Sulfate Concentration   	265
          7.3.4 Results and Discussion  	268
                7.3.4.1  Northeast   	271
                7.3.4.2  Mid-Appalachians  	279
                7.3.4.3  Southern Blue  Ridge Province	280
                7.3.4.4  Conclusions	280
                                             VI

-------
                                  CONTENTS (Continued)
                                                                                     Page

8    LEVEL I STATISTICAL ANALYSES	285
     8.1  INTRODUCTION	285
          8.1.1  Approach	285
          8.1.2  Statistical Methods	286
     8.2   RELATIONSHIPS  BETWEEN  ATMOSPHERIC DEPOSITION  AND SURFACE
          WATER CHEMISTRY   	291
          8.2.1  Introduction  	291
          8.2.2  Approach	291
          8.2.3  Results and Discussion   	292
               8.2.3.1  Northeast	292
               8.2.3.2  Southern Blue Ridge Province	292
               8.2.3.3  Summary	292
     8.3  DERIVED HYDROLOGIC PARAMETERS	295
          8.3.1  Soil Contact (Darcy's Law)  	295
               8.3.1.1  Introduction  	295
               8.3.1.2  Results and Discussion	299
          8.3.2  Geomorphic/Hydrologic Parameters  	302
               8.3.2.1  Introduction  	302
               8.3.2.2  Results and Discussion  	310
          8.3.3  TOPMODEL Parameters  	316
               8.3.3.1  Introduction  	317
               8.3.3.2  Results and Discussion  	317
               8.3.3.3  Summary	326
     8.4  MAPPED BEDROCK GEOLOGY	326
          8.4.1  DDRP Bedrock Sensitivity Scale	327
          8.4.2  Results	328
               8.4.2.1  Sulfate  and Percent Retention	332
               8.4.2.2  Sum of Base Cations, ANC, and pH   	335
          8.4.3  Summary   	336
     8.5  MAPPED LAND USE/VEGETATION	337
          8.5.1  Introduction  	337
          8.5.2  Data Sources	337
          8.5.3  Statistical Methods	338
          8.5.4  Sulfate and Percent Sulfur Retention  	338
               8.5.4.1  Northeast  	338
               8.5.4.2  Southern Blue Ridge Province	347
               8.5.4.3  Regional Comparisons	347
          8.5.5  ANC. Ca plus Ma. and pH  	347
               8.5.5.1  Northeast	347
               8.5.5.2  Southern Blue Ridge Province	349
               8.5.5.3  Regional Comparisons	349
          8.5.6  Summary and Conclusions	351
     8.6  MAPPED SOILS	351
          8.6.1  Introduction  	351
          8.6.2  Approach	352
          8.6.3  Sulfate and Sulfur Retention	 354
               8.6.3.1  Northeast	360
               8.6.3.2  Southern'Blue Ridge Province	362
               8.6.3.3  Regional Comparisons	365
                                            vii

-------
                             CONTENTS (Continued)


     8.6.4  ANC. Ca plus Mg. and pH  	367
          8.6.4.1  Northeast	367
          8.6.4.2  Southern Blue Ridge Province	369
          8.6.4.3  Regional Comparisons	377
     8.6.5  Summary and Conclusions	378
8.7 ANALYSES OF DEPTH  TO BEDROCK	379
     8.7.1  Introduction  	379
     8.7.2  Approach	379
     8.7.3  Sulfate and Percent Sulfur Retention   	381
          8.7.3.1  Northeast	381
          8.7.3.2  Southern Blue Ridge Province	381
          8.7.3.3  Comparison of Regions  	381
     8.7.4  ANC. Ca plus Mq and pH	385
          8.7.4.1  Southern Blue Ridge Province	 385
          8.7.4.2  Comparison of Regions	386
     8.7.5  Summary and Conclusions	386
8.8 INTEGRATED ANALYSIS OF ALL MAPPED VARIABLES	388
     8.8.1  Introduction	388
     8.8.2  Approach	388
     8.8.3  Sulfate and Sulfur Retention  	388
          8.8.3.1  Northeast	388
          8.8.3.2  Southern Blue Ridge Province	390
          8.8.3.3  Regional Comparisons	392
     8.8.4  ANC. Ca plus Mq. and pH  .	393
          8.8.4.1  Northeast	393
          8.8.4.2  Southern Blue Ridge Province	395
          8.8.4.3  Regional Comparisons	398
     8.8.5  Summary and Conclusions	398
8.9 SOIL PHYSICAL AND CHEMICAL CHARACTERISTICS	,	399
     8.9.1  Introduction  	399
     8.9.2  Approach	399
          8.9.2.1  Statistical Methods  	400
     8.9.3  Aggregation of Soil Data	402
          8.9.3.1  Introduction  	402
          8.9.3.2  Aggregation of Soil Data  	403
          8.9.3.3  Assessment of the DDRP Aggregation Approach	404
          8.9.3.4  Estimation of Watershed Effect  	406
          8.9.3.5  Evaluation of Watershed Effect  	407
     8.9.4  Regional Soil Characterization	407
     8.9.5  Sulfate and Sulfur Retention	413
          8.9.5.1  Northeast	418
          8.9.5.2  Southern Blue Ridge Province	421
     8.9.6  Ca plus Mq  (SOBC). ANC. and pH  	421
          8.9.6.1  Northeast	421
          8.9.6.2  Southern Blue Ridge Province	425
     8.9.7  Evaluation of Alternative Aggregation Schemes  	426
     8.9.8  Summary and Conclusions	426
          8.9.8.1  Alternative Aggregation Schemes	426
          8.9.8.2  Sulfate and Sulfur Retention	429
          8.9.8.3  Ca plus Mg (SOBC), ANC, and pH  	429
     8.9.9  Summary Conclusions  	430
                                       viii

-------
                                 CONTENTS (Continued)
                                                                                   Page

     8,10  EVALUATION  OF  ASSOCIATIONS BETWEEN  WATERSHED  ATTRIBUTES  AND
          SURFACE WATER CHEMISTRY  	430
          8.10.1  Introduction	430
          8.10.2  Approach	431
          8.10.3  Regional Characterization of Watershed Attributes	431
               8.10.3.1  Northeast Subregions	431
               8.10.3.2  Northeast and Southern Blue Ridge Providence	435
          8.10.4  Sulfate and Sulfur Retention	436
               8.10.4.1  Northeast   	436
               8.10.4.2  Southern Blue Ridge Province  	436
          8.10.5  Ca plus Mq (SOBQ. ANC.  and pH	437
               8.10.5.1  Northeast   	437
               8.10.5.2  Southern Blue Ridge Province  	437
          8.10.6  Summary and Conclusions  	450
               8.10.6.1  Sulfate and Sulfur Retention  	450
               8.10.6.2  Ca  plus  Mg (SOBC), ANC, and pH  	450
          8.10.7  Summary Conclusions		450

9    LEVEL II ANALYSES - SINGLE FACTOR  RESPONSE TIME ESTIMATES  	452
     9.1  INTRODUCTION	,	.452
     9.2  EFFECTS OF SULFATE ADSORPTION ON WATERSHED SULFUR RESPONSE TIME . 453
          9.2.1  Introduction	453
          9.2.2  Section Objectives	454
          9.2.3  Approach	455
               9.2.3.1  Model Description   	455
               9.2.3.2  Data Sources	'.	456
               9.2.3.3  Model Assumptions and Limitations	456
               9.2.3.4  Adsorption Data	458
               9.2.3.5  Evaluation of Aggregated Data and Model Outputs  	461
               9.2.3.6  Target Populations for Model Projections	 462
          9.2.4  Results	464
               9.2.4.1  Comparison of Northeast and Southern Blue Ridge Province Isotherm
                       Variables  	,	464
               9.2.4.2  Model Results - Northeastern United States   	466
               9.2.4.3  Model Results - Southern Blue Ridge Province  	479
               9.2.4.4  Uncertainty Analyses and Alternative Aggregation Approaches	493
               9.2.4.5  Summary of Results  from the Southern Blue Ridge Province   	501
          9.2.5  Summary Comments on  Level II Sulfate Analyses	502
          9.2.6  Conclusions	504
     9.3  EFFECT OF CATION EXCHANGE AND  WEATHERING ON SYSTEM RESPONSE .... 506
          9.3.1  Introduction  	506
               9.3.1.1  Level II Hypotheses	506
               9.3.1.2  Approach		509
          9.3.2  Descriptions of Models	512
               9.3.2.1  Reuss Model	512
               9.3.2.2  Bloom-Grigal Model   	527
          9.3.3  Model  Forecasts   	533
               9.3.3.1  Reuss  Model	535
               9.3.3.2  Bfoom-Grigal Model	577
          9.3.4  Comparison of the Bloom-Grigal and Reuss Model Projections	605
          9.3.5  Summary and Conclusions	612
                                           IX

-------
                                 CONTENTS (Continued)


10   LEVEL III ANALYSES - DYNAMIC WATERSHED MODELLING	618
     10.1  INTRODUCTION	,	618
     10.2  DYNAMIC WATERSHED MODELS	,	,	620
          10.2.1 Enhanced Trickle Down (ETD) Model	622
          10.2.2 Integrated Lake-Watershed Acidification Study (ILWAS) Model  	627
          10.2.3 Model of Acidification of Groundwater in Catchments (MAGIC)	628
     10.3  OPERATIONAL ASSUMPTIONS  	629
     10.4  WATERSHED PRIORITIZATION	629
          10.4.1 Northeast		629
          10.4.2 Southern Blue Ridge Province	  632
          10.4.3 Effects of Prioritization on Inclusion Probabilities  	632
     10.5  MODELLING DATASETS		634
         . 10.5.1 Meteorological/Deposition Data	 ,	  634
          10.5.2 DDRP Runoff Estimation	634
               10.5.2.1  Annual Runoff	634
               10.5.2.2  Monthly Runoff  	635
          10.5.3 Morphometry  	636
          10.5.4 Soils	636
          10.5.5 Surface Water Chemistry  	637
          10.5.6 Other Data	637
          10.5.7 Chloride Imbalance	  637
     10.6  GENERAL APPROACH		639
     10.7  MODEL CALIBRATION	  642
          10.7.1 Special  Interest Watersheds' -	642
               10.7.1.1  Northeast	  643
               10.7.1.2  Southern Blue Ridge Province	643
          10.7.2 General Calibration Approach	644
          10.7.3 Calibration of the Enhanced Trickle Down Model	644
          10.7.4 Calibration of the Integrated Lake-Watershed  Acidification Model  	647
          10.7,5 Calibration of the Model of Acidification of Groundwater in Catchments	650
          10.7.6 Calibration/Confirmation Results	652
     10.8  MODEL SENSITIVITY ANALYSES	  656
          10.8.1 General Approach	 .  657
          10.8.2 Sensitivity Results	667
     10.9  REGIONAL PROJECTIONS REFINEMENT	"	 .  658
          10.9.1 Enhanced Trickle Down	658
          10.9.2 Integrated Lake-Watershed Acidification Study	659
          10.9.3 Model of Acidification of Groundwater in Catchments	659
          10.9.4 DDRP Watershed Calibrations	"	661
               10.9.4.1  Integrated Lake-Watershed Acidification Study	661
               10.9.4.2  MAGIC	664
               10.9.4.3  Southern Blue Ridge Province	664
     10.10  MODEL PROJECTIONS	.668
          10.10.1 General Approach	 .  668
          10.10.2 Forecast Uncertainty	  672
               10.10.2.1 Watershed Selection .	 .  672
               10.10.2.2 Uncertainty Estimation Approaches	  673
               10.10.2.3 Relationship Among Approaches  	674
               10.10.2.4 Confidence Intervals	678

-------
                                 CONTENTS (continued)


     10.11  POPULATION ESTIMATION AND REGIONAL FORECASTS  	678
          10.11.1  Northeast Regional Projections  	678
               10.11.1.1  Target Population Projections Using MAGIC	,	 678
               10.11.1.2  Target Population Projections Using MAGIC and ETD	687
               10.11.1.3  Restricted Target Population Projections Using All Three Models .  . . 796
          10.11,2 Southern Blue Ridge Province	,	723
               10.11,2.1  Target Population Projections Using MAGIC	723
               10.11.2.2  Restricted Target Population Projections Using ILWAS and
                        MAGIC	749
          10.11.3 Regional Comparisons	765
               10.11.3.1  Northeastern Projections of Sulfate Steady State	 765
               10.11.3.2  Southern Blue Ridge Province Projections of Sulfate
                        Steady State  	771
               10.11.3.3  ANC and Base Cation Dynamics	 771
     10.12  DISCUSSION	 790
          10.12.1  Future Projections of Changes in Acid-Base Surface Water Chemistry  .... 790
          10.12.2 Rate of Future Change	 790
               10.12.2.1  Northeast	790
               10.12.2.2  Southern Blue Ridge Province	792
          10.12.3 Uncertainties and Implications for Future Changes in Surface Water
                 Acid-Base Chemistry	795
               10.12.3.1  Deposition Inputs		795
               10.12.3.2  Watershed Processes	797
     10.13  CONCLUSIONS FROM LEVEL III ANALYSES	799

11   SUMMARY OF RESULTS	801
     11.1  RETENTION OF ATMOSPHERICALLY DEPOSITED SULFUR	801
          11.1.1  Current Retention	801
          11.1.2  Projected Retention  	801
     11.2   BASE CATION SUPPLY 	805
          11.2.1  Current Control	  . . 805
          11.2.2  Future Effects		80S
     11.3 INTEGRATED EFFECTS  ON SURFACE WATER ANC  	806
          11.3.1  Northeast Lakes	807
          11,3.2  Southern Blue Ridge Province	 . 814
     11.4  SUMMARY DISCUSSION	,	820

12   REFERENCES	823

13   GLOSSARY	856
     13.1  ABBREVIATIONS AND SYMBOLS 	856
          13.1.1  Abbreviations	856
          13.1.2  Symbols	 858
     13.2  DEFINITIONS	.862

APPENDICES	,	888
                                           XI

-------
                                         TABLES
1-1.    Lakes in the NE Projected to Have ANC Values <0 and <50 fjeq L~1
       for Constant and Decreased Sulfur Deposition	  14
1-2.    SBRP Stream Reaches Projected to Have ANC Values <0 and <50 peq L"1
       for Constant and Increased Sulfur Deposition   	  17

3-1.    Major Rock Forming Minerals and Their Relative Reactivities 	  44

5-1.    Sampling Structure for Phase I, Region 1 (Northeast), Eastern Lake Survey   	  57
5-2.    Sample Structure for the Direct/Delayed Response Project -Northeastern Sample  .....  61
5-3.    ANC Group, Lake Identification, ELS-I Phase I ANC, Weight and Inclusion
       Probabilities for the Direct/Delayed Response Project Northeast Sample Watersheds ...  62
5-4.    Lake Identification and Name, and State and Latitudinal/Longitudinal Location
       of the Northeast Sample Watersheds	.,..,,.  66
5-5.    Lake Identification and Name, Sorted by State - Northeast Sample Watersheds	  69
5-6.    Stream Identification, Weight, and Inclusion Probabilities for the Southern
       Blue Ridge Province Direct/Delayed Response Project Sample Watersheds 	  78
5-7.    Stream Identification and Name, and State and Latitudinal/Longitudinal Location
       of the Southern Blue Ridge Province Sample Watersheds	  79
5-8.    Stream Identification and Name, Sorted by State -- Southern Blue Ridge Province
       Sample Watersheds	  80
5-9.    DDRP Reclassiflcation  of Northeastern Lakes Classified as "Seepage" or "Closed"
       by the NSWS		,	,	  83
5-10.   Depth-to-Bedrock Classes and Corresponding Level of Confidence	  100
5-11.   Interpretation Codes for Northeast Map Overlays - Land Use/Land Cover,
       Wetlands, and Beaver Activity	  106
5-12.   Northeast Watersheds Studied for Independent Field Check of Land Use and
       Wetland Photointerpretations	  109
5-13.   Chi-Square Test for General Land Use Categories	  110
5-14.   Comparison of Field Check (Matched) General Land Use Determinations with
       Office Photointerpretations	  111
5-15.   Chi-Square Test for Detailed Wetland Categories	  113
5-16.   Comparison of Field Check (Matched) Detailed Wetland Determinations with
       Office Photointerpretations	  114
5-17.   Comparison of Beaver Dam Number, Breached and Unbreached Status,
       and Lodges, Identified via Field Check and Office Photointerpretation Methods ......  115
5-18.   Aggregated Land Use Data for Northeast Watersheds	  117
5-19.   Watershed No. 1E1062 Soil Mapping Units	  130
5-20.   Land Use Codes Used as Map Symbols   	  138
5-21.   Percent Land Use Data for Southern Blue Ridge Province  Watersheds  	  144
5-22,   Laboratory Analysis of DDRP Soil Samples	  158
5-23.   Analytical Variables Measured in the DDRP  Soil Survey	  160
5-24.   Data Quality Objectives for Detectability and Analytical Within-Batch Precision  	  163
5-25.   Detection Limits for Contract Requirements, Instrument Readings,
       and System-Wide Measurement in the Northeast  	  165
5-26.   Detection Limits for the Contract Requirements, Instrument Readings,
       and System-wide Measurement in the Southern Blue Ridge Province 	  166
5-27.   Attainment of Data Quality Objectives by the analytical laboratories as
       determined from blind audit samples for the Northeast	  168
5-28.   Attainment of Data Quality Objectives by the Analytical Laboratories as Determined
       from Blind Audit Samples for the Southern Blue Ridge Province	  170


                                            xii

-------
                                    TABLES (Continued)


5-29.   Quality Assurance and Quality Control Checks Applied to Each Data Batch   	  176
5-30.   Medians of Pedon-Aggregated Values of Soil Variables for the DDRP
       Regions and Subregions	  189
5-31.   Monthly Values of Leaf Area Index Used to Apportion Annual Dry Deposition to
       Monthly Values	  202
5-32.   Ratios of Coarse-to-Fine Particle Dry Deposition	  205
5-33.   Ratios of Dry Deposition to Wet Deposition for DDRP Study Sites for the
       Typical Year Deposition Dataset	  207
5-34.   Deposition Datasets Used in DDRP Analyses	  225
5-35.   DDRP texture classes and saturated hydraulic conductivity (K) for the NE
       study systems	  229
5-36.   SCS slope classifications	  233
5-37.   Mapped and calculated geomorphic parameters collected for the NE study sites. ....  236
5-38.   Mapped and calculated geomorphic parameters collected for the SBRP study sites.  . .  240

7-1.    Summary of Computed Sulfur Retention by In-lake Reduction for Lake
       Systems in the Eastern United States	  250
7-2.    Intensively Studied Sites  Used in Surface Water Chemistry Uncertainty Analysis	  257
7-3.    Summary Statistics on Percent Differences Between  Flow-weighted Average
       Annual Sulfate Concentration and the Fall/Spring Flow-weighted Averages	  261
7-4.    Bedrock Geology  Maps Used in the DDRP Internal Sources of Sulfur
       Bedrock Geology  Analyses	  263
7-5.    Potential for Sulfur Contribution by Geologic Type 	  264
7-6.    Summary of Watersheds (by  ELS and NSS Subregion) Dropped Due
       to Suspected Internal Sources of Sulfur Identified by Steady-State Analysis	  270
7-7.    Percent Sulfur Retention  - Summary Statistics by Region	  273
7-8.    Summary of Sulfur Retention Status and of Watershed Variables
       Contributing to Sulfur Retention for 42 Watersheds in the Northeastern  United States .  278

8-1.    Surface Water Chemistry and Percent Sulfur Retention Summary Statistics
       for the Northeastern DDRP Sample of 145 Lake Watersheds	  287
8-2.    Surface Water Chemistry and Percent Sulfur Retention Summary Statistics
       for the DDRP Sample of 35 SBRP  Stream Watersheds 	  288
8-3.    Summary Statistics for Wet and Dry Deposition on the DDRP Sample
       of 145 Northeastern Lake Watersheds  	  289
8-4.    Summary Statistics for Wet and Dry Deposition on the DDRP Sample of 35
       SBRP Stream Watersheds  	  290
8-5.    Results of Regressions Relating Surface Water Chemistry to Atmospheric Deposition
       in the Northeast Region	  293
8-6.    Results of Regressions Relating Surface Water Chemistry to Atmospheric Deposition
       in the Southern Blue Ridge Province  	  294
8-7.    Estimated Population-Weighted Summary Statistics on the Darcy's Law Estimates
       of Flow Rate and  the Index of Flow Relative to Runoff 	  296
8-8.    Estimated Population-Weighted Summary Statistics for Northeastern Geomorphic/
       Hydrologic Parameters   	  303
8-9.    Estimated Population-Weighted Summary Statistics for Southern Blue Ridge
       Province  Hydrologic/Geomorphic Parameters   	  304
8-10.   Mapped and Calculated Geomorphic Parameters Collected for
       the Northeastern Study Sites   	  305
8-11.   Mapped and Calculated Geomorphic Parameters Collected for the SBRP Study  Sites .  308
8-12.   Stratification Based on Sulfur Deposition   	  311
                                            XIII

-------
                                    TABLES (Continued)


8-13.   Results of Stepwise Regression Relating Surface Water Chemistry
       versus Geomorphic/Hydrologic Parameters for the Entire NE	  312
8-14,   Stepwise Regression Equations for Surface Water Chemistry and Hydrologic/
       Geomorphic Parameters Based on Sulfur Deposition Stratification	  313
8-15,   Results   of  Stepwise  Regression   Relating  Surface  Water   Chemistry   and
       Geomorphic/Hydrologic Parameters for the SBRP	,  314
8-16,   Population-Weighted Summary Statistics for ln(a/KbTanB) for the Northeast	  318
8-17.   Population-Weighted Summary Statistics for ln(a/TanB) for the Southern Blue
       Ridge Province 	  319
8-18.   Spearman's Correlation Coefficients  Between  ln(a/KbTanB)  and  Surface  Water
       Chemistry	  320
8-19.   Pearson's Correlation Coefficients Between ln(a/TanB) and NSS Pilot Chemistry  .....  325
8-20.   Tabulation of the Generic Bedrock Types Used to Classify the Mapped Units
       Identified on State  Map Legends	  329
8-21.   Tabulation of the Generic Bedrock Types Used to Classify the Mapped Units
       Identified on State  Map Legends	  330
8-22.   Regional and Subregional Statistics for the Bedrock Sensitivity Code Variables   	  331
8-23.   Results of Regressions of Surface Water Chemistry on Bedrock Sensitivity
       Code Statistics and Deposition Estimates for Northeast	  333
8-24.   Results for SBRP of Regressions of Surface Water Chemistry on Bedrock
       Sensitivity Code Statistics and Deposition Estimates	  334
8-25.   Land Use and Other Environmental Variables Related to Surface Water
       Chemistry of Northeastern Lakes	  339
8-26.   Factor Loadings for First 13 Principal Components after Varimax Rotation of
       the Correlation Matrix of Land Use and other Environmental Variables for
       Northeastern Lakes 	  340
8-27.   Interpretation of the First 13 Principal Components After Varimax Rotation of the
       Correlation Matrix of Land Use and Other Environmental Variables for Northeastern
       Lakes	  342
8-28.   Land Use and Other Environmental Variables Related to Surface Water Chemistry of
       Southern Blue Ridge Province Streams	  343
8-29.   Composition of First 11 Principal Component Analysis (PCA) Factors Land
       Use and Other Environmental Variables Related to Surface Water Chemistry
       of Southern Blue Ridge Province Streams	  344
8-30.   Interpretation  of the First  11  Principal  Components  of  Land   Use and  Other
       Environmental  Variables for Southern Blue Ridge Province Streams  	  345
8-31.   Results of Regressions Relating Surface Water Chemistry of Northeastern Lakes
       to Land Use and Other Environmental Data	  346
8-32.   Results of Regressions Relating Sulfate and  Percent Sulfur Retention of
       Southern Blue Ridge Province Streams to Land Use Data  	  348
8-33.   Results of Regressions Relating ANC, Ca plus Mg,  and pH of Southern Blue
       Ridge Province Streams to Land Use Data  	  350
8-34.   Summary Statistics for Percent Area Distribution of the 38 Soil Sampling
       Classes and the 4  Miscellaneous Land Areas on the DDRP Sample of 145
       NE Lake Watersheds	  355
8-35.   Summary Statistics for the  Percent Area Distribution of the 38 Soil Sampling
       Classes and the 4  Miscellaneous Land Areas in the GIS 40-ft Contour on the
       DDRP Sample of 145 NE Lake Watersheds	  356
8-36.   Summary Statistics for the  Percent Area Distribution of the 38 Soil Sampling
       Classes and the 4  Miscellaneous Land Areas in the Combined GIS  Buffers on the
       DDRP Sample of 145 NE Lake Watersheds	  357


                                            xiv

-------
                                   TABLES (Continued)
                                                                                    Paoe

8-37.   Summary Statistics for the Percent Area Distribution of the 12 Soil Sampling
       Classes and the 2. Miscellaneous Land Areas on the DDRP Sample of 35 SBRP
       Stream Watersheds .,.,,..,	358
8-38.   Summary Statistics for the Percent Area Distribution of the 12 Soil Sampling
       Classes and the 2 Miscellaneous Land Areas in the 100-Meter Linear GIS Buffer
       on the DDRP Sample of 35 SBRP Stream Watersheds  	359
8-39.   Lake Sulfate and Percent S Retention Regression Models Developed for NE Lakes
       Using Deposition, Mapped Soils and Miscellaneous Land Areas as Candidate
       Independent Variables	361
8-40.   Regression Models of Sulfate in SBRP Streams, Developed Using Deposition,
       Mapped Soils and Miscellaneous Land Areas as Candidate
       Independent Variables	  363
8-41.   Regression Models of Percent Sulfur Retention In SBRP Stream Watersheds
       Developed Using Deposition,  Mapped Soils,  and Miscellaneous  Land Areas as
       Candidate Independent Variables	  366
•8-42.   Lake ANC and the Sum of Lake Calcium and Magnesium Regression Models
       Developed for NE Lakes Using Deposition, Mapped Soils, and  Miscellaneous Land
       Areas as Candidate Independent Variables	  368
8-43,   Lake pH Regression Models Developed for NE Lakes Using  Deposition,
       Mapped Soils, and Miscellaneous  Land Areas as Candidate
       Independent Variables	  370
8-44.   Regression Models of ANC in SBRP Stream  Watersheds, Developed Using
       Deposition, Mapped Soils, and Miscellaneous Land Areas as Candidate
       Independent Variables	372
8-45.   Regression Models of Calcium Plus Magnesium in SBRP Streams, Developed
       Using Deposition, Mapped Soils, and Miscellaneous Land Areas as a Candidate
       Independent Variables	  373
8-46.   Regression Models of SOBC  in SBRP Streams,  Developed Using Deposition,
       Mapped Soils, and Miscellaneous  Land Areas as Candidate
       Independent Variables .	375
8-47.   Regression Models of Stream pH  in SBRP Streams, Developed  Using Deposition,
       Mapped Soils, and Miscellaneous  Land Areas as Candidate
       Independent Variables	  376
8-48.   Depth-to-Bedrock Classes for the Northeast  and the Southern Blue Ridge Province  . .  380
8-49.   Regional and Subregional Statistics for Percentage of Watershed Coverage of the
       Depth-to-Bedrock Classes	  382
8-50.   Results for NE of Regressions of Surface  Water Chemistry on Depth-to-Bedrock
       Classes and Deposition Estimates	  384
8-51.   Results for SBRP of Regressions of Surface  Water Chemistry on Depth-to-Bedrock
       Classes and Deposition Estimates	  387
8-52.   Regression Models of Surface Water Sulfate and Sulfur Retention in the
       NE Lake Watersheds   	  389
8-53.   Regression Models of Surface Water Sulfate and Sulfur Retention in the SBRP
       Stream Watersheds	  392
8-54.   Regression Models of Surface Water ANC, Ca plus Mg, and pH in the NE Lake
       Watersheds	  394
8-55.   Regression Models of Surface Water ANC, Ca plus Mg, and pH in the SBRP
       Stream Watersheds   	  397
8-56.   Standard Deviations Within and Among Northeast Sampling  Classes Estimated
       from B Master Horizon Data	  405
8-57.   Means and Standard  Deviations of Soil Characteristics by Aggregation
       Method and Region	.. .  408
                                           xv

-------
                                    TABLES (Continued)


8-58.   Population Means and Standard Errors for Selected Variables, by Subregion/Region
       and Aggregation (Watershed Adjusted Data)	  411
8-59.   Non-parametric Correlations Between Lake Chemistry Variables and Selected Soil
       Properties for the NE DDRP Watersheds	  414
8-60.   Non-parametric Correlations Between Stream Chemistry Variables and Selected
       Soil Properties for the SBRP  DDRP Watersheds . . .-	  416
8-61.   Results of Stepwise Multiple Regressions for DDRP Lake and Stream Sulfate
       Concentrations Versus Soil Physical and Chemical Properties  	  419
8-62.   Results of Stepwise Multiple Regressions for DDRP Watershed Sulfur Retention
       Versus Soil Physical and Chemical Properties  	  420
8-63.   Results of Stepwise Multiple Regressions  for DDRP  Lake Calcium plus Magnesium
       Concentrations and Stream Sum of Base Cation Concentrations Versus Soil Physical
       and Chemical  Properties	  422
8-64.   Results of Stepwise Multiple Regressions for DDRP Lake and Stream ANC
       Versus Soil Physical and Chemical Properties  	  423
8-65.   Results of Stepwise Multiple Regressions for DDRP Lake and Stream pH
       Versus Soil Physical and Chemical Properties  	  424
8-66.   Results of Stepwise Multiple Regressions for DDRP Lake and Stream ANC
       Versus Unadjusted and Watershed Adjusted Soil Physical and Chemical Properties .  .  427
8-67.   Results of Stepwise Multiple Regressions for DDRP Lake and Stream Sulfate
       Versus Unadjusted and Watershed Adjusted Soil Physical and Chemical Properties .  .  428
8-68.   Population Means and Standard Errors for Selected Variables, by Subregion/
       Region and Aggregation	  432
8-69.   Non-parametric Correlations Between Lake Chemistry Variables and Selected
       Watershed Attributes for the NE DDRP Watersheds  	  438
8-70.   Non-parametric Correlations Between Stream Chemistry Variables and Selected
       Watershed Attributes for the SBRP DDRP Watersheds  	  442
8-71.   Results of Stepwise Multiple Regressions for DDRP Lake and Stream Sulfate
       Concentration  Versus Watershed Attributes	  445
8-72.   Results of Stepwise Multiple Regressions for DDRP Watershed Sulfur Retention
       Versus Watershed Attributes	  446
8-73.   Results of Stepwise Multiple Regressions for DDRP Lake Calcium Plus Magnesium
       Concentrations and Stream Sum of Base Cations Versus Watershed Attributes  	  447
8-74.   Results of Stepwise Multiple Regressions  for DDRP  Lake and Stream ANC  Versus
       Watershed Attributes 	  448
8-75.   Results of Stepwise Multiple Regressions for DDRP Lake and Stream Air Equilibrated
       pH Versus Watershed Attributes  	  449

9-1.    Comparison of Summary Data for Sulfate  Adsorption Isotherm Data for Soils in the
       Northeastern United States and Southern Blue Ridge Province	  465
9-2.    Summary Statistics for Modelled Changes in  Sulfate Concentration,  Percent Sulfur
       Retention, and Delta Sulfate  for Northeast Watersheds Using  Long-Term Average
       Deposition Data	  470
9-3.    Summary Statistics for Modelled Changes  in Sulfate  Concentration, Percent Sulfur
       Retention, and Delta Sulfate for Northeast Watersheds Using Typical Year
       Deposition Data	„	  471
9-4.    Comparison of Measured and Modelled Base Year (1985) Sulfate Data for SBRP
       Watersheds, Using Long-Term Average Deposition Data	  482
9-5.    Comparison of Modelled Rates of Increase for  [SO42'] in DDRP Watersheds in the
       SBRP with Measured Rates of Increase in Watersheds in the  Blue Ridge and
       Adjoining Appalachians	  484
                                            xvi

-------
                                    TABLES (Continued)


9-6.    Summary Statistics for Modelled Changes in Sulfate Concentration, Percent Sulfur
       Retention, and Delta Sulfate for Watersheds in the Southern Blue Ridge Province,  Using
       Long-Term Average Deposition Data	  488
9-7.    Summary Statistics for Modelled Changes in Sulfate Concentration, Percent Sulfur
       Retention, and Delta Sulfate for Watersheds in the Southern Blue Ridge Province,
       Using Typical Year Deposition Data	  489
9-8.    Summary Comparison of Watershed Sulfur Status and Model Forecasts in the
       Northeastern United States and Southern Blue Ridge  Province	  503
9-9.    List of the Chemical Species and Reactions Considered Within the Reuss
       Model Framework	  515
9-10.   Effect of pCO2 on Changes Projected to Occur in Surface Water ANC Values at
       50 and  100 Years Using the Reuss Model.  Deposition Used in the Model is LTA  .  . .  524
9-11.   List of Input Data for the Bloom-Grigal Soil Acidification Model	  534
9-12.   Summary Statistics for the Population Estimates of Current ANC Conditions for
       Lakes in the NE Region for Five Different Deposition or Soils Aggregation
       Schemes   .  .	  541
9-13.   Descriptive Statistics of the  Population Estimates for Changes
       in Lake  Water ANC for Systems in the NE  	  546
9-14.   Summary Statistics Comparing the Projections Regarding Changes in Surface
       Water ANC Values Obtained Using  Different Soils Aggregation Schemes   	  549
9-15.   Summary Statistics of the Differences Between the Population Estimates for
       Future ANC Projections Made Using the Constant Level and Ramped
       Deposition Scenarios	  550
9-16.   Summary Statistics for the Population Estimates of Current ANC Conditions for Stream
       Reaches in the SBRP for Four Different Deposition Scenarios  	  552
9-17.   Descriptive Statistics of the  Population Estimates for Changes in Stream Reach
       ANC Values for Systems in  the SBRP	  555
9-18.   Summary Statistics of the Differences Between the Population Estimates for Future
       ANC Projections Made  Using the Constant Level and Ramped Deposition  Scenarios for
       Stream  Reaches in the SBRP	  559
9-19.   Summary Statistics of the Projected Changes in Soil Base Saturations in the NE
       Region, Obtained Using the  Different Deposition Scenarios  or Soil
       Aggregation Schemes	  562
9-20.   Summary Statistics of the Projected Changes in Soil pH in the NE Region, Obtained
       Using the Different Deposition Scenarios  or Soil Aggregation Schemes	  563
9-21.   Summary  Statistics of the Projected Changes in Soil Base Saturations in  the SBRP,
       Obtained Using the Different Deposition Scenarios	  571
9-22.   Summary Statistics of the Projected Changes in Soil pH  in the SBRP, Obtained
       Using the Different Deposition Scenarios	  572
9-23.   Comparison of the Changes in Soil Base  Saturation and Soil pH that Are Projected to
       Occur in the NE and SBRP	  576
9-24.   Regionally Weighted Median Values of Initial Annual Deposition Inputs to the Bloom-
       Grigal Model for the Northeastern Region and the Southern Blue Ridge Province ....  579
9-25.   Regionally Weighted Median Values of Annual Initial Soil Chemical Values Input
       Into the Bloom-Grigal Model for the Northeastern Region and the Southern Blue
       Ridge Province  	  581
9-26.   Bloom-Grigal Model Regional Projections for the Change in Soil pH in the Northeastern
       United States. Organic  Soil  Horizons Included	  585
                                            XVII

-------
                                   TABLES (Continued)
                                                                                     Page

9-27.   Bloom-Grigal Model Regional Projections of the Change in Percent Base Saturation in
       the Northeastern United States. Organic Soil Horizons included	  587
9-28.   Bloom-Grigal Model Regional Projections of the Change in Soil pH in the Northeastern
       United States.  Organic Soil Horizons Included  	  592
9-29.   Bloom-Grigal Model Regional Projections for the Change in Percent Base Saturation in
       the Northeastern United States. Organic Soil Horizons Included 	  594
9-30.   Bloom-Grigal Model Regional Projections for the Change in Soil pH  in the Southern
       Blue Ridge Province.  Organic Soil Horizons Included	  598
9-31.   Bloom-Grigal Model Regional Projections for the Change in Percent Base Saturation
       in the Southern Blue Ridge Province.  Organic Soil Horizons  Included  	  600
9-32.   Summary  of the  Bloom-Grigal Projected  Changes in Soil  pH and Percent Base
       Saturation in the NE and SBRP Under Constant LTA Deposition 	  603
9-33.   Comparison of the Results from the Reuss and Bloom-Grigal
       Models with Regard to the Magnitude  of Changes in Soil pH and Base Saturation
       Projected in Soils of the NE   	  607
9-34.   Comparison of the Results from the Reuss and Bloom-Grigal Models with Regard to
       the Magnitude of Changes in Soil  pH  and Base Saturation Projected in Soils of the
       SBRP		  613

10-1.   Major Processes Incorporated in the Dynamic Model Codes	  621
10-2.   Meteorological Data Required by the Dynamics Model  Codes	  623
10-3.   Chemical Constituents in Wet and Dry Deposition Considered by the MAGIC, ETD, and
       ILWAS Codes   	  624
10-4.   Chemical Constituents  Included in Soil Solutions
       and Surface Water for  the MAGIC,  ETD, and ILWAS Codes	 .  625
10-5.   Definitions of Acid Neutralizing Capacity (ANC) Used by the MAGIC, ETD,
       and ILWAS Codes (Brackets indicate concentration in  molar or molal units, and R',
       R", and R'" represent mono-, di-, and  triprotic organic acids, respectively,)  ANC
       Simulated by All Three Models is Equivalent to the Modified Gran ANC  	  626
10-6.   Level 111  Operational Assumptions	 , , ,	  630
10-7.   Comparison of Calibration/Confirmation RMSE for Woods Lake Among ETD, ILWAS, and
       MAGIC Models, with the Standard Error of the Observations	, .  653
10-8.   Comparison of Calibration/Confirmation RMSE for Panther Lake Among ETD,
       ILWAS, and MAGIC Models, with the Standard Error of the Observations  	  654
10-9.   Comparison of  Calibration RMSE for  Clear Pond Among ETD, ILWAS,  and MAGIC
       Models,  with the Standard Error of the Observations   	  655
10-10.  Percent  Change in RMSE for MAGIC and ETD for a Ten Percent Change in Parameter
       Values   	  658
10-11.  Watersheds, by Priority Class, for Which Calibration Criteria Were Not Achieved	  671
10-12.  Deposition Variations Used in Input Uncertainty Analyses	  675
10-13.  Target Populations for  Modelling Comparisons and Population Attributes  	  679
10-14.  Descriptive Statistics of Projected ANC, Sulfate, pH,  Calcium Plus  Magnesium, and
       Percent  Sulfur Retention for NE Lakes  in Priority Classes A -  I Using MAGIC for Both
       Current  and Decreased Deposition	  682
10-15.  Change in Median ANC and Sulfate Concentrations Over a 40-Year Period as a
       Function of the  Initial ELS-Phase I or NSS Pilot Survey ANC Groups	  690
10-16.  Descriptive Statistics of Projected ANC, Sulfate, and Percent Sulfur  Retention for NE
       Lakes in Priority Classes A - E Using MAGIC and ETD for Both Current and Decreased
       Deposition	  797
                                           XVIII

-------
                                    TABLES (Continued)


10-17.  Descriptive Statistics for Projected ANC, Sulfate, Percent Sulfur Retention,
       and Calcium Plus Magnesium for NE Lakes in Priority Classes A and B Using
       ETD, ILWAS, and MAGIC for Both Current and Decreased Deposition	  716
10-18.  Descriptive Statistics of Projected ANC, Sulfate, and Percent Sulfur Retention, and
       Calcium and Magnesium for SBRP Streams in Priority Classes A -E Using MAGIC for
       Both Current and Increased Deposition	  744
10-19.  Descriptive Statistics of Projected ANC, Sulfate, Percent Sulfur Retention,
       and Calcium Plus Magnesium for SBRP Streams in Priority Classes A and B Using
       ILWAS and MAGIC  for Both Current and Increased Deposition	  756
10-20.  Effects of Critical Assumptions on Projected Rates of Change	  896
11-1.   Weighted Median  Projected Change  in ANC at 50 Years  for Northeastern DDRP
       Lakes	  809
11-2.   Lakes in the NE Projected to Have ANC Values <0 and <50 peq L"1 for
       Constant and Decreased Sulfur Deposition  	  812
11-3.   Weighted Median Projected Change in ANC at 50 Years for DDRP SBRP
       Stream Reaches  	  816
11-4.   SBRP Stream Reaches Projected to Have ANC Values <0 and <50 ^eq L1 for
       Constant and Increased Sulfur Deposition	  819
                                            XIX

-------
                                          FIGURES
1-1.    Steps of the Direct/Delayed Response Project (DDRP) approach	   6

2-1.    Activities of the Aquatic Effects Research Program within the National Acid
       Precipitation Assessment Program	  25

3-1.    Diagram of sulfur cycle in forest ecosystems	  33
3-2.    Diagram of terrestrial base cation cycle	  41

4-1.    Steps of the Direct/Delayed Response Project (DDRP) approach	  50

5-1.    Representation of the point frame sampling procedure for selecting NSS
       Stage I  reaches	  59
5-2.    DDRP site locations for Subregion 1A. 	  72
5-3.    DDRP site locations for Subregion 1B	  73
5-4.    DDRP site locations for Subregion 1C	  74
5-5.    DDRP site locations for Subregion 1D	  75
5-6.    DDRP site locations for Subregion 1E	  76
5-7.    The pH-ANC relationship for (A) lakes of the ELS Phase I sampling in the Northeast
       and (B) DDRP study lakes in the Northeast,	  90
5-8.    The pH-ANC relationship for samples with ANC <400 jueq L taken at the downstream
       nodes of stream reaches sampled in the NSS	  94
5-9.    Location of Northeast field check sites and other DDRP watersheds	  108
5-10.   Example of digitization log sheet	  125
5-11.   Example of attribute entry log sheet	  126
5-12.   Definition of soil sampling classes for the DDRP Soil Survey in the  Northeast	  149
5-13.   Definition of soil sampling classes for the DDRP Soil Survey in the Southern
       Blue Ridge Province	  151
5-14.   Selection of watersheds for sampling	  152
5-15.   Selection of starting points for sampling	  153
5-16.   Field selection of a sampling point for sampling class on a watershed	  154
5-17.   Major steps and datasets from the  DDRP database	  173
5-18.   Calculation percentage of regional or subregional area in each soil sampling	  179
5-19.   Relative areas of sampling classes  in the Northeast subregions.	  180
5-20.   Relative areas of sampling classes  in the entire Northeast and Southern
       Blue Region Province	  181
5-21.   Aggregated soil variables for individual pedons in the Northeast	  182
5-22.   Aggregated soil variables for individual pedons in the Southern Blue Ridge Province.  .  184
5-23.   Calculation of cumulative distribution function for a soil variable in a region
       or subregion	  186
5-24.   Cumulative distribution functions for pedon aggregated soil variables for the
       Northeast and the Southern Blue Ridge Province.	  187
5-25.   Sulfur deposition scenarios for the NE and SBRP for Level II and III Analyses  	  191
5-26.   Example of average annual runoff map for 1951-80  	  226
5-27.   Flow chart of Darcy's Law soil contact calculation as applied to the DDRP
       study sites	  235

7-1.    Estimated percent sulfur retention by in-lake processes in drainage lakes
       in ELS Region 1 (northeastern United States)	  252
7-2.    Percent sulfur retention for intensively studied sites in the United States and
       Canada relative to the southern extent of the Wisconsinan glaciation  	  254
                                             xx

-------
                                     FIGURES (Continued)


7-3.    Model of flow-weighted average concentration calculations for Biscuit Brook	   259
7-4.    Flow chart for the determination of internal sources of sulfur using the
       steady-state sulfate concentration	   267
7-5.    Scatter plot of the Monte Carlo calculated standard deviation versus the
       calculated  mean  [SO/"]^  	   269
7-6.    Comparison of percent sulfur retention calculated using (A) modifled-LTA
       deposition  and (B) modified-LTA deposition adjusted with a 20 percent increase
       in dry  deposition	   272
7-7.    Population-weighted distribution of projected percent sulfur retention
       (upper and lower bounds for 90 percent confidence interval):  (A) Northeast;
       (B) Mid-Appalachians, and (C) Southern Blue Ridge Province	   274
7-8.    Supplemental watersheds mapped for special evaluation of sulfur retention	   276
7-9.    Population-weighted distributions of projected percent  sulfur retention, with
       upper and lower bounds for 90 percent confidence intervals, for additional NSS
       subregions:  (A) Southern Appalachian Plateau, (B) Mid-Atlantic Coastal Plain,
       (C) Catskills/Poconos, and (D) Piedmont.	   281
7-10.   Combination regional population-weighted distributions of projected percent
       sulfur  retention, with upper and lower bounds for 90 percent confidence intervals,
       for the Northeast, Mid-Appalachians, and Southern Blue Ridge Province  	   282

8-1.    Distribution of estimated contact rate using Darcy's Law calculation	   297
8-2.    Distribution of index of contact using Darcy's Law calculation	   298
8-3.    Scatter plot of ANC versus contact rate calculated using Darcy's Law	   300
8-4.    Scatter plot of ANC versus index of soil contact calculated using Darcy's Law	301
8-5.    Scatter plot of ANC versus ln(a/KbTanB)	 .   321
8-6.    Scatter plot of Ca plus Mg versus  ln(a/KbTanB)	   322
8-7.    Scatter plot of pH versus ln(a/KbTanB)	,	   323
8-8.    Data and regression model development flow diagrams	   353
8-9.    Model development procedure	   401
8-10.   Histograms of unadjusted and adjusted watershed means for selected SBRP soils
       variables.	   409
8-11.   The mean  pH ฑ  2 standard errors for the SBRP watersheds estimated using the
       common aggregation and the watershed effects adjusted aggregation the lack of
       variation among the common aggregation values	   410

9-1.    Schematic diagram of extended Langmuir isotherm fitted to data points from
       laboratory  soil analysis	   459
9-2.    Comparison of measured lake (NE) or stream (SBRP)  sulfate concentration with
       computed  soil solution concentration	   462
9-3.    Historic deposition inputs and modelled output for soils in a representative
       watershed  in the northeastern United States.	 .   466
9-4.    Schematic of surface water response to changes in sulfur inputs	   467
9-5.    Comparison of measured, modelled and steady-state sulfate for Northeast lake
       systems  in 1984	   472
9-6.    Projected changes in percent sulfur retention and sulfate concentration for
       soils in northeastern lake systems  at 10, 20, 50 and 100 years	   474
9-7.    Box-and-whisker  plots showing changes in sulfate concentration, percent sulfur
       retention, and change in sulfate concentration for soils in northeastern lake
       watersheds, using long-term average deposition data	   475
                                             XXI

-------
                                     FIGURES (Continued)


9-8.    Box-and-whisker plots showing changes in sulfate concentration, percent sulfur
       retention, and change in sulfate concentration for soils in northeastern lake
       watersheds, using Typical Years deposition data	   476
9-9.    Projected time to steady-state concentration  for sulfate in northeastern lakes (A)
       at current deposition and (B) after end of decreasing input in ramp scenario	   478
9-10.   Historic deposition inputs and modelled output for soils in stream systems in the
       Southern Blue Ridge Province	   480
9-11.   Comparison of measured, modelled, and steady-state  sulfate for stream  systems in
       the Southern Blue Ridge Province in 1985  	   483
9-12.   Comparison of forecasts based on two sulfur deposition datasets for soils in SBRP
       watersheds	   485
9-13.   Projected changes in percent sulfur retention and in sulfate concentration for stream
       systems in the Southern Blue Ridge Province at 0, 20, 50, 100 and 140  years	   487
9-14.   Box and whisker plots showing changes in sulfate concentration, percent sulfur
       retention, and change in sulfate concentration for soils in watersheds of the Southern
       Blue Ridge Province. Data are shown for Typical Year deposition data  	   490
9-15.   Box and whisker plots showing changes in sulfate concentration, percent sulfur
       retention, and change in sulfate concentration for soils in watersheds of the Southern
       Blue Ridge Province.  Data are shown for Typical Year deposition data	   491
9-16.   Projected time to 95 percent of steady-state  sulfur concentration of Southern
       Blue Ridge Province stream systems	   492
9-17.   Comparison of model simulation results for DDRP Southern Blue Ridge
       watersheds	   495
9-18.   Projected base year sulfate concentration with upper and lower bounds  for 90
       percent confidence intervals for Southern  Blue Ridge Province watersheds	   496
9-19.   Projected time to sulfur  steady  state with upper and lower bounds for 90
       percent confidence intervals in  Southern Blue Ridge Province watersheds	   497
9-20.   Effects of data aggregation on  simulated watershed sulfur response for soils
       in DDRP watersheds of  the Southern  Blue Ridge Province	   499
9-21.   Evaluation of alternate soil aggregation procedures for soils in SBRP watersheds. .   . .   500
9-22.   Schematic diagram of the principal process involved in the cycling of base
       cations in surficial environments	   513
9-23.   Plot of the log of the activity  of AI3+ vs. soil solution pH for individual soil
       samples collected for DDRP	   518
9-24.   Plot of the log of the selectivity coefficient for the calcium-aluminum  exchange
       reaction vs. the measured base saturation in A/E horizons in the NE	   520
9-25.   Histograms of the (unweighted for the population estimates) projected
       present-day ANC values for lakes in the NE	   521
9-26.   Histograms of the (unweighted for the population estimates) projected, present-day
       ANC values for lakes in the NE	   523
9-27.   Flow diagram for the one-box Bloom-Grigal soil simulation model	   529
9-28.   Cumulative distribution  of projected, present-day ANC values for lakes  in the  study
       population in the NE as projected using Reuss  cation exchange model	538
9-29.   Scatter plot of the projected, present-day ANC values for lakes in the NE, obtained
       using the Reuss model  vs. observed (ELS) values	   539
9-30.   Scatter plot of the present-day  lake ANC values projected using the Reuss model in
       conjunction with the Watershed-Based Aggregation (WBA) soils data vs. observed (ELS)
       ANC values	   542
9-31.   Cumulative distribution  of the projected surface water ANC values projected for the
       study population  of lakes in 50 years in the  NE	   544
9-32.   Cumulative distribution of the projected surface water  ANC values projected for the
       study population  of lakes in 100 years in the NE	   545

                                             xxii

-------
                                    FIGURES (Continued)
                                                                                       Page

9-33.   Schematic illustration of the titration-like behavior displayed by soils in response to
       constant loadings of acidic  deposition	,	  547
9-34.   Cumulative distribution of projected present-day ANC values for stream reaches
       in the study population in the SBRP, as projections using Reuss's cation exchange
       model	  551
9-35.   Scatter plot of the projected present-day ANC values for stream reaches in the SBRP,
       obtained using the Reuss model, vs. observed (NSS) values	  553
9-36.   Cumulative distribution of projected changes (at  50 years) in surface  water ANC
       obtained using the Reuss model for stream reaches in the SBRP	  556
9-37.   Cumulative distribution of  projected changes (at 100  years) in surface water ANC
       obtained using the Reuss model for stream reaches in the SBRP	  557
9-38.   Comparison of measured vs. calculated soil pH values for the 580 aggregated master
       horizons in the NE	,	  561
9-39.   Cumulative distribution of projected (a) base saturations and (b) soil pH values for soils
       in NE. Projections made using the Reuss model	  564
9-40.   Cumulative distribution of projected (a) base saturations and (b) soil pH values for soils
       In the NE. Projections were made using the Reuss model	  565
9-41.   Plot  of  the  measured  (ELS) ANC values for lakes  in the NE vs. the  estimated,
       watershed-level base saturations for mineral horizons in those watersheds	  566
9-42.   Plot of the changes in surface water ANC values at (a) 20, (b)  50, and (c) 100 years
       as projected by the Reuss model vs. the estimated, present-day, watershed-level base
       saturations for mineral horizons  in those watersheds. .  .	  567
9-43.   Plot of the projected changes in soil base saturations vs. he observed, present-day,
       aggregated base saturations for mineral horizons in the NE. The projections were made
       with the Reuss model	  568
9-44.   Cumulative frequencies of changes in (a) soil base saturation and (b) soil pH for the
       population of soils in the SBRP		  573
9-45.   Cumulative frequencies of changes in (a) soil base saturation and (b) soil pH for the
       population of soils in the SBRP	  574
9-46.   Cumulative distributions  of aggregate initial soil pH and percent base saturation in
       the NE and SBRP, with and without organic horizons	  582
9-47.   Regional CDFs of the projected change in the pH of soils on NE lake watersheds under
       constant and ramp  down (30 percent  4-) deposition scenarios  after  20, 50, and 100
       years of LTA, LTA-rbc, and  LTA-zbc deposition.  Organic horizons included	583
9-48.   Regional CDFs of the projected change in the percent base saturation of soils
       on NE lake watersheds under constant and ramp down (30 percent •!•) deposition
       scenarios after 20, 50, and  100  years of LTA, LTA-rbc, and  LTA-zbc  deposition.
       Organic horizons included	  584
9-49.   Regional CDFs of the projected change in the pH of soils on NE lake watersheds under
       constant and ramp down (30% i) deposition scenarios after 20, 50, and 100 years of
       LTA, LTA-rbc, and LTA-zbc  deposition.  Organic  horizons are excluded	  590
9-50.   Regional CDFs of the projected  change in the percent base saturation of soils on NE
       lake watersheds under constant and ramp down (30% 4-) deposition scenarios after 20,
       50, and 100 years of LTA,  LTA-rbc,  and LTA-zbc deposition.  Organic horizons
       excluded	  591
9-51.   Regional CDFs of the projected change in the pH of soils on SBRP stream watersheds
       under constant and ramp up (20%  t) deposition  scenarios after 20, 50, 100, and 200
       years of LTA, LTA-rbc, and  LTA-zbc deposition.  Organic horizons included	597
9-52.   Regional CDFs of the projected change in the percent base saturation of soils on SBRP
       stream watersheds under constant and ramp up (20% t) deposition scenarios after 20,
       50, 100, and 200 years  of LTA, LTA-rbc, and LTA-zbc deposition. Organic horizons
       included	  598

                                            xxiii

-------
                                    FIGURES (Continued)
                                                                                      Page

9-53.   Cumulative distributions of changes  in soil  base saturation for the population of
       watersheds in the NE	,	  608
9-54.   Cumulative distributions of changes in soil pH for the population of watersheds
       in the NE	,	  609
9-55.   Scatter diagrams of the projected changes in base saturation for individual
       systems (not population weighted) in the NE  obtained from the Reuss and
       Bloom-Griga! models	  610
9-56.   Scatter diagrams of the  projected changes  in soil pH for  individual  systems (not
       population weighted) in the NE obtained from the Reuss and Bloom-Grigal models.  . .  611
9-57.   Cumulative distributions of changes in soil base saturation for the population of
       watersheds in the SBRP	  614
9-58.   Cumulative distributions of changes in soil pH for the population of watersheds
       in the SBRP	  615

10-1.   Modelling priority decision tree:  Northeast	  631
10-2.   Modelling priority decision tree; Southern Blue Ridge Province	, .  633
10-3.   Decision tree used to identify watersheds with net chloride export and procedures for
       determining chloride imbalance	  638
10-4.   Approach used  in performing long-term projections of future changes in surface water
       chemistry	  640
10-5.   Schematic of modelling approach for making  long-term projections.	  641
10-6.   Representation  of horizontal segmentation of Woods Lake,  NY, watershed for MAGIC
       and ETD.	,	  645
10-7.   Representation  of vertical layers of Woods Lake Basin  for ETD	  646
10-8.   Representation  of horizontal segmentation of Woods Lake Basin for ILWAS	648
10-9.   Representation  of vertical layers of Woods Lake Basin  for ILWAS	  649
10-10.  Representation  of vertical layers of Woods Lake,  NY, watershed for MAGIC	  651
10-11.  Comparison of  population histograms for simulated versus observed (Eastern  Lake
       Survey Phase I  1984 values) ANC for ILWAS and MAGIC	,	  662
10-12.  Comparison of  population histograms for simulated versus observed (Eastern  Lake
       Survey -  Phase I 1984 values)  sulfate concentrations for ILWAS and MAGIC, Priority
       Classes A and B	  663
10-13,  Comparison of  population histograms for simulated versus observed (Eastern  Lake
       Survey Phase  I 1984  values)  ANC  and sulfate concentrations for  MAGIC, Priority
       Classes A - E	  665
10-14.  Comparison of  population histograms for simulated versus observed (Eastern  Lake
       Survey Phase I 1984 values ) ANC and sulfate concentrations for  MAGIC, Priority
       Classes A - I	  666
10-15.  Comparison of population histograms for simulated versus observed (NSS Pilot Survey
       values) ANC, Priority Classes A and B using ILWAS and MAGIC	  667
10-16.  Comparison of population histograms for simulated versus observed (NSS Pilot Survey
       values) sulfate concentrations, Priority Classes A and B using ILWAS and MAGIC. . . .  677
10-17.  Comparison of population histograms for simulated versus observed (NSS Pilot Survey
       values) ANC and sulfate concentrations, Priority Classes A - E using MAGIC.  ......  678
10-18.  Comparison of projection standard errors as a function  of ANC (top figure) and sulfate
       (bottom figure)  concentrations for the NE uncertainty analysis watersheds using ETD
       and MAGIC	  685
10-19.  Projections of ANC and sulfate concentrations for NE lakes, Priority Classes
       A -1, using MAGIC for 20, 50, and 100 years, under current deposition and a
       30 percent decrease in deposition	  689
10-20.  pH projections for NE lakes, Priority Classes A - I, using MAGIC for 20, 50,
       and 100 years,  under current deposition and  a 30 percent decrease  in deposition.   . .  692

                                           xxiv

-------
                                    FIGURES (Continued)


10-21.  Box and whisker plots of ANC distributions at 10-year intervals for NE
       Priority Classes A -1 using MAGIC	  686
10-22.  Box and whisker plots of sulfate distributions at 10-year intervals for NE
       Priority Classes A - I using MAGIC	  687
10-23,  Box and whisker plots of pH distributions at 10-year intervals for NE
       Priority Classes A -1 using MAGIC	  688
10-24.  Comparison of population histograms for ANC under current levels of deposition
       and a 30 percent decrease in deposition for NE lakes, Priority Classes  A -1,
       using MAGIC		  691
10-25.  Comparison of population histograms for sulfate concentrations at current
       levels of deposition and a 30 percent decrease for NE lakes, Priority Classes
       A -1, using MAGIC	  692
10-26.  Comparison of MAGIC and  ETD projections of ANC for NE lakes, Priority
       Classes A - E, under current and decreased deposition	  693
10-27.  Comparison of  MAGIC and ETD projections of sulfate concentrations  for NE  lakes,
       Priority Classes A - E, under current and decreased deposition	  694
10-28.  Comparison of MAGIC and  ETD projections of pH for NE lakes, Priority
       Classes A -E, under current and decreased deposition.  ,	  695
10-29.  Comparisons of projected change in ANC under current and decreased
       deposition for NE Priority Classes A - E, using ETD and MAGIC	  699
10-30.  Comparisons of  projected change  in  sulfate  concentrations  under  current and
       decreased  deposition for NE Priority Classes A - E, using  ETD and MAGIC	  700
10-31.  Comparisons of projected change in pH under current and decreased
       deposition for NE Priority Classes A - E, using ETD and MAGIC	  701
10-32.  Box and whisker plots of ANC distributions projected using ETD in 10-year
       intervals for NE lakes, Priority Classes A - E	  702
10-33.  Box and whisker plots of sulfate distributions projected using ETD in
       10-year intervals for NE  lakes, Priority Classes A - E	  703
10-34.  Box and whisker plots of pH projected using ETD in 10-year intervals for
       NE lakes, Priority Classes A - E.	  704
10-35.  Box and whisker plots of ANC distributions in 10-year intervals using MAGIC
       for NE lakes, Priority Classes A - E	  705
10-36.  Box and whisker plots of sulfate distributions in 10-year intervals using
       MAGIC for  NE lakes, Priority Classes  A - E	  706
10-37.  Box and whisker plots of pH in 10-year intervals using MAGIC for NE lakes,
       Priority Classes A - E	  707
10-38.  ETD ANC distributions at year 10 and year 50 for NE lakes, Priority
       Classes A - E, under current and decreased deposition	  708
10-39.  MAGIC ANC distribution at year 10 and year 50 for NE lakes, Priority
       Classes A - E, under current and decreased deposition	  709
10-40.  ETD sulfate distributions at  year 10 and year 50 for NE lakes, Priority
       Classes A - E, under current and decreased deposition	  710
10-41.  MAGIC sulfate distributions  at year 10 and year 50 for NE lakes, Priority
       Classes A - E, under current and decreased deposition	  711
10-42.  Comparison of ANC projections using ETD,  ILWAS, and MAGIC for NE lakes,
       Priority Classes A and B, under current and decreased deposition	  713
10-43.  Comparison of sulfate projections using ETD, ILWAS, and MAGIC for NE lakes,
       Priority Classes A and B, under current and decreased deposition	 .  714
10-44,  Comparison of pH projections using ETD, ILWAS, and MAGIC for NE lakes,
       Priority Classes A and B, under current and decreased deposition.   ,	  715
                                            xxv

-------
                                    FIGURES (Continued)
                                                                                       Page

10-45.  Comparison of ANC projections under current and decreased deposition for
       NE lakes, Priority Classes A and B, at year 20 and year 50 using ETD, ILWAS,
       and MAGIC	   720
10-46.  Comparison of sulfate projections under current and decreased deposition for
       NE lakes, Priority Classes A and B, at year 20 and year 50 using ETD, ILWAS,
       and MAGIC	 .   721
10-47.  Comparison of pH projections under current and decreased deposition for NE
       lakes, Priority Classes A and B, at year 20 and year 50 using ETD, ILWAS,
       and MAGIC	   722
10-48.  Box and whisker plots of ANC distributions in 10-year intervals projected
       using ETD for NE lakes, Priority Classes A and B	   724
10-49.  Box and whisker plots of ANC distributions in 10-year intervals projected
       using ILWAS for NE lakes, Priority Classes A and B	   725
10-50.  Box and whisker plots of ANC distributions in 10-year intervals projected
       using MAGIC for NE lakes,  Priority Classes A and B	   726
10-51.  Box and whisker plots of sulfate distributions in  10-year intervals projected
       using ETD for NE lakes, Priority Classes A and B	   727
10-52.  Box and whisker plots of sulfate distributions in  10-year intervals projected
       using ILWAS for NE lakes, Priority Classes A and B	   728
10-53.  Box and whisker plots of suifate distributions in  10-year intervals projected
       using MAGIC for NE lakes,  Priority Classes A and B	,	 .   729
10-54.  Box and whisker plots of pH distributions in 10-year  intervals projected
       using ETD for NE lakes, Priority Classes A and B	,	   730
10-55.  Box and whisker plots of pH distributions in 10-year  intervals projected
       using ILWAS for NE lakes, Priority Classes A and B	,	   731
10-56.  Box and whisker plots of pH distributions in 10-year  intervals projected
       using MAGIC for NE lakes,  Priority Classes A and B.	   732
10-57.  ETD ANC population distributions at year 10 and year 50 for current and
       decreased deposition	   733
10-58.  ILWAS ANC population distributions at year 10 and year 50 for current and
       decreased deposition	   734
10-59.  MAGIC ANC population distributions at year 10 and year 50 for current  and
       decreased deposition	   735
10-60.  ETD sulfate population distributions at year 10 and year 50 for current and
       decreased deposition	   736
10-61.  ILWAS sulfate population distributions  at year 10 and year 50 for current and
       decreased deposition	   737
10-62.  MAGIC  sulfate population  distributions at year 10  and year  50  for  current arid
       decreased deposition	   738
10-63.  MAGIC ANC and sulfate projections for SBRP streams, Priority Classes A - E,
       at year 20, year 50, year 100, and year 200 under current and increased
       deposition		, . ,,	   740
10-64.  MAGIC pH projections for SBRP streams,  Priority Classes A - E,  at year 20,
       year 50,  year 100, and year 200 under current and increased deposition	   742
10-65.  Box and whisker plots of ANC distributions in 10-year intervals projected
       using MAGIC for SBRP streams, Priority Classes A - E, for current and
       increased deposition	   746
10-66.  Box and whisker plots of sulfate distributions in  10-year intervals projected
       using MAGIC for SBRP streams, Priority Classes A - E, for current and
       increased deposition	   747
                                            XXVI

-------
                                    FIGURES (Continued)
                                                                                       Page

10-67.  Box and whisker plots of pH distributions in 10-year intervals projected
       using MAGIC for SBRP streams, Priority Classes A - E, for current and
       increased  deposition	  748
10-68,  MAGIC ANC population distributions at year 10 and year 50 for current and
       increased  deposition, SBRP streams, Priority Classes A - E	,  750
10-69.  MAGIC sulfate population distributions at year 10 and year 50 for current
       and increased deposition, SBRP streams, Priority Classes A - E	  751
10-70.  Comparison of ILWAS and MAGIC projections for ANC at years 0, 20, and 50
       for SBRP streams, Priority Classes A and B, under current and increased  deposition.  .  753
10-71.  Comparison of ILWAS and MAGIC projections for sulfate concentration at  years
       0, 20, and 50 for SBRP streams, Priority Classes A and B, under current
       and increased deposition	  754
10-72,  Comparison of ILWAS and MAGIC projections for pH at years 0, 20, and 50 for
       SBRP streams, Priority Classes A and B, under current and increased deposition.  . .  .  755
10-73,  Box and whisker plots for ANC distributions in 10-year intervals projected
       using ILWAS for SBRP streams, Priority Classes A and B, for current and
       increased  deposition	  759
10-74,  Box and whisker plots for ANC distributions in 10-year intervals projected
       using MAGIC for SBRP streams, Priority Classes A and B,  for current and
       increased  deposition	  760
10-75.  Box and whisker plots for sulfate distributions  in 10-year intervals
       projected using ILWAS for SBRP streams, Priority Classes  A and B, for  current
       and increased deposition,  .	  761
10-76.  Box and whisker plots for sulfate distributions  in 10-year intervals projected
       using MAGIC for SBRP streams, Priority Classes A and B,  for current and
       increased  deposition	  762
10-77,  Box and whisker plots for pH distributions in 10-year intervals projected
       using ILWAS for SBRP streams, Priority Classes A and B, for current and
       increased  deposition	  763
10-78.  Box and whisker plots for pH distributions in 10-year intervals projected
       using MAGIC for SBRP streams, Priority Classes A and B,  for current and
       increased  deposition	,	  764
10-79.  ILWAS ANC population distributions at year 10 and year 50 for current  and
       increased  deposition, SBRP Priority Class A and B streams	  766
10-80,  MAGIC ANC population distributions at year 10 and year 50 for current and
       increased  deposition, SBRP Priority Class A and B streams	  767
10-81.  ILWAS sulfate population distributions at year  10 and  year 50 for current and
       increased  deposition, SBRP Priority Class A and B streams	  768
10-82.  MAGIC sulfate population distributions at year 10 and year 50 for current  and
       increased  deposition, SBRP Priority Class A and B streams	,  .  769
10-83.  Comparison of projected sulfate versus sulfate steady-state concentrations
       using ETD, ILWAS, and MAGIC for NE lakes	  770
10-84,  Comparison of projected sulfate concentrations under decreased deposition
       with the current sulfate steady-state concentrations using ETD, ILWAS, and
       MAGIC for NE lakes	  772
10-85.  Comparison of projected sulfate concentrations between models for NE lakes
       after 50 years under current and decreased deposition	  773
10-86.  Comparison of projected sulfate versus sulfate steady-state concentrations
       for SBRP streams using ILWAS and MAGIC under both current and increased
       deposition,	  774
10-87.  Comparison of projected ANC between models in NE lakes after 50 years
       under current and decreased deposition	  775

                                            xxvii

-------
                                   FIGURES (Continued)


10-88.  Projected changes in ANC as a function of changes in sulfate for NE lakes
       using ETD, ILWAS, and MAGIC for current and decreased deposition	  776
10-89.  Comparison of pH - ANC relationship for each of the models	  777
10-90.  Comparison of projected pH values between models for NE lakes after 50 years
       under current and decreased  deposition	  779
10-91.  Comparison  of projected changes in calcium and magnesium versus changes in
       sulfate using ILWAS and MAGIC for NE lakes	  780
10-92.  Change  in median ANC, calcium and magnesium, and sulfate concentrations
       projected for NE lakes using MAGIC under current and decreased deposition	  781
10-93.  Comparison of the change in  pH after 50 years as a function of the initial
       calibrated pH for  MAGIC, ETD and ILWAS  on northeastern lakes	  782
10-94.  Comparisons of projected ANC and sulfate concentrations and pH between
       ILWAS and MAGIC after 50 years for SBRP streams	  793
10-95.  Comparison of projected AANC and Asulfate relationships in SBRP Priority
       Class A  and B streams using  ILWAS and MAGIC	  785
10-96.  Comparison of projected AANC and Asulfate relationships and A(calcium and
       magnesium)  and  Asulfate relationships for  SBRP Priority Class A - E streams
       using MAGIC	  786
10-97.  Comparison of projected A(calcium and  magnesium) and Asulfate relationships
       for SBRP Priority  Class A and B streams using ILWAS and MAGIC	  787
10-98.  Change  in median ANC, calcium and magnesium, and sulfate concentrations
       projected for SBRP streams under current  and increased deposition using MAGIC.  . .  788
10-99.  Comparison of the change in  pH after 200 years as a function of the initial
       calibrated pH for  MAGIC on SBRP streams, Priority Classes A - E	  789
10-100. Comparison of projected MAGIC change in pH versus derived pH after 50 years
       for NE lakes	  793
                                          XXVIII

-------
                                          PLATES
1-1.    Direct/Delayed Response Project study regions and sites	   3
1-2.    Sulfur retention and wet sulfate deposition for National Surface Water Survey
       subregions in the eastern United States	   9
1-3.    Changes in sulfur retention in the Southern Blue Ridge Province as projected
       by MAGIC for constant sulfur deposition	  11
1-4.    Change in median ANC of northeastern lakes at 50 years as projected by MAGIC ....  13
1-5.    Change in median ANC of Southern Blue Ridge  Province stream reaches at 50 years
       as projected by MAGIC  	  16

2-1.    Direct/Delayed Response Project study regions and sites	  26

5-1.    Northeastern subregions and ANC map classes,  Eastern Lake Survey Phase I	  56
5-2.    ANC of DDRP  lakes by ANC group	  77
5-3,    DDRP stream reach study sites in the Southern Blue Ridge Province	  81
5-4.    Final DDRP classification of lake hydrologic type - Subregion 1A.	  84
5-5.    Final DDRP classification of lake hydrologic type - Subregion 1B	  85
5-6.    Final DDRP classification of lake hydrologic type - Subregion 1C	  86
5-7.    Final DDRP classification of lake hydrologic type - Subregion 1D	  87
5-8.    Final DDRP classification of lake hydrologic type - Subregion 1E	  88
5-9.    Example of watershed soil map   .,	   119
5-10.   Example of watershed vegetation map	   120
5-11.   Example of depth-to-bedrock map	   121
5-12.   Example of watershed land use  map	   122
5-13.   Example of watershed geology map	   123
5-14.   Example of 40-ft contour delineations on a  15' topographic  map	   131
5-15.   Example of combination  buffer: (A)  stream and 30-m  linear buffer for streams, (B)
       wetlands and 30-m linear buffer for wetlands, (C) elevational buffer for lake, and (D)
       combination of all preceding buffers	   133
5-16.   ADS and NCDC sites linked  with DDRP study sites for NE Subregion 1A	   194
5-17.   ADS and NCDC sites linked  with DDRP study sites for NE Subregion 1B	   195
5-18.   ADS and NCDC sites linked  with DDRP study sites for NE Subregion 1C	   196
5-19.   ADS and NCDC sites linked  with DDRP study sites for NE Subregion 1D	   197
5-20.   ADS and NCDC sites linked  with DDRP study sites for NE Subregion 1E	   198
5-21.   ADS and NCDC sites linked  with DDRP study sites for the  SBRP	   199
5-22.   DDRP study sites relative to  distance from Atlantic Coast  	   204
5-23.   Pattern of typical year sulfate deposition for the DDRP NE study sites	   209
5-24.   Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1A.  . .   210
5-25.   Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1 B.  . .   211
5-26.   Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1C.  . .   212
5-27.   Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1D.  . .   213
5-28.   Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1E.  . .   214
5-29.   Pattern of typical year sulfate deposition for the DDRP SBRP study sites	 .   215
5-30.   Pattern of LTA sulfate deposition for the DDRP NE study sites	   217
5-31.   Pattern of LTA sulfate deposition for the DDRP study sites  in Subregion 1A	   218
5-32.   Pattern of LTA sulfate deposition for the DDRP study sites  in Subregion 1B	   219
5-33.   Pattern of LTA sulfate deposition for the DDRP study sites  in Subregion 1C	   220
5-34.   Pattern of LTA sulfate deposition for the DDRP study sites  in Subregion 1D	   221
5-35.   Pattern of LTA sulfate deposition for the DDRP study sites  in Subregion 1E	   222
5-36.   Pattern of LTA sulfate deposition for the DDRP SBRP study sites	   223
                                            XXIX

-------
                                    PLATES (Continued)


7-1.    Sulfur retention and wet sulfate deposition for National Surface Water Survey
       subregions in the eastern United States	  275
7-2,    Regional percent sulfur retention by major land resource area (MLRA) based
       on target populations (ELS and NSS sites). ,	  283

11-1.   Sulfur retention and wet sulfate deposition for National Surface Water Survey
       subregions in the eastern United States	  802
11-2.   Changes in sulfur retention in the Southern  Blue Ridge Province as  projected by
       MAGIC for constant sulfur deposition	  804
11-3.   Change in median ANC of northeastern lakes at 50 years as projected  by MAGIC .  .  ,  808
11-4.   ANCs of northeastern  lakes versus time, as projected by MAGIC for constant sulfur
       deposition	  810
11 -5.   ANCs of northeastern lakes versus time, as projected by MAGIC for decreased sulfur
       deposition	  811
11-6.   Changes in median pH of northeastern lakes at 50 years as projected by MAGIC .  .  .  813
11 -7.   Change in median ANC of Southern Blue Ridge Province stream reaches at  50 years
       as projected by MAGIC	  815
11-8.   ANCs of Southern Blue Ridge Province stream reaches versus time, as projected by
       MAGIC for constant sulfur deposition	 ,	  817
11-9.   ANCs of Southern Blue Ridge Province stream reaches versus time, as projected by
       MAGIC for increased sulfur deposition	  818
11-10.  Changes in pH of SBRP stream reaches as projected by MAGIC	  821
11-11.  Changes in pH of SBRP stream reaches as projected by ILWAS	  822
                                            xxx

-------
PRIMARY CONTRIBUTORS TO THE DDRP REPORT

     The Direct/Delayed Response Project and this Review Draft Report represent the efforts of many
scientists, technical and support staff.  The primary contributors to this report are noted here.

Section  1:  Executive Summary
     M. R. Church, U.S.  Environmental Protection Agency

Section  2:  Introduction
     M. R. Church, U.S.  Environmental Protection Agency

Section  3:  Processes of Acidification
     P. W. Shaffer, NSI Technology Services Corp.
     G. R. Holdren, NSI  Technology Services Corp.
     M. R. Church, U.S.  Environmental Protection Agency

Section  4:  Project Approach
     M. R. Church, U.S.  Environmental Protection Agency

Section  5:  Data Sources and Descriptions1
     L J. Blume, U.S. Environmental Protection Agency
     G. E. Byers, Lockheed Engineering and Sciences Co.
     W. G. Campbell, NSI Technology Services Corp.
     M. R. Church, U.S.  Environmental Protection Agency
     D. A. Lammers,  U.S.D.A. Forest Service
     J. J. Lee, U.S. Environmental Protection Agency
     L H. Liegel, U.S.D.A.  Forest Service
     D. C. Mortenson, NSI Technology Services Corp.
     C. J. Palmer, NSI Technology Services Corp.
     M. L. Papp, Lockheed Engineering and Sciences Co.
     B. P. Rochelle, NSI Technology Services Corp.
     D. D. Schmoyer, Martin Marietta Energy Systems, Inc.
     K. W. Thornton, FTN & Associates, Ltd.
     R. S. Turner, Oak Ridge National Laboratory
     R. D. Van Remortel, Lockheed Engineering and Sciences Co.

Section  6:  Regionalization of Analytical Results
     D. L. Stevens, Eastern Oregon State University
     K. W. Thornton, FTN & Associates, Ltd.

Section  7:  Watershed Sulfur Retention
     B. P. Rochelle, NSI Technology Services Corp.
     M. R. Church, U.S.  Environmental Protection Agency
     P. W. Shaffer, NSI Technology Services Corp.
     G. R. Holdren, NSI  Technology Services Corp.

Section  8:  Level I  Statistical Analyses
     M. G. Johnson,  NSI Technology Services Corp.
     R. S. Turner, Oak Ridge National Laboratory
     D. L. Cassell, NSI Technology Services Corp.
     D. L. Stevens, Eastern Oregon State University
     M. B. Adams, Automated Systems  Group, Inc.
     C. C. Brandt, Oak Ridge National Laboratory
     W. G. Campbell, NSI Technology Services Corp.
     M. R. Church, U.S.  Environmental Protection Agency
     G. R. Holdren, NSI  Technology Services Corp.
     L H. Liegel, U.S.D.A.  Forest Service
                                             XXXI

-------
Section 8: Level i Statistical Analyses (continued):
      B. P. Rochelle, NSI Technology Services Corp.
      P. F. Ryan, University of Tennessee
      D, D. Schmoyer, Martin Marietta Energy Systems, Inc.
      P. W. Shaffer, NSI Technology Services Corp.
      D. A. Wolf, Martin  Marietta Energy Systems, Inc.

Section 9:  Level II  Single-Factor Time Estimates1
      G. R. Holdren, NSI Technology Services  Corp.
      M. G. Johnson, NSI Technology Services Corp.
      C. I. Lift, Utah State University
      P. W. Shaffer, NSI Technology Services Corp.

Section 10:  Level III Dynamic Watershed Models
      K. W. Thornton, FTN & Associates, Ltd.
      D. L Stevens, Eastern Oregon State University
      M. R. Church, U.S. Environmental  Protection Agency
      C. I. Lift, Utah State University
           Extramural Cooperators Providing Modelling Expertise and Support:
                 C. C. Brandt,  Oak Ridge National  Laboratory
                 B. J.  Cosby, University of Virginia
                 S. A.  Gherini,  Tetra-Tech, Inc.
                 G. M.  Hornberger, University of Virginia
                 M. Lang, Tetra-Tech, Inc.
                 S. Lee, University of Iowa
                 R. K.  Munson, Tetra-Tech, Inc.
                 R. M. Newton, Smith College
                 N. P. Nikolaidis, University of Connecticut
                 P. F.  Ryan, University  of Tennessee
                 J. L. Schnoor, University of Iowa
                 R. S.  Turner, Oak Ridge National  Laboratory
                 D. M. Wolock, U.S. Geological Survey

Section 11:  Integration and  Summary
      M. R. Church, U.S. Environmental  Protection Agency
      P. W. Shaffer, NSI Technology Services Corp.
1 Contributors to this section listed alphabetically
  Beginning on this line, remaining contributors listed alphabetically
                                               xxxii

-------
                                    ACKNOWLEDGMENTS

      The performance of this  portion of the  Direct/Delayed  Response  Project  (DDRP)  and the
preparation of this report have required the efforts of hundreds of scientists and support personnel. We
acknowledge here a few of those persons who made particularly  outstanding contributions. To all the
others who helped us,  but who are not named here, we also extend  our sincere thanks.

      William Ruckleshaus led the way in calling for the initiation of the DDRP and  Lee Thomas showed
a continued  and very patient interest in seeing that it was completed  properly.  We thank them for their
foresight and leadership.

      Courtney Riordan and Gary Foley of the EPA Office of Research and Development (ORD)  provided
much encouragement and support for the Project throughout its development and  implementation. We
thank them for their appreciation of the technical  complexity of the task.

      Rick Linthurst, the first Director  of the Aquatic Effects  Research  Program (AERP), played an
absolutely critical role in the development and nurturing of the Project during  its early years.  We greatly
appreciate his early and continuing commitment to the DDRP.  Dan McKenzie, as Director of the AERP,
provided important continuing support for the  Project and  we thank him for his efforts in helping guide
this phase of the Project to its conclusion.

      Tom Murphy, Laboratory Director for EPA's Environmental Research Laboratory-Corvallis (ERL-C),
and Ray Wilhour, Bob  Lackey and Spence Peterson, Branch Chiefs  for ERL-C, have all supported the
Project and  its staff from the first to the last.  We thank them for their support.

      Dwain Winters and Brian McLean from the Office of Air and Radiation at EPA-Headquarters provided
insight and suggestions for analyses of particular relevance to questions of  Agency policy. We thank
them  for their interest and assistance.

      Dixon  Landers, Technical Director of the National Surface Water Survey, Jay Messer, Technical
Director of the Pilot Stream Survey, and Phil Kaufmann, Technical Director of the National Stream Survey
and their staffs all provided valuable help in interpreting and correctly using their surface water chemistry
data.  We thank especially Tim  Sullivan,  Joe Eilers,  Jim  Blick, Mark DeHaan,  Alan  Herlihy and Mark
Mitch.

      Jim Omernik (EPA),  Andy Kinney (NSI) and Andy Herstrom (NSI) provided many interesting hours
of instruction and discussion on the topics of physical geography and the proper use and application
of Geographic Information Systems.  Our efforts in these technical areas have certainly profited from
their valuable advice and counsel.

      Bill Fallen  (ORD), Chuck Frank (EPA) and his staff, Linda Looney (EPA), and  Cindy Burgeson (NSI
Technology  Services Corp.) all have provided much administrative assistance to help keep the Project
moving in the right direction and at the pace required. We thank them all for their efforts and assistance.

      Many  landowners and state and government agencies allowed us  to  map  and sample  soils on
their properties.  We thank them for permission to do so.
                                             XXXIII

-------
      The cooperation  of the U.S.  Department of Agriculture (USDA) Soil Conservation Service (SCS)
was essential to the completion of the DDRP Soil Survey. People in the SCS state offices who were
responsible for mapping of DDRP watersheds and obtaining the soil descriptions and samples included
Ed  Sautter, Roy Shook (Connecticut and Rhode Island); Gene Grice,  Steve Hundley (Massachusetts);
Dick Babcock, Bob Joslin, Kenny LaFlamme (Maine); Sid Pilgrim, Henry Mount (New Hampshire); Fred
Gilbert, Keith  Wheeler, Will Hanna (New York);  Garland Lipscomb,  George Martin (Pennsylvania); Dave
Van Houten (Vermont); Talbert Gerald,  Bob Wilkes (Georgia); Horace Smith, Andy Goodwin  (North
Carolina); Darwin Newton, David Lewis (Tennessee); Niles McLoda (Virginia). In addition, more than 100
soil scientists were involved in mapping and sampling.

      Regional consistency and comparability was greatly assisted by the efforts of people at the SCS
National Technical  Centers, especially Oliver Rice, Ted Miller (Northeast) and Larry Ratliff (South). The
continuing support of DDRP activities by Milt Meyer, Ken Hinkley, and Dick Arnold of the SCS National
Office was extremely helpful.

      John Warner, former  SCS  Assistant State  Soil  Scientist  for  New  York  was  the  Regional
Correlator/Coordinator  of the Soil Survey for both the Northeast and Mid-Appalachian Regions. Hubert
Byrd, former State  Soil  Scientist for North Carolina, served as RCC for the SBRP Soil Survey.

      Elissa Levine and Harvey Luce (University of Connecticut), Bill Waltman and Ray Bryant (Cornell
University), Cheryl Spencer and Ivan Fernandez (University of Maine), Steve Bodine and Peter Veneman
(University of Massachusetts), Bill Smith and Lee Norfleet (Clemson University),  and Dave Litzke and
Marilew Battling  (University of Tennessee) supervised the operation of the soils preparation laboratories
for  the DDRP Soil Survey.

      A large and dedicated staff at EPA's Environmental Monitoring and Systems Laboratory-Las Vegas
(EMSL-LV) played an absolutely crucial role in  support of the DDRP Soil Survey.  Gareth Pearson and
Bob Schonbrod  provided supervisory guidance for the DDRP Soil Survey activities at EMSL-LV.  Lou
Blume (EPA)  served as Technical Monitor for the program and was responsible for delivery of verified
field, soil preparation laboratory, and analytical  databases.  Lou Blume  was responsible for contracting
and management of soil preparation laboratories and analytical  laboratories and for the delivery of
operations reports, quality assurance reports, methods manuals and field sampling manuals for the Soil
Survey.  Mike Papp of Lockheed Engineering  and  Sciences Corporation (LESC) was responsible for
delivery of verified field, soil preparation and analytical databases for the  Soil Survey. Rick Van Remortel
(LESC) assisted  in  the verification of the SBRP  analytical database  and  in the preparation of laboratory
operations and quality assurance reports.  Bill Cole (LESC) was the Task Lead for the verification of the
analytical  database for  the NE and  assisted  in the preparation of the  methods manual and quality
assurance report for the NE Soil Survey. Gerry Byers (LESC) assisted in the preparation of methods
manuals  and  quality assurance reports for the  NE and SBRP.  Marilew Bartling (LESC)  served as the
Task Lead for the verification of Soil Survey data for the SBRP, served as a manager of a soil preparation
laboratory for the SBRP Soil Survey and contributed to the operations and quality assurance reports for
the SBRP.  Rod  Slagle  (LESC) served as the DDRP soils database  manager at EMSL-LV.  Steve Simon
and Dan Hillman (LESC) assisted in methods  development and  project  implementation early in the
Project.  Craig  Palmer of the Environmental Research Center of the University of Nevada-Las Vegas
provided invaluable technical assistance on quality assurance of soils analytical data.
                                             xxxiv

-------
      Deborah Coffey (NSI) played a critical role in ensuring the quality of the watershed and soils data
gathered for the Project.  She either had a major responsibility for, or assisted in, the development of
data quality objectives,  field sampling  manuals,  laboratory methods  manuals, field operations reports,
field quality assurance reports and  numerous other facets of the  Soil Survey.  We thank  her for her
unswerving attention to  detail.  Jeff Kern (NSI) has also assisted in helping to assure the quality of field
and laboratory data.

      Other scientists who made major contributions to the design of the soil survey activities included
Stan  Buol (North  Carolina State University), John  Ferwerda (University of Maine-Orono),  Maurice
Mausbach (Soil  Conservation  Service), Ben  Hajek (Auburn  University), John Reuss  (Colorado State
University), Mark David  (University of Illinois), and Fred Kaisaki (Soil Conservation Service).

      Phil Arberg  (EPA) and Dave  Williams (LESC) of EMSL-LV were responsible for acquisition and
interpretation of aerial photography of the DDRP watersheds.

      Numerous extramural cooperators assisted in this Project.  Jack Cosby, George Hornberger, Pat
Ryan and David Wolock (University of Virginia), Jerry Schnoor, Tom Lee, Nikolaos  Nikolaidis,  Konstantine
Georgakakos and  Harihar Rajaram (University of Iowa), Steve Gherini, Ron Munson and Margaret Lang
(Tetra-Tech, Inc.), Carl  Chen  and Louis  Gomez (Systech,  Inc.)  all  assisted in watershed modelling
analyses. Bob Newton  of Smith College assisted in gathering supplementary watershed data for  use in
calibrating the models to the Special interest lake/watersheds in the Adirondacks.  John Reuss and Mark
Walthali of Colorado State University and Tom Voice of Michigan State University performed investigations
of processes of base cation supply and sulfate adsorption, respectively, that assisted  us in interpreting
our Soil Survey data and in modelling soil responses.  Warren Gebert, Bill Krug, David Graczyk and Greg
Allord of  the U.S. Geological  Survey (Madison,  Wisconsin) supplied runoff data and maps that were
crucial to the Project.  Wayne Swank and Jack Waide of the USDA Forest Service cooperated with the
Project  in allowing us to use data gathered by the Coweeta Hydrologic Laboratory.   Jack Waide also
provided many insights into the workings of watersheds in the Southern Blue Ridge and  in the application
of watershed simulation models. Tony Olsen, Sally Wampler and Jeanne Simpson of Battelle Pacific
Northwest Laboratories  provided a great deal of information on estimates  of wet deposition to sites of
interest in the Eastern United States.  Tony Olsen also assisted in editing text describing analyses of the
wet deposition data. Robin Dennis and Terry Clark of the EPA's Atmospheric and Exposure Assessment
Laboratory-Research Triangle Park and Steve Seilkop  of Analytical Services, Incorporated, provided key
information on estimates of atmospheric dry deposition. Steve Lindberg of Oak Ridge National Laboratory
and  Bruce Hicks  and Tilden  Myers of the  National  Oceanographic and Atmospheric Administration
provided considerable assistance in the form of discussions and preliminary data on rates of atmospheric
dry deposition. We thank all of these cooperators for their assistance.

      No project of the magnitude of the DDRP can  be  successfully completed  without the assistance
of peer reviewers.  The DDRP benefited  immensely from peer review comments all  the way from its
inception to the completion of this report.

      The following scientists served as reviewers of the initial Review Draft Report:  David Grigal of the
University of Minnesota, Peter Chester,  R. Skeffington and D. Brown of the Central Electricity Generating
Board (London),  Jerry  Elwood  of Oak Ridge National  Laboratory,  John  Melack of  the University of
California - Santa Barbara, Phil Kaufmann  of  Utah State  University, Bruce Hicks of the  National
Oceanographic and Atmospheric Administration,  Gary Stensland of the Illinois State Water Survey, Jack

                                              xxxv

-------
Waide of the USDA Forest Service,  David Lam  of the National  Water Research Institute  (Burlington,
Ontario), Nils Christophersen of the Institute of Hydrology (Wallingford  Oxon, Great  Britain), Bill McFee
of Purdue University, Steve Norton  of the University of Maine, Scott Overton of Oregon State University,
Ken  Reckhow of Duke University, Dale Johnson  of the Desert  Research Institute (Reno,  Nevada), and
Gray Henderson of the University of Missouri.  We thank these scientists for their efforts in  reviewing a
long and complex  document.  We especially thank Dave Grigal (Chairman), Jerry Elwood, John Melack
and  Phil Kaufmann who served  on the Overview  Committee of reviewers. This report benefitted greatly
from the comments and constructive  criticisms of all of these reviewers.

      Numerous other scientists also served  as  reviewers over the  years of individual aspects of the
Project  or of the Project as a whole.   We  thank them also for  helping  us to improve the quality of the
work that we performed.

      Dave Marmorek, Mike Jones, Tim Webb and Dave Barnard  of ESSA, Ltd. provided much valuable
assistance in the  planning of various phases of the DDRP.  Their assistance in this  planning was
invaluable.

      John Berglund of InstaGraphics, Inc. prepared many of the figures that appear in this  report.  We
thank him for the fine job  that he did.

      A majority of the word processing throughout the DDRP and, especially, for this report was done
by Carol Roberts  of NSI.  We thank  Carol for  her many, many hours of diligent work and for her
forbearance  in helping us in our attempts to get everything "exactly right".  Significant word processing
support was also provided by Laurie Ippoliti (NSI), Amy Vickland (NSI),  Lana McDonald, Rose  Mary Hall
and Deborah Pettiford of Oak Ridge National Laboratory, and Eva Bushman and Suzanne Labbe of Action
Business Services.

      Penelope Kellar and Perry Suk  of Kilkelly Environmental Associates performed  truly amazing tasks
in editing both the  Review Draft and Final  Draft of this report.  The job could not have been completed
on time without their efforts.  Ann  Hairston (NSI), Amy Vickland  (NSI),  Susan Christie (NSI) and Linda
Allison (ORNL) also provided  important editorial assistance.

      The DDRP Technical Director sincerely thanks all of the Project staff and extramural cooperators
for their unquenchable enthusiasm and dedication to seeing that this very tough job was done  correctly.
Good work gang...thank you.
                                             xxxvi

-------
                                           SECTION 7

                               WATERSHED SULFUR RETENTION

7.1  INTRODUCTION

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

      The purpose of the watershed sulfur retention component of the DDRP  Level  I Analyses  Is to
estimate the current status of annual sulfur retention in watersheds of the eastern United States, with
primary emphasis on the NE, Mid-Appalachian, and SBRP Regions. The  Mid-Appalachian Region provides
important information for the Interpretation of sulfur retention patterns from the NE to the SBRP.  Specific
objectives of this section are to

      •    examine the influence of in-iake sulfur retention on watershed sulfur retention  estimates;

           assess the contributions of internal sources of sulfur to (and the possible influences on) sulfur
           input/output budget calculations;

      *    characterize current average annual  input/output budgets in the NE, Mid-Appalachians, and
           SBRP using (1)  data from intensively studied  sites  and (2) estimates computed  using
           regionally extensive datasets;

      •    compare annual  sulfur retention  patterns within and among regions to determine possible
           trends relative to water chemistry, soils, and atmospheric deposition; and

      •    conduct an uncertainty analysis  of the  sulfur retention  estimates based on the associated
           uncertainties of the factors used  in the  input/output budget calculations.
                                               247

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

7.2.1   Introduction

      Section 3.3  describes several processes that can cause sulfur to be retained within watersheds.
One of the processes considered is retention by sulfate reduction in wetlands and/or lakes. Retention in
these environments occurs principally by dissimilatory reduction, with sulfate used as an electron acceptor
and with hydrogen sulfide, organic sulfur,  or metal sulfides as end products (Rudd et al., 1986; Brezonik
et al.,  1987).

      The occurrence of sulfate reduction in  anaerobic hypolimnetic waters in  lakes  has long been
recognized,  but has  been considered  unimportant in long-term sulfur budgets  because sulfides  are
reoxidized during  lake overturn.  Recent studies  in several locations have shown, however, that sulfate
reduction in (anaerobic) sediments overlaid by oxic lake waters can be a major sink for sulfur (e.g., Cook
et al.,  1986; Baker et al., 1986a). Reduction rates are approximately first order for sulfate concentration,
and in-lake rates are apparently limited by diffusion rates into sediments (Baker et al., 1986b; Kelly et al.,
1987).   Sulfides  produced in  lake sediments are  largely retained within the sediment  profile on  a
permanent basis, with little reoxidation or volatilization (Rudd et al., 1986; Brezonik et al., 1987).  Because"
sulfate reduction is rate limited (i.e., by diffusion of sulfate) rather than capacity limited (Rudd et al., 1986),
reduction will likely continue roughly at current rates  (expressed as percent retention)  on a long-term
basis.

      Measured and computed  mass transfer coefficients for sulfate vary over a relatively narrow range
(Baker et al., 1986b; Kelly et al., 1987),  but the importance of in-lake sulfur retention on lake/watershed
sulfur  budgets is highly variable and is greatly influenced by hydraulic residence times of lakes.  Sulfur
retention within lakes  has been discussed and modelled by Baker et  al.  (1986b)  and by Kelly et  al.
(1987), who developed identical equations to predict sulfate  retention  in lakes:

                                  I/-    * inn
          % SO4   Retention =     KSQ4   IUU                                          (Equation 7-1)
                                (ZAJ  + kso4

where:                kSO4  = sulfate mass transfer coefficient (m yr"1 )
                       Z     = lake depth (m)
                        ^   = hydraulic residence time (yr)

Baker et al.  (1986b) and  Kelly et al. (1987) computed mass transfer coefficients using sulfur input/output
budgets from the  literature and  determined average constants of 0.54 and 0.46 m yr"1 , respectively.

      Transfer of sulfate  from the  water-sediment interface to the anoxic zone of the sediments occurs
principally by diffusion. Thus, absolute transfer rates are relatively low, with the result that reduction in
sediments is a small component of lake sulfur fluxes except in lakes with long hydraulic residence times.
High sulfur retention has been  reported for a diverse group of seepage  lakes and other lakes with long
hydraulic residence (e.g., Baker et al., 1988; Schindler et al., 1986b; Lin and Schnoor, 1986).  In contrast,
Shaffer and  Church (1989) evaluated in-lake alkalinity production [to which sulfate reduction is the largest


                                               248

-------
contributor (Schindler,  1986; Brezonik et al., 1987)] and sulfur retention for regional lake populations in
subregions of the  Eastern  Lake Survey (ELS)  (Linthurst  et  al.,  1986a), and concluded  that  in-lake
processes  have only  a minor  effect  on  ANC  and  sulfur budgets  for  most drainage lakes  in  the
northeastern  United States, Upper Midwest (UMW), and Southern  Blue Ridge  (SBR).  [For this section
only, SBR refers to lakes in ELS subregion 3A, which  includes portions of the  Piedmont and Ridge and
Valley provinces (Linthurst et al., 1986a), and encompasses a larger  geographic  area than the stream
systems within the Pilot Stream survey region (Messer et al., 1986a) of the SBRP  considered elsewhere
in the DDRP.]

      Dynamics of sulfur  in freshwater wetlands have been  studied  in detail  at  only a few sites and
probably cannot be described  effectively at regional scales by relationships such as the in-lake retention
expression (Equation 7-1). Rates of sulfur reduction  in wetlands can  be  very high (Weider and  Lang,
1988) and, even small wetland  areas,  depending on their location within a  watershed, can retain a
substantial fraction of watershed sulfur inputs (Calles,  1983; Weider and Lang,  1988).  Generalization of
wetland area -  sulfur  budget relationships is difficult,  however, because the importance of wetland
retention on watershed sulfur budgets depends on the location of the  wetland  in the watershed and the
portion  of  watershed  runoff flowing through it.   Also, sulfur reactions in wetlands and wet soils can
change seasonally or  in wet/dry years.  Wetlands and  wet soils can  act as sulfur sinks (reduction of
sulfur) during wet periods when the system is anaerobic, but  can become major sulfate  sources  due to
reoxidation of sulfides  upon drying  (Bayley et al., 1986;  Nyborg,  1978).

      In this  section, we use the sulfur retention model of Baker et al.  (1986b) with hydrologic data from
the ELS  (Linthurst et al.,  1986a;  Kanciruk et al., 1986a) to estimate  sulfur retention  in  drainage lakes
(including reservoirs) in the northeastern United  States and the Southern  Blue Ridge. Because we lack
models to make direct  estimates  of sulfur reduction in wetlands, regression  analyses are used to describe
relationships  between watershed  sulfur input/output budgets and wetlands for DDRP watersheds. Results
of these analyses are described  in Sections 7.4 and 8.5.

7.2.2 Approach

      The ELS characterized lake depth and hydraulic residence time for a statistically representative set
of lakes in selected areas of the eastern United  States,  including the  Northeast (Linthurst et al.,  1986a;
Kanciruk et al., 1986a). For these analyses,  we used a subset of the ELS population comprised of all
drainage lakes  and reservoirs with  lake areas <2000 ha.  Target populations are listed in Table 7-1.
Using ELS data with Equation 7-1  and assuming a value of 0.5 m yr"1   for ks04  (Baker et al.,  1986b;
Kelly et al., 1987), we estimated sulfur retention by in-lake reduction for  drainage lakes in the northeastern
United States and  for DDRP watersheds.  Due to major uncertainties  in defining hydrologic boundaries
for seepage and closed lakes and resulting uncertainties in hydrologic and chemical budgets, estimates
of in-lake sulfur retention  were made only for drainage lakes and reservoirs.   Based on the sampling
design described by Linthurst et  al.  (1986a), we extrapolated results from sampled lakes  to obtain target
population estimates for each  region and for the five ELS subregions  in the  Northeast.
                                               249

-------
Table 7-1. Summary of Computed Sulfur Retention by In-Iake Reduction for Lake Systems in the
Eastern United States. Data for the Southern Blue Ridge and Upper Midwest are from Shaffer and
Church (1989)
Region
percent
Drainage Lakes8
# %
twb Computed % S retention
(yr) median 90 %ile maximum
% of lakes with > 10
computed S retention
ELS Region 1             (88)
   1A
   1B
   1C
   1D
   1E
.20
3.1
1091
1421
1276
1071
1429
(87)
(96)
(86)
(81)
(94)
.23
.25
.17
.18
.23
2.8
3.9
2.5
3.7
3.0
11.1

 9.0
128
 7.9
12.5
11.1
38.8

25.6
38.8
19.4
21.1
26.6
1Z5

 7.7
19.1
 7.9
17.3
14.4
NE DDRP lakes   137    (94)
.46
4.2
12.5
25.6
18.6
SBR (ELS 3A)     250    (97)      .10     1.2         4.0       5.4

UMW(ELS2)     4404   (52)      .48     5.3        13.0      19.3
                                             23.2
a  "Drainage Lakes" indicates drainage lakes and reservoirs; # is target population, % is percentage of all lakes in the ELS
   target population in each region,

b  Hydroiogic retention time (yr).
                                              250

-------
7.2.3 Results-

      Estimates of sulfur retention for drainage lakes and reservoirs in ELS Region 1 (northeastern United
States, Plate 5-1) are summarized in Table 7-1  and in Figure 7-1.  Computed  in-lake retention was
generally low, with a median retention in the NE of 3.1 percent and more than 10 percent sulfur retention
in  only 12.5 percent  of northeastern drainage systems.  Maximum computed retention  in northeastern
lakes was 39 percent. Retention in individual ELS  subregions was comparable to the region as a whole;
retention in  Subregions  1A (Adirondacks) and 1C (Central New England) was slightly  lower than the
regional distribution, and retention in Subregions 1B (Poconos/Catskills) and 1D  (Southern New England)
slightly higher.  Because drainage lakes and reservoirs comprise 88 percent of lakes in the region and
at  least 81 percent of target lakes in Individual subregions, retention data summarized here represent by
far the majority of target lake systems  in the region. Computed retention for DDRP lakes is  generally
comparable  to, but is slightly higher than, that of the regional target lake population.   The fraction  of
drainage systems in  the DDRP  lakes is higher than for the ELS population  estimate, due principally  to
reclassification of several DDRP lakes (from closed or seepage to drainage) based on data from DDRP
watershed mapping activities (Section 5.3).

      For comparison, data from Shaffer and Church (1989) for two other ELS regions are also included
in  Table 7-1. Lakes in the SBR  (ELS Subregion 3A) are dominated by drainage systems and reservoirs,
which have  very short hydraulic residence times and are consequently projected to have very low in-
lake sulfur retention.  Median  computed retention in SBR  lakes is only 1.2  percent,  and  maximum
retention is 5.4 percent.  Estimated retention in lakes of the UMW (ELS Region 2) is somewhat higher
than in the  NE, with median projected  retention  of 5.3  percent and more than 10 percent retention  in
almost one-fourth of  drainage systems.  An important difference between the NE and UMW lies in the
relative abundance of lake hydrologic types; seepage and closed lakes account for almost half of all lakes
in  the UMW and in-lake processes are probably an important sulfur sink in most of these lake systems.

      The estimates of low sulfur retention In northeastern lake systems are consistent with independent
lines of evidence regarding watershed sulfur budgets and in-lake processes. Our estimates of low sulfur
retention,  consistent  with sulfur input/output  budgets developed  by Rochelle  and  Church (1987) and
discussed in Section 7.3, show lake/watershed systems in the region to be, on average, very close  to
steady state. Data presented here also are consistent with  estimates of Shaffer  et al.  (1988) and Shaffer
and  Church  (1989), based on watershed-to-lake area ratios for ELS watersheds, which suggest that in-
lake processes (principally suifate reduction) are a minor contributor to ANC budgets in most northeastern
lake/watershed systems.

      The relative importance of in-lake  suifate reduction  to  basin sulfur budgets in most systems  is
largely determined by two factors: (1) absolute rates of suifate reduction and (2) lake hydrologic variables
(more explicitly, the volume of water from which suifate is removed or the annual discharge per unit lake
area).  Rates  of suifate "reduction  (as KSO4) apparently vary among lakes over a  fairly narrow range
(Rudd et al., 1986; Kelly et  al., 1987; Brezonik et al., 1987) and in typical  drainage lakes of the eastern
United States are probably comparable to rates  measured in systems in which reduction is a major
component of sulfur budgets (Brezonik et al., 1987; Kelly et al., 1987). Hydraulic  residence times of lakes,
however, vary greatly among regions.  For example, at the Experimental  Lakes Area in Ontario and  in
many seepage lakes  (e.g., Schindler et al., 1986b; Lin and Schnoor, 1986; Baker et al., 1986a), residence

                                              251

-------
      0
                                 Percent Sulfur Retention
Figure  7-1.  Estimated  percent sulfur retention by  in-lake processes in drainage lakes  in ELS
Region 1 (northeastern  United States).  Retention was computed using the model of Baker et al.
(1986b).
                                          252

-------
times are long and sulfur budgets are greatly influenced by in-lake reduction.  By contrast, residence
times in most drainage lakes of the northeastern United States are short, averaging about two months
(Linthurst et al., 1986a), The relatively minor role of in-lake reduction in drainage lakes of the northeastern
United States is  a consequence of short hydraulic residence times, rather than of low inherent sulfate
reduction rates.  The importance of residence time is explicit in the models of Baker et al. (1986b) and
of Kelly et al.  (1987).  Those authors concluded that in  lakes with short hydraulic residence times (one
year or less), including most lakes In the  northeastern United States, in-lake processes have little  net
effect on watershed sulfur budgets.

7.3  WATERSHED SULFUR RETENTION

      Our first investigation of the regional patterns of  sulfur retention consisted of a  review of sulfur
input/output  budgets at intensively studied sites (Rochelle et al., 1987).   Figure 7-2 summarizes  the
findings from this review.  Definitive statements  about sulfur retention on regional scales could not be
made because of lack of spatial coverage by the intensively studied sites and inconsistencies in data used
for budget calculations. There are trends, however, in sulfur retention from North to South in the eastern
United States, especially relative to the extent of the Wisconsinan glaciation, with higher retention in  the
southern areas (Figure 7-2).  The DDRP Level I sulfur retention analysis examines these apparent trends
in more detail using regionally consistent sulfur input and output data (Section 5) for the surface water
sites sampled by the Eastern Lake Survey  (ELS) and National Stream Survey (NSS).

7,3.1  Methods

7.3.1.1  Input/Output Calculation

      In the Level ! sulfur retention analysis, we used an annual mass balance approach to estimate
percent retention.  The general equation used to calculate percent sulfur retention is:

           % Retention      =     (((Sw + Sd) - (R *  Ss ))/(S „ + Sd))*100           (Equation 7-2)

where:           Inputs
                      Sw   =    wet  sulfur deposition (mass length"2 time"1)
                      Sd   =    dry sulfur deposition (mass length"8 time"1)
                 Outputs
                      R    =    runoff  (length time"1)
                      Ss   ==    surface water sulfur (mass length"3)

Equation 7-2 relates the total sulfur input (on  a mass basis) to each watershed to the total sulfur output.
We applied this  equation to  each of the  study watersheds examined in the Level I  sulfur retention
analyses (ELS and NSS sites).
                                              253

-------
en
               Figure 7-2.  Percent sulfur retention for intensively studied sites in the United States and Canada
               relative to the southern extent of the Wisconsinan glaciation (adapted from Rochelle et al. (1987)).

-------
7.3.1.2 Data Sources

7.3.1.2.1 Inputs -

      Wet sulfur deposition was estimated for each site using chemistry data from the National Trends
Network/National Acid Deposition Program (NTN/NADP) network and precipitation data from the NOAA
National  Climatic Data Center  (NCDC)  network (Section 5.6).   Briefly,  wet sulfate concentrations and
precipitation were kriged to each site, and wet deposition was calculated (see Wampler and Oisen, 1987,
for a detailed description of the calculation).  Dry sulfur deposition was estimated based on output from
the Regional Acid Deposition Model  (RADM)  (see Section 5,6).'

7.3.1.2.2 Outputs -

      We used estimates of annual runoff for the 30-year period  of 1951-80 (see Section 5.7 for details).
For the purpose of these analyses we assumed that the vast majority of sulfur leaves the watershed in
the form of dissolved sulfate  (David and Mitchell, 1985; Mitchell et al., 1986).  Section 5,3 discusses the
chemistry data used in these analyses.  For additional  information concerning the ELS arid NSS surface
water suifate estimates, see Linthurst et al. (1986a), Messer et al. (1986a), and Kaufmann et al. (1988).
Seepage lakes and closed  lakes were excluded from the analyses.

7.3.2  Uncertainty  Estimates

7.3.2.1 Introduction

      We used a Monte Carlo approach to evaluate the uncertainty associated with estimates of annual
average sulfur retention.  (The specific Monte Carlo procedure used is very similar to that described in
Section 6.3.) The critical step in applying the Monte Carlo routine is developing error rates on each of
the input/output variables used in calculating  percent sulfur retention  (see Equation 7-2).  We determined
an uncertainty  distribution for each  of these variables. The uncertainty distributions were  propagated
through the retention equation to determine an  estimate of the overall uncertainty of the percent sulfur
retention calculations.

7.3.2.2 Individual Variable Uncertainties

7.3.2.2.1 Input variables -

      Two variables are used to estimate the total sulfur input to each surface water system - wet and
dry sulfur deposition.  The determination of uncertainty estimates for these variables is discussed in
Section 10.10.  For the sulfur retention uncertainty analyses, we used relative standard deviation (RSD)
estimates of 0.25 for Sw  and 0.50 for Sd.
                                               255

-------
7.3.2.2.2  Output variables -

7.3.2.2.2.1  Runoff -

      The sulfur output from each watershed is a product of the estimated annual average surface water
sulfate concentration and the annual runoff.  Rochelle et al. (in press) determined that runoff for individual
watersheds could be estimated from the map of Krug et al. (in press) within ฑ15 percent.  Based on this
determination, we used an RSD of  0.15 for  runoff in the sulfur retention uncertainty analysis.

7.3.2.2.2.2  Surface water sulfate  concentration  -

      We estimated the annual average surface water sulfate concentration from the single fall index value
for the northeastern lakes (Section 5.3) or  an average of 3 (Pilot  Stream Survey) or 2 (NSS Phase I)
spring baseflow samples for  the SBRP and Mid-Appalachian streams, respectively (Section 5.3).  As
described below, we used extensive temporal data from intensively studied sites to estimate the variability
arising from using an index to represent average annual sulfate concentrations.  Table 7-2 lists the sites
from which data were available and the frequency of data collection at each site.

      First, we calculated flow-weighted annual averages for each year for each site and a  spring and/or
fall flow-weighted average concentration.  The fall and  spring flow-weighted  averages were  calculated
using sulfate concentrations for samples  collected during  periods that corresponded to the sampling
windows used in the ELS (mid-September to early November) (Linthurst et ai., I986a) and NSS (March
15 to May 15)  (Messer et al., 1986a; Kaufmann et al., 1988).  An  additional criterion defining the NSS
sampling window was to sample prior to spring leaf-out.  The spring samples collected by  the NSS were
non-event samples (i.e., baseflow).  To maintain consistency we checked the weekly data used from the
intensively studied SBRP and Mid-Appalachian sites (Table 7-2) to ensure that no samples that were
unduly influenced by events were included in the spring flow-weighted average calculations.

      Sulfur budgets  for  SBRP  watersheds might  be biased to  some  extent  by  their reliance on
streamwater sulfate concentration  data  collected during spring  baseflow periods.  Spring baseflow
chemistry closely approximates weighted  mean annual chemistry computed  from  weekly  grab  samples
for many southeastern systems. However, data representing  preclpitatlon/snowmelt episodes were not
collected as part of the Pilot Stream Survey. For the few watersheds in the Southeast for  which at least
some  episodes have been  characterized,  there  has been  a consistent  trend  of  increased sulfate
concentration during storm episodes [Deep Run and White Oak  Run, VA (Hendrey et al., 1980;  P.W.
Shaffer, unpublished  data),  Fernow, WV (D. Helvey, personal communication),  Walker Branch, TN
(Johnson and Henderson, 1979), Coweeta, NC (Swank and Waide, 1988), Panola Mountain, GA (N. Peters
and R. Hooper, personal communication)].  Due to  the highly variable extent of episodic sulfate increases
and  the  extremely  limited data available for  the  region, the episodic bias  in sulfur budgets for the
Southeast cannot be quantified.  In one system In the SBRP for which detailed sulfate export  budgets
have  been determined (Coweeta WS #2,  three years of data),  sulfate export calculated from  flow-
proportional sampling was 19 percent higher than export calculated from weekly grab sample data (Swank
and Waide,  1988).  The only  other watershed in the region for which comparable analyses have been
completed is Panola Mountain, GA. Panola, located in  the Piedmont near Atlanta, is physiographically
and  climatically different from the DDRP watersheds in  the SBRP, and is subject to extreme  episodic

                                             256

-------
Table 7-2.  Intensively Studied Sites Used in Surface Water Chemistry Uncertainty
Analysis
Site Name

Northeast
1LWAS
Woods Lake
Panther Lake
Sagamore Lake
R1LWAS
Arbutus Lake
Black Pond
Bub/Sis Lakes
Darts Lake
Moss Lake
Pancake Hall Creek
Rondaxe Lake
Townsend Lake
West Lake
Windfall Pond
Clear Pond
Heart Lake
Otter Lake
SBRP and Mid-Appalachian
Coweeta 34
Coweeta 36
Deep Run
White Oak Run
Fernow
Biscuit Brook
Shenandoah Nat. Park
(52 streams)
Years of
Study


5
5
5

3
4
3
3
2
1
4
1
4
3
4
2
2
Streams
6
11
6
5
10
4
1

a Samples were collected between 10 to 13 times
b Biscuit Brook is an episodic
Sample Frequency
for Chemistry


weekly
weekly
weekly

monthly8
monthly
monthly
monthly
monthly
bimonthly
monthly
monthly
monthly
monthly
monthly
monthly
monthly

weekly
weekly
weekly
weekly
weekly
variable13
bimonthly

per year.
Reference


R. Goldstein,
pers. comm.


C. Driscoll,
pers. comm.













J. Waide,
pers. comm.
J. Galloway,
pers. comm.
D. Helvey,
pers. comm.
Lynch and
Dise, 1985

study site. Samples collected periodically
    through out each year, however, during selected events extensive water
    chemistry samples were taken; often on an hourly time basis.
                                         257

-------
increases in sulfate following  prolonged  dry periods.   Estimation of annual sulfate export from spring
baseflow samples at Panola appears to underestimate total annual sulfate export by as much as 50
percent (N. Peters and R. Hooper, unpublished data).  The bias observed at Panola should be regarded
as an  upper bound that might be approached  by a few SBRP  systems.  The climate and moisture
regimes of SBRP watersheds are more similar  to those at Coweeta than to the more xeric conditions at
Panola and at Walker Branch, TN (which also experiences large episodic  increases in sulfate but with
uncertain effects on  sulfate export budgets; Johnson and Henderson, 1979), suggesting that the 19
percent bias  observed at Coweeta is  probably not atypical of SBRP watersheds  (J. Waide, personal
communication).

      Data from 2  extensively studied  sites, Biscuit Brook, NY, and Shenandoah National Park (SNP),
VA, required  special  considerations.  The Biscuit Brook data were collected  as part of a program  to
monitor events. As a result, the dataset contained sections of very extensive temporal data (hourly) along
with  more infrequent sampling through the year.  For this dataset, the flow-weighted annual average
sulfate concentration was estimated by calculating the area under the hydrograph to properly weight the
influence of  any particular event flow value on the overall average (Figure 7-3).  We were not able  to
perform a complete hydrograph separation analysis [such  as described by Dunne and Leopold (1978)]
due to the highly variable temporal sampling of the flow measurements.  To determine the spring flow-
weighted average we used a  flow of 10  cfs as the maximum flow that could be regarded as equal  to
baseflow. The 10-cfs "limit" was determined after examination (simple hydrograph separation) of 4 years
of available data.

      The SNP dataset contains bimonthly flow and  water chemistry data for 52 steams  for one year.
We calculated the flow-weighted annual average  using the 6 flow and concentration measures.  Two of
the 6 values fell within or were close to the March 15  to May 15 time frame used to calculate estimates
of spring baseflow sulfate concentration.  The 2 samples were collected near the beginning and  the end
of the period (March 15-19 and May 17-20, respectively).  Although the March sample was barely within
the period, there was evidence that the flows were higher than the usual spring  baseflow values for the
SNP area (P. Shaffer,  personal communication).   The May sample  was well after leaf-out and the
concentration values were low compared to more extensive data available for Deep Run and White Oak
Run.  [These two watersheds, located in the SNP, are included in the 52-site SNP dataset. They also
have been extensively monitored as part  of the Shenandoah Watershed Acidification Survey (SWAS) (P.
Shaffer, personal communication).]  Although the March sample had very high flows, as noted above, we
used  it,  rather than the sample from May  15,  for  the  error analysis.   The March  sample  sulfate
concentration was closer than the May sample  to the spring flow-weighted average sulfate concentrations
for Deep Run and White Oak Run.

      After  we calculated the 2 flow-weighted averages (annual average  and fall  average for lakes  or
spring average for streams) for all years of data for the intensive study sites listed in Table 7-2, we then
calculated an  estimate of the percent difference (%Diff) between the  two averages, as described  in
Equation 7-3.

                      %Diff       =    ((lnd_Avg - Ann_Avg)/lnd_Avg)*100            (Equation 7-3)
                                              258

-------
                    I	I
      t = time
      f = flow
      c = sulfate concentration
       Avg.c =
Figure 7-3.  Model of flow-weighted average concentration calculations for Biscuit Brook.
                                             259

-------
where:                 lnd_Avg    =    flow-weighted average sulfate concentration for the
                                        index sample time frame (spring or fall)
                       Ann_Avg   =    the flow-weighted annual average sulfate concentration

      In the final step,  we used  the  estimate of %Diff for each watershed and year to determine an
estimate of the uncertainty associated with using the fall index value or the spring  baseflow estimate to
represent average annual chemistry for the sulfur retention analysis.  First, we determined the distribution
of %Diff  for  each set  of  data  (e.g., ILWAS, RILWAS, Fernow).  Next,  we  estimated  an appropriate
uncertainty estimate to be used in the uncertainty calculations for the sulfur retention analyses using the
standard  deviations around the mean  %Diff for each of the intensively studied datasets (Table 7-3).  The
combined dataset had a mean value only slightly above zero and was slightly skewed to higher values.
Both of these aspects can be attributed to the SNP data, which are probably somewhat high because
March data were used.  The  overall  distribution was approximately bell-shaped, with over 95 percent
inside  ฑ2 standard deviations of zero.  Therefore, a lognormal distribution with an BSD of 9.4 percent
was used to describe the uncertainty.

7.3.2.3 Uncertainty Calculation - Monte Carlo Analysis

      Once uncertainty estimates were obtained for each of the input/output variables, the next step was
to combine the information to obtain an overall estimate on uncertainty on the percent retention  estimate.
We did this using Monte Carlo analysis.  The basic strategy employed in the Monte Carlo analysis was
to randomly  select for each  iteration a  value  for each input/output  variable  (e.g., runoff)  from  the
distribution of possible values as determined by  the  associated uncertainty for that variable.  These
randomly  selected input/output variables were then used to calculate  an estimate  of percent sulfur
retention for that particular iteration.   We randomly selected 11  watersheds from the study regions and
ran the Monte Carlo uncertainty using 10,000 iterations.  (This number of iterations was chosen based
on the simplicity of the sulfur retention equation and the computer CPU required. We performed several
tests to evaluate the  influence of the number of iterations  on convergence of the  mean and  standard
deviation and found that results were generally  the same with significantly less than 10,000 iterations.)
The overall uncertainty of the percent retention estimate was determined from the variance of the percent
sulfur retention estimates calculated from the Monte Carlo iterations.

      Based on the results of the Monte Carlo, we determined that a multiplicative normal distribution best
described the percent sulfur retention uncertainty. Finally, we plotted the standard deviation of the Monte
Carlo runs for each watershed  against the average percent retention from the runs.  Equation 7-4 presents
the results of the linear regression that describes this relationship.

                             Std. Dev.   =  30.1  - 0.30 (Avg.)                          (Equation 7-4)
                                  R2    =  0.99
                               Prob>F  =  0.0001
                                  MSE  =  0.15

This relationship, along with estimated variance, was used to calculate a 90 percent confidence interval
about the percent sulfur retention population  estimates presented in Section 7.3.4  (see Figures 7-7 and
7-10).

                                               260

-------
Table 7-3.  Summary Statistics on Percent Differences Between Flow-weighted
Average Annual Sulfate Concentration and the Fall/Spring  Flow-weighted Averages
Study Site

NE
RILWAS
ILWAS
Na


12
32
Mean


-1.1
-3.0
Median


-3.6
-3.0
Std.
Dev.

10.0
8.4
Min


-9.7
-36.4
Max


15.6
8.9
SBRP & Mid-
Appalachian
Fernow
Coweeta-34
Coweeta-36
Biscuit
SNP
SWASb


10
6
11
3
52
11


-6.7
-8.5
-1.8
4.2
8.9
-3.3


-5.7
-7.5
-1.2
1.5
8.6
-2.4


7.5
6.0
10.5
5.6 ,
5.4
5.5


-19.0
-15.9
-28.0
0.4
0.0
-14.7


4.5
0.0
8.7
10.6
23.3
5.0
                137           1.4         1.5         9.5        -36.4        23.3
  N is a combination of the number of years of data and the number sites.
  SWAS includes White Oak Run
  Watershed Acidification Study.
b SWAS includes White Oak Run and Deep Run and stands for the Shenandoah
                                               261

-------
7.3.3  Internal Sources of Sulfur

7.3.3.1  Introduction/Approach

      Sources  of sulfur within a watershed can be  important factors affecting the interpretation of the
annual percent sulfur retention estimates calculated using the input/output budget analyses. In the DDRP
we are interested in percent sulfur retention relative to sulfur deposition. Sulfur from sources such as acid
mine drainage^ natural weathering of sulfide-bearing bedrock, or sulfate-containing sedimentary rocks can
increase the surface water concentration of sulfate,  thus biasing the results of the annual input/output
budget calculations. We have used 2 approaches to identify watersheds with internal sources of sulfur.
The  first approach  uses information on bedrock geology to identify watersheds associated with sulfur-
bearing bedrock. In the second  approach we determine an estimated surface water sulfate concentration
for each site that, if exceeded, indicates (at a designated probability level) that at least some sulfate is
derived from internal sources. This calculation is based on the determination of theoretical steady-state
sulfate concentrations.  This section describes the methods used for, and the results of, (1) the bedrock
geology analyses and  (2) the computation of an upper limit steady-state sulfate concentration.

7.3.3.2 Bedrock Geology

      The first step in the bedrock analysis was to Identify the types of bedrock within each of the DDRP
watersheds. The DDRP subset of watersheds (NE=145, SBRP=35) within the ELS lake and NSS  stream
populations was selected to test whether the approach could be used to identify systems with potential
internal sources of sulfur.  Using the GIS, we overlaid  watershed boundaries onto state geology maps
(Section 5.4.1.7.3.1 and 5.4.2.7.2.1) and then identified the mapped bedrock units within the  boundaries
(Plate 5-13).  State geology maps used in the analysis are listed in Table 7-4.

      After we identified the mapped units associated with a watershed,  we then assessed the potential
for each unit to contribute sulfur to surface waters.  We developed a 3-level stratification for  classifying
each bedrock type.  Mapped units  with high probabilities for contributing sulfur were assigned the value
"Y".  Primarily, these units consisted of calcareous rocks or of rocks identified in the  state map legends
as "sulfitic", "pyrite-bearing", or a similar description.  Bedrock units containing potentially large amounts
of sulfur, but with more limited contact with surface waters, were assigned a value of "P".  These units,
consisting  of black and gray shales, sulfitic slates, fossiliferous sediments (potential carbonate sources),
and  "rusty weathering" metasediments,  all probably contain substantial amounts of  mineral sulfides.
Because of limited permeabilities, however, the units in most cases will retain the native sulfur unless the
bedrock has been  disturbed (e.g., by quarrying or  mining operations).   All other bedrock types were
assigned a classification of "N", indicating a low potential for supplying sulfur to local surface waters.
Table 7-5  summarizes  the classification scheme.

      These are  2 caveats to the  above  classification  system.  First, we  assigned the classifications
independent of potential weathering rates. Although both the rapidly weathering bedrock (e.g.,  limestone)
and  the more resistant material (e.g., sulfitic  schist) are assigned  the same code,  the more highly
weatherable rock yields a higher flux of sulfur  per unit time.  Second,  we assigned the classifications
based  on data compiled at a state map level. This latter fact causes 2 potential problems.  First, because
                                               262

-------
Table 7-4.  Bedrock Geology Maps Used in the DDRP Internal
Sources of Sulfur Bedrock Geology Analyses
State                  Scale                   Reference
 CT               1:250,000             Rodgers (1985)
 GA               1:500,000             Pickering and Murray (1976)
 MA               1:250,000             Zen (1983)

 ME               1:500,000             Osberg et al. (1985)

 NH               1:250,000             Billings (1980)
 NY               1:250,000             Isachsen and Fisher (1970)
 NC               1:500,000             Brown (1985)
 PA               1:250,000             Miles (1980)

 Rl                1:250,000             Quinn (1971)
 SC               1:250,000             Overstreet and Beli (1965)
 TN               1:250,000             Hardeman (1966)
 VT               1:250,000             Doll et al. (1961)

 WV               1:250,000             Cardweli et al. (1968)
                                    263

-------
Table 7-5.  Potential for Sulfur Contribution by Geologic Type
Sulfur Contribution                         Geologic Type
   Potential
        Y                          Calcareous
                                    Limestones
                                    Dolostones
                                    Sulfitic
                                    Marbles
                                    Carbonaceous
                                    Pyrlte-bearlng

        P                          Black/gray shales
                                    Fossiiiferous
                                    Rusty weathering (schists)

        N                          All other types  (includes sandstones,
                                    conglomerates, most metamorphics,
                                    igneous, etc)
                                     264

-------
of the scales of the state geology maps, local concentrations of sulfide-bearing bodies are frequently not
delineated; therefore, potential local sulfur sources in individual watersheds are  not always Identified.
Second, as a result of correlation difficulties, the location of contacts between contiguous units might not
be depicted accurately on the watersheds. This could result in the mis-identification of the presence (or
absence) of sulfur-bearing units on a particular watershed.

      Using the above classification scheme, we formulated  and tested the hypothesis that watersheds
having large areal percentages of bedrock falling into the "Y" and "P" groups would more  likely have
excess sulfur appearing  in the input/output budgets (Le,, net negative retention of sulfur).  Evaluation
of this hypothesis, however, indicated no significant correlation between net sulfur retention and the group
classification.

      We attribute  the lack  of a  correlation  between these variables to several  factors. First, DDRP
watersheds were selected and  stratified based on lake ANC (Section 5.2). No systems with ANC values
greater  than 400 /ueq L"1 were included in  the northeastern  sample  population, thereby  effectively
eliminating from the sample most watersheds with carbonate-bearing bedrock. In those watersheds with
carbonate-bearing bedrock, the fraction of areal coverage is generally sufficiently small to mask any
internal  contributions to the sulfur  budgets.   As  a result,  because  of  the  restrictions of  our target
population and,  thus, sample, we do not get an evenly balanced sample of the different bedrock types.
Second,  DDRP watersheds acting  as large net sources for sulfur (e.g., 1D1-093,  1E1-123) generally are
associated with major disturbances (e.g., quarrying operations). This observation suggests local  sources
for  the  sulfur and,  thus, information not identified on  state geology maps.   As  noted above,  the
disturbances probably enhance the flux of sulfur from bedrock to the surface  water,  magnifying  the
internal  contributions to the sulfur budgets. Finally, watersheds exhibiting modest excess sulfur fluxes,
but associated with "N"-type bedrock, probably reflect unidentified local  sources of sulfur.  Again,  the
discrepancy  could result from scale problems with the state maps, or could simply reflect local
concentrations of sulfur-bearing minerals.

      In  summary, at the level of  resolution  currently available, bedrock geology does not explain a
significant portion of the  high sulfate outputs found in the sulfur input/output budgets.  Although  In  many
Instances, local  sources  for sulfur are bedrock-related, it  Is not  possible to isolate those sources  using
information compiled for state geology maps.  More detailed investigations (outside of the scope of this
Project) would be required to isolate and identify these sources and resolve the discrepancies.

7.3.3.3  Upper Limit Steady-State Sulfate Concentration

7.3.3.3.1  Introduction -

      The second approach  selected to determine an estimate of the number of systems with internal
sources of sulfur was based on an estimated steady-state sulfate concentration. As discussed previously
(Section  7.1), steady state is obtained when sulfur outputs from a watershed equals inputs.  The sulfate
concentration of the surface water at that point is the steady-state sulfate concentration, and an estimate
can be computed from the inputs and the runoff, as noted in Equation 7-5  below:
                                               265

-------
               [S042']ss     =     (Sw+Sd)/R                                       (Equation 7-5)

where:         [SO42~ ]ss     =     steady-state sulfate - [average annual
                                   concentration (mass  length"3)]
                       Sw   =     wet sulfur deposition (mass length"2 yr"1)
                       Sd   =     dry sulfur deposition  (mass length"2 yr"1 )
                        R   =     runoff depth (length yr"1)

Steady-state sulfate concentration can be computed for any individual site for which we have estimated
inputs and runoff.  If the observed (i.e., by ELS or NSS) sulfate concentration at a site is greater than
the computed steady-state concentration, a source of sulfur internal to the watershed is suspected.

      As discussed  previously (Section 7.3.2.3),  each of the estimates of inputs and  runoff has  an
associated uncertainty. The computed steady-state  sulfate concentration has an  uncertainty that is a
function of these input uncertainties.  Thus, we can compute for each surface water (i.e.,  lake or stream
reach) an  upper limit steady-state  sulfate concentration  that,  if exceeded,  serves to indicate  the
occurrence (with known probability) of an internal source of sulfur.

7.3.3.3.2  Objectives -

      The objectives  of the steady-state sulfate concentration analysis are

      (1)       to apply an uncertainty analysis (of the type presented in Section 7.3.2) to determine an
               estimate of the steady-state sulfate concentration and associated uncertainty, and

      (2)       to calculate an upper limit steady-state sulfate concentration that, if exceeded, indicates
               the presence of internal sources of sulfur.
7.3.3.3.3  Calculation of steady-state sulfate -

7.3.3.3.3.1  Data -

      We used the long-term annual average estimates of wet and dry sulfur (Section 7.3.1.2.1) and the
30-year average annual runoff  (Section 5.7) to calculate estimates of  steady-state sulfate concentration
for each watershed (Equation 7-5).

7.3.3.3.3.2 Monte Carlo analysis —

      Figure 7-4 presents  a flow chart of the  steady-state  sulfate analysis and subsequent use of the
steady-state sulfate concentration to identify internal sources of sulfur.  Briefly, the first step in developing
the upper limit steady-state sulfate concentration is to determine an estimate of the uncertainty associated
with the steady-state calculation. We conducted a Monte Carlo analysis similar to the one discussed in
Section 7.3.2.3 using the parameter uncertainty estimates for Sw , Sd , and  R. We performed Monte
Carlo simulations (10,000 iterations) for 34 watersheds selected at random  from the NE, SBRP, and Mid-
Appalachian  study sites  (NSWS).  The results  of the Monte Carlo analysis provided an estimate of the
                                               266

-------
                                     Run Monte Carlo simulations (10,000)
                                            for random watersheds
                                                    T
                                        Determine regression relationship
                                     between Monte Carlo estimates of mean
                                       [SO 4 ] ss & standard deviation (sd)
                                   Use regression equation to calculate standard
                                     deviation on steady-state calculations for
                                             watersheds (all NSWS)
                                                    I
                                        Compute [SO 4 ] ss upper limit
                                            =  [SO4 ]    + 2sd
                                                      ss
                  Suspected internal sulfur sources;
                  watershed removed from analysis
No internal sulfur sources based
   on steady-state analyses
Figure 7-4.  Flow chart for the determination of internal sources of sulfur using the steady-state
sulfate concentration.
                                             267

-------
standard deviation around the mean steady-state sulfate using the uncertainty estimates for each of the
34 watersheds.

7.3.3.3.3.3  Development and application of the regression equation —

      We then  plotted  the  standard deviation versus  the  calculated  mean steady-state  sulfate
concentration  (based on 10,000 runs) for the 34 watersheds (Figure 7-5) and determined the  regression
equation with the standard deviation as the independent variable.

                Est. Std. Dev.  = - 4.9 + 0.339 mean [SO42" ]ss                        (Equation 7-6)
                             R2 = 0.99, p = 0.0001

We then substituted  the computed  (or nominal) value of steady-state  sulfate concentration for each
watershed into Equation 7-6 to calculate an associated standard deviation applicable to each site (note
that there is an individual estimate for each site). Analyses of the Monte Carlo runs for the 34 watersheds
indicate that a log normal distribution best  describes the uncertainty associated with  steady-state sulfate
concentration. In applying the regression equation to each watershed, we conducted a log transformation
of the prediction procedure to  reflect the observed distribution of the uncertainty in  steady-state sulfate
concentration.

      The final step in the analysis of steady-state internal sources was to apply the calculated standard
deviation on steady-state sulfate concentration to determine an upper limit.  We added twice the estimated
standard deviation (97.5  percent confidence interval) to the computed steady-state sulfate concentration
and then compared the result to the measured sulfate concentration.  If the computed upper limit steady
state  sulfate concentration was equal to, or greater than, the measured sulfate concentration, then we
assumed no significant internal sources of sulfur.  Conversely, if it was less than the measured sulfate
concentration, we strongly suspected that some source of sulfur was contributing to the surface waters
in addition to that estimated from atmospheric deposition.

      This analysis does not work well for  sites  in regions that retain large amounts of deposited sulfur
(e.g.,  the SBRP). In such regions, statistical outlier analyses  (e.g., see Section 8) need to be performed
to identify unusually disturbed or affected sites.

      A summary by region of the number of ELS and NSS watersheds removed from the average annual
percent sulfur retention analysis is given in Table 7-6. These sites were identified using the  upper limit
steady-state sulfate  concentration estimates.  The additional sulfur Is  probably from some  internal
weathering source (as discussed above) or possibly could be due to a very localized emissions  source.

7.3.4 Results and Discussion

      We calculated percent sulfur retention for sites located in the NE,  Mid-Appalachians, SBRP, and
several adjacent regions. Sites identified as having internal sulfur sources through the steady-state sulfate
concentration  analysis were  eliminated (Table 7-6).   Also, 3 SBRP  sites sampled  as part of the Pilot
Stream  Survey were dropped due to  outlier surface  water chemistry (i.e., ANC > 1000 /ueq L"1).
                                               268

-------
   100 n
 "ซ  60
 CD
 Q
 T3

 I
 CO
 CO

 ง  40
   20-
                                100                      200
                                    Mean Sulfate (jieq L'1)
300
Figure 7-5.  Scatter plot of the Monte Carlo calculated  standard  deviation versus the calculated
mean [SO^^ (based on 10,000 runs per watershed) n=34.
                                            269

-------
Table 7-6. Summary of Watersheds (by ELS and NSS
Subregion) Dropped Due to Suspected Internal Sources
of Sulfur Identified by Steady-State Analysis
Region                             # of Watersheds
Eastern Lake Survey
       1A                                3
       1B                                5
       1C                                15
       ID                                12
       1E                                18
       3A (SBRP)                         2

Pilot Stream Survey                        0

National Stream Survey
       1D                                12
       2BN                               9
       2CN                               23
       2X                                7
       3A                                2
       3B                                7
                             270

-------
      We used a  modified version of the long-term  annual  average (LTA) sulfur deposition  in these
analyses.  This modified LTA sulfur deposition does not include the 20 percent increase in dry deposition
discussed in Section 5.6 for the TY and standard LTA deposition data. The TY and standard LTA data
were only available for the primary DDRP study sites (NE=145, SBRP=35).  The sulfur retention analyses
use surface water chemistry from  approximately 1,000 sites (lakes and stream reaches) sampled as part
of the ELS and NSS.  The dry sulfur deposition data provided by AREAL-RTP (see Section 5.6) were the
only internally consistent dry sulfur deposition data for all of the ELS and NSS sites.  These dry sulfur
deposition estimates  were combined with the long-term  wet sulfur deposition  estimates to form the
modified-LTA deposition dataset.  To test the overall effects of not using the 20 percent increase in these
analyses,  we adjusted the modified-LTA  data with a 20 percent increase in  dry sulfur for the  ELS
Northeast sites. This adjustment created a dataset analogous to the TY and standard LTA data.  We then
calculated percent sulfur retention using the adjusted data and compared the results to the unadjustified
modified-LTA sulfur retention results (Figure 7-6).  An inspection of Figure 7-6 indicates that there is  only
a slight shift In the distribution of  percent  sulfur retention between the two datasets.  This slight shift is
unimportant relative to the principal conclusions drawn from these  regional sulfur retention analyses.
Similarly, analyses using the TY dataset yield results very  close to those computed using the modified-
LTA dataset.  Thus, the latter dataset is used for the remainder of the analyses presented in this section.

7.3.4.1  Northeast

      Results from analyses using the modified-LTA deposition data indicate that lake systems in the NE
are generally at or near zero percent net sulfur retention (Table 7-7; Figure 7-7A; Plate 7-1).  Rochelie and
Church (1987) conducted  a sulfur retention analysis using runoff and deposition data for the water year
prior to ELS and NSS sampling and showed similar results. Also, we examined sulfur retention patterns
in the NE for individual  ELS subregions.  Although lakes in Subregion 1B  have the highest percent net
retention,  lakes in all  subregions  are, on average,  very close to zero percent net retention (Table 7-7).

7.3.4.1.1  Evaluation of sulfur retention  mechanisms in NE watersheds -

      Although most  NE watersheds  are  near sulfur  steady  state, a small number of watersheds are
characterized by high apparent sulfur retention.  During development of preliminary sulfur input/output
budgets for lakes  in the northeastern United States, we identified a subset of watersheds for which budget
analyses indicated  significant sulfur retention.   Because analyses of sulfate adsorption at that time
suggested that adsorption was likely to delay sulfur response in NE watersheds for a very limited time,
it was unclear how sulfur was being retained in this subset of watersheds.  In an effort to understand
sulfur retention in these systems  and In an effort to evaluate potential future sulfate increases  at those
sites, we  identified for additional  analysis a group of 45  NE watersheds  having  high computed sulfur
retention.   Soils, vegetation, land  use, depth  to bedrock, and  bedrock geology were mapped on 44  of
the watersheds (permission to map was denied for one) during the fall of 1987 and spring and summer
of 1988 (Figure 7-8).  Watersheds  were mapped by the USDA SCS according to protocols developed for
the original NE DDRP soil survey (Section 5.4),  except that mapping criteria were modified to require
discrete mapping of wetland  areas  2 acres or larger, rather than the 6-acre map unit delineations used
for other soils and for vegetation.  After completion of mapping, soil map units were correlated to the soil
sampling classes defined for the initial DDRP NE  soil survey, except for soils on parts of two watersheds
in Pennsylvania.  Those soils were  correlated to  sample classes defined for Mid-Appalachian soils,  and

                                              271

-------
                -100-90 -60 -70 -60 -50 -40 -30 -20 -10  0 10  20 30 40  50 60 70  60 90 100
                                       Midpoint Percent Sulfur Retention
                                                                                   B
           a
           •8 10-

                -100-90 -60 -70 -60 -50 -40 -30 -20 -10  0 10  20 30 40  50 60 70  80 90 100
                                       Midpoint Percent Sulfur Retention
Figure 7-6.  Comparison of percent sulfur retention calculated using (A) modified-LTA deposition
and (B)  modified-LTA deposition adjusted with a 20 percent increase in dry deposition.
                                                 272

-------
Table 7-7.  Percent Sulfur Retention - Summary Statistics by  Region
Region
NE
ELS Rg1
1A
1B
1C
1D
1E
Mid-App
NSS 2Bn
NSS 2Cn
SBRP
PSS
ELS Rg. 3a
Misc.b
S. App. Pla.
NSS2X
Piedmont
NSS 3A
Mid Atlantic Coastal
NSS 3B .
Poconos/Catskills
NSS 1D
Na

5,828
1099
1285
1190
966
1288

12,580
6,478

2,031
247


4,329

7,199
Plain
9,535

2,724
Mean

-5.0
-11.9
8.9
-3.8
-11.5
-9.4

27.9
-4.0

67.5
68.0


43.4

68.4

30.5

-21.5
Median

-5.2
-13.9
7.6
-7.2
-8.9
-11.7

39.6
3.1

75.4
78.6


50.3

78.0

34.2

-29.2
Std. Dev.

27.6
22.5
25.9
27.3
29.9
26.5

43.1
31.8

23.2
32.8


37.5

24.1

38.4

31.2
Min.

-69.8
-63.5
-65.6
-65.7
-69.8
-61.9

-82.6
-82.7

-54.3
-64.4


-63.6

-9.8

-60.4

-71.2
Max.

73.3
60.6
73.3
63.3
53.9
51.0

90.6
55.2

88.0
92.9


86.0

91.9

92.7

67.1
a Estimated target population calculated using NSWS weights (see Linthurst et al., 1986a; Messer et al,, 1986a;
  Kaufmann et al., 1988) for information on weights.

b Additional regions sampled as part of NSS Phase i  (see Plate 7-1),
                                                    273

-------
                                                  NE
                                   1.0r
                                _g  0.8
                                k.
                                o
                                a.
                                2  0.6
                                a.
                                
                                ฃ  0.4
                                CO
                                O
                                             Upper Bound
                                             Projected
                                             Lower Bound
                                  0.2
                                  0.0
                                   -100     -50      0      50     100
                                        Percent  Sulfur  Retention
                                                  SBRP
                                   1.0
                                .2  0.8
                                o
                                Q.
                                S 0.6
                                o.

                                
-------
Plate 7-1. Sulfur retention and wet sulfate deposition for National Surface Water Survey subregions
in the eastern United States.
                                             275

-------
                        NSWS  SUBREGIONS
                 MEDIAN  %  SULFUR   RETENTION
                AND  WET  SULFATE  DEPOSITION
                                                           2.25
MEDIAN  PERCENT
SULFUR  RET-ENTION





H  20  - 40

Q  40  - 60

H  60  - 80

         100
2.00
Average Annual
Wet Sulfate       f
Deposition (g nf2 yr~')*
                                                       Eostern Lake Survey
                                             -2.25
                                                                Median
                                                       Subregion  % Retention
                                                         1A
                                                         IB
                                                         1C
                                                         ID
                                                         IE
                                              -14
                                               8
                                              -7
                                              -9
                                              -12
                                                  2.00
                                   Notionol Stream Survey

                                            lied inn
                                   Subregion  X Retention
                                                         2Cn
                                                         2Bn
                                                         3B
                                                         n
                                               3
                                              40
                                              34
                                              SO
                                              75
                                              78
                                             "Deposition for 1980 - 1984
                                              (A. Olsenf Personal Communi cai i on)

-------
                                   DDRP  STUDY SITES
                            Supplemental Watershed Mapping
       * Supplemental Mapping Sites

       •*• Study Sites
Figure 7-8.  Supplemental watersheds mapped for special evaluation of sulfur retention.
                                            276

-------
the watersheds were dropped  from the analysis.  We  have used these mapping data to assist in an
analysis of retention in these watersheds.

7.3.4.1.1.1  Approach ~

      Watersheds for this analysis were selected from the NE lakes sampled in Region 1 as part of the
ELS,  using preliminary watershed sulfur input/output budgets developed with 1984 Water Year data.
Criteria for watershed inclusion  were  (1) lake type - limited  to drainage lakes and  reservoirs;  (2)
watershed area -  less  than 3000 ha; and  (3)  watershed  sulfur  budgets  (1984 Water  Year data)
characterized by  one or more of the following:

      •    at least 20 percent sulfur retention;

      *    a 20 /^eq L" or greater difference between lake sulfate and steady-state sulfate concentration;

      ป    lake sulfate concentrations  at or below the tenth  percentiie of sulfate concentrations in the
           respective ELS Subregion.

The budget/concentration criteria were not intended as independent selection criteria; rather, multiple
criteria were defined to ensure inclusion  of lake systems with high apparent absolute and/or relative sulfur
retention.   With few exceptions, watersheds met at  least 2 of the sulfur  budget/concentration  criteria,
and most met all three.

      We assessed watershed sulfur budgets using procedures and  uncertainty estimates as described
in Section 7.3.  Based on uncertainty analyses presented  In Section 7.3, we  determined that retention
should be regarded as significant if computed percent sulfur retention exceeds 37.5 percent.

7.3.4.1.1.2  Results and  Discussion

      Table 7-8 summarizes  sulfur budget status for the 42 NE watersheds considered, and also  lists
computed in-lake sulfur retention and proportions of wet soils on each watershed.  Using the criterion
of 37.5 percent to define significant retention, 27 of the 42 watersheds had significant (positive) sulfur
retention.   If actual  retention  was not significant for any of the 535 lakes in the ELS sample (all regular
ELS Region 1  drainage lakes and reservoirs, excluding DDRP lakes  and lakes with watersheds  > 3000
ha) from which the 42 watersheds of concern were identified, an expected 13 lakes would fall above the
95  percent confidence window of 37.5  percent Assuming that  retention  estimates for each watershed
are independent,  if there were in fact no lakes with significant sulfur  retention  in the sample population,
the probability of observing significant (computed) retention in > 27 watersheds is 0.00057.  From these
results we can  conclude that although  sulfur retention in many of the 42 watersheds is not statistically
significant, a small proportion of watersheds in the NE target population are characterized by significant
positive sulfur retention.

      Indirect evaluation of soils data for the NE virtually rules out the  possibility that significant net sulfate
adsorption is presently occurring in these watersheds. Response times for the 38 NE soil sample classes
are comparable to  NE watershed responses presented in Section 9.2; none of  the sample classes is

                                               277

-------
   Table 7-8.  Summary of Sulfur Retention Status and  of Watershed Variables Contributing to
   Sulfur Retention for 42 Watersheds in the Northeastern United States
LAKE ID

1A1-019
1A1-037
1A2-001
1A2-036
1A2-038
1A2-056
1A2-057
1A3-018
1B1-004
1B1-006
1B1-007
1B1-008
1B1-038
1B2-059
1 B2-069
1B3-003
1B3-013
1B3-029
1B3-068
1C1-046
1C1-069
1C2-055
1C2-061
1 C3-034
1D1-007
1D1-011
1D1-021
1D1-058
1 D2-006
1D2-013
1 D2-087
1 D3-004
1 D3-083
1E1-010
1E1-017
1E1-023
1E1-036
1E1-060
1 E1 -072
1E1-097
1 E2-004
1E2-046
ws area WA:LA
(ha)
94
105
225
574
91
78
232
153
106
124
146
131
501
57
168
105
125
37
120
422
274
295
368
136
165
641
180
464
138
192
148
185
21
217
418
234
148
84
197
49
198
634
Computed using

11.4
14.2
15.2
20.8
4.3
7.1
4.3
9.6
11.8
5.0
12.4
3.5
78.2
4.0
15.5
7.3
13.4
6.6
10.2
17.5
25.9
8.9
11.7
12.3
18.2
14.3
6.8
12.0
20.2
8.6
10.4
15.0
3.0
17.3
44.9
28.2
6.8
8.3
1.2
1.7
11.0
12.8
rtn.
time (yr)
0.24
0.94
0.08
0.11
0.34
1.17
1.31
0.55
0.32
0.59
0.18
2.25
0.02
0.59
0.20
1.81
0.26
0.62
0.22
0.43
0.06
0.99
0.25
0.25
0.13
0.32
0.25
0.10
0.12
0.30
0.14
0.16
0.86
0.35
0.07
0.04
0.18
1.31
6.60
1.36
0.08
0.25
sulfate
lake
80.6
61.1
64.8
84.3
78.3
42.5
84.0
54.0
91.7
104.5
79.5
102.4
62.6
72.3
121.8
118.3
93.1
67.4
73.9
61.2
50.2
65.5
59.1
52.4
93.7
74.3
79.5
69.3
58.8
93.5
67.7
81.6
95.6
33.6
29.1
36.8
40.9
35.1
38.2
36.5
38.4
44.2
(uea L'1)
s-s
98.7
105
95.9
126
106
108
104
119
179
183
187
184
143
179
216
144
149
199
147
100
108
109
75.6
92.2
124
128
98.5
120
128
130
130
124
143
50.4
50.3
53.8
65.9
64.3
65.6
67.3
60.8
90.1
ws sulfur
rtn (%)
18.3
41.9*ฐ
32.4
33.2
25.9
60.6*
19.0
54.7*
48.8*
43.0*
57.4*
44.2*
56.2*
59.6*
43.5*
17.8
37.6*
66.2*
49.7*
39.0*
53.7*
39.8*
21.8
43.2*
24.3
42.0*
19.3
42.1*
53.9*
28.2
47.8*
-34.3
33.0
33.2
42.2*
31.7
38.0*
45.3*
41.7*
45.7*
37.0
51.0*
In-lake
rtn(%)a
4.7
4.9
4.3
4.5
12.0
8.0
13.0
8.1
7.1
14.0
6.9
18.0
0.9
16.0
6.2
11.0
5.4
12.0
6.8
4.4
2.5
8.3
4.5
5.6
4.1
4.5
7.7
5.5
3.7
7.6
7.0
4.9
16.0
3.5
1.4
2.3
8.9
7.0
23.0
22.0
6.4
5.4
adj. ws
rtn (%)b
13.6
37.0
28.1
28.6
13.5
52.6*
6.5
46.5*
41.7*
29.0
50.5*
26.4
55.2*
43.2*
37.3
7.2
32.2
54.7*
42.9*
34.5
51.2*
31.5
17.3
37.6*
20.2
37.5
11.6
36.7
50.2*
20.6
40.9*
29.4
17.3
29.8
40.8*
29.4
29.1
38.3*
18.6
23.7
30.6
45.5*
% Of
watershed
wetland H02 + H03
8.1
15.9
2.3
4.4
4.5
10.7
0.0
0.0
0.0
7.7
2.7
9.9
0.6
7.0
0.0
7.2
3.3
0.0
4.3
9.7
7.9
14.2
13.3
13.6
14.7
10.6
19.2
21.6
17.2
8.8
17.6
2.5
0.0
9.6
8.4
36.5
46.7
24.4
17.0
5.4
90.4
20.0
19.3
10.3
3.9
8.0
4.1
15.4
0.7
4.1
0.0
1.9
4.9
3.6
17.2
7.0
0.0
0.3
0.3
0.9
1.5
4.9
4.2
4.8
9.1
5.8
8.0
5.5
12.0
4.8
12.8
0.6
8.9
1.9
0.0
9.6
6.1
21.7
26.0
16.6
7.9
4.2
12.7
16.2
area
I25
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
38.6
51.4
17.4
34.8
9.8
46.1
3.2
8.4
15.3
0.0
11.1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Equation 7-1.
Asterisks indicate significant retention (<* = 0.05).
                                                   278

-------
projected to be retaining sulfate or to have solution sulfate concentrations less than steady state under
current conditions (based on the historic deposition sequences used In Section 9).  Therefore, it is highly
unlikely that the observed retention on these watersheds can be explained by adsorption.

      Direct estimation of in-Iake sulfur  retention, using the model of Baker et al. (1986b) (Section 7.2),
suggests that  in-Iake processes are a minor  sulfur sink in most NE lake systems.  For many of the 42
watersheds under consideration here, however, the relative importance of in-Iake reduction Is apparently
much higher (Table 7-8).  Computed retention for the 42 lakes ranged from 0.9 to 23.1 percent, exceeding
10 percent for 10 watersheds and 20 percent for 2 watersheds.  If watershed sulfur budgets are adjusted
by computed in-Iake retention,  retention in 12 of 27 watersheds drops  below the 37.5 percent threshold
that is used to define  significant retention.   In other  words, for almost half  of the 27 watersheds with
significant computed sulfur retention, in-Iake reduction has a significant influence  on watershed sulfur
retention status.

      Wetlands and wet soils  might also contribute  significantly  to sulfur retention in many of these
watersheds.   Table 7-8  lists  percentages of watershed areas covered  by wetlands (SCS land  use
classifications) and by soils In sample classes H02 and H03 (wetlands) and I25 (deep, very poorly to
somewhat  poorly drained  aquepts).   These data indicate that  all but 3 of the  15 watersheds with
significant sulfur  retention  (after adjusting for In-Iake  retention) have  at least  10 percent coverage of
wetlands or wet  soils (these  areal proportions are  not additive on watersheds;  much of the area
designated as wetlands on SCS land use maps is also classified in soil  sample classes H02 or H03);
wetland area exceeds  10 percent on 19 of the  42 watersheds with maximum coverage exceeding 90
percent.  Evidence of net sulfur retention In wetlands  Is inferential; actual sulfur retention in  these soils
depends not only on the extent but also the  location of these areas within a  watershed and the fraction
of watershed runoff flowing through them.

      Analyses to date suggest that there is  significant sulfur retention in a small proportion  of NE lake
watersheds.  Evaluation of Level II modelling data (Section  9) also indicates that adsorption is unlikely
to play a significant role in that retention.  The  most likely processes contributing to retention in this
group of watersheds are a combination of In-Iake retention, which is important In those lakes having long
hydrologic retention times (Section 7.2), and reduction in wetlands/wet soils. Unlike adsorption, reduction
In lake sediments and wet soils is a rate-limited, rather than a  capacity-limited process; retention by
reduction mechanisms can therefore continue at  current rates indefinitely because no capacity exists to
be filled or exhausted.  Reduction in lakes provides a permanent sink for sulfur, but the extent of retention
In wetlands and wet soils can  change on an annual or even seasonal basis.  During dry periods, soils
in wetlands and  other anaerobic  areas could reoxidize, resulting in oxidation  of  reduced sulfur and,
potentially, its  release as sulfate.  The role of wetlands and wet soils can consequently shift from that of
sulfate sink to source during dry periods; the potential for long-term  retention  in such systems is thus
dependent on watershed hydrologic conditions.

7.3.4.2  Mid-Appalachians

      The Mid-Appalachian Region does not  present as clear a picture of percent sulfur retention as the
NE (Table 7-7;  Figure 7-7B).   For this study  we  have defined the Mid-Appalachian  Region as  a
combination of NSS Subregions 2Bn and 2Cn (Plate 7-1). Kaufmann et al. (1988) defined these regions
as the Valley and  Ridge and Northern Appalachians, respectively.  We found that percent retention was

                                               279

-------
more evenly distributed with no strong patterns of low or high percent net retention.  In general, for
both subregions, percent net retention is low with average values less than 30 percent.  Subregion 2Cn
has  significantly lower percent retention  estimates on the average than Subregion 2Bn.  Subregion
2Cn  receives higher sulfur deposition than does Subregion 2Bn  (Plate 7-1).  Although Subregion 2Cn
probably has a high incidence of potential acid mine drainage influence, systems identified by Kaufmann
et at. (1988) as having potential internal sulfur sources were also subsequently identified in our steady-
state analysis  (Section 7.3.3.3) and dropped from this analysis and presentation of results.

      The Southern Appalachian Plateau and the Mid-Atlantic Coastal Plain have percent retention on the
average of 30 to 40 percent (Table 7-7; Figure 7-9). In these regions there is a pattern toward higher
net retention,  although a large amount of scatter  in percent  sulfur retention remains.  The Catskills/
Pocono Region has a median net sulfur retention of -21,5 percent.  This region is a transitional area from
the NE, where glaciated  soils predominate, to the Mid-Appalachians, where older and more weathered
soils predominate.

7.3.4.3  Southern Blue Ridge Province

      Median  net sulfur retention for the SBRP  is  approximately  75 percent (Table 7-7;  Figure 7-7C).
Rochelle and Church (1987), working  with sulfur deposition data  from Water Year 1984, found similar
results.  The average percent sulfur retention for the  Piedmont Region (adjacent to the SBRP) is also high
compared to the NE and  Mid-Appalachians (median = 78.0, Table 7-7 and Figure 7-9).

7.3.4.4  Conclusions

      When collectively examined, definite spatial trends in net sulfur retention are evident among the NE,
the Mid-Appalachian Region, and the SBRP.  Percent sulfur retention generally increases from North to
South in the eastern United States (Figure 7-10; Plate 7-1). Plate 7-2 provides an additional view of the
North to South regional patterns  of percent sulfur  retention.  Using the broad major land use resource
area (MLRA) classes (USDA, 1981) to stratify the NSWS study sites, Plate 7-2 indicates again that the
SBRP and adjacent areas are retaining higher amounts of incoming sulfur deposition when compared to
the Mid-Appalachian Region.

      Also, indications are that net retention of sulfur in the NE on the  average is zero or close to zero.
Net  sulfur  retention in the Mid-Appalachian  Region appears to be in  transition between the NE  and
SBRP. A simple analysis  of variance indicates that, on the average, percent net retention is significantly
different among these 3 regions.

      We attribute the spatial patterns in sulfur retention shown here to 2 key factors: (1) soil type, and
(2) sulfur deposition.  Whereas soils of the SBRP are predominately weathered Ultisols and Inceptisols
that  tend to have high sulfate adsorption  capacities, the NE Region is dominated by Spodosols, which
characteristically have low sulfate adsorption  capacities [e.g., see  discussion by Rochelle  et al. (1987)].
Soils of the Mid-Appalachian Region are predominately Inceptisols and Ultisols. Given the current patterns
of wet sulfur deposition (Plate 7-1) and assuming that the Mid-Appalachian region has received elevated
levels of sulfur deposition for a considerable period of time, it is apparent that this region is in transition
                                               280

-------
           Southern Appalachian Plateau
        to
     O  0.8
        0.6
o
CL
O
CL


                                                         *= 0.4
                                                         s.
                                                          3
                                                          E

                                                         O ฐ-2
                                                            0.0
                                                        -100
                                                                   Upper Bound
                                                                   Projected
                                                                   Lower Bound
                                                                 -50      0       50
                                                             Percent Sulfur Retention
                                    100
                 Catskills/Poconos
                                                                     Piedmont
                                         D
     c
     .2
     ฃ3
     o
     Q-
     o
     CL
        1.0r
   0.8
   0.6
     35 0.4
     ซ
     3
     E

     O ฐ-2
       0.0
                         Upper Bound
                         Projected
                         Lower Bound
        -100
                 -50       0       50
              Percent  Sulfur  Retention
                                    100
                                                            1.0
                                                       0.8
o
CL
2 0.6
o.
JSS
3

o
                                                       0.4
                                                            0.0
                                                             -100
                                                                       Upper Bound
                                                                       Projected
                                                                       Lower Bound
                                                                 -50      0       50
                                                             Percent Sulfur Retention
                                    100
Figure 7-9. Population-weighted distributions of projected percent sulfur retention, with upper and
lower bounds for 90 percent confidence intervals, for additional NSS subregions:  (A) Southern
Appalachian Plateau, (B) Mid-Atlantic Coastal Plain, (C) Catskills/Poconos, and (D) Piedmont.
                                               281

-------
 c
 o
 "•E
.a.
 0)
     0.8-
     0.6
 IS   0.4-
 3
 E
 3
 O
     0.2 -I
     0,0
                    Northeast
                    Mid-Appalachians
                    Southern Blue Ridge Province
       -100     -75      -50      -25       0        25       50

                                 Percent Sulfur Retention
75
100
Figure 7-10.  Combination regional population-weighted distributions of projected percent sulfur
retention, with upper and lower bounds for 90 percent confidence intervals, for the Northeast, Mid-
Appalachians, and Southern Blue Ridge Province (taken  from Figure 7-7).  See Plate  7-1 for
delineation of the three regions.
                                          282

-------
Plate 7-2.  Regional percent sulfur retention by major land resource area (MLRA) based on target
populations  (ELS and NSS sites).
                                           283

-------
                  MAJOR   LAND  RESOURCE  AREAS
                  MEDIAN  %  SULFUR  RETENTION
                 AND  WET  SULFATE  DEPOSITION
MEDIAN PERCENT
SULFUR RETENTION
    0 -  20


    20 -  40


    40 -  60


0  60 -  80


H  80 -  100



UAJOR LAND RESOURCE AREAS

R - Northeastern Forage  and
   Forest Region

S - Northern Atlantic Slope
   Diversified Farming  Region

N - East and Central Farming
   and Forest Region

P - South Atlantic and Gulf
   Slope Cash Crops, Forest,
   and Livestock Region
   2-5CK


  2.25
2.00-
                      Average Annual

                      Wei Sulfaie           2.75--

                      Deposiiion (g ra"2 yr"')*  3.00-
                                  3.25
                               3.50s
                                                     2.00
                                                               2-25
                                                                      -2-25
                                                            ULRA     UEDIAN
                                                           REGION  X RETENTII
                                                                    -12

                                                                     25

                                                                     41

                                                                     73
                                               'Deposition for 1980 - 1984

                                                (A. 01 sen, Personal Communicaiion)

-------
toward a situation of lowered percent net sulfur retention and significantly elevated surface water sulfate
concentrations.  We feel that this change  is a  direct consequence of elevated  atmospheric sulfur
deposition. The SBRP is probably undergoing such a transition but with a lag, or "delay", in time.  The
dynamics of transitions in the NE and SBRP are the subject of DDRP analyses in Section 9 and 10 of
this report.  Analyses for the Mid-Appalachian Region will be examined in subsequent DDRP activities.
Relationships among sulfur deposition, edaphic characteristics, and sulfur retention in the NE and SBRP
are examined in Section 8.
                                              284

-------
                                          SECTION 8

                               LEVEL I STATISTICAL ANALYSES

8.1  INTRODUCTION

      The chemistry and  quality of  surface waters in natural settings  are  the  result  of  inputs from
deposition, terrestrial processes, and in-lake or in-stream processes.  In this section we consider the
relationships between subtending surface water chemistry and  inputs from deposition and the physical
and chemical attributes of the catchments.  The scope of these analyses includes the DDRP sample of
northeastern lakes and streams in the Southern Blue Ridge Province (SBRP).  We will not, however,
consider in-lake or in-stream processes explicitly in this analysis because data are  not available for these
processes on a regional basis.

      Level I Analyses are designed primarily to address the first two DDRP objectives (see Section 2.2):
(1) regional description of soil and watershed characteristics and (2) characterization of the relationships
between watershed attributes and surface water chemistry. These Level I  Analyses are of particular
importance to the DDRP because they are designed to corroborate some of the fundamental assumptions
of the DDRP on a regional basis.  Previous research has generally been limited to observations from a
small sample of sites.  To make model-based  regional projections of future surface water chemistry, it
is important to determine whether or not the findings of previous studies on watershed and surface water
chemistry relationships can also be observed on a regional basis.  If they  cannot, other approaches may
need to be taken. At the same time, it is critical to know if we are  overlooking important relationships
that  should be included in the Level II and Level 111 Analyses.

      The principal objective of the analyses in this section is to  determine  which soil  and watershed
characteristics are most strongly related to surface water chemistry.   Some  of the questions we hope
to answer are the following:  Can surface water chemistry be linked to specific watershed  and soil
characteristics?   Are there  controls  on surface water chemistry  that are not yet identified?   Which
deposition and/or watershed factors explain  most of the observed variability  in surface water  sulfate
concentrations?   Do the characteristics of the near-stream  or near-lake areas have a greater Influence
on surface water chemistry than the watershed as a whole?

      We realize that many of the results of these analyses may only provide further evidence to support,
relationships already known to exist. However, because of the quality, consistency,  and extent of the data
used in these analyses, new relationships between watershed characteristics and surface water chemistry
are likely to be identified,  and at the  same time previously observed relationships will be reaffirmed.

8.1.1 Approach

      The approach used in this section is an empirical, statistical evaluation of the relationships between
selected  watershed attributes gathered for the DDRP  sample of watersheds and the chemistry of the
surface water draining these watersheds. The principal dependent variables considered in this analysis
include surface water sulfate  concentrations, percent watershed sulfur retention (% S retention), surface
water acid neutralizing capacity (ANC), the  sum  of surface water concentrations of calcium  and

                                              285

-------
magnesium (Ca plus Mg), and surface water pH.  Although there are a number of other variables that
could be considered,  these  are of primary interest to the DDRP.  With the exception of % S retention,
each of the dependent variables is a direct measure of surface water chemistry. Percent S retention is
computed as the ratio of the difference between watershed sulfur inputs (from deposition) and surface
water sulfur concentrations  to  sulfur inputs (see Section 7).  Percent S retention  is a measure of the
amount of sulfur arriving via deposition that is retained by the watershed. A summary of the dependent
variable data from the northeastern sample of 145 lakes and SBRP sample of 35 streams is presented
in Tables 8-1 and 8-2,  respectively.

      The deposition data used in this section are the "long-term annual  average" (LTA) deposition data.
These data  have annual resolution and  represent atmospheric deposition as of the early to mid-1980s.
The LTA deposition dataset  is described more fully in Section 5.6.3.2 and summary statistics are given
in Tables 8-3 and 8-4 for the 145 northeastern and 35 SBRP sample watersheds, respectively.

      For this analysis we have grouped catchment attributes  into  six groups. The variables in these
groups  serve  as the  independent or explanatory variables.  The  groups  are: (1) derived  hydrologic
variables (Section 8.3), (2) mapped bedrock geology (Section 8.4), (3) land  use and vegetation (Section
8.5), (4) mapped .soils (Section 8.6), (5)  depth to bedrock (Section 8.7), and (6) measured chemical and
physical soil properties (Section 8,9).  Variables in  each of the groups  are thought to have significant
influence on some aspect of surface water chemistry.  We consider deposition with each of the attributes
to identify  the key relationships between the  dependent variables and each attribute.  We include
deposition in each of these analyses  because it is inextricably linked to surface water chemistry.  Failure
to include deposition  would in all likelihood result in inclusion  of surrogate deposition variables in the
regression  models. As a separate analysis we also consider  the relationship between the deposition
variables and surface  water  sulfate concentrations and ANC (Section 8.2).

      Because none of these attribute groups alone  can fully account for the observed variability in the
dependent variables, we also consider them in combination.  In Section 8.8 we combine the deposition
and the mapped variables (groups 1-5), excluding the measured chemical and physical properties; and
then in  Section 8.10 we integrate deposition and all  of the watershed attributes.

8.1.2 Statistical Methods

      In Section 8 there are tables presenting descriptive statistics of the explanatory variables, as well
as tables presenting results of regression analyses.  In each case, the descriptive statistics are population-
weighted, unless otherwise noted. Population weighting provides estimates of the parameters in the target
population,  rather than  estimates for the DDRP sample only.   None of  the regression analyses in this
section  is weighted.   Based on the discussion  in DuMouchel and  Duncan  (1983) and on the  similarity
of the across-strata relationships among the variables, weighted regressions were deemed  unnecessary.

      Additionally,  in the tables of regression results we have included a plus (+) or minus (-) sign to
indicate the direction  of significant  relationships, rather than a numeric  estimate of the  regression
parameter. These statistical analyses should be considered descriptive rather than predictive. Regression
estimates have been excluded to discourage their use in predictive  equations or naive computations of
the relative importance  of the explanatory variables.

                                              286

-------
Table 8-1.  Surface Water Chemistry and Percent Sulfur Retention Summary Statistics
for the Northeastern DDRP Sample of 145  Lake Watersheds
 Variable8
Mean
Std.  Dev.
Min.
Q11
Median  Q3C
Max.
Suifate
% S retent
ANC
Ca+Mg
pH
112.6
-9.7
126.3
223.1
6.9
45.2
41.3
113.6
126.4
0.8
33.8
307.7
-53.3
35.0
4.5
82.4
-23.3
33.3
125.3
6.7
105.5
-6.5
97.3
191.8
7.2
130.8
14.9
213.0
292.6
7.5
303.6
61.1
391.6
560.3
8.0
a Units on sulfate, ANC, and Ca-f Mg are /weq L"1. Sulfur retention is expressed as a percent.  pH is unitless.

b Q1 is the 25th percentile, and Q3 is the 75th percentiie.
                                              287

-------
Table 8-2. Surface Water Chemistry and Percent Sulfur Retention Summary Statistics
for the  DDRP Sample of 35 SBRP Stream Watersheds
 Variable8        Mean       Std Dev.       Min.       Q1b      Median     Q3b    Max.
Sulfate
% S retent.
ANC
Ca+Ma
SOBCT
pH
40.3
65.1
286.8
285.4
371.0
7.1
34.1
26.0
447.9
455.1
466.2
0.41
14.7
-60.5
16.2
46.0
92.8
6.4
19.8
60.1
98.8
85.8
156.0
6.9
23.6
74.9
126.5
117.2
223.4
7.0
42.2
79.1
171.1
189.4
244.7
7.2
178.6
88.6
1710.5
1841.6
1958.5
8.4
8 Units on sulfate, ANC, Ca+Mg, and SOBC are j/eq L"1. Sulfur retention is expressed as a percent. pH is unitless.

b Q1 is the 25th percentiie, and Q3 is the 75th percentile.

0 SOBC = Sum of base cations (Ca+Mg + Na+K)
                                             288

-------
Table 8-3. Summary Statistics for Wet and Dry Deposition on the DDRP Sample
of 145 Northeastern Lake Watersheds (units are jjeq m  )
   Variable
Mean
Std. Dev.  Min.
Median    Q3a
Max.
SO4-WET
SO4-DRY
H-WET
H-DRY
CA+MG-WET
CA+MG-DRY
44900
22800
46800
24600
8200
10600
10300
10100
12000
16300
3600
3300
26700
9300
24500
1600
4800
3000
35200
16000
36000
11000
5800
8500
46100
20100
47300
23900
7300
10100
53800
26100
57800
32300
9200
13300
62300
60400
67300
77400
24100
19500
1 Q1 is the 25th peroantile, and Q3 is the 75th percentile.
                                           289

-------
Table 8-4.  Summary Statistics for Wet and Dry Deposition on the DDRP Sample of 35
SBRP Stream Watersheds (units are jieq m  )
   Variable
Mean    Std. Dev.   MIn.
Median   Q3a
Max.
SO4-WET
SO4-DRY
H-WET
H-DRY
CA+MG-WET
CA+MG-DRY
52400
33000
45700
23300
10600
18000
4600
4300
3900
5400
800
3900
40800
20400
36300
11100
8500
6800
49800
30900
42300
19800
10200
16500
52900
33400
45200
22100
10700
19600
54900
34700
48000
24800
11100
20200
69400
42400
61100
36500
13200
22400
   Q1 is the 25th pereentile, and Q3 is the 75th percentile.
                                         290

-------
      The standard statistical approach used in this section begins with a stepwise regression of the
explanatory variables on the surface water chemistry. This approach  enables us to select explanatory
variables in a  way  that  avoids  serious  problems  with  collinearity.   The stepwise regression was
implemented in SAS (SAS Institute Inc., 1985, 1987), using a value of 0.15 for both the significance level
for entry and the significance level for remaining in the model Mallows' C  statistic was used as a model
selection criterion. Significance levels for the explanatory variables are given in tables in each section.

      The selected model was then run as a standard linear regression  to perform  residual  analyses,
checking for outliers, leverage points, and problems with  standard regression  assumptions (Belsley et
al,, 1980).  Cook's D statistic was used to identify leverage points (Madansky,  1988), i.e., observations
that might  exert an extreme influence on the estimates of the regression parameters.  In addition, the
effect on the regression parameters was assessed using the calculated DFBETAs (Belsley et al., 1980).

      Plots of the studentized residuals were used to check for  outliers, as well as homoseedasticity
(constant variance of the residuals across the range of the dependent  variable).  Specific  instances
where log  transformations of ANC  or  Ca  plus Mg were  necessary to produce homoseedasticity are
discussed in Sections 8.3 and 8.7. If outliers or leverage points were found to be affecting the regression,
the stepwise regression and subsequent residual analyses were performed again without the problematic
observations.

      Specific exceptions to this approach are discussed in the individual sections where the exceptions
occur.  In Section 8,5, the standard statistical approach is applied to rotations of principal components
extracted from the original explanatory variables,  rather than the variables themselves.  In Section 8.8,
Mallow's C statistic could not be  used as  the  model selection criterion in the SBRP,  and Akaike's
information criterion was  used instead.

8.2  RELATIONSHIPS BETWEEN ATMOSPHERIC DEPOSITION AND SURFACE WATER CHEMISTRY

8.2.1  Introduction

      Atmospheric deposition and its effects on surface water chemistry have been extensively studied
for several decades.  Smith and Alexander (1986) found a strong positive correlation between sulfur
emissions and surface water sulfate concentration on a regional  basis.  Neary and Dillon (1988)  found
that sulfur deposition had a positive relationship with lake sulfate  and  a negative relationship with ANC
for a sample of  1168 Canadian lakes.  Sullivan  et al. (1988b) found  significant  correlations between
median lake sulfate concentrations  and wet  sulfate deposition for the National Surface Water  Survey
(NSWS) sites,  in this section we  examine such relationships for the DDRP sample of watersheds using
the wet and dry atmospheric  deposition data for the Project (Section 5.6).

8.2.2  Approach

      Surface water sulfate concentration and ANC are the two primary variables linked to the influence
of sulfur deposition on surface water chemistry,  and hence these two variables are the focus of this
analysis.  For explanatory variables, we used the LTA estimates of wet and dry deposition (discussed
in Section 8.1.2).  In addition to the individual wet and dry deposition estimates, we also used total sulfate

                                               291

-------
deposition  and total, hydrogen deposition.  In each case the total deposition value is the sum of the
appropriate wet and dry deposition values.  The statistical analyses are discussed In Section 8.1.2.

8.2.3  Results and Discussion

8.2.3.1  Northeast

      The statistical analyses show a significant positive relationship between lake sulfate and total sulfate
deposition  (Table 8-5).  Residual analysis of this regression revealed two strong outliers with lake sulfate
levels much higher than predicted.  These sites have quarry pits and will be discussed in Section 8.6.3.1.
Removing  these two sites with apparent internal  sources of sulfur increases the amount of explained
variability to 38 percent (Table 8-5).

      There is a weaker relationship between ANC and deposition  (Table 8-5).  Wet and  dry sulfate
together explain only 18 percent  of the variability in  ANC  in  the northeastern lakes.  Notice that the
parameter  estimates for wet and dry sulfate have opposite signs.  In the stepwise regression  used to
select a model, wet sulfate deposition was selected in  the first step, and then dry sulfate deposition- was
included as the deposition variable with the best relationship to the residuals from the first step.  Residual
analysis indicates  that this is an adjustment in the model to correct for areas  with high deposition and
high ANC,  such as Subregion  1B  (the Poconos/Catskills, see Plate 5-1).  The  size of Rz for ANC is not
surprising,  because ANC  is strongly dependent upon mechanisms of ANC generation within watersheds
(see Section 3).

8.2.3.2 Southern Blue Ridge Province

      In the  SBRP,  sulfate  deposition variables  were  not  significantly  related to  stream  sulfate
concentration (Table 8-6).  Because the stepwise regression used a 0.15 level of significance for entry
into the model, this result indicates that the relationship between deposition and surface water sulfate is
very weak.

      The  only deposition  variable related to ANC was dry  hydrogen deposition, but the parameter
estimate is positive and  is not significant at the 0.05 level (Table 8-6).  The  fact that the relationship is
positive instead of negative suggests  that  dry hydrogen deposition  may be acting as a  surrogate  for
some  other factor.  Dry  hydrogen deposition is  significantly  negatively  correlated  with runoff,  so this
result could represent a dilution effect due  to increased runoff.

8.2.3.3  Summary

      There is a significant relationship between surface water sulfate concentration and deposition in the
NE, but not  in the SBRP.  Nonparametric  statistical analysis shows  that median sulfur retention is not
significantly different from zero in the NE, but is significantly greater than zero in the SBRP.  Rochelle and
Church (1987) support this  conclusion.  Thus, watersheds are approximately at steady state with respect
to sulfur deposition in the NE but  not in the  SBRP, as discussed in  Section 7.3.  Soils in the NE have
little remaining sulfate  adsorption capacity,  so the lake sulfate concentrations  reflect the deposition
gradient (see Section 9.2).  In the SBRP, the watersheds are still retaining sulfur to varying degrees.

                                                292

-------
Table 8-5.  Results of Regressions Relating Surface Water Chemistry to Atmospheric
Deposition  in the Northeast Region (n - 145)
Water
Chemistry
Variable
               Adjusted
                 R2
            Deposition
            Variable
            in Model
               Regression
                 Sign
Signif.'
 Level
Sulfate               0.27             0.27         total sulfate

    second model (omitting two outliers)

                     0.38             0.38         total sulfate
ANC
0.18
0.16
wet sulfate
dry sulfate
 ***
 ***
     = significant at the 0.001 level
                                             293

-------
Table 8-6.  Results of Regressions ReJating Surface Water Chemistry to Atmospheric Deposition
in the Southern Blue Ridge Province (n = 32)
Water
Chemistry
Variable
            Adjusted
R2            R2
                      Deposition
                      Variable
                      in Model
                  Regression  Signif.3
                    Sign      Level
Sulfate
            none selected
ANC
0.10
0.07
dry hydrogen
 deposition
 S = significant at 0.15 level, but not at 0.05 level
                                             294

-------
Watershed processes, e.g., adsorption by soils, are the primary controls on stream chemistry, so a clear
relationship does not exist between deposition and stream concentration.

      In neither region do the deposition estimates explain much of the variability in current ANC. This
observation does not mean that sulfur deposition is unimportant in causing long-term surface water
acidification (i.e.,  loss of ANC), but rather, highlights the important roles that watershed and soil factors
play in determining current surface water ANC. These relationships are explored further in Section 9 and
the remainder of  Section 8.

8.3  DERIVED HYDROLOGIC PARAMETERS

      Hydroiogic processes are important factors affecting the response of surface waters to acidic
deposition (Chen et al.,  1984;  Peters  and Driscoll,  1987).  The  flowpaths  followed by water  moving
through the terrestrial  portion of a  watershed have been  hypothesized as important in controlling the
chemistry of surface waters (Chen et al., 1984; Newton and April, 1982).  Acidic  deposition that rapidly
moves through the watershed system will have limited contact with the soil, resulting in reduced potential
for neutralization. In this part of the Level I Analyses, we test for relationships among mapped hydrologic,
empirically modelled, and physically modelled data and selected surface water chemistry for the DDRP
northeastern lake watersheds and SBRP stream watersheds.  The objectives of these analyses are to
identify watershed characteristics that are related to surface water  chemistry and to infer the influence of
potential flowpaths.

8-3.1  Soil Contact (Oarcv's Law)

8.3.1.1  Introduction

      An  estimate of the annual  flow rate of water moving through the soil and an Index of soil  contact
time were calculated for  each drainage lake watershed in the DDRP sample (n = 136).  Details of the
calculation are presented  in Section 5.7.  Briefly, the estimate of soil-water flow rate and the index of soil
contact are calculated using Darcy's Law.

                       Q = KAS

            where:      Q = lateral soil flow
                       K =  estimate of saturated hydraulic conductivity
                       A =  cross  sectional area of flow
                       S =  hydraulic gradient

The Index of  soil contact is calculated by dividing Q by the average annual runoff (R).   Figure 5-27
outlines the steps involved in the Darcy's Law calculation.  In this  application, we have attempted to use
the Darcy's Law approach to model flow and index of contact time at watershed scales. The resulting
estimates of flow and index of contact are essentially estimates of the theoretical maximum  potential for
runoff to contact soil in a watershed.

      Table 8-7 and Figures 8-1  and 8-2 summarize the results of  the flow rate and index of soil contact
calculations.  The estimated flow rate and index of contact were  less than 0.87  m yr"1 and 1.10 m yr"1,
respectively, for approximately 90 percent of the study watersheds.  Of the remaining 14 watersheds
(approximately 10 percent), 11  are  located in Subregion 1D (see Plate 5-1). This region encompasses
                                               295

-------
Table 8-7.  Estimated Population-Weighted Summary Statistics on the
Darcy's Law Estimates of Flow Rate and the Index of Flow Relative to
Runoff
Variable
Rate (m yr"1 )
Index (yr)
Mean
0.45
0.76
Median
0.09
0.14
Std.
Dev.
2.34
4.32
Min.
0.002
0.003
Max.
18.2
35.8
                                      296

-------
       30 -i
       20 -
    a
    in
    en
       10 -
I
           I


               i

1

            I

        i


                                          1
•

            0  0.05 0.1  0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75  0.8 0.85 0.9 0.95  1


                                             MIDPOINT


                                Estimated flow contact rate, meters yr1
Figure 8-1.  Distribution of estimated contact rate using Darcy's Law calculation.
                                                 297

-------
40 ~]
*an „

o
z
til *3A
FREQUE
{
,
10 -



0 -



i
1
^
1
I
^S
1
1
i
n



ll
Sss
1
t
^
n
1

0.05




A^
1
I
^
1
1

0.1





1
i
^
1
i
i
015






I
^ซซ, ^
iiซ ซ. I
1
m** ฎ** I
02 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
MIDPOINT
Flow rate divided by runoff












Figure 8-2.  Distribution of index of contact (yr) using Darcy's Law calculation.
                                              298

-------
southern New England and is comprised mainly of Massachusetts, Connecticut, and Rhode Island.  These
watersheds have a high proportion of sandy soils that probably resulted in the high flow rate and index
estimates.  These sites  also have low ANC concentrations, however, with 8 of the 11  1D watersheds
having ANC values less than 50 #eq L"1.   The resulting  chemistry is  probably a function of the high
deposition and the limited neutralizing capacity of the sandy soils found on many of the watersheds.

      For the DDRP Level I  Analyses, we have tested for correlations  between the estimated flow rate
and index of contact time and ANC,  sulfate, sulfur retention, pH  and Ca plus Mg on  a regional and
subregional level.  We have excluded eight watersheds with large rate and index values (discussed above)
from the general analysis because these sites represent a special situation in the  NE and resulted in large
outlier estimates.

8.3.1.2 Results and Discussion

      Examining the DDRP northeastern region, we found very little correlation between the calculated
estimates of the Percy's Law flow rate and index of soil contact time and sulfate, percent sulfur retention,
Ca plus Mg, ANC, or pH (R2 ranging from 0.003 to 0.03).  We also looked at correlations between  the
Darcy's Law  calculations and  the  surface water chemistry  variables on a  subregional  level.   The
subregions used were defined as part of the sampling strategy for the NSWS (see  Section 5.7).   We
determined that there was very little correlation at the subregional  level.  Figures 8-3 and 8-4 show
bivariate scatter plots of the  rate versus ANC and the index versus ANC.  Bivariate plots of the rate and
index versus the other  surface water chemistry parameters are similar.  Figures 8-3 and 8-4 indicate a
large amount of scatter in the chemistry relative to the rate and  index values, particularly at the low values
where most of the data  tend to be concentrated.

      Peters and  Murdoch (1985)  observed a strong relationship between the Darcy's Law  index of
hydrologic contact and  surface water chemistry  in the two systems (Woods and Panther Lakes) they
studied as part of  the  Integrated  Lake/Watershed Acidification Study (ILWAS).  Our  results differed
significantly.  One difference between the DDRP  study and ILWAS is the heterogeneity of the systems
studied.  ILWAS  involved two watersheds that had similar physical characteristics such as basin area,
relief, lake area, percent forest cover, lake size, and lake volume (Murdoch et al., 1984).  The major
difference between the two watersheds was  depth to bedrock with the acidic system having very shallow
soils (low index contact; Woods Lake) and the circumneutral system (high index contact; Panther Lake)
having very deep  soils.  As indicated in Section  5.7,  soil  depth is a key factor in the Darcy's Law
calculation.  These two  watersheds probably represent the possible extremes in soil depth. There  are
significant variations in  many of the  above-mentioned characteristics among the  regional  DDRP
watersheds.  As an example, the DDRP lakes  range in size from approximately  40 to 3000 ha.  Another
factor that has been identified as having significant effects on surface water chemistry is sulfur deposition
(NAS,  1984).  Wampler and Olsen (1984) found that wet sulfur deposition varied  in the NE with a general
southwestern to northeastern decreasing gradient.  The DDRP study watersheds are located across the
NE, and thus are subject to a high degree  of  variability in  sulfur loading.   The ILWAS  watersheds,
however, are only a few  kilometers apart and receive very similar sulfur deposition (Murdoch et al., 1984).
                                              299

-------
      400 •
      300 -a
  ,^~" 200-
   cr
   s
   O
      100-fl
      -100'
                                             1 0
                                                                                 20
                                   Contact Rate (meter yr1)
Figure 8-3.  Scatter plot of ANC versus contact rate calculated using Darcy's Law.
                                              300

-------
        400-
           13




        3001!
        200-
     CT
     CD
                B  B
    O
        lOO-ffn
         o -H B


           1
            e
           I
       -100'
                                               10
                                       Index of Contact (yr)
Figure 8-4.  Scatter plot of ANC versus index of soil contact calculated using Darcy's Law.
                                              301

-------
      A second factor affecting the Darcy's Law result is the precision of the data available for the DDRP
study watersheds.  The parameters used  in  calculating the lateral soil flow (Q) were estimated as
watershed averages.  For all three of the major parameters, hydraulic conductivity (K), soil depth (D,
used in estimating cross-sectional area), and slope (S), an area-weighted average was calculated based
on mapping data provided by the DDRP Soil Survey. By calculating areal averages some smoothing of
the data occurred, which might  not have accurately reflected the values of  these variables  where the
main hydrologic activity in the watersheds occurs.  Identifying which soils and depth-to-bedrock classes
are most important in affecting the basin hydrology is difficult without extensive field measurements.

8.3,2   Geomorphic/Hydrotoqic  Parameters

8.3.2.1  Introduction

      A significant amount  of work has attempted to  relate hydrologic characteristics with mapped
watershed geomorphic parameters for forested watersheds (Hewlett  and Hibbert,  1967; Dingman,  1981;
Carlston, 1963; Lull and Sopper, 1966; Vorst and Bell, 1977; Woodruff and Hewlett, 1970).  In general,
most previously reported research is at the event level or covers short time periods (i.e., days or weeks).
In this study we  are using the NSWS  index chemistry value (see Section 5.3, LJnthurst et al., 1986a;
Messer et al., 1986a; Landers et al., 1988); therefore, hydrologic response should be viewed as an annual
representation.  We assume that if a system can be interpreted as a quick response,system based on
geomorphfc/hydrologic information, then the system is, on the average annual basis, a quick response
system.  As discussed in Section 8.3, quick response systems should have  less solhrunoff interaction,
resulting in reduced potential for neutralization of acidic  inputs.

      In  this part of Level  I hydrologic  analyses,  we test  for apparent relationships among mapped
watershed hydrologic and geomorphic parameters that might affect (or be related to) hydrologic response
and selected surface water chemistry variables for 144 lake watersheds in the NE and 32 stream
watersheds  in the SBRP. Three watersheds with  ANC  >  1000 jueq L"1  were not included in SBRP
analyses.  We are testing for correlations between  chemistry and watershed factors on a large regional
scale  in the NE and SBRP  (see Section 8.1.2 for discussion of statistics).  Tables 8-8 (NE)-and 8-9
(SBRP) contain  summary statistics of the  geomorphie/hydrologic  parameters used for this analysis.
Tables 8-10  (NE)  and 8-11 (SBRP) contain variable  names, descriptions, and units.  Detailed information
on database development is included in Section 5.7 (also, see Rochelle et al., in  press-a).

      Because we were specifically interested in the relationships between hydrologic/geomorphic factors
and surface  water chemistry, we chose not to include other independent variables (e.g., soils, deposition)
that could influence or control surface water chemistry.  In particular, for the NE, deposition  explains a
large proportion of variability in some of the surface water chemistry (see Section 8.2).  In  some cases
the removal of deposition as  a  variable might have resulted in some variables  acting as deposition
surrogates.  We will discuss those cases as appropriate.

      We have,   however, also  performed  analyses on northeastern watersheds stratified  by sulfur
deposition (wet plus dry).  In these analyses, we used a simple stratification procedure based on the
distribution of sulfur deposition for our study sites.  We defined four classes based  on the 25th and 75th
                                              302

-------
Table 8-8.  Estimated Population-Weighted Summary Statistics for Northeastern
Geomorphic/Hydrologic Parameters
Variable3
MIN EL
RTO
VOL
RUNOFF
WS_LA
AW
AL
AK
H2O WS
B LEN
B WIDTH
MAX REL
REL RAT
B PERIM
PERIMRAT
TOTSTRM
PERIN
INT
STRMORDER
DDENSITY
PER DD
B SHAPE
ELONG
ROTUND
COMPACT
M PATH?
WM PATH
Mean
319.0
0.7
2.1
64.0
19.8
5.4
0.4
0.5
0.1
2.7
1.6
134.3
0.05
10.2
3.6
3.1
2.3
0.8
2.9
0.6
0.4
1.9
0.9
0,5
1.4
765.2
1701.6
Median
327.7
0.4
0.5
64.0
11.5
3.4
0.2
0.3
0.09
2.5
1.3
103.7
0.05
9.0
2.9
0.9
0.3
0.0
3.0
0.4
0.1
1.8
0.9
0.4
1.3
489.1
1433.6
Minimum
Value
1.5
0.03
0.04
49.1
2.6
0.15
0.02
0.02
0.01
0.3
0.26
10.7
0.003
1.7
0.76
0.0
0.0
0.0
0.0
0.0
0.0
0.2
0.5
0.06
1.1
48.3
59.5
Maximum
Value
791.0
5.7
57.0
77.6
110.1
30.2
4.6
6.4
0.4
9.5
5.4
604.7
0.2
31.4
9.9
32.8
29.3
11.8
4.0
3.2
1.9
5.1
2.4
1.3
3.3
3618.6
8125.4
 See Table 5-37 for variable names, variable descriptions, and units.
                                              303

-------
Table 8-9.  Estimated Population-Weighted Summary Statistics for Southern Blue Ridge
Province Hydrologic/Geomorphic Parameters
Variable3
B LEN
B PERIM
B SHAPE
B WIDTH
COMPACT
TOT DD
ELONG
AUG EL
M PATH
MAX PEL
PEL RAT
ROTUND
RUNOFF
TOTSTRM
STRMORDER
WM PATH
WS AREA
Mean
4.8
13.9
3.0
1.7
1.4
2.6
0.7
831.5
2398.3
539.1
0.1
0.7
82.0
11.3
2.04
2548.0
9.6
Median
4.5
12.5
2.9
1.4
1.4
2.3
0.7
716.3
1951.4
538.0
0.1
0.7
86.3
7.7
2.0
2091.8
7.3
Minimum
Value
1.8
4.9
1.9
0.8
1.1
0.8
0.5
448.8
888.4
132.9
0.02
0.5
38.1
0.0
0.0
888.4
1.5
Maximum
Value
10.8
31.5
5.2
3.5
1.9
5.3
0.8
1409.7
5611.4
1368.6
0.2
1.3
114.3
41.4
4.0
5862.7
30.0
 See Table 5-38 for variable names, variable descriptions, and units.
                                             304

-------
Table 8-10. Mapped and Calculated Geornorphic Parameters Collected for
the Northeastern Study Sites (Same as Table 5-37)
Parameter
        Description
Units
Measured

BJDENT


B_LEN



B_PERIM




AH


INT
L_CENT


L_PERIM

MAX_EL

MIN_EL

PERIN



SUB_BAS(n)


STRMORDER
Drainage basin centroid expressed as
an X,Y coordinate

Length of drainage basin; air-line
distance from basin outlet to farthest
upper point in basin

The length of the line which defines
the surface divide of the drainage
basin

Area of all open water bodies in drainage
basin

Total length of intermittent streams
as defined from USGS topographic maps of
aerial photos

Area of the primary lake

Primary lake centroid expressed as
X,Y coordinates

Perimeter of primary basin lake

Elevation at approx. highest point

Elevation of primary lake

Total perennial stream length as defined
from USGS topographic maps and aerial
photographs

Area of each  subcatchment  in the
drainage basin

Maximum stream order (Morton) of streams
in the watershed (aerial photos used to aid
in reducing coding problems between 7.5-
and 15-minute maps)
 km



 km



 km2


 km



 km2




 km

 m

 m

 km



 km2
                                                              continued
                                             305

-------
Table 8-10.  (Continued)
Parameter
Description
Units
TOTSTRM        Total stream length; combination of
                perennial and intermittent

Ayy             Total watershed area

Calculated

B_SHAPE        Basin shape ratio;
                B_LEN 2/WS_AREA

B_WIDTH        Average basin width;
                WS_AREA/B_LEN

COMPACT       Compactness ratio; ratio of perimeter
                of basin to the perimeter of a circle
                with equal area;
                (PERIM)/(2 X (a- X Aw)'5)

DDENSITY       Drainage density;
                TOTSTRM/WS_AREA

ELONG          Elongation ratio;
                (4 X WS_AREA)/L_BEN

H20_WS         Ratio of open water bodies area to
                total watershed area
                H2O_AREA/WS_AREA

MAX_REL        Maximum relief;
                MAX_ELEV - M1N_ELEV

M_PATH         Estimate of mean flow path

PER_DD         Drainage density calculated from
                perennial streams only;
                PERIN/WS_AREA

PERIMRAT       Ratio of the lake perimeter
                to the watershed perimeter;
                Lake Perimeter/B_PERIM

REL_RAT        Relief ratio;
                (MAX_ELEV-MI N_ELEV)/B_LEN
                                       km
                                       km*
                                       km
                                       m
                                       m
                                                           continued
                                          306

-------
Table 8-10.  (Continued)
Parameter
        Description
Units
ROTUND


WM_PATH


WS_LA


Additional

RT..
Rotundity ratio;
(B_LEN)2/(4 x WS_AREA)

Estimate of weighted  mean flow
path

Ratio of the total watershed area to
the area of the primary lake
Lake retention time

Volume of the primary lake

Average annual  runoff; interpolated
to each site from Krug et al. (in press)
runoff map
m
yr

106m3
                                                                  cm
                                             307

-------
Table 8-11.  Mapped and Calculated Geomorphic Parameters Collected for the
SBRP Study Sites.
Parameter
        Description
Units
Measured

B_CENT         Drainage basin centroid expressed as
                an X,Y coordinate

B_LEN          Length of drainage basin; air-line                   km
                distance from basin outlet to farthest
                upper point In basin

B_PERIM        The length of the line which defines                 km
                the surface divide of the drainage
                basin

MAX_EI_         Elevation at approx. highest point                  m

MIN_EL         Elevation at watershed outlet                       m

SUB_BAS(n)     Area of each subcatchment in the                  km2
                drainage basin

STRMORDER    Maximum stream order (Norton) of streams
                in the watershed  (aerial photos used to aid
                in reducing coding problems between 7.5-
                and 15-minute maps)

TOTSTRM       Total stream  length; perennial                      km

WS AREA       Total watershed area                              km2
Calculated

AVG_EL


B_SHAPE


B WIDTH
Average elevation;
(MAX_ELEV + MIN_ELEV)/2

Basin shape ratio;
BJ.EN 2/WS_AREA

Average basin width;
WS_AREA/B_LEN
m
km
                                                             continued
                                            308

-------
Table 8-11 (Continued)
Parameter
        Description
Units
COMPACT




DDENSITY


ELONG


MAX_REL


M_PATH

REL_RAT


ROTUND


TOT_DD


WM PATH
Compactness ratio; ratio of perimeter
of basin to the  perimeter of a circle
with equal area;
(PERIM)/(2 x (* x Aw)'5)

Drainage density;
TOTSTRM/WS_AREA

Elongation ratio;
(4 x WS_AREA)/B_LEN

Maximum relief;
MAX_ELEV - MIN_ELEV

Estimate of mean flow path

Relief ratio;
(MAX_ELEV-MIN_ELEV)/B_LEN

Rotundity ratio;
(BJ_EN)2/(4  X WS_AREA)

Estimated drainage density based on
crenulations identified on topographic map

Estimate of weighted mean flow path
m
m
                                                               m
Additional

R
Average annual runoff; Interpolated
to each site from Krug et al. (in press)
runoff map
                                                               cm
                                           309

-------
percentiles and the median value of the deposition data (Table 8-12). We did not analyze sulfur retention
based on the stratified watersheds because deposition is a component of retention.

8.3,2.2  Results and Discussion

8.3.2.2.1  Sulfate and Sulfur Retention -

8.3.2.2,1.1  Northeast -

     We found negative  relationships between surface water sulfate concentration and stream order,
runoff, and maximum relief for the NE (Table 8-13). Northeastern watersheds that had low order streams
(first and  second order) were associated with  high suifate concentrations.  Watersheds dominated by
lower order  streams tend  to be headwater systems that are more likely to be dominated by quickflow
runoff.  Quickflow results in less potential for soil-runoff interaction and subsequent neutralization of acidic
inputs.  Also, we found that the watersheds with lower order streams tended to be located at the higher
elevations.  These systems typically receive higher sulfur deposition due to deposrtional patterns in the
NE, particularly Adirondack watersheds.  The combination of high sulfur deposition and reduced potential
for  soil  interaction  due  to  increased  percent quick runoff  leads  to  higher surface water sulfate
concentrations. We  found no significant  relationships for the entire NE between sulfur retention and the
geomorphic/hydrologic parameters.

8.3.2.2.1.2  Northeast - stratified  by sulfur retention -

     The results of statistical analyses between the hydrologie/geomorphic parameters and in-Iake sulfate
are presented in Table 8-14.  Although some individual parameters were significantly related to sulfate for
deposition classes 1  and 4, these were not consistent. None of the parameters appeared as a significant
predictor in  more than one of the deposition classes.

8.3.2.2.1.3  Southern Blue Ridge Province -

     We identified  no  significant  correlations   between  sulfate  or  sulfur  retention  and   the
hydrologie/geomorphic parameters for the SBRP (Table 8-15).  A probable explanation for the lack of
significant correlations is  the relative homogeneity  of the SBRP watersheds in terms  of both sulfur
chemistry data  and the hydrologie/geomorphic parameter values.

8.3.2.2.1.4  Regional comparison ~

     in. the NE,  we identified stream order,  runoff, and maximum relief as significant predictors for
surface  water sulfate concentration. These findings suggest that headwater streams are associated with
high surface water  sulfate concentrations due to a  higher  percentage of quick  runoff.  A  higher
percentage  of quickflow would result in less  soil interaction and, consequently, higher surface water
sulfate.  We found no significant relationships between the hydrologic/geomorphic parameters and sulfate
concentration or sulfur retention in the SBRP.
                                               310

-------
Table 8-12.  Stratification Based on
Sulfur Deposition (Wet and  Dry)
Class                Deposition
                       (g Hi'2)
  1                     <2.46

  2                  2.46 < 3.33

  3                  3.33 < 3.74

  4                      >3.74
                             311

-------
Table 8-13.  Results of Stepwise Regression Relating
Surface Water Chemistry versus Geomorphic/Hydrologic
Parameters for the Entire NEa
ANC
Ca + Mg SO42' pH
B LEN
COMPACT
DDENSITY + +
ELEV
H20 WS
MAXREL - +
PERIN
RT
RUNOFF - -
STRMORDER + +
Adjusted R2 0.1 5b
0.11 0.29 0.20
  ANC, Ca + Mg, and pH: n = 141
  SO4 : n = 142

b Significant at the 0.15 level
                                             312

-------
Table 8-14.  Stepwise Regression Equations for Surface Water Chemistry and
Hydrologic/Geomorphic Parameters Based on Sulfur Deposition Stratification
Class =1 ANC
COMPACT
DDENSITY +
H2O WS
MAXREL
PER DD
PERTMRAT +
RT
RUNOFF
Adjusted R2 0.49
Class =2 ANC
B PERIM +
B~WIDTH +
COMPACT
MAXREL +
STRORDER
WSAREA
Adjusted R2 0.39
Class =3 ANC
ELEV
DDENSITY +
H20 WS
MEAWATH
PER DD
VOL~ +
WSAREA
Adjusted R2 0.68
Class =4 ANC
COMPACT
ELEV
ELONG
REL RAT
RUNOFF
Adjusted R2 0.36
pH Ca + Mg SO42'
+ + +
0.36 0.54 0.31
pH Ca + Mg SO42"
:
0.27
pH Ca + Mg SO42"
+
0.31 0.73
pH Ca + Mg SO42"
-
0.36
                                        313

-------
Table 8-15.  Results of Stepwise Regression  Relating Surface Water Chemistry
and Geomorphic/Hydrologic Parameters for the SBRPa
TOT DD
REL_RAT
RUNOFF
                ANC Log   (Ga  + Mg)  SO42"        pH    Sulfur Retention
Adjusted R2       0.39        0.15       b            0.16
8 ANCX Ca + Mg, and pH: n = 32
 SO4   and sulfur retention: n = 31

b No variables met the 0.15 significance level for entry into the model.
                                             314

-------
8.3.2.2.2  pH, ANC, and (Ca plus Mg) -

8.3.2.2.2.1  Northeast -

      Runoff was the only geomorphic/hydrologlc parameter that was related to Ca plus Mg.  The
     jnship is negative with an increase in runoff
relationship is probably due to a dilution effect.
relationship is negative with an increase in runoff resulting in a decrease in Ca plus Mg (R2 = 0.11).  This
      As was the case with Ca plus Mg, we found  significant relationships between ANC and the
geomorphic/hydrologic parameters (Table 8-13, R2 = 0.15). We found that drainage density and stream
order were positively related with ANC and runoff was negatively related to ANC.  As discussed above,
stream order is probably reflective of the relative position of the watershed (i.e., headwater), with higher
stream order systems tending to have smaller percentage contributions of quick  runoff to total runoff.
The negative relationship with runoff might be due to dilution effects.

      Drainage density, maximum relief, and stream order were positively related to pH (R2 = 0.20). The
positive relationships between pH and stream order and pH and drainage density are probably functions
of relative proportion of quickflow runoff associated with the high stream order systems.  As discussed
above, the low stream order systems tended to be located at high elevations and have a greater potential
for quickflow runoff and high sulfur  deposition.   The high stream order systems we studied in the
northeastern typically are low elevation systems with gentler slopes and  larger watershed areas. These
systems  probably have a greater potential for  soil interaction and subsequent  neutralization of acidic
inputs.   Drainage density  was  relatively low for most  of  the northeastern watersheds since these
watersheds are primarily lake  watersheds. The higher drainage densities are generally found in the lower
elevation areas where stream development is more advanced.

8.3.2.2.2.2 - Northeast - stratified by sulfate deposition class -

      The results of statistical analyses between the hydrologic/geomorphie parameters, stratified  by
sulfate deposition class, and  pH, ANC, and  Ca plus Mg  were presented in Table 8-14.  No consistent
relationships were found between the  hydrologic/geomorphie  parameters and pH or Ca plus Mg. None
of the parameters appeared as a significant  predictor in more than one  of the deposition classes.

      We found a significant positive correlation between ANC and drainage density in deposition classes
1 and 3.  A significant negative correlation between ANC and runoff was found in deposition  classes 1
and 4.  No  other consistent relationships were found.  These findings are  consistent with those for the
entire NE and were discussed more fully in Section 8.3.2.2.2.1.

8.3.2.2.2.3  Southern Blue Ridge Province -

      The results of statistical analyses between the hydrologic/geomorphie parameters and pH, ANC,
and Ca plus Mg were presented in Table 8-15.  A log-transformation of Ca plus Mg was used in this
analysis to make the variance of the residuals constant.  We found no significant relationships between
pH and the hydrologic/geomorphie parameters in the SBRP.
                                              315

-------
      We found relationships between ANC and the hydrologic/geornorphic parameters in the SBRP.
ANC was negatively correlated with runoff and relief ratio and positively correlated with drainage density.
Drainage density was based on crenulations identified on a topographic map.  The negative correlation
between ANC and runoff suggests that  higher runoff results in lower ANC streams.  This relationship
probably reflects a dilution effect.   Relief ratio  was negatively correlated with ANC.  High relief ratio
watersheds tend to be headwater streams with a higher  percentage of quick runoff, which would lead to
less interaction of water with the soil matrix and, hence, lower ANC.  The positive relationship between
ANC and drainage density may also be a function of relative position of the watershed within the region.

      We  also found  limited relationships  between  Ca  plus  Mg and the hydrologic/geornorphic
parameters.  As with ANC,  Ca plus Mg was negatively correlated with runoff.  As discussed previously,
the negative correlation between Ca plus Mg and runoff is probably due to a dilution effect.

8.3.2.2.2.4  Regional comparisons -

      We found similar hydrologic/geornorphic predictors for ANC and Ca plus Mg in the NE and SBRP.
Although we found significant predictors for pH  in the NE, no  significant correlations were found  in the
SBRP. In the NE, stream order and drainage density were positively related to ANC.  Lower stream order
watersheds are primarily headwater systems that have  a high percentage of quickflow  and, therefore,
would tend to have lower ANC.  Similarly, drainage density is a measure of position within the watershed.
Streams with lower drainage densities tended to  be  headwater streams while lower elevation watersheds
tended to have a more developed drainage network.  In the SBRP, ANC was also positively correlated
with drainage density. Additionally, relief ratio was  negatively correlated with ANC.  Similar to drainage
density  and  stream order in  the NE,  relief ratio is probably a function of the relative  position of the
watershed.  Watersheds with high relief ratios tend to be headwater systems and,  therefore, have lower
ANC due to increased quickflow.

      Runoff is a second factor that appeared to influence ANC and Ca plus Mg in both the NE and
SBRP.  Significant negative relationships were found for both ANC  and Ca  plus Mg in both  regions.
These relationships are probably due to the increased  dilution of  stream and lake  chemistry in  areas
where runoff is high.

      Other significant predictors were found in the NE but not in the SBRP. These  predictors included
basin perimeter for ANC,  and drainage  density,  maximum  relief,  and  stream  order for pH.   The
identification of a larger number of predictors in the NE may be a function of either the larger sample
size (141 in the NE vs. 32 in the SBRP)  or the relative homogeneity of the SBRP.

8.3.3  TOPMOPEL Parameters

      The hydrologic model TOPMODEL, which  is based on the variable source area concept, was used
to characterize flow path partitioning of  the DDRP  watersheds. TOPMODEL was chosen because the
model uses readily available topographic and soils information,  and  it  predicts internal states that can be
used to partition streamflow.  A more complete  description of TOPMODEL is given in Section 5.7.2.1.
                                              316

-------
8.3.3.1  Introduction

      TOPMODEL characterizes flowpath partitioning for each watershed by characterizing the spatially
aggregated distribution function of ln(a/KbTanB)  In the NE and ln(a/TanB)  in the SBRP where "a" is the
area drained per unit contour, 'TanB" is the local  slope, "K" is the hydraulic conductivity, and "b" is depth
to bedrock (Beven and Kirkby,  1979; Beven, 1986; Wolock et al., 1989). Details of the calculations are
presented in Section 5.7.2.1.1.3. Values of Hn(a/KbTanB)  and ln(a/TanB) have been correlated with the
likelihood of producing surface runoff.  Surface runoff is defined as saturation-excess ("return") flow rather
than Infiltration-excess f Hortonian") flow. High values of In(a/KbTanB) or In(a/TanB) suggest areas within
a watershed that  are likely to produce surface runoff.  These areas would typically be characterized as
topographically convergent, low transmissivity areas. Conversely, low ln(a/KbTanB) or ln(a/TanB) values
represent areas that have low potential for  surface runoff generation (e.g., well-drained soils draining little
upslope area).  The mean of ln(a/KbTanB) or In(a/TanB) is the critical parameter for  characterizing an
individual watershed (Wolock et al., 1989). In the NE, four watersheds were  deleted from the analysis
due  to a lack of relief as portrayed  in the  1:250,000-scale digital elevation models (DEM), resulting in a
total of 141 study watersheds.  In the SBRP, we eliminated three watersheds with ANC  >  1000 /*eq L"1
from the analyses resulting In a total of 32 watersheds.

      For the NE, values  of ln(a/KbTanB) are summarized in Table 8-16.  Mean ln(a/KbTanB) values
ranged from -3.38 to 3.40 with  a regional  mean of 1.03.  Subregional means were highest in subregion
1B (2.40), followed by Subregions 1E (1.48), 1A  (0.91), 1C (0.77), and 1D (-0.67). For the SBRP, values
of ln(a/TanB) are summarized in Table 8-17.  Mean ln(a/TanB) values  ranged from  7.34 to 8.89 with a
regional  mean  of 7.81.   Within  Level I  Analyses, we  have tested  for  correlations between  mean
!n(a/KbTanB) or ln(a/TanB) values and ANC,  sulfate, sulfur retention,  pH, and  Ca plus  Mg on a regional
scale in the NE and in the SBRP.  We used Spearman's correlation coefficient rather than Pearson's, as
the scatter plots did not suggest a bivariate normal distribution.  Spearman's correlation coefficient does
not require normality.

8.3.3.2  Results and Discussion

8.3.3.2.1  Northeast

      Statistical correlations between ln(a/KbTanB) and surface water  chemistry are given in Table 8-
18.   We  found  no  significant correlations between ln(a/KbTanB) and  sulfur retention or sulfate
concentration.  Noisy but significant positive  relationships were found between values of In(a/KbTanB)
and  ANC (r = 0.28), Ca plus Mg (r  = 0.31), and  pH (r =  0.27).  Scatter plots for these relationships are
shown in  Figures 8-5, 8-6, and 8-7, respectively.  The relationship  between ln(a/KbTanB) and pH  is
particularly noisy (Figure 8-7).

      The positive correlations between values of In(a/KbTanB) and ANC and Ca plus Mg are difficult to
explain.  High watershed mean values of  ln(a/KbTanB) suggest that a larger percentage of storm flow
originates from  quickflow  mechanisms (e.g., return flow),  whereas watersheds with  low values of
In(a/KbTanB) are dominated by subsurface storm flow.   A larger proportion of quickflow should result
in less overall contact of water with the soil matrix and, hence, lower ANC and Ca plus  Mg.
                                               317

-------
Table 8-16.  Population-Weighted Summary Statistics for In(a/KbTanB)
for the Northeast
Region
Mean
Std.
Dev.
Min.
Max.
Northeast                    1.03      1.08       -3.38      3.40
1A                           0.91       0.86       -0.73      3.04
1B                           2.40      2.42        1.34      3.40
1C                           0.77      0.75       -1.20      1.71
1D     .                     -0.67      -0.20       -3.38      1.33
1E                           1.48      1.63       -0.59      3.18
                                        318

-------
Table 8-17.  Population-Weighted Summary Statistics for ln(a/TanB)
for the Southern Blue Ridge Province
Region               Mean      Median     Min.       Max.


SBRP                 7.81       7.74       7.34       8.89
                                       319

-------
Table 8-18.   Spearman's  Correlation Coefficients Between  ln(a/KbTanB) and Surface Water
Chemistry
Region          n          ANC        SO4       S Ret.        pH         Ca plus Mg



NE             141        0.28a        0.05       0.28          0.27*           0.31a
 Significant at p = 0.10
                                            320

-------
   400 -i
   300-
   200-
I
o
   100-
     o -
  -100-
      -4
                                                     B
                                                   i  a
                                                                o
                                                             a   B
   m



   a
                                           %
         a  0

       B
    B    m
        B

   a   e
 ma         a
 0,0      n"a   a

Ls      Q   Q
                                               13  %
                 B ,    B


              B   sa  m


               D
                       -2
                                     ln(a/KbTanB)
Figure 8-5. Scatter plot of ANC versus ln(a/KbTanB).  TOPMODEL was used to calculate values
of In(a/KbTanB).
                                           321

-------
   600 -i
   500 -
CT
01

A

D>
CO
O
300 -
   200 -
   100-
      -4
                                                                 0   Q
                                                a


                                                B
                                       B     Iff    B
                                          B    rani  B D
                                                      Si
               B



               B
                                                OR,   ป
                                                    BB
                        -2
                                      ln(a/KbTanB)
 Figure 8-6. Scatter plot of Ca plus Mg versus ln(a/KbTanB).  TOPMODEL was used to calculate
 values of ln(a/KbTanB).
                                             322

-------
   9 1
   8 -
   7 -
JC
Q.
   5 H
                                                                ua

                                                                 B
    -4

                                                   E
                                                              Eja B
                                          B   B   H         H

                                               D
                      -2
                                    In(ayKbTanB)
 Figure 8-7.  Scatter plot of pH versus ln(a/KbTanB). TOPMODEL was used to calculate values of
 ln(a/KbTanB).
                                            323

-------
      Positive correlations between In(a/KbTanB) and ANC and Ca plus Mg would seem to contradict
the findings reported earlier (e.g., see Section 8,3.2). As discussed  in Section 8.3.2, we found stream
order to be consistent predictor variable for ANC. The relationship between stream order and ANC was
positive, thus lower ANC tended to be associated with lower order streams. These lower order streams
are generally high elevation with small drainage areas and, therefore, have a higher potential for quickflow
runoff, resulting in low ANC. The positive correlation between values of ln(a/KbTanB) and ANC suggests,
however, that more quickly responding systems result in higher ANC,

      One possible explanation for the positive correlations  between values of ln(a/KbTanB) with ANC
and Ca plus Mg is given in Wolock et al. (1989).  In watersheds with  high mean values of In(a/KbTanB)
watersheds, less water passes through the soil matrix  during  high  flows,  as compared to low mean
ln(a/KbTanB) watersheds, which are dominated by subsurface storm flow. Throughout the hydrochemical
history of the catchments,  more water has passed through those with  the low mean ln(a/KbTanB) values
than through  those with high mean ln(a/KbTanB) values, thereby  consuming more of the buffering
capacity of the low mean catchments. If the buffering capacity of all  catchments were initially the same
and finite, then the low  mean catchments should be more depleted  of buffering capacity.  Low mean
catchments should, therefore, have lower ANC.  Given this hydrochemical scenario, and assuming that
ANC represents subsurface flow chemistry, then catchments with high ln(a/KbTanB) values  should have
high ANC.

      Other factors may  explain why ln(a/KbTanB) was not  significantly  related  to surface water
chemistry.  First, there are numerous sources of uncertainty in the calculations of ln(a/KbTanB). Digital
elevation models at 1:250,000 scale were  used to compute values of "a" and "TanB".   The OEMs are
generalized to a large degree when  compared to a watershed mapped at a  scale of  1:24,000, which
tends to become particularly critical on smaller watersheds. Additional uncertainties are  functions of the
errors associated with the DDRP Soli Survey information  (e.g., error in map unit description, aggregation,
depth-to-bedrock estimates).  Second, there are many controls on surface water chemistry that were not
considered within this analysis (e.g., watershed processes, sulfur deposition).  For example,  the physical
and chemical characteristics  of  soils within "low" versus "high"  In(a/KbTanB) areas are undoubtedly
different.  The spatial variability of soils within a catchment, however, were  not considered  within these
analyses.  Finally, because TOPMODEL characterizes the partitioning  of storm flow through the concept
of variable source areas, it may be more suitable as an  event model.  Variable source areas tend to be
active only during storm events and would not  be expected to contribute a significant amount of runoff
during baseflow conditions. Because NSWS surface water chemistry more accurately represents baseflow,
it may be unrealistic to  expect an index  of variable source areas to be correlated with surface water
chemistry.

8.3.3,2.3  Southern Blue  Ridge Province -

      Statistical correlations between ln(a/TanB) and surface water chemistry are  shown in  Table 8-19.
We found no significant correlations between ln(a/TanB) and sulfate, sulfur retention, pH, ANC, or Ca plus
Mg. The possible factors responsible for lack of a significant correlation between ln(a/TanB) and surface
water chemistry are discussed more fully in Section 8.3.3.2.1.
                                              324

-------
Table 8-19.  Pearson's Correlation Coefficients Between ln(a/TanB) and NSS
Pilot Chemistry
Region     n         ANC        SO4       S Ret.        pH        Ca plus Mg


SBRP      32        0.28        -0.15       0.08         -0.07         0.18
                                      325

-------
8.3.3.3  Summary

      In the Level I hydrologic analyses we attempted to relate empirically physically modelled parameters
and  mapped geomorphic/hydrologic  parameters to surface water chemistry.  The objective of these
analyses was to use  indirect measurements of hydrology,  which can be obtained relatively easily, to
describe surface  water  chemistry.  These measurements include estimates of soil contact  based on
Darcy's Law, output parameters from the hydrologic model TOPMODEL, and mapped measurements of
geomorphology and hydrology.

      •    We did not  determine any significant relationships between the  Darcy's Law estimates and
           surface water chemistry. The major factor determining this lack of relationship is the probable
           large error associated with watershed soil depth, hydraulic conductivity, and slope  estimates.

      •    Although  a  limited significant relationship was  identified  between TOPMODEL output and
           surface water chemistry, this result was not necessarily explainable nor consistent with theory.
           One probable explanation for the lack of correlation  is that TOPMODEL is based  on the
           variable source area concept and is more appropriately an event level model.

      •    Relationships between the mapped geomorphic/hydrologic parameters and surface water
           chemistry were identified.  The major variables that were significantly related were runoff,
           stream order, and an estimate of basin shape.

      These findings  suggest that hydrologic/geomorphic characteristics are related to surface water
chemistry, although specific processes cannot be identified.  Although we found little correlation between
Darcy's Law and  TOPMODEL with surface water chemistry, we chose to include these  analyses within
this report for documentation purposes.  Our conclusions neither  confirm nor repudiate the findings of
April and Newton (1985) and  Chen et al. (1984).

8.4  MAPPED BEDROCK GEOLOGY

      A parameter hypothesized  to be  important in controlling the  composition  of  surface waters is
bedrock geology.  Different  lithologies  exhibit different reactivities.   Some,  such  as limestones  or
dolostones, are highly reactive. Waters in contact with these rock types quickly attain  equilibrium with
the carbonate regardless of the  acidity  of the incident deposition.   Other lithologies  are, effectively,
unreactive. For example, quartzites will  modify the composition of incident deposition only slightly. As
a result, waters evolving from quartzite systems tend to  strongly reflect the  composition of the incident
deposition.

      In addition to lithology, a number of other factors  contribute to the extent of interaction between
bedrock and soil and ground waters.  Porosity and permeability of the bedrock, in conjunction  with other
parameters (e.g., hydraulic head),  control water contact times and the rates of infiltration  through ground
water. Longer contact times provide greater opportunity for the water to react with the bedrock, thereby
increasing cation concentrations and ANC. Structural considerations, such as the strike of a bedrock unit
relative  to the aspect of the watershed could influence water infiltration and  contact times as well.
Unfortunately, quantifying these non-lithological characteristics  of bedrock  was not possible from the

                                              326

-------
sources used for this study. As a result, the analyses here focus on bedrock lithology as the variable of
interest for evaluating statistical relationships between bedrock and surface water chemistry.

      The first step in the bedrock analysis was to identify the types of bedrock within each of the DDRP
watersheds. Using an ARC/INFO Geographic Information System (GIS) (see  Section 5.4.1.7), watershed
boundaries were overlaid onto state geology maps and the bedrock units mapped within the boundaries
identified.

      For the  145 watersheds located in the NE region, a total of 136 different mapped  bedrock units
were  identified. The  large  number  of bedrock types relative to the  number  of watersheds results in
insufficient degrees of freedom for a reasonable statistical evaluation of the relationship between individual
bedrock types and surface water chemistry. Therefore, it was necessary to group the different units into
more generic classes in order to perform the analyses. This classification was accomplished in a two-
step process.  The first step was to assign each mapped unit to a generic bedrock type.  Then,  we
assigned a relative reactivity to each of  these generic rock types.

8.4.1   PPRP Bedrock Sensitivity Scale

      A number of studies  have been undertaken to evaluate the relationship between bedrock geology
and surface water  composition (Hendrey et al., 1980;  Rapp et al., 1985; Shilts, 1981). These studies have
been  used on regional scales to help identify areas that are potentially sensitive to the effects of acidic
deposition.

      Hendrey et al. (1980) used a 4-point scale to delineate rocks of different reactivities. Highly reactive
rocks, such as limestones,  dolostones, or highly fossiliferous rocks, were assigned scores of (4). As  the
reactivity of the rocks decreased, the reactivity score was decreased. The reactivity scale of Shilts (1981)
was developed along similar lines,  except that he used a value of (1)  to designate the  most reactive
lithologies. Some structural considerations were implicitly included in these rankings.  For example, marine
shales are cation-rich, but  because of limited permeability and the presence of pyrite, these units were
assigned reactivities of (2)  on the Hendrey et al. scale and (3)  on Shilts' scale. These ranking systems
have  proven useful for Identifying regions potentially  sensitive to acidification.

      Rapp et ai.  (1984)  developed a  10-point scale to evaluate  the  relationships between bedrock
geology  and surface water chemistry  for lakes located in the  Upper Midwest.  On their  scale,  (1)
represents the most reactive bedrock types (limestones, marbles, calcareous tills), while (10) represents
the least reactive units (e.g., quartzites, organic deposits). Significant correlations were found  between  the
amounts of nonreactive bedrock and surface water chemistry in their  study area.

      In  attempting to use the above  scales in  the DDRP Level I analyses, several difficulties were
encountered. The Rapp et al. scale was developed for the Upper Midwest. As such, it does not contain
the range of lithologies encountered in the  DDRP and is not appropriate for use here.  In working with
the other scales, a major problem has been the lack of resolving power for distinguishing  the different
contributions of weathering to the range of compositions observed among lakes. The watershed sample
used  in the DDRP was selected based  on lake water ANC. Watersheds with surface water ANC > 400
    L"1 were excluded from the study, and the majority of systems have surface waters with ANC < 200

                                              327

-------
peq L".   Using the  Hendrey et al. scale,  the  majority of systems  included  in DDRP have bedrock
sensitivity scores  of  (2).   Similar limitations  have been encountered  using  the other scales  in this
evaluation. For this reason, we wanted to develop a sensitivity or reactivity scale that would allow us to
distinguish the relative ANC generating capacities of a group of iithologically different but,  otherwise,
moderately unreactive rock types.

      With this goal in  mind, a 6-point scale was developed with the  intent  of  separating rocks with
slightly different reactivities into different categories.  The top two categories,  (5) and (6), are reserved
for the reactive and  highly reactive lithologies of Hendrey  et al. (1980), corresponding  to their classes
(3) and (4). Within our classes (1) through  (4), we attempted, then, to distinguish rock  types that have
only slightly different reactivities with surface or ground waters.

      Classification of individual mapped bedrock units was accomplished in a two-step process. First,
each mapped  unit was classified according to  a generic  rock  type.  Table 8-20 lists  the rock types
considered. Once this step was completed, a reactivity score was assigned to each of the generic rock
types. Table  8-21 summarizes the reactivities assigned.  These assignments were reviewed by  both
project participants and a limited  number of individuals external to the project.  Consensus was usually,
but not universally, attained for each of the scores. In all cases, project participants made final decisions
concerning the selection of the relative reactivity score. The decisions regarding final reactivity
assignments were  made independent of any  knowledge of the ANC of the surface waters associated with
the specific bedrock units.

8.4.2 Results

      For the DDRP samples In the NE and SBRP, multiple estimates of the aggregated bedrock reactivity
were synthesized.  In the following analyses, the variable Mean is the weighted average of the sensitivity
codes for a watershed, where the weights are the areal proportions  of the watershed  covered  by the
bedrock type.  The variable Max is the maximum,sensitivity code  observed on a watershed. The variable
HSup is the percent of the watershed covered by bedrock with sensitivity codes that are at least 5.0. The
statistical  analyses on the DDRP  data used  standard regression  procedures discussed in Section 8.1.2.

      As can be seen in Table 8-22, the variables Mean and Max do not differ much between subregions
in the NE, or between the NE and the SBRP. The average and maximum  sensitivity codes for watersheds
are highest in  Subregion 1A and lowest in Subregion 1B,  but  it is unlikely that these differences are
significant. HSup.  the percent of  the watershed with sensitivity codes of 5 or 6, has the highest average
in Subregion 1E.  This result means  that more watersheds in this subregion  are classified  as  having
significant percentages of reactive to highly reactive bedrock types.  The data for the  SBRP indicate that
the estimated sensitivity codes are similar to those for the northeastern subregions with lower proportions
of the highly reactive bedrock types.

      The  measurement error analyses  show that there are highly significant  relationships between
bedrock  geology  and surface water chemistry,  particularly ANC  and  sum of base cations.   These
relationships may be masked in analyses performed on the DDRP watersheds,  since measurement error
models cannot be  used with the more detailed geological information available on these watersheds. The
possible masking of  existing relationships should be kept In mind when reviewing these results.

                                               328

-------
Table 8-20.  Tabulation of the Generic Bedrock Types Used to Classify the Mapped Units Identified
on State Map Legends
  Symbol
Description
Symbol
Description
   101          alkali feldspar granite
   IQ2          granite
   I03          quartz porphyry
   I04          granite porphyry
   I05          granophyre
   I06          pegmatite
   I07          aplite
   I08          syenite
   I09          quartz syenite
   110          alkali feldspar syenite
   111          granodiorite
   112          tonalite
   113          monzonite
   114          quartz monzonite
   115          diorite
   116          quartz diorite
   117          alkali feldspar rhyolite
   118          rhyolite
   119          dacite
   I20          obsidian/pumice
   121          diorite porphyry
   I22          andesite
   123          latfte
   124          trachyte
   125          phonolite
   126          gabbro
   127          anorthosite
   128          norite
   129          diabase
   130          basalt
   131          charnockite
   I32          ultramafic(s)

   C23         Organic deposits/peat
   C22         mixed iimestone/dolostone
   C21         dolomltes/dolostones
   C20         limestones
   C19         interfingering Is/elastics
                                   M01
                                   MQ2
                                   M03
                                   M04
                                   M05
                                   M06
                                   MQ7
                                   M08
                                   M09
                                   M10
                                   M11
                                   M12
                                   M13
                                   M14
                                   M15
                                   M16
                                   M17
                                   M18
                                   C01
                                   C02
                                   COS
                                   C04
                                   COS
                                   C06
                                   C07
                                   COS
                                   C09
                                   C10
                                   C11
                                   C12
                                   C13
                                   C14
                                   C15
                                   C16
                                   C17
                                   C18
              mixed metamorphies
              quartzite
              schist
              phyllite
              slate
              gneiss
              granitic gneiss/granofel
              greenstones
              amphibolites
              serpentinites
              chlor/amphib/epid schist
              marble
              sulfldic schist
              calc-silicates
              leucocratic gneisses
              migmatites
              mixed metaclastics
              mixed types
              quartz sandstone
              sulfidic pelite/shale
              chert
              iron formations
              pelite/mud stone
              shale
              argillite
              conglomerate
              sandstone
              arenite/arkose
              graywacke
              siltstone
              mixed elastics
              calcareous shale
              calcareous siltstone
              calcareous sandstone
              calcareous arenite
              calcareous conglomerate
                                              329

-------
Table 8-21.  Tabulation of the Generic Bedrock Types Used to Classify
the Mapped Units Identified on Slate  Map Legends
Reactivity
  Score
Explanation
Generic bedrock types
  (from Table 8-22)
   5


   6
minimally reactive


slightly reactive


modestly reactive


moderately reactive



reactive


highly reactive
C01 C02 C03 C04 C23
M02 M13
I06  I07 I09

COS COG C07 COS C09 C13
M03 M04 M05 M07 M16 M17
101  I02 I03 I04 I05 I08

C10 C11 C12 C14
M01 M06 M11
111  112 117 118 119 I24 131

C15
M08 M09 M14 M15
113  114 115 116 I20 121 I22 I23
I27  I28 I29 I30

C19
I25  I26 I32

C16 C17 C18 C20 C21 C22
M12
                                    330

-------
Table 8-22.  Regional and Subregional Statistics for the Bedrock Sensitivity Code Variables

Entire Region 1
Mean
Max
H5up
Subregion 1A
Mean
Max
H5up
Subregion 1B
Mean
Max
H5up
Subregion 1C
Mean
Max
H5up
Subregion 1D
Mean
Max
HSup
Subregion 1E
Mean
Max
HSup
Entire Region 2
Mean
Max
HSup
Average
2.6
3.3
3.6
3.1
3.7
0.8
2.1
2.4
0.02
2.7
3.8
3.4
2.3
2.9
1.3
2.7
3.5
14.5
(SBRP)
2.2
2.7
0.3
Median
2.5
3.0
0.0
3.0
3.5
0.0
2.0
2.0
0.0
2.8
3.5
0.0
2.3
2.3
0.0
2.0
2.5
0.0
2.0
2.0
0.0
Min.
1.0
1.0
0.0
2.35
2.5
0.0
1.0
1.0
0.0
1.0
1.0
0.0
1.8
2.0
0.0
1.0
1.0
0.0
1.0
1.0
0.0
Max.
6.0
6.0
100.0
4.0
6.0
18.8
3.0
5.0
0.4
3.6
6.0
43.8
3.0
5.0
18.6
6,0
6.0
100.0
' 3.4
6.0
8.2
                                             331

-------
8.4.2.1  Sulfate and  Percent Retention

8.4.2.1.1  Northeast  -

      As discussed in Section 8.2, sulfate deposition appears to be the dominant source of sulfate in
northeastern surface waters.  After sulfate deposition is taken into account, no bedrock geology variable
appears in the model (Table 8-23).  This  observation is not surprising, because the reactivity of a rock
type is not necessarily related to its sulfur-bearing potential.  We would not expect most of the DDRP
bedrock types to act as internal sources for sulfur.  Some of the more reactive rock types,  especially
limestones and dolostones, will release sulfate by the dissolution of gypsum.  However, we anticipate that
"disturbed" systems, e.g., mining operations for coal or base metals,  would serve as the primary internal
sources of sulfur for surface waters.  These operations expose fresh, unweathered sulfide minerals that
can  be oxidized and therefore  have a high potential for contributing to the sulfur and hydrogen  ion
budgets in these systems.

      The stepwise regressions suggest both the  Mean and Max variables exhibit  negative correlations
with percent  sulfur retention. That  is,  watersheds with the  least reactive bedrock types tend to retain
higher percentages of sulfate than do the watersheds containing more reactive bedrock types.  The
reasons for this correlation are not immediately obvious.  One possible explanation might be  that since
the more  reactive  bedrock types may act as  minor internal sources for sulfur, watersheds containing
highly reactive lithologies may have soils that  carry higher  ambient  loads  of sulfur. As a result, these
soils may allow a greater percentage of sulfur  delivered to the watersheds via deposition to pass on to
surface waters.  If this were a small effect in watersheds containing  bedrock with  sensitivity codes of 5
or 6, then it is possible that these internal sources might  not have been large enough to identify in  the
sulfate regression analysis, and  yet be of sufficient magnitude to have a measurable effect on the sulfur
budgets.   An  alternative explanation could be that the more acidic rock types provide  conditions
conducive to the generation  of oxides in soils,  hence increasing the  sulfate adsorption capacities of the
soils.  Details of this  hypothesis will be addressed in the section on soil chemical properties and their
relationship to surface water chemistry (see Section 8.8.4.)

8.4.2.1.2  Southern Blue Ridge Province -

      When we performed the regressions on the 32 SBRP watersheds, the only explanatory variable that
appeared was H5up.  i.e., the percentage of the watershed covered by bedrock with sensitivity codes of
5 or 6 (Table 8-24).  H5up was  positively correlated with surface water sulfate and  negatively  correlated
with percent  sulfur retention.  On analysis of the residuals, however, these effects could be  attributed
primarily to one watershed (2A08808),  which has high stream sulfate and very low  % S retention.  When
this site and watershed 2A07827 (Table 8-24) were excluded from the regression, no regressor variables
were identified as significant.

8.4.2.1.3  Comparison of Regions -

      In both  regions, the  more reactive  bedrock types are associated both with higher surface water
sulfate and with lower percent sulfur retention.   In this regard, the effect of bedrock on sulfur  dynamics
within the watershed appears to be similar across both regions. Although there is no a priori  reason to

                                               332

-------
Table  8-23.   Results of Regressions  of Surface Water Chemistry on Bedrock  Sensitivity Code
Statistics and Deposition Estimates for Northeast
Water
Chemistry
Variable
SuIfate
Percent
Sulfur
Retention
ANC
Ca+Mg
pH

Adjusted
R2 R2
0.3618 0.3573
0.1370 0.1169


0.0558 0.0491
0.0566 0.0432
0.0878 0.0683

Variable
in Model
total suIfate
total suIfate
Mean
Max
H5up
H5up
total H
Max
Mean
H5up
Regression Signif.8
Sign Level
.j. ***
4_ ***
**
+ S
. **
+ *
+ *
+ **
*
+ S
  S = Not significant at 0.05 level
  * = Significant at 0.05 level
  ** = Significant at 0.01 level
  *** = Significant at 0.001 level
                                                 333

-------
Table 8-24.  Results for SBRP of Regressions of Surface Water Chemistry on Bedrock Sensitivity
Code Statistics and Deposition Estimates
Water
Chemistry
Variable
Sulfate
Model 2
Percent
Sulfur
Retention
Model 2
ANC
Ca + Mg
Model 2
Model 3
PH
Adjusted
R2 R2
0.2526 0.2277
none significant
0.2790 0.2550
none significant
0.0859 0.0554
0.2546 0.2297
0.0791 0.0450
none significant
none significant
Variable Regress. Signif.3 Watershed13
in Model Sign Level Removed
H5up + ** 2A08808(B)
2A07827(B)

H5up - ** 2A08808(B)
2A07827(B)

H5up + S
H5up + ** 2A07827(L)
2A07813(O)
2A08808(L)
H5up + S 2A07826(O)
2A07833(L)


  S - not significant at 0.05 level
  * = significant at 0.05 level
  *• = significant at 0.01 level

  (L) = site removed is a leverage point
  (O) = site removed is an outlier
  (B) =site removed is  both a leverage point and an outlier
                                                  334

-------
expect a relationship between the sensitivity scale and sulfur dynamics,  our results do suggest that the
most reactive bedrock types also act as (minor) internal sources for sulfate, which influences the way in
which a watershed will respond to the effects of elevated sulfur deposition.

8.4.2.2  Sum of Base Cations, ANC, and pH

      As discussed in Section 8.1, the sum of calcium and magnesium is used to represent base cations
In these analyses. This representation was necessary because of the non-local sources for sodium (e.g.,
sea salt and road salt) to the surface waters  in many of the study watersheds.

8.4.2.2.1  Northeast -

      The stepwise regression analyses indicate positive relationships between  surface water ANC and
the regressor variable H5up.   The positive relationship with the  percent of watershed covered by high
bedrock sensitivity codes indicates that the bedrock is contributing significant amounts of ANC through
weathering.

      In conjunction with these analyses, the regressions show that surface water pH has statistical
relationships with the variables Max. Mean and H5up.  The positive relationship with  Max suggests that
watersheds with higher  bedrock sensitivity codes have higher pH  values.  The  relationships with  Mean
and H5up may indicate correction factors for particular watersheds with high or low  bedrock sensitivity
codes.

      We find  a  strong positive relationship between  Ca  plus Mg and the sensitivity code for the
watershed. This finding suggests that there is a relationship between the presumed reactivities assigned
to the bedrock  types and the rate of cation supply to  surface waters. The higher reactivity rankings are
associated with higher weathering  rates and,  hence, stronger internal sources for base cations.

      The stepwise regressions indicate that  other variables contribute to the regulation of base cation
concentrations  in surface waters.  In particular,  there is a positive relationship  between surface  water
calcium plus magnesium and  the total hydrogen ion deposition.  If this  correlation has any significance
in terms of ecological processes, then two  explanations can be  offered.   First, the relationship may
indicate possible leaching of base cations from the soil  exchange complex  in excess of the  mass
contributed by  primary mineral weathering. The  alternative explanation,  especially for those watersheds
containing some carbonate bedrock (e.g., limestones), would be that the higher incident acidic deposition
allows for additional dissolution of the  carbonate and  hence contributes to the base  cation budget.

8.4.2.2.2  Southern Blue Ridge Province -

      The stepwise regression for the sum of calcium and magnesium concentration showed a positive
correlation with H5up. as shown in Table 8-24. Residual analysis indicated that watershed 2A08808 was
a strong leverage point.  This site has already been  discussed  as an  internal source  of sulfur.   Upon
removing this site, as well as two other watersheds, from the analysis, the stepwise regression procedure
still selected HSup. This correlation, however, was  no longer significant at the 0.05 level.
                                               335

-------
      The stepwise regression for ANC showed a positive relationship between this variable and H5up.
As in the calcium and magnesium model, the stepwise procedure selected H5up. but it was not significant
at the 0.05  level.  This regression  and the previous one suggest that the higher bedrock sensitivity
numbers are somewhat associated with increased base cation  concentrations and ANC.

      In the SBRP, the analyses do not show any consistent relationship between the bedrock sensitivity
numbers and the pH of the surface waters across the region.  The stepwise regression for pH selected
none  of the deposition or bedrock geology variables, presumably due to the lack of variability in the
deposition gradient across the region.

8.4.2.2.3  Comparison of regions -

      Hydrogen ion deposition appeared to be strongly related to base cation concentration in the NE,
but not in the SBRP. This is probably due to the much smaller deposition gradient in the SBRP. In the
NE, we  observe  an increase in  base cation  export from watersheds with  increasing hydrogen ion
deposition, but in the SBRP, the change in deposition is smaller, so that the change  in base cations is
not significant.

      In both the NE and SBRP, positive relationships were observed between bedrock sensitivity codes,
and ANC and  base cation concentrations.   In both regions, the regressions explained between 5 and 9
percent of the variability in the surface water variables, but due to a larger  sample size, the regressors
for the NE were highly significant.

      The smaller sample size in the SBRP may also explain  why no significant  correlations between
bedrock  lithology and pH were observed there, while such relationships are observed in the NE.

8.4.3   Summary

      Results of the studies of the relationships between the relative reactivities of the different bedrock
types found within the  DDRP watershed population and the associated surface water properties indicate
several  pertinent  factors.  About two-thirds of the variability  associated  with the assignment of the
sensitivity numbers is attributable to measurement error. This means that our data are somewhat "noisy,"
and so relationships may be obscured or  minimized.  Nonetheless, there  are significant  relationships
between  the relative reactivities assigned to watersheds and associated surface water characteristics, in
particular, base cation concentrations and surface water ANC values. These relationships do not appear
to be as strong as we expected.  In addition to measurement error the absence of strong  relationships
might be related to the population of systems being studied.  In  essence, most  of the watersheds
included  in the study are underlain by nonreactive  bedrock types, so many of the differences  observed
in the surface  water chemistry might be more strongly controlled at this level  by factors such as depth
to bedrock or selected soil properties. The multiple regression studies will address  these issues (see
Section 8.8).
                                              336

-------
8.5  MAPPED LAND USE/VEGETATION

8.5.1  Introduction

      The effects of vegetation and land use on  surface water chemistry are both general and  site
specific.  For example, species differences in root density, depth, and morphology affect how nutrients
cycle from the soli to forest vegetation as well as the physical and biological processes that influence soil
water infiltration and percolation.   Both evergreen and deciduous  vegetation strip or scavenge acidic
deposition materials from the atmosphere before they reach ground waters and surface waters. Long-
term effects of acidic deposition can be either beneficial or adverse to the nutrient status of forest soils
and to forest health; the deciding factors are local site nutrient status, ongoing sllvlcultural practices,
present forest species mix, and both the amount and type of atmospheric inputs received at specific
localities (Johnson et al., 1982a, and Johnson et al., 1988).

      Usually, surface waters within forested watersheds have lower turbidity and temperature and  have
lower nutrient loadings than water from agricultural or urban watersheds (Simmons, 1976; Chang et al.,
1983; Comeau and Bellamy, 1986; Morgan and Good, 1988).  One  exception is forest land  subjected to
clear-cut harvesting and/or extensive site preparation (Pritchett and  Fisher, 1987).  The magnitude of
water chsmistry changes within or outside harvested watersheds is dependent upon  clear-cut  intensity
(Tiedmann et al., 1988) and ionic  species (Lawrence and Driscoll,  1988).

      Within  these broad generalizations substantial  site-to-site variation occurs  because of  inherent
natural spatial and temporal variability across the landscape.  For example, in upland headwater forested
watersheds receiving acidic deposition, surface water sulfate can predominate in areas having minimal
vegetation and soil development; however,  surface water concentrations of  bases generally  increase
downstream where interactions of forest  species composition, soil  depth, and geochemical weathering
are greater (Jeffries et al., 1988; Driscoll et al.,  1987). In other situations, riparian zone vegetation reduces
chemical concentrations  in soil water (Schnabel, 1985) and lowers suspended  sediment loads leaving
agricultural watersheds (Cooper et  al., 1986).  Spatially, wetland  position is also important:   wetland
fringes bordering water bodies seem to  be  more effective in  modifying water quality than are upland
wetlands remote from major downstream lakes (Johnston et al.,  1988).  Finally, significant temporal
alterations in stream  water chemistry have been attributed to both beaver activity (Driscoll et al., 1987b;
Naiman et aL, 1986)  and  changing historical or recent land use patterns (Buso et al., 1985; Hunsaker et
al,, 1986b).

      Although northeastern lake and SBRP stream watersheds were primarily undisturbed and forested,
significant amounts of other land  uses were present.  It was also  known that many northeastern lakes
had varying amounts of beaver activity and wetlands.  The main DDRP objective  in mapping land use
and forest vegetation cover types  was to determine  whether any land uses were consistently associated
with specific surface water chemistry variables. Section 8.5 examines those relationships that were found.

8.5.2  Data Sources

      Land use and land cover data for northeastern lakes were  obtained by interpretation of recent
1:12,000 color infrared (CIR)  photography specifically acquired for DDRP (Section 5.4.1.6).  For SBRP

                                              337

-------
watersheds, SCS personnel determined land use from older (late 1970s) alternate black and white and
CIR, quad-sized  National High Altitude  Photography (NHAP) photos (see Section 5.4.2.7),  Forest cover
types were determined during soil  mapping activities (see Sections 5.4.1.3, 5.4.2.3).  All land use, forest
cover, and wetland data were entered  into GIS (see Sections 5.4.1.7, 5.4.2.8) so that information could
be analyzed by percent watershed area or actual hectare area in desired land use classes.  Select data
on acidic  deposition, precipitation, and runoff were also included.

8.5.3 Statistical Methods

      Relationships between water chemistry  variables and many  environmental variables have  been
examined  via normal regression techniques for small (Osborne and Wiley, 1988) and very large (Hunsaker
et al., 1986a) data  sets.   Some  of the problems  with regression approaches are:  selection of an
appropriate and  parsimonious subset  of regressors for the model; multicollinearity  of the  regressor
variables;  peculiar distributions of some of the regressors, particularly when  some variables have many
zero entries; and practical interpretability of results when many regressors appear in any one model.
      For  these reasons, we used principal component analysis or PGA (Johnson and Wichern, 1982; SAS
Institute Inc., 1985, 1987) to analyze mapped land use and vegetation data. For land use data in the NE,
the correlation matrix of the 42 regressor variables in Table 8-25 was used to generate the  principal
components.  Thirteen principal components  had eigenvalues greater  than  one.  These factors (Table
8-26) were retained for  further analysis.  Together these principal components explained 81  percent of
the variability in the correlation matrix.  We used a varimax rotation of the original factors to improve the
interpretability of the factors (Table 8-27).  Then we performed stepwise regressions of the surface water
chemistry  variables on the rotated factors and examined the residuals for leverage points and outliers after
verifying the standard assumptions of regression analyses (see Section 8.1.2).
    1                                                                                    i
      For  SBRP watersheds, initial analysis showed three watersheds with ANC > 1000 peq L  , due to
local carbonate bedrock rich in calcium and magnesium. We excluded these three  watersheds from all
subsequent analyses. A correlation matrix of 39 regressor  variables (Table 8-28) was used to  generate
the principal components.  Eleven principal components had eigenvalues greater than one. These factors
(Table 8-29) were retained for further analyses, because the correlation  matrix was used to generate the
components.  Together,  these principal components explained 93 percent  of  the  variability in  the
correlation matrix.  We  used a varimax rotation of the original  factors  to improve factor interpretability
(Table 8-30).  Finally, we performed regressions of the surface water chemistry variables on the rotated
factors for the SBRP watersheds, after  examining residuals for leverage points and outliers (see Section
8.1.2).

8.5.4 Sulfate and Percent Sulfur Retention

8.5.4.1  Northeast

      Lake sulfate was  positively correlated with deposition (Section 8.2) and watershed development
but negatively correlated with beaver activity, wetland percent, and precipitation and runoff factors (Table
8-31).  The adjusted R2  of  0.50 was the highest  for all  five water chemistry  variables investigated.
Because NE watersheds have low sulfate adsorption capacity and are assumed to be at sulfur steady
                                               338

-------
Table 8-25.  Land Use and Other Environmental Variables Related to Surface Water Chemistry
of Northeastern Lakes
Variable
  Kind
Variable
 Name
     Explanation of Variable Name
Photointerpretation B DAM
B~LODGE
CT
C H
CIBIN
E
EH
G~
G H
H~
H H
IM" H2O
L H
L~
M .
M H
N~
N H
O~DAM
OW H
P ~
P H
UC
UC H
U OAM
ur
III H
UV
UV H
W~
W H
total number of breached beaver dams
total number of beaver lodges
percent area in cropland
area (ha) in cropland
total number of cabins counted
percent area in forest
area (ha) in forest
percent area in pasture
area (ha) in pasture
percent area in horticulture
area (ha) in horticulture
percent impounded water
area (ha) in waste disposal land
percent area in waste disposal land
percent area in cemeteries
area (ha) in cemeteries
percent area in barren land
area (ha) in barren land
total number of old beaver dams
area (ha) in open water
percent area in pits or quarries
area (ha) in pits and quarries
percent area in urban commercial land
area (ha) in urban commercial land
total number of unbreached beaver dams
percent area in urban industrial
area (ha) in urban industrial
percent area in urban residential
area (ha) in urban residential
percent area in wetlands
area (ha) in wetlands
SCS land and forest
  cover type
Other data
CON
HWD
MIX
LV  WET
SCSJDPN

H_D
H W
ELEV
PRECIP

RUNOFF

SO4_W
SO4 D
percent area in conifers
percent area in hardwood forest
percent area in mixed forest
percent open-wet area
percent area in open (non-forest) land

dry H deposition, g m"a
wet H deposition, g  m"2
elevation  in m from USGS topo maps
precipitation in cm from National Climatic
   Data Center, Asheville, NC
mean annual runoff,  in inches
   from Krug et al. (1985) (see Section 5.7.1)
wet sulfate deposition,  g m"2
dry sulfate deposition, g m "a
                                            339

-------
Table 8-23.  Factor Loadings for First 13 Principal Components after Varimax Rotation of
the Correlation Matrix of Land Use and other Environmental Variables for Northeastern
Lakes
Environmental
Variables
L-H
L
P H
UV H
UC H
CABIN
H H
P
H D
H W
SO4 W
S04 D
ELEV
CON
U DAM
O DAM
IM H20
B LODGE
B DAM
SCS OPN
G
C H
C
G H
LV WET
W
N
UC
N H
OW H
E rf
W~H
M H
M~"
UV
E
Ul H
Ul
ELS WET
H
SCS WET
PRECIP
RUNOFF
MIX
HWD
1
98*
98*
94*
92*
81*
75*
41*
38*
-5
6
5
5
-3
-7
-4
-1
-1
-2
-1
10
•4
6
-1
3
2
2
0
12
-3
4
14
20
0
-1
17
-17
0
0
_2
23
-2
5
-4
2
0
2
-3
-3
0
9
6
10
-31*
4
88*
88*
86*
81*
61*
-67*
-1
-5
-8
-8
6
5
13
-6
5
2
4
3
7
9
1
-14
-10
-14
4
4
8
-10
-4
11
.4
-29*
10
27*
-14
-22
61*
Principal Components*
345
-1
-1
0
-2
4
-2
24
1
-1
-3
-8
-9
-7
-5
89*
88*
87*
78*
75*
-7
-9
41*
4
5
2
2
0
-6
37*
10
63*
46*
-1
-1
-8
9
-2
-3
-2
2
5
5
-1
17
-7
0
0
-1
4
14
4
20
-3
17
1
-2
23
-8
0
-4
2
9
3
-2
89*
86*
73*
73*
71*
0
1
0
9
-9
0
2
8
0
0
0
-58*
-1
0
-6
15
4
-18
-15
-8
-27
3
3
-4
3
-2
6
-1
-13
4
6
10
11
-17
15
2
-3
7
3
-5
4
5
-8
-6
1
96*
96*
-5
-6
5
-7
-1
34*
-2
-4
-9
-28*
6
-4
37*
-9
37*
11
-12
-9
-10
6
1
1
-4
-1
46*
-3
0
-8
4
6
8
6
-5
-2
8
9
-1
4
-8
9
1
-3
-1
-4
_4
.4
97*
93*
61*
1
-1
1
-1
-1
-3
-10
3
0
-1
0
-5
9
-5
-1
0
7
-1
-1
-9
13
7
37*
25*
-18
0
-10
-20
-2
-16
-4
16
-8
12
37*
-9
3
4
-3
-18
39*
-1
-1
-2
.4
10
80*
68*
61*
0
-1
-4
3
19
-6
9
3
6
-1
13
5
-2
                                                          continued
                                            340

-------
Table 8-26.  (Continued)
Environmental
Variables
L-H
L
P H
UV H
UC H
CABIN
H H
P
H D
H~W
SO4 W
S04 D
ELEV
CON
U DAM
O DAM
IM H2
B LODGE
B DAM
SCS OPN
G
C H
C
G H
LV WET
W
N
UC
N H
OW H
E H
W H
M H
M
UV
E
U! H
Ul
ELS WET
H
SCS WET
PRECIP
RUNOFF
MIX
HWD
8
0
0
-1
-1
1
1
3
-2
4
7
1
3
-11
-4
0
-1
1
-2
-2
2
3
-4
-6
8
.4
.4
-1
-1
0
-5
7
1
99*
99*
-4
2
-1
-2
13
3
-7
0
-6
4
-1
Principal Components3
9 10 11 12
-5
-5
12
23
-10
27*
4
35*
-2
-1
16
9
-39*
-13
-1
3
0
-1
-11
16
11
-8
-9
-1
-1
-1
-2
-5
14
-3
-10
-2
-2
-1
87*
-67*
2
-3
-10
27*
9
19
-27*
-4
5
-3
-3
11
-2
-4
-5
14
27*
3
4
11
9
-15
5
-7
-1
-2
0
0
12
-6
-1
-1
-8
1
1
1
0
4
4
7
15
-2
-1
-6
-2
86*
86*
-6
5
-2
12
-13
2
-9
-2
-2
-2
7
3
18
38*
0
-4
-9
-6
-5
-13
-25*
-12
6
-1
-4
14
6
-18
20
11
-8
14
14
0
6
-19
8
2
14
3
3
5
-15
0
-2
63*
62*
59*
12
12
10
0
1
1
8
-7
0
-14
21
16
-13
35*
27*
-13
23
34*
5
-7
6
-1
-2
-6
-10
3
-6
-10
_1
-1
2
2
-3
5
3
8
-2
-2
-5
5
8
-8
-1
19
10
74*
72*
-8
-17
13
1
1
1
2
-5
1
-20
1
-16
~*Q
-7
-11
-3
-14
15
0
-3
15
-2
5
10
-21
-6
6
-3
-3
-3
-8
20
2
2
4
2
3
-6
_-|
3
1
-5
10
13
-2
0
89*
-62*
 ' Printed values are multiplied by 100 and rounded to the nearest integer. Values
  greater than 0.29 have been flagged by an asterisk.
                                                      341

-------
Table 8-27.  Interpretation of the First 13 Principal Components After Varimax
Rotation of the Correlation Matrix of Land Use and Other Environmental
Variables for Northeastern Lakes
Principal
Component
  Rank
General Interpretation of Principal Component
PC1

PC2
PC3
PC4
PCS
PC6
PC7
PCS
PC9
PC10
PC11
PC12
PC13
developed land: waste disposal, pits and quarries,
cabins, urban residential, and urban commercial
overall wet and dry deposition
beaver activity, wetlands, and cropland
pasture land and cropland; less forest
wetlands
barren and urban commercial land
open water, forest, and wetlands
cemeteries
cabins, urban residential, and pits and quarries; less forest
urban industrial land and pits and quarries
wetlands  and horticulture
precipitation  and runoff
more  mixed and  less hardwood forest
                                    342

-------
Table 8-28. Land Use and Other Environmental Variables Related to Surface Water Chemistry
of Southern Blue Ridge Province Streams
Variable
  Kind
Variable
 Name
Explanation of Variable Name
SCS
  interpretations
Forest cover type
Other data
C
C_H
E
E H
F
F-H
G
G_H
H
H_H
L
L H
K~
K_H
M
M_H
N
N H
O
O_H
R
R_H
U
U H
W~H
Z~
Z_H

CON
HWD
MIX
OPEN

CAMG D
CAMG~W
H_D  ~
H W
PRECIP

RUNOFF

SO4_W
SO4 D
 percent area in cropland
 area (ha) in cropland
 percent area in grazed forest land
 area (ha) in grazed forest land
 percent ungrazed forest land
 area (ha) in ungrazed forest land
 percent area in managed  or native pasture
 area (ha) in managed or native pasture
 percent area in horticulture
 area (ha) in horticulture
 percent area in waste disposal
 area (ha) in waste disposal
 percent area in rock outcrop
 area (ha) in rock outcrop
 percent area in cemeteries
 area (ha) in cemeteries
 percent area in pits and quarries
 area (ha) in pits and quarries
 percent area in miscellaneous land use
 area (ha) in miscellaneous use
 percent area in wetlands
 area (ha) in wetlands
 percent area in urban land
 area (ha) in urban land
 area (ha) in open water
 percent area in ridge top  barren land
 area (ha) in ridge top barren land

 percent area in conifers
 percent area in hardwood forest
 percent area in mixed  forest
 percent (dry) areas without forest or wetlands

 dry CA +  Mg deposition,  g  m"s
 wet Ca +  Mg deposition,  g  m"2
 dry H deposition, g m"2
 wet H deposition, g  m"2
 precipitation in cm from National Climatic
   Data Center, Asheville,  NC
 mean annual runoff,  in inches
 from Krug et al. (1985) (See Section 5.7,1)
 wet suifate deposition, g m"2
 dry suifate deposition,  g m"2
                                              343

-------
Table 8-29.  Composition of First 11  Principal Component Analysis (PCA) Factors After
Varimax Rotation of the Correlation Matrix of Land Use and Other  Environmental Variables
Related to Surface Water Chemistry of Southern Blue Ridge Province Streams
Environmental
Variables
1
HWET
S04WET
K
K H
CAMGWET
W H
PPT
L
M H
L H
M
H
H H
OPEN
C
C H
F
U H
R
R H
U
MIX
CAMGDRY
HWD
CON
S04DRY
HDRY
O
O H
G~H
G
RUNOFF
Z H
Z
F H
N
N H
E
E H
89*
88*
88*
87*
85*
72*
69*
9
9
9
9
-2
-1
-3
7
7
4
1
-8
-8
-6
-9
31*
6
13
-8
-5
-12
-10
-21
-12
41*
-3
-3
26
2
2
-7
-13
2
22
21
-14
-14
18
31*
10
99*
99*
99*
99*
-7
-8
18
64*
66*
-17
-4
-1
-1
.4
-9
9
-1
2
2
0
-1
0
8
17
9
.1
-1
-11
-2
-2
.4
-4
3
6
5
-4
-4
1
-2
-5
4
4
4
4
98*
98*
80*
74*
70*
-81*
9
-1
-1
16
-5
-7
-31*
-8
-1
6
-4
-3
10
14
-18
-4
-4
-17
-3
-3
-1
0
4
-10
-10
0
0
-2
5
-17
-2
-2
-2
-2
-3
-4
33*
-2
-2
-33*
98*
98*
98*
95*
-18
-28
2
-4
-17
14
-2
-2
19
15
-23
-2
-2
-4
-2
-2
-6
-8
5
13
18
-28
-27
17
-14
4
0
0
0
0
5
3
8
2
-1
-7
-9
-6
-6
-11
89*
60*
-88*
9
58*
-17
15
14
-23
14
-11
4
4
-44*
6
6
0
-14
PCA Factors
678
9
9
-15
-13
17
39*
44*
0
0
0
0
-7
-8
-14
12
19
10
-1
-7
-7
-1
3
51*
-19
84*
-59*
-83*
3
2
-5
-14
25
-3
-3
16
2
2
-11
-15
-12
-17
14
15
-19
4
-28
-1
-1
-1
_-j
0
0
-8
-4
-5
7
0
-1
-1
-2
20
3
-14
-3
-2
-15
95*
94*
-7
-23
-23
-4
-4
49*
-3
-3
-5
-1
-21
-22
5
7
-26
18
-31*
5
5
5
5
-4
0
38*
-4
-4
-39*
11
12
12
5
-8
-14
-8
-8
5
13
-9
-8
87*
81*
-50*
-7
-7
7
-4
-4
2
6
9
6
3
-1
-1
-4
-5
-4
-1
-1
-1
-1
0
0
,-9
-3
-3
8
-1
0
0
-3
4
12
4
-13
7
-14
-4
-4
-4
-13
4
98*
98*
49*
• -2
-2
-4
.4
10
5
8
-5
-5
11
-9
2
-1
-1
-1
-1
-1
-1
.4
-3
-4
3
-2
0
0
-3
3
13
-2
3
9
-3
-3
-2
-3
-9
-7
ป2
-2
-8
99*
99*
-4
-2
11
.4
-4
-6
-5
-3
-4
-10
-2
-2
-2
-2
-2
-4
1
1
2
-8
-4
-3
-3
-6
12
-9
11
1
20
27
.4
-2
-2
9
-8
-3
-3
-9
-3
-3
96*
94*
  Printed values are multiplied by 100 and rounded to the nearest integer. Values
  greater than 0.29 have been flagged by an asterisk.
                                             344

-------
Table 8-30.  Interpretation of the First 11  Principal Components after Varimax
Rotation of the Correlation Matrix of Land Use and Other Environmental
Variables for Southern Blue Ridge Province Streams
Principal
Component
Rank
General Interpretation of Principal Component
PC1
PC2
PCS
PC4
PCS
PC6
PC7
PCS
PC9
PC10
PC11
overall wet and dry deposition and precipitation
cemeteries, cropland, and waste disposal land
cropland, horticulture, and open land
open land, urban lands, and wetlands
mixed forest and dry Ca plus Mg and SO4 deposition
open water and dry Ca plus Mg deposition
miscellaneous and ungrazed forest land
open land and pasture; less precipitation
ridge top barren land and ungrazed forest
pits and quarries
grazed forest land
                                     345

-------
Table 8-31.  Results of Regressions Relating Surface Water Chemistry of Northeastern Lakes to
Land Use and Other Environmental Data*
Water
Chemistry
Variable
ANC

Ca + Mg

pH




Factor
R2 No.
0.37 4
12
0.42 4
12
0.32 2
4
7
11

Regr. Signif.b
Sign Level
.(. ***
***
+ ***
***
**
_j_ ***
+ *
+ *


Factor Explanation
agriculture: SCS open dry, G, C
precipitation and runoff
agriculture: see above
precipitation/runoff
deposition
agriculture: see above
open water and wetlands
wetlands and horticulture: H, ELS wet,
SCS_wet


Suifate
             0.50
12

 2
 3
 4
 5
 7
 9
12
                                             ***
                                             **
                                             *
                                             ***

                                             s
precipitation and  runoff

deposition
beaver activity, water, wetlands
agriculture: see above
wetlands: SCS, LV, ELS
open water and wetlands
development: cabins, P,  Uv, H; less forest
precipitation/runoff
Percent
Sulfur
Retention
0.19 5
+ *** wetlands (see above)
  **

  ***
= 143
= Significant at 0.15 level
= Significant at 0.05 level
= Significant at 0.01 level
= Significant at 0.001 level
                                               346

-------
state (Section 8.2), a positive correlation between surface water sulfate and sulfate deposition is expected.
increased sewage and animal or chemical waste loadings to streams from agricultural and residential
development also lead to greater overall surface water sulfate levels.  Although sulfate deposition was
associated with  surface water sulfate levels,  amounts  in surface waters  were less  in small  watersheds
when beaver activity and wetland percentage were high. Low downstream sulfate concentrations, caused
by  increased anaerobic  conditions  and  suifate retention behind  beaver  impoundments  has been
documented by others (Driscoll  et al., 1987; Goldstein et al., 1987),  especially during low-flow summer
months.

      Percent sulfur retention was positively related to wetland percent (Table 8-31). Anaerobic wetland
conditions favor suifate reduction processes that in turn foster increased sulfur retention.

8.5.4.2  Southern Blue  Ridge Province

      Singular land uses acted as either leverage points or outliers  and  influenced regressions relating
land use to both sulfate and percent sulfur  retention.   For sulfate,  important land uses were  pits  and
quarries, open land, and pasture.   Eliminating watersheds with  these land uses left no significant land
use factors in regression models (Table 8-32).  Evidently, the watershed with pits and quarries land use
had  an internal source of sulfur.   Agricultural  practices on  open land  and pasture, including  soil
amendments and animal husbandry, may result in increased sulfate  loadings.

8.5.4.3  Regional Comparisons

      In the  NE, sulfate is  strongly and positively correlated with deposition and agricultural and urban
development  Since soils  in the region have little remaining  suifate adsorption capacity  (Section  7,
Rochelle et al., 1989), incoming  sulfur deposition or within-watershed generated sulfur quickly circulates
to surface waters after storm events.  Exceptions are small watersheds with beaver activity or wetlands.
In these watersheds, sulfate reduction  processes are  the probable  cause of decreased surface water
sulfate concentrations and  increased  percent sulfur retention.

      In the SBRP region, surface water sulfate and percent sulfur retention are both influenced by local
internal sulfur sources from pits and  quarries and pasture land. However, when such watersheds are
eliminated from  regression models, the  homogeneous  nature of the  region  stands out:  upland forested
watersheds  with little agricultural or urban development.  Under such conditions, land use is unrelated
to either surface water sulfate or percent sulfur retention.  Instead, both variables are more controlled by
high sulfate adsorption capacity of  soils in the region (Section 7).

8.5.5  ANC. Ca plus Mo.  and pH

8.5.5.1  Northeast

      Lake  ANC was  positively correlated with agricultural land use  and  negatively correlated  with
precipitation/runoff (Table  8-31); both factors in  the regression  were highly significant (p < 0.001).   In
watersheds having a greater percentage of agricultural, urban, or other disturbed land (Buso et al., 1985),
ANC values of surface waters are generally higher than those found  in mostly-forested, small-headwater

                                               347

-------
Table 8-32,   Results  of Regressions Relating Sulfate and Percent  Sulfur Retention  of Southern
Blue Ridge Province Streams to Land Use Data
Dependent
Variable
Sulfate


Percent
Sulfur
Retention
Regress.
R2 n Sign
0.78 32 +
+
0.80 31 -f
30
0.76 32
30
Significant Factor/
Land Uses Included
10/pits & quarries
8/open land & pasture
10/pits & quarries
no significant factors
10/pits & quarries
no significant factors
Significance
Level3
***
S
***
—
***
Watersheds
Removedb
2A07813(L)
2A08808(L)
—
2A08808(L)
2A07823(O)
 : S = Significant at 0.15 level
  *** = Significant at 0.001 level

   (L) = Leverage point removed from regression
   (O) = Outlier point removed from regression
                                                 348

-------
watershed systems (Hunsaker et al., 1986a; Jeffries et al., 1988). Where precipitation and runoff are high,
ANC in surface waters is reduced  because of dilution effects.

      Lake Ca plus  Mg  was positively correlated with agricultural land use but negatively  related to
precipitation and runoff (Table 8-31); both factors in the regression were highly significant (p < 0.001).
Successful farming and related activities are generally located on deeper and  higher base status soils
unless low soli pH and poor fertility are offset by applying lime and fertilizers (Tisdale and Nelson, 1975).
Where acidic deposition is high, soil bases can be leached from the soil and replaced by hydrogen and
aluminum ions (Section 3); bases leached from the soil are flushed rather quickly from lakes associated
with high runoff.  The  positive correlation of Ca plus Mg with agriculture (Comeau and Bellamy, 1986)
but negative correlation with greater precipitation and runoff is indicative of these relationships.

      Surface water pH was positively correlated with agriculture, wetlands, and horticulture but negatively
correlated with  precipitation and  runoff (Table 8-31); all factors were significant (p  < 0.05) in  the
regression.  Agricultural and lowland (cranberry) horticultural land uses could be associated with higher
pH in  surface waters  via fertilizer inputs.   Wetlands and water  impounding  via beaver activity  also
contribute to  sulfate reduction (Driscoll  et  al.,  1987a) and an increase in pH (Section  7.2).   Where
precipitation and runoff are high, lake pH will be reduced  because of dilution effects.

8.5.5.2 Southern Blue Ridge Province

      All regression  models relating ANC,  Ca  plus Mg, and  pH  of SBRP  streams  to land use factors
were strongly influenced by leverage  points (Table 8-33).   Since  Ca plus Mg and ANC are chemically
related surface water variables, those land uses that had potential and significant impact on one also
influenced the other variable. In all instances, the significant land uses were those which allowed within-
watershed  inputs  of base elements to SBRP  streams.   Deleting all the leverage points  removed  all
significant land uses from the ANC and Ca plus Mg models.  For  pH, removing only one leverage point
with open land and pasture  left no significant land use in  the regression.

      As stated  in Section 8.5.4.2, the SBRP region is very homogeneous in terms of  forest and  land
cover; overall, there is little  agricultural or urban development.  Where anthropogenic development or
disturbance is present, it has very  marked and significant impacts  on ANC,  Ca plus Mg,  and pH of local
streams.

8.5.5.3 Regional Comparisons

      Agricultural land  uses, particularly cultivated land and pasture were positively correlated  with ANC,
Ca plus Mg, and pH in both the NE and SBRP.  In the small SBRP region, single land  uses were usually
leverage points or outliers in the overall analysis. Removing SBRP watersheds with leverage points and
outliers from  the analysis produced  a  more  homogeneous  data set comprised mostly of forested
watersheds  with little  urban or other development.   Under these conditions, land use  was not readily
correlated with ANC, Ca plus Mg,  or pH.

      For northeastern lakes, sulfur deposition was negatively correlated with pH.  Via suifate reduction
processes under anaerobic conditions, northeastern wetlands mitigate the effects of fiigh sulfur deposition

                                               349

-------
Table 8-33,  Results of Regressions Relating ANC, Ca plus Mg, and pH of Southern Blue Ridge Province
Streams to Land  Use  Data
Dependent Regr. Significant Factors/ Significance Watersheds
Variable n R2 Sign Land Uses Included Level8 Removed13
ANC 32 0.11 + 8/ open land and pasture S
. 30 0.50 + 2/cemeteries, wasteland *
+ 3/cropland, horticulture, open land ***
+ 4/open land, urban areas, wetland S
27 0.21 + 4/open land, urban areas, wetland S
+ 10/pits and quarries *
26 no significant factors
Ca-f-Mg 31 0.28 + 10/pits and quarries *
30 0.21 + 10/pits and quarries **
28 0.18 + 3/cropland, horticulture, open land S
+ 10/pits and quarries
26 0.35 + 2/cemeteries, waste land **
+ 4/open land, urban areas, wetlands *
25 no significant factors
pH 31 0.12 + 8/open land and pasture S
30 no significant factors —
2A07827(L)
2A07813(L)
2A07802(L)&
2A07826(L)&
2A07830(L)
2A08808(L)
...
2A08808(L)
2A07813(L)
2A07826(L)
2A07827(L)
2A07802(L)
2A07830(L)
—
2A07813(L)
—
  **
  ***


 '(L)
  (O)
Significant at 0,15 level
Significant at 0.05 level
Significant at 0.01 level
Significant at 0.001 level

Leverage point removed from regression
Outlier point removed from regression
                                                     350

-------
and are associated with higher lake pH. In SBRP upland forested watersheds, there are few wetland or
riparian zones to mitigate deposition effects on stream pH.  Presently, however, high sulfate adsorption
capacity of SBRP soils does help minimize deposition effects on stream water chemistry.

8.5.6  Summary and Conclusions

  The major findings of this section are:

      *    In the NE,  surface water  sulfate is positively  correlated with  deposition  and
           extent of agricultural and urban development.

           In small  northeastern watersheds with beaver activity and wetlands, sulfate reduction
           processes decrease surface water sulfate concentrations and increase percent sulfur
           retention and pH.

      •    In the SBRP region, surface water sulfate and sulfur retention are influenced by local
           internal sulfur sources from pits and quarries.

      ซ    Agricultural land uses,  particularly cultivated land and pasture were  correlated with
           ANC, Ca plus Mg, and  pH in both the NE and  SBRP.  However, in the SBRP  region,
           removing outlying  or  influential sites produced a homogeneous dataset In which land
           use was not readily correlated with ANC, Ca plus Mg, and pH.

      •    In  both  the NE  and  SBRP,  forest cover is not directly related to  surface water
           chemistry; in the NE,  greater developed land (and less forest) is correlated with higher
           surface water ANC, Ca plus Mg, pH, and sulfate.
8.6  MAPPED SOILS

8.6.1  Introduction

      Soils are an important component of terrestrial ecosystems.  They are the principal source of plant
nutrients and provide a rooting medium for aboveground vegetation; they are the major site  of within
watershed decomposition reactions.  Soils host a plenitude of chemical reactions, including adsorption,
desorption, ion exchange, weathering, and precipitation reactions.  These chemical reactions can affect
the composition and quality of soil water and consequently subtending.surface and ground waters.  Soil
physical properties, such as structure  or architecture, the flowpath of soil water,  the  soil  particle-size
distribution,  the depth to impermeable layers, and .soil  bulk density, are also  important.  In natural
settings, the chemical and physical attributes of soils  are inseparable.  The objective of this analysis is
to identify the relationships that exist between mapped soils and surface water chemistry on a regional
basis.

      Some soils  are known to attenuate some of the effects of chronic sulfur deposition principally
through  sulfate  adsorption and  base  cation  supply reactions  (e.g., cation exchange  and  mineral
weathering).   These reactions are important in isolation  at the atomic level, however, as the scales
become coarser (i.e., atomic to  micro,  micro to meso, meso to watershed,  watershed to  regional) the
number of simultaneous, overlapping processes increases.  At the regional scale the relationship between
soil  properties and soil water chemistry involves thousands of hectares of soils and the composition of
                                               351

-------
a large number of lakes or many kilometers of stream reaches. The relatively simple set of relationships
(at the atomic level) becomes a complex set of diffuse relationships as the scale expands to the region.
Recognition of the fact that soils per se integrate a large number of physical and chemical processes is
the basis of the DDRP mapped soils analysis.  In this analysis we use the proportion of different kinds
of soils in watersheds, at a well-defined but regional scale,  to identify relationships that exist between
soils and the  chemical composition of subtending surface waters.

8.6.2 Approach

      As discussed  in the Sections 4.1 and 4.2 and described in Lee et al. (1989a), a stratified random
sample of watersheds was  selected and mapped in the DDRP.  Mapping followed  strict protocols, and
soil map units were regionally defined and correlated across the respective  regions.   The details of
watershed selection, map unit  correlation, and mapping can be found In Sections 5.2 and 5.4.

      In the NE 592 kinds  of  soils were  identified.  These  soils are the  components of the 338  map
units used to  map the soils in the NE.  In the SBRP 286 components and  176 map units were identified.
Because it was not tractable to characterize this large number of soils it became apparent that a smaller
set of soil units were  needed to make regional  soil characterization and  sampling  feasible.  The result
was the development  of the  soil "sampling  classes".  Soil components considered to have similar
chemical and physical characteristics were grouped into unique classes that we termed a soil sampling
class.  In the NE, 38  different sampling classes were identified, and in the SBRP there were  12.  Soil
sampling classes were the basis for soil sampling and analytical characterization and served as our main
link between  the analytical  data and the  soils of the regions.  They also serve as the  basic units for
relating mapped soils  to surface waters in this analysis.

      All watersheds are not  completely  covered by soils.  Other non-soil cover  is present and  can,
sometimes, extend over large areas.  To completely assess the relationships between soils and surface
water chemistry, such areas that occurred on  our sample  of watersheds were also  Identified during the
mapping and were termed "miscellaneous land areas". Because these areas may influence the quantity
and quality of surface waters they are Included In this analysis.  In the NE these include:  rock outcrop
(M01); pits, gravel (M02); rubble land (M03); and pits, quarry (M04).  In the SBRP  there were only two
miscellaneous land  areas:   rock outcrop (MRO)  and  quarry pits (MPQ).

      An overview of how this analysis was conducted is  presented in Figure 8-8.  After the soil maps
were digitized, a summary of the soil map units and  their  extent on each  watershed was obtained  from
the GIS for each region.  The  relative proportion of each  map unit component had been estimated for
the regions and entered into a mapping data  file.  Each map unit component  had  been assigned to a
sampling class and, therefore, the proportion of each  sampling class in the respective watersheds could
be calculated. For  example, 112 ha of map unit 134A was mapped on a  particular  watershed and map
unit 134A was defined by three components (a, b, and c) with the following percentages:  80 percent
component a, 15 percent component b, and 5 percent component c. Component a therefore accounts
for 89.6 ha (112 ha x  0.80) of  the 112 ha of map unit 134A  while b and c account  for 16.8  ha  and 5.6
ha, respectively. This calculation was repeated for each map unit on a watershed basis and the results
were pooled by sampling class.  The proportions of the watersheds in the various soil sampling classes
                                              352

-------
                Data flow
             Map watersheds
               Digitize maps
           Use GIS to calculate
            soil map unit areas
         Parse map unit areas into
           sampling class areas
            Calculate watershed
          sample class proportions
:Sampling class and miscellaneous
             land areas dataset
            Deposition dataset
        Dependent variables datasets
                                                  Modelling dataset
         f   Start   j
 Run stepwise selection procedure
    Use Mallow's Cp to select
    unbiased model variables
    Run standard regression
   including residual analysis
    Outliers/ ^x   NO
leverage points?
                                                                          Remove outliers and leverage points
Figure 8-8.   Data and regression model development flow diagrams.
                                                   353

-------
and  miscellaneous  land  areas  were calculated  by dividing the extent of the  soil sampling class or
miscellaneous land area by the total area of the  watershed.

      With the GIS we can  dissect or subdivide the watersheds to test various hypotheses.  We are
particularly interested in evaluating the effect of the riparian zone (near-lake or near-stream) on surface
water chemistry.   In the NE, two watershed  buffer zones were considered  in addition to the  whole
watershed area. One is limited to the area within the first 40-ft contour interval (called the 40-ft contour
buffer zone) above the sampled lake.  This buffer zone is used to delineate the near-lake  soils  and
wetlands.  The other Includes the same 40-ft contour buffer zone plus a 30-m linear buffer on either  side
of any perennial (blue-line) stream  and around  contiguous wetlands.   It also includes a 40-ft contour
buffer zone around any other lakes  or ponds that are on the watershed in addition to the sampled lake.
Due to contour map distortions or errors there are only 144 watersheds in the 40-ft buffer zone dataset
and  143 in the combined buffer dataset   Because the resource of  interest in the  SBRP is streams,
elevational or contour buffer zones are not suitable, so linear buffer zones were used.  These include the
area within 100 m of the blue-line streams on the DDRP sample of watersheds in the SBRP.  Because
2 of the 35 SBRP streams are not perennial,  the SBRP buffer zone dataset  has  a sample size  of 33.
Tables 8-34 through 8-38 summarize the distributions of the soil sampling classes and miscellaneous  land
areas on the DDRP sample  of watersheds.  Table 8-34 is for whole watersheds in the  NE, Table  8-35
is for the  land within the 40-ft GIS contour buffer zone, and Table 8-36 is for the combined GIS buffer
zones.  Table 8-37 is for the whole  watersheds in the SBRP, and Table 8-38 is for the GIS 100-m linear
buffer zones.

      Soils alone cannot explain all of the variation in  surface water chemistry.  Other factors such as
deposition and in-lake or in-stream processes also influence surface water chemistry.  Regional data on
in-Iake and in-stream processes do not exist, but deposition data do.  For this analysis we include six
variables from the long-term annual  average  data sets.  The details of how these data  were complied are
described in  Section 5.6.3.  The specific variables used in this analysis are discussed in Section 8.1.1.
A total of 48 independent variables are used the  NE (38 soil sampling classes  plus 4 miscellaneous  land
areas plus 6 deposition variables) and 20 in the SBRP (12 soil  sampling classes plus 2 miscellaneous
land areas plus 6  deposition variables).  The dependent variables include four surface  water chemical
measurements, sulfate Qปeq  L"1), ANC (in peq L"1), Ca plus Mg (CAMG in fj,eq  L"1), and pH.   A fifth
variable, % S retention,  is a calculated variable derived from deposition and surface water  chemistry
values (See Section 7 for details on how the percent sulfur retention values were  calculated).

8.6.3  Sulfate and Sulfur Retention

      The retention  of sulfur by terrestrial ecosystems Is an Important mechanism that can delay the
acidification of subtending surface  waters.   In the biogeochemical sulfur cycle there are two principal
soil or sediment mediated sulfur retention mechanisms, sulfate adsorption and sulfate reduction.  These
mechanisms  have been characterized and discussed in Sections 3.3,  7, and  9.2 in detail.  Soils low in
organic matter content having significant amounts of hydrous  oxides of iron and aluminum will tend  to
retain sulfate via adsorption. Soils or sediments that are  sufficiently wet to have anaerobic conditions
retain sulfate via sulfate reduction.
                                               354

-------
Table 8-34.  Summary Statistics for Percent Area Distribution of the 38 Soil Sampling Classes
and the 4 Miscellaneous Land Areas on the DDRP Sample of 145 NE Lake Watersheds
SMPLCLAS
E02
E03
EOS
E06
H01
H02
H03
101
I02
I05
I06
I09
110
111
121
I25
I29
130
133
137
138
140
141
(42
146
S01
S02
SOS
S09
S10
S11
S12
S13
S14
S15
816
S17
S18
M01
M02
M03
M04
MEAN
0.4
5.4
2.1
0.2
2.3
1.0
3.6
1.5
3.2
1.5
1.8
1.4
3.1
1.3
0.3
4.3
3.2
0.5
6.0
0.3
1.2
1.1
0.2
0.2
1.6
0.3
2.8
0.7
8.7
1.9
4.8
6.8
7.0
9.0
1.3
2.5
1.5
1.6
3.3
0.1
0.0
0.0
STD_DEV
1.0
19.6
2.5
1.0
4.2
3.2
5.2
3.3
5.2
4.6
5.5
6.0
10.2
5.0
3.3
11.1
9.0
1.9
15.7
0.8
3.2
7.0
1.0
1.2
5.7
1.0
11.5
2.0
12.9
5.9
9.9
8.8
9.0
12.7
5.6
7.6
7.1
5.5
5.1
0.7
0.0
0.1
MIN
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
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
0.0
0.0
0.0
0.0
0.0
0.0
Qla
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
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
0.0
0.0
0.1
0.0
0.0
0,0
MEDIAN
0.0
0.0
1.4
0.0
1.0
0.0
1.4
0.3
0.6
0.0
0.0
0.0
0.0
0.0
0,0
0.0
0.0
0.0
0.0
0.0
0.1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.1
0.0
0.7
3.3
3.9
1.2
0.0
0.0
0,0
0.0
1.6
0.0
0.0
0.0
Q3a
0.1
0.0
2.8
0.0
3.0
0.1
5.7
1.8
3.7
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.3
0.9
0.0
0.0
0.0
0.0
0.1
0.0
0.1
13.2
0.1
3.0
12.4
12.7
15.5
0.0
0.0
0.0
0.0
4.1
0.0
0.0
0.0
MAX
6.6
99.9
14.4
7.3
33.9
19.2
42.3
26.7
22.9
27.0
26.6
48.8
78.0
39.8
38.1
47.0
48.2
13.2
65.1
4.8
27.9
79.9
11.3
13.4
35.6
7.8
99.8
17.6
55.8
46.0
56.3
36.4
60.5
52.4
49.5
34.6
56.7
29.5
28.4
7.0
0.3
0.9
' Q1 is the 25th percentile, and Q3 is the 75th percentiie.
                                           355

-------
Table 8-35. Summary Statistics for the Percent Area  Distribution of the 38  Soil Sampling
Classes and the 4 Miscellaneous Land Areas in the CIS 40-ft Contour on the DDRP Sample of
145 NE Lake Watersheds
SMPLCLAS
E02
E03
E05
E06
H01
H02
H03
101
I02
I05
I06
!09
110
111
121
125
I29
I30
!33
I37
I38
I40
141
I42
146
$01
S02
SOS
S09
S10
S11
S12
S13
S14
S15
S16
S17
S18
M01
M02
M03
M04
MEAN
1.6
6.3
1.0
0.2
1.2
2.0
10.1
2.6
3.7
1.0
1.1
1.3
2.4
1.1
0.3
6.2
2.0
0.3
4,8
1.0
2.3
1.2
0.3
0.7
2.6
0.9
6.6
1.1
6.0
2.7
4.4
3.4
3.3
8.0
0.6
1.6
1.1
1.0
1.6
0.1
0.0
0.0
STD_DEV
4.1
19.9
1.6
1.2
3.0
6.8
12.5
6.4
7.3
3.6
4.0
5.4
8.0
4.4
3.0
16.8
6.2
1.3
13.8
2.9
5.6
7.4
1.4
3.3
10.1
2.9
19.1
3.9
13.0
8.9
10.3
6.5
6.7
12.3
3.4
5.6
6.8
3.7
3.4
1.1
0.1
0.0
MIN
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
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
0.0
0.0
0.0
0.0
0.0
0.0
Qla
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
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
0.0
0.0
0.0
0.0
0.0
0.0
MEDIAN
0.0
0.0
0.2
0.0
0.0
0.0
3.0
0.5
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
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.2
0.0
1.1
0.0
0.0
0.0
0.0
0.2
0.0
0.0
0.0
Q3a
0.1
0.0
1.6
0.0
0.8
0.0
18.9
1.9
3.1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.8
1.4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
4.6
0.1
2.1
3.1
3.8
12.4
0.0
0.0
0.0
0.0
1.6
0.0
0.0
0.0
MAX
20.2
100.1
11,1
8.2
19.0
48.8
60.0
40.9
39.9
22.8
22.2
34.9
73.9
37.6
34.9
77.5
33.0
10.9
75.8
22.5
43.9
80.1
11.5
31.3
66.1
18.0
99.9
30.2
69.8
53.9
49.2
35.4
59.4
51.8
38.0
39.3
60.9
23.6
24.8
10.2
1.2
0.0
' Q1 is the 25th percentile, and Q3 Is the 75th pereentite.
                                          356

-------
Table 8-36.   Summary Statistics for the Percent Area  Distribution of the 38 Soil Sampling
Classes and the 4 Miscellaneous Land Areas in the Combined  CIS Buffers  on the DDRP
Sample of 145 NE Lake Watersheds
SMPLCLAS
E02
E03
EOS
E06
H01
H02
H03
101
102
105
106
109
110
111
121
I25
I29
I30
133
137
138
140
141
142
146
S01
S02
SOS
S09
S10
S11
S12
S13
S14
S15
S16
S17
S18
M01
M02
M03
M04
MEAN
1.5
6.0
1.1
0.2
1.4
2.4
9.6
2.9
4.4
1.0
1.1
1.1
2.5
1.0
0.2
6.9
1.7
0.2
4.6
0.9
2.5
1.3
0.2
0.4
2.7
0.9
5.6
1.1
6.1
2.5
4.1
3.7
3.5
8.3
0.7
1.7
1.1
1.0
1.7
0.1
0.0
0.0
STD_DEV
3.8
19.7
1.7
1.4
3.2
7.0
11.5
7.0
7.7
3.5
3.8
4.9
8.6
3.5
2.4
17.7
5.3
1.0
12.6
2.2
5.6
7.4
1.2
1.7
10.3
2.7
17.4
3.8
12.8
7.7
9.6
6.6
6.7
12.3
2.8
5.6
7.0
3.7
3.5
0.7
0.1
0.0
MiN
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
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
0.0
0.0
0.0
0.0
0.0
0.0
Q1a
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
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
0.0
0.0
0.0
0.0
0.0
0.0
MEDIAN
0.0
0.0
0.3
0.0
0.0
0.0
3.3
0.4
0.8
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
0.0
0.3
0.0
0.1
0.2
0.0
1.0
0.0
0.0
0.0
0.0
0.3
0.0
0.0
0.0
Q3a
0.1
0.0
1.6
0.0
1.1
0.1
17.3
1.9
4.1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.7
1.7
0.0
0.0
0.0
0.0
0.2
0.0
0.0
6.1
0.0
2.2
5.0
4.7
12.4
0.0
0.0
0.0
0.0
1.6
0.0
0.0
0.0
MAX
19.4
100.1
11.5
11.2
20.4
49.0
60.0
41.2
39.9
23.0
22.0
34.9
73.9
26.4
27.9
74.4
32.7
8.1
62.5
14.2
43.5
80.1
11.5
17.1
65.6
16.4
100.2
30.2
69.8
53.9
51.9
35.4
59.4
53.1
24.4
40.6
64.9
23.6
25.1
6.7
0.8
0.6
1Q1 is the 25th percentile, and Q3 is the 75th percentile.
                                          357

-------
Table 8-37. Summary Statistics for the Percent Area Distribution of the 12  Soil Sampling
Classes and the 2  Miscellaneous Land  Areas on  the  DORP Sample of 35  SBRP Stream
Watersheds
SMPLCLAS
ACC
ACH
ACL
FL
FR
MSH
MSL
OTC
OIL
SHL
SKV
SKX
MPQ
MRO
MEAN
17,7
5.9
32.9
2.5
4.2
2.8
11.2
2.5
4.9
7.4
5.1
1.6
0.1
1.2
STD_DEV
28.3
9.7
26.3
3.1
10.2
7.3
16.4
10.5
9.2
8.7
8.2
3.3
0.5
1.9
MIN
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
Q13
0.0
0.0
1.8
0.0
0.0
0.0
0.0
0.0
0.2
0.0
0.0
0.0
0.0
0.0
MEDIAN
0.7
0.5
30.7
0.9
0.0
0.0
0.0
0.0
1.1
5.1
0.6
0.0
0.0
0.7
Q3a
28.8
8.5
57.1
4.0
0.0
0.0
21.9
0.0
3.4
15.0
7.0
1.3
0.0
1.5
MAX
80.1
37.8
78.2
10.7
43.2
48.1
61.7
52.3
37.6
30.0
36.0
18.1
2.8
15.0
 ' Q1 is the 25th percentile, and Q3 is the 75th pereentile.
                                          358

-------
Table 8-38.   Summary Statistics for the  Percent  Area Distribution  of the 12  Soil Sampling
Classes and the 2 Miscellaneous Land Areas in the 100-Meter Linear GIS Buffer on the DDRP
Sample of 35 SBRP Stream Watersheds
SMPLCLAS
ACC
ACH
ACL
FL
FR
MSH
MSL
OTC
OIL
SHL
SKV
SKX
MPQ
MRO
MEAN
5.1
10.6
24.6
11.1
2.9
4.4
10.1
6.8
5.2
6.4
10.7
1.0
0.2
0.9
STD_DEV
10.9
14.8
20.7
12,9
7.2
10.1
14.2
19.9
6.7
8.3
14.4
2.3
1.0
1.5
MIN
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
Q1a
0.0
0.0
0.1
1.0
0.0
0.0
0.0
0.0
0.6
0.2
0.0
0.0
0.0
0.0
MEDIAN
0.7
0.9
25.0
8.1
0.0
0.0
0.0
0.0
2.0
2.8
3.0
0.0
0.0
0.1
Q3a
1.8
24.9
36.6
15.8
0.0
0.0
17.6
0.0
6.1
12.8
16,4
0.0
0.0
2.3
MAX
40.7
43.7
71,0
54,8
27.6
59.0
42.0
79.1
24.4
28.8
50.8
11.7
5.8
10.2
a Q1 is the 25th percentile, and Q3 is the 75th pereentile.
                                         359

-------
      Sulfate concentrations In the DDRP SBRP sample of streams are, in general, much lower than in
the sample of lakes in the NE. However, the region as a whole has somewhat higher sulfur deposition
than the NE.  The deposition in  the  SBRP is more uniform than in the NE.  Because the SBRP is a
relatively small region with  relatively uniform deposition, there is not a significant sulfate deposition
gradient as  in the NE.  Unlike the NE  the watersheds in the SBRP are not at  sulfur steady  state.   A
principal difference between the two regions is the soils.  In the NE the soils are relatively young, having
less profile and secondary mineral development.  In the SBRP most soils are relatively older and more
deeply weathered with abundant accumulations of secondary mineral phases (hydrous oxides of iron and
aluminum).

8.6.3.1  Northeast

      Following the procedures described above  and in  Section 8.1,  several regression models were
developed for the relationship between the independent variables (mapped soils and deposition) and
lake sulfate.  The results are presented  in Table 8-39.  The first whole  watershed model for lake suifate
         p
had an R  of 0.73.   Two observations, however, had unusually  high lake sulfate concentrations and
wereidentified as  being  outlier and/or  leverage  points.   These watersheds also had the  only two
occurrences  of the miscellaneous land  area, M04 — quarry  pits.  This finding is consistent with  the
observation  that  major watershed disturbances,  such as  mining, may  outweigh any  surface  water
chemistry effects due to acidic deposition.  Removing the  watersheds with quarry pits and one other
leverage point from the whole watershed analysis  resulted in the model R  statistic dropping from 0,73
to 0.64.  The variables in the  regression model, however, remained the same, except for M04 which  is
now excluded.  Table 8-39 shows the variables that were included in the lake sulfate regression models,
indicated by the sign of their respective parameter estimates.  These signs indicate either positive or
negative correlation to the dependent variable.  The inclusion of both wet and dry sulfate deposition in
the regression models for the NE  is not surprising.  In the 142 observation, whole  watershed model these
two variables account for about 45 percent of the variability in lake sulfate.  As discussed In Section 7,
the NE is almost at sulfur steady state (i.e., input ซ output) which explains why these two variables make
such a large contribution to the explanatory power of the model.

      In addition to sulfur deposition variables, eight soil sampling classes are Included in the best whole
watershed regression model for  lake sulfate.   Sampling classes E03, H03, and 133  were consistently
included in  the lake  sulfate  regression models.   ฃ03 is positively correlated with lake sulfate.   This
sampling class is characterized by coarse texture, poor development, and low sulfate adsorption  capacity.
These  soils  are excessively drained  and seem to be a non-interacting  conduit  for drainage waters.
Likewise, 106 and 111  are well drained and positively related with lake sulfate. These three soil  sampling
classes, however,  generally occur only in Subregion 1D. Therefore, it  is also plausible that these three
sampling  classes  (EOS,  I06,  and I)  are  surrogates  for sea-salt sulfate contributions that  may be
underestimated in  the LTA deposition dataset.  Their inclusion in the regression model, may have little
to do with their actual chemical and physical properties.

      H03 was consistently negatively related to lake sulfate.  The H03  soils are deep,  wet, organic soils
characterized by low  pH (dysic).  Because of the negative correlation,  these soils are thought  to be an
active site of sulfur retention via  biological processes (e.g., sulfate reduction), and would be  expected
to have a positive  relationship with percent sulfur retention.  The analysis indicates that sampling  class

                                               360

-------
Table 8-39.  Lake Sulfate and Percent S Retention Regression Models Developed
for NE Lakes Using Deposition, Mapped Soils (as a Percentage of Watershed Area
in Soil Sampling Classes) and Miscellaneous Land Areas  (as a Percentage of
Watershed Area) as Candidate Independent Variables3
Lake Sulfate
Whole 40-ft Comb.
SO4-Wet + +
SO4-Dry + + +
Ca+Mg-Dry
EOS + + +
H01
H03
I06 +
111 +
I25 + +
ISO +
I33 ...
S02
S12 + +
S17
S18 -f +
R2 0.64 0.61 0.59
adjusted R2 0.61 0.59 0.57
n-lakesb 142 141 141
p-modelฐ 10 7 7
O/Ld 332
Percent S Retention
Whole 40-ft Comb.

ซ ซ _

.
+ + +




+ +


+ +
-
0.31 0.39 0.44
0.29 0.36 0.41
141 142 140
466
423
* + and - refer to positive and negative parameter estimates, respectively
                                         the regression model
= number of observations
                         to
 . p-model = number of regressor variables in model
 O/L = number of outlier or leverage points omitted
                                              361

-------
125 is positively correlated with lake sulfate concentrations and that 133 is negatively related.  Fragipans
occur in the soils of both of these classes within  100 cm of the soil surface.  The principal difference
between these sampling classes is that I25 soils are very poorly to somewhat poorly drained while I33
soils are somewhat  poorly drained to well drained and deep.  Anaerobic conditions occur throughout the
upper 100 cm of soil in the I25 class during some part of the year, intuitively, this would suggest that
this class of soils should have a negative  relationship with lake  sulfate concentrations rather  than  a
positive one,  because of the potential for sulfate reduction during the anaerobic periods.  Other factors
such as the landscape  position of these soils and the timing and  nature of the anaerobic periods may
be responsible for the observed relationship.

      The S12 class of soils is well drained, moderately deep, coarse-loamy  Spodosols with relatively low
base saturation and pH.  Water  moves rapidly through  this class of soils and moves downhill at the
bedrock contact with little opportunity for sulfur retention. There is a positive relationship between S12
soils and lake sulfate.  The  reason for the negative relationship between the S02 class and lake sulfate
is unclear from this analysis; S02 soils may be a surrogate for another attribute.

      In the NE the percent % S retention values range from  -22 to +60, with  a  median value of -4.
In contrast, in the SBRP the range is from -60 to  +89, with a median value of 75  percent.  In the NE
the systems are almost at steady state with respect to sulfur.  In contrast, the SBRP is effectively retaining
most of the sulfur inputs. We would expect  markedly different  results in the regression analysis.

      In the  NE  (see Table 8-39) the  best  whole watershed model explains only 31 percent of the
variation in percent sulfur retention with a four-variable model.  The best model is a six-variable model
using the combined buffer data with an R2 of 0.44.  The sum of dry Ca and Mg deposition,  H01, H03,
and S18 was included in all three  models; H03 soils have a  positive parameter estimate.  The H01 soils
are thin  (< 30 cm), organic soils overlaying bedrock or fragmental material that is freely drained. They
are not wetland soils.   They are  probably active sites of organic  matter decomposition and  contribute
sulfur from organic  matter to the  surface waters.  The class H03, as described earlier, includes wetland
soils that presumably retain sulfur via  a sulfate reduction mechanism.   Sampling  class S18 soils are
shallow and somewhat  excessively drained.  They are likely to be  non-interacting conduits for drainage
waters.   The I33 and S17 classes are  positively related to  percent sulfur  retention in the buffer zone
models. The  relationship with I33 was not as expected  as  was discussed with the lake sulfate results
above.  The reason for inclusion of dry Ca plus Mg deposition in  the models Is not known.  It may be
an  artifact of the deposition data compilation or it may be functioning  as a  surrogate for another
deposition variable.

8.6.3.2   Southern Blue Ridge Province

      The regression models developed for the stream sulfate concentrations are summarized in Table
8-40.  The results for seven models  are included:  four using  the soils and miscellaneous land area
distributions on the whole watersheds and three for the distributions in the 100 meter buffers.

      The whole  watershed  model with 35  observations includes five  variables that are all positively
correlated with stream  sulfate.  One  is the  miscellaneous land area, MPQ.  As noted in the NE, the
occurrence of quarries  on watersheds can have  a significant  effect on the subtending surface water
chemistry.  The effect on the surface water is dependent upon the type of geological strata being mined.
                                               362

-------
Table 8-40.  Regression Models of Sulfate in SBRP Streams, Developed Using
Deposition, Mapped Soils (as a Percentage of  Watershed Area in Soil Sampling
Classes) and Miscellaneous Land Areas (as a  Percentage of Watershed Area) as
Candidate Independent Variables8
Whole Watersheds
Ca+Mg-Wet
OTC +
ACH ;
MSH
SHL + + +
OIL + + +
MSL + + +
MPQ +
R2 0.84 0.63 0.45 0.30
adjusted R2 0.82 0.58 0.42 0.28
n-streamsb 35 33 32 31
p-modelฐ 5421
O/Ld none 234
100-m buffer zones
-
+
-
-
+ +
+ + +
+ +

0.82 0.66 0.57
0.79 0.59 0.52
33 31 30
553
none 2 3
* + and - refer to positive and negative parameter estimates, respectively
  n-lakes = number of observations used to develop the regression model
* p-model = number of regressor variables in model
  O/L =  number of outlier or leverage points omitted
                                               363

-------
In the SBRP sample population of watersheds, there was one occurrence of MPQ, The surface water
in this watershed had the highest observed value of stream sulfate concentration  and the lowest % S
retention.  It is likely that the mining on this watershed  has exposed sulfur-bearing materials which have
subsequently oxidized and impacted the surface water.  The Anakeesta Formation (King et al.,  1968),
which contains sulfur-bearing  minerals is common in parts of the SBRP.  Watershed disturbances, such
as road construction and landslides, may expose these materials as well.

     The SHL and OTL soils, as defined In  Section 5, are both well drained and have relatively low
organic matter content.  SHL soils are shallow and OTL soils are deep.  In general, the soils in these
sampling classes have low sulfate adsorption  capacities.  The OTC soils, however, only occur on three
watersheds in the SBRP DDRP sample, and  In one Instance they cover more than 50 percent  of the
watershed. These soils, while high in  secondary clay minerals, have pH values that are unfavorable for
suifate adsorption and therefore do not retain  sulfate to any significant degree.  The watershed with the
high OTC coverage also had the second highest stream sulfate concentration.  In addition to low sulfate
adsorption potential, it is likely that the calcareous parent material of the OTC is interbedded with  sulfur-
bearing materials.   In the 33 observation, whole watershed model, the watersheds with more than 50
percent OTC coverage and the MPQ site are  not included. The resulting model is the same as the 35
observation model,  except that it no longer has MPQ and the sign on the OTC variable is negative. The
sign reversal is probably caused by the low abundance of OTC on the two remaining  OTCs. The 32
observation  model  has  one OTC site remaining with 0.1 percent  OTC coverage.   In this mode, OTC
was not a significant explanatory variable in this model.  The 31 observation  model  only  had  one
significant explanatory variable, the SHL sampling class.  The soils in the SHL sampling class account
for 30 percent of the variability in stream sulfate concentrations alone. These soils  are well drained, low
in clay,  have moderate to rapid permeability, and are less than 50 cm deep.  These four properties are
characteristic of soils with short hydrologic contact times that have little or no effect on the chemistry of
drainage waters passing through them. Surface waters in watersheds with an abundance of shallow soils
will be more susceptible to acidification than watersheds with deep, moderately well-drained soils.

      The soils in the MSL sampling class were consistently selected with positive parameter estimates
in the stream sulfate regression models.  This suggests that  as the proportion of MSL soils on a
watershed increases, the concentration of sulfate in the subjacent stream also increases.  The soils in the
MSL sampling class are, by definition,  derived  from metasediments and have low organic matter content.
Chemically, they have only intermediate sulfate adsorption potential.  The positive parameter estimates
indicate, however, that these  soils may be associated with sulfur-bearing materials.

      The 33 observation, 100-m buffer model Is similar to the 35 observation, whole watershed model,
but has lower R2 and adjusted Ra values.  In the 31 observation  model, the ACH and MSH  sampling
classes both have negative parameter estimates.  These soils as a group have relatively higher organic
matter content in the surface layer than their ACC and MSL counterparts. This implies that, when these
sampling classes occur in the near-stream zone, sulfate is retained.  The 30 observation model has only
three variables, SHL, OTL, and MSL,  and explains approximately 60 percent of the variation In stream
sulfate concentrations.

      The regression  model results for sulfur retention in the SBRP are presented in Table 8-41.  The
whole watershed model with 35 observations has an R2 of 0.86. This R2 is highly inflated by the presence

                                              364

-------
of one watershed.  This watershed was the only observation with a negative net  sulfur retention  (-66
percent).  It is also the watershed with the quarry (MPQ), which serves as a source of sulfur.  Omitting
this watershed  from the analysts  produces a model with only  one variable and an R2 of 0.34.   The
remaining variable is the MSL sampling class, which is negatively correlated to percent sulfur  retention.
As proposed in the stream sulfate discussion above, the MSL sampling class soils are associated  with
sulfur-bearing  parent materials  that  function as a  source of sulfur.   Omitting  five  more possible
outlier/influence points only decreases the model R2 from 0.34 to 0.33. The resulting model contains  only
the MSL sampling class.   On a  regional basis, these soils explain  one third of the variation in sulfur
retention and appear to be an important source of sulfur.

      Three 100-m buffer,  sulfur retention regression models are also presented  in Table 8-41.   The
model with 33  observations and  R2 of 0.84 is biased by the watershed with the quarry.  Omitting  that
watershed results in a three-variable model that accounts for 52 percent of the variation in sulfur retention.
The variables in this model are all negatively correlated with sulfur retention.  They include the OTC, SKX,
and MSL sampling classes.  The OTC and MSL have low and intermediate sulfate adsorption potentials,
respectively, and may be sources of sulfur.  The SKX  soils are coarse textured,  excessively drained
Inceptisols formed in metasedimentary residuum.  They  may be a minor sulfur source, but more likely
they are non-interacting soils with short hydrologic contact times. Omitting a second watershed, this one
with 79 percent of the 100-m buffer zone area in the OTC sampling class, produces a two-variable model
that  explains 45  percent  of the  variation  in sulfur retention.   In  addition to the MSL sampling class
(negative  parameter estimate), the MSH sampling class is included in this model.  Based on the soil
sampling class  definitions, the only difference between these two sampling classes is their organic matter
content and thickness of the surface  layers; the MSH is  high in organic matter and the MSL is low.  In
general, the MSH soils are well drained; however, where they occur in the near-stream areas (within 100
m of the stream) they may be saturated with water at depths 100 cm or more below the soil surface for
a sufficient period to create anaerobic conditions that can potentially retain sulfur via sulfate reduction.
Because of the distribution of the MSH sampling class soils, they may also be a surrogate for watersheds
with high  sulfate adsorption capacity  soils.

8.6.3.3 Regional Comparisons

      In the NE sulfur deposition explains the majority of the variability in lake sulfate concentrations.
The central tendency for watersheds in the NE is to be at sulfur steady state where sulfur input ซ sulfur
outputs.  The capacity of these systems to retain sulfur effectively is inherently low (Section 9.2), and has
been exhausted (i.e., low or negative sulfur retention).  In the SBRP where sulfur retention is high and
the watersheds are retaining most of the sulfur inputs, sulfate deposition is not yet significantly related
to stream sulfate concentrations.

      In the NE there is evidence to  suggest that localized sources  of  sulfur deposition, not accounted
for in  the  LTA deposition dataset, may be contributing to higher  sulfate  concentrations in  the near-
coastal watersheds in  Subregion 1D.  This additional sulfur deposition is probably derived from wind-
blown sea-salt aerosols.  In both regions there were also  indications that at least one of the soil sampling
classes is functioning as a sulfur source or is a surrogate for a source of sulfur.
                                               365

-------
Table 8-41.  Regression Models of Percent Sulfur Retention in SBRP
Stream Watersheds Developed Using  Deposition, Mapped Soils (as a
Percentage of Watershed Area in  Soil Sampling Classes), and
Miscellaneous Land Areas (as a Percentage of Watershed Area) as
Candidate Independent Variables"
Whole Watersheds
OTC
SKX
MSH +
MSL ...
MPQ
Ra 0.86 0.34 0.33
adjusted R2 0.84 0.32 0.30
n-streamsb 35 34 29
p-modelc 411
O/Ld none 1 6
100-m buffer zones
-
_
+
.
-
0.84 0.52 0.45
0.81 0.47 0.41
33 32 31
432
none 1 2
  •f and - refer to positive and negative parameter estimates, respectively
  n-lakes =  number of observations used- to develop the regression model
c p-mode! = number of regressor variables in model
d O/L = number of outlier or leverage points omitted
                                      366

-------
      Very poorly drained soils and acidic Histosols were positively related to sulfur retention;  sulfate
reduction  is the likely mechanism. A stronger relationship was found  when these soils occurred in the
near-lake or near-stream areas in the NE. These soils may be responsible for most of the sulfur retention
in the NE (See Sections 8.5 and 9,2).

      Shallow soils with short hydrologic contact times serve as non-interacting drainage water conduits
in both  regions.  It  was also noted in both regions that the  surface water in watersheds with a major
watershed disturbance, such as a  quarry, have  higher  concentrations of sulfate.  The extent of this
effectdepends on the nature of the geologic strata being disturbed and the magnitude and location  of
the disturbance.

8.6.4 ANC. Ca plus Ma. and pH

      Acid neutralizing capacity (ANC)  is an important measure of the potential of surface waters  to
buffer the input of acidic deposition.  Systems with zero or negative  ANC are already acidic and with
low  ANC  are  likely  to be  vulnerable to acid inputs.   Systems with  high ANC are  strongly buffered
(capacity  protected)  against acid inputs and are therefore not likely to become acidic, even at current
levels of deposition,  for some time, possibly centuries.

      ANC is  the principal  indicator of surface water buffering.  Related to ANC are the sum of base
cations  (Ca, Mg, K, Na) and the.surface water pH. In this analysis, the  sum of the principal base cations,
Ca and  Mg, is considered.

8.6.4.1  Northeast

      In the NE region, regression models were developed that explain approximately one-half of the
variability  in ANC (Table 8-42).  The best whole watershed model has six variables, including wet suifate
deposition.  Sulfur deposition is negatively correlated  with ANC. The  remaining variables are sampling
class variables and are all positively related to lake ANC.  Soils in sampling classes 101, 111,  I25, and I46
(very poorly drained and poorly drained  Inceptisols) are  among  the classes with the highest base
saturation and pH values.  It follows that these soils are sources of base cations and supply subtending
surface  waters with  base cations and  buffer lake ANC.  The soils in sampling class I06 are shallow, and
in general, low in pH and base  saturation.  Their  contribution to  ANC is questionable; they may be
functioning as a surrogate for another variable.

      The best ANC  model is developed with combined buffer data, explaining 54 percent of the variation
in lake ANC with nine variables. Sulfate deposition and the I05 sampling class are the only variables in
the model negatively correlated with  ANC. I05 occurs mainly in Subregion 1D and may  serve as a
surrogate for sea-salt contributions of sulfur. Sampling classes lOi, 111,125, I46, and 106 are in this model
as well  as the whole watershed ANC model.  This model also includes sampling classes S01 and S18
as variables, both having positive parameter estimates.  The soils in both of these sampling classes have
intermediate base saturation (-20 percent) and pH (~4.6). S01  soils are deep and are widely distributed
across the region in small amounts.   S18 soils are very shallow with a lithic or paralithic contact within
50 cm.  The S18 sampling  class occurs only in Subregion 1E.
                                               367

-------
Table  8-42.   Lake ANC and  the Sum  of  Lake  Calcium and Magnesium Regression Models
Developed for NE Lakes Using Deposition, Mapped Soils (as a Percentage of Watershed Area in
Soil Sampling Classes) and Miscellaneous Land Areas (as a Percentage of Watershed Area) as
Candidate Independent Variables*
                       Lake ANC                       Lake Ca plus Mg
                  Whole      40-ft         Comb.      Whole     40-ft         Comb.


SO4-Wet
H-Wet
E06                  '         +                                 +           +
H01
H02                           +                                              +
101                +          +           +          +                       +
I05                                        ....
I06                +                      +          +          +           +
I09                            +
111                +          +           +          +          +           +
I25                +          +           +          +          +           +
I46                +          +           +          +          +           +
S01                                       +
S18                           +           +                     +           +
M04                                                  +


R2                0.48        0.47         0.54        0.56       0.49         0.55

adjusted R2       0.46        0.43         0.51        0.54       0.46         0.52

n-lakesb          145         144         142        145        144         143

p-modelc          6          10          9          8          8           10

O/Ld             none       none        1          none       none        none


* + and - refer to positive and negative parameter estimates, respectively
  n-lakes = number of observations used to develop the regression model
ฐ p-model = number of regressor variables in model
  O/L = number of outlier or leverage points omitted
                                             368

-------
      Because lake Ca plus Mg is strongly related to ANC (r = 0.94), the regression models developed
for Ca plus Mg are similar to those for ANC and have comparable R2 values.  Sampling class 105 and
wet hydrogen deposition  (H-Wet) have  negative parameter estimates.  105 was discussed above.   H-
WET is strongly correlated with wet sulfate deposition  (r  = 0.92) and was substituted for the wet sulfate
deposition variable  included in the  ANC models.   In  the whole  watershed model  of Ca plus Mg, the
miscellaneous land  area M04 (quarries) was included in the model. These watershed disturbances also
increase the amount of Ca and Mg in the subtending  surface waters in addition to increasing the levels
of sulfate.  In fact, the levels of Ca plus Mg and sulfate are quite similar in these watersheds,  and both
have fairly high ANCs.  Ca and  Mg  appear to be the cations accompanying the mobile anion sulfate.
As long as the soils are not being depleted of base cations, this  situation is little cause for concern.

      The lake pH regression models were simpler than the ANC  and Ca plus Mg models because they
have fewer variables and similar  R2 values fTable 8-43).  In ail three models wet sulfate deposition had
negative parameter estimates while the sampling classes 109 and 125  had positive parameter estimates.
These results seem to  be  reasonable.   Soils in  the 125  sampling class have a  relatively high base
status.which accounts for the positive correlation with lake pH. The 109 soils are also positively related
to pH, but are lower base status soils than the 125 soils. H03 is also included in the whole watershed
model with a positive parameter  estimate. Soils in the H03 sampling  class are deep, wet, organic soils
principally located in wetlands. These soils are dysic, meaning that the pH of undried  samples is less
than 4.5 (in 0.01  M CaCl2). Because pH is an intensity variable  (i.e., concentration) the pH of the last
soil that drainage water passes through before it reaches the lake may initially control the pH of the lake
water. If there are extensive wetlands surrounding a lake including H03 soils, the H03 soils may be the
last soil that the drainage waters pass through; the pH of the lake will therefore be  similar to the pH of
the  H03.

      The combined buffer model for lake pH had the  highest R2 and adjusted R2 values, 0.49  and 0.46,
respectively, of the three  regression models.  As in the whole watershed model, the combined  buffer
model includes wet sulfate deposition and the sampling classes 109 and 125.  Additionally, it has ฃ05 and
I05 with negative  parameter estimates and E06 and H02 with positive parameter estimates, H03 was not
included as an important variable in either of the buffer models. The soils in the EOS sampling  class are
poorly developed, very shallow (< 25 cm), underlain by  hard  bedrock, and also have one of the lowest
aggregate pH values. Because of their chemical and  physical characteristics, an increasing abundance
of ฃ05 soils will lead to lower lake pH values. The characteristics of the soils in the ฃ06 sampling class
are the direct result of human activities.  They are deep soils that lack pedogenic development due to
significant anthropogenic disturbance such as road construction.  They are classed as Udorthents.  In
general, they have moderate to high base saturation and moderate pH.  Because of recent disturbance,
they may have abundant fresh weatherable mineral faces that supply base cations at a higher rate than
other soils in the region.

8.6.4.2 Southern Blue Ridge Province

      ANC in the SBRP  is generally higher than in the  NE.  The median ANC value for the  region is
120 /ieq L"1.  In the NE the median is 56 /neq L"1.  Because of the chemical characteristics of the soils
in the SBRP (i.e., higher sulfate adsorption capacities), these systems are not close to steady state with
respect to sulfur inputs and outputs.  The soils in the region are retaining a significant proportion of the

                                              369

-------
Table 8-43.  Lake pH  Regression Models Developed
for NE Lakes Using Deposition, Mapped Soils (as a
Percentage of Watershed Area in Soil Sampling
Classes) and Miscellaneous Land Areas (as a
Percentage of Watershed Area) as Candidate
Independent Variables8
                           LakepH

              Whole         4pt           Comb.
SO4-Wet
EOS
E06
H02
H03
105
109
125
R2            0.45           0.45           0.49

adjusted R2   0.44           0.42           0.46

n-Iakesb      144           144            142

p-mode!c      467

O/Ld         1              none          1
b
+ and - refer to positive and negative parameter estimates, respectively
n-Iakes = number of observations used to develop the regression model
ฐ p-model = number of regressor variables in model
 O/L = number of outlier or leverage points omitted
                               370

-------
sulfur in deposition, and at present, appear to be delaying the acidification of the surface waters in the
region.  The ANC of these surface waters and the factors that control it are, therefore, very important

      Regression models of the SBRP stream ANCs using the mapped soils, mapped miscellaneous land
areas, and wet and dry deposition are presented in Table 8-44. The whole watershed regression models
with 35, 34, and  33 observations all have very high R2 values, 0.92, 0.86, and 0.943, respectively.  These
large  R2 values are dueto the presence of observations with very high ANC (>1000 peq L"1 ) values.
Two of the three sites are associated with the OTC sampling class and its calcareous parent materials.
These three systems are probably capacity protected against acidification.   Omitting them from the
analysis and leaving 32 observations results in  a three-variable model that explains 47 percent of the
variation in stream ANC.  One of the 32 watersheds has 0.1 percent OTC.  Because of the strong positive
relationship between calcareous materials and ANC, OTC has been included by the  stepwise procedure
as a variable.  The FI_ sampling class is also included with a positive parameter estimate. The soils  In
this sampling class occur on flood plains.  Compared to the other sampling classes in the SBRP, the soils
in the FL sampling class have the third  highest base saturation and the second highest pH.  Omitting a
fourth influence point results in a one-variable regression model for stream ANC that has an R2 of 0.40.
The single variable is the FL sampling class.

      Unlike the NE, there is little or no Indication that the DDRP sample of streams in  the SBRP are
contaminated with Na from road  salt or  sea salt  additions.  Therefore, in addition to considering the
sum of stream Ca and  Mg concentrations as a dependent variable, we have included an analysis of the
sum of the four  principal base cations (Ca -f Mg + Na + K), the sum of base  cations  (SOBC).  In this
section and the two sections that follow, additional analyses of the relationship between these watershed
attributes and SOBC are included.

      The  initial 100-m buffer models of stream ANC are similar  to the whole  watershed models.
Dropping  outlier/influence points  results  in a two-variable model that accounts for 92 percent of the
variation in stream ANC.  The two variables are OTC and FL   The OTC accounts for most of the
explanatory power.  Omitting all of the high ANC sites from this analysis does not produce an unbiased
model.

      The results of the Ca plus Mg analysis are presented in Table 8-45.  Inclusion of all 35 observations
results in  a four-variable regression model with an Ra of 0.90.  However, as  in the ANC analysis this
model is strongly influenced by three observations with exceptionally high values of  Ca plus Mg.  These
are the same three with high ANC. Omitting them from the analysis results in a two-variable model with
an  Ra of 0.42. The two variables  are FL and MPQ.  Analysis of the residuals and Influence diagnostics
indicate that the MPQ  is an influence point.   Omitting it  and  developing a model  based  upon 31
observations results in a one-variable model.  The variable is the FL sampling class.  This  is the identical
model developed for ANC with the same 31  observations.

      The  results for  the 100-m  Ca plus Mg  model  follow the  pattern  set by the whole  watershed.
Inclusion of all 33 observations results  In a model with a high R2 but  with two strong influence points.
Omitting these two observations and rerunning the analysis leads to  a higher R2 model that -has four
variables.   At the same  time another influence  point is identified.  Omitting this observation results in a
two-variable model with yet another influence point.  This time, however, eliminating it and proceeding
with the analysis does not produce an  unbiased model.
                                              371

-------
Table 8-44.  Regression Models of ANC in SBRP Stream Watersheds, Developed
Using Deposition, Mapped Soils (as a Percentage of Watershed Area in Soil
Sampling Classes) and Miscellaneous Land Areas (as a Percentage of Watershed
Area) as Candidate Independent Variables"
Whole Watersheds
Ca+Mg-Dry
OTC + + + +
SKV
SKX
FL + + +
MSH
SHL
OIL + +
MSL + +
R2 0,92 0.86 0.943 0.47 0.40
adjusted R2 0.90 0.84 0.938 0.41 0.38
n-streamsb 35 34 33 32 31
p-modelฐ 44331
O/Ld none 1234
100-m buffer zones

+ + +
-
+
+
-
-
+ +
+ +
0.91 0.933 0.924
0.88 0.926 0.919
33 32 31
732
none 1 2
* + and - refer to positive and negative parameter estimates, respectively
  n-lakes =  number of observations used to develop the regression model
0 p-model = number of regressor variables In model
  O/L ~ number of outlier or leverage points omitted
                                              372

-------
Table 8-45.  Regression Models of Calcium Plus Magnesium in SBRP Streams,
Developed Using  Deposition, Mapped Soils (as a Percentage of Watershed Area
in Soil Sampling Classes) and Miscellaneous Land Areas (as a Percentage of
Watershed Area) as a Candidate Independent Variables
Whole Watersheds
OTC +
SKV
SKX
FL + +
MSH
OIL +
MSL +
MPQ +
R2 0.90 0.42 0.25
adjusted R2 0.89 0.38 0.22
n-streamsb 35 32 31
p-modelc 4 2 1
O/Ld none 3 4
100-m buffer zones
+

+
+ +
-
+ +
+
+
0.88 0.96 0.922
0.86 0.95 0.916
33 32 30
542
none 1 3
a + and - refer to positive and negative parameter estimates, respectively
  n-Iakes =  number of observations used to develop the regression model
*j p-model = number of regressor variables in model
  O/L =  number of outlier or leverage points omitted
                                               373

-------
     The results of the  whole watershed and  100-m buffer zone regression analyses for  SOBC are
presented in Table 8-46.  The whole watershed model, including all 35 SBRP systems, produced a four-
parameter model with  an R2 of 0.91.  included In this  model is the calcareous sampling  class OTC
(positively related to SOBC).  It follows that the presence of significant amounts of calcareous material
can increase the levels of base cations that may be transported to the surface water.  Systems with high
SOBCs and ANCs are likely to be capacity protected against acidification. Analysis of the residuals found
that three systems, all with  ANC and SOBC greater than 1000 /^eq L~1 were strong influence points.
Dropping these and rerunning the analysis produces a three-parameter model with an R2 of  0.50.  This
model  also had significant influence points remaining.  Omitting these resulted In a five-parameter model
built on the data from 28 systems.  This unbiased model explains 75 percent of the  observed variation
in SOBC.  Five independent variables were included in this model. Ca and Mg in dry deposition and the
SKX and ACL soils were  included with negative parameter estimates. The FL and ACC soils were also
included but with positive parameter estimates.

     The soils  in both the SKX and ACL sampling classes tend to be low base status. The SKX soils
are formed in residuum and  the ACL soils are formed in either residuum or alluvium.  The presence of
these soils on a watershed is indicative of highly weathered, low base status soils, and lower base status
surface waters.   In contrast,  the soils in the FL class are relatively high base status and are  associated
with the higher base status  surface waters.  The ACC soils are very similar to the ACL soils and are
differentiated  by their particle-size families:  the  ACC soils are clayey and  the ACL soils are  either fine-
loamy  or coarse-loamy. The presence of the ACC soils may be indicative of readily weatherable primary
minerals, while the FL  soils  may represent hydrologic convergence zones where base cation enriched
drainage waters  and sediments accumulate.

     Table 8-46 includes three regression models developed  for SOBC using deposition and the soils
and miscellaneous land areas within 100 m  of the sampled stream.  The FL sampling class is included,
with a  positive parameter estimate, in the three models. The unbiased model (i.e., without outliers and/or
leverage points)  had four parameters, all with positive parameter estimates and an R2 of 0.79. Included
were the FR,  FL, MSL, and OTL sampling class soils.  All parameter estimates were  positive, indicating
that the soils in these classes are all associated with higher base status surface waters. The FR and MSL
are typically low base status soils, while FL and OTL are some of the highest  base status soils in the
region. Because of their low base status and positive correlation to SOBC,  the FR and MSL classes may
be surrogates for other watershed attributes  that supply base cations to the streams.

     The buffer zone model explains slightly more of the variability in SOBC than the whole watershed
model.  This lends support to the hypothesis that the  near-channel soils may have the greatest effects
on surface water chemistry for some variables.

     Stream pHs  in the  SBRP  are higher than the northeastern lake pHs with a central tendency near
circumneutrality.   The regression  models  for stream  pH developed with all  35  whole  watershed
observations  and with all 33  100-m buffer  observations are  identical in that  they  include the same
variables.  As with  stream ANCs and the concentrations of Ca plus Mg, stream pH is strongly influenced
by the presence of calcareous  soils. The results of the stream pH  analysis are presented in Table 8-
47.  Omitting the  two highest  OTC sites and one other  observation  with  high influence diagnostics,
produces  a model with two variables that explains 38 percent of the observed variation in stream pH.

                                              374

-------
Table 8-46.  Regression Models of SQBC* in SBRP Streams, Developed Using
Deposition, Mapped Soils (as a Percentage of Watershed Area in Soil Sampling
Classes) and Miscellaneous  Land Areas  (as a Percentage of Watershed Area)
as Candidate Independent Variables'1
Whole Watersheds
Ca+Mg-Dry
FR
OTC +
SKV
SKX
PL + +
MSL +
OIL + +
ACL
ACC +
MPQ +
R2 0.91 0.50 0.75
adjusted R2 0.90 0.44 0.70
n-streamsc 35 32 28
p-modeld 435
O/Le none 3 4
100-m buffer zones
-
+ +



+ + +
+ +
+ +


+ +
0.41 0.74 0.79
0.36 0.65 0.74
30 28 25
275
358
a SOBC = sum of base cations (Ca + Mg + Na + K)
  + and - refer to positive and negative parameter estimates, respectively
c n-streams = number of observations used to develop the regression model
  p-model = number of regressor variables in model
e O/L = number of outlier or leverage points omitted
                                               375

-------
Table 8-47.  Regression Models of Stream pH in SBRP Streams,  Developed  Using
Deposition, Mapped  Soils (as a Percentage of Watershed Area in Soil Sampling
Classes) and Miscellaneous  Land Areas  (as a Percentage of Watershed Area) as
Candidate Independent Variables"
              Whole Watersheds
100-m buffer zones
SO4-Wet
Ca+Mg-Dry
FR
OTC +
SKV
SHL +
OTL + +
MSL +
ACC
R2 0.67 0.42
adjusted R2 0.61 0.39
n-streamsb 35 33
p-modelc 5 2
O/Ld none 2
-
-
-
+ + +
-

+ + +
+
-
0.68 0.38 0.39 0.47 0.29
0.63 0.34 0.31 0.45 0.23
33 31 30 29 28
52312
none 2345
* + and - refer to positive and negative parameter estimates, respectively
  n-lakes =  number of observations used to develop the regression model
0 p-model = number of regressor variables in model
  O/L =  number of outlier or leverage points omitted
                                              376

-------
The two variables are the OTL and SHL sampling classes. After the OTC sampling class, the OIL class
has the highest base saturation  in the region.  The SHL class  consists of shallow Inceptisols and
Ultisolswith moderate base saturation (-11%) and alone only accounts for approximately 3 percent of
the variation in stream  pH.  By far, the OTL sampling class soils are more important in explaining the
variability in stream pH.

      Using only the soils and miscellaneous land areas within 100 m of the streams and the deposition
data results in a  final model with two  variables that has an  R2 of 0.29,  One of the variables is the
sampling class FR, and the other Is the  sum of the dry Ca and  Mg deposition.  Both have negative
parameter estimates. The negative parameter estimate on the FR is because, in the region as a whole,
the soils in the FR sampling class have the lowest pHs and base saturation.  The negative relationship
with  Ca and Mg  in dry deposition does  not seem reasonable.  This variable may be a surrogate for
another independent variable.

8.6.4.3  Regional Comparisons

      Because ANC, Ca  plus Mg, and pH are interrelated, the resulting regression models within each
region are often similar.  The median ANC and pH values in the northeastern lakes are lower than in the
SBRP streams even though the median Ca plus Mg concentrations are higher.  This is a  direct result
of elevated lake sulfate concentrations in  the HE.

      Results from both regions show that soils  with high  base saturation, especially those derived from
calcareous parent materials (SBRP), are associated  with surface waters that have higher values of ANC,
base cations, and pH. Both lake and stream resources are susceptible to the effects of major watershed
disturbances (e.g., quarries). When these disturbances are present, the subtending surface waters will
have elevated  base cation concentrations as well as elevated sulfate levels. In most cases the increase
in sulfate is balanced with concomitant increases in  base cations.  Therefore, the ANCs are not generally
negatively impacted  by the water  disturbance.

      In the NE,  poorly  drained (wetland), organic  soils  that are acidic appear to decrease lake pH.
This is probably due in part to organic acids from these soils.  In the SBRP frigid soils (FR sampling
class) are associated with lower pH surface waters.  These soils are low pH and low base status.  As
was observed in the northeastern sulfate  analysis above, the coastal watersheds appear to have occult
sources of deposition that result in lower surface water ANCs, pHs, and Ca  plus Mg.

      In the NE  the whole watershed  regression models generally had about the same explanatory
power as the  models developed using buffer zone data.  In the SBRP, however, the models developed
using buffer zone data usually had more explanatory power; the only exception was for stream pH. This
suggests that  in  stream  watersheds, the near-channel zones  have a greater effect  on  surface water
chemistry for some variables than  the rest of the watershed. To test this hypothesis definitively, however,
would require  finer resolution mapping  data than those obtained in the  DDRP.
                                              377

-------
8.6.5 Summary and Conclusions

      The difference in soils between the two regions accounts for most of the observed differences
seen in  sulfate and sulfur retention.  Compared to the soils of the SBRP the soils in the NE are young,
shallow, less developed, and have a lower overall  capacity to retain sulfur. In contrast, the soils in the
SBRP are, in general, deep and highly weathered, with abundant secondary mineral  phases that provide
considerable sulfur retention capacities.
      The major conclusions of this analysis are:
           In the NE, where sulfur retention on average is low, sulfur deposition explains more of the
           observed variation in lake sulfate concentrations than any other independent variable.

           In the SBRP, where the majority of sulfur inputs  are retained by watershed soils,  sulfur
           deposition is not yet significantly related to stream  sulfate concentrations.

           Local sources of sulfur  deposition from sea salt may be negatively affecting the surface
           water chemistry in the near-coastal watersheds  in Subregion 1D of the NE,

           Wetland soils or soils that are wet part of the year promote sulfur retention via sulfate
           reduction reactions.

           Shallow soils with short  hydrologic contact times and low sulfate adsorption capacities do
           not interact sufficiently with drainage waters to affect their chemistry.  Watersheds with these
           types of soils in areas of high sulfur deposition are likely to be susceptible to surface water
           acidification.

           Soils with high base saturation are associated with higher surface water ANC, pH, and base
           cations.

           Poorly drained, acidic organic soils in the NE and frigid soils in the SBRP are associated
           with lower pH surface waters.  In the NE this relationship may be due in part to the result
           of the organic acids in these soils.  The frigid soils are low pH and low  base status.

           Using  only mapped soils information from  the  near-stream areas in the SBRP to  develop
           regression models  generally produced models with  more explanatory  power  than  those
           developed using only information from the whole watershed.
      This type of analysis has been shown to be a useful tool for regionally assessing the relationships
between surface waters and soils, miscellaneous land areas, and deposition. Although these attributes
alone cannot account for all the variability in the observed data, there are some Instances in which they
do  account for  most of  it.  Care must  be exercised  in  evaluating the  resulting regression  models.

                                               378

-------
Inclusion of outliers and leverage points may result in models that are heavily biased. Sample populations
with small sample sizes are particularly susceptible to bias.

      This analysis demonstrates the utility of soil sampling classes in characterizing the soils across
large geographic regions.  It has helped us to assess the concept of soil sampling classes and may
lead to some revisions in the way classes are differentiated.

8.7  ANALYSES  OF DEPTH TO BEDROCK

8.7.1 Introduction

      One of the important findings of the Integrated Lake/Watershed Acidification Study (ILWAS) (Newton
and April, 1982; Goldstein et al., 1984) was that the depth of soil and surficial geological materials have
a significant effect on  the  quality of subtending surface  waters.  The ILWAS scientists found that the
difference in lake pH between Woods (pH 4.4-5.1) and Panther (pH 5-7.5) Lakes could be attributed to
the difference in the depths of surficial materials on these watersheds.

      In  addition  to depth, the chemical  and physical characteristics of the surficial  materials are also
important.  The latter affects the hydrologic flow path and the hydrologic contact time, which in turn
affect the length of time to react with the  drainage waters. Short hydrologic contact times produce little
alteration in the chemistry of the drainage waters.  The chemical characteristics of the surficial material
are also  important.  Materials without weatherable primary minerals will  have little beneficial effect on
acidic inputs, even though they may be deep.  Our objective in this analysis is to  test this depth-to-
bedrock  hypothesis on a regional basis.

8.7.2 Approach

      A depth-to-bedrock map was prepared during the mapping phase  of the DDRP  by  assigning a
depth-to-bedrock class to all soil map delineations.  The procedure used in the  NE was presented in
Section 5.4.1.2 and for the SBRP in Section 5.4.2.2.  In  addition to the depth-to-bedrock maps, depth
to  bedrock  was  recorded  for each soil  component in the soils database  for each region  (Sections
5.4.1.1.2  and 5.4.2.1.2).  The soils databases, therefore,  provided  an  alternative approach  to estimate
the extent of depth-to-bedrock classes on watersheds and subsequently subregions and  regions.  Using
these data rather than the data from the depth-to-bedrock maps  provides a more precise method to
estimate  the proportion of depth-to-bedrock classes.

      The depth-to-bedrock classes developed for this analysis from the soils data for the NE and SBRP
are listed in Table 8-48.  Note that the numbering of the  classes proceeds from the rock outcrop (I) to
the very  deep (VI), and that depth classes V and VI indicate deeper soils for the SBRP than for the NE.
In both regions, depth class I represents rock  outcrop on the watersheds.

      The depth of each soil (component) is recorded in the soil component file. As described in Section
8.6.2, the mapped soils are linked  to the  soil component file.  This file contains component-specific
information, including soil depth.  Using this soil depth information we calculated the percentage of each
                                              379

-------
Table 8-48.  Depth-to-Bedrock Classes for the Northeast
and the Southern Blue Ridge Province
                 Northeast Region
Class      Depth range (cm)      Definition
NE 1
NE II
NE HI
NE IV
NE V
NE VI

10 -
25 -
50 -
100
150

25
50
100
- 150
+
Rock outcrop
Very shallow
Shallow
Moderately deep
Deep
Very deep
           Southern Blue Ridge Province
Class      Depth range (cm)      Definition
SE I
SE II
SE III
SE IV
SE V
SE VI

10 -
25 -
50 -
100
200

25
50
100
- 200
- 500
Rock outcrop
Very shallow
Shallow
Moderately deep
Deep
Very deep
                                       380

-------
watershed In each of the depth categories.  These percentages are used as the independent depth-to-
bedrock variables in the following analysis.  The LTA sulfate and hydrogen deposition estimates, both
wet and dry, are also used as candidate explanatory variables.

      The descriptive statistics on the proportion of these depth classes for both NE and  SBRP are
presented in Table  8-49. In general, the NE has higher proportions of shallower soils than does the
SBRP. The proportions of deeper soils are not strictly comparable between the regions, because depth
classes V and VI are not the same across regions.

      Within  the  NE, Subreglons  1A  and  1E  have  the highest percentages of rock  outcrop, and
Subregions 1C and 1D have the lowest.  Subregion 1A has the highest percentage of very shallow and
shallow soils, while Subregions 1D and 1B have the lowest.  Subregions 1D and 1C  have the highest
proportions of the very deep soils, and Subregion 1A has the lowest proportion.

      The statistical analyses used in the section are discussed in Section 8.1.2.   Residual analysis
revealed heteroscedasticity in the residuals for ANC and  base cations for both regions.  We, therefore,
log-transformed these dependent variables in the analyses for depth-to-bedrock relationships.

8-7.3  Sulfate and  Percent Sulfur Retention

8.7.3.1  Northeast

      In the  NE,  depth to bedrock seems to have  little effect on  surface water sulfate (Table 8-50).
Surface water sulfate is  dominated by wet and dry sulfate deposition.  The positive correlation between
percent sulfur retention and dry sulfate deposition may represent a spurious correlation due to the
formulation for percent retention.

8.7.3.2 Southern Blue Ridge Province

      In the SBRP, depth to bedrock has a significant effect on sulfate dynamics. The percent of shallow
soils  (SEJII)  has a strong  positive  relationship  with  surface  water  sulfate  and a  strong negative
relationship with percent sulfur retention (Table 8-51). This suggests that, as the percent of shallow soils
increases and the percent  of deep soils decreases, the amount  of  sulfate adsorption  decreases.  This
decrease in sulfate adsorption  may be due to several factors.   The deep soils have  more  mass and,
hence, more total sulfate adsorption capacity. The deep soils may also have higher contact times and
different flowpaths for the soil water.

8.7.3.3 Comparison of Regions

      It appears that in-lake sulfate in the NE is predominantly controlled by atmospheric  deposition
and  not by the depth of surficial material. In the SBRP, the shallow (25 - 50 cm) category of depth to
bedrock accounts for about 32 percent of the variability in observed stream sulfate concentrations and
more than 40 percent of the variability in watershed sulfur retention  estimates.  These results imply that
shallow soils play an important role in regional sulfur dynamics  in the SBRP and that they are often
associated  with higher stream water sulfate concentrations and lower watershed sulfur retention.

                                              381

-------
Table 8-49.  Regional and Subregional Statistics for Percentage of
Watershed Coverage of the Depth-to-Bedrock Classes
NE
NE I
NE 11
NE 111
NE IV
NE V
NE_VI
Subregion 1A
NE I
NE II
NE Hi
NE IV
NE V
NE_VI
Subregion 1 B
NE I
NE II
NE III
NE IV
NE V
NE_Vl
Subregion 1 C
NE I
NE II
NE III
NE IV
NE V
NE VI
Average
3.3
4.2
10.8
13.2
0.4
67.8
Average
4.8
8.1
17.9
17.8
0.0
51.5
Average
3.5
1.8
7.3
20.6
0.0
66.7
Average
1.9
3.4
9.9
10.1
0.0
75.5
Median
1.6
2.3
10,3
11.2
0.0
70.0
Median
2.8
6,8
15.6
16.0
0.0
55.6
'Median
0.0
0.0
3.8
14.5
0.0
67.5
Median
1.7
2.8
10.6
9.9
0.0
73.3
Minimum
0.0
0.0
0.0
0.0
0.0
2.8
Minimum
0.0
0.0
0.0
0.0
0.0
1Z1
Minimum
0.0
0.0
0.3
6.7
0.0
2&1
Minimum
0.0
0.0
0.0
0.0
0.0
3Z7
Maximum
28.4
42.8
60.5
56.7
64.8
100.0
Maximum
18.1
21.3
34.9
36.4
0.0
100.0
Maximum
24.0
12.7
26.6
48.2
0.0
92.8
Maximum
8.6
10.0
22.6
3Z2
0.1
100.0
                                                      continued
                                           382

-------
Table 8-49. (Continued)
Subregion 1 D
NE I
NE II
NE ill
NE IV
NE V
NE_VI
Subregion 1 E
NE I
NE II
NE III
NE IV
NE V
NE_VI
SBRP
SE I
SE I!
SE~in
SE IV
SE V
SE VI
Average
2.2
0.9
4.1
5.7
0.0
86.5
Average
4.0
5.7
124
11.3
1.9
64.8
Average
1.2
0.0
10.1
23.7
14.6
50.2
Median
0.3
0.1
1.0
1.4
0.0
96.6
Median
1.6
2.0
10.8
7.6
0.0
70.0
Median
0.7
0.0
6.6
19.3
7.1
38.2
Minimum
0.0
0.0
0.0
0.0
0.0
33.2
Minimum
0.0
0.0
0.0
0.0
0.0
2.8
Minimum
0.0
0.0
0.0
0.0
0.0
126
Maximum
128
4.4
224
27.0
0.0
100.0
Maximum
28.4
428
60.5 -
56.7
64.8
10QO
Maximum
15.0
2.6
30.0
53.8
58.3
99.0
                                           383

-------
Table  8-50.   Results for NE of Regressions of Surface  Water Chemistry  on Depth-to-Bedrock
Classes and Deposition Estimates
Water
Chemistry
Variable
Sulfate


Adjusted
R2 R2
0.2723 0.2621


Variable
in Model
wet SO4 dep.
dry SO4 dep.

Regression Signif.8
Sign Level
_i_ **
+ *
Percent
Sulfur
Retention
0.1051     0.0983
dry SO4 dep.
Log(ANC+100)   0.2603      0.2446
                            wet SO4 dep.
                            dry SO4 dep.
                            NE II
                                  ***
                                  ***
                                  *
Log(Ca+Mg) 0.2481 0.2211




pH 0.3203 0.3058


dry SO4 dep. +
wet SO4 dep.
NE_V
NEJI
NE_V!
wet SO4 dep.
dry SO4 dep. +
NE II
***
***
**
**
S
***
***
**
  S = significant at 0.15 level
  * = significant at 0.05 level
  ** = significant at 0.01 level
  *** = significant at 0.001 level
                                               384

-------
8.7.4 ANC. Ca Plus Mq and pH

      In this part of the depth-to-bedrock analysis we consider the relationships between the proportion
of watershed coverage In the various depth-to-bedrock categories and the non-sulfur dependent variables.
Unlike the NE, there is  little  or  no indication that the  DDRP sample  of streams in the SBRP are
contaminated with Na from road salt or sea salt. Therefore, in addition to  considering the sum of stream
Ca plus Mg as a dependent variable, we have included an analysis of SOBC (Ca + Mg + Na + K).  Due
to the behavior of the residuals of the regressions, both ANC and base cations were log-transformed to
remove heteroscedastlclty.  One hundred (100) was added to the ANC before transforming, in order to
avoid problems in taking the logarithm of non-positive numbers.

      Wet sulfate deposition was negatively correlated with ANC, and dry sulfate deposition was positively
correlated with ANC (Table 8-50).  Wet sulfate deposition was introduced in the regression model  first,
and  it presumably represents decreases in alkalinity of the surface waters with increasing deposition of
sulfate.  The second deposition variable may be a correction to an overfilling with wet sulfate deposition,
or it  may be a surrogate for some explanatory variable not included in the model.  Since there are some
high ANC sites in Subregion 1B, it could also represent a geographic effect, as discussed in Section 8.2,
The  very shallow soils represented by NE_11 (10-25 cm) are negatively  correlated with ANC (Table 8-
50).  This results suggests that as the proportion of soils deeper than 25  cm increases, the capacity for
cation exchange increases and ANC of the surface waters increases.  It also  suggests that these soils
may have short hydrologic contact and therefore little  or  no effect on drainage water chemistry.

      A similar  behavior in the explanatory variables is seen for Ca  plus Mg.  Dry sulfate deposition  is
positively correlated with in-lake base cations. This correlation may represent increased  cation exchange
and  leaching due to acidic  deposition in a system at or near sulfur steady state.  A second deposition
variable, wet sulfate, is introduced with a negative parameter in the model  later.  As discussed previously,
this  may be  a  surrogate for some other variable or variables estimate or possibly a geographic effect.
Three of the depth-to-bedrock classes were included with negative parameter estimates. Included were
the deep (NE_V, 100  - 150 cm), the very deep (NE_VI, > 150 cm), and the very shallow (NEJI, 1  - 25
cm)  depth categories.  This result  is contrary to the hypothesis suggested by the ILWAS project outlined
in Section 8.7.1, that the deeper the surficial geologic material (i.e., the deeper the depth to bedrock) the
higher the pH,  ANC, and base  cation  status of the  surface water.  Our result  implies  the opposite
relationship, on a regional basis.  Overall, our model accounts for only about 25 percent of the variability
in Ca plus Mg.  Therefore, factors other than depth  to  bedrock are likely to account for  most of the
variability  in Ca plus Mg.

      The regression  model  for lake pH contained the same set of parameters as the model developed
for ANC.  This model, however, had a higher R2 (0.32) than the one developed for ANC (0.26).  Again,
as wet sulfate deposition increases,  surface water pH decreases.   Likewise,  as the proportion of the
watershed  with  shallow soils increases, we can expect lower lake water pH.

8.7.4.1  Southern Blue Ridge Province

      Due to the behavior  of  the residuals of the regressions, the dependent variables ANC, Ca plus
Mg,  and SOBC were  log-transformed to remove heteroscedasticity.  The regression models developed

                                              385

-------
for these transformed variables are presented in Table 8-51.  In each of these models, the depth class
SE_V (100 - 200 cm) was Included with a negative parameter estimate.  In the SOBC model it was the
only parameter Included and explained 22 percent of the observed variability.  Both the ANC and Ca plus
Mg models included a  deposition variable.  The estimate of dry hydrogen ion deposition was included
(positive parameter estimate) In the ANC model, whereas wet sulfate deposition was included (negative
parameter estimate)  in the  Ca plus Mg model.  The reasons these variables were included in these
models are unclear.

      The negative relationships between depth class SE_V (100 - 200 cm)  and ANC, Ca plus Mg, and
SOBC suggest that  this depth class represents surficial material that is highly weathered and therefore
deep, with little or no weatherable minerals.  In the SBRP,  because the soils  and surficial materials are
old and highly weathered,  unweathered primary minerals  may  be prevalent only at  the  bedrock:soil
interface, in  the saprolite.  It is reasonable to assume that  as these saprolites get farther from the soil
surface, the weathering rates (cation supply rates)  may actually decrease  because they are farther
removed from diurnal and other environmental influences.  Because of this, watersheds with abundant
deep, highly weathered soils,  will  probably be associated with lower ANC, pH, and base status surface
waters.

      The regression model for stream pH  only included the estimate of dry hydrogen deposition and
only accounts for about  15 percent of  the  observed variability in stream  pH.  Dry deposition  has a
positive parameter estimate, implying that as it increases so does stream pH.  This apparent relationship
is unreasonable; therefore, dry hydrogen deposition  is probably functioning as a surrogate for another
variable that is positively  related to stream pH.

8.7.4.2 Comparison of  Regions

      In the NE, wet and dry sulfate deposition are  important factors included in the depth-to-bedrock
regression models developed for ANC, Ca plus Mg, and pH.  In these models, the very shallow (NEJI,
1-25 cm) depth-to-bedrock categories were consistently negatively related to the dependent variables.
This Implies that as  the proportion of the watershed in the very shallow depth categories increases (i.e.,
lower proportion of  deeper material), we can expect  the ANC, Ca  plus Mg, and pH to decrease.  This
is a reasonable  result  since shallower surficial materials are generally indicative of lower base  cation
supply capacities.

      In the SBRP the  deep depth-to-bedrock category was negatively related to ANC, Ca plus Mg, and
SOBC. This result suggests that this depth class represents surficial material that is highly weathered and
deep, with little or no  weatherable minerals.  As the proportion of the watershed in this deep material
increases, we can expect ANC, Ca plus Mg, and SOBC to be lower. The regression model for stream
pH did not include any depth-to-bedrock variables.

8.7.5  Summary and Conclusions

      Depth to bedrock appears to have an Important effect on sulfate dynamics in the SBRP, but not
in the NE. An important reason for this difference is that, in general, NE watersheds are at sulfate steady
                                              386

-------
Table 8-51.  Results for SBRP of Regressions of Surface Water Chemistry on  Depth-to-Bedrock
Classes and Deposition Estimates
Water
Chemistry
Variable
Sulfate
Percent
Sulfur
Retention
Adjusted
R2 R2
0.3200 0.2966
0.4211 0.4004
Variable
in model
SEJII
SEJII
Regression Signif.a
Sign Level
j_ ***
***
Log(ANC)
0.3140     0.2667
Log(Ca+Mg)     0.2740     0.2239


Log(SOBC)       0.2202     0.1933

pH              0.1494     0.1210
dry H dep.
SE_V

SE_V
wet SO4 dep.

SE_V

dry H dep.
*
*

**
S
       = significant at 0.15 level
       = significant at 0.05 level
       = significant at 0.01 level
       = significant at 0.001 level
                                              387

-------
state, whereas the SBRP sites are not. In both regions, depth to bedrock appears to be related to cation
supply dynamics but in opposite ways.  In the NE the shallower surficial material is associated with lower
base status surface waters.  In the SBRP the deeper material is also related to lower base status surface
waters.  In the NE as the proportion of the watershed in the very shallow depth categories increases (i.e.,
lower proportion of deeper material) we can  expect the ANC, sum of Ca plus Mg concentrations, and
pH to decrease. In the SBRP it is hypothesized that the deep class of surficial material represents highly
weathered materials with little or no  weatherable minerals remaining.  As the proportion of the watershed
in this class of material  increases, lower base status surface waters can be expected.

8.8  INTEGRATED ANALYSIS OF  ALL MAPPED VARIABLES

8.8.1   Introduction

Each of the preceding sections has considered the relationship between an isolated set of independent
watershed variables and the chemistry of the subtending surface waters.  These analyses considered the
relationships of surface water chemistry to atmospheric deposition only (Section 8.2), derived hydrologic
parameters (Section  8.3), mapped bedrock geology (Section 8.4), mapped land use/vegetation (Section
8.5), mapped soils (Section 8,6), and depth to bedrock (Section 8.7). None of these attributes alone can
explain  all of the variability in the observed chemistry.  The chemistry of surface waters is the integrated
result of many interacting factors, including those just mentioned.

In this analysis we combine the data from Sections 8.2-8.7 to develop  regression models that more fully
account for the variability in the observed dependent variable data. Our objective is to identify the most
important relationships that exist between watershed physical characteristics and surface water chemistry.
In Section 8.10 we include the soil chemical and  physical data.  In the analysis presented in this section,
we do not consider any of the watershed buffer zone data.

8.8.2 Approach

The approach used in this analysis follows that described in Sections  8.1.1  and 8.1.2 with, the following
exception.    Because the  number of explanatory variables in this analysis  exceeded the  number of
watersheds in  the SBRP, Mallow's Cp statistic could not be used as a model selection criterion; Akaike's
information criterion was used instead.  After each mode! was developed we performed residual analysis
on it, checking for leverage points  and  outliers,  as well as for the standard regression assumptions as
described in Section 8.1.2.

8.8.3  Sulfate and Sulfur Retention

8.8.3.1  Northeast

In the NE there is a strong  relationship  between wet sulfate deposition and sulfate concentration in the
lakes (Table 8-52). Because the watersheds are in general at sulfur steady state, the surface waters tend
to reflect the  sulfur  chemistry of atmospheric deposition.  The negative relationship between aquatic
sulfate and precipitation indicates dilution effects caused by increased rainfall and runoff.
                                               388

-------
Table 8-52. Regression Models of Surface Water Sulfate and Sulfur Retention In the
NE Lake Watersheds Using Deposition, Derived Hydroiogic  Parameters, Bedrock
Geology Reaction Classes, Depth To Bedrock, Mapped Landuse/Vegetation, and
Mapped Soils as Candidate Regressor  Variables
Adjusted
Variable8 R2 R2
Sulfate 0.7223 0.6962
(n = 141)










S Retention 0.4710 0.4360
(n = 129)





Variable
in Model
WET SULFATE DEP.
M04
PRECIPITATION
FACTOR9
H03
FACTOR4
E06
H2O WS
M03
PERIN
TOTSTRM
REL RAT
FACTORS
I46
NE II
FACTOR12
ATKBMEAN
MAXREL
H03
Regression SIgnlf.b
Sign Level
_i_ ***
> ***
***
.}_ ***
***
.j. **
+ S
+ S
S
S
+ S
S
.J. **
***
***
***
1 ***
+ *
+ *
  n = number of observations included in regression model
b S = significant at 0.15 level
  * = significant at 0.05 level
  ** = significant at 0.01 level
  *** = significant at 0.001 level
                                      389

-------
      Anthropogenic factors also strongly affect the sulfate concentrations. The miscellaneous land area
M04 (quarries) and the soil sampling  class  E06 (made land) both represent sources  of  sulfur from
anthropogenic watershed disturbances (Table 8-52). The positive correlation with Factor 9 also represents
anthropogenic sources:  Factor 9 indicates  increasing cabin count,  urbanization,  and  quarries.  The
positive correlation between lake sulfate and Factor 4 (agricultural land, and cropland  and pasture land)
results from  some combination  of anthropogenic amendments (e.g.,  lime, fertilizers) to the soil and
preference for conducting agricultural activities on fertile soils, which are likely to have higher pH and thus
reduced anion adsorption capacities.  If these soils are  limed or amended with phosphate, displacement
of sulfate  from adsorption sites may result in  increased sulfate moving into surface  waters.

      The correlations with the soil sample class H03 indicate that reduction  of sulfate and  retention of
sulfur by wetlands (Table 8-52) are also important.  The positive relationship  with H2O_WS (the water
bodies to watershed  area ratio)  suggests that in-lake  sulfate reduction has  a greater effect on sulfur
budgets in those watersheds with high watershed to lake area ratios and long lake hydrologic residence
times  (see Section  3.3.7.2).    An   alternative explanation  could  be  that   this  relationship  reflects
lessopportunlty for precipitation to contact soils and hence more control of sulfate concentration by the
deposition.

      The  first variable selected by the stepwise regression for sulfur  retention  is Factor 5,  which
represents wetlands  (Table 8-52).   This  correlation  reiterates  the  importance  of wetlands  in the
biogeochemical sulfur cycle on  a regional basis.  The same rationale applies to the inclusion of soil
sampling class H03, a wetland soil, with a positive parameter estimate.

      The very shallow (10-25 cm), NEJI, depth-to-bedrock class was included in the sulfur retention
model with a negative parameter estimate.  This result implies that as  the proportion of watershed
coverage  in very shallow surficiai materials increases, watershed sulfur retention  decreases.   This is
apparently due to a  concomitant decreased watershed  sulfate adsorption  (sulfur  retention) capacity.
Alternatively,  NEJI is highly correlated with the presence of both H01 and ฃ05 on a watershed.  H01 is
a  relatively dry Histosol in which mineralization of organic matter and consequently sulfur production
could occur.  The H01  soils are  usually associated with high  elevations and may be indicative of cloud
interception (i.e., increased deposition).  Factor 12 (rainfall and  runoff) was  included with a negative
relationship with  retention (Table 8-52), suggesting a dilution effect due to increased runoff.

8.8.3.2 Southern Blue Ridge Province

      The first variable selected by the  stepwise regression procedure is SEJII, in the regression model
of stream sulfate concentration (the shallow depth class) (Table 8-53). This  is indicative of the overall
reduced sulfate adsorption capacity of  shallower soils.  The two depth classes SE_VI (very  deep soils)
and  SE_V (deep soils) which are  brought  in  later  in the  stepwise  procedure  probably represent
corrections to the overfilling of SEJII in the regression.  Since the amount of adsorption is  not  linearly
related to the proportion of shallow soils on a watershed, it is reasonable that the extents of SE_VI and
SE_V are  used to correct for the regression estimate for SEJII.

      Runoff has a negative relationship with stream  sulfate  concentration, indicating a dilution effect
from increased precipitation.   The sample class  MSL has a positive relationship  with  stream  sulfate

                                               390

-------
concentration (Table 8-53). This same sample class also has a negative relationship with sulfur retention
in the first two SBRP sulfur retention models.  Considered together, these results suggest that MSL may
be  related to or indicative of sulfur-bearing  parent  material.   This relationship  was  also  noted and
discussed in Section 8.6.  Factor 8 (open land and pasture) has a positive relationship with stream sulfate
concentrations, suggesting that anthropogenic additions or the activities of livestock are impacting stream
sulfate concentrations as discussed in the previous section.

      The sample class ACG has a negative relationship  with stream sulfate concentration.  Soils in this
sample are derived from acid crystalline parent materials.  They are clayey and  have high  sulfate
adsorption capacities.   Thus, we expect the extent of these soils to be negatively related to In-stream
sulfate. MAX, the maximum bedrock sensitivity number  on a watershed,  Is negatively related to sulfate
in the  subtending waters  (Table 8-53).  In the SBRP this relationship is expected  because deeper soils
are associated with more extensively weathered parent materials, which in turn results  in increased
amounts of iron and aluminum oxides, the principal sites of sulfate adsorption. More weatherable bedrock
produces more of the deeper, finer textured soils abundant in  iron and aluminum.

      In the three successive  models for sulfur retention, we see  explanatory variables similar to those
selected for the stream  sulfate concentration regression models. The first model for sulfur retention was
a model developed  with 32 SBRP watersheds.  The residual analysis identified one watershed as a strong
leverage point due to its unusual  negative sulfur retention and the singular presence of MPQ  (quarry) on
the site.   In the second model another  site was  identified  as  an  outlier and was also excluded.  This
watershed  also appears to have an internal source of sulfur.  The variables that appear in the first two
models and not in the third model are probably site-specific or are included due to correlations with other
variables.

      The miscellaneous land  class MPQ (miscellaneous pits and  quarries)  is negatively correlated with
sulfur  retention in the first  model  (Table 8-53), indicating an internal source  of sulfur, as  previously
discussed. The sample classes MSL and FL are also negatively related to percent sulfur retention in the
first two models  (Table 8-53).   This  result  may indicate that one  or both of these sample  classes
occasionally has weatherable sulfur-bearing parent material. The soils in the FL sampling class have low
sulfur retention (adsorption) capacities.

      The depth-to-bedrock classes SEJII and SE_VI are negatively related to sulfur retention in all three
models, and  SE_V  is also negatively  related to retention in the final model  (Table 8-53).  As discussed
previously, SEJII may be indicative of the lower capacity of shallower soils to adsorb sulfate,  and the
signs for SE_V and SE_VI suggest that they appear in  the regressions as  nonlinearity corrections for
overestimating the regression  parameter for SEJII.

      The bedrock  geology variable H5up is positively related to sulfur retention in the first two  models,
but not in the third  (Table 8-53).  This result indicates, as does the negative relationship between sulfate
concentration and MAX. that  more weatherable bedrock geologies tend to produce deeper soils with
higher sulfate adsorption capacities.  HSup does not appear in the 30-observation regression model.
                                               391

-------
Table 8-53.  Regression Models of Surface Water Sulfate and Sulfur Retention in
the SBRP  Stream Watersheds Using Deposition, Derived Hydrologic Parameters,
Bedrock   Geology    Reaction   Classes,   Depth   To   Bedrock,    Mapped
Landuse/Vegetation, and Mapped! Soils as  Candidate Regressor Variables
Variable8 R2
Sulfate 0.8496
(n = 31)








S Retention 0.9297
(n = 32)






S Retention 0.6893
(n - 31)





Adjusted Variable Regression
R In Model Sign
0.7744 SE III +
RUNOFF
MSL +
SE VI +
FACTORS +
ACC
REL RAT
SE V +
MAX
SKX
0.9052 MPQ
MSL
SE 111
SE VI
DRY SULFATE DEP. +
H5UP +
DRY H DEP.
FL
0.5948 MSL
SE III
SE VI
DRY SULFATE DEP. +
H5UP +
DRY H DEP.
FL
Signlf.b
Level
***
S
***
***
*
S
**
**
*
S
***
***
***
S
**
**
*
S
***
***
S
**
**
*
S
S Retention
(n=30)
0.5835
0.5168
 SE  11
SE VI
SE V
ACC
***
**

S
S
  n = number of observations included in regression model
  S = significant at 0.15 level
  * = significant at 0.05 level
  ** = significant at 0.01 level
  *** = significant at 0.001 level
                                      392

-------
      The first two  models suggest some possible indicators of internal sources of sulfate  (e.g., MSL,
MPQ), and the final model indicates the Importance of soil depth and soil type. The sample class ACCIs
positively related to sulfur retention, indicating that very clayey soils, derived from acid crystalline parent
materials, are strong adsorbers of sulfate.

8.8.3.3  Regional Comparisons

      In the NE, the surface water sulfate concentrations are strongly affected by  sulfur deposition.  In
the SBRP,  however, the watersheds are not at sulfur steady state, and hence do  not mirror trends in
deposition as readily as the northeastern sites do.  In both regions, watershed disturbance and agricultural
practices may outweigh the  effects of deposition  on surface water chemistry.  Some soils have distinct
relationships with stream sulfate concentrations and watershed sulfur retention and may be indicative of
internal watershed sulfur sources.  The northeastern lakes display more obvious effects of wetlands than
do streams in the SBRP, where extensive wetlands are  relatively uncommon.  Effects due to soil depth
and  bedrock geology are more pronounced  in  the SBRP.  In the NE, sulfur "retention  seems  to be
primarily controlled  by extent and type of wetlands.  In the SBRP, sulfur retention is  controlled by the soil
mass (i.e.,  oxyanion adsorption capacity) available to adsorb sulfate and the extent of types of soils that
adsorb more strongly.

8.8.4 ANC. Ca plus Mq. and pH

8.8.4.1  Northeast

      The  regression model for ANC in northeastern watersheds  indicates that surface water ANC is
primarily driven by  watershed-specific variables.  The first variable in the model is  Factor 4, the extent
of pasture  and cropland in  the watershed,  which has  a positive  relationship with surface  water ANC
(Table 8-54).  This relationship may represent an increase in base cations through the  use  of soil
amendments (i.e., lime), and  It may also be  indicative of the selection of high base status soils for
agricultural activities.  The soil sampling class I46 was also included  with a positive relationship with
ANC  (Table 8-54),  which Is expected, because the soils  in this sampling class are  high base  status.
Their typical base saturation is over 75  percent, and their average pH is the  highest of all northeastern
soil sampling classes.  Factor  12 (precipitation and runoff) has a negative relationship with surface water
ANC  (Table 8-54), indicating a dilution  effect in the surface waters produced  by increased runoff.

      MAX. the highest value of the bedrock sensitivity code on a watershed, has a positive  relationship
with ANC (Table 8-54).  The higher bedrock sensitivity numbers are associated with lithologies, such as
carbonates, that can buffer soils and surface water against changes in ANC.

      The first variable in the  regression model of Ca plus  Mg is  Factor 4 (pasture and cropland), as
in the model for ANC (Table 8-54).   Again,  this result probably indicates  preference for  agricultural
development on  higher base status  soils and  the  introduction of soil amendments.    Factor 12
(precipitation and runoff) again  has a  negative relationship with surface water  ANC  and Ca plus  Mg,
indicating a chemical dilution.
                                               393

-------
Table 8-54. Regression Models of Surface Water ANC, Pa plus Mg, and pH In the
NE Lake Watersheds Using Deposition,  Derived Hydrologic Parameters, Bedrock
Geology Reaction Classes, Depth To Bedrock,  Mapped Landuse/Vegetation, and
Mapped Soils as Candidate Regressor Variables
Adjusted
Variable8 R2 R2
ANC 0.4860 0.4666
(n = 138)



Ca+Mg 0.5877 0.5662
(n = 142)





pH 0.4621 0.4383
(n = 143)




Variable
in Model
FACTOR4
146
FACTOR12
ATKBMEAN
MAX
FACTOR4
FACTOR12
I46
M04
125
M01
H01
FACTOR4
FACTOR12
WET SULFATE DEP.
H03
FACTOR1
DRY SULFATE DEP.
Regression Signif.b
Sign Level
+ ***
+ ***
**
+ **
4- *
+ ***
***
_(. ***
_l_ ***
_j. ***
+ ***
*
4. ***
*
***
***
_j_ **
4- *
* n = number of observations included in regression model
b S = significant at 0.15 level
 * = significant at 0.05 level
 ** = significant at 0.01 level
 *** = significant at 0.001 level
                                      394

-------
      The soil sampling classes 146 and 125 are positively related to Ca plus Mg (Table 8-54).  Both of
these sampling classes have soils with high base status and high pH.  Soil-water flowing through these
soils would be expected to have more exchange of acid cations for base cations and to contribute base
cations to the surface waters. The miscellaneous land area M04 is also positively related to surface water
Ca plus Mg (Table 8-54).  M04 (pits, quarries) has been shown previously to be related to increases in
lake  and stream sulfate concentrations (Section 8.6 and 8.8.3).  Apparently base cations accompany
sulfate to the surface waters.

      The sample class H01 is negatively related to Ca plus Mg  (Table 8-54).  The soils that make up
this class are thin mantles of organic material overlaying rock outcrop or rock fragments.  These soils
generally occur on  steep  mountain slopes at high elevations. Precipitation falling on these soils flows
rapidly from these areas to drainage ways that feed directly into the surface waters.  The lower mass
available for cation  exchange and the reduced soil contact could account for the lower base cations in
the surface waters.

      Surface water pH is positively related to Factor  4 (pasture and cropland) and negatively related
to Factor 12  (precipitation and runoff), as are  ANC  and  Ca plus Mg  (Table 8-54).  As  discussed
previously, Factor 4 probably reflects preference for agricultural development on higher  base status soils
and  Introduction of soil  amendments.   Factor 12 again  indicates the  dilution effect  of  increased
precipitation. Wet sulfate  deposition  has  a negative relationship with surface water pH (Table 8-54),
indicating the ability of increased sulfate deposition to  lower the pH  of subtending surface waters.

      The soil sampling class H03  has a  negative relationship with lake pH  (Table 8-54).  The  soils in
this soil sampling class are dysic (low pH) wetland soils.  By definition, they have low pH, typically less
than 4.5.  These soils may be contributing organic acids and thus affecting the pH of the surface waters.
Also, these soils can be the last soil that the drainage waters pass through before reaching the lake.
Because of their position in  the watershed, these soils may therefore  have a significant ultimate effect on
surface water chemistry.

      The model developed for lake pH includes Factor 1 with a  positive parameter estimate. Factor 1
represents developed land.  This factor Incorporates effects due to waste disposal sites, pits and quarries,
cabins, urban commercial land, and urban residential land.  The  positive relationship with pH (Table 8-
54) may indicate that the  relationship is primarily driven by the base cation influx associated  with some
anthropogenic  disturbance, such as pits or quarries.  The positive  relationship between surface water
pH and dry sulfur deposition suggests that this explanatory variable is a surrogate for some other factor.

8.8.4,2 Southern Blue Ridge Province

      The three SBRP sites with high ANC are excluded in order to make the analysis more comparable
with the results from the NE and to adhere to the DDRP design. If these sites were included the squared
correlations for ANC and Ca plus Mg would have been over 90 percent, and the squared  correlation
for pH  would have been  over 70  percent.  These increases  in explained variability are all due to the-
presence of highly weatherable bedrock with large amounts of carbonates and calcareous soils on these
watersheds.
                                               395

-------
      The model for surface water ANC shows a positive relationship with the sample class FL and a
negative relationship with runoff (Table 8-55).  The sample class FL is composed of flooded soils that
are near the stream channels and have fairly high base status and pH compared to the other soils in
the region. These soils may be the last soil that the drainage waters pass through before reaching the
surface waters. In the near-stream channel position, these soils may have a significant effect on stream
chemistry.  The negative relationship with runoff suggests chemical dilution.

      Two regression models were developed for Ca plus Mg in SBRP streams. The first was developed
with 32 observations. The second is based upon 29 observations after residuals analysis of the first
model identified one outlier and two leverage points. In both models, the sample class FL has a positive
relationship with stream base cations (Table 8-55), as it does with ANC.  Runoff is negatively related to
stream base cations (Table 8-55).  Both of these have been discussed previously.

      In the first Ca  plus Mg  model, HSup is positively  related to base cations.  The higher values of
the sensitivity  scale are associated with carbonate bedrocks that weather more easily and contribute
base cations to drainage waters.  In both models for Ca plus Mg, there is a positive relationship with
Factor 3 (Table 8-55). Factor 3 represents larger proportions of cropland, land under horticulture, and
open land as defined by the SCS. As in the NE,  this result may  reflect the impact of soil amendments
on surface water chemistry, and it may indicate that agriculture is conducted on fertile, high base status,
flood plain soils.  In the second  model,  Factor  4  (open land, urban development, and wetlands) is
positively related to  stream base cations.  The  relationship with SCS open  land and  development
indicates anthropogenic sources of base cations to the surface waters.  DDENS1TY, a measure  of
drainage density, is negatively related to stream base cation concentration. Higher drainage densities
usually indicate a faster runoff response and, hence, lower soil interaction. With less soil contact, the
base cation supply would tend to  be lower. In the second model, wet sulfate deposition is negatively
related to base cation concentration. This may be a surrogate for  increased precipitation and thus might
represent a dilution effect.

      Unlike the NE, there is little or no indication that the DDRP sample of streams in the SBRP is
contaminated with sodium (Na) from road-salt or sea-salt additions. Therefore, in addition to considering
the stream Ca plus Mg as a dependent variable,  we have included an analysis of the sum of the four
principal base cations, Ca + Mg  + Na + K,  (SOBC),  The regression model developed for SOBC
explains  about 92 percent of the observed variation in SOBC  and contains three highly significant
variables with positive parameter estimates.  These are Factor 3 (larger  proportions of cropland,
horticultural activities, or open land), SE_VI (very  deep depth-to-bedrock category,  200 - 500 cm), and
the OTL soil sampling class. As mentioned previously, Factor 3  is indicative of the preference of  high
base status soils for agricultural purposes, which tend to be located near streams in the flood plain.  In
conjunction with this, soil amendments may result in increased surface water base cation concentrations.
The very deep depth-to-bedrock category (SE_VO 's synonymous with near channel, flood plain  soils.
These zones are also where base cation enriched drainage waters and sediments accumulate. The soils
in the OTL soil sampling class are generally very high base status soils, and are therefore associated with
higher base status surface waters.

      In the model for surface water pH, there is a negative relationship with runoff (Table 8-55),  The
soil sampling class FR is negatively related to  stream pH (Table 8-55).  The sampling class FR  is
composed of the frigid soils, which have the lowest pH and base status soils in the region.  When they
                                             396

-------
Table 8-55.  Regression Models of Surface Water ANC, Ca plus Mg, and pH in the
SBRP  Stream Watersheds  Using  Deposition,  Derived  Hydrologic  Parameters,
Bedrock   Geology   Reaction    Classes,    Depth   To    Bedrock,   Mapped
Landuse/Vegetation, and Mapped Soils as Candidate Regressor Variables
Adjusted
Variable8 R2 R2
ANC 0.4531 0.4154
(n = 32)
Ca+Mg 0.6714 0.6227
(n = 32)


Ca+Mg 0.7101 0.6471
(n = 29)



SOBC 0.9285 0.8927
(n = 25)






pH 0.4312 0.3470
(n-32)


Variable
in Model
FL
RUNOFF
FL
RUNOFF
H5UP
FACTORS
FL
FACTORS
FACTOR4
DDENSITY
WET SULFATE DEP.
FACTORS
SE VI
OTt
WET SULFATE DEP.
WET H DEP.
ACC
MSL
SKV
RUNOFF
FR
STRMORD
DRY H DEP.
Regression Signif.b
Sign Level
4. ***
**
i **
**
i ***
+ *
_j. ***
_i_ **
+ S
*
S
_J. ***
_J. ***
1 ***
**
+ *
*
+ *
S
*
*
+ *
+ S
 n = number of observations included in regression model
b S = significant at 0.15 level
 * = significant at 0.05 level
 ** = significant at 0.01  level
 *** = significant at 0.001 level
                                      397

-------
occur, the drainage water pH would be expected to be reduced.  Also, the proportion of these soils is
negatively correlated with the proportion of FL, the flooded soils with high base status and high pH.

     STRMORD, the maximum Norton stream order on the watershed, is positively  related to stream
pH  (Table 8-55),  Larger values of stream order tend to be associated with larger watersheds.  These
sites have longer flow paths and more soil contact, which would elevate the pH of the drainage waters.
Dry hydrogen deposition is positively related to stream pH in the regression model (Table 8-55).  The sign
of the relationship indicates that dry hydrogen deposition is probably a surrogate for another variable.

8.8.4.3  Regional Comparisons

     In  both regions,  watershed-specific factors appear  to be more important  than atmospheric
deposition on the base status and pH of  surface water.  The effects of bedrock lithology and presence
of agricultural land appear across both regions.  The base status of soils and their  contact time also
affect the surface water ANC.

8.8.5 Summary and Conclusions

The  specific conclusions of these analyses are:

      •     The  effect of deposition on surface water chemistry is much more distinct in the NE than
           in the SBRP.

      •     Major watershed disturbances, such as quarries and urbanization, result in increased surface
           water sulfate concentrations.   They also produce higher base status surface waters.

      •     Land use, especially  near-lake or near-stream agricultural activities (e.g.  lime and fertilizer
           amendments) may outweigh the effects of deposition  on  surface water chemistry.

      •     in the NE, wetland  soils are associated with sulfur retention.

           Shallow  soils are negatively related to sulfur retention in both the NE and SBRP. This is
           probably caused by their decreased  capacity to adsorb  sulfate.

      *     In the SBRP,  easily  weathered parent  materials produce  abundant iron and  aluminum
           .oxyhydroxides.  Soils formed  in these types of parent materials are usually deep and have
           large sulfate adsorption capacities.

      •     In the SBRP, the very deep depth-to-bedrock category  of surficial material is synonymous
           with near channel,  flood  plain soils.   These zones are  also where  base cation enriched
           drainage waters and sediments accumulate.  These  zones are therefore associated with
           higher ANC surface waters.
                                              398

-------
8.9  SOIL PHYSICAL AND CHEMICAL CHARACTERISTICS

8.9.1 Introduction

      This section evaluates the relationships between surface water chemistry and the soil physical and
chemical characteristics that were measured by the analytical laboratories as part of the DDRP soil survey.
Section  2 outlined  the  hypothesized  basis for control of surface water chemistry by soil  chemical
characteristics, i.e., sulfate retention and base cation supply. Section 3 discussed the influences of these
soil  chemical characteristics in greater detail.   This  section uses an  empirical approach to  evaluate
whether  the  hypothesized  mechanisms of soil  chemical  influence on  surface water chemistry  are
supported by relationships between measured soil chemical and physical data and water chemistry data.

      Relationships between soil characteristics and surface water chemistry are evaluated in this section
using btvariate correlations and multiple regressions. The dependent variables are discussed in Section
8.1.2,

8.9.2 Approach

      The candidate independent or explanatory variables considered in this section are those that were
measured at  the soil  analytical laboratories  on soil samples taken  during the DDRP soil survey.  A
complete list of the measured physical and chemical characteristics was given in Table 5-22.  Summary
statistics for the subset of those variables that were used in this section are  given in Section 8.9.4.

      The soil samples  analyzed  by the DDRP were from individual subhorizons  of pedons"sampled
randomly from  areas  of occurrence of predefined  sampling classes as described in Section 5.5.  As
many as seven and as few as  zero pedons were sampled on each watershed.  In order for the data to
be used in the empirical analyses, they were aggregated through the sampling class framework and single
values calculated for  each watershed according to the mass  and the area of occurrence (from  the
mapped  data, Section 5.4) of each sampling class on each watershed.  This procedure is  described in
more detail in Section 8.9.3.

      There are questions about how data  should be aggregated from single points in a heterogeneous
watershed or landscape to represent the entire study area. The related issues have been discussed in
detail by Johnson et al. (1988b).  For these Level I analyses the aggregation should yield a value that
is representative of the  soils that  influence the chemistry of water draining into the lake or stream as
measured by the index  sample.  The  Index sample (defined  in  Section 5.3) represents water that  has
passed through the watersheds over different time periods and along different flow paths.  For example,
some portion of the water in northeastern lakes passed slowly through the deeper soils of the watersheds
and  entered the lakes or streams draining into the lakes as baseflow; another portion flowed rapidly
through shallow soils as quickflow draining directly into the lakes or streams because the deeper soils
were saturated. Thus, under some hydrologic conditions, characteristics of the deeper soils on much or
all of each watershed might be relevant; under spring runoff,  frozen, or storm conditions, the attributes
of the shallow soils  or soils closest to the lakes or  streams might be more important. Since the SBRP
stream samples were collected during baseflow conditions, the influence of shallow hydrologic flowpaths
                                              399

-------
should be relatively less important than In the HE, and characteristics of deeper soils over most of the
watersheds should be most relevant.

      The soils data have been aggregated two different ways to evaluate whether the characteristics
of soils over the entire watershed or soils closest to the lake or stream are more closely associated with
the surface water chemistry.  The first aggregation  results in watershed  values, weighted by area of
occurrence of  each sampling class,  representing all of the soils  on the watershed.  The second
aggregation results in watershed values representing the area of occurrence of each sampling class
within  mapped  buffer zones around  the lakes and streams.  The development  of the buffer zones is
discussed in  Section 5.4.1.7.5.  The aggregation procedures are discussed in more detail  in Section
8.9.3.

      The concept of capacity and intensity variables needs to be considered in these Level I analyses.
Capacity variables Include the  pool of  exchangeable calcium, cation exchange capacity,  or  sulfate
adsorption capacity,  for example.  They represent either pools of available ions that may be exchanged
for other ions in solution or sinks that may remove Ions from solution.  The size of these pools or sinks
determines how long a process such as base cation leaching or sulfate adsorption can occur before the
pool or sink  is depleted and  other processes begin to occur.   Intensity  variables, such as  pH, base
saturation, and  equilibrium soil solution sulfate concentration, represent concentrations of ions that are
readily exchangeable and that quickly reach equilibrium with water in contact with the soil.   In the
absence of in-stream or in-lake changes and deposition directly to  the stream  or lake, surface water
should reflect the values of the intensity variables of the soil with which it was last In contact. For the
correlation and  regression analyses presented here, both capacity and intensity variables were selected
as candidate variables to evaluate the importance of each in relationships with the index chemistry
variables.

8.9,2.1  Statistical Methods

      A multiple linear regression modelling approach was used to estimate the value of a response or
dependent variable as a linear function of a  set of predictor variables.  Figure 8-9 illustrates the steps
used  to  develop the regression models.  This  section provides a brief summary  of the modelling
approach.

      The DDRP database  contains information  on 145 lake  watersheds In  the NE and  35 stream
watersheds in the SBRP.  Prior to regression analysis, the distributions of the selected dependent (i.e.,
surface water chemistry) variables were examined for obvious outliers.  Based on this examination, two
northeastern  watersheds with  high  lake  sulfate  concentrations were dropped.   In the  SBRP three
watersheds were eliminated due to  high stream  alkalinity  and an additional watershed was removed
because of high sulfate.  Each of the watersheds deleted due to high  sulfate concentrations had open
pits or quarries on a small  portion of the watershed.  The three SBRP watersheds  had ANC > 1200
fj&q L"1 probably due to the presence  of carbonate bedrock.

      Candidate explanatory variables were chosen in a two-stage procedure. First, explanatory variables
common to both the NE and SBRP were  selected to facilitate comparison of the regression models for
the two regions. This selection was based on hypothesized  relationships and nonparametrie correlations

                                               400

-------
    f    Enter   j-
                                   Eliminate outlier watersheds
                                   Select variables based on
                                 hypothesized relationships and
                                       correlation analysis
                                 Candidate explanatory variables
                                 Perform collinearity diagnostics
                                       Reduced candidate
                                      explanatory variables
                                Perform multiple linear regression
                                  analysis and model evaluation
Figure 8-9.  Model development procedure.
                                               401

-------
between the dependent and predictor variables. At least one soil base cation, pH, sulfate, aluminum, and
particle size variable was included in the initial set of predictors.  Ammonium  chloride-extracted base
cations were selected  over the ammonium acetate extractions to be consistent  with the Level II and III
Analyses.  Concentrations were used instead of pools because of the collinearity  introduced into the pool
estimates  when multiplying the concentrations by a common mass.

      In building and interpreting multiple regression models  ft is important to recognize  that few
Independent, i.e.,  explanatory, variables  in watersheds are statistically independent.   Soil  pH,  base
saturation, and  exchangeable calcium are usually correlated with each other,  as are total carbon,
extractable aluminum, extractable sulfate, and  sulfate  isotherm  parameters.   Candidate variables were
selected from the list in Table 5-22 to eliminate highly  correlated variables (those  having  |rj > 0.95).

      The  second  step in variable selection  used the collinearity diagnostics from the regression
procedure (REG) in SAS (SAS Institute, Inc., 1988) to  identify highly collinear predictor variables in the
initial set.  When a predictor variable is nearly a linear combination of other predictor variables, parameter
estimates  for these variable coefficients are unstable and have high uncertainty (Draper and Smith, 1981).
The collinearity diagnostics available In the REG procedure test for near-linear  dependencies among sets
of predictors.  The intercept  was not included in  the  analysis  because zero values for the predictor
variables were generally not within the range of the data (Freund  and L'rttell, 1986).  The diagnostics were
applied "rteratively to the initial set of predictors.   At each step, the maximum condition number was
examined and "rf ft exceeded 30,  one of the identified collinear variables was dropped.  Preference was
given to keeping a collinear variable that (1) was more mechanistic, i.e., potentially  causal, than  other
collinear variables; (2)  was considered a more reliable measure;  and (3) was the only remaining variable
of its type (e.g., hydrologic, deposition, vegetation) (Hunsaker et a!.,  1986a).  Stepwise regression was
then performed, as described in Section 8.1.2.

8.9.3  Aggregation of Soil Data

8.9.3.1  Introduction

      Previous  regional analyses of relationships between watershed characteristics  and surface water
chemistry in areas with varying levels of acidic deposition have generally  been data limited so aggregation
within watersheds was not an issue (e.g.,  Rapp et al., 1985; Nair 1984;  Eilers et  al., 1983).  Hunsaker et
al. (1986a), however,  used more intensive soils data and maps for the  Adirondacks and found that
different aggregation procedures resulted in different associations between soil characteristics and surface
water chemistry.

      There are no universally accepted or generalized procedures for aggregating watershed components
to obtain  a  weighted watershed average  or  characteristic value.  Therefore, there is a variety  of
aggregation procedures that might satisfy the objective of the Level I Analyses.

      One issue considered In aggregating data for modelling relationships between  soil chemistry and
surface water chemistry was the distinction between intensity and capacity variables. Water chemistry
at any point In time is controlled by intensity variables such as soil  pH, base saturation, or aluminum
solubility.  The effect of intensity variables on water chemistry is dependent on the relative cross-sectional

                                               402

-------
area of the soil through which water flows just prior to emerging as surface water.  Therefore, aggregation
of intensity variables should give greater weight to that portion of the soil last encountered  by the water.
Because of the difficulty in quantifying lateral versus vertical flow through watershed soils, we have not
succeeded in defining an ideal aggregation scheme for intensity variables, and the method for capacity
variables (below) was used.

      Changes in water chemistry over time are dependent on capacity variables  such as soil cation
exchange capacity, amounts of weatherable minerals present, or amount of soluble aluminum present.
Unlike intensity variables, the effect of  capacity variables is proportional to the mass of soil which the
water contacts before emerging as surface water. Consequently, the capacity variables were aggregated,
weighting by the mass of soil contacted by the water.

      Because  aquatic  chemistry  represents the  integrated response of  an entire watershed,  one
aggregation  approach  was to define the  watershed-level quantity as a weighted  combination  of the
sampling  classes that  occur on the entire watershed.  This weighting scheme used  the  percentage
composition of the watershed in terms  of the sampling classes (i.e., each sampling  class was weighted
by its area! fraction on the watershed).

      Another aggregation approach was to consider only those  soils in the  immediate vicinity of the
lake or stream.  Physical and chemical characteristics of the soils in these zones might exhibit a much
stronger relationship  with water chemistry than the aggregation of all watershed soils.  For the NE a
combined buffer around the lake and streams draining into the lake was delineated; in the SBRP the area
within a 100-m zone  along each side of the stream was determined (see Section 5.4.1.2 for details on
development of the buffers).  In both regions the  sampling class composition of the buffers was calculated
and the areal fractions were used as weights in calculating aggregated soil chemistry data for  the buffers.

      It should be noted that there  are a number of approaches in addition to the  two described here
that could be used to obtain an aggregated watershed estimate.  Possible approaches include weighting
by hydrologic group, bedrock type,  or vegetation type.  However, given the sample  design used  for the
DDRP, the aggregation approaches used for these Level I Analyses all involve weighting by  the area of
the sampling classes on all or part of the watershed.

8.9.3.2  Aggregation  of Soil Data

      Extensive discussion  among  the  DDRP  investigators  resulted  in the formulation of  a common
aggregation  approach that appeared to be  applicable for each level of analysis (Johnson  et al.,  I988b)
This approach was to

      (1)   weight each horizon by its  mass per unit area [thickness x bulk density x (1 - coarse
           fragments)] to obtain a mass-weighted average for each pedon,

      (2)   weight the  pedon values by their  mass  per unit area to obtain  a sampling  class
           weighted average, and
                                              403

-------
      (3)   weight the sampling  class value by the product  of  mass per unit area and  areal
           proportion of the sampling class on the watershed to obtain a watershed-weighted
           average.

Mass weighting was necessary for capacity variables (e.g., cation exchange capacity, sulfate adsorption
capacity) because these variables represent the amount of soil potentially available to react with acidic
deposition.   Mass weighting was also used for  aggregation  of intensity variables  (e.g.,  pH, base
saturation) because a more appropriate method was not obvious.

      Coefficients for sulfate isotherms describe  the  partitioning of sulfate between adsorbed and
dissolved phases within the soil.  Because the coefficients are derived from a function fitted to a set of
observations, the techniques  used to obtain watershed estimates for these coefficients differ from  the
aggregation methods described above.

      The procedure involved fitting the extended Langmuir equation to isotherms for individual samples
using a nonlinear least squares routine.  Estimates of net adsorbed sulfate at a set of reference points
were obtained for each sample using the fitted function, and these estimates were mass weighted to
sampling class.   An  isotherm  was fit to the sampling class values, and net adsorbed sulfate was
estimated at the set of reference points.  The net adsorbed sulfate values generated using the sampling
class isotherm coefficients were aggregated for each watershed, using the product of the sampling class
mass and the areal fraction of the sampling class on the watershed as a weight.   Finally, an isotherm
was fit to the watershed estimates and the coefficients were derived from  the fitted function.

8.9.3.3  Assessment of the DDRP Aggregation Approach

      There are several assumptions  Inherent in the sampling class approach to soil characterization
described In Section 5.5.1.  One important assumption  is that soil components within a sampling class
are sufficiently similar so that any sample from a particular class can be used to  characterize that class.
A consequence is that there may be a significant sample location effect that could inflate the estimate
of the sampling class variance.  The following two sections  describe procedures for evaluating  the
occurrence and importance of watershed effects.  Additional discussion of  these results can be found In
Turner et at, in review.

      The sampling class definitions grouped soils having  similar taxonomy or physical properties with
the assumption that chemistry of soils in a sampling class would also be similar.  Comparison of  the
variance within sampling  classes to the variance between sampling classes, as estimated by a variance
components analysis, revealed  that for most soil variables the within-class variance was equal  to or
greater than the between-class  variance (Table 8-56).  Subsequent aggregation to watersheds resulted
in very little variance among watersheds, i,e., the watershed values for most chemical parameters were
very similar for most watersheds.  The significance of  this result  depends on the spatial scale of  the
variation.  If the observed within-sampling-class variance occurred on the  scale of meters (i.e., as if all
pedons for a sampling class were sampled on the same watershed), then the sampling class aggregation
scheme has accomplished a desirable smoothing of the data and it would appear that soils In the DDRP
regions are fairly uniform, especially in the SBRP.   If, on the other hand, the observed within-sampling-
class variation occurred on the scale of kilometers, then aggregating through sampling class to watershed

                                              404

-------
Table 8-56.  Standard Deviations Within and Among Northeast Sampling
Classes Estimated from B Master Horizon Data.
Variable8
SAND
CLAY
FRAG
AC KCL
CA_CL
SBC CL
BS_CL
CEC_CL
AC_BACL
PHJ)1M
PHJH2O
C_TOT
N_TOT
S_TOT
AL_KCL
AL_PYP
AL_CD
SO4_H2O
SO4~PO4
SO4 Blb
SO4_B2b
SO4_XINb
SO4_SLPb
Within
Replicate
5.437
1.557
6.747
0.155
0,213
0.155
0.121
0.095
0.156
0.139
0.144
0.270
0.310
0.152
0.162
0.142
0.135
0.118
0.173
Within
Sampling
Classes
12.861
4.032
12.221
0.359
0.552
0.402
0.354
0.228
0.295
0.369
0.416
0.327
0.305
0.266
0.423
0.265
0.259
, 0.200
0.338
0.247
0.187
0.228
0.307
Among
Sampling
Classes
15.642
4.752
9.184
0.326
0.538
0.409
0.346
0.214
0.310
0.272
0.370 *
0.375
0.305
0.195
0.367
0.274
0.286
0.219
0.381
0.212
0.055 *
0.228
0.297
Percent Variation
Explained by
Sampling Class
60
58
36
45
49
51
49
47
52
35
44
57
50
43
43
52
55
55
56
42
8
50
48
  Variable labels and units are found in Table 8-59. All variables
  except SAND, CLAY, and FRAG are !og10.

 ' Within replicate estimates not available.

  Within variation significantly larger than among variation (p = 0.05).
                                      405

-------
has averaged out real watershed-to-watershed differences.  Under this assumption, the unWbrmfty of the
watershed estimates indicates that they are biased toward the regional mean.

      The DDRP sampling design was  not  intended to directly answer the question  of the  scale of
variation. DDRP soli sampling was statistically designed to characterize sampling classes, not watersheds.
Given the available data we can, however, ask whether there is a watershed effect, i.e., do the values for
a specific variable from  all pedons sampled on a watershed tend to be above or below their respective
sampling class  means?   Analyses  described below  revealed significant watershed effects for  most
variables in both the NE and SBRP.

8.9.3.4  Estimation of Watershed Effect

      A weighted, unbalanced analysis of variance model that partitions the variability of a given soil
parameter into a sampling class effect, a watershed effect, and a residual error was used to assess the
watershed-specific effect on  each  variable.  The statistical model used in this analysis was:

                 yy - 3j  +  b, + By,                                               (Equation 8-1)

where yป is the pedon value for a given soil parameter from sampling class i on watershed j, a, and b,
are estimates of the sampling class and watershed effects, and By is the residual error.

      Horizon data were aggregated to  the  pedon prior to  watershed effect analysis in order  to avoid
the occurrence of missing values which would result from using only subhorizon or master horizon data,
since not all  pedons  sampled had all horizons.   Weighted  pedon averages for capacity and  intensity
variables were calculated using the aggregation approach described  in Section 8.9.3.3.

      It should be noted that this model does not contain a term for the sampling class by watershed
interaction.  Since only one pedon was sampled in a sampling class on a watershed, there were not
enough data to estimate the interaction term.  Furthermore, only a small  percentage of the  possible
sampling class by watershed combinations was actually sampled in each region. Also, the model does
not contain an Intercept in order to avoid  the difficulties encountered in using an  intercept model  with
unbalanced data (Searle, 1987).

      For the Northeast, there were 38 a( effects,  one for each sampling  class. The OTC sampling class
was not included in the SBRP analysis, because the watersheds which contained OTC were outliers with
respect to stream alkalinity and were dropped from the analysis (see Section 8.9.2).  Therefore, the SBRP
model contained 11 sampling class effects.  The parameterization of the  model required that the number
of watershed effects,  b,, be  one less than the number of sampled watersheds.  This parameterization
ensured that the model was of full rank and  that the estimates of sampling  class and watershed effects
were unbiased.  For the Northeast this resulted in 135 b. terms, since only  136 of the  145 watersheds
were sampled. In the SBRP, three  watersheds were dropped due to extreme values for stream ANC and
the model for this region contained 31 watershed effects.

      The analyses were conducted using the SAS REG regression procedure (SAS Institute, Inc.,  1988).
Binary Indicator variables (0 or 1)  were used to classify each pedon into the appropriate sampling class
and watershed.  The sampling class estimates obtained from the regression model were aggregated to
watershed, weighting by the areal fraction of sampling class on the watershed and, for capacity variables,


                                              406

-------
the product of the areal fraction and the sampling  class mass.   The resulting unadjusted watershed
estimate was modified by adding the estimate of the watershed  effect to give an adjusted watershed
value.  The adjusted watershed values were then used as explanatory variables in the analyses described
in Sections 8.9.4 to 8.9.6.

8.9.3.5 Evaluation of Watershed Effect

      There was a  significant  watershed effect  for most variables  (Table  8-57), and therefore the
watershed effect adjustment was applied uniformly to all of the data.  The watershed effect adjustment
had little effect on the means of the watershed estimates but the variance was generally much greater
for the adjusted values (Table  8-57).  This  result was expected  given the large withln-sampiing-class
variance.  Variance that had been averaged  out in the sampling class aggregation was relntroduced as
a  watershed effect.  The  variability in  the  distributions  of  the  adjusted  values was more  like our
expectations of the variability of natural systems {Figure 8-10).  Figure 8-11  illustrates the difference in
the watershed means and standard errors for pH in 0.01 M CaCI2.  Note that the adjusted watershed
means are more variable from watershed to watershed than the unadjusted means. The uncertainty of
the adjusted means, however, is higher than that of the unadjusted means. The actual variance probably
lies between these two estimates.

      Because the watershed effect was significant, the watershed-effect-adjusted  soil chemistry was
used in the following Level  I regression analyses.  The large uncertainty of the adjusted estimates limits
the predictive power of the  soil variables  in the regression analyses.  Future surveys should be designed
to reduce this uncertainty.

8-9.4  Regional Soil Characterization

      Soil physical and chemical properties were expected to vary  between the NE and  SBRP and
among the subregions of the NE.  In this section, soils are characterized using data for measured soil
variables regionalized to the target  populations.  Means and  standard errors for these variables are
presented for  each of the northeastern subregions,  for the NE  as a whole, and for the SBRP in Table
8-58.  The regionalized means are averages of  the adjusted watershed values weighted by the inverses
of the watershed inclusion  probabilities.  The standard error of the regionalized mean is the weighted
standard error calculated from the adjusted watershed values weighted by the inverses of the watershed
inclusion probabilities.  Values were calculated for the whole watershed, for the combined buffer zone in
the NE, and for a 100-m buffer zone in the SBRP (see Section 5.4.1.2 for detailed description of the buffer
zones). For base cations, only values from the  1.0 N NH4 Cl extraction were used in these analyses, as
these are the values of interest  to the modelling efforts.  Data obtained using the 1,0 N NH4 Cl  and 1.0
N NH4 OAc extractions were found to be highly  correlated, so similarities may be  inferred. Values In
Table 8-58, as well as the  cumulative distribution frequencies shown  in Section 5.5.6,  can be  used to
characterize the DDRP soils.

      Watersheds  from the five subregions of the  NE differ in the primary soil properties that were
hypothesized to affect surface water chemistry (Sections 2 and 3, Church and Turner, 1986).   For the
whole-watershed  aggregation,  base saturation (BS_CL)  ranges  from  17  to  30 percent, with the
Adirondacks (1A) and Southern New England (1D) soils having the lowest mean base saturation.  Cation
                                               407

-------
Table 8-57.  Means and Standard  Deviations of Soil Characteristics by
Aggregation Method and Region

Variable

SAND *
CLAY*
SOILDEN *
CA CL
MQ CL*
BS"CL*
CEC CL *
PH 01M *
AL PYP
C TOT*
SO4 H2O
SO4_PO4

NE

Unadjusted
Mean
65.5
5.17
1.27
1.92
0.45
20.3
6.40
4.02
0.29
4.00
9.66
29.0
Std. Dev.
13.2
3.82
0.17
1.52
0.37
9.72
2.93
0.13
0.12
2.70
3.82
10.7
Mean
65.0
5.21
1.27
2.10
0.39
19.6
7.11
4.28
0.29
4.08
9.57
28.9

Adjusted
Std.Dev.
17.9
6.33
0.22
5.81
1.09
20.9
6.75
0.42
0.19
4.72
8.22
18.9


Variable

SAND *
CLAY*
SOILDEN *
CA CL
MG CL*
BS CL*
CEC CL*
PH 01M *
AL PYP
CTOT
SO4 H2O
SO4 PO4

SBRP

Unadjusted
Mean
54.7
18.3
1.31
0.26
0.23
11.03
6.81
4.32
0.25
0.93
8.82
84.9
Std. Dev.
2.13
2.56
0.07
0.06
0.03
1.98
0.86
0.06
0.09
0.44
1.28
10,6
Mean
55.1
16.7
1.27
0.37
0.24
12.3
7.30
4.37
0.29
1.21
9.42
82.8

Adjusted
Std.Dev.
12.1
5.69
0.13
0.40
0.15
9.04
3.19
. 0.20
0.21
1.13
4.61
36.0
* Watershed effect significant at p < 0.01,
                                  408

-------
       2.6
       T.S
      12.6
      17.i
      22. 6
      27.5
      32.8
      37.5
      42.5
      47.S
        t
        3
        S
        7
        a
       11
       13
       15
       17
       19
                 Unadjusted
                                Total Clay (percent wt)
           Adjusted
                                                  2.5
                                                  7.6
                                                 12.6
                                                 17.6
                                                 22.5
                                                 27.5
                                                 32.5
                                                 37.5
                                                 42.S
                                                 47 .5
                   Frequency                                  Frequency
                           Base Saturation, Nf-yci (percent)
                 Unadjusted                                   Adjusted
 2.S
 7.5
12.5
17.6
22.5
27.5
12.5
37.5
42.S
47.5
                    10
                         15
                              20
                                   25
                   Frequency                                  Frequency
                    Cation Exchange Capacity, NHซCI (meq/100g)
                Unadjusted
           Adjusted
              5     10    16   20    25
                  Frequency
17
19
                                   pH, 0.01 M CaCI2
       S    10    15    20   25
           Frequency
                Unadjusted
           Adjusted
               5    10    15   20
                   Frequency
                                   25
             10    is
            Frequency
                                                                         20
                                                                              25
Figure 8-10. Histograms of unadjusted and adjusted watershed means for selected SBRP soils
variables.  The values on the vertical axes denote interval midpoints of the soil variables.
                                          409

-------
            5 . 5-
          
-------
Table 8-58.  Population Means and Standard Errors for Selected Variables, by Subregion/Region
and Aggregation (Watershed Adjusted Data)


Variable/

Mean ฑ
Standard Error by DDRP Subregion/Region
NE Subregion8

Aggregation " 1A

1B

1C 1D

1E
Overall

NE

SBRP

ง111


WX,X:M:ซ:™K:XS^
SAND
WS
BUF
SILT
WS
BUF
CLAY
WS
BUF
FRAS
WS
BUF
THKA
WS
BUF
SOILDEN
WS
BUF
Illi
CA CL
~ WS
BUF
MG CL
~WS
BUF
KCL
WS
BUF
NA CL
~ WS
BUF
as CL
~ WS
BUF
CEC CL
"ws
BUF
AC BACL
~ WS
BUF
Ptl 01M
" WS
BUF
PH H2O
~ WS
BUF
AL AO
WS
BUF

c 69.60 ฑ 3.03
73,87 4 2,45

27.65 ฑ 2.38
23.55 ฑ 1.93

2.69 ฑ 1,01
2.68 ฑ 0.82

19.88 ฑ 3.03
20.85 4 2.45

70,14 ฑ 5.68
85.19 ฑ 4,44

1.21 ฑ 0.03
1,21 ฑ 0,03
IllllSllilllliil

4,87 4 2.10
4,95 ฑ 1.72

0,47 ฑ 0.32
0.69 4 0,26

0.06 4 0.01
0.06 * 0.01

0.03 ฑ 0.02
0.08 ฑ 0.02

18.44 4 3.94
23.30 * 3,19

9.89 ฑ 1,60
10.82 ฑ 1,31

19,82 * 2,88
19.72 ฑ Z36

4,22 ฑ 0,09
4,38 ฑ 0,07

4.80 4 0.10
4.97 * 0,08

0,86 ฑ 0.06
0.66 4 O.OS

. 45.52 4 1.77
48.47 4 1.78

41.67 4 1.39
41.84 4 1.39

12.85 4 0.59
12.80 ฑ 0.58

25.41 4 1,88
24.32 * 1.88

90^2 ฑ 3.35
101,35 ฑ 4,14

1.60 ฑ 0.02
1.50 4 0.02


68.15 ฑ 2.82 78,15 ฑ 2,19
71.11 ฑ 3.10 79.42 ฑ 2.59

28.30 ฑ 2.22 19.72 ฑ 1.73
24.58 ฑ 2.44 18,81 ฑ 2.04

3.68 ฑ 0,95 2,08 ฑ 0.73
4.33 ฑ1.04 1.75 ฑ 0.87

25.28 ฑ ฃ62 19.10 ฑ 2.25
27.39 ฑ ฃ83 18.47 ฑ 2.59

77.18 4 5,08 108,33 * 4.99
81,84 ฑ 5,25 111.11 4 5.68

1.16 ฑ 0,03 1.41 ฑ 0.03
1.O8 ฑ 0.04 1.38 ฑ 0.03


53.56 ฑ 2.18
54.70 ฑ Z64

37.64 ฑ 1.72
35.28 ฑ 2.08

8.69 ฑ 0.73
9.80 ฑ 0.89

21.47 ฑ 2.05
22,92 4 2.72

76.SS ฑ 3,87
81.03 ฑ 4.74
_
1.23 ฑ 0.03
1,15 ฑ 0.03


62,17 * 1.32
64.04 ฑ 1.51

31.67 ฑ 1.04
29.37 ฑ 1.18

8,15 ฑ 0.44
6.53 ฑ 0.51

22.38 * 1.27
23.08 ฑ 1.47

82.42 ฑ 2.53
90.19 ฑ 2.69

1.28 ฑ 0.02
1.25 ฑ 0.02
:•x"x-x•^:v:•:.Xv^:.:.x.x•x;:•:::^:-:vX-^r-^x•^::ft>-:X^xxx:x-:;x:x;Xv^^

2.B8 4 1,15
3.31 4 1.15

0.74 ฑ al7
0.83 4 0.17

0.10 4 O.01
0.10 ฑ 0.01

0.05 4 0.01
0.08 ฑ 0.01

28.10 ฑ 3.22
35.17 ฑ 3,30

8.93 ฑ 0.88
8.59 * 0.88

10.68 4 1.58
11.18 4 1.57

4.37 ฑ 0.08
4.47 4 0.07

5.01 4 O.07
5.14 ฑ 0.07

0.23 ฑ 0.04
O.21 4 0.03

1,36 ฑ 2.00 2.96 4 1.34
2.40 4 1.98 3.94 ฑ 1.55

0.2O ฑ 0,30 O.S8 4 O.20
0,58 * 0.30 0.40 4 0.23

O.O8 ฑ 0.01 0.05 ฑ O.01
0.10 ฑ 0.01 0.05 ฑ 0.01

0,05 ฑ 0.02 0.03 ฑ 0.01
tt10 ฑ 0.02 0,04 4 0.01

22.25 ฃ 3.77 17.49 * 3.73
30,21 ฑ 3.90 20,13 ฑ 4.24

6.27 ฑ 1.52 3.98 * 1.03
9.44 ฑ 1.50 5.08 4 1.18

18.95 ฑ 2,73 10.04 ฑ 1.83
84.57 ฑ 2.71 12.46 4 2.12

4,40 ฑ 0,08 4,35 ฑ O.O8
4.50 ฑ 0.08 4.38 ฑ 0.09

4.S6 ฑ 0.09 4.83 ฑ 0.09
5,08 ฑ 0.09 4,86 * 0,10

0.54 ฑ 0.06 0.24 ฑ 0.04
0,39 ฑ 0.06 0.23 4 0.05

2.04 ฑ1.35
2,82 ฑ1.52

0.53 ฑ 0,20
0.93 ฑ 0.23

0.13 ฑ 0.01
0.14 ฑ 0.01

0,07 ฑ O.O1
0.12 ฑ 0.02

25.39 4 2.85
30.49 ฑ a58

7.92 ฑ 1.03
10.64 ฑ 1,15

17.67 ฑ 1.85
23.06 ฑ 2.07

4,43 ฑ 0.06
4.53 ฑ 0.08

5.01 ฑ 0.07
B.10 ฑ 0.08

0.43 ฑ 0,04
0.36 s O.OS

2,65 4 0.87
3.44 ฑ 0.85

0.44 4 O.13
0.73 ฑ0.13

0.09 * 0.01
0.09 ฑ. O.01

0.04 ฑ 0.01
0.08 4 0.01

22.80 ฑ 1.91
28.28 ฑ2.11

7.60 ฑ 0.66
9.43 ฑ 0.68

16.03 ฑ 1.20
18.94 ฑ 1,16

4.36 ฑ 0.04
4.46 4 0.04

4.83 ฑ 0.04
5.O4 ฑ 0.05

0.48 4 0.03
0.38 ฑ 0.03
l;S:ป;?i^x:Bi!;iP;!lilS

50.35 ฑ 1,48
5ฃ19 ฑ 1.46

30.04 ฑ 0.95
30.89 ฑ 0,93

19.61 4 1.03
laS2 4 1,01

19.40 ฑ 1,89
25.32 ฑ 1.71

102.31 ฑ 7.73
103.79 ฑ 7.05

1.30 ฑ 0.02
1.25 4 0.02
||ll|l|||||;i|li|i

0.30 ฑ 0.07
0.41 ฑ 0.07

QXl * 0.03
0.21 ฑ 0.03

0.15 4 0.01
0.14 ฑ 0.01

0.02 4 0,00
0,03 ฑ 0.00

11,83 ฑ 1.35
12.27 ฑ 1.23

7.17 ฑ 0.48
7,86 ฑ 0.45

10.60 4 0.81
11.90 ฑ 0.80

4.34 ฑ 0.05
4.38 ฑ 0.05

5,08 ฑ 0.08
5,12 ฑ 0.05

o,ae ฑ 0.03
0.32 4 0,03
                                                                                      continued
                                            411

-------
Table 8-58.   (Continued)


Mean
ฑ Standard Error by DDRP Subregion/Region
NE Subregion
Variable/
Aggregation
ALJ3O
WS
BUF
AL_PVP
WS
BUF
ALPOT
WS
BUF
UMEPOT
WS
BUF
CJTOT
WS
BUF
SCM_H2O
WS
BUF
SO4_PO4
WS
BUF
SO4_EMX
WS
SO4.B2
WS
SO4_X!N
WS
SO4_SLP
WS

1A

0.47 ฑ 0.05
0.37 ฑ 0.04

0.43 ฑ 0.04
0.36 ฑ 0.03

7.48 ฑ 0.22
7.71 ฑ 0.17

2.80 ฑ 0.09
2.76 * 0.07

5,21 ฑ1.35
5.62 ฑ 1.11

7.77 ฑ 4.10
8.27 ฑ 3.35

28.87 ฑ 6.08
24.90 ฑ 4.96
d
3052.2 ฑ .
d
870.8 ฑ .
d
120.93 ฑ .
d
3.15 ฑ .

IB

0.20 ฑ O.O3
0.1B ฑ O.O3

0.18 ฑ 0.02
. 0.18 ฑ 0.02

7.82 ฑ 0.15
8.11 ฑ 0.18

2.68 ฑ 0.06
2.80 ฑ 0.06

1.90 ฑ 0.74
2.56 a 0,74

13.84 ฑ 2.25
13.82 ฑ 2.24

23.20 ฑ 3.32
21.75 ฑ 3.31

2165.3 ฑ .

952.7 ฑ ,

289.84 ฑ .

1.74 ฑ .

mm
SO416
WS
SO4_NRET
~ws
CAMG16
WS
SOBC
WS
CM JO
WS
ALKA e
WS
PHEQ11
WS


115.48 ฑ 3.54

-0.12 ฑ 0.04

183.63 ฑ 20.80

230.82 4 24.75

1.00 ฑ 0.12

80.81 ฑ 18.29

6.50 ฑ 0.16

155.16 ฑ 8.52

0.11 ฑ 0,05

327.00 ฑ 32.82

441 .98 ฑ 4441

1.63 ฑ 0.17

191.83 ฑ 31.27

7.15 ฑ 0.18

1C

0.31 ฃ 0.05
0.25 ฃ 0.05

0.27 ฑ 0.03
0.23 * 0.03

7.B4 ฑ0,18
7.83 * 0.20

2.71 ฑ- 0.08
2.86 * 0.08

4.68 ฑ 1.28
7.07 ฃ 1.27

8.35 ฑ 3.90
9.B9 ฑ 3.86

28.90 ฑ 5.76
31.98 ฑ 5.70

2823.7 ฑ .

891.1 ฑ ,

125.49 ฑ ,

2,81 ฑ ,

ID

0.15 ฑ 0.03
0.15 ฑ 0.04

0.19 ฑ 0.02
0.18 ฑ O.O3

8.18 ฑ 0.19
8.32 ฑ 0.21

2.73 ฑ a08
2.79 ฑ 0.09

3.18 ฑ 0.86
4.39 ฑ1.00

11.98 ฑ 2.65
13.07 ฑ 3.08

40.53 ฑ 3.86
41.63 ฑ 4.48

1750.4 ฑ .

1116.8 ฑ .

255,83 ฑ .

1.31 ฑ ,

1E

0.26 ฑ 0.03
0.23 ฑ 0,04

0.24 ฑ 0.02
0.22 ฑ O.O3

7.48 ฑ 0.15
7.83 ฑ 0.18

2.74 ฑ 0.08
2.86 ฑ 0.07

3.84 ฑ 0.87
5.97 ฑ 0.97

7.24 ฑ 2.64
8.88 ฑ 2.85

25.71 ฑ 3.89
30.29 ฑ 4.37

2338.1 ฑ .

872.2 ฑ .

133.59 ฑ .

2.3B ฑ .
Overall

NE

0.29 ฑ 0.02
0.24 4 0.02

0.27 ฑ 0.01
0.24 ฑ 0.01

7.66 ฑ 0.10
7.93 ฑ 0.10

2.69 ฑ 0.04
2.82 ฑ 0.04

3.86 ฑ 0.58
5.26 ฑ 0.55

9.49 ฑ 1.70
10.51 ฑ 1.66

28,24 ฑ 2.52
29.54 ฑ 2.45

2436,5 ฑ .

925.6 ฑ .

175.22 a ,

2,33 ฑ ,

SBRP

ft45 ฑ 0.03
0.42 ฑ 0.03

0.26 ฑ O.O3
0.32 ฑ 0.03

7.22 ฑ 0.19
7.41 ฑ 0.18

2.51 ฑ 0.05
2.58 ฑ 0.05

1.16 ฑ 0.20
1.42 ฑ 0.20

9.68 ฑ 0.63
10.28 ฑ 0.62

87,91 ฑ7.12
82.10 ฑ 7.02

S362.4 ฑ .

175.4 ฑ .

38.59 ฑ .

32.75 ฑ .
S:pฅS:;S:Sgi8^^

82.33 ฑ 4.99

-0,00 ฑ 0.07

191.85 ฑ 19.46

SSS.30 ฑ 24.59

1.06 * 0.13

117.04 ฑ 19.33

7,01 ฑ 0.13

129.28 ฑ 6.48

-0.13 ฑ 0.05

2O6.01 ฑ 26.23

507.77 ฑ 42.57

1.03 ฑ 0.18

94.60 ฑ 23.07

8.73 ฑ O.Z1

73.90 ฑ 4.21

-0.13 ฑ 0.07

205.03 ฑ 19.68

292.48 ฑ 18.68

1.23 ฑ 0.13

140.02 s 18.60

7.16 ฑ 0,09

109.54 ฑ 3.58

-0.05 ฑ 0.03

220.22 ฑ 11.47

337,53 ฑ 16.12

1,19 ฑ 0.07

125.60 ฑ 10.45

8.93 ฑ 0.07

29.88 ฑ 4.00

0.72 ฑ 0,03

121.55 * 15.17

204.63 ฑ 22.40

0.69 ฑ 0.11

128.78 ฑ 18.26

7.28 a 0,05
  1A is the Adirondacks, 1B is the Poconos/Catskills, 1C is Central New England, 1D is Southern New England, and 1E is Maine.
b Variable labels and units are found in Table 8-60.
0 For each variable, WS refers to the entire watershed and  BUF refers to the buffer zone.
  Error estimates were unavailable.
e ALKANEW in the NE, ALKA11 in the SBRP
                                                      412

-------
exchange capacity (CEC_CI_) is lowest in Southern New England, with a mean approximately half that
of the northeastern regional mean. The highest levels of water-extractable sulfate (SO4_H2O) are found
In the two more southern subregions (Poconos/Catskills, 1B, and Southern New England); phosphate-
extractable sulfate Is highest in the Southern New England soils. Sulfate isotherms also differ among the
subregions.  Suifate adsorption capacity (SO4JEMX) is highest for soils in the Adirondack Subregion and
lowest in Southern New England. The Southern New England soils are also characterized by the highest
half-saturation constant  (SO4_B2) and  the second largest  equilibrium soil  solution  concentration
(SO4_XIN).  Thus, the adsorption curve for the Southern New England soils is flatter and lower than that
for the other subregions. Soils In the Poconos/Catskills Subregion  have similar isotherm parameters to
the Southern New England soils, except for a  significantly higher sulfate adsorption capacity.  Sulfate
isotherms for soils from the three northern subregions are distinct from those of the southern subregions.

      Other soil properties also vary among the subregions.  Exchangeable acidity (AC BACL) Is relatively
high in the Adlrondacks soils, which  are also characterized by  the highest sum of base cations and the
highest  cation exchange capacity.   In general, extractable aluminum (AL_AO,  AL_CD, AL_PYP) also is
highest  in the Adirondacks soils.  Soils of the Poconos/Catskills Subregion are finer-textured relative to
the other subregions and have a higher mean bulk density (SOILDEN).  Soils in the Southern New
England watersheds have higher sand content. The relatively low mean CEC may be related to the
higher sand content of these soils. Soils in the Maine (1E) Subregion are similar to the Adirondacks and
Central  New England  (1C) soils, with  relatively high levels of exchangeable  acidity and total carbon.
Spodosols represent a large proportion of the soils in these three northern subregions, which may partially
explain these observations.  Soil pH  varies relatively little among the five  subregions.

      Comparing regional watershed means for the NE and SBRP,  a few differences are notable.  Soils
in the NE are characterized by higher concentrations of bases and  base saturation, higher acidity, and
much lower clay content and phosphate-extractable sulfate.  Carbon content of northeastern soils Is also
higher than SBRP soils. Sulfate isotherm parameters also differ significantly between  the two regions, with
the SBRP exhibiting significantly higher maximum adsorption capacity and significantly lower equilibrium
soil solution sulfate concentrations than the northeastern soils.

      Mean values for soils within the buffer zones are similar to the whole-watershed means for most
variables. Differences exist for the base cation variables (CA_CL, MG_CL, K_CL, NA_CL, BS_CL), where
levels are higher in  the  buffers relative to the  whole watershed.   As these buffers represent areas  of
convergent flow (variable hydrologic source areas, riparian zones), this is as expected.  Soils in the buffer
zones have  higher total carbon  content and slightly higher water-extractable sulfate levels.  In the Central
New England and Maine Subregions, the mean extractable acidity of the buffer soils is higher relative to
the whole watershed mean.  Differences between buffers and whole watershed values are generally larger
in the NE than in the SBRP.

8.9.5 Sulfate and Sulfur Retention

      This section and Section  8.9.6  discuss the statistical relationships between measured  soil physical
and chemical properties and water chemistry for the DDRP watersheds.  These relationships are also
evaluated in terms of potential cause-effect controls on water chemistry.  Tables 8-59 and 8-60 show the
                                               413

-------
Table 8-59. Non-parametric  Correlations  Between Lake Chemistry Variables and Selected  Soil
Properties for the  NE  DDRP  Watersheds
Variable
Spearman Correlation Coefficients Significant at 0.05 / Significance LeveF / N = 143

               Units         SO416a      SO4  NRET    CAMG16     ALKANEW    PHEQ11

Soil Physical Properties
SAND
Sand, total
SILT
Silt, total
CLAY
Clay, total
FRAG
Fragments > 2mm diameter
THKA
Thickness adjusted for FRAG
SOILDEN
Bulk density
percent
percent
percent
percent
0.22660
cm 0.0065
0.18434 0.16098
g/cc 0.0275 0.0548
-0.34086
0.0001
0.31267
0.0001
0.30097
0.0003


0.20719
0.0130
-0.33012
0.0001
0.31272
0.0001
0.25367
0.0022


0.18355
0.0282
-0.31760
0.0001
0.30011
0.0003
0.24658
0.0030


0.18659
0.0257
Soil Chemical Properties
CA_CL
Exchangeable calcium (NH4 Cl)
MG_CL
Exchangeable magnesium (NH4 Cl)
K_CL
Exchangeable potassium (NH4 Cl)
NA_CL
Exchangeable sodium (NH4 Cl)
SBC_CL
Sum of base cations (NH4 Cl)
BS_CLM
Base saturation
CEC_CL
Cation exchange capacity
meq/100g
meq/100g
meq/100g
meq/100g
-0.20633
meq/100g 0.0134
percent
meq/100g
0.22554
0.0068
0.24922
0.0027
0.33975
0.0001

0.17816
0.0333
0.35370
0.0001

0.20727
0.0130
0.22041
0.0082
0.35981
0.0001

0.19769
0.0179
0.32607
0.0001

0.21764
0.0090
0.21303
0.0106
0.34094
0.0001

0.20094
0.0161
0.31892
0.0001

a SO416 is the lake sulfate concentration, SO4_NRET is watershed sulfur retention, CAMG16 is the lake sum of base cation
  concentration, ALKA11 is the lake acid neutralizing capacity, and PHEQ11 is the air-equilibrated stream pH.
                                                                                                continued
                                                  414

-------
Table 8-59.  (Continued)
Spearman Correlation Coefficients Significant at 0.05 / Significance Level / N = 143
Variable Units SO416 SO4 NBET CAMG16 ALKANEW
AC.BACL
Acid'rty, total exchangeable meq/100g
PH 01 M -0.26116
pH '(0.01 M CaC12 ) 0.0016
PH_H2O -0.24556
pH (delonized water) 0.0031
AL_AO
Aluminum, acid oxalate extr. percent
AL_CD
Aluminum, citrate dithionite extr. percent
AL_PYP
Aluminum, pyrophosphate extr. percent
ALPOT
Aluminum potential (pH - */3 pAI)
LIMEPOT -0.22764
Ume potential (pH - 14pCa) 0.0063
C_TOT
Carbon, total percent
SO4_H2O
Suifate, water extractable mg S/kg
SO4_PO4
Suifate, phosphate extractable mg S/kg
SO4_EMX
Adsorption asymptote ueq/kg
SO4_B2 0.26265
HaW .saturation constant ueq/L 0.0015
SO4_XIN 0.32235
Zero net adsorption concentration ueq/L 0.0001
SO4_SLP -0.17347
Zero net adsorption, slope L/kg 0.0383

0.20045
0.0164
0.19831
0.0176
-0.22581
0.0067
-0.30041
0.0003
-0.31004
0.0002

0.18844
0.0242

0.21355
0.0104

-O.17203
0.0399




0.34515
1 0.0001
0.39852
0.0001
-0.29118
0.0004
-0.17679
0.03470
-0.23970
0.0039

0.31191
0.0001



-O.31713
0.0001
0.20137
0.0159
0.33774
0.0001
-0.34564
0.0001

0.50310
0.0001
0.55995
0.0001
-0.22259
0.0075
-0.19756
0.0180
-0.24274
0.0035

0.45409
0.0001


'-0.16690
0.0463
-0.26774
0.0012

0.17842
0.0330
-0.23617
0.0045
PHEQ11

0.50717
0.0001
0.56159
0.0001
-0.22549
0.0068
-0.21248
0.0108
-0.25233
0.0024

0.45649
0.0001



-0.26759
0.0012

0.17229
0.0396
-0.23432
0.0049
                                            415

-------
Table 8-60. Non-parametric  Correlations  Between  Stream Chemistry Variables and Selected Soil
Properties for the SBRP DDRP Watersheds      	
Variable
Spearman Correlation Coefficients Significant at 0.05 / Significance Level / N = 31

               Units        SO416a      SO4 NRET     SOBC       ALKA11
          PHEQ11


SAND
Sand, total percent
SILT 0.42863
Silt, total percent 0.0161
CLAY
Clay, total percent
FRAG
Fragments > 2mm diameter percent
THKA -0.37742
Thickness adjusted for FRAG cm 0,0363
SOILDEN
Bulk density g/cc




-0.40161 -0.45121
0.0251 0.0108

^
Sott Chemical Properties
CA_CL 0.58790 -0.40766
Exchangeable calcium (NH4 Cl) meq/100g 0.0005 0.0228
MG_CL
Exchangeable magnesium (NH4 Cl) meq/IOOg
0.51532 0.41331 0.51734
0.0030 0.0208 0.0029
0.49587 0.44315 0.35363
0.0045 0.0125 0.0510
K_CL
Exchangeable potassium (NH4 Cl)     meq/100g
NA_CL
Exchangeable sodium (NH4 Cl)

SBC_CL
Sum of base cations (NH4 Cl)

BS_CLM
Base saturation
               meq/100g

                           0.48226
               meq/100g    0.0060
                                percent
                                                      0.54597
                                                      0.0015
0.52823
0.0023
CEC_CL
Cation exchange capacity (NH4 Ci)   meq/IOOg
0.50444
0.0038
a SO416 is the stream suifate concentration, SO4_NRET is watershed sulfur retention, SOBC is the stream sum of base cation
  concentration, ALKA11  is the stream acid neutralizing capacity, and PHEQ11 is the air-equilibrated stream pH.

                                                                                                 continued
                                                   416

-------
Table 8-60.  (Continued)
Variable
Spearman Correlation Coefficients Significant at 0.05 / Significance Level / N =  31

                Units          SO416        SO4.NRET     SOBC       ALKA11
                                                                                                      PHEQ11
AC_BACL
Total exchangeable acidity

PH_01M
pH  (0.01 M CaCI2)

PHJH2O
pH  (detenized water)

AL_AO
Aluminum, acid  oxalate extr.

AL_CD
Aluminum, citrate dithionite extr.

AL_PYP
Aluminum, pyrophosphate extr.

ALPOT
Aluminum potential (pH - */3 pAI)

LIMEPOT
Lime potential-(pH - 1/4pCa)

C_TOT
Carbon, total

SO4_H2O
Sulfate, water extractable

SO4_PO4
Sulfate, phosphate extractable

SO4_EMX
Adsorption asymptote

SO4_B2
Half saturation constant

S04_XIN
Zero net adsorption concentration

SO4_SLP
Zero net adsorption, slope
                meq/1 OOg
                percent
                percent
                percent
                                                          -0.35524
                                                          0.0499
                                                          -0.48589
                                                          0.0056
                             0.36573
                             0.0430
-0.38790
0.0311
                percent

                             0.43427
                mg S/kg      0.0146
                mgS/k

                            -0,49395
                ueq/kg       0.0047
                ueq/L


                ueq/L
 0,45976
 0.0077
               -0.44032
                0.0132
                                                          0.41129
                                                          0.0215
                            -0.36169
                             0.0456

                             0.37298
                             0.0388
                            -0.36653
                             0.0426
                             -0.48710
                              0.0055
-0.44476
 0.0122

-0.42702
 0.0166

 0.44758
 0.0116
                                                                                   0.46169
                                                                                   Q.Q089-
           -0.41774
            0.0194
                                                                                   0.39194
                                                                                   0.0292
                                                       417

-------
nonparametric Spearman correlations between selected soil properties and each of the water chemistry
variables considered.  Results of stepwise multiple regressions for sulfate and sulfur retention are given
In Tables 8-61 and 8-62.

8.9.5.1  Northeast

8.9.5.1.1  Whole watershed aggregation -

      The coefficients  of determination, or R2, range from 0 to 0.56 for sulfate and from 0.12 to 0.64 for
sulfur retention in the northeastern subregions.  Bivariate correlations between soil properties and sulfate
or sulfur retention are generally not high.  The strongest correlation is between lake sulfate concentration
and the zero net adsorption concentration (or equilibrium soil solution  sulfate  concentration) from the
sulfate isotherms.  This relationship makes sense mechanistically, i.e., since northeastern watersheds are
generally near steady state with respect to sulfur deposition, soil and lake sulfate concentrations both tend
to reflect  deposition.  The highest correlation with sulfur retention is a negative one with  extractable
aluminum.   Soils in the NE appear to be approaching a new equilibrium with lower sulfate  deposition;
soils rich in extractable aluminum have a large adsorbed sulfate pool that is now desorbing,  resulting in
an  inverse relationship between extractable aluminum and sulfate retention. The correlations of sulfate
and sulfur  retention with  soil pH and  lime potential also fit this  scenario.  As would be expected, the
correlations for sulfate and sulfur retention tend to be opposite in sign. A similar pattern of relationships
is apparent in the multiple regression results,  with  some  variation among the subregions.   In the
Poconos/Catskiils (1B) and Southern New England (1D), sulfate is positively correlated with soil pH; in
Central New England  (1C) and in the  region as a whole, however, the relationship with pH is negative.
In the Adirondacks (1A)  and the Poconos/Catskiils, the two subregions with highest sulfate deposition,
sulfate is correlated with the half-saturation concentration, a sulfate isotherm intensity factor that is highly
correlated with the concentration at zero net adsorption.  This relationship is consistent with the  bivariate
correlations.  In the Poconos/Catskiils, Central  New England, and  Southern New England, sulfate is
correlated with extractable  aluminum.   This is consistent with the  hypothesis that previously adsorbed
sulfate  may be desorbing from  these soils.  This scenario  also is supported by the sulfur retention
regressions; in most subregions and the NE overall, the greater the extractable aluminum in the soil, the
lower the net retention.

8.9.5.1.2  Combined  buffer aggregation -

      Variables selected  by the stepwise regressions for the northeastern watersheds aggregated for the
combined buffers  around the lakes and streams were the same as or similar to those  selected for the
whole watershed  aggregation.  The R2 for sulfate improved significantly, with fewer variables in the
model,  but the R2 for the sulfur retention model is lower.  Buffer zone models were  not run for the
northeastern subregions.   These  results alone  do not allow a conclusion to be drawn regarding the
relative merits of each aggregation for these analyses.
                                               418

-------
Table 8-61.  Results of Stepwise Multiple Regressions for DDRP Lake and Stream
Sulfate Concentrations (SO416) Versus  Soil Physical and  Chemical Properties
Variable6
SAND
CLAY
FRAG
THKA
SOILDEN
CA CL
MG CL
SBC CL
BS CL
CEC CL
AC BACL
PH 01M
AL AO
AL CD
AL PYP
ALPOT
C TOT
SO4 H2O
SO4 PO4
SO4 EMX
SO4 B2
SO4 XI N
SO4_SLP
R2
Whole Watersheds Buffer Zone
Subregion" Region Region
1AC 1B 1C 1D 1E NE SBRP NE SBRP
5
3
1 1
1 2(-) 2 2(-)
3
1 4
2 63
3 4(-) 4(-)
3(-) 1(-) 3 3
2 2(-) 2 2(-)
1 3 1
1
0.35 0.56 0.52 0.46 None 0.27 0.66 0.43 0.62
Selected
a 1A is the Adirondacks, 1B is the Poconos/Catskills, 1C is Central New England, 1D is Southern New England,
  and  1E is Maine.
  Variable labels and units are found in Table 8-60.
c Numbers indicate order of entry into stepwise model. (-) indicates a negative parameter estimate.
                                              419

-------
Table 8-62.  Results of Stepwise Multiple Regressions for DDRP Watershed Sulfur
Retention (SO4_NRET) Versus Soil Physical and Chemical Properties
Variable"
                                     Whole Watersheds
                           Subregiona
                                  Region
               1AC
1B
1C
1D
1E
NE
SBRP
                                                Buffer Zone
                                           Region
NE
SBRP
SAND
CLAY
FRAG
THKA
SOILDEN
CA CL
MG CL
SBC"CL
BS_CL
CEC_CL
AC BACL
PH_01M
AL AO
AL CD
AL_PYP
ALPOT
C TOT
S04 H2O
SO4 PO4
SO4 EMX
SO4 B2
SO4 XIN
SO4 SLP
                               5
                               4
        2
        6
               4

               2
R              0.12    0.41    0.64    0.30    0.37    0,22    0.44     0.16    0.44


a 1A is the Adirondacks, 1B is the Poconos/Catskills, 1C is Central New England, 1D is Southern New England,
  and 1E is Maine.
  Variable labels and units are found in Table 8-60.
c Numbers indicate order of entry into stepwise model.  (-) indicates a negative parameter estimate.
                                         420

-------
8.9.5.2  Southern Blue Ridge Province

8.9.5.2.1  Whole watershed aggregation -

      Exchangeable magnesium and maximum sulfate adsorption capacity are most strongly related to
sulfate and sulfur retention in the SBRP in the multiple linear regressions (Tables 8-61 and 8-62). R2 for
sulfate is  0.66, and 0.44  for sulfur retention.  Higher exchangeable magnesium  (and  calcium, in the
bivariate correlations) in the soil is correlated with higher sulfur in the water and lower sulfate retention
by the soil; higher  sulfate adsorption capacity is correlated with lower sulfate in the water and higher
retention in the soil.  Higher water-extractabie sulfate in the soil is correlated  with higher water sulfate.
Higher base saturation soils are correlated with greater sulfur retention in the soil (though this may be
spurious because there is  no bivariate correlation between these variables).  In this region where the soils
have not yet reached equilibrium with atmospheric sulfur deposition, the adsorption capacity of the soil
is  a good explanatory  variable of both retention and concentration in  the  drainage water.   Water-
extractable sulfate is a readily mobilized pool of sulfate, acting in the SBRP as the soil intensity variable
associated with sulfate  in the water.  The reason for the strong  relationship between exchangeable
magnesium (and calcium) and sulfate is possibly due to higher base status soils generally having higher
pH and hence lower sulfate adsorption, although there is no correlation between  base saturation or pH
and sulfate. Another possibility would be a sulfur-rich  bedrock source that is weathering both bases and
sulfur.   This is  supported  by the correlation between the low organic  meta-sedimentary MSL sampling
class soils and stream sulfate (Section 8.6.3.2).

8.9.5.2.2  100-m buffer aggregation -

      There is virtually no difference in the  models selected for the 100-m  buffer aggregation from those
for the whole watershed in the SBRP. This suggests that estimated  chemistry  for the soils at the stream
sides is not more strongly associated with spring baseflow chemistry than those in the whole watershed.
Stream  chemistry measured during stormflow, a time  when the near-stream soils would be expected to
be  more hydrologically active, might be more strongly associated with 100-m  buffer soil chemistry.

8.9.6 Ca plus Mq (SOBC). ANC. and PH

      Results of stepwise  multiple regression for Ca plus Mg concentrations (sum of base cations in the
SBRP),  ANC, and pH are given in Tables 8-63 through 8-65.  This section summarizes the results and
discusses potential  cause-effect controls on  surface water chemistry.  The dependent water chemistry
variables are all highly correlated with each other and therefore have very similar  associations with soil
physical and chemical  properties.   Of the four, pH  is the most  dissimilar  because of  its nonlinear
relationship with ANC (Figure 5-7),

8.9.6.1  Northeast

8.9.6.1.1  Whole watershed aggregation -

      The R2 values range from 0.31 to 0.84 for the northeastern subregions.  Soil pH is most commonly
and most strongly associated with the water chemistry in most northeastern subregions and in the region

                                               421

-------
Table 8-63.  Results of Stepwise Multiple Regressions for DDRP Lake Calcium plus
Magnesium Concentrations (CAMG16) and Stream Sum of Base Cation Concentrations
(SOBC) Versus Soil Physical  and Chemical Properties

Variable6 1AC
SAND
CLAY
FRAG
THKA
SOILDEN
CA CL
MG CL
SBC CL 3
BS CL
CEC CL
AC BACL
PH 01M 2
AL AO
AL CD
AL PYP
ALPOT
C TOT
S04 H2O
SO4 PO4
SO4 EMX
SO4 B2
SO4 XI N 1
SO4 SLP
Whole Watersheds Buffer Zone
Subregion" Region Region
1B 1C 1D . 1E NE SBRP NE . SBRP
1
6
3
1 111
235
2
141 2 4
2
5(-) 5
1
4(-) 2
                0.71
0.57    0.49
0.81
0.59    0.40
0.44
0.38    0.48
a 1A is the Adirondacks, 1B is the Poconos/Catskills, 1C is Central New England, 1D is Southern New England,
  and 1E is Maine.
  Variable labels and units are found in Table 8-60.
c Numbers indicate order of entry into  stepwise model.  (-) indicates a negative parameter estimate.
                                            422

-------
Table 8-64.  Results of Stepwise Multiple Regressions for DDRP Lake and Stream ANC
(ALKANEW and ALKA11) Versus Soil Physical and Chemical Properties
Variable15
SAND
CLAY
FRAG
THKA
SOiLDEN
CA CL
MG CL
SBC CL
BS CL
CEC CL
AC BACL
PH 01 M
AL AO
AL CD
AL PYP
ALPOT
C TOT
SO4 H2O
SO4 PO4
SO4 EMX
SO4 B2
SO4 XIN
SO4 SLP
Whole Watersheds Buffer Zone
Subregion* Region Region
1AC 1B 1C 1D 1E NE SBRP NE SBRP
3(-)
1
2 4
3(-) 3(-) 5(-)
3
1 1 1
2 5
4
21 12 2
10
4 5(-)
2
4(-)
1(-)
3(-)
5(-) 3(-)
4,(-)
1 2(-) 3
R              0.75     0.62    0,53     0.83    0.47    0.43     0.44      0.36    0.47


8 1A is the Adirondacks, 1B is the Pooonos/Catskills, 1C is Centra! New England, 1D is Southern New England,
  and-1E is Maine.
  Variable labels and units are found in Table 8-60,
c Numbers indicate order of entry into  stepwise model.  (-) indicates a negative parameter estimate.
                                             423

-------
Table 8-65.  Results of Stepwise Multiple Regressions for DDRP Lake and Stream pH
(PHEQ11) Versus Soil Physical and Chemical Properties
Variable
                                     Whole Watersheds
                           Subregiorf
                                                        Region
                                                                      Buffer Zone
Region
               1AC    1B     1C      1D     1E     NE    SBRP     NE     SBRP
SAND
CLAY
FRAG
THKA
SOILDEN
CA_CL
MG_CL
SBC_CL
BS  CL
CEC_CL
AC_BACL
PH  01M
AL AO
AL_CD
AL_PYP
ALPOT
C TOT
S04JH2O
SO4_PO4
SO4_EMX
SO4_B2
SO4_XlN
SO4 SLP
               1        1               1
                              2
                              K-)
R2
               0.33    0.84    0.64    0.46    0.71    0.31     0.45      0.30     0.48
  1A is the Adirondacks, 1B is the Poconos/Catskills, 1C is Central New England, 1D is Southern New England,
  and 1E is Maine.
  Variable labels and units are found in Table 8-60.
0 Numbers indicate order of entry into  stepwise model.  (-) indicates a negative parameter estimate.
                                         424

-------
overall.  Exchangeable calcium, base saturation, and the sum of soil base cations are highly correlated
with each other and are also commonly associated with the water chemistry,  in the Poconos/Catskills
Subregion, cation exchange capacity was selected by the stepwise regressions for ANC and Ca plus Mg;
this region has the highest mean base saturation.  In Southern New England, cation exchange capacity
was selected for lake pH, but with a negative sign; that subregion has the lowest base saturation, I.e.,
it has a greater proportion of acidic  cations on its exchange sites.  The bivariate correlations (Table 8-
59) also match the pattern seen In the multiple regressions; heavier, clay-rich soils high in exchangeable
bases, pH, and base saturation are  strongly correlated with higher bases, ANC, and pH in the water.
These relationships lend support to the hypothesis that exchangeable bases in soils are important controls
on the base cation supply to, and ANC of, surface waters.

      A group of soil sulfate-related variables also is correlated with  base cations, ANC, and  pH of the
northeastern DDRP  lakes.  The variables include intensity and capacity isotherm parameters, water and
phosphate extractatde sulfate, the different forms of extractable aluminum, and possibly exchangeable
magnesium.  The variables appear in different combinations and with different signs in the regressions
for the different subregions. The bivariate correlations (Table 8-59) show strong positive correlations with
sulfate concentration at zero net adsorption; I.e., high equilibrium sulfate concentration (which is correlated
with sulfate deposition) is associated with  high base cation supply for the region overall.  The sulfate
isotherm variables SO4_EMX and SO4_SLP are correlated with low base  cation supply, ANC, and pH.
In the regressions for the northeastern region as a whole, most of the sulfate-related chemical parameters
are replaced  by  soil texture variables.  The sandier soils are associated with lower  base cations, ANC,
and pH in surface water; the soils with higher  clay content are associated with higher base  cations,
ANC.and  pH in  the water.   Further work is  needed to detail possible  mechanisms and  subregional
differences in these relationships.

8.9.6.1.2  Combined buffer aggregation -

      The variables selected by the stepwise regressions for the buffer zone aggregation are more similar
to the variables selected for the subregion models than to those selected by the models for the whole
region.  However, they have slightly lower R2 values than the whole-watershed  models,  and much lower
R2 values than the subregion models. Buffer zone models were not run for the subregions. These limited
results suggest that the buffer zone aggregation does not help in explaining variability in the surface water
chemistry.

8.9.6.2  Southern Blue Ridge Province

8.9.6.2.1  Whole watershed  aggregation -

      Base saturation  is most highly associated with these SBRP stream chemistry variables.   Other
associated variables for pH include  aluminum potential and the isotherm half-saturation constant.   R2
values range from 0.44 to 0.45, slightly higher than for the NE.  The bivariate correlations are consistent
with the multiple regression  results.  As for the NE, these relationships  support the  hypotheses that
exchangeable bases and mobile sulfate are important regulators of surface water chemistry.
                                               425

-------
8.9.6.2.2  100-m  buffer aggregation -

      The models selected for the buffer zone  aggregation are very similar to those for the entire
watershed aggregation.

8-9-7 Evaluation of Alternative Aggregation Schemes

      In order to examine the effect of the different aggregation schemes on the Level I Analysis results,
we ran several regressions using soil chemistry variables from the unadjusted aggregation scheme. The
results of these regressions are compared with the results from the watershed adjusted data in Tables
8-66 and 8-67.  Prior to regression analysis, a collinearity analysis was conducted.  Variables dropped
as a  result of this analysis are  marked by X's  in  the tables.   The  remaining  variables were used in
stepwise regressions with ANG and sulfate as the response variables.

      Examination of Tables 8-66 and 8-67 shows that many more candidate explanatory variables had
to be dropped from  the unadjusted data than from the adjusted data.  There were fewer instances of
multi-collinearity when using the  watershed aggregation.  Second, the regression models based on the
adjusted data generally explained more variance in the response variables than did the models based on
the unadjusted data. The only  exception to this  result is  SO416 in the SBRP.  The  adjustment for
watershed effect generally appears to Increase the explanatory  power of the soil chemistry variables.

8.9.8 Summary  and Conclusions

8.9.8.1 Alternative Aggregation Schemes

      The DDRP soil sampling and common aggregation scheme (unadjusted data) probably characterizes
regional and subregional means of soil properties well.  The common aggregation scheme appears to
have limitations, however, in characterizing the regional distribution of soil properties or the soil properties
of individual watersheds.  The common aggregation scheme biases individual watershed values toward
the regional mean value.  An alternative aggregation approach that uses a regression model to adjust for
watershed effects appears to adjust the  problem of bias toward the  regional mean but adds additional
uncertainty to the estimates of watershed soil chemistry.

      The common aggregation  scheme was used for  most Level II and III modeling because it was
the only data available at the time. The correlations and regressions conducted here used the watershed-
effects-adjusted data because they  have the most explanatory power for surface water chemistry.
Additional field work  would be needed to assess which aggregation scheme most closely mimics reality.
The characteristics of each aggregation  scheme  must be  kept  in mind when interpreting the results of
the models.

      Although the buffer zone and whole-watershed aggregation schemes do result in slightly different
values for some of the soil physical and chemical  variables, most differences are  probably not significant.
The buffer zone aggregation  does not result in  improved regression relationships for either the NE or
SBRP, thus the advantage of using one aggregation  scheme over the other for explaining index chemistry
is unclear.  The buffer zone aggregation was hypothesized to be more representative because it implicitly

                                              426

-------
Table 8-66. Results of Stepwise Multiple Regressions for DDRP Lake and Stream ANC (ALKANEW
and ALKA11) Versus Unadjusted and Watershed Adjusted Soil Physical and Chemical Properties
                                 NEb                             SBRPb
   Variable8         Unadjusted       Adjusted         Unadjusted       Adjusted
SAND
CLAY
FRAG
THKA
SOILDEN
CA CL
MG CL
SBC CL
BS CL
CEC CL
AC BACL
PH 01 M
AL AO
AL~CD
AL PYP
ALPOT
C TOT
SO4 H2O
SO4 PO4
SO4~EMX
SO4~B2
SO4 XIN
SO4 SLP
X







1
2(-)
X

X

X

X
X




X
3(-) X
1 - X
X
X
X
4
5(-) X

1
X
X
2

X
X
X
X

X


X
2








1
X












X
    R2                   0.33             0.43              0.29            0.44



    a Variable labels and units are found in Table 8-60.
     X's indicate variables dropped in collinearity analysis. Numbers indicate order of entry into stepwise model.  (-) Indicates
    a negative parameter estimate.
                                               427

-------
Table 8-67.  Results of Stepwise Multiple Regressions for DDRP Lake and Stream Sulfate (SO416)
Versus Unadjusted and Watershed Adjusted Soil Physical and Chemical Properties
                                  NEb                               SBRPb
   Variable8         Unadjusted       Adjusted          Unadjusted       Adjusted
SAND
CLAY
FRAG
THKA
SO1LDEN
CA CL
MG CL
SBC CL
BS CL
CEC CL
AC BACL
PH 01 M
AL AO
AL CD
AL~PYP
ALPOT
C TOT
SO4 H2O
SO4~PO4
SO4 EMX
SO4 B2
SO4 XIN
SO4 SLP
X
4(.) 5

5(.)
3





X
2(-)
X
4
X 6

X
X
3(-)
2
1
1
X
X
X
X
X
X

X 1


X X
X


X
X
X
X
3
X
K-) 29

X
2 X
    R2                    0.47             0.27              0.32             0.62



    a Variable labels and units are found in Table 8-60.
    b X's indicate variables dropped in collinearity analysis. Numbers indicate order of entry into stepwise model.  (-) indicates
    a negative parameter estimate.
                                                428

-------
weighted watershed values to take into account convergent flow and last hydrologic contact with the soil.
However, from our analyses, the importance of these characteristics appear to be minor for explaining
index water chemistry.  This may be  due to insufficient characterization of the buffer zones.  Only soil
mapping units greater than 6-10 acres were mapped; the effective buffer zones may be much smaller in
size,  A more  thorough  soil  characterization and evaluation of watershed hydrology  is necessary before
the importance of buffer zones in controlling stream chemistry can be determined.

8.9.8.2  Sulfate and Sulfur Retention

      The regression analyses indicated that the  sulfate isotherm parameters are strongly related  to
surface  water  sulfate. In the NE the important parameters are the equilibrium sulfate concentration and
the half saturation constant.  This is consistent with  the hypothesis that northeastern soils are near steady
state with respect to sulfate adsorption. In the SBRP the adsorption asymptote and the extractable sulfate
are important  explanatory variables for stream sulfate concentration.  These variables indicate soils that
are actively adsorbing sulfate.  Too much emphasis should not be placed on which particular isotherm
parameters  are selected  in the regressions,  since the  isotherm parameters  are themselves strongly
correlated.  It is significant, however, that the isotherm parameters are selected in both the NE and SBRP.
Even in  a region near steady state, the sulfate isotherm parameters yield  information about concentrations
of sulfate in the surface waters.

      Variables relating to soil acidity and base status are also important but do not  enter the regression
models  for the regions and subregions in a consistent manner. The relationship of surface water sulfate
concentration  and soil pH varies among the subregions  of the NE and is not  statistically significant for
the SBRP  watersheds.   In general  in  the NE, high concentrations of  sulfate in  surface  waters  are
associated  with low  pH and high extractable aluminum concentrations in the  soils.   In the SBRP, high
sulfate concentrations are associated with high concentrations of base cations in the soils.  The fact that
the two regions are approaching soil sulfate equilibrium from different directions  (declining deposition and
desorption  in the NE versus adsorption in the SBRP) may be  responsible for the variability observed in
the soil  chemical relationships.

      In general, the same soil  variables that  are  important in  predicting sulfate concentration  are
important In the regressions for sulfur net  retention,  but the coefficients of the variables  have opposite
signs.  Values of sulfur  retention are significantly higher in the SBRP then in the NE. This is consistent
with the lower observed equilibrium sulfate concentration (Table  8-58).

8.9.8.3  Ca plus Mg (SOBC), ANC,  and pH

      Soil pH, exchangeable base cations, and texture are most strongly  related  to lake ANC, pH, and
base cation concentrations in the NE.  Soil base saturation has the strongest relationships in the SBRP.
The sulfate isotherm parameters are more common as explanatory variables in the NE than the SBRP.
This is  consistent with  the mobile  anion hypothesis.  The drainage water sulfate concentration, and
therefore the sulfate isotherm parameters, is relatively less important In controlling ANC in the SBRP than
in the NE.

      Mean concentrations of Ca plus Mg are significantly higher in the NE than the SBRP, as expected
since soils in the SBRP are older and more highly weathered.  Northeastern soils also have a higher base

                                              429

-------
saturation on average than those in the SBRP.  The regressions and  measured soil and surface water
attributes support the hypothesis that soil base cation availability has a stronger effect on surface water
ANC than other soil chemical properties.

8.S.9  Summary Conclusions

      *     Soil variables important in explaining surface water sulfate  and watershed sulfur
           retention Include soil sulfate concentration and adsorption capacity, extractable
           aluminum, soil pH, and texture.

      •     Soil variables important in explaining ANC,  pH, and Ca plus Mg (sum of base
           cations)  in  surface water include  soil  base  saturation,  pH,  soil  sulfate
           concentration, and texture.

      •     Using a multiple regression modelling approach, measured soil chemical and
           physical properties alone can account for one quarter to  three quarters of the
           variance In ANC, sulfate, and base cations in the lake and stream waters of the
           DDRP regions and subregions.

      *     The DDRP soils data aggregation scheme using soil sampling classes masks a
           significant watershed  effect.  The aggregation  scheme  probably accurately
           characterizes regional mean values, but it draws all data toward the mean, and
           may affect the distribution of modelling results. Those results also will be drawn
           toward the mean, underestimating the possible response of watersheds having
           soil characteristics at the sensitive end of the distribution.

      ป     Aggregating soils by buffer zones near the lakes and streams does not generally
           result in  better correlations with index values  of surface water chemistry.
           Stronger  associations  would  likely  be  observed  between buffer  zone  soil
           characteristics and stormflow chemistry, when those soils are more hydrologically
           active.

8.10  EVALUATION  OF ASSOCIATIONS  BETWEEN  WATERSHED ATTRIBUTES AND  SURFACE
      WATER CHEMISTRY

8.10.1  Introduction

      This section evaluates the relationships between surface water chemistry and all of the watershed
attributes  measured in the DDRP.   Many watershed characteristics have  been shown to explain  a
significant portion of the variance in surface water chemistry when considered individually or in groups
of related variables (see Sections 8.2 through 8.9). The analyses In this section are designed  to integrate
and evaluate the various watershed attributes in explaining the variability in surface water chemistry. The
results are important in assessing whether the DDRP Levels II and  III  modelling efforts are  considering
the variables most important in controlling surface water chemistry.
                                              430

-------
8.10.2  Approach

      The candidate explanatory variables considered in this section include soil physical and chemical
properties, climate and deposition,  geology,  hydrology, physiography, vegetation,  and  land  use
characteristics.  Two basic categories of watershed attributes were used:  average attribute values for a
watershed and areal proportions of a watershed meeting specified criteria. Average attributes for each
watershed  include  means for depth to  bedrock, soil  base saturation,  soil  permeability,  deposition,
precipitation, and runoff values, among others. Mean watershed attributes were calculated by averaging
the values associated with mapped  areas on a watershed and weighting by the areal fraction of the
mapped area.  An overview of the procedure for aggregating soil variables is given in Section 8.9.3,  and
a description for the other watershed attributes  can be found in Turner et al. (1989).

      Although average values provide an integrated estimate of an attribute at the watershed level, such
values  do not  provide much information  about the  distribution  of an attribute  on a  watershed.
Furthermore, mean values cannot  be calculated for many attributes such as vegetation cover type or
geomorphic position.  Therefore a second category of attributes was developed in  order to estimate the
proportion of watersheds meeting specified  criteria.   Watershed proportions were  derived  from the
mapped data by summing the areal percentages of those mapping units on each watershed that satisfy
the specified criterion.

      Summary statistics for the subset of watershed attributes that were used in these analyses are
given in Tables  8-58 and 8-68.   Data derived  from field mapping activities are  described in Section
5.4.1.3,  and the land use/wetland data obtained from photointerpretation are explained in Section 5.4.1,6.
Deposition, precipitation, and runoff data were obtained as described in Sections 5,6 and 5.7.   The
regression modelling approach described in Section 8.9.2.2 was also used here.

8.10.3  Regional Characterization of Watershed Attributes

      Characteristics of the sampled  watersheds differ among the five subregions  and between the two
study regions.  The characteristics can be grouped into four categories: climate/deposition variables,
geologic parameters,  hydroiogic/physiographic  descriptors, and land use/vegetation variables.  Means
and standard errors for these means are  presented for each of the northeastern subregions, for the NE
as a whole, and for the SBRP in Table 8-68.

8.10.3.1 Northeast Subregions

      Long-term atmospheric deposition of ions  varies among the five subregions, despite approximately
equal precipitation amounts.  Sodium and chloride deposition are highest in the Southern New England
subregion, probably due to sea-salt deposition. The highest levels of calcium and magnesium deposition
are found in the Adirondacks and Southern New  England, and deposition of hydrogen ions and suifate
is highest in the Poconos/Catskills Subregion.
                                              431

-------
Table 8-68.  Population Means and Standard Errors for Selected Variables,  by Subregion/Region
and Aggregation


Variable/
Aggregation

Mean ฑ
: Standard Error by DDRP
Subregion/Region
NE Subregtorf
•> 1A
1B
1C
1D
1E
Overall
NE
SBBP

iiiii

5?S:Sg5:ySii*sP-i:iปJ*:;S
xffffffSfffilitiffyiKliff^fiA-A

ifllillltflllll

SSii:Ssi:ixl5l5SSI|^??::iil

CA_LTD c
~ WS
MG LTD
WS
NA LTD
~ WS
KLTD
WS
CM LTD
~ws
NH4 LTD
~WS
H LTD
WS
SO4 LTD
"ws
NO3 LTD
"ws
CLLTD
WS
PREC L
WS
RNOFT
WS
TMP AVQ
"ws
COASTD
WS
illili
SftwiSSSii:-?
SEO SEN
'ws
GEO MAX
"ws
SEO GT4
"ws
lllllll
ELMIN
~ WS
MAXREL
WS
SLP
WS
BUF
0.15 ฑ 0.01

0.07 ฑ 0.00

0,07 ฑ 0.01

0,03 ฑ 0.00

0.22 ฑ 0.01

0,25 ฑ 0.01

0.77 ฑ 0.01

0.73 ฑ 0.02

0.56 ฑ 0.01

0.05 ฑ 0.00

108.30 ฑ 1.27

70,26 ฑ 1.80

4.84 ฑ 0.14

297.83 x 6.25
||||||||1|||||||
%f$$ฃ$ฃ3J$!iij@%iljgi&

2.71 ฑ0.13

3.53 ฑ 0.23

0.82 x 0.61
0.10 * O.O1

0.06 ft 0.01

0.10 ฑ 0.01

0.02 ฑ 0.00

0.18 ft 0.01

0.21 ft 0.01

1.14 ฑ O.O2

0,97 ฑ 0.02

0.60 X 0.01

0.10 ฑ 0.01

111.52 * 1.29

65.99 ฑ1.69

8.29 X 0.27

128.57 ฑ11.81
lllil|lli;;il:||llll;i
;งSS™P"s5jSฅ:;:;:;:ii:i$i:lg;:i:

2.32 ft 0.15

2.70 ft 0.26

4.26 ฑ 3.55
0.08 x 0,01

0.08 ฑ 0.01

0.14 ฑ O.O2

0.02 X 0.00

0.16 * 0,01

0.15 x 0.01

0.66 x 0.01

0.61 x 0.02

0.41 x 0.01

O.O9 X 0.01

109.05 X 1.39

83.00 * 1.79

6,45 ฑ 0,28

100.62 ft 6.87



2,70 ฑ0.14

3.95 X 0.27

3.59 ft 1.99
0.10 x 0.01

0.15 ฑ 0.01

0.49 ฑ 0.07

0.02 ft 0,00

0.28 ft 0.02

0.14 ft 0.01

0.72 ft 0.02

0.71 ft 0.02

0.43 ft 0.02

0.48 ft 0.05

117.78 * 1.51

62.03 ft 1.69
"
9.18 ft 0.26

22.54 ft 5.03
ttlllilllii


2.15 ft 0.08

2.67 ft 0.27

1,44 ฑ 1.16


530.33 ฑ 19.74

155.43 * 15,82

20,02 ฑ 0,81
12,17 ฑ 0.89

398,32 ฑ 24.88

86.37 ft 16,88

9.98 ft 0.95
7.65 X 0.84

294.97 ฑ 26.02

220.97 ฑ25.11

14.39 ft 0,83
9.19 ฑ 0.82

89,05 ft 22.2S

42.36 X 6.16

10.59 ft 1.02
9.87 X 1.13
0.06 ฑ 0.01

0.10 x 0.01

0,26 ft 0.03

0.02 x 0.00

0,17 x 0.01

0.11 x 0.01

0.42 X 0.01

0.46 X 0.02

0.28 x 0,01

0.19 X 0.02

110.30 X 1.61

69.69 at 1.19

5.82 X 0.22

69.80 ฑ 13.74
illlBllll
•risS?:;Sfi:;??;:;'l;;;;;S:;

2,80 X 0.24

3.74 X 0.34

20.33 ft 6.21
0.10 ฑ 0.00

0.09 ft 0.00

0.20 ft 0.01

0.02 ft 0.00

0.19 X 0.00

0.17 ft 0.00

0.72 ฑ 0.01

0.88 ฑ 0.01

0.45 X 0.01

0.16 ft 0.01

110,88 4 0.65

64.71 ft 0.89

e.ea * 0.17

128.63 ft 9.58
ii;llliili;iiiii.
9:MSWfiM:-K™:::iKi.

2.58 ฑ 0.08

3.40 ฑ 0.14

6.87 X 1.80
0,20 ฑ 0.02

0,09 ฑ 0.01

0,14 x 0.01

0.03 ft 0.00

0.29 x 0,02

0.22 * 0.01

0.68 X 0.03

0.84 X O.O3

0.43 X 0.01

0.12 x 0.01

145.40 X 1.18

82.O6 X 3.75

13.15 X 0,26

. i;



2.15 ft 0.12

2.82 ft 0.25

.22 ft 0.32
mOsmm vs^i^^tt^MSSS^^^syA- ''•••' '.'•'**ฃ

160.08 ฑ 26.80

124,83 ฑ 15.41

13,01 ฑ 1.35
9.32 ft 1.04

301.63 X 17.98

134.80 ft 9.73

13.89 X 0,57
9.66 ฑ 0,45

566,48 X 37.48

537.01 X 83.21

37.75 ft 3.40
34.64 X 2.89
a 1A is the Adirondacks, 1B is the Poconos/Catskills, 1C is Central New England, 1D is Southern New England,
  and  1E is Maine.
  Variable labels and units are found in Table 8-70.
0 For each variable, WS refers to the entire watershed and BUF refers to the buffer zone.
                                                                                                 continued
                                                  432

-------
Table 8-68.  (Continued)


Mean :
t Standard Error by DDRP Subregion/Region
NE Subregion
Variable/
Aggregation
ATNMEAN
WS
ATKBMEAN
WS
GMP FTN
WS
BUF
LOW
WS
BUF

HYD SLW
~WS
BUF
DRN SLW
"ws
BUF
PERM
WS
BUF
PRM SLW
Iws
BUF
DEPTH
WS
BUF
BRD SHL
~WS
BUF
IPD SHL
~WS
BUF
AREA TEH
WS
AREA H2O
WS
WAIA
WS
VOL .
WS
DDENSfTY
WS
STRORDER
WS
lllill
FOREST
WS
BUF
CULTO
WS
BUF
PASTURE
WS
BUF
DISTURB
WS
BUF
WETLAND
WS
BUF
1A

7,90 ฑ 0.12

0.91 ft 0.13

20.31 ฑ3,10
46.29 ft 4.89

5.65 x 1.10
17.15 ft 3.30
Illlllllll

63.79 X 3.31
70.49 ft 4.48

12.06 ฑ 1.79
27.41 ft 3.28

6.87 + 1.07
7.75 x 1.68

33.57 ft 3.71
39.08 X 5.46

3.08 ft 0.76
4.73 x 0.93

46.48 X 3.96
25.66 ft 4.48

32.41 X 2.85
19.53 ft 2.88

356.54 x 67.42

46.31 ft 12.79

19.38 ฑ 4.99

2.09 ฑ1.33

O.48 X 0,11

1,92 ft 0.06
'••&ซ.. w.UWKwSViWAW
liiiiilllll

96,13 x 0.88
ฑ .

0.00 ft 0.00
. ฑ .

0.11 ft 0.08
. ft .

0.41 ft 0.21
ft .

3.35 ft 0.70,
. ft .
1B

8,47 ft 0,14

2.4O x 0.11

19.97 ft 3.08
40.15 ฑ 5.44

8.81 ฑ1.48
19.02 ฑ 4.14


9O.67 ft 1.76
92.87 x 2.08

42.82 ฑ 4.99
61, BO ft 443

2.36 ft 0.52
1.81 x 0.44

61.57 ft 5,94
70,80 ft 6.00

1,74 ft 0.30
2.23 ฑ 0,35

33.15 ft 4.57
18.84 ฑ 3.06

46.3O ft 4.16
52,03 ft 5.17

328,22 ft 78.84

• 26,87 ft 6,12

18.64 ft 5.75

0.87 ft 0.25

0.68 ft 0.18

2.88 + 0.07
;>Xv:ปBxซS-™-K: ,: .-ivX'SaW-foJ


75.88 x 471
ft .

1.OB ft 0.68
. ft .

1 1.62 X 4.24
. ฑ ,

6.46 ft 1.61
ft .

4.82 ft 1.11
. ฑ .
1C 1D

8.31 ft 0.09 8.40 ft 0.14

0.77 + 0.11 -0.82 + 0.37

28.28 ft 4.00 24.00 ฑ 5.21
52.05 + 4.88 36.36 + 6.32

6.68 x. 1.68 10.54 ft 2.19
19.34 + 3.11 17.4O ft 3.65
ซlllllllllllll^!iiillisi

65.45 ฑ 3.68 32.87 ft 6.60
68.85 ft 4.59 33.25 ft 6.69

23.31 ft 3,34 11.75 * 2.26
43.28 X 3,85 19.4O X 3,48

5.22 ft 0.71 21.74 ft 3.62
8.19 ft 1.70 22.63 ft 3.34

48.41 ft 3.50 . m33 ft 5.44
42.53 ft 450 10.34 ft 3.96

4.68 ft 0.87 15.44 ft 2.12
7.18 ft 1.18 17.15 X 1.97

25.39 ft 2.32 13.22 x 430
12.24 ft 2.31 8.16 ft 3.21

22,93 ft 3.71 8.98 ft 2.60
15.4O ft 3.03 6.13 X 2.03

671,77 X 128.58 190.82 ft 34.91

43.76+7.11 31.67 ft 7.98

28.75 + 4.85 8.68 ft 1.47

1.39 ฑ 0.58 0.74 ft 0.19

0.86 ft 0.18 0.32 ft 0.12

3.00 ft 0.00 3.00 ft 0.00
^f^^^f^^^.<-if^^!'fMf(gl:
i^iimmf^miKfMi^miiB

91.41 ft 1,48 75,65 ft 3.90
ft . ft .

0.44 ft 0.17 • 2.55 ft 1.41
. ft . . ft .

2.67 ft 0.72 2.30 x 1.42
. 4 . . ft ,

0.57 ft 0.15 13.99 ft 3.54
ft . ft ,

4.92 ft 1.23 5.51 ฑ 1.26
. ft . . ft .
1E

8.29 ft 0.18

1.47 ft 0.19

31.99 ft 3.98
52.94 ฑ 4.86

5.38 ft 0.94
15.82 ft 2.26


73.20 X 4.55
73.58 s 5.23

28.00 * 3,59
43.75 ft 4.89

5.81 ft 1.21
6.59 ft 1.56

43,71 ft 406
44.82 ฑ 4.86

3.17 S 0.62
4,55 ft 0.81

33.55 ft 4.83
21.09 + 448

33.21 ft 4.62
26.75 ft 421

681.50 + 138.86

95.61 ft 25.33

15.26 ft 2.29

4.18 ฑ 1.81

0.71 ft 0.14

3.68 ft 0.09
|iii|ง|S|ii|||ii:i

89.37 ft 1.78
ft .

3.38 ft 1.00
. ft .

1,72 ft 0.55
. ft .

O.8O ft 0.23
ft ,

4.73 ft 0.90
. ft .
Overall
NE

8.26 ft 0.06

1.03 ft 0.12

24.92 ft 1.79
46.58 ฑ 2.41

6.72 ft 0.87
17.68 ft 1.46


67.01 X 2.40
69.76 ft 2.69

24.04 ft 1.81
40.05 ft 2.26

7.52 ฑ 0.87
8.58 ft 1.00

41.18 ft 2.38
42.99 ฑ 2.78

4.96 s 0.58
6.52 ft 0.65

31.91 ft 2.09
17.90 ฑ 1.81

29.73 ft S.OO
24.5B ft 2.14

476.84 ft 51.34

52.00 ft 7.39

18.82 ft 2.02

2.03 ft 0,56

O.63 ft 0.07

2.92 ft 0.06


86.79 ft 1.36
ft .

1.48 ft 0.36
. ft .

3.51 ft 0.87
. ft .

3.59 ft 0.74
ft .

4.63 ft 0.47
.. ft .
SBBP

7,81 ft 0.09

, ft .

3.17 * 1.40
9.20 ฑ 3.34

2.33 ft 0.71
11.74 ft 2.90
il;lixK:l;':'-;i:1i:

5.32 ft 1.78
4.65 ft 1.38

1.89 ft 0.60
7.87 ft 2.28

5.08 ฑ 0.45
5.16 + 0.34

0.22 ft 0.16
0.69 ft 0.43

1,59 ft 0.13
2.01 ft 0.15

9.43 ft 2.26
8.69 ft 2.03

ft .
ft .

966.91 ft 213.37

0.58 ฑ 0.38

. ft .

ft .

1.03 X 0.17

2.23 ft 0.21
litiiL •••• -.-2

9O.2O ft 3.00
84.44 ft 4.51

1.77 ft 2.06
3.19 ft 2.73

6.84 ft 1,83
10.50 ft 3.28

1.10 ft 0.69
1.57 S 0.88

0.03 ft 0.07
0.31 ft 0.55
                                                                                             continued
                                                433

-------
Table  8-68.  (Continued)
                                      Mean ฑ  Standard Error by DDRP  Subregion/Region
                                             NE  Subregion                                              Overall
Variable/
Aggragation	1A	IB	1C	ID	IE	NE	SBRP

VQTJ3NF
   ~WS       16.00 * 2,51        6.18 ฑ  1.46       18.18 ฑ 3.17       27.20 ฑ 6.ซ       32.81 ฑ 4.83       20.56 ฑ 1.97        4.71 ฑ 1.55
    BUF      31.48 ฑ 4.52        11.55 ฑ  3.05       30.33 ฑ 5.16       26.31 ฑ 6.ZS       43.68 ฑ 6L01       30.19 ฑ 2,84        4.71 ฑ 2.04
vaT_oco
   ~WS       72.80 ฑ 3.49        7487 ฑ  4.85       45.21 ฑ 5.74       44.34 ฑ 7.88       33.21 ฑ 4.82       53.25 ฑ 2.82       38.75 ฑ 7.69
    BUF      44.15 ฑ 4.94        66.25 ฑ  4,44       28.50 ฑ 542       42.93 ฑ 7.87       24.31 ฑ 4.68       39.91 ฑ 2.78       37.07 ฑ 5.71
VQT_DRY
   ~WS        0.46 ฑ 0.27        16.96 ฑ  4.86        4.28 ฑ 1.15        9.80 ฑ 3.60        7.84 ฑ 1.82        7.48 ฑ 1.S2        9.49 ฑ 3.00
    BUF       0.48 ฑ 0.37        15.44 ฑ  3.91        4.30 ฑ 1.58        6.62 ฑ 1.64        404 ฑ 1.13        5.84 ฑ 0.96       14.94 ฑ 4.53
VGTWET
   ~WS        1.BO ฑ 0.57        2.58 ฑ  1.18        2.48 * 0.89        4.27 ฑ 1.26        Z20 * 0.59        2,50 ฑ 0.37        0.03 ฑ 0.08
    BUF       7.88 ฑ 2.45        6.28 ฑ  2.67        8.57 ฑ 2,30        6.71 ฑ 1.74        7.08 ฑ 1.71        7.07 * 1.01        0.12 ฑ 0.19
                                                                434

-------
     Although differences in bedrock sensitivity (GEO_SEN), expressed using the DDRP weatherability
index (see Section 8.4), are not large, in the Maine Subregion nearly 20 percent of the watershed area
is of sensitivity class > 4 (GEO_GT4).  This is larger than for any other subregion.

     Elevation of the sample lakes (EL_MIN)  is highest in the Adirondacks and lowest in Southern New
England.  Maximum relief (MAXREL) is highest in the Central  New England Subregion and lowest in
Southern New England.  The percentage of watershed area in foot or toe slope (GMP_FTN) is highest
in Maine. The highest percentage of land with  poor drainage and permeability characteristics (HYD_SLW,
DRN_SLW, PERM_SLW) Is consistently found  In the Poconos/Catskills Subregion and the lowest in  the
Southern New England Subregion. Soils in Southern New England have greatest mean depth to bedrock
(DEPTH), relatively few shallow (< 50  cm) Impermeable layers (IPD_SHL), and are  generally coarser-
textured than soils on watersheds in other subregiqns  (see Section 8.9.4).

     Vegetation and land use characteristics vary somewhat among the subregions. No appreciable
area of cultivated land occurs in the Adirondacks  watersheds.   The Southern New England watersheds
have the largest  proportion of urban or disturbed  land. Watersheds of the Southern  New England and
Maine Subregions contain the most coniferous cover,  and deciduous forest coverage is greatest in  the
Adirondacks and Poconos/Catskills watersheds. The percent of open, dry vegetation (open, non-forested
land that is not wetlands) is greatest in the Poconos/Catskills Subregion; the Adirondacks Subregion  has
almost  no area of open, dry vegetation.   The open,  dry vegetation class  often indicates pasture or
abandoned farm land.

8.10.3.2 Northeast and Southern Blue Ridge Providence

     Current atmospheric deposition  is  higher  in the SBRP than in  the NE.  The mean annual
temperature in the SBRP Is  nearly twice that of  the  NE.  Northeastern  watersheds generally contain
more weatherable bedrock than the SBRP.  Elevation, maximum relief, and  slope are all higher in  the
SBRP.   However, the  minimum elevation  in the  SBRP is approximately  equal  to the minimum of  the
Adirondacks subregion.  The  northeastern  watersheds include  a higher percentage  of "wet" or poorly
drained  soils (based  on HYD_SLW, DRN_SLW, and PERM_SLW).  Despite  a  greater mean depth of
bedrock, the NE has a higher percentage of watershed  area overlying shallow  (<50 cm) bedrock.
Watersheds are  generally larger  in the SBRP and  contain  larger percentages of area  in pasture;
watersheds in the Ni have a greater percentage of wetlands and more coniferous vegetation.

     The buffer zones are characterized by lower slopes, a higher percentage of foot and toe slopes,
and a slightly higher  percentage of land with "slow"  drainage.   For the Adirondacks and Poconos/
Catskiils Subregions, the buffers also included more soils with hydrologic group C or D and permeability
of < 3.  In general, soils in the buffer zones  are  slightly deeper, with a lower percentage of area with
shallow  bedrock relative to the whole watershed.  The buffer zones  contain a higher percentage of
lowlands and more coniferous vegetation in the NE. In  the NE the percent  open, dry vegetation is slightly
less in the buffer zones than in  the whole watersheds, although  the percent open,  wet vegetation is
higher.  In the SBRP the percent area with open, dry vegetation is greater in the  buffer zones than in the
whole watersheds.
                                             435

-------
8.10.4  Sulfate and Sulfur Retention

      This section and Section 8.10.5 discuss the statistical relationships between selected watershed
attributes and the water chemistry for the DDRP regions. These relationships are evaluated in terms of
potential cause-effect controls on surface water chemistry and  identification of any important controlling
factors  which  are  not accounted for.   Tables  8-69 and  8-70 show the nonparametric Spearman
correlations between selected watershed attributes and each of  the water chemistry variables considered.
Correlations with soil properties were shown in Tables 8-59 and 8-60.  Results of the stepwise multiple
regressions for suifate and sulfur retention are given In Tables  8-71 and 8-72.

8.10.4.1  Northeast

      The coefficients of determination, or R2 values, range from 0.45 to 0.83 for suifate and from 0.34
to 0.82  for sulfur  retention in the NE.   In general,  watershed attributes that had higher bivariate
correlations with  water chemistry were selected as explanatory variables in the stepwise regressions.
For the northeastern region as a whole, the strongest association is between lake suifate concentration
and long-term total suifate deposition.  Precipitation amount and runoff are also  highly associated, with
a  negative  sign, which  probably indicates  a  dilution effect.   The  suifate  isotherm  half-saturation
concentration is the most highly associated soil chemistry variable, consistent with the regression results
using only soil chemistry variables (Table 8-61). Watersheds having greater areas of poorly drained foot
and toe slopes and lowlands generally have lower lake suifate concentrations; these may be areas of
suifate reduction. Watersheds with shallow bedrock or shallow impermeable soil  layers have higher lake
suifate concentrations.  Open dry vegetation is correlated with  high lake suifate;  open wet vegetation is
correlated with lower lake suifate.  Sandy soils are associated with higher lake suifate.  There are some
differences in the variables selected in the regressions for the  subregions, but  most are correlated with
those selected for the region as a whole.

      These  results  are  consistent  with those  discussed elsewhere in Section  8.   Lake  suifate
concentration is largely dependent on atmospheric deposition of suifate as modified by amount of runoff,
suifate adsorption-desorption characteristics of the soil, and soil depth and texture.  Suifate reduction in
wetlands and/or flooded soils can also reduce sulfur concentrations and therefore affect budgets in some
northeastern watersheds.  Suifate retention or  release resulting from  wetting and drying of soils during
seasonal cycles or over longer periods of wet or drought years can substantially influence watershed
sulfur status based  on measurements made at one point in time. The extent to  which these processes
and thus sulfur budgets are in equilibrium with atmospheric deposition  or are acting as long-term sinks
cannot  be determined with certainty from these analyses; observed relationships  suggest reduction may
provide long-term watershed sulfur sinks (See also Section 7).

8.10.4.2  Southern Blue  Ridge Province

      The first variables selected by the stepwise regressions for the SBRP are identical to those selected
for the  soil properties alone (Tables 8-61 and 8-62).  These  are  exchangeable  magnesium, water-
extractable suifate, and adsorption capacity (negative) for stream suifate, and exchangeable magnesium
(negative), base saturation,  and adsorption capacity for sulfur  retention.  For stream  suifate, additional
watershed attributes  selected  are  runoff and  soil permeability (both  negative), and  slope,  generally
consistent with relationships  seen in the NE,  No additional variables were selected for sulfur retention.

                                               436

-------
The bivariate correlations also show  relationships between stream sulfate and precipitation  (negative),
shallow soils, and foot and toe slope soils.  The latter is the opposite of that seen  in the NE; these areas
in the SBRP are much better drained than comparable areas of northeastern watersheds, and may be
retaining less  sulfur through  sulfate  reduction.   They also  have high exchangeable magnesium  and
calcium, and may be adsorbing  less sulfate.  In the SBRP, sulfate deposition data are not related to
stream suifate data, as discussed in Section 8.2.

8.10.5  Ca Plus Ma (SOBC). ANC. and pH

      Results of stepwise multiple regressions for Ca  plus Mg,  ANC, and pH are given in Tables 8-73
through 8-75.   This section  summarizes the  results and discusses potential cause-effect controls on
surface water chemistry.   As discussed  in Section 8.9.6, these water  chemistry  variables  are highly
correlated with each other and often show similar relationships with explanatory variables in the multiple
regressions and in the bivariate correlations (Tables 8-69 and 8-70).

8.10.5.1  Northeast

      The coefficients of determination, or R2 values, for multiple regressions on  these water chemistry
variables range from 0.50 to 0.91 for the northeastern subregions.   Percentage area on the watershed
having open  dry vegetation  is often  selected as the first variable  in the models.  These are  areas in
pasture, cultivation, urban, and other  disturbed land. They generally coincide with  high base saturation
and deep, relatively fiat soils, all  of which correlate well with these  dependent variables.   The open dry
areas often  are disturbed, exposing  fresh weathering surfaces, have been limed or fertilized, or are
associated with land uses that  contribute base cations  to runoff.  Precipitation  amount is inversely
correlated with all three dependent variables, probably a dilution effect. Bedrock  weatherability index is
related to these dependent variables as seen in several of the regressions and the bivariate correlations.
Soil pH, base status, extractable aluminum,  and several  sulfate isotherm parameters are also  related,
along with soil texture.   These  have been discussed in more detail in Section  8.9.6.1.   Watershed
attributes including soil depth, permeability, and area in low geomorphic position are commonly correlated
explanatory variables. Where water can infiltrate rapidly and follow deep flow paths to  contact with high
base saturation soils or weatherable minerals, base cation supply is high. There does not seem to be
any single good index of these characteristics that is common to all subregions,  but combinations of
several  indices  in  the  multiple  regressions  lead  to  reasonably  good  explanatory  power for  most
subregions.  This suggests the importance of knowing hydrologic characteristics of a watershed to explain
water chemistry.

8.10.5.2 Southern Blue  Ridge  Province

      In the  SBRP as in the NE,  the variable  most  strongly  associated with these water chemistry
variables is area in open  dry vegetation.  As  in the NE, the geologic weatherability index is related to
base  cation supply.   Runoff,  Ca+Mg deposition (probably a surrogate for precipitation), permeability,
elevation, slope, and relief all have regression  estimates or bivariate correlations with negative signs; the
more water that  passes quickly  through the  watersheds into the  streams, the  lower the base cation
concentrations and pH. The R2 for these models ranges from 0.65 to 0.85.
                                               437

-------
Table 8-39. Non-parametric Correlations Between Lake Chemistry Variables and Selected Watershed
Attributes for the NE DDRP  Watersheds
Variable
Spearman Correlation Coefficients Significant at 0.05 / Significance Level / N = 143

               Units         SO416"      SO4 NRET    CAMG16    ALKANEW   PHEQ11


CAJ.TD
Calcium deposition, long-term
MGJ.TD
Magnesium deposition, long-term
NAJ.TD
Sodium deposition, long-term
KJ.TD
Potassium deposition, long-term
CM_LTO
Calcium-)- magnesium, long-term
NH4J.TD
Ammonium deposition, long-term
H_LTD
Hydrogen ion deposition.long-term
SO4_LTD
Sulfate deposition, long-term
NO3_LTD
Nitrate deposition, long-term
CLJ.TD
Chloride deposition, long-term
PRECJ.
Precipitation, long-term
RNOF_T
Runoff, long-term
TMP_AVG
Avg. Temp., long-term
COASTD
Distance to coast
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
cm
cm
C
km

iiiiiiiHliiili^^!iiii
0,41920 -0.16242
0.0001 0.0526

-0.24646
0.0030

0.17295
0.0389
0.49008
0.0001
0.55209 0.21190
0.0001 0.0111
0.57387 0,22209
0.0001 0.0077
0.53753
0.0001
0.17202
0.0399

-0.32362 -0.30330
0.0001 0.0002
0.24862 0.29674
0.0028 0.0003
0.21602
0.0096

-0.34929
0.0001
-0.29001 -0.22988
0.0004 0.0057

-0.26888 -0.26041
0.0012 0.0017
-0.33078 -0.41435
0.0001 0.0001
-0.22532
0,0068

-0.20025
0.0165
-0.26048
0.0017

-0.32169 -0.30643
0.0001 0.0002
-0.24870
0.0027


-0.34727
0.0001
-0.21387
0.0103

-0.24346
0.0034
-0.40138
0.0001
-0.23133
0.0054

-0.20518
0.0140
-0.26870
0.0012

-0.30034
0.0003



a SO416 is the lake surf ate concentration, SO4_NRET is watershed sulfur retention, CAMG16 is the lake calcium + magnesium
  concentration, ALKANEW is the lake acid neutralizing capacity, and PHEQ11 is the air-equilibrated fake pH.

                                                                                               continued
                                                  438

-------
Table 8-69.  (Continued)
Spearman Correlation Coefficients Significant at 0.05 / Significance Level
Variable
Units
SO416 S04 NRET
CAMG16
/ N = 143
ALKANEW

PHEQ11


GEO SEN
Geological weather.
GEO MAX
Geological weather.
GEO GT4
Geological weather.

index, mean

index, max.

index > 4 percent

EL AVG
Elevation, average
MAXREL
Relief, maximum
SLP
Slope, mean
ATNMEAN
ln(a/tan f>), mean
ATKBMEAN
ln(a/kbtan ,3), mean
GMP FTN
Footslope, toeslope,
LOW
Lowlands

m

m

percent





flood plain percent

percent

HYD SLW
Hydrologic group C
DRN SLW
Drainage class < =
PERM
Permeability, mean
PRM SLW

or D percent

3 percent

cm/hr

-0.17501
0.0368
-0.18158
0.0300
-0.21270
0.0108
;;;:tt:;:;$-i::%>ฃ^
•KiWi-i'i^rWSIOPr^pny ™ : -ffxSS^:::::^^^
•:^::-y:$:**:^^
0,27059
0.0011
-0.22345
0.0075


0.27666
0.0008


-0.27776
0.0008


mmmmmmmmmmmmmm


0.16782
0.0451



Permeability class <— 3 percent
0.22226
0.0076
0.27023
0.0011
0.26025
0.0017





-0.18579
0.0263


0.29688
0.0004





0.29712
0.0003
0.36531
0.0001
-0.34268
0.0001
0.32347
0.0001
0.22715
0.0064
0.31602
0.0001
0.30655
0.0002









0.27479
0.0011
0.19420
0.0201


-:-:-:•:-:•;.;....-.-;-,-: -:-:-:-:•:-;•:•:•:•:•;-••:-:••• :•:•;-
xฃ:::':':'::i:;:;::::"::!':-:'i:::X'!:;-X':''-:': -,: '
:';XvXฃ:;I :•:•'--'.;•-;. vX, v>. ;.;.;-; ;';* ;.;>':''
0.24800
0.0028
0.34550
0.0001
-0.33047
0.0001
0.39821
0.0001
0.21625
0.0095
0,31118
0.0002
0,30405
0.0002
;if'::;-'- '•' :^;s;ฅl:
-0,16687
0,0464






0.26179
0.0019
0.18822
0.0244


Sll^:/.;tl!
0.22654
0.0065
0.33196
0.0001
-0.31449
0.0001
0.38357
0.0001
                                                                                        continued
                                              439

-------
Table 8-69.  (Continued)
Spearman
Variable
DEPTH
Bedrock depth, mean
BRDJ.T2
Bedrock class <= 2 (100 cm)
BRD_SHL
Bedrock <= 50 cm
IPD_SHL
impermeable layer <= 50 cm
AREA_TER
Area, terrestrial
AREA_H2O
Area, water
WALA
Watershed area/lake area
VOL
Lake volume
DDENSITY
Drainage density
STRORDER
Stream order, maximum

FOREST
Forested land
CULTIV
Cultivated land
PASTURE
Pasture/grazed land
DISTURB
Disturbed land
WETLAND
Wetland
VOT_CNF
Vegetation, coniferous
VGT_DCD
Vegetation, deciduous
Correlation Coefficients Significant
Units SO416
-0.18515
m 0.0268
percent
percent
percent
-0.17159
ha 0.0404
ha
-0.18537
0.0267
m3

-0.37781
0.0001

percent
percent
percent
0.26987
percent 0.0012
percent
-0.28421
percent 0.0006
0.33794
percent 0.0001
at 0.05 / Significance Level /
SO4 NRET CAMG16

-0.18181
0.0298
-0.18022
0.0313
0.27970
0.0007
0,21169
0.0111



0.18521
0.0273
0.16568 0.22011
0.0488 0.0085

-0.23145 -0.32069
0.0056 0.0001
0.20787
0.0131
0.19619 0.39902
0.0193 0.0001
0.33590
0.0001
0.26558
0.0014


N = 143
ALKANEW



0.21578
0.0096
0.26421
0.0014
0.20137
0.0159


0.22157
0.0081
0.35154
0.0001
Iliiillilillll
-0.22830
0.0063
0.24694
0,0030
0.44050
0.0001
0.28701
0.0005




PHEQ11



0.20136
0.0159
0.26093
0.0016
0.22224
0.0076

0.17854
0.0335
0.21157
0.0115
0.35029
0,0001

-0.20495
0,0144
0.24764
0,0030
0.42920
0.0001
0,28373
0.0006



                                                                                        continued
                                              440

-------
Table 8-69.  (Continued)
                  Spearman Correlation Coefficients Significant at 0.05 / Significance Level / N = 143

Variable	Units	SO416	SO4 NRET    CAMG16     ALKANEW    PHEQ11

VGTJDRY                                                   0.19488        0.39888       0.43434     0.41717
Vegetation, dry open               percent                    0.0197         0.0001        0.0001      0.0001

VGT.WET                                     -0.19436       0.18205
Vegetation, wet open               percent      0.0200        0.0295
                                                    441

-------
Table  8-70.  Non-parametric  Correlations  Between  Stream  Chemistry  Variables  and  Selected
Watershed Attributes for the SBRP DDRP Watersheds
Spearman Correlation Coefficients Significant at 0,05 / Significance Level / N = 31
Variable Units SO4168 SO4 NRET SOBC ALKA11 PHEQ11


CA_LTD
Calcium deposition, long-term
MGJ.TD
Magnesium deposition, long-term
NAJ.TD
Sodium deposition, long-term
KJ.TD
Potassium deposition, long-term
CM_LTD
Calcium -1- magnesium deposition
NH4J.TD
Ammonium deposition, long-term
HJ.TD
Hydrogen ion deposition, long-term
SO4J.TD
Sulfate deposition, long-term
NO3J.TD
Nitrate deposition, long-term
CL_LTD
Chloride deposition, long-term
PREC_L
Precipitation, long-term
RNOFJ
Runoff, long-term
TMP_AVG
Avg. Temp., long-term

keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
keq/ha
cm
cm
C





-0.47863
0.0065
0.40484
0.0239

-0.38306
0,0334
-0.36371
0.0443
-0.44960
0.0112
-0.46914
0.0078

a SO416 is the stream sulfate concentration, SO4_NRET is watershed sulfur retention, SOBC is the stream sum of base cation
  concentration, ALKA11 is the stream acid neutralizing capacity, and PHEQ11 is the air-equilibrated stream pH.


                                                                                             continued
                                                 442

-------
Table 8-70.  (Continued)
                  Spearman Correlation Coefficients Significant at 0,05 / Significance Level / N = 31

Variable                          Units         SO416        SO4 NRET     SOBC       ALKA11      PHEQ11
QEO_SEN
Geological weather. Index, mean

GEO_MAX
Geological weather, index, max.

GEO_GT4
Geological weather, index > 4
percent


EL_AVG
Elevation, average
EL_MAX
Elevation, maximum
MAXREL
Maximum relief
SLP
Slope, mean
ATNMEAN
In (a/tan p), mean
GMP_FTN
Footslope, toeslope, flood plain
LOW
Lowlands
?^?f^^S^ft^^^'^^^ฎS^^^ft^?S^i;S^:SSS^
If^^^^^l^S^KRK^iii'flS^^^^tillif^l^
m
m
m
percent

0.51896 -0.43147
percent 0.0028 0.0154
percent
iggjSsi^
HYD_SLW
Hydrologic group C or D
DRN_SLW
Drainage class <= 3
PERM
Permeability, mean
PRM_SLW
Permeability class <= 3
percent
percent
cm/hr
percent
Ifllllli
-0.38149
0.0342
-0.40040
0.0256
-0,36411
0.0440
-0.40282
0.0247

0.40300
0.0246
0.50570
0,0037


0.36819
0.0416
-0,40524
0.0237
0.46569
0.0083
|||;||||||l|l|:;;*s;
-0.36072
0.0462
-0.40484
0.0239
-0.39194
0.0292
-0.43185
0.0153


0.47039
0.0076
iiilpilllisi

0.37350
0.0385
-0.45887
0.0094
0.48854
0.0053
iij^jgi



-0.34798
0.0551

0.49028
0.0051
0.53204
0.0021
ISiftii:---;- *-;ป:f?il|i

0.41309
0.0209
-0.36089
0.0461
0.56103
0.0010
                                                                                                   continued
                                                    443

-------
Table 8-70.  (Continued)
Spearman
Variable
DEPTH
Bedrock depth, mean
BRDJ.T2
Bedrock class <= 2 (100 cm)
BRD_SHL
Bedrock <= 50 cm
IPD_SHL
Impermeable layer <= 50 cm
AREAJTER
Area, terrestrial
AREA_H2O
Area, water
DENSITY2
Drainage density, NSS
STRORDER
Stream order, maximum

FOREST
Forested land
CULTIV
Cultivated land
PASTURE
Pasture/grazed land
DISTURB
Disturbed land
WETLAND
Wetlands
VGT_CNF
Vegetation, coniferous
VGTJ3CD
Vegetation, deciduous
VGT_DRY
Vegetation, dry open
VGT_WET
Vegetation, wet open
Correlation Coefficients Significant at 0.05 / Significance Level / N = 31
Units SO416 SO4 NRET SOBC ALKA11 PHEQ11
m
0.37192
percent 0.0394
0.37383 -0.38794
percent 0.0383 0.0310
percent
ha
ha
0.35286
0.0515

fcxmsmss;^^

-0.62067 -0.65401 -0.60673
percent 0.0002 0.0001 0.0003
0.42443 0.49383 0.53198
percent 0.0173 0.0048 0.0021
0.54318 0.56287 0.49295
percent 0.0016 0.0010 0.0048
percent
ha
percent
percent
0.60522 0.61150 0.55585
percent 0.0003 0.0003 0.0012
percent
                                           444

-------
Table 8-71.    Results  of  Stepwise Multiple Regressions  for  DDRP  Lake and  Stream  Sulfate
Concentration (SO416) Versus Watershed Attributes

Soil Physical
Properties



Soil Chemical
Properties














Deposition/
Climate





Geology
Physiography





Hydrology







Vegetation


Subrealon8 Reaion
Variable" 1AC 1B 1C 1D 1E NE SBRP
SAND 23 8
CLAY 3 4
FRAG 1(-) 5(-)
THKA 7(-)
SOILDEN
CA CL
MCTCL 1
SBC" CL 4(-) 4(-)
BS CL
CEC CL
AC SACL
PH~01M
AL~AO 3 4
AL~CD 1
AL~PYP
CTOT
SD4 H2O 3
SO4 PO4 2
SOTEMX 2(-)
SO4~B2 3
SO4~XIN
CMITD
NH4 LTD
H LTD
SD4 LTD 1
PREC L 2(-)
TMP AVG
COASTD 1 5
GEO MAX 2(-)
EL AVG 2
MAKREL 6(-)
SLP 6
ATNMEAN 4(-)
ATKBMEAN
GMP FTN 9(-)
PERKS 5(-)
DEPTH
BRD SHL
IPD "SHL 4 5
AREA TER 3(-)
AREA~H2O 5
WAUT
DDENSITY
VGT CNF
VGTDRY 1 3 5
VGTTWET K-)
                                         0.45
0.76      0.83      0.78
                                                                               0.64
0.64
                                                                                                 0.73
  1A is the Adirondacks, 1B is the Poconos/Catskills, 1C is Central New England, 1D is Southern New England,
  and 1E is Maine.
 1 Variable labels and units are found in Table 8-69.
  Numbers indicate order of entry into Stepwise model,  (-) indicates a negative parameter estimate.
                                                  445

-------
Table 8-72.   Results of  Stepwise Multiple  Regressions for DDRP  Watershed Sulfur  Retention
(SO4  NRET) Versus Watershed Attributes	
                                                       Subreaion
                                                  Region
                      Variable13
1AC
18
1C
1D
1E
NE
SBRP
Soil Physical
Properties



Soil Chemical
Properties














Deposition/
Climate






Geology
Physiography





Hydrology







Vegetation


SAND 5{-)
CLAY
FRAG
THKA 4
SOILDEN
CA CL
MG~CL 2(-)
SBC" CL 42
BS CL 3
CEC CL
AC SACL
PH~01M
AL~AO
AL~CD 2(-)
AL PYP 1(-) 5(-)
CTOT
S04 H2O
SO4 PO4 5(-) 2
SO4~EMX 5 4 1
SO4~B2 1(-)
SO4"XIN
CMT.TD
NH4~ LTD
H LTD
SD4 LTD 3
PREC L 4(-) 3
RNOFT 3(-) 2(-)
TMP AVG 1
COASTD 2(-)
GEO MAX 3(-) 4(-) 1
EL AVG 1(-) 3(-)
MAXREL
SLP
ATNMEAN 5
ATKBMEAN
GMP FTN
PERW
DEPTH
BRD SHL
IPD BHL 2(-)
AREA TER
AREA~H2O
WAUT
DDENSITY
VGT CNF
VGTTDRY 3(-)
VGTWET 4(-) 1
                       R2
0.64      0,76      0.82      0.76      0.69      0.34      0.44
a 1A is the Adirondacks, 1B is the Poconos/Catskills, 1C is Central New England, 1D is Southern New England,
  and  1E is Maine.
  Variable labels and units are found in Table 8-69.
c Numbers indicate order of entry Into  stepwise model. {-) indicates a negative parameter estimate.
                                                  446

-------
Table 8-73.  Results of Stepwise Multiple Regressions for DDRP  Lake Calcium Plus Magnesium
Concentrations (CAMG16) and Stream Sum of Base Cations (SOBC) Versus Watershed Attributes

Soil Physical
Properties



Soil Chemical
Properties














Deposition/
Climate






Geology
Physiography





Hydrology







Vegetation


Subreqton" Reoion
Variable15 1AC 1B 1C 1D 1E NE SBRP
SAND 4(-)
CLAY
FRAG
THKA 3(-) 5(-)
SOILDEN
CA CL
MG~CL 2
SBC" CL
BS CL
CEC CL
AC BACL
PH~01M 5 3
AL~AO 2 11(-)
AL~CD
AL~PYP 3(-)
CTOT
SD4 H2O 4
SO4"PO4
SO4~EMX 5(-) 3(-)
SO4~B2
SO4~XIN 3 2(-) 4
CM "LTD 4(-)
NH4~ LTD
H LTD
SC4 LTD
PREC L 1(-) 2(-)
RNOF~T 5(-)
TMP AVG 4(-) 1(-)
COABTD 6
GEO MAX 1 12
EL AVG 2(-) 10(-)
MAXREL
SLP
ATNMEAN
ATKBMEAN
GMP FTN
PERI51 3(-) 9(-) 5(-)
DEPTH 5 8
BRD SHL
IPD BHL 2 7
AREA TER
AREA~H2O 3(-) 4
WALA~
DDENSITY
VGT CNF 5
VGTDRY 1 1211
VGT~WET 4
                                       0.88
                                                0.91
0.83
                                                                  0.81
                                                                           0.83
                           0.68
                                                                                             0.85
a 1A is the Adirondacks, 1B is the Poconos/Catskills, 1C is Central New England, 1D is Southern New England,
  and 1E is Maine.
  Variable labels and units are found in Table 8-69.
0 Numbers indicate order of entry into stepwise model. (-) indicates a negative parameter estimate.
                                                447

-------
Table 8-74.  Results of Stepwise Multiple Regressions for DDRP Lake and Stream ANC
(ALKA11, ALKANEW) Versus Watershed Attributes

Soil Physical
Properties



Soil Chemical
Properties














Deposition/
Climate






Geology
Physiography





Hydrology







Vegetation


Variable"
SAND
CLAY
FRAG
THKA
SOILDEN
CA CL
MG~CL
SBC" GI-
BS CL
CEC CL
AC SACL
PK~01M
AL~AO
AL~CD
AL~PYP
CTOT
SD4 H2O
SO4 PO4
SO4 EMX
SO4~B2
SO4~XIN
CM "LTD
NH4~ LTD
H LTD
SD4 LTD
PREC L
RNOF~T
TMP AVG
COABTD
GEO MAX
EL ff/G
MAXREL
SLP
ATNMEAN
ATKBMEAN
GMP FTN
PERW
DEPTH
BRD SHL
IPD "BHL
AREA TER
AREA~H2O
WALA~
DDENSITY
VGT CNF
VGT~DRY
VGTWET
Subreaiona Reaion
1AC 1B 1C 1D 1E NE SBRP
3(-) 40

4
6(-)

3 7


2


43 4
6(->
5(-)



8(-)


2 2(->



2
4f) 2(-) 2(-)

5(-) 1(-)

1 5 3
3(-)
5







2

4(.)

5f)
3 1f)
1 1311
4
                      R2
                                        0.85
0,84
0.76
0.87
0.78
0.61
                                              0.82
  1A is the Adirondaeks, 1B is the Poconos/Catskills, 1C is Centra! New England, 1D is Southern New England
  and 1E is Maine.
  Variable labels and units are found in Table 8.9.5-1.
0 Numbers indicate order of entry into stepwise model.  (-) indicates a negative parameter estimate.
                                                 448

-------
Table 8-75.  Results of Stepwise Multiple Regressions for DDRP Lake and Stream Air Equilibrated
pH (PHEQ11) Versus Watershed Attributes

Soil Physical
Properties



Soil Chemical
Properties














Deposition/
Climate






Geology
Physiography




Hydrology







Vegetation


Subreaion3 Reaion
Variable" 1AC 18 1C 1D 1E NE SBRP
SAND 8(-)
CLAY
FRAG
THKA
SOILDEN
CA CL
MG~CL 4
SBCJCL
BS CL
cec CL
AC BACL
PH~01M 1 3412
AL AO 2 3(-)
AL CD 4
AL PYP 2(-)
CTOT
804 H2O 1(-)
SO4~PO4
SO4 EMX 2(-) 2(-)
SO4~B2
SO4~XIN 4(-)
CMITD
NH4~ LTD
H LTD
S04 LTD
PREC L 2(-) 4f)
RNOF~T 4(-)
TMP AVG 4(-) 1(-)
COASTD
GEO MAX 1 5
MAKREL 6
SLP 3
ATNMEAN
ATKBMEAN
GMP FTN 5(-) 3
PERSB
DEPTH
BRD SHL 7(-)
IPD BHL 2
AREA TER 3 5
AREA H2O
WALA~
DDENSITY
VGT CNF 1(-)
VQT~DRY 3 3 1
VGT~WET
                       FT                0.65     0.87      0.77      0.56       0.85      0.50      0.65

a 1A is the Adirondacks, 1B is the Poconos/Catskills, 1C is Central New England, 1D is Southern New England,
  and  1E is Maine.
  Variable labels and units are found in Table 8.9.5-1,
c Numbers indicate order of entry into stepwise model.  (-) indicates a negative parameter estimate.
                                                  449

-------
8.10.6  Summary and Conclusions

8.10.6.1, Sulfate and Sulfur Retention

      In the NE, sulfate deposition, precipitation amount, and watershed hydrologic characteristics have
the strongest associations with surface water sulfate concentration and watershed sulfur retention.  In the
SBRP, soil chemical variables (sulfate adsorption capacity, exchangeable magnesium, base saturation, and
water-extractable sulfate) have the strongest associations with stream sulfate concentration and watershed
sulfur retention.  For stream  sulfate concentration, watershed hydrologic variables also entered the
regressions In the SBRP.  Since soils in the NE are near steady-state with respect to sulfur adsorption,
current sulfate deposition is a associated with lake sulfate as discussed in Section 8.2.3.  Soils in the
SBRP appear to be actively adsorbing sulfate,  and the stream sulfate is controlled by soil chemistry (see
Section 9.2).

8.10.6.2 Ca plus Mg (SOBC), ANC, and pH

      For the  base cation-related water chemistry variables in both the NE and the SBRP, the percent
of land with open, dry vegetation  consistently Is among the first variables selected in the regressions.
These are areas in pasture, cultivation, urban, and other disturbed-land  uses.  This variable is notably
absent in the  regression for the Adirondack Subregion,  which  contains almost  no  land  with open, dry
vegetation and also has the lowest mean lake ANC values. Conversely, the Poconos/Catskills Subregion,
with the highest proportion of open, dry vegetation, has the highest mean lake alkalinity. Area of open,
dry vegetation is the most strongly associated  variable in the ANC models for the Poconos/Catskills and
Southern New England Subregions.

      Areas of open, dry vegetation have usually been disturbed by the activities of man in some way.
The strong relationship  between  these  areas and surface water base  cations may  result from the
disturbances (plowing, fertilization, liming, excavation leading to faster bedrock weathering, waste disposal)
or from characteristics that predispose the areas to disturbance (low slopes, fertile soils, etc.).  Generally
these areas coincide with high base saturation,  and deep relatively flat soils, all of which correlate well
with the dependent variables.  The dependent variables  are also correlated with bedrock weatherability
and surface water sulfate concentrations overall between the NE and SBRP. Such correlations within
each region, however, are weak or do not occur.

8.10.7  Summary Conclusions

      •    A significant proportion of the variability in  surface water chemistry can  be explained  by
           watershed and soil characteristics.

      •    Deposition alone does  not explain the large  variability seen In surface water  chemistry.
           Sulfate deposition is an important explanatory variable for surface water sulfate concentrations
           In the NE, but not in the SBRP.  Additional information on watershed attributes is essential
           for explaining index water chemistry.

      •    Variables found to  be associated  with surface water sulfate and watershed sulfur retention
           include:   sulfur deposition and  soil solution sulfate concentration (in the NE);  soil adsorption

                                               450

-------
capacity and base status (in the SBRP); watershed disturbances such as pits and quarries;
amounts of precipitation and runoff; extent of wetlands and flooded soils; and soil depth and
particle-size distributions.

Variables found to  be associated with  surface water ANC, pH, and Ca plus  Mg  (sum of
base cations) include: bedrock type; watershed disturbances such as area of agriculture or
pits and quarries; levels of precipitation and runoff; soil  base saturation and pH; soil suifate
concentration;  atmospheric  deposition;  and  soil  characteristics  Involving  particle-size
distribution,  permeability, and depth.

Surface water chemistry may be significantly influenced by watershed disturbances or the
extent of sulfate-reducing and acidic organic soils.  The Level II and 111 models do  not deal
explicitly with these variables.  One model assumption  is that no land use change occurs
during the period being  modelled; the  available  data and model  structures do not permit
assessment of  potential watershed changes that may  occur as disturbed lands  revert to
natural conditions as  is happening today in many areas of the eastern United  States.  The
extent  of sulfate-reducing  wet soils  Is handled  implicitly in  model  calibration as in-Iake
reduction of suifate, and the extent of acidic organic soils is handled  by the aggregation of
soil chemistry through sampling classes.

In general, the relationships found In the regressions are the postulated relationships that
are incorporated in the Level II  and  III models.   Given  the caveats discussed in this
document,  the Level  II and  III models incorporate  the variables that are most  strongly
associated with surface water chemistry.
                                    451

-------
                                          SECTION 9

             LEVEL II ANALYSES - SINGLE FACTOR RESPONSE TIME ESTIMATES

9.1  INTRODUCTION

      Although a number of watershed processes are recognized as influencing surface water chemistry
(Sections 2  and 3),  only a few are believed to represent the major  controls on short- and long-term
changes in watershed response to acidic deposition. The NAS Panel on Processes of Lake Acidification
(NAS, 1984) focused on  sulfate adsorption and base cation exchange by soils as  critical time-varying
processes that might contribute to a delayed response to acidic deposition.  The NAS Panel  recognized
scientific uncertainties in the present and  potential long-term  role of these two processes.  Mineral
weathering, cation uptake by vegetation, etc., are rate-limited processes, the magnitudes of which are not
likely to change substantially over the period of the DDRP projections (50 years).  Sulfate adsorption and
cation exchange, on the other  hand, are capacity-limited processes.   As adsorption  sites become
occupied or as exchangeable cations are leached from the soil, the buffering capacity of watershed soils
decreases, resulting in increased probability of acidification.  The projected time frame of such changes
is  believed to vary widely and is thought to be a function of soil physical and chemical properties.  In
watersheds  with thin or  very coarse-textured soils, buffering  of acidic  deposition by  adsorption  or
exchange would be  very limited and some systems would  respond almost immediately.  Alternatively,
watersheds with deep soils and high adsorption capacities and/or large exchangeable base cation pools
might experience significant changes in soil leachate chemistry only after decades to centuries of  high
acidic deposition loadings.

      This section presents results of Level  II Analyses, which involve simulations  of the temporal
response of individual watershed  processes considered in isolation.  Sulfate adsorption  and cation
exchange are examined  as mechanisms  contributing  to delays in surface water  acidification  for
watersheds  in the Northeast (NE) and Southern Blue Ridge Province  (SBRP).  The  analyses are based
on models that consider  only adsorption or exchange within the upper regolith (<  1.5 m in the NE, <
2  m in the SBRP).   These analyses assess the influence of adsorption and exchange on present soil
and/or surface water chemistry and project probable future changes in adsorption and exchange.  Soil
chemistry data collected during the DDRP Soil Surveys and models are used to project future (1) changes
in  suifate mobility controlled by sulfate adsorption and (2) changes in base cation leaching, soil pH, and
cation exchange pools controlled by base  cation exchange.  By considering base cation exchange but
not resupply (i.e., through mineral weathering),  the models presumably overestimate  the potential rate of
base cation  leaching  from the soil; this overestimate results In underestimates of response times for future
changes in soil  and solution chemistry.

      Because these analyses only consider temporal response of single processes  to acidic deposition
in  a  portion  of the watershed (i.e., the upper 1.5 -  2.0 m of watershed soils), model results  should not
be interpreted as integrated projections of  watershed response time.  Rather,  they represent a set  of
bounding estimates  of the relative importance, now and in the future, of the role of adsorption  and
exchange within soils as delay mechanisms. The results of  model simulations in some cases, however,
allow inferences about other processes not considered  in the Level II Analyses (e.g., contributions  of
mineral weathering to watersheds with ANC > 100 peq  L").  Section 7 provides a partial assessment

                                              452

-------
of the role of processes other than adsorption in mediating sulfate mobility in watersheds, of the NE and
SBRP, Level III modelling (Section 10) provides projections of changes in surface water chemistry based
on integration of adsorption and exchange with other processes.

9,2  EFFECTS OF SULFATE ADSORPTION ON WATERSHED SULFUR RESPONSE TIME

9.2.1  Introduction

      As discussed in Sections 2 and 3, the  DDRP has focused  on sulfate as the principal anion in
acidic deposition and as the major mobile anion affecting chronic surface water acidification at  sites in
the eastern United States.  The extent and duration of sulfate retention within watersheds varies widely
within and among regions, depending on deposition history and physical and chemical properties of soils.
Sulfate retention, therefore, has  been identified as one of the most important variables  influencing the
rate of watershed  chemical response (i.e., changes in ANC) to acidic deposition  (Johnson and Cole,
1980; Galloway et  al., 1983a; NAS, 1984).

      At the  start  of the DDRP, soils in the glaciated northern areas of North America  were generally
believed to have  low sulfate adsorption  capacity, resulting in negligible  sulfate retention  by soils and
watershed sulfate budgets at or near steady state.  In contrast, watersheds in the southeastern  United
States were believed to be characterized by high net sulfur retention, attributable to the moderate to high
sulfate adsorption  capacities of deep,  highly weathered soils (NAS, 1984).   Site-specific  and regional
analyses of watershed sulfur budgets (Rochelle et al., 1987;  Rochelie and Church, 1987;  discussed in
Section 7)  have confirmed  differences  in regional sulfur budgets.  These studies  did not, however,
evaluate causal mechanisms, nor did they project a time frame for changes in the Southeast.

      Studies of sulfur retention  processes  in watersheds,  summarized by  Church and Turner (1986)
and discussed in Section 3.3, suggest that  adsorption is the most important net retention mechanism
in typical terrestrial systems  in the NE and SBRP.   Process  studies have consistently identified
Iron/aluminum hydrous oxide content and soil texture (clay content or surface area) as variables that are
positively correlated with adsorption, and pH and  organic  content as variables that  are negatively
correlated with adsorption. These findings, coupled with  the observed differences in these  soil variables
between the  two regions, are consistent with (and have  contributed significantly to the development of)
the paradigm that northeastern soils have low retention capacity and are near sulfur steady state, whereas
southeastern soils  have high adsorption capacity and high watershed sulfur retention.

      Previous regional  soil comparisons  (e.g.,  Johnson  et  al.,  1980; Johnson  and  Todd,  1983)
documented  regional differences in sulfate pools and adsorption and correlated them with differences in
soil pH, hydrous oxide and organic content of  soils.  These  comparisons  provided no  direct basis,
however, for assessing sulfate dynamics in soils of a region and no means of forecasting response to
continued or altered loadings of sulfate.  Within the DDRP, assessments of sulfur budgets (Section 7),
summary descriptions of soil chemistry data, and empirical linkages of soil chemical variables with surface
water chemistry (Section 8) provide important incremental results and an improved understanding of
processes  controlling sulfate in  these watersheds.   These results  are  generally consistent with the
hypothesis  that the mobility in watersheds  of  suifate derived from  acidic deposition is controlled by
adsorption. The principal DDRP objectives, however, lie not just in identifying processes but in predicting

                                               453

-------
the dynamics of sulfate in study regions, specifically in projecting future changes in surface water sulfate
response to continued sulfur deposition  at current or  altered deposition levels.  Level II Analyses are
designed to project response of individual watershed processes; this section describes the procedures
for and results of projecting sulfur dynamics in soils of the NE and the SBRP.

9.2.2  Section  Objectives

      Analyses  in this section are limited to  consideration of changes In sulfate mobility in DDRP
watersheds and  regions attributable to sulfate adsorption (and desorption) by soils. Controls on sulfate
by other processes (Section  3.3  and Section 7)  are  of  relatively minor importance in most DDRP
watersheds and in the regions as a whole.  The goal of Level II  Analyses of sulfate  is to  assess the
importance of sulfate adsorption in influencing delays in surface water acidification  in the NE and SBRP.
Specific objectives of Level II Analyses are to:

           characterize  and compare sulfate pools and sulfate adsorption  capacity of soils in the NE
           and SBRP;

           estimate the  response time of soils in DDRP watersheds to changes in sulfur deposition using
           an adsorption-based model;

      •    estimate time to steady state under current deposition loadings and project response time
           to future  increases (SBRP) or decreases (NE)  in deposition for systems  not presently at
           steady  state, but for which  sorption  is regarded as  an  important  control mechanism.
           Extrapolate results to obtain  regional projections; and  to

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

The  results related to the fourth objective also provide data for evaluating  and comparing the relative
importance of sorption and other processes considered by DDRP models (e.g., cation exchange).  Such
comparisons, however, are not made in this section.

      It is  important to  recognize that procedures and models used for this analysis treat sorption
processes in isolation.  Processes affecting watershed chemistry other than those directly involving sulfate
sorption are not considered, and except for sulfur deposition, watershed conditions (e.g., soil  mass, soil
pH) and fluxes are assumed to be static over the duration of the projections.  It is equally important to
recognize that the projections and estimates of time to steady state made here apply only  to sulfate.
Although change in sulfate mobility is one of the principal factors  driving changes in base cations and
ANC, non-sulfur processes also play critical roles in  such changes.  Rates of change in  ANC, and
particularly projected times to reach zero ANC (i.e., become acidic), are not necessarily coincident with
times to sulfur steady state.  Systems can reach an acidic state prior to, concurrently with, or after sulfur
steady-state conditions are reached.  The relationship between changes in suifate  and  changes in ANC
is characterized  as part of Level III Analyses and discussed  in Section 10.
                                               454

-------
9.2.3 Approach

      Level II sulfate analyses are based on model-based projections of future sulfate dynamics in soils
of watersheds in the DDRP NE and SBRP Regions.  Projections were made using soil chemistry data
generated by the DDRP Soil Surveys (Section 5.5).  The  principal soil variables used for these analyses
are sulfate adsorption isotherms generated for individual soils collected in the surveys and aggregated
to the watershed level.  Projections were made using a modified version of the sulfate subroutine in the
Model of Acidification of Groundwater in Catchments (MAGIC) (Cosby et at,  1985a,  1986b).

9.2.3.1  Model Description

     A critical early decision in this analysis was the selection of one or  more models to describe
sulfate retention In watersheds.   The DDRP was conceived and developed as a relatively  short-term
assessment  project.   Consequently,  project design dictated  use of existing  models  rather than
development of new sulfur cycling  models.  This constraint restricted options  for model selection; for
instance, no model available in 1985 effectively described sulfur cycling or net retention  in soil organic
sulfur pools, and  only  very fragmentary data  existed  on transformation  rates  for  organic  pools.
Furthermore, many integrated watershed  models were  developed  for systems with negligible sulfur
retention.  For  these models, terrestrial  sulfur retention was set  to zero (e.g., the Trickle Down Model,
Schnoor et al., 1984, 1986b), or was described by empirical relationships that served principally to fit
seasonal or hydrologically-driven variability in dissolved  sulfate, without linkage  to  specific  processes
(e.g., the Birkenes  model, Christophersen and Wright, 1981).  After consideration of available models that,
had  adsorption routines, the sulfate subroutine of the MAGIC was viewed as the most straightforward
and  least data-intensive alternative,  and  was selected for use.

     The model uses a deterministic, mass-balance approach that considers only adsorption  as a sulfur
retention process  by soils  (Cosby et al.,  1985b,c; 1986b).  Sulfate partitioning between dissolved and
sorbed phases is  defined by an hyperbolic (Langmuir)  isotherm.  The  original  MAGIC subroutine has
been modified to accommodate multiple soil horizons (up to  10, although either 2 or 3 were used for this
study).  Soil horizons are treated as a series of continuously stirred tank reactors (CSTRs); all inputs of
precipitation and sulfur (wet and dry) deposition are to the top mineral soil horizon (organic horizons are
not  considered in the model, because  sorption is negligible in the O horizon).   Evapotranspiration
implicitly occurs in the top soil horizon.  All flow is then routed  sequentially through each soil horizon.
Data are input to the model using annual time steps.  The projected  surface water sulfate concentration
is defined by  (set equal to)  the equilibrated  solution sulfate concentration  in  the lowest soil horizon.
Because sorption  is essentially  an  instantaneous process,  reaction kinetics are not considered and
equilibrium between solution  and sorbed phases is assumed to occur in all cases.

      For these analyses, model simulations were run starting 140 years prior to the base year (1984 for
NE lakes, 1985 for SBRP  streams).  Soil  and streamwater  surface water  sulfate concentrations were
initialized at the start of simulations by assuming both to be at steady state with respect to deposition.
Simulations were run either 140 years (NE) or 300 years (SBRP) into the future, allowing projected sulfate
concentrations to reach steady state for all watersheds. Data sources for model simulations are described
below (Section 9.2.3.2).
                                               455

-------
9.2.3.2  Data Sources

      Input requirements for the  sulfur  model include current sulfur inputs and outputs (precipitation,
runoff, total sulfur deposition, and sulfate concentration in runoff), scenarios of historic and future sulfur
deposition, and soil  variables to describe sulfate  partitioning  and  adsorption  capacity  (adsorption
isotherms, soil  mass).  Data sources are identified in Section  9.2.3.3;  procedures for generation of
adsorption isotherms and for aggregation of soil chemistry data are described in Section 9.2.3.4.

      Procedures used to estimate precipitation and sulfur deposition are described in  Sections 5.6.3;
both typical year (TY, annual values) and long-term average (LTA) estimates of total sulfur deposition
were  used for NE and SBRP watersheds. Runoff estimates, based on interpolation of 30-year average
USGS runoff maps, were generated as described in Section 5.7.1.  Current lake sulfate concentrations
were from the EPA's Eastern Lake Survey (Linthurst et al., 1986a) and the Pilot Stream Survey (Messer
et al., I986a) (Section 5.3).

      Initial  sulfate  inputs  (year  140) were set to 5 percent of  current deposition; estimated sulfur
deposition between initial and base years (i.e., 1844 to 1984 in the NE and 1845 to 1985 in the SBRP)
was based on emission estimates of Gschwandtner et al. (1985).  Estimates of historic deposition for the
NE and SBRP are based on emission estimates for Federal Regions I and II (CT, MA,  ME, NH, NJ, NY,
Rl, VT) and  Region IV (GA, NC, SC, TN), respectively; linear interpolation between the initial simulation
year and 1900 was used. The historic emission pattern was used as a scaling factor for each watershed,
which was a procedure that assumed that the relationship between regional emissions and  site-specific
deposition over the last 140 years was constant.

      Two scenarios of future sulfur deposition were used for each region as characterized in Section
5.6.1. The first scenario for each region was constant deposition  through the entire simulation period.
For the NE,  the alternative scenario is constant deposition for 10 years, followed by a linear  decrease in
deposition for 15 years (by 2 percent  per year), then constant  deposition  at 70 percent of current
deposition for the remainder of the simulation period. The alternative scenario for the SBRP  also begins
with constant deposition for 10 years, followed by a linear increase in deposition between years 10 and
25, then constant deposition (at 120 percent of current levels) for the remainder of the simulation period.

      Mapping  of soils and quantification of the areal extent of various soils  on DDRP watersheds are
described in Section 5.4  Sampling  and chemical/physical analyses of soils are  described in Section
5.5.  For each mineral soil  horizon, sulfate adsorption data were  used to compute adsorption isotherms
which were then aggregated with soil mass (computed from horizon thickness, bulk density,  and coarse
fragment content) to obtain sample class and watershed values.  Procedures for derivation of adsorption
isotherms and for aggregation of adsorption data are described in Section 9.2.3.4.

9.2.3.3  Model Assumptions and Limitations

      Several critical assumptions are encompassed by the choice of model and  by methods of data
collection. These in turn impose limitations on the scope of model  projections. Key model assumptions
and their Implications for data Interpretation  include:
                                              456

-------
Sorption Is the only watershed process  affecting sulfate mobility and watershed response
time.  As noted previously, this decision was intentional and is believed to provide the most
effective means of assessing the  significance of adsorption by soils as a process delaying
surface water acidification.  To  the  extent that other terrestrial  processes sequester  or
generate sulfate on a net basis, model projections will under- or overestimate the time and/or
magnitude of the projected response. As noted earlier, the net role of other processes in
most  DDRP watersheds is believed to be small.  (The importance  and influence of in-Iake
processes on sulfur budgets and response time in northeastern lakes is addressed in Sections
7.2 and 10.)

The analytical approach used to define sulfate partitioning  in the soil (hyperbolic isotherms
defined by batch equilibrium methods using  air-dried soils) adequately describes sulfate
partitioning by soils under field conditions. Recent findings (Hayden, 1987) support the use
of hyperbolic isotherms and batch equilibrium  methods.  A preliminary evaluation of effects
of soil drying suggested small, non-systematic effects on adsorption; however, subsequent
study (Hayden, 1987) suggests that the measured adsorption capacity of soils increases upon
drying.  This issue is currently being thoroughly assessed by a separate EPA project.

Soil and watershed  conditions influencing adsorption (e.g., soil pH,  Fe, Al, and  organic
content) are static over the life of model projections.  Potentially, pH is the  most important
of these variables since adsorption is strongly pH  dependent.  If soil pH were to change
significantly, the projections of adsorption could be substantially altered.  However, soil pH
is strongly buffered at  low values in  most of the NE and SBRP soils considered by DDRP,
and substantive changes in soil pH are not expected.

Hydrologic routing is simple, representing the soil as a series  of CSTRs; all flow is routed
sequentially through each  horizon.   The "perfect" hydrologic contact represented by  a
simplified  flowpath such as that  used here does not realistically  reflect how lateral  flow,
macropore flow, etc., occur in the soil. However, data to objectively define alternate flowpaths
are lacking.   (The effect  of  flow bypassing  upper or lower  soil  horizons under natural
conditions would result in projections of higher initial sulfate leaching (part of the input signal
would not be attenuated by sorption on the soil),  but a more  gradual (in terms of change
in concentration with time) subsequent  sulfate response.)   The responses projected  here
represent an upper bound  on initial response time assuming complete contact between the
soil and flow through the soil, and a lower bound on time  to steady state.

Because the model runs on an annual time step and uses  identical precipitation and runoff
data from year to year, projections do not reflect the variability of natural systems.  The lack
of "realistic" variability  in the projections is recognized, but should have  little effect on the
primary objective of projecting long-term changes attributable to chronic sulfur deposition.
If there were any long-term trends in precipitation or runoff, they would not, of course,  be
represented by model  projections.
                                    457

-------
9.2.3.4  Adsorption Data

      Data describing  sulfate  partitioning by soils, which  are  used to develop  partitioning  functions
(isotherms) of sulfate adsorption capacity  of soils, were generated as part  of the DDRP soil survey
(Section 5.5),  Adsorption isotherms were developed for each soil, as were soil thickness,  bulk density,
and coarse fragment adjustments.  Adsorption isotherms were then  aggregated from data  for  individual
soils to watershed values using a mass-weighting procedure. Determination of isotherm coefficients and
aggregation of data from individual soils to watershed values are described below.

      In the design of the DDRP, emphasis was placed on projecting dynamics of sulfate and other
ions at  regional scales,  rather than  on a watershed-by-watershed  basis.   The design  mandated  that
procedures for sampling and  aggregating soils data were  targeted at describing soils  for the region.
Using the sample  classes described in Sections  5 and 8,  soils for each  sample class were sampled
approximately eight times across their area of occurrence, which in  many cases included several states
and covered substantial sulfur deposition gradients.  Aggregated sample class chemistry provides a
representative value for that sample class across  the region, but  probably does  not optimally estimate
soil characteristics at the Individual watershed level, and thus does not enable optimal projections for
individual  watersheds.  As an  example, sulfate in  northeastern lake systems is roughly at steady state
across the region; observed lake  sulfate concentrations are proportional to sulfur deposition and decrease
by over 50 percent from ELS  Subregion 1B (Poconos/Catskills) to 1E  (Maine) (Linthurst et al.,  1986a).
Aggregated soil chemistry for  sulfur variables in a sample class that extends from New York to Maine
are the same for all soils in the class,  however, and thus presumably would  underestimate concentrations
in New York while overestimating them for  Maine.

      An alternative approach for sampling and data aggregation would have been to focus sampling to
enable characterizations of individual watersheds.  This watershed focus would generate more intensive
sampling  of points likely to be  representative of each sampled  watershed, but would have allowed
sampling of fewer catchments  in the  region, with the risk of describing  less of the soil variability across
the region.  A watershed focus  also would have  resulted in fewer  sites for extrapolation  of results to
obtain regional population projections.  The watershed approach thus is regarded as less effective than
the regional sample class approach for describing most soils that occur on watersheds in each region
and for generating  regional projections. While the  sample class approach describes the central tendency
and most  of the range  of watersheds, it does not,  however,  provide  precise watershed-level projections,
especially for extreme  watershed values in the population.   Soils data were  mapped  and sampled on
specific watersheds and aggregated to watershed values in order to allow explicit linkage  between soil
characteristics and surface  water chemistry.  To  deal with uncertainties in projections, uncertainty
estimates  for major input variables for Level II models (soil mass and  Isotherm coefficients for  Level II
sulfate analyses) were generated, and Monte Carlo analyses were  used  to describe uncertainty In model
projections for a subset of watersheds.

      Adsorption isotherms were generated from data for soil-water slurries equilibrated with six different
amounts  of sulfate  (0, 2, 4, 8,  16, and  32 mg S L"1) described  as  SO4_0, SO4_2, etc., in Section
                                              458

-------
5,5.4,2.1. For each of the six samples, net sulfate adsorbed by the soil was computed from the change
in dissolved sulfate.  For example, for the 8 mg L"1 sample:

            SO4  8n =  (SO4 a - SO4 8f) * k                                        (Equation 9-1)
               _          _        _     J3
where:      SO4_8j  =  dissolved sulfate concentration prior to equilibration (^eq L )
            SO4_8f  =  final dissolved sulfate concentration after equilibration (^eq L*1)
                 L  =  volume of liquid  (~ 0.050 L)
                 S  =  mass of volume  of soil (~ 0.010 kg)


An extended Langmuir isotherm ("extended" by addition  of a third variable to describe the non-zero
Y-intercept)  was then fit to the six data points for each soil  (final dissolved sulfate and net adsorbed
sulfate)  (Figure 9-1) (Hayden,* 1987).  The equation used to model sulfate partitioning  has the form:

                                   E0  =     B1   *  C   + B3                       (Equation 9-2)
                                            B2   +  C

where:      B1  = maximum sulfate adsorption (meq kg")
            B2  = half saturation constant Gueq L")
            B3  = Y-lntercept (meq kg"1)
            C   = dissolved sulfate (^eq L"')
            Ec  = net adsorbed sulfate at [SO42~] =  C (meq kg"1)

The parameters B.,, B2,  and B3 were estimated using  non-linear least squares, using the  Fletcher-Powell
(1963) method to minimize the sum of squares function.  The Fletcher-Powell method  uses a second
order algorithm that iterativeiy constructs an  estimate  of the  inverse Hessian matrix.  This matrix, in
conjunction  with  the  residual  sum  of squares, provides an estimate of the covariance matrix for the
estimated parameters.

      Several  approaches were evaluated for aggregating data from individual soils, including weighted
averaging of isotherm coefficients or alternatively fitting a single isotherm to all data  points for all  soils
in an aggregation group (e.g.,  all Individual soils in a pedon/master  horizon or sample  class/horizon).
Both  approaches were  rejected because they provided a poor description  of the average partitioning
coefficient (isotherm slope) along the isotherm.   As  an  alternative, after fitting isotherms for individual
soils, values of net adsorbed sulfate corresponding to several concentrations of dissolved sulfate (0, 10,
25, 40, 75, 125, 200, 500,  1000, 2000 /*eq L"1) were computed for each soil.  For each value of dissolved
sulfate,  the  mass-weighted average of the  corresponding concentrations of adsorbed sulfate  was
generated for  all soils in an  aggregation group (typically all soils with  the same master  horizon
designation  in a sample class).  Finally, a new isotherm was fitted to the set of weighted  averages.  This
isotherm was defined as the aggregate isotherm and was used to describe sulfate partitioning for that
group of soils.

      This approach provides a very good estimator of the weighted average soil partitioning coefficient
(isotherm slope) over the  range of dissolved sulfate of interest to this  project (0 - 200 jueq L"1), even for
groups of soils in which coefficients for individual soils are highly variable.  Aggregation was conducted
in three steps, with any missing data assigned the aggregate average for other soils in its pedon/horizon:
(1) individual soil (sub)horlzons to master horizon within a pedon (mass weighting); (2) pedon/horizon

                                               459

-------
  CO
 V)
 "D
  0
  o
  CO
  CD
 2
       0
                                                  EXTENDED LANGMUIR ISOTHERM
                                                                          B,
                                                Where  C = Dissolved Sulfate
                                                        Ec = Net adsorbed sulfate at C
                                                        B1 = Maximum Adsorption
                                                        BZ = Half-saturation constant
                                                        B3 = Y-intercept
                                                        EMAX = B1 + B3
                                                        ESSS (Equilibrium Soil Solution
                                                          Sulfate) = Dissolved Sulfate at
                                                          Ec = 0(jieqL-1)
Figure 9-1. Schematic diagram of extended Langmuir isotherm fitted to data points from laboratory
soil analysis.
                                             460

-------
to sample class/horizon (mass weighting);  and (3) sample class/horizon to watershed/ horizon (mass
and area weighting of each sample class occurring on each watershed).  For routine uses, data for soil
master  horizons  were used directly in  the model  and  were not  aggregated.   For  certain model
applications, data were aggregated to 1  or 2 horizons per pedon using a comparable mass-weighting
approach.

      Because  the aggregation approach was not conducive to direct  computation  of  parameter
uncertainty, uncertainties for the original  isotherm fits were retained; a Monte Carlo procedure was  used
during each step of aggregation to generate estimates of uncertainty in aggregated coefficients at the
sample class and watershed level.  The uncertainty in the sulfate isotherms was propagated through the
aggregation procedure using the Monte Carlo technique described in general in Section 6.3.  Application
of the procedure to  sulfate isotherm aggregation  proceeded through  steps similar to those used for
aggregation of  other variables.  The aggregation from individual subhorizons to sample class master
horizon was repeated 100 times, each time selecting a randomly perturbed set of coefficients for  each
subhorizon isotherm. The perturbation of B1 was selected first from a normal distribution with a standard
deviation obtained from  the residual sum of squares and the inverse  Hessian  matrix from the nonlinear
least  squares.  The perturbed  value of  B.,, along  with the correlation of B1  and B2 from  the inverse
Hessian matrix, were used to estimate the conditional expectation of B2 given B.,. This conditional value
was then perturbed by a value drawn from a  normal distribution with the conditional standard deviation
of B2 given Bv  A similar procedure was used to perturb Bg, except that the mean and variance  were
adjusted for both B1  and  B2.  The mean values, standard  deviations,  and  correlation  matrix of the
coefficients were summarized at the sample class level. These values were then passed to the watershed
level  aggregation algorithm.   The uncertainty  calculation was conducted  as above, except  that the
correlations were derived from the sample class Monte Carlo study rather than from an  inverse Hessian
matrix.

      The rationale for the mass-weighting aggregation approach described above is consistent with the
common aggregation approach discussed in Johnson et al. (1988b).  Several  alternative approaches to
aggregation of  soil chemical data were  discussed  in that document,  including weighting schemes that
would represent watershed factors such as hydrologic flowpaths, landscape  position, etc.   Ultimately,
alternative aggregations  for capacity variables, including sulfate adsorption capacity were rejected.  This
decision was based principally on  the  lack of objective criteria for setting  weighting coefficients to
describe hydrologic routing or other watershed factors (including unsuccessful attempts to empirically
determine statistically significant coefficients). The mass weighting approach used here provides unbiased
estimates of the pools and/or capacities (e.g., sulfate adsorption capacity, exchangeable base cation
pool)  for capacity variables in soils of the DDRP watersheds. Hydrologic routing, incomplete soil contact,
landscape position, etc., influence the degree of interaction between acidic deposition and the soil, and
might alter the rate at which soil pools or capacities are affected. In the absence of quantitative estimates
of routing coefficients, however, the unbiased  pool estimates generated by the  mass-weighting approach
provides the best description of soil pools and  capacities for the Level II models used here.

9.2.3.5  Evaluation of Aggregated  Data and Model Outputs

      Several approaches were used  to evaluate aggregated soil sulfate data and  model outputs. An
initial  assessment of isotherm data and aggregation procedures was made by comparing the equilibrium

                                              461

-------
soil solution sulfate of Isotherms aggregated to the watershed level  (B  horizons) to measured  surface
water suifate concentrations  in the NE and  SBRP.   If adsorption by the  soil were the sole process
influencing sulfate mobility and if aggregation procedures were perfect, a 1:1  correlation between soil and
lake/stream sulfate concentrations would be obtained.  Realistically (due  to contributions of factors such
as hydrologic routing, heterogeneity  of natural soils, uncertainties  introduced  by soil sampling  and
analysis, and effects  of regionally  focused aggregation), a high correlation between  soil  solution  and
surface water sulfate was not expected.  The purpose of this comparison was to evaluate whether the
two sets of values were comparable and whether major biases existed that would invalidate the entire
approach.

      Results of this comparison (Figure 9-2), show that for the most part agreement between computed
soli solution and  surface water sulfate concentrations is good.  Although  the data have  considerable
scatter, points for NE Subregions A, C, and E and for the SBRP generally plot near the 1:1 line.  The
effects of aggregating data collected along a deposition gradient (noted in Section 9.2.3.4) are clearly
apparent for NE Subregions A, C, and E. Although the lake sulfate concentrations range from  roughly
50 to 150 /*eq L"1, aggregate soil solution concentrations are clustered near 100 #eq L"1. For watersheds
in NE Subregions B and D, computed soil solution sulfate concentrations are  consistently biased high.
The  difference  between these watersheds and other systems in the NE and  SBRP is believed to be
related to differences  in soil type rather than geographic location. The difference might be an artifact of
soil handling  (air-drying) procedures. This bias, although substantial, occurs in  only a subset of the data
and, in any case, is not sufficiently  large to  invalidate the data or the aggregation approach.  It is  also
important to note that equilibrium soil solution sulfate  is not used  directly in the Level II models.  Related
isotherm variables that affect model projections  (reflected by isotherm slope)  appear to be much  less
sensitive to effects of air-drying.

      In addition to the evaluation of aggregated data described  above,  several approaches were taken
to evaluate model outputs.  Model  projections were  compared to observed surface water chemistry in
several ways.  Model simulations  start 140 years in the  past and run through the  present, allowing
projections for  the base  year to be compared to observed lake or stream chemistry, and means and
distributions of  the two datasets to be compared for  biases.  Preliminary evaluation of model results for
soils on northeastern watersheds indicated very rapid  time to sulfate steady state and showed that model
Inputs (isotherm coefficients)  in many cases could be varied  by almost  an  order of magnitude  without
significantly  changing the  projected  sulfate concentration for the base year.   Evaluation of  model
projections also was done for the SBRP.  Using both mean values and sample distributions for the SBRP
target population, modelled and measured sulfate concentrations were compared, as well as modelled
vs. observed percent sulfate retention.   Projected  rates of increase in  dissolved sulfate for DDRP
watersheds  also  were  compared   to  available  data on  measured  rates  of  increase for sulfate in
southeastern watersheds.

9.2.3.6 Target Populations  for Model Projections

      For both  the NE and SBRP, projected changes in sulfate are presented for lake (NE) and stream
(SBRP) populations at regional scales.  Model runs were made using data  for  the DDRP watersheds In
the respective regions, then extrapolated to regional target population projections, using weights  defined
by  the National Surface Water Survey (Linthurst et al., 1986a; Messer et al., I986a).  In the NE, model

                                              462

-------
      350
                                                         O  N.E. - Subregion A.C.E
                                                            N.E. - Subregion BปD
                                                         A  SBRP
                    50
100
150
200
250
300
350
                              Surface water sulfate (|ieq L ~1)
Figure  9-2.  Comparison of measured lake  (NE) or stream (SBRP)  sulfate  concentration with
computed soil solution concentration.

                                          463

-------
input data were prepared for all watersheds, and the model run for all watersheds in Priority Classes A
through G, i.e., lakes in classes H (seepage lakes) and 1  (significant internal sulfur sources) were deleted
from the analyses.  (Priority classes are described in Section 10.4.) After initial assignment of priority
classes, the lake type of one northeastern lake (1D2-036)  was changed from closed to impoundment, and
four additional northeastern watersheds (1C2-068, 1E1-025, 1E1-040, and 1E3-040) were identified as
having probable significant internal sulfur sources.   Data for lake  1D2-036  were then included in the
analyses, and the four lakes with putative internal sulfur sources were deleted. The final dataset used
for generating watershed sulfur projections included 131 NE watersheds, representing a regional target
lake population of 3,314 lakes.

      In the SBRP, all of the 35 DDRP stream watersheds were included in the analysis, except a single
watershed  in Priority Class E (2A08808), which had significant internal  sulfur sources.   Using weights
defined during the Pilot Stream Survey,  results for the SBRP watersheds were extrapolated to describe
a regional target population of 1,492 stream reaches.

9.2.4

9.2.4.1  Comparison of Northeast and Southern Blue Ridge Province isotherm Variables

      Before presenting and discussing  model projections generated as  part  of the Level  II Analyses for
sulfate, a comparison of data used as model inputs is useful, including adsorption Isotherm data for soils
of the two regions  and secondary data derived from the  isotherms.  Table 9-1  summarizes isotherm
data by soil horizon and soil order for the NE and by soil horizon for the SBRP. In addition to isotherm
coefficients, data in the table include several derived variables that provide a  more convenient basis for
comparing the potential for  sulfate  adsorption by  soils in the 2  regions.   Derived  parameters were
computed using a dissolved sulfate concentration of 100 /*eq L"1 to facilitate comparison. The  derived
parameters  include isotherm  slope (soil-water partitioning coefficient), adsorbed sulfate  (change  in
adsorbed sulfate per kg soil  as sulfate  concentration increases from 0  to 100 /ieq L"1),  and adsorbed
sulfate for soil horizons, which  couples  adsorption with  soil mass to describe potential sorption  by the
pedon.

      Examination of the isotherm data reveals differences in adsorption capacities of soils within the
NE and very pronounced differences between soils in the NE and SBRP. Within the Northeast, Entisols
have the lowest adsorption capacity, whereas potential  sorption capacity of Inceptisols and Spodosols
is roughly equal.  For all three northeastern soli orders, adsorption capacity and  isotherm partitioning
coefficients are lowest in the poorly developed  C horizon soils.   Comparison of NE and SBRP data
consistently suggests higher  adsorption by SBRP soils; maximum adsorption capacities  are higher for
each SBRP horizon than for any of the northeastern soils, and the partitioning coefficient  (slope) is two-
to tenfold higher for SBRP  soils than for the same horizon  in northeastern soils.  These differences are
reflected in adsorbed sulfate pools; on a unit mass basis, sulfate pools at 100 peq L"1 are typically three-
to tenfold higher for SBRP soils than  for those in  the NE.  When the greater  mass of SBRP soils
(especially in the B horizon) is considered, the 100 /zeq L"1 adsorbed sulfate pool in SBRP soils is about
10 times as large as those for northeastern Inceptisols and Spodosols and 30 times that for northeastern
Entisols.
                                               464

-------
Table 9-1.  Comparison of Summary Data for Sulfate Adsorption Isotherm Data for Soils in the
Northeastern United States and Southern Blue Ridge Province
Isotherm Coef8
Region/
Order
Northeast
Entisols




Inceptisols




Spodosols




Southern Blue
Ridge Province
Soil
Horizon


A/E
B
C


A/E
B ,
C


A/E
B
C



B,
{meq kg"1)


2.37
1.05
0.76


3.15
3.68
1.63


2.72
5.13
1.19



Halfsat'n
(/*eq I'1)


1641
997
994


1560
1017
1007


1117
893
970



Slopeฎ
100 fj.eq L"1


1.08
1.21
0.63


1.58
2.59
0.96


1.28
4.33
0.98



Adsorbed Sulfate
Pool @ 100 fj,eq L"1b
(meq kg"1) (keq ha"1)


0.114
0.133
0.088
s'

0.214
0.308
0.172


0.279
0.483
0.154





0.09
0.27
0.54
^-0.90

0.17
1.73
0.99
2.89

0.15
1.66
0.83
2.64


All
A/E 5.89
B 7.13
C 4.80

1199
322
361

3.39
12.18
6.45

0.541
2.657
1.837

0.98
20.9
5.86
27.74
  Coefficients for Langmuir isotherm of the form:
       Adsorbed SO4
                         B,
                        haffsat'n + C

   where c = dissolved sulfate concentration

  Computed pools of adsorbed sulfate using the equation listed in footnote a.
                                                 465

-------
      The observed differences in adsorption characteristics of northeastern and SBRP soils are generally
as expected.  Retention capacity of soils in the SBRP, expressed as adsorbed sulfate for the pedon, is
much higher than for soils in the NE.  Two principal reasons for this difference are apparent.  The first
is related to differences in sulfate adsorption capacity of A and B soil  horizons  in the two regions.
Comparison of soil chemistry characteristics for the two regions suggests that differences are not, as has
been suggested (e.g., NAS, 1984), attributable solely to differences in soil age and degree of weathering.
Although upper horizons of northeastern soils have lower clay content than SBRP soils, the  northeastern
soils do have substantial concentrations of extractable iron and aluminum.  Extractable aluminum is often
higher in northeastern soils than in those of the  SBRP.  Northeastern soils, however,  also have much
higher organic content than SBRP soils, and organic blocking is likely to reduce anion adsorption capacity
of northeastern  soils substantially and to  account for much  of  the regional difference in adsorption
capacity of upper soil  horizons (Chao et  al., 1964a;  Johnson and Todd,  1983).  The  second factor
affecting total  pedon adsorption  capacity is explicitly tied to soil age and extent of weathering.  Soils in
the NE have  typically undergone significant weathering only to a depth  of 30-50 cm;  subsoils  are
minimally weathered and have few clays or hydrous oxides and thus little effective substrate for sorption.
In the SBRP,  by contrast,  most soils are extensively weathered to a depth of well over a meter, and
subsoils have  abundant clays and hydrous oxides and very low organic content, resulting in high anion
adsorption capacity.  SBRP soils thus not only have higher adsorption capacity per unit soil mass than
soils in the NE,  but also have a much greater mass of those soils with high adsorption capacity. This
results in potential sulfate retention capacities for SBRP soils that are 10- to 30-fold higher than for typical
northeastern soils and leads to differences in projected response times to sulfur deposition for the two
regions.

9.2.4.2  Model Results - Northeastern United States

      Based on model projections using long-term average deposition data, sulfur response times for soils
in northeastern watersheds are very rapid In all cases.  For typical systems in the NE, the projected lag
between changes in deposition  and surface water response is on the order  of a  decade.  For some
watersheds the delay is as short as five years, and the  longest projected lags are less than 15 years.
For all of the 131 northeastern watersheds modelled as part of the Level H Analyses for sulfate, response
times are sufficiently short that, during periods of  higher deposition prior to 1975,  sulfate concentrations
exceeded steady state  with  1984 levels  of deposition.  Concentrations  are projected to  be declining
currently in response to  reduced deposition over  the past decade (Figure 9-3).

      Based on the results shown in Figure 9-3, it  is apparent that the suifate model used for this analysis
predicts very short  lags in sulfate response time  and thus significant deviation from sulfur  steady state
for soils in northeastern watersheds only  during periods when sulfur inputs are changing  rapidly. When
deposition  inputs are decreased, projected surface water sulfate concentrations  are also  projected to
decrease rapidly; during the period of re-equilibration to the lower  deposition level, soils release (desorb)
sulfate and the watershed has negative sulfur retention (i.e., watershed output exceeds input; Figure 9-
4). Conversely, as Figure 9-4 also shows, during the lag phase when deposition  is increased, soils adsorb
sulfate and there is positive retention by the watershed. As  used  in this section, steady state for sulfur
                                               466

-------
                                 Northeast Lakes
     180-


     160-


     140-
  x
   :  1001
 CO
 "55   60*
 DC

      40-
      20-


       0
                                          Deposition Input
                                          Lake Output
                                            Median Response
                                            Range
          1850
1900
                                   1950
   2000
Year
                                     2050
2100
Figure 9-3.  Historic deposition inputs and modelled output for soils in a representative watershed
in the northeastern United States. The historic deposition  pattern is based on emission estimates
of Gschwandtner et al. (1985).  Sulfur flux is expressed on a relative scale, 1984 deposition flux
= 100. The base year (1984) is Indicated by the arrow.  Note that because precipitation and runoff
are constant throughout the simulation period, changes in flux correspond to proportional changes
in projected lake sulfate concentration.
                                          467

-------
                       SULFATE CONCENTRATION
       c
       
-------
refers solely to sulfur input/output budget status;  no inference Is  intended  regarding  stasis of the
biogeochemical sulfur cycles within the watersheds.  Percent sulfur steady state is computed as
          Percent Sulfur Steady State  =  (	) 100                            (Equation 9-3)
                                           q
                                           ^input
and is related to percent sulfur retention by

                                          o      . g
                                           input     output
              Percent Sulfur Retention  =  ( —	-— ) 100                    (Equation 9-4)
                                      =  100 - Percent Sulfur Steady State


9.2.4.2.1  Evaluation of base year data, calibration of model inputs -

      Sulfur input-output budgets, calculated for DDRP lakes using ELS sulfate concentration data and
long-term average deposition data, were computed as described in Section 7.3. Percent sulfur retention
ranges widely among northeastern lakes,  from -60 to +70 percent, with a mean  of -2.5 percent (Table
9-2).  In contrast to the computed percent  retention, the range of model  projections for 1984 is much
narrower due to the short response times for northeastern watersheds. The short response times, coupled
with decreases in  deposition since 1975,  result  in  model forecasts for the 1984  base year (when
northeastern lakes were sampled for the ELS) of slight to  moderate negative retention for all northeastern
lakes (Tables 9-2 and 9-3, Figure  9-5).  For long-term average  (LTA) deposition data, modelled retention
in 1984 varied  from -19.3 to -1.3 percent, with a population median  of -7.1  percent.  Estimates using
typical year (TY) deposition data were almost identical, ranging from -18.9  to -0.1  percent retention, with
a median of -6.8 percent (Table 9-3).

      Although computed  and modelled  percent sulfur retention  differ  considerably, the range  and
distribution of  measured and  modelled  sulfate concentrations for  1984 are  very  similar, and are
comparable to  steady-state sulfate concentrations (Figure 9-5),  As indicated by percent retention  data,
modelled concentrations slightly  exceed  steady-state concentrations for  all  systems, whereas  (on a
watershed-by-watershed basis)  measured  sulfate often deviates substantially and unsystematically from
steady-state concentrations. Because DDRP objectives are focused at the regional population level, the
close overlap of measured and steady-state sulfate concentrations is  reassuring,  in that it suggests the
sulfur data used for model projections provide a good representation of the regional  population of lake
sulfate concentrations.

      Results discussed in the preceding paragraphs are based on  use of soils data without adjustment
or model calibration.  Projections of  sulfate concentration and percent sulfur retention are  essentially
unbiased, but the range of percent sulfur retention projected  by the models  is  much smaller than the
range  of  measured percent  sulfur retention.   Sensitivity analyses  indicate that,  because  of  the  rapid
response times of northeastern systems, projected lake sulfate concentrations  for  1984  remain near
steady state even if the principal  model inputs (isotherm coefficients and/or soil mass) are adjusted by
a factor of 2.  Model projections of base year sulfate concentrations and percent sulfur retention remained

                                               469

-------
Table 9-2.   Summary Statistics for Modelled  Changes in  Sulfate Concentration,  Percent  Sulfur
Retention, and Delta Sulfate for Northeast Watersheds Using Long-Term Average Deposition Data
Sulfate Concentration
Scenario
Year
Lake S04
Constant Depn,
0
10
20
50
100
Steady State
Decreased Depn,
20
50
100
Steady State

Scenario
Year
Lake SO4
Constant Depn,
0
10
20
50
100
Decreased Depn.
20 .
50
100

Mean
110.0

120.3
111.6
110.7
110.5
110.5
110.5

103.8
77.5
77.4
77.4


Mean
-2.5

-7.9
-0.7
-0.1
0.0
0.0

-17.0
-0.1
0.0

Std. Dev.
39.5

46.7
41.0
39,9
39.7
39,7
39.7

38.4
28.0
27,8
27.8



Min.
33,8

54.7
51.1
50.8
50.8
50.8
50.8

47.8
35.6
35.6
35.6
Percent Sulfur

Std. Dev. Min.
24.9

4.0
1.0
0.2
<0.1
0.0

1.7
0.3
<0.1
-60.0

-19.3
-4.7
-1.1
>-0.1
>-0.1

-21.9
-1.4
>-0.1

P_25
81.7

83.7
77.9
77.5
77,5
77.5
77.5

72,6
54.3
54.2
54.2
Retention

P_25
-20.9

-10.0
-0.8
-0.1
0.0
0.0

-20.9
-0.8
>-0.1

Median
105,4

114.3
106.2
106.0
106.0
106.0
106.0

99.1
74.2
74.2
74.2


Median
-3.1

-7.1
-0.3
>-0.1
0.0
0.0

-19.4
>-0.1
0.0

P_75
130.7

142.7
126.0
126.0
125.5
125.5
125.5

118.0
87.9
87.8
87.8


P_75
15.6

-5.2
-0.1
>1.0
0.0
0.0

-15.2
>-0.1
0.0

Max,
213.8

249.3
218.8
211.7
209.6
209.6
209.6

204,1
148.6 ~-
146.7
146.7


Max,
61.1

-1.3
0.0
0.0
0.0
0.0

-13.1
0.0
0.0
Scenario
Mean
 Delta Sulfate (Change from Year 0 to n)

Std, Dev,     Min.      P 25      Median
P 75
Max.
Constant Depn.
0-10
0-20
0-50
0-100
Decreased Depn.
0-20
0-50
0-100

-8.7
-9.6
-9.7
-9.7

-16.5
-42.8
-42.9

6.5
7.8
8.2
8.2

8.7
19.0
19.3

-30.6
-37.6
-39.7
-39.7

-45.2
-100.7
-102.6

-11.1
-12.0
-12.1
-12,1

-19,0
-52.5
-52.5

-6.4
-6.8
-6.8
-6.8

-14.1
-39.1
-39.1

-4.1
-4.2
-4.2
-4.2

-10.6
-29.4
-29.5

-0.7
-0,7
-0.7
-0.7

-5.8
-16.9
-16.9
                                            470

-------
Table 9-3.  Summary Statistics for Modelled Changes in Sulfate Concentration, Percent Sulfur
Retention, and Delta Sulfate for Northeast Watersheds Using Typical Year Deposition Data
Sulfate Concentration
Scenario
Year
Lake SO4
Constant Depn.
0
10
20
50
100
Steady State
Decreased Depn.
20
50
100
Steady State

Scenario
Year
Lake SO4
Constant Depn.
0
10
20
50
100
Decreased Depn.
20
50
100

Mean
110.0

127.8
118.7
117.7
117.6
117.6
117.6

110.6
82.7
2.6
82.3


Mean
4.2

-7.7
-0.7
>-0.1
-0.1
0.0

-17.4
-0.7
-0.6

Std. Dev.
39.5

54.7
48.3
47.2
46,9
46.9
46.9

45,0
32.8
32,6
32.9


Std. Dev.
19.9

3.8
0.9
0.2
<0.1
<0.1

1.6
1.1
1.1

Min.
33.8

52.4
50.2
50.1
50.1
50.1
50.1

47.3
36.2
36.2
35.1
Percent Sulfur

Min.
-53.6

-18.9
-4.4
-1.0
>-0.1
>-0.1

-21.7
-3.2
-3.2

P 25
81.7

79.6
75.7
75.7
75.7
75.7
75.7

71.1
54.2
54.2
53.0
Retention

P_25
-4.6

-9.5
-0.7
>-0.1
0.0
0.0

-18.5
-0.9
>-0.1

Median
105.4

118.2
111.6
111.2
111.2
111.2
111.2

103.7
77.8
77.8
77.8


Median
0.0

-6.8
-0.4
>-0.1
0.0
0.0

-17.4
-0.1
>-0.1

P_75
130.7

157.6
146.6
146.0
146.0
146.0
146.0 -

137.0
102.2
102.2
102.2


P_75
17.0

-5.1
-0.2
0.0
0.0
0.0

-16.1
>-0.1
0.0

Max.
213.8

281.3
247.6
240.3
238.8
238.8
238.8

231.3
168.7
167.2
167.2


Max.
68.6

-0.1
0.0
0.0
0.0
0.0

-13.3
0.0
0.0
Scenario
Mean
   Delta Sulfate (Change from Year 0 to n)

Std. Dev.       Min.      P  25      Median
P 75
Max.
Constant Depn.
0-10
0-20
0-50
0-100
Decreased Depn.
0-20
0-50
0-100

-9.1
-10.0
-10.2
-10.2

-17.6
-45.1
-45.2

7.1
8.6
8.9
8.9

10.1
22.1
22.4

-33.7
-41.1
-43.0
-43.1

-50.1
-112.7
-114.5

-12.0
-12.8
-12.9
-12.9

-20.2
-55.2
-55.3

-6.6
-7.0
-7.0
-7.0

-14.5
-40.4
-40.4

-4.1
-4.3
-4.3
-4.3

-10.4
-27.7
-27.7

-0.1
-0.1
-0.1
-0.1

-3.3
-15.4
-15.4
                                            471

-------
                                  Northeast Lakes
                                   Year 0 Sulfur
                   Deposition = Long Term Average, Constant
                   1.0
                   0.8
                  .
                 o
                   0.4-
                 o
                   05-
                   0.0
                              50      100      150      200
                                 Sulfate Concentration (jieq Lr1)
                                250
                   1.0
                   0.8.
                 fo.6
                 I
                   0.4
                O
                   0.0
                    -80
-40          0         40
   Percent Sulfur Retention
80
Figure 9-5.  Comparison of measured, modelled and steady-state sulfate for Northeast lake systems
in 1984.
                                           472

-------
unbiased, but the distribution of percent retention again was  small.  Changes made in deposition or
rainfall/runoff  ratios resulted in changes in the steady-state  sulfur concentrations  and in   modified
projections of sulfate concentration, but the range of modelled percent sulfur retention was again virtually
unchanged.  The net  result of changing deposition/hydrologic fluxes was the introduction of bias  in
projected sulfate  concentrations, without an expansion in the range of projected percent sulfur retention
to match observed distributions.  Systematic changes in soils or  deposition data that increased the ranges
of suifate response (percent retention in the base year) could not be identified without introducing bias
in projected average sulfate concentrations  or percent sulfur retention.

      An alternative calibration  approach of adjusting soil chemistry data for individual watersheds was
also considered in order to match model projections with measured sulfate concentrations and percent
retention.  The success of this  approach, attempted  for a subset of  northeastern watersheds, was
marginal.  For watersheds with sulfur retention less than about -25 percent (output >  125 percent  of
input), no combination  of adsorption  parameters could match  observed retention,  unless historic
deposition sequences were altered.  For watersheds with positive computed  retention, matching modelled
to measured suifate concentrations required increases in adsorption capacity (isotherm Emax and/or soil
mass) by a factor of 14 to 24.  Concurrent  sensitivity analyses for SBRP watersheds (Section 9.2.4.3.1)
indicated that no adjustment of isotherm parameters was necessary or appropriate.  It  was concluded
that model inputs should not be calibrated, based on these results, that is (1) the lack of  bias in average
projections for the HE, (2)  substantial adjustments to isotherm  data required for matching mean  values
and ranges of projected  sulfate concentrations and percent retention with  measured  distributions, and
(3) absence of similar needs for such adjustments for the SBRP,  suggesting  that there were no systematic
biases. This conclusion, in turn, led to the conclusion that, to the extent that significant  deviations from
steady state are  currently  observed for  sulfur in the NE (especially positive retention),  they should be
attributed to uncertainties in sulfur input/output budgets or to other retention processes  such as  in-lake
retention or sulfate reduction in wetlands.

9.2.4.2.2 Projections of future sulfate concentrations -

      Projections of future suifate concentrations and percent sulfur retention  in soils  of northeastern
watersheds, for periods ranging from 10 to 100 years,  are presented in Figures 9-6, 9-7, and 9-8, and
in Tables 9-2  and 9-3.  As previously noted, the reliability of  model projection decreases with  longer
projection periods.  Model projections for periods longer than 50 years are included principally to provide
bounds on potential change after the 50-year period that is the focus of DDRP.  Results  using both LTA
and TY deposition  scenarios, and for both  the constant and ramped future deposition  sequences, are
included. Projected sulfate concentrations  and percent retention based on the LTA  and TY deposition
datasets are very similar; in order to avoid redundancy, therefore, discussion is limited to results based
on the LTA deposition dataset, except to note differences between the two sets of projections.

      As expected on  the basis of discussion in the preceding  section,  projected  changes for sulfate in
the NE are rapid  and times to steady state are short.  If current levels of deposition are  maintained, the
only projected changes are small declines in sulfate concentrations as watersheds come  to steady state.
Within 10 years,  sulfate  is  projected to  decrease from a median  of 107.9 to 100.7 percent of steady-
state concentration, and  maximum concentration is projected to decrease from 119.3 to 104.7 percent
of steady-state concentration. The corresponding median and maximum declines in suifate concentrations

                                               473

-------
                   Northeast Lakes
           Percent Watershed Sulfur Retention
       Deposition = Long Term Average, Constant
        1.0,
        0.8.
        0.6.
     Q.
     0
        0.4.
     I
        02.,
        0.0
_w~—
	
OYeปrs
10Yun
20Yซซim
SOYMM
100Yซra
              Northeast Lakes
     Percent Watershed Sulfur Retention
       Deposition = Long Term Average
           Ramped 30% Decrease
         -25
                 -20
                        -15     -10      -5
                     Percent Sulfur Retention
  1.0..



  0.8.

C
4ฃ
•K
ง" ฐ'6'
a.
i
I 0.4
3
o

  0.2.


	
OYoti*
10 Yean

100 Y*n
                                                     0.0. _
                                                       -25
                                                                                         7
           •20     -15     -10     -5

                Pencent Sulfur Retention
                   Northeast Lakes
         Changes In Lake Sulfate Concentration
       Deposition = Long Term Average, Constant
              Northeast Lakes
    Changes In Lake Sulfate Concentration
       Deposition a Long Term Average
           Ramped 30% Decrease
       1.0,
       o.s,
      irO.6.
       0.4.
       02.
       0.0
                                                     1.0,
                                                     0.8.
                                                    -0.6.
10.4J
                                                     0.2.
        -125
                -100
                       -75     -50
                      A Sulfate fcieqL-1
                                      -25
                                                     0.0
                                                      -125
                                                             -100
                  -75     -50
                  A Sulfate JieqU'1)
                                                                                    -25
Figure 9-6.   Projected  changes  in  percent  sulfur retention and sulfate concentration  for soils in
northeastern lake  systems at 10, 20, SO and  100 years.   Data  are shown for long-term average
deposition for constant and decreased inputs.
                                                   474

-------
                   250-




                   200-




                Ii  150—

                g"
                A


                g  loo--




                   So-
                     Measured
                                       SULFATE CONCENTRATION
                                      10    20
                                                50
                                        Simulation Year
                                       Constant Depositor!
100 Steady
    State
1    50   10(

   Simulation Year
 Decreased Deposition
                  80 —



                  6O —



                  40
                .1
                 -4O-



                 -6O —
                                     PERCENT SULFUR RETENTION
                                         Slrmilatfon Year
                                        Constant Depostfori
               20    eo     100
                  Simulation Year
                Decm&sod Deposition
                8.
                                   CUMULATIVE CHANGE IN SULFATE
                                 0-20     o-so
                                   Simulation Year
                                  Convtant Dopostlon
         O-2O     O-SO     O-10O

               Emulation Yซar
             D*crซaaซd Deposition
Figure 9-7.   Box-and-whisker  plots  showing  changes in  sulfate concentration,  percent sulfur
retention, and change in sulfate concentration for soils in northeastern lake watersheds, using long-
term average deposition data.
                                                   475

-------
                  300-




                  250-




               _ 200-
                  150-
               8
                  100-
                  50—
                    Measured
                                     SULFATE CONCENTRATION
                                  10   20
                                            SO
                                      Simulation Year
                                     Constant Depostion
1OO Steady
    State
  50   100
  Simulation Year
Decreased Deposition
                                   PERCENT SULFUR RETENTION
                 100—i
ซ• so—
EX.
DC
"-50-
"J-OO™

1=
ซ

ฅ * s&

Measured 0 10 20 SO 1OO 20 SO 10O
Simulation Year Simulation Year
Constant Depostfon Decreased Deposition
                                CUMULATIVE CHANGE IN SULFATE
o-
-20-
f7 -40-
S^
1-
•100-
S i
I
4



0-10 0-
3 c

.



20 0-
r -i
3 t

•



SO 0-
r -i
J p


4


q
d
>











*

3
L





100 0-20 0-50 0-100
SimulatkDn Year
Constant Deposfion
Simulation Year
Decreased Deposition
Figure  9-8.   Box-and-whisker  plots  showing changes in  sulfate concentration, percent  sulfur
retention, and change in sulfate concentration for soils in northeastern lake watersheds, usina TY
deposition data.
                                              476

-------
are 8.7 fjteq L"1 (from 114.3 to 106.2  peq L"1) and 30.6 neq L'1 (249.3 to 218.8 jueq  L"1), respectively.
Within 20 years, at constant deposition, sulfate in all northeastern systems is projected to be within 2
L"1, or 1  percent, of steady state.

      For the scenarios of a  ramped 30 percent decrease in sulfur deposition, similarly short response
times are projected. Most watersheds are projected to have virtually reached steady state with current
deposition by year 10; they begin to respond almost immediately to the reductions in  deposition.  As
inputs decrease, watersheds begin to re-equilibrate by desorbing sulfate, and projected percent retention
becomes negative. In year 20,  during the period of decreased deposition, projected watershed sulfur
outputs are roughly 15-20 percent above inputs. At year 25, when the decrease in inputs ends, however,
systems  again quickly re-equiiibrate, and at 50 years, projections of watershed sulfur retention are within
1 percent (and  2 /*eq L"1) of the new, lower steady-state concentration  for all northeastern watersheds.

      At  year  50,  projected  changes in sulfate concentration  in the  ramped deposition scenario are
considerably larger than for the constant deposition scenario. The projected decrease in median sulfate
is 39.1 ;ueq L"1, with a range of  16.9 to 100.7 /ieq L"1.   Projected changes for model runs using the TY
deposition dataset occur over time frames comparable to those for LTA deposition and are slightly larger
(median  and maximum of 40 and 114 peq L"1, respectively) due to the slightly higher sulfur inputs defined
for  most watersheds by the TY dataset.  These results indicate that if deposition were reduced, a rapid
and proportional decrease in sulfate leaching in soils and reduced sulfate flux to surface waters would
occur in  the NE. Because projected changes in sulfate concentration would result in equivalent increases
in ANC and/or decreases in  base cation leaching from watersheds, decreased deposition would result
in substantial increases in ANC or deceleration of base cation removal.

      Projected sulfate concentrations and percent retention approach steady state asymptotically,  and
thus  the response times discussed here  (although short)  are overestimates.   The projected  annual
changes in sulfate concentration and percent retention decrease exponentially as the systems come to
steady state,  and  rates  of  change  become increasingly  small  for  the last few years.  Given the
uncertainties in hydrologic and sulfate measurements and the annual variability in watershed sulfate fluxes,
95  or 105  percent of steady state  is regarded as indistinguishable from steady state.  Time to reach 95
or 1 05 percent of  steady-state concentration is a useful means of describing and comparing watershed
response to altered sulfur deposition. For the current period,  in which sulfate concentrations are declining
in response to reduced deposition, 48 percent of the  systems are projected to be within 5 percent of
steady state at the end of the  base year, 75 percent within 2 years,  and  100  percent within 9 years
(Figure 9-9). Following the decrease in deposition in  the ramp scenario, the most rapidly responding
systems  are projected to have sulfate concentrations within  5 percent of steady state only 3 years after
the end  of the  reductions; projected median and  maximum times are only 6 and  15 years, respectively,
after the end of the decrease in deposition.

9,2.4.2.3 Summary of results for the Northeast -

      Model projections  for  the northeastern United  States, using two deposition datasets and  two
scenarios of future deposition, uniformly indicate rapid soil response to past and potential future changes
In sulfur  deposition to watersheds. At present, watershed sulfur concentrations are projected to be slightly
higher than steady-state concentrations and are decreasing due to recent decreases in deposition.  About
                                               477

-------
                                  Northeast Lakes
                            Time To Sulfur Steady-State
                     1.0,
                     0.0
                                        6        9
                                       Year After 1984
12
                    1.0,
                    0.8,
                    0.6.
                  O-
                  .s o-4-l
                  d
                    0.2.
                    0.0
                                                                B
                                3       6        9       12
                               Year After End of Ramped Decrease
         15
Figure 9-9.  Projected time to steady-state concentration for sulfate in northeastern lakes (A) at
current deposition and (B) after end off decreasing input in ramp scenario.  Results for long-term
average deposition are shown.
                                            478

-------
half of the watersheds in the NE are estimated to have soils within 5 percent of steady state in the 1984
base year, and all are projected to be within 5 percent of steady state within 9 years of the base year.
For a hypothesized future  decrease in deposition, median time to reach steady state (within ฑ 5%) was
projected to be only 6 years following the end of the decrease  in deposition, and maximum projected
response time was only 15 years.  These projections lead to the following conclusions:

      •     To the extent that watershed sulfur budgets in the NE deviate significantly from steady state
           (particularly if they are retaining sulfate), the deviations are probably not the result of sorption
           reactions in soils, but should be attributed to uncertainties in sulfur input/output budgets,  non-
           sorption  sulfur sources (e.g., sulfide mineral weathering), or alternative retention  processes
           (e.g.,  reduction in lakes or wetlands).   It is emphasized that  this Level II  Analysis has
           considered  sorption by soils as the only  process regulating sulfur mobility in watersheds.
           Other processes  are recognized  as having  the  potential to  influence sulfur budgets
           significantly in at least a small proportion of watersheds, but their consideration  is beyond
           the scope of this analysis.

      •     Watersheds in the  NE should be  regarded as direct response systems in terms of suifate
           dynamics.   Soils have low  sorption capacities and  therefore can buffer changes in sulfur
           inputs for only a very few years.   If  deposition is  reduced, watersheds are expected  to
           respond  with a rapid and proportional decrease in sulfur output.

9.2.4.3  Model Results -  Southern Blue Ridge Province

9.2.4.3.1  Evaluation of base  year data, model calibration -

      In contrast  to sulfate chemistry and dynamics in northeastern soils of  watersheds, stream systems
in the SBRP are characterized by a wide range of sulfate concentrations and wide variability  in percent
sulfur retention.  Figure 9-10 shows the deposition sequence used for SBRP watershed modelling, the
sulfate response of a typical SBRP watershed, and the range in  projected sulfate responses for stream
systems in the region. The historic deposition sequence used for the SBRP differs considerably from that
used  for the  NE.  Significant increases in sulfur input  began relatively  recently  in the SBRP, and
deposition reached 50 percent  of current levels only about 25 years ago. Unlike the historic deposition
scenario for the NE,  historic sulfur inputs have never significantly exceeded current levels of deposition.
The lower cumulative deposition to SBRP watersheds and the  high sulfate adsorption capacity of many
soils in the region are the  most important factors affecting the current sulfur budget status of watersheds
in the region.  Typical watersheds in the SBRP presently retain over 50 percent of sulfur inputs, but as
shown in Figure 9-10, sulfate concentrations in SBRP watersheds are now projected to be increasing at
a substantial  rate (proportional to changes in  sulfate flux from the watershed).  Measured increases in
sulfate concentration, at rates comparable to those projected for  SBRP  watersheds in this analysis,  have
been reported for several stream systems in the  region and have been summarized by Church et al.  (in
review). The  range of watershed response rates is much broader than that for the NE: a few watersheds
are projected to be already close to steady state, while sulfate concentrations in others are just beginning
to increase and are not likely to reach steady state during the 140-year period.
                                               479

-------
                               SBRP Streams
     120-
   100-


 x

LL 80-

 Z3
**—

Cฃ) 60-

 CD
  JES
  CD
  DC
    40-
      20-
       0
                   Deposition  Input
                   Stream Output
                     Median Response
                     Range
           1850     1900     1950     2000     2050     2100

                                       Year

Figure 9-10. Historic deposition Inputs and modelled output for soils in stream systems in the
Southern Blue  Ridge Province.  Note the much slower response  compared to systems in the
Northeast, shown In Figure 9-3.  Historic deposition input is based on emission estimates of
Gschwandtner et al. (1985).  Sulfur flux is expressed on a relative  scale, 1985 deposition flux =
100.  The base year for projections (1985) Is indicated by the arrow. Note that because annual
precipitation and runoff are constant throughout the simulation period, changes in flux correspond
to proportional  changes in projected stream sulfate concentration.
                                      480

-------
      An early issue in evaluating SBRP model projections was calibration of the model and data for the
region.   Because both the response times projected  for watersheds and  the  range of base  year
projections are much wider in the SBRP than In the NE, the need for and effects of calibrating input
data are much more obvious for SBRP systems.  As for the NE, the model  runs were begun at -140
years, continued through the base year (1985, when SBRP streams were sampled in the Pilot Stream
Survey), and then for 300 years into the future. Starting with uncallbrated isotherm data, measured and
modelled sulfate concentrations and percent sulfur retention were compared to evaluate bias  and their
distribution.  The measured and  modelled data were in close agreement for both  sulfate concentration
and percent sulfur retention (Table 9-4 and Figure 9-11).  Model projections essentially are unbiased for
both parameters with the average modelled concentration differing by only 2 peq  L"1 from the average
for measured data (39.0 vs  36.8 peq L"1) and average retention differing by only 3 percent  Ranges and
standard deviations of model projections  also closely approximate those  of measured data.  Modelled
sulfate concentrations are slightly higher than measured concentrations over most of the observed range,
although modelled  concentrations are slightly lower at the high  and low  ends of the distribution.
Corresponding relationships for percent sulfur retention indicate lower modelled retention over most of
the range.  Overall, the two sets of data are very similar, and a small systematic  adjustment  could be
made to one  of the isotherm coefficients to completely eliminate bias.  Using LTA deposition data,
differences between measured and modelled  base year sulfate concentration and percent sulfur retention
are very small and are comparable to differences in  projections of base year sulfate concentrations and
percent retention using different  deposition datasets (LTA, TY).  There Is thus no compelling rationale
for adjusting either the data or the model.  For all subsequent projections, therefore, isotherm data were
used without adjustment.

      Comparison of modelled rates of increase in sulfate concentration (for base year 1985) In the DDRP
watersheds to measured rates of increase for watersheds in the region (Table 9-5)  Indicates  generally
good agreement. The range of rates generated by the model for DDRP watersheds encompasses all of
the measured  rates; observed rates (except those for watersheds 2  and 18 at Coweeta) are between the
25th and 75th percentlles of the 34 DDRP  sample watersheds.  The close agreement between  observed
and modelled rates of increase provides additional  support for the use of isotherm data without  extensive
calibration and provides a useful check on the model  projections generated from those data.

      Concurrent with assessments of the need for data adjustments during model calibration, differences
in model projections resulting from use of different deposition datasets were evaluated.  Comparison of
projections based on  LTA  and  TY deposition data (Figure  9-12) reveals systematic  but very small
differences.  For year zero data (expressed as percent sulfur retention) retention is marginally higher for
the LTA data,  whereas projected concentrations are 5 to 10  /*eq  L"1  higher for the  TY data; times to
steady state for the two sets of  projections are very similar.  Time to sulfur steady state is typically 3-
4 years shorter for projections based on TY than for  LTA deposition data.  Given these small differences,
the balance of this discussion will focus only on the long-term results, except to note differences between
the two sets of projections.

9.2.4.3.2  Projections of future sulfate dynamics -

      Projections of future sulfur dynamics for the SBRP differ in almost every respect from those for the
NE.  Projected sulfate concentrations for the  NE  are  slightly above steady state,  and the projected

                                              481

-------
Table 9-4.  Comparison of Measured and Modelled Base Year (1985)  Sulfate
Data for SBRP Watersheds, Using Long-Term Average Deposition Data.  Values
Represent Population-Weighted Mean ฑ  1 Standard Deviation
Parameter                          Measured Value        Modelled Value
Sulfate concentration (fj.eq L"1)          36.8 ฑ 25.7            39.0 ฑ 21.0

Percent sulfur retention                 68.3 ฑ 16.0            64.8 ฑ 17.5
                                    482

-------
                                  SBRP Streams
                                   Year 0 Sulfur
                  Deposition = Long Term Average, Constant
                     -ฐi Concentration
                    0.8.
                   o
                   cu
                   L0.6.
                    0.4.
                  O
                    0.2.
                    0.0
                                50
    100
Syifate 4ieq L'1)
150
200
                     1.0.  Percent Sulfur Retention
                    0.8
                    0.6
                   e
                   a.
                    0.4
                  3
                    0.2.
                    0.0
                                25        50         75
                                   Percent Sulfur Retention
                        100
Figure 9-11.  Comparison of measured, modelled, and steady-state sulfate for stream systems in
the Southern Blue Ridge Province in 1985.
                                          483

-------
Table 9-5.  Comparison of Modelled Rates of Increase for [SO42 ] in DDRP Watersheds
in the SBRP with Measured Rates of Increase in Watersheds in the Blue Ridge and
Adjoining Appalachians
Site
Period of Record
                                           iso/i
L-1)
                                                             Rate of
                                                              2-1
               [SO4 ] Increase
                     I-'1 yf1)
                    References3
DDRP watersheds
model-based
estimates
Cataloochee Cr.,  NC   1968-1981
  15-119

Q, = 0.80
Qa = 1.90

   26
median = 1.21
range 0.2-2.9
                                       1.0
                              this study
Coweeta, NC
WS 2
WS 18
WS 27
WS 36
Deep Run, VA
Madison Run, VA
Fernow, WV
WS 4

1974-1983
II
it
11
1980-1986
1968,1982

1970-1985

13
13
29
24
100
70

85-90

0.7
0.6
0.8
0.8
1.7
1.3

1.0

b



c
d

e
    References:  (a) Smith and Alexander, 1986; (b) Swank and Waide, 1988; J. Waide, personal communication;
    (c) P. Ryan, Univ. of VA, personal communication; (d) USGS, 1969, 1970; Lynch and Dise, 1985; (e) D.
    Helvey, personal communication.
                                               484

-------
                                   SBRP Streams
                             (DDRP watersheds only)
                    1.01
                        Modelled year 0 percent sulfur retention
                                                       Long Term Average
                                                       Typical Year
                               20       40       60      80

                                    Percent Sulfur Retention
100
                    1.01
                    0.8
                         Modelled year 0 sulfate concentration
                  o
                 t
                    0.6"
                 .1

                 O
                    0.4-
                    0.2-
                    0.0
                                                  	Long Term Average
                                                      Typical Year	
                              25     50     75     100     125

                             Modelled Suifate Concentration fpeq L"1)
150
Figure 9-12.  Comparison of forecasts based on two sulfur deposition datasets for soils In SBRP
watersheds.  Modelled-sulfate concentration and percent sulfur retention for the 1985 base year
are shown for long-term average and TY data.
                                            485

-------
response is small decreases in sulfate concentration as systems move toward steady state over the next
decade. In contrast, most SBRP watersheds are presently far below steady-state concentrations; however,
moderate to large increases in sulfate concentrations are projected over time frames of several  decades
to over a century.  As previously noted, the reliability of model projections decreases with the duration
of the projections.  The time Interval of primary interest to DDRP is 0 to 50 years; projections for longer
periods (>100 years) are included principally to characterize the magnitude of potential change following
the 50-year projection period.  Results using LTA and TY deposition data (Figure 9-13, Tables 9-6 and
9-7), indicate significant increases in sulfate concentration and corresponding decreases in percent sulfur
retention for  most SBRP watersheds within 20 years.  At current  deposition,  the projected increase in
median sulfate concentration at year 20 is 24 /xeq L"1, with an additional increase of 25 //eq L"1 by year
50. The range of the increase is 5 to 48 peq  L"1 at 20 years and  15 to 93 //eq L"1  at year 50.  By year
100, when the average projected total increase for stream sulfate is 66 ฃieq L"1,  most watershed sulfur
budgets are projected to be near steady state; increases after year 100 will be restricted to a small subset
of systems with very long projected response times.

      Between years 0 and 20, percent sulfur retention decreases by about 20 percent for soils in most
watersheds, and only a few watersheds approach zero percent retention.  Decreases during this period
appear to be controlled by  deposition/sorption capacity relationships.   After year 20, however, a
substantial number of watersheds are at or very  close to steady state, and by year 50 over  half of the
SBRP  watersheds have less than  10  percent  sulfur retention.  By year  100, over 75 percent of the
watersheds are within 5 percent of steady state, and most have projected retention  of 1 percent or  less.
Only a few systems, with very long response times, remain below steady-state concentration by year 140.

      Box and whisker diagrams (Figures 9-14 and 9-15) summarize changes in sulfate concentration,
percent sulfur retention, and delta sulfate between 0 and 140 years. These diagrams  illustrate not  only
the trends for these parameters, but also the relationships among them.   In particular, sharp increases
in sulfate  concentration and In delta sulfate  are shown at 20 and 50 years.   The increases slow by  year
100  as percent sulfur retention approaches  zero,  constraining further changes In sulfate concentration.

      Using the ramped deposition sequence, no differences in  status at  year 0 are projected,  and
differences in sulfate concentration between constant and ramped scenarios at year 20 are insignificant
(1 neq L"1 or less).  Because increases in sulfur input are not matched by enhanced sulfur  output at  year
20, percent sulfur retention for year 20 is higher for the increased deposition scenario than for constant
deposition.  Major effects of the increased deposition are seen in year 50 projections.  Projected sulfate
concentrations for year 50 are typically 12-15 ^eq L'1 higher for the scenario with increased deposition
than with constant deposition, whereas percent sulfur retention is only slightly higher for the projections
with increased deposition.  By year 1 00, almost all of the increase in deposition can be observed as an
increase in projected sulfate concentration; percent sulfur retention is comparable to, and in most cases,
actually lower for the increased deposition scenario forecasts than for the constant deposition projections
(Tables 9-6 and  9-7).

      Figure 9-16 illustrates projected time to sulfur steady state (ฑ 5 percent of steady state) for current
and increased deposition scenarios.  At current deposition levels, soils in SBRP watersheds are projected
to  reach steady state in as  little as 16 years after the base year, with a roughly linear increase  in the


                                               486

-------
                SBRP Streams
       Percent Watershed Sulfur Retention
        Deposition = Long Term Average,
                   Constant
               SBRP Streams
     Percent Watershed Sulfur Retention
      Deposition = Long Term Average,
           Ramped 20% Increase
                                                  1.0
                                                  0.8-
                                                i
                                                8-0.6
                                                8
                                                QL
                                                $
                                                |0.4
                                                n
                                                O

                                                  0,2
            20
                    40      60
                Percent Sulfur Retention
                                          100
                                                  0.0
           20
   40      60
Percent Sulfur Retention
                                                                                 80
                                        100
                SBRP Streams
       Changes in Sulfate Concentration
       Deposition = Long Term Average,
                   Constant
   1.0
   0.8'
  8.0.6
  e
  a.
 •50.41
 E
   0.0
                                  Year 0-20
                                  YearO-SO
                                  YoarO-100
               SBRP Streams
      Changes in Sulfate Concentration
      Deposition = Long Term Average,
           Ramped 20% Increase
                                                  1.01
                                                  0.8-
.2

1.0J6-
o
0.
                                                ฃ
  O.4
                                                o
                                                  0.2'
            40
                    80      120
                  ASuHatefjieqL:1)
                                  160
                                         200
                                                  0.0
                                                           40
                  80     120
                 & SuHate 4ieq L'1)
                                                                                 160
                                                                                        200
Figure 9*13.  Projected changes in percent sulfur retention and In sulfate concentration for stream
systems In the Southern Blue Ridge Province at 0, 20, 50, 100 and 140 years.  Data for long-term
average deposition, at constant and Increased deposition, are shown.
                                              487

-------
Table 9-6.  Summary Statistics for Modelled Changes in Sulfate Concentration,
Percent Sulfur Retention, and  Delta Sulfate for Watersheds in the Southern Blue
Ridge Province,  Using  Long-Term Average Deposition Data
Sulfate Concentration
Scenario
Year
Stream SO4
Constant
0
20
50
100
140
Steady State
Increased
20
50
100
140
Steady State
Scenario
Year
Stream SO4
Constant
0
20
50
100
140
Increased
20
50
100
140

Mean
36,8

39.0
62.7
88,2
104.8
108.9
110.5

63.0
100.9
126.3
131.4
132.6

Mean
68.3

64.8
43.8
21.1
5.7
1.6

50.1
24.8
5.1
1.0

Std. Dev.
25.7

1.0
31.0
33.7
28.6
25.8
24,7

131.3
40.1
33.8
30.4
29.7

Std. Dev.
16.0

17.5
24.3
22.7
10.6
3.5

21.7
23.0
9.6
2.2

Min.
14.7

12.0
17.2
31.0
65.7
69.5
69.5

17.2
33.1
82.0
83.4
83.4
Percent
Min.
23.7

21.0
3.2
<0.1
<0.1
<0.1

13.0
1.2
<0.1
<0.1

P_25
19.8

21.7
36.7
65.4
86.2
86.7
94.9

36.8
72.8
103.5
106.7
113.8
Sulfur
P_25
65.1

54.3
26.2
4.3
0.2
<0.1

34.5
8.6
0.3
<0.1

Median
23.6

35.3
62.3
89.6
103.1
103,6
103.6

62.4
101.4
124.2
124.3
124.3
Retention
Median
74.9

69.1
42.8
9.6
0.6
0.1

49.3
13.2
0.6
0.1

P_75
40.8

57.5
86.9
111.1
127.0
127.6
127.8

87.6
130.6
152.2
153.1
153.3

P_75
79.1

78.9,
65.0
32.5
3,9
0.8

69.0
36.9
3.4
0.6

Max.
119,2

85,5
134.0
154.0
184.4
189.8
190.4

135.0
179.7
222.5
228.2
228.5

Max.
85,9

83.8
81.9
67.4
30.8
10.0

84.0
71.0
28.0
6.3
Scenario
             Delta Sulfate (Change from Year 0 to n)

Mean  Std. Dev.    Min.     P 25      Median     P 75
Max.
Constant
0-20
0-50
0-100
0-140
Increased
0-20
0-50
0-100
0-140

23.7
49.2
65.8
69.9

24.0
61.9
87.5
92.4

11.5
20.4
23.6
24.3

11.7
25.3
27.9
8.1

5,2
14.6
14.7
14.7

5.2
21.1
28.6
28.6

15,0
39.0
48.2
48.3

15.1
48.1
66.4
66.5

23.6
43.8
67.0
71.3

24.0
58.4
85.2
94.8

29.4
60,7
73.5
74.4

30.4
78.5
101.8
101.9

48.4
93.2
149.1
154.5

49.5
113.9
187.2
192.9
                                            488

-------
Table 9-7.  Summary Statistics for Modelled Changes in Sulfate Concentration,
Percent Sulfur Retention, and Delta Sulfate for Watersheds in the Southern Blue
Ridge Province, Using Typical Year Deposition Data
Sulfate Concentration
Scenario
Year
Stream SO4
Constant
0
20
50
100
140
Steady State
Increased
20
50
100
140
Steady State
Scenario
Year
Stream SO4
Constant
0
20
50
100
140
Increased
20
50
100
" 140

Scenario
Constant
0-20
0-50
0-100
0-140
Increased
0-20
0-50
0-100
0-140

Mean
36.8

43.5
71.1
99.8
116.8
120.5
121.7

71.6
115.0
141.2
145.3
146.0

Std. Dev.
25.7

22.7
32.5
34.3
27.0
23.9
23.0

32.8
41.0
31.5
28.2
27.6

Min.
14.7

12.5
18.6
35.4
77.3
86,3
86.4

18.6
38.1
97.8
103.6
103.7

P_25
19.8

25.2
45.0
81,7
93.1
97.0
104.0

45.1
92.4
112.3
120.3
124.8
Percent Sulfur
Mean
70.8

64.4
41.9
18.8
4.5
1.1

48.4
22.1
3.8
0.6

Mean
27.7
56.4
73.4
77.1

28.2
71.6
97.7
101.8
Std. Dev.
16.7

17.9
24,8
22.4
8.9
2.4

22.2
22.7
7.5
1.3
Delta
Std. Dev.
12.3
22.4
25,4
26.3

12.6
27.1
29.0
29.5
Min.
17.3

19.6
2.6
0.3
<0.1
<0.1

12.3
1.0
<0.1
<0.1
Sulfate
Min.
6.1
16.8
16.9
16.9

6.1
25.7
34.1
34.1
P_25
66.0

54.6
•21.6
2.7
<0.1
<0.1

30.5
5,6
0,1
<0.1

Median
23.6

40.8
72.4
106,5
120.2
120.2
120.2

72.7
123.7
144.2
144.2
144.2
Retention
Median
78.2

68.4
38.9
7,0
0.3
0.1

45.8
10.9
0.3
0.1
(Change from Year 0 to
P_25
19.3
41.3
61,4
61.7

19.9
60,2
81.7
81.8
Median
30.9
56.5
66.9
80.8

31.1
69.8
93.2
102.1

P_75
40.8

57.6
92.4
119.1
131.2
133.3
133.5

92,9
140.3
159.0
159.6
160.2

P_75
82.1

79.6
64.2
31.1
3.1
0,4

68.4
33.4
2.4
0.3
n)
P_75
34.8
74.1
88.0
88.4

35.4
89.8
113.4
113.9

Max.
119.2

106.5
134.1
171.9
199.2
203.1
203.1

136.3
197.8
240.8
243.8
243.8

Max.
87.0

88.0
82.2
66.0
25.7
6.8

84.2
69.5
21.7
3.7

Max.
49.9
122.5
164.3
168.5

50.5
148.3
205.9
209.5
                                            489

-------
                 300-1
                 250~
               A 150-


               8
                 100-



                  50 —
                                       SULFATE CONCENTRATION
                    Measured
                                   20  50   100  140

                                     Simulation Year
                                     Constant Depostion
    20   50   100  140

          Simulation Year
         Increased Deposition
100-
8O-
i
Sz. 60 —
w


"** ป

rti , r-*-.
50 100 140
Fl


If




20









•

rri _ —
SO 10Q 140"
Simulation Year • Simulation Year
Constant Deposb'on Increased Deposition
                 200—i
                 1 SO-
               Sฃ 100-

               ฃ
               ••a
                                 CUMULATIVE CHANGE IN SULFATE
                        20    50     100    140
                               Simulation Year
                              Constant Depositor*
20     50     100     140
       Simulation Year
      Increased Deposition
Figure  9-14.   Box  and  whisker  plots  showing changes in sulfate  concentration, percent sulfur
retention, and change in sulfate  concentration for soils in watersheds of the Southern Blue Ridge
Province.  Data are shown for long-term average deposition data.
                                                490

-------
               300-





               250-





             ^200-

             lj

             9
             A 150-
             O
             w
               100-
                50-
                     SULFATE CONCENTRATION
                  Maasured
              100—1
               80 —
             it60'
             ฅ
             'fS

             I
             ฃ40-

             J3

             I
              20-
   Measured






 3OO —



 250-



• 200-
                                  Simufatton Year
                                 Constant Dcposfion
                                             140
                                           _   _   __, 40 Steady

                                                 Simulation Year
                                                Increased Deposition
                                PERCENT SULFUR RETENTION
                                20    SO    100

                                Simulation Year
                                Constant Dopostion
140    20    SO    100   140
           Simulation Year
          Increased Deposition
                              CUMULATIVE CHANGE IN SULFATE
             1


             8
               100-
                50—
ฃ E
' — .
] E
4
t
20 50 100 140
Simulation Year
Constant Dopostion
                                                     20     50     100
                                                           Simulation Year
                                                          Increased Deposition
                                                                          140
Figure  9-15.  Box  and  whisker  plots  showing changes in suifate concentration,  percent  sulfur
retention, and change in suifate concentration for soils in watersheds of the Southern Blue  Ridge
Province.  Data are shown for TY deposition  data.
                                                481

-------
                                    SBRP Streams
                                 Time To Steady-State
                  ' •""!  Deposition = LTA Constant
                  0,8-
                .9

                foe
                CL
                O
                 10.4-
                o
                  0.2-
                  0.0
                          20    40    60     80    100    120   140
                                Years To 95% Steady-State
                  1.0-
                  0.8
                .2

                f.0.6
                to
                  0.4
                3
                O
                  0.2
                  0.0
                      Deposition ~ LTA, Ramped 20% Increase
                          20    40    60     80    100
                                Years To 95% Steady-State
120   140
Figure 9-16.  Projected time to 95 percent of steady-state sulfur concentration of Southern Blue
Ridge  Province  stream systems.   Results  for long-term average  deposition, for constant and
increased deposition scenarios, are shown.
                                            492

-------
proportion  of systems at steady state until year 75, when about 75 percent wl be at steady state.
Following year 75, the increase in the number of systems at steady state is slower, with about 95 percent
of watersheds reaching steady state by the final year of model forecasts, year 140. For the systems that
reach steady state in more than about 60 years, increased deposition negligibly changes times to steady
state.  For those systems projected to reach steady state in less  than 60  years, especially those that
respond most quickly, increased deposition delays time to steady state.  Higher deposition, coupled with
modest delays in increased watershed sulfur output, maintain these systems below steady state for as
long as two decades.  The  results for these watersheds  do  not  correspond to lower stream sulfate
concentrations.  Higher input simply results in a higher input to output ratio;  projected stream suifate
concentrations are in all cases the same or higher for the increased deposition  scenario than for current
deposition.

      The magnitude and consequences of the projected changes in sulfate over the next 20 to  100 years
on overall stream chemistry in  the  SBRP are substantial.   The  projected changes represent 50 to 100
percent increases in sulfate concentration within 20 years and, on average, about a threefold increase in
sulfate concentration when systems  reach steady state. Increases in sulfate  of this magnitude will cause
major changes in surface water base cations and/or ANC and will accelerate base cation leaching from
soils, as discussed in Sections 9.3 and 10.

9.2.4.4   Uncertainty Analyses and Alternative Aggregation Approaches

9.2.4.4.1   Uncertainty analyses -

      As noted in Section 9.2.3.4, partitioning coefficients (isotherm slopes) for sulfate equilibrium between
soil and solution phases are nonlinear; because the coefficients of these nonlinear isotherms are highly
correlated,  generation of weighted  averages of isotherm coefficients is not an effective or appropriate
means of aggregating isotherm data for soils.  The procedures for aggregation of isotherm date described
in Section 9.2.3.4 are not conducive  to direct estimation of uncertainty for adsorption isotherm coefficients
or for derived variables such as isotherm slope.  The development of uncertainty estimates for Level II
sulfate projections thus required an  alternative approach to data aggregation and use in model runs.

      Section 9,2.3.4 described a Monte Carlo procedure for generating uncertainty estimates  for model
projections. The procedure,  however, also  involved  derivation of new isotherm coefficients and model
projections for each  DDRP watershed that were  developed independently of projections that used  data
aggregated by the original mass-weighting approach.  Because the two sets of independent coefficients
and  the  projections  generated from  them  could significantly differ, the initial concern  in uncertainty
analyses was to assess comparability of the two sets of model outputs. Direct comparison of coefficients
was  not possible, since the uncertainty analysis generated a new aggregate  isotherm for each of the 100
model runs.  Moreover, such an analysis would have been inconclusive  since two sets of very different
isotherm coefficients  can describe virtually identical partitioning curves over the range of dissolved sulfate
concentrations of interest here (0 to  300 peq L"1).   Instead, the  mean  and median values (which are
virtually identical) from the  Monte Carlo simulations for  each SBRP  watershed were compared  to
projections generated using the aggregate isotherm from the original mass-weighting aggregation.
Comparisons were made for projected sulfate concentration and  projections  of time to  sulfur steady
state for  several reference years.

                                              493

-------
      Results indicate very close agreement between the two sets of projections for the base year and
time to steady state (Figure 9-17).  Results for sulfate concentrations at other time intervals also were
similar.  For the comparisons of concentration at year 0 and for projected time to steady state, slopes
and  intercepts of the two lines are virtually  equal to  1.0 and 0,  respectively, and the coefficients of
determination exceed 0.99  in  both cases.   These results  are  important for two  reasons:   (1) they
demonstrate no fundamental Inconsistencies between the aggregation and uncertainty procedures used
to generate the two sets of projections and that the uncertainties developed using  the Monte Carlo
approach can  be  used  to characterize uncertainty  for projections  and  summaries that  use data
aggregated by the routine aggregation approach; and (2) they also suggest that the adsorption isotherms
and  the  projections generated from those isotherms are  highly constrained, i.e., two  different  and
independent data aggregations generate virtually identical projections.

      Mean values  and confidence intervals for projected base year sulfate concentrations for stream
systems in the SBRP are shown in Figure 9-18. Uncertainties are generally modest in magnitude, and
upper and lower confidence intervals are almost symmetrical  and are within 10 to 15 ^teq L"1 of the mean
sulfate concentration. Uncertainties increase very little with mean projected sulfate concentration.  Only
uncertainties in sulfate adsorption  capacity (including those in both the original least squares fitting of
isotherms to raw data points and data aggregation) and soil mass were considered.  Separate analyses
of the components of uncertainty for four SBRP watersheds indicate that uncertainty in soil mass  is the
primary contributor to the total variability in base year projections of sulfate concentration; upper and
lower confidence bounds are within 5 percent  of the median sulfate concentration for the four sets of
Monte Carlo simulations  in which soil mass was held constant.

      Projections of  mean time to sulfate  steady state  in SBRP watersheds (with 5 and 95 percent
confidence intervals) are shown in Figure 9-19.  Similar to base year results, confidence intervals are
almost symmetrical;  but In contrast, uncertainties  in  response time increase with  mean  projected
concentration. Relative uncertainties are  smaller than those observed for the base year concentration,
averaging only about 20 percent of the mean time interval. As was the case for the base year evaluation,
uncertainty in soil mass is the largest contributor to uncertainty  in the  projections;  confidence intervals
for projected time to steady state vary  by less than 5 percent of the  median value in Monte  Carlo
analyses for which soil mass was held constant.  This  result provides additional confirmation that the
isotherm fits are highly constrained and also highlights the important influence of uncertainty in soil mass
on projections involving changes in capacity variables.

9.2.4.4.2  Alternative aggregation approaches -

      Uncertainties in aggregated data associated with the method of aggregation also were considered.
Use of the  sulfate subroutine in MAGIC requires aggregating data for the entire  watershed  into one
compartment per soil  horizon  per watershed.   Aggregating data for a variety of  soils with differing
chemistry, vegetation, hydraulic contact times, etc., inevitably introduces errors and  uncertainty. Johnson
et al.  (1988b) discussed the  rationale for several aggregation  procedures and  described the  mass
weighting approach  based on soil sample  classes that  was used for routine data aggregation for this
analysis.  They did not address the possibility of adjusting aggregated data to account for differences in
soil  chemistry  at  intermediate spatial scales,  nor  did  they  address other approaches  to  describe
"watershed  effects" on aggregated soil  chemistry  (Section 8,9).  As one means of assessing possible

                                               494

-------
                                  SBRP Streams
                             (DDRP watersheds only)
                   Deposition = Long Term Average, Constant
            as
            Q
            c
              100-1
               80-1
            o>
            O3
            O)
            O)

            c  ซH
            o
            E


            O  40-
            cr
            =L  20"
            O
            CO
                  Modelled Year 0 Sulfate Concentration
                                                   trt
                                            y = 0,103 + 1.009SX
                                            r 2= 0.3981
                           20         40         60         80  "

                         SO4 (neq t1 ) - Medial of Monte Carlo Simulation
                                                         100
              1401
            O
            1 100 i
o>

f
o
E
o
O
I
S
               80-
               60-
               40-
                  Time to Sulfur Steady-State
                                                        y = -0.258 + 0.9918x
                                                        r2 = 0.9923
                        20      40      60      80      100     120
                            Years - Median of Monte Carlo Simulation
                                                          140
Figure 9-17. Comparison of model simulation results for DDRP Southern Blue Ridge watersheds.
Data generated by the mass-weighting common aggregation approach and median projected values
from Monte Carlo uncertainty analyses are shown.
                                           495

-------
                              SBRP Streams
                              Year 0 Sulfate
               Deposition = Long Term Average, Constant
    0.81
  c
CL

>
1
3
E
3
o
    0.6-
    0.4-
    0.2
    0.0
        0
                                                   Upper Bound
                                                   Projected Distribution
                                                   Lower Bound
                  20         40          60          80
                      Sulfate Concentration (}ieq L"1)
100
Figure 9-18. Projected base year sulfate concentration with upper and lower bounds for 90 percent
confidence intervals for Southern Blue Ridge Province watersheds.
                                     496

-------
                             SBRP Streams
                       Time to Sulfur Steady State
               Deposition = Long Term Average, Constant
        0.8-
     8.0.6
     o
     CL
     Q>

     Jl 0.4-
     1
     ZJ
     o
        0.21
        0.0
            0
                                            Upper Bound
                                            Projected Distribution
                                            Lower Bound
      50              100
Years to 95% Sulfur Steady State
150
Figure 9-19.  Projected time to sulfur steady state with upper and lower bounds for 90 percent
confidence Intervals in Southern Blue Ridge Province watersheds.
                                    497

-------
watershed  effects, data for the  pectens  sampled on  individual DDRP watersheds in the SBRP were
aggregated and were used with LTA deposition data to make projections.   Measured chemistry and
projections using the standard  aggregation approach described by  Johnson  et ai. (1988b)  differed
considerably  (Figure  9-20).   Projections using the pedons-on-the-watershed  approach substantially
overpredlcted year zero sulfate concentration (modelled mean SO42"  = 67.4 /*eq L"1 vs. measured mean
SO42" =  36.0 peq L"1) and underpredicted percent sulfur retention (median retention of 38 percent vs.
75 percent for  measured retention and 69 percent for standard aggregation  projections). Projections
using the alternative approach also indicated that almost 20 percent  of SBRP watersheds are  already at
sulfate steady state, in contrast to measured watershed sulfur retention, which indicates that more than
20 percent are at steady state.   Additionally, model  simulations  based on the standard aggregation
approach project all watersheds  to have  >20 percent  retention in  the  base  year and no watersheds to
reach steady state for 16 years.

      The overpredictlons using the alternate aggregation (pedon-on-the-watershed) result from the  soil
sampling design in the SBRP. The design called for approximately equal numbers of samples for each
soil sampling class, even though the spatial areas covered by the classes are widely variable. As a result,
for this aggregation approach, in which data were arithmetically averaged, soils collected from sample
classes with below-average areal spatial coverage were assigned artificially high weights. Because sample
classes with  small areal  coverage  (e.g.,  shallow or flooded soils)  often have low sulfate adsorption
capacity, averaged adsorption capacities are biased low and corresponding projected response times are
short.  This component of uncertainty Is  not introduced In projections using  the common aggregation
approach,  because data are  area weighted; the alternative aggregation approach, moreover, raised
questions about sampling design and the magnitude of uncertainties in parameter estimates.  For these
reasons,  it was  dropped from consideration for this analysis.  The issue of watershed effects (watershed-
to-watershed  differences  in  soil  chemistry) Is under active investigation,  and changes in aggregation
procedures for  future analyses remain a possibility.

      A second question concerns  the  number  of soil  horizons used for aggregation.  The model
formulation treats soils as a series  of continuously stirred  tank  reactors (CSTRs) in which reactions
proceed  to equilibrium. This treatment results in model projections that are sensitive to the number of
CSTRs.   At one extreme, for a one-CSTR model,  projected output concentrations respond immediately
to changes in inputs and responses  are sustained over a long period  of time. At the other extreme, a
model with an  infinite number of small CSTRs (having a  total soil  mass equivalent  to that  for the 1-
compartment model) would act much like a chromatography column:  output  remains constant  until the
breakthrough of the front through the soil column,  at which time output  concentration increases as a
square wave (ignoring dispersion) to steady state.  The number of  horizons, which can be  varied for the
Level II Analysis, will affect the timing of projected changes as well as the concentration at any point in
time (base  year in this case). Results of running the  model with  soil chemistry data aggregated to 1,
2, and 3 horizons (A/E, B, and C) are displayed  in Figure 9-21.  Few differences between projections
for 2- and 3-horizon aggregations are evident, but 1-horizon projections do differ. The close agreement
between  2- and 3-horizon projections was not  unexpected; the A/E horizon Is thin and  has relatively
low adsorption capacity (i.e., the A/E horizon CSTR has a very short response time), so combining it with
B horizon data has little  effect on projections.  On the other hand,  combining  all data in one horizon
results In a system that  responds immediately to  altered inputs (thus resulting In projections of higher
                                              498

-------
             t.Oi
             0.8-
          E
          O
          tf

          %.
          2
          CL
0.6-
             0.4-
          =>
          E
          =J
          O
             0.2-
             0.0
                                  SBRP Streams
                            (DDRP watersheds only)
                 Deposition = Long Term Average, Constant
                Year 0 Sulfate Concentration
                                                      Measured
                                                      Modelled
                                                       Sample Class agg.
                                                       Pedon on Watershed agg.
                      20
                             40
60
                               SO
100
                                                          120
                          140
                                                            160
                                      Sulfate Oxeq L1)
             0.8-
             0.0
                Years to Sulfur Steady-State
                                                      Sample Class agg. •
                                                      Pedon on Watershed agg.
                      —,—
                       20
                  —i—
                   40
60
           —,—

           80
    —i—

     100
                        —I—
                         120
                                                                         140
                                 Years to 95% Steady-State
Figure 9-20.  Effects of data aggregation on simulated watershed sulfur response for soils in DDRP
watersheds of the Southern Blue Ridge  Province.   Results for the common (sample class)
aggregation procedure and for an alternative aggregation using pedons sampled on each watershed
are shown.
                                          499

-------
                                  SBRP Streams
                             (DDRP watersheds only)
                  Deposition = Long Term Average, Constant
                  10,  Modelled year 0 percent sulfur retention
                  0.8
                _o

                0.0.61
                S
                O_
                
-------
base year sulfate concentrations); it also results In a system that responds more gradually than a multiple-
horizon model, projecting longer times to steady state. Model data provide no real  basis for choosing
among 1-, 2-,  or 3-horizon models for routine modelling efforts.  Based on differences in soil chemistry
among A/E, B, and C horizons and on the good fit between measured and modelled projections of base
year sulfate  concentration and percent sulfur retention using  the three-horizon model, the three-horizon
model was chosen for routine analysis.

9.2.4.5  Summary of Results from the Southern Blue Ridge Province

     The response of soils and surface waters in the SBRP to sulfate deposition represents a "textbook
example" of delayed response watersheds.  The response to major sulfur deposition increases that have
occurred over the last two to three decades has been  high watershed sulfur retention, with only modest
increases in stream sulfate concentrations for most SBRP watersheds.  Measured data summarized by
Church et al. (in review) and model projections indicate, however, that the delay is now ending and that
surface water  sulfate concentrations  are increasing at  rates  projected to  accelerate over the next few
decades.  Major increases in stream sulfate concentration are projected for SBRP streams  in the next
20 years, with continued increases for at least 50 years for most watersheds.  When SBRP watersheds
come to steady state for sulfate (at projected times ranging from 16 to >150 years) sulfate concentrations
will be, on average, about three times current concentrations. The  projected changes in stream sulfate
will result In  substantial changes in streamwater chemistry  and could substantially accelerate base cation
leaching from  soils.

     The results of these analyses are generally consistent with those of other DDRP analyses.  Model
projections of base year sulfate in soils of northeastern and SBRP  watersheds are consistent with, and
provide a mechanistic explanation for, analyses by Rochelle and Church  (1987).  Their analyses,
summarized in Section 7.3, show watersheds in  the northeastern United States to be at or near sulfur
steady state, whereas SBRP watersheds have high net sulfur  retention.  The very short sulfate response
times projected for the NE are also consistent with results of regression analyses presented  in Sections
7 and 8, which indicate that deposition is the principal control  on surface water sulfate in the NE and that
significant sulfur retention (where observed) is probably attributable to sulfate reduction in lakes and/or
wetlands rather than to sorption.

     The short sulfur response times projected for northeastern soils in this analysis are comparable to
watershed response times projected by integrated models  (Section 10). Projected response times for the
SBRP are roughly comparable to those generated for SBRP watersheds by MAGIC (Section 10,11),
although the projections of time to steady  state generated in Level II Analyses for  SBRP systems are
generally somewhat shorter  than the MAGIC forecasts. Two factors are believed to contribute to the
differences in projections:

            Hydrologic routing in the two models is different.  The Level II  projections used a simplified
            routing in which all water was routed through all  soil horizons, while a substantial portion of
            runoff in MAGIC simulations bypassed either the upper or lower soil compartment.

            Soil depth was treated differently by the two  modelling efforts.  Level II  models considered
            adsorption within the top 1.5 to 2 meters of the regolith while the Level III models assigned

                                              501

-------
           the adsorption capacity (and other chemical properties) of the lower soil horizon to all
           material between the B-C horizon boundary and the estimated depth to bedrock. The MAGIC
           projections are therefore based on a larger mass of soil, having a larger integrated pedon
           adsorption capacity, which ultimately results in a slower projected response to changes in
           sulfur deposition.

Despite the differences between  the two modelling  approaches (Level  II limited to adsorption in the
developed  soils  as  compared  to Level  III  models which integrate  hydrologic  processes  with
blogeochemical processes in the  entire catchment), the magnitude of differences between the two sets
of projections was generally small. Results are viewed as mutually supportive, both of the two modelling
approaches and of the projections generated by them.

9-2-5  Summary Comments on  Level II Sulfate Analyses

      At the start of the DDRP, it was widely believed that soils in the northeastern United States had low
sulfate adsorption capacity, resulting in rapid sulfate response to changes in sulfur deposition and further
resulting in watershed sulfur  budgets near  steady  state.  Conversely,  observed  sulfur retention in
southeastern  watersheds was attributed to high sulfate  adsorption capacity of soils in that  region.
Measured sulfate data for the two regions and model forecasts summarized in Table 9-8 are consistent
with previous soil sulfate data and provide strong support for this paradigm of regional sulfur dynamics.
DDRP model projections also Suggest fundamental  differences in future  sulfate dynamics of the  two
regions.  Northeastern watersheds are very close to steady state; assuming constant deposition at current
levels for the future,  only small  changes in sulfate concentration are  anticipated as systems reach
equilibrium with deposition inputs. If deposition were to change in the future, model projections suggest
very rapid response by watersheds in the region, with systems projected to reach steady state with the
altered deposition inputs  In 5 to 15  years.

      In the  SBRP, sulfate adsorption by soils has delayed  effects of  acidic deposition, but model
projections indicate that soils and  watersheds in the region are  now moving into a more dynamic phase,
in which relative adsorption by soils will decline and stream sulfate concentrations will increase  sharply
in the coming decades.  Major changes in stream water sulfate are projected for the next 20-50 years.
If and when they occur,  equivalent  changes in surface  water base cations or ANC are  inevitable,  and
enhanced leaching of soil base cations is likely to occur.

      In a qualitative sense,  the differences of the current status and projected future sulfate dynamics
for the two regions are unequivocal. Sulfur budget status and soil adsorption isotherm data document
clear differences  in soil  and surface water chemistry for the  two regions, and projections of future
response times differ by roughly an  order of magnitude.  In making such comparisons, it is important to
recognize that the models embody a variety of assumptions and approximations and that the projections
carry significant uncertainty.  Given the magnitude of the differences in projected responses for the two
regions,  however, their responses to sulfur  deposition undoubtedly are  also very different.   Sulfate
retention appears to  have been  a  minor contributor to  delays in surface water response  to acidic
deposition in  the NE, but has been and continues to  be a  critical process delaying effects of deposition
in the SBRP.   In considering the  projected responses for  sulfur, especially in the SBRP, it is important
to recognize that projections presented here apply only to  sulfate and are based on the assumption that

                                               502

-------
Table 9-8. Summary Comparison of Watershed Sulfur Status and Model Forecasts in
the Northeastern United States and  Southern Blue Ridge Province.  Model Projections
are Based on Long-Term Average Deposition Data, Assuming Constant Future Deposition
                                NE Lakes
SBRP Streams
Median Range
CURRENT STATUS
Steady State Sulfate
Lake/Stream Sulfate
Percent Sulfur Retention
MODEL FORECASTS
Sulfate Cone, (^eq L"1)
Year 0
Year 20
Year 50
Year 100
Year 140
Percent Sulfur Retention
Year 0
Year 20
Year 50
Year 100
Year 140
Delta Sulfate (/^eq L"1 yr"1)
Year 0-20
Year 0-50
Year 0-100
Yoar n.ldfl

106.0
105.4
-3.1


114.3
106.0
106.0
106.0

-7.9
-0.1
>-0.1
>-0.1
	

-6.8
-6.8
-6.8

50.8 -
33.8 -
-60.0 -


54.7 -
50.8 -
50.8 -
50.8-

-19.3
-1.1
-0.2
>-0.1
...

-37.6
-39.7
-39.7

209.6
249.3
61.1


249.3
211.7
209.6
209.6
- -1.3
- 0.1
- 0.1
- 0.1
— .

- -0.7
- -0.7
- -0.7
Median

103.5
23.6
74.9


35.3
62.3
89.6
103.1
103.6
69.1
42.8
9.6
, 0.6
0.1

23.6
43.8
67.0
71 3
Range

69.5 -
14.7 -
23.7 -


12.0 -
17.2 -
31.0 -
65.7-
69.5 -
21.0 -
3.2 -
<0.1 -
<0.1 -
<0.1 -

5.2 -
14.6 -
14.7 -
1/1 7 -

190.4
119.2
85.9


85.5
134.0
153.9
184.4
189.8
83.8
81.9
67.4
30.8
10.0

48.4
93.2
149.1
1 fiAfi
                                          503

-------
adsorption and desorption are the only processes influencing watershed sulfur retention.  Finally, readers
should be cognizant of the complexity in the relationships among sulfate, base cations, and ANC, which
are Influenced by several processes. The timing and magnitude of changes In ANC need not be directly
correlated with changes in sulfate.  In particular, time to steady state for sulfate should  not  be equated
to time to zero (or to any other threshold value) ANC.

9.2.6 Conclusions

      Watersheds in the northeastern United States can be characterized as direct response systems in
terms of sulfate dynamics mediated by sorptlon in soils. Northeastern watersheds are near sulfur steady
state and  are  projected to respond quickly to changes in sulfur deposition.

      •      In the base year (1984), median measured percent sulfur retention was -3.1  percent for LTA
            deposition, 0 percent for TY deposition data.

      •      Modelled percent sulfur retention for the  base year was slightly negative for both LTA and
            TY data, -7.1 and -6.8 percent,  respectively.

      •      If deposition continues at current levels, all northeastern watersheds are projected to be within
            5  percent of steady state in less than 10 years. Median sulfur concentrations will  decrease
            to within one percent (2 /xeq L"1) of steady state in 20 years.
           At current deposition, changes In median sulfate concentration as watersheds reach steady
           state will  be small  (7 /^eq L"1) with a maximum of 40 /^eq  L"1 (for LTA deposition); for TY
           deposition,  median and  maximum  projected  changes  in sulfate  are 7 and  43 peq L"1,
           respectively. Changes will have little impact on overall water chemistry of most lakes in the
           region.

           If deposition is decreased 30 percent, the magnitude of changes would be much greater, with
           a decrease In  median sulfate of 39 or 40  /ieq L"1 (LTA and TY deposition, respectively).
           Maximum projected decreases are 103 and 115 
-------
      Watersheds in the Southern Blue Ridge Province should be  characterized as delayed response
systems. Sulfate adsorption by soils has  minimized the effects of  acidic deposition on surface water
chemistry in the region. Sulfate concentrations in SBRP watersheds are projected to increase significantly
in the next 20 to 50 years, however, as the adsorption capacity of soils is exhausted.

      •     Median measured sulfate retention in SBRP watersheds for the 1985 base year was  74.9
            percent for LTA deposition, 78.2 percent for TY deposition data.  The percent retention varies
            from 24 to 86 percent for LTA deposition.

      ซ     Median modelled retention for the base year is also high: 69.1 and 68.4 percent for LTA and
            TY deposition  datasets, respectively.   The range  of modelled percent retention for  LTA
            deposition was 21 to 84 percent.

      ป     Time to steady state at current deposition varies from 16 to more than 150 years; median
            projected time is 61  years.   At 20, 50, 100, and 140 years from  the base  year, projected
            median percent sulfur retention is 43, 9.6, 0.6, and 0.1 percent. Maximum projected retention
            for the  same periods is 82,  67, 31, and 10 percent.

      *     As soils in SBRP watersheds reach steady state, average sulfate concentrations in watershed
            runoff  will increase  roughly threefold.  The median sulfate concentration  is projected  to
            increase from 35 peq L"1 to 62,  90, 103, and 104  /zeq L"1  at 20, 50, 100,  and 140 years.
            Maximum projected increases for sulfate for the same periods are  48, 93, 149, and 155
           For the 20 percent increase in deposition In the SBRP, times to steady state increase by up
           to 20 years for the watersheds with short retention times, but are almost unchanged for most
           watersheds.   The increase  in  deposition  has almost  no  effect  on projected sulfate
           concentrations at year 20, but results  in significant increases  in delta sulfate at later years
           compared to the constant  deposition scenario. Projected  increases  in  median sulfate
           concentration for the increased deposition scenario at years 50, 100, and 140 are 58, 85, and
           95 peq L"1' maximum projected increases are 114, 187, and 193 ^eq L"1.

           Model projections indicate that adsorption  of sulfate by  soils has played a major role in
           delaying potential adverse effects of sulfur deposition on surface waters in the SBRP.  Most
           SBRP watersheds will not reach sulfate steady state for several decades, but  significant
           increases in sulfate concentration are projected for the next 20 to 50 years.

           The large increases in sulfate concentration projected for  the  next 20 to 50 years will have
           major implications for overall surface water chemistry and are likely to accelerate base cation
           leaching from soils.
                                               505

-------
9.3  EFFECT OF CATION EXCHANGE AND WEATHERING ON SYSTEM RESPONSE

9.3.1  Introduction

      During the development of the NAS panel report (NAS, 1984), much discussion was devoted to the
role of cation exchange and mineral weathering in "protecting" watersheds from acidification.  One group
of panel members argued that cation exchange in most watersheds has a large capacity to buffer against
potential changes caused by acidic deposition. Therefore, they argued, if cation exchange is an important
process within a specific watershed, then the future effects of acidic deposition are  probably not a
concern.  Another group of panel members argued that the buffering capacity of soils was finite, and that
continued exposure to current levels of acidic deposition would have long-term, adverse effects on water
quality in  some  systems.  The conclusion of the committee as  a whole was that the role of cation
exchange in buffering against the effects of acidic deposition is an area of considerable uncertainty, and
that these processes need to be considered when attempting to project future effects of acidic deposition
on aquatic ecosystems.

      Toward this goal,  the Level H base cation studies were designed to determine the role of base
cation exchange in controlling future changes  in surface  water  chemical composition.  The specific
objectives were to

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

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

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

      Background information concerning weathering and base cation exchange processes is presented
in Section  3.4.   Given the objectives stated above .and relying  on  our  current understanding of the
processes presented in Section 3.4, a number of hypotheses were developed that were used to guide
the investigations of  the role of cation  supply processes in regulating surface water chemistry in
representative watersheds in  the DDRP study regions.

9.3.1.1  Level II Hypotheses

      Five hypotheses guided the investigations in the Level II analyses:

      (1)   Cation exchange processes determine surface water composition.

      (2)   Soils delay surface water acidification.

      (3)   Increased deposition induces net cation leaching.

                                              506

-------
      (4)   Cation resupply rate is slow.

      (5)   Soil chemistry Is an Indicator of soil response to acidic deposition.

It was not possible in all cases to test the hypotheses with the survey data collected for the Project.  For
example, testing of the fourth hypothesis requires time series data, the collection of which was beyond
the scope of the DDRP.  As a result, this hypothesis was treated as a "system-level" assumption for the
analyses, the implications of which are discussed below.

9.3.1.1.1  Cation exchange processes determine  surface water composition -

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

      The  hypothesis is  tested  by comparing surface water  composition  projected using  soil  cation
exchange models with observed values.  A close correspondence between  the observed and projected
values suggests that soil exchange processes have a major role in regulating surface water chemistry.
Major discrepancies  between  observed and projected  values  would  provide  information  regarding
alternative controlling processes. For example, if the outputs from the soil models  suggest that soils are
strong hydrogen ion buffers, i.e., if the aggregated model  results fall into narrow ranges  of pH and ANC,
this would suggest that  other  processes,  such as primary mineral  weathering, are serving as major
sources (or, for cation accretion into biomass, sinks) for base cations in the population of watersheds
being studied.  Examining  this hypothesis, therefore, provides bounds for  arguments  regarding which
processes are primarily Involved In regulating observed surface water composition.

9.3.1,1.2  Soils delay surface  water acidification -

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

      When a perturbation such as acidic deposition is imposed on a system, the system (in this case
the soil) evolves toward a new state of balance.  The  rate at which changes take place depends both
on the sizes of the cation reservoirs in the system and on the flux of material between reservoirs. If the
transfer rates of material  between reservoirs is slow, or if the mass of material in the affected reservoirs

                                              507

-------
is large relative to the transfer rates, then rates of evolution toward the new balance point tend to be
slow.  Conversely, if  reservoirs are  small  relative to the size of the flux  between reservoirs, then
adjustments to a  new system state can occur rapidly.

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

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

9.3.1.1.3  Increased deposition Induces net cation leaching -

      The  third   major  hypothesis  is  that  increased levels   of  deposition,  specifically  increased
concentrations of sulfate and  nitrate in deposition, increase the rate of  cation  leaching from the soil
exchange complex by way of the mobile anion effect (Johnson et al., 1980; Seip, 1980). Two factors are
considered when  evaluating this hypothesis.  First, the average base status of the soil exchange complex
represents a balance among the various supply and demand processes in  the ecosystem. For example,
under steady-state conditions, weathering should supply sufficient base cations to meet the demands of
vegetative uptake while maintaining soil  solution  concentrations in equilibrium  with the soil exchange
complex.  Perturbations to the system, such as changed deposition, will alter this steady-state condition.

      Second, charge balance requirements  need to be maintained between the soil exchange complex
and  soil  solutions.  Maintaining charge balance,  coupled  with the increased anion loads provided by
acidic deposition, requires that total (acid  plus base) cation  concentrations  in soil solutions increase. The
ratio of the base  to acid cations will not change  dramatically, however, at least during the initial stages
of leaching.  The  higher concentrations of base cations in  soil solutions lead to a net depletion of base
cations from the exchange complex.  If this Increased leaching is not matched by an increased level of
supply (e.g., from weathering), then the overall effect will be a net depletion of the base cations from the
exchange complex.

      We have tested this hypothesis using a modelling approach. As for  the second  hypothesis,  model
runs  are conducted that  enable the determination of whether  the increased anion concentrations in
deposition will, indeed, result in an increased  rate of leaching of the base cations.  The mass balance
computations,  in combination with the equilibrium  mass action descriptions of the system, should permit
an unequivocal evaluation of this concept.
                                              508

-------
9.3.1.1.4  Cation resupply is slow -

      The fourth hypothesis,  that the rates of cation  resupply to the soil exchange complex are slow
relative  to the rates of base cation stripping, is not being tested  directly in this  study.  Rather,  the
hypothesis is being subsumed in the models as an assumption, or,  more accurately, the assumption is
that exchange reactions provide sufficient buffering such that resupply rates are not an issue for the time
scales of concern to the study.

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

9.3.1.1.5  Soil chemistry as indicators of soil  response to acidic deposition -

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

      The hypothesis is being tested using two approaches.  First, the results from the Level I statistical
analyses (see Section 8.8.4) have been presented. These results suggest significant relationships between
present day soil properties and observed  surface water chemistry.  These observations  support  the
contention  of the relationship between  basic soil properties and the response  of the system to acidic
deposition.  Second, as part  of the Level II modelling activities, relationships will be examined between
current  soil  properties  and the magnitude of projected changes in soil and surface water composition.
While these results will not Integrate the roles of multiple processes, e.g., weathering and ion  exchange,
they should provide some additional evidence for examining the hypothesis.

9.3.1.2  Approach

      As previously discussed (see  Section 9.1), the approach used for  Level II base cation analyses is
model-based.  The primary processes believed to regulate exchange processes are  known, and models
have been developed that describe these processes in internally consistent manners.  As such, existing
model formulations are used  extensively in conducting these studies.

      Data used in running the  models were collected specifically  for this study.   Section 5 provides
details of the type,  quantity, and level of information gathered. In collating the data for use in the models,
certain  decisions were made regarding  how data from  individual soils and  watersheds  would be
condensed,  or aggregated, for use  in the models.  Because the primary goal of the DDRP is to make
regionally representative projections about future changes in surface water chemistry as a direct result

                                               509

-------
of acidic deposition, a decision was made to aggregate the data, first into groups of soil with similar
chemical and physical characteristics and, then, to the watershed level.

      Because this aggregation approach was used, projections regarding individual watersheds will not,
necessarily, be accurate reflections of the chemistry observed in that watershed. On a population basis,
however, the models should provide useful information about the anticipated behavior of the soils in the
DDRP study  regions.  Details regarding the soil  aggregation procedures are outlined below (Section
9.3,1.2.2).

      Finally, the rationale used  to define the time scales over which simulations were  executed  is
presented.  Model projections are  inherently uncertain.  As the durations of simulations increase, the
associated overall uncertainty increases. Therefore, there are practical limits to the usefulness of long
time frame projections.   Section 9.3.1.2.3 provides a brief discussion about the  trade-offs between
uncertainty and information gained.

9.3.1,2.1  Off-the-shelf models -

      In designing the Level  II base cation studies, one of the issues considered was the  selection of
models.  As discussed in Section 2,  a decision was made during the planning stages of the DDRP to use
only published, peer-reviewed, and publicly available models.  A primary advantage of this decision was
that the data requirements for these models were known, so the field programs could be developed to
collect the appropriate data required. A second advantage was that minor modifications or improvements
could be incorporated into the model codes  in a  timely manner.  Because of concerns relative to field
design issues, and because the report from  the NAS  (NAS,  1984)  indicated that models describing the
major  soil  processes controlling  base  cation dynamics were available, only published and publicly
available models were selected for application in the Level II base cation studies.  The selected  models
are described in Reuss  (1983), Reuss and Johnson (1985), and Bloom and Grigai (1985)  (see Section
9.3.2).

9.3.1.2.2  Aggregated soil chemistry data -

      Having selected models for use in the Level II Analyses, the next major issue was preparation of
data for use in the models. Soil physical and chemical data were gathered on a representative sampling
of soils in the NE and SBRP (Section 5.5.1).   These data were obtained from individual pedons and soil
horizons.  To transform these data into a form  usable by the  models, the data were aggregated to
produce  information that was representative  of whole  watersheds.

      Details  of the aggregation procedures were presented by Johnson et al. (1988b). Briefly, the steps
taken to  produce the aggregated  data depend on the structure of the model to be applied. In general,
data are first averaged within the  master horizons (i.e., O, A/E, B or C horizons)  of individual soil
sampling classes. Then, if required by the models (e.g., those that describe the soil as a single "box"),
results from the master horizons are  averaged to yield values of parameters representative of the sampling
class as a whole. The procedures used to average soil chemical and physical  properties at the horizon
and sampling class levels varied slightly in accordance with the model for which the data were being
developed. For models that use capacity variables as inputs, e.g., the Bloom-Grigal model, soil properties

                                               510

-------
were averaged using  mass weighting procedures.  For models using intensity variables as Inputs, an
intensity weighting scheme (Johnson et al., 1988b) was developed that preferentially weighted the lowest
subhorizon in  generating values for master horizons, and then employed straight numerical averages to
produce sample class/pedon data.

      Finally, data from individual sampling classes were averaged, using areal weighting, to produce soils
data representative of the watershed as a whole.  The weighting used In this last aggregation step was
strictly related  to the relative occurrence of the sampling class on a particular watershed.  The weighting
precludes bias based on the location of  the  soil  on a watershed.  For  example,  although it might be
argued that riparian soils have a greater influence on the composition of surface waters than do ridge-
crest soils, riparian zone soils and those soils immediately adjacent to the lakes  are not preferentially
weighted  relative to upland soils.   The decision  to use the uniform weighting  approach was based
primarily on the difficulty of developing uniform, broadly based algorithms to apply  preferential weighting
to specific soils based on geomorphie considerations.

9.3.1.2.3  Scale of temporal forecasts -

      Another decision to be made in implementing the Level II Analyses was the time scale over which
to run the model  simulations.   In the  near term, dramatic,  permanent  changes to  surface water
composition are not expected to occur on annual  time  scales.  Acidic deposition is a phenomenon that
has probably affected eastern North America for at least several decades. Rapid responses to changing
deposition, If they were to occur, have probably already taken place.

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

      Using these procedures as guidelines for bounding the time intervals to be modelled, simulations
for 20, 50, and 100 years were selected for the NE Region. For the SBRP, simulations for 20, 50, 100,
and 200 years were selected. The 20- and 50-year projections provide information about relatively near-
term changes  that might be anticipated and are relevant time frames  with regard to the implementation
of regulatory controls.

      The 100- and 200-year projections are included as "worst-case" results.  Such projections will allow
policymakers to understand  the magnitude of changes that could  occur.  The 200-year simulations are
included for the SBRP  primarily because major changes to sulfate mobility in soils in this region are
expected to occur during the next century (see Section 9.2).  By extending the model simulations for an
additional  100  years,  the full effect of changes in  mobile anion  concentrations  will become  evident.
Inasmuch  as the  NE is, essentially, at steady  state with regard to sulfur deposition (see Section  7), this
additional  time is not required.
                                              511

-------
9.3.2  Descriptions of Models

     The model originally selected to conduct the base cation analyses was one developed by Reuss
(1983) and Reuss and Johnson  (1985).  This  model uses a mass action approach to  modelling soil
exchange processes.   As  such, the model  requires a  broad range of data  as input, which  was
incorporated into the design of the field program for the Project.  An additional factor in the selection of
the Reuss model for use in the Level II Analyses was the fact that its data requirements are compatible
with those of some  of the Level Hi models to be used in the study.

     In addition to the Reuss model, one other model was incorporated into the Level II base cation
studies,  a model developed by Bloom and Grigal (1985). This model describes soil exchange processes
based on observed  relationships  between the cation exchange pool and soil pH.  This model, therefore,
not only expands the model base from which  the Level II Analyses are conducted, but also provides
an alternative approach for describing soil exchange reactions.

9.3.2.1  Reuss Model

9.3.2.1.1  Model description -

     The Reuss model was originally developed by  Reuss (1983) and coworkers (Reuss and Johnson,
1985; Johnson and  Reuss, 1985). The model is an equilibrium-based, mass balance  model in which the
solubility of a gibbsite-like phase is assumed to control the concentration of aluminum.  Subsequently,
exchange reactions are used to  partition the cations AI3*, Ca2*, Mg2*, Na"1"  and K+ between the solid
and solution phases.  Figure 9-22 presents schematically the processes considered  in the model.  The
model computes soil  pH, soil solution ANC, and base  cation and dissolved aluminum concentrations.
The model then "re-equilibrates"  soil solutions with atmospheric carbon dioxide and computes surface
water composition.                            _

      Reuss's approach has several advantages for  modelling exchange reactions in soil environments
over the use of simple exchange  reactions. First, the charge balance requirement of the code makes the
model responsive to ionic strength. In forested soils, composition of the soil solutions have been shown
to depend on ionic strength (Richter et al., 1988).  Therefore, this aspect of the  model permits a more
realistic simulation  of natural exchange reactions than do the less involved computations.  Second, the
model allows the user to specify  the partial pressure  of carbon dioxide (pCO2) in the soil  gas.  Although
the pCO2 in forested soils rarely  exceeds about 1 percent (Fernandez and Kosian, 1987; Solomon and
Ceiling, 1987), these levels can be high enough to significantly affect soil solution composition  (Reuss
and Johnson, 1985, 1986).  Third, by relying on a gibbsite-like phase to regulate aluminum activities, one
degree  of freedom  in  the  solution composition is  effectively  constrained.   Finally, the mass balance
constraints allows the user to track cation depletion  from  the exchange complex as a function of time,
hydrogen ion loading, and the imposed physicochemical environment.

     The Reuss model focuses on soil exchange reactions.  The model does not consider other cation
source/sink processes such as mineral weathering, nitrogen transformations, or afforestation, even though
these processes may  have  equal or greater Influence in regulating  surface water composition in certain
ecological settings.  Models have been developed that  include these processes,  and thus yield an

                                             512

-------
       Top
    Horizon
        n
    Middle
  Horizon(s)
    Bottom
   Horizon

                         ETPan
                    AI(OH)3
AI(OH)3
                   Ai(OH)3
                               OK)
Figure 9-22.  Schematic diagram of the principal process involved in the cycling of base cations
in surficial environments. Arrows indicate the major pathways through which ions are interchanged
among the reservoirs. Ho attempt is made to distinguish the relative fluxes among the different
reservoirs.  The heavier lines, however, indicate those processes that serve as the focus of the
Level II modelling efforts presented here.
                                      513

-------
integrated system response (Cosby et al., 1985a,c, 1986a,b; Gherini et al., 1985; Galloway et al., 1983a;
Nikolaidis et al.,  1988) to the imposed deposition. These integrated models, however, cannot be used
effectively to understand the contributions of individual processes, such as soil cation exchange, to the
buffering responses of watersheds.

9.3.2.1.2  Model Formulation  -

      The original versions of Reuss's code were written in BASIC.  Table 9-9 lists the chemical species,
principal reactions, and related  chemical definitions used in developing the computer code.  The original
versions were written for a one-horizon setting in which the water flux, rather than time increment, was
used in scaling the step sizes for time-series simulations.  Reuss's codes  also ignored ammonium inputs
to soils,  thus effectively using  H+ as the surrogate  for NH4+ deposition.   In  incorporating the Reuss
model into the DDRP, a number  of modifications were made.

      The model was adapted for this study by receding in FORTRAN, which enabled greater execution
speeds and, thus, an ability to  handle more simulations.  The formats of the input and output datasets
were  revised to  better accommodate the needs of  this study.  In receding the  model, a number of
operational changes were implemented.   For example, the data of May et al. (1979) instead of those
employed by Reuss and Johnson (1985) were selected to describe aluminum speciation.  in addition, the
algorithms  used  to  partition ions between solution and the solid-phase exchangers were modified to
provide more accurate  mass action expressions for thin and low base saturation horizons.  Given these
changes, rigorous one-to-one comparisons of results obtained from the  FORTRAN and BASIC versions
of the model have not been possible, as the two models yield slightly different results.  A more substantial
modification to the code entailed the use of the Vanselow exchange formulation.  In the original versions,
the Reuss code employed the Gaines-Thomas formulation for cation exchange processes.  Comparisons
of three exchange formulations (Holdren et al., 1989), including the Gaines-Thomas, Vanselow, and Gapon
models, suggested the Vanselow model provided results more representative of field data than the other
two models. Differences among  the three models, in general, were small, but  significant.

      The selectivity coefficients for the Vanselow formulation are based on a mass action expression of
the form:

                      {Nn+}m.  [Xw]n
             Ky =   	                                         (Equation 9-5)

                      { Mm+ >n '  [ XN ]m
where [XM] and [XN] are the mole fractions of the solid species indicated, m and n are the appropriate
stoichiometrie coefficients; the species enclosed in the braces {i}  indicate activity  of the i*  aqueous
species.  The specific mass action expressions for the exchange reactions considered are listed in Table
9-9.
                                              514

-------
Table 9-9.  List of the Chemical Species and Reactions Considered Within the Reuss
Model Framework
Aqueous Species:    H+, OH", Na+, K"1", NH,+, Ca  ,  Mg  , CI", NCL~, HCO-", SO4  ,
                    A13\ AI(OH)2+, AI(OH)2+, and Ai(OH)4"
Solid Phases:
                    AI(OH)3, Exch-AI, Exch-Ca, Exch-Mg, Exch-Na, and Exch-K
Mass Action Equations:

                    Kw

                    Kc

               KSD(gibb)
                K

                K
                    K,

                      AI
                     ex

                    K  c"1
                    Kex
                    K
                     ex
                    K
                     ex
                               =  {H+HHC03"}/(pC02)
                               =  {AI(OH)2+}{H+}/{AI3+}

                               =  {A1(OH)2+}{H*}2/{AI3+}

                               =  {AI(OH)4-}{H+}4/{AI3+}

                               =  {Ca2+}3[X-AI]2/{AI3+}2[X-Ca]3

                               =  {Ca2+}[X-Mg]/{Mg2+}[X-Ca]

                               =  {Ca2*}[E-Na]2/{Na+}2[X-Ca]

                               =  {Ca2""}[E-K]2/{K"h}2[X-Ca]
Charge Balance Equation:
H+  + Na+
                       NH4+  + AI(OH)2* + 2.0*(Ca
                                                 2+
                                                              AI(OH)2+) + 3.0* AI3+
    OH" + HCO3" + CT + NO3" + Ai(OH)4" + 2.0 * SO4:
                                                   2-
ANC or Alkalinity:

    ANC =  (OH")  + (HCO3")  + (Ai(OH)4" - (H+) - (Ai(OH)2+) - 2*(Ai(OH)2+) - 3*(A13+)

                    or

              fc "4- v   , /t/""^"Y
    ANC =  (Na*) + (K*) + (NH/) + 2*[(Ca^) +
                                                        - (CO - (N0al - 2*(S04=)
    Exch - A! = Exchangeable aluminum
    Exch - Ca = Exchangeable calcium
    Exch - Mg = Exchangeable magnesium
    Exch - Na = Exchangeable sodium
    Exch - K =  Exchangeable potassium
                                     515

-------
      Other modifications to  the  code were also incorporated.  The model was expanded to allow
inclusion of up to four horizons.  With this expansion, a provision was made to allow the user to route
soil water  from  any horizon  directly to  surface water.   This "routed  water"  is assumed to not be
equilibrated with  soils deeper in the pedon. Rather, it is "mixed" on a volume-weighted basis with waters
derived from other horizons plus all water draining from the bottom of the pedon.  New pH values and
aluminum  concentrations are then  computed  for  the  "surface water"  assuming  equilibration  with
atmospheric pCO2.

      Three options were incorporated into our version of the model regarding the treatment of input
nitrogen chemistry.  In the original versions of the model ammonium in deposition was ignored, and H+
effectively served as a surrogate for NH4+. This treatment was retained as one option in our code. The
second option (used in all model runs for this report) was based on the presumed reaction:

                                    NO3~ +  NH4+  =  org-N                       (Equation 9-6)

in which  the two nitrogen species  are accreted into the  organic nitrogen  pool  on  an  equivalent basis.
If nitrate concentrations exceed ammonium concentrations in deposition, then the  excess nitrate is passed
through the soil as a mobile anion.  Conversely, if ammonium concentrations exceed nitrate in deposition,
the excess ammonium is presumed to be replaced by H+.  The third  option is  based  on the reaction:

                         3 NOg" + 5 NH4+  = 4 N2(g) + 9 H2O + 2 H+            (Equation 9-7)

in which  nitrate and ammonium combine to form nitrogen gas, water,  and hydrogen ion. This process
originally was conceived to occur if the organic nitrogen pool attained  steady state.  As with the second
 option, excess nitrate is passed on to the soil as a mobile anion, or the excess ammonium is presumed
to be replaced by H+.

      These options are not sufficiently comprehensive for modelling of nitrogen species distributions or
concentrations In surface waters. As previously indicated, the purpose  of the Reuss model Is to examine
soil exchange phenomena with  regard to the effects of acidic deposition, and  not to provide detailed
information concerning the effects of nitrogen transformations in the soil environment.  The different
routines,  however,  provide the  user with some degree of flexibility in the treatment of nitrogen
transformations.

      Finally, the time series computations were converted from deposition volume-controlled increments
to time related steps, largely as a matter of convenience for dealing with units and to facilitate use of the
model by others.

9.3.2.1.3  Assumptions -

      As with any model, assumptions are necessary regarding certain processes, the soil environment,
and the characteristics of* certain reactions.  These assumptions and their justifications are outlined below,
along with  an  assessment of the effect they have on model  predictions.
                                              516

-------
9.3.2.1.3.1  Gibbsite solubility controlling soil AI3+ concentrations —

      One of the main features of Reuss's  model is that the solubility of a gibbsite-like [AI(OH)3 ] phase
controls the concentrations of dissolved aluminum in soil solutions.  Objections have been raised to this
assumption  on several grounds.  First, in many acidic forest soils, gibbsite is probably not present as a
separate or distinct  phase,  and thus cannot regulate  concentration of an aqueous species.  Second,
investigators have noted that aluminum activities, {AI3+}, in forest soil solutions do not behave according
to classic gibbsite solubility dynamics In response to changing  H+ activities (Johnson, 1986; Bloom et
al., 1979a,b).  Theoretically,  aluminum activities should decrease by three orders of magnitude for each
unit increase in pH,  i.e.,

                                   Iog10  {Al3*  }  =   C - 3 pH                      (Equation 9-8)

where C  is an arbitrary constant related  to the solubility product of  gibbsfte.  In  the studies cited,
however,  aluminum  activities appear to be independent of soil  pH, or vary in  ways different from  the
above relationship.   A third  concern focuses on the variability in aluminum solubility behavior observed
in natural materials.  In natural systems, gibbsite  is expected to display a range  of solubilities, based on
the grain size, crystallinity of the parent material,  and conditions under which it was formed.  This issue
is irrelevant in the  context  of the  Reuss  model, as apparent solubility products are computed on a
sample-by-sample basis using field  data to constrain the aluminum behavior.

      Despite these  concerns, the solubility of a  gibbsite-like phase is assumed in the present analyses
to control aluminum activities  in soil solutions.  Figure 9-23 shows  pH vs. Iog10 {AI3+} (both measured
in 0.002 M CaCI2) for all samples collected  during the DDRP HE  Soil Survey.  The solid line indicates the
theoretical solubility of gibbsite (C = 8.774; May et al., 1979). In computing the aluminum activities, only
the hydroxide complexes of aluminum were included in the speciation model.  Contributions from sulfate,
fluoride,  or  organic  ion pairs  or  complexes were not  incorporated into the  speciation model because
data on the  counter ion species were not available from the analytical solutions.  The contributions of the
sulfate, fluoride,  and  organic  complexes  to total dissolved aluminum  concentrations  increase with
increasing pH, so, effectively,  aluminum activities should  be increasingly overestimated at successively
higher pH values.

      For soil samples with pH values  greater than about 4.0,  i.e., all  B and  C horizon samples, and
about half of  the A/E horizon samples, gibbsite solubility appears to  provide a  reasonable model of
aluminum solubility.  Regression of the data with pH values greater than 4.0 yields a  slope of -1.4.  For
soils with pH  values between 4  and 5, predicted aluminum activities are generally within an order of
magnitude of  measured values.  Soils with higher pH values generally display high aluminum activities.
These results are attributed to the inability to incorporate organic  complexes of  aluminum into  the
speciation model.

      For soils with pH values less  than 4.0, i.e.,  all O  horizon soils and about half of the A/E horizons,
aluminum activities appear to be independent of soil pH.  This observation is interpreted as an indication
that the mass of rapidly exchangeable aluminum  available on soil exchange sites is limited.  Most of the
soil buffering is expected to be derived from mineral horizons.   Given the behavior of soils illustrated in
Figure 9-23, and  considering the limitations  of the aluminum  speciation model used to estimate the

                                               517

-------
     -3
    -5-
CO
 < -6-
 O)
 o
    -7-
    -8-
    -9
                  a  D
                                                                 o o
           Range of data regressed
                 4.0
-------
aluminum activities indicated, the evidence supports the model of a gibbsite-Iike solubility behavior to
describe aluminum availability in soils.

      The use of the limited aqueous speciation model  to compute aluminum activities  potentially
introduces one other problem.  If  aluminum  activities  are significantly overestimated, then selectivity
coefficients computed using the artificially high values should similarly be  too large.  (The value Is
calculated as:  Kexao =  {AI3+  }2  [X^ ]/{Ca2+  }3 [X^ ].  This effect, however, partially compensates
for an opposite effect, namely that induced by not considering  rational activity coefficients for the solid-
phase exchangers (see next subsection).

9.3.2.1.3.2 Constancy in  selectivity coefficients as functions of base saturation —

      As detailed in Section 9.3.2.1.2, the Reuss model describes exchange  reactions using Vanselow-
type mass action equations. The equations are developed and used  based on data  derived from the
soils. This approach is reasonable as long as the changes in the base saturation of the  soils under study
are limited.   Problems may be encountered, however, during time  dependent simulations  if significant
changes in base saturation are projected to occur.

      Problems may arise  because selectivity  coefficients are not true thermodynamlc constants.   As
presently formulated,  the selectivity coefficients do not incorporate rational activity coefficients  for the
solid-phase exchangers. Therefore, the selectivity coefficients are only approximately constant, and then
only for narrow ranges of base saturation around those levels for which they were calculated. As base
saturation declines, the selectivity coefficients would be expected to vary accordingly.

      As an example of this behavior, Figure 9-24 illustrates the relationship  between log10(Kexac ) (the
selectivity coefficient for the Ca/AI exchange reaction)  and base saturation for aggregated  A/E horizon
samples used In the watershed runs.  For base saturations  between 30 and 40 percent, selectivity
coefficients average slightly more than 100.  As base saturation decreases, selectivity coefficients increase
such that for samples with base saturations between  10 and  12 percent the constants have average
values of about 1000.  As indicated, this change is a direct result of not having incorporated the rational
activity coefficients into the  expression for the solid-phase  exchangers.

      Obviously, the apparent change in the selectivity coefficients as a function  of base saturation is of
concern in terms of the model  results.  To determine what the effects of varying selectivity coefficients
might be, a  modelling experiment  was undertaken in which the selectivity  coefficients for the Ca/AI
exchange reaction were both increased and decreased by an order of magnitude for each of the master
horizons in the 145 watersheds.  The model was run using these inputs, and the projected surface water
ANC was determined. Results for the present-day ANC values are summarized  in Figure 9-25.

      Changing  the selectivity  coefficients by an  order of magnitude introduces about a  10 /jeq L"1
change in the projected ANC for any particular system.  This change is small,  as the predicted total base
cation concentrations for most systems fall in the 100 to 200 #eq L"1 range.  Therefore,  errors introduced
by having selectivity coefficients that are off by as much as an order of magnitude are approximately 5
to 10 percent.   This change  is small  enough not to affect the long-term  projection of  depletion of
buffering capacity significantly in most systems.

                                               519

-------
        3-
   x
   03
        2-
                          o?
                                       a      na   a D a   Q

                                               a
          0
10
20            30

% Base Saturation
40
50
Figure 9-24.   Plot of the log of the  selectivity coefficient for the calcium-aluminum  exchange
reaction  vs. the measured base saturation in A/E  horizons in the  HE.  The Increase  in the
selectivity coefficients with decreasing  base saturation is a direct result of not incorporating rational
activity coefficients into the mass action expression used to estimate  selectivity coefficients.
                                            520

-------
          50
                              0               50              100
                                Predicted ANC (jieq L'1
150
Figure 9-25.  Histograms of the (unweighted for the population estimates) projected present-day
ANC values for lakes in  the NE.  The three curves were  generated  by varying the selectivity
coefficient for the calcium-aluminum exchange reaction by * an order of magnitude from the value
estimated from the soils data.  Varying the  selectivity coefficient by a factor of 10 changes the
projected ANC values for any system by about 10 j*eq L  for present-day conditions.  This is not
of sufficient magnitude to have a significant effect on the projected rates of depletion of buffering
capacity for the vast majority of these systems.
                                          521

-------
9.3.2.1.3.3  Soil gas pCO2 -

      The last major assumption made in running the Reuss model was the selection of an average
annual soil gas pCO2. For the model runs reported here, the partial pressure of CO2 for all horizons and
all soil classes was assumed to  be 0.005 atm on an average annual basis.  Soil pCO2 concentrations
vary  with a  number  of factors,  including temperature,  soil  productivity, and  moisture  content.
Unfortunately, there are not enough available data for the range of soils included in the DDRP to be able
to model values accurately. Data that are available (Solomon and Cerling, 1987; Fernandez and Kosian,
1987; Lam et al., 1988) indicate soil gas CO2 can exceed 0.5 percent, and In some  soils at some times
of the year, levels are less than 0.5 percent. A partial pressure of 0.005 atm was selected simply because
it appears to be a  representative value  for forested soils, based on available data.

      Reuss model outputs are  responsive to Increases in soil  gas CO2 concentrations.  Figure 9-26
illustrates present-day surface water ANC values predicted using CO2 concentrations of 0.001, 0.005, and
0.025 atm CO2.  As the  partial pressure of  CO2 increases in the  soil gas, the ANC of the associated
surface water also Increases.  It is possible, therefore, to adjust predicted present day surface water ANC
values up (or down) simply by adjusting the CO2.  In making these adjustments,  however, there is a
trade-off.  By increasing soil gas CO2 concentrations to increase the predicted ANC values, the  rate of
base  cation  leaching from the soil exchange complex is increased significantly.   Therefore, the  rate at
which soil buffering capacity becomes depleted is also increased.  This is illustrated in Table 9-10, where
changes predicted for  surface water ANC at 50 and  100 years are given for the  three soil gas CO2
scenarios.

      Finally, it should be clarified that no attempt was made, in selecting the 0.005 atm value for pCO2,
to use this parameter to  "calibrate" the  Level II models.  Admittedly,  had the  pCO2  been  adjusted on a
watershed-by-watershed basis, a much  better frt to observed  so(l pH values could have been obtained.
However, the purpose of this  part of the modelling exercise was to determine the magnitude of possible
responses to acidic deposition.  In the absence of more specific data on soil gas CO2 levels in individual
watersheds,  the approach taken  here yields  the least controversial, and most widely applicable,  results
possible.

9.3.2.1.4  Limitations -

      The Reuss model  focuses on soil exchange reactions.  The model does  not  consider other
probesses such as sulfate adsorption, mineral weathering, nitrogen transformations, or afforestation, even
though these processes may have equal  or greater influence  in regulating surface water composition in
certain ecological settings (Likens et al.,  1977; Johnson et al., 1988b).  The purpose of this part of the
study, however,  was specifically to examine exchange processes and their contribution in regulating
surface water composition and buffering against  changes caused by acidic deposition.

      Most of the  other  model limitations were  alluded to in Section 9.3.2.1.3.  The most significant
among these is that no  provisions are made to consider organo-cations, and especially organo-aluminum
interactions.   The data  are not available to include these interactions in our modelling efforts.
                                              522

-------
                                                          Pco-
                                                          Pco,
                                                          PCO,
= 0.001 atm
= 0.005 atm
= 0.025 atm
                            0               50              100
                               Predicted ANC Qieq  L
            150
Figure 9-26,   Histograms of the (unweighted for the population estimates) projected, present-day
ANC values for lakes in the NE.  The three curves were generated by varying the soil gas partial
pressure  of carbon dioxide by plus or minus a factor of 5 from the assumed value of 0.005
atmospheres.  Decreasing the partial pressure of CO2 reduces the projected ANC by about 10 /*eq
L~1 on the average, but does  not dramatically affect the projected rates of depletion of buffering
capacity of the systems in the NE. Increasing the partial pressure dramatically increases both the
projected mean ANC values for the lakes and the rates of cation depletion from the soil exchange
complex.
                                         523

-------
Table 9-10.  Effect of pCO2 on Changes Projected to Occur in Surface
Water ANC Values at 50 and 100 Years Using the Reuss Model. Deposition
Used in the Model is LTA.  Values Are  Given as the Mean, Population-
Weighted ANC Values for the NE (see Section 9.3.3.1 for details)
   Time Step                     ANC@         ANCฎ        ANC@
                                0.001 atm      0.005 atm     0.025 atm
Present day ANC                   -6.6           10.0           74.5

A-ANC @ 50 years                  -8.5          -13.7          -50.7

A-ANC @ 100 years                -21.3          -32.1          -97.0
                                    524

-------
      Another limitation of the Reuss model Is the implicit assumption that the reactions considered can
be modelled  on an equilibrium basis.  Clearly, most ongoing chemical processes in watersheds are
subjected to rapidly and constantly changing chemical environments.  Fluctuations in temperature, fluid
flow, and external cation  demands occur on  daily, weekly, and seasonal bases.  Few if any processes
actually attain chemical equilibrium. Nonetheless,  the Reuss model assumes exchange processes can
be modelled using an equilibrium approach. Given the relatively rapid nature of exchange reactions and
the annual time step used in most computations, this assumption is probably not unreasonable.

9.3.2.1.5  Model  inputs -

9.3.2.1.5.1  Deposition and associated data -

      The model requires deposition data including precipitation quantity (cm yr"1) and average annual
concentrations of the major ions in precipitation (SO42", CT, NO3", NH4+, Ca2+, Mg2+, Na+, and K+).
The  atmospheric flux  of each  ion  was  the combined wet plus dry average annual deposition.
Evapotranspiration (% ET) data are required to adjust the concentrations of the non-reactive tracers (e.g.,
CT) between  deposition and  runoff.  This parameter also helps define the ionic strength of  the soil
solutions, thereby influencing solution composition.

      As described in Section 5.6, a number of deposition scenarios are used for model simulations. The
LTA deposition is  used as the baseline against which other results are compared.  LTA data are the best
available estimates of total deposition occurring in each watershed.  Typical year (TY) deposition data
have also been compiled for these watersheds and  are used in model simulations.  Two  reduced dry
deposition scenarios have been examined as part  of these efforts. The first scenario, long-term annual
average-reduced  or LTA-rbc, assumes fluxes  of  base cations (Ca +, Mg  , Na+, and K+)  in dry
deposition to be half the  values used in LTA.  This scenario was  included because concerns had been
raised over the large particle deposition rates incorporated into the LTA baseline estimates.  The second
scenario, LTA-zbc, assumes zero flux for dry deposition base cations.  The LTA-zbc deposition dataset
yields maximum H+ deposition estimates for each watershed and has been included in the Reuss model
analyses to ascertain the magnitude of  error  potentially caused by uncertainty in dry  deposition data.

      In addition to  the above  data, which  assume constant depositional inputs to the watersheds over
the course of the  simulations, ramped datasets have been constructed, in these ramped datasets, total
(wet plus dry)  sulfate  and hydrogen ion  depositional  fluxes  are varied  during the course of each
simulation.  Changes are assumed to occur over a 15-year period, from year 10 in the simulation to year
25. The change is linear during this period, and the value attained in year  25 is maintained to the end
of the simulation.   In the NE Region, the ramp decreases sulfate depositional values to 70 percent of
current estimates; in the SBRP, the ramp increases sulfate fluxes  to 120 percent of current estimates.

9.3.2.1.5.2 Soils data -

      The model requires physical and chemical information about each of the horizons included  in the
simulations.  Required physical parameters are horizon thickness, bulk density, percent coarse fragments,
and a hydrologic runoff parameter.  Required  chemical parameters  include cation exchange capacity
(CEC), base cation concentrations on the exchange complex, selectivity coefficients for the Ca/AI, Ca/Mg,

                                             525

-------
Ca/Na, and Ca/K exchange reactions, soil gas pCO2, the apparent solubility product for AI(OH)3(s), and
the stoichiometric coefficient for H+ to be used in describing the dissolution of the aluminum solid phase.
Multiple-horizon versions of the model require the above information for each of the horizons to be
considered. Some minor adjustments were required to incorporate these parameters in the model.

      Soil bulk densities in the DDRP database  were entered on a coarse fragment-free basis.  As a
result, two adjustments to the associated field data were  necessary.  First, the percent coarse fragments
parameter,  was assigned a value of 0 in all cases. The contribution of coarse fragments is subtracted
from the  bulk density and soil (horizon) thickness, since these fragments are essentially unreactive mass.
Second,  to retain the proper reactive soil mass,  the horizon thicknesses were adjusted to remove the
contribution of the rock fragments.  The fragments not only add mass to a horizon, but also contribute
to the overall thickness. Had this correction had not been made, the  reactive masses of the individual
horizons  would have been larger than those actually measured.

      Another fixed parameter in the input datasets was the soil gas pCO2 concentrations. As discussed
in Section 9.3.2.1.3, a uniform value of 0.005 atm (0.5 percent) was used for all model  computations.
Finally, for this report,  the  stoichiometric coefficient for H+  used to describe the  dissolution of the
aluminum oxyhydroxide phase is assumed in all  cases to be 3.00.  Although the model can adjust this
parameter  (e.g., In  response to observed aluminum  behavior  in O  and A/E horizons;  see  Section
9.3.2.1.3), we retained the gibbsite-like solubility behavior because of data limitations regarding aluminum
behavior  in individual soil samples.

      Other data used as model input were taken directly from the DDRP soil chemistry database.  For
most of the simulations  discussed in the report, data were aggregated according to the procedures and
protocols presented in Section 9.3.1.2.2.  That is,  data from the six to ten pedons in each sampling class
were aggregated (Johnson et al., 1988b) to a master horizon level (O, A/E, B and C horizons). Extensive
parameters, such as horizon thickness,  were aggregated by simple  arithmetic averaging.  Intensive
parameters, such as soil pH or CEC, were aggregated using mass weighting procedures.

      in  addition to the  sampling class-based design, a study was undertaken to evaluate an alternative
aggregation procedure.  Results from some of the muitivariate analyses (see Section 8.3)  suggested a
significant watershed-specific component to observed variances.  Simulations were conducted using only
data collected on each individual watershed to model that watershed.  Because soils were  not sampled
on all watersheds, and  because difficulties were encountered with some of the analytical data, complete
coverage of the DDRP watersheds was not possible using the alternative aggregation scheme. Data were
collected on 129 of the 145 systems included in the study, however.  Results from this effort are  used
to determine if changing the aggregation scheme would significantly affect conclusions.

9.3.2.1.6 Model outputs -

      For each simulation, the model  generates two results files,  one containing projections for surface
water composition and the other describing soil and soil solution composition for major chemical species.
Results are compiled for the first and final years of the computation and at user-specified intervals during
the simulation.  For example, if the user were running a 13-year  simulation  and requested output  at 5-
year intervals, the result files would contain data  for years  1, 5, 10, and  13.

                                              526

-------
      Occasionally,  results are not available for all soils or watersheds for  the  duration requested
(typically,  100-year simulations). This occurs when the model fails to converge with a particular set of
input  parameters, which at most is about 7 percent of the simulations.  Initially, the failure-to-converge
rate was considerably higher than 7 percent.  However, by adjusting convergence criteria, the loss of
results minimized without sacrificing significant numerical accuracy.

      Information on many variables is retained in the two output files.  For surface waters, data on pH,
ANC,  SO42", NO3', CI", Ca2*, Mg2*, Na+, K+, AI3+, sum-(AI)aq, and ionic strength are captured.  For
soils,  information on soil pH, base saturation, and exchangeable Ca, Mg, Na and K are retained for the
solids, and ANC, Ca2"1", Mg2*,  Ha*, and K+ data are retained  for the soil solutions.  For this report,
analyses focus on a few select parameters,  namely surface water pH and ANC and soil  pH and base
saturations, because these parameters are believed to provide the most easily interpretable indicators of
system responses to continued exposure to  acidic deposition.

9,3.2.2  Bloom-Grigal Model

      In the DDRP,  surface water is  the principal  resource of interest. However,  soils play a vital role in
maintaining the quality of surface waters because drainage waters entering  lakes  and streams pa>ss
through soils.  Soils can buffer drainage waters against changes in several ways, as discussed in other
parts  of this report.  If soils in the study regions were to change dramatically (e.g., become more acidic),
these changes would ultimately be reflected in the subtending surface waters and  in the status and health
of forest vegetation. Characterizing the status of the soils in the DDRP regions and considering the effect
of chronic acidic deposition on them is, therefore, important.

      Two very different simulation  models have been  included  in the  DDRP to assess the impact of
acidic deposition on surface waters.  The Reuss model was discussed in  the  previous section.   This
section describes the Bloom-Grigal model.

9.3.2.2.1  Model description -

      The impact of acidic deposition on soils can be modelled following one of two approaches (Bache,
1983). The first approach (used in the Reuss model) is to view the Interaction of precipitation with soils
as a perturbation of the equilibrium  between ions in the soil solution and ions on the soil  ion exchange
complex.  Following the perturbation, the system returns to equilibrium according to the theories of ion
exchange equilibria. The second approach is to view this interaction as a simple mass action  (non-
equilibrium) exchange reaction.  Following this approach, the amount of acidity in deposition replaces an
equivalent amount  of base cations  in the soil.   The  Bloom-Grigal model is a form of this second
approach.

      The Bloom-Grigal model estimates the  loss of base cations on an annual basis using the following
equation:
                                          S =  i - A - C                             (Equation 9-9)
                                              527

-------
where  S is the sum of base cations, I is the amount of effective acidity in deposition, A is the acid
leached from the soil, and C is the correct ion factor for the decrease in acidity due to protonation of
bicarbonate.  The model is presented graphically In Figure 9-27.

     The Bloom-Grigal model is a simple semi-empirical computer simulation model created to project
the effects of acidic deposition on soils (Bloom and Grigal, 1985). The model tracks soil pH and base
saturation. Unlike the Reuss model, the Bloom-Grigal model is not formulated to project the chemistry
of subtending surface waters.  The model does, however, follow the  concentrations of aluminum in the
soil solution during simulation runs which  can serve as an indicator of possible changes in surface water
chemistry and forest health.  The Bloom-Grigal model was initially formulated to assess the effects of
acidic deposition on forested soils in  northeastern Minnesota.  Because  the model is based on  widely
applicable principles,  we believe that it  can  be meaningfully applied to project the effects of  acidic
deposition on the soils in the DDRP study regions.

9.3.2.2.2  Model formulation -

     The Bloom-Grigal model is formulated around the assumption that, in steady-state ecosystems,
acidic deposition depletes base cations on the soil ion exchange complex.  The model's simplicity lies
in the  fact that soils are treated as a single homogeneous unit  or compartment and  all  incoming
deposition reacts completely with the soil  in the compartment.  Soils,  however, are much more complex.
The Bloom-Grigal model seems to  be an  appropriate tool for assessing the impact of acidic deposition
on forested soils.

     The Bloom-Grigal model assumes that the acidity in deposition reacts completely with the soil. In
other words, the model  makes no provision for deposition to be routed around the soil and directly into
the surface water or into the subsoil strata.  The amount of exchangeable  base cations removed from
the soil compartment is calculated as the  difference between the input acidity and the output of H+ and
Al3"1", corrected for the protonation  of bicarbonate. The amount of base cations lost is subtracted from
the pool of exchangeable base  cations and  a new base  saturation is calculated.  The  Bloom-Grigal
model then calculates a new soil pH based on an equation  that relates soil pH to base saturation. After
adjusting parameters, the model then  simulates the next year of deposition (see  Figure 9-27).

     This model was created to assess  the effect of acidic deposition on  non-sulfate adsorbing soils.
Soils that  adsorb sulfate have lower base  cation removal rates than soils that do not.  In this regard, the
Bloom-Grigal model Is probably more appropriate for application to the soils in the NE than in the  SBRP.

     Another feature of the Bloom-Grigal  model is that it incorporates the  input of nitrogen in deposition.
Because forested soils are generally deficient in available  nitrogen,  inorganic nitrogen in  deposition is
removed by plants and organisms in the soil (Bloom and Grigal, 1985). When plants assimilate nitrogen
in the form of nitrate (NO3"), they release  hydroxyls (OH") to the soil, which is a non-acidifying reaction.
However, when plants assimilate nitrogen  as ammonium (NH4+), they release protons (H+).  Ammonium
uptake is  an acidifying reaction.  The  biological oxidation of NH4"1' to NO3" produces one  H+ for every
molecule of NH^ oxidized. The Bloom-Grigal model incorporates these processes in calculating the net
or effective acidity of deposition.
                                              528

-------
                                                                                  Preclp/runoff
                                                                                Soil chemistry
 wRecord watershed ID, year,|
*•%! soil pH, % base saturation
    	Wป/
                                                                                 Modelling
                                                                                  dataset
,,
Increment year


                                     Calculate protonation
                                        of bicarbonate
                                    Calculate loss of bases
                                    Calculate new % base
                                    saturation and soil pH
Figure 9-27. Flow diagram for the one-box Bloom-Grigal soil simulation model.
                                                  529

-------
     The original versions of the Bloom-Grigal model were coded in FORTRAN and BASIC; the version
used in this analysis is coded in a high speed, compilable form of BASIC.  In addition to optimizing the
code for speed, code has been added that allows the input data to be processed in a batch mode.  The
output  is now formatted to magnetic media to simplify the data reduction process.  Additional lines  of
code have been added to calculate the various deposition scenarios automatically as the model  is
running.  The fundamental equations in the original  model have not  been altered, however.

9.3.2.2.3  Assumptions ~

     A number of assumptions are made in modelling the effect of acidic deposition on soils with the
Bloom-Grigal  model.  Some are implicit to the model, others are made to meet the needs of our current
application.  The assumptions used in implementation of the Bloom-Grigal model are  itemized below
including additional explanatory discussion or comments.

9.3.2.2.3.1 Sulfate adsorption --

     The Bioom-Grigal  model assumes that sulfate is  not adsorbed by the soil and  is treated as a
completely mobile anion.   As mentioned previously, in  soils that have net  sulfate adsorption, this
assumption may lead to an overestimation of the amount of base cations actually leached from the soil.

9.3.2.2.3.2 Input acidity —

     The total effective  acidity (H+totai) in deposition is equal to:

                                  H+totai = H+ + NH4+ - N03'                     (Equation 9-10)

9.3.2.2.3.3 Extent of reaction -

     The effective acidity in deposition reacts completely with the soil.

9.3.2.2.3.4 Depth of soil -

     The depth of reactive soil  material equals  the  mean  aggregated thickness of the soil sampling
classes represented by the  types of soils on the specific watersheds.

     In their original paper, Bloom and Grigal (1985)  assumed that only the  top 25 cm of soil are
affected by acidic deposition. We consider the effect, however, on the whole soil compartment. Our soil
chemistry input data are aggregated to represent the central tendency of the soil chemical characteristics
of the  whole soil compartment.   The effect of acidic  inputs on data aggregated  in  this way, thus,
represents a mean effect.  At the same time, this assumption allows for the water that flows in  cracks
or root channels to lower soil horizons before reacting with the soil.
                                              530

-------
9.3.2.2.3.5  Volume of drainage water -

      The volume of water moving though the watersheds in each year of simulation is equal to the long-
term annual average runoff.

9.3.2.2.3.6  Partial pressure of CO2 -

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

9.3.2.2.3.7  Activity of A!3*  -

      To calculate the  amount of input acidity that is converted to output acidity by aluminum buffering,
the activity of AI3+  in soil solutions is calculated using the following equation:

                                 log(A!3+) = 2.60 -  1.66 * soil pH                   (Equation 9-11)

This equation is the empirical part of the Bloom-Grigal  model.  In developing their model, Bloom and
Grigal had a fundamental problem with using the solubility of AI(OH)3 to describe the variation in AI3+
with pH.   They state that in very acidic soils, such as forested soils, Al3+ is undersaturated with respect
to the precipitation of AI(OH)3. Therefore, Al(OH)3 solubility is a poor model for the pH-AI3+  relationship.
To establish a more realistic relationship between AI3+ and pH, they developed the above equation from
laboratory measurements of AI3+ in artificially acidified soils. Although not appropriate for all soils, Bloom
and  Grigal believe that model  results from which their equation was generated were reasonable for
selected forested soils  of northeastern Minnesota.

9.3.2.2.3.8  Relating soil solution pH to base saturation —

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

                                pH = pKa +  n  * log [BS/(1 -  BS)]                  (Equation 9-12)

where pKa is the apparent acidity constant for soil (i.e., aggregate watershed/soil compartment) and n
is an empirical constant. This equation is an extended form  of the Henderson-Hasselbach  equation.

      The Bloom-Grigal model used here calculates pKa and n for each watershed using the  input values
of soil pH and base saturation.  These parameters describe the  relationship between soil pH and base
saturation and are unique for each watershed.

9.3.2.2.3.9  Base  cation uptake -

      The model assumes no net accretion of base cations in biomass.  The uptake of base cations by
forest vegetation is an  acidifying process by which H* is exchanged for an equivalent amount  of base

                                               531

-------
cations to maintain charge neutrality.  At the same time, through litterfall and decomposition, base cations
are released to soils.  The Bloom-Grigal model only tracks the flux of base cations that are leached from
the soil.  This no-accretion assumption implies that the uptake of base cations by vegetation is exactly
equal to the amount recycled to the  soil.

9.3.2.2,3.10  Mineral weathering -

      Mineral weathering  is the ultimate source of base cations, and the Bloom-Grigal model has a
subroutine that calculates the contribution of base cations to the soil solution via mineral weathering.  The
rate of mineral weathering for these simulations, however, is set to zero for two reasons. First, assuming
no  base  cation resupply a "worst-case" base cation loss scenario is evaluated, thereby bounding the
projections.   Second, the relationships between weathering and soil  solution  pH  are  not sufficiently
established to provide accurate parameters for the weathering equations.  One complication, in particular,
is that mineral weathering rates are a. dynamic function of the chemical weathering environment.

9.3.2.2.3.11  Cation exchange capacity ~

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

9.3.2.2.3.12  Time steps  -

      The time step  for simulations is annual.  For assessment purposes, yearly time steps are a useful
increment.  From a modelling  standpoint any shorter  time step  (e.g.,  daily)  is  data  intensive  and
computationally demanding. Shorter time steps may provide more accurate projections,  however.

9.3.2.2.4  Limitations -

      Soils are highly complex and no simulation models exist that accurately depict the  flux of energy
and matter in soil systems. As with any attempt to project future events, the Bloom-Grigal  soil simulation
model is not without limitations.  Some of the limitations  are due to the state  of the  science and others
are have been imposed by the  DDRP.

      The scientific  limitations  center around the  factors  that  control aluminum solubility  and the
relationship of soil pH to base saturation.  Bloom and Grigal  (1985) empirically developed equations to
describe this relationship for a selected set of northeastern Minnesota forested soils. As described in their
paper, the equations appear appropriate for forested soils in Minnesota, in the DDRP, the equations are
assumed to be widely applicable and they are not independently verified.   It  is doubtful,  however,  that
these equations are  universally true due to vast differences  in soils and vegetation.
                                              532

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

      Individual soil processes are inextricably linked to a number of other processes and considering
a single  process (e.g.,  base cation  flux)  in  isolation  may distort projections.  In the DDRP Level  II
Analyses, processes are isolated in order to focus on the principal soil reactions associated with surface
water acidification.  It is recognized that  some  of the uncertainty in  assessing effects is due to this
approach of isolating facets of the whole ecosystem.

9.3.2.2.5  Model inputs -

      The Bloom-Grigal  model was designed not to be data intensive.  The data required to run the
Bloom-Grlgal model fall into four categories: (1) deposition data, (2) precipitation data, (3) soil chemistry
data,  and (4) fixed parameters.  The  deposition data are described in Sections 5.6 and 9.3.3.  Table 9-
11 lists the specific data requirements.

      The soil chemistry data used in these simulations has been aggregated to the single compartment,
watershed level.  These procedures  are described  in detail in  Johnson et al.  (1988b).  The capacity
variables, sum of base cations (SOBC) and CEC are capacity weighted. Soil pH is intensity weighted.

9.3.2.2.6  Model outputs -

      The Bloom-Grigal  model simulates soil processes relevant to the assessment of impacts of acidic
deposition on soils. During model simulation runs, soil pH, soil base cation status (i.e., base saturation),
and soil  solution AI3+ are tracked. Principal interest for this analysis is soil pH and base saturation.

      During 200-year simulations, soil pH and percent base saturation are recorded (see Figure 9-27)
at years 0, 20, 50, 100, and 200. The results are converted to change in soil pH and  change in percent
base saturation by subtracting the initial values from the projected values.  Because the initial values are
higher than the projected values, the reported results are all negative numbers, reflecting  a decrease.

      The projected changes in soil pH  and  percent  base saturation are  presented as cumulative
distribution functions (CDF)  for graphical comparisons.   The  CDFs represent  regionally  weighted
projections for soils on the target population  of watersheds. Summary statistics for the CDFs also are
presented for numerical  comparisons.

9.3.3  Model Forecasts

      Level II base cation analyses were conducted using Reuss's (Reuss, 1983;  Reuss and  Johnson,
1985) cation  exchange model and Bloom and Grigal's (1985) cation depletion model.  Results from the
individual models are presented  in this section along with a  comparison of the  projections made using
the two models.
                                              533

-------
Table 9-11.  List of Input Data for the Bloom-Grigal Soil Acidification Model
Input Variables
   Units
Annual average runoff

Annual H+, NH4+, NOa", and SO42" in wet deposition

Annual H4, NH4+, NOa", and SO42~ in dry deposition

Soil pH

Sum of soil base cations (0.1 M NH.pl)

Soil cation exchange capacity (0,1 M NH4CI)
cm
keq ha"1

keq ha"1
keq ha"1

keq ha"1
Fixed Parameters
   Value
Length of simulation


Partial pressure of CO2

Activity coefficient of AI3+

Activity coefficient of Al(OH)2*
NE = 100 years
SBRP = 200 years

0.005 atm

0.82

0.92
                                            534

-------
9.3.3.1  Reuss Model

9.3.3.1.1  Data sources -

      Summaries and examples of the various datasets used in running the Reuss model are presented
in Section 5.  A brief summary of the data used for the simulations also  is given below.

      The data fall into two categories: deposition data and soils data.  Four deposition datasets were
used in making population estimates of watershed responses.  As described in Section 9.3.3.1.1.1, these
datasets were used in model  simulation  runs assuming constant levels of deposition for the future and
in conjunction with a ramping function that adjusted deposition downward by 30 percent in the NE and
upward by 20 percent in the  SBRP (see Section 5.6).  Similarly, soils  data were aggregated using two
approaches.  The sampling class-based aggregation described in Section 9.3.1.2.2 was  used with each
of the deposition scenarios.  In this approach, soils data were aggregated to  master horizon/watershed
level.   The second approach (watershed-based aggregation)  was initially undertaken  because some
preliminary Level I Analyses indicated  a substantial "watershed effect". That is,  some combination of local
variables indicated that a soil from a given watershed was more similar to other soils in the watershed
than it was to other soils in the region from the same  sampling class.  While this preliminary observation
was not substantiated by additional investigations (see Section 8.8.1), the watershed-based aggregation
procedure was further examined to determine whether substantial  differences  in the results would be
observed.  Results from this examination are presented in Section 9.3.3.1.2.1.

      Given the number of deposition scenarios and soils aggregation approaches available, 16 distinct
sets of results could be generated for the NE. Because the purpose of examining the scenarios and the
aggregation schemes was to determine the sensitivity of model results to different conditions, discussions
are limited to nine  combinations of  deposition scenarios and soils aggregation schemes.  AH of the
constant and  ramped deposition scenarios are run in conjunction with the master horizon/watershed  soils
aggregation scheme. The two soils aggregations are run using the constant level, LTA deposition. Thus,
results obtained using LTA deposition and the master horizon/watershed soils  aggregation scheme serve
as the baseline dataset against which other results are compared.

9.3.3.1.1.1 Deposition  data  --

      Four deposition datasets were used.  The dataset considered to be most representative of "actual"
deposition Is  the LTA dataset, derived from 5-year averages of species concentrations in deposition and
30-year averages of precipitation quantities (see Section 5.6).

      Except for the TY dataset, which is based on data obtained from a year with mid-range depositional
values (see Section 5.6), other deposition datasets are variations of LTA.  In constructing LTA,  transport
and deposition of large  particles (> 20 pm)  were integral components of the dry deposition estimates.
The uncertainty in the long-range transport of these larger particles (concern that  net  H+  fluxes to
watersheds might be underestimated) prompted construction of two additional deposition datasets.  LTA-
rbc Is essentially identical to LTA, except that the estimated dry deposition of base cations (Ca2+, Mg2+,
Na+,  and K+)  is reduced  by 50 percent.  LTA-zbc assumes zero  net dry deposition of  base cations.
Dataset LTA-zbc, as a result, yields the highest hydrogen ion fluxes to watersheds, and, in fact,  probably

                                              535

-------
significantly overestimates net H+ fluxes.  In this context,  LTA-zbc can be viewed as a "worst-case"
deposition  scenario.

9.3.3.1.1.2  Soils data -

      Soils data were aggregated using one of two approaches.  The primary aggregation scheme uses
the soil sampling  class  concept around which  DDRP was designed  (see Section  5.5).   The other
aggregation was based on locale, and is described in more detail in this section.

      The aggregation scheme routinely used In the DDRP is the master horizon/watershed aggregation.
Data representing each of the four master horizons (O, A/E, B, and C) are obtained.  For each master
horizon, data are first averaged to within sampling class using protocols described in Johnson et at.
(1988b), which are then averaged using areal weighting to obtain estimates for a watershed.  Hydrologic
routing of water is considered if at least one of the sampling classes has a lower-most horizon that is
anything other than a C horizon.  For example, overland flow of water for the watershed is set equal to
the percentage of precipitation falling directly on rock outcrops and Is routed directly to the surface water
without  equilibration  with any of the soil horizons.  As another  example, for watersheds having soils in
sampling class  H01   (which  has only an O horizon), that fraction of soil  water equal to the areal
percentage cover  of the watershed by H01  is  routed to surface water after equilibrating with the O
horizon.   While this approach is oversimplified,  watershed hydrologic characteristics  are spatially
distributed, and  adequate representation of the complexity in natural systems cannot be accomplished
in the current formulation. The hydroiogic routing was established for these analyses in full cognizance
of its limitations.  Bedrock outcrops tend to occur along ridgelines, so incident precipitation will not run
off  directly into the surface water.   Histic soils, on the other hand, tend to be concentrated in riparian
zones,  Histic soils can have extremely low permeabilities,  and unless they are dry, Incident precipitation
will tend to run off from their surfaces. Nonetheless, the model equilibrates incident deposition with these
soils. The  model also does not consider any aspect of lateral flow, and therefore  downward  percolation
is likely to  be considerably, overestimated especially on steeper slopes.   Considering the various trade-
offs, we feel that the hydrologic routing, as described, yields a  reasonable approximation for modelling
these complex, spatially-related processes.

      Second, a watershed-based aggregation of soils data was undertaken in order to obtain  information
concerning the sensitivity of model results to the aggregation method.  For this approach, only data from
those soils sampled  on a particular watershed were used to describe the watershed.  Therefore, if the
only two soils sampled on a watershed were a Histosol and a Spodosol, the data from those two soils
were used to represent the watershed regardless of the actual areal coverage. The potential problem with
this aggregation is that,  for watersheds on which sample  classes are minor proportions of the total
watershed  area, the soils sampled  may not be  representative of the actual local environment.  As
described in Section  9.3.3.1, however, preliminary concerns had suggested that, even with this limitation,
the aggregation  might be more representative of the population of soils in each  of the regions than is
the sampling class-based information (see Sections 5.2 and 8.8).
                                               536

-------
9.3.3.1.2  Projections of surface water ANC -

9.3.3.1.2.1  Northeast -

  9.3.3.1.2.1.1  Prediction of current conditions —

      The distribution of current surface water ANC values projected for the NE using the Reuss model
is  illustrated in Figure 9-28, along with upper and lower bounds for 90 percent confidence intervals
associated with the projection.  The ANC values for  each of these lakes, as measured by the Eastern
Lake  Survey (Linthurst et al., 1986a), are listed in Table 5-3 for comparative purposes.  An  obvious
feature of these projections is the extremely tight  clustering of the results in the range of -25 to +50
/*eq L"1. This clustering has been observed on virtually ail model runs conducted to date, including those
runs using data aggregated at the watershed level and those conducted on individual sampling classes.
For the individual sampling classes, the upper limit for ANC values exceeds 200 peq L"1, while for the
other 37 classes in the NE an upper limit of 80 #eq L"1  is observed.

      These results are consistent with the hypothesis that soil  exchange reactions can buffer soil and
surface water ANC values and that the buffering occurs in the low ANC range.  Although surface waters
with  higher ANC values occur in the NE, they are not typical of the region.  Soil exchange reactions,
therefore,  do  not  adequately explain the observed  distribution of surface  water ANC.   Figure 9-29
illustrates  the relationship between observed and projected ANC values. Clearly, the tight clustering of
the predicted values near zero indicate no significant correlation.

      In order to explain the observed distribution of ANC values in the population of lakes sampled for
this study, it is necessary to invoke some mechanism other than base cation exchange to  produce ANC
values greater than 100  peq L"1.  Uptake of cations by aggrading vegetation is a possible mechanism,
but if cation uptake were a significant process in these watersheds, the observed ANC values would be
lower than those computed by the model.  The other major process that could explain the distribution
is  primary mineral weathering, which can significantly alter cation balance.  Release of base cations and
ANC through reactions such as those listed in Section 3.4 can increase surface water ANC to values well
above the 100 #eq L"1 limit apparently imposed by soil exchange processes.  Other processes that could
increase ANC to the levels observed in the lakes are presently unidentified.

      For lakes exhibiting  ANC values  exceeding 100  pey L"1, mineral  weathering apparently  is the
dominant watershed process controlling ANC. For systems with ANC values less than 100 t*eq L"1, either
mineral weathering or soil exchange processes could be regulating the observed levels. Given available
methods, however, determining which process accounts for the observed ANC values  is not  possible.

      The implications of these findings are significant in terms of projected future changes in surface
water chemistry.  If  mineral weathering is, in fact,  regulating ANC levels in  those systems with ANC
greater than 100 fj.eq L"1, then these systems probably will not experience significant future declines in
ANC at current levels of deposition.  Inasmuch as present trends  in the NE indicate stable or declining
hydrogen ion deposition, lakes with  ANC values exceeding 100 #eq L"  are  probably not at risk with
regard to future acidification.

                                              537

-------
                 1,0 r
                                   NE  Lakes
                             Deposition   =  LTA
                                   Year  =  1
                               Model  =  Reuss
                                               Upper Bound
                                               Projected
                                               Lower Bound
-25     0     25
     ANC
                                               50
75
100
Figure 9-28. Cumulative distribution of projected, present-day ANC values for takes in the study
population In the NE as projected using Reuss's cation exchange model. LTA deposition was used
in making these projections. The error bounds on the plot are the 90 percent confidence intervals
and were obtained using a Monte Carlo approach, assuming that errors on individual Input
parameters to the model are normally distributed, and that the only source of error is In those input
parameters.
                                       538

-------
         300
         200^
        -100
                     n  =  145
                     r2  =  0.03
           -100
0           100          200

   Measured ANC ([ieq L ~1
300
Figure 9-29.  Scatter plot of the projected, present-day ANC values for lakes in the NE, obtained
using the Reuss model vs. observed (ELS) values.  The heavy diagonal line indicates the 1:1, or
perfect correspondence, line.  As is apparent, the model projects that current ANC values should
cluster at values that are slightly in excess of 0 Ateq L'1.  This is interpreted as indicating the
importance of mineral weathering in controlling observed surface water composition for the majority
of systems in the NE.
                                         539

-------
      Soil exchange processes might regulate ANC in systems exhibiting ANC levels less than 100
L" .  If so some of these systems might currently be experiencing an increase in base cation leaching
rates  in response to  acid anion  inputs from acidic deposition.   In the future, these systems could
experience significant  ANC decline.  Unfortunately, given the current state of the science, distinguishing
between those  systems in which ANC is controlled  by  mineral  weathering and  those  in which ANC is
controlled by soil exchange processes is not currently possible.

      To provide an upper bound  on the number of  systems that may experience additional declines in
ANC,  summary  information from the Eastern Lake Survey can be examined (Linthurst et al., 1986a). Data
from this survey suggest that about 1,038 lakes  (about 15 percent of the total ELS target population in
the NE) have ANC values in the range of 0 to 50 /*eq  L"1.  The largest population of lakes that might
be adversely affected by changes  to  the soil  exchange  buffering  capacities  is In the Adirondacks
(Subregion 1A), where 321 lakes (25 percent of the target population) have ANC between 0 and 50
L*1.  The  Poconos/Catskiils Subregion (1B)  has the fewest lakes in  this ANC class:  116 lakes (7.8
percent).  As noted, the proportion of these systems that may actually experience future declines in ANC
cannot be determined. Some proportion  of the systems that currently have low ANC values, however,
will probably experience adverse changes.

      An issue of concern regarding these conclusions is the sensitivity of the results to the input data
used  in the simulations.  To address  this issue, several different versions  of input data were  used in
running the simulations:  four deposition scenarios and two soil aggregation schemes (see  Sections
9.3.2.1 and 9.3.3.1), Summary results  from these model runs for projected present-day ANC values  are
given in Table 9-12.  For the four deposition scenarios, the differences among projected ANC are minimal,
with projected population-weighted,  mean lake ANC values of 9 ฑ 1 ^eq  L"1; medians,  maxima, and
standard deviations are equally comparable.  The largest differences are  observed  for the projected
minima.  The LTA-zbc deposition scenario results  in an ANC value that is 10 ^eq  L"1  less than that
projected  using the LTA and 15 peq L"1 less than that projected using the TY.

      The greatest observed differences occur with the use of the different soil aggregation schemes.
For the data listed in Table 9-12, the columns  under LTA and WBA were obtained using the same
deposition data, but different soils aggregation schemes.  The data under the LTA column were obtained
using the  master horizon/watershed  aggregation  scheme, whereas those  under the WBA  (or  the
Watershed Based Aggregation approach) column were aggregated based on  soils  collected from
individual  watersheds  and used to describe only those watersheds.  The WBA data indicate moderate
changes in the means and medians for the present-day ANC values.  The extremes, however, represent
a much broader range of values than  are actually represented by the field data.  Figure 9-30 illustrates
the relationship between the observed and  projected ANC values obtained  using the WBA scheme.
Fifteen of  the 129 lakes in the sample have projected ANC values exceeding  100 /*eq L"1.  Despite the
wider range of projected  values, the WBA scheme does not improve the correlation between observed
and projected values.  This finding is  not surprising, since the soils sampled on any given watershed were
not selected to be representative of the soils on that watershed, but rather to  be representative of a group
of soils in the region (see Section 5.2.4.1). Therefore, although the WBA scheme may more accurately
portray the variability of individual soils in the  regions, it does not demonstrably provide a more accurate
means for explaining observed surface water composition.

                                              540

-------
Table 9-12.  Summary Statistics for the Population Estimates of
Current ANC Conditions lor Lakes in the NE Region for Five
Different Deposition of Soils Aggregation Schemes (Refer to the
text for explanation of the different input scenarios)
                 LTA      TY      LTA-rbc    LTA-zbc     WBA
Mean
Std Dev.
Median
P25
P75
Max
Min             -52.0     -46.8     -56.7       -61.3     -121.1
10.0
18.4
8.3
0.34
21.7
70.8
8.3
19.0
7.4
-1.8
20.6
67.1
9.4
18.9
7.8
-0.7
21.0
70.7
8.8
19.5
7.4
-0.2
21.0
70.6
35.5
87.4
18.9
0.3
43.3
863.7
                              541

-------
         300
         200-
      CD
     O

     <
     .9
      CO
     O
         100-
        -100
                     n = 129
                     r2 = 0.01
           -100
0           100         200         300
   Measured ANC (jxeq L ~1)
Figure 9-30.  Scatter plot of the present-day lake ANC values projected using the Reuss model
In conjunction with the Watershed-Based Aggregation (WBA) soils data vs. observed (ELS) ANC
values. The range of ANC values projected using this approach is much greater than obtained
using the sampling class/watershed-based approach.  However, the correlation is not improved.
                                                      -i
Three projected points with ANC values in excess of 300 /*eq L  are not shown on this plot.
                                        542

-------
  9.3.3.1.2.1.2  Projected future conditions -

      In order to project the magnitude of changes in ANC that might occur in the NE, as well as the
time frame over which such changes might occur, the Reuss model was run using its mass balance
component. The mass balance component of the model tracks the loss (or gain) of base cations from
soil exchange  sites through time.  For these simulations, precipitation quantity (cm  yr"1) as well as the
depositional fluxes were  used  to specify the total loadings of ions delivered to  the soil.  Annual time
steps were used in making these computations.  For the NE,  model simulations  were run for a total of
100 years, with results of the computations being collected at 10-year intervals. Results are reported only
at the 20-, 50-, and 100-year time increments.

      Projected,  time-dependent changes in ANC values for the population-weighted results are illustrated
in Figures 9-31 and 9-32 and summary statistics are  given in  Table 9-13. The Reuss model  considers
only the effects  of the soil cation exchange process In making these projections.  Mineral weathering
reactions would, in  general,  further delay the  response  of  these systems to  the effects  of  acidic
deposition.  At 20  and 50 years, most systems in the NE are projected to experience  minimal change
in ANC. Apparently, the  soil buffering capacity in these systems is sufficient to moderate the effects of
acidic deposition over these time scales.  Only a small percentage  of the watersheds (about 10 percent)
is projected to experience losses of ANC that exceed about 25 neq L."1  within the 50-year time frame.

      The 100-year projections for changes in ANC (Figure 9-32) suggest a bimodal distribution in the
way watersheds  respond to the effects of acidic deposition. About half  of the watersheds in the region
are projected to experience minimal changes (<-13 ^eq L  ) over the 100-year time frame.  The other
half is projected to experience a median  change in ANC of about  -50 ^eq L"1 and a maximum change
of almost -200 fj,eq L"1.  The magnitude of these  changes is of concern, if mineral weathering reactions
do not control ANC.   A closer examination of the results (Table 9-13) suggests that projected changes
in the ANC values through time are not linear,  but rather accelerate to a point where  the  buffering
capacity of  soils is depleted.  Soils response to acidic deposition is analogous  to a buffer  effectively
being titrated by acidic deposition.  As such, any given soil behaves in the same way a dissolved buffer
in an aqueous system behaves (Figure 9-33).  Assuming that the  system is not  yet near to or beyond
the inflection point of the titration curve, the initial response  of a soil to continued loadings  of acidic
deposition will be a gradual, and almost linear, decline in projected ANC for some period of time.  Once
the system reaches the inflection point, however, the rate of decline in ANC dramatically accelerates until
the buffering capacity of  the system is depleted.

      For  the  soils examined to date, these observations have two major implications.  First, minimal
changes observed  in lake water ANC values do not necessarily preclude the possibility that more dramatic
changes will occur in the future.  If the buffering capacity  of a soil is currently being depleted, the full
effect might not  be immediately apparent. Rates of change in system response can increase with time,
unless the process is being moderated  by mineral weathering.  Second, for the soils included in DDRP,
dramatic changes in system response to acidic deposition are projected only for those systems with lower
ANC values.  Most of the titration curves deviate from  relatively  flat slopes to  steeper slopes as the
inflection points approach ANC in the range of -20 to +20 peq L"1.  Therefore, the systems that are most
                                              543

-------
              c
              o
o
Q_
O
i_
Q_

CD
                  1.0
    0.8
                 0.6
              *3 0.4
              JT3
              3
              E

              O ฐ-2
                 0.0
                   -75
                                  NE  Lakes
                             Deposition  =  LTA
                                 Year  =  50
                               Model  = Reuss
                 Upper Bound
                 Projected
                 Lower Bound
                  -50          -25
               A  ANC  (jieq  L.-1)
Fiaure 9-31. Cumulative distribution of the projected surface water ANC values projected for the
study population of lakes in SO years In the NE.  The model runs were conducted using LTA
(constant level) deposition.
                                    544

-------
O
O
CL
O
i_
CL

Q)
T*z>
JS
I
3
O
                 1.0
                 0.8
                 0.6
                 0.4
                 0.0
                   -75
                                  NE  Lakes
                             Deposition  =  LTA
                                Year  =  100
                              Model  =  Reuss
                               Upper  Bound
                               Projected
                               Lower  Bound
                  -50          -25
               A ANC (jxeq L"0
Figure 9-32. Cumulative distribution of the projected surface water ANC values projected for the
study population of lakes in 100 years in the NE. The model runs were conducted using LTA
(constant level) deposition.
                                    545

-------
Table 9-13.  Descriptive Statistics of the Population Estimates for
Changes in Lake Water ANC for Systems in the NE.  Mean, Median,
Standard Deviations for the Population and the Maximum Changes
Projected Are Presented for Each of the Four Deposition Scenarios
at the Time Increments 20, 10, and 100 Years
                           LTA     TY      LTA-rbc    LTA-zbc
ANC (0) (Mean)             10.0      8.3         9.4        8.8

  ANC (20)
  Mean                    -6.1     -6.4        -6.5       -6.9
  Std                      16.4     18.0        17.4       18.3
  Median                   -2.0     -2.4        -2.3       -2.5
  Max                    -101.9   -118.3      -107.1     -110.0

  ANC (50)
  Mean                    -13.7    -16.1       -15.5      -17.5
  Std                      23.6     26.4        26.4       30.0
  Median                   -5.2     -6.0        -6.0       -6.4
  Max                    -127.5   -138.8      -140.0     -160.0

  ANC (100)
  Mean                    -32.1    -43.1       -39.4      -44.7
  Std                      36.1     51.5        43.8       49.6
  Median                   -13.9    -22.0       -16.4      -20.6
  Max                    -185.4   -231.7      -207.5     -228.7
                               546

-------
       20
   cr-20-
   0)
   O
      -40.
   T3
   J)

   .H -60.
   T3
   Du
      -80.
     -100
          n  a
                        o   n
                        O
                        CM
o
^t-
O
00
O
O
                                        Time (Years)
Figure 9-33. Schematic illustration of the titration-like behavior displayed by soils in response to
constant loadings of acidic deposition.  Initially, soils respond slowly, showing only minor changes
in ANC  as the base status of the soil is reduced.  However,  once the  base cations have been
sufficiently depleted from the exchanger, rapid and dramatic changes in ANC values from the soils
can take place.  This example was computed using soil sampling class S14 and a  mid-range
deposition.
                                           547

-------
vulnerable to dramatic future changes in ANC are those that currently have an associated surface water
ANC of about 0 neq L ,

      The alternative soils aggregation scheme, WBA, yields results that are qualitatively similar to those
obtained using the master horizon/watershed aggregation (Table 9-14).  Quantitatively,  the changes
projected  using the WBA scheme are two to three times as large as those  projected with the routine
aggregation method.  Also,  the WBA  scheme projects substantial  changes in a  small number  of
watersheds during the early phases of the simutations. These results substantially shift the mean values
of the changes to more negative numbers.  Because there is a lower limit to values that ANC  can attain
within the framework of this model, the magnitude of changes that can occur in the population  means
is limited.

      The last group of simulations addresses the effects that ramped deposition has on projected future
changes.  As discussed in Section 9.3.3.1.1.1, the three LTA deposition datasets, as well as the TY data,
were modified using a ramp function  that decreased sulfate  and hydrogen ion depositional fluxes by 30
percent in the NE between years 10 and 25 of the simulations. Differences between the projections made
using ramped and constant deposition are presented in Table 9-15. Not surprisingly, differences between
the two scenarios  are minor at the 20-year point.  By year  50, the median declines  projected for ANC
using ramped deposition  are only half as large as  those projected using constant deposition.  After 100
years, the differences in the medians are less.  Ramped deposition results in changes in surface water
ANC that are two-thirds the  magnitude of those for constant deposition.  Differences in the means are
more uniform for both year 50 and year 100. At both years, ramped deposition results in  changes that
are about 60 percent as  large as those obtained using constant deposition,

      Incorporation of mineral weathering effects  into these results would suggest smaller differences
between the constant and ramped  depositions than those reported here.   A supply of  cations from
weathering would tend to minimize the changes projected by both datasets,  but such effects would be
larger for the constant deposition scenario than for the ramped deposition scenario.

9.3.3.1.2.2  Southern Blue Ridge Province ~

  9.3.3.1.2.2.1  Prediction of current conditions

      The distribution of current surface water ANC values projected using the Reuss model for the SBRP
Is illustrated in Figure 9-34. These values can be compared to the actual distribution of ANC measured
for these stream reaches  during the Pilot Stream Survey (Messer et al., 1986a) (see Table 5-6).  As with
the northeastern results,  the  extremely tight clustering of the results around an ANC value  of zero is
notable.  Mean and median values for each of the four deposition scenarios (Table  9-16) are between
2 and 4 /*eq L"1 , and the total range for the four scenarios is about -15 to 23 neq L'1.
      As for the northeastern data, these results are interpreted as an indication that the soils of the
region are characterized by strong buffering.  Additionally, the results suggest a dominant role for mineral
weathering in  regulating the observed surface water composition,  since neither sulfate adsorption nor
cation accretion into biomass can readily explain the differences between observed and projected  ANC
                                              548

-------
Table 9-14.  Summary Statistics Comparing the
Projections Regarding Changes in Surface Water
ANC Values Obtained Using Different Soils
Aggregation Schemes
                        LTAa            WBAb
ANC (0)               10.0              35,5
A-ANC (20)
   Mean                -6.1             -25.1
   Std                -16.4              42.0
   Median              -2.0              -5.2
   Max              -101.9            -216.4

A-ANC (50)
   Mean              -13.7             -43.9
   Std                 23.6              55.1
   Median              -5.2             -14.3
   Max              -127.5            -241.4

A-ANC (100)
   Mean              -32.1             -66.9
   Std                 36.1              67.6
   Median            -13.9             -36.5
   Max              -185.4            -275.6
  The LTA data have been obtained using a sampling-
  class-based aggregation, in which soils from the
  whole region are used to describe specific soils
  on the watershed (see Section 5.5.1).

  The WBA is based on data obtained from only those
  soils sampled on the watersheds being described.
  The text contains details of the procedures used.
                                         549

-------
Table  9-15.  Summary Statistics of the Differences Between the
Population Estimates  for Future ANC Projections Made Using the
Constant Level and Ramped Deposition Scenarios*
                  LTA            TY        LTA-rbc       LJA-zbc
ANC (0)
   Mean           0.0            0.0            0,0            0.0

A-ANC  (20)
   Mean           2.1            2.B            2.1            2.2
   Std.  Dev.        1.0            1.6            1.3            1.3
   Median          1.7            1.9            1.6            1.7
   Max             9.9           10.7           10.6           12.1

A-ANC  (50)
   Mean           5.7            7.0            5.9            6.3
   Std.  Dev.        4.8            6.6            5.5            6.4
   Median          3,0            3.8            3.1            2.9
   Max           25.8           27.5           29.2           33.2

A-ANC  (100)
   Mean         11.4           15.6           12.7           12.2
   Std.  Dev,      10.8           14.2           11,3           11.5
   Median          4.7           11,3            5.4            4.5
   Max           79.9           56.7           57.4           54.8
   The values were computed as the difference between ramped and constant
   deposition. The magnitude of the values can be compared to the descriptive
   statistics presented in Table 9-13 to obtain estimates of the absolute values of the
   changes incurred with the rarnped datasets.  Standard deviations are presented as
   absolute values.
                                 550

-------
                          SBRP  Stream  Reaches
                             Deposition  =  LTA
                                   Year  =   1
                               Model  =  Reuss
             0
                 to r
0.8-
O
Q.
o 06
ol
CD
*= 0.4
"5
E
~* no
o u-^




On
.u
_c





-


;
_•*
•
.*
.* j
/ /'
t #T_ . . !

iO -25 0
AN
[Ij



;



	 upper tsouna



	 Lower Douna
i i i i

25 50 75 100
1C (jieq L-1)
Figure 9-34.  Cumulative distribution of projected present-day ANC values for stream reaches In
the study population in the SBRP, as projections using Reuss's cation exchange model.  Long-
term average (LTA) deposition was used in making these projections.  The error bounds on the
plot are the 90 percent confidence intervals and were obtained using the parameter error estimates
developed for northeastern region soils.  Then, as completed in the NE, a Monte Carlo approach
was used to obtain population estimates of the errors.
                                       551

-------
Table 0-16. Summary Statistics for the Population Estimates of Current
ANC Conditions for Stream Reaches In the SBRP for Four Different
Deposition Scenarios (Refer to the text for explanation of the different
input scenarios)
                       LTA           TY         LTA-rbe      LTA-zbc
Mean                  3,9            2.2           3.7          3.4
Std. Dev.              5.8            6.1           6.0          6.2
Median                2.9            2.3           2.9          2.9
  P25                  -0.55          -1.5          -0.55         -0.55
  P75                  7.0            4.7           6.7          6.5
Max                  21.2           23.0          21.2         20.8
Min                   -12.8          -14.1         -15.3         -17.7
                               552

-------
     100
      75.
      50.
   cr
   o
      25.
   1  0
   o
DO  a        f,  _     a


  ฐฐ  "•"J  •  ฐ  •   "
      -25.
      -50
         o
         ID
         o
         o
o
o
CJ
 o
 in
 CM


-1
O
o
co
o
w
CO
o
o
                                Measured ANC (jj,eq L  )
Figure 9-35. Scatter plot of the projected present-day ANC values for stream reaches in the SBRP,

obtained using the  Reuss model, vs. observed (NSS)  values.  The heavy diagonal line indicates

the 1:1, or perfect correspondence, line.  As is apparent, the model projects that current ANC

values should cluster at  values that are  slightly in excess of 0 /zeq L .  This is  interpreted as

indicating the importance of mineral weathering in controlling observed surface water compositions

for the majority of systems in this region.
                                           553

-------
values (Figure  9-35).   Mineral weathering  also  might explain  why  the  observed ANC values are
considerably higher than the model results.

  9.3,3.1.2.2.2  Projected future conditions

      As described for the NE, simulations of the time-dependent responses of the ANC in the study
population stream  reaches In the SBRP were conducted. Annual time steps were employed for these
runs,  and results were collected at 10-year intervals; data are summarized here for the 20-, 50-,  100-,
and 200-year increments only.

      Changes  in the projected surface water ANC  values are summarized in Table 9-17.  During the
first 50 years of these simulations, the Reuss model results suggest that changes,  even the maximum
changes, are trivial relative to our ability to measure  representative ANC values.  As base cation supply
becomes depleted, these changes become much more dramatic, but this depletion is projected to occur
on a century-long time scale.  Mean and median changes for this region are  estimated to be -20 ฑ 5
L"1  on a 100-year time  scale.  Over two hundred years, these changes Increase by a factor of 5 to
approximately -100 ฑ 20 ^eq L"1.  These changes are  projected to occur regardless of the selected
deposition scenario.  These results are illustrated in Figures 9-36 and 9-37 for the LTA deposition.

      Watersheds  in the SBRP  are projected to respond relatively uniformly to the different deposition
scenarios, unlike the NE, for which a range  of  responses to acidic loadings was displayed.  This
observation can be explained by several factors.  First, the watersheds in the SBRP were selected from
a geographically more limited area than those in the NE.  Second, the number of stream reaches studied
in the SBRP  is considerably smaller than the lake  study population in the  NE.   This smaller subset of
systems will limit the observed variability simply because of the reduced sample size being examined.

      In examining the changes projected for surface waters in  the SBRP,  it is important to remember
that the Reuss model Is a cation exchange  model, and it does not consider the  effects of increasing
anion mobility.  At present, the soils in the SBRP  are retaining significant  percentages of sulfur being
deposited in  the region  (see  Sections 7 and 9.2).  As a result,  rates of base cation leaching from the
soil exchange pool are probably less than those presented  above because the total anion concentration
in soil solutions are lower than considered in the model.  The rates of leaching will  increase as the soils
approach zero net retention  of sulfur and will approach the projected levels asymptotically.  Therefore,
the magnitude of observed changes should be  some non-linear combination of the time frames involved
in base cation leaching  and changes in sulfur retention.

      Mineral weathering would even further delay any anticipated changes in observed surface water
ANC values.  As weathering proceeds, additional  cations are  provided both to the exchange complex
and to surface waters.   As in the NE,  it Is not  possible with the data and models currently available to
isolate the separate  effects of weathering and cation  exchange.  In  a qualitative  sense, however, we
conclude base cation-related changes  in surface water ANC in the SBRP should occur only on century-
long time scales once the effects of weathering are incorporated into the projections.

      The last  major issue concerns  the effects  of ramped  deposition  datasets  on the response of
watersheds to acidic deposition. As discussed in Section 5.6, the ramping functions  increased deposition

                                              554

-------
Table 9-17.  Descriptive Statistics of the Population Estimates for
Changes in Stream Reach ANC Values for Systems in the SBRP. Mean,
Median, and Standard Deviations for the Population and the Maximum
Changes Projected Are Presented for Each of the Four Deposition
Scenarios at the Time Increments 20, 50, 100,  and 200 Years
                     LTA
TY
LTA-rbc
LTA-zbc
ANC(0) (Mean)
A-ANC (20)
Mean
Std. Dev.a
Median
Max.
A-ANC (50)
Mean
Std. Dev.
Median
Max.
A-ANC (100)
Mean
Std. Dev.
Median
Max.
A-ANC (200)
Mean
Std. Dev.
Median
Max.
3.9

-1.2
0.4
-1.2
-1.8

-3.0
1.1
-2.9
-5.2

-14.6
6.0
-14.8
-27.4

-81.2
24.5
-77.8
-134.8
2.2

-1.2
0.7
-2.2
-2.2

-3.7
2.0
-4.0
-7.4

-23.0
16.2
-18.5
-58.8

-97.8
36.2
-103.3
-161.7
3.7

-1.2
0.5
-1.9
-1.9

-3.5
1.5
-3.7
-5.9

-20.3
12.8
-18.3
-48.0

-97.3
33.1
-103.9
-154.9
3.4

-1.2
0.4
-2.0
-2.0

-5.3
2.8
-4.8
-10.9

-33.7
23.6
-24.4
-80.4

-120.1
39.1
-122.7
-185.3
  Standard deviations are reported as absolute values.
                              555

-------
                  1.0
              O  0.8
              o
              CL
              O
              l_
              Q.
0.6
              *= 0.4
              E
              O
                 0.0 <—
                   -75
                          SBRP  Stream  Reaches
                             Deposition  =  LTA
                                  Year  =  50
                               Model =  Reuss
                               Upper Bound
                               Projected
                               Lower Bound
               -50          -25
           A  ANC  (|ieq  L
Figure 9-36.   Cumulative distribution of projected changes (at 50 years) in surface water ANC
obtained using the Reuss model for stream reaches in the SBRP. The deposition scenario usec
in making these projections was LTA. Confidence intervals around the distribution are based or
uncertainty estimates of the individual parameters used in the model.
                                    556

-------
                  1.0
              O 0.8
              O
              CL
              O
                 0.6
              Q_

              0)

              •4=  0.4

              "5
              E

              O  ฐ-2
                 0.0
                          SBRP  Stream  Reaches
                             Deposition  =  LTA
                                 Year  =  100
                               Model  =  Reuss
                   -75
                                Upper  Bound
                                Projected
                                Lower  Bound
   -50           -25
A ANC  (|ieq  l_-
Figure 9-37.   Cumulative distribution of projected changes (at 100 years) in surface water ANC
obtained using the Reuss model for stream reaches in the SBRP. The deposition scenario used
in making these projections was LTA.  Confidence intervals around the distribution are based on
uncertainty estimates of the individual parameters used in the model. The "choppiness"  of the
curve is due, in  part, to the smaller number (n=20) of watersheds for which 100-year prelections
were obtained.
                                      557

-------
by 20 percent during the 10- to 25-year time interval of each simulation. This ramping function was used
in conjunction with each of the  deposition datasets.

      Differences in the projections of surface water ANC between the ramped and constant scenarios
are given in Table 9-18.  Not surprisingly, projected differences are minor during the first 50 years of the
simulations, although the increased levels of deposition in the ramped dataset nearly double the median
projected changes at 50 years  (from -2.9 ^teq L"1  to -5.0 fjteq L"1).  At 100 years, the projections using
the ramped deposition are double those for  constant levels of deposition.  Median changes between
the deposition scenarios are not as large, but the ramped scenario projections result in changes that are
50 percent larger than those for the constant deposition.  At 200 years,  the medians of the population
projections for the ramped and constant deposition scenarios continue to diverge.   However,  the
differences In the population means have not changed substantially from those observed  at 100 years,
suggesting that the limiting values proscribed by the composition of the  deposition are  being attained.

9.3.3.1.2.3 Comparison of results from the Northeast and Southern  Blue Ridge Province ~

      Comparison of the  effects projected  by the Reuss model in  the  two regions indicates both
similarities and differences between the two regions.  In both regions the soils behave initially as strong
buffers for surface water ANC. Also, in both regions the projected present-day ANC values are generally
substantially less than the actual observed values.  These observations are interpreted to  indicate the key
role that mineral weathering plays in regulating ANC in surface waters of the two regions.

      The soils in the two regions, however, are projected to respond differently to continued exposure
to acidic deposition.  At  present levels of deposition, soils in the NE appear to be more susceptible to
significant changes in the future than are the soils in the SBRP.  In a  sense, this conclusion is counter-
intuitive because the soils In the NE tend to exhibit higher levels of base  saturation  (see  Section
9.3.3.1.3.1.3). The soils in the NE, however, are also younger than those in the SBRP, and as a result,
tend to  have less clay-size materials. Because the bulk of the exchange  capacity is associated with fine
particles (see  Section 8.8.1) and because the soils in the  NE tend to be shallower than those in the
SBRP, soils in the NE apparently have a lower overall capacity to supply base cations to surface waters
from exchange processes.

9.3.3.1.2.4 Summary -

      Several conclusions can  be drawn from the observations made using the Reuss model and the
projected behavior of watersheds in both the NE and  SBRP.

      •    For  lakes !n the NE currently exhibiting  ANC values in excess of 100 /neq L"1, mineral
           weathering is probably the dominant watershed process controlling observed  ANC values.

      ซ    At present levels of deposition, NE  lakes with ANC values in excess of 100 /^eq  L"1 will
           probably not experience declining ANCs in the  foreseeable future.
                                              558

-------
Table 9-18.  Summary Statistics of the Differences Between the
Population Estimates  for Future ANC Projections Made Using the
Constant Level and Ramped Deposition  Scenarios for Stream Reaches
in the SBRP"
                        LTA
TY
LTA-rbc
LTA-zbc
ANC (0)
Mean
A-ANC (20)
Mean
Std. Dev.a
Median
Maximum
A-ANC (50)
Mean
Std. Dev.
Median
Maximum
A-ANC (100)
Mean
Std. Dev.
Median
Maximum
A-ANC (200)
Mean
Std. Dev.
Median
Maximum

3.9

-0,9
0.1
-1.0
-1.1

-1.5
0.6
-2.1
-6.9

-8.7
7.1
-6.3
-41.1

-24.3
11.5
-29.1
-95.9

2.2

-1.0
0.0
-0.9
-1.1

-3.3
1.7
-3.2
-8.7

-20.6
17.3
-11.9
-94.5

-25.9
3.1
-19.6
-64.9

3.7

-1.0
0.0
-1.0
-0.9

-2.9
1.4
-2.2
-18.1

-15.6
8.9
-11.7
-35.9

-26.8
3.2
-20.7
-90.2

3.4

-1.0
0.0
-1.0
-0.9

-3.6
2.5
-2.6
-30.8

-25.8
12.0
-31.5
-53.3

-23.3
0.7
-18.3
-74.0
   The values were computed as the difference between ramped and constant deposition.
   The magnitude of the values can be compared to the descriptive statistics presented
   in Table 9-17 to obtain estimates of the absolute values of the changes incurred with
   the ramped datasets. Standard deviations are reported as absolute values.
                                 559

-------
      •    For lakes in the NE currently exhibiting ANC values of less than 100 jteq L"1, soil exchange
           processes may be regulating the observed ANCs, although in most systems, the observed
           levels are probably controlled by a combination of cation exchange and mineral weathering.

      *    As an upper limit, over 1000 additional lakes in the NE region could become acidic (i.e.,
           ANC < 0 #eq  L"1) within a 50- to  100-year time frame. This is four times the number of
           lakes that are  currently acidic.  This number is  considered to  be extreme  because the
           contribution of weathering is not included in  these projections. However, some lakes are
           expected to become acidic during the next several decades.

      *    In the SBRP, changes in observed  ANC values due to changes in the base status of soils
           during the next century should be minimal. Observed changes in this region will be driven
           primarily by changes in anion mobility in these soils (see Sections 7.3.4 and 9.2.3.2.3).

9.3.3.1.3  Projections of soil pH and percent base saturation -

      Another concern regarding the effects  of acidic deposition is  the  changes in  soil  pH and  base
saturation status.   As discussed in  Section 9.3.1.1, soils can be used as indicators  of potential future
changes. As with  the ANC  results, these model results are presented on a regional basis.

9.3.3.1.3.1 Northeast -

      Unlike the ANC projections, for which the correspondence between observed and predicted values
was only a secondary concern, the Reuss model should be able to predict observed  soil pH values with
a reasonable degree of accuracy. (Present day base saturation is an input to the model and, as such,
cannot be used in this type of an analysis.) Figure 9-38 illustrates the correlation between the observed
and predicted soil pH values for all of the master horizon/watershed combinations considered in the NE.
Two features are immediately apparent from this plot.

      First, there Is a high correlation between the observed and predicted values. In general, the model
tends to over-predict individual observations. For measured pH values greater than about 4.0, the model
results exceed measured values by 0.20 ฑ 0.10 pH units. The divergence between  the two increases
substantially  at pH  values below  4.0. Therefore, the Reuss model reasonably predicts the relative
differences in soil pH among soils (for pH values exceeding 4.0).

      Second, the model predicts very few  soil pH values  of less than 4.0, and, in fact, the data appear
to reach a plateau at soil pH values of about 4.25 ฑ 0.25.  Effectively, the lower limit to soil-water acidity
is defined by  hydrogen ion content of deposition after it has undergone evapotranspirative concentration.
This lower limit is about 3.8 in the NE region (precipitation with a pH of 4.2, concentrated by 40 percent
through evapotranspiration). Within the Reuss formulation,  no provisions are available to address acidity
generated by organic processes, and only limited acidity can be added to soil solutions by the exchange
of base cations in deposition for acid cations on soil exchange sites. For  these reasons,  the model has
difficulty predicting the extremely low pH values observed in most O horizons and in the organic-rich A
                                              560

-------
       6.0
       5.3-
    _o

    (34.7

    X
    Q.
       4.0-
       3.4
                                      580
                                      0.768
                                      1.35
                                      0.948
          3.4
4.0
  4.6
pH water
5.2
5.8
Figure 9-38. Comparison of measured vs. calculated soil pH values for the 580 aggregated master
horizons in the NE. The heavy diagonal line Is the 1:1, perfect correspondence line. In general, the
model slightly over-projects soil pH values.
                                        561

-------
Table 9-19.  Summary Statistics of the Projected Changes in Soil Base
Saturations in the NE Region, Obtained Using the Different Deposition Scenarios
or Soil Aggregation Schemes. The Time Increments Included in the Table Are 20,
50, and 100 Years
                         LTA        TY        LTA-rbc     LTA-zbc     WBA
% BS (Initial)
Mean
Std. Dev.a
Median
A-%_BS (20 years)
Mean
Std. Dev.
Median
Max.
A-%_BS (50 years)
Mean
Std. Dev.
Median
Max.
A-%_BS (100 years)
Mean
Std. Dev.
Median
Max.

20.9
10.5
17.6

-1.4
0.9
-1.3
-6.0

-3.5
1.7
-3.4
-11.0

-7.6
3.2
-7.5
-17.0

20.9
10.5
17.6

-1.4
0.9
-1.3
-6.0

-3.7
2.0
-3.5
-13.0

-7.9
3.4
-8.1
-21.0

20.9
10.5
17.6

-1.4
0.9
-1.3
-6.0

-3.7
1,8
-3.5
-11.0

-8.0
3.3
-7.9
-18.0

20.9
10.5
17.6

-1.4
0.9
-1.3
-6.0

-3.8
1.8
-3.6
-12.0

-8.4
3.6
-8.1
-20.0

24.4
20.3
16.7

-1.7
1.6
-1.5
-5.0

-4.2
3.9
-4.3
-20.0

-7.5
6.5
-6.4
-33.0
   Standard deviations are reported as absolute values.
                                     562

-------
Table 9-20.  Summary Statistics of the Projected Changes in Soil  pH
in the NE Region, Obtained Using the Different Deposition Scenarios
or Soil Aggregation Schemes. The Time  Increments Included in the
Table are 20, 50, and 100 Years
                       LTA       TY       LTA-rbe     LTA-zbc   WBA
Soil pH (initial)
Mean
Std. Dev.a
Median
A-Soi! pH (20 years)
Mean
Std. Dev.
Median
Maximum
A-Soil pH (50 years)
Mean
Std. Dev.
Median
Maximum
A-Soil pH (100 years)
Mean
Std, Dev.
Median
Maximum

5.32
0.194
5.34

-0.075
0.140
-0.032
-0.68

-0.167
0.187
-0.086
-0.88

-0.355
0.278
-0.272
-1.10

5.30
0.206
5.33

-0.075
0.132
-0.040
-0.67

-0.181
0.198
-0.105
-0.90

-0.385
0.298
-0.326
-1.12

5.32
0.194
5.34

-0.078
0.143
-0.036
-0.71

-0.181
0.194
-0.108
-0.91

-0.389
0.295
-0.299
-1.12

5.32
0.194
5.34

-0.081
0.145
-0.037
-0.73

-0.192
0.203
-0.114
-0.94

-0.418
0.310
-0.344
-1.15

5.30
0.206
5.33

-0.046
-0.131
-0.011
-0.65

-0.116
0.177
-0.048
-0.84

-0.289
0.274
-0.210
-1.01
  Standard deviations are reported as absolute values.
                                     563

-------
                                       NE  Lakes
                                  Deposition  =  LTA
                                     Year  =  50
                                    Model  =  Reuss
ฃ 1.0 ~|
O
         Q.
         O
           0,6 -
         > 0.4 -
           0.2 -
         O
  o.o
   -30.00
                     95 % Conf. Limit
                     Predicted Distribution
                     5 % Conf. Limit
                                -20.00              -10.00
                              A  Base  Saturation  (%)
                                                              o.oo
                                       NE  Lakes
                                  Deposition  =  LTA
                                     Year  =  50
                                    Model  =  Reuss
         c
         QL
         O
         ft
  0.6 -
         |0.4-i

         15
         rs 0.2
         E
                     95 % Conf. Limit
                     Predicted Distribution
                     5 % Conf, Limit
                                                                    B
-0.75
                                -0.50
                                          A  pH
                                          -0.25
0.00
Figure 9-39.  Cumulative distribution of projected (a) base saturations and (b) soil pH values for
soils in NE. Projections were made using the Reuss model in conjunction with the LTA (constant
level) deposition. The results are presented for 50 years.
                                          564

-------
                                       NE  Lakes
                                  Deposition =  LTA
                                     Year  =  100
                                    Model  =  Reuss
 c 1.0-1
 o
         Q_
         O
           0.6
         > 0.4 -
         _
         rso.2 ~J
         E
         oฐ
         ^ -30
                     95 % Conf. Limit
                     Predicted Distribution
                     5 % Conf. Limit
      .00
-20.00
 Base
           -10,00
Saturation  (%)
                                                               o.oo
                                       NE  Lakes
                                  Deposition =  LTA
                                     Year  =  100
                                    Model  =  Reuss
 c 1.0
 o
 Q_
 O
         > 0.4 -
 E
 Z5
O
           0.0
                     95 % Conf. Limit
                     Predicted Distribution
                     5 % Conf. Limit
                                                                            B
             -0.75
                       -0.50
                                             pH
                    -0.25
                               0.00
Figure 9-40.  Cumulative distribution of projected (a) base saturations and (b) soil pH values for
soils in  the NE. Projections were made using  the Reuss  model in  conjunction with the LTA
(constant level) deposition. The results are presented for 100 years.
                                         565

-------
         400
         300-
o 2001

O
         ioo
       CO
       cd
       o
           o-
        -100
             0
                       no
                           a    D
                     a     a    a
                          a
               Dฐ
                      fflD
                     nn   a o
                      _*  ฐ
                          nฐ
                          20
     Aggregated
    Mineral Horizon
         0n|V
                                                    n  = 145
                                                    m = 6.10
                                                    b  = 29.3
                                                    r2  = 0.310
40
60
                                        %BS
Figure 9-41.  Plot of the measured (ELS) ANC values for lakes in the NE vs. the estimated,
watershed-level base saturations for mineral horizons in those watersheds.
                                      566

-------
              T   -5-
              _l
               cr


              O
              z .10
              < -10
                 -15
                     O

                                                       Aggregated
                                                      Mineral Horizon
                                                          Only
                                                        2O Years
                      2O
-4O
6O
                                          %BS
               -
              1
              o
                 -20
                 -40
    -6O-
                 -SO
                                                        Aggregated
                                                       Mineral Horizon
                                                           Only
                                                         CSO Years
                                                                           B
                                    20
                                          %BS
                                                       6O
                    O
             —.  -SO
cr
i.
i
                -100
             1 -150
                -20O
                                  -i.
                            ซ^ S
                                            Aggregated
                                          Mineral Horizon
                                              Only
                                                          1OO Years
                                    20
                                           %BS
                                                       6O
Figure 9-42. Plot of the changes in surface water ANC values at (a) 20, (b) 50, and (c) 100 years
as projected by the Reuss model vs. the estimated, present-day, watershed-level base saturations
for mineral horizons in those watersheds. The deposition used in computing these differences is
the LTA deposition.
                                            567

-------

CO
CD
"53




u
-1-
-2
-3-
-4-
-5-
-6-
-7-
-8-
-9-
-10-
-11-

C
a
ฃ
'" - • ."s"-1* ' D
fff M IS n
%, B= * &,"ซ ฐฐ™ n • •
of " o ฐฐ " ^ " " B "
B"ซf B D
0 " a ฐ
n o 0
ra
a
Change After 50 Years
a
m~= 0.045 ฑ0.016
0 b = -4.47 ฑ1.79
r2 =0.060
e

I 10 20 30 40 50 &
%BS
                    100
                     90
                     80
                  c
                  o

                  1  70

                  li
                  w  60
                  0)
                  1
                  CQ  so
                  _c

                  oป 40
                  CO

                  O  30
20


10


 0
                                                 Changes After 50 Years
                                  IS fi

                               -~.il "a."
                               *     %
                                    B n
                               o   a on a    Do

                                •  •    *:*:
                                       *  •> a
                                                    B
10      20     30      40
              %BS
                                                              50
                                                 60
Figure 9-43,  Plot of the projected changes in soil base saturations vs. the observed, present-day,
aggregated base saturations for mineral horizons in the NE. The projections were made with the
Reuss model using LTA deposition. The data are presented for 50-year projections. Data in plot
(A) illustrate the absolute changes that are projected using the model.  Plot (B) illustrates the
relative changes that are projected to occur.
                                             568

-------
horizons.  For most other horizons, however, the relationship between observed and predicted soil pH
values are acceptable.

      Projections regarding future changes in base saturation and pH of soils in the NE are listed in
Tables 9-19 and 9-20, respectively.  The projections are illustrated in Figures 9-39 and 9-40 for 50 and
100 years,  respectively.  Mean and median  changes In soil base  saturation  exhibit uniform rates of
depletion of about 0.75 percent ฑ 0.05 percent per year throughout the simulations regardless  of the
deposition scenarios used. The rates of depletion are slightly higher for the reduced base cation loading
scenarios, as expected. Extreme values are only about three times the magnitude of the mean changes
observed for the population  of systems being studied. Soil pH values show similar time-dependent
changes (Table 9-20). Soil pH values decline at a mean rate of about 0.04 pH units per year throughout
the simulation,  with  only minor,  but consistent, differences projected  among the different  deposition
scenarios.

      The data  presented here are based on the results aggregated from mineral horizons only. An issue
of concern with these results, therefore, is the possible  effect that organic horizons might have on the
magnitude  or direction of changes projected by the model. To evaluate this issue, the model  runs using
data aggregated both with and without the presence of organic layers would  need to  be conducted.
These model runs have not been performed.  However, this issue, is addressed  in Section 9.3.3.2 for the
Bloom-Grigal model.  The importance of organic horizons in regulating changes to soil chemistry are
presented there.

9.3.3.1.3.2  Soils as  an indicator of possible future changes  In ANC ~

      Soils  may serve  as indicators of future changes occurring because of acidic deposition.  An
analysis of this hypothesis is useful for  identifying those systems that are most susceptible  to adverse
changes.   This  information also  could  be used in the design  phases  of a monitoring program.  To
conduct this analysis, aggregated, watershed-level estimates of  mineral horizon base saturations were
obtained for the 145 watersheds in the NE. These data were plotted  against (1) the observed lake water
ANC values for each of the lakes  and (2) the  projected changes  in ANC at 20, 50 and 100 years.  Figure
9-41 shows the relationship observed between aggregated soil  base saturations and surface water ANC.
These data support a significant relationship between these variables (see Section 8.8.1). Although there
is considerable scatter in the results,  lakes with  lower ANC  values tend to have soils  with  lower
aggregated base saturations.

      The  relationship  between  current  base saturation  and  projected  changes in  ANC is  more
pronounced, as illustrated in Figure 9-42.   In this analysis,  the projected magnitude of change in ANC
at 20-, 50-, and 100-year intervals is related to  the current, aggregated watershed base saturation. At each
of these time steps, watersheds with aggregated soil base saturations in excess  of 20 percent exhibit little
or no significant decline in projected ANC over the course of the simulations. As  the base  saturations
decrease below 20 percent,  however, there is a marked increase in the magnitude of the response of
individual systems to the effects of acidic deposition. These results suggest that systems with aggregate
base  saturation of less than 20 percent should be most susceptible to the effects  of acidic  deposition,
at least in terms of projected changes in surface water ANC.
                                               569

-------
      An alternative approach is to examine changes in soil base saturations as a function of the current
state of the systems. Figure 9-43A shows the relationship between current, aggregated, mineral soil base
saturations and projected changes in base saturation at 50 years. A significant relationship does not exist
between  the magnitude of the projected  changes and the current base saturation of the systems being
studied.  This  result is Interesting, especially In light of the rather strong  relationship observed between
base  saturation and the projected change in  surface water ANC.  The  observation suggests that  the
largest changes In soil base saturation (In the absence of weathering) occur independently of the present
base  status of soils. The magnitude  of the changes may be mediated by physical factors,  such as  the
thickness or bulk density of the soils. There are some chemical limitations on these changes as well.

      Current  base saturation is related to the relative magnitude of changes expected to occur in  the
base  status of these soils (Figure 9-43B). Although the data are scattered somewhat (probably due to
differences in soil physical  parameters and to variations in soil exchange  properties) the lower the initial
base  saturation, the greater the projected relative depletion of base cations from the soil exchange
complex.  This result is consistent with  the observations  concerning surface water ANC changes and
demonstrates that the soils are behaving In an internally consistent manner.

      As noted throughout this section,  the Level  II models are, by and large, single-process models,
used  in this context to determine the contribution of individual processes to the integrated responses of
watersheds as complete systems.  The suggestion that  systems with base saturations in excess of 20
percent are  at minimal risk to  future change  needs to be considered In the context of the complete
system. Therefore, watersheds with higher aggregate base  saturation could  experience  significant
acidification  if  other processes, such as  hydrologic routing of water within the soils and  ground water,
restrict the degree of interaction between  soils and soil-water. Similarly, the present base saturation status
of soils probably plays only a  limited role in  regulating episodic acidification (as opposed to chronic
acidification, the principal issue of concern in  this report).

      Conversely,  soils with base saturations  of less than 20 percent might  not  experience significant
chronic depletions in ANC  if related processes, such as mineral weathering, were able to sustain current
base  saturations. The above analysis, however, suggests that these systems are more  susceptible to
adverse  changes.  Programs  designed  to  monitor future changes should  consider using soil base
saturation status as one criterion for site selection.

9.3.3.1.3.3 Southern Blue Ridge Province -

      Summaries of the results for the SBRP are given in Tables 9-21 and 9-22 and in Figures 9-44 and
9-45.  Soils in the SBRP currently have  base saturations that are half as large as those in  the NE (see
Section 5.5.1.3). For this reason, it is reasonable to expect both larger and  more rapid responses to the
effects of acidic deposition  In the SBRP compared to the NE. Examination of the model results, however,
suggests that the soils in the SBRP respond more slowly to acidic deposition than do the soils in the  NE.

      At 50 years, the average base  saturations in SBRP soils have declined by between 20 percent and
30 percent, depending on the deposition  scenario considered. These declines are equivalent to absolute
changes  in base  saturation of 2 to 3 percent.  By 100 years, the average  base  saturation  for the soils
                                               570

-------
Table 9-21.  Summary Statistics of the Projected Changes in Soil
Base Saturations in the SBRP, Obtained Using the Different
Deposition Scenarios. The Time Increments Included in the Table
Are 20, 50, 100, and 200 Years

%_BS (initial)
Mean
Std. Dev.a
Median
A-%_BS (20 years)
Mean
Std, Dev.
Median
Max
A-%_BS (50 years)
Mean
Std. Dev.
Median
Max
4-%_BS (100 years)
Mean
Std. Dev.
Median
Max
A-%_BS (200 years)
Mean
Std. Dev.
Median
Max
LTA

10.5
5.7
9.3

-0.49
0.27
-0.44
-1.09

-1.89
0.37
-1.90
-2.80

-5.16
0.76
-5.06
-7,24

-8.83
0.99
-8.78
-12.41
TY

10.5
5.7
9.3

-0.55
0.29
-0.51
-1.18

-2.41
0.58
-2.52
-4.26

-6.04
1.14
-5.64
-9,22

-9.03
1.38
-9.10
-12.41
LTA-rbc

10.5
5.7
9.3

-0.59
0.30
-0.51
-1.18

-2.37
0.37
-2.42
-3.28

-6.00
0.70
-5.84
-7.93

-9.36
1.45
-9.44
-13.03
LTA-zbc

10.5
5.7
9.3

-0.70
0.32
-0.59
-1.26

-2.94
0.49
-2.96
-3.82

-7.16
0.99
-7.15
-8,90

-9.44
1.32
-9.68
-12.41
  Standard deviations are reported as absolute values.
                                      571

-------
Table 9-22.  Summary Statistics of the Projected Changes in Soil
pH in the SBRP, Obtained Using the Different Deposition Scenarios.
The Time Increments Included in the Table Are 20, 50, 100, and 200
Years

Soil pH (initial)
Mean
Std. Dev.a
Median
A-Soii pH (20 years)
Mean
Std. Dev.
Median
Maximum
A-Soil pH (50 years)
Mean
Std. Dev.
Median
Maximum
A-Soi! pH (100 years)
Mean
Std. Dev.
Median
Maximum
A-Soil pH (200 years)
Mean
Std. Dev.
Median
Maximum
LTA
5.15
0.10
5.13
-0,03
0.01
-0.03
-0.06
-0.10
0.03
-0.10
-0.19
-0.34
0.09
-0.35
-0.49
-0.66
0.15
-0.67
-0.82
TY
5.12
0.10
5.12
-0.03
0.01
-0.03
-0.06
-0.13
0.05
-0.15
-0.21
-0.40
0.14
-0.47
-0.64
-0.65
0.16
-0.64
-0.81
LTA-rbc
5.15
0.10
5.12
-0.03
0.01
-0.04
-0.06
-0.13
0.05
-0.14
-0.21
-0.41
0.12
-0.45
-0.58
-0.69
0.15
-0.71
-0.84
LTA-zbc
5.15
0.10
5.13
-0.04
0.01
-0.04
-0.06
-0.19
0.06
-0.20
-0.27
-0.52
0.14
-0.57
-0.68
-0.74
0.17
-0.81
-0.86
   Standard deviations are reported as absolute values.
                                      572

-------
                              SBRP  Stream  Reaches
                                 Deposition =  LTA
                                   Year  =  50
                                  Model  =  Reuss
         o
         Q.
         O
         v_
         Q_
           0.8 -
  0.6 -
         > 0.4 -
J2


1
D
           0.2-
           0.0
            -30
                              Limit
                    Predicted Distribution
                             Limit
      .00
                              -20.00              -10.00
                             A  Base  Saturation  (%)
                                                                     0.00
                             SBRP Stream  Reaches
                                Deposition  =  LTA
                                   Year  =  50
                                  Model  =  Reuss
           1.0 -

         CL
         O
         D
         30.2 -
         E
        o
                    95 % Conf. Limit
                    Predicted  Distribution
                    5 % Conf. Limit
                                                                        B
-0.75
                               -0.50
                                           pH
                                          0.25
0.00
Figure 9-44.  Cumulative frequencies of changes in (a) soil base saturation and (b) soli pH for
the population of soils in the SBRP. The projections are for year SO and have been computed
using LTA deposition data.
                                       573

-------
           1.0 -i
         Q-
         O
         *- 0.6 -
        Q_

         > 0.4 -

        15
           0.2-
         E
         3
           o.o
            -30.00
                              SBRP  Stream Reaches
                                 Deposition  =  LTA
                                    Year =  100
                                  Model  = Reuss
                    QC <5/
                    3*j /o
                    Predicted Distribution
                    5 ฐ,
                                                              A
                              -20.00             -10.00
                             A  Base  Saturation  (%)
                                                            o.oo
         c 1.0 i
CL
O
*-
OL
           0.8
           0.6 -
         > 0.4 -
        .4—
        _o
         30.2 -j
         E
        o
                              SBRP  Stream  Reaches
                                 Deposition  =  LTA
                                   Year  =  100
                                  Model  = Reuss
            95 % Conf. Limit
            Predicted Distribution
            5 % Conf. Limit
-0.75
                                                                    B
                               -0.50
                                           pH
                                         -0.25
0.00
Figure 9-45.  Cumulative frequencies of changes in (a) soil base saturation and (b) soil pH for
the population of soils in the SBRP The projections are for year 100 and have been computed
using LTA deposition data.
                                       574

-------
in this region have declined by 60 percent ฑ10 percent, and by 200 years, by approximately 90 percent.
Again, these projections are made with the assumption that weathering is not supplying base cation to
the soils of the  region. Clearly,  primary mineral weathering supplies  base cations to these soils and,
hence, soil acidification will be slower than the rates projected here.

      Changes in soil  pH projected using the Reuss model are  parallel to those projected for the soil
base cations.  Changes are minimal at 20  years, with an  absolute magnitude of the projected  changes
ofQ.04 pH units, regardless of the deposition scenario  used. By 50 years, the changes are significant.
Within this time frame, soil pH values have declined by an average of about 0.13 pH units, depending
on the deposition scenario. The  rate of decline in soil pH increases between 50 and 100 years. At this
point, the soil is projected to be losing much of the buffering capacity, with a resultant drop in soil pH.
By 200  years, when much of the soil buffering  capacity has  been depleted, the average soil pH has
declined to values near the minimum that can be reached In the context of the Reuss model framework.

9.3.3.1.3,4  Regional comparisons —

      Results from the Reuss modelling effort have led to many observations concerning the soil behavior
In the two regions and how that behavior affects the ANC of waters  passing through those soils  (see
Table 9-23).  First, the absolute rate of cation depletion  is slower In the SBRP than it is in the NE. Within
the first 50 years, mean base saturations have declined by about 3.5 percent in the NE, while they have
declined by only slightly  less than 2 percent in the SBRP. However, in terms  of the percentage  of
available cations, cation depletion is severe In the SBRP. After 50 years, between 20 and 30 percent  of
the cations on soil exchange sites have been  lost through leaching, whereas in the NE only about 15-
20 percent of the available cations are lost during the same time period. These trends continue  at 100
years. In the NE, base saturations have declined by about 7.5 percent,  or about one third of the total
supply of available cations on soil exchange sites. In the  SBRP, base  saturations have declined by only
slightly more than 5 percent.  However, this decline constitutes more than half of the available  buffering
capacity.

      For soil pH, parallel trends to those describe above are observed. In the NE, soil pH values decline
by about 0.2 pH units during the first 50 years of the simulations,  while in the SBRP, the average change
is on the order of 0.1 pH units. However, because soil pH values  in the NE are initially higher than those
in the SBRP by an average of about 0.15 pH units, the observed differences result primarily in a lessening
of the disparity between the two regions in terms of their characteristic  pH  values. By 100 years, changes
in soil pH in the  SBRP have started to accelerate  such that the absolute magnitude of the differences
observed between the two regions are, again, equal  to about 0.15 pH units. We interpret this observation
as an indication that the loss of buffering capacity occurs later in the SBRP relative to the NE.  This
difference is attributable to differences in soil physical properties, such as soil thickness and bulk density,
rather than to differences in soil chemical characteristics. The absolute magnitude  of  the  changes
projected for the two regions is equal to about 0.35-0.4 pH units  (depending on the deposition scenario
considered).

      Results from the Reuss model suggest that, in the absence of mineral weathering, both regions will
sustain substantial losses  of base cations  from their soils. In translating these changes into the  effects
on surface water chemistry, the model results  suggest that the largest effects (on the time scale  of 100

                                              575

-------
Table 9-23.  Comparison of the Changes in Soil Base Saturation and Soil pH that
Are Projected to Occur in the NE and SBRP. The  Projections Have Been
Obtained Using Reuss's Cation Exchange Model and Are Presented for Two
Deposition Scenarios, the LTA and TY Depositions
                              LTA
TY

A_ANC (year 50)
Mean
Std. Dev.a
A_%BS (Year 50)
Mean
Std. Dev.
A_SoiI pH (Year 50)
Mean
Std, Dev.
A_ANC (Year 100)
Mean
Std. Dev.
A_% BS (Year 100)
Mean
Std. Dev.
A_SoiI pH (Year 100)
Mean
Std. Dev.
NE
-13.7
23.6
-3.5
1.7
-0.17
0.19
-32.1
36.1
-7.6
3.2
-0.36
0.28
SBRP
-2.96
1.05
-1.9
0.4
-0.10
0.03
-14.6
6.04
-5.2
0.8
-0.34
0.09
NE
-16.1
26.4
-3.7
2.0
-0.18
0.20
-43.1
51.5
-7.9
3.4
-0.39
0.30
SBRP
-3.7
2.0
-2.4
0.5
-0.13
0.05
-23.0
16.2
-6.0
1.1
-0.40
0.14
  Standard deviations are reported as absolute values.
                              576

-------
years) will be observed in the NE.  Larger changes are projected to occur much earlier in the NE. For
example, after 50 years, the mean change in projected surface water ANC in the SBRP is less than -3
fj,eq  L"1, whereas it is more than -13 /*eq L"1  in the NE.  At 100 years, the  rate of ANC decline has
increased  in the SBRP. At this point, the projected change for the SBRP is about -15 #eq L"1  (using the
LTA deposition; this change is about -23/jeq L"1 for the TY deposition). However, this change is still only
half the magnitude of that projected to occur in the NE, regardless of the deposition scenario.  Therefore,
larger relative changes  in the base  cation pool are  projected for the  soils  in the  SBRP;  the larger
projected  effects of those changes appear in the surface waters of the NE region.

9.3.3.1.3.5  Summary —

     A number of observations  and conclusions can  be drawn from  results obtained using  the Reuss
model to evaluate changes in soil pH and base saturations in the NE  and SBRP regions.

      •      In the absence of mineral weathering,  significant depletions of base cations are projected
            for the soils of both  the NE and SBRP regions.

      *      The absolute magnitude of base cation depletion is greater in the NE than it is in  the SBRP.
            The relative projected changes, however, are greater in the SBRP.

      •      Current base saturation of soils in  the regions can be used as indicators of potential future
            change in surface water ANC. Soils with base saturations currently in  excess of about 20
            percent appear to undergo  minimal changes on the time scale of the  next 100  years. For
            soils with base saturations less than 20 percent, however, projected changes in surface water
            ANC appear to increase with decreasing aggregate base saturation.  This effect is more
            pronounced in the NE region than it is in the SBRP.

      *      Current base saturation can be used as an indicator of the anticipated relative changes that
            might occur in the soil base status over the next 100 years. The percentage decline in  base
            saturation increases  with decreasing base saturation, although other factors, such as soil
            thickness or bulk density, probably influence the relationship  as well.

9.3.3.2  Bloom-Grigal  Model

9.3.3.2.1  Data  sources -

     In the DDRP, the basic unit of investigation is the watershed.  Instead of characterizing  the effects
of acidic  deposition on individual  soils, the research focus Is the integrated effect  of the  soils  on  a
particular  watershed.  Consequently,  all  of the Bloom-Grigal  modelling input data are at the  watershed
level.  Because the DDRP sample of watersheds  serve as  the  basic link to the target  population of
watersheds, watershed level results can  be extrapolated to the target population of watersheds.

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

                                              577

-------
9.3.3.2.1.1  Deposition data ~

      The deposition data are from four sources: (1) Typical Year (TY), (2) Long-Term Annual Average
(LTA), (3) LTA Reduced  Base Cation (LTA-rbe)--LTA with  a 50 percent reduction in dry base cations,
and (4) LTA Zero Base Cation (LTA-zbc)-LTA with a 100  percent reduction in dry base cations.  Both
of these reductions in dry base cations are offset by concomitant increases in dry H"1".  The details on
the acquisition/generation of the DDRP deposition data sets are given in  Section 5.6.

      A summary of the regionally weighted median deposition inputs in the  four deposition  data sets
(LTA,  LTA-rbc, LTA-zbc, and TY) used in the Bloom-Grigal modelling is presented In Table 9-24 by region.
In the NE there  appears to be little  difference between LTA and TY.  A  priori, we  expect to see only
minor differences in the forecasts made with these two deposition data sets.  The SBRP TY median value
of H+ is 22 percent greater than the  LTA value.  The NH4+  is, however,  lower  and NO3" is greater.
Consequently, the total effective acidity (H+total = H+ + NH4+ - NOg") is only slightly larger.

      The largest differences in H+tota) are between the LTA and the reduced (LTA-rbc) and zero (LTA-
zbc) deposition data sets.  In the NE the difference  between the median H+tota| in the LTA and median
value  of H+tota| In the LTA-zbc is 0.19 keq ha"1, while In the SBRP this difference is 0.24 keq ha"1. Such
differences should  result In differences In projections, especially for the higher levels of H+tota|.

9.3.3.2.1.1.1  Deposition scenarios --

      The Level  II base cation models are run with  three deposition scenarios.  The scenario common
to both the NE  and SBRP is the constant deposition scenario.   In this scenario  the annual load of
deposition is held constant for the duration of the simulation.

9.3.3.2.1.1.2 Northeast ~

      In addition to the constant deposition scenario in the  NE, a ramp down scenario is used to simulate
a 30 percent decrease  in wet and dry SQ42" deposition. Deposition is held constant for the first 10 years
of the simulation.  Beginning with the eleventh year, deposition is decreased  by 2 percent per year for
15 years for a total decrease of 30 percent.  This new ievef is then  held constant for the  duration of the
simulation.

9.3.3.2.1.1.3 Southern Blue Ridge Province ~

      In addition to the constant deposition scenario in the  SBRP, a ramp up scenario is used to simulate
a 20 percent increase in wet and dry SO42" deposition.  Deposition is held constant for the first 10 years.
Beginning with the eleventh year, deposition is increased  by  (20/15) percent per  year for 15 years for
total increase in deposition of 20 percent.  This new level is then held  constant for the duration of the
simulation.
                                              578

-------
Table 9-24. Regionally Weighted Median Values of Initial Annual Deposition
Inputs to the Bloom-Grigal Mode! for the Northeastern Region and the
Southern Blue Ridge Province*

NE
LTA
LTA - rbc
LTA - zbc
TY
SBRP LTA
LTA - rbc
LTA - zbc
jy
H+

0.71
0.79
0.91
0.78
0.67
0.82
0.97
0.82
NH/

0.15
0.15
0.15
0.14
0.22
0.22
0.22
0.16
N03-

0.44
0.44
0.44
0.45
0.42
0.42
0.42
0.46
Total Acid Inputb

0.43
0.49
0,62
0.44
0.47
0.61
0.77
0.51
  Values are in keq ha"1 yr~1

  Total Acid Input = [H+ + NH4+ - NOa"]
                                             579

-------
9.3.3.2.1.2  Soils data -

      The Bloom-Grlga! model uses one value for the following soil chemistry variables to depict the
soil chemistry of a particular watershed:  soil pH, cation  exchange capacity (CEC), and  the  sum  of
exchangeable base cations (SOEBC).  To obtain results that represent the central tendency of the DDRP
regions, a large  number of observations for these variables were aggregated to obtain values for each
of the DDRP watersheds.  Combining or aggregating these data can be accomplished in several ways.
It is not correct to use a simple average for all variables; rather, capacity and intensity variables should
be weighted differently.  Of the variables used  in the BIoom-Grigal  model simulations, soil pH  was
aggregated using an intensity variable aggregation method, whereas CEC and SOEBC were aggregated
using a capacity variable aggregation method. The details of these methods are provided in Johnson
et ai.  (1988b).

      To evaluate the role of soil organic  horizons (Oa, Oe, and Oi) in the chemistry of soils, the  soils
data for the BIoom-Grigal data were aggregated two ways:  (1)  including organic horizons and
(2) excluding organic horizons.

      A summary of the regionally weighted median values of the Bloom-Griga! soil chemistry input  data
(aggregated with and without organic horizons) is presented in Table 9-25.  In the NE, inclusion of the
organic horizons decreases the median pH by 0.30 and base saturation by slightly more than 1 percent.
In the SBRP the changes are  even more negligible.  Although  the pH and SOEBC values are similar
between the regions, CEC in the SBRP is more than twice that in the NE. Simply stated, the soils in the
SBRP have greater exchangeable acidity than those in the NE with similar SOEBC.

      The regional initial soil pH and  percent base saturation  with and without organic horizons are
presented in Figure 9-46 as cumulative distribution functions (CDFs).  This manner of presentation allows
interregional and intraregional differences to be easily observed.  The soil pH in the SBRP is less affected
by the exclusion of the organic horizons than in the NE.

9.3.3.2.2  Projections of soil ph and  percent base saturation -

      In all, BIoom-Grigal model simulations representing more than 300,000 years were needed to obtain
the results for the four deposition  data sets  and  different deposition scenarios.  A  subset  of these is
presented below by region, and a regional comparison follows in Section 9.3.3.2.3.

9.3.3,2.2.1  Northeast -

      The results of the BIoom-Grigal simulations in the NE with LTA, LTA-rbc, and LTA-zbc are presented
in Figures 9-47 and 9-48 for the change in soil pH and percent base saturation, respectively.  Statistical
summaries of the CDFs are presented  in Tables 9-26 and 9-27.

      The projected changes in soil pH and percent base saturations using the constant LTA deposition
scenario are quite small (Figure 9-47),  The median change after  100 years is only -0.04.  Of the systems
in the target population, less that 25 percent of the watersheds have a projected decrease in  soil pH
greater than -0.10. The largest decrease is projected to be -0.35. Most of these changes are probably

                                              580

-------
Table 9-25.  Regionally Weighted Median Values of Annual Initial Soil
Chemical Values Input into the Bloom-Grigal Model for the Northeastern
Region and the  Southern Blue Ridge Province8.  With and Without Organic
Soil Horizons
                           pH          SOEBC           CEC            BS
      WJthb               4.62         40.04            183.8           21.98
w/o
SBRP
With
W/O
4.92 34.11

4,85 40.42
5.01 40.62
177.4

433.3
436.4
20.60

9.22
9.20
pH            =     intensity weighted soil pH
SOEBC        =     mass weighted sum of exchangeable base cations
CEC           =     mass weighted cation exchange capacity
BS            =     base saturation [(SOEBC/CEQ*100]

a All values in keq ha"1 except BS which  is percent.
  "With" means that organic soil horizons were Included. "W/O" means that organic soil horizons
  were excluded.
                                      581

-------
       5 0,8 •

       r
       o
       Q.
       O 0.8
         0.4

       2
       =3

       ง0.1
       o
         0.0
                       Soil  pH
               Organic Horizons Included
                      NE
                      SBRP
                  4.5     5.0      5.5
                       Soil pH
                                                             o.o
                                                 Percent Base  Saturation
                                                Organic Horizons Included
NE
SBRP
                                                   Base  Saturation (%)
                       Soil  pH
               Organic Horizons Excluded
       O
       CL
       O0.6 -

       a.
       o
         o.o
                                        NE
                                        SBRP
           4.0
4.5     5.0      5.5
     Soil pH
                                                 Percent Base  Saturation
                                                Organic Horizons Excluded
                                                                          NE
                                                                          S8RP
                                                                    tO   20   30   40  50   I
                                                                     Base  Saturation  (%)
Figure 9-4i. Cumulative distributions of aggregate initial soil pH and percent base saturation in the
NE and SBRP, with and without organic horizons.
                                              582

-------
                 NE Lake Watersheds
               Deposition = LTA Constant
              Organic Horizons  = Included
               Dry Base Cations =  100 %
CL
O0.6

O.

IV
         L
         o
          0.0 -
           -075
     	Yr. 20
      	Yr, 50
     	Yr. 100
                       A Soil pH
                                                           NE Lake Watersheds
                                                     Deposition =  LTA Ramp 30%  Decrease
                                                       Organic Horizons =  Included
                                                         Dry Base  Cations = 100 %
                                                           ง0.8-

                                                           O
                                                           O-
                                                           oo.ซ •
                                                           a.

                                                           a>

                                                           Si 0.4 H
                                                           5
                                                                 — Vr, 20
                                                                 - - Ifr. SO
                                                                 — Yr. 100
                                                              -0.50      -0.2S
                                                                A S05t  pH
                                                                                         0.00
                 NE Lake Watersheds
               Deposition =  LTA Consfanf
              Organic Horizons  — Included
               Dry Base Cations =   50 %
         O
         CL
         O0.6
                • Yr. 20
                " Yr. 5O
                • Yr. 100
                     -0.'SO      -t>.*2S
                       4 Soil pH
                                        o3o
                                                           NE Lake Watersheds
                                                     Deposition =  LTA Ramp 30%  Decrease
                                                       Organic Horizons =  Included
                                                         Dry Base  Cations =  50  %
                                                           ง0.8-
                                                  o
                                                  a.
                                                  O 0.6 -
                                                          _O
                                                           m
                                                          o
                                                    0.0-
                                                     -07S
                                                       	 Yr. 20
                                                        	Yr. SO
                                                                A Soil  pH
                 NE  Lake Watersheds
               Deposition —  LTA Constant
              Organic Horizons  = Included
               Dry Bass Cations =   0 %
         5ฐ-*

         o
         CL
         OO.S
         01
         •Sซ.ซ
         J2

         la.
         O
               	Yr. 2O
               - - Yr. 50
               — Yr. 100
                     -0.50      -0,25
                      A Soil pH
                                        0.00
                                                           NE Lake Watersheds
                                                     Deposition =  LTA Ramp 30% Decrease
                                                       Organic Horizons =  Included
                                                         Dry  Base  Cations =  0  %
                                                  o
                                                  Q.
                                                  00.6-
                                                       	Tr. 20
                                                       	Yr. 50
                                                       	Yr. 100
                                                              -0.50       -0.25
                                                                A Soil  pH
Figure  9-47. Regional CDFs of the  projected  change in the pH of soils on NE lake  watersheds
under constant and ramp down (30  percent 4-) deposition scenarios after 20, 50, and 100 years of
LTA, LTA-rbc, and LTA-zbc deposition.  Organic horizons are Included.
                                                   583

-------
                    NE Lake Watersheds
                 Deposition =  LTA Constant
                Organic Horizons  = Included
                 Dry Base Cations =  100 %
           o
           o.
           go.ซ -
           CL
           _O
           3
       	Tr. 20
       - - Tr. SO
       	Tr. IOO
   -M.OO     -20,00     —10.00      0.00
        A Base  Saturation  (%)
                                                          NE Lakซ Watersheds
                                                    Deposition = LTA  Ramp 30% Decrease
                                                      Organic Horizons  = Included
                                                       Dry Base Cations = 100 %
                                                            o
                                                                  — Yf. 20
                                                                  - — Yr. SO
                                                                  — tr. too
                                                              -3000     -20.00      -10.00      O.OO
                                                                    A Base Saturation  {%)
                   NE Lake Watersheds
                 Deposition =  LTA Constant
                Organic Horizons  = Included
                 Dry Base Cations =  50  %
o
O.
O0.6


-------
Table 9-26.  Bloom-Grigal Model Regional Projections for the Change in Soil pH in the Northeastern
United  States. Projections Made Using LTA, LTA-rbc, and LTA-zbc Deposition with Constant and
30% Ramped Down Deposition Scenarios at Three Levels of Base Cations in  Dry  Deposition.
Results Reported for 20-, 50-, and 100-Year Projections.  Organic Soil Horizons Included


                  Deposition = Constant ** Dry Base  Cations = 100% ** LTA

YEAR       MEAN     STD DEV   MIN       P 25      MEDIAN    P 75      MAX
20
50
100
-0.02
-0.04
-0.06
0.02
0.05
0.07
-0.10
-0.19
-0.35
-0.03
-0.06
-0.10
-0.01
-0.03
-0.04
0.00
0.00
0.00
0.00
0.00
0.00
                Deposition = Constant ** Dry Base Cations = 50% ** LTA - rbc

YEAR       MEAN     STD DEV   MiN       P 25      MEDIAN    P 75      MAX
20
50
100
-0.03
-0.06
-0.09
0.03
0.05
0.07
-0.13
-0.22
-0.38
-0.04
-0.09
-0.13
-0.02
0.05
-0.08
-0.01
-0.02
-0.04
0.00
0.00
0.00
                Deposition = Constant ** Dry Base Cations = 0% ** LTA - zbc

YEAR       MEAN     STD DEV   MIN       P 25       MEDIAN   P 75      MAX
20
50
100
-0.05
-0.10
-0.14
0.04
0.06
0.08
-0.17
-0.27
-0.44
-0.08
-0.14
-0.18
-0.05
-0.09
-0.14
-0.03
-0.06
-0.10
0.00
0.00
0.00
                Deposition = 30% Decrease ** Dry Base Cations = 100% ** LTA

YEAR       MEAN     STD DEV   MIN       P 25      MEDIAN   P 75      MAX
20
50
100
-0.01
-0.02
-0.02
0.02
0.02
0.03
-0,07
-0.11
-0.17
-0.02
-0.03
-0.03
-0.01
0.01
-0.01
0.00
0.00
0.00
0.00
0.00
0.00
                                                                               continued
                                          585

-------
Table 9-26.  (Continued)
             Deposition = 30% Decrease ** Dry Base Cations = 50% ** LTA - rbc
YEAR       MEAN     STD DEV   MIN      P 25      MEDIAN   P 75      MAX
20
50
100
-0.02
-0.03
-0.04
0.02
0.03
0.04
-0.11
-0.14
-0.22
-0.03
-0.04
-0.05
-0.02
-0.02
-0.03
-0.01
-0.01
-0.01
0.00
0.00
0.00
              Deposition = 30% Decrease ** Dry Base Cations = 0% ** LTA - zbc
YEAR       MEAN     STD DEV   MIN       P 25      MEDIAN   P 75      MAX
20
50
100
-0.05
-0.06
-0.07
0.03
0.04
0.05
-0.15
-0.17
-0.27
-0.07
-0.07
-0.10
-0.04
-0.05
-0.07
-0.02
-0.03
-0.04
0.00
0.00
0.00
                                         586

-------
Table 9-27.  Bloom-Grigal Model Regional Projections of the Change in Percent Base Saturation
in the Northeastern United States.  Projections Made Using LTA, LTA-rbc, and LTA-zbc Average
Deposition with Constant and 30% Ramped Down Deposition Scenarios at Three Levels of Base
Cations in Dry Deposition. Results Reported for 20-, 50-, and 100-Year Projections.  Organic Soil
Horizons Included


                  Deposition = Constant ** Dry Base Cations = 100% ** LTA

YEAR       MEAN      STD DEV   MIN      P 25      MEDIAN   P 75       MAX
20
50
100
-0.97
-2.00
-3.15
1.11
2.34
3.79
-5.00
-11.26
-18.70
-1.58
-3.49
-5.83
-0.57
-1.05
-1.46
0.00
0,00
0.00
0.00
0.00
0.00
                Deposition = Constant ** Dry Base Cations = 50% ** LTA - rbc

YEAR       MEAN     STD DEV   MIN      P 25      MEDIAN   P 75       MAX
20
50
100
-1.43
-2.90
-4.45
1.20
2.47
3.95
-5.56
-12.11
-19.88
-2.09
-4.14
-6.96
-1.25
-2.63
-3.68
-0.49
-0.96
-1.26
0.00
0.00
0.00
                Deposition = Constant ** Dry Base Cations = 0% ** LTA - zbc

YEAR       MEAN     STD DEV   MIN       P 25      MEDIAN   P 75       MAX
20
50
100
-2.32
-4.52
-6.58
1.41
2.80
4.33
-6.63
-13.72
-21.94
-3.05
-5.74
-8.82
-2.24
-4.29
-6.16
-1.50
-2.70
-3,36
0.00
0.00
0.00
                Deposition = 30% Decrease ** Dry Base Cations = 100% ** LTA

YEAR       MEAN     STD DEV   MIN      P 25      MEDIAN   P  75       MAX
20
50
100
-0.74
-0.90
-1.14
0.90
1.20
1.69
-4.38
-6.86
-10.03
-1.32
-1.43
-1.57
-0.38
-0.38
-0.38
0.00
0.00
0.00
0.00
0.00
0.00
                                                                               continued
                                          587

-------
Table 9-27.  (Continued)
             Deposition = 30% Decrease ** Dry Base Cations = 50% ** LTA - rbe



YEAR        MEAN     STD DEV   MIN      P 25      MEDIAN   P 75      MAX
20
50
100
-1.18
-1.49
-1.91
1.07
1.54
2.26
-4.94
-8,36
-12.66
-1.69
-2.25
-2.55
-1.00
-1.03
-1.03
-0.31
-0.31
-0.31
0.00
0.00
0.00
              Deposition = 30% Decrease ** Dry Base Cations = 0% ** LTA - zbc



YEAR        MEAN     STD DEV   MIN      P 25      MEDIAN   P 75      MAX
20
50
100
-2.02
-2.68
-3.49
1.29
1.97
2.95
-6.02
-10.01
-15.27
-2.71
-3.57
-4.83
-1.86
-2.50
-2.85
-1.19
-1.21
-1.21
0.00
0.00
0.00
                                     588

-------
within the uncertainty of the model and are not significant.  With the LTA-rbc and LTA-zbc deposition
larger decreases in soil pH  in a greater proportion of the systems is evident.   Yet, even at the highest
level of acidic input (LTA-zbc), the median change  in soil pH is only -0.14.  Larger decreases are evident
in a few systems (<10 percent).

      Ramping the deposition down by 30  percent reduces the projected declines significantly.  With LTA-
zbc the median decline in soil pH is -0.07, one half that projected for the constant LTA scenario. The
results for the projected change in base saturation  are similar to those for pH.  However, when the initial
median base saturation is only 17 percent, a decrease of 6 percent (LTA-zbc) after 100 years  (to  11
percent) is projected.  The 30 percent decrease in deposition results in smaller changes.

      Excluding  the organic horizons  results in an amplified decrease in soil  pH and base saturation
(Figures 9-49 and 9-50, Tables 9-28 and  9-29),  Without the contribution of the organic horizons, the
median change in soil pH and percent base saturation  after only 20 years is nearly equal to or greater
than the 100-year projections for soils with organic  horizons. This result is misleading, however. The
initial median pH of the soils without the organic horizons is 4,92, and after 100 years of LTA deposition
the median change is -0.21. For the soils with the organic horizons the initial median pH is 4.62, and
after 100 years the median  change  is only -0.04.   Thus, although pH of the  soils without the organic
horizons had greater projected changes, their pH  values were still projected to be higher at the end of
the 100-year  simulation.

      As for pH, the decrease in percent base saturation for the soils  with the organic horizons is greater
than for the  soils without the organic  horizons.  However,  because percent base saturation  is  initially
lower for the soils without the organic horizons, the projected percent base  saturation is much lower
than for the soils with the organic horizons.

      There are two principal explanations  for the above  results.  First, soils without organic horizons have
higher initial  pH values.  At  higher pH values less Al is available to buffer the losses of base cations.
Recalling Equation 9-9 (S =  I - A  - C), the tendency of a system to lose bases  (S) increases if the inputs
of acidity (I)  are held constant  and the  buffering of Al (A) and protonation of bicarbonate (C) are
decreased.  Such  is the case at higher pH values.  Second, the  large decreases in soil pH result from
low base saturation, as reflected  by the equation  that relates soil pH to base saturation (see Equation
9-12).  For low base saturation (<20 percent), the slope of the pH versus percent base saturation line
increases dramatically and small changes  in base saturation result in large changes in pH.  Because the
systems without organic horizons have higher pH  values, their base cation  losses are greater than  for
other soils with  lower pH values {e.g., the soils  with  the  organic  horizons)  assuming  all other soils
characteristics are  the same.   The loss rate of base cations  decreases,  however, as the soil pH
decreases. Turchenek et ai. (1987) and  Turchenek et al. (1988), also using the BIoom-Grigal  model,
demonstrated similar results.

      The median  change in base saturation after  50 years of constant LTA deposition on soils without
organic horizons  is -4.38, and the pH  change is -0.12.   After an  additional 50 years, the percent base
saturation decreases by an  additional -2.44 and the pH by -0.09.
                                               589

-------
                    NE Lake Watersheds
                 Deposition =  LTA Constant
                Organic Horizons =  Excluded
                 Dry Base Cations = 100 %
           ฃ.ฐ*
           t
           o
           a.
           oo.e
           a.
           o
                   - Yr. 20
                   - Yr. SO
                   • Vr. <ฉQ
                       -D.9O     -O.2S
                        A Soil  pH
                                        0.00
                                                        NE take Watersheds
                                                  Deposition = LTA Ramp 30% Decrease
                                                    Organic Horizons =  Excluded
                                                     Dry Base Cations = 100 %
                                                            A Soil pH
                   NE Lake  Watersheds
                 Deposition = LTA Constant
                Organic Horizons = Excluded
                 Dry Base Cations =   50 %
t
o
o.
     	Yr. SO
     	Yr. SO
     	Yr. 100
A
                                     i
                        jo      -oss
                        A Soil  pH
                                        o.&o
                                                        NE Lake Watersheds
                                                  Deposition = LTA Ramp 30% Decrease
                                                    Organic Horizons =  Excluded
                                                     Dry Base Cations =  50 %
                                                          o
                                                          CL
                                                          O0,t
                                                                	W. 20
                                                                	Vr. 50
                                                                — Yr. 100
                                                           -030     -0.25
                                                            A Soil pH
                   NE Lak* Watersheds
                 Deposition  — LTA Constant
                Organic  Horizons = Excluded
                 Dry Base Cations =   0 %
                                                        HE Lake Watersheds
                                                  Deposition = LTA Ramp 30% Decrease
                                                    Organic Horizons =  Excluded
                                                     Dry Base Cations =  0 %
           -oSo
             & Soil  pH
                                                                      -eia
                                                                        A Soil pH
Figure 9-49. Regional CDFs of the projected change in the pH of  soils  on NE  lake watersheds
under constant and ramp down (30% 4.) deposition scenarios after 20, 50, and 100 years of LTA,
LTA-rbc, and LTA-zbc deposition.  Organic horizons are excluded.
                                                   590

-------
                     NE Lake Watersheds
                   Deposition = LTA Constant
                  Organic Horizons = Excluded
                   Dry Base Cations =  100 %
             — O.4.
             30.2
             O
                    - Yr. 20
                    - fr. SO
                    - Tr. 100
               -3O.OO     -20.00     -10.00      0.60
                    A Base Saturation  (%)
       NE Lake Watersheds
 Deposition =  LTA  Ramp  30% Decrease
   Organic Horizons = Excluded
     Dry  Base  Cations =  100 %
                                                             o
                                                             a.
   	 Vr. 2O
   	fr. SO
   	Vr, 100
   -20',00     ~10*.00
A Base Saturation
                       (%)
                     NE Lake Watersheds
                   Deposition  = LTA Constant
                  Organic Horizons = Excluded
                   Dry  Base Cations —   50 %
             O
             a.
             ฃftซ-
             
-------
Table 9-28. Bloom-Grigal Model Regional Projections of the Change in Soil pH in the Northeastern
United States.  Projections Made Using LTA, LTA-rbc, and LTA-zbc Deposition with Constant and
30% Ramped Down Deposition Scenarios at Three Levels of Base Cations in  Dry Deposition.
Results Reported for 20-, 50-, and 100-Year Forecasts.  Organic Soil Horizons Excluded


                  Deposition  = Constant ** Dry Base Cations  = 100% ** LTA

YEAR      MEAN     STD DEV      MIN     P 25    MEDIAN      P 75      MAX
20
50
100
-0.08
-0.16
-0.22
0.07
0.12
0.15
-0.33
-0.45
-0.55
-0.13
-0.27
-0.34
-0.05
-0.12
-0.21
-0.02
-0.06
-0.10
0.00
0.00
0.00
                Deposition = Constant ** Dry Base Cations = 50% ** LTA - rbc

YEAR        MEAN     STD DEV    WIN       P 25      MEDIAN   P 75      MAX
20
50
100
-0.10
-0.21
-0.28
0.07
0.13
0.14
-0.38
-0.47
-0.60
-0.16
-0.33
-0.40
-0.09
-0.18
-0.27
-0.04
-0.08
-0.15
0.00
0.00
0.00
                Deposition = Constant ** Dry Base Cations = 0% ** LTA - zbc

YEAR        MEAN     STD DEV    MIN      P 25      MEDIAN    P 75      MAX
20
50
100
-0.14
-0.27
-0.35
0.09
0.15
0.16
-0.45
-0.53
-0.69
-0.22
-0.41
-0.48
-0.14
-0.31
-0.39
-0.05
-0.11
-0.19
0.00
0.00
0.00
                Deposition = 30% Decrease ** Dry Base Cations = 100% ** LTA

YEAR        MEAN     STD DEV    MIN      P 25      MEDIAN    P 75      MAX
20
50
100
-0.06
-0.07
-0.08
0.05
0.06
0.07
-0.28
-0.28
-0.29
-0.10
-0.11
-0.12
-0.04
-0.06
-0.08
-0.02
-0.03
-0.03
0.00
0.00
0.00
                                                                              continued
                                          592

-------
Table 9-28.  (Continued)
             Deposition = 30% Decrease ** Dry Base Cations = 50% ** LTA - rbc



YEAR         MEAN    STD  DEV    MIN      P 25      MEDIAN     P 75      MAX
20
50
100
-0.09
-0.12
-0.15
0.06
0.08
0.09
-0.37
-0.37
-0.37
-0.13
-0.19
-0.22
-0.08
-0.10
-0.13
-0.03
-0.06
-0.08
0.00
0.00
0.00
              Deposition = 30% Decrease ** Dry Base Cations = 0% ** LTA - zbc



YEAR        MEAN    STD  DEV    MIN      P 25      MEDIAN   P 75      MAX
20
50
100
-0.13
-0.19
-0.24
0.08
0.10
0.11
-0.44
-0.44
-0.48
-0.19
-0.28
-0.33
-0.13
-0.20
-0.25
-0.05
-0.09
-0.13
0.00
0.00
0.00
                                         593

-------
Table 9-29.  Bloom-Grigal Model Regional Projections for the Change in Percent Base Saturation
in the Northeastern United States.  Projections Made Using LTA, LTA-rbc, and LTA-zbc Average
Deposition with Constant and 30% Ramped Down Deposition Scenarios at Three Levels of Base
Cations in Dry Deposition. Results Reported for 20-, 50-, and 100-Year  Projections. Organic Soil
Horizons Excluded


                  Deposition = Constant ** Dry Base Cations = 100% ** LTA

YEAR       MEAN     STD  DEV     MIN        P 25      MEDIAN    P 75      MAX
20
50
100
-2.40
-4.79
-6.93
1.58
3,11
4.54
-6.74
-14.34
-23.19
-3.56
-6.51
-8.95
-2.16
-4.38
-7.08
-1.41
-2.72
-3.87
0.00
0.00
0.00
                Deposition = Constant ** Dry Base Cations = 50% ** LTA - rbc

YEAR       MEAN     STD DEV     MiN       P 25     MEDIAN    P 75      MAX
20
50
100
-3.12
-6.05
-8.56
1.57
3.02
4.39
-7.64
-16.04
-24.40
-4.21
-7.87
-9.99
-2.81
-5.59
-8.65
-2.04
-4.47
-6.06
0.00
0.00
0.00
                 Deposition = Constant ** Dry Base Cations = 0% ** LTA - zbc

YEAR       MEAN     STD  DEV     MiN        P 25      MEDIAN   P 75      MAX
20
50
100
-4.14
-7.66
-10.36
1.89
3.51
4.97
-8.79
-18.03
-26.41
-5.55
-9.72
-12.25
-4.04
-7.51
-9.86
-2.49
-5.28
-7.03
0.00
0.00
0.00
                Deposition = 30% Decrease ** Dry Base Cations = 100% ** LTA

YEAR       MEAN     STD  DEV     MIN        P 25      MEDIAN    P 75      MAX
20
50
100
-1.96
-2.45
-3.07
1.31
1.83
2.65
-5.98
-9.04
-14.13
-2.66
-3.20
-4.37
-1.87
-2.19
-2.46
-1.06
-1.07
-1.08
0.00
0.00
0.00
                                                                               continued
                                          594

-------
Table 9-29. (Continued)
             Deposition = 30% Decrease ** Dry Base Cations = 50% ** LTA - rbc
YEAR      MEAN      STD DEV     MIN       P 25      MEDIAN    P 75      MAX
20
50
100
-2.79
-3.87
-5.06
1.43
2.16
3.12
-7.00
-1 1 .30
-16.92
-3.83
-5.03
-6.35
-2.50
-3.48
-4.68
-1.85
-2.64
-2.97
0.00
0.00
0.00
              Deposition = 30% Decrease ** Dry Base Cations = 0% ** LTA - zbc
YEAR        MEAN    STD DEV     MIN       P 25      MEDIAN   P 75      MAX
20
50
100
-3.81
-5.69
-7.52
1.75
2.68
3.78
-8.13
-13.55
-19.69
-5.07
-7.34
-9.06
-3.68
-5.34
-7.21
-2.32
-3.90
-5.01
0.00
0.00
0.00
                                      595

-------
      Organic horizons apparently influence the soil chemistry in at least two ways.  First, because
organic horizons have abundant base cations, they increase the size of the exchangeable base cation
pool (see Table 9-25).  Because of the concomitant addition of CEC, however, the relative magnitude
of the median percent base saturation remains the same.  Second,  because organic  horizons are
inherently acidic, the lower soil pH values decrease the rate of base cation removal from the soil cation
exchange complex.  At lower soil pH values, potentially toxic  acid cations, such as AI3+ and Mn2+
become more prevalent and may be transported in drainage waters to surface water or groundwater.

      The BIoom-Grigal modelling results using the TY deposition in the NE are similar to those using
the LTA deposition.  For this reason they are not presented here.

9.3.3.2.2.2  Southern Blue  Ridge Province -

      Although the median  aggregated soil pH values are higher in the SBRP target population  of
watersheds than in the NE,  SBRP soils have dramatically lower percent base  saturation.  Because of
these chemical properties, and for the larger reasons described in the  preceding section, the soils in
the SBRP are projected to experience decreases in pH and percent base saturation than soils in the NE.

      The changes projected for the soils without the organic horizons differ only slightly from those for
the soils with the organic horizons.  Unlike the NE, omitting the organic horizons does not appreciably
affect the  initial aggregate soil  pH and percent base saturation.  As for the NE,  the forecasts using the
TY deposition data are only  slightly higher than those using the LTA.  (These data are not presented.)

      The CDFs for the projected changes in soil pH and percent  base  saturation using the  LTA, LTA-
rbc, and LTA-zbc deposition data sets are presented in Figures  9-51 and 9-52.   The summary statistics
for these CDFs are presented in Tables 9-30 and 9-31.  These results are for the soils with the organic
horizons included.

      After 50 years under the constant deposition scenario, the median predicted change in soil pH is
-0.16.  After 100 years it is -0.24.  From year 100 to year 200 the change is only -0.07.  The  change in
percent base saturation after 100 years is -3.22, and after 200 years of the change is only -3.39.  These
results imply that between year 100 and 200 the buffering mechanism these soils shifts with the latter
mechanism buffering soil pH to more acidic  levels.

      Projected changes with the increased  acid loadings of the LTA-rbc and LTA-zbc are much more
rapid.  After 50 years under constant LTA-rbc deposition,  the projected change in soil  pH equals that
under the LTA deposition after 100 years.  With the  LTA-zbc,  an equivalent projected change occurs in
less than 50 years.  The 20 percent ramped Increase in deposition further Increases the rates of projected
change:  increased acid inputs increase the initial rate  of change,  i.e., the decrease in  base  saturation
and soil  pH.  The convergence of the CDFs for 50, 100, and  200 years demonstrates these results.
These results are  explained by the initial conditions:   the greater the pH and the lower the base
saturation, the faster the base  cation depletion rate (see Section 9.3.3.2.2.2).

      This convergence of  the CDFs  may  represent the limit of change, at  least for the next two
centuries. In order to consider the limit of change, assume that the median value of change represents
the central tendency for change.  The limit for change in soil pH, therefore, is approximately  -0.40 and
                                              596

-------
                  SBRP  Stream  Watersheds
                 Deposition = LTA Constant
                Organic  Horizons = Included
                 Dry Base Cations =  tOO %
1.0 -
งซ•
t
a
Q.
oo.ซ-

0}


o
"^
ฃ(j_2_

o

	 Yr. JO _. .
	 Yr. 50 .--"7 '/
	 Yr. 100 ,' f ' I
	 Yr. 200 : ,' ' I
I'll
/.' '
// I


./ / '
!/' /
i ' i
// *
* t S J









•/ > (
/,' .
/' '
• ' ' (

— ซ&***"''"' )
-O.75 -O^SO -0*25 Q.QQ
A Soil pH
                                           SBRP Stream Watersheds
                                      Deposition = LTA Ramp 20% Increase
                                         Organic Horizons = Included
                                          Dry Base Cations  = tOO %
l.O -
„
O0"8'
t
O
CL
O0.6 -
Q_

Iซ.4-
3
3
o


	 Iff, 20
	 	 Vr. 50 f - ' •"
---- Yr. 100 ;.'"• "
— — Vr. 200 •"! ^
\ \
* i
,' * *
/ '
• t
;"' * t
\* '
'•* t
a ,' i
/,' ; f
ป' :
// r
^,-i'-- * i
_^*f* ",--'' r
y














                                                             O.7S      -0^0
                                                                         Soil pH
                  SBRP Sfreom Watersheds
                 Deposition  = LTA Constant
                Organic Horizons  = Included
                 Dry Base Cations =  50  %
                                           SBRP Sfream Watersheds
                                      Deposition = LTA Romp 20% Increase
                                         Organic Horizons = Included
                                          Dry Base Cations  =  50 %
           O
           CL
           O0.6
-O.SO
  A Soil pH
                                   t
                                    o
                                    o.
                                                          S
                                                          3
                                                          E.
                                                                     -OSO      ~0.2S
                                                                       A Soil pH
                                                                                    ~  s
                                                                                    •  X
                                                                                     J
                                                                                    /
                                                                                       0.00
                  SBRP Stream Watersheds
                 Deposition  = LTA Constant
                Organic Horizons  = Included
                 Dry Base  Cations =  0  %
                                           SBRP Stream Watersheds
                                      Deposition = LTA Ramp 20% increase
                                         Organic Horizons — Included
                                          Dry Base Cations  =  0 %
           o.
           oo.ซ
                	Yr. Jป
                	Yr, 50
                	Yr. 100
                   Yr. 200
                                    o

                                    g(M-
                                    O.
                               -
                          Soil pH
                                               -0-M      -0.15
                                                 & Soil pH
Figure 9-51. Regional CDFs of the projected change in the pH of soils on SBRP stream watersheds
under constant and ramp up (20% t) deposition scenarios after 20, SO, 100, and 200 years of LTA,
LTA-rbc, and LTA-zbc deposition.  Organic horizons are  included.
                                                  597

-------
                    S8RP Stream Watersheds
                   Deposition  — LTA Constant
                  Organic Horizons  = Included
                   Dry Base Cations = 100 %
tJQ -



C


O
CL
00.6-
O_

5
3
Eo^-

o

OJB •
	 Yr. ป
	 Yr. SO f,
	 Yr. 100 ,"'
	 Yr. ZOO f
ft
a,
1*
j I
1,'
1
E
1
J
I
. f
.-* 1
J* |
* .X"' * '
""* -' * '
                                                    SBRP Stream Watersheds
                                                Deposition =  LTA Ramp 20%  Increase
                                                  Organic Horizons  = Included
                                                   Dry  Base  Cations = 100  %
LO-


CI.. fl .
_Oฐ'ฐ "
*C
O
(X
J3Q.6-
Q-
0}
^0.4 -
O
3
O

. 	 trr. 28 ^
- — - Yr. 50 ft
— — Yr. 100 f
— yr. 200 ,
1
1
/
r






^
•? '
.•i /
.----/ ,' /
                        	      	
                      Base Saturation (%)
                                                       -zo.oa       .
                                                      Base Salutation (%)
                    SBRP Stream Watersheds
                   Deposition =  LTA Constant
                 Organic Horizons —  Included
                   Dry Base Cations =  50 %
                 	Yr. ป
                 	Yr. 50
                 	Yr. <00
                 	Yr. 100
-jojป     -moo     -10.00     o.oo
     A  Base Saturation (%)
                                                    SBRP Stream Watersheds
                                               Deposition = LTA Ramp 20% Increase
                                                  Organic Horizons = Included
                                                   Dry Base Cations  =  50 %
-


o8,-*-
o
a.
oo^ -
a.


^0.4-
ฃ
o


Oft .
	 Yr. 20
	 Yr. 50
	 Yr. 100
	 Yr. 200






;
:
:
l
i
4 j
.--— 1*


ff-














.
















                                                              -M.OO     -20.00     -10.00      0.00
                                                                   A Base Saturation  (%)
                    SBRP Stream Watersheds
                  Deposition —  LTA Constant
                 Organic Horizons =  Included
                  Dry Base Cations =  0 %
1,0-
go*-
O
Q.
oo.e -
a.
ฎ
"5
o
O.O -
	 Yr. m
	 Yr. $0
-. 	 Yr" 200 /
|
M
|
/?
	 7" } .





                                                    SBRP Stream Watersheds
                                               Deposition =  LTA Ramp 20% Increase
                                                  Organic Horizons  = Included
                                                   Dry Base  Cations =  Q %
1,6 •

go*-
t
O
CL
oe.fi -
Q-
fl>
^O.4 -
s.
iซ-
o


	 Yr. SO j.
	 Yn 1OO 1
	 Yr. MO ,f,
t|
1
I
i
i
I
ft
f
It
I*
|J
^** -
1 '
_^.^-- — v*"* i
— ' ' J















                       -,       --
                      Base Saturation (%)
                                                       -        -
                                                      Base  Saturation (%)
Figure 9-52. Regional CDFs of the projected  change  in  the percent base saturation of soils on
SBRP stream watersheds under constant and  ramp up (20% t) deposition scenarios after 20, 50,
100, and 200 years of LTA, LTA-rbc, and LTA-zbc deposition.  Organic horizons are included.
                                                  598

-------
Table 9-30. Bloom-Grigal Model Regional Projections for the Change in Soil pH in the Southern
Blue Ridge Province.  Projections Made Using LTA, LTA-rbc, and LTA-zbc Deposition with Constant
and 20% Ramped Up Deposition Scenarios at Three Levels  of Base Cations in Dry Deposition.
Results Reported for  20-, 50-,  100-, and 200-Year Projections. Organic Soil Horizons Included
YEAR
YEAR
YEAR
  Deposition = Constant ** Dry Base Cations = 100% ** LTA

 MEAN      STD DEV  MIN       P 25       MEDIAN    P 75
Deposition = Constant ** Dry Base Cations = 50% ** LTA - rbc

 MEAN     STD DEV     MiN     P 25        MEDIAN    P 75
Deposition = Constant ** Dry Base Cations = 0% ** LTA - zbc

 MEAN    STD DEV     MIN     P 25        MEDIAN   P 75
MAX
20
50
100
200
-0.07
-€.16
-0.24
-0.28
0.03
0.06
0.09
0.10
-0.14
-0.32
-0.55
-0.62
-0.10
-0.21
-0.29
-0.32
-0.07
-0.16
-0.24
-0.27
-0.05
-0.11
-0.17
-0.19
-0.02
-0.05
-0.08
-0.08
 MAX
20
50
100
200
-0.11
-0.25
-0.34
-0.36
0.04
0.08
0.10
0.09
-0.21
-0.48
-0.61
-0.66
-0.15
-0.32
-0.40
-0.42
-0.11
-0.24
-0.31
-0.35
-0.09
-0.20
-0.27
-0.29
-0.03
-0.06
-0.12
-0.19
 MAX
20
50
100
200

YEAR
20
50
100
200
-0.16
-0.33
-0.41
-0.43
Deposition
MEAN
-0.08
-0.24
-0.35
-0.38
0.05
0.10
0.10
0.09
= 20%
STD_
0.03
0.08
0.11
0.10
-0.29
-0.61
-0.66
-0.68
Increase ** Dry
DEV MIN
-0.15
-0.46
-0.68
-0.71
-0.22
-0.42
-0.47
-0.50
Base Cations =
P_25
-0.11
-0.30
-0.42
-0.44
-0.14
-0.32
-0.41
-0.43
= 100% **
MEDIAN
-0.08
-0.23
-0.33
-0.36
-0.13
-0.29
-0.36
-0.37
LTA
P_75
-0.06
-0.18
-0.28
-0.30
-0.03
-0.07
-0.14
-0.26

MAX
-0.02
-0.06
-0.13
-0.21
                                                                               continued
                                          599

-------
Table 9-30.   (Continued)
YEAR
Deposition
MEAN
= 20% Increase
STD_DEV
** Dry
MIN
Base Cations
P_25
= 50% ** LTA
MEDIAN
- rbc
P_75
MAX
  20
  50
 100
 200
-0.13
-0.33
-0.43
-0.45
0.04
0.10
0.10
0.09
-0.23
-0.60
-0.72
-0.73
-0.16
-0.42
-0.49
-0.51
-0.12
-0,32
-0.41
-0.43
-0.10
-0.28
-0.37
-0.37
-0.03
-0.07
-0.15
-0.27
YEAR
Deposition = 20% Increase ** Dry Base Cations = 0% ** LTA - zbc
MEAN STD_DEV MiN P_25 MEDIAN P_75
MAX
  20
  50
 100
 200
-0.17
-0,41
-0.49
-0.50
0.06
0,12
0.11
0,09
-0.31
-0.71
-0,75
-0.75
-0.23
-0,51
-0,55
-0.57
-0.16
-0.41
-0.49
-0.49
-0.14
-0.37
-0.43
-0.43
-0.03
-0.08
-0.17
-0.33
                                                  600

-------
Table 9-31.  Bloom-Grigal  Model Regional Projections for the Change in Percent Soil  Base
Saturation in the Southern Blue Ridge Province. Projections Made Using LTA, LTA-rbc, and LTA-
zbc Deposition with Constant and 20% Ramped Up Deposition Scenarios at Three Levels of Base
Cations in Dry Deposition.  Results Reported for 20-, 50-, 100-, and 200-Year Projections. Organic
Soil Horizons Included
YEAR
YEAR
YEAR
YEAR
  Deposition  = Constant ** Dry Base Cations = 100% ** LTA

 MEAN     STD DEV     M!N     P 25       MEDIAN    P 75
Deposition = Constant ** Dry Base Cations = 50% ** LTA - rbc

 MEAN     STD DEV     MIN     P 25        MEDIAN    P 75
Deposition = Constant ** Dry Base Cations =  0% ** LTA - zbc

 MEAN    STD DEV     MIN     P 25       MEDIAN    P 75
Deposition = 20% increase ** Dry Base Cations = 100% ** LTA

  MEAN     STD DEV   MIN      P  25       MEDIAN    P 75
 MAX
20
50
100
200
-1.20
-2.44
-3.50
-4.14
0.35
0.79
1.33
2.20
-2.21
-4.67
-7.27
-11.16
-1.30
-2.98
-3.96
-4.10
-1.23
-2.43
-3.22
-3.39
-0.88
-1.83
-2.69
-2.77
-0.46
-0.68
-0.74
-0.74
 MAX
20 -
50
100
200
-1.75
-3.43
-4.53
-5.09
0.35
0.78
1.41
2.55
-2.82
-5.55
-7.71
-13.94
-1.99
-3.73
-4.59
-4.62
-1.72
-3.36
-4.11
-4.23
-1.39
-2.71
-3.52
-3.62
-1.15
-1.54
-1.57
-1.57
 MAX
20
50
100
200
-2.32
-4.31
-5.28
-5.77
0.40
0.84
1.61
2.89
-3.41
-6.34
-9.19
-16.35
-2.59
-4.67
-5.16
-5.17
-2.33
-4.25
-4.70
-4.75
-1.99
-3.65
-4.27
-4.31
-1.67
-2.02
-2.03
-2.03
MAX
20
50
100
200
-1.32
-3.35
-4.67
-5.28
0.36
0.84
1.48
2.74
-2.39
-5.65
-8.19
-15.15
-1.45
-3.75
-4.94
-4.97
-1.34
-3.20
-4.29
-4.39
-0.98
-2.69
-3.53
-3.71
-0.57
-1.5
-1.67
-1.67
                                                                               continued
                                          601

-------
Table 9-31. (Continued)
               Deposition = 20% Increase ** Dry Base Cations = 50% ** LTA-rbc
YEAR            MEAN     STD DEV   WIN      P 25      MEDIAN   P 75      MAX
20
50
100
200
-1.89
-4.27
-5.40
-5.93
0,36
0.84
1.65
3.08
-2.99
-6.45
-9.24
-17.57
-2.13
-4.64
-5.29
-5.30
-1.87
-4.16
-4.78
-4.91
-1.52
-3.59
-4.32
-4.37
-1.28
-2.08
-2.10
-2,10
              Deposition = 20% Increase ** Dry Base Cations = 0% ** LTA - zbc
YEAR            MEAN     STD DEV   M!N      P 25      MEDIAN   P 75      MAX
20
50
100
200
-2.46
-5.00
-5.91
-6.41
0.42
0.92
1.88
3.43
-3.57
-7.01
-10.86
-19.96
-2.74
-5.21
-5.78
-5.78
-2.48
-4.97
-5.28
-5.31
-2.12
-4.34
-4.84
-4.85
-1.80
-2.41
-2.42
-2.42
                                         602

-------
Table 9-32.  Summary of the Bloom-Grigal Projected Changes fn Soil pH and
Percent Base Saturation In the NE and SBRP Under Constant LTA Deposition
                                                  Change in Parameter
                                                  After Selected Years
Region         Parameter          Initial    	
                                  Value   .    20          50        100
NE               pH               4.62       -0.01       -0.03       -0.04
SBRP             pH               4.85       -0.07       -0.16       -0.24
NE               % B,S.          21.98       -0.57       -1.05       -1.46
SBRP             % B.S.           9.22       -1.23       -2.43       -3.22
                              603

-------
for base saturation is about -4.75. The results of subtracting these values from the current median values
suggest that the new median pH value will be about 4.5 and for percent base saturation about 4.5. Both
of these values are quite low considering that they represent aggregate values--i.e., the weighted average
of all soil horizons.

      Such changes are likely to affect surface waters.  While Al buffering prevents the occurrence of
even lower soil pH values, Al3* and other acid cations (e.g., Mn2+ and Fe3"1") will become the dominant
cations in the soil.  These elements are toxic to plants and soil microbes and are also potentially toxic
in the aquatic environment.

      These projections may represent  the worst-case estimates of the effects of acidic deposition on
soils of the NE and SBRP.  Several key points should, however,  be reiterated:  (1) these  projections
were made in the absence of mineral weathering and biomass accretion; (2) sulfate was treated as a
completely mobile anion; (3) the projected changes are sensitive to the relationships between soil pH and
percent base saturation, and these relationships were empirically derived for a selected subset of soils
outside the DDRP regions; and (4) many types of soils were aggregated to derive a single value for initial
soil pH, cation exchange capacity, and  the sum of base cations.

9.3.3,2.3  Regional comparisons

      Soils in the NE currently are somewhat lower in pH than soils in the  SBRP.   Soils in the SBRP,
however, have much  lower  percent base  saturation.   These two  differences lead to very different
projections of the change in soil pH and percent base saturation with the Bloom-Grigal  model.  The
median estimates of total effective acidity (H+tota|   = H+ +  NH4+  - N03")  inputs in LTA deposition
datasets for the NE and SBRP are similar (see Table 9-24).  The output from the simulations using these
two datasets, therefore, can be compared (Table 9-32).

      The regional response of soils to acidic deposition (changes in soil pH and percent base saturation)
differ.  Because the  soils in the SBRP are older and more extensively weathered, their initial percent base
saturation Is  markedly lower than that of the younger, less weathered soils in the NE.  Aggregate  soil
pH values for the SBRP are, at the same time,  slightly higher, which may be due to lower organic matter
content.  These two conditions, moderate to high  pH  (high for forested soils) and low percent base
saturation result in rapid and severe projected decreases in soil pH and percent base saturation in soils
that  are already low In base cations. Only  minor changes in soil pH and  percent  base saturation  are
projected for the NE.

      A series of buffer ranges proposed by Ulrich (1983), assist, in part, with the interpretation of these
results. He suggested that soil-water pH is indicative of the mineral phases that  buffer the soil.  He
proposed five distinct buffer ranges:

      (1)   Calcium carbonate (pH > 6.2)
      (2)   Silicate (pH 6.2  - 5.0)
      (3)   Cation  exchange (pH 5.0 - 4.2)
      (4)   Aluminum  (pH 4.2 - 2.8)
      (5)   Iron (pH 3.8 - 2.4)
                                               604

-------
In the NE, the soils are generally in the Al buffer range (as defined by Ulrich), which is consistent with
the model predictions.  In the SBRP, soils are principally in Uirlch's cation exchange buffer range. When
the pool of exchangeable base cations is depleted, the cation exchange buffer is exhausted, and the
buffering of the system becomes controlled by Al.  In Section 9.3.3.2.2.2, the convergence of the CDFs
were suggested as bounding the change  in soil  pH and percent  base saturation, and pH 4.5  was
proposed as the limit.  The apparent buffering of the NE against changes in soil pH (despite significant
acid inputs) suggests that soil pH values  near 4.5 are likely to be in the Al buffer range rather than the
cation buffer range. The Al buffer range should be extended from pH 4.2 - 2.8 (as suggested  by Ulrich)
to a range of pH 4.5 -  2.8.

9.3.3.2.4  Summary and conclusions

      Based on model  projections, the soils in the NE appear  to be buffered against changes in soil pH
and percent base saturation by an Al buffering mechanism.  Soils In the SBRP may experience significant
decreases in  soil pH and percent base saturation because of their current status and  the level of acid
inputs.  While currently buffered against changes in pH via cation  exchange buffering, the effectiveness
of this buffer will be exceeded with the current levels of acid input.  pH  of these soils is projected  to
decrease until changes in soil pH become controlled by the Al buffering system.The major conclusions
of this analysis  using the BIoom-Grigal model are:

      •    Organic horizons contribute sufficient base cations to increase the size of the base cation
           pool, which slows the rate of acidification.

      •    In the NE, organic horizons contribute acidity and base cations, which results in lower cation
           leaching rates.

      •    Soils in the NE are buffered against changes in soil pH and percent base saturation via an
           Al buffering mechanism.

           Soils in the SBRP may  experience  significant  decreases In soil pH and percent  base
           saturation.   The median soil pH could decrease as much as 0.5 pH units, and the median
           percent base saturation may decrease from Its current level of 9.2 percent to 4.5 percent.
           They are thought to be worst case estimates because sulfate is considered to be a mobile
           anion in this analysis. The extent of change  in the SBRP soils will be limited by the Al buffer
           range,

      •    The soil pH buffer ranges by Ulrich (1983) provide a good basis for interpreting the model-
           based projections.

9.3.4   Comparison of the Bloom-Grigal and Reuss Model Projections

      Results from two soil  cation exchange models  have been presented in  detail.  The behavior
modelled by  the two formulations is  remarkably  different  in  some  respects and more comparable  in
others.  A summary of the median, mean,  and  maximum changes In percent base saturation and soil
pH in the NE is presented in Table 9-33 for 50- and 100-year projections.  Specific comparisons between
the models can be made at two levels. First, model results can be compared for the entire population

                                              605

-------
of lakes in the NE or stream reaches in the SBRP.  Because the primary purpose  of the DDRP is to
obtain such regional estimates, this comparison is of particular importance.  On a more detailed level,
model  results can  be compared for individual lakes  or stream  reaches.   While  a high degree of
correspondence between the model outputs for individual systems should be expected, the comparison
at this level may help to increase our understanding  of the behavior of the individual models.

     With respect to the estimates for the changes expected at the population level, several observations
are of interest here.  The dynamics of the two models are quite different. Initially, the Bloom-Griga! model
projects substantially larger changes  in percent  base saturations than does the Reuss model (Table 9-
33 and Figure 9-53).  Both the mean and median values for changes in percent base saturation projected
using the Bloom-Griga!  model are larger at 20 and 50 years than  those  projected using the Reuss
formulation. At 100 years, however, the relative magnitude of the changes projected  by the two models
is reversed. At 100  years, both the mean and median changes projected by the Reuss model are larger
than those projected by the Bloom-Grigal model.  Overall, the CDFs for the projected changes in  soil
base saturation for systems in the NE using the two models are reasonably  similar.

      In  contrast to model  behavior  for percent base saturation, the  models project  quite different
distributions for the response of soil  pH to acidic deposition.   Results from the Reuss model suggest
that the rate of change in soil pH increases over the course  of the  100-year simulation.  The Bloom-
Grigal model results, on the other hand,  suggest that any changes in soil pH  over this period  are
generally linear.   Another major difference between  the two models is  that, with the Reuss model,  a
small number of  systems experience extreme changes in soil pH during the simulation period.   The
Bloom-Grigal model results, in contrast, suggest that extremes should  not be observed. The effect of the
longer tail on the Reuss model population distributions is an increase in the mean projected change in
soil pH over a 100-year period.  Although the population .medians, as projected by both models, are more
similar than the medians, they still differ significantly  as  illustrated for the NE (Figure 9-54).

      Comparison of the results for individual systems in the NE supports and reinforces the information
obtained from the population-level evaluations.  Figure 9-55  shows  a scatter  plot  of the  changes in
percent base saturation projected for individual systems by the two models at 50- and 100-year intervals.
Surprisingly,  no  correlation between  the two model outputs is evident.  Clearly, the two approaches
used to model cation loss from the soli exchange complex differ. Nevertheless, when integrated over
the population of systems in the NE,  the differences between the models become  sufficiently small to
yield similar population estimates.

      A greater degree of correlation appears to exist between the models for soil pH projections (Figure
9-56).  At 50 years, the Reuss  model appears to project smaller changes  in soil pH for most of the
systems.  However, the Reuss model projects some extreme changes for small number of systems, relative
to the magnitude of changes projected  by the  Bloom-Grigal model.   After an additional 50 years, the
Reuss  model projections have  increased in magnitude relative to those made with the Bloom-Grigal
model.  The  number of systems projected to have extreme changes  in soil pH also increases.

      The patterns for individual systems and for populations  observed in the NE also are observed in
the SBRP.  Because the simulations  were extended to 200 years in  the SBRP, however,  some of the
differences are more pronounced. Table 9-34 summarizes the results  for changes in soil pH and percent
                                              606

-------
Table 9-33.  Comparison of the Results from the Reuss and
Bloom-Grigal Models with Regard to the Magnitude of Changes
in Soil pH and Base Saturation Projected in Soils of the NE.
Results Are Shown for 50 and 100 Years
                       Median     Mean        s,d.a    Maximum
A_% BS (20 years)
Reuss
Bloom-Grigal

-1.3
-2.2

-1.4
-2.4

0.9
1.6

-6.0
-6.7
A_%  BS (50 years)
   Reuss                -3.4        -3.5         1.7       -11.0
   Bloom-Grigal          -4.4        -4.8         3.1       -14.3

A_% BS (100 years)
   Reuss                -7.5        -7.6         3.2       -17.0
   Bloom-Grigal          -7.1        -6.9         4.5       -23.2

A_soil pH (20 years)
   Reuss                -0.03       -d.08       0.14       -0.68
   Bloom-Grigal          -0.05       -0.08       0.07        0.33

A_SoIl pH (50 years)
   Reuss                -0.09       -0.17       0.19       -0.88
   Bloom-Grigal          -0.12       -0.16       0.12       -0.45

A_Soil pH (100  years)
   Reuss                -0.27       -0.36       0.28       -1.10
   Bloom-Grigal          -0.21        0.22       0.15       -0.55
  Standard deviations are repotted as absolute values.
                               607

-------
                                      NE  Lakes
                                   Deposition = LTA
                                     Year = 50
                                    Model = Reuss
               c 1.0-,
               o
              a
              o
              D-
              ฃ
              3.
• 95 % Cent Limit
 Predicted Distribution
 5 % Cant. Limil
                                -20.00           -10.00
                               A Base  Saturation  (%)
                                      NE  Lakes
                                  Deposition  =  LTA
                                    Year =  100
                                   Model  =  Reuss
              o
              a.
              o
              •-O.6 -
             OL
             5<>.2-

              E

              = 01
             *-' -3000
                       95 X Conl, LimU
                       Predlclod Distribution
                       S % Conf. Limll
                                                     B
                                -20.00            -10.00
                                 Base Saturation  (%)
                                  NE  Lake Watersheds
                                Deposition =  LTA Constant
                               Organic Horizons  = Excluded
                                Dry Base Cations =100 %
                         Jo.s

                         t
                         o
                         CL
                         OO.6
                                 • Yr. 2O
                                 • Yr. SO
                                 • Yr. 100
                            -3O.OO     -20.00     -10.00      O.QO
                                  A Base Saturation  (%)
Figure 9-53.  Cumulative distributions of changes in soil base saturation for the population of
watersheds in the NE: (A) Illustrates changes projected by the Reuss model at 50 years;   (B)
indicates those  changes projected after 100 years, again using the Reuss model; and  (C)
shows the results at 20, 50, and 100 years, as projected using the Bloom-Griga! formulation.
                                            608

-------
                                       NE Lakes
                                   Deposition  = LTA
                                     Yซar  =  50
                                    Model  =  Reuss
               c
               CL
               o
               *- 0.6 -
              O.
               o
              •50.2-
               E
               ,0.0
                        95 % Con<- Limit
                        Predicted Distribution
                        5 % Conf. LimH
                                                                             A
                 -ซ.7S
                                           pH
                                      NE  Lakes
                                   Deposition =  LTA
                                      Year = 100
                                    Mode! = Reuss
               a.
               o
               30.2-
               E
               3
               O
1 0.0
                         95 7. Cotlf. Limit
                         Predicted OlslfibufiOEl
                        • 5 % Conf. Limil
                                                             B
                  -0.75
                                  -0.50
                                         A  pH
                                                 -0.25
                                                                 0.00
                                  NE  Lakes Watersheds
                                Deposition =  LTA Constant
                               Organic Horizons  = Excluded
                                Dry Base Cafions =  100 %
                         JJ0.8

                         t
                         o
                         Q.
                         O0.ซ
                         a.
                         0)
                         o
                               	Yr. 20
                               	Yr. SO
                               	Yr. 100
                                     -O.SO       -O.2S
                                       A Soil pH
Figure 9-54.  Cumulative distributions of changes in soil pH for the population of watersheds
in the NE: (A) illustrates changes projected by the Reuss model at 50 years; (B) indicates those
changes projected after 100 years, again using the Reuss model; and (C) shows the results at
20, 50, and 100 years, as projected using the Bloom-Grigal formulation.
                                            609

-------
              15
              0X0
                    -5-
              o 
-------
                    0.0
                    -0.1
               fa
                 8

               CQ
                    -0.3-
                    -0.4-1
                      -0.6
                                                   ฐB
     -0.4  . ,   ....  -0.2
          del sot! pH
         Reuss model
                           0.0
                     o.o
                     -0.2
                •E Q.
                II
                ~T3
                CD
                     -0.4
                     -0.6
                                                        f "
                                 NWYnare
                          n  - 116    b -  -fl.157
                          m  - Q322.   f -  0.191
                                                                           B
                       -1.0
-0.8
-0.6
-0.4
-0.2
0.0
                                            del soil pH
                                           Reuss model
Figure 9-56.  Scatter diagrams of the projected changes in soil pH for individual systems (not
population  weighted)  in  the  NE  obtained  from  the  Reuss  and  Bloom-Grigal  models:
(A)compares results  from the two models after a 50-year  simulation  and (B) illustrates the
relationship observed after 100 years.
                                         611

-------
base saturation obtained with the two models for this region,   in the SBRP, the Bloom-Grigal model
initially projects larger changes in both soil pH and percent base saturation than does the Reuss model.
As the simulations progress, however, the changes projected by the Reuss model Increase more rapidly,
so that by 200 years, substantially larger changes for both percent  base saturation and soil pH are
projected.  Figures 9-57 and 9-58 illustrate the changes projected for the population of soils in the SBRP
at 50 and 100 years.

      In summary,  soil cation exchange models were used to explore possible changes in soil chemical
properties potentially occurring as a  result of acidic deposition.  Overall, the two models yield similar
results with regard  to projected changes for the NE and the SBRP.  The major differences between the
models appear to be that the Bloom-Grigal model projects  more rapid initial changes to the soil chemical
environment, whereas results obtained using the Reuss model indicate that changes  should occur more
rapidly as the soil exchange pool becomes depleted.  Information needed for more critical evaluation of
the two models currently is not available.

9.3.5   Summary and Conclusions

      Results from two soil cation exchange models have  been presented.  These models focus on the
role of cation exchange processes in regulating pH and percent base saturation In  soils.  The models
do  not consider processes such as primary mineral weathering, uptake of cations by vegetation, sulfate
dynamics, or detailed hydrologic flow regimes; nor do they address the deep regolith (e.g., soil depths
> ~2  m).  Consequently, these model results are not directly comparable to the integrated watershed
process models presented in Section 10.  The models do, however, provide considerable information
concerning how base cation pools may respond to continued acidic deposition.

      The two models provide slightly different types of information about the soils and their associated
surface waters. The Reuss model projects changes in both surface water chemistry and soil chemistry.
In contrast, the Bloom-Griga! model provides information about the magnitude of projected changes only
in soil chemical properties. The models employ markedly different algorithms in making these projections.
The Reuss formulation uses a mass action approach.  This approach allows each of the soil reactions
to proceed independently, while simultaneously allowing individual soil properties to vary in an internally
consistent manner. The Bloom-Grigal model relies on empirically derived  relationships to define  time-
varying behavior of individual soil  parameters.   Each approach has certain  advantages, making it
important to determine how the two models differ  in projected changes to the population of systems in
the NE and SBRP  regions.

      While these  models do not explicitly  consider many processes, it is  possible to understand
qualitatively how non-modelled processes would affect the projections presented here.  For example,
cation accretion in biomass is a net base cation  sink, and thus has an acidifying  effect  on the  soils.
Conversely, mineral weathering is a net source for base cations.  Incorporation of a weathering term in
these models would delay the projected response times of individual systems.   Unfortunately, regionally
based estimates of the  magnitude of these processes is  unavailable.  Despite  these limitations, model
results  do provide  information  to possible  watershed responses.   For systems with long projected
response times, future changes in the quality of surface waters likely will not be large.   However, for those
                                              612

-------
Table 9-34. Comparison of the Results from the Reuss and Bloom-
Grigal Models with Regard to the Magnitude of Changes in Soil pH
and Base Saturation Projected in Soils of the SBRP.  Results Are
Shown for 50 and 100 Years
                       Median      Mean      s.d.      Maximum
A_% BS (50 years)
   Reuss               -1,9        -1.9        0.4        -2.8
   Bloom-Grigal         -3.1         -3.1        0.8        -5.1

A_% BS (100 years)
   Reuss               -5.1         -5.2        0.8        -7.2
   Bloom-Grigal         -3.9        -4.4        1.3        -7.9

A_% BS (200 years)
   Reuss               -8.8        -8.8        1.0       -12.4
   Bloom-Grigal         -4.4        -5.1        2.2       -11.7
A_Soil pH (50 years)
   Reuss                -0.10       -0.10       0.03       -0.19
   Bloom-Grigal          0.23       -0.23       0.07       -0.40

A_Soil pH (100  years)
   Reuss                -0.35       -0.34       0.09       -0.49
   Bloom-Grigal          -0.35       -0.35       0.10       -0.67

A_Soil pH (200  years)
   Reuss                -0.67       -0.66       0.15       -0.82
   Bloom-Grigal          -0.38       -0.39       0.10       -0.75
                               613

-------
                                SBRP  Stream  Reaches
                                  Deposilion  =  LTA
                                    Year =  50
                                   Model =  Reuss
c '-o
o
              CL
              O
             Jo., -I
              O
              30.2 -
              ฃ
             o
 o.o
  -30,00
                        95 'Z Con<, UmH
                        Predicted Distribution
                        5 TS. Coal. UmH
                                -2O.OO            -1O.OO
                              A Base Saturation  (%)
                                                                 o.oo
                                                       SBRP  Stream Reaches
                                                         Deposition = LTA
                                                            Year = 100
                                                           Model = Reuss
                                     O
                                     OL
                                     O
                                       O.8-
                                     E
                                           ซ	S5 % Con*. LimH
                                           	   Predicted Distribution
                                           —— S 5C Conf. Urn!)
                                       -30.OO
                                                                               B
                                                       —20.00            -10,00
                                                      4 Base  Saturation  (%)
                                                                                        o.oo
                      SBRP Stream Watersheds
                     Deposition =  LTA Constant
                    Organic Horizons = Excluded
                     Dry Base Cations =  100 %
1JJ-
go*-
T.
o
a.
|
-------
                                    S8RP Siream Reaches
                                       Deposition  = LTA
                                         Year  =  50
                                        Mode!  =  Reuss
                   c
                   CL
                   O
                   -0.6H
                  ฃ
                            95 7. Cant. Umll
                            Predicted Oislribulion
                            5 % Conf. Limif
                                     -0.50
                                               pH
                                                    -0.25
                                    S8RP  Stream  Reaches
                                      Deposition =  LTA
                                        Year-= 100
                                       Model = Reuss
                  a.
                  o
                  E
                            95 7. Conf. Limit
                            Predicted Dislribulion
                            5 % Conf. Limit
                                                                                  B
                                            A  PH
                                                    -0,25
                                 SBRP  Stream Watersheds
                                Deposition =  LTA Constant
                              Organic  Horizons  = Excluded
                                Dry Base Cations =  tOO %
                         o
                         o.
                         O0.6
                         _a
                         3
                         30.2
                         O
                                 • Yr. 20
                                 - Yr, 50
                                 • Yr. 100
                                 • Yr. ZOO
                                              -OJS
                                       A Soil  pH
                                                        o.oo
Figure 9-58.  Cumulative distributions of changes in soil pH for the population of watersheds In
the SBRP: (A) illustrates changes projected by the Reuss model at 50 years; (B) indicates those
changes projected after 100 years, again using the Reuss model; and (C) shows the results at 20,
50 and 100 years, as projected using the Bloom-Grigal formulation.
                                                615

-------
systems with short projected response times, additional information about the magnitude of other potential
sources or sinks for base cations is essential for describing the responses of these systems accurately.

      Detailed results from the  models have been  presented in Section 9.3.3.  Major findings, first for
surface waters and then for soils, are summarized below for both the NE and the  SBRP.

           For  lakes in the NE currently exhibiting ANC values  in  excess of 100 /ieq L"1, mineral
           weathering is  probably the dominant watershed process controlling observed ANC values.

           At present levels of  deposition, lakes in  the NE with ANC values in excess of 100 /ieq L"1 will
           probably not experience declining ANCs  in  the foreseeable future.

      •    For  stream  reaches in the SBRP projected to exhibit ANC values  in excess of 50 jueq L"1
           (after having attained a  state of net zero  sulfate retention), mineral weathering will probably
           be the  dominant watershed process controlling ANC  values  for  systems with chemistry
           currently dominated by  sulfur dynamics.

      •    Stream  reaches in the SBRP with projected  ANC values  in excess of 50 /ieq L"1  (after having
           attained a state of net zero sulfate retention) will probably not become  acidic (ANC < 0 /ieq
           L"1) at current  or slightly elevated levels of deposition. The capacity of weathering processes
           to mitigate  the effects of acidic deposition could  be overwhelmed in those  systems with
           marginal (ANC < 100 /ieq  L"1) contributions  from  weathering, substantial  increases in the
           levels of acidic deposition were to occur.

           For  lakes in the NE exhibiting ANC values of less than 100  /ieq L"1, soil exchange processes
           may be regulating  the  observed ANCs,  although  in most systems, the  ANC is probably
           controlled by a combination  of cation exchange and mineral weathering.

           As an upper limit, approximately 15 percent or over 1000 lakes (four  times the  number of
           currently acidic lakes) in the  NE with current positive ANC values could become acidic (i.e.,
           ANC <  0 jueq L'1) within 50 to 100 years.  The projection is extreme,  because the contribution
           of weathering  is not considered.   However,  some fraction of this  number  of lakes will
           probably become acidic during the next  several decades.

           In the SBRP, changes in observed ANC  values that occur because of changes in the  base
           status of soils during the next century  should  be minimal.

           For systems in the  NE and  SBRP that have  ANC values  in the range of  0 to 50 jueq L"1,
           rates of system  response are projected to  increase  with continued exposure to acidic
           deposition.  The increased rates coincide with the  depletion of soil buffering capacity.

           In the absence of mineral weathering,  significant depletion of  base cations is projected to
           occur in the soils of both the NE and SBRP regions.

           The absolute magnitude of base cation depletion is greater in the NE than it is  in the SBRP.
           The relative projected changes are, however,  greater in the SBRP.

                                              616

-------
Current percent base saturation of soils in the regions can be used as indicators of potential
future change in surface water ANC. Soils with base saturation currently in excess of about
20 percent appear to undergo minimal changes on the time scale of the next 100 years.  For
soils with base saturation less than 20 percent, however, projected changes In surface water
ANC appear to increase with decreasing  aggregate percent  base saturation, an effect that
is more pronounced in the NE than in the SBRP.

Current percent base  saturation  can  be  used as an Indicator of the anticipated  relative
changes that might occur in the  soil  base status over the  next 100 years.  The  relative
percentage decline in percent base saturation [(current - projected)/current] x 100 increases
with decreasing percent base saturation, although other factors, such as soil thickness or bulk
density, probably also  influence the relationship.
                                   617

-------