Sections 1-6
Preprint
July 31, 1989
Future Effects of Long-Term Sulfur Deposition
on Surface Water Chemistry
in the Northeast and Southern Blue Ridge Province
CD
cn
CVJ
(Results of the Direct/Delayed Response Project)
by
M. R. Church, K. W. Thornton, P. W. Shaffer, D. L. Stevens, B.
G. R. Holdren, M. G. Johnson, J. J. Lee, R. S. Turner, D. L Casgeil,
D. A. Lammers, W. G. Campbell, C. I. Liff, C: C. Branjdt, t. H.
G. D. Bishop, D. C. Mortenson, S. M. Pierson, D.;l>^JS
A Contribution to the
National Acid Precipitation AssessimntsProgfahi
U.S. Environmental Protection Agency
Office of Research and Development, Washingtoi
Environmental1 Research Laboratory, Corvallis, Oegon:
-------
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 xtx
Plates .' xxvii
Contributors xxix
Acknowledgments xxxi
1 EXECUTIVE SUMMARY *. 1-2
1.1 INTRODUCTION 1-2
1.1.1 Project Background 1-2
1.1.2 Primary Objectives 1-3
1.1.3 Study Regions 1-4
1.1.4 Time Frames of Concern 1-4
1.2 PROCESSES OF ACIDIFICATION 1-6
1.2.1 Sulfur Retention 1-6
1.2.2 Base Cation Supply 1-7
1.3 GENERAL APPROACH 1-7
1.3.1 SoH Survey 1-8
/ 1.3.2 Other Regional Datasets . : 1-8
t 1.3.3 Scenarios of Atmospheric Deposition 1-10
1.3.4 Data Analysis 1-10
1.4 RESULTS 1-11
1.4.1 Retention of Atmospherically Deposited Sulfur 1-11
1.4.1.1 Current Retention 1-11
1.4.1.2 Projected Retention 1-12
1.4.2 Base Cation Supply 1-15
1.4.2.1 Current Control 1-15
1.4.2.2 Future Effects 1-15
1.4.3 Integrated Effects on Surface Water ANC 1-16
1.4.3.1 Northeast Lakes 1-16
1.4.3.2 Southern Blue Ridge Province 1-20
1.5 SUMMARY DISCUSSION 1-23
1.6 REFERENCES 1-24
2 INTRODUCTION TO THE DIRECT/DELAYED RESPONSE PROJECT 2-1
2.1 PROJECT BACKGROUND 2-1
2.2 PRIMARY OBJECTIVES 2-2
2.3 STUDY REGIONS 2-3
2.4 TIME FRAMES OF CONCERN 2^3
2.5 PROJECT PARTICIPANTS 2-6
2.6 REPORTING 2-6
3 PROCESSES OF ACIDIFICATION 3-1
3.1 INTRODUCTION 3-1
3.2 FOCUS OF THE DIRECT/DELAYED RESPONSE PROJECT 3-3
iii
-------
CONTENTS (continued) Page
3.3 SULFUR RETENTION PROCESSES 3-3
3.3.1 Introduction 3-3
3.3.2 Inputs 3-4
3.3.3 Controls on Sulfate Mobility within Forest/Soil Systems 3-5
3.3.3.1 Precipitation/Dissolution of Secondary Sulfate Minerals 3-7
3.3.3.2 Sulfate Reduction in Soils and Sediments 3-7
3.3.3.3 Plant Uptake 3-8
3.3.3,4 Retention as Soil Organic Sulfur 3-9
3.3.3.5 Sulfate Adsorption by Soils 3-10
3.3.4 Models of Sulfur Retention 3-14
3.3.5 Summary 3-16
3.4 BASE CATION SUPPLY PROCESSES 3-17
3.4.1 Introduction 3-17
3.4.2 Factors Affecting Base Cation Availability 3-20
3.4.2.1 Mineral Weathering 3-21
3.4.2.2 Cation Exchange Processes 3-25
3.4.3 Modelling Cation Supply Processes 3-28
3.4.3.1 Modelling Weathering 3-28
3.4.3.2 Modelling Cation Exchange Processes 3-29
4 PROJECT APPROACH 4-1
4.1 INTRODUCTION 4-1
4.2 SOIL SURVEY 4-3
4.2.1 Watershed Selection 4-3
4.2.2 Watershed Mapping 4-3
4.2.3 Sample Class Definition 4-3
4.2.4 Soil Sampling 4-4
4.2.5 Sample Analysis 4-4
4.2.6 Database Management 4-4
4.3 OTHER REGIONAL DATASETS 4-4
4.3.1 Atmospheric Deposition : 4-5
4-3.2 Runoff Depth 4-5
4.4 DATA ANALYSIS 4-6
4.4.1 LevelJ Analyses 4-6
4.4.2 Level II Analyses 4-6
4.4.3 Level III Analyses 4-7
4.4.4 Integration of Results 4-8
4.4.5 Use of a Geographic Information System 4-9
5 DATA SOURCES AND DESCRIPTIONS 5-1
5.1 INTRODUCTION 5-1
5.2 STUDY SITE SELECTION 5-1
5.2.1 Site Selection Procedures 5-1
5.2.2 Eastern Lake Survey Phase I Design 5-1
5.2.3 Pilot Stream Survey Design 5-2
5.2.4 DDRP Target Population 5-6
5.2.4.1 Northeast Lake Selection 5-6
5.2.4.2 Southern Blue Ridge Province Stream Selection 5-25
5.2.4.3 Final DDRP Target Populations 5-25
5.3 NSWS LAKE AND STREAM DATA 5-25
5.3.1 Lakes in the Northeast Region 5-25
5.3.1.1 Lake Hydrologic Type 5-25
5.3.1.2 Fall Index Sampling 5-30
5.3.1.3 Chemistry of DDRP Lakes 5-37
IV
-------
CONTENTS (continued) Page
5.3.2 Streams in the Southern Blue Ridge Province Region 5-37
5.3.2.1 Spring Baseflow Index Sampling 5-37
5.3.2.2 Chemistry of DDRP Stream Reaches 5-40
5.4 MAPPING PROCEDURES AND DATABASES 5-40
5.4.1 Northeast Mapping 5-42
5.4.1.1 Soils 5-43
5.4.1.2 Depth to Bedrock 5-49
5.4.1.3 Forest Cover Type 5-51
5.4.1.4 Bedrock Geology 5-51
5.4.1.5 Quality Assurance 5-52
5.4.1.6 Land Use/Wetlands 5-58
5.4.1.7 Geographic Information Systems Data Entry 5-73
5.4.2 Southern Blue Ridge Province Mapping 5-90
5.4.2.1 Soils 5-93
5.4.2.2 Depth to Bedrock 5-97
5.4.2.3 Forest Cover Type/Land use 5-98
5.4.2.4 Bedrock Geology 5-98
5.4.2.5 Drainage 5-98
5.4.2.6 Quality Assurance 5-100
5.4.2.7 Land Use/Wetlands 5-105
5.4.2.8 Geographic Information Systems Data Entry 5-106
5.5 SOIL SAMPLING PROCEDURES AND DATABASES 5-111
5.5.1 Development/Description of Sampling Classes 5-111
5.5.1.1 Rationale/Need for Sampling Classes 5-111
5.5.1.2 Approach Used for Sampling Class Development 5-112
5.5.1.3 Description of Sampling Classes 5-113
5.5.2 Selection of Sampling Sites 5-117
5.5.2.1 Routine Samples 5-117
5.5.2.2 Samples on Special Interest Watersheds 5-122
5.5.3 Soil Sampling 5-122
5.5.3.1 Soil Sampling Procedures 5-122
5.5.3.2 Quality Assurance/Quality Control of Sampling 5-123
5.5.4 Physical and Chemical Analyses 5-124
5.5.4.1 Preparation Laboratories 5-124
5.5.4.2 Analytical Laboratories 5-126
5-5.5 Database Management 5-140
5.5.5.1 Database Structure 5-140
5.5.5.2 Database Operations 5-143
5.5.6 Data Summary 5-148
5.5.6.1 Summary of Sampling Class Data 5-148
5.5.6.2 Cumulative Distribution Functions 5-150
5.6 DEPOSITION DATA 5-150
5.6.1 Time Horizons of Interest 5-161
5.6.1.1 Current Deposition 5-161
5.6.1.2 Future Deposition 5-161
5.6.2 Temporal Resolution 5-161
5.6.2.1 Level I Analyses 5-161
5.6.2.2 Level II Analyses 5-161
5.6.2.3 Level III Analyses 5-163
5.6.3 Data Acquisition/Generation 5-163
5.6.3.1 Level III Analyses - Typical Year Deposition Dataset 5-164
5.6.3.2 Level I and II Analyses - Long-Term Annual Average Deposition Dataset . . . 5-191
5.6.4 Deposition Datasets Used in DDRP Analyses 5-200
-------
CONTENTS (continued) Page
5.7 HYDROLOGIC DATA 5-200
5.7.1 Runoff 5-200
5.7.1.1 Data Sources 5-200
5.7.1.2 Runoff Interpolation Methods 5-203
5.7.1.3 Uncertainty Estimates 5-203
5.7.2 Derived Hvdroloolc Parameters 5-204
5.7.2.1 TOPMODEL 5-204
5.7.2.2 Soil Contact (Darcy's Law) 5-209
5.7.2.3 Mapped Hydrologic Indices 5-211
6 REGIONAL POPULATION ESTIMATION 6-1
6.1 INTRODUCTION 6-1
6.2 PROCEDURE 6-1
6.2.1 Use of Variable Probability Samples 6-1
6.2.2 Estimation Procedures for Population Means 6-2
6.2.3 Estimators of Variance 6-4
6.2.4 Estimator of Cumulative Distribution Function 6-5
6.3 UNCERTAINTY ESTIMATES 6-6
6.4 APPLICABILITY 6-8
7 WATERSHED SULFUR RETENTION 7-1
7.1 INTRODUCTION 7-1
7.2 RETENTION IN LAKES AND WETLANDS 7-2
7.2.1 Introducjign 7-2
7.2.2 Approach 7-4
7.2.3 Results 7-6
7.3 WATERSHED SULFUR RETENTION 7-9
7.3.1 Methods 7-9
7.3.1.1 Input/Output Calculation 7-9
7.3.1.2 Data Sources 7-11
7.3.2 Uncertainty Estimates 7-11
7.3.2.1 Introduction 7-11
7.3.2.2 Individual Variable Uncertainties 7-12
7.3.2.3 Uncertainty Calculation - Monte Carlo Analysis 7-17
7.3.3 Internal Sources of Sulfur 7-19
7.3.3.1 Introduction/Approach 7-19
7.3.3.2 Bedrock Geology 7-20
7.3.3.3 Upper Limit Steady-State Sulfate Concentration 7-24
7.3.4 Results and Discussion 7-29
7.3.4.1 Northeast 7-31
7.3.4.2 Mid-Appalachians 7-41
7.3.4.3 Southern Blue Ridge Province 7-41
7.3.4.4 Conclusions 7-43
8 LEVEL I STATISTICAL ANALYSES 8-1
8.1 INTRODUCTION 8-1
8.1.1 Approach 8-2
8.1.2 Statistical Methods 8-7
8.2 RELATIONSHIPS BETWEEN ATMOSPHERIC DEPOSITION
AND SURFACE WATER CHEMISTRY 8-9
8.2.1 Introduction 8-9
8.2.2 Approach 8-9
8.2.3 Results and Discussion 8-9
8.2.3.1 Northeast 8-9
8.2.3.2 Southern Blue Ridge Province 8-11
8.2.3.3 Summary 8-11
vi
-------
CONTENTS (continued) Page
8.3 DERIVED HYDROLOGIC PARAMETERS 8-13
8.3.1 Soil Contact (Darcv's Law) 8-13
8.3.1.1 Introduction 8-13
8.3.1.2 Results and Discussion 8-18
8.3.2 Geomorphic/HvdrolQQic Parameters 8-21
8.3.2.1 Introduction 8-21
8.3.2.2 Results and Discussion 8-22
8.3.3 TQPMODEL Parameters 8-37
8.3.3.1 Introduction 8-38
8.3.3.2 Results and Discussion 8-41
8.3.3.3 Summary 8-48
8.4 MAPPED BEDROCK GEOLOGY 8-48
8.4.1 DDRP Bedrock Sensitivity Scale 8-50
8.4.2 Results 8-51
8.4.2.1 Sulfate and Percent Retention 8-54
8.4.2.2 Sum of Base Cations, ANC, and pH ^. 8-59
8.4.3 Summary 8-61
8.5 MAPPED LAND USE/VEGETATION 8-62
8.5.1 Introduction 8-62
8.5.2 Data Sources 8-63
8.5.3 Statistical Methods 8-63
8.5.4 Sulfate and Percent Sulfur Retention 8-64
8.5.4.1 Northeast 8-64
8.5.4.2 Southern Blue Ridge Province 8-73
8.5.4.3 Regional Comparisons 8-73
8.5.5 ANC. Ca plus Ma. and oH 8-75
8.5.5.1 Northeast 8-75
8.5.5.2 Southern Blue Ridge Province 8-76
8.5.5.3 Regional Comparisons 8-76
8.5.6 Summary and Conclusions 8-78
8.6 MAPPED SOILS 8-78
8.6.1 Introduction 8-78
8.6.2 Approach 8-79
8.6.3 Sulfate and Sulfur Retention 8-88
8.6.3.1 Northeast 8-88
8.6.3.2 Southern Blue Ridge Province 8-92
8.6.3.3 Regional Comparisons 8-97
8.6.4 ANC. Ca plus Ma. and pH 8-102
8.6.4.1 Northeast 8-109
8.6.4.2 Southern Blue Ridge Province 8-111
8.6.4.3 Regional Comparisons 8-113
8.6.5 Summary and Conclusions 8-113
8.7 ANALYSES OF DEPTH TO BEDROCK 8-113
8.7.1 Introduction 8-113
8.7.2 Approach . 8-113
8.7.3 Sulfate and Percent Sulfur Retention 8-115
8.7.3.1 Northeast 8-115
8.7.3.2 Southern Blue Ridge Province 8-119
8.7.3.3 Comparison of Regions 8-119
8.7.4 ANC. Ca plus Mo and pH 8-119
8.7.4.2 Southern Blue Ridge Province 8-122
8.7.4.3 Comparison of Regions 8-123
8.7.5 Summary and Conclusions 8-123
vii
-------
CONTENTS (continued) Page
8.8 INTEGRATED ANALYSIS OF ALL MAPPED VARIABLES 8-124
8.8.1 Introduction 8-124
8.8.2 Approach 8-124
8.8.3 Sulfate and sulfur retention 8-125
8.8.3.1 Northeast 8-125
8.8.3.2 Southern Blue Ridge Province 8-127
8.8.3.3 Regional Comparisons 8-130
8.8.4 ANC. Ca plus Ma. and pH 8-131
8.8.4.1 Northeast 8-131
8.8.4.2 Southern Blue Ridge Province 8-134
8.8.4.3 Regional Comparisons 8-137
8.8.5 Summary and Conclusions 8-137
8.9 SOIL PHYSICAL AND CHEMICAL CHARACTERISTICS 8-138
8.9.1 Introduction 8-138
8.9.2 Approach 8-138
8.9.2.1 Statistical Methods 8-140
8.9.3 Aggregation of Soil Data 8-143
8.9.3.1 Introduction 8-143
8.9.3.2 Aggregation of Soil Data 8-144
8.9.3.3 Assessment of the DDRP Aggregation Approach 8-145
8.9.3.4 Estimation of Watershed Effect 8-148
8.9.3-5 Evaluation of Watershed Effect 8-149
8.9.4 Regional Soil Characterization 8-155
8.9.5 Sulfate and Sulfur Retention 8-157
8.9.5.1 Northeast 8-157
8.9.5.2 Southern Blue Ridge Province 8-164
8.9.6 Ca plus Ma (SOBCV ANC. and pH 8-165
8.9.6.1 Northeast 8-169
8.9.6.2 Southern Blue Ridge Province 8-170
8.9.7 Evaluation of Alternative Aggregation Schemes 8-170
8.9.8 Summary and Conclusions 8-171
8.9.8-1 Alternative Aggregation Schemes 8-171
8.9.8.2 Sulfate and Sulfur Retention 8-174
8.9.8.3 Ca plus Mg (SOBC), ANC, and pH 8-175
8.9.9 Summary Conclusions 8-175
8.10 EVALUATION OF ASSOCIATIONS BETWEEN WATERSHED ATTRIBUTES
AND SURFACE WATER CHEMISTRY 8-176
8.10.1 Introduction 8-176
8.10.2 Approach 8-176
8.10.3 Regional Characterization of Watershed Attributes 8-177
8.10.3.1 Northeast Subregrons 8-177
8.10.3.2 Northeast and Southern Blue Ridge Providence 8-182
8.10.4 Sulfate and Sulfur Retention 8-182
8.10.4.1 Northeast 8-192
8.10.4.2 Southern Blue Ridge Province 8-193
8.10.5 Ca plus Mg (SOBC). ANC. and pH 8-193
8.10.5.1 Northeast 8-193
8.10.5.2 Southern Blue Ridge Province 8-197
8.10.6 Summary and Conclusions 8-197
8.10.6.1 Sulfate and Sulfur Retention 8-197
8.10.6.2 Ca plus Mg (SOBC), ANC, and pH 8-198
8.10.7 Summary Conclusions 8-198
viii
-------
CONTENTS (continued) Page
9 LEVEL II ANALYSES - SINGLE FACTOR RESPONSE TIME ESTIMATES 9-1
9.1 INTRODUCTION 9-1
9.2 EFFECTS OF SULFATE ADSORPTION ON WATERSHED SULFUR RESPONSE TIME . . 9-2
9.2.1 Introduction 9-2
9.2.2 Section Objectives : 9-4
9.2.3 Approach 9-5
9.2.3.1 Model Description 9-5
9.2.3.2 Data Sources 9-6
9.2.3.3 Model Assumptions and Limitations 9-8
9.2.3.4 Adsorption Data 9-9
9.2.3.5 Evaluation of Aggregated Data and Model Outputs 9-14
9.2.3.6 Target Populations for Model Projections 9-17
9.2.4 Results 9-18
9.2.4.1 Comparison of Northeast and Southern Blue Ridge Province Isotherm
Variables 9-18
9.2.4.2 Model Results - Northeastern United States 9-20
9.2.4.3 Model Results - Southern Blue Ridge Province 9-35
9.2.4.4 Uncertainty Analyses and Alternative Aggregation Approaches 9-51
9.2.4.5 Summary of Results from the Southern Blue Ridge Province 9-59
9.2.5 Summary Comments on Level II Sulfate Analyses 9-62
9.2.6 Conclusions 9-64
9.3 EFFECT OF CATION EXCHANGE AND WEATHERING ON SYSTEM RESPONSE 9-66
9.3.1 Introduction 9-66
9.3.1.1 Level II Hypotheses . 9-67
9.3.1.2 Approach 9-71
9.3.2 Descriptions of Models 9-75
9.3.2.1 Reuss Model 9-75
9.3.2.2 Bloom-Grigal Model 9-94
9.3.3 Model Forecasts 9-103
9.3.3.1 Reuss Model 9-105
9.3.3.2 Bloom-Grigal Model 9-154
9.3.4 Comparison of the Bloom-Griaal and Reuss Models 9-185
9.3.5 Summary and Conclusions 9-196
10 LEVEL III ANALYSES - DYNAMIC WATERSHED MODELLING 10-1
10.1 INTRODUCTION 10-1
10.2 DYNAMIC WATERSHED MODELS 10-3
10.2.1 Enhanced Trickle Down (ETD) Model 10-6
10.2.2 Integrated Lake-Watershed Acidification Study flLWASl Model 10-7
10.2.3 Model of Acidification of Groundwater in Catchments (MAGIC) 10-13
10.3 OPERATIONAL ASSUMPTIONS 10-14
10.4 WATERSHED PRIORITIZATION 10-14
10.4.1 Northeast 10-16
10.4.2 Southern Blue Ridge Province 10-18
10.4.3 Effects of Prioritization on Inclusion Probabilities 10-20
10.5 MODELLING DATASETS 10-20
10.5.1 Meteorological/Deposition Data 10-21
10.5.2 DDRP Runoff Estimation 10-22
10.5.2.1 Annual Runoff 10-22
10.5.2.2 Monthly Runoff : 10-22
10-5.3 Morphometrv 10-24
10.5.4 Soils 10-25
10.5.5 Surface Water Chemistry 10-25
10.5.6 Other Data 10-25
10.5.7 Chloride Imbalance 10-25
10.6 GENERAL APPROACH 10-28
IX
-------
CONTENTS (continued) Page
10.7 MODEL CALIBRATION 10-30
10.7.1 Special Interest Watersheds 10-30
10.7.1.1 Northeast . . - 10-33
10.7.1.2 Southern Blue Ridge Province 10-33
10.7.2 General Calibration Approach 10-34
10.7.3 Calibration of the Enhanced Trickle Down Model 10-35
10.7.4 Calibration of the Integrated Lake-Watershed Acidification Model 10-38
10.7.5 Calibration of the Mgdel of Acidification of Groundwater in Catchments 10-42
10.7.6 Calibration/Confirmation Results 10-44
10.8 MODEL SENSITIVITY ANALYSES 10-49
10.8.1 General Approach 10-50
10.8.2 Sensitivity Results 10-51
10.9 REGIONAL PROJECTIONS REFINEMENT 10-53
10.9.1 Enhanced Trickle Down 10-53
10.9.2 Integrated Lake-Watershed Acidification Study 10-54
10.9.3 Model of Acidification of Groundwater In Catchments 10-54
10.9.4 DDRP Watershed Calibrations 10-56
10.9.4.1 Integrated Lake-Watershed Acidification Study 10-56
10.9.4.2 MAGIC 10-59
10.9.4.3 Southern Blue Ridge Province 10-61
10.10 MODEL PROJECTIONS 10-66
10.10.1 General Approach 10-66
10.10.2 Forecast Uncertainty 10-67
10.10.2.1 Watershed Selection 10-69
10.10.2.2 Uncertainty Estimation Approaches 10-70
10.10.2.3 Relationship Among Approaches 10-74
10.10.2.4 Confidence Intervals 10-76
10.11 POPULATION ESTIMATION AND REGIONAL FORECASTS 10-76
10.11.1 Northeast Regional Projections 10-77
10.11.1.1 Target Population Projections Using MAGIC 10-77
10.11.1.2 Target Population Projections Using MAGIC and ETD 10-91
10.11.1.3 Restricted Target Population Projections Using All Three Models .... 10-113
10.11.2 Southern Blue Ridae Province 10-141
10.11.2.1 Target Population Projections Using MAGIC 10-141
10.11.2.2 Restricted Target Population Projections Using ILWAS and MAGIC ... 10-155
10.11.3 Regional Comparisons 10-174
10.11.3.1 Northeastern Projections of Sulfate Steady State 10-174
10.11.3.2 Southern Blue Ridge Province Projections of Sulfate Steady State .... 10-178
10.11.3.3 ANC and Base Cation Dynamics - 10-178
10.12 DISCUSSION 10-195
10.12.1 Future Pro)ectionsj>f Changes in Acid-Base Surface Water Chemistry 10-195
10.12.2 Rate of Future Change 10-197
10.12.2.1 Northeast 10-197
10.12.1.2. Southern Blue Ridge Province \ 10-202
10.12.3 Uncertainties and Implications for Future Changes in
Surface Water Acid-Base Chemistry 10-204
10.12.3.1 Deposition Inputs 10-205
10.12.3.2 Watershed Processes 10-207
10.13 CONCLUSIONS FROM LEVEL III ANALYSES 10-210
11 SUMMARY OF RESULTS 11-1
1.1 RETENTION OF ATMOSPHERICALLY DEPOSITED SULFUR 11-1
11.1.1 Current Retention 11-1
11.1.2 Projected Retention 11-3
11.2 BASE CATION SUPPLY 11-6
11-2.1 Current Control 11-6
11.2.2 Future Effects 11-7
-------
CONTENTS (continued) Page
11.3 INTEGRATED EFFECTS ON SURFACE WATER ANC 11-8
11.3.1 Northeast Lakes 11-9
11.3.2 Southern Blue Ridae Province 11-17
11.4 SUMMARY DISCUSSION : 11-26
12 REFERENCES 12-1
13 GLOSSARY 13-1
13.1 ABBREVIATIONS AND SYMBOLS 13-1
13.1.1 Abbreviations 13-3
13.1.2 Symbols 13-6
xi
-------
TABLES
TABLE PAGE
1-1 Lakes in the NE Projected to Have ANC Values <0 and <50 pieq L"1 for Constant
and Decreased Sulfur Deposition 1-19
1-2 SBRP Stream Reaches Projected to Have ANC Values <0 and <50 /*eq L for
Constant and Increased Sulfur Deposition 1-22
3-1 Major Rock Forming Minerals and Their Relative Reactivities 3-22
5-1 Sampling Structure for Phase I, Region 1 (Northeast), Eastern Lake Survey 5-4
5-2 Sample Structure for the Direct/Delayed Response Project - Northeastern Sample 5-8
5-3 ANC Group, Lake Identification, ELS-I Phase I ANC, Weight and Inclusion
Probabilities for the Direct/Delayed Response Project Northeast Sample Watersheds ... 5-9
5-4 Lake Identification (ID) and Name, and State and Latitudinal/Longitudinal
Location of the Northeast Sample Watersheds, Sorted by Lake ID 5-13
5-5 Lake Identification (ID) and Name, Sorted by State - Northeast Sample Watersheds ... 5-16
5-6 Stream Identification (ID), Weight, and Inclusion Probabilities for the Southern
Blue Ridge Province Direct/Delayed Response Project Sample Watersheds 5-26
5-7 Stream Identification (ID) and Name, and State and Latitudinal/Longitudinal
Location of the Southern Blue Ridge Province Sample Watersheds, Sorted by Stream ID 5-27
5-8 Stream Identification (ID) and Name, Sorted by State - Southern Blue Ridge
Province Sample Watersheds 5-28
5-9 DDRP Rectification of Northeastern Lakes Classified as 'Seepage" or "Closed"
by the NSWS 5-31
5-10 Depth-to-Bedrock Classes and Corresponding Level of Confidence 5-50
5-11 Interpretation Codes for Northeast Map Overlays - Land Use/Land Cover, Wetlands,
and Beaver Activity 5-59
5-12 Northeast Watersheds Studied for Independent Field Check of Land Use and
Wetland Photointerpretations 5-63
5-13 Chi-Square Test for General Land Use Categories 5-65
5-14 Comparison of Field Check (Matched) General Land Use Determinations with
Office Photointerpretations 5-66
5-15 Chi-Square Test for Detailed Wetland Categories 5-67
5-16 Comparison of Field Check (Matched) Detailed Wetland Determinations with
Office Photointerpretations 5-68
5-17 Comparison of Beaver Dam Number (#), Breached (B) and Unbreached
(U) Status, and Lodges (L), Identified via Field Check and Office Photointerpretation
Methods 5-70
5-18 Aggregated Land Use Data for Northeast Watersheds 5-72
5-19 Watershed No. 1E1062 Soil Mapping Units 5-87
5-20 Land Use Codes Used as Map Symbols 5-99
5-21 Percent Land Use Data for Southern Blue Ridge Province Watersheds 5-107
5-22 Laboratory Analysis of DDRP Soil Samples 5-125
5-23 Analytical Variables Measured in the DDRP Soil Survey 5-127
5-24 Data Quality Objectives for Detectability and Analytical Within-Batch Precision 5-131
5-25 Detection Limits for Contract Requirements, Instrument Readings, and
System-Wide Measurement in the Northeast 5-133
5-26 Detection Limits for the Contract Requirements, Instrument Readings, and
System-wide Measurement in the Southern Blue Ridge Province 5-134
5-27 Attainment of DQO's by the analytical laboratories as determined from blind
audit samples for the Northeast 5-136
5-28 Attainment of DQO's by the Analytical Laboratories as Determined from Blind
Audit Samples for the Southern Blue Ridge Province 5-138
5-29 Quality Assurance and Quality Control Checks Applied to Each Data Batch 5-146
5-30 Medians of Pedon-Aggregated Values of Soil Variables for the DDRP Regions
and Subregions 5-160
xii
-------
TABLES (continued) Page
5-31 Monthly Values of Leaf Area Index (LAI) Used to Apportion Annual Dry Deposition
to Monthly Values 5-176
5-32 Ratios of Coarse-to-Fine Particle Dry Deposition 5-180
5-33 Ratios of Dry Deposition to Wet Deposition for DDRP Study Sites for the
Typical Year (TY) Deposition Dataset 5-182
5-34 Deposition Datasets Used in DDRP Analyses 5-201
5-35 DDRP texture classes and saturated hydraulic conductivity (K) for the NE study systems. .5-206
5-36 SCS slope classifications 5-212
5-37 Mapped and calculated geomorphic parameters collected for the NE study sites 5-215
5-38 Mapped and calculated geomorphic parameters collected for the SBRP study sites 5-219
7-1 Summary of Computed Sulfur Retention by In-lake Reduction for Lake Systems in the
Eastern United States. 7-5
7-2 Intensively Studied Sites Used in Surface Water Chemistry Uncertainty Analysis 7-13
7-3 Summary Statistics on Percent Differences Between Flow-weighted Average Annual
Sulfate Concentration and the Fall/Spring Row-weighted Averages 7-18
7-4 Bedrock Geology Maps Used in the DDRP Internal Sources of Sulfur Bedrock
Geology Analyses 7-21
7-5 Potential for Sulfur Contribution by Geologic Type 7-23
7-6 Summary of Watersheds (by ELS and NSS Subregion) Dropped Due to Suspected
Internal Sources of Sulfur Identified by Steady-State Analysis 7-30
7-7 Percent Sulfur Retention - Summary Statistics by Region . . . . 7-33
7-8 Summary of Sulfur Retention Status and of Watershed Variables Contributing
to Sulfur Retention for 42 Watersheds in the Northeastern United States 7-39
8-1 Surface Water Chemistry and Percent Sulfur Retention Summary Statistics
for the Northeastern DDRP Sample of 145 Lake Watersheds 8-3
8-2 Surface Water Chemistry and Percent Sulfur Retention Summary Statistics
for the DDRP Sample of 35 SBRP Stream Watersheds 8-4
8-3 Summary Statistics for Wet and Dry Deposition on the DDRP Sample
of 145 Northeastern Lake Watersheds 8-5
8-4 Summary Statistics for Wet and Dry Deposition on the DDRP Sample of 35
SBRP Stream Watersheds 8-6
8-5 Results of Regressions Relating Surface Water Chemistry to Atmospheric
Deposition in the Northeast Region (n = 145) 8-10
8-6 Results of Regressions Relating Surface Water Chemistry to Atmospheric
Deposition in the Southern Blue Ridge Province (n = 32) 8-12
8-7 Estimated Population-Weighted Summary Statistics on the Darcy's Law Estimates
of Flow Rate and the Index of Flow Relative to Runoff 8-15
8-8 Estimated Population-Weighted Summary Statistics for Northeastern
Geomorphlc/Hydrologic Parameters 8-23
8-9 Estimated Population-Weighted Summary Statistics for Southern Blue
Ridge Province Hydrologic/Geomorphic Parameters 8-24
8-10 Mapped and Calculated Geomorphic Parameters Collected
for the Northeastern Study Sites (Same as 5-37) 8-25
8-11 Mapped and Calculated Geomorphic Parameters Collected for the
SBRP Study Sites 8-28
8-12 Stratification Based on Sulfur Deposition (Wet and Dry) 8-30
8-13 Results of Stepwise Regression Relating Surface Water Chemistry versus
Geomorphic/Hydrologic Parameters for the Entire NE 8-31
8-14 Stepwise Regression Equations for Surface Water Chemistry and
Hydrologic/Geomorphic Parameters Based on Sulfur Deposition Stratification 8-33
8-15 Results of Stepwise Regression Relating Surface Water Chemistry
and Geomorphic/Hydrologic Parameters for the SBRP 8-34
8-16 Population-Weighted Summary Statistics-for ln(a/KbTanB) for the Northeast 8-39
8-17 Population-Weighted Summary Statistics for ln(a/TanB) for the Southern Blue
Ridge Province 8-40
xiii
-------
TABLES (continued) Page
8-18 Spearman's Correlation Coefficients Between ln(a/KbTanB) and Surface Water Chemistry 8-42
8-19 Pearson's Correlation Coefficients Between ln(a/TanB) and NSS Pilot Chemistry 8-47
8-20 Tabulation of the Generic Bedrock Types Used to Classify the Mapped Units
Identified on State Map Legends 8-52
8-21 Tabulation of the Generic Bedrock Types Used to Classify the Mapped Units
Identified on State Map Legends 8-53
8-22 Regional and Subregional Statistics for the Bedrock Sensitivity Code Variables 8-55
8-23 Results of Regressions of Surface Water Chemistry on Bedrock Sensitivity
Code Statistics and Deposition Estimates for Northeast 8-56
8-24 Results for SBRP of Regressions of Surface Water Chemistry on Bedrock
Sensitivity Code Statistics and Deposition Estimates 8-58
8-25 Land Use and Other Environmental Variables Related to Surface Water
Chemistry of Northeastern Lakes 8-65
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 8-66
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 8-68
8-28 Land Use and Other Environmental Variables Related to Surface Water Chemistry of
Southern Blue Ridge Province Streams 8-69
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 8-70
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 8-71
8-31 Results of Regressions Relating Surface Water Chemistry of Northeastern Lakes to
Land Use and Other Environmental Data 8-72
8-32 Results of Regressions Relating Sulfate and Percent Sulfur Retention of
Southern Blue Ridge Province Streams to Land Use Data 8-74
8-33 Results of Regressions Relating ANC, Ca plus Mg, and pH of Southern
Blue Ridge Province Streams to Land Use Data 8-77
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 8-83
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 8-84
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 Bufferson the DDRP
Sample of 145 NE Lake Watersheds 8-85
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 8-86
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 8-87
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 Candidate Independent Variables . . 8-89
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 Variables 8-93
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 Candidate Independent Variables . 8-96
xiv
-------
TABLES (continued) Page
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 Candidate Independent Variables . . . . ' 8-99
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 Candidate Independent Variables 8-101
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 Candidate Independent Variables 8-104
8-45 Regression Models of Calcium Pius 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 Candidate Independent Variables 8-106
8-46 Regression Models of SOBC in SBRP Streams, Developed Using Deposition,
Mapped Soils (as a Percentage of Watershed Area in Soil Sampling Classes) and
Miscellaneous Land Areas as Candidate independent Variables 8-107
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 Candidate Independent Variables 8-110
8-48 Depth-to-Bedrock Classes for the Northeast and the Southern Blue Ridge Province .... 8-114
8-49 Regional and Subregional Statistics for the Depth-to-Bedrock Classes 8-116
8-50 Results for NE of Regressions of Surface Water Chemistry on Depth-to-Bedrock Classes
and Deposition Estimates 8-118
8-51 Results for SBRP of Regressions of Surface Water Chemistry on Depth-to-Bedrock
Classes and Deposition Estimates 8-120
8-62 Regression Models of Surface Water Sutfate and Sulfur Retention 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 8-126
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 8-128
8-54 Regression Models of Surface Water ANC, Ca 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 . .: 8-132
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 8-135
8-56 Standard Deviations Within and Among Northeast Sampling Classes Estimated
from B Master Horizon Data ; 8-147
8-57 Means and Standard Deviations of Soil Characteristics by Aggregation Method
and Region 8-150
8-58 Population Means and Standard Errors for Selected Variables, by Subregion/
Region and Aggregation (Watershed Adjusted Data) 8-153
8-59 Non-parametric Correlations Between Lake Chemistry Variables and Selected
Soil Properties for the NE DDRP Watersheds 8-158
8-60 Non-parametric Correlations Between Stream Chemistry Variables and Selected
Soil Properties for the SBRP DDRP Watersheds 8-160
8-61 Results of Stepwise Multiple Regressions for DDRP Lake and Stream Sulfate
Concentrations (SO416) Versus Soil Physical and Chemical Properties 8-162
8-62 Results of Stepwise Multiple Regressions for DDRP Watershed Sulfur Retention
(SO4_NRET) Versus Soil Physical and Chemical Properties 8-163
xv
-------
TABLES (continued) Page
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 8-166
8-64 Results of Stepwise Multiple Regressions for DDRP Lake and Stream ANC
(ALKANEW and ALKA11) Versus Soil Physical and Chemical Properties 8-167
8-65 Results of Stepwise Multiple Regressions for DDRP Lake and Stream pH (PHEQ11)
Versus Soil Physical and Chemical Properties 8-168
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 8-172
8-67 Results of Stepwise Multiple Regressions for DDRP Lake and Stream Sulfate
(S0416) Versus Unadjusted and Watershed Adjusted Soil Physical and Chemical
Properties 8-173
8-68 Population Means and Standard Errors for Selected Variables, by Subregion/
Region and Aggregation 8-178
8-69 Non-parametric Correlations Between Lake Chemistry Variables and Selected
Watershed Attributes for the NE DDRP Watersheds 8-183
8-70 Non-parametric Correlations Between Stream Chemistry Variables and Selected
Watershed Attributes for the SBRP DDRP Watersheds 8-187
8-71 Results of Stepwise Multiple Regressions for DDRP Lake and Stream Sulfate
Concentration (SO416) Versus Watershed Attributes 8-190
8-72 Results of Stepwise Multiple Regressions for DDRP Watershed Sulfur Retention
(SO4_NRET) Versus Watershed Attributes 8-191
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 8-194
8-74 Results of Stepwise Multiple Regressions for DDRP Lake and Stream ANC
(ALKA11, ALKANEW) Versus Watershed Attributes 8-195
8-75 Results of Stepwise Multiple Regressions for DDRP Lake and Stream Air
Equilibrated pH (PHEQ11) Versus Watershed Attributes 8-196
9-1 Comparison of Summary Data for Sulfate Adsorption Isotherm Data for Soils in
the Northeastern United States and Southern Blue Ridge Province 9-19
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 9-25
9-3 Summary Statistics for Modelled Changes in Sulfate Concentration, Percent Sulfur
Retention, and Delta Sulfate for Northeast Watersheds Using Typical Year Deposition
Data 9-26
9-4 Comparison of Measured and Modelled Base Year (1985) Sulfate Data for SBRP
Watersheds, Using Long-Term Average Deposition Data 9-38
9-5 Comparison of Modelled Rates of Increase for [SO4 ] in DDRP Watersheds in the
SBRP with Measured Rates of Increase In Watersheds in the Blue Ridge and
Adjoining Appalachians 9-41
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 9-45
9-7 Summary Statistics for Modelled Changes in Sulfate Concentration, Percent Sulfur
Retention, and Delta Sulfate for Watersheds in the Southern Blue Ridge Province, .... 9-46
9-8 Summary Comparison of Watershed Sulfur Status and Model Forecasts
in the Northeastern United States and Southern Blue Ridge Province Using
Typical Year Deposition Data 9-63
9-9 List of the Chemical Species and Reactions Considered Within the Reuss Model
Framework. 9-78
xvi
-------
TABLES (continued) Page
9-10 Effect of pCO2 on Changes Projected to Occur in Surface Water ANC Values at
50 and 100 Years Using the Reuss Model 9-90
9-11 List of Input Data for the Bloom-Grigal Soil Acidification Model 9-104
9-12 Summary Statistics for the Population Estimates ol Current ANC Conditions for
Lakes in the NE Region for Five Different Deposition or Soils Aggregation Schemes .... 9-113
9-13 Descriptive Statistics of the Population Estimates for Changes in Lake Water
ANC for Systems in the NE 9-118
9-14 Summary Statistics Comparing the Projections Regarding Changes In Surface
Water ANC Values Obtained Using Different Soils Aggregation Schemes 9-122
9-15 Summary Statistics of the Differences Between the Population Estimates for
Future ANC Projections Made Using the Constant Level and Ramped Deposition
Scenarios 9-123
9-16 Summary Statistics for the Population Estimates of Current ANC Conditions for
Stream Reaches in the SBRP for Four Different Deposition Scenarios 9-126
9-17 Descriptive Statistics of the Population Estimates for Changes in Stream Reach
ANC Values for Systems in the SBRP 9-128
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 9-133
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 9-138
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 9-139
9-21 Summary Statistics of the Projected Changes in Soil Base Saturations in the
SBRP, Obtained Using the Different Deposition Scenarios 9-147
9-22 Summary Statistics of the Projected Changes in Soil pH in the SBRP,
Obtained Using the Different Deposition Scenarios 9-148
9-23 Comparison of the Changes in Soil Base Saturation and Soil pH that Are
Projected to Occur in the NE and SBRP 9-152
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 9-156
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 9-159
9-26 Bloom-Grigal Model Regional Projections of the Change in Soil pH In the
Northeastern United States 9-163
9:27 Bloom-Grigal Model Regional Projections of the Change in Percent Base Saturation
in the Northeastern United States 9-165
9-28 Bloom-Grigal Model Regional Projections of the Change in Soil pH in the
Northeastern United States : 9-170
9-29 Bloom-Grigal Model Regional Projections for the Change in Percent Base Saturation
In the Northeastern United States 9-172
9-30 Bloom-Grigal Model Regional Projections for the Change in Soil pH in the
Southern Blue Ridge Province 9-178
9-31 Bloom-Grigal Model Regional Projections for the Change in Percent Soil Base
Saturation in the Southern Blue Ridge Province 9-180
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 9-183
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 9-187
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 9-193
xvii
-------
TABLES (continued) Page
10-1 Major Processes Incorporated in the Dynamic Mode) Codes 10-5
10-2 Meteorological Data Required by the Dynamics Model Codes 10-8
10-3 Chemical Constituents in Wet and Dry Deposition Considered
by the MAGIC, ETD, and ILWAS Codes 10-9
10-4 Chemical Constituents Included in Soil Solutions
and Surface Water for the MAGIC, ETD, and ILWAS Codes 10-10
10-5 Definitions of Acid Neutralizing Capacity (ANC) Used by the MAGIC, ETD,
and ILWAS Codes 10-11
10-6 Level III Operational Assumptions 10-15
10-7 Comparison of Calibration/Confirmation RMSE for Woods Lake Among
ETD, ILWAS, and MAGIC Models, with the Standard Error of the Observations 10-45
10-8 Comparison of Calibration/Confirmation RMSE for Panther
Lake Among ETD, ILWAS, and MAGIC Models, with the Standard Error
of the Observations 10-46
10-9 Comparison of Calibration RMSE for Clear Pond Among ETD,
ILWAS, and MAGIC Models, with the Standard Error of the Observations 10-47
10-10 Percent Change in RMSE for MAGIC and ETD for a Ten Percent Change in
Parameter Values. Parameters are Ranked in Descending Order of Sensitivity
from Left to Right 10-52
10-11 Watersheds, by Priority Class, for which Calibration Criteria
Were not Achieved 10-68
10-12 Deposition Variations Used in Input Uncertainty Analyses 10-72
10-13 Target Populations for Modelling Comparisons and Population Attributes 10-78
10-14 Descriptive Statistics of Projected ANC, Sulfate, pH, Calcium Plus Magnesium,
and Percent Sulfur Retention for NE Lakes in Priority Classes A -1 Using
MAGIC for Both Current and Decreased Deposition 10-81
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 10-89
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 10-96
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,
JLWAS, and MAGIC for Both Current and Decreased Deposition 10-117
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 10-146
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 10-159
10-20 Effects of Critical Assumptions on Projected Rates of Change 10-206
11-1 Weighted Median Projected Change in ANC at 50 Years for Northeastern DDRP
Lakes 11-11
11-2 Lakes in the NE Projected to Have ANC Values <0 and <50 ;ieq L"1 for
Constant and Decreased Sulfur Deposition 11-14
11-3 Weighted Median Projected Change in ANC at 50 Years for DDRP SBRP
Stream Reaches 11-19
11-4 SBRP Stream Reaches Projected to Have ANC Values <0 and <50 ^eq L"1 for
Constant and Increased Sulfur Deposition 11-23
xviii
-------
FIGURES
FIGURE PAGE
1-1 Steps of the Direct/Delayed Response Project (DDRP) approach 1-9
2-1 Activities of the Aquatic Effects Research Program within the National Acid
Precipitation Assessment Program 2-4
3-1 Diagram of sulfur cycle in forest ecosystems 3-22
3-2 Diagram of terrestrial base cation cycle 3-18
4-1 Steps of the Direct/Delayed Response Project (DDRP) approach 4-2
5-1 Northeastern subregions and ANC map classes, Eastern Lake Survey Phase I 5-3
5-2 Representation of the point frame sampling procedure for selecting NSS Stage I
reaches 5-5
5-3 DDRP site locations for Subregion 1A 5-19
5-4 DDRP site locations for Subregion 1B 5-20
5-5 DDRP site locations for Subregion 1C 5-21
5-6 DDRP site locations for Subregion 1D 5-22
5-7 DDRP site locations for Subregion 1E 5-23
5-8 DDRP stream reach study sites in the Southern Blue Ridge Province 5-29
5-9 The pH-ANC relationship for (A) lakes of the ELS Phase I sampling in the
Northeast and (B) DDRP study lakes in the Northeast , 5-38
5-10 The pH-ANC relationship for samples with ANC < 400 /*eq L*1 taken at the
downstream nodes of stream reaches sampled in the NSS 5-41
5-11 Location of Northeast field check sites and other DDRP watersheds 5-62
5-12 Example of digitization log sheet 5-81
5-13 Example of attribute entry log sheet 5-82
5-14 Definition of soil sampling classes for the DDRP Soil Survey in the Northeast 5-114
5-15 Definition of soil sampling classes for the DDRP Soil Survey in the Southern
Blue Ridge Province 5-116
5-16 Selection of watersheds for sampling 5-118
5-17 Selection of starting points for sampling 5-119
5-18 Field selection of a sampling point for sampling class on a watershed 5-120
5-19 Major steps and datasets from the DDRP database 5-141
5-20 Calculation percentage of regional or subregional area in each soil sampling 5-149
5-21 Relative areas of sampling classes in the Northeast subregions 5-151
5-22 Relative areas of sampling classes in the entire Northeast and Southern
Blue Region Province 5-152
5-23 Aggregated soil variables for individual pedons in the Northeast 5-153
5-24 Aggregated soil variables for individual pedons in the Southern Blue
Ridge Province 5-155
5-25 Calculation of cumulative distribution function for a soil variable in a
region or subregion 5-157
5-26 Cumulative distribution functions for pedon aggregated soil van'ables for the
Northeast and the Southern Blue Ridge Province 5-158
5-27 Sulfur deposition scenarios for the NE and SBRP for Level II and III Analyses 5-162
5-28 Example of average annual runoff map for 1951-80 5-202
5-29 Flow chart of Darcy's Law soil contact calculation as applied to the DDRP
study sites 5-213
xix
-------
FIGURES (continued) Page
7-1 Estimated percent sulfur retention by in-lake processes in drainage lakes in
ELS Region 1 (northeastern United States) 7-7
7-2 Percent sulfur retention for intensively studied sites in the United States and
Canada relative to the southern extent of the Wisconsinan glaciation 7-10
7-3 Model of flow-weighted average concentration calculations for Biscuit Brook 7-16
7-4 Row chart for the determination of internal sources of sulfur using the
steady-state sulfate concentration 7-26
7-5. Scatter plot of the Monte Carlo calculated standard deviation versus the
calculated mean ISO/!,, 7-28
7-6. Comparison of percent sulfur retention calculated using (A) modrfied-LTA
deposition and (B) modified-LTA deposition adjusted with a 20 percent increase
in dry deposition 7-32
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 7-34
7-8. Supplemental watersheds mapped for special evaluation of sulfur retention 7-36
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 7-42
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 7-44
8-1 Distribution of estimated contact rate using Darcy's Law calculation 8-16
8-2 Distribution of index of contact (yr) using Darcy's Law calculation 8-17
8-3 Scatter plot of ANC versus contact rate calculated using Darcy's Law 8-19
8-4 Scatter plot of ANC versus index of soil contact calculated using Darcy's Law 8-20
8-5 Scatter plot of ANC versus ln(a/KbTanB) 8-43
8-6 Scatter plot of Ca plus Mg versus ln(a/KbTanB) 8-44
8-7 Scatter plot of pH versus ln(a/KbTanB) 8-45
8-8 Data and regression model development flow diagrams 8-81
8-9 Model development procedure 8-141
8-10 Histograms of unadjusted and adjusted watershed means for selected SBRP
soils variables 8-151
8-11 The mean pH ± 2 standard errors for the SBRP watersheds estimated using
the common aggregation (bars) and the watershed effects adjusted aggregation
(lines) illustrate the lack of variation among the common aggregation values 8-152
9-1 Schematic diagram of extended Langmuir isotherm fitted to data points from
laboratory soil analysis 9-12
9-2 Comparison of measured lake (NE) or stream (SBRP) sulfate concentration with
computed soil solution concentration 9-16
9-3 Historic deposition Inputs and modelled output for soils in a representative
watershed in the northeastern United States 9-22
9-4 Schematic of surface water response to changes in sulfur inputs 9-23
9-5 Comparison of measured, modelled and steady-state sulfate for Northeast lake
systems in 1984 9-27
9-6 Projected changes in percent sulfur retention and sulfate concentration for soils
in northeastern lake systems at 10, 20, 50 and 100 years 9-30
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 9-31
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 TY deposition data 9-32
xx
-------
FIGURES (continued) Page
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 9-34
9-10 Historic deposition inputs and modelled output for soils in stream systems in
the Southern Blue Ridge Province 9-36
9-11 Comparison of measured, modelled, and steady-state sulfate for stream systems
in the Southern Blue Ridge Province in 1985 9-39
9-12 Comparison of forecasts based on two sulfur deposition datasets for soils in SBRP
watersheds 9-42
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. . . . 9-44
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 9-47
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 9-48
9-16 Projected time to 95 percent of steady-state sulfur concentration of Southern Blue
Ridge Province stream systems 9-50
9-17 Comparison of model simulation results for DDRP Southern Blue Ridge watersheds. .... 9-53
9-18 Projected base year sulfate concentration with upper and lower bounds for
90 percent confidence intervals for Southern Blue Ridge Province watersheds 9-54
9-19 Projected time to sulfur steady state with upper and lower bounds for 90 percent
confidence intervals in Southern Blue Ridge Province watersheds 9-56
9-20 Effects of data aggregation on simulated watershed sulfur response for soils in
DORP watersheds of the Southern Blue Ridge Province 9-57
9-21 Evaluation of alternate soil aggregation procedures for soils in SBRP watersheds 9-60
9-22 Schematic diagram of the principal process involved in the cycling of base
cations in surficial environments. . . : 9-76
9-23 Plot of the log of the activity of Al vs. soil solution pH for individual soil
samples collected for DDRP 9-83
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 9-86
9-25 Histograms of the (unweighted for the population estimates) projected present-
day ANC values for lakes in the NE 9-87
9-26 Histograms of the (unweighted for the population estimates) projected, present-
day ANC values for lakes in the NE 9-89
9-27 Flow diagram for the one-box Bloom-Grigal soil simulation model 9-97
9-28 Cumulative distribution of projected, present-day ANC values for lakes in the
study population in the NE as projected using Reuss's cation exchange model 9-109
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 9-110
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 9-115
9-31 Cumulative distribution of the projected surface water ANC values projected for
the study population of lakes in 50 years in the NE 9-116
9-33 Schematic illustration of the titration-like behavior displayed by soils in response
to constant loadings of acidic deposition 9-117
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: 9-125
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 9-127
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 9-130
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 9-131
xxi
-------
FIGURES (continued) Page
9-38 Comparison of measured vs. calculated soil pH values for the 580 aggregated
master horizons in the NE 9-136
9-39 Cumulative distribution of projected (a) base saturations and (b) soil pH values for
soils in NE 9-140
9-40 Cumulative distribution of projected (a) base saturations and (b) soil pH values for
soils in the NE 9-141
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 9-143
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 9-144
9-43 Plot of the projected changes in soil base saturations vs 9-145
9-44 Cumulative frequencies of changes in (a) soil base saturation and (b) soil pH
for the population of soils In the SBRP 9-149
9-45 Cumulative frequencies of changes in (a) soil base saturation and (b) soil pH for
the population of soils in the SBRP 9-150
9-46 Cumulative distributions of aggregate initial soil pH and percent base saturation
in the NE and SBRP, with and without organic horizons 9-160
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 9-161
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 4-) deposition
scenarios after 20, 50, and 100 years of LTA, LTA-rbc, and LTA-zbc deposition 9-162
9-49 Regional CDFs of the projected change in the pH of soils on NE lake
watersheds under constant and ramp down (30% *) deposition scenarios
after 20, 50, and 100 years of LTA, LTA-rbc, and LTA-zbc deposition 9-168
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 9-169
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 9-176
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. . . . 9-177
9-53 Cumulative distributions of changes in soil base saturation for the population of
watersheds in the NE 9-188
9-54 Cumulative distributions of changes in soil pH for the population of watersheds
in the NE 9-190
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-Grigal models 9-191
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 9-192
9-57 Cumulative distributions of changes in soil base saturation for the population of
watersheds in the SBRP 9-194
9-58 Distributions of changes in soil pH for the population of watersheds in the SBRP 9-195
10-1 Modelling priority decision tree: Northeast 10-17
10-2 Modelling priority decision tree: Southern Blue Ridge Province 10-19
10-3 Decision tree used to identify watersheds with net chloride export and procedures
for determining chloride imbalance 10-26
10-4 Approach used in performing long-term projections of future changes in surface
water chemistry. 10-29
10-5 Schematic of modelling approach for making long-term projections 10-31
xxii
-------
FIGURES (continued) Page
10-6 Representation of horizontal segmentation of Woods Lake, NY, watershed for
MAGIC and ETD 10-36
10-7 Representation of vertical layers of Woods Lake Basin for ETD 10-37
10-8 Representation of horizontal segmentation of Woods Lake Basin for ILWAS 10-39
10-9 Representation of vertical layers of Woods Lake Basin for ILWAS 10-40
10-10 Representation of vertical layers of Woods Lake, NY, watershed for MAGIC 10-43
10-11 Comparison of population histograms for simulated versus observed (Eastern
Lake Survey Phase I 1984 values) ANC for ILWAS and MAGIC 10-57
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 10-58
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 10-60
10-14 Comparison of population histograms for simulated versus observed (Eastern
Lake Survey Phasee I 1984 values ) ANC and sulfate concentrations for MAGIC,
Priority Classes A -1 10-62
10-15 Comparison of population histograms for simulated versus observed (NSS Pilot
Survey values) ANC, Priority Classes A and B using ILWAS and MAGIC 10-63
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 10-64
10-17 Comparison of population histograms for simulated versus observed (NSS Pilot
Survey values) ANC and sulfate concentrations, Priority Classes A - E using MAGIC . . . 10-65
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 10-75
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 10-79
10-20 pH projections 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 10-84
10-21 Box and whisker plots of ANC distributions at 10-year intervals for NE Priority
Classes A -1 using MAGIC 10-85
10-22 Box and whisker plots of sulfate distributions at 10-year intervals for NE Priority
Classes A -1 using MAGIC 10-86
10-23 Box and whisker plots of pH distributions at 10-year intervals for NE Priority
Classes A -1 using MAGIC 10-87
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 10-90
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. 10-92
10-26 Comparison of MAGIC and ETD projections of ANC for NE lakes, Priority
Classes A - E, under current and decreased deposition 10-93
10-27 Comparison of MAGIC and ETD projections of sulfate concentrations for NE lakes,
Priority Classes A - E, under current and decreased deposition 10-94
10-28 Comparison of MAGIC and ETD projections of pH for NE lakes, Priority Classes
A - E, under current and decreased deposition 10-95
10-29 Comparisons of projected change in ANC under current and decreased
deposition for NE Priority Classes A - E, using ETD and MAGIC 10-99
10-30 Comparisons of projected change in sulfate concentrations under current and
decreased deposition for NE Priority Classes A - E, using ETD and MAGIC 10-100
10-31 Comparisons of projected change in pH under current and decreased deposition
for NE Priority Classes A - E, using ETD and MAGIC 10-101
10-32 Box and whisker plots of ANC distributions projected using ETD in 10-year
intervals for NE lakes, Priority Classes A - E 10-103
xxiii
-------
FIGURES (continued) Page
10-33 Box and whisker plots of sulfate distributions projected using ETD in 10-year
intervals for NE lakes, Priority Classes A - E 10-104
10-34 Box and whisker plots of pH projected using ETD in 10-year intervals for NE lakes,
Priority Classes A - E 10-105
10-35 Box and whisker plots of ANC distributions in 10-year intervals using MAGIC for
NE lakes, Priority Classes A - E 10-106
10-36 Box and whisker plots of sulfate distributions in 10-year Intervals using MAGIC
for NE lakes, Priority Classes A - E 10-107
10-37 Box and whisker plots of pH in 10-year intervals using MAGIC for NE lakes,
Priority Classes A - E 10-108
10-38 ETD ANC distributions at year 10 and year 50 for NE lakes, Priority Classes A -
E, under current and decreased deposition 10-109
10-39 MAGIC ANC distribution at year 10 and year 50 for NE lakes, Priority Classes
A - E, under current and decreased deposition 10-110
10-40 ETD sulfate distributions at year 10 and year 50 for NE lakes, Priority Classes
A - E, under current and decreased deposition 10-111
10-41 MAGIC sulfate distributions at year 10 and year 50 for NE lakes, Priority Classes
A - E, under current and decreased deposition 10-112
10-42 Comparison of ANC projections using ETD, ILWAS, and MAGIC for NE lakes,
Priority Classes A and B, under current and decreased deposition 10-114
10-43 Comparison of sulfate projections using ETD, ILWAS, and MAGIC for NE lakes,
Priority Classes A and B, under current and decreased deposition 10-115
10-44 Comparison of pH projections using ETD, ILWAS, and MAGIC for NE lakes,
Priority Classes A and B, under current and decreased deposition 10-116
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. 10-122
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. 10-123
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. 10-124
10-48 Box and whisker plots of ANC distributions in 10-year intervals projected using
ETD for NE lakes, Priority Classes A and B 10-126
10-49 Box and whisker plots of ANC distributions in 10-year intervals projected using
ILWAS for NE lakes, Priority Classes A and B 10-127
10-50 Box and whisker plots of ANC distributions in 10-year intervals projected using
MAGIC for NE lakes, Priority Classes A and B 10-128
10-51 Box and whisker plots of sulfate distributions in 10-year intervals projected using
ETD for NE lakes, Priority Classes A and B 10-129
10-52 Box and whisker plots of sulfate distributions in 10-year intervals projected using
ILWAS for NE lakes, Priority Classes A and B 10-130
10-53 Box and whisker plots of sulfate distributions in 10-year intervals projected using
MAGIC for NE lakes, Priority Classes A and B 10-131
10-54 Box and whisker plots of pH distributions in 10-year intervals projected using
ETD for NE lakes, Priority Classes A and B 10-132
10-55 Box and whisker plots of pH distributions in 10-year intervals projected using
ILWAS for NE lakes, Priority Classes A and B 10-133
10-56 Box and whisker plots of pH distributions in 10-year intervals projected using
MAGIC for NE lakes, Priority Classes A and B 10-134
10-57 ETD ANC population distributions at year 10 and year 50 for current and
decreased deposition 10-135
10-58 ILWAS ANC population distributions at year 10 and year 50 for current and
decreased deposition 10-136
10-59 MAGIC ANC population distributions at year 10 and year 50 for current and
decreased deposition 10-137
10-60 ETD sulfate population distributions at year 10 and year 50 for current and
decreased deposition 10-138
XXIV
-------
FIGURES (continued) Page
10-61 ILWAS sulfate population distributions at year 10 and year 50 for current and
decreased deposition 10-139
10-62 MAGIC sulfate population distributions at year 10 and year 50 for current and
decreased deposition. 10-140
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. ... 10-142
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 10-143
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. . 10-149
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 10-150
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 10-151
10-68 MAGIC ANC population distributions at year 10 and year 50 for current and
increased deposition, SBRP streams, Priority Classes A - E 10-153
10-69 MAGIC sulfate population distributions at year 10 and year 50 for current and
increased deposition, SBRP streams, Priority Classes A - E 10-154
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 10-156
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 10-157
10-72 Comparison of JLWAS and MAGJC projections for pH at years 0, 20, and 50 for
SBRP streams, Priority Classes A and B, under current and increased deposition 10-158
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 10-164
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 10-165
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 10-166
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 10-167
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 10-168
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 10-169
10-79 ILWAS ANC population distributions at year 10 and year 50 for current and
increased deposition, SBRP Priority Class A and B streams 10-170
10-80 MAGIC ANC population distributions at year 10 and year 50 for current and
Increased deposition, SBRP Priority Class A and B streams. 10-171
10-81 ILWAS sulfate population distributions at year 10 and year 50 for current and
increased deposition, SBRP Priority Class A and B streams 10-172
10-82 MAGIC sulfate population distributions at year 10 and year 50 for current and
increased deposition, SBRP Priority Class A and B streams 10-173
10*83 Comparison of projected sulfate versus sulfate steady-state concentrations using
ETD, ILWAS, and MAGIC for NE lakes 10-175
XXV
-------
FIGURES (continued) Page
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 10-176
10-85 Comparison of projected sulfate concentrations between models for NE lakes
after 50 years under current and decreased deposition 10-177
10-86 Comparison of projected sulfate versus sulfate steady-state concentrations for
SBRP streams using ILWAS arid MAGIC under both current and increased deposition. . . 10-179
10-87 Comparison of projected ANC between models in NE lakes after 50 years under
current and decreased deposition 10-180
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 10-181
10-89 Comparison of pH - ANC relationship for each of the models 10-183
10-90 Comparison of projected pH values between models for NE lakes after 50 years
under current and decreased deposition 10-184
10-9T Comparison of projected changes in calcium and magnesium versus changes in
sulfate using ILWAS and MAGIC for NE lakes 10-185
10-92 Change in median ANC, calcium and magnesium, and sulfate concentrations
projected for NE lakes using MAGIC under current and decreased deposition 10-186
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 10-188
10-94 Comparisons of projected ANC and sulfate concentrations and pH between
ILWAS and MAGIC after 50 years for SBRP streams 10-189
10-95 Comparison of projected MNC and Asulfate relationships in SBRP Priority
Class A and B streams using ILWAS and MAGIC 10-190
10-96 Comparison of projected MNC and Asulfate relationships and ^(calcium and
magnesium) and Asulfate relationships for SBRP Priority Class A - E streams
using MAGIC 10-191
10-97 Comparison of projected A(calcium and magnesium) and Asulfate relationships
for SBRP Priority Class A and B streams using ILWAS and MAGIC 10-193
10-98 Change in median ANC, calcium and magnesium, and sulfate concentrations
projected for SBRP streams under current and increased deposition using MAGIC 10-194
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 10-196
10-100 Comparison of projected MAGIC change in pH versus derived pH after 50
years for NE lakes. 10-201
xxvi
-------
PLATES
PLATE PAGE
1-1 Direct/Delayed Response Project study regions and sites 1-5
1-2 Sulfur retention and wet sulfate deposition for National Surface Water Survey and
National Stream Survey regions in the eastern United States 1-13
1-3 Changes in sulfur retention in the Southern Blue Ridge Province as projected by
MAGIC for constant sulfur deposition 1-14
1-4 Changes in median ANC of northeastern lakes at 50 years as projected by MAGIC . . 1-18
1-5 Changes In median ANC of Southern Blue Ridge Province stream reaches at 50
years as projected by MAGIC 1-21
2-1 Direct/Delayed Response Project study regions and sites 2-5
5-1. ANC of DDRP lakes by ANC group 5-24
5-2 Final DDRP classification of lake hydrologic type - Subregion 1A 5-32
5-3 Final DDRP classification of lake hydrologic type - Subregion 1B 5-33
5-4 Final DDRP classification of lake hydrologic type - Subregion 1C 5-34
5-5 Final DDRP classification of lake hydrologic type - Subregion ID 5-35
5-6 Final DDRP classification of lake hydrologic type - Subregion 1E 5-36
5-7 Example of watershed soil map (including pedon site location) 5-75
5-8 Example of watershed vegetation map 5-76
5-9 Example of depth-to-bedrock map 5-77
5-10 Example of watershed land use map 5-78
5-11 Example of watershed geology map 5-79
5-12 Example of 40-ft contour delineations on a 15' topographic map 5-89
5-13 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 5-91
5-14 ADS and NCDC sites linked with DDRP study sites for NE Subregion 1A. 5-166
5-15 ADS and NCDC sites linked with DDRP study sites for NE Subregion 1B 5-167
5-16 ADS and NCDC sites linked with DDRP study sites for NE Subregion 1C 5-168
5-17 ADS and NCDC sites linked with DDRP study sites for NE Subregion 1D 5-169
5-18 ADS and NCDC sites linked with DDRP study sites for NE Subregion 1E 5-170
5-19 ADS and NCDC sites linked with DDRP study sites for the SBRP 5-171
5-20 DDRP study sites relative to distance from Atlantic Coast (<10 km, 10-50 km, >50 km). 5-178
5-21 Pattern of typical year sulfate deposition for the DDRP NE study sites 5-184
5-22 Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1A. ... 5-185
5-23 Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1B. ... 5-186
5-24 Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1C. ... 5-187
5-25 Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1D. ... 5-188
5-26 Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1E. ... 5-189
5-27 Pattern of typical year sulfate deposition for the DDRP SBRP study sites 5-190
5-28 Pattern of LTA sulfate deposition for the DDRP NE study sites 5-193
5-29 Pattern of LTA sulfate deposition for the DDRP study sites in Subregion 1A 5-194
5-30 Pattern of LTA sulfate deposition for the DDRP study sites in Subregion 18 5-195
5-31 Pattern of LTA sulfate deposition for the DDRP study sites in Subregion 1C 5-196
5-32 Pattern of LTA sulfate deposition for the DDRP study sites in Subregion 1D 5-197
5-33 Pattern of LTA sulfate deposition for the DDRP study sites in Subregion 1E 5-198
5-34 Pattern of LTA sulfate deposition for the DDRP SBRP study sites 5-199
xxvii
-------
PLATES (continued) Page
7-1 Median percent sulfur retention by NSWS Subregion 7-35
7-2 Regional percent sulfur retention by major land resource area (MLRA) based on
target populations (ELS and NSS sites) 7-45
11-1 Sulfur retention and wet sulfate deposition for National Surface Water Survey subregions
in the eastern United States 11-2
11-2 Changes in sulfur retention In the Southern Blue Ridge Province as
projected by MAGIC for constant sulfur deposition 11-5
11-3 Changes in median ANC of northeastern lakes at 50 years as projected by MAGIC . . . 11-10
11-4 ANCs of northeastern lakes versus time, as projected by MAGIC for
constant sulfur deposition. 11-12
11-5 ANCs of northeastern lakes versus time, as projected by MAGIC for
decreased sulfur deposition 11-13
11-6 Changes in median pH of northeastern lakes at 50 years as projected by MAGIC ... 11-16
11-7 Changes in median ANC of Southern Blue Ridge Province stream reaches at 50
years as projected by MAGIC 11-18
11-8 ANCs of Southern Blue Ridge Province stream reaches versus time, as projected
by MAGIC for constant sulfur deposition 11-21
11-9 ANCs of Southern Blue Ridge Province stream reaches versus time, as projected
by MAGIC for increased sulfur deposition 11-22
11-10 Changes in pH of SBRP stream reaches as projected by MAGIC 11-24
11-11 Changes in pH of SBRP stream reaches as projected by ILWAS 11-25
xxviii
-------
PRIMARY CONTRIBUTORS TO THE DORP 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
Q. 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.DA 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. Rochelie, 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. Rochelie, 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
xxix
-------
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. Uff, Utah State University
P. W. Shaffer, NSI Technology Services Corp.
Section 10: Level HI Dynamic Watershed Models
K. W. Thornton, FTN & Associates, Ltd.
D. L Stevens, Eastern Oregon State University
M. ft. Church, U.S. Envfronmentaf Protection Agency
C. I. Uff, 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.
Contributors to this section listed alphabetically
2 Beginning on this line, remaining contributors listed alphabetically
XXX
-------
ACKNOWLEDGMENTS
The performance of this portion of the Direct/Delayed Response Project (DORP) 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 DORP and Lee Thomas showed
a continued and very patient interest in seeing that tt 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.
XXXI
-------
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.
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
xxxii
-------
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 (Qemson University), and Dave Litzke and
Marilew Bartling (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 Remodel
(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
XXXIII
-------
the SBRP. Rod Slagle (LESC) served as the DORP 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.
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
qualfty 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
Walthall 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
xxxrv
-------
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 Seflkop 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
Waide of the USDA Forest Service, David Lam of the National Water Research Institute (Burlington,
Ontario), Nils Christophersen of the Institute of Hydrology (Wailingford 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
xxxv
-------
and Phi! 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 DORP. 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 (USDA Forest Service), 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 (USDA Forest Service), 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
-------
PREFACE TO THE EXECUTIVE SUMMARY
This Executive Summary contains both summary results of Project analyses and overview
information on the Project background and approach. Those readers wishing a synopsis only of major
Project results may turn directly to Section 1.4. Because of the complexity of design and approach of
this Project, however, we encourage readers to review Sections 1.1 through 1.3 of this Executive
Summary.
M. Bobbins Church, Technical Director
Direct/Delayed Response Project
-------
SECTION 1
EXECUTIVE SUMMARY
1.1 INTRODUCTION
1.1.1 Project Background
Much scientific interest and public debate surround the effects of acidic deposition on freshwater
ecosystems (e.g., Schindler, 1988; Mohnen, 1988). A comprehensive chemical survey (the National
Surface Water Survey - NSWS) of the lakes and streams of the United States considered to be most
vulnerable to acidic deposition (i.e., those with the lowest acid neutralizing capacity or ANC) was recently
completed by the U.S. Environmental Protection Agency (EPA) (LJnthurst et al., 1986a; Kaufmann et a!.,
1988). Analysis of these and other lake and stream chemistry data, together with data on temporal and
spatial patterns of atmospheric deposition, indicates that long-term deposition of sulfur-containing
compounds originating from the combustion of fossil fuels has acidified (i.e., decreased the ANC of) some
surface waters in eastern North America (Altshuller and (Jnthurst, 1984; NAS, 1986; Sullivan et al., I988b;
Neary and Dillon, 1988; Asbury et al., 1989). Transport of mobile anions (primarily sulfate) through
watershed soils and closely associated cation leaching are the most widely accepted mechanisms of this
acidification process (Seip, 1980; Galloway et al., I983a; Driscoil and Newton, 1985; Church and Turner,
1986). in addition, acidic deposition apparently has shifted the nature of some very low ANC or naturally
acidic surface waters in the Northeast from organic acid "dominance" to mineral acid "dominance*
(Driscoil et al., 1988; Driscoil et al., 1989a). This process is, perhaps, best explained as the effective
titration of naturally occurring humic substances by suffuric acid deposition (Krug and Frink, 1983; Krug
et al., 1985; Krug, 1989). In both cases, the net effect of atmospheric deposition of sulfuric acid on
surface water chemistry is a shift toward aquatic systems more dominated by mineral acidity and more
likely to contain inorganic forms of aluminum, which are toxic to aquatic organisms.
Given that acidification of some surface waters has occurred, critical scientific and policy questions
focus on whether acidification is continuing in the regions of concern, whether it is just beginning in other
regions, how extensive effects might become, and over what time scales effects might occur. EPA is
examining these questions through the activities of the Direct/Delayed Response Project (DDRP) (Church
1-2
-------
and Turner, 1986; Church, in press). The Project was begun in 1984 at the specific request of the EPA
Administrator following a meeting of the Panel on Processes of Lake Acidification of the National Academy
of Sciences. Principal among the conclusions of the Panel was that atmospheric deposition of sulfur-
containing compounds is the major source of long-term surface water acidification in eastern North
America (MAS, 1984). The Panel also debated at length the dynamic aspects of the acidification process.
The DORP was designed to focus on the topic of acidification dynamics and draws its name from
consideration of whether acidification might be immediate (or immediately proportional to levels of
deposition) (i.e., "direct") or whether it would lag in time (i.e., be "delayed") because of edaphic
characteristics. A compilation and discussion of the processes of long-term surface water acidification
and methods for its Investigation were presented by Church and Turner (1986) at the beginning of the
Project. A relatively brief and more current discussion of processes relevant to this Project is presented
in Section 3 of this report.
Although recent research has indicated the potential importance of deposition of nitrogen-
containing compounds to both the episodic (Galloway et al., 1987; Driscoll et al., 1987a) and long-term
(Henriksen and Brakke, 1988) acidification of surface water, the ODRP does not address these effects.
Such effects are the focus of developing or ongoing research within EPA's Aquatic Effects Research
Program.
1.1.2 Primary Objectives
The DDRP has four technical objectives related to atmospheric/terrestrial/aquatic interactions:
(1) to describe the regional variability of soil and watershed characteristics,
(2) to determine which soil and watershed characteristics are most strongly related to surface
water chemistry,
(3) to estimate the relative importance of key watershed processes in moderating regional effects
of acidic deposition, and
(4) to classify a sample of watersheds with regard to their response characteristics to inputs of
acidic deposition and to extrapolate the results from this sample of watersheds to the study
regions.
The fourth objective is the critical "bottom line" of the Project.
1-3
-------
It was never the intent of the DDRP to serve as a "research" project to investigate exact
mechanisms and processes of surface water acidification. Rather, the principal mandate of the Project
was to make regional projections of future effects of sulfur deposition on long-term surface water
chemistry based on the best available data and most widely accepted hypotheses of the acidification
process. In-depth investigations Into processes of soil and surface water acidification are being
conducted as part of other projects within the National Acid Precipitation Assessment Program.
1.1.3 Study Regions
The Project focuses on three regions of the eastern United States where low ANC surface waters
are located and where levels of atmospheric deposition (relative to other U.S. regions) are greatest: (1)
the Northeast (NE), (2) upland areas of the Mid-Atlantic (referred to here as the Mid-Appalachian Region),
and (3) the mountainous section of the Southeast called the Southern Blue Ridge Province (SBRP) (Plate
1-1). Initiation of the Project depended on the availability of the regional surface water chemistry data
of the NSWS. Thus, the Project focused Its initial work on the lake resources of the NE (L'nthurst et al.,
1986a) and the stream resources of the SBRP (Messer et al., 1986a). The results for these two regions
are presented in this report. Complete results of subsequent work In the Mid-Appalachian Region will be
reported at a later date.
1.1.4 Time Frames of Concern
The DDRP focuses on the potential effects of acidic deposition on surface water ANC at key annual
"Index" periods. These index periods were defined by NSWS sampling periods (i.e., fall period of
complete mixing for lakes and spring baseflow for streams - see Section 5.3). The primary time horizon
for DDRP analyses is 50 years. This horizon was selected on the basis of the projected lifetimes of
existing power plants and the potential implementation of additional emissions controls relative to those
lifetimes. Where possible and reasonable, some time-dependent analyses are extended beyond this 50-
year horizon to better evaluate process rates and changes and potential future effects.
1-4
-------
Plate 1-1. Direct/Delayed Response Project study regions and sites.
1-5
-------
DDRP STUDY REGIONS
Northeast
- .
Mid-Appalachian
Region
DDRP Lake Study Sites
ODRP" Stream Study Sites
Southern Blue Ridge
Province
-------
1.2 PROCESSES OF ACIDIFICATION
As discussed in Section 1.1, the NAS Panel identified (1) the retention of deposited sulfur within
watersheds and (2) the supply of base cations from watersheds to surface waters as the most important
watershed processes affecting or mediating long-term surface water acidification (NAS, 1984). These
processes, thus, have become the focus of the DORP. Factors other than sulfur retention and base
cation supply affect surface water acidification, but were either deemed by the Panel to be relatively less
important in long-term acidification or could not be addressed completely within the scope of the DORP
due to time, budgetary, or logistical constraints. Several of these alternative factors are discussed briefly
in Section 3.1 of this report.
1.2.1 Sulfur Retention
During the past decade there has been an increased recognition that surface water acidification
is controlled not only by rates of hydrogen ion deposition, but also by the mobility of associated anions
through the ecosystem. Galloway et at. (1983a) and the 1984 NAS Panel identified controls on anion
mobility, specifically on sulfate adsorption, as one of the two dominant variables affecting the rate and
extent of surface water acidification by atmospheric deposition of mineral acids.
Almost three decades ago, Nye and Greenland (1960) recognized the importance of anions as
"carriers'1 for cations in solution. The "mobile anion" paradigm they proposed [more recently applied to
surface water acidification (Johnson and Cole, 1980; Seip, 1980)] suggests that a variety of processes
act more or less independently to control the concentrations of individual anions in solution, whereas
exchange and weathering processes control the relative quantities of cations. Controls on, and changes
in, anion mobility can thus be viewed as the proximate controls on rates of cation leaching from soils
and, coupled with rates of cation resupply processes, on surface water acidification.
Within the DDRP the primary issue with regard to anion mobility lies in forecasting temporal
changes in dissolved sulfate. Sulfur retention processes are discussed further in Section 3.3.
1-6
-------
1-2-2 Base Cation Supply
The MAS Panel identified rates of base cation supply from watersheds as the second dominant
factor determining the rate and ultimate acidification of surface waters by acidic deposition. Supply of
base cations occurs principally from mineral weathering (as the 'original' source) and cation exchange
In soils. The exchange of cations from the soil complex to the soil solution is a rapid process whereas
the supply of base cations from mineral weathering to the exchange complex proceeds much more
slowly. The balance between these rates and the rate of cation leaching by mobile anions is a critical
factor in determining the rate of soil and surface water acidification. Mineral weathering and cation
exchange are discussed further in Section 3.4. Projections of rates of cation leaching from the exchange
complex are presented in Section 9.3 and are incorporated in watershed modelling studies presented in
Section 10.
1.3 GENERAL APPROACH
As H.B.N. Hynes (1975) once noted, "We must not divorce the stream from its valley in our
thoughts at any time. If we do we lose touch with reality.* Although surface waters can be affected by
acidic deposition originating from emissions many miles distant, the concept of the watershed as a unit
is critical in understanding current and future aquatic effects. Indeed, for drainage lake and reservoir
systems in the Northeast, Upper Midwest, and Southern Blue Ridge Province, most ANC production
occurs as a result of biogeochemical processes within the surrounding watershed (Section 7.2; Shaffer
et al., 1968; Shaffer and Church, 1989).
Because of the importance of watershed processes (especially those occurring In soils) in
determining future aquatic effects, new data on these processes and on related soil pools and capacities
were required. Initially, we considered using existing regional soils data for the DDRP analyses. Existing
soils databases, however, were limited with respect to their application to address surface water
acidification issues. First, such data are available primarily from lowland agricultural regions, whereas
surface water acidification occurs principally in relatively undisturbed upland systems. Second, such
databases generally do not include a number of key variables relevant to soil chemical interactions with
acidic deposition. We subsequently decided that a new regional soils database was required for the
1-7
-------
Project, thus necessitating a major soil survey (Sections 5.1 - 5.5; also see Lee et al., I989a). We further
concluded that this survey should allow the specific soils (and specific soil types) to be linked with the
existing NSWS databases that describe the chemistry of low ANC lakes and streams. Accordingly, we
adopted the approach outlined in this section and illustrated in Figure 1-1.
1.3.1 Soil Survey
DDRP watersheds were selected as a high interest subset of lake and stream systems surveyed
in the NSWS [for details see Section 5.2 and Lee et al. (I989a)j. The watersheds were chosen as
probability samples to ensure that results could be extrapolated to a specified target population (see
Section 6).
Maps of soils, vegetation, land use, and depth to bedrock were prepared for each DDRP watershed
by the USDA Soil Conservation Service (SCS) (see Section 5.4). Soil sample classes were defined for
each DDRP region, and soils selected from these classes were sampled and analyzed for physical and
chemical characteristics. Soils were aggregated within sampling classes to develop characterizations (e.g.,
dass means and variances) that were used to 'rebuild" or represent (e.g., by mass or area weighting)
the characteristics of study watersheds. Details of the sample class selection, sampling, and soil analysis
are provided in Section 5.5.
1.3.2 Other Regional Datasets
The regional nature of the Project required estimates of precipitation, atmospheric deposition (wet
and dry), and surface water runoff (as runoff depth) developed in a standardized manner across the
eastern United States. Study sites for the DDRP were selected statistically, and most sites had no direct
information for deposition and runoff. The development of these datasets for the DDRP is described in
Sections 5.6 and 5.7, respectively.
1-8
-------
Watershed Selection
Watershed Mapping
Development of Soil *
Sampling Classes
Soil Sampling and
Field Measurements
Soil Preparation
Chemical/Physical
Laboratory Analysis
Data Analysis
Supporting Regional
Datasets
Database Management
Figure 1-1. Steps of the Direct/Delayed Response Project (DDRP) approach. Asterisks denote
steps that received significant support from Geographic Information Systems (GlS)-based activities
(Campbell and Church, 1989; Campbell et al., in press).
1-9
-------
1.3.3 Scenarios of Atmospheric Deposition
The major question driving the DORP concerns the response of surface water chemistry to
atmospheric deposition in the future. Within the DORP we were requested by the Agency's Office of Air
and Radiation to evaluate two sulfur deposition scenarios for each study region. The first deposition
scenario for each region was that of constant deposition at current levels. For the Northeast, the second
scenario was for sulfur deposition to remain constant at current levels for 10 years, then to ramp down
for 15 years to a level 30 percent below current and to remain at that level. For the Southern Blue Ridge
Province, the second scenario was for sulfur deposition to remain constant at current levels for 10 years,
then to ramp up for 15 years to a level 20 percent above current and to remain at that level.
1.3.4 Data Analysis
A variety of complementary data analyses were performed within the project (see Section 4.4 for
more details). The most basic of these analyses is the statistical evaluation of interrelationships among
atmospheric deposition, mapped watershed characteristics, soil chemistry, and current surface water
chemistry. The principal goal of these analyses is to verify that the processes and relationships
incorporated in the subsequent modelling analyses reasonably represent the systems under study. The
results of these statistical analyses are presented in Section 8.
Watershed retention of atmospherically deposited sulfur is an important consideration within the
Project. Current regional retention is evaluated in Section 7, and the dynamics of retention via soil sulfate
adsorption are considered in Section 9. Also considered in Section 9 are "single-factor" models (Bloom
and Grigal, 1985; Reuss and Johnson 1985,1986) of the influence of acidic deposition on the supply of
base cations from soils to surface waters. The purpose of this modelling is to evaluate the potential
relative importance of cation exchange as a process mediating surface water acidification.
Watershed models are used in the DDRP to project future integrated effects of atmospheric sulfur
deposition on surface water chemistry. Three models specifically developed to investigate the effects of
acidic deposition on watersheds and surface waters are being applied: (1) the Model of Acidification of
Qroundwater in Catchments (MAGIC) (Cosby et al., 1985a,b; 1986a,b), (2) the Enhanced Trickle Down
MO
-------
(ETD) Model (Lee, 1987; Nikolaidis et al., 1988; Schnoor et al., 1986b); and (3) the Integrated
Lake-Watershed Acidification Study (ILWAS) Model (Chen et al., 1983; Gherini et a!., 1985). The three
models are being run using common datasets for forcing functions (e.g., rainfall, runoff, atmospheric
deposition) and data aggregated from the DDRP soils database for state variables (e.g., soil physical and
chemical variables). Projections of changes in annual average surface water chemistry are being made
for each region for at least 50 years for the two scenarios of atmospheric sulfur deposition described in
Section 1.3.3. Results of these modelling analyses are presented In Section 10.
1.4 RESULTS
This section presents an overview of the results of the DDRP analyses. DDRP statistical analyses
(see Section 8) of the interrelationships among deposition, edaphic factors, and surface water chemistry
generally supported the postulated relationships incorporated into both the single factor models (for
sulfate adsorption and cation supply) and the integrated watershed models. For example, soil depth, soil
chemical characteristics, and watershed hydrology factors will appear as important explanatory variables
in the regressions that we performed. Additionally, wetlands in northeastern watersheds appear to have
an important role in influencing sulfur dynamics (see Sections 7 and 8). Wetland effects are not explicitly
represented in the integrated watershed models. Atmospheric deposition is an important explanatory
variable for current surface water chemistry (especially sulfate concentrations) in northeastern lakes but
not for chemistry of SBRP stream reaches. In both regions, watershed disturbances, especially
agricultural activities, play important roles in affecting surface water chemistry and in masking
interrelationships with acidic deposition.
1.4.1 Retention of Atmospherically Deposited Sulfur
1.4.1.1 Current Retention
At present (for watersheds not having apparent significant internal sources of sulfur; see Section
7), net retention of atmospherically deposited sulfur appears to be approximately at steady state (i.e.,
inputs equal outputs) in the NE. Median net retention is about 75 percent in the SBRP. These
observations are qualitatively consistent with theory (Galloway et al., 1983a).
1-11
-------
The Mid-Appalachian Region is a zone of transition between the NE and SBRP in terms of observed
current sulfur retention. Because of the similarities between soils in this region and the SBRP, it is
possible that this region at one time retained as much of the elevated sulfur deposition as is now evident
in the SBRP (i.e., 70 - 80 percent). It is also possible that sulfur deposition has decreased this retention
[perhaps very dramatically in the westernmost area (Subregion 2Cn of the National Stream Survey), which
now has median percent sulfur retention of only 3 percent (Plate 1-2)] and has ted to the low ANC and
acidic stream reaches (excluding stream reaches affected by acid mine drainage) identified there by the
National Stream Survey (Kaufmann et al., 1988). To further address this issue, in-depth soil sampling and
analyses are being conducted in the Mid-Appalachian Region as part of the DDRP.
1.4.1.2 Projected Retention
Projections of sulfur retention were performed for the deposition scenarios described previously.
Results discussed here are from MAGIC (as are discussions in Section 1.4.3 on projected changes In
surface water ANC). Northeastern watersheds are projected to respond relatively rapidly (I.e., with a lag
of 10 - 20 years) to changes in sulfur deposition. For the scenario of constant deposition, the median
sulfate concentration in northeastern lakes is projected to decrease approximately 10 jteq L*1 over the
next 50 years. Under the scenario of decreased sulfur deposition, the projected decrease in median
sulfate concentration is roughly 40 Meq L*1 over the next 50 years.
Responses are projected to be slower but much more dramatic in the SBRP. Under the constant
deposition scenario, the percent sulfur retention Is projected by MAGIC to decrease to less than 50
percent in 20 years and to less than 30 percent in 50 years (Plate 1-3). The response is very similar
under the increased deposition scenario, in terms of percent sulfur retention. These results correspond
to an increase in median sulfate concentration of roughly 38 jueq L'1 at 50 years for the constant
deposition scenario and 55 jueq L*1 at 50 years for the increased deposition scenario. Such changes
will be accompanied by decreases in surface water ANC to an extent dependent upon the relative
leaching of acids and base cations from watershed soils.
1-12
-------
Plate 1-2. Sulfur retention and wet sulfate deposition for National Surface Water Survey and
National Stream Survey regions in the eastern United States.
1-13
-------
NSWS SUBREGIONS
MEDIAN % SULFUR RETENTION
AND WET SULFATE DEPOSITION
2-25
MEDIAN PERCENT
SULFUR RETENTION
|f 0 - 20
H 20 - 40
40 - 60
60 - 80
80 - 100
Average Annual
Wei Sulfate > 2.75-
Deposition (g nf2 yr~')* j.o(K
3.25
Eastern lake Sumy
2.50.
2.25-^.
2.00-
-2.25
lledign
Subrijitn I {tiniti'tn
H
IB
1C
ID
IE
-14
-7
-J
-12
*2.00
Hotioofll Strum Surrey
id ion
Sabrefion I Retention
2Cn
2Bn
18
U
2As
3A
J
40
34
SO
?5
78
'Deposition for 1980 - 1984
(A. Olsen. Personal Communication)
-------
Plate 1-3. Changes in sulfur retention in the Southern Blue Ridge Province as projected by MAGIC
for constant sulfur deposition (see Section 1.3.4 for definition of deposition scenarios used).
1-14
-------
% SULFUR RETENTION
Model = MAGIC
Deposition = Constant
?
1
1
_ };« Ouorlile +
(1.5 i Interquartile Konje)
5
I
3rd Ouortile
Dean
Median
1st 'J'jorlife
111 Quortile - ..
(1,5 i Inltrqumlile Range)
0 • KSS Samp I.
1 Kol lo Htttt ttlttmt tol»«.
-------
1.4.2 Base Cation Supply
1.4.2.1 Current Control
•s
Base cations are supplied from watersheds to surface waters by two processes acting in concert.
The initial source is mineral weathering, which is a 'slow* process that supplies base cations to the soil
exchange complex. Equilibrium between the exchange complex and soil water (and thus waters delivered
to lakes and streams) is reached quickly. Inasmuch as current rates of acidic deposition in the eastern
United States are unlikely to lead to significant decreases in soil pH, weathering rates are likely to
increase only negligibly due to this effect. If weathering supplies base cations to surface waters at rates
equal to or greater than rates of acid anion deposition, then systems are relatively "protected". If
weathering rates are low and exchange dominates base cation supply rates, then the rate of depletion
of the exchange complex becomes very important in determining rates of surface water acidification. Our
analyses indicate that surface waters with ANC >100 /ieq L'1 are not explained by the cation exchange
model of Reuss and Johnson (1986); thus, ANC generation appears to be dominated by weathering in
these systems and they presumably are relatively protected against loss of ANC (Section 9). Surface
waters with ANC <100 Ateq L'1 are likely controlled by a mix of weathering and cation exchange but the
exact proportion of the mix is very difficult to determine.
1.4.2.2 Future Effects
In general, applying the model of Reuss and Johnson (1986), we performed a "worst-case" analysis
by assuming that the supply of base cations was totally controlled by cation exchange. This analysis
indicated that depletion of base cations from the exchange complex would occur under the sulfur
deposition scenarios simulated. The effect on surface water ANC was initially slight but was not
negligible. The magnitude of soil base cation depletion was projected to accelerate in the future. At
current levels of deposition, about 15 percent of the lakes in the ELS target population are potentially
susceptible to significant depletion of exchangeable cations and, thus, depletion of associated surface
water ANC. The greatest portion of such changes are projected to occur on a time scale of about 50
years. In the SBRP, a greater percentage of systems are projected to be susceptible to adverse effects,
but at longer time scales (I.e., about 100 years) than in the NE.
1-15
-------
In general, effects of base cation depletion would be superimposed upon effects resulting from
changes in sulfate mobility in soils. The combined effects were simulated using the integrated watershed
models and are presented in the next section.
1.4.3 Integrated Effects on Surface Water ANC
The three watershed models were used to project the Integrated watershed and surface water
responses to the sulfur deposition scenarios. Results from MAGIC are presented here because this model
was successfully calibrated to the largest number of watershed systems in the two regions (i.e., 123 of
the 145 DDRP sample watersheds, representing a target population of 3,227 systems In the NE; and 30
of the 35 ODRP sample watersheds, representing a target population of 1,323 stream reaches in the
SBRP). Results among the models were generally very comparable (see Sections 10 and 11). For
example, for northeastern lakes with ANC < 25 jxeq L~1, the three models projected changes In median
ANC (under the decreased deposition scenario) within 3 /ieq L"1.
As discussed in Section 10, the watershed modelling analyses make use of watershed soil
representations as aggregated from the DORP Soil Survey. Because the focus of the DORP is on
regional characteristics and responses, soils data were gathered and aggregated so as to capture the
most important central tendencies of the study systems. As a result, extremes of individual watershed
responses probably are not fully captured in the analyses presented here (see Section 8 for further
discussion). Those systems that are projected to respond to the greatest extent or most quickly to
current or altered levels of sulfur deposition might, in fact, respond even more extensively or more quickly
than indicated here. This possibility should be kept in mind when reviewing the simulation results
presented in this section.
1.4.3.1 Northeast Lakes
Results of the projections for both deposition scenarios are given in Plate 1-4 and Table 1-1. Plate
1-4 illustrates the projected change in the median ANC at 50 years for lakes classified into four ANC
groups (i.e., <0 jueq L*1, 0-25 /ieq L"1, 25-100 /ieq L"1, and 100-400 jieq L"1). These projections indicate
a generally very slight decline in ANC over the 50-year period under the current deposition scenario
1-16
-------
and an increase of roughly 5-15 /ieq L*1 in ANC for ail groups under the decreased sulfur deposition
scenario. Plate 1-4 shows the changes projected by MAGIC. Changes projected by the ETD and ILWAS
models are quite comparable.
Table 1-1 presents the population estimates (with 95 percent confidence intervals) of northeastern
lakes having values of ANC <0 /ieq L*1 and <50 /ieq L'1 at 20 and 50 years as projected by MAGIC
for the two deposition scenarios. The ANC = 0 jueq L'1 value is used to define acidic systems, and the
ANC value of 50 /ieq L"1 (for index values as sampled in the NSWS, see Section 5.3) has recently been
suggested as useful in approximating the level at or below which systems are susceptible to severe
episodic acidification (i.e., brief periods of ANC down to very low or negative values) (Eshleman, 1988)
with consequent adverse effects on biota. It is extremely important to keep in mind that these values
only serve as indices in an otherwise smooth continuum of surface water chemistry conditions and
responses to acidic deposition. It is also important to remember that adverse biological effects occur
at higher ANCs (i.e., greater than 50 Meq L'1) in systems that previously (i.e., prior to the advent of
acidic deposition) were adapted to more circumneutral conditions (Schindler, 1988).
Model projections indicate a mixed response of northeastern lake systems at current levels of sulfur
deposition. Although slight decreases in median ANC for all ANC groups are projected, as is a slight
increase in the number of systems with ANC <0 jieq L'1, the total number of systems having ANC <
50 Meq L'1 (and thus potentially susceptible to episodic acidification) is projected to decrease (Table
1-1). Projected responses to decreased sulfur deposition show a clearer pattern; MAGIC projects surface
water ANCs to Increase and the number of lakes with ANC <0 /ieq L'1 and ANC <50 Meq L*1 to
decrease. Such a response would be consistent qualitatively with reported changes in the chemistry
of lakes near Sudbury, Ontario, following reductions of sulfur dioxide emissions from the Sudbury smelter
(Dillon et a)., 1986; Hutchinson and Havas, 1986; Keller and Pitbaldo, 1986).
1-17
-------
Plate 1-4. Changes in median ANC of northeastern lakes at SO years as projected by MAGIC (see
Section 1.3.4 for definition of the deposition scenarios used).
1-18
-------
CHANGE IN MEDIAN ANC
Year 10 to Yeor 50
Model = MAGIC
-------
Table 1-1. Lakes in the NE Projected to Have ANC Values <0 and <50 peq L*1
for Constant and Decreased Sulfur Deposition**
Time from
Present (yr)
0 #°
%
20 #
50 #
Constant
ANC <0
162d
5
161 (134)
5(4)
186 (143)
5(4)
Deposition
ANC <50
880d
27
648 (246)
20(8)
648 (246)
20(8)
Decreased
ANC <0
162d
5
136 (124)
4(4)
87 (100)
3(3)
Deposition
ANC <50
880d
27
621 (242)
19 (18)
586 (237)
18(7)
a Projections are based on 123 lake/watersheds successfully calibrated by MAGIC.
b See Section 1.3.4 for definition of the deposition scenarios used.
c # is the number of lakes; % is percent of the target population of 3,227 lakes; (} indicate 95 percent confidence
estimates.
d Indicates estimate from NSWS Phase I sample for the same 123 lakes; target population = 3,227 lakes
1-19
-------
Because of the organic nature of some soils in the NE, the exact nature of chemical "recovery" of
northeastern lakes is uncertain. Under decreased sulfur deposition scenarios, organic acidity leached
from soils could replace mineral acidity associated with sulfur deposition (Krug and Frink, 1983; Krug et
al., 1985; Krug, 1989). Available evidence from catchment manipulations indicates that this process
partially occurs under extreme conditions but the effect probably is not regionally important In regions
such as the NE (Wright et al., 1988). Even if there was an appreciable increase in organic acid leaching
as a response to reduced deposition acidity, the net effect would be beneficial to aquatic biota inasmuch
as it would most likely be accompanied by reductions in surface water concentrations of inorganic
monomeric aluminum, which is highly toxic to fish.
Thus, although the exact chemical response of the DDRP NE systems is unknown, projections
consistently indicate some improvement in surface water quality as a consequence of reduced sulfur
deposition in the region.
1.4.3.2 Southern Blue Ridge Province
Plate 1-5 illustrates the projected changes (MAGIC) in median ANC at 50 years for stream reaches
in the SBRP. In this analysis, MAGIC was successfully calibrated to 32 of the 35 DDRP SBRP stream
reach watersheds. Two stream reaches had ANC values > 1 ,000 neq L*1 and were dropped from this
analysis. The remaining 30 stream reaches had ANC values > 25 Meq L'1 and < 400 fieq L.'1 and
represent a target population of 1,323 stream reaches in the SBRP. The projected changes in median
ANC have been computed for the same ANC groups (25-100 fieq L,*1 and 100-400 jueq L*1) as for the
NE (Plate 1-3).
Table 1-2 presents the population estimates (with 95 percent confidence intervals) of SBRP stream
reaches having ANC < 0 Meq L'1 and <50 jueq L'1 at 20 and 50 years as projected by MAGIC for the
two deposition scenarios. The 95 percent confidence intervals about these projections are broad but
understandable given the tow number of systems available for simulation (30) and the Inherent
uncertainties involved in such a complex simulation of environmental response.
1-20
-------
Plate 1-5. Changes in median ANC of Southern Blue Ridge Province stream reaches at 50 years
as projected by MAGIC (see Section 1.3.4 for definition of the deposition scenarios used).
1-21
-------
CHANGE IN MEDIAN ANC
Year 10 to Year 50
Model = MAGIC
- * Deposition
-------
Table 1-2. SBRP Stream Reaches Projected to Have ANC Values <0 and <50 0eq L"1
for Constant and Increased Sulfur Deposition8*
Time from
Present (yr)
0 #c
%
20 #
50 #
* Projections are basi
Constant
ANC <0
0"
0
0
0
129 (195)
10 (15)
id on 30 stream /watershe
Deposition
ANC <50
3d
0.2
187 (228)
14(17)
203 (236)
15 (18)
ds successfultv
Increased
ANC <0
Od
0
0
0
159 (213)
12 (16)
calibrated bv MJ
Deposition
ANC <50
3d
0.2
187 (228)
14(17)
340 (286)
26 (22)
M3IC.
b Sea Section 1.3.4 for definition of the deposition scenarios used.
6 # is the number of lakes; % is percent of the target population of 1,323 stream reaches; () indicate 95 percent
confidence estimates.
d Indicates estimate from NSWS Pilot Stream Survey sample for the same 30 streams; target population = 1,323
stream reaches.
1-22
-------
Model projections for the SBRP stream reaches indicate decreased surface water quality under
scenarios of either current or increasing sulfur deposition. Due to the fact that soils in this region are
much less organic in nature than those in the NE [e.g., wetlands in the SBRP are virtually non-existent;
maximum stream DOC at lower stream reach nodes = 2.0 mg L"1, mean = 0.8 mg L"1 (Kaufmann et
al., 1988)], these model projections are uncomplicated by any potential effects of organic acid leaching.
Model projections for the increased sulfur deposition scenario indicate the potential for about one-fourth
of the target population of stream reaches in the SBRP to reach an ANC of < 50 peq L*1 in 50 years,
and thus to have the potential to be acidified to an ANC of ~0 jtteq L'1 during storm event episodes
(Eshleman, 1988). As noted In Section 1.4.1.2, responses to changes in sulfur deposition levels in the
SBRP are projected to be slower than those in the NE; I.e., there is a considerable lag in the response
of the systems due to the storage of sulfur in the soils. The result Is that there is a delay not only in the
acidification of surface waters in the region, but also in any potential chemical recovery if sulfur deposition
were to be decreased.
Projections of stream water quality response for the DDRP SBRP target population clearly indicate
future adverse effects of sulfur deposition at increased or current levels.
1.5 SUMMARY DISCUSSION
The NE is currently at sulfur steady state, and sulfate concentrations in surface waters would
respond relatively rapidly to decreases in sulfur deposition. Associated with these changes would be
small increases in surface water ANC. Continued sulfur deposition at current levels is gradually depleting
the cation exchange pool in northeastern soils with consequent decreases in surface water ANC. Such
changes are relatively slow and minor, however, relative to direct effects on surface water chemistry of
increased sulfate mobility in watersheds.
Watersheds in the SBRP are currently retaining nearly three-quarters of the atmospherically
deposited sulfur on the average, but soils are projected to become more saturated with regard to sulfur.
Sulfate concentrations are projected to increase in the surface waters of the region. This response is
1-23
-------
projected to be marked over the next 50 years at either current or increased levels of sulfur deposition,
as are decreases In stream water ANC. Superimposed upon this effect is a relatively minor acidification
effect of base cation depletfon.
Results from all levels of DDRP analyses are (1) consistent internally, (2) consistent with theory
(Galloway et al., 1983a), and (3) consistent with observations of lakes monitored during changing sulfur
deposition regimes (Dillon et al., 1986; Hutchinson and Havas, 1986; Keller and Pitbaldo, 1986).
1.6 REFERENCES
Altshuller, A.P., and R.A. Unthurst. 1984. The Acidic Deposition Phenomenon and Its Effects: Critical
Assessment Review Papers. EPA/600/8-83/016bf. U.S. Environmental Protection Agency,
Washington, DC.
Asbury C.E., F.A. Vertuccl, M.D. Mattson, and G.E. Likens. 1989. Acidification of Adirondack lakes.
Environ. Sci. Technoi. 23:362-365.
Bloom, P.R., and D.F. Grigal. 1985. Modeling soil response to acidic deposition In nonsulfate adsorbing
soils. J. Environ, dual. 14:489-495.
Campbell, W.G., and M.R. Church. 1989. EPA uses G1S to study lake and stream acidification. Federal
Digital Cartography Newsletter 9:1-2.
Campbell, W.G., M.R. Church, G.D. Bishop, D.C. Mortensen, and S.M. Pierson. in Press. The Role for
a Geographic Information System in a large environmental project. Internal. J. GIS.
Chen, C.W., S.A. Gherini, J.D. Dean, R.J.M. Hudson, and RA Goldstein. 1983. Modeling of Precipitation
Series, Volume 9. Ann Arbor Sciences, Butterworth Publishers, Boston, MA. 175 pp.
Church, M.R., and R.S. Turner. 1986. Factors Affecting the Long-term Response of Surface Waters to
Acidic Deposition: State of the Science. EPA/600/3-86/025. NTIS PB 86 178 188-AS. U.S.
Environmental Protection Agency, Corvallis, OR. 274 pp.
Church, M.R. In Press. Predicting the future long-term effects of acidic deposition on surface water
chemistry: The Direct/Delayed Response Project. Eos, Transactions, American Geophysical Union.
Cosby, B.J., G.M. Homberger, J.N. Galloway, and R.F. Wright. I985a. Modeling the effects of acid
deposition: Assessment of a lumped parameter model of soil water and streamwater chemistry.
Water Resour. Res. 21:51-63.
Cosby, B.J., G.M. Hornberger, J.N. Galloway, and R.F. Wright. I985b. Time scales of catchment
acidification: A quantitative model for estimating freshwater acidification. Environ. Sci. Technoi.
19:1144-1149.
Cosby, B.J., G.M. Homberger, E.B. Rastetter, J.N. Galloway, and R.F. Wright. 1986a. Estimating catchment
water quality response to acid deposition using mathematical models of soil ion exchange
processes. Geoderma 38:77-95.
Cosby, B.J., G.M. Hornberger, R.F. Wright, and J.N. Galloway. 1986b. Modeling the effects of acid
deposition: control of long-term suJfate dynamics by soil sulfate adsorption. Water Resour. Res.
22:1283-1292.
1-24
-------
Dillon, P.J., R.A. Reid, and R. Girard. 1986. Changes in the chemistry of lakes near Sudbury, Ontario
following reductions of SO2 emissions. Water, Air, Soil Pollut. 31:59-65.
Driscoll, C.T., C.P. Yatsko, and F.J. Unangst. I987a. Longitudinal and temporal trends In the water
chemistry of the Norm Branch of the Moose River. Biogeochemistry 3:37-61.
Driscoll, C.T., and R.M. Newton. 1985. Chemical characteristics of Adirondack lakes. Environ. Sci.
Technd. 19:1018-1024.
Driscoll, C.T., R.D. Fuller, and W.D. Schecher. I989a. The role of organic acids in the acidification of
surface waters In the eastern U.S. Water, Air, Soil Pollut. 43:21-40.
Driscoll, C.T., N.M. Johnson, G.E. Likens, and M.C. Feller. 1988. Effects of acidic deposition on the
chemistry of headwater streams: A comparison between Hubbard Brook, New Hampshire, and
Jamieson Creek, British Columbia. Water Resour. Res. 24:195-200.
Eshleman, K.N. 1988. Predicting regional episodic acidification of surface waters using empirical models.
Water Resour. Res. 24:1118-1126.
Galloway, J.N., SA Norton, and M.R. Church. 1983a. Freshwater acidification from atmospheric deposition
of sulfuric acid: a conceptual model. Environ. Sci. Technof. 17:541-545.
Galloway, J.N., G.R. Hendrey, C.L Schofield, N.E. Peters, and A.H. Johannes. 1987. Processes and
causes of lake acidification during spring snowmelt in the west-central Adirondack Mountains, New
York. Can. J. Fish. Aquat. Sci. 44:1595-1602.
Gherini, SA, L Mok, R.J. Hudson, G.F. Davis, C.W. Chen, and RA Goldstein. 1985. The ILWAS model:
Formulation and application. Water, Air, Soil Pollut. 26:425-459.
Henriksen, A., and D.F. Brakke. 1988. Increasing contributions of nitrogen to the acidity of surface
waters in Norway. Water, Air, Soil Poilut. 42:183-201.
Hutchinson, T.C., and M. Havas. 1986. Recovery of previously acidified lakes near Coniston, Canada
following reductions in atmospheric sulphur and metal emissions. Water, Air, Soil Pollut. 28:319-
333.
Hynes, H.B.N. 1975. Edgardo Baldi memorial lecture: The stream and its valley. Verh. Intemat. Verein.
Limnol. 19:1.
Johnson, D.W., and D.W. Cole. 1980. Anion mobility in soils: Relevance to nutrient transport from forest
ecosystems. Environ. Intemat. 3:79-90.
Kaufmann, P.R., AT. Heriihy, J.W. Elwood, M.E. Mitch, W.S. Overton, M.J. Sale, J.J. Messer, K.A. Cougar,
D.V. Peck, K.H. Reckhow, A.J. Kinney, S.J. Christie, D.D. Brown, C.A. Hagley, and H.I. Jager. 1988.
Chemical Characteristics of Streams in the Mid-Atlantic and Southeastern United States, Volume I:
Population Descriptions and Physico-Chemical Relationships. EPA/600/3-88/021 a. U.S.
Environmental Protection Agency, Washington, DC. 397 pp.
Keller, W., and J.R. Pitbaldo. 1986. Water quality changes in Sudbury area lakes: A comparison of
synoptic surveys in 1974-1976 and 1981-1983. Water, Air, Soil Pollut. 29:285-296.
Krug, E.C. 1989. Assessment of the theory and hypotheses of the acidification of watersheds. Illinois
State Water Survey Division, SWS contract Report 457. 252 pp.
Krug, E.C., and C.R. Frink. 1983. Acid rain on acid soil: A new perspective. Science 221:520-525.
Krug, E.C., P.J. Isaacson, and C.R. Frink. 1985. Appraisal of some current hypotheses describing
acidification of watersheds. J. Air Pollut. Control Assoc. 35:109-114.
1-25
-------
Lee, J.J., D.A. Lammers, M.G. Johnson, M.R. Church, D.L Stevens, D.S. Coffey, R.S. Turner, LJ. Blume,
LH. Liegel, and G.R. Holdren. I989a. Watershed surveys to support an assessment of the regional
effect of acidic deposition on surface water chemistry. Environ. Mgt. 13:95-108.
Lee, S. 1987. Uncertainty Analysis for Long-term Acidification of Lakes in Northeastern USA. Ph.D. Thesis.
University of Iowa, Iowa City.
Linthurst, R.A., D.H. Landers, J.M. Eilers, D.F. Brakke, W.S. Overton, E.P. Meier, and R.E. Crowe. 1986a.
Characteristics of Lakes in the Eastern United States, Volume I: Population Descriptions and
Physico-Chemical Relationships. EPA/600/4-86/007a. U.S. Environmental Protection Agency,
Washington, DC. 136 pp.
Messer, J.J., C.W. Ariss, J.R. Baker, S.K. Drouse, K.N. Eshleman, P.R. Kaufmann, R.A. Unthurst, J.M.
Omernik, W.S. Overton, M.J. Sale, R.D. Schonbrod, S.M. Stambaugh, and J.R. Tuschali Jr. 1986a.
National Stream Survey Phase I, Pilot Survey. EPA/600/4-86/026. U.S. Environmental Protection
Agency, Washington, DC. 179 pp.
Mohnen, V.A. 1988. The challenge of acid rain. Scientific American 259:30-38.
NAS. 1984. Acid Deposition: Processes of lake acidification. Summary of Discussion. National Research
Council Commission on Physical Sciences, Mathematics, and Resources. Environmental Studies
Board, Panel on Processes of Lake Acidification. National Academy Press, Washington, DC. 11 pp.
NAS. 1986. Acid Deposition: Long-Term Trends. National Research Council Commission on Physical
Sciences, Mathematics, and Resources. Environmental Studies Board. National Academy Press,
Washington, DC. 506 pp.
Neary, B.P., and P.J. Dillon. 1988. Effects of sulphur deposition on lake-water chemistry in Ontario,
Canada Nature 333:340-343.
Nikolaidis, N.P., H. Rajaram, J.L Schnoor, and K.P. Georgakakos. 1988. A generalized soft water
acidification model. Water Resour. Res. 24:1983-1996.
Nye, P.M., and D.J. Greenland. 1960. The Soil Under Shifting Cultivation. Commonwealth Bureau of Soils
Tech. Comm. No. 51. Commonwealth Agricultural Bureaux, Farnham Royal, Bucks.
Reuss, J.O., and D.W. Johnson. 1985. Effect of soil processes on the acidification of water by acid
deposition. J. Environ. Qual. 14:26-31.
Reuss, J.O., and D.W. Johnson. 1986. Acid Deposition and the Acidification of Soils and Waters.
Ecological Studies Volume 59. Springer-Verlag, Inc., New York, NY.
Schindler, D.W. 1988. The effects of acid rain on freshwater ecosystems. Science 239:149-157.
Schnoor, J.L, N.P. Nikolaidis. and G.E. Glass. 1986b. Lake resources at risk to acidic deposition in the
Upper Midwest. J. Water Pollut. Control Fed. 58:139-148.
Seip, H.M. 1980. Acidification of freshwaters - sources and mechanisms, pp. 358-366. In: D. Drabl$s and
A. Tollan, eds. Ecological Impact of Acid Precipitation, Proceedings of an International Conference,
Sandefjord. March 11-14. SNSF Project, Oslo-As, Norway.
Shaffer, P.W., R.P. Hooper, K.N. Eshieman, and M.R. Church. 1988. Watershed vs. in-lake alkalinity
generation: A comparison of rates using input-output studies. Water, Air, Soil Pollut. 39:263-273.
Shaffer, P.W., and M.R. Church. 1989. Terrestrial and in-lake contributions to the alkalinity budgets of
drainage lakes: An assessment of regional differences. Can. J. Fish. Aquat Sci. 46:509-515.
Sullivan, T.J., J.M. Eilers, M.R. Church, DJ. Bltck, K.N. Eshleman, D.H. Landers, and M.S. DeHaan. 1988b.
Atmospheric wet sulphate deposition and lakewater chemistry. Nature 331:607-609.
1-26
-------
Wright, R.F., E. Lotse, and A. Semb. 1988. Reversibility of acidification shown by whole-catchment
experiments. Nature 334:670-675.
1-27
-------
SECTION 2
INTRODUCTION TO THE DIRECT/DELAYED RESPONSE PROJECT
2.1 PROJECT BACKGROUND
Much scientific interest and public debate surround the effects of acidic deposition on freshwater
ecosystems (e.g., Schindler, 1988; Mohnen, 1988). A comprehensive chemical survey (the National
Surface Water Survey - NSWS) of the lakes and streams of the United States considered to be most
vulnerable to acidic deposition (i.e., those with the lowest acid neutralizing capacity or ANC) was recently
completed by the U.S. Environmental Protection Agency (EPA) (Unthurst et al., 1986a; Kaufmann et a*.,
1988). Analysis of these and other lake and stream chemistry data, together with data on temporal and
spatial patterns of atmospheric deposition, indicates that long-term deposition of sulfur-containing
compounds originating from the combustion of fossil fuels has acidified (i.e., decreased the ANC of) some
surface waters in eastern North America (Altshuller and LJnthurst, 1984; NAS, 1986; Sullivan et a)., I988b;
Neary and Dillon, 1988; Asbury et al., 1989). Transport of mobile anfons (primarily sulfate) through
watershed soils and closely associated cation leaching are the most widely accepted mechanisms of this
acidification process (Seip, 1980; Galloway et a)., I983a; Driscoll and Newton, 1985; Church and Turner,
1986). In addition, acidic deposition apparently has shifted the nature of some very low ANC or naturally
acidic surface waters in the Northeast from organic acid "dominance" to mineral acid "dominance"
(Driscoll et al., 1988; Driscoll et at., 1989a). This process is, perhaps, best explained as the effective
titration of naturally occurring humic substances by sulfuric acid deposition (Krug and Frink, 1983; Krug
et al., 1985; Krug, 1989). In both cases, the net effect of atmospheric deposition of sulfuric acid on
surface water chemistry is a shift toward aquatic systems more dominated by mineral acidity and more
likely to contain inorganic forms of aluminum, which are toxic to aquatic organisms.
Given that acidification of some surface waters has occurred, critical scientific and policy questions
focus on whether acidification is continuing in the regions noted, whether it Is just beginning in other
regions, how extensive effects might become, and over what time scales effects might occur. EPA is
examining these questions through the activities of the Direct/Delayed Response Project (DDRP) (Church
2-1
-------
and Turner, 1986; Church, in press). The Project was begun in 1984 at the specific request of the EPA
Administrator following a meeting of the Panel on Processes of Lake Acidification of the National Academy
of Sciences (NAS). Principal among the conclusions of the Panel was that atmospheric deposition of
sulfur-containing compounds is the major source of long-term surface water acidification in eastern North
America (NAS, 1984). The Panel also debated at length the dynamic aspects of the acidification process.
The DDRP was designed to focus on this question and, thus, draws its name from consideration of
whether acidification might be immediate (or immediately proportional to levels of deposition) (i.e., 'direct")
or whether it would lag in time (i.e., be "delayed") because of edaphic characteristics. A compilation and
discussion of the processes of long-term surface water acidification and methods for its investigation
were presented by Church and Turner (1986) at the beginning of the Project. A relatively brief and more
current discussion of processes relevant to this Project is presented in Section 3 of this report.
Although more recent research has indicated the potential importance of deposition of nitrogen-
containing compounds to both the episodic (Galloway et al., 1987; Driscoll et al., I987a) and long-term
(Henriksen and Brakke, 1988) acidification of surface water, the DDRP does not address these effects.
Such effects are the focus of developing or ongoing research within EPA's Aquatic Effects Research
Program.
2.2 PRIMARY OBJECTIVES
The DDRP has four technical objectives related to atmospheric/terrestrial/aquatic interactions:
(1) to describe the regional variability of soil and watershed characteristics,
(2) to determine which soil and watershed characteristics are most strongly related to surface
water chemistry,
(3) to estimate the relative importance of key watershed processes in moderating regional effects
of acidic deposition, and
(4) to classify a sample of watersheds with regard to their response characteristics to inputs of
acidic deposition and to extrapolate the results from this sample of watersheds to the study
regions.
The fourth objective is the critical "bottom line" of the Project.
2-2
-------
The relationship of the DORP to other projects within the Aquatic Effects Research Program (AERP)
of the National Acid Precipitation Assessment Program (NAPAP) is shown in Figure 2-1. It was never the
intent of the DORP to serve as a "research" project to investigate exact mechanisms and processes of
surface water acidification. Rather, the principal mandate of the Project was to make regional projections
of future effects of sulfur deposition on long-term surface water chemistry (principally ANC) based upon
the best available data and most widely accepted hypotheses of the acidification process. Further
watershed modelling activities within the NAPAP Integrated Assessment (see Figure 1-2) will investigate
a variety of sulfur deposition scenarios and potential future effects on biologically relevant surface water
chemistry (e.g., pH, and concentrations of calcium and Inorganic monomeric aluminum).
2.3 STUDY REGIONS
The Project focuses on three regions of the eastern United States where low ANC surface waters
are located and where levels of atmospheric deposition (relative to other U.S. regions) are greatest:
(1) the Northeast (NE), (2) upland areas of the Mid-Atlantic (referred to here as the Mid-Appalachian
Region), and (3) the mountainous section of the Southeast called the Southern Blue Ridge Province
(SBRP) (Plate 2-1). Initiation of the Project depended on the availability of the regional surface water
chemistry data of the NSWS. Thus, the Project focused its work initially on the lake resources of the NE
(LJnthurst et al., I986a) and the stream resources of the SBRP (Messer et al., !986a). The results for
these two regions are presented in this report. Complete results of subsequent work in the
Mid-Appalachian Region will be reported at a later date.
2.4 TIME FRAMES OF CONCERN
The DDRP focuses on potential effects of acidic deposition on surface water ANC as evaluated
at key annual "index" periods. These index periods follow the sampling schemes of the NSWS (i.e., fall
period of complete mixing for lakes and spring baseflow for streams - see Section 5.3). "Episodic"
acidification (e.g., due to snowmeit or intense rainstorms) is not considered within the DDRP but is the
primary consideration of a companion project within the AERP, the Episodic Research Project (Figure
2-1).
2-3
-------
Eastern Lake Survey*
• Survey of lake index
chemistry
Watershed Manipulation Project
• Watershed process research
• Watershed manipulation by acid
addition
• DDRP model testing
National Stream Survey*
• Survey of stream reach index
chemistry
Episodic Response Project
• Evaluation of episodic
acidification of streams
Direct/Delayed Response Project
• Projections of future effects of
long-term sulfur deposition on
surface water chemistry
- Northeastern lakes and
Southern Blue Ridge Province
streams*
- Mid-Appalachian streams
State of Science
• Comprehensive analysis of
evidence for aquatic effects
Integrated Assessment
• Synthesis of aquatic effects
state of science
• Comparative evaluation of
aquatic effects for various
emissions control scenarios
Critical Loads Project
• Evaluation of the effects of
long-term nitrogen and sulfur
deposition on surface water
chemistry
Rgure 2-1. Activities of the Aquatic Effects Research Program within the National Acid Precipitation
Assessment Program. Completed projects are designated by asterisks.
2-4
-------
Plate 2-1. Direct/Delayed Response Project study regions and sites.
2-5
-------
DORP STUDY REGIONS
Northeost
Mid-Appalachian
Region
DDRP Lake Study Sites
DDRP Stream Study Sites
Southern Blue Ridge
Province
-------
The primary time horizon for DDRP analyses is 50 years. This horizon relates to the projected
lifetimes of existing power plants and the potential implementation of additional emissions controls relative
to those lifetimes. Where possible and reasonable, some time-dependent analyses are extended beyond
this 50-year horizon to better evaluate process rates and changes and potential future effects.
2.5 PROJECT PARTICIPANTS
The DDRP was designed and implemented at EPA's Environmental Research Laboratory - Corvallis
s
(ERL-C) and is a very large effort involving many participants. The Project involves two other EPA
laboratories, the Atmospheric Research and Exposure Assessment Laboratory - Research Triangle Park
(AREAL-RTP) and the Environmental Monitoring and Systems Laboratory - Las Vegas (EMSL-LV). The
DDRP is assisted by three other federal agencies, the U.S. Department of Agriculture (including the Forest
Service and the National Office, two National Technical Centers, and 12 state offices of the Soil
Conservation Service), the U.S. Geological Survey, and National Oceanographic and Atmospheric
Administration. Two national laboratories [Oak Ridge National Laboratory (ORNL) and Battelle - Pacific
Northwest Laboratories (PNL)], five state and private universities, and four consulting firms also have
participated in this Project. In all, over 200 field, laboratory, database management, scientific, and
management personnel have contributed to this effort.
2.6 REPORTING
This report documents and discusses the data analyses performed for the NE and SBRP Regions.
It does not contain a complete list of all data used or all results produced in the analyses. The complete
list and documentation will be available at a later date. Section 5 of this report, however, does contain
appropriate summary and example data.
During the course of the Project many of its activities have been documented, externally peer
reviewed and approved, and published as EPA reports. Any reference used in this report that has an
EPA publication number is the final, externally peer-reviewed product of this (or another) EPA project.
Usually, such documents contain more complete descriptions and details of the work undertaken than
2-6
-------
can be presented within this report. Copies of these cited EPA published reports are available upon
request from the Project Technical Director, M. Bobbins Church, at ERL-C,
\
Project participants have published descriptions of activities and results of the DDRP in the
peer-reviewed literature. Published papers and manuscripts in review are cited throughout the report and,
like the published EPA reports, can be obtained by request from the Technical Director As of this
writing, many additional peer-reviewed publications that document the activities and results of the DDRP
are in preparation or are planned. Other preliminary results and discussions of the Project have been
presented at meetings and workshops of the American Geophysical Union (fall, 1987); Association of
American Geographers (November 1987); Biometric Society (July 1986); International Society of Ecological
Modelling (August 1987); North American Lake Management Society (November 1986); and the Soil
Science Society of America (December 1987 and 1988).
2-7
-------
SECTION 3
PROCESSES OF ACIDIFICATION
3.1 INTRODUCTION
As discussed in Section 2.1, the Direct/Delayed Response Project was developed as a result of
the conclusions of the NAS Panel on Processes of Lake Acidification (NAS, 1984) concerning the most
important watershed processes affecting or mediating long-term surface water acidification. The Panel
identified these processes as (1) the retention of deposited sulfur within watersheds and (2) the supply
of base cations from watersheds to surface waters. These processes have therefore become the focus
of tiie DDRP. The purpose of this section is to review these processes briefly in the context of the DDRP
watershed and soil survey (Section 5) and the analyses that follow (Sections 6-10).
Factors other than sulfur retention and base cation supply affect surface water acidification, but
were either deemed by the Panel to be relatively less important in long-term acidification or could not
be addressed completely within the scope of the DDRP due to time, budgetary, or logistical constraints.
Several of these alternative factors (nitrate deposition, land use, leaching of organic acids from soils, and
hydrologic flowpaths) are discussed briefly below in the context of the design and objectives of the
DDRP, but they are not addressed in detail in this report.
Leaching of nitrate from soils has been identified as a potential source of acid transport to surface
waters during spring snowmelt events (Galloway et al., 1987; Driscoll et al., I987a). Inasmuch as the
DDRP focuses on long-term acidification, this effect is not considered here. Although nitrogen appears
to be retained almost entirely in most forested watersheds by biological uptake and accretion in biomass
(Abrahamsen, 1980), recent studies suggest that nitrogen throughput (with leaching as nitrate) is a
significant contributor to long-term acidification at some selected sites with mature forests (C. Driscoll,
personal communication). Evidence for such chronic effects was not available when the DDRP began
and is not considered within the analyses presented here. It will likely be the focus of future studies by
the EPA.
3-1
-------
Land use and changing land use [e.g., forest growth; Krug and Prink (1983)] can affect both the
chemistry of surface waters and the interaction of acidic deposition with soils, which, in turn, can affect
surface water acidification. Apparent influences of land use are discussed in Section 8. Projection of
changes In land use and projection of changes in surface water chemistry associated with such alterations
(either on the DDRP study watersheds or in the DDRP study regions) are outside of the scope of the
DDRP analyses.
Krug and Frink (1983) discussed the importance of natural soil acidification processes and
hypothesized that acidic deposition could lower the pH of soils proximate to surface waters, thereby
decreasing the dissociation of humic acids and decreasing the mobility of organic 'carrier anions."
Reverse conditions could occur under a scenario of decreased deposition acidity. Although LaZerte and
Dillon (1984) have presented evidence that the Krug and Frink hypothesis is not supported in studies of
acidified lakes in Ontario, Canada, such changes could affect both the pH and buffering capacity of
surface waters. Potential dynamic effects on watershed soils and surface water chemistry (as presented
by Krug and Frink, 1983) are not discussed here in detail, nor are they considered explicitly within the
DORP except when such interactions are incorporated in the Integrated Lake/Watershed Acidification
Study (ILWAS) Model (see Section tO).
The route that water follows within watersheds plays a very important role in the determination of
surface water chemistry and in the response of soils and surface waters to acidic deposition (e.g., see
Chen et al., 1984; Peters and Driscdl, 1987). Such interactions and effects were reviewed by Church and
Turner (1986) at the outset of the DDRP, and these discussions are not repeated here. The in-depth
determination of ffowpaths within individual DDRP study watersheds was not within the time, budgetary,
or logistical scope of the Project. The apparent associations between watershed hydrologic parameters
or indicators is presented in Section 8. Assumptions concerning flowpaths and their effects on analyses
are presented in Section 9, and descriptions of the hydrologic modules of integrated watershed models
are presented in Section 10.
3-2
-------
3.2 FOCUS OF THE DIRECT/DELAYED RESPONSE PROJECT
During the past decade there has been an increased recognition that surface water acidification
is controlled not only by rates of hydrogen deposition, but also by the mobility of associated anions
through the ecosystem. A conceptual model of surface water acidification (Galloway et al., I983a) and
the 1984 NAS Panel identified two dominant variables affecting the rate and extent of watershed
acidification: (1) control on anion mobility, specifically on sulfate adsorption and (2) rates of base cation
supply from watersheds.
Almost three decades ago, Nye and Greenland (1960) recognized the Importance of anions as
"carriers" for cations in solution. The "mobile anion" paradigm they proposed, more recently applied to
surface water acidification (Johnson and Cole, 1980; Seip, 1980), suggests that a variety of processes
(e.g., adsorption of sulfate and phosphate, biological uptake of nitrate, pH- and pCO2-dependent
dissociation of carbonic acid) act more or less independently to control the concentrations of individual
anions in solution, whereas cation exchange and weathering processes control the relative quantities of
cations. Controls on, and changes in, anion mobility can thus be viewed as the proximate controls on
rates of cation leaching from soils and, coupled with rates of cation resupply processes, on surface water
acidification. Within the DDRP the primary issue with regard to anion mobility lies in forecasting temporal
changes in dissolved sulfate. Sulfur retention processes are further discussed in the following section.
Rates of base cation supply from watersheds were identified as the second dominant factor
determining the rate and ultimate acidification of surface waters by acidic deposition. Supply of base
cations occurs principally from mineral weathering (as the "original" source) and cation exchange in soils.
These processes are discussed further in Section 3.4.
3.3 SULFUR RETENTION PROCESSES
3.3.1 Introduction
Watershed sulfur budgets and regional summaries of sulfur Input/output budgets indicate substantial
regional differences in sulfate mobility between the Northeast (NE) and the Southern Blue Ridge Province
3-3
-------
(SBRP) (Rochelle et al., 1987; Rochelle and Church, 1987). Understanding and characterizing these
differences are important objectives of the DDRP, and efforts toward fulfilling these objectives are
discussed in Sections 7.3 and 9.2. This section provides a background for those efforts, summarizes
the current understanding of controls on sulfate mobility in soils and watersheds, and assesses the
relative importance of the control processes. Discussion Is focused on sulfate adsorption, which is
regarded as a major process in sulfur retention by forest soils (Johnson and Todd, 1983; MAS, 1984;
Fuller et al., 1985).
3.3.2 Inputs
The upper limit on sulfate concentration in surface waters is controlled by sulfur inputs to the
system, i.e., the deposition flux to the watershed and the generation of additional sulfate within that
system. The principal concern with regard to acidification, and often the only sulfur source considered,
is deposition of atmospheric sulfur derived from anthropogenic sources. Sea salt can supply a significant
amount of sulfate to watersheds and surface waters in near-coastal areas. Concentrations and deposition
fluxes of sulfate from natural sources other than sea salt (e.g., biogenic emissions, volcanoes) in "clean*
areas, however, are roughly an order of magnitude lower than the anthropogenicalty enhanced fluxes in
parts of eastern North America, which are heavily influenced by acidic deposition (Olsen and Watson,
1984).
A second potential contributor of sulfate to watersheds is oxidation of sulfides in soils or bedrock.
Net mineralization of organic matter, if it occurs, provides a significant source of sulfate, although it
represents release of sulfur sequestered by biomass at some previous time rather than "new" sulfur.
Oxidation of minerals such as pyrfte is more common and the most important internal sulfur source.
Sulfide oxidation typically is not quantified in watershed studies, except inferentially from detailed sulfur
input/output budgets. In the absence of specific sulfide oxidation data or of other strong evidence for
internal sulfur sources (e.g., net sulfur efflux, geologic data), watershed sulfur sources are typically
ignored altogether (e.g., Christophersen and Wright, 1981; Helvey and Kunkle, 1986; Jeffries et al., 1986)
or are assumed to be unimportant contributors to sulfur budgets (e.g., Dillon et al., 1982; Schafran and
3-4
-------
Driscoll, 1987). Cyclic reoxidation of reduced sulfur from wetlands and/or flooded soils during dry periods
can generate substantial transient sulfate effluxes (deGrosbois et al., 1986; Bayley et al., 1986), but should
be recognized as a recycling of previously retained sulfate rather than as a true source of "new" sulfur
to a watershed.
Dissolution of sulfate minerals (e.g., gypsum) is another potential watershed source of sulfate.
Mineral sulfates occur In soils in arid to semi-arid climates In association with other evaporites, including
carbonates, in bedrock, sulfates are also associated with carbonates (coprecipitated with) (Doner and
Lynn, 1977; Hurtbut and Klein, 1977). Because of the co-occurrence of mineral sulfates with carbonates,
and because even small amounts of carbonate provide substantial ANC to receiving waters, watersheds
with significant inputs of sulfate from sulfate mineral dissolution likely will have high ANC and thus will
not be sensitive to acidification.
In the DDRP, there has been an extensive effort to quantify atmospheric deposition to the study
watersheds (Section 5.6). Both direct and indirect efforts have been made to assess Internal sulfur
sources to watersheds based on mapped lithology and on analysis of uncertainties in watershed sulfur
input/output budgets (Section 7).
3.3.3 Controls on Sulfate Mobility within Forest/Soil Systems
The sulfur cycle in forest ecosystems is strongly influenced by both inorganic and biologically
mediated processes (Figure 3-1). The forest canopy acts as a collection surface for dry deposited sulfur,
both for paniculate sulfate aerosols and for gaseous sulfur dioxide. Precipitation subsequently washes a
large portion of dry deposited sulfur along with, In some cases, sulfate leached from leaf surfaces in the
canopy (Lindberg et al., 1986; LJndberg and Garten, 1988). The increase in sulfur flux in throughfall
compared to that of incident precipitation has been used as an estimator of the amount of dry deposition
to a system (Khanna et al., 1987; LJndberg et al., 1986; LJndberg and Garten, 1988).
3-5
-------
CO O
0)
V
§•
§
o>
I
E
S
o>
1
°»
u.
3-6
-------
Within the soil, solution concentrations of sulfate are strongly regulated by sorption reactions,
which may add (desorb) or remove (adsorb) sulfate from solution, depending on the sulfur status of soil
and concentration of sulfate In incoming solution. A variety of other inorganic and biologically mediated
processes also occur within forest ecosystems and are discussed briefly below.
3.3.3.1 Precipitation/Dissolution of Secondary Sulfate Minerals
Along with sorption reactions, secondary mineral phases of aluminum or iron can control sulfate
concentrations in solution (Adams and Rawajfih, 1977; Nordstrom, 1982; Khanna et al., 1987). Evidence
for occurrence and control of sulfate (and/or aluminum) concentrations by these phases Is usually indirect
(saturation indices) and thus is not unequivocal. These minerals are likely to occur only in soils with very
low pH (ca. 4.0 or lower) and with high sulfate concentrations in which formation of jurbanite (AIOHSO^,
basaiutninite (AI4(OH10)SO4), or alunite ((K,Na)AJ3(OH)(SO4)2) is likely. Although control of dissolved
sulfate by jurbanite apparently occurs at two sites in Germany, both sites are characterized by very high
sulfur fluxes apparently due to internal sources at Goettingen (Weaver et al., 1985) or to extremely high
atmospheric deposition (up to 40 kg S ha"1 yr*1) at Soiling (Khanna et al., 1987). At more typical sites
where acidification is a concern (soil pH >4.0, wet sulfate deposition <. 15 kg S ha'1 yr'1), soil solutions
are not likely to be saturated with respect to secondary AI-OH-SO4 phases. Saturation index data should
be interpreted with caution, however, because little is known about solution chemistry in most soils under
dry or unsaturated conditions, and there is a possibility of cyclic formation/dissolution of AI-OH-S04
mineral phases during dry and wet periods (Nordstrom, 1982; Weaver et al., 1985). For the vast majority
of watersheds in the eastern United States, including DORP watersheds, control of solution sulfate by
aluminum sulfate mineral phases cannot be ruled out, but is unlikely.
3.3.3.2 Sulfate Reduction in Soils and Sediments
In intermittently or permanently anaerobic soils, in wetlands, and in lake sediments, significant
reduction of sulfate can occur. The principal reduction products in soils are metal sulfides, which are
evident as gleying or mottling of the soil. Little is known about the overall magnitude of sulfur retention
in such soils. Long-term retention of sulfur by reduction occurs only in soil environments that are
3-7
-------
permanently anaerobic. In seasonally reduced zones, sulfides are quickly reoxidized upon drying of the
soil (Nyborg, 1978). Partially oxidized sulfur species (i.e., oxidation states between -2 and +6) also occur
in soils, but usually represent a very small fraction of total soil sulfur (Freney, 1961; David et al., 1982),
and are likely to occur mostly as labile redox intermediates in the oxidation of mineral or organic sulfides.
In wetlands and in anaerobic lake sediments, sulfate is used as an electron acceptor by bacteria,
with reduced sulfur sequestered as metal or organic sulfides. Retention is first order with respect to
sulfate concentration (Baker et al., 1986b; Kelly et al., 1987). Total retention within lakes Increases with
hydrologic retention time, so the relative importance of in-lake processes varies as a function of
watershed-to-lake area ratio and other lake hydrologic characteristics. In seepage lakes and other
systems with long hydrologic retention times, in-lake reduction is a critical factor in sulfur budgets. In
lakes with short retention times (one year or less, see Baker et al., I986b), including the great majority
of lakes in the DORP regions, in-fake retention has a very minor influence on sulfur budgets (Norton et
al., 1988; Shaffer et al., 1988; Shaffer and Church, 1989; see also Section 7.2).
3.3.3.3 Plant Uptake
Sulfur Is an essential plant nutrient and is extensively cycled through vegetation. Soils in certain
areas of the world have serious sulfur deficiencies (Turner et al., 1980), but deposition in areas receiving
acidic deposition typically provides sulfur far in excess of plant requirements (Johnson et al., 1982a).
Recent studies of sulfur cycling at 10 U.S. and German sites, summarized by Johnson et al. (I982a),
Indicated annual sulfur biomass accretion of 0.5-1.6 kg S ha"1 yr"1 and standing sulfur biomass of 19-
98 kg S ha'1. Accretion at the 10 sites averaged less than 10 percent of wet deposition (including data
for two sites in the western United States with low deposition), and biomass was equivalent to only one
to four times annual deposition. In young forests with aggrading litter mass, significant sulfur accretion
within the litter can also occur (Switzer and Nelson, 1972; D. Johnson, personal communication).
Although such data suggest that biomass accretion is a relatively small net sink for sulfur, the significance
of biological sulfur cycling in the soil cannot be overlooked. Fluxes of sulfur through vegetation and
3-8
-------
between soil pools are very dynamic and play important roles in storage and translocation of sulfur within
the soil.
3.3.3.4 Retention as Soil Organic Sulfur
Probably the most controversial issue regarding sulfur retention in soils is the role of soil organic
sulfur. Organic sulfur, largely contained in or derived from litter, represents by far the largest sulfur pool
In forest soils (>90 percent of total sulfur in many northern soils, and well over half the total sulfur at
Walker Branch, TN, the only southeastern system for which adequate data are available) (Bettany et al.,
1973; David et al., 1982; Schindler et al., 1986a; Johnson et al., I982b). Several recent studies have
documented rapid uptake of sulfate by soil bacteria and conversion to ester sulfate (R-O-SO3 linkages)
and to reduced sulfur (C-S bonds), suggesting a major role for organic forms as net watershed sulfur
sinks (e.g., Fitzgerald et al., 1982; Swank et al., 1984; David et al., 1984; Schindler et al., 1986a). Initial
transformations of Inorganic sulfur, primarily to ester sulfate, are rapid and extensive. Ester sulfate is then
mineralized rapidly to form inorganic sulfate (Houghton and Rose, 1976; Fitzgerald and Johnson, 1982;
Schindler et al., 1986a). Formation of carbon-bonded (reduced) sulfur from ester sulfates occurs more
slowly, but turnover is also much slower. Carbon-bonded sulfur, along with the reduced sulfur generated
by vegetation and stored in litter, represents a large pool of sulfur that turns over very slowly.
Because of the numerous pods, transformations, and kinetic variables in the soil organic sulfur
cycle, the magnitude of net organic sulfur retention is unclear. Watersheds in Coweeta, NC, are
characterized by high net sulfur retention (Swank and Waide, 1988). Field and laboratory studies have
been used to assess contributions of adsorption and organic sulfur formation to watershed sulfur
retention. Short-term uptake indicated high potential flux into organic pools in upper soil horizons (Swank
et al., 1984); later studies have shown high adsorption by soils at Coweeta, with subsequent
transformation of a portion of the adsorbed sulfate to organic forms (Strickland and Fitzgerald, 1984;
Strickland et al., 1987). Strickland and Fitzgerald (1984), Strickland et al. (1987), and Fitzgerald and
Watwood (1987) have concluded that both adsorption and organic accumulation are important
contributors to net watershed sulfur retention. Studies at Coweeta have focused principally on
3-9
-------
transformations in the upper (O and A) soil horizons, however, so considerable uncertainty remains
concerning net fluxes to adsorbed and organic pools in the integrated soil pedon.
In contrast to the conclusions of Fitzgerald and coworkers regarding Coweeta, a recent model of
sulfur transformation kinetics that considers sorption, immobilization, and mineralization rates suggests
a different conclusion for two northeastern sites. Fuller et ai. (1986a) concluded that overall uptake and
mineralization of sulfur are of comparable magnitude, and that the overall net sulfur budgets at Huntington
Forest, NY, and at Hubbard Brook, NH, are near steady state (i.e., sulfur input equals output). Separate
analyses of sulfur isotope data for the Hubbard Brook Experimental Forest led Fuller et al. (1986b) to
conclude again that Hubbard Brook soils have negligible net sulfur retention. A broader evaluation of
sulfur input/output budgets for the NE (Rochelle and Church, 1987; see also Section 7.3) showed that
watershed sulfur budgets for the region are, on average, at or near steady state, suggesting little or no
net retention as organic or other forms of sulfur in typical watersheds of the region.
3.3.3.5 Sulfate Adsorption by Soils
Adsorption has long been recognized as an important process affecting sulfate mobility in soils and
availability to plants (early research on sulfate retention focused on sulfate deficiencies in agricultural
soils). Pioneering work by Chao and coworkers in the early 1960s identified adsorption as a principal
retention mechanism, identified key soil variables affecting adsorption capacity, and used nonlinear
isotherms (Freundlich) to describe partitioning between dissolved and adsorbed phases (Chao et ai.,
1962a,b; 1964a,b).
Research during the late 1960s and 70s suggested two distinct mechanisms of adsorption,
commonly referred to as (1) non-specific adsorption, an electrostatic bonding at positive charge sites on
the soil surface, and (2) specific adsorption, which involves ligand exchange (with OH* or OH2) and ionic
bonding (Hingston et ai., 1967, 1972). Subsequent work by Rajan (1978, 1979) and by Parfltt and Smart
(1978) demonstrated that specific sorption could involve exchange of one or two surface ligands, with
3-10
-------
the latter resulting in "bridging" and formation of an M-O-S(02)-O-M ring, in which M is a metal ion,
usually iron or aluminum, incorporated in a polymeric hydrous oxide or on the edge of a clay lattice.
3.3.3.5.1 Factors affecting adsorption by soils -
Adsorption capacity of soils is influenced by a variety of physical and chemical variables. The
amount of adsorbing substrate (iron and aluminum hydrous oxides, day), soil organic content, and pH
is usually regarded as the most important of these variables. Hydrous oxides of iron and aluminum are
probably the most important substrates for sulfate adsorption in soils. These materials are precipitated
as amorphous or poorly crystalline coatings on particle surfaces in the soil and are positively charged
at low pH, providing anion adsorption sites. Adsorption occurs by exchange with OH' or OH2, and can
involve a single ligand or pair of ligands, depending on surface charge and the abundance of one-
coordinated (La, linked to a single metal atom) hydroxyl or aquo ligands (Parfitt and Smart, 1978).
Several studies, under field and laboratory conditions, have demonstrated high positive correlations
between sulfate adsorption and iron and/or aluminum content of soils (e.g., Chao et al., 1964b; Johnson
and Todd, 1983; Fuller et al,, 1985).
Clay content of soils has been correlated with sulfate adsorption by soils, although it is regarded
as a minor adsorber (Johnson and Todd, 1983). In part, the correlations result from occurrence of
positive charge sites for anion adsorption on clay edges. Perhaps more important, day content is often
highly correlated with non-silicate iron and aluminum content of soils and can serve as a surrogate for
the oxides in regression analyses. Several investigators (e.g., Ndler, 1959; Chao et al., 1962b; Johnson
et al., 1980) have found positive correlations between day content (or surface area, which is in turn
correlated with bom clay content and hydrous oxide content) and adsorbed sulfate. Others (e.g., Haque
and Walmsley, 1973; Johnson and Todd, 1983; Fuller et al., 1985) failed to find such a correlation,
although, as noted above, significant correlations have been observed between adsorbed sulfate and
hydrous oxides.
3-11
-------
Interactions of soil organic matter, sulfate, and adsorbing substrates have received increasing
attention in recent years. Chao et a!. (1964a) noted that the presence of a variety of organic acids
reduced sulfate adsorption in laboratory studies; the most pronounced reduction was by those acids
forming very strong complexes with metals (i.e., oxalate, tartrate). Negative correlations between soil
organic matter and sulfate adsorption have also been noted by Barrow (1967; correlation was with soil
organic nitrogen) and Haque and Walmsley (1973). More recently, organic "blocking" of sulfate adsorption
has been hypothesized to occur in forest soils and has been suggested as a major factor contributing
to regional differences in sulfate mobility and surface water acidification in forest systems receiving acidic
deposition (Johnson et a)., 1980; Johnson and Todd, 1983). This hypothesis is consistent with observed
regional (NE vs. SBRP) differences in sulfur budgets. Northeastern soils typically have higher organic
content than those from the Southeast, but have lower adsorption capacities despite having iron and
aluminum concentrations comparable to those of southeastern soils (Johnson and Todd, 1983).
Fitzgerald and Johnson (1982) have suggested that blocking is a result of competition for anion
adsorption sites by fulvic acids. Similarly, Davis (1982) noted that introduction of fulvic acids resulted in
reduced anion phosphate adsorption by alumina in laboratory studies. He concluded that preferential
sorption of the organic acids was the principal blocking mechanism. Although the occurrence of blocking
is now widely accepted and sorption of organic acids is the most likely process, there has not been a
rigorous evaluation of this or other hypothesized blocking mechanisms (e.g., coating of iron and aluminum
surfaces; Couto et al., 1979).
Along with the amounts of adsorbing substrates and of competing anions, pH is a major, albeit
indirect, control on sulfate adsorption by soils. Chao et al. (1964b) initially demonstrated effects of pH
on adsorption, using fresh hydrous oxides of iron and aluminum, and demonstrated that adsorption
increased as soil pH was lowered. Subsequent investigators (e.g., Hingston et al., 1967, 1972; Couto et
al., 1979; Nodvin et al., 1986) showed similar effects of soil pH on adsorption. Surface charge on iron
and aluminum hydrous oxides is amphoteric. The ratio of OH2 to OH* ligands increases as pH is reduced,
resulting in increased positive surface charge and enhanced anion adsorption capacity. Reduced pH also
3-12
-------
decreases dissociation of organic acids (Stevenson, 1982), minimizing the interference or blocking effect
of organic matter on sulfate adsorption.
The specific soil properties cited above, as well as sulfate adsorption, have been associated with
a variety of qualitative variables. Shriner and Henderson (1978) suggested that differences in net sulfate
retention at Coweeta, NC (high), Walker Branch, TN (intermediate), and Hubbard Brook, NH (negligible),
were related to cumulative acidic (sulfur) deposition, or more specifically to relative saturation of sulfate
adsorption sites. Barrow et al. (1969) noted significant differences in sulfate adsorption by soils formed
over different parent rock. They also noted that the soils had different pH, texture, and hydrous oxide
properties related to mineralogy of the parent material. Barrow et al. (1969), Hasan et al. (1970), and
Johnson and Henderson (1979) have also noted correlations between adsorbed sulfate and degree of soil
weathering, which were In turn related to age and/or annual rainfall. Those investigators pointed out that
differences in composition of the parent material and/or degree of weathering lead to differences in soil
pH and hydrous oxide content, which are probably actually controlling sulfate adsorption. Although It is
important to remember that the quantitative soil properties (iron and aluminum hydrous oxides, organics,
pH, etc.) that control sulfate adsorption are the end products of their environment, and therefore reflect
parent material, weathering history, vegetation, climate, and the influences of man.
3.3.3.5.2 Sorption kinetics -
Kinetics of sulfate adsorption have usually been reported to be very rapid, with soil solution sulfate
concentrations reaching 95-97 percent of steady state within 5 to 15 minutes after addition of sulfate to
soil-water slurries, and steady state within one to three hours (Rajan, 1979; Chao et al., I962a; Bolan et
al., 1986). In a few cases, slower equilibration has been reported, with gradual changes in sulfate for 50
days or more (Barrow and Shaw, 1977; Singh, 1984; Hayden, 1987). Hayden (1987) attributed the slow
changes in her batch experiments to physical alteration (grinding) of surfaces (because no equivalent
"slow* equilibration was observed in concurrent column experiments using the same soils). In the other
reported cases, it appears likely that slow "adsorption* was similarly attributable to treatment effects
and/or to microbially mediated sulfate uptake.
3-13
-------
3.3.3.5.3 Desorption -
Although adsorption of sulfate has been extensively studied, relatively little attention has been paid
to desorption. Reported reversibility of sorption ranges widely, from less than 10 percent (Bomemisza
and Uanos, 1967) to complete desorption (e.g., Weaver et al., 1985; Sanders and Tinker, 1975). Several
factors influencing reversibility have been identified: aging of sulfate on the soil (decreased desorption
with time since adsorption), temperature (less desorption for soils held at higher temperatures), and
characteristics of the adsorbing substrate. Other factors, especially the mechanism of adsorption and
number of ligands, also may contribute to the effects noted above. Desorption kinetics have not been
extensively characterized, but are apparently similar to those for adsorption (Barrow and Shaw, 1977;
Rajan, 1979). The extent of sorption reversibility for soils from the NE and SBRP is currently being
evaluated as part of an ongoing EPA-funded project. Results of that project will contribute significantly
to our understanding of sorption processes and to our ability to project rates of soil and surface water
sulfate response to changes in atmospheric deposition.
3.3.4 Models of Sulfur Retention
Several models have been developed to describe components of watershed sulfur cycles, but to
date there has not been a single model that incorporates all the major terrestrial and aquatic processes
of concern. Both equilibrium and kinetic expressions are incorporated in existing models. Baker et at.
(1986b) and Kelly et al. (1987) developed essentially identical kinetic models to describe in-iake alkalinity
generation. Both models include equations that describe rates of sulfate retention (principally reduction)
as a first-order process with respect to sulfate concentration, and annual percent sulfur retention as a
function of lake hydrotogic retention time. The models use a sulfur mass transfer coefficient based on an
average of field measurements from a variety of sites in North America and Europe. Neither model
considers terrestrial or wetland processes, and both are limited to in-lake retention.
Fuller et al. (1986a) have developed a relatively complete kinetic model to describe soil sulfur
transformations. The model includes reversible sorption reactions and reversible, first-order
immobilization/mineralization reactions for both ester sulfate and carbon-bonded sulfur. Although this
3-14
-------
model was developed in part to provide a set of sulfur-cycling subroutines for incorporation In integrated
watershed chemistry modefs, there are at present insufficient data for its general usage. Rate constants
have been defined for only a few sites under very limited conditions, and supporting soil chemistry data
(e.g., quantification of soil organic sulfur pools) do not exist except for a few research sites.
Several dynamic watershed chemistry models have been developed to describe or project
watershed acidification, and all consider sulfate retention in some way. Jenne et al. (in press) have
recently evaluated and compared process representation, including sulfur processes, for the three models
used in the DORP. The Model for Acidification of Groundwater in Catchments (MAGIC), developed by
Cosby et al. (1985a,b; I986b), uses a nonlinear isotherm (Langmuir) to partition sulfate between dissolved
and sorbed phases in the soil; phase equilibrium is assumed at each time step. For simulations of lake
chemistry, MAGIC optionally incorporates the Baker et al. (1986b) model of intake retention. The
Integrated Lake/Watershed Acidification Study (IIWAS) model (Chen et al., 1983; Gherini et al., 1985) can
use either a linear or Langmuir function to describe inorganic partitioning in the soil; a first-order in-lake
retention component also can be Included as appropriate. The Enhanced Trickle Down (ETD) model of
Schnoor and coworkers (Schnoor et al., I986b; Lee, 1987) was originally developed for seepage lakes
In the Upper Midwest; early versions assumed steady state for sulfate in the terrestrial system and used
an empirically defined zero-order function for in-lake retention. Current versions of the model include a
linear isotherm to describe adsorption by soils. Application of these models in DDRP Level III Analyses
is discussed in Section 10.
Along with the models used in the DDRP, the Birkenes model (Chrlstophersen and Wright, 1981;
Christophersen et at., 1982) has been used for simulation of watersheds in Norway and Canada. The
Birkenes model was developed by Christophersen and coworkers for a catchment in Norway having thin
soils with high organic content and low adsorption capacity. Sulfate transformations in soils are
represented by empirically derived equations (fit to stream sulfate concentrations). The upper soil horizon
includes a constant net mineralization term, while transformations in the lower soil compartment are
described by an exponential function with an empirically derived half-time (45 days) and equilibrium
3-15
-------
value. The exponential function was not designed to describe a specific process, but is believed to
represent some combination of adsorption and microbially mediated transformations.
3.3.5 Summary
Sulfur input/output budgets for individual sites and for regional lake or stream populations indicate
major differences in sulfur mobility in watersheds of the NE and the SBRP (Rochelle et al., 1987; Rochelle
and Church, 1987). Although several terrestrial and in-lake processes may contribute to the observed
differences in budget status, two processes are believed to dominate sulfate control. The first process,
accumulation of soil organic sulfur, apparently does not contribute significantly to net sulfur retention in
most northeastern watersheds (Fuller et al., I986a,b; Rochelle et al., 1987), but may be a net sulfur sink
in the SBRP. Due to a lack of data describing soil organic sulfur pools and a paucity of kinetic data,
however, the actual importance of organic transformations as sulfur sinks cannot be evaluated at regional
scales for regions with significant net sulfur retention.
The second process presumed to have significant influence on sulfate mobility in the two regions
is adsorption. Until recently, it was often assumed that differences in regional sulfur budget status resulted
from differences in soil age between the Northeast and Southeast (young, glaciated northern soils are
developed to only a relatively shallow depth, have few hydrous oxides and secondary clay minerals, and
thus low sorption capacity; in contrast, older southeastern soils are often developed to a depth of several
meters and have abundant secondary minerals, hydrous oxides and clay). Recent data (e.g., Johnson
et al., 1980; Johnson and Todd, 1983) suggest the previous assumption was partly correct. The glaciated
northern soils have much lower day content and poorer development of the C horizons than typical
southeastern soils, but B horizons of many northern soils have iron and aluminum contents and adsorbed
sulfate concentrations comparable to those of southern soils and have significant capacity to adsorb
additional sulfate (Johnson et al., 1980; Johnson and Todd, 1983; Fuller et al., 1985). An important, but
more recently recognized difference between the regions is the higher organic content of many northern
soils, which acts to inhibit or "block" adsorption (Couto et al., 1979, Johnson et al., 1980; Johnson and
Todd, 1983).
3-16
-------
Differences in soil physico-chemical variables related to adsorption, coupled with significant
differences in historic sulfur loadings to the two regions (Gschwandtner et al., 1985), probably account
for most of the observed difference in sulfate mobility between the NE and SBRP. Although many SBRP
watersheds are retaining a major portion of incident sulfur deposition, soils have a finite sorption capacity,
and there are recent observations of increasing sulfate in many streams of the region (e.g., Smith ami
Alexander, 1983; Swank and Waide, 1988). These trends suggest that effects of acidic deposition are likely
to increase in softwater systems of the region over the next few decades. Conversely, because most
northeastern systems are already at or very near steady state for sulfur, changes in sulfate concentration
under current deposition loadings will be small, if deposition were to change in the NE, the relatively low
sorption capacity of typical soils in the region suggests that resulting increases or decreases in surface
water sulfate would also occur quickly, probably within a few decades or less. Predicted responses of
watersheds in the two regions to continued loading at current or altered levels of deposition are
addressed in both Level II and HI Analyses in the DDRP, and are described in Section 9.2 (sulfate only)
and Section 10 (sulfate and associated changes in cations).
3.4 BASE CATION SUPPLY PROCESSES
3.4.1 Introduction
The second major group of processes affecting surface water acidification is composed of those
processes or reactions responsible for supplying base cations (i.e., Ca2+, Mg2+, Na+, and K+) to surface
waters. Recently, Driscoll et al. (1989b) have demonstrated that a good correlation exists between
changes in base cation deposition at Hubbard Brook Experimental Forest and changes in cation fluxes
in streams. Based on these results, these authors have hypothesized that the deposition of base cations
may be a primary factor in regulating surface water acidification. However, we (Hofdren and Church, in
review) feel that the processes that have previously been suggested as primary factors controlling surface
water composition (Galloway et al., 1983a; Reuss and Johnson, 1986) are sufficient to explain the results
of Driscoll et al. (1989b). As such, the focus of the remainder of this section is on primary mineral
weathering and cation exchange (Figure 3-2). The weathering of primary minerals is the ultimate source
for base cations. Cations released during weathering, especially K+, Ca2+, and Mg2+ are extensively
3-17
-------
0)
fits
o
£
o>
5
3-18
-------
cycled between the actively growing biomass and the forest litter layer. Within the solum, cations may
precipitate as secondary minerals. The extent to which the base cations are released into solution plays
a major role in determining the response of a system to acidic deposition.
In evaluating the potential for weathering or cation exchange to neutralize incident deposition, it
is critical to be aware of the time frames over which the two processes operate and to understand the
potential for depletion of buffering capacity. In general, weathering is a slow process that releases base
cations and silica to soil and surface waters at a more or less constant rate over long periods of time.
The capacity for weathering to supply base cations is dictated by the exposed surface area of cation
release is large. The primary concern about weathering, therefore, is not capacity, but rather the rate at
which the reactions occur. The rate of weathering depends on the exposed surface areas of reactive
minerals In soils or aquifers and on the hydraulic contact between minerals and soil waters. Rates can
vary widely and, in some watersheds, may not be sufficient to neutralize incident deposition. These
systems are thus potentially susceptible to adverse effects of acidic deposition.
In those systems in which the weathering rates are low, cation exchange reactions can ameliorate,
at least transiently, the effects of acidic deposition. Unlike weathering, exchange reactions are rapid,
usually approaching steady-state conditions within several hours in static systems. The ability of exchange
processes to neutralize incident deposition depends both on the size of the exchange reservoir and on
its exchange properties. In the regions of interest to this study, the exchange reservoir is probably small.
The northeastern soils are young and have low clay mineral content, and soils in the SBRP are highly
weathered. While the organic horizons of soils in these regions do have substantially larger cation
exchange capacity (CEC) values than do their underlying mineral horizons, these organic components
of the soils tend to be quite acidic (i.e., pH < 4.0). As such, exchange reactions involving organic
horizons do not contribute substantially to the ANC-generating capacity of the soils as integrated systems.
In soils with low weathering rates, base cations will be depleted from the exchange complex as a direct
consequence of acidic deposition.
3-19
-------
The exchange reaction, in which acid cations (H+ or AI3+) replace base cations on the exchange
complex, is essentially a buffering process. This reaction affords some degree of protection on soil and
surface waters in terms of limiting changes in pH and ANC. If the cation resupply rate from weathering
is less than the rate at which the increased acidic deposition removes cations from the exchange
complexes, then through time, as the reservoir of base cations is depleted, the observable effects on the
soils and associated surface waters increase. Initially, when base saturation is high, changes in the
projected surface water ANC are relatively small. When the base saturation for the soils is reduced to
only a few percent, however, the projected changes in ANC are much greater (resulting In extremely low
projected ANC) per unit change in base saturation (Cosby et al., 1985b; 1986a).
In the regions examined in DORP, certain soils could experience significant depletions of base
cations over the next 50 to 100 years. If soils undergo a significant reduction in base saturation, the
associated soils and surface waters will experience parallel declines in pH and ANC. Such changes have
already been well documented in the northeastern United States, eastern Canada, and Scandinavia. The
major concerns, then, focus not on whether changes have occurred, but rather on the possible extent
of changes anticipated to occur in selected regions over the next several decades.
3.4.2 Factors Affecting Base Cation Availability
Base cations actively cycle through virtually all ecosystems. In forested watersheds, the cations
can be delivered via deposition, both wet deposition and various forms of dry deposition. Alternatively,
cations can be derived by the weathering of bedrock underlying these systems. In the ecosystem, base
cations actively participate in a number of cycles. Vegetation actively cycles Ca2+, Mg2+, and K*
through the upper portion of the soil (Likens et al., 1977; Johnson et al., 1988a). Easily observable
changes in the concentrations of cations in the exchange pools occur on a seasonal basis as a result
of these processes. On a slower time scale, the base cations participate in the formation (and,
subsequently, the degradation) of secondary minerals (Garrels and Mackenzie, 1967). Although in young
soils these secondary minerals might represent only a small fraction of the mass of cations cycling
through the ecosystem, they serve as a reasonably accessible reservoir for cations on time scales of
3-20
-------
weeks to months. Cations can be removed from watersheds by any of a number of processes. Surface
water runoff and deep groundwater percolation both export cations from watersheds, and cations can be
transported fluvially via suspended solids loads. Aggrading forests also act as a sink for base cations,
and, as such, have an acidifying effect in forest soils (Nilsson et al., 1982; Johnson et al., I988a).
Two processes provide primary buffers against adverse effects of acidic deposition: mineral
weathering and base cation exchange. Other processes (e.g., cation uptake by vegetation), however, can
be quantitatively important aspects of the elemental cycles in watersheds (Likens et al., 1977).
Perturbations or disruptions, such as logging, to the biogeochemical cycles not only have dramatic effects
on the cation balances in these systems, but also play a major role in soil acidification (Nilsson et al.,
1982).
3.4.2.1 Mineral Weathering
The weathering of rock-forming minerals is the primary source for base cations in surficial
environments. Because of differences in composition and reactivity, different minerals contribute in varying
degrees to the ability of soils or watersheds to neutralize incident deposition. A summary of some the
principal factors and processes is given below.
3.4.2.1.1 Primary rock-forming minerals and their rates of weathering -
A list of the major rock-forming minerals is provided in Table 3-1 along with information on their
relative weathering rates and responsiveness of the rates to changes in pH. Reactions Involving the
primary minerals listed in Table 3-1 are, for all practical purposes, unidirectional. These minerals are
unstable in soil environments. They weather to form secondary phases, such as the clay minerals, that
are thermodynamically more stable and kinetically favored for formation.
Secondary minerals, such as kaolinite, smectites, or allophane, are not Included in the table. The
clays may contribute to ANC through both the degradation of their silicate frameworks and their ion
3-21
-------
Table 3-1. Major Rock Forming Minerals and Their Relative Reactivities
Reaction Bate Hydrogen Ion
Mineral (moles m s ) Rate Coefficient
OlMne 1.2 x 10'12 (aH+)0'6
Pyroxenes 1.0 x 10°° (aH+)0'8
Amphiboles 1.4 x 10'10 (aH+)0'7
Muscovite 2.6 x 10'13 (aH+)a10
Feldspars m (a"
Albite 1.2 x 10'10
Oligoclase 2.0 x 10""
Anorthite 5.6 x 10"9
Microcline 5.0 x 10
Orthodase 1.7 x 10"12
Quartz 4.1 x 10"14 (aH+)°
3-22
-------
exchange properties. Discussion about how these minerals contribute to the transient buffering of soil pH
and ANC in the absence of primary mineral weathering is presented in Section 3.4.3.
As indicated In Table 3-1, different minerals weather at different rates. Some, such as calcite,
weather rapidly and provide considerable ANC to soils and surface waters. Watersheds underlaid by
limestone, therefore, are rarely at risk to acidification (Hendrey et al., 1980). Conversely, some
rocks/minerals are slow to weather and generate little or no ANC. For example, watersheds underlaid
by the quartzites are usually considered to be sensitive to the adverse effects of acidic deposition
(Hendrey et al., 1980; Rapp et al., 1985; Shilts, 1981).
The rates listed in Table 3-1 were obtained from laboratory studies. As such, these rates are the
maximum values expected to occur in field settings. In most cases, rate estimates obtained from field
studies are one to three orders of magnitude slower than those listed in the table (Velbel, 1986b; Paces,
1973). Although up to about one order of magnitude might be due to temperature effects, reasons for
these discrepancies are currently not well understood. Other potential processes contributing to rate
suppression include poisoning of active surfaces by organic coatings or mineral precipitates and non-
continuous reactions caused by the wetting/drying cycles in soils (Velbel, I986a).
Given that it is difficult to infer rates of weathering of primary minerals from field studies, it is
virtually Impossible to obtain information concerning how changes in the soil environment influence
estimated rates. Essentially all of the data available on the effects of pH, organic interactions, and
temperature are derived from laboratory studies. The major results from these efforts are summarized
below.
3.4.2.1.2 Laboratory studies on rates and mechanisms -
Over the past three decades, considerable efforts have been made to determine the rates and
mechanisms of weathering of most of the major rock-forming minerals. Of the factors affecting observed
3-23
-------
rates, pH, organic interactions, and temperature probably exert the most significant influence in
determining the rates that are realized in the field. Other factors, specifically the nature and extent of
mineral surface complexation by inorganic anions and the postulated presence (or lack thereof) of
leached layers on weathered mineral surfaces, are likely to be important determinants of dissolution rates
in soil environments. At this point, however, there is not sufficient information concerning these processes
to understand how changing environmental conditions would influence observed rates through these
processes.
3.4.2.1.2.1 Dependence on pH -
Over the past two decades a number of laboratory studies have been undertaken to determine
changes in the reaction rates of various common rock-forming minerals as functions of hydrogen ion
activity (e.g., Wollast, 1967; Helgeson et al., 1984; Chou and Wollast, 1985; Holdren and Speyer, 1985).
Results from these studies are summarized in Table 3-1. As might be expected, different minerals
respond differently to changes in solution pH. Reaction rates for some minerals, such as quartz, are only
marginally affected by pH in acidic to circumneutral solutions. At the other extreme, calcite and dolomite
reaction rates are quite responsive to changes in hydrogen ion activity. For the major soil-forming
minerals present in the study regions, i.e., the feldspars, various micas, and hornblende, observed reaction
rates tend to vary as functions of (aH*)0"2 to (aH+)0'5. Therefore, since soil pH values are expected to
change by only a few tenths of a pH unit in response to acidic deposition, the magnitude of this pH
effect on the rates of mineral weathering in soils should be affected by less than a factor of about 2.
3.4.2.1.2.2 Organic interactions -
One of the more poorly understood aspects of weathering has to do with the effects of organic
iigands on the rates of reaction. Comparisons between laboratory-generated data and field observations
have resulted in a clear understanding of the role of organics in the weathering process. If the role of
organics on the rates of weathering is poorly understood, then our understanding of how acidic
deposition will affect these rates is even less understood. With the information currently available, two
reasonable hypotheses can be developed.
3-24
-------
First, if organics do not have a major influence on the reaction rates of primary minerals (Mast and
Drever, 1987), then the effect of acidic deposition on the reaction rates will be limited to the direct effects
caused by changes in the hydrogen ion activities of the solutions bathing the soil particles. Under these
conditions, weathering rates would most likely increase slightly in response to imposed environmental
conditions.
On the other hand, if organics play an active role in weathering, then the interaction between
acidic deposition and the organics could suppress mineral weathering. It has recently been hypothesized
(Krug and Frlnk, 1983; Krug et a)., 1985; Sullivan et al., in press) that the mobility of natural organics is
depressed in more acidic environments. The decreased mobility, and hence concentration, coupled with
their effect on reaction rate could conceivably cause net decreases in weathering rates in certain
environments, it should be stressed that little is actually known about the effects of organics on
weathering rates under field conditions. The above scenarios are, at best, speculative, but they do
present the range of expected effects under different conditions.
3.4.2.1.2.3 Temperature -
The third major environmental influence on observed reaction rates is temperature. Very little
experimental work has been undertaken at environmentally representative temperatures (i.e., in the 0 °C
to 10 "C range). Results from a number of studies suggest, however, that the activation energies for
dissolution for most common silicate minerals are in the range of 60 to 80 kj moi*1. Assuming activation
energy in this range and a mean average annual temperature of 4 °C, dissolution rates are probably
seven to eight times slower in the field than those observed in laboratory settings (see Table 3-1).
3.4.2.2 Cation Exchange Processes
The second major base cation-related process contributing to watershed buffering is base cation
exchange. Exchange pools are dynamic reservoirs. Under steady-state conditions, the base cation content
of the exchange pool represents a dynamic balance between supply from mineral weathering and organic
matter mineralization and removal processes, including uptake by vegetation and leaching to ground and
3-25
-------
surface waters. On annual time scales, the soil exchange complex is in equilibrium with the soil solutions.
Hence, the net uptake or net desorption of cations from soil exchange complexes should change only
slowly in response to long-term weathering processes. In the absence of system perturbations and under
steady-state conditions, base cations derived from weathering should be effectively passed through, either
to accreting biomass or to ground or surface waters. The exchange pool reflects the concentrations
observed in soil solutions, but again, in the absence of a system perturbation, is neither a source nor sink
for the base cations on a long-term basis.
With increased levels of H+ inputs, this balance changes. The increased acidity of the deposition
increases the leaching of base cations from the exchange complex, replacing them with acid cations,
namely H* or AT3* (Reuss et al.t 1987). In addition, the Increase in the total anionic content of the soil
water requires an increased total cationic flux from the soil (Johnson and Cole, 1980; Seip, 1980). The
increased leaching resulting from the increased acidity is a transient phenomenon. Eventually, a new
steady state is attained that reflects the properties of the exchange complex and the increased anionic
concentrations in soil solutions. In the short term, then, the surface water pH and ANC are buffered by
the increased leaching of base cations from the soil exchange complex. Concurrently, the soil pH and
base saturation of the soil are reduced. As the exchange approaches the new steady state, the balance
between the flux of H* to the ecosystem and the average primary mineral weathering rate will determine
the final pH and ANC values for soils and the associated surface waters.
3.4.2.2.1 Types of exchangers -
The soil exchange complex is composed of essentially three types of exchangers: day minerals,
organics, and metal oxides. Within each of these broad categories of exchangers, several types of sites
can actively participate in exchange reactions. For example, clay minerals can have both pH-dependent
surface charges and permanent, structurally based sites acting as exchange sites. The two types of
exchangers that are of most concern are the clays and the organic exchangers. The metal oxides, at
the pH values of forested soils, typically have positively charged surface sites. As such, they represent
sites for anion exchange (see discussion in Section 3.3) rather than for cation exchange.
3-26
-------
3.4.2.2.2 Factors affecting the exchange process -
A number of factors that affect exchange processes can be most easily described when the
process is conceived in terms of an exchange reaction. For example, for the reaction:
3Ca2+ + 2(=S-)AI = 2AI3+ + 3(=S-)Ca (Equation 3-1)
where the (=S-) indicates the surface exchange site, the reaction characteristics can be estimated in
terms of a mass action equation:
= {AI3+}2 {XCa]3/{Ca2+}3 py2 (Equation 3-2)
where the species in braces, {x}, are the activities of the aqueous species, and those in the bracket,
[x], are the mole fractions of the associated solid exchangers. The selectivity coefficient, Kexac, is not a
thermodynamic constant because no attempts have been made to include the rational activity coefficients
for the solid phase exchangers. Nevertheless, this expression can be used to understand the effects
that various perturbations might have on the system.
It should also be pointed out that, in the above expressions, aluminum is being used as a surrogate
for the hydrogen ion. In soils, AT3"1" comprises the bulk of the acid cations on exchange sites. In addition,
AI3+ activities, while having a major role in exchange reactions, are frequently regulated by other
reactions such as the dissolution of gibbsite-Jike phases (Reuss, 1983).
Soils exposed to increased hydrogen ion activities undergo a number of possible changes. For
example, aluminum activities increase at lower soil pH values simply because the solubility of gibbsite
increases with decreasing pH. In response to the changing hydrogen ion regime, the activity of calcium
in soil solutions would have to increase, or the ratio of calcium to aluminum on the exchange sites
would have to decrease (i.e., there would be a net replacement of aluminum for calcium on exchange
sites).
3-27
-------
tn addition to acidic deposition, other anthropogenic activities can affect the cation balance of soil
exchange pools. Perhaps one of the better documented activities is the effect of whole tree harvesting
(Johnson et al., !988a; Reynolds et ai., 1988). Biomass is a major and dynamic reservoir for base cations
in most watersheds. Cations are absorbed from the solum during the spring and summer growing
seasons and then partially recycled to the soil in the fall and winter with leaf fall and organic matter
mineralization. Afforestation places an increased demand on the cation supply in soils as cations are
retained in the aggrading biomass. This process has an additional acidifying effect on forest soils.
3.4.3 Modelling Cation Suoolv Processes
3.4.3.1 Modelling Weathering
In general, weathering models used to describe watershed-scale processes have been developed
along one of two conceptual lines: whole watershed/mass balance and kinetics. The most commonly
used models are the watershed/mass balance-type models (Bricker et al., 1968; Cleaves et al., 1970;
Garrets and Mackenzie, 1967; Clayton, 1986; Creasey et al., 1986; VelbeJ, 1985, 1986a; Dethier, 1986).
These models are based on selected sets of reactions and are calibrated to specific systems. The models
work best in systems with simple mineralogies. However, the application of this type of model for studying
the impacts of acidic deposition is limited because, in general, the models do not distinguish between
primary mineral weathering and transient, enhanced leaching of base cations from soil exchange sites.
Therefore, watershed models are most applicable to systems at steady state with regard to incident
deposition.
More recently, kinetic models have begun to appear (e.g., Furrer et ai., 1989). As discussed
previously, application of these models to field situations is only now becoming possible. Discrepancies
between laboratory and apparent field rates of weathering for individual primary minerals result in poorly
constrained models. As more Is learned about processes controlling rates of weathering in the field,
kinetic models should play an increasing role in projecting the effects of acidic deposition.
3-28
-------
In spite of the state of weathering models, several integrated watershed-type models (e.g., Cosby
et al., 1985a,c; Gherini et al., 1985; Nikolaidis et al., 1988) incorporate weathering 'modules" within their
frameworks. In some models, primary mineral weathering reactions are lumped with exchange processes
to yield net cation transfer rates (Nikolaidis et al., 1988). Other models treat the processes independently
(Cosby et at.. 1985a,c; Gherini et al., 1985). In either case, the weathering modules tend to be used
primarily in calibrating stream or lake compositions, because data needed to determine these parameters
for individual soils and watersheds are not generally available in sufficient detail to set the values a priori.
3.4.3.2 Modelling Cation Exchange Processes
In contrast to the situation with mineral weathering, cation exchange processes have been examined
in detail and have been modelled extensively. Two types of models have been used in describing
exchange processes: mass action models and heuristic models. Both types of models, it should be
stressed, are empirical and depend on obtaining appropriate descriptive data from the field sites being
studied.
The mass action models are based on specific reactions such as the one illustrated in Equation
3-1. For example, Reuss (1983) and Reuss and Johnson (1985, 1986) have developed soil exchange
models incorporating the effects of soil gas pCO2 and soil solution ionic strength as well as the properties
of the exchange reactions. Reuss's approach has the advantage of being responsive to a wide range of
environmental conditions. The models, however, generally tend to be data intensive.
Heuristic models, in contrast, are based on known or observed relationships between various soil
parameters. For example, Bloom and Grigal (1985) developed a model based on the relationship between
soil pH and base saturation in selected Minnesota soils. These models have the advantage of providing
reasonably accurate descriptions of closely related soils or horizons, and they are less data intensive than
the mass action-type models. They are not as flexible, however, in modelling the effects of perturbations
to a soil (e.g., changes in soil pCO2).
3-29
-------
In DDRP, both types of models are being used to examine the effects of acidic deposition on the
base saturation status of soils in forested watersheds in the NE and SBRP of the United States. Details
regarding mode) formulations are presented in Section 9.
3-30
-------
SECTION 4
PROJECT APPROACH
4.1 INTRODUCTION
As H.B.N. Hynes (1975) once noted, "We must not divorce the stream from its valley in our
thoughts at any time. If we do we lose touch with reality." Although surface waters can be affected by
acidic deposition originating from emissions many miles distant, the concept of the watershed as a unit
is critical in understanding current and future aquatic effects, indeed, for drainage lake and reservoir
systems in the Northeast, Upper Midwest, and Southern Blue Ridge Province, most ANC production
occurs as a result of biogeochemical processes within the surrounding watershed (Section 7.2; Shaffer
et al., 1988; Shaffer and Church, 1989).
Because of the importance of watershed processes (especially those occurring in soils) in
determining future aquatic effects, new data on these processes and on related soil pools and capacities
were required, initially, we considered using existing regional soils data in the DDRP analyses. Existing
soils databases, however, have serious deficiencies with respect to the needs of the Project. First,
because of the economic importance of croplands, such data are available primarily for lowland
agricultural regions; surface water acidification, however, occurs principally in relatively undisturbed upland
systems. Second, such databases generally do not include chemical characterizations of a number of
key variables relevant to soil chemical interactions with acidic deposition (e.g., sulfate adsorption capacity
and unbuffered cation exchange capacity).
After consideration of these factors, we decided that a new regional soils database was required,
thus necessitating a major soil survey effort (Sections 5.1 - 5.5; also see Lee et al., 1989a; Church, in
press). We further concluded that this survey should enable the specific soils (and specific soil types)
to be linked with the NSWS databases that describe the chemistry of low ANC lakes and streams.
Accordingly, we adopted the approach outlined in this section and illustrated in Figure 4-1.
4-1
-------
Watershed Selection
Watershed Mapping
T
Development of Soil *
Sampling Classes
Soil Sampling and
Reid Measurements
Soil Preparation
Chemical/Physical
Laboratory Analysis
Supporting Regional
Datasets
Database Management
Figure 4-1. Steps of the Direct/Delayed Response Project (DDRP) approach. Asterisks denote
steps that received significant support from Geographic Information Systems (GtS)-based activities
(Campbell and Church, 1989; Campbell et al., in press).
4-2
-------
4.2 SOIL SURVEY
4.2.1 Watershed Selection
The selected DDRP watersheds comprise a high interest subset of lake and stream systems
surveyed in the NSWS. A sufficient number of watersheds was selected to allow for (1) reasonably
broad regional coverage and (2) statistical examination of interrelationships (deposition:watershed
characteristics:surface water chemistry) and model projections of response. Because watersheds were
selected as probability samples, results can be extrapolated to a specified population of interest. Further
details on watershed selection are provided in Section 5.2 (also see Lee et al., 1989a). Regional
population estimation is discussed in Section 6.
4.2.2 Watershed Mapping
Maps of soils, vegetation, land use, and depth to bedrock were prepared for each DDRP watershed
by the USDA Soil Conservation Service (SCS). Bedrock geology was obtained from existing state
geology maps. SCS mapping was at a scale of 1:24,000 and was at a 'second order* intensity
(comparable to most county soil surveys). An important part of this mapping was the regional correlation
of map unit names and definitions, a common procedure at the county or state level but a much greater
challenge at the regional scales of this Project. Additional maps of land use and wetlands were
developed by interpreting infrared stereo aerial photographs at a scale of 1:12,000, with land use
delineated to 2.5 ha and wetlands to 0.4 ha (LJegel et al., in review). Watershed mapping is discussed
in detail in Section 5.4.
4.2.3 Sample Class Definition
Because many soil components were mapped in the study regions (e.g., about 600 in the NE),
characterizing each one physically and chemically was not feasible. Instead, sample classes were defined
for each region, and individual soils were assigned to those classes based on (1) expert knowledge of
the soils mapped and (2) expectations of the potential responses of those soils to acidic deposition. Soils
selected from these classes were sampled across the study regions. Soils were aggregated within
sampling classes to develop characterizations (e.g., class means and variances) that were used to
4-3
-------
"rebuild* or represent (e.g., by mass or area weighting) the characteristics of study watersheds. Details
of the sample class selection are provided in Section 5.5.1 and by Lee et al. (1989b).
4.2.4 Soil Sampling
We developed a procedure that allowed random selection of soil sampling sites within the context
of expert classification. This procedure was designed to ensure that adequate and complete coverage
was obtained of both the sampling classes and the watersheds across the regions. Details are given In
Sections 5.5.2 and 5.5.3.
4.2.5 Sample Analysis
Samples were analyzed by independent soil laboratories under contract through EPA's
Environmental Monitoring and Systems Laboratory - Las Vegas (EMSL-LV). A rigorous quality assurance
program was implemented to ensure the quality of these analyses. Sample analyses are discussed in
Section 5.5.4.
4.2.6 Database Management
Management of the soil survey databases involved operations at the Environmental Research
Laboratory - Corvallis (ERL-C), EMSL-LV, and Oak Ridge National Laboratory (ORNL). Centralized
database management was maintained at ORNL with backup at ERL-C. Database management activities
in the DDRP are further discussed in Sections 5.4 and 5.5.
4.3 OTHER REGIONAL DATASETS
Because of the regional nature of the Project, we required estimates of precipitation, atmospheric
deposition (wet and dry), and surface water runoff (as runoff depth) that were generated in a standardized
manner across the eastern United States. Study sites for the DDRP were selected statistically, and most
sites had no direct information for the above variables. Furthermore, time and budgetary constraints
precluded the instrumentation of sites and, thus, the direct acquisition of such data. These estimates,
therefore, had to be developed within the Project.
4-4
-------
4.3.1 Atmospheric Deposition
The acquisition/development of internally consistent regional datasets on atmospheric inputs was
a challenging task. In effect, two types of datasets were developed. One dataset representing 'long-
term average annual" conditions was constructed for use in the correlative analyses in the Project (Section
4.4.1). The temporal resolution of this dataset is annual. A second atmospheric deposition dataset was
constructed for use in the watershed modelling analyses of the Project (Section 4.4.3). This dataset
provides daily estimates of precipitation (to "drive" the hydrologic subroutines of the watershed models)
and monthly inputs of atmospheric deposition.
For both datasets, precipitation amount and chemical concentrations were estimated from the Acid
Deposition System (ADS) network (Wampler and Olsen, 1987). Wet deposition was determined as the
product of these measures. Dry sulfur deposition was estimated from simulations using the Regional Acid
Deposition Model (RADM) (R. Dennis and S. Seilkop, personal communication and unpublished internal
report, 1987; Clark et al., 1989). Estimates of dry deposition for other ions were not directly available
from any source and had to be developed within the Project. The atmospheric data acquisition and
development are described in Section 5.6. To our knowledge, this is the first time that such a complete
deposition database has been developed on such an extensive regional basis.
4.3.2 Runoff Death
Because direct runoff measurements were lacking for the selected watersheds, we relied upon
regional maps of annual runoff depth. Investigation of the maps available at the start of the Project
yielded no single map with a resolution finer than five inches of runoff depth. We therefore enlisted the
U.S. Geological Survey (Madison, Wl) to produce an annual runoff map for the period 1951-80 (Krug et
al., in press), corresponding to long-term precipitation records used to estimate deposition. As part of
this work we performed a quantitative uncertainty analysis of estimates of long-term runoff from the Krug
et al. map (Rochelle et al., in press-b). Details of the development and application of these runoff
within the Project are given in Section 5.7.1.
4-5
-------
4.4 DATA ANALYSIS
A variety of analyses have been undertaken within the Project. Many analyses were performed by
the EPA and contractor staff at ERL-C. Others were performed by extramural cooperators in close
coordination with ERL-C staff. Data analyses within the Project are classified into three "levels", according
to the complexity of the analyses and the degree of reliance upon knowledge, or hypotheses, of process
interaction within watersheds. For example, Level I Analyses presuppose the least about our knowledge
of the way watersheds "operate", whereas Level III Analyses depend upon more comprehensive
knowledge of system behavior.
4.4.1 Level I Analyses
Level I Analyses include constituent input/output budget estimates and statistical analyses. The
leaching of the mobile anion sulfate is considered to be a key process in long-term acidification.
Accordingly, one part of the Level I Analyses is to determine retention of atmospherically deposited sulfur
within watersheds. We examined annual watershed input/output budgets for sulfur, based on detailed
studies at a few sites and relatively sparse data from many sites. These analyses and results are
presented in Section 7 (for interim results see Rochelle et al., 1987; Rochelle and Church, 1987).
The other part of Level I Analyses is the statistical evaluation of interrelationships among
atmospheric deposition, mapped watershed characteristics, soil chemistry, and current surface water
chemistry (e.g., see Rochelle et al., in press-a). One goal of this evaluation is to verify that the processes
and relationships incorporated in the Level II and III Analyses reasonably represent the systems under
study. These analyses (presented in Section 8) are complicated by the fact that the ANC range of the
study systems is relatively narrow.
4.4.2 Level II Analyses
The Level II Analyses use relatively restricted models of key processes that regulate the dynamics
of (1) base cation supply and (2) watershed retention of atmospherically deposited sulfur. The models
are used to project how these processes might affect conditions in the DDRP watersheds and in the
4-6
-------
surface waters that drain them under continuing or altered future levels of atmospheric sulfur deposition.
The models used to investigate and project base cation supply are the "Bloom-Grigal" model (Bloom and
Grigal, 1985) and the "Reuss-Johnson" model (Reuss and Johnson, 1985, 1986). Watershed sulfur
retention is modeled as sulfate adsorption according to the approach presented by Cosby et ai. (1986b).
The models are run independently of one another and of other watershed factors, such as forest
accretion, that might affect watershed response. The analyses and results are given in Section 9.
4.4.3 Level III Analyses
In the DDRP Level III Analyses, integrated watershed models are used to project future effects of
atmospheric sulfur deposition on surface water chemistry. Three models specifically developed to
investigate the effects of acidic deposition on watersheds and surface waters are being applied: (1) the
Model for Acidification of Groundwater in Catchments (MAGIC) (Cosby et al., 1985a,b,c; Cosby et al.,
1986a,b); (2) the Enhanced Trickle Down (ETD) Model (Lee, 1987; Nikolaidis et al., 1988; Schnoor et ai.,
1986a); and (3) the integrated Lake-Watershed Acidification Study (ILWAS) Model (Chen et ai., 1983;
Gherini et al., 1985).
These three models were selected on a competitive, externally peer-reviewed basis via EPA's
standard Cooperative Agreement funding mechanism. A sequence was followed that included a public
announcement of the Request for Proposals, committee review of pre-proposals, and external peer review
of full proposals. Candidates were requested to submit for review only those models that met the
following criteria:
The model to be applied must be capable of time-variable predictions of the effects of acidic
deposition on the chemistry of waters delivered from terrestrial systems to streams and lakes.
A simplified mechanistic or process-oriented approach is preferred. As such, the model
should include representation of those processes commonly considered to be the most
important within soil systems (e.g., anion retention, cation exchange, mineral weathering, CO2
dynamics). It Is not required that interactions of deposition within vegetative canopies be
simulated, nor is it required that within-lake or within-stream interactions be included (e.g.,
sulfate reduction in anoxic hypolimnia or in sediments, exchange reactions within sediments).
The model must contain its own soil hydrologic component, but significant lumping within
this component is allowable (e.g., a standard two-compartment representation).
4-7
-------
Although further testing and refinement of the model structure and application is encouraged
and expected during the course of the project, the model to be used must be reasonably
complete and tested, and in the possession .of the applicant at the time of proposal
submission.
The three models are run using common datasets for forcing functions (e.g., rainfall, runoff,
atmospheric deposition) and state variables (e.g., soil physical and chemical variables). Projections of
changes in annual average surface water chemistry are being made for the Northeast (NE) and the
Southern Blue Ridge Province (SBRP) for at least SO years for two scenarios of atmospheric sulfur
deposition: (1) continued deposition at current levels (for both regions) and (2) altered deposition over
the next 50 years, i.e., a decrease in the NE and an increase in the SBRP (see Section 5.6). Because
the models are being applied to watersheds having sparse, but internally consistent regional datasets,
reliability checks are being performed using much more complete (in terms of time and space) data from
intensively studied watersheds. Such analyses for three of the ILWAS/RILWAS lakes in the NE (Chen et
al., 1983) are presented in Section 10. Additionally, confirmation activities continue for White Oak Run,
VA (Cosby et al., I985c) and Coweeta watersheds 34 and 36 (Swank and Crossley, 1987) and will be
presented In the DORP Mid-Appalachian Report. The Level III Analyses and results are presented in
Section 10.
4.4.4 Integration of Results
To a large extent, de facto integration of interim results has taken place during the course of the
Project with feedback occurring among all levels of analyses. As noted in Section 2, the principal
"bottom line" of the DDRP (i.e., time dynamic projections of the long-term effects of sulfur deposition on
regional surface water chemistry) comes from the dynamic watershed simulations performed in the Level
111 Analyses. The manner in which the results from the Level I and II Analyses support and expand upon
the Level III findings is presented in Section 11 of this report.
4-8
-------
4.4.5 Use of a Geographic Information System
A Geographic Information System (GI8) has played an integral part in the DDRP (Campbell and
Church, 1989; Campbell et al., in press). Initial CIS-based activities were data entry (Section 5.4.1.7),
display, and spatial analysis of the watershed mapping data from the Soil Survey. Activities have been
greatly expanded, however, to include data aggregation, analysis, and display at a variety of scales and
projections. The GIS outputs are particularly useful in communicating results to a variety of audiences.
4-9
-------
SECTION 6
REGIONAL POPULATION ESTIMATION
6.1 INTRODUCTION
The purpose of this section is to describe the procedures used to extrapolate analyses on individual
watersheds to the target populations In the study regions. This process of extrapolation is called
population estimation.
6.2 PROCEDURE
6.2.1 Use of Variable Probability Samples
Probability samples were selected for lake watersheds in the Northeast and stream watersheds in
the Southern Blue Ridge Province (SBRP). Any quantity that can be defined for a sample unit (i.e., for
each watershed) can be extended to a corresponding population quantity through the probabilistic
structure of the sample. The quantity can be a measured variable or a model-based estimate. It can
be a number, a vector, or a function. In the Eastern Lake Survey (ELS), most quantities were measured
values, and the measurement error tended to be small relative to the sampling variation, in contrast to
the ELS, many of the quantities produced in the DDRP are model outputs believed to have significant
uncertainty associated with them. The population estimation techniques provided below apply to any
probability sample with defined inclusion probabilities. Thus, they are applicable to any identifiable subset
of the DDRP sample. Explicit provision is made for including uncertainty associated with the quantity
that is extended to the regional population.
In the ELS and, hence, the DDRP, the size of the target population is not precisely known. The
sampling frame for the ELS consisted of designated lakes on USGS maps. In some cases during field
sampling in the ELS, a field visit to the sample lakes selected from this frame indicated that some water
bodies designated as lakes on the map actually were not lakes, but rather marshes or old beaver ponds,
for example. When these "non-lakes" were subsequently excluded from the sample, a similar proportion
6-1
-------
of lakes also had to be excluded from the target population, effectively reducing its size. Thus, the size
of the target population is estimated from the sample size. This presents no particular difficulty as long
as each unit in the sample has a known inclusion probability.
The design of the surface water surveys and the DDRP also permits arbitrary subsetting of the
sample. In some cases, the subsetting would correspond to a redefinition of the target population (e.g.,
the exclusion of seepage lakes). In such cases, the Inclusion probabilities for the remaining sample units
do not change, which, as can be seen from Equation 6-1 below, implies a smaller target population.
In other cases, the subset should be viewed as a subsample. In these cases, a smaller sample is being
used to make an inference about the same target population, and the inclusion probabilities do change.
This might occur if a selected lake could not be sampled or simulated for some reason. Inferences can
still be made about the same target populations, but the inclusion probabilities would change.
6.2.2 Estimation Procedures for Population Means
The structure of the DORP sample is almost identical to the structure of the ELS Phase II sample.
The differences are primarily in the conditional probability of inclusion in the second phase of the sample:
the DDRP sample was reduced by exclusion of lakes with large watersheds and the Phase II sample was
reduced at random. The estimation procedures are parallel to those detailed in the ELS Phase II Data
Analysis Plan (Overton.1987). Let n be the size of the sample selected from the target population, let
p, be the probability that sample unit I was included in the sample, and tet p,. be the joint inclusion
probability of units i and j. For sample unit i, let y; be the "true" quantity, and let z, be the observed
quantity, i.e., the unknown true value with an associated error 6j. The error may be an observation error
or a measurement error; it could also be a prediction error. In each case we assume that the
characteristics of the error distribution are known, and that the uncertainty in the observed values is
characterized by that error distribution. The basic estimation procedures will follow the Horvitz-Thompson
estimator (Cochran, 1977) for variable probability samples; some details, however, will depend on
assumptions made about the observation error. Several distinct error models are treated below.
6-2
-------
In one case, the uncertainty is due to an additive error term, so that the magnitude of the
uncertainty is constant over the range of the response. The observation is related to the true value
through the equation Z| = y, + e,. Two distributions were available to handle this case: the error term
was assumed to have either a normal distribution with mean 0 and variance a2 or a uniform distribution
over the interval (-a,a). For this uniform distribution, the mean is 0 and oz = a2/3.
In a second case, the magnitude of the uncertainty depends on the magnitude of the response.
This can be modelled with a multiplicative error term, where the uncertainty is proportional to the
response, so that z, = y { e;. We assumed that the uncertainty followed a log-normal distribution with
a mean value of 1 and a variance a2 = RSD2 where RSD was the relative standard deviation.
An implication of the above multiplicative model is that the uncertainty goes to 0 along with the
response. In some instances, however, there was appreciable uncertainty even when the response was
0. For these cases, we assumed that the uncertainty was proportional to the sum of the response plus
an offset (h), so that the observation equation was Zj = ys + (y( + h)e j = V| (e, + 1) + h6j. The mean
value of the error term was 0, and the a2 = RSD2. As above, a log-normal distribution was used for
this case.
The error structure affects only the variance of the population total, the variance of the population
mean, and the estimator of the cumulative distribution function and its associated variance. The estimator
of the target population size and population total take the same form under all of the above error
structures.
yv
Estimator of population total, T :
T = \ z,/p, (Equation 6-1)
6-3
-------
Estimator of the size of the target population, N.
N =ns 1/p. (Equation 6-2)
Estimator of population average, Y:
? = T/N. (Equation 6-3)
Both t and N are random variables, and both are unbiased estimators of the respective population
quantities. However, 7, similar to most ratio estimators, is a slightly biased estimator of the population
average.
6.2.3 Estimators of Variance
A
For all three error models, the estimator of the variance of T has the form
—
Varfn = £ <1 - Pjjzi + Ss'LJEilfiEilS) + g(e,z) (Equation 6-4)
where g(e,z) is a function that depends on the error model and the sample data. For the additive model,
g(e,z) = a2 N ; for the multiplicative model, g(e,z) = a2 ZZj2/pj, and for the multiplicative model with
offset, g(e,z) = a2 S(z, + h)2/p,, where h is the offset.
The variance of N is estimated by
Var(N) = i: 0 - Pi) + SSn.(JPjLLEiEiL) + 9(e,z) (Equation 6-5)
6-4
-------
The joint inclusion probabilities py are determined by the structure of the DDRP sample. They are
computed according to the algorithm in the ELS Phase II Analysis Plan (Overton, 1987).
Finally, the variance of the estimator of the population average was obtained from a first-order variance
propagation using Equations 6-4 and 6-5:
Var(?) - Var(T)/N2 + T2Var(N)/N4 - TCovCT.NJ/N2, (Equation 6-6)
where
~" n n
Cov(T,N) = ZZ fop, - p,,)(l/pi - l/p,)(z,/p7 - zj/pf)
i i>j
Confidence intervals will be derived from the usual normal theory, e.g., a 95 percent Cl on the population
average is given by
Y ± 1.967 Var(\).
6.2.4 Estimator of Cumulative Distribution Function
Let N(y) be the total number in the population with the value of Y less than or equal to y, so that
the cumulative distribution function of Y is F Y (y) = N(y)/N. An estimator of N(y) is
= Sv5(y)/ps,
where
10, z, > y
6-5
-------
An estimator of the cumulative distribution function of Y is
FY(y) = N(y)/N
/s
The variance of FY has both a sampling component and a component due to measurement uncertainty.
The variance of the N(y) and co-variance of N(y) and N are needed to calculate the sampling variance
of FY. These are given by
Var(N(y)) = FY(y)(1-FY(y))S 1/pf + F^MVarfN)
and
Cov(N,N(y)) = FY(y)Var(N).
Then a first order variance propagation formula gives
Var(N(y))/N2 + N2(y)Var(N)/N4 - N(y)Cov(N(y),N)/N2
for the sampling variance. A Monte Carlo procedure was used to calculate the measurement variance.
The sampling variance and the measurement variance were added to obtain total variance.
^
The median and quintiles of the distribution of Y were estimated by the linear interpolation of FY.
6.3 UNCERTAINTY ESTIMATES
The quantities displayed in this report are the end result of a sequence of operations, beginning
with collection of a physical sample in the field and ending with the production of a table or graph. A
variety of steps were conducted, including chemical analyses, data aggregation, data reduction, and
processing of the data through various mathematical models. The final result contains an element of
uncertainty that has its origin in the design, in the implementation of the field protocol, and in the
precision of the basic measurement process (e.g., the chemical analytic precision). The uncertainty on
the final result can be quantified by propagating the uncertainty (or its mathematical analog) through
the same sequence of operations as were the data.
6-6
-------
In the DDRP, several techniques have been used to propagate uncertainty through a functional
relationship (which could be a complex simulation model as well as an explicit function). Let f(x1 ,x2
xfl) be a function of the variables x1,x2l... .x,, with uncertainties e, e2 en, respectively. The
probability distributions (or at the least the variances) of the uncertainties are presumed known. If the
functional relationship is such that partial derivatives can be easily obtained, then the variance of
functional values can be estimated using a first-order linear approximation to the functional relationship:
Var(f)
In the case of a simulation model, the function is the model itself, and the partial derivatives cannot
be calculated explicitly. An approximation to the partials can be obtained by perturbing the Xg's in turn.
If a suitably small perturbation is chosen, then the ratio of the change in output to the perturbation is an
estimate of the partial derivative. These estimates can then be used in a first-order propagation as above.
A disadvantage of both of the above techniques is that they Ignore possible correlations among
the uncertainties. One way to account for such correlations is to propagate not only variances but also
co-variance terms. The "first-order, second-moment" technique used in the Enhanced Trickle Down
uncertainty analysis is a means of doing exactly that. A first-order approximation is made to the model,
and Katman filtering techniques are used to build up an estimate of the state variable variance-covariance
matrix. A final method that was used in uncertainty assessment was Monte Carlo. The Monte Carlo
method is applied by repeatedly calculating the value of f, each time perturbing the value of each x, by
a random quantity drawn from the respective uncertainty distribution. Monte Carlo is most easily applied
when uncertainties are statistically independent, but can also be applied when correlations exist. A variant
of Monte Carlo, called "fuzzy optimization", was used in the uncertainty analyses for the Model of
Acidification of Groundwater in Catchments.
6-7
-------
6.4 APPLICABILITY
This section discusses the procedures for the Level I, II, and III population estimation approaches
for DORP, including the statistical formulas that will be used to estimate population means, variances, and
cumulative frequency distributions. The population estimation procedures are generic and do not depend
on the level of analysis. The specific target populations for inference, however, do depend on the
analyses performed. Not all DDRP watersheds were used at each level of analysis so the target
population will vary. The explicit target populations being considered in the analysis are discussed in
Sections 8, 9, and 10. The generic uncertainty estimation procedures introduced in this section also
are more explicitly discussed for each of the individual analyses in Sections 8, 9, and 10.
6-8
-------
SECTION 5
DATA SOURCES AND DESCRIPTIONS
5.1 INTRODUCTION
The purpose of Section 5 is to present sufficient information concerning the design of the Project
and data acquisition within the DDRP to familiarize the reader with the characteristics of the regions
studied and to allow the reader to evaluate the analyses performed in Sections 7-10. Many data have
been generated and used by the Project during its course. Although a complete listing of the data is
not presented here, descriptions of the way the data were gathered within the Project or obtained from
other sources are presented along with pertinent examples or summaries of the data. As indicated in
Section 2.6, a complete listing of the DDRP databases will be presented late in 1989.
5.2 STUDY SITE SELECTION
In selecting study sites, the intent was to focus on regions with watersheds potentially sensitive to
acidic deposition (Section 2.3), but exhibiting a wide contrast in both soil and watershed characteristics
and levels of deposition.
5.2.1 SiteT.Selectiorii Procedures
The procedures for selecting the DDRP sample watersheds differed somewhat between the NE and
the SBRP, primarily because of the differences in the Eastern Lake Survey (ELS) and Pilot Stream Survey
sampling designs. Some background on the design of these two surveys is provided here because of
their influence on the DDRP design. Complete details are provided by Linthurst et al. (1986a) and Messer
et ai. (I986a).
5.2.2 Eastern Lake Survey Phase I Design
The ELS Phase I, conducted in the fall of 1984, sampled over 1,600 lakes in the eastern United
States, including over 760 lakes in the NE. The sampling approach of the ELS was to use a stratified
design with about 50 lakes per stratum. For purposes of the survey, the Northeast Region was divided
5-1
-------
into subregions based on physiographic features. Each subregion, in turn, was divided into three mapped
strata based on the surface water ANC expected to dominate different areas (Figure 5-1). The expected
values for ANC in each stratum were based on a national map of surface water ANC that indicated areas
with low ANC and, therefore, areas potentially susceptible to acidic deposition (Omernik and Powers,
1983). Stratum 1 had projected ANC < 100 /ieq L"1, stratum 2 had projected ANC of 100-200 Meq L*
1, and stratum 3 had projected ANC > 200 /ieq L'1 . A probability sample of about 50 lakes in each
stratum was selected from a list of all lakes identifiable on USGS 1:250,000-scale maps using a systematic
sample with a random start. Some of the sample fakes were subsequently classified as non-target and
eliminated from the sample. The ELS strata included lake populations of differing sizes and, therefore,
the inclusion probability for any given lake in the target population varied among strata (Table 5-1).
5.2.3 Pilot Stream Survey Design
The National Stream Survey (NSS) began with a Pilot Survey in the SBRP, the purpose of which
was to establish the methodology for conducting a broad regional survey of streams. The Pilot Stream
Survey framed the target population by defining a stream reach as the length of "blue-line" stream on
USGS 1:250,000-scale maps that lies between the downstream and upstream confluences with other blue-
line streams, or the upper stream boundary if no upper confluence is present (Figure 5.2.3-1).
A two-stage sampling approach was used for selecting NSS streams. In the first stage, a point
frame was used to select the sample of stream reaches. A rectangular grid of points, separated by a
scaled distance of approximately 13 km (8 mi), was positioned at random over a 1:250,000 topographic
map. The first stream reach intersected by a line from each point drawn downslope perpendicular to the
contour lines was included in the first stage sample. If any portion of the reach extended outside the
study region, if the reach drained into a reservoir, or if a watershed was too large ( > 155 km2 (60 ml2)),
the reach was designated non-interest and dropped from the sample. The inclusion probability for a reach
in the target population was proportional to the watershed area that drains directly into the reach
compared to the total area of the grid square (i.e., -164 km2 ) (Figure 5-2). This is the area in which
a grid point had to fall for the reach to be selected.
5-2
-------
ANC (w LH)
B < 100
• 100 - 200
ffl > 200
Adirondack* (1A)
Maine <1E)
Central
New England (1C)
Southern
New England (ID)
Figure 5-1. Northeastern subregions and ANC map classes, Eastern Lake Survey Phase I (Unthurst
et al., I986a).
5-3
-------
Table 5-1. Sampling Structure for Phase I, Region 1 (Northeast), Eastern
Lake Survey
Stratum
w
N
SE
A1 711
A2 542
A3 431
B1 208
B2 96
B31682
C1 631
C2 752
C3 650
D1 443
D2 656
D31568
E1 1038
E2 606
E3 744
N* =
n =
p
W
Nf
SE »
57 0.1038
51 0.1199
47 0.1488
49 0.3133
48 0.6770
47 0.0368
63 0.1278
54 0.0931
47 0.1117
47 0.1522
43 0.1448
37 0.0515
89 0.1239
48 0.1198
41 0.0968
No. of takes identified on the
No. of lakes sampled
Inclusion probability for each
Weight or no. of lakes in the
that lake. Defined as 1/p.
Estimated no. of lakes in the
9.633
8.338
6.719
3.192
1.477
27.209
7.822
10.743
8.953
6.572
6.905
19.426
8.070
8.344
10.333
maps
lake in stratum
target population '
stratum (n*w)
549.08
425.24
315.79
156.41
70.90
1278.82
492.79
580.12
420.79
308.88
296.92
718.76
718.23
400.51
423.65
* represented by
33.08
26.13
22.13
9.29
3.00
90.37
27.31
36.20
34.59
23.00
31.14
85.22
39.71
31.80
41.55
Standard error of the estimate
5-4
-------
Non-Head wot er Reach
Headwater Reach
\
Figure 5-2. Representation of the point frame sampling procedure for selecting NSS Stage I
reaches. Area a, represents the direct drainage area to the lower node of non-headwater reaches,
or the total drainage area to the lower node of headwater reaches. Area az is the total drainage
area to the upper node of non-headwater reaches.
5-5
-------
The first-stage sample was used to establish physical characteristics of the stream reach population
(e.g., distribution of reach lengths and drainage areas). A second-stage sample for chemical sampling
was chosen by selecting reaches corresponding to every other grid point.
5.2.4 DDRP Target Population
The DDRP data were obtained from 145 lake watersheds in the NE (a subsample from the ELS
Phase I) and 35 stream watersheds in the SBRP (a subset of the streams surveyed in the NSS Pilot
Survey).
5.2.4.1 Northeast Lake Selection
At the time the DDRP subsample was selected, lakes for'the detailed sampling phase (Phase II)
of the ELS also were being chosen. Preliminary data from the ELS Phase I were used to identify lakes
of low interest, such as high ANC lakes (ANC > 400 peq L*1 ), shallow lakes (<1.5 m deep), or
anthropogenically disturbed lakes. These lakes were excluded from consideration as DDRP or Phase II
lakes. Very large lakes (surface area > 2000 ha) were placed in a reserved category and also excluded
from sampling for the present ELS Phase II or DDRP studies. Logistical considerations for both the
DDRP and the Phase II sample limited the number of lakes/watersheds that could be adequately
characterized to a total of about 150 lakes. Statistical precision requirements indicated that a sample size
of about 50 lakes was required for any subset for which estimates were desired. In order to satisfy these
constraints, the remainder of the ELS Phase I sample was split into three groups using cluster analysis
on the Phase I chemical data. After examination of the clusters using variables that described the
chemical, physical, and pollution status of the lake, the lakes were split into groups based only on ANC.
The final division defined the three groups as (1) ANC < 25 peq L'1; (2) 25-100 peq L'1; and (3) 100
- 400 peq L"1.
Although the DDRP and Phase II both required a sample size of about 50 lakes per cluster, the
DDRP had an additional constraint: watersheds with an area greater than 3000 ha could not be
adequately mapped during the DDRP soil survey phase (Section 5.4). To accommodate this constraint,
5-6
-------
60 lakes were selected from each of the three ANC groups. The lakes were selected from the clusters
using a fixed size, variable probability systematic sampling scheme that resulted in approximately equal
inclusion probabilities within groups. The selection probabilities were inversely proportional to Phase I
inclusion probabilities, so that the total inclusion probabilities were nearly uniform within groups. (Some
Phase I sample lakes had inclusion probabilities sufficiently small, relative to group size and sample size,
that they were entered with probability one. These lakes disrupted the within-group uniformity.)
Additionally, the Phase II selection was made before final disposition of the ELS Phase I sample lakes,
and some lakes were subsequently reclassified. Because this step changed the Phase I inclusion
probabilities, it also changed the total DORP inclusion probabilities. The conditional selection probabilities
were fixed by the list at the time of selection and did not change. After the sample was selected, lakes
with large watersheds were dropped from the DDRP sample, which resulted in a redefinition of the DDRP
target population. The 60-lake sample was randomly reduced to 50 for the Phase II sample. Thus,
although there is considerable overlap of DDRP and Phase II (ca. 85 percent), there are lakes in ELS
Phase II whose watersheds were not studied by DDRP, and vice versa.
Several lakes also were eliminated from the sample because access was denied to the watershed
for mapping or soil sampling. This process was treated as a random deletion, which decreased the
sample size but left the target population unchanged. This step resulted in a total sample size of 145
lakes. Subsequent to this sample determination the NSWS recalculated lake ANC values because of
some slight errors in the original fitting of the Gran's titration data (J. Eilers, personal communication).
The resultant recalculation generally decreased the computed lake ANC values resulting in a shift En
sample size for the ANC groups. Again, this affected the sample size but not the target population. The
final structure for the DDRP sample is given in Table 5-2; lake ID's, inclusion probabilities, and weights
are given in Table 5-3. Further identification of the NE DDRP lake/watersheds is given in Tables 5-4 and
5-5, Figures 5-3 through 5-7, and Plate 5-1.
5-7
-------
Table 5-2. Sample Structure for the Direct/Delayed Response Project
Northeastern Sample
ANC Group3 n N
1 55 796
2 46 1100
3 44 1772
Subtotal 145 3368
Reserved 262
Total 768
n = No. of DORP lakes sampled
from the ANC group.
N - Estimated no. of lakes in
the DORP target population.
' Group 1 = ANC < 25 peq L'1.
Group 2 = 25-100 peq L
Group 3 » ANC > 100 peq L'1
based on recalculated ANC values (see text for explanation)
from the ELS - Phase I (Unthurst et al., 1986a)
b Reserved lakes in the ELS -1 population that were of
low interest (e.g. ANC > 400 peq L'1) and were
placed in a reserved category
5-8
-------
Table 5-3. ANC Group, Lake Identification, ELS-I
Phase I ANC, Weight and Inclusion Probabilities for the
Direct/Delayed Response Project Northeast Sample Watersheds
ANC
Group
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Phase I
Lake ID (
1A1-003
1A1-012
1A1-017
1A1-020
1A1-028
1A1-039
1A1-049
1A1-057
1A1-061
1A1-066
1A1-073
1A2-002
1A2-004
1A2-041
1A2-042
1A2-045
1A2-046
1A2-048
1A2-052
1A2-054
1A3-028
1A3-046
1A3-048
1A3-065
181-010
1B1-043
1B2-028
1B3-052
1B3-056
1B3-059
1C1-068
1C2-037
1C2-041
1 02-048
1C2-054
1C2-057
1C3-055
1D1-031
1D1-034
101-037
101-046
101-056
101-067
101-068
102-027
ANC
/ieq I/1)8
-21.7
11.4
-7.4
6.0
1.8
-1.7
-30.3
-18.0
-53.0
1.8
-28.1
1.8
-32.0
22.4
6.2
7.8
12.9
-5.3
1.1
-14.7
-4.3
18.2
7.3
0.5
-23.9
12.1
14.6
16.4
-6.0
-4.4
-43.1
5.5
2.2
11.5
-6.9
19.6
-35.2
2.9
9.8
5.3
13.1
3.5
3.6
-16.9
-6.0
Weight
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
22.4929
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
22.4929
12.2850
12.2850
12.2850
12.2850
12.5230
27.2090
27.2090
27.2090
12.2850
12.2850
12.2850
22.4929
12.2850
22.4929
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.2850
12.0620
Inclusion
Probability
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.04445860
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.04445860
0.08140010
0.08140010
0.08140010
0.08140010
0.07985310
0.03675250
0.03675250
0.03675250
0.08140010
0.08140010
0.08140010
0.04445860
0.08140010
0.04445860
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08140010
0.08290500
continued
5-9
-------
Table 5*3. (Continued)
ANC
Group
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
Lake ID
1D2-036
1D2-094
1D3-002
1D3-029
1E1-009
1E1-011
1E1-106
1E1-111
1E2-038
1E2-049
1A1-014
1A1-038
1A1-046
1A1-064
1A2-006
1A3-001
1A3-040
1A3-042
1B1-023
181-055
1B3-025
1B3-041
1C1-031
1C1-050
1C1-084
1C1-086
1C2-002
1C2-012
1C2-028
1C2-033
1C2-Q35
1C2-050
1C2-062
1C2-064
1C2-066
1C3-030
1D1-027
101-054
102-025
1D2-074
1D3-044
1E1-025
1E1-040
1E1-050
1E1-054
Phase I ANC
( Meq L'1)a
0.1
-5.6
1.6
-15.2
11.1
19.5
22.7
6.3
9.4
-3.7
30.0
97.2
56.3
82.9
33.5
76.9
69.2
30.0
33.3
52.9
30.4
89.7
62.9
63.6
41.7
25.7
69.2
71.5
51.7
97.3
64.7
45-0
36.4
86.2
67.7
86.8
67.1
63.4
71.8
80.3
41.5
98.0
36.8
43.8
33.4
Weight
12.0620
12.0620
19.4260
19.4260
12.4160
22.7327
22.7327
12.4160
12.0540
12.0540
22.4929
40.4692
22.4929
22.4929
22.4929
22.4929
22.4929
22.4929
22.4929
22.4929
27.7643
27.7643
22.4929
22.4929
22.4929
22.4929
22:4929
22.4929
22.4929
40.4692
22.4929
22.4929
22.4929
22.4929
22.4929
22.4929
22.4929
22.4929
22.0837
22.0837
22.0031
40.9000
22.7327
22.7327
22.7327
Inclusion
Probability
0.08290500
0.08290500
0.05147740
0.05147740
0.08054120
0.04398960
0.04398960
0.08054120
0.08296000
0.08296000
0.04445860
0.02471010
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.03601750
0.03601750
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.02471010
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.04445860
0.04528230
0.04528230
0.04544820
0.02444990
0.04398960
0.04398960
0.04398960
continued
5-10
-------
Table 5-3. (Continued)
ANC
Group
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Phase 1 ANC
Lake ID ( /ieq L'1)a Weight
1E1-061
1E1-062
1E1-073
1E1-074
1E1-077
1E1-082
1E1-092
1EM23
1E2-007
1E2-056
1E2-063
1A1-029
1A1-033
1A2-037 '
1A2-039
1A2-058
1A3-043
1B1-029
163-004
1B3-012
183-019
1 83-021
183-032
183-043
183-051
1B3-053
183-060
183-062
1C1-009
1C1-017
1C1-018
1C1-021
1C2-016
1C2-056
1C2-068
1C3-031
1C3-063
1D2-049
1D2-084
102-093
1 03-003
103-020
1D3-025
1D3-033
1E2-002
1E2-030
1E2-054
66.0
86.2
52.3
70.2
81.0
89.0
77.5
74.0
75.4
58.8
25.6
111.9
183.2
161.2
140.9
391.6
238.4
166.0
342.7
342.2
218.5
380.8
332.5
143.0
275.3
245.8
190.9
376.4
105.7
325.9
173.3
122.5
128.8
213.0
285.5
122.4
325.9
107.8
142.5
221.1
154.4
104.0
162.1
368.5
256.7
174.3
228.9
22.7327
22.7327
22.7327
22.7327
22.7327
22.7327
22.7327
22.7327
22.0694
22.0694
22.0694
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
39.7325
39.7325
39.7325
39.5892
39.5892
39.5892
39.5892
39.7075
39.7075
39.7075
Inclusion
Probability
0.04398960
0.04398960
0.04398960
0.04398960
0.04398960
0.04398960
0.04398960
0.04398960
0.04531160
0.04531160
0.04531160
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02516830
0.02516830
0.02516830
0.02525940
0.02525940
0.02525940
0.02525940
0.02518420
0.02518420
0.02518420
continued
5-11
-------
Table 5*3. (Continued)
ANC
Group
3
3
3
3
3
3
3
3
Lake ID
1E2-069
1E3-022
1E3-040
1E3-041
1E3-042
1E3-045
1E3-055
1E3-062
Phase I ANC
{ Meq L'V
238.1
222.1
229.1
349.4
162.5
141.7
299.5
153.7
Weight
39.7075
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
40.4692
Inclusion
Probability
0.02518420
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
0.02471010
Recalculated values (see text for explanation)
5-12
-------
Table 5-4. Lake Identification (ID) and Name, and State
and Latitudinal/Longitudinal Location of the Northeast Sample
Watersheds, Sorted by Lake ID
Lake ID
1A1-003
1A1-012
1A1-014
1A1-017
1A1-020
1A1-028
1A1-029
1A1-033
1A1-038
1A1-039
1A1-046
1A1-049
1A1-057
1A1-061
1A1-064
1A1-066
1A1-073
1A2-002
1A2-004
1A2-006
1A2-037
1A2-039
1A2-041
1A2-042
1A2-045
1A2-046
1A2-048
1A2-052
1A2-054
1A2-058
1A3-001
1A3-028
1A3-040
1A3-042
1A3-043
1A3-046
1A3-Q48
1A3-065
1B1-010
1B1-023
1B1-029
1B1-043
1B1-055
1B2-028
1B3-004
1B3-012
1B3-019
1B3-021
Lake Name State
Hawk Pond
Whitney Lake
Wilmurt Lake
Constable Pond
Fourth Lake (Bisby Lakes)
Dry Channel Pond
Middle Pond
Kiwassa Lake
Nicks Pond
John Pond
Partlow Lake
Middle South Pond
Hitchcock Lake
Wolf Lake
Mt. Arab Lake
Woodhull Lake
Gull Lakes (South)
St. John Lake
Duck Lake
Lake Frances
Fish Ponds (Northeast)
Oxbow Lake
Mud Lake
North Branch Lake
Woods Lake
Nine Corner Lake
No Name
Chub Lake
Trout Lake
Trout Lake
Nate Pond
Curtis Lake
Zack Pond
Cheney Pond
Unknown Pond
Long Pond
Grass Pond
South Lake (East Branch)
Ganoga Lake
Twin Lakes (Brink P)
No Name (Wilson Creek Dam)
Penn Lake
Rock Hill Pond
Mill Creek Reservoir
Guilford Lake
Little Butler Lake
Hartley Pond
Cord Pond
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
PA
PA
PA
PA
PA
PA
NY
PA
PA
PA
Latitude
0 i .
43
43
43
43
43
44
44
44
44
44
44
43
43
43
44
43
43
43
43
44
43
43
43
43
43
43
43
43
43
44
43
43
43
43
43
43
43
43
41
41
41
41
41
41
42
41
41
41
57
35
25
50
34
21
20
17
8
6
0
59
51
37
11
35
51
26
14
41
32
26
20
18
15
11
7
15
20
21
51
20
56
52
49
38
41
30
21
23
17
6
18
15
24
51
39
39
25
15
45
0
15
10
20
45
35
45
15
22
0
45
18
30
22
30
8
45
50
30
26
45
10
45
39
30
48
47
30
10
0
40
10
15
35
38
30
0
30
45
49
45
45
45
30
10
Longitude
O 1 U
74
74
74
74
74
74
74
74
74
74
74
75
75
74
74
74
74
74
74
74
74
74
74
74
74
74
74
74
74
75
74
74
74
74
74
74
75
74
76
74
75
75
75
75
75
75
75
75
57
33
43
47
58
26
22
9
58
45
50
1
2
39
36
59
49
3
27
19
3
29
27
47
19
33
35
31
42
16
5
57
11
9
17
17
3
53
19
54
14
46
0
45
30
37
42
51
30
45
30
45
15
15
45
30
5
50
0
6
30
15
3
13
15
40
9
30
40
0
14
40
0
0
20
50
50
8
30
40
0
45
0
20
40
32
15
15
20
10
58
0
0
40
30
0
continued
5-13
-------
Table 5-4. (Continued)
Lake ID
1B3-025
1B3-032
1B3-041
1B3-043
1B3-051
1B3-052
1B3-053
1B3-056
1B3-059
1B3-060
183-062
1C1-009
1C1-017
1C1-Q18
1C1-021
1C1-031
1C1-050
1C1-068
1C1-084
1C1-086
1C2-002
1C2-012
1C2-016
1C2-028
1C2-033
1C2-035
1C2-037
1C2-041
1C2-048
1C2-050
1C2-054
1C2-056
1C2-057
1C2-062
1C2-064
1C2-066
1C2-068
1C3-030
1C3-031
1C3-055
1C3-063
1D1-027
1D1-031
1D1-033
1D1-037
1D1-046
1D1-054
1D1-056
1D1-067
Lake Name
Trout Lake
Wixon Pond
East Stroudsburg Reservoir
Trout Lake
Barrett Pond
No Name
No Name (Snowflake Lake)
Riga Lake
Island Pond
Sly Lake
Bassett Pond
Upper Baker Pond
Welhern Pond
Decker Ponds (Eastern)
Clear Pond
Hunt Pond
Billings Pond
Lincoln Pond
Upper Beech Pond
Star Lake
Iron Pond
Black Pond
Trafton Pond
Sunset Lake
Long Pond
Smith Pond
Mendums Pond
Juggernaut Pond
Cranberry Pond
Moores Pond
Lake Wamponoag
Drury Pond
Babbidge Reservoir
Pemigewasset Lake
Hancock Pond
Turtle Pond
Quimby Pond
Pelham Lake
Sadawaga Lake
Darrah Pond
Martin Meadow Pond
School House Pond
Kings Pond
Rocky Pond
Ezekiel Pond
Robbins Pond
Upper Millpond
Little West Pond
Round Pond
State
NY
NY
PA
PA
NY
NY
PA
CT
NY
PA
PA
NH
ME
ME
ME
ME
NH
MA
NH
NH
ME
ME
ME
NH
NH
NH
NH
NH
NY
MA
MA
ME
NH
NH
ME
NH
ME
MA
VT
NH
NH
Rl
MA
MA
MA
MA
MA
MA
Rl
Latitude
0 > »
41
41
41
41
41
41
41
42
41
41
41
43
45
45
45
44
43
42
43
43
45
44
43
43
43
43
43
42
42
42
42
44
42
43
44
43
44
42
42
42
44
41
41
41
41
41
41
41
41
35
23
4
0
26
29
54
1
15
49
35
54
12
11
6
5
17
40
38
27
27
8
50
28
12
9
10
57
44
39
37
42
56
36
57
15
59
42
47
49
26
24
54
53
48
42
43
55
58
10
45
0
15
4
23
18
18
26
25
33
30
45
45
30
0
0
10
54
43
30
45
45
15
14
15
30
35
40
20
2
15
5
55
20
15
27
0
0
52
30
0
40
10
15
20
51
17
17
Longitude
O i u
74
73
75
75
73
74
75
73
74
75
75
71
70
69
69
71
71
71
71
72
70
70
70
71
71
72
71
72
73
72
71
70
72
71
69
71
70
72
72
71
71
71
70
70
70
70
70
70
71
40
44
10
20
44
32
24
29
8
20
42
59
29
56
59
0
56
54
12
3
22
48
53
18
48
1
4
0
26
20
57
14
13
35
59
31
44
53
52
26
36
40
42
41
36
6
7
42
46
50
5
0
30
25
20
37
0
25
14
40
30
40
15
15
0
30
45
15
20
30
0
30
0
43
45
0
45
0
50
45
30
0
45
10
0
31
30
30
40
30
0
15
45
45
40
0
24
20
continued
5-14
-------
Table 5-4. (Continued)
Lake ID
101-068
1D2-025
1D2-027
1D2-036
1D2-049
1D2-074
1D2-084
1D2-093
1D2-094
1D3-002
1D3-003
1D3-020
1D3-025
1D3-029
1D3-033
1D3-044
1E1-009
1E1-011
1E1-021
1E1-040
1E1-050
1E1-054
1E1-061
1E1-062
1E1-073
1E1-074
1E1-077
1E1-082
1E1-092
1EM06
1E1-111
1E1-123
1E2-002
1E2-007
1E2-030
1E2-038
1E2-049
1E2-054
1E2-056
1E2-063
1E2-069
1E3-022
1E3-040
1E3-041
1E3-042
1E3-045
1E3-055
1E3-062
Lake Name
Little Sandy Pond
Little Quittacas Pond
Sandy Pond
Micah Pond
Spring Grove Pond
Stetson Pond
Goose Pond
Ashland Reservoir
Snows Pond
Dykes Pond
Sandy Pond
Little Alum Pond
Long Pond
Killingly Pond
No Name
Middle Farms Pond
Peep Lake
Fourth Davis Pond
Bean Ponds (Middle)
Lt. Greenwood Pond (West)
Lower Oxbrook Lake
Duck Lake
Little Seavey Lake
Long Pond
Georges Pond
Craig Pond
Parker Pond
Stevens Pond
Great Pond
Greenwood Pond
Long Pond
First Pond
No Name
Fairbanks Pond
Round Lake
Nelson Pond
Gross Pond
Brettuns Pond
Peabody Pond
Kalers Pond
No Name
Number Nine Lake
Nokomis Pond
Round Pond
Sand Pond
McClure Pond
Togue Pond
Cain Pond
State
MA
MA
MA
MA
Rl
MA
MA
MA
MA
MA
MA
MA
CT
CT
CT
NY
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
Latitude
O i «
41
41
41
41
41
42
41
42
41
42
42
42
42
41
41
41
44
45
45
45
45
45
44
44
44
44
44
44
44
45
44
44
45
44
45
44
44
44
43
44
46
46
44
44
44
44
46
44
47
47
46
38
54
1
41
14
45
36
33
7
1
51
39
16
54
15
48
22
17
9
56
55
37
35
22
22
36
32
32
22
59
23
1
24
3
23
56
6
7
25
52
44
34
29
56
29
47
30
20
20
35
40
38
22
30
15
45
45
15
45
30
30
30
30
45
0
0
0
15
0
0
0
20
0
3
7
2
10
40
21
0
55
30
30
32
29
27
0
15
20
10
0
2
32
Longitude
O i ii
70
70
70
70
71
70
70
71
70
70
71
72
71
71
73
71
67
69
69
69
67
68
67
68
68
68
68
69
68
69
68
68
69
69
67
70
69
70
70
69
68
68
69
69
70
68
68
68
36
55
39
22
39
49
0
27
51
43
33
9
49
47
11
58
53
23
11
24
50
6
38
16
14
40
42
18
17
13
10
36
47
49
16
15
23
15
41
25
46
3
18
13
7
57
53
58
13
0
15
45
0
39
28
52
10
46
15
15
0
45
30
40
30
40
30
30
30
0
0
11
30
0
30
0
0
58
13
0
0
52
0
45
35
0
13
22
45
0
0
30
10
50
31
3
continued
5-15
-------
Table 5-5. Lake Identification (ID) and Name, Sorted by State
- Northeast Sample Watersheds
State
CT
CT
CT
CT
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
Lake ID
1B3-056
1D3-025
1D3-029
1D3-033
1C1-068
1C2-050
1C2-054
1C3-030
1D1-031
1D1-034
1D1-037
1D1-046
1D1-054
1D1-056
1D1-068
1D2-025
1D2-027
1D2-036
1D2-074
1D2-084
1D2-093
102-094
103-002
1D3-003
1D3-020
1C1-017
1C1-018
1C1-021
1C1-031
1C2-002
1C2-012
1C2-016
1C2-056
1C2-064
1C2-068
1E1-009
1E1-011
1E1-025
1 El -040
1E1-050
1E1-054
1E1-061
1E1-062
1E1-073
1E1-074
1E1-077
1E1-082
1E1-092
Lake Name
Riga Lake
Long Pond
Killingly Pond
No Name
Lincoln Pond
Moores Pond
Lake Wamponoag
Pelham Lake
Kings Pond
Rocky Pond
Ezekiel Pond
Robbins Pond
Upper Millpond
Little West Pond
Little Sandy Pond
Little Quittacas Pond
Sandy Pond
Micah Pond
Stetson Pond
Goose Pond
Ashland Reservoir
Snows Pond
Dykes Pond
Sandy Pond
Little Alum Pond
Welhern Pond
Decker Ponds (Eastern)
Clear Pond
Hunt Pond
Iron Pond
Black Pond
Trafton Pond
Dairy Pond
Hancock Pond
Quimby Pond
Peep Lake
Fourth Davis Pond
Bean Ponds (Middle)
Lt. Greenwood Pond (West)
Lower Oxbrook Lake
Duck Lake
Little Seavey Lake
Long Pond
Georges Pond
Craig Pond
Parker Pond
Stevens Pond
Great Pond
continued
5-16
-------
Table 5-5.
State
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
(Continued)
Lake ID
1E1-106
1E1-111
1E1-123
1E2-002
1E2-007
1E2-030
1E2-038
1E2-049
1E2-054
1E2-056
1E2-063
1E2-069
1E3-022
1E3440
1E3-041
1E3-042
1E3-045
1E3-055
1E3-062
1C1-009
1C1-050
1C1-084
1C1-086
1C2-028
1C2-033
1C2-035
1C2-037
1C2-041
1C2-057
1C2-062
1C2-066
1C3-055
1C3-063
1A1-003
1A1-012
1A1-014
1A1-017
1A1-020
1A1-028
1A1-029
1A1-033
1A1-038
1A1-039
1A1-046
1A1-049
1A1-057
1A1-061
1A1-064
1A1-066
Lake Name
Greenwood Pond
Long Pond
First Pond
No Name
Fairbanks Pond
Round Lake
Nelson Pond
Gross Pond
Brettuns Pond
Peabody Pond
Kalers Pond
No Name
Number Nine Lake
Nokomls Pond
Round Pond
Sand Pond
McClure Pond
Togue Pond
Cain Pond
Upper Baker Pond
Billings Pond
Upper Beech Pond
Star Lake
Sunset Lake
Long Pond
Smith Pond
Mendums Pond
Juggernaut Pond
Babbidge Reservoir
Pemigewasset Lake
Turtle Pond
Darrah Pond
Martin Meadow Pond
Hawk Pond
Whitney Lake
Wilmurt Lake
Constable Pond
Fourth Lake (Bisby Lakes)
Dry Channel Pond
Middle Pond
Kiwassa Lake
Nicks Pond
John Pond
Partlow Lake
Middle South Pond
Hitchcock Lake
Wolf Lake
Mt. Arab Lake
Woodhull Lake
continued
5-17
-------
Table 5-5. (Continued)
State
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
Rl
Rl
Rl
VT
Lake ID
1A1-073
1A2-002
1A2-004
1A2-006
1A2-037
1A2-039
1A2-041
1A2-042
1A2-045
1A2-046
1A2-048
1A2-052
1A2-054
1A2-058
1A3-001
1A3-028
1A3-040
1A3-042
1A3-043
1A3-046
1A3-048
1A3-065
1B3-004
1B3-025
1B3-032
1B3-051
1B3-052
1 63 -059
1C2-048
1D3-044
181-010
1B1-023
181-029
1 81 -043
181-055
1B2-028
1B3-012
183-019
183-021
1B3-041
1B3-043
183-053
1B3-060
183-062
1D1-027
101-067
102-049
1C3-031
Lake Name
Gull Lakes (South)
St. John Lake
Ouck Lake
Lake Frances
Fish Ponds (Northeast)
Oxbow Lake
Mud Lake
North Branch Lake
Woods Lake
Nine Corner Lake
No Name
Chub Lake
Trout Lake
Trout Lake
Nate Pond
Curtis Lake
Zack Pond
Cheney Pond
Unknown Pond
Long Pond
Grass Pond
South Lake (East Branch)
Guilford Lake
Trout Lake
Wixon Pond
Barrett Pond
No Name
Island Pond
Cranberry Pond
Middle Farms Pond
Ganoga Lake
Twin Lakes (Brink P)
No Name (Wilson Creek Dam)
Penn Lake
Rock Hill Pond
Mill Creek Reservoir
Little Butler Lake
Hartley Pond
Cord Pond
East Stroudsburg Reservoir
Trout Lake
No Name (Snowflake Lake)
Sty Lake
Bassett Pond
School House Pond
Round Pond
Spring Grove Pond
Sadawga Lake
5-18
-------
SUBREGIOH U
DORP SITE LOCATIONS
• DOIP Uk»i
+ Nti-ttRP US IS Ltkn
Figure 5-3. DDRP site locations for Subregion 1A.
5-19
-------
SUBREGION IB
DDRP SITE LOCATIONS
Subrtglon
Location
• DDIP litit
+ Nu-ftORP ISIS lain
Figure 5-4. DDRP site locations for Subregion 1B.
5-20
-------
SUBREGION 1C
DDRP SITE LOCATIONS
• DDRP Ukn
+ Nit-DARP NSIS liku
Figure 5-5. DDRP site locations for Subregion 1C.
5-21
-------
SUBREGION ID
DORP SITE LOCATIONS
Subrtglon
Location
• DORP Ltkti
+ H««-I»W NSIS loktt
IJ-W
Figure 5-6. DDRP site locations for Subregion 1D.
5-22
-------
SUBREGION IE
DDRP SITE LOCATIONS
• DORP Uktt
+ Noi-DDRP KSIS Uiit
Figure 5-7. DDRP site locations for Subregion 1E.
5-23
-------
Plate 5-1. ANC of DDRP lakes by ANC group.
5-24
-------
DDRP STUDY SITES
Lake ANC
ANC (ueq L'1)
• < 25
B 25 - 100
Q 100 - 400
-------
5.2.4.2 Southern Blue Ridge Province Stream Selection
Fifty-one stream reaches were sampled for water chemistry in the Pilot Stream Survey. Of these,
only 35 had watersheds less than 3000 ha (as defined based on the downstream sampling node), the
maximum size suitable for mapping within the DDRP. All of these 35 stream reaches were included in
the DDRP. As for the NE, eliminating streams with large watersheds has the effect of re-defining the
target population. The sampling structure stream ID's, inclusion probabilities, and weights for DDRP SBRP
streams are given in Table 5-6. Further information is provided in Tables 5-7 and 5-8 and in Figure 5-
8.
5.2.4.3 Final DDRP Target Populations
5.2.4.3.1 Northeast -
The final DDRP target population for the northeastern lakes represents 3,668 lakes, based on a
sample size of 145 lakes subsampled from the ELS Phase I target population. The target population
represents lakes with watershed areas greater than 4 ha and less than 3,000 ha and ANC concentrations
less than 400 jieq L*1 . The comparable ELS Phase I target population represented 7,157 lakes.
5.2.4.3.2 Southern Blue Ridge Province -
The final DDRP target population for the SBRP represents 1,531 streams based on a sample size
of 35 watersheds from the NSS Pilot Survey that satisfied the DDRP selection criteria. The SBRP target
population represents stream reaches with watershed areas less than 3,000 ha and ANC concentrations
less than 400 jueq L'1 . The comparable NSS Pilot target population represented 2,021 stream reaches.
5.3 NSWS LAKE AND STREAM DATA
5.3.1 Lakes in the Northeast Region
5.3.1.1 Lake Hydrologic Type
The NSWS classified lakes of the NE by hydrologic type, as described in the following paragraph.
"Lakes were classified by hydrologic type (Wetzel, 1983) through visual
examination of their morphology on the largest-scale topographic maps available.
"Seepage" lakes were defined as those lakes having no inlet or outlet. "Closed" lakes
5-25
-------
Table 5-6. Stream Identification (ID), Weight, and Inclusion Probabilities for the
Southern Blue Ridge Province Direct/Delayed Response Project Sample Watersheds
Stream ID
Inclusion
2A07701
2A07702
2A07703
2A07802
2A07803
2A07805
2A07806
2A07811
2A07812
2A07813
2A07816
2A07817
2A07821
2A07823
2A07826
2A07827
2A07828
2A07829
2A07830
2A07833
2A07834
2A07835
2A07882
2A08801
2A088Q2
2A08803
2A08804
2A08805
2A08806
2A08808
2A08810
2A08811
2A08901
2A08904
2A08906
ANC (/ieq L'1)
89.3
1,218.8
145.2
219.5
1710.5
98.8
104.4
16.2
102.7
371.7
56.5
30.4
126.5
102.5
347.7
234.7
48.2
64.8
217.2
211.8
43.2
96.3
106.5
1497.7
87.8
171.1
58.6
118.2
164.3
202.8
138.0
121.3
120.5
186.5
72.7
Weight
15.44026
30.33178
32.65306
17.46253
64.64653
93.43066
22.85709
26.72230
43.68692
13.34720
10.82906
12.36710
50.39373
16.93120
31.37254
32.32320
17.39136
15.12998
23.65990
21.47648
28.44442
11.99629
57.65760
75.73965
58.44749
50.39373
129.29292
38.67072
213.33337
39.87533
68.08512
99.22483
17.08941
25.49798
25.19680
Probability
0.06477
0.03297
0.03063
0.05727
0.01547
0.01070
0.04375
0.03742
0.02289
0.07492
0.09234
0.08086
0.01984
0.05906
0.03188
0.03094
0.05750
0.06609
0.04227
0.04656
0.03516
0.08336
0.01734
0.01320
0.0171 1
0.01984
0.00773
0.02586
0.00469
0.02508
0.01469
0.01008
0.05852
0.03922
0.03969
* Recalculated values (see text for explanation}
5-26
-------
Table 5-7. Stream Identification (ID) and Name, and State and
Latitudinal/Longitudinal Location of the Southern Blue Ridge
Province Sample Watersheds, Sorted by Stream ID
Stream ID Stream Name
2A07701
2A07702
2A07703
2A07802
2A07803
2A07805
2A07806
2A07811
2A07812
2A07813
2A07816
2A07817
2A07821
2A07823
2A07826
2A07827
2A07828
2A07829
2A07830
2A07833
2A07834
2A07835
2A07882
2A08801
2A08802
2A08803
2A08804
2A08805
2A08806
2A08808
2A08810
2A08811
2A08901
2A08904
2A08906
Sugar Cove Creek
Childers Creek
Hall Creek
Puncheon Creek
Chestnut Flats Branch
Cosby Creek
Roaring Fork
False Gap
Correll Branch
Little Sandymush
Eagle Creek
Forney Creek
Grassy Creek
Brush Creek
Henderson Creek
Welch Mill Creek
White Oak Creek
Catheys Creek
Mud Creek
Allison Creek
Brush Creek
Middle Saluda River
Little Branch Creek
Perry Creek Tributary
Dunn Mill Creek
Owenby Creek
Bear Creek
Weaver Creek
Kiutuestia Creek Trib.
White Path Creek
Bryant Creek
Hinton Creek
Persimmon Creek
She Creek
Deep Creek
State
TN
TN
NC
NC
NC
NC
NC
NC
NC
NC
TN
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
SC
NC
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
Latitude
0 1 «
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
34
34
34
34
34
34
34
34
34
34
34
34
19
11
5
54
46
47
49
41
40
42
29
30
27
19
22
11
13
12
15
7
6
7
26
57
56
59
49
52
51
44
36
29
54
50
40
20
25
44
36
48
37
17
59
33
12
54
48
51
8
42
6
33
48
17
17
50
14
59
37
57
13
28
16
32
15
35
7
47
6
37
Longitude
0 < "
84
84
84
82
83
83
82
83
83
82
83
83
82
83
82
83
83
82
82
83
83
82
83
84
84
84
84
84
84
84
83
84
83
83
83
6
29
19
32
47
14
53
23
5
45
45
33
16
31
23
53
37
47
30
28
15
32
3
44
26
8
33
18
1
25
59
25
30
20
27
1
23
32
56
47
22
33
2
19
38
49
28
55
0
5
38
7
9
2
28
28
19
50
13
18
47
58
0
25
59
57
17
7
42
22
5-27
-------
Table 5-8. Stream Identification (ID) and Name, Sorted
by State - Southern Blue Ridge Province Sample Watersheds
State
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
SC
TN
TN
TN
TN
Stream ID
2A08801
2A08802
2A08803
2A08804
2A08805
2A08806
2A08808
2A08810
2AQ8811
2A08901
2A08904
2A08906
2A07703
2A07802
2A07805
2A07806
2A07811
2A07812
2A07813
2A07817
2A07821
2A07823
2A07826
2A07827
2A07828
2A07829
2A07830
2A07833
2A07834
2A07882
2A07835
2A07701
2A07702
2A07803
2A07816
Stream Name
Perry Creek Tributary
Dunn Mill Creek
Owenby Creek
Bear Creek
Weaver Creek
Kiutuestia Creek Tributary
White Path Creek
Bryant Creek
Hinton Creek
Persimmon Creek
She Creek
Deep Creek
Hall Creek
Puncheon Creek
Cosby Creek
Roaring Fork
False Gap
Correll Branch
Little Sandymush
Forney Creek
Grassy Creek
Brush Creek
Henderson Creek
Welch Mill Creek
White Oak Creek
Cathey's Creek
Mud Creek
Allison Creek
Brush Creek
Little Branch Creek
Middle Saluda River
Sugar Cove Creek
Chtlders Creek
Chestnut Rats Branch
Eagle Creek
5-28
-------
SOUTHERN BLUE RIDGE PROVINCE
SITE LOCATIONS
DDKP Sitei
+ Pilot StridB Sir«ey Situ
Figure 5-8. DDRP stream reach study sites in the Southern Blue Ridge Province.
5-29
-------
were those with inlets and no outlets. Lakes with outlets but no inlets or with both were
termed "drainage" lakes. A fourth category comprised artificial lakes or "reservoirs"." (From
Linthurst et al., 1986a; Section 2.4.3 "Lake Type")
During the course of DDRP field mapping, aerial photo-interpretation, and field auditing and
checking (see Section 5.4), we found that numerous lakes classified by the NSWS as seepage or closed
actually fit the NSWS classification of drainage lakes. Lakes falling in the DDRP sample that were
originally classified by the NSWS as seepage or closed are indicated in Table 5-9 along with their final
classification by the DDRP. The final DDRP classification of lake hydrologic type for the entire DDRP
sample is shown in Plates 5-2 through 5-6. This classification is often important in determining to which
lakes or watersheds certain analyses are applied (e.g., see Section 7.2.2).
5.3.1.2 Fall Index Sampling
As discussed in Section 5.2, the DDRP was designed in a manner consistent with and dependent
upon the NSWS sampling of lakes and streams. The ELS Phase I was based on the concept of "index*
sampling. This conceptual basis was the result of much consultation among chemical limnologists and
statisticians specializing in sampling statistics. The approach was exhaustively reviewed prior to the
NSWS sampling and has proven to be a very powerful tool for answering the types of regional questions
posed by the NSWS. The NSWS index sampling approach is described below.
"A critical issue in the design of the ELS-I was the representation of a selected lake.
If a single water sample can adequately represent the chemistry of a lake to satisfy the
specific objectives of a study, a large number of lakes can be sampled. If multiple water
samples are needed on a single occasion, then a reduction in the number of sample lakes
must be considered. If multiple occasions are needed to represent the chemistry of a single
lake, the number of sample lakes must be reduced proportionally.
It is obvious that one sample, from one location, at one time of the day, in a specific
season of a particular year, cannot characterize the complex chemical dynamics of a lake.
Such a sample is justified only in the sense that it is an index to the essential characteristics
of the lake.. But even if two samples are taken, or three, they remain only indices because
understanding the dynamics of a single lake requires far more detailed study. This study was
designed to describe populations of lakes. Therefore, each lake must be represented in that
population description in a manner that captures its essence, but such that the number of
lakes that can be sampled is maximized. The single index sample maximizing both lake
number and spatial coverage on a large geographic scale was therefore deemed the most
appropriate choice for addressing the collective objectives of the ELS-I.
To enhance the utility of the index sample, careful consideration was given to location
and season. The sampling window was designated as the fall season, just after turnover.
Spatial variation within the lake is reduced at this time. Sampling at the apparently deepest
part of the lake was intended to provide a sample from the dominant water mass. Therefore,
5-30
-------
Table 5-9. DDRP Reclassification of Northeastern Lakes Classified as "Seepage"
or 'Closed' by the NSWS
Lake ID
1A1-039
1A1-066
1A2-006
1A2-058
1 A3 -028
1C1-018
1C1-031
1C1-050
1C1-068
1C2-056
1C2-066
1C3-055
1D1-027
1D1-034
1D1-037
1D1-068
1D2-036
1D2-084
1D3-044
1E1-009
1E2-007
1E2-049
1E2-069
S = Seepage lake
C = Closed lake
D = Drainage lake
Original
NSWS Class
S
C
S
C
S
S
S
S
C
C
S
S
S
S
S
S
C
S
S
S
S
S
S
Final
DDRP Class
D
D
S
D
S
D
D
D
D
D
D
S
D
D
S
S
D
S
D
S
S
D
D
5-31
-------
Plate 5-2. Final DDRP classification of lake hydrologic type - Subregion 1A.
5-32
-------
SUBREGION 1A
LAKE HYDROLOGIC TYPE
HYDROLOGIC TYPE
Dra i nage
Reservoi r
Closed
Seepage
-------
Plate 5-3. Final DDRP classification of lake hydrologic type - Subregion IB.
5-33
-------
SUBREGION IB
LAKE HYDROtOGIC TYPE
Subreg t on
Locat i on
HVOROLOGIC TYPE
J Dra i nage
[HJ Reservoir
|H Closed
jpj Seepage
-------
Plate 5-4. Final DDRP classification of lake hydrologic type - Subregion 1C.
5-34
-------
SU6REGION 1C
LAKE HYDROLOGIC TYPE
SubregI on
Locst ion
HYDROLOGIC TYPE
1] Drainage
[H] Re s e r v o i r
Closed
Seepage
-------
Plate 5-5. Final DDRP classification of lake hydrologic type - Subregion 10.
5-35
-------
SUBREGION ID
LAKE HYDROLOGIC TYPE
Subregi on
Locai i on
HYDROLOGIC TYPE
J Drai nage
[pl Reservo i r
If] Closed
{£] Seepage
v-
101-084
-------
Plate 5-6. Final DDRP classification of lake hydrologic type - Subregion 1E.
5-36
-------
SUBREGION IE
LAKE HYDROLOGIC TYPE
HYDROLOG3C TYPE
Dra i nage
Reservoir
Cl osed
Seepage
-------
the combination of a fall season sampling period and collecting a sample near the lake
center at the apparently deepest part, appeared to be the best protocol to provide the
needed sampling characteristics.
The perspective that each lake is represented by an index chemistry, rather than, for
example, -mean chemistry or some other integration over time and space, is important in
interpreting the results presented in this report. The population descriptions represent and
characterize the chemistry of a population of lakes, as though every lake in the population
had been sampled in the same manner as the sampled lakes. Thus the resulting frequency
and area! distributions for the chemical parameters (Sections 4.2 and 4.3) represent an index
to water mass chemistry for the population of lakes that can be interpreted only through
study of the predictive capacity of that index." (From Lfnthurst et al., 1986a; Section 2.1.2
"Lake Representation").
5.3.1.3 Chemistry of DDRP Lakes
The complete chemistry of the lakes of the DDRP watersheds in the NE has been given by
Kanciruk et al. (1986a) and will not be repeated here. The pH-ANC relationship for ELS Phase I lakes
falling in the DDRP target population (i.e., ANC < 400 /zeq L~ , Section 5.2.4.1) is shown in Figure 5-9.
Also shown in Figure 5-9, for comparison, is the pH-ANC relationship for the DDRP study lakes by
themselves. The ANC referenced for DDRP lakes is the modified Gran for ANC.
5.3.2 Streams in the Southern Blue Ridge Province Region
5.3.2.1 Spring Baseflow Index Sampling
The index sampling concept for Phase I of the NSS Is described in the following paragraphs.
'Like the ELS-I components of the NSWS, the NSS-I relied on samples taken during
an appropriate season from a representative sample of water bodies to provide an index of
the chemical characteristics of the target population (Messer et al., 1986). In the Eastern
and Western Lake Surveys (Linthurst et al., 1986; Landers et al., 1987), a single mid-lake
sample taken during well-mixed conditions at fall turnover provided a reasonably good spatial
representation of the nonlittoral lake water volume. Furthermore, this fall index sample for
lakes can be related to water quality during other seasons of the year when chemical
conditions may be more critical for biota (Driscoll and Newton, 1985; Newell, 1987). In lakes,
relatively long hydraulic residence times (low flushing rates) tend to integrate the inputs of
water and dissolved materials from the lake watershed, which reduces that portion of the
chemical variability caused by changes in input rates. Streams generally exhibit greater within-
and among-season variability than do lakes. Since streams have little temporal integrative
capacity within their channels, it is necessary to draw an index sample during a period of
the year that is expected to exhibit chemical characteristics most closely linked to acidic
deposition or to its most deleterious effects. Sampling the relatively stable chemistry of late
summer baseflows dominated by groundwater, for example, would provide a poor index of
potentially limiting conditions during winter and spring periods when the stream water is
poorly buffered against pH changes. The choice of the spring index sampling period for
streams was based on a literature search followed by a series of meetings with hydrologists,
biochemists, and fishery experts in Pennsylvania, Virginia, North Carolina, Florida, and
5-37
-------
si Total Northeast NSWS
-100
too
ANC
200
-1)
300
400
8-
7-
*••
5-
4-
-1<
Northeast DDRP . B
*..•• •"" """" "
X-rc"
l"
f
A
•L
•
X) 0 100 200 300 400
ANCftieqL-1)
Figure 5-9. The pH-ANC relationship for (A) lakes of the ELS Phase I sampling In the Northeast
and (B) DORP study lakes in the Northeast.
5-38
-------
Arkansas to discuss ongoing projects involving stream chemistry and fisheries in the
proposed NSS-I study areas {U.S. EPA, 1984b). The choice involved a trade-off between
minimizing within-season and episodic chemical variability and maximizing the probability of
sampling during chemical conditions potentially limiting for aquatic organisms.
A number of sources of stream chemistry data from several geographic areas support
the choice of a spring index sampling period for observing prolonged periods of low pH and
ANC. Ford et al. (1986), for example, summarized the results of four recent (1984-1985)
studies of seasonal and short-term variability in six second- and third-order streams in the
Catskill Mountains of New York (Murdoch, 1986), the Laurel Hills of Pennsylvania (Witt and
Barker, 1986), the Southern Blue Ridge Province of North Carolina and Tennessee (Oiem,
1986), and the Ouachtta Mountains of Arkansas (Nix et al., 1986). Minimum flow-weighted pH
values and concentrations of base cations and ANC occurred during the spring at almost all
sites. Those sites with minimum values during the winter had spring values nearly as low.
For a spring index sampling period to be biologically relevant, however, sensitive
life-stages of aquatic biota must also be present during the sampling period. Studies have
indicated that all life stages of fish are not equally sensitive to acidity and chemical
constituents that accompany low pH conditions in surface waters. Some of these studies
involved observations of acidic lakes and streams in which viable eggs were found together
with older age classes of fish that appeared to be spawning successfully, but in which young
age classes were absent (e.g., Beamish et al., 1975; Muniz and Leivestadt, 1980; Kelso and
Gunn, 1982; Gunn and Keller, 1984; Sharpe et al., 1984). Such a population structure
suggests more pronounced effects of acidity on larval fish than on egg hatching or adult
survival. These field observations are in agreement with laboratory bioassays that also
indicate greater sensitivity of fry to low pH conditions, relative to other fish life stages
(Schofield, 1976; Haines, 1981). Fry of the most important sport fish are present in the NSS-I
study area during the March 15 - May 15 period. Fry of some trout (Salmo spp.) populations
may also be present at other times of the year.
In summary, spring appears to be the most appropriate index sampling period for
streams, because ANC is typically low, and life stages of aquatic biota that are sensitive to
low pH are likely to be present at this time. The low ANC during the season minimizes
buffering against episodic pH changes accompanying high runoff. Although pH and ANC
depressions can also occur during other seasons, they may be more pronounced during
the spring because short hydraulic residence times in the soil during the spring minimize
acid neutralization. Also, acid-sensitive, swim-up fry of key fish species are typically present
in streams during the spring in many parts of the United States. The index sampling period
for the NSS-I thus was chosen as the time period following snowmelt but prior to leafout
(mid-March to mid-May, depending on the subregion). Results of the NSS-I Pilot Survey in
the Southern Blue Ridge showed very little difference in separate population distributions of
pH, ANC, and major cations and anions based on three successive spring baseflow samples
during this sampling window (Messer et al., 1986, 1988). The occurrence of large episodic
chemical changes over the course of hours or days during storm runoff, however, makes the
use of spring samples for indexing water chemistry difficult, unless sampling during such
events is avoided (Messer et al., 1986). To avoid alterations in index chemistry caused by
atypical stormflow samples, the NSS-I avoided sampling within 24 hours following significant
rain events (> 0.2 inches).
Unlike lakes, for which a single mid-lake sample taken during well-mixed conditions
at fall turnover can provide a reasonably good spatial representation of the nonlittorai
lakewater volume, a sample taken at a single point on a stream reach would not adequately
describe chemistry for the whole length of the reach (Messer et al., 1986). Streams were
expected to exhibit substantial trends in chemistry over their length at any given time during
the spring index period. To incorporate this variability and to establish a basis for quantifying
relationships between upstream and downstream chemistry on sample reaches, samples from
both ends of the reaches were collected in the NSS-I." (From Kaufmann et al., 1988; Section
2.5 "Index Sampling*)
5-39
-------
As discussed in Section 5.2.4.2, DDRP study watersheds in the SBRP were defined based upon
the downstream nodes of the reaches sampled.
5.3.2.2 Chemistry of DDRP Stream Reaches
The complete chemistry at the downstream nodes of the stream reaches used to define the DDRP
study watersheds in the SBRP has been given by Messer et al. (1986a) and Sale et al. (1988) and will
not be repeated here. The pH-ANC relationship for samples taken at the downstream nodes of the
reaches for ANC < 400 fjeq L*1 is given in Figure 5-10. Also given, for comparison, is the pH- ANC
relationship for the samples taken at the downstream nodes of the DDRP reaches. Only the relationship
for ANC < 400 /Jeq L*1 is shown. The ANC referenced for the stream reaches is the modified Gran for
ANC.
5.4 MAPPING PROCEDURES AND DATABASES
The first step in gathering the terrestrial information required to characterize the study watersheds
was to map them. This mapping was designed to include all the major characteristics thought to be
important in determining the response of surface waters to acidic deposition for watersheds selected to
represent the study region. Existing terrestrial databases were examined and found to be highly limited
(Lee et al., I988a).
Specific resource inventories of soils, geology, depth to bedrock, drainage, forest cover type, and
land use were designed within the Project and implemented through the assistance of the USDA Soil
Conservation Service (SCS) in the NE and SBRP Regions.
The performance and direction of field activities in the Soil Survey were modelled after the
organization of the National Cooperative Soil Survey (Soil Survey Staff, 1983). The Mapping Task Leader
for the DDRP, located at the ERL-C, had overall responsibility for mapping and coordinated all mapping
activities. A Regional Coordinator/Correlator (RCC), an independent contractor, provided quality
assurance/quality control (QA/QC) for the field mapping. The RCC maintained a uniform, consistent
5-40
-------
8-
7-
5-
4-
-1(
A
*
• " • • •
• •
» 0 100 200 300 400
ANC&ieqL-')
8-
7-
£6-
5-
4-
-1(
B
» 0 100 200 300 400
ANC (jieq L-1)
Figure 5-10. The pH-ANC relationship for samples with ANC < 400 peq L'1 taken at the
downstream nodes of stream reaches sampled in the NSS. Shown are the relationships for all
such samples from the NSS and samples for the downstream nodes of DDRP study reaches in the
SBRP. Samples are the average of either two or three samples each, with samples taken during
events excluded.
5-41
-------
regional mapping legend, participated in at least one field review of mapping procedures for each state,
ensured regional consistency of field procedures, and evaluated mapping activities to assure quality.
Each State Soil Scientist (USDA-SCS), with the support of the State Soils Staff, was responsible for
mapping activities in that state. This included supervising and coordinating field and support crews,
forwarding maps and notes to the Mapping Task Leader, performing at least one field review of each
field crew, and working with the RCC to ensure regional consistency. The field crews, led by an
experienced soil scientist, mapped the watersheds, described the soils and soil map units, and transected
each watershed to determine the correctness of the mapping.
Map products delineated at a scale of 1:24,000 were digitized in separate layers and entered into
a Geographic Information System (GIS) as described in Sections 5.4.1.7 and 5.4.2.8 of this report. The
soil map legend, map unit composition, characteristics of the soils, and soil transect data described from
the mapping were entered into an interactive microcomputer data management system. The survey of
each region was independent, in that a single unified and consistent legend was developed and correlated
within each individual region.
5.4.1 Northeast Mapping
Mapping of soils, watershed drainage, geology, forest cover type, and depth to bedrock on 145
watersheds in the NE was initiated on April 15, 1985, and completed before July 5, 1985; the total area
mapped was about 75,000 ha (185,000 acres). Soil mapping activities and quality assurance of the
mapping data were described in depth by Lammers et al. (I987b).
Before field work started, mapping protocols were written, a preliminary regional soil legend was
developed, and a plan of operations was prepared for each state. Mapping protocols, used by personnel
involved in the mapping to maintain quality and consistency, were described in detail by Lammers et al.
(1987b, Appendix C). The preliminary soil legend was based on soil map units that had been mapped
within the region. These map units, therefore, had established soil-landscape relationships and were
5-42
-------
expected to be applicable to much of the area to be mapped. A plan of operations was prepared by
the SCS State Soil Scientist in each state to direct the flow of personnel and mapping products. USGS
topographic quadrangle maps at a scale of 1:24,000 were acquired for the watershed areas and prepared
for field use. For areas where 7.5' maps were not available, 15' topographic quadrangle maps were
photographically enlarged to an approximate scale of 1:24,000. When available at SCS field offices,
aerial photographs were used to assist with landscape interpretation and map delineation. Just prior to
the start of mapping, State Soils Staff, Field Soil Scientists, and Mapping Task Leaders met at a workshop
to review mapping protocols and to clarify instructions.
5.4.1.1 Soils
Soils were mapped using standards and procedures specified in the National Soils Handbook (Soil
Survey Staff. 1983) and Soil Survey Manual (Soil Survey Staff, 1981). Soils were classified according to
Soil Taxonomy (Soil Survey Staff, 1975). Soils map units were delineated directly on topographic
quadrangle base maps and identified with a unique map symbol. Each map unit represented a collection
of areas defined and named the same in terms of their soil components, miscellaneous areas, or both.
Units that consisted of one dominant component (consociation) and units with two or more dominant
components (complexes) were mapped. Although most soil components of map units were phases of
soil series, some components were phases of soil families or higher categories of taxonomic classes.
The soil map units and soil components that make up the map units were described with the following
characteristics:
name and symbol of the map unit
regional landform
local landform
geomorphic position
slope configuration
percent composition of map unit components
characteristics of the soil components
name of the soil component
phase (i.e., slope, texture, rock fragments)
drainage class
parent material, origin and mode of deposition
depth to bedrock
depth to impermeable layer
taxonomic classification
5-43
-------
Soil map units were cartographic delineations of the landscape that reflected the dominant soil
conditions of a landscape element or segment. Soil mapping was based on enough observations to
determine soil/landscape relationships and confirm predictions of soil occurrence established from these
relationships. Each map delineation was visited in the field. The minimum size of individual map
delineations was 2.7-4.5 ha (6-10 acres). Inclusions may have been larger if the soils were similar and
there were no readily observable landscape features to use for delineation. The proportion of small areas,
inclusions In map unit delineations, were estimated to the nearest 5 percent and were, thereby, included
in aggregated values for a watershed. Soil map unit boundaries were delineated directly on USGS
topographic quadrangle base maps at a scale of 1:24,000. Perennial and intermittent drainages not
indicated on the USGS maps were drafted onto the base maps.
5.4.1.1.1 Soil Correlation -
Soil correlation is the process of maintaining consistency in naming, classifying, and interpreting
soils and units delineated on maps. Thus, there are two main elements of soil correlation: (1) the
correlation of an individual soil pedon or groups of soil pedons with a soil series, or with some higher
level soil taxonomic class, and (2) the correlation of map units. Correlation requires consistent methods
of observation and measurement among all participants, as well as the use of consistent conventions and
terminology. The Soil Survey Manual (Soil Survey Staff, 1981) and Soil Taxonomy (Soil Survey Staff,
1975) provide the conventions and guidelines for defining and naming map units, and for defining
diagnostic properties and taxonomic classes of soils used in the National Cooperative Soil Survey. Soil
series are defined by official soil series descriptions.
The soil correlation process started with the development of the preliminary regional identification
legend and continued throughout the progress of the mapping phase. The preliminary identification
legend was based on soil map units that had been mapped previously within the NE. These map units,
therefore, had been tested for soil-landscape relationships and were expected to be applicable to much
of the area to be mapped. Consistent breaks for slope phases for map units and the use of the most
common soil texture phase for a soil series were established by the preliminary legend, in which 623 units
5-44
-------
were listed. Soil map units were not limited to those in the preliminary legend and map units were
redefined and added as necessary during the progress of the field mapping; 89 map units were added
to the identification legend during the field mapping. From the total of 712 map units in the preliminary
legend and those added to the legend, 398 map units were used in the mapping.
The soil scientist responsible for mapping each watershed performed the first level of correlation
of the soils and map units. The descriptions of official soil series were adopted to represent those soil
series for the region. At each point along a traverse, the soil was examined and evaluated for
characteristics that were within the range of a soil series or that were similar to an established series.
Soils that were dissimilar to all recognized soil series were classified at the family level of Soil Taxonomy
(Soil Survey Staff, 1975). Brief descriptions were made of the different kinds of soil to document what
was observed and to compare or correlate with the official series description or other field descriptions.
The descriptions of the different recognized kinds of soil were further evaluated by the State Soils Staff
during the progress field review for consistent correlation within each state.
In addition to the proper recognition and classification of soils observed, the soil scientist also
determined the relative proportion of each kind of soil on a landscape segment. In this manner, soil map
units were defined. The soil scientist then correlated the composition of soils with one of the map units
in the preliminary legend or proposed an additional map unit. The map units were reviewed by the
State Soils Staff to correlate map units on all watersheds within each state. The RCC controlled the
mapping legend and correlated soil map units throughout the NE Region.
During the week of July 8-12, 1985, soil scientists representing the SCS from all of the states
involved in mapping the DDRP NE Region met at Saranac Lake, NY, with the RCC, task leaders from the
ERL.-C, and the Data Management Leader from ORNL. Objectives of the meeting were to correlate soils
and soil map units for the region and to complete descriptions of the map units. Each of the 398 map
units used during the mapping was reviewed. Descriptions of the map units and the characteristics and
taxonomic classification of the major components of each map unit were checked and completed.
5-45
-------
A few units mapped in more than one state were found to be similar, and they were combined.
Other map units were represented by just a few hectares and were combined with the most similar map
unit in the legend. When transect data or field notes indicated that the map unit was not correct, the
description was adjusted or the map unit was combined with another that better fit the soils recognized.
Map unit descriptions, defined during the mapping and correlated within states, and summaries of the
mapping transects were the basis for correlation and map unit description decisions. The state with the
greatest area of each map unit took the lead responsibility for providing a description of the map unit.
Transect summaries from every state mapped were summarized on a regional basis to determine a
consensus description. When transect data did not appear to accurately represent the map unit, soil
scientists with experience in mapping that unit were asked to alter the description. Most often, the
alterations were based on the kinds and percentages of minor soil components in the map unit.
Following the regional correlation review, 356 map units remained in the regional soils legend. After the
area of each map unit was more precisely determined from the digitized data in the GIS, additional map
units with only a few acres were combined with other similar map units by the Mapping Task Leader.
This resulted in a final soil map legend of 338 map units. A few small map units remained in the legend,
if there were no similar map units with which they could be combined.
The soil taxonomic class, drainage class, depth to bedrock, and estimated depth to a slowly
permeable or impermeable layer were obtained from the official soil series for the major components of
each map unit.
5.4.1.1.2 Soils database -
The mapping phase of the DDRP NE Soil Survey generated vast amounts of data. In order to
verify, validate, and analyze these data, the data were entered into computer database files. Data
products generated by the mapping included the identification legend, descriptions of the soil map units,
descriptions of the soil taxonomic units (components of the map units), soil transect information, and the
map products. The map products included maps of the soils, vegetation, depth to bedrock, and geology
of the 145 watersheds. This section describes the database files developed for the ODRP mapping data
5-46
-------
and the procedures and QA/QC checks used during the computerization of the DORP data. Both ORNL
and EPA's ERL-C were involved with management of the mapping data. Most of the data were double
entered by ORNL, using the Statistical Analysis System (SAS) installed on tandem IBM 3033 computers.
ORNL also performed most of the data checking. ERL-C had overall responsibility for the quality of the
data and contributed to the development of the database files. The maps were digitized for input to a
GIS at ERL-C. An overview of mapping databases is provided by Turner et al. (in review).
The preliminary soil identification legend, including additions and corrections to the legend made
during the mapping, was reviewed by the RCC during the regional correlation workshop held in Saranac
Lake, NY, in July 1985. Map units not used were marked for deletion, and map units that were combined
were noted on the legend. The legend data were input using dBase III software on an IBM PC at ERL-
C and also double entered at ORNL by in-house data entry center personnel, with the resulting files
transferred to SAS files on the IBM 3033 system. Legends from each watershed map also were entered
into the GIS as the maps were digitized at ERL-C.
The legend data from the GIS were transferred to a dBASE III file where they were summarized
for the region and then compared to the regional soil identification legend. Discrepancies were then
resolved and the map unit names were checked with the descriptions of the map units for validity. Map
units from the GIS database showing less than 8 ha (20 acres) were then combined with another similar
map unit where possible. Usually this procedure involved either including the major component of the
minor map unit with another slope phase of the same soil and adjusting the slope range or showing an
inclusion of the soil on the different slope. The ERL-C version of the identification legend was then
compared with the ORNL version and discrepancies were resolved.
The soil identification legend database file for the DDRP NE Soil Survey, NEIDLGD, contained the
following information for each map unit: map symbol; map unit name, including the name of major soil
component(s), texture modifier (e.g., gravelly, mucky), texture phase, slope phase, and other phase (e.g.,
very stony, rocky); regional landform; local landform; geomorphic position; slope shape across; slope
5-47
-------
shape down; and area in acres (determined from the GIS database). This file contains 338 records,
one for each map symbol in the soils legend.
A soil map unit worksheet was used to record Information about each map unit. This worksheet
included the map symbol, map unit name, information about the landscape, major soil components, minor
soil components, the proportion of each component in the map unit, and information about the major
components including the taxonomic classification. Originally the minor components or Inclusions were
only listed by name and percent composition. After the soil correlation workshop at Saranac Lake, NY,
the map unit worksheet data were entered into a database file using dBASE III software on an IBM PC.
Inasmuch as soil data analyses must be made on kinds of soil or classes of soils, it was immediately
evident that individual components of map units must be recognizable in the database, not the map units
themselves. In some map units, the minor components (inclusions) collectively made up more than 30
percent of the map unit and were found to be important for data analyses. Also, a major soil component
in a consociation may have the same attributes as a major component in a complex or minor component
in another map unit. The information from the map unit worksheet was therefore separated into two
files, a map unit composition file and a soil components file. Each unique soil component was assigned
a component code to aid in accessing all the attributes of a soil component with one code. The map
unit composition file, NECMPOS contains the map symbol, the component code for every component
in the map unit, and the percent composition of each of the components. There are 1381 data records
in the NECMPOS file.
The soil components file was named NECMPON and has 594 records. Each record includes the
component code; soil name, texture, and slope of the component; five characteristics of the soil:
drainage, depth to bedrock, depth to impermeable layer, origin, and mode of deposition of the parent
i-
material; and the taxonomic class. The sampling class code for the class with which the soil component
was grouped for sampling was also included in the record for each component in this database file.
5-48
-------
The records from the three database files, NEIDLGD, NECMPOS, and NECMPON, were merged
for printout of a computer-generated map unit worksheet. Copies of these computer-generated
worksheets were sent to the SCS State Soil Scientist in each of the northeastern states for review.
Instructions were to review the map units used in their respective states, make corrections, and fill in
data blanks wherever possible. Data from these corrected map unit worksheets were entered into the
SAS files at ORNL The updates were entered into a change file containing the record identifier, variable
name, and old value for each record in the database. Only when all three items matched an observation
in the database was the old value updated. This method of correcting the database virtually eliminated
the possibility of updating the wrong observation or variable.
After the updates were completed, ORNL generated frequency tables of the coded variables and
compared these tables with lists of valid codes. The frequency tables were also used to build code
translation tables containing the codes and definitions. These translation tables were stored as SAS
format libraries and are a part of the database. The final step in editing the map data files involved
labeling variables and, where necessary, modifying variable names and labels to ensure consistency
among the various mapping data files.
5.4.1.2 Depth to Bedrock
Depth to bedrock maps were prepared on mylar overlays of base maps at a scale of 1:24,000
during soil mapping. Soil depth was observed while traversing all map unit delineations and at an
average of 100 transect stops in each watershed. Soil scientists usually examined the soil to a depth of
1.5 m, or depth to bedrock or dense till. Because of this direct observation, soil scientists were highly
confident in the reliability of depth-to-bedrock estimates within the depth of observation. Estimates of
depths greater than 2 m were based on road cuts, stream incisements, and knowledge of the landscape;
the confidence in the reliability of these estimates was lower. Each soil map delineation was assigned
to one of six depth classes and a depth-to-bedrock map was prepared by combining contiguous
delineations of the same class. The six depth-to-bedrock classes and qualitative estimated reliability In
determining the correct class are shown in Table 5-10.
5-49
-------
Table 5-10. Depth-to-Bedrock Classes and
Corresponding Level of Confidence
Class Depth Midpoint Estimated
Reliability
I < 0.5 m - High
II 0.5 - 1 m 0.75 High
III 1 - 2 m 1.50 High
IV 2-5 m 3.50 Moderate
V 5 - 30 m 17.50 Low
VI > 30 m - Low
5-50
-------
Standard seismic refraction techniques were employed to estimate depth to bedrock along selected
transects in 15 of the 145 watersheds. Depth to bedrock estimated from soil mapping and from seismic
techniques could not be directly compared due to differences in the two approaches. Of the 696 seismic
readings, 83 percent were within one class of that on the depth-to-bedrock map. Means of the seismic
determined depths increased with increasing mapped depth class for all classes.
The extent of depth-to-bedrock classes on a watershed could also be estimated directly from the
soils database. This approach, is described in Section 8.7.2 and was used in the regression analysis
reported in that section. The percent (in intervals of 5 percent) of exposed bedrock or soil having a
designated depth to bedrock was estimated for each soil map unit and recorded with the map unit
composition in the soils database. In this manner, areas of rock outcrop or of soil with a depth different
depth than that of the major component of the map unit and too small to be delineated at the scale of
mapping were accounted for in the analysis.
5.4.1.3 Forest Cover Type
During the soil mapping, soil scientists also made vegetation cover type maps, at a scale of
1:24,000 for each watershed. The vegetation map units were based on Society of American Foresters
(SAP) cover types described by Eyre (1980). Open areas containing poorly drained soils were delineated
as "open areas-wet," while other open areas were delineated simply as "open." Delineation was made
by air photo imagery and topographic and landscape features. Delineations were confirmed by field
observation during the course of soil mapping.
5.4.1.4 Bedrock Geology
Reid soil scientists obtained the best available bedrock geology map and sketched delineations
of the formations on an overlay of the base map for each watershed. Geology data extracted from
several different sources were found to be extremely difficult to correlate; therefore, bedrock geology
was digitized from state geology maps as discussed in Section 5.4.1.7.
5-51
-------
5.4.1.5 Quality Assurance
A rigorous plan for QA/QC was implemented from the beginning of the mapping and maintained
at every level of authority. QA/QC activities included field review, point transects of the watersheds, and
independent evaluation of mapping on selected watersheds. The transect data were evaluated to
determine the correctness of the soil map units.
5.4.1.5.1 Field reviews by the Soil Conservation Service -
Field reviews were conducted by the SCS, the State Soil Scientist, or another member of the State
Soils Staff for each soil mapping crew in their respective states. During a review, the watershed was
visited and a number of map unit delineations were traversed. The mapping was evaluated and the
following items were checked: adherence to protocols, identification of soils, placement of map unit
boundaries, identification of soil map units, and clarity and legibility of the field notes and maps. There
were 34 different soil mapping crews responsible for the mapping, and field reviews were conducted on
watersheds mapped by 32 of the crews. A written field review report was submitted to the Mapping Task
Leader and the RCC following the field review. Field review reports for each of the watersheds are
summarized by Lammers et al. (1987b, Appendix E).
None of the field review reports indicated that the mapping was unacceptable. Field reviews
clearly resulted in improved mapping on the watersheds in which the reviews were conducted. Although
it cannot be quantified, undoubtedly the communication and feedback resulting from the field reviews
improved mapping on the other study watersheds.
5.4.1.5.2 Field reviews by the Regional Coordinator/Correlator -
The RCC participated in a field review of at least one watershed in each state (except Vermont
for which there was only one watershed). The purpose of the RCC participation was to coordinate the
mapping throughout the region and to control the quality of the mapping. The RCC facilitated better
communication among states to ensure consistency and improve correlation of the soils and map units.
5-52
-------
The RCC participated in 13 field reviews and submitted an independent narrative report of the
results of each review to the Mapping Task Leader. The mapping was judged acceptable on all of the
watersheds after discrepancies were corrected. Field review reports by the RCC are contained in
Appendix F of Lammers et al. (1987b).
5.4.1.5.3 Evaluation of mapping by the Regional Coordinator/Correlator -
The RCC independently evaluated the mapping of 15 of the 145 watersheds in the Northeast Soil
Survey. These 15 watersheds were selected from the top of a random list of all 145 watersheds, with
the constraints that watersheds visited by the RCC for progress field reviews during the mapping would
not be revisited for independent evaluation, and no more than one watershed would be evaluated for each
mapping team.
Mapping was evaluated by examining stereoscopic pairs of aerial photographs, when available.
Relationships between soils and landform segments were scrutinized and questionable areas marked for
further examination on the ground by traversing and transecting. About one-third of the delineations on
the soil map were evaluated on the ground and about one-half as many transect points as in the routine
mapping were examined. A report of the results of the mapping evaluation was submitted to the EPA
ERL-C.
The soil mapping activity carried out by soil scientists within the framework of the National
Cooperative Soil Survey was in reality the art of sketching the landscape portrait to show a location of
areas; with defined kinds and distribution of soils, the soil map units. Although map correctness was
judged by how well the map unit descriptions fit the soils in the mapped areas, the utilitarian correctness
had to be judged with the DDRP in mind. For this project, depth to bedrock, depth to slowly permeable
or impermeable layer, drainage class, taxonomic family, slope, and stoniness were selected as important
soil characteristics.
5-53
-------
Of the 15 watersheds evaluated by the RCC, the mapping was judged acceptable on 13 and
unacceptable on the other 2. Mapping that was unacceptable did not mean that all the mapping on
that watershed was incorrect, but of the delineations checked, nearly one-half had an inappropriate map
symbol. Mapping in both of the unacceptable watersheds was corrected by the mapper and the State
Soils Staff.
The number of watersheds, other than the 15 evaluated by the RCC, that might have unacceptable
mapping errors could not be determined. Mapping errors were most likely associated with individual soil
scientists, with watersheds where soil mapping was dominantly based upon published soil surveys, or with
soils that were difficult to map.
Summaries of mapping evaluation results of the 15 watersheds selected for evaluation by the RCC
were reported in Appendix G by Lammers et al. (I987b). Significant watershed boundary errors were
identified on 3 watersheds during the mapping evaluation. The mapped area was adjusted as appropriate
to correct these errors. Watershed boundaries were difficult to determine in topography common to the
glaciated region of the Northeast.
5.4.1.5.4 Evaluation of soil transect data -
As specified In the mapping protocols (Lammers et al., 1987b), point observations were made at
30-m (100-ft) intervals along transects across each watershed. Transects were located to pass through
as many map unit delineations as was practical. A "yes" response was recorded when the soil was
similar to a named soil of the map unit in which the transect stop was located. In complex map units,
the name of the major soil was recorded with a "yes" response and the name of a dissimilar soil was
recorded with a "no" response. "Routine" transects were conducted by soil scientists with the soil survey
party, and additional "RCC" transects were made by the RCC in 15 of the 145 watersheds. The transect
data were entered into a SAS database at ORNL and were verified at ERL-C.
5-54
-------
A number of analyses could be made with the transect data, which were used to evaluate the
correctness of the described map units. For both the routine and RCC transect data, the proportion of
major components in map units, "yes" responses in the transect data, was compared to the estimated
percent composition in the map unit description, NECMPOS file. Routine transects were compared to
RCC transects of the same map unit for watersheds in common. Finally, soil components observed at
transect points were assigned the proper class for sampling, and map unit correctness was evaluated with
respect to the sampling class composition. This evaluation of the proportion of sampling classes in map
units was especially relevant to judging the "correctness" of map units for the purposes of the DDRP.
5.4.1.5.4.1 Analysis of major components in map units with routine transects —
Transect data "yes" responses, representing 274 of the 338 map units in the regional legend, were
compared to the proportion of major map unit component(s) in the map unit descriptions. Of these 274
map units, 39 were found to have significantly different proportions. Seven of the 39 map units had 100
or more transect observations and had a difference between the proportion from the transect and the
proportion estimated in the NECMPOS database of less than .09, or 9 percent.
Another 18 of the 39 map units had significantly different proportions, but fewer than 30
observations. These differences could be indicative of unrepresentative transects, rather than incorrect
expected proportions of major components. Given that transect points were not independent random
observations and that individual transect segments had not yet been analyzed for problems, the number
of map units with significantly different composition was reasonable.
A transect segment union was defined as all transect stops in the same map unit on a watershed.
Fifty-two transect segment unions, about 3 percent of all transect segment unions in the database, had
proportions significantly different from that estimated. When these 52 transect segment unions were
excluded from the dataset and map unit composition was analyzed, 29 map units were found to have
a significantly different proportion of "yes" responses compared to the estimated proportion of major
5-55
-------
component(s). Of these 29 map units, 13 had fewer than 30 observations and 9 had a proportion
difference of 9 percent or less.
5.4.1.5.4.2 Analyses of major components in map units with RCC transects -
As with the routine transects, the proportion of major map unit components from the "yes"
responses on transects conducted by the RCC on 15 watersheds was compared to the estimated
proportion in the NECMPOS data file. Of the 47 map units examined, 12 map units had proportions
that were significantly different. Ten of the 12 map units had less than 30 transect observations.
Similarly, the RCC transects were examined for transect segment unions with significantly different
proportions at the .01 level of significance. After significant transect segment unions were excluded, there
were five map units with significantly different proportions. Two of these were based on fewer than five
observations for which the power of the hypothesis test was limited. The percent of map units with
significantly different proportions was about the same for routine transects as for RCC transects.
These comparisons suggest that the correctness of the estimated composition of major map unit
components was about the same when analyzed with routine transects as when analyzed with transects
conducted by the RCC.
5.4.1.5,4.3 Comparison of the routine and RCC transects -
The analyses in the previous section compared the proportion of major components predicted
from the routine transects and those of the RCC with the proportion estimated in the NECMPOS data
file. The analysis in this section compares the proportion of "yes" responses in the routine dataset with
the proportion of "yes" responses in the RCC dataset for the same watershed and map unit
There were 94 watershed/map unit combinations, of which 47 had observations from both RCC
and routine transecting that could be compared. Five map units were found to have significantly different
proportions. If the true proportions were the same for each of the 47 comparisons, about two or three
5-56
-------
combinations would still appear to be significant. Furthermore, some of these significant combinations
were based on few observations, and RCC transects were not in the same places on the watersheds as
the routine transects. For these reasons, the RCC and routine transects are considered reasonably
consistent.
5.4.1.5.4.4 Analysis of the map units by sampling class -
The soil series in the transect data were assigned the appropriate sampling class. Sampling class
composition of the map units predicted from the transect data was compared to sampling class
composition estimated during the mapping by soil scientists and entered in the NECMPOS data file.
(Sampling classes are described in Section 5.5.1.3.) The Bonferroni (Johnson and Wichern, 1982)
inequality was used to handle the error rate of the simultaneous hypothesis tests within each map unit.
There were 38 map units, analysis for which the proportion of one or more sampling class(es)
differed significantly from the proportion estimated in the NECMPOS data file. Of these 38 map units,
17 had fewer than 30 transect observations, and 3 more map units, all with more than 100 observations,
had a difference of less than 9 percent. These 20 map units were therefore removed from the analysis,
leaving 18 map units with 30 or more transect observations, for which the difference in proportion of at
least one sampling class in the map unit was 15 percent or greater. At the .05 level of significance, we
would expect about 14 map units to have at least one sample class with a significantly different
proportion, suggesting that the sample class proportion from the NECMPOS data file was not noticeably
different from to that predicted from the transect data for most map units.
5.4.1.5.4.5 Summary of the transect analyses -
For the most part, the routine transect data were used to make estimates of map unit composition
in the NECMPOS data file. The RCC transects were independent of the NECMPOS data file estimates;
map unit component proportions, however, were consistent with proportions from routine transects. The
proportion of each sampling class in the map units, estimated in the NECMPOS data file, was within
reasonable mapping precision for more than 95 percent of the map units. Although transects were used
5-57
-------
as a measure of map unit composition correctness, it was recognized that soil scientist experience may
in some cases provide the better composition estimate.
5.4.1.6 Land Use/Wetlands
5.4.1.6.1 Data acquisition -
information on land use and wetlands for the NE DDRP watersheds was obtained via interpretation
of aerial photography. Details of the procedures used and evaluation of the results are presented by
Liegel et al. (in review).
During April and May 1986, leaf-off color infrared (CIR) stereo photography, 1:12,000 scale, was
obtained for 145 watersheds from Lockheed Engineering and Sciences Company, Las Vegas, NV (LESC-
contract no. 68-03-3245). The selected film, Kodak aerochrome 2443 or equivalent, was kept frozen until
a few hours before actual use. Two subcontractors were responsible for the actual photography; Zeiss
cameras and Zeiss B and D filters were used. Exposed film was packed in styrofoam mailers and
shipped to the contractor by next-day air courier service. The contractor used Kodak 1594 film
processing to make 23 x 23 cm contact prints.
Contractor staff made overlays of land use and wetlands from office photointerpretation of CIR
stereo film positive negatives. Information was transferred to 1:24,000-scale (7.5') USGS topographic base
map overlays. When 7.5' maps did not exist, photographic enlargements of 15' maps were used. On
the land use overlay, 12 general land use classes (Table 5-11) were mapped to a resolution of 2.5 ha.
On the second overlay, detailed wetlands, using modified National Wetland Inventory (Cowardin et al.,
1979) subcategories, were mapped to a resolution of 0.4 ha; greater resolution for wetlands was tied to
the suspected greater influence of wetlands on ameliorating surface water chemistry in areas of high
acidic deposition. Also, five point classes summarizing beaver activity were included (Table 5-11).
5-58
-------
Table 5-11. Interpretation Codes for Northeast Map Overlays -
Land Use/Land Cover, Wetlands, and Beaver Activity
Overlay
Type Class Subclass
Land use/ Cropland
land cover Forest land
Pasture land
Horticulture land
Cemeteries
Waste disposal land
Barren land
Gravel pits/quarries
Urban-commercial
Urban-industrial
Urban-residential (#)
Wetlands
Map
Unit
C
E
G
H
M
L
N
P
u
u°
u
w
Detailed Aquatic bed
wetlands
Emergent
Forested
algal AB1
aquatic moss AB2
rooted vascular AB3
floating vascular AB4
unknown submergent AB5
unknown surface AB6
persistent EM1
nonpersistent EM2
broad-leaved deciduous F01
needle-leaved deciduous F02
broad-leaved evergreen FO3
needle-leaved evergreen F04
dead F05
deciduous F06
evergreen F07
open water/unknown OW
rocky bottom RB
rocky shore RS
continued
5-59
-------
Table 5-11. (Continued)
Overlay Map
Type Class Subclass Unit
Scrub/Shrub broad-leaved deciduous SS1
needle-leaved deciduous SS2
broad-leaved evergreen SS3
needle-leaved evergreen SS4
dead SS5
deciduous SS6
evergreen SS7
Beaver unbreached dam U
activity breached dam B
old beaver dam O
beaver lodge A
impounded water IM
open water OW
5-60
-------
5.4.1.6.2 Field check protocols -
Independent QA/QC activities were required to evaluate both base map and air photo overlays and
to assess inherent photointerpretation discrepancies. To meet this requirement, field checks were made
of office interpretations for 15 watersheds, a 10 percent subsample (Figure 5-11, Table 5-12). Sample
watersheds were representative of DDRP watershed sizes and maximized mapped land use, wetland,
beaver activity, and stream variability found across the 145 NE watersheds. For example, for several
similar-sized watersheds, those with diverse land uses and wetlands were chosen over ones that had a
few wetlands or forest cover as the sole land use.
Staff from the Center for Earth and Environmental Science, State University of New York,
Pittsburgh (SUNY-P), performed the field QA/QC check of office-generated land use and wetland maps.
The QA/QC work involved two distinct phases (Boguckj et al., 1987): Phase I, field checks, and Phase
II, photointerpretation and evaluation. In Phase I, SUNY-P staff verified existing point and area land use
delineations at 5 to 12 sites per watershed that had been targeted for field checking by ERL-C staff.
Between specific check points, detailed observations also were made to characterize land use, wetlands,
and beaver activity existing across the landscape. Field checking was conducted during October and
November 1986, when most leaves had fallen from the trees.
In Phase II, SUNY-P staff independently mapped land use and wetlands on CIR stereo pair overlays.
Notes also were made on imagery quality factors that adversely affected photointerpretation, watershed
disturbances that could affect surface water chemistry (e.g., recent or historical logging), and interpretation
problems that were encountered.
ERL-C staff analyzed differences between office and field maps by Chi-square goodness-of-fit tests
to determine those categories that were significantly more difficult to map. The null hypothesis for each
test was that differences between office and field maps were proportional over all classification categories
used in the tests (Sokal and Rohlf, 1969). In the first test, the null hypothesis was that all general land
use categories were likely to have equal differences in interpretation. In the second test, the null
5-61
-------
Field Check Sites
Study Watersheds
NORTHEAST REGION
Field Check Sites
Figure 5-11. Location of Northeast field check sites and other DDRP watersheds.
5-62
-------
Table 5-12. Northeast Watersheds Studied for Independent Field
Check of Land Use and Wetland Photointerpretations
Lake ID
1A1-033
1A2-004
1A2-039
1A2-058
1B1-029
1B1-055
1B3-059
1C1-017
1C2-012
1C2-028
1C3-Q30
1D2-025
1D3-003
1D3-033
1E1-092
Name/State
Kiwassa, NY
Duck Lake, NY
Oxbow Lake, NY
Trout Lake, NY
No Name, PA
Rock Hill, PA
Island Pond, NY
Welhern Pond, ME
Black Pond, ME
Sunset Lake, NH
Pelham Lake, MA
Little Quittacus, MA
Sandy Pond, MA
No Name, CT
Great Pond, ME
Size
(ha)
415
111
1165
444
486
194
330
341
460
1348
1082
298
531
337
2511
TopO
Name
Saranac Lake
Caroga Lake
Lake Pleasant
Piseco Lake
Edwards
Bigelow
S. Edwards
Hermon
Promised Land
Peck's Pond
Monroe
Sloatsburg
Tim Mtn.
Stratton
Pleasant Mtn.
N. Waterford
Gilmantown
Winnepesaukee
Rowe
Heath
Assawompset
Ayer
LJtchfield
Hancock
Eastbrook
Scale
(min)
15'
7.5'
15'
7.5'
7.5'
7.5'
7.5'
7.5'
15'
7.5'
15'
7.5'
7.5'
7.5'
7.5'
7.5'
County
Franklin
Hamilton
Hamilton
St. Lawrence
Pike
Pike
Rockland
Franklin
Oxford
Bel knap
Franklin
Plymouth
Middlesex
LJtchfield
Hancock
5-63
-------
hypothesis was that all detailed wetland categories were likely to have equal differences in interpretation.
This method is probably more appropriate than using a simple contingency table to measure how well
office map units matched the field landscape (George, 1986).
5.4.1.6.3 Results and discussion -
For general land use, differences in interpretations between office and field maps were significant
(Table 5-13). The forest class (E) had the highest percentage (87 percent) of matches, whereas the least
percentages of matches were for wetlands (W), 50 percent, and for an aggregated "disturbance" class
that included waste disposal sites (L), gravel pits and quarries (P), barren land (N), horticulture (H), and
cemeteries (M), 52 percent (Table 5-14). Mismatches might have been unusually high for several general
land use classes. First, the high discrepancy in wetland matches was due to field check identification
of more very small (+_ 0.4 ha) wetland areas than were mapped by office procedures. Second, matches
for the combined/cropland pasture (C/G) class were low because both unimproved and improved pasture
were often confused with plowed land on the leaf-off imagery taken during the very wet spring. Third,
a high percentage of mismatches for the disturbed class (L/P/N/H/M) probably reflects the very Intensive
field check work done by SUNY-P staff who classified specific land use that appeared different on aerial
photos [e.g., horticulture (cranberries, H) instead of wetland (W); waste disposal land (L) instead of urban
land (U); and forest land (E) instead of barren land (N)].
For detailed wetlands, differences between field check and office determinations were also
significant (Table 5-15). The easiest subclasses to identify were emergents (EM), broad-leaved evergreen
shrubs (SS3), and open water (OW), which all had about 80 percent matches (Table 5-16). Forest
subclasses (FO1/FO2, FO4, and F05) were the most difficult to map. Office maps usually had only one
National Wetlands Inventory (NWI) subclass for each wetland delineation; SUNY-P field maps had a
greater number of dual NWI subclass modifiers for wetland delineations.
5-64
-------
Table 5-13. Cni-Square Test for General Land Use Categories
Land Use Classes9
C/G E L/P/N/H/M W
U
* See Table 5-11 for explanation of land use codes.
OW
Totals
0= Match
E=Em
(O-Ef/E
46
47.030
0.023
0= Mismatch
E=Exp.
(O-Er/E
Chi-squared,
d.f.
Table value
25.970
0.041
73
' 109
80.531
10.064
16
44.469
18.226
125
11
13.529
0.473
10
7.471
0.056
21
170
220.977
11.760
173
122.022
21.297
343
300
278.960
1.587
133
154.039
2.874
433
25
19.972
1.266
6
11.028
2.293
31
Z-fOj-E^/Ej = 70.759
= 5
= 16.748 (for p
= 0.005)
5-65
-------
Table 5-14. Comparison of Field Check (Matched)
General Land Use Determinations with Office Photointerpretations
Land Use Classes9
C/G E P/P/N/H/M W U OW
Matched 46 106 11 170 300 25
Totals 73 125 21 343 433 31
Matched
Percent 63 87 52 50 69 81
a See Table 5-11 for explanation of land use codes.
5-66
-------
Table 5-15. Chi-Square Test (or Detailed Wetland Categories
Wetland Categories8
F01/F02 F04
Totals
O= Match
E=Exp.
(0-E)2/E
O= Mismatch
E=Exp
(0-E)2/E
50 26
88.865 50.466
16.998
112
73.135
20.654
162
Chi-squared,
d.f.
Table value
=,(0| -=
11.861
66
41.533
14.412
92
7
F05 SS1
6
14.810
5.241
21
12.189
6.369
27
130.89
20.276
142
127.260
1.706
90
104.730
2.073
232
(for p - 0.005)
SS3
55
38.398
7.178
15
31.601
8.721
70
SS4/SS5
17
15.908
0.075
12
13.092
0.091
29
EM ON
102 37
74.054 252
10.545 54
33 9
60.945 20fl
12.814 6.7
135 46
8 See Table 5-11 for explanation of detailed wetland codes.
5-67
-------
Table 5-16. Comparison of Field Check (Matched) Detailed
Wetland Determinations with Office Photointerpretations
Wetland Categories8
FO1/F02 FO4 F05 SS1 SS3 SS4/SS5 EM OW
Matched 50 26 6 142 55 17 102 37
Totals 162 92 27 232 70 29 135 46
Matched
Percent 31 28 22 61 79 59 76 80
* See Table 5-11 for explanation of detailed wetland codes.
5-68
-------
Open water matches were lower than expected because SUNY-P delineations followed CIR imagery
water body boundaries shown on CIR aerial photos, whereas office map delineations tended to follow
shorelines shown on USGS topographic maps. Topographic maps ranged from 15 to more than 40 years
old and Included a large number of smaller scale, less detailed 15' base maps. A low percentage of
matches for forest and shrub subclasses was probably due to the combined effects of (1) office
photointerpreters less familiar with northeastern forest and wetland vegetative patterns and (2) the poor
quality CIR imagery used.
Based on the overall mapping accuracy observed, general land use data, but not detailed wetland
data, were digitized for all 145 DDRP NE watersheds. Two factors influenced this decision. First,
although the office maps excluded many small wetlands, they did include large wetlands that generally
coincided with field delineations; such wetlands were primarily adjacent to lakes or along major streams
that flowed into them. Small wetlands excluded on office maps were usually in remote parts of the
watershed and not adjacent to either the perimeter of lakes or major streams flowing into them. The
larger wetlands contiguous to lakes and streams are probably much more important in influencing surface
water chemistry (Johnston et al., 1984; Cooper et al., 1986; Osborne and Wiley, 1988). Second, although
office maps had delineations with one rather than two or three NWI wetland subclasses, total wetland area
on office maps agreed well with that found on field maps.
5.4.1.6.3.1 Beaver activity -
Three of the 15 field check watersheds had no beaver activity, 2 had ancient or recent dams not
mappable from photo imagery, and 10 had many examples of unbreached (U) and breached (B) dams
and in-lake lodges (L). Generally, office maps identified beaver activity in watersheds where it existed
but underestimated the total number of dams present (Table 5-17). Field work characterized extent of
beaver activity, identified bank lodges not seen on aerial photos, verified two roads mistaken for beaver
dams, and identified a large rock mapped as an in-lake lodge. Based on experiences from prior studies
(Bogucki et al., 1986), some discrepancies in beaver activity between office and field maps were probably
5-69
-------
Table 6-17. Comparison of Beaver Darn Number (#),
Breached (B) and Unbreached (U) Status, and Lodges (L),
Identified via Field Check and Office Photointerpretation Methods
Office
Lake ID/Name
lA2-004/Duck Lake, NY
lA2-039/Oxbow Lake, NY
1A2-058/Trout Lake, NY
IBS-OSS/Island Pond, NY
lCl-017/Welhern Pond, ME
102-012/Black Pond, ME
1C2-028/Sunset Lake, NH
lC3-030/Pelharn Lake, MA
1D3-033/No Name, CT
lE1-092/Great Pond, ME
Totals
#
4
4
1
1
7
2
27
9
1
20
76
B
1
0
1
1
0
0
4
1
0
4
12
U
3
4
0
0
7
2
23
8
1
16
64
L
2
1
0
0
0
1
5
1
0
5
5
#
6
15
3
0
14
2
42
22
2
42
148
Field
B
0
12
0
-
14
1
32
22
2
32
115
U
6
3
3
-
0
1
10
0
0
10
33
L
3
1
0
-
1
1
5
2
2
5
20
5-70
-------
due to (1) the five- to six-month time difference between photography and field check dates and (2)
variability in photointerpreter experience in distinguishing beaver dam activity on large-scale photography.
5.4.1.6.3.2 Map scale and imagery quality -
The CIR 1:12,000 scale imagery was ideal for mapping land use, wetlands, and water bodies.
Detailed NWI wetland subclasses were readily identifiable as were beaver dams and lodges. Compared
to 1:24,000-scale photography, however, the 1:12,000 scale was somewhat undesirable for mapping
watershed boundaries, particularly in large areas of the NE where local relief is either minimal or very
great. Also, although lots of wetland detail was seen on the 1:12,000-scale photos, considerably greater
time and money were spent on photointerpretation time and map control to prepare photo and base map
overlays. Imagery quality was generally poor (e.g., excessive shadows, variable color quality, excessive
vignetting, and considerable tonal variation) on photos for 9 of the 15 field check watersheds. Only in
a few instances, however, did imagery quality seriously affect mapping quality at the NWI subclass and
land use discrimination levels used in the Project. Limiting photo acquisition to one subcontractor and
imposing stricter quality control on "minimally acceptable* photo products would have improved image
quality for all photos.
5.4.1.6.4 Land use digitization -
Photointerpretation of 1:12,000 CIR photographs allowed characterization of land use and land cover
found across the NE. Thus, general land use data from all 145 NE watersheds were digitized via GIS
(Section 5.4.1.7). Finally, some watershed land use attributes (e.g., particularly small beaver dams and
some pasture/cropland distinctions) were only detectable by conscientious ground-truthing rather than
careful photointerpretation of large-scale conventional imagery.
5.4.1.6.5 Land use/land cover summaries -
More urban land exists in Subregion 1D, which includes the heavily built-up portions of Connecticut
and Massachusetts (Table 5-18). Subregion 1B, northeastern Pennsylvania and southeastern New York,
had the second highest amount of urban land as well as the highest amount of agricultural land. These
5-71
-------
Table 5-18. Aggregated8 Land Use Data for Northeast Watersheds
Subregions
Land Use iA 18 lE TS TE
(38) (20) (30) (24) (32)
Water (ha) 44 29 44 31 104
Urban (%) 1
Agriculture (%) -
Forest (%) 96
Wetlands (%) 3
Sum (%) 100
6
11
78
5
100
0.5
2.5
91
6
100
15
4.5
75
5
99.5
0.5
4
91
4.5
100.0
Aggregated land use classes, as described in Table 8-27. Percents are based on
terrestrial areas of watersheds.
Water =* a!l OW
urban = P + I. + Uy + U, + UC + M
agriculture = C + G + H
forest = E
wetlands = W
5-72
-------
areas comprise the Pocono and Catskill Mountains, respectively, both of which have large commercial
and private retreat camps for East Coast city residents. Valleys between rolling hills contain gentle
topography and fertile soil that is suited to agriculture. Three subregions had forest land j> 91 percent;
even Subregions 1B and 1D, with the highest urban areas, had forest percentages >. 75 percent. These
results are not surprising because NE watersheds were selected to eliminate very urbanized and/or
disturbed watershed systems.
Although the mean size of water bodies in Subregion 1E, Maine, was two or three times greater
than mean lake size in the other subregions, average percent of area in wetlands was not greater. In
all subregions, total wetlands averaged 5 percent. The range in wetland was also fairly constant, 0 to
16 percent for all but Subregion 1C, comprising Vermont/New Hampshire. Although much of Subregion
1A includes the mountainous and heavily forested Adirondack State Park, work by Bogucki et al. (1986)
showed a 34 percent increase in beaver activity between 1978 to 1985. We therefore expected mean
wetland percent for the subregion to be greater. However, in another study covering 10 watersheds In
the Adirondacks, mean wetland cover was. also low: 2 percent (Cronan et al., 1987).
5.4.1.7 Geographic Information Systems Data Entry
5.4.1.7.1 Introduction -
Upon receipt from the mapping contractors, the mapped watershed Information was entered into
a GIS. The GIS is designed to automate, manipulate, analyze, and display geographical data in digital
form, and was used in the DDRP as a spatial tool for technical analysis and for effective communication
(Campbell et al., 1989).
5.4.1,7.2 Northeast databases -
The DDRP obtained data from contract mappers and from existing information. The USDA-SCS,
in cooperation with the U.S. EPA, mapped soils, vegetation, and depth to bedrock at a scale of 1:24,000
for each watershed (Lammers et al., 1987b; Lee et al., 1989a) (see Sections 5.4.1.1 through 5.4.1.5).
Land use was mapped at the same scale by the U.S. EPA - EMSL-LV (LJegel, in review) (Section 5.4.1.6).
5-73
-------
Streams and bedrock geology were extracted from existing maps published by the USGS: streams from
7.5' or 15' topographic maps and geology from appropriate state geology maps. Contour lines for the
elevational buffers (Section 5.4.1.7.5.1) were obtained from the same topographic maps as the streams.
These data were all entered in a GIS using ARC/INFO software. Examples of GIS maps of watershed
characteristics are given for a specific watershed (1E1-062, Little Seavey Lake) in Plates 5-7 through 5-
11 (see also Plate 5-13).
Upon receipt of the SCS and EMSL-LV manuscript maps, the map overlays were prepared for GIS
entry. A minimum of four registration marks were placed on each overlay to geographically coordinate
the watershed to the surface of the earth. These registration marks also ensure manuscript-to-manuscript
registration.
Because the topographic reference maps used for the manuscript maps were produced differently
by the SCS and EMSL-LV, separate datasets were developed. The SCS primarily used diazo prints of
topographic maps, and EMSL-LV used original or photographically enlarged topographic maps. This
made the scaling of the reference maps somewhat different. In addition, the two sources were mapped
during different years; thus, the lake delineations were inherently slightly different. The same geographic
coordinates for registration within each watershed were used for both sources, however, and were drafted
onto each manuscript map. These registration marks were independently checked by another technician
before the map was digitized.
5.4.1.7.2.1 Digital entry of manuscript maps -
The manuscript maps were digitally entered into the computer using two basic steps - digitization
and attribute entry. Once entered, the digitized map is referred to as a coverage in ARC/INFO. The
digitization process enters the lines (arcs) of each layer into separate coverages. Attribute entry relates
the map classifications to a specific polygon area.
5-74
-------
Plate 5-7. Example of watershed soil map (including pedon site location).
5-75
-------
LITHE SEAVEY LAKE
SOILS
Scale 1i'24,000
• 1 Padon saapl* ait*
SOIL CLASS AND % SLOPE
ISSil Water
0326 Bray ion fin*
•and/ lo«»
n
3-8%
0-3%
3-8%
I 5-25%
8-I53C
3-15%
outcrop eonplsx 15-45%
3-8%
053A Chocorua
mucky paat
0548 Co(ton gravelly
loamy sand
054D Coltofl jr«v«lly
I oaity sand
086C Hornon sandy
loan
I06C Lyaan-Rock
out crop coeip I »*
I 158 Mario* fin*
sandy loan
I ISC Mar low fin*
sandy loan
1150 Mario* f/n«
sandy Iosn
146A Peachan Buck
MSB Psru fine
sandy loan
215C Tunbridgi
cotip I *x
226A Waskish p.a-l
8-15%
15-25%
0-3X
3-BX
8-15%
0-3X
-------
Plate 5-8. Example of watershed vegetation map.
5-76
-------
LITTLE SEAYEY LAKE
VEGETATION
Scale 1«24.000
VEGETATION CLASS
[jigfe;] Water
13 Black spruce - Tamarack
|~^\ J 16 Aspen .
19 Gray birch - Red maple
21 Eastern white pine
31 Red spruce ~ Sugar maple
- Beech
33 Red spruce - Balsam fir
lpT~p| 37 Northern white cedar
60 Beech - Sugar maple
ffi$g\ 108 Red maple
f [ 997 Open
998 Open wet
-------
Plate 5-9. Example of depth-to-bedrock map.
5-77
-------
LD
DEPTH TO BEDROCK CLASS
Water
1 < 0.5 meiers (m)
2 0.5 to 1.0 m
3 1.0 to 2.0 m
4 2.0 to 5-0 n>
5 5.0 to 30. m
LITTLE SEAVEY LAKE
DEPTH. TO BEDROCK
Seal* 1*24,000
-------
Plate 5-10. Example of watershed land use map.
5-78
-------
LITTLE SEAVEY LAKE
LAND USE
Scale 1 '24. 000
'Sf^f^y"f ij-S
LAND USE CLASS
H
Woier
Forest land
Pasture la«d
Hor tIcuIt ur a I land
Grave I pits
Wet land
-------
Plate 5-11. Example of watershed geology map.
5-79
-------
LITTLE SEAVET LAKE
GEOLOGY
Scale 1«24,000
GEOLOGY CLASS
Water
D9-D6
331 0£cr
-------
To ensure consistency in data entry among personnel over the course of the Project, a log sheet
of the steps necessary to create each of the coverages was developed. These steps were
(1) sequentially digitizing the arcs of each layer into separate coverages, (2) adding labels to each
polygon, (3) editing any necessary changes, and (4) establishing the coverage into a workable database.
Upon completion of each task, the log sheet was initialed by the operator (Figure 5-12).
5.4.1.7.2.1.1 Digitization
To ensure consistent delineation among the SCS coverages, a template coverage with only the lake
and watershed boundaries was created. This template was used as a starting base for the soil,
vegetation, and depth-to-bedrock coverages. Because of the watershed and lake boundary differences
discussed previously, the EMSL-LV land use did not use this template. The remaining polygons and
labels were then digitized into their corresponding coverages, using the specified criteria for high quality
digitizing.
5.4.1.7.2.1.2 Polygon error and digitizing quality check
After the coverages had been completely digitized, a plot of any polygon errors was made using
an internal editing function of ARC/INFO. These plots indicated polygon errors, including unclosed
polygons, unlabeled polygons, or polygons with more than one label point. Any errors found were
corrected before continuing. A new plot displaying all the digitized polygons and labels was then made
at the same scale as the manuscript maps. This plot was used as the first quality check of the digitized
arcs. The plot was overlaid with the mylar or acetate manuscript maps on the light table. If any light
appeared between the digitized line and the drafted line, the digitized line was corrected, and the polygon
error check was repeated. If there were no line errors, the attributes were written on the map next to
the polygon identification number for attribute entry.
5.4.1.7.2.1.3 Attributes entry and quality control procedures
Log sheets listing the ARC/INFO commands were created (Figure 5-13) to promote consistency
in adding attributes. The attributes for each polygon were added, in code form, into an ASCII file.
5-80
-------
yATERSHED NO.
Date
1. $ CREATE/DIR [DDELAY. ]
i.e., CREATE/DIR [DDELAY.1A1003]
2. $ SET DEF [DDELAY.
3. $ ARC
4. ARC: ADS BASEDG
Digitized tics, watershed boundary,
and all water bodies.
5. ARC: CLEAN BASEDG BASECN .25
6. ARC: EDITPLOT BASECN BASE.PLT
7. ARC: ARCEDIT
Make corrections, as needed.
Repeat steps 5 - 7, as needed.
8. ARC: COPY BASECN BASE
9. ARC: COPY BASE (SOILDG)
(VEGDG)
COPY SOIL (DEPDG)
10. ARC: ADS (SOILDG)
(VEGDG)
Digitize the remaining arcs and labels.
11. ARC: ARCEDIT
Delete arcs from SOIL to create DEPDG.
Add polygon IDs.
12. ARC: CLEAN (SOILDG) (SOILCN) .25
(VEGDG) (VEGCN)
(DEPDG) (DEPDG)
13. ARC: EDITPLOT (SOILCN) (SOIL.PLT)
(VEGCN) (VEG.PLT)
(DEPCN) (DEP.PLT)
14. ARC: ARCEDIT
Make corrections, as needed.
Repeat steps 12 - 14, as needed.
15. ARC: COPY (SOILCN) (SOIL)
(VEGCN) (VEG)
(DEPCN) (DEP)
INITIALIZE UPON
COMPLETION
BASEDG
BASECN
BASE.PLT_
BASECN
BASE
SOILDG
VEGDGj
DEPDG
SOILDG
VEGDG ~
DEPDG
SOILCN
DEPCN
SOIL.PLT
VEG.PLT_^
DEP.PLT
SOILCN
VEGCNj
DEPCN_
SOIL
DEP
Figure 5-12. Example of digitization log sheet.
R.Q1
-------
WATERSHED NO.
DATE
INITIALIZE UPON
COMPLETION
1. $ SET DBF [DDELAY.
2. $ COPY [DDELAYJBASE.DAT (SOIL.DAT)
(VEG.DAT)
(DEP.DAT)
3. $ VP (SOIL.DAT)
(VEG.DAT)
(DEP.DAT)
type in data
cntl/z
4. $ ARC
5. ARC: INFO
USER NAME> ARC
ENTER COMMAND> ADIR [DDELAY.SKELETON.INFO]
ENTER COMMAND> TAKE ARC *
ENTER COMMAND> SEL SOILTAB
ENTER COMMAND> ADD FROM [DDELAY. ]SOIL.DAT
SOIL.DAT
VEG.DAT_]
DEP.DAT
SOIL.DAT
DEP.DAT
SOILTAB
ENTER COMMAND> SEL VEGTAB
ENTER COMMAND> ADD FROM [DDELAY.
ENTER COMMAND> SEL DEPTAB
ENTER COMMAND> ADD FROM [DDELAY.
•]VEG.DAT
•]DEP.DAT
VEGTAB
DEPTAB
ENTER COMMAND> Q STOP
8. ARC: JOINITEM SOIL.PAT SOILTAB SOIL.PAT SOIL-ID
SOIL-ID
SOIL.PAT
9. ARC: JOINITEM VEG.PAT VEGTAB VEG.PAT VEG-ID VEG-ID
VEG.PAT
10. ARC: JOINITEM DEP.PAT DEPTAB DEP.DAT DEP-ID DEP-ID
DEP.PAT
11. ARC: INFO
USER NAME> ARC
ENTER COMMAND> SEL (SOIL.PAT)
(VEG.PAT)
(DEP.PAT)
ENTER COMMAND> LI
check .PAT files
ENTER COMMAND> Q STOP
SOIL.PAT
VEG.PATJ
DEP.PAT
Figure 5-13. Example of attribute entry log sheet.
-------
This list was rechecked for accuracy. The file was then merged with the corresponding coverage file
and scanned for errors. Corrections were made, and any necessary QC procedures were repeated.
5,4.1.7.2.2 Quality control plotting -
The final plots containing the arcs and attributes of each coverage for each watershed were
produced. These plots were then compared with the original manuscript maps over a light table and
checked for accuracy. If any light passed between the digitized arc and the drafted line, the arc was
corrected and the necessary QC procedures were repeated. If the arcs appeared to be correct, the
attributes were checked for accuracy. Each individual attribute was checked against the drafted map,
thereby double checking for any attributes that might have been misunderstood due to plotting resolutions
(e.g., 1's and Ts, o's and "O"s, "G"s and "C"s). This procedure was performed Independently by two
individuals. Agreement, by signature, was required before the coverage was accepted.
5.4.1.7.2.3 Projection -
Following digitization, the coverages were projected into Universal Transverse Mercator (UTM)
coordinates relating the watersheds to the surface of the earth and enabling comparison of databases.
A series of procedures was used to ensure accurate transformation from the original digitized coordinates.
First, the projected DORP coverages were interactively displayed and visually checked for consistent
location with coordinates of each lake corresponding to the NSWS (LJnthurst et al., 1986a; Landers et
al., 1988). The NSWS point locations data represent NSWS sites estimated independently from 1:24,000
topographic maps and projected into a UTM projection, and should appear near the center of the lake
for each DDRP coverage. Next, the original digitized coverages were overlaid with the projected
coverages and checked for consistency in area representation. Watershed areas were then calculated
for the original digitized coverages versus the projected coverages. Variations of greater than 5 percent
were flagged and rechecked for accuracy. Finally, to detect locations! errors or any major differences
in watershed or lake delineations, the SCS template coverage was visually compared to the EMSL-LV land
use coverage. No errors were found during this final check.
5-83
-------
5.4.1.7.3 Databases derived from existing maps -
Additional information was obtained from existing maps. The bedrock geology and streams
databases were collected directly from published USGS maps.
5.4.1.7.3.1 Bedrock geology -
State bedrock geology maps were used to generate the bedrock geology coverages. These were
the only maps available that provided a consistent geologic classification scheme within each state. The
scale of the geology maps is 10-20 times smaller than the other coverages previously described
(1:125,000 for Connecticut and Rhode island; 1:250,000 for Massachusetts, New Hampshire, New York,
Pennsylvania, and Vermont; 1:500,000 for Maine) (Billings, 1980; Doll et al.( 1961; Isachsen et al., 1970;
Miles, 1980; Osberg et al., 1985; Quinn et al., 1971; Rodgers, 1985; Zen, 1983). The SCS attempted to
map the geology, but the map legends for the various reference maps used were inconsistent, and it was
difficult to distinguish the classes into a single, usable map legend (see Section 5.4.1.4).
It was only necessary to digitize portions of the state geology maps corresponding to each
particular watershed. A state scale map containing the template coverage and watershed identification
number for all watersheds within that state was plotted. This plot was used as an overlay to the state
geology maps to focus on the DDRP watershed areas.
Following digitization, the geology coverages were clipped, or "cookie cut", with the template
coverage. This process creates a geology coverage with the same watershed delineation as the template
coverage. Each new coverage was examined individually to ensure that the geology was complete for
the entire watershed area. If it was incomplete, the missing area was added and clipped again. All water
bodies from the template coverage were then added to the geology coverage.
To check the digitized arcs against the original map, the geology coverages within each state were
plotted at the scale of the original state maps. The plots were then overlaid onto the state geology map
and independently compared by two technicians. Enlargement of the geology coverages were also plotted
5-84
-------
to Increase the visibility of all polygons. Attributes were written on the enlarged versions and checked
independently for accuracy. Attributes were added into the coverage file, checked, and plotted at a scale
of 1:24,000. The final QC procedure consisted of two independent comparisons of the new plots for
accuracy. If there were any discrepancies in the attributes, they were corrected, plotted, and, again,
checked twice independently until agreement was reached that the information was correct.
5.4.1.7.3.2 Streams-
The USGS topographic maps provided the most consistent drainage information available.
(Independent interpretation of aerial photos was attempted as a means of identifying perennial and
intermittent streams, but was not successful (see Section 5.4.1.1).) The 7.5' maps were used whenever
possible; otherwise 15' maps were used. Because the resolution of 15' maps defines very few intermittent
streams, only perennial streams were digitized for the entire stream dataset.
A log sheet was created for consistency in stream data entry. Four classes were used to
categorize the streams: perennial inflow streams, perennial inflow streams through wetlands, outlet
streams of the lake, and outlet streams of the lake draining through wetlands.
The stream coverages were plotted with the template coverage, and were overlaid with the
topographic map on a light table and checked for discrepancies. If any light passed between the
digitized arc and the stream on the map, the digitized arc was corrected. Stream classification was also
checked for accuracy. This QC procedure was performed independently by two individuals until both
agreed that the information was accurate.
5.4.1.7.4 Final quality control check and output generation -
After the mapped information was digitized and checked for accuracy, computer programs were
written in ARC/INFO to check for consistency and to calculate a usable output for the data. These
programs created lists of the classifications used and the calculated area, and also generated reports.
This information was then transferred to other computer systems for data analysis within the project.
5-85
-------
5.4.1.7.4.1 Classification -
A sorted list of all the attributes used to classify map units was generated to check for consistency
in data entry from watershed to watershed. Any unnecessary spaces or data entry errors (e.g., O's and
"O"s, or capital letters used in one watershed and lower case letters used in another) were easily
detected. Corrections were made within the coverage file.
5.4.1.7.4.2 Area -
The total watershed and lake areas were calculated and compared on a per-watershed basis for
all the polygon coverages. Soils, vegetation, depth-to-bedrock, and geology coverages were exactly the
same within a particular watershed, because these coverages used the same template coverage. Because
land use was digitized independently of the template coverage, its coverage was similar, but not identical.
If the differences were greater than 5 percent, the watershed was re-examined, and any necessary
changes were made. When a change occurred after the original QC check, the watershed was subjected
to an additional independent quality check until agreement was reached that the information was accurate.
Additional changes were recorded.
5.4.1.7.4.3 Reports -
When all the QC requirements were met, the data were used to create land area summaries, or
reports, using programming within INFO software. These reports list the classification, description of that
class, area in hectares, and percentage for each watershed (see Table 5-19 as an example). Water
bodies were not included in the percentage calculation. This output was then released for data analysis
within the Project.
5.4.1.7.5 Buffers -
Low lying areas adjacent to the lakes and streams potentially have a more direct influence on the
chemistry and response of the study lakes than upland areas. To examine this phenomenon, elevational
5-86
-------
Table 5-19. Watershed No. 1E1062 Soil Mapping Units
Class Soil Mapping Unit
Slope Hectares
Percent
ooow
032B
053A
054B
0540
086C
106C
106E
115B
115C
1150
146A
1483
21 5C
226A
Water
Brayton Fine Sandy Loam
Chocorua Mucky Peat
Cotton Gravelly Loamy Sand
Cotton Gravelly Loamy Sand
Hermon Sandy Loam
Lyman-Rock Outcrop Complex
Lyman-Rock Outcrop Complex
Marlow Fine Sandy Loam
Marlow Fine Sandy Loam
Marlow Fine Sandy Loam
Peacham Muck
Peru Fine Sandy Loam
Tunbridge-Lyman Complex
Wakish Peat
3-8
0-3
3-8
15-25
8-15
3-15
15-45
3-8
8-15
15-25
0-3
3-8
8-15
0-3
42.1
170.3
38.7
29.2
4.4
108.5
59.7
3.3
14.6
43.7
6.8
6.2
172.1
23.8
2.4
0.0
24.9
5.7
4.3
0.6
15.9
8.7
0.5
2.1
6.4
1.0
0.9
25.2
3.5
0.4
725.7
100.0
5-87
-------
buffers were digitized for study lakes and generated linear buffers around streams and wetlands to
capture the most proximal characteristics. [The wetlands information was mapped in the land use
coverage (see Section 5.4.1.6)]:
5.4.1.7.5.1 Elevational buffers -
An elevational buffer was developed to provide a topographically and hydrologically meaningful
buffer around each lake. Such buffers tend to include low lying wetlands areas and exclude sheer cliffs.
The 40-foot contour above the outlet lake and any other lake connected to the outlet lake by a perennial
stream was selected and digitized using topographic maps (Plate 5-12). The contour interval for the maps
was 6 m, 10 ft, or 20 ft. Depending on the elevation of the lake, the actual elevation change from the
lake to the digitized contour varied from 7 to 12 m (23 ft to 39 ft) on 6-m interval maps, 31 to 40 ft on
10-ft interval maps, and 21 to 40 ft on 20-ft interval maps. For example, if the elevation of the lake was
1219 ft on a 20-ft contour interval map, the digitized contour was 1240 ft, making the elevation change
only 21 ft. If the elevation of the lake was 1200 ft, the digitized contour was still 1240 ft, making the
elevation change 40 ft.
A log sheet of the steps necessary to create this coverage was designed to promote consistency
in digitizing among personnel. It also provided additional information, such as lake elevation, the digitized
contour, the contour interval of the map, and the number of islands or hills over 40 ft within the contour.
The 40-ft contour was digitized into a copy of the template coverage for consistent lake delineation and
registration.
When digitization of the contours and labels was completed, the coverage was plotted. The plot
overlaid with the topographic map on a light table. Any discrepancies were corrected, and the QC
procedure repeated. As described previously for the other coverages, two technicians independently
checked the plot for accuracy.
was
5-88
-------
Plate 5-12. Example of 40-ft contour delineations on a 15' topographic map.
5-89
-------
IW i
l«04
UNITED STATES
DEPARTMENT OF THE ARMY
CORPS OF ENGINEERS
-^ / J* f<*-\_^'» « ^^*-^ks-\ /-*^
SCALE 1:62500
1BOOO 21000 FEET
=3 1 -..-.3
5 KILOMETERS
Contour interval 20 feet
Datum is mean sea level
-------
5.4.1.7.5.2 Combination buffers -
The 40-ft contour buffers, 30-m linear stream buffers, and 30-m linear wetlands buffers were
combined to make a continuous hydroiogic buffer. The 30-m stream and wetlands buffers were
generated using ARC/INFO software {Environmental Systems Research Institute, 1986). No digitizing was
necessary to develop this coverage. Extensive editing was required because isolated buffers needed to
be deleted and areas surrounded by buffers needed special labeling so that these areas were not
included in the buffer (Plate 5-13).
The combination buffer coverages were plotted, overlaid with the topographic map on a light table,
and checked for the inclusion of pertinent buffers, exclusion of irrelevant buffers, and any special labeling.
To check the wetlands information, the plots were also overlaid with the land use manuscript map.
Discrepancies were corrected, and the QC procedure repeated. Two independent checks were made of
each plot to ensure consistency.
5.4.1.7.6 Summary -
The DDRP has a complex geographic database and uses a GIS to store, manipulate, analyze,
and display these data. Extensive QC procedures were developed to ensure the data were entered as
consistently and accurately as the original mapped information allowed. These procedures included
checking the accuracy of the information before, during, and after digitizing. Two independent checks
were performed before any data were accepted for use within the DDRP. (For more extensive details
concerning the GIS, see Mortenson (1989a).)
5.4.2 Southern Blue Ridoe Province Mapping
Mapping of soils, forest cover type and land use, depth to bedrock, geology, and watershed
drainage was initiated in the SBRP during the week of October 15, 1985. Thirteen field soil scientists
were responsible for mapping 35 watersheds, an area of about 46,730 ha (115,430 acres). Soil mapping
activities and quality assurance of the mapping data were described in depth in a report by Lammers et
al. (1987a).
5-90
-------
Plate 5-13. 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.
5-91
-------
A. Stream Buffers
B. Wetlands Buffers
C- Elevational Buffer
D. Combination Buffers
-------
The survey was implemented through interagency agreements between the EPA and the SCS. The
80S has a professional staff of trained soil scientists located throughout the DORP region, who are
capable of producing high quality mapping products over large geographical regions within a short time
frame.
Mapping protocols were developed in cooperation with soil scientists who worked in the SBRP
and were familiar with standards and procedures used in the National Cooperative Soil Survey. The
mapping phase was scheduled for October 15, 1985, to December 20, 1985, in order to accomplish
compilation and correlation tasks before the sampling phase, scheduled to begin in April 1986. A
preliminary regional soils identification legend was developed from existing soil survey legends within
major land resource areas in the region. The State Soil Scientist in each state prepared a work plan
to arrange for personnel and equipment to conduct the mapping. National high altitude aerial
photographs at a scale of 1:24,000 were ordered for most of the watersheds, to be used as field base
maps.
Prior to the start of mapping, the Mapping Task Leader, Regional Coordinator/Correlator, State
Office Soils Staff and Field Soil Scientists involved in the mapping attended a workshop in order to review
and practice the mapping protocols. The purpose of the workshop was to promote more consistent
interpretation and application of the protocols.
Start of field work was delayed in North Carolina due to other mapping commitments, and an
early fall snow storm terminated work in high elevation watersheds until spring. Field mapping continued
through the winter months, and all watersheds except one (2A07816) were completed before the soil
correlation workshop on March 3-6, 1986. Mapping was accomplished on most of the watersheds by
two-person teams, each led by an experienced soil scientist. The soil scientist responsible for the
mapping of each watershed, along with the watershed identification number, the name of the watershed,
and the state responsible for the mapping, are listed in Lammers et al. (1987a). Because wilderness
restrictions prohibited the use of any type of motorized vehicle in some watersheds, and several of the
5-92
-------
watersheds did not have roads, these watersheds could only be accessed by hiking. Access to some
small areas within watersheds was denied by private landowners, but was not a major problem during
the mapping phase of the survey. Generally, some other part of the landscape or a similar landscape
was accessible and could be investigated. The mapping was extrapolated to the inaccessible areas by
aerial photograph interpretation, other soil maps, or observation from a distance.
Map cartography and map compilation usually occurred In a field office. Some of the states
arranged for the final cartography and compilation work to be performed by one person at a central
location. Area of each map unit was estimated by a dot grid or with a planimeter. The map symbol,
map unit name, and area of each map unit in acres was listed in a legend on each watershed map.
Rough estimates of the area of map units were used during correlation and selection of classes of soils
for sampling. A more precise measurement of the area of the map units was obtained when the maps
were digitized and entered into the computerized GIS.
Aerial photographs at a scale of 1:20,000 were used for mapping one watershed (2A07826). Three
watersheds (2A07828,2A07833, and 2A07834) were mapped using 1:12,000-scale orthophotographs. Map
overlays were rectified to 1:24,000-scale film positives of orthophotographs on scale stable film after
mapping was completed. Film positives of 7.5', 1:24,000-scale topographic quadrangle maps were used
for the rectified base where orthophotographs were not available.
5.4,2.1 Soils
Soils were mapped using the same standards and procedures described in Section 5.4.1.1 and
the mapping protocols in Appendix A of the report by Lammers et al. (I987a). Soil scientists made soil
maps on mylar overlays of base maps at a scale of 1:24,000.
5.4.2.1.1 Soil correlation -
The soil correlation process was described in Section 5.4.1.1.1 of this report.
5-93
-------
A preliminary identification legend was developed based on soil map units that had been mapped
within the SBRP. There were 210 map units listed in the preliminary legend, and about 90 map units
were added during the field mapping. From the total of 300 map units, about 200 were actually used
in the mapping. During the week of March 3-6, 1986, soil scientists representing the SCS from all of the
states involved in mapping the DDRP SBRP Region met in Corvallis, OR, with the RCC and task leaders
from ERL-C and ORNL The objectives of the meeting were to correlate the soils and soil map units for
the region and to complete descriptions of the map units. Each one of the 200 map units used during
the mapping was reviewed. The characteristics and taxonomic classification of the major components
of each map unit were checked and completed.
A few map units mapped in more than one state were found to be similar and were combined.
Other map units were represented by just a few hectares and were combined with the most similar map
unit in the legend.
Fifty of the map units were randomly selected for collection of transect data. Four available
transects were randomly selected to represent each of the 50 map units. A transect consisted of
examining and documenting the kind of soil or miscellaneous area at 10 points at equally spaced intervals
across the map unit delineation. •
Map unit descriptions were again reviewed during an Exit Meeting held at Park City, UT, on July
15-17, 1986. Transect data, available at that time, were examined and used for making adjustments to
map unit composition and for correlation. When transect data did not appear to accurately represent the
map unit, soil scientists with experience in mapping that unit were asked to make a "best estimate" of
the composition. Most often, the alterations were based on the kinds and percentages of minor soil
components in the map unit. After the area of each map unit was more precisely determined from the
digitized data in the GIS, additional map units with only a few acres were combined with other similar
map units by the Mapping Task Leader. This resulted in a final soil map legend of 176 map units. A
5-94
-------
few small map units remained in the legend, if there were no similar map units with which they could be
combined.
The soil taxonomic class, drainage class, depth to bedrock, and estimated depth to a slowly
permeable or impermeable layer were compared to the official soil series for the major components of
each map unit.
5,4,2.1,2 Soils database -
The mapping phase of the DORP SBRP Soil Survey generated vast amounts of data. In order to
verify, validate, and analyze these data, they were entered into computer database files. Data products
generated by the mapping included the identification legend, descriptions of the soil map units,
descriptions of the soil taxonomic units (components of the map units), soil transect information, and the
map products. The map products included maps of the soils, land use/vegetation, depth to bedrock,
geology, and drainage of the 35 watersheds. This section describes the database files developed for the
DDRP mapping data and the procedures and QA/QC checks used during the computerization of the
DDRP data. Both ORNL and EPA's ERL-C were involved with management of the mapping data. Most
of the data were entered at ERL-C using a Mapping Data Management program with dBase III plus
software. Correlation corrections to the mapping data were also entered at ORNL using SAS. ORNL
performed most of the data comparisons to identify discrepancies. ERL-C had overall responsibility for
the quality of the data and validation of data. The maps were digitized for input to a G1S at ERL-C as
described in Section 5.4.2.6. An overview of mapping databases is presented by Turner et al. (in review).
5.4.2.1.2.1 Soil identification legend -
A preliminary soil identification legend was developed from existing soil survey legends within the
major land resource area of the Southern Blue Ridge Mountains. Map symbols were assigned to map
units in the legend by the RCC. Corrections and additions were approved by the State Soil Scientist and
RCC and then entered into the identification legend (SE_MP_UN) file at ERL-C. Map unit symbols used
in the mapping as shown on individual watershed maps were entered into a dBASE III file during a
5-95
-------
regional correlation workshop. Map units not used were marked for deletion, and map units that were
combined due to small extent or similar soils were noted on the legend. Legends from each watershed
map were also entered into the GIS as the maps were digitized at the ERL-C.
The legend data from the GIS were transferred to a dBASE 111 file where they were summarized
for the region and then compared to the regional soil identification legend. Discrepancies were then
resolved, and the map unit names were checked with the descriptions of the map units for validity. The
soil identification legend database file for the DDRP SBRP Soil Survey named SE_MP_UN contained the
following information for each map unit: map symbol; map unit name, including the name of dominant
soil component(s), texture modifier (e.g., gravelly, mucky), texture phase, slope phase, and other phase
(e.g., very stony, rocky); regional landform; local landform; geomorphic position; slope shape across;
slope shape down; and area in acres (determined from the GIS database). This file was accessed
through the Southern Blue Ridge Mapping Database Management (SEDBMNT) program developed at ERL-
C, as demonstrated by Lammers et al. (1987C). The initials of the person making changes to the legend,
the date the changes were made, and records marked for deletion were automatically recorded.
5.4.2.1.2.2 Soil map units and soil taxonomic units ~
In some map units, the minor components (inclusions) collectively made up more than 30 percent
of the map unit and were found to be important for project analyses. Also, a major soil component in
a consociation may have had the same attributes as a major component in a complex or minor
component in another map unit. The information from the map unit worksheet was, therefore, separated
into two files, a map unit composition file and a soil components file. Each unique soil component was
assigned a component code to aid in accessing all the attributes of a soil component with one code.
The map unit composition file named SE_MP_CM contains the map symbol, the component code for
every component in the map unit, and the percent composition of each of the components.
The soil components file was named SECMPNT. Each record in the SECMPNT file includes the
component code; soil name, texture, and slope of the component; five characteristics of the soil:
5-96
-------
permeability, drainage, depth to bedrock, origin, and mode of deposition of the parent material; and the
taxonomic class. The sampling class code for the class with which the soil component was grouped for
sampling was also included in the record for each component in this database file. The records from
the three database files, SE_MP_UN, SE_MP_CM, and SECMPNT, were merged In the SEDBMNT, to
display information about each map unit on a soil map unit worksheet. This worksheet included the map
symbol, map unit name, information about the landscape, major soil components, minor soil components,
the proportion of each component in the map unit, and information about the major components including
the taxonomic classification.
Copies of these computer-generated worksheets were reviewed by soil scientists in each of the
Southern Blue Ridge states. Corrections to these map unit data sheets were reviewed at the Exit Meeting
in Park City, UT, in July 1986. Data from these corrected map unit worksheets were entered into the SAS
files at ORNL and corrections entered into the SEDBMNT files at ERL-C. The two databases were
compared to identify discrepancies. This method of correcting the database virtually eliminated the
possibility of updating the wrong observation or variable.
After the updates were completed, ORNL generated frequency tables of the coded variables and
compared these tables with lists of valid codes. The frequency tables were also used to build code
translation tables containing the codes and definitions. These translation tables were stored as SAS
format libraries and are a part of the database. The final step in editing the map data files involved
labeling variables and, where necessary, modifying variable names and labels to ensure consistency
among the various mapping data files.
5.4.2.2 Depth to Bedrock
Depth-to-bedrock maps were made during the course of soil mapping on mylar overlays of base
maps at a scale of 1:24,000. Bedrock was in some cases weathered bedrock designated as a Cr horizon
in the soil description. Each delineation on the soil map was assigned to one of the six depth classes.
Complexes in which depth of soil over bedrock spans two depth classes were described with dual
5-97
-------
classes, with the dominant depth to bedrock listed first. Such dual classes were used where the soils
were in adjacent depth classes, e.g., classes II and III, III and IV. From this information, a depth-to-
bedrock map on a mylar overlay was made by combining contiguous delineations of the same class.
Depth to bedrock was estimated from all available information, including soils data, road cuts, and stream
incisements.
5.4.2.3 Forest Cover Type/Land use
Forest cover type and land use were mapped together on overlays of base maps at a scale of
1:24,000. The forest cover types mapped were those published by the Society of American Foresters
and described by Eyre (1980). Open areas not having a forest cover type were designated by the land
use. Land use was designated by one of the land use codes used by the SCS for soil site description,
as listed in Table 5-20.
Forest cover and land use were delineated from interpretation of aerial photograph imagery and
from observation of topographic and landscape features. Delineations were confirmed by field observation
during the course of soil mapping.
5.4.2.4 Bedrock Geology
Bedrock geology maps were made on a mylar overlay of USGS topographic quadrangle or National
High Altitude Photography (NHAP) base maps at a scale of 1:24,000. Bedrock geology was obtained
from current geology maps. Field crews noted any obvious departures from mapped bedrock geology
(Section 5.4.2.8.3.1).
5.4.2.5 Drainage
Watershed drainage was drafted on a mylar overlay of USGS topographic or NHAP base maps
at a scale of 1:24,000. Drainages not shown on topographic maps were added to the drainage overlay
(Section 5.4.2.8.3.2).
5-98
-------
Table 5-20 Land Use Codes Used as Map Symbols
C cropland
E forest land, grazed
G pasture land
L waste disposal land
P rangeland, grazed
R wetlands
T tundra
I cropland, irrigated
F forest land, not grazed
H horticultural land
N barren land
S rangeland, not grazed
Q wetlands, drained
U urban and built-up land
5-99
-------
5.4.2.6 Quality Assurance
A rigorous plan for QA/QC, similar to that used in the NE mapping, was implemented during all
phases of mapping activities. QA/QC activities included field reviews, independent evaluation of the
mapping, and random transecting of selected soil map units in the watersheds. The approach used in
the transecting was different than that used in the NE. The transect data were evaluated to determine
the correctness of the soil map units.
5.4.2.6.1 Field reviews by the Soil Conservation Service -
Field reviews were conducted by the SCS State Soil Scientist, or another member of the SCS State
Soils Staff, for each soil mapping crew in their respective states. There were 11 different soil mapping
crew:; responsible for the mapping, and field reviews were conducted on watersheds mapped by all 11
crews.
The purpose and conduct of the field reviews were the same as described for the NE Region in
Section 5.4.1.5.1 of this report, and the same aspects of the mapping were evaluated. Field reviews
were conducted at various stages of mapping; on some the mapping had been completed, and on others
the mapping was just beginning. A written progress review report was submitted to the RCC and
Mapping Task Leader following each field review. Field review reports from Georgia, North Carolina, and
Tennessee are in Appendix D of the QA/QC report on soil mapping activities by Lammers et al. (1987a).
The field review reports documented about 45 problems or mapping errors on 32 watersheds.
Problems included understanding of mapping protocols, identification of available transects, and
correlation with the regional legend. Mapping errors included use of the wrong map symbol,
inappropriate map unit, poor location of the map delineation, mapping of small areas of less than
2.7 ha (6 acres), incorrect depth-to-bedrock class, and incorrect vegetation cover type. Most of the
problems were resolved during the review and mapping errors were corrected, or the responsible soil
scientists agreed to make corrections or conduct additional investigations to resolve the discrepancies.
5-100
-------
Mapping was judged acceptable for the purposes of the survey on all watersheds for which a review was
conducted.
5.4.2.6.2 Field reviews by the Regional Coordinator/Correlator -
The RCC was required to participate in the field review of at least one watershed in each of the
three states responsible for the mapping. Eight watersheds • three in Georgia, three in North Carolina,
and two mapped by Tennessee - were reviewed by the RCC. The purpose of the RCC participation was
to coordinate the mapping throughout the region and to control the quality of the mapping. The RCC
facilitated better communication among states to effect consistency and improve correlation of the soils
and map units.
A narrative report on the findings and the discussion resulting from the field reviews for each of
the states was submitted to the Mapping Task Leader at ERL-C. After discrepancies were corrected,
the mapping was judged acceptable on all of the watersheds. Field review reports by the RCC are in
Appendix E of the report by Lammers et a). (I987a).
5.4.2.6.3 Evaluation of mapping by the Regional Coordinator/Correlator -
The RCC evaluated the mapping on 7 of the 35 watersheds in the SBRP Soil Survey. These 7
watersheds were selected from the top of a random list of all 35 watersheds with the constraint that no
more than one watershed would be evaluated by the RCC for each mapping team.
Mapping was evaluated by examining stereoscopic pairs of aerial photographs. Relationships
between soils and landform segments were scrutinized and questionable areas marked for further
examination on the ground by traversing and transecting. About one-third of the soil map delineations
were traversed on the ground and the soils in five map units on each watershed were documented at
10 points along a transect. A report of the results of the mapping evaluation was submitted to the
Mapping Task Leader at ERL-C. Mapping was judged "acceptable" on ail 7 watersheds that were
5-101
-------
evaluated. Summaries of the mapping evaluation by the RCC are reported by Lammers et al. (1987a,
Appendix E).
5.4.2.6.4 Evaluation of soil transect data -
Observations were made at 10 equally spaced stops along transects across selected map unit
delineations in each watershed. Of the 176 map units, 50 were selected for transecting by soil mappers.
Four transects were randomly selected from the list of available transects for all of the 50 map units
except for 5 for which there were only 3 available transects. The RCC conducted five 10-point transects
in 7 of the 35 watersheds. At each transect stop, the soil name and a few important differentiating
characteristics were recorded.
5.4.2.6.4.1 Management of the transect data -
The watershed number, transect number, map symbol, soil name, slope, and notes at each transect
stop were entered into a database file at ERL-C and into a SAS file at ORNL The two files were
compared and discrepancies resolved. Due to correlation of soils and soil map units, soil names and
map unit symbols were different on the final regional soils legend than on the transect data forms. The
transect soil name and map symbol entries were corrected to agree with the final correlation.
5.4.2.6.4.2 Analyses of the transect data -
The transect data were used to evaluate the correctness of the described map units. The
correctness of the map units was evaluated by comparing the proportions of soils transected with the
expected proportions in the map unit composition (SE_MP_CM) database. Routine transect results were
compared to RCC transects for watersheds mapped by both. Finally, soil components observed at
transect points were assigned the proper sampling class, and map unit correctness was evaluated with
respect to the sampling class composition. The latter evaluation was especially relevant to judging the
correctness of map units for the purposes of DORP. All hypothesis tests were conducted using two-
sided alternative hypotheses.
5-102
-------
5.4.2.6.4.2.1 Analysis of major components in map units with routine transects
The soils that were major (named) components of each map unit were pulled from the map unit
file (8E_MPJJN). For the major components of each map unit, the transect points were treated as
observations from a binomial distribution with the population proportion p, where p was calculated from
the SE_MP_CM data file. The sample proportion used was the total number of transect points containing
a major component, divided by the total number of transect points. This analysis was similar to the
analysis described in Section 5.4.1.5.4.1 and subject to the same conditions.
The proportion of major components was significantly different than the estimated proportion in the
SE_MP_CM file for 25 of the 188 transects used in the analysis. Corrections made during correlation
accounted for differences in 13 of these transects. At the 0.01 level of significance, we would expect
about 10 transects to have a significantly different proportion.
These significantly different transects were excluded from the dataset, and the variability in
proportion of major components between transects was calculated, to determine if some soils were
uniformly different from the expected proportions or just highly variable. The proportion of major
components was calculated for each transect that was not significantly different. The variance of these
proportions was then calculated for each map unit with two or more transects. Since the distribution of
these variances was asymmetric, a robust data analysis technique was used.
A boxplot (also called a box-and-whiskers plot; Velleman and Hoaglin, 1981) of the variances of
proportions was drawn. Boxplots use the interquartile range (IQR) (i.e., the distance between the 75th
and 25th percentiles). Points more than 1.5 times the IQR away from the median are considered outliers,
and points more than 3 times the IQR away from the median are considered strong outliers. For this
dataset, there were no map units with variance of proportion outside the 1.5 IQR and only seven above
the 75th percentile.
5-103
-------
5.4.2.6.4.2.2 Analysis of major components with RCC transects
RCC transects were examined, at the 0.01 level of significance, for the proportion of major map
unit components significantly different from the proportion estimated in the SE_MP_CM data file. There
were 4 transects out of 31 that were found to have significantly different proportions. Two of the
transects had names of soils recorded that were similar to the major components, which most likely
results from difference in soil scientists rather than a difference in map unit composition.
These comparisons suggest that the quality of the mapping evaluated by the routine transects
was about the same as when evaluated from transects performed by the RCC. About 7 percent of the
transects were as significantly different at a significance level of 0.01.
5.4.2.6.4.2.3 Comparison of the routine and RCC transects
When routine and RCC transects were compared, as in Section 5.4.1.5.4.3, five map units were
found to have significantly different proportions. Three of the map units were due to significantly different
RCC transects, and the other two map units were situations in which the routine transect proportion was
slightly above the expected proportion and the RCC transect proportion was slightly below it. Neither
the RCC proportion nor the routine proportion was significantly different from the expected proportion,
but in both cases, they were far enough apart to be observed. This strong agreement suggested that
the routine transecting and transecting done by the RCC were comparable.
5.4.2.6.4.2.4 Analysis of the map units by sampling class
Because soils were grouped into 12 classes for sampling and analytical characterization, the
correctness of mapping these classes was important to evaluate the mapping for project assessment.
The soil at each transect stop was assigned the appropriate sampling class (described in Section
5.5.1.3.2) to compare transect-determined sampling class composition with map unit description sampling
class composition. Because there were several hypothesis tests for each map unit, the Bonferroni
inequality was used to handle the error rate of the simultaneous hypothesis tests within each map unit.
5-104
-------
There were nine map units in which the sampling class proportion differed significantly from the
proportion estimated by soil scientists. Six of the nine map units had an estimated sampling class
proportion of 100 percent. Five of these units had an actual difference of less than 7 percent, due to
the estimated proportion being 100 percent. All five were actually quite close to the estimated proportion.
At the .05 significance level we would expect about five significant map units, if the null hypothesis were
true throughout and no estimated proportions were equal to 100 percent. This observation suggests that
the estimated sampling class proportions are accurate for most map units.
5.4.2.6.4.2.5 Summary of the transect analyses
For the most part, the transect analysis indicated that the mapping was good. There were a few
map unit delineations that may have been mismapped. On the other hand, the differences between
transect data and estimated map unit composition could have been due to unrepresentative transects,
incorrect estimated proportions, or soils that were difficult to map. No map units had unusually high
variability that may have indicated problems with the transecting or the map unit definitions. After the
significant transects were removed, the map units were reasonably consistent, suggesting that some of
the problem may have been the estimated proportions. The sampling classes matched the estimated
proportions very well, indicating that there were few problems in the mapping as far with regard to the
needs of DORP.
5.4.2.7 Land Use/Wetlands
5.4.2.7.1 Data acquisition -
Information on land use and wetlands was obtained via interpretation of existing, but older, NHAP
photography (1:24,000) and field observations during soil mapping (Section 5.4.2.3). Current photography
was not required nor were specialized photointerpretation/field checking activities performed as they were
for the NE watersheds (Section 5.4.1.6). General land use, forest cover type, and wetland data for all
SBBP watersheds were digitized via GJS (Section 5.4.2.8).
5-105
-------
5.4.2.7.2 Land use/land cover summary -
The predominant land use in all but one watershed was ungrazed forest; in the one exception
(2A07826), the predominant land use was horticulture (Table 5-21). Hardwood forests, ranging from 50
to 93 percent, predominated over mixed and coniferous forest. Coniferous vegetation (42 to 44 percent)
was predominant only in two watersheds. Wetlands were absent in all but one watershed (2A07802).
Urban development was absent or minimal in alt but one watershed and ranged from 1 to 6
percent; watershed 2A07802 had 19 percent of its area in unban land use. Agricultural development was
also limited. Little area was devoted to cropland in any watershed except 2A07826. However, all but
10 watersheds had managed or unimproved native pasture and 10 watersheds had pasture percentages
.>. 10 percent.
5.4.2.7.3 Regional comparisons -
Although methods for determining land use and wetlands were different for the NE and SBRP
Regions, certain generalizations are possible. Similarities between NE lake and SBRP stream watersheds
are the predominance of forest land use and little agricultural or urban development - both are results
of the overall DORP field design to work with "undeveloped" watersheds. The greatest dissimilarity is the
overall lack of wetlands, beaver activity, and (lowland) horticulture in the SBRP region, reflecting large
differences in physiography and landform features of the two regions.
5.4.2.8 Geographic Information Systems Data Entry
5.4.2.8.1 Southern Blue Ridge Province databases -
The DDRP obtained data from contract mappers and from existing information. The SCS, in
cooperation with the EPA, mapped soils, vegetation/land use, depth to bedrock, and streams at a scale
of 1:24,000 for each watershed (Lammers et al., 1987a; Lee et al., 1989a) (see Sections 5.4.2.1 through
5.4.2.6). Bedrock geology and additional stream information were extracted from existing maps published
by the USGS, geology from appropriate state geology maps, and streams from 7.5' or 15' topographic
maps. These data have been entered into the GIS.
5-106
-------
Table 5-21. Percent Land Use Data for Southern Blue Ridge Province Watersheds
D
lo.
01
02
03
13
16
35
01
02
03
04
05
06
08
10
11
01
04
06
03
02
05
06
11
12
17
21
23
26
27
28
29
30
33
34
82
St.
TN
TN
TN
TN
TN
SC
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
C
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.2
0.0
0.0
0.0
0.0
0.0
0.0
1.8
3.8
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.0
11
0.9
1.0
0.0
11
0.7
0.0
0.0
E
0.0
0.0
18
0.0
0.0
0.0
0.0
3.5
2.6
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
3.2
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.8
0.0
0.0
F
100
91
63
74
99
94
96
86
91
100
84
80
95
100
100
94
89
85
94
63
95
88
99
100
98
87
95
34
91
97
97
75
87
98
94
G
0.0
9
16
24
0.0
0.8
4.3
10
5
0
16
20
2
0
0
4
3
13
6
.15
0
9
0
0
2
12
5
8
7
2
0.0
13
11
2
0.0
H
0.0
0.0
0.0
1.7
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.6
0.0
42
0.9
0.0
1.3
0.4
0.5
0.0
0.0
K
0
0
0
0
0
4
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
0
0
L
0
0
0
0
0
0
0
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
M
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
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
N
0
0
0
0
0
0
0
0
0
0
0
0
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
o
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
0
1
0
0
0
0
0
0
0
0
0
0
0
0
R
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
u '
0.0
0.3
2.7
0.0
0.0
1.6
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.9
4.0
0.6
0.0
19
0.0
0.0
0.0
0.0
0.0
0.0
0.0
3.6
0.0
0.0
0.0
0.1
0.0
0.0
6.1
z
0
0
0
0
.4
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
»: For explanation of land use symbols, consult Table 8-37.
5-107
-------
Soils, vegetation/land use, depth to bedrock, and streams were mapped in the field by the SCS
between fall 1985 and spring 1986. The SCS used 7.5' USGS orthophoto film positives or topographic
film positives as their reference maps. These field maps were transferred onto mylar overlays following
DDRP specifications (Lammers et al., 1987a; Lee et al., I989a). These overlays are the final manuscript
maps and were entered into the GIS.
When the SCS streams were overlaid with the topdogbal maps, 14 of the watersheds were found
to have been incorrectly transferred to the reference map. The field mapped information drawn on the
area! photographs were traced directly onto the mylar without first adjusting, or "rectifying," the information
to the reference map. Because the SCS was unavailable to make the corrections, these corrections were
contracted to Oregon State University, Department of Geography. The overlays were hand transferred
to the reference map using topographic features. For example, ridge top soils were placed on ridge tops,
valley bottom soils were placed in valley bottoms. Because the soils overlay directly relates to the depth-
to-bedrock overlay, the depth-to-bedrock overlay was adjusted by using the soils overlay as a mapping
guide. Vegetation/land use information used features found on topographic maps and orthophotos such
as open areas, vegetation changes, and riparian zones along with proximity to watershed boundaries and
soil classes as guides.
Because the 14 unrect'rfied watersheds were not identified until after the original database was
completed, the corrections for these 14 watersheds were updated into the original database, making
their entry procedure slightly different.
The original database followed the same digitizing procedures as the NE database. Because this
information was completed and projected into Universal Transverse Mercator coordinates, the 14
unrectified watersheds were digitized directly into the same projection.
5-108
-------
5.4.2.8.2 Database preparation and digital entry -
The SBRP databases were prepared and digitized similarly to the NE databases (see Sections
5.4.1.7.2.1 through 5.4.1.7.2.2), with two minor exceptions. First, as mentioned in the previous section,
the remapped watershed information was digitized directly into the UTM projection. Secondly, the SCS
combined the land use with the vegetation overlay. Codes depicting land use were entered as an
attribute to the vegetation coverage, rather than as a separate coverage as in the NE.
5.4.2.8.2.1 Projection -
The same series of procedures was used in the SBRP as the NE to ensure accurate transformation
from the original digitized coordinates to the UTM coordinates (see Section 5.4.1.7.2.3). Once the entire
database was in UTM coordinates, the DDRP coverages were interactively displayed and visually checked
for consistent location with coordinates of the downstream sampling node of each stream corresponding
to the NSS (Messer et a!., 1986a).
5.4.2.8.3 Databases derived from existing maps -
5.4.2.8.3.1 Bedrock geology -
As in the NE database, state bedrock geology maps were used to generate the bedrock geology
coverages. These were the only maps available that provided a consistent geologic classification scheme
within each state (see Section 5.4.1.7.3.1). The scale of the geology maps is 10-20 times smaller than
the other coverages previously described (1:125,000 for South Carolina and Tennessee; 1:500,000 for
Georgia and North Carolina) (Brown, 1985; Hardeman, 1966; Overstreet and Bell, 1965; Pickering and
Murray, 1976). The same procedures were used to enter the geology information as in the NE (see
Section 5,4.1.7.3.1).
5-109
-------
5.4.2.8.3.2 Streams -
The streams manuscript maps provided by the SCS were used for the original database. In order
to correct the streams for the unrectified watersheds, the streams and topological features from the USGS
topographic maps were used as a guide. A log sheet was created for consistency in stream data entry.
The stream coverages were plotted with the template coverage. These plots were overlaid with
the topographic map on a light table and checked for discrepancies, if any light was evident between
the digitized line and the delineated stream, the digitized line was corrected. Stream classification was
also checked for accuracy. This QC procedure was performed independently by two individuals until both
agreed that the information was accurate.
5.4.2.8.4 Final quality control check and output generation -
After the mapped information had been digitized and checked for accuracy, computer programs
similar to those used in the NE were written in ARC/INFO to check for consistency and to calculate a
usable output for the data. These programs created lists of the classifications used and the calculated
area, and also generated reports (see Sections 5.4.1.7.4.1 through 5.4.1.7.4.2). This information was then
transferred to other computer systems for data analysis within the DDRP.
5.4.2.8.5 Summary -
Much of the SBRP database was developed similarly to the NE database. Unlike the NE, the SBRP
received mapped information from one contractor rather than two. The SBRP did, however, follow the
same QC procedures as the NE to ensure the data were entered as consistently and accurately as the
original mapped information allowed. These procedures include checking the accuracy of the information
before, during, and after digitizing. Two independent checks were performed at critical stages of the
digitization before the data were accepted for use within the DDRP. (For more extensive details
concerning the GIS, see Mortenson (1989b).)
5-110
-------
5.5 SOIL SAMPUNG PROCEDURES AND DATABASES
Soils were described and sampled to provide the morphological, physical, and chemical data
needed for the three DORP levels of analysis. In the NE, 306 pedons were described and 2000 samples
(i.e., about six horizons per pedon) were taken. (A pedon is the smallest volume of soil that has all the
characteristics by which a specific soil is defined. Operationally, It is usually taken to be about a meter
square in cross section to depth of 1.5 m or to bedrock, whichever is shallower.) The corresponding
numbers for the SBRP were 110 pedons and 1000 samples. Soil survey activities have been described
in some detail by Lee et al. (1989a). Much of this section draws from that report and from the detailed
description of sampling class development in the NE by Lee et al. (1989a).
5-5.1 Development/Description of Sampling Classes
5.5.1.1 Rationale/Need for Sampling Classes
In the NE, about 600 soils (mostly phases of soil series) were identified during mapping of 145
watersheds. In the SBRP, about 300 soils were identified on 35 watersheds. Because of the large number,
it was impractical to sample each soil enough times to obtain statistically adequate estimates of the
means and variances of the relevant soil properties. As a practical alternative, the soils identified during
mapping were combined into a tractable number of groups, or sampling classes, that were either known
or expected to have similar chemical and physical characteristics with respect to their responses to acidic
deposition. The development and characteristics of these classes .have been described by Lee et al.
(1989b,c) and Lammers et al. (in review).
Each of these sampling classes was sampled across several watersheds, so that the mean and
variance of the characteristics of each sampling class could be computed for the region. These regional
means and variances are then used in conjunction with the soil maps to build area or volume weighted
estimates, with error estimates, of the characteristics of each watershed. This same approach can be used
for specific portions of watersheds, such as poorly drained soils near lakes. When using this approach,
however, a given soil sample does not represent the specific watershed from which it was sampled.
Instead, it contributes to a set of samples that, collectively, represent a specific sampling class on all
5-111
-------
DDRP watersheds within the region for which the sampling class is defined. Because the DORP is
designed to estimate the uncertainties of its projections and conclusions, it is necessary to know the
probable range of expression of a given characteristic for a sampling class within a region, and not just
the value associated with the central concept of the class.
The soil sampling classes were used for statistical stratification for sampling, and for aggregation
of data for analysis. Stratification and aggregation were necessary to obtain soils information on a very
extensive area from sampling of only a limited number of pedons. If, for example, we had used a purely
random scheme had been selected with each pedon representing 40 ha (larger than some DDRP
watersheds), sampling would have been required on about 1700 pedons in the NE. By using soil
mapping to determine the kinds of soils and their spatial distribution on the DDRP watersheds. Project
objectives can be satisfied with approximately 300 sampled pedons in the NE.
5.5.1.2 Approach Used for Sampling Class Development
Sampling classes were developed at workshops (Lee et al., 1989b,c; Lammers et al., in review) by
the field soil scientists responsible for soil mapping, in cooperation with the modelers and statisticians
who would be using the data. Soils were split into different classes based on characteristics the soil
scientists thought might be important for determining the responses of watersheds to acidic deposition.
Characteristics considered included mineralogy, iron and aluminum oxides, organic matter content,
texture, versus oxidizing chemistry, cation exchange capacity, base saturation, drainage (wetness), depth,
hydraulic conductivity, role as a source area for surface waters. The schemes were tailored to each
region to best distinguish these characteristics in the field.
Adequate resolution for the modelling and analysis tasks within the Project. The underlying
rationale was that if the classes are in fact distinct, better resolution is attained by separating them. If,
however, they turn out not to be distinct, then we have paid a small price in terms of precision; that is,
the allocation of samples is not as efficient as ft might have been. It is possible (and, in fact, expected)
that soils were split into finer groups than were needed.
5-112
-------
5.5.1.3 Description of Sampling Classes
5.5.1.3.1 Northeast -
The flowchart defining sampling classes in the NE is shown in Figure 5-14. Spodosols were
separated because of the accumulation of aluminum oxides, iron oxides, and organic matter in spodic
horizons. These would affect cation exchange capacity (CEC) and sulfate adsorption, two important
processes that influence ANC (Section 3; Altshuiler and Linthurst, 1984; NAS, 1984). Aluminum is also of
interest as the toxin primarily responsible for the adverse effects of acidification on aquatic organisms
(Altshuiler and Unthurst, 1984).
The primary split on mode of deposition of parent material was glacial till versus glaciofluvial,
implying differences in the degree of sorting of parent material. This distinction was made because
particle-size distribution correlates with many properties of interest to DDRP, such as CEC and hydraulic
conductivity.
The wettest soils (e.g., Aquods, Aquepts, aquic subgroups, non-folist Histosols) were separated
because of their likely role as source areas for surface waters. They are also likely to differ from other
soils in having a reducing rather than oxidizing chemical environment, which is especially Important for
sulfur retention. (Histosols were different from the other soils in most properties of interest.) Because
approximately equal numbers of pedons were sampled for all sampling classes, separating the wettest
soils resulted in more sampling of those soils in closest proximity to the surface waters.
The use of drainage classes to define sampling classes was considered by workshop participants.
The consensus was to use aquic vs. non-aquic instead because these taxonomic terms are better defined
and used more consistently by soil scientists than are the somewhat subjective drainage classes. The
aquic vs. non-aquic split was made at the suborder (e.g., Aquepts vs. Ochrepts) and subgroup (e.g.,
Typic Dystrochrepts vs. Aquic Dystrochrepts) levels.
5-113
-------
CO
S
o
(O
Q.
oc
Q
O
I
S
S
s
jo
u
f
Q.
i
s
o
E
i
D
£
I
5-114
-------
One property used for defining groups was soil depth. This split reflects the ILWAS hypothesis
(Gherini et al., 1985; Newton and April, 1982) that soil depth may be the most relevant soil property in
the NE. For example, soils in groups S15, S17, and S18 are all non-aquic Orthods formed on similar
parent materials, and are likely to have very similar chemical and physical properties, especially in the
upper horizons. The distinction among them is soil depth: deep, moderately deep, and shallow,
respectively. Aquic soils were not split by depth because the consensus was that these soils were
hydrologically similar in that most water flow would be through the upper horizons; also, almost all of the
aquic soils were deep.
As another example two classes I37 and 12, both contain wet, non-acidic Inceptisols formed from
similar parent materials. They differ in family particle size: sandy vs. coarse-loamy. In other words, soils
in 137 contain greater than 50 percent sand, and those in 12 less than 50 percent sand in the particle-
size control section. If these soils range far from the 50 percent breakpoint, then these classes are likely
to differ in properties of interest to DDRP. If, however, they cluster near the 50 percent breakpoint, the
classes might not be separable; more importantly, there would be no reason to separate them. The fatter
case is an example of the conservative approach to defining classes.
5.5.1.3.2 Southern Blue Ridge Province -
The flowchart defining sampling classes in the SBRP is shown in Figure 5-15. The frigid soils occur
only at the highest elevations in the region. They were separated because they have soil organic matter
contents greater than average for the region, and might differ chemically because of the effects of
temperature and vegetation on pedogenic processes. The calcareous soils occur only as inclusions in
the SBRP. They were separated because their calcareous nature might have an effect on surface waters
disproportionate to their small area of occurrence. Skeletal soils were separated because of the short
residence time and limited amount of soil fines available for reaction with precipitation. The concave
skeletal soils are the main conduits for waterflow, and represent the most probable path for the majority
of water delivered to the streams. The convex skeletal soils occur at the upper extreme of watershed
5-115
-------
I
&
SHALLC
[SHL
v /
O
r
SOILS OF THE SOUTHERN BLUE RIDGE PROVINCE |
11
. , t
FRIGID
(FR j
NON-SK
NON-FRIGID
1
NON-CALCAREOUS CALCAREOUS
IOTCI
ELETAL SKELETAL
CONCAVE CONVEX
fsKVj (SKXJ
FLOODED NON-FLOODED
[FL )
LOW
ORGANIC
MATTER
1
>W OT
)
THER ME
SEDIME
"S ^
OTLi IMJ
HIGH
ORGANIC
MATTER
I
,__ ACID META-
1ER CRYSTALLINE SEDIMENTARY
UCH ij (MSHJ
TA- ACID
NTARY CRYSTALLINE
3L I CLAYEY OTHER
nitl°n °' S°" Sampl1"9 classes for the DDRP Son Survey in the Southern Blue
5-116
-------
slopes and serve as intakes of precipitation. Flooded soils were separated because of their proximity to
surface waters. They serve as the final conduit of water from the watershed to the stream.
The break on high vs. low organic matter refers to the thickness of an organic rich surface horizon,
which may affect organic content, aluminum forms, and other characteristics of the lower soil horizons.
The break on soils formed from acid crystalline (e.g. light-colored, siliceous, granite, gneiss, and schist)
vs. metasedimentary (e.g., phyllite, metasandstone, quartzite, slate) parent materials reflects a probable
difference In the amounts of HIV clays; the latter group is likely to have the greater amounts. These clays
can serve as sinks for aluminum in solution (Bud et al., 1980), an important consideration for the
biological effects of acidification of surface waters. Gibbsite and kaolinite are common in soils from either
parent material.
The separations on soil depth and family particle size in the SBRP were made for the same reasons
as in the NE.
5.5.2: Selection of Sampling Sites
5.5.2.1 Routine Samples
There is a strong tendency for soil scientists to select typical soils for sampling. Although this is
proper for most applications, it would not have been appropriate for the statistically based sampling
scheme used by the DDRP. To ensure an unbiased sample for estimating means and variances of the
characteristics of sampling classes over the regions, the DDRP used a unique three-part scheme of
randomly selecting sampling sites for each sampling class: (1) random selection of watersheds from
those in which the desired sampling class occurred (Figure 5-16); (2) random location of potential
sampling sites on soils maps within delineations in which the class occurred (Figure 5-17); and (3)
random selection of transect direction if the field crew found that the desired sampling class did not occur
within 5 m of the potential sampling site (Figure 5-18).
5-117
-------
f Enter j-
List watersheds on
which sampling class occurs
Assign statistical weight to each
watershed as inverse of number
of sampling classes occurring
Randomly select 8 -10
watersheds
All sampling
classes
selected?
NO
YES
Redistribute samples to approach
uniformity and completeness of
allocation of samples to watersheds
i
i
Final allocation of samples I
to watersheds
Figure 5-16. Selection of watersheds for sampling.
5-118
-------
NO
Obtain soil map for
watershed selected for sampling
Designate map unit delineations
with at least 20% sampling class
Randomly position dot grid
on soil map
Number all points falling
on designated delineations
Determine percent composition (X)
of sampling class in map unit
Randomly select an Integer (Y)
where 1SY S100
Reject point
ME
!
r
Assign number 1 to
first point selected
Assign numbers 2-5 according
to order of selection
ir
As;
"ISSRP
Assign numbers 2 - 5 according
to distance from point 1
Overlay vegetation map to determine
vegetation cfass for each point
Final map showing pro-selected
I sites & vegetation das
Starting
' points selected"
for all watersheds
Jor samptng
class?
YES
All
sampling
classes
completed?
Figure 5-17. Selection of starting points for sampling.
5-119
-------
Obtain watershed map showing
pre-selected starting points
and veae
Proceed to point 1
Examine area within
5 meters of point
Desired soil and
vegetation found?
Randomly select
direction (N, NE, E •••)
Proceed to next
point
Proceed 10 meters In
selected direction
150 meters
from starting
point?
Desired soil
and vegetation
found within 5 meters
of new point?
5 directions
tried?
A1IS
pre-selected
ints tried?
Describe and
sample soil
Cannot sample class
on watershed
Figure 5-18. Field selection of a sampling point for sampling class on a watershed.
5-120
-------
The first step in choosing sampling sites for a given class was to list every DDRP watershed in
which that class occurred (Figure 5-16). Thus, watersheds were identified that had at least one
delineation of any map unit for which soils in that class occupied at least 20 percent of the area.
Watersheds for sampling were randomly selected from this list, at the rate of one watershed for each
desired sample of that class (typically a total of 8-10 per class). For the purposes of this selection, each
watershed was given a statistical weight equal to the inverse of the number of sampling classes occurring
on the watershed. After watersheds were selected, samples were reallocated to approach the following
conditions: (1) equal numbers of samples per watershed and (2) no more than one sample of a given
sampling class in any watershed. Details of the selection process were described by Lee at al. (1989c).
After a watershed was selected for sampling of a particular class, potential sampling sites (usually
five) were determined by random selection from grid points that fell on delineations of map units with
at least 20 percent of their area occupied by soils within the class (Figure 5-17). The vegetation map unit
(i.e., SAF cover type; Eyre, 1980) at each selected point was noted and classified into one of five broad
groups (conifer, hardwood, mixed, open dryland, open wetland). A detailed description of the selection
of potential sampling sites was documented by Lee et al. (1987b).
In each watershed selected for sampling, the field crews proceeded to the first of the potential
sampling sites and determined whether a soil within the desired class occurred within 5 m (Figure 5-18).
If there was any such soil, and if the vegetation at the site felt into the broad group identified from the
vegetation map, the soil was sampled. Otherwise, the crew leader used a random number table to select
a transect direction. The crew proceeded in this direction, stopping at regular intervals to determine if a
suitable soil was present. They sampled the soil at the first site they found that met the criteria for the
sampling class and for the broad vegetation class. If no such site was found on the first transect, another
direction was selected (see Figure 5-18). If the desired combination of soil sampling class and broad
vegetation group was not found after five transects, the crew proceeded to the second pre-selected
potential sampling point, until all pre-selected points on the watershed were exhausted. The instructions
given to the crews for selecting sampling sites were documented in Coffey et al. (1987a,b).
5-121
-------
5.5.2.2 Samples on Special Interest Watersheds
The Special Interest Watersheds (SIW) serve a different purpose than the routine watersheds
(Sections 4 and 11), so a different approach to soil sampling was taken. The five sampling sites in each
SIW were selected to be representative of that watershed. DDRP scientists, in coordination with watershed
modellers, located sampling sites on the soil maps based on hydrologic relation to the lake or stream,
extent of soils in the watershed, and distribution of sampling sites across the watershed. Field crews went
to each site and sampled a soil that they considered to be representative of soils in that portion of the
watershed.
5.5.3 Soil Sampling
The USDA Soil Conservation Service conducted the soil sampling activities for the DDRP. State
offices involved were Connecticut, Maine, Massachusetts, New Hampshire, New York, and Pennsylvania
for the NE, and Georgia, North Carolina, Tennessee, and Virginia for the SBRP.
5.5.3.1 Soil Sampling Procedures
Protocols for DDRP soil sampling were developed for each region (Coffey et al., 1987a,b) by
adapting the procedures of the National Cooperative Soil Survey (Soil Survey Staff 1975,1983,1984). To
enhance regional consistency, standard supplies and equipment were provided to the field crews through
regional centers, specifically the Soil Preparation Laboratories established in cooperation with Agricultural
Experimental Stations. After the crews delivered soil samples to these laboratories, they obtained new
supplies for the next sampling. Laboratory personnel inspected the samples for obvious problems (e.g.,
inadequate sample volume, poor labeling, possible contamination), thereby providing an additional check
on regional consistency.
The protocols gave detailed instructions on the randomized procedure for locating sampling sites
(see Figure 5-18), for excavating pedons in difficult situations, and for documenting the site and pedon
with notes and photographs. Soil profile descriptions were entered onto a form (SCS 232) designed to
facilitate entry into a database. Field estimates of percent rock fragments, included in the profile
5-122
-------
descriptions, were used to correct for non-soil volume during data aggregation (see Sections 8.8.3. 9.2,
9.3).
Crews sampled every horizon thicker than 3 cm thick down to bedrock or to 1.5 m (NE) or
2.0 m (SBRP). Thick horizons were split for sampling. Wherever possible (about 50 percent of horizons),
clods were gathered and coated with Saran in the field, for subsequent determination of bulk density.
Samples were cooled to 4*C within 12 hours, and then taken to the preparation laboratories.
5.5.3 2 Quality Assurance/Quality Control of Sampling
The purpose of the QA/QC tasks for sampling was to ensure and document that the samples were
collected and handled in a consistent, proper manner, and that the chain of custody for each sample was
properly tracked. The QA/QC procedures for sampling were described by Bartz et al. (1987); an
evaluation sampling based on these procedures was documented by Coffey et al. (1987a,b).
Crews were trained at regional workshops prior to sampling. During sampling, every crew was
audited by the State Soils Staff and the RCC, who were responsible for consistency within each state
and within each region, respectively. At least one site per state was audited jointly by the State Soils
Staff and the RCC.
Each crew also was audited by a member of the DORP QA staff. As an independent evaluation,
the EPA auditor used a detailed checklist to document adherence or deviation from protocols as given
in the DDRP sampling manuals. As noted above, regional consistency was also promoted by feedback
from the preparation laboratories.
The QC activities also provided unique information on the variability of pedon descriptions prepared
by different soil scientists. The State Soils Staff and the RCC each performed independent descriptions
of pedons that also had been described by the sampling crews. Thus, for some pedons, up to three
independent descriptions were available. The primary purpose was not to decide which soil scientist was
5-123
-------
"right," but to document the variability inherent in a procedure that is somewhat subjective. Comparison
of descriptions also was useful to promote consistent application of soils concepts within states and
regions.
As an additional QA/QC check, the pedon descriptions were reviewed for consistency by the 80S
state offices and by EMSL-LV staff. Discrepancies were documented and resolved by consulting the field
crews.
Every day, each crew sampled one horizon in duplicate by placing alternate trowelfuls of soil into
two sampling bags. Discrepancies in the laboratory analyses of these samples would indicate probable
contamination at some point in the chain of custody (i.e., sampling, transportation, preparation laboratory,
analytical laboratory). The variability of these samples was documented by Byers et al. (1989) and Van
Remortel et al. (1988).
5-5.4 Physical and Chemical Analyses
The chemical and physical analyses performed on DDRP soil samples are summarized in Table
5-22.
5.5.4,1 Preparation Laboratories
Preparation laboratories acted as intermediaries between the sampling crews and the analytical
laboratories. They were established at Agricultural Experiment Stations at locations within driving distance
of the sampling sites. Four preparation laboratories were established in the NE, and two in the 8BRP:
NE SBRP
University of Connecticut Clemson University
Cornell University University of Tennessee
University of Maine
University of Massachusetts
5-124
-------
Table 5-22. Laboratory Analysis of DDRP Soil Samples
Chemical Analyses
1. pH (distilled water; 0.01 M CaO2 ; 0.002 M CaCI2 )
2. Total carbon*
3. Total nitrogen
4. Total sulfur
5. Cation exchange capacity
a. 1 N NH, OAc, pH = 7.0
b. 1 N NH4 Q, unbuffered
6. Exchangeable bases (Na, K, Mg, Ca)
a. extraction by 1 N NH4 OAc. pH = 7.0
b. extraction by 1 N NH4 Cl, unbuffered
c. extraction by 0.002 M CaCI2
' 7. Exchangeable acidity
a. Bad, -TEA method, pH = 8.2
b. 1 N KCI - effective acidity, exchangeable Al
8. Extractable iron and aluminum
a sodium pyrophosphate
b. ammonium oxalate
c. citrate-dithionite
d. 0.0002 M CaCI2
9. Extractable sulfate
a. water soluble
b. phosphate extractable
10. Sulfate adsorption isotherms (six points)
11. Specific surface area
Physical Analyses
1. Particle size (5 sand fractions, 2 silt fractions, clay)
2, Bulk density
3. Moisture content
* A qualitative test for inorganic carbon is also performed. In the two completed regions, only two samples (out of approximately
3000) tested positive.
5-125
-------
5.5.4.1.1 Responsibilities -
The preparation laboratories received the samples from the crews and provided the crews with
supplies. Soil samples were air dried, sieved (2 mm), subsampled, packaged, and shipped to the
analytical laboratories by the preparation laboratories. Two to four audit samples supplied by the DDRP
QA staff were included in each batch shipped to the analytical laboratories. In addition, one soil sample
was split by the preparation laboratory and included as two samples, called "prep lab duplicates". The
audit samples and preparation laboratory duplicates were packaged and labeled in the same way as
routine samples, and were not identifiable by the analytical laboratories.
The preparation laboratories also were responsible for determining the coarse fragment and
moisture content of samples, for performing a qualitative test for carbonates, and for determining bulk
density from the clod samples. The procedures followed by the preparation laboratories were documented
by F'app and Van Remortel (1987) and Haren and Van Remortel (1987).
5.5.4.1.2 Quality assurance/quality control of physical and chemical analyses -
Preparation laboratories were audited by DDRP QA staff before becoming operational and again
while operational. The QA/QC procedures for the preparation laboratories were described by Bartz et
al. (1987). QA/QC results were evaluated by Papp and Van Remortel (1987) and Maren and Van Remortel
(1987), who concluded that soil sample Integrity was maintained at the preparation laboratories.
5.5.4.2 Analytical Laboratories
5.5.4.2.1 Analyses -
The analytical laboratories were contracted to perform the physical and chemical analyses listed
in Table 5-22 and described in Table 5-23. More complete descriptions of the procedures used by the
analytical laboratories were given by Cappo et al. (1987).
In addition to the parameters listed in Table 5-23, a number of calculated variables were derived
for use in various analyses. Derivation of these variables is described in sections where the variables
5-126
-------
Table 5-23. Analytical Variables Measured in the DDRP Soil Survey (Van Remodel et al., 1988)
Variable
Description of Variable
MOIST Percent air-dry soil moisture measured at the analytical laboratory and expressed as a percentage on an oven-
dry weight basis. Mineral soils were dried at 105°C, organic soils at 60°C.
SP SUR Specific surface area determined by a gravimetric method of saturation with ethylene glycol monoethyl ether
(EGME).
SAND Total sand is the portion of the sample with particle diameter between 0.05 mm and 2.0 mm. It was calculated
as the summation of percentages for individual sand fractions: VCOS + COS + MS + FS + VFS.
VCOSi Very coarse sand is the sand fraction between 1.0 mm and 2.0 mm. It was determined by sieving the sand
which had been separated from the silt and clay.
COS Coarse sand is the sand fraction between 0.5 mm and 1.0 mm. It was determined by sieving the sand which
had been separated from the silt and clay.
MS Medium sand is the sand fraction between 0.25 mm and 0.50 mm. It was determined by sieving the sand which
had been separated from the silt and clay.
FS Fine sand is the sand fraction between 0.10 mm and 0.25 mm. It was determined by sieving the sand which
had been separated from the silt and clay.
VFS Very fine sand is the sand fraction between 0.05 mm and 0.10 mm. It was determined by sieving the sand
which had been separated from the silt and clay.
SILT Total silt is the portion of the sample with particle diameter between 0.002 mm and 0.05 mm. It was calculated
by subtracting from 100 percent the sum of the total sand and clay.
COSI Coarse silt is the silt fraction between 0.02 mm and 0.05 mm. It was calculated by subtracting the fine silt
fraction from the total sitt.
FS1 Fine silt is the silt fraction between 0.002 mm and 0.02 mm. It was determined by the pipet method (USDA/SCS,
1984) and was calculated by subtracting the clay fraction from the less than 0.02 mm fraction.
CLAY Total clay is the portion of the sample with particle diameter of less than 0.002 mm and is determined using
the pipet method.
PH_H20 pH determined in a'deionized water extract using a 1:1 mineral soil to solution ratio and 1:5 organic soil to
solution ratio. The pH was measured with a pH meter and combination electrode.
PH_002M pH determined in a 0.002M calcium chloride extract using a 1:2 mineral soil to solution ratio and 1:10 organic
soil to solution ratio. The pH was measured with a pH meter and combination electrode.
PH 01M pH determined in a 0.01 M calcium chloride extract using a 1:1 mineral soil to solution ratio and 1:5 organic soil
to solution ratio. The pH was measured with a pH meter and combination electrode.
CA_CL Exchangeable calcium determined with an unbuffered 1M ammonium chloride solution. A 1:26 mineral soil to
solution ratio and 1:52 organic soil to solution ratio were used. Atomic absorption spectrometry or inductively
coupled plasma atomic emission spectrometry was specified.
MG_OL Exchangeable magnesium determined with an unbuffered 1M ammonium chloride solution. A 1:26 mineral soil
to solution ratio and 1:52 organic soil to solution ratio were used. Atomic absorption spectrometry or inductively
.coupled plasma atomic emission spectrometry was specified.
K_CL •- Exchangeable potassium determined with an unbuffered 1M ammonium chloride solution. A 1:26 mineral soil to
solution ratio and 1:52 organic soil to solution ratio were used. Atomic absorption spectrometry was specified.
NA_CL Exchangeable sodium determined with an unbuffered 1M ammonium chloride solution. A 1:26 mineral soil to
solution ratio and 1:52 organic soil to solution ratio were used. Atomic absorption spectrometry or inductively
coupled plasma atomic emission spectrometry was specified.
(continued)
5-127
-------
Tablo 5-23. Continued
Variable
Description of Variable
CA_O>\C Exchangeable calcium determined with 1M ammonium acetate solution buffered at pH 7.0. A 1:26 mineral soil
to solution ratio and 1:52 organic soil to solution ratio were used. Atomic absorption spectrometry or inductively
coupled plasma atomic emission spectrometry was specified.
MG_OAC Exchangeable magnesium determined with 1M ammonium acetate solution buffered at pH 7.0. A 1:26 mineral
soil to solution ratio and 1:52 organic soil to solution ratio were used. Atomic absorption spectrometry or
inductively coupled plasma atomic emission spectrometry was specified.
K OAC Exchangeable potassium determined with 1M ammonium acetate solution buffered at pH 7.0. A 1:26 mineral soil
to solution ratio and 1:52 organic soil to solution ratio were used. Atomic absorption spectrometry was specified.
NA_OAC Exchangeable sodium determined with 1M ammonium acetate solution buffered at pH 7.0. A 1:26 mineral soil
to solution ratio and 1:52 organic soil to solution ratio were used. Atomic absorption spectrometry or inductively
coupled plasma atomic emission spectrometry was specified.
CEC CL Cation exchange capacity determined with an unbuffered 1M ammonium chloride solution is the effective CEC
which occurs at approximately the fietd pH when combined with the acidity component. A 1:26 mineral soil to
solution ratio and 1:52 organic soil to solution ratio were used. Samples were analyzed for ammonium content
by one of three methods: automated distiltation/titration; manual distillation/automated titration; or ammonium
displacement/flow injection analysis.
CECJDAC Cation exchange capacity determined with 1M ammonium acetate solution buffered at pH 7.0 is the theoretical
estimate of the maximum potential CEC for a specific soil when combined with the acidity component. A 1:26
mineral soil to solution ratio and 1:52 organic soil to solution ratio were used. Samples were analyzed for
ammonium content by one of three methods: automated distillation/titration; manual distillation/automated
titration; or ammonium displacement/flow injection analysis.
AC_KCL Effective exchangeable acidity determined by titration in an unbuffered 1M potassium chloride extraction using
~ a 1:20 soil to solution ratio.
AC BACL Total exchangeable acidity determined by titration in a buffered (pH 8.2) barium chloride triethanolamine extraction
using a 1:30 soil to solution ratio.
AL_KCL Extractable aluminum determined by an unbuffered 1M potassium chloride extraction using a 1:20 soil to solution
~ ratio. Atomic absorption spectrometry or inductively coupled plasma atomic emission spectrometry was specified.
CA Cl-2 Extractable calcium determined by a 0.002M calcium chloride extraction. A 1:2 mineral soil to solution ratio and
1:10 organic soil to solution ratio were used. The calcium is used to calculate lime potential. Atomic absorption
spectrometry or inductively coupled plasma atomic emission spectrometry was specified.
MG CL2 Extractable magnesium determined by a 0.002M calcium chloride extraction. A 1:2 mineral soil to solution ratio
and 1:10 organic soil to solution ratio were used. Atomic absorption spectrometry or inductively coupled plasma
atomic emission spectrometry was specified.
K CL2 Extractable potassium determined by a 0.002M calcium chloride extraction. A 1:2 mineral soil to solution ratio
and 1:10 organic soil to solution ratio were used. Atomic absorption spectrometry or inductively coupled plasma
atomic emission spectrometry was specified.
NA_CI-2 Extractable sodium determined by a 0.002M calcium chloride extraction. A 1:2 mineral soil to solution ratio and
1:10 organic soil to solution ratio were used. Atomic absorption spectrometry or inductively coupled plasma atomic
emission spectrometry was specified.
FE_CL2 Extractable iron determined by a 0.002M calcium chloride extraction. A 1:2 mineral soil to solution ratio and
1:10 organic soil to solution ratio were used. Atomic absorption spectrometry or inductively coupled plasma
atomic emission spectrometry was specified.
AL_CL2 Extractable aluminum determined by a 0.002M calcium chloride extraction. A 1:2 mineral soil to solution ratio
and 1:10 organic soil to solution ratio were used. The aluminum concentration obtained from this procedure is
used to calculate aluminum potential. Atomic absorption spectrometry or inductively coupled plasma atomic
emission spectrometry was specified.
(continued)
5-128
-------
Table 5-23. Continued
Variable Description of Variable
FE PYP Extractable iron determined by a 0.1 M sodium pyrophosphate extraction using a 1:100 soil to solution ratio.
The pyrophosphate extract estimates organically-bound iron. Atomic absorption spectrometry or inductively
coupled plasma atomic emission spectrometry was specified.
AL_PYP Extractable aluminum determined by a 0.1M sodium pyrophosphate extraction using a 1:100 soil to solution
~ ratio. The pyrophosphate extract estimates organically-bound aluminum. Atomic absorption spectrometry or
inductively coupled plasma atomic emission spectrometry was specified.
FE AO Extractable iron determined by an ammonium oxalate-oxalic acid extraction using a 1:100 soil to solution ratio.
The acid oxalate extract estimates organic and amorphous iron oxides. Atomic absorption spectrometry or
inductively coupled plasma atomic emission spectrometry was specified.
AL AO Extractable aluminum determined by an ammonium oxalate-oxalic acid extraction using a 1:100 soil to solution
ratio. The acid oxalate extract estimates organic and amorphous aluminum oxides. Atomic absorption
spectrometry or inductively coupled plasma atomic emission spectrometry was specified.
FE_CQ Extractable iron determined by a sodium citrate-sodium dithionite extraction using a 1:30 soil to solution ratio.
The citrate dithionite extract estimates non-silicate iron. Atomic absorption spectrometry or inductively coupled
plasma atomic emission spectrometry was specified.
AL_CD Extractable aluminum determined by a sodium citrate-sodium dithionite extraction using a 1:30 soil to solution
ratio. The citrate dithionite extract estimates non-silicate aluminum. Atomic absorption spectrometry or inductively
coupled plasma atomic emission spectrometry was specified.
SO4JH2O Extractable sulfate determined with a double deronized water extract. This extraction approximates the sulfate
which will readily enter the soil solution and uses a 1:20 soil to solution ratio. Ion chromatography was specified.
SO4_?O4 Extractable sulfate determined with a 0.016M sodium phosphate (500 mg P/L) extract. This extraction
approximates the total amount of adsorbed sulfate and uses a 1:20 soil to solution ratio. Ion chromatography
was specified.
SO4jD Sulfate remaining in a 0 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio and 1:20
organic soil to solution ratio. The data are used to develop sulfate isotherms. Ion chromatography was specified.
SO4j2 Sulfate remaining in a 2 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio and 1:20
organic soil to solution ratio. The data are used to develop sulfate isotherms. Ion chromatography was specified.
SO4_4 Sulfate remaining in a 4 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio and 1:20
organic soil to solution ratio. The data are used to develop sulfate isotherms. Ion chromatography was specified.
SO4ji Sulfate remaining in a 8 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio and 1:20
organic soil to solution ratio. The data are used to develop sulfate isotherms. Ion chromatography was specified.
SCMJ6 Sulfate remaining in a 16 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio and
1:20 organic soil to solution ratio. The data are used to develop sulfate isotherms. Ion chromatography was
specified.
SO4J32 Sulfate remaining in a 32 mg S/L solution following equilibration with a 1:5 mineral soil to solution ratio and
1:20 organic soil to solution ratio. The data are used to develop sulfate isotherms. Ion chromatography was
specified.
C_TOT Total carbon determined by rapid oxidation followed by thermal conductivity detection using an automated CHN
analyzer. Total carbon can be used to characterize the amount of organic material in the soil.
N_TOT Total nitrogen determined by rapid oxidation followed by thermal conductivity detection using an automated
CHN analyzer. Total nitrogen can be used to characterize the organic material in the soil.
S_TOT Total sulfur determined by automated sample combustion followed by infrared detection or titration of evolved
sulfur dioxide.
5-129
-------
are first used (e.g., sulfate adsorption isotherms in Section 9.2.3 and cation exchange selectivity
coefficients in Section 9.3.2).
5.5.4.2.2 Selection of analytical laboratories -
The solicitation process (Van Remortel et al., 1988) began with preparation of a detailed statement
of work that defined the analytical and QA/QC requirements in.contractual format, followed by preparation
and advertisement of an invitation for bid (IFB). All laboratories that responded to the IFB were sent
performance evaluation soil samples (PE) to analyze according to DDRP procedures; these samples had
been previously characterized for DDRP. PE bidding laboratories were rated using a scoring sheet
developed by Bartz et al. (1987). All laboratories that passed the PE sample evaluation were then audited
to verify their ability to meet the contractual requirements. Laboratories that passed these on-site
evaluations were awarded contracts for analytical services.
5.5.4.2.3 Quality assurance/quality control of analytical laboratories -
The QA/QC procedures used for evaluating the analytical laboratories were described by Bartz
et al. (1987). Evaluations of analytical laboratory performance were documented by Byers et ai. (1989)
and Van Remortel et al. (1988).
A priori data quality objectives (DQO) were established for all analyses performed by the analytical
laboratories (Table 5-24). DQOs are statements of the levels of uncertainty that a data user is willing to
accept for the planned purposes of the data. The wide variety of data uses planned by the DDRP made
it difficult to set user-specific DQOs. The approach adopted was to set them at levels of precision that
could be expected from good laboratory practices, based on review of available literature and the
experience of DDRP scientists and cooperators. The DQOs were translated into detection limits and
precisions that the analytical laboratories were required to meet (Tables 5-25 and 5-26).
5-130
-------
Table 5-24. Data Quality Objectives for Detectabilfty and Analytical Within-Batch
Precision (Van Remodel et al., 1988)
CRDLa
Precision
Variable
MOIST
SP SUR
SARD"
SILT0
CLAY
PH H2O
PH~002M
PH~01M
CA CL
MG~CL
K Cl
N/A_CL
CA OAC
MG~OAC
K 0~AC
NA_OAC
CEC CL
CEC~OAC
AC RCL
AC'BACL
AL_KCL
CA CL2
MG~CL2
K Cl2
NA CL2
FE~CL2
AL~CL2
Reporting
Units
wt%
m/9
wt %
ii
H
pH units
M
M
meq/100g
H
H
H
meq/100g
*
a
H
meq/100g
H
H
•
•
meq/100g
Units
—
—
—
—
—
—
—
0.003
0.011
0.003
0.006
0.006
0.011
0.006
0.006
0.002
0.002
0.11
0.75
0.80
_„
0.0007
0.0002
0.0004
0.0005
0.0001
mgL'1
_.
._
—
—
—
—
...
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.01d'e
0.01d'°
0.40e
0.25*
0.10
_f
0.05
0.05
0.05
0.05
0.05
Lower (SD)
—
1.0
1.0
1.0
0.15
0.15
0.15
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.25
0.25
0.50
0.50
0.50
mmm
_.
...
—
...
Upper (BSD)
_.
—
—
—
—
—
15%
15%
15%
15%
15%
15%
15%
15%
10%
10%
20%
20%
20%
5%
10%
10%
10%
10%
10%
knot
—
...
...
—
—
—
_.
0.20
0.20
0.20
0.20
0.20
0.20
0.20
0.20
2.5
2.5
2.5
2.5
2.5
—
...
...
—
—
— ~
continued
5-131
-------
Table 5-24. (Continued)
Variable
Reporting
Units
CRDLa
Precision
Units
i -1
mg L" Lower (SD) Upper (BSD)
knot
FE PYP
AL PYP
FE~AO
AL AO
FE"CD
AL_CD
S04 H20
SO4 PO4
SO4~0-32
CJTOT
NJOT
S_TOT
wt %
H
" I
M
«
H
mg S/kg
H
mg S/L
wt %
•
M
0.005
0.005
0.005
0.005
0.002
0.002
2.0
2.0
0.10
0.01 9
0.019
0.01 9
0.50
0.50
0.50
0.50
0.50
0.50
0.10
0.10
0.10
0.010
0.010
0.010
0.05
0.05
0.05
0.05
0.05
0.05
1.0
1.0
0.05
0.05
0.01
0.01
15%
15%
15%
15%
15%
15%
10%
10%
5%
15%
10%
10%
0.33
0.33
0.33
0.33
0.33
0.33
10.0
10.0
1.0
0.33
0.10
0.10
* Contract-required detection limit in reporting units and parts per million, respectively
Precision objectives below and above the knot separating the lower tier (standard
deviation in reporting units) and the upper tier (relative standard deviation in percent);
the knot is in reporting units
* DQOs were not established for size fractions of this parameter
Units are meq L"1 for this parameter for flow injection analysis
* Units in meq for this parameter for titration
CRDL reported as standard deviation of ten non-consecutive blanks
9 Units are weight percent (wt %) for this parameter
5-132
-------
Table 5-25. Detection Limits for Contract Requirements, Instrument Readings, and System-Wide
Measurement in the Northeast (Byers et al., 1989)
Variable
CRDL" Calc IDLb
Conv IDL°
SDLa
%RS>SDL*
CACL
MfJCL
K Cl
M_CL
CA OAC
MCfOAC
K CAC
NAJDAC
CEC CL
CECfOAC
AC ffiCL
AC~BACL
AI/KCL
CA CL2
MG CL2
K C12
MA CL2
FI:~CL2
AI.JCL2
FZ: PYP
AI.~PYP
FIE~AO
AI.~AO
FK~CO
AI.~CD
SO4 H2O
SOI PCM
SO4_0
C TOT
N~TOT
SlTOT
0.05 mg/L
0.05 '
0.05 "
0.05 "
0.05 mg/L
0.05 "
0.05 "
0.05 '
0.01 meq/L
0.01 '
0.25 '
0.40 -
0.10 mg/L
-• mg/L
0.05 "
0.05 "
0.05 '
0.05 "
0.05 "
0.50 mg/L
0.50 '
0.50 "
0.50 "
0.50 "
0.50 "
0.10 mgS/L
0.10 "
0.10 "
0.01 wt %
0.01 "
0.01 "
0.0333 mg/L
0.0174 "
0.0285 '
0.0343 '
0.0275 mg/L
0.0278 '
0.0282 '
0.0279 "
0.0861f meq/L
0.1086' '
0.0693 "
0.3374 '
0.1235 mg/L
0.0208 mg/L
0.0144 "
0.0258 *
0.0343 "
0.0183 "
0.0295 *
0.1941 mg/L
0.2880 "
0.1972 "
0.2238 -
0.1739 "
0.2697 "
0.0250 mgS/L
0.0725 "
0.0306 "
0.0387 Wt %
0.0776 "
0.0045 "
0.0043 meq/100g
0.0037 '
0.0019 •
0.0039 •
0.0036 meq/100g
0.0059 "
0.0019 '
0.0032 "
0.1722 meq/100g
0.2172 "
0.1386 "
1.0122 "
0.0274 "
0.0002 meq/100g
0.0002
0.0001
0.0003
0.0002
0.0007
0.0020 wt %
0.0029
0.0021
0.0022
0.0006
0.0009
0.1669 mgS/kg
0.6050 "
—
«»
—
0.0237 meq/100g
0.0058 "
0.0090 "
0.0149 '
0.0215 meq/100g
0.0126 "
0.0163 "
0.0319 "
0.6032 meq/100g
0.8541 "
0.2400 "
3.6072 '
0.1267 "
0.0939 meq/100g
0.0023
0.0022
0.0081
0.0014
0.0058
0.0200 wt %
0.0603
0.0193
0.0457
0.0653
0.0223
1.1905 mgS/kg
3.2985 '
0.1319 "
0.0478 wt %
0.0058 "
0.0051 "
88.5
88.5
95.5
71.7
89.5
83.0
84.4
44.2
92.2
96.5
82.1
78.3
84.0
99.9
93.7
93.8
87.4
45.2
71.7
92.5
85.1
96.9
94.6
95.4
97.2
99.1
90.7
98.8
97.1
88.5
72.7
* Contract-required detection limit.
Calculated instrument detection limit; estimated as three times the pooled standard deviation of a low
level OL-QCCS.
e Converted instrument detection limit; based on the specified reporting units.
System detection limit; estimated as three times the pooled standard deviations of the lowest 10 percent
of field duplicates; independent of the CRDL
* Percent of routine samples exceeding the system detection limit.
Estimated by averaging laboratory-reported IDLs for incomplete DL-QCCS data.
9 CRDL reported as standard deviation of ten non-consecutive blanks.
NOTE: Detection limits were not applicable for the physical parameters, soil pH, and the remainder of
the sulfate isotherm parameters. Detailed discussions of the attainment of DQOs were given by Byers et al.
(1989) and Van Remortel et al. (1988).
5-133
-------
Table 5-26. Detection Limits for the Contract Requirements, Instrument Readings,
and System-wide Measurement in the Southern Blue Ridge Province (Van Remortel et al., 1988)
Variable
CROL"
Gate IDLb
Conv IDLC
SDLa
%RS>SDL*
CIA CL 0.05 mg/L 0.0524 mg/L 0.0068 meq/100g 0.0311 meq/100g
MdfCL 0.05 mg/L 0.0369 mg/L 0.0079 meq/100g 0.0328 rneq/100g
K Cl 0.05 mg/L 0.0364 mg/L 0.0024 meq/100g 0.0423 meq/100g
MA_CL 0.05 mg/L 0.0415 mg/L 0.0046 meq/100g 0.0195 meq/100g
CA OAC 0.05 mg/L 0.0314 mg/L 0.0041 meq/100g 0.0725 meq/100g
MG"OAC 0.05 mg/L 0.0121 mg/L 0.0026 meq/100g 0.0220 meq/100g
K OAC 0.05 mg/L 0.0330 mg/L 0.0022 meq/100g 0.0363 meq/100g
rVA_OAC 0.05 mg/L 0.0448 mg/L 0.0051 meq/IOOg 0.0098 meq/100g
CEC CL 0.01 meq/L 0.0153 meq/L" 0.0306 meq/100g 1.0724 meq/IOOg
CEC OAC 0.01 meq/L 0.0155 meq/L* 0.0311 meq/100g 0.5609 meq/IOOg
AC KCL 0.25 meq/L 0.0060 meq/L9 0.0188 meq/100g 0.3870 meq/IOOg
AC BACL 0.40 meq/L 0.1840 meq/L* 0.3681 meq/100g 3.7750 meq/IOOg
AL_KCL 0.10 mg/L 0.0840 mg/L 0.0186 meq/100g 0.4780 meq/IOOg
CA CL2 — mg/L1 0.6071 mg/L 0.0160 meq/100g 0.0565 meq/IOOg
MQ CL2 0.05 mg/L 0.0187 mg/L 0.0003 meq/100g 0.0041 meq/100g
K Cl2 0.05 mg/L 0.0335 mg/L 0.0002 meq/100g 0.0020 meq/IOOg
NA CL2 0.05 mg/L 0.0560 mg/L 0.0005 meq/100g 0.0031 meq/100g
FE~CL2 0.05 mg/L 0.0402 mg/L 0.0004 meq/100g 0.0021 meq/100g
AL_CL2 0.05 mg/L 0.0616 mg/L 0.0014 meq/100g 0.0071 meq/100g
FE PYP 0.50 mg/L 0.1434 mg/L 0.0015 wt % 0.0273 wt %
AL PYP 0.50 mg/L 0.2278 mg/L 0.0023 wt % 0.0220 wt %
prAO 0.50 mg/L 0.1941 mg/L 0.0019 wt % . 0.0509 wt %
AL AO 0.50 mg/L 0.2282 mg/L 0.0023 wt % 0.0547 wt %
FirCD 0.50 mg/L 0.1340 mg/L 0.0004 wt % 0.1449 wt %
AL_CD 0.50 mg/L 0.1998 mg/L 0.0006 wt % 0.0426 wt %
S04 H2O 0.10 mgS/L 0.0141 mgS/L 0.2828 mgS/kg 1.7394 mgS/kg
SO4~PO4 0.10 mgS/L 0.0367 mgS/L 0.9186 mgS/kg 3.2539 mgS/kg
S'D4~0 0.10 mgS/L 0.0494 mgS/L — 0.0759 mgS/L
C TOT 0.010 wt % 0.0105 wt % — 0.0821 wt %
N'TOT 0.010 wt% 0.0114wt% — 0.0247 wt %
S^TOT 0.010 wt % 0.0026 wt % — 0.0178 wt %
89.8
92.4
90.0
69.1
77.5
96.1
92.2
92.0
99.9
100
92.1
89.8
83.1
99.6
99.7
99.6
98.9
12.7
51.3
93.8
99.5
93.7
96.3
98.5
99.3
92.0
99.7
91.4
96.7
71.2
44.6
* Contract-required detection limit.
Calculated instrument detection limit, estimated as three times the pooled standard deviation of a low level DL-QCCS.
0 Converted instrument detection limit, based on the specified reporting units.
System detection limit, estimated as three times the pooled standard deviations of the lowest 10 percent of field.
duplicates, independent of the CRDL; Percent of routine samples exceeding the system detection limit.
* Estimated by averaging laboratory-reported IDLs for incomplete DL-QCCS data.
CROL reported as standard deviation of ten non-consecutive blanks.
NOTE: Detection limits not applicable for the physical parameters, soil pH, and the remainder of the sulfate isotherm parameters
5-134
-------
5.5.4.2.3.1 Audits -
Each analytical laboratory was audited twice. The first audit was conducted after evaluating the PE
sample data, before the laboratories became operational. The second audit occurred after sample analysis
had begun, and included review of data from audit and QC samples.
5.5.4,2.3.2 Quality control samples -
QC samples were created and used by the analytical laboratories to maintain random and
systematic errors within specified limits (see Tables 5-25 and 5-26). They were used to evaluate the
calibration and standardization of instalments and to identify problems such as contamination or analytical
interference. QC samples included calibration blanks, reagent blanks, QC check samples, detection limit
QC check samples, matrix spikes, analytical duplicates, and ion chromatography standards. Failure to
meet the specified quality limits could result in rejection of a batch. Detailed descriptions of the use of
QC samples was given by Byers et al. (1989) and Van Remortel et al. (1988).
5.5.4.2.3.3 Audit samples -
Audit samples differed from QC samples in that they were submitted as blind samples to the
analytical laboratories. These were samples of soils that had been well characterized before the DDRP
analyses began. The preparation laboratories inserted a pair of audit samples into batches so that their
identities and composition were unknown to the analysts. Thus, data from these samples provided an
independent assessment of data quality and a means for monitoring the QC procedures. As with the
preparation laboratory duplicates and the field duplicates, the audit samples provided a measure of
precision (i.e., standard deviation) that could be compared to the DQOs. Tables 5-27 and 5-28
summarize the attainment of DQOs in the Northeast and Southern Blue Ridge Province, respectively, as
indicated by data from the audit samples.
Examples of cases in which DQOs were not attained are listed in Tables 5-27 and 5-28.
Parameters for which the within-batch standard deviations exceeded their respective DQOs in both regions
included SAND, SILT, K_CL2, NA_CL2, FE_CL2, S04_0, and N_TOT. In retrospect, the DQOs for SAND
5-135
-------
Table 5-27 Attainment of DQO's by the analytical
laboratories as determined from blind audit samples
for the Northeast.
Variable
Lower limit
Attainment of DQO
Upper limit
SAND
SILT
CLAY
PH H20
PH~002M
PHJNM
CA CL
MG~CL
K Cl
NA_CL
CA OAC
MG'OAC
K OAC
NA_OAC
CEC CL
CEC"OAC
AC RCL
AC~BACL
AC~KCL
CA CL2
MG~CL2
K Cl2
NA CL2
FE~CL2
AL~CL2
FE PYP
AL~PYP
FE AO
AL~AO
FE CD
AL~CD
N
N
S
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
N
Y
Y
S
Y
N/A
N/A
N/A
N/A
N/A
N/A
Y
Y
S
Y
Y
Y
N/A
N/A
N/A
N/A
N/A
N/A
Y
Y
N
Y
Y
Y
S
Y
Y
Y
Y
Y
Y
N
Y
N
N
N
S
Y
Y
Y
Y
Y
Y
Continued
5-136
-------
Table 5-27. (Continued)
Variable Attainment of DQO
Lower limit Upper limit
S04 H20
S04 PO4
SO4 0
S04~2
SO4 4
SO4~8
SO4~16
SO4~32
C TOT
N TOT
S-TOT
Y
N
N
N/D
N/D
N/D
N/D
N/D
Y
Y
Y
Y
N
N
Y
Y
Y
Y
Y
Y
N
Y
Notes: Y = Met DQO.
N = Did not meet DQO.
S = Slightly exceeded DQO.
N/A = Not applicable because no DQO set.
N/D = No data.
5-137
-------
Table 5-28. Attainment of DQO's by the Analytical Laboratories
as Determined from Blind Audit Samples for the Southern Blue
Ridge Province.
Variable
Lower limit
Attainment of DQO
Upper limit
SAND
SILT
CLAY
PH H20
PH 002M
PH_01M
CA CL
MG CL
K CL
NA_CL
CA OAC
MG~ OAC
K O~AC
NAJDAC
CEC CL
CEC~OAC
AC KCL
AC~BACL
AC~KCL
CA CL2
MG~CL2
K CL2
NA CL2
FE CL2
AL_CL2
FE PYP
AL PYP
FE AO
AL~AO
FE CD
AL CD
N
N
Y
Y
Y
Y
N
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
S
Y
N/A
N/A
N/A
N/A
N/A
N/A
Y
Y
S
Y
Y
Y
N/A
N/A
N/A
N/A
N/A
N/A
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
N
N
N
N
Y
Y
Y
Y
Y
Y
Y
Continued
5-138
-------
Table 5-28. (Continued)
Variable Attainment of DQO
Lower limit Upper limit
SO4 H20
S04 PO4
SO4 0
SO4 2
SO4~4
SO4~8
SO4 16
SO4~32
C TOT
N TOT
S-TOT
Y
Y
S
N/D
N/D
N/D
N/D
N/D
Y
Y
Y
Y
Y
S
Y
Y
Y
Y
Y
Y
N
Y
Notes: Y = Met DQO.
N = Did not meet DQO.
S = Slightly exceeded DQO.
N/A = Not applicable because no DQO set.
N/D - No data.
5-139
-------
and SILT (+. 1 percent SD) and for the CL2 cations (+.10 percent BSD) seem excessively restrictive,
especially for the extremely low concentrations in the extracts obtained with 0.002 M CaCI2. For the
latter, it would have been better to have specified a lower-limit DQO, as was done for the OAC and CL
cations, rather than specifying a RSD to be applied for all concentrations.
S.S.S Database Management
The DDRP database was developed by ORNL on IBM and VAX computers using the SAS statistical
software system and the ARC/INFO GIS. Detailed descriptions of database development and
management are contained in the DDRP Database Users' Guide (Turner et al., in review). This section
quotes extensively from that document.
5.5.5.1 Database Structure
The data in the database consist of two fundamental types: alphanumeric (attribute) and
geographic (map). The alphanumeric data were tabulated on IBM personal computers using dBase III and
PC SAS software systems. All tabular data eventually were incorporated into a series of SAS files on
mainframe computers. The map data were digitized and stored as ARC/INFO files (Section 5.4.1.7). In
general, the database design and implementation used by the NSWS-ELS-I, described by Kanciruk et al.
(I986b) was followed.
Figure 5-19 shows the major steps and datasets that led to the final validated database. The final
database is composed of five groups of data files (mapping, field, laboratory, enhanced laboratory, and
synthesis). The mapping, field (pedon description), and laboratory data files contain data that were
collected specifically for this Project. The enhanced laboratory data files have missing, zero, and negative
values replaced by duplicate values or imputed from the remainder of the data; this database was not
used in the analyses presented in this report. The synthesis data files contain data that were summarized
or calculated from the mapping, field, and Deposition Program/National Trends Network (atmospheric
deposition), the USGS (runoff and topographic attributes), and the NSWS (lake and stream chemistry).
5-140
-------
Sampling dass
development
Soil description
and sampling
Mapping data files |
"Sampling data (lies |
Laboratory data files
Raw dataset
(datasetl)
Sampling data tiles |
Laboratory data files 1
Verified dataset
(daiaset 2)
Mapping data files
Sampling data files \-
Laboratofy data files
I
Data aggregation
to pedon, sample
dass, and
watershed
/ Imparted data files
Atmospheric deposition
runoff
Topographical attributes
NSWS - lakes & streams
(validated)
f
Synthesis data files
r
Enhanced
laboratory data files
Validated dataset
(dataset 3)
\
Figure 5-19. Major steps and datasets from the DDRP database.
5-141
-------
The data acquired in each of the above activities were recorded on appropriate data input forms.
The data forms were scanned visually for obvious errors; where possible, these were corrected before
data entry through consultation with DDRP staff or the outside collaborators who had completed the
forms;. The data were double entered from the forms by two different keyboard operators and the files
electronically compared and edited to produce one file with minimized input error. The edited files were
then converted to SAS files as the "Raw Dataset" (Dataset 1).
Verification procedures were designed to ensure that QC goals were met and to evaluate and
quantify sources of error in data collection and handling. Verification included evaluation of precision and
accuracy, representativeness, completeness, and comparability. The specific checks varied with the type
/
of data input. Data that did not meet the QA/QC criteria specified in the DDRP DQOs (Bartz et al., 1987;
Coffey et al., I987a,b) were flagged and then reviewed for field, laboratory, transcription, or data entry
errors. Completion of the verification procedures resulted in the "Verified Dataset" (Dataset 2).
Validation of the data was an extension of verification, but from a larger perspective. For example,
values that appeared reasonable in isolation or when compared with other values in the dataset for that
variable could be distinct outliers within their particular pedon, sampling class, or watershed. Various
graphical and statistical techniques were applied to the verified database to identify and check expected
patterns within pedons, sampling classes, soil taxonomic classes, watersheds, and geographical regions.
Flags, assigned to the laboratory data during verification and validation were translated to a level of
confidence for each laboratory data value to enable subsequent data analysis. The "Validated Dataset"
was Dataset 3.
Data from the mapping, field, and laboratory files were linked and calculations made to aggregate
the data into weighted-average values for each pedon, sample class, and watershed. These summary
data are included in the synthesis data files. Aggregation methods are documented in Sections 8.8.3,
9.2, and 9.3. Data from outside sources, previously verified and validated, were also merged into the
5-142
-------
synthesis data files. These files were checked for transmission errors and compliance with documented
format and contents as each file was merged into the database.
5.5.5.2 Database Operations
5.5.5.2.1 Field data (pedon descriptions) -
5.5.5.2.1.1 Entry of pedon descriptions -
Upon completion of field sampling in the NE, data from the field forms were entered into the
database using ORNL in-house double-entry procedures and then converted to SAS files. For the SBRP
field data, a custom dBase III data entry program was developed (Jones et al., 1986). The data were
double entered, once by SCS staff and once by DDRP staff, using the dBase III program. The two
versions of the data were converted into SAS files and compared using SAS procedures. Corrections
were made to the data using the same transaction-checking procedure described for the mapping data
(Section 5.4.1.2).
For all regions, the field data were entered as two linked files: base and horizon. The base file
contained one record for each pedon. Data pertinent to the entire pedon, such as identifier, date
sampled, location, taxonomic classification, and physiographic and other site information, were stored in
this file. The horizon file contained the detailed horizon descriptions. Information such as horizon depth,
thickness, color, structure, and other features specific to each horizon within each pedon was stored in
this file. For the SBRP the log data (i.e., notes by field crews) were entered into a separate file. Log
data for the NE were recorded in log books by the northeastern field teams. These data were not
entered into the database.
5.5.5.2.1.2 Verification and validation of pedon descriptions -
When the pedon description forms (SCS 232 forms, Coffey et al., 1987a,b) (see Section 5.5.3.2)
were returned from the field, they were evaluated by the QA staff for completeness, legibility, valid codes,
and consistency of entries for each sampling team. After data entry, frequency tables of coded variables
were generated and compared against lists of valid codes. With each of these steps, discrepancy forms
5-143
-------
were returned to the SCS state offices for resolution. Updates from the SCS were entered into a change
file and integrated into the database in the manner described for the mapping files (Section 5.4.1.1.2).
Pedon description data that still were questionable after these checks were flagged in the database.
5.5.5,2.2 Laboratory data -
5.5.5.2.2.1 Entry of laboratory data -
The soil samples were processed in batches consisting of up to 42 samples. The laboratory data
were reported on two preparation laboratory forms and up to 67 analytical laboratory forms (see Cappo
et al., 1987; Van Remodel et al., 1988; Byers et al., 1989). In addition, cover letters from the laboratories
often contained pertinent data or data qualifiers. Some of these were added to the database as
laboratory data tags accompanying the affected data and were considered in the verification process.
The laboratory data forms were sent concurrently to the DDRP data entry staff and the QA staff.
The forms were logged into a tracking and filing system that facilitated entry of data into the computer
as well as evaluation of progress. All forms were visually scanned for completeness, legibility, obvious
error:; of omission, and improper reporting units. Problems were noted and referred back to the
laboratories for resolution.
The data were entered using a customized dBase 111 program (Schmoyer et al., in review)
developed specifically for the DDRP. The double-entered data files were compared using a dBase III file
comfjarison program. Discrepancies were corrected and the files were visually compared with the data
sheets before being converted to SAS files.
. The routine laboratory data were stored in three files containing 72 analytical and identifier variables.
These in turn were linked to nine files of QA/QC data. Labels were assigned to all variables and, where
necessary, variable names and labels were modified to ensure consistency among the various data files.
A detailed listing of the laboratory data file contents is found in Chapter 7 of the DDRP Database Users'
Quids (Turner et al., in review).
5-144
-------
5.5.5.2.2.2 Verification of laboratory data -
Three types of data evaluation were performed for laboratory data verification: (1) QC samples
were used by the laboratories and the QA staff to maintain systematic and random error within tolerable
limits; (2) QA samples that were blind to the laboratories were used by the QA staff as an independent
evaluation of laboratory performance; and (3) internal consistency checks of routine data were performed
to identify outliers or potentially bad data.
Upon receipt by the QA staff, each data batch underwent the QA/QC checks shown in Table 5-
29. QA and QC samples that did not meet the required limits were flagged in the database (these flags
are listed in Chapter 7, Turner et al., in review). Based on the results of the QA/QC checks, the QA staff
prepared a verification report for each batch of samples submitted. A letter was sent to each laboratory
describing potential problems with the reported data. The letters suggested where errors such as
transposed numbers, erroneous dilution factors, or improper calculations may have occurred. The
laboratories were asked to respond with confirmation or reanalysis of the parameters in question.
Reanalysis data were evaluated in the same manner as the original data and, if they met the required
limits;, were entered into the database during the verification process. QA and QC flags that remained
in the database Indicated data outside of specified limits, but the deviations were not considered serious
enough to request reanalysis.
An internal consistency analysis was performed to identify outliers in the routine data (Byers et al.,
1989; Van Remortel et al., 1988). A correlation matrix was generated for all laboratory variables
measured. Highly correlated variables were regressed against one another using a weighted linear
regression model. Outliers were identified using a variety of influence diagnostics. In some cases the
outliers could be attributed to data entry errors, transcription errors, batch-wide calculation errors, and
laboratory-specific calculation errors. If no discrepancies were found, the values were flagged with an
"X4" flag (see Chapter 7, Turner et al., in review). In some cases, the values for one variable would not
correlate well with values for any other variable. In those cases, the highest and lowest 10 percent of
values for that variable were checked for errors.
5-145
-------
Table 5-29. Quality Assurance and Quality Control Checks Applied to Each Data Batch (From Bartz
et all., 1987; Byers et al., 1989)
1. , Audit sample pairs were checked with a standard chart (template) that gave expected values and
their acceptance windows.
2. The percent relative standard deviations (% RSD) of all duplicate pairs were checked.
3. The audit pairs were also subjected to consistency checks of the standard analyte relationships:
Soil pH PH_H20 > PH_002M > PHJD1M
CEC CEC_OAC > CEC_CL
Extractable SO4 SO4_PO4 > SO4JH2O
Acidity AC_KCL < AC_BACL
SO4 Isotherms SO4J) <2<4<8<16<32
Particle Size SAND + SILT + CLAY = 100 ± .1%
Organic Soil > 12% C_TOT
4. Blank concentrations were checked for compliance with the Contract Required Detection Limits
(CRDL).
5. Instrument Detection Limits (IDLs) were checked for contract compliance.
6. Matrix spikes were checked for compliance in preparation (i.e., concentrations were ten times the
CRDL or twice the endogenous level, whichever was greater). Data were checked to ensure a
recovery rate of 100% ± 15%.
7. QC Check Sample (QCCS) data were checked for compliance with the specified control limits.
8. Non-blank corrected data and blank corrected data were checked for proper calculations.
5-146
-------
The data verification procedures also included an evaluation of the database with respect to data
precision, accuracy (interlaboratory differences), representativeness, completeness, and comparability.
Results of these evaluations were detailed by Byers et ai. (1989) and Schmoyer et al. (1989) for the NE,
and by Van Remodel et al. (1988) for the SBRP.
5.5.5.2.2.3 Validation of laboratory data -
Validation of the DORP laboratory data checked relationships among the routine samples in the
context of sampling classes, pedons, and horizons. These checks were made subsequent to the
verification activities that evaluated batch-level laboratory QA/QC data and a limited number of internal
consistency checks on routine data. Because the internal consistency checks conducted as part of the
verification process were limited by time, more of these were added to the validation activity.
Numerous values of a given variable that appeared reasonable when correlated or regressed
against all of the values in the database for that variable could appear as distinct outliers when compared
with other values for that variable within a sampling class, pedon, or horizon. To check for these outliers,
the data for each variable were grouped by sampling class and master horizon and evaluated using a
custom SAS program that performed a box-and-whisker outlier test (Velleman and Hoaglin, 1981). Values
that iell outside the interquartile range (IQR) ± three times the IQR, as well as values with QA/QC or
verification flags that fell outside the IQR ±1.5 times the IQR, were flagged as outliers.
All of the data flagged as outliers were evaluated individually by DDRP soil scientists. Outliers
were evaluated with respect to (1) data for the same variable in the horizons above and below,
(2) variability within the sampling class for that variable, (3) values for related parameters (such as CEC
in acetate and chloride extraction solutions), (4) pedon horizon descriptions, (5) notes accompanying the
pedon description data, (6) QA/QC data such as field and preparation laboratory duplicates, and
(7) verification flags that had been assigned to the data. A validation flag (V1 to V9 or DH) was assigned
to each outlier (see Chapter 7, Turner et al., in review).
5-147
-------
A number of additional internal consistency checks also were run for the individual routine sample
data as part of the validation activity. These were primarily checks of expected relationships that were
requested by DORP scientists prior to establishing confidence levels for the data.
The final step of data validation was to assign confidence levels of zero (good or high confidence)
to four (bad or low confidence) to all data values, based on the assigned number and type of laboratory,
QA, QC, verification, and validation flags. In assigning levels of confidence, values with DH flags were
automatically given level V4; these are most certainly bad data. Values with V5 flags are those that
appeared as outliers on the box-and-whisker test, but are probably valid data based on all the available
information; they are the expected outliers in the dataset. Values with V3 flags are probably also good
data; they probably appeared as outliers due to the way the data were grouped, or aggregated. For
example, Bs and Cg horizons are very different from other B and C subhorizons, but were included with
other B and C horizons in the evaluations. V3 and V5 flags were defined as informational only. Several
pedons with samples that received V7 flags were later deleted from further use In the DDRP. The data
for these samples are probably valid, but the pedons were probably contaminated, making them not
representative of the established DDRP sampling classes.
Only data with levels of confidence of zero, one, and two have been used in most DDRP analyses.
Data with levels of confidence of three or four have been discarded from most analyses, Including all data
aggregation schemes.
5.5.6 Data Summary
5.5.6.1 Summary of Sampling Class Data
The percentage area of each sampling class in the target population was calculated using the
procedure shown in Figure 5-20. First, the area of each sampling class on each watershed was estimated
from the area and composition of each map unit. The regional or subregional area of a sampling class
was estimated as a weighted sum over watersheds, using the inverse of the watershed inclusion
probability as a watershed weight. Total area was calculated by summing over all sampling classes.
5-148
-------
f Enter j—*\
- ^
Designate- sampling class
Designate a watershed in
region or subregion
7
NO
Designate a map unit
on watershed
Area of map unit
on watershed
Area of sampling dass
on watershed: A$«r
Assign weight W». equal to inverse
of watershed inclusion probability
Calculate sampling dass area (At)
by weighted sum over watersheds
As-IWWAsw
All sampling
classes
signaled?
YES
Calculate total area by
summing over sampling classes
Ar-IAs
Percent of sampling
in map unit dass
Calculate percentage areas
7
Figure 5-20. Calculation of percentage of regional or subregional area in each soil sampling.
5-149
-------
Percentages were calculated by dividing the area of each sampling class by the total area. This procedure
yieldiKi unbiased estimates (Figures 5-21, 5-22) of the relative areas of sampling classes in the target
population; that is, all watersheds in the regions that meet the conditions stated in Section 5.2.4.
Depending on the intended use, data from individual soil samples were aggregated to horizons,
pedons, sampling classes and watersheds (Sections 8.8.8, 9.2.2.3). For every routine pedon included in
the DORP database for the NE, Figure 5-23 shows the pedon-aggregated values of pH (water, 0.01 M
CaClj), CEC (NH4CI), base saturation, clay content, extractable sulfate (water, PO^, and the slope and
x-intercept of the sulfate isotherms. The corresponding data for the SBRP are shown in Figure 5-24.
5.5.6.2 Cumulative Distribution Functions
Cumulative distribution functions (CDFs) of the variables included in Figures 5-23 and 5-24 were
calculated for the target population using the procedure shown in Figure 5-25. Sampling class means
were given weights equal to the percentage of the area of the target population occupied by the
corresponding class. CDFs for the NE and SBRP (Figure 5-26) were obtained by ordering the sampling
class means and summing the weignts. Table 5-30 shows medians of these variables by region and also
by subregion.
5.6 DEPOSITION DATA
The regional nature of the Project required estimates of precipitation and atmospheric deposition
(wet and dry) developed in a standardized manner across the eastern United States. Study sites for the
DDRP were selected statistically and had no direct information for deposition. Furthermore, time and
budgetary constraints precluded the instrumentation of sites and, thus, the direct acquisition of any
deposition data. As discussed in Sections 2.1, 3.1, and 4.3.1, the DDRP was designed to focus on the
long-term effects on surface water chemistry of deposition of sulfur. Although sulfur is the primary
deposition variable of interest, complete deposition chemistry is required for the Level I statistical analyses
(Section 8), the Level II base cation analyses (Section 9.3), and the Level III watershed modelling (Section
10).
5-150
-------
0.4
0.4
, 0,3
O.V
Oj»
«•
l
.^...-fSgS&S
«w«i«»4»«a«>ww
Sampling Class: Subr«gion1A
0.1
0.0 ••
n Kl -in
>«»«9«9«9<00««
Sampling Class: Subregion IB
0.3
0.1
Sunpling Claw: Subregion 1C
0.4
, OJ
0.1
OJJ
SempCng Chun: Subregion 10
0.4
I
0.1
0.0
S«nipOnfl Clatc Subrogion IE
Figure 5*21. Relative areas of sampling classes in the northeastern subregions.
5-151
-------
0.31
to
CD
0.2-
o
o>
fS
'E
o
f
o.H
COO) CO CO CO CO CO CO CO
Sampling Class: Northeast Region
0.3i
OJ
|0.2-
-------
4.
. ..:
it:.'. ;.
m * '
Sampling Class
Sampfing Class
to.
; i
Sampling Class
MX!
£J
ii-
•' ii8'
Sampling Oass
10.
0.
•
•
•
• •
_ •
• • »
1 . "'
II1! '•«•!
1
. •
• :
•
i f.«
* ••
: ' •
i . :
• •
: •
.•" .1
i"T ii
«
• **
:«:
. • •:;
• *
• I *
\\i\'' .
Sampfing Class
Figure; 5-23. Aggregated soil variables for individual pedons in the Northeast (continued).
5-153
-------
£ too,
&
soo
g«o
£
, too
8
Sampling Class
40
—M
' Tj *
A
a»
i
• £ ie
. * i
• •:•.* 8ej
Jil';J!J:'JJij-
&mt&%izin
Sampling C5as»
Sampling Class
I-'-'. • ' . -.-"•
• i.. -!!t5i.'.'il'iil!i.,;5.'iin.ui,i
Sam ptog Class
Figura 5-23. (Continued).
5-154
-------
=
£
I ' !
a.
3
i 1 ! - B i i
•
«
3 ri E
5 *
g 1 1 i
Sampling Class
M
*9
SI
|SO
o*
s*
20
10
to*.
c
I »
; ' . !••
i i i i : i \ ! ; i i i
s i i < i 1 I i * i i i
Sampling Class
*
• • " * •
• • • • m * ,
I i i i S i ! : s II i
Sampling Class
so.
g*oj
i-1
i !
Sampling Class
Figure 5-24. Aggregated soil variables for individual pedons in the Southern Blue Ridge Province
(continued).
5-155
-------
*>
£
8
• i
• i
i I i
Sampling Class
i • *
5 : 8
i 1 8
Sampting Ctass
aoo
8
•
' "
MM.
| MOO.
-4 *-
-» 1 I
* E
1 8
M
-« 1 1
Sampdng Class
Figure 5-24. (Continued).
5-156
-------
G-nter ] »H
/ , ».
Designate a sampling class
Catcutate aggregated soil
variable for pedon
Ipedons
in sampling class
ignated?
Calculate mean of soil
variable for sampling dass
Assign weight as percentage area
of dass in target population
NO
Alt sampling
asses designated?
YES
Order sampling dass means
from lowest to highest
i
t
{Calculate COF by accumulating
sampling class weights
Figure 5-25. Calculation of cumulative distribution function for a soil variable in a region or
subregion.
5-157
-------
§»..
o
Q.
po.«
u
0"
o
Q-
oa.«
_
L
o
4.O S.C
pH in Water
U JjO U 4.0 *A *•» >-» «•«
pH in 0.01 m Calcium Chloride
"
a
o«.c-
CL
1..J
MC
SBRP
_£*•
.o
O.
po.«
I-
_O
3
I"
O
to m 30 a M
CEC (Ammonium Chloride) (meq/IOOg)
o M «> «a «o
Base Saturation (Ammonium Chloride)
o
a.
po.*
0 t to tS 20 IS
Percent Ctoy
Figure 5-26. Cumulative distribution functions for pedon aggregated soil variables for the Northeast
and the Southern Blue Ridge Province (continued).
5-158
-------
1.0 i
_
o
o.
d.
£0.4
u
0.0
1.0
o
a.
oo.c -
£«-H
NC
SBRP
u
6 50 100 150
Water-Exfractable Sulfate (mg/kg)
0.0
HI
SBRP
M'^'M'IMH*'!"*!'!""*"''***'*!*'*""**'1!
0 48 80 120 ISO
Phosphale-Exlractable Sulfafe (mg/kg)
i.o
§»•«
o
a.
OO.C
a.
I-
to,
-------
Table 5-30 Medians of Pedon-Aggregated Values of Soil Variables for the DDRP
Regions and Subregions.
Variable Units Median for (Sub)Region
1A IB i£ID IE NE SBRP
pH (water)
pH (CaCI2)
CEC
BS
Gay
S04 (water)
SO, (P04)
Isotherm slope
Isotherm intercept
_._
—
meq lOOg
%
%
mg
mg
—
mg
4
kg
kg'1
r1
5,0
4.5
7.0
9.0
3.0
9.5
24
2.9
101
4.9
4.5
7.0
20
11.5
9.5
18
1.7
285
5.2
4.3
5.0
9.0
3.5
7.0
23
2.8
103
4.9
4.4
2.0
7.0
2.0
9.5
27
1.2
262
5.2
4.5
6.0
18
4.5
6.5
22
2.8
106
4.9
4.5
6.0
10
4.0
7.0
23
2.8
148
5.1
4.3
7.0
10
16.0
8.0
82
21.4
30
5-160
-------
5.6.1 Time Horizons of Interest
5.6.1.1 Current Deposition
Current deposition is of interest within the DDRP because of the Level I Analyses to determine (1)
the current status of sulfur retention within watersheds (Section 7) and (2) the current relationship among
atmospheric deposition, watershed and soil factors, and surface water chemistry (Section 8). This
interest/requirement led to the development of a deposition dataset that represents atmospheric
deposition as of the early to mid 1980s. This deposition dataset, the "long-term annual average" (LTA)
datasiet, is described more fully in Section 5.6.3.2.
5.6.1.2 Future Deposition
The major question driving the DDRP concerns the response of surface water chemistry to
atmospheric deposition in the future. Within the DDRP we were requested by the U.S. EPA's Office of
Air to evaluate two sulfur deposition scenarios for each study region. The first deposition scenario was
constant deposition at current levels. For the NE, the second scenario was for sulfur deposition to
remain constant at current levels for 10 years, then to ramp down for 15 years to a level 30 percent
below current, and to remain at that level for the duration of all Level II and III simulations. For the
SBRP, the second scenario was for sulfur deposition to remain constant at current levels for 10 years,
then to ramp up for 15 years to a level 20 percent above current, and to remain at that level for the
duration of the Level II and III simulations. These scenarios are illustrated in Figure 5-27.
5.6.2 Temporal Resolution
5.6.2.1 Level I Analyses
The Level I Analyses were performed as static analyses of current relationships and thus required
data at only an annual resolution. The LTA dataset fulfilled this requirement.
5.6.2.2 Level II Analyses
,. The simulations of the Level II Analyses were performed using annual time steps and thus required
deposition at the same resolution. The LTA dataset was used in a repetitive fashion for this work, i.e.,
5-161
-------
w
.2
'SI
CO
WF
C
0)
o
CO
C
*Z5
•mm
O
Q_
•Ac
Q
2
"5
CO
c
•
> K
**
X
,«** Current
»*
\
\
\
\
\
\
»,
1
3 20
SBRP +20%
(Base Case)
NE -30%
1 • 1 • • ,
30 40 5
1 5
- 1.4
- 1.3
-12
™* I mmm
- 1.1
•
1 0
"
- 0.9
"
- 0.8
•
- n 7
VI. •
- 0.6
0.5
0
Time (yr)
Figure! 5-27. Sulfur deposition scenarios for the NE and SBRP for Level II and iil Analyses. Ration
of total sulfur deposition at time t (St) to current total sulfur deposition (Sc ).
5-162
-------
the year was repeated for each year of the simulation and was adjusted appropriately for the increase
and decrease scenarios (see Section 5.6.3.2). The LTA dataset matched the resolution of the watershed
runoff dataset (Section 5.7).
i-
i'
5.6.2.3 Level 111 Analyses
The watershed models used in the Level III Analyses required a fine time resolution of precipitation
data lor the calibration of the hydrologic portions of those models (see Sections 10.5.1 and 10.5.2). This
requirement necessitated the development of a finer resolution deposition dataset for the Level III
Analyses. This dataset, termed the "typical year" (TY) dataset, has a daily resolution of precipitation and
a monthly resolution of deposition and was used exclusively in the Level III Analyses. It was also used
as a comparative check against the LTA dataset in (1) the Level I Analyses for sulfur retention (Section
7) and (2) the Level II Analyses for sulfate adsorption (Section 9.2) and base cation depletion (Section
9.3).
5.6.3 Data Acquisition/Generation
Where possible we attempted to use deposition data (wet and dry) as available from specialized
deposition projects within the National Acid Precipitation Assessment Program (NAPAP). A very difficult
constraint of the DORP analyses, however, was that the datasets used had to be complete in terms of
chemical composition (i.e., all major ions), regional coverage, and internal consistency (e.g., charge
balance). Such datasets were not available within NAPAP. Thus, as explained below, we had to generate
such data ourselves as best possible. In the course of this data generation, we consulted at length with
available authorities (both within and external to NAPAP) regarding the reasonableness of our
assumptions, methods, and data generated. Because the deposition datasets for the Level III Analyses
were the most complex and in some cases were the basis for construction of other datasets, we begin
with a description of the Level III typical year dataset.
5-163
-------
5.6.3.1 Level III Analyses - Typical Year Deposition Dataset
As noted in Section 5.6.2.3, the TY dataset was designed to provide a daily resolution of
precipitation and a monthly resolution of deposition in order to be consistent with the hydrologic and
model time step requirements of the Level III models (see Section 10.5). The TY dataset was designed
to represent a yearly precipitation regime that was, indeed, "typical" of current climatological conditions
for the study regions. The dataset was used repetitively (i.e., for each year) for the Level III simulations
with expropriate adjustments during the increase or decrease scenarios (Figure 5-27).
5.6.3.1.1 Wet deposition -
An approach for determining wet deposition data was developed through close consultation with
A. Olsen and his staff who manage the Acid Deposition System database (ADS) at Battelle-Pacific
Northwest Laboratories (PNL). The ADS database is comprised of data from all of the major wet
deposition monitoring networks in the United States. After the approach was developed, the actual
component datasets were developed by A. Olsen and his staff.
o
Initially we investigated the use of wet deposition data derived by spatial interpolation (kriging) of
deposition monitored at ADS sites. Several factors immediately acted to dissuade us from this approach.
First was the relatively poor spatial coverage by the ADS sites, which are widely scattered geographically.
As a test of interpolation, we kriged wet sulfate deposition, in and about the area of the Adirondack State
Park (NY) and visually compared the spatial patterns of wet deposition to these sites with the patterns
of sulfate flux from watersheds in the Adirondacks (see Section 7 for a thorough discussion of
computation of sulfur input/output budgets). Previous work has indicated that sulfur inputs probably are
in balance with sulfur outputs in the Adirondacks (Rochelle et al., 1987; Rochelle and Church, 1987).
Visual comparison indicated that the wet input patterns poorly coincided with the output patterns. As a
comparison, we computed wet suifate inputs by multiplying together wet sulfate concentration kriged from
ADS sites with precipitation kriged from the much denser network of sites of the National Oceanographic
and Atmospheric Administration (NOAA) National Climatic Data Center (NCDC). Patterns produced by
5-164
-------
this procedure were much closer in agreement to observed patterns of sulfate outflux from the Adirondack
watersheds.
A second important consideration was the efficacy of interpolating monthly values of deposition
or wet concentration from ADS sites. The geographic sparseness of the ADS network and the occasional
paucity of monthly data (e.g., during extremely dry months or months during which samples were not
acceptable due to contamination) argued strongly against this approach (A. Olsen, personal
communication).
A third consideration was that daily precipitation data, needed as inputs to the hydrotogic models
of the Level III Analyses were not available from the ADS sites.
As a result, we decided to develop for each individual DDRP study site (1) an appropriate typical
year of wet concentration chemistry obtained from a nearby linked ADS site and (2) a daily precipitation
dataset for a nearby linked NCDC site for the same year as selected for the typical year deposition
chemistry for the linked ADS site. Wet deposition at the DDRP site is then the product of the wet
chemistry and precipitation datasets. This type of multiplicative approach (in genera!) has been discussed
and endorsed by Vong et al. (1989).
Sites for wet deposition chemistry (ADS) and daily precipitation (NCDC) were carefully selected
for each DDRP study site based on geographic location, elevation, and terrain. This selection was made
by DDRP staff in close coordination with A. Olsen and project cooperators involved in the Level III
modelling who were familiar with the requirements of the models and the need for appropriate linkages
between the precipitation inputs and hydrologic outputs from the study watersheds. The ADS and NCDC
sites selected for pairing with the DDRP study sites are shown in Plates 5-14 through 5-19.
5-165
-------
Plate 5-14. ADS and NCDC sites linked with DORP study sites for NE Subregion 1A. The
"concentration zone" indicates to which DDRP sites the appropriate typical year of wet
concentration chemistry from the linked ADS site was applied. The "precipitation zone" indicates
to which DDRP sites the appropriate precipitation (i.e., same year as selected for the linked ADS
site and intersecting concentration zone) from the linked NCDC site was applied. See text for
further description.
5-166
-------
SUBREGION 1A
DDRP SITE LOCATIONS
* PRECIPITATION SITE"
• WET DEPOSITION SITE*
--PRECIPITATION ZONE
(SEE CAPTION)
-— CONCENTRATION ZONE
(SEE CAPTION)
,' ! +1AI-806 \
\ UU057+
\ i Big Maott
\ QIIBiaah^ ... ,
+1kI-Stf
-------
Plate 5-15. ADS and NCDC sites linked with DDRP study sites for NE Subregion 1B. The
"concentration zone" indicates to which DDRP sites the appropriate typical year of wet
concentration chemistry from the linked ADS site was applied. The "precipitation zone" indicates
to which DDRP sites the appropriate precipitation (i.e., same year as selected for the linked ADS
site and intersecting concentration zone) from the linked NCDC site was applied. See text for
further description.
5-167
-------
SUBRE6ION IB
DORP SITE LOCATIONS
Subreg ion
Locat ion
* PRECIPITATION SITE*
• WET DEPOSITION SITE"
--PRECIPITATION ZONE
(SEE CAPTION)
-—CONCENTRATION ZONE
(SEE CAPTION)
*PrecIp!iaiion and wei
deposition siie ID's ere
given only for siies used
-------
Plate 5-16. ADS and NCDC sites linked with DDRP study sites for NE Subregion 1C. The
"concentration zone" indicates to which DDRP sites the appropriate typical year of wet
concentration chemistry from the linked ADS site was applied. The "precipitation zone" indicates
to wiiich DDRP sites the appropriate precipitation (i.e., same year as selected for the linked ADS
site and intersecting concentration zone) from the linked NCDC site was applied. See text for
further description.
5-168
-------
SUBREGION 1C
DDRP SITE LOCATIONS
SubregI on
Local: ion
x PRECIPITATION SITE*
• WET DEPOSITION SITE*
-- PRECIPITATION ZONE
(SEE CAPTION)
-— CONCENTRATION ZONE
(SEE CAPTION)
nv l II « Statio
•30 •
x 'Precipitation and wet
N deposition site ID's are
i given,only for sites used
-------
Plate 5-17. ADS and NCDC sites linked with DDRP study sites for NE Subregion 1D. The
"concentration zone* indicates to which DDRP sites the appropriate typical year of wet
concentration chemistry from the linked ADS site was applied. The "precipitation zone" indicates
to which DDRP sites the appropriate precipitation (i.e., same year as selected for the linked ADS
site and intersecting concentration zone) from the linked NCDC site was applied. See text for
further description.
5-169
-------
SUBREGiON ID
DORP SUE LOCATIONS
Subreg i on
LocatI on
x PRECIPITATION SITE*
• VET DEPOSITION SITE*
--PRECIPITATION ZONE
(SEE CAPTION)
-—CONCENTRATION ZONE
(SEE CAPTION)
\
i ; *~ P ^ ,^-N
I i \ (
V "^~ ^ m» - X .».!»,.
l^ >*" - \ c«-i" -. v~£ 'M^ik
i
I
*Precip!iatIon and wei
deposition site ID's are
given only for sites used
-------
Plato 5-18. ADS and NCDC sites linked with DDRP study sites for NE Subregion 1E. The
"concentration zone* indicates to which DORP sites the appropriate typical year of wet
concentration chemistry from the linked ADS site was applied. The "precipitation zone" indicates
to which DDRP sites the appropriate precipitation (i.e., same year as selected for the linked ADS
site and intersecting concentration zone) from the linked NCDC site was applied. See text for
further description.
5-170
-------
SUBRE6ION IE
DDRP SITE LOCATIONS
x PRECIPITATION SITE*
• WET DEPOSITION SITE*
--PRECIPITATION ZONE
(SEE CAPTION)
-— CONCENTRATION ZONE
(SEE CAPTION)
(*>?&,
* ' \
^ \ _
(+tik-«« VjrVv
\ x^ ^"Mi
IB«^A.II^- It**^*^"1!1* yl
"^ -fltt-MBX \
+II1-HJ / ,
*Prec(pI tat Ion end wet
deposition site ID's are
given only for sites used
-------
Plato 5-19. ADS and NCDC sites linked with DDRP study sites for the SBRP. The "concentration
zono" indicates to which DDRP sites the appropriate typical year of wet concentration chemistry
from the linked ADS site was applied. The "precipitation zone" indicates to which DDRP sites the
appropriate precipitation (i.e., same year as selected for the linked ADS site and intersecting
concentration zone) from the linked NCDC site was applied. See text for further description.
5-171
-------
SOUTHERN BLUE RIDGE PROVINCE
ODRP SITE LOCATIONS
SBRF Study Area
x PRECIPITATION SITE*
• WET DEPOSITION SITE*
-—PRECIPITATION ZONE
(SEE CAPTION)
-—CONCENTRATION ZONE
(SEE CAPTION)
*—*s-*.\ \ ' ' ,
***^ +• i*^ ""^ X /
^ ** t * sisasifx '
*Prec I piiat I on and wet
deposition site ID's are
given only for si.tes used
-------
5.6.3.1.1.1 Wet deposition chemistry -
Precipitation chemistry data were obtained from the ADS database. For each ADS site the entire
history (usually less than five years) of daily or weekly data was obtained. The annual cumulative
distribution functions (CDFs) for each individual year were compared with the summary CDFs of data
for till years. The typical year was selected as the year that compared best to all years for sulfate
concentration, nitrate concentration, and precipitation. After the typical year was selected, monthly wet
deposition chemistry was computed using the procedures recommended by the Unified Database
Committee (Olsen et al., 1989). Their quarterly criteria were applied to each month. When monthly data
for the typical year selected did not meet the criteria, an alternate typical year was used.
5.6.3.1.1.2 Daily precipitation -
The same year chosen as the typical year for deposition chemistry was used as the typical year
for precipitation at the linked NCDC site.
In a few cases precipitation data were not available for the ADS typical year. In this event the
closest years with respect to sulfate concentration, nitrate concentration, and precipitation were used for
which precipitation data were available. An additional advantage of using the NCDC sites was that
long-term data are available for these sites allowing adjustments of individual years and days of data to
a long-term norm for the location. In this case daily precipitation at each NCDC site was adjusted using
a nearby site with 30-year normal monthly and annual data. Sites and data were obtained from the
NCDC tape TD9641: Monthly Normals of Temperature, Precipitation, and Heating and Cooling Degree
Days 1951-80.
Daily precipitation during a month was adjusted to match the 30-year normal for the month. Each
daily value was multiplied by the ratio of the 30-year normal for the month and the monthly total for the
typical year selected. This procedure also ensured that the typical year annual total matched the annual
30-year normal.
(Information on data completeness and quality for the ADS sites is available from A. Olsen, PNL)
5-172
-------
5.6.3.1.2 Dry deposition -
The determination of representative typical year estimates of dry deposition at the DDRP study sites
was a very difficult task. The accurate measurement of dry deposition to watersheds is a developing
art and for the purposes of the DDRP, no network of sites existed that was able to provide the regionally
consistent information that our analyses required. Instead, we had to rely on estimates from a variety
of modelling and inferential techniques. (Note that we used the term "estimates" (as opposed to "data")
to describe derived values for dry deposition for all variables. To describe the grouping of these
estimates, however, we use the term "database".) Information on estimates of dry sulfur deposition from
the Regional Acid Deposition Model (RADM) (Chang et al., 1987) was obtained from R. Dennis (AREAL-
RTP) and S. Seilkop (Analytical Sciences, Inc.) as was information on possible annual-scale relationships
among fine particle dry deposition of base cations and chloride and wet deposition of those same ions.
We combined this information with other information on (1) dry deposition to surfaces, (2) canopy
scavenging, (3) throughfall, and (4) pertinent information on interactions among atmospheric deposition
and watershed ion budgets to construct complete (major ions) suites of dry deposition to represent a
typical year for each of the DDRP study sites.
5.6.31.2.1 Sulfur-
Interim or first-stage dry sulfur deposition estimates for DDRP study sites were provided based
upon output available from RADM (R. Dennis and S. Seilkop, personal communication and unpublished
internal report, 1987a; Clark et al., 1989). Previous site estimates made by the Regional Lagrangian Model
of Air Pollution (RELMAP) (Eder et al., 1986) appeared to suffer from an over-smoothing problem and
were judged inadequate for our work. (R. Dennis and S. seilkop, personal communication). Because
this over-smoothing problem could not be corrected in time to provide estimates for the DDRP, RADM
was used.
The first-stage estimates were based on the simulation of six three-day episodes and the results
averaged to establish regional dry deposition. "Ground-truth" data on dry sulfur deposition from a sparse
number of measurement sites (Hicks et al., 1986; Hosker and Womack, 1986) were used to
5-173
-------
geographically adjust (spatially calibrate) the RADM output. Output from RADM is to points on an 80 x
80 km grid. Estimates at those points were then kriged to individual DDRP study sites.
We performed an evaluation of the first-stage dry sulfur deposition estimates by combining them
with the wet deposition data and constructing sulfur input/output budgets for sites in the NE. Use of the
first-stage dry sulfur estimates resulted in computed inputs that were slightly lower than outputs (for a
Subregion 1A). In the NE, and especially in this Subregion there is good reason to believe that
watersheds are in a steady-state situation with regard to sulfur (i.e., inputs = outputs) (see Section 7 and
Galloway et al., 1983a; Rochelle et al., 1987; Rochelle and Church, 1987). An increase in the first-stage
dry sulfur deposition estimates of 20 percent slightly increased the estimated total sulfur inputs and
brought the input/output budgets for Subregion 1A more closely to the steady-state point. [Note,
however, that this slight adjustment had no effect on conclusions on regional patterns of sulfur retention
in eastern watersheds (see Section 7).] The uncertainty associated with the RADM outputs could
conceivably have bias of 20 percent. Indeed, ground-truth data from the only two northeastern sites
available (West Point, NY, and Whiteface Mountain, NY) indicated an underestimate by RADM at these
sites by 40 and 20 percent, respectively, even after geographic spatial calibration (R. Dennis and S.
Seilkop, personal communication and unpublished internal report, 1987). Consideration of these factors
by the DDRP staff in close coordination with the Level 111 modellers (J. Cosby, J. Schnoor, S. Gherini;
see Section 10) led to the joint decision that the first-stage dry sulfur deposition estimates from RADM
should be adjusted upward by 20 percent annually in the NE. This adjustment was made to the annual
estimates for the DDRP NE study sites, and ail subsequent manipulations (as described in this section)
to the estimates of dry sulfur deposition were performed on this adjusted or second-stage dataset. No
comparable watershed data were available in the SBRP to check the deposition estimates because the
»
SBRP is a region where atmospherically deposited sulfur generally is strongly retained (Galloway et al.,
1983a; Rocheile et al., 1987; Rochelle and Church, 1987). Thus, no such comparable adjustments were
mads to the annual dry sulfur deposition estimates provided for the SBRP.
5-174
-------
The next step was to apportion the dry sulfur deposition on a monthly basis. Because scavenging
of dry sulfur deposition should be a function of canopy development, we used the watershed vegetation
information from the DDRP mapping (Section 5.4.1.3) to adjust for monthly partitioning. This was a
two-step process. First, we assigned a leaf area index (LAI) (Table 5-31) to each vegetation type
(coniferous, deciduous, and open), based, in part, on values used by Gherini and Goldstein (1984) (see
Table 5-31). We used two variations on this approach: (1) we assigned an LAI of 0.25 to deciduous
vegetation during the months of November through March and (2) we partitioned our "mixed" vegetation
class as half deciduous and half coniferous. Second we applied an iterative predictor-corrector technique
to apportion the monthly deposition so that its sum closely approximated the second-stage annual dry
sulfur deposition totals. Application of these procedures provided a third-stage (final) dry sulfur deposition
dataset for which the annual sum of the monthly dry deposition was within two percent of the second
stage annual value on the average for any watershed.
5.6.3.1.2.2 Base cations and chloride -
Computation of individual watershed values for dry deposition of base cations (Caz*. Mga+, Na*,
K+) find chloride (CT) involved quite a number of considerations and computational steps. At the heart
of the computation was the development of a technique (Eder and Dennis, in revision) that used
regression analysis between measured annual wet deposition and the annual geometric means of ambient
air concentration (used with deposition velocities to compute dry deposition). Data used In the
development of the technique and relationships were obtained from the Ontario Ministry of the
Environments Acid Precipitation in Ontario Study (APIOS) (For a description of the network and its data
collection and analysis techniques see Chan et al., 1982 and Tang et al., 1986.) Because of the manner
in which the ambient concentrations were measured, these relationships probably apply only to fine
particle (< 2 pm) dry deposition. For the purpose of computing annual fine particle dry deposition, it
5-175
-------
Table 5-31. Monthly Values of Leaf Area Index (LAI) Used to Apportion
Annual Dry Deposition to Monthly Values
Month LAI d LAI c LAI o
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
0.25
0.25
0.25
0.5
1.0
2.5
4.0
4.5
4.5
1.0
0.25
0.25
12
12
12
12
13
14
15
15
15
15
14
13
1
1
1
1
1.5
1.5
1.5
1.5
1.5
1
1
1
LAI ci = leaf area index for deciduous vegetation
LAI c = leaf area index for coniferous vegetation
LAI o = leaf area index for open areas
5-176
-------
was assumed that deposition velocities were roughly equivalent among the base cations at 0.8 cm sec*1
for heavily forested vegetative situations (Eder and Dennis, in revision) (this is the condition for the DORP
watersheds, inasmuch as most have vegetative coverage of at least 80 percent).
This approach was developed from data at inland stations and probably is inappropriate for use
In near-coastal situations. Watersheds close to the coast can be strongly influenced by sea salt inputs
of wet deposition (considered in the selection of ADS sites for computation of typical year wet
deposition), but development of fine particle dry deposition using the approach outlined above would
probably lead to overestimates. The relative proximity of DDRP study sites to the coast is shown in Plate
5-20. Sites located within 10 km of the coast are probably strongly influenced by sea salts but sites
greater than 50 km from the coast are probably negligibly influenced (Sullivan et al., I988a). Because
of this likely effect, we substituted annual wet deposition data from adjacent but more inland ADS sites
for this computation for 20 near-coastal DDRP sites.
The annual fine particle dry deposition had to be partitioned into monthly components. The annual
values first were partitioned based upon 13 28-day months (Eder and Dennis, in revision), and then they
were repartitioned by DDRP staff into the 12 months comprising the TY dataset.
Coarse particle (> 2 m) dry deposition also can be an important contributor to vegetative
canopies and, thus, to watersheds (Lindberg et al., 1986; Stensland, personal communication). A major
debate currently exists as to whether "inputs" of dry base cations and chloride originate within or external
to watersheds (of the size studied by DDRP) (Hicks, personal communication). We feel that a majority
of such inputs arise externally and thus we applied a ratio of coarse-to-fine particle dry deposition to
account for this influence. We estimated these rations based on a number sources of information (1)
information presented by Lindberg et al. (1986), (R. Munson, personal communication), and (3)
consideration in input/output budget calculations. To a large degree, these values were derived iteratively
in concert with the other computations we for estimating dry deposition of these ions. The ratios used
5-177
-------
Plate 5-20. DORP study sites relative to distance from Atlantic Coast (< 10 km, 10-50 km, >50 km).
5-178
-------
DDRP STUDY SITES
Distance From Coast
Kilometers
H 0-10
• 10-50
H > 50
10 km
50 km
-------
are shown in Table 5-32. Computed coarse-particle dry deposition values were added to the values of
fine-particle dry deposition.
We next applied an adjustment for scavenging using the monthly LAIs indicated in Table 5-31. We
again assumed that the "mixed" vegetation class was half coniferous and half deciduous. Application
of these LAIs, however, resulted in values of total dry deposition that appeared much too large in relation
to output fluxes of the ions from the study watersheds, i.e., inputs of base cations appeared to nearly
equal outputs and inputs of chloride greatly exceeded outputs. Assuming that outputs of base cations
from watersheds usually significantly exceed outputs and that inputs of chloride should roughly equal
outputs (in undisturbed locations, see Section 10.5.7), then the values obtained using the above
procedure were unrealistically high.
Scavenging of dry deposition by vegetative canopies (especially coniferous canopies) is subject to
a pronounced "edge effect" whereby lower windspeeds and ambient concentrations within interior
canopies result in markedly lower effective dry deposition to those interior regions (Dasch, 1987;
Grennfelt, 1987). We reasoned that this process could be represented by a function of the form of the
well-known Michaelis-Menten equation and used to adjust the "effective" coniferous canopy scavenging.
In this way, scavenging of base cations and chloride within our watersheds would not be computed using
total coniferous LAIs, but rather the coniferous LAIs would be adjusted (in effect) so that as the areal
coniferous coverage of a watershed increased, its effect on scavenging reached an effective plateau
rather than increasing linearly. We used this approach to adjust total dry deposition until the chloride
budgets approximately balanced in undisturbed NE sites. The final equation used in the adjustment was
% CON = (30 * % CON)/(15 + % CON) (Equation 5-1)
6
where % CON = mapped percent coniferous coverage
i % CON = effective percent coniferous coverage
6
5-179
-------
Table 5-32. Ratios of Coarse-to-Fine Particle Dry Deposition
Ion Ratio
Calcium 1.5
Magnesium 1.0
Sodium 1.0
Potassium 1.0
Chloride 0.2
5-180
-------
These computations of dry base cation deposition leave a great deal to be desired. The final values,
however, relate well to (1) estimates of dry deposition-to-wet deposition ratios observed by Lindberg et
al. (1986) for a southeastern forested catchment, and (2) previously modelled estimates at Woods and
Panther Lakes in the Adirondack Mountains (R. Munson, personal communication).
5.6.3.1.2.3 Nitrate and ammonium -
We had no objective or mechanistic approach to use for estimating dry deposition of nitrate and
ammonium. Instead, we assumed that total dry deposition of nitrate was equal to wet deposition and
that total dry deposition of ammonium was equal to one-half wet deposition. These ratios approximate
values measured by Lindberg et al. (1986) in an eastern forested watershed.
5.6.3.1.2.4 H* -
We computed dry H deposition as the difference between dry anions and other dry cations.
When the sum of other dry cations was greater than the sum of dry anions, we set dry H* to zero.
5.6.3.1.2.5 Ion ratios -
The ratios of dry deposition to wet deposition for all ions for the NE and SBRP study sites for the
TY dataset are shown in Table 5-33
5.6.3.1.2.6 Comparisons with Direct Measurements
Although extensive data do not exist with which to compare the DDRP estimates, there is some
limited information) obtained as a personal communication from Dr. Bruce Hicks, Atmospheric Turbulence
and Diffusion Division, Environmental Research Laboratories, NOAA) that can be used for this purpose.
For example, preliminary NOAA estimates of wet and dry sulfur deposition for the NE (sites in
central Pennsylvania, Whiteface Mountain and Howland, Maine) and the SBRP (Oak Ridge) are highly
comparable to regional averages of the DDRP estimates. The regional average of DDRP estimates of wet
5-181
-------
Table 5-33. Ratios of Dry Deposition to Wet Deposition for DDRP Study Sites for the Typical Year
(TY) Deposition Dataset
NE
SO42'
Ca2+
Mg2+
Na+
K*
cr
*NOg
*NH4*
H+
SBRP
SO42'
Ca2+
Mg2+
Na+
K+
cr
*N03"
*NH4'1"
H"
Median
0.44
1.13
1.92
1.29
1.56
0.38
1.0
0.5
0.47
Median
0.62
1.72
1.83
1.14
1.48
0.40
1.0
0.5
0.50
Mean
0.48
1.12
1.82
1.29
1.66
0.33
1.0
0.5
0.46
Mean
0.60
1.54
1.69
1.06
1.36
0.36
1.0
0.5
0.52
Standard Deviation
0.12
0.42
0.72
0.61
0.71
0.12
0.23
Standard Deviation
0.12
0.39
0.45
0.35
0.36
0.10
-
-
0.16
* nitrate set to 1.0. ammonium set to 0.5
5-182
-------
sulfur deposition in the NE is roughly 15 percent greater than the NOAA estimate for the sites examined
and the DDRP estimate of dry sulfur deposition is about 25 percent less than the NOAA estimate. The
estimates of total sulfur deposition in the NE are virtually identical; the total and even the individual
component estimates (i.e., wet and dry) in the SBRP are within 4 percent (B. Hicks, personal
communication).
A comparison of regional (i.e., NE and SBRP) dry/total deposition ratios as obtained from the DDRP
estimates and "as quantified by NOAA for the same region" shows remarkable agreement for both regions
for suifate, nitrate and ammonium. No NOAA values are available for chloride so no comparison can be
made for that ion. The DDRP estimates of the dry/total ratio for base cations is generally just over twice
as high as the NOAA values for the NE and ranges from 4 to 10 times as great in the SBRP. This
difference is due at least partly to the fact that DDRP estimates for base cation deposition include an
estimate for large particle dry deposition, whereas the NOAA values do not. For example, a comparison
of the average DDRP estimates of small particle base cation deposition for five watershed sites in
proximity to the NOAA West Point station show good agreement (i.e., within 20 percent for calcium
sodium and magnesium; within 50 percent for potassium) with the measured NOAA values at that site
(B. Hicks, personal communication). To account in part for the uncertainties associated with the DDRP
estimates of base cation dry deposition, we performed sensitivity analyses (see Section 9) with datasets
having much reduced base cation values (Section 5.6.3.2.4). More formal uncertainty analyses were
performed with the integrated watershed models (Section 10). In general, DDRP analyses and the
conclusions drawn from them were not sensitive to these uncertainties (see Sections 9 and 10).
5.6.3.1.3 Sulfur deposition scenarios -
Typical year total sulfur deposition (as suifate) is shown in Plates 5-21 through 5-27. As described
in Section 5.6.1 (see Figure 5-27), the DDRP was requested to examine the effects of scenarios of both
current and altered sulfur deposition in the NE and SBRP. The sulfur increases and decreases were
performed as suifate with both dry and wet deposition altered at equal and constant percentages (of the
total) each year. It seemed appropriate that only wet and dry H* should be adjusted to coincide with
5-183
-------
Plate 5-21. Pattern of typical year sulfate deposition for the DDRP NE study sites.
5-184
-------
TYPICAL YEAR
SULFATE DEPOSITION
SO,
2" -
Q 1 - 2
H .2 - 3
B 4 - 5
• > 5
-------
Plate 5-22. Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1A.
5-185
-------
SUBREGION 1A
TYPICAL YEAR
SULFATE DEPOSITION
-------
Plate 5-23. Pattern of typical year sulfate deposition for the DORP study sites in Subregion 18.
5-186
-------
Subregi on
ion
SUBREGION 18
TYPICAL YEAR
SULFATE DEPOSITION
SO," - g rrf2
Q 0 - 1
n i - 2
r-^-.t f\ -y
yy c - o
• 3-4
• 4-5
> 5
-------
Plate 5-24. Pattern of typical year sulfate deposition for the DORP study sites in Subregion 1C.
5-187
-------
SUBREGION 1C
TYPICAL YEAR
SULFATE DEPOSITION
Subregi on
Locat ion
-------
Plate 5-25. Pattern of typical year sulfate deposition for the DORP study sites in Subregion 1D.
5-188
-------
SUBRE6ION ID
TYPICAL YEAR
SULFATE DEPOSIT
Subregion
Locat ion
SO,2" - g m-
D 0- 1
n i - 2
El 2 - 3
• 3-4
B 4 - 5
• > 5
v-
1D2-OM
-------
Plate 5-26. Pattern of typical year sulfate deposition for the DDRP study sites in Subregion 1E.
5-189
-------
SUBREGION IE
TYPICAL YEAR
SULFATE DEPOSITION
-------
Plate 5-27. Pattern of typical year sulfate deposition for the DDRP SBRP study sites.
5-190
-------
SOUTHERN BLUE RIDGE PROVINCE
TYPICAL YEAR
SULFATE DEPOSITION
-------
these changes (A. Olsen, R. Dennis, personal communication) and that was the procedure followed. Wet
H was adjusted equal to the wet sulfate adjustment and dry H was recomputed so that the sum of
dry cation inputs was not less than the sum of dry anion inputs (on an equivalent basis) In any month.
5.6.3.2 Level I and II Analyses - Long-Term Annual Average Deposition Dataset
As discussed in Section 5.6.1.1, it was appropriate to develop a dataset of atmospheric deposition
at an annual resolution to represent "current" deposition as of the early-to-mid 1980s. This dataset was
called the long-term annual average (LTA) dataset and was used in the Level I and II Analyses (Section
5.6.2).
5.6.3.2.1 Wet deposition -
The objectives for developing LTA wet deposition estimates were to produce at each DDRP site
annual wet deposition representative of (1) current (early to mid 1980s) atmospheric chemistry conditions
and (2) average regional spatial deposition patterns. Our approach is to use 5-year average precipitation
chemistry available from six regional and national wet deposition networks in conjunction with 30-year
normal (1951-1980) annual precipitation available at a much greater spatial density. Annual wet deposition
is computed as the product of annual precipitation amount and annual precipitation chemistry. This
approach was discussed in detail in Section 5.6.3.1.1. We considered the estimation of annual wet
deposition at a site to be computed by developing an annual estimate of precipitation at the site, an
estimate of annual wet deposition chemistry (precipitation-weighted concentration) at the site, and taking
their product.
5.6.3.2.1.1 Wet deposition chemistry -
Wet deposition chemistry data were provided by A. Olsen (PNL) from the ADS database of regional
and national wet deposition monitoring networks. Annual precipitation-weighted concentration was
estimated for each watershed based on 1982-1986 data from the monitoring networks. The process for
developing the estimates is similar to those discussed by Wampler and Olsen (1987) and described
further by Vong et al. (1989). Briefly, the procedure used the following steps. For each wet deposition
5-191
-------
monitoring site, annual summaries for each ion species were computed using the procedures described
by the Unified Deposition Database Committee (UDDC) (Olsen et al., 1989). An average over 1982-1986
was completed for all sites that had three or more years that met the UDDC data quality rating. Simple
kriging (applied to moving areas to minimize trend effects) provided wet deposition chemistry estimates
at each DDRP site.
5.6.3.2.1.2 Annual precipitation -
Annual precipitation estimates were provided by A. Olsen. The estimates are derived from 30-year
(1951-1980) normal precipitation data obtained from the National Climatic Data Center. Simple kriging
applied to subregions of the United States was used to estimate annual precipitation at each DDRP site.
The subregions were developed to maximize the homogeneity of precipitation spatial patterns and improve
the model used within the kriging estimation procedure.
5.6.3.2.2 Dry deposition -
Dry deposition for all ions for the LTA dataset was computed from LTA wet deposition using the
dry/wet ion ratios developed in the TY dataset for each ion for each DDRP watershed.
5.6.3.2.3 Sulfur deposition scenarios -
The annual sulfur deposition (as sulfate) for the LTA dataset is shown in Plates 5-28 through 5-34.
The altered sulfur deposition scenarios were computed by decreasing or increasing (as appropriate) the
sulfate values in the original LTA dataset In the same manner as described in Section 5.6.3.1.1 for the
TY dataset and then adjusting dry H+ as described in Section 5.6.3.1.2.4.
5.6.3.2.4 Decreased base cation LTA datasets -
Because the dry deposition of base cations in the TY and LTA datasets appeared higher than might
be expected from sparse measured data (B. Hicks, personal communication), we produced additional LTA
datasets having base cation dry deposition set at 50 percent and 0 percent of the original LTA values.
This was done to test the sensitivity of the Level II watershed base cation modelling analyses presented
5-192
-------
Plate 5-28. Pattern of LTA sulfate deposition for the DDRP NE study sites.
5-193
-------
LONG TERM ANNUAL AVERAGE
SULFATE DEPOSITION
SO,2" - 9 rr
ED 2 - 3
• > 5
-------
Plate 5-29. Pattern of LTA sulfate deposition for the DORP study sites in Subregion 1A.
5-194
-------
SUBREGION 1A
LONG TERM ANNUAL AVERAGE
SULFATE DEPOSITION
-------
Plate 5-30. Pattern of LTA sulfate deposition for the DDRP study sites in Subregion 1B.
5-195
-------
SUBREGION IB
LONG TERM ANNUAL AVERAGE
SULFATE DEPOSITION
SubregI on
Local I on
D o - i
n i - 2
• 3-4
• 4-5
• > 5
-------
Plate 5-31. Pattern of LTA sulfate deposition for the DDRP study sites in Subregion 1C.
5-196
-------
SUBRE6ION 1C
LONG TERU ANNUAL AVERAGE
SULFATE DEPOSITION
Subregi on
Location
-)
i \
Qicj-ttt \
I
} 0ict-«oi
uiV»ii®
©IC1-9J5
Ul-lW
0i«-*iiv
N
*J \
k
\
t
\
/ \
1
\
f1
t
\
(
J_
>
t
(
®1C1-OII
ftiei-oio^1
<91Ci-t)>
®1C1-OJJ
_ ®ICI-»«I
®1C»-0!7
"i
HJ-8J7
et-oii ® j
^
/
®1C3-O^J
"-^
\ -
IC^04I
I'-'SittSi^"-^
>iei-03t
U1-0
C"
-------
Plate 5*32. Pattern of LTA sulfate deposition for the DORP study sites in Subregion 1D.
5-197
-------
SUBREGION ID
LONG TERM ANNUAL AVERAGE
SULFATE DEPOSITION
Subregion
Location
SO,*" - g trf2
n o -1
n i - 2
02-3
• 4-5
• > 5
1D2-QI4
-------
Plate 5-33. Pattern of LTA sulfate deposition for the DDRP study sites in Subregion IE.
5-198
-------
SUBREGION IE
LONG TERM ANNUAL AVERAGE
SULFATE DEPOSITION
-------
Plate 5-34. Pattern of LTA sulfate deposition for the DDRP SBRP study sites.
5-199
-------
SOUTHERN BLUE RIDGE PROVINCE
LONG TERM ANNUAL AVERAGE
SULFATE DEPOSITION
-------
in Section 9.3. In these datasets dry H* again makes up the difference between the sum of dry anions
and the sum of other dry cations. These datasets are referred to here as LTA-reduced base cations
(LTA-rbc) and LTA-zero base cations (LTA-zbc), respectively. Analysis using these datasets might be of
special Interest relative to recent hypotheses presented by Driscoll et al. (1989b) concerning the potential
role of base cation deposition in controlling the chemistry of dilute surface waters in the NE (for further
discussion see Section 3).
5.6.4 Deposition Datasets Used in DDRP Analyses
Table 5-34 presents a summary of the analyses to which the deposition datasets described above
were applied in the DDRP. For a discussion of quantitative uncertainties associated with the use of these
data see Section 10.10.
5.7 HYDROLOGIC DATA
5.7.1 Runoff
An estimate of average annual runoff for the DDRP study sites is necessary for all three levels of
analysis. Given the site selection procedures used in the DDRP, it is not surprising that the DDRP study
sites are ungaged and measured values of annual runoff are not available. Three options existed for
obtaining estimates of runoff. The first, gaging the systems, would not have been practical to obtain the
estimates of runoff needed, given the relatively short time frame of the DDRP and the large number of
sites. The second option was to use an interpolation method, such as kriging, to estimate runoff at each
site. Large variability in topography across the regions, and in other features that influence runoff, limited
the applicability of this method. The third option was selected for estimating runoff to each DDRP site:
(1) interpolations were made based runoff contour maps developed with existing runoff data and (2)
expert judgment of hydrologists experienced in runoff mapping.
5.7.1.1 Data Sources
Working in cooperation with the USGS, a runoff contour map of average annual runoff for 1951-
80 (Figure 5-28, Krug et al., in press) was developed to use for interpolating runoff at the DDRP sites.
5-200
-------
Table. 5-34. Deposition Datasets Used in DORP Analyses
Dataset
TY
LTA
LTA-rbc
LTA-zbc
Sulfur
Retention
(Section 7)
X
X
-
-
Level 1
Statistics
(Section 8)
-
X
-
-
Level II
Base Cation
(Section 9.2)
X
X
X
X
Sulfate Adsorption
(Section 9.3)
X
X
-
-
Level III
Modelling
(Section 10)
X
-
-
-
TY = Typical Year
LTA = Long-Term Annual Average
LTA-rbc = Long-Term Annual Average - Reduced Dry Base Cations (dry base cations = 50% of LTA)
LTA-zbc = Long-Term Annual Average - Zero Dry Base Cations (dry base cations set to 0)
5-201
-------
ANNUAL RUNOFF 1951 - 1980
(From Krug et d, in press)
Figure 5-28. Example of average annual runoff map for 1951-80 (Krug et al., in press).
5-202
-------
The map was developed to encompass the NE, Mid-Appalachian, and SBRP Regions of the eastern
United States (Figure 5-28). Average annual runoff data for the 30-year period was taken primarily from
watersheds of less than 2,590 km2 having no diversions or regulations. If a gaging station did not have
a complete set of records for the 30-year period, then Krug et al. (in press) calculated a 30-year estimate
using standard correlation methods described by Matalas and Jacobs (1964). Runoff contours were
plotted at a 1:500,000 scale at 5.1-cm (2-in.) intervals up to 76.2 cm (30 in.) and at 12.7-cm (5-in.)
contour intervals for runoff greater than 76.2 cm (30 in.). (Krug et al. (in press) have provided more
specific information on map development and quality assurance.)
5.7.1.2 Runoff Interpolation Methods
A simple nearest-contour linear interpolation method was used to estimate runoff for each DDRP
site. The Krug et al. (in press) map was digitized Into a GIS system by the USGS. Using the GIS
(Campbell et al., in press), we the DDRP study sites were overlaid onto the runoff contour maps and
runoff was interpolated at each DORP site to the nearest one inch based on the nearest contour to a site.
The nearest contour was determined using an engineer's scale to measure a line from the station location
perpendicular to the contour (Rochelle et al., in press).
5.7.1.3 Uncertainty Estimates
Determining a quantifiable estimate of the uncertainty associated with the runoff interpolations is
important to the effective use of the runoff data in the Levels I, II, and III Analyses. Working with the
USGS, we conducted an analysis to estimate the uncertainties in using a runoff contour map to determine
runoff at a specific site. This analysis was incorporated into the development of the 1951-80 runoff map
(Rochelle et al., in press; Krug et al., 1988). We randomly selected a subset of the total USGS sites
available for map development and withheld these sites from use in map development. Then we used
the runoff contour map to interpolate runoff at these sites and compared the interpolated values to the
actual long-term measured values. We determined that runoff could be estimated, on the average, within
approximately 8.9 cm (3.5 in. or 14.9 percent) of the actual measured runoff. [See Rochelle et al. (in
press b) for a complete discussion of the uncertainty analysis.]
5-203
-------
A second analysis was conducted to test the consistency of interpolating runoff using the hand-
linear interpolation method described above. For the NE region, 883 NSWS watersheds were plotted on
the 1951-80 runoff contour map, and runoff was interpolated to each site. A 146-watershed subset of the
883 NSWS sites was plotted onto the runoff contour maps, and runoff was interpolated at the test sites
a second time. We compared the two independent estimates to check for consistency in using the hand-
linear interpolation method. We found that 11 percent of the sites had a 2.5-cm (1-in.) difference between
the two interpolations (5 percent runoff difference) and 1 percent had a 5.1-cm (2-ln.) difference between
the two runoff interpolations. The results of a paired t-test indicate that the hand interpolation method
is reasonably consistent with no significant differences in runoff between the two iterations (t=0.65,
p=0.51). Rochelle et al. (in press b) provide a full description of all uncertainty analyses.
5.7.2 Derived Hvdroloaic Parameters
The hydrologic pathway followed by precipitation in reaching surface waters is an important factor
affecting the processes that control the response of surface water chemistry to acidic deposition.
Determining the hydrologic pathways in a watershed is difficult without extensive hydrologic information.
Often collecting such data is expensive and requires long periods of data collection to yield hydrologicaily
meaningful information. We have attempted to use other indirect methods to describe the hydrology of
the DDRP study watersheds and to, in turn, relate these characteristics to surface water chemistry. We
have included hydrologic/geomorphic parameters from three sources for use in the DDRP: (1)
parameters calculated by the hydrologic model TOPMODEL (Seven and Kirkby, 1979; Beven, 1986), (2)
empirical index of soil contact calculated using Darcy's Law, and (3) mapped hydrologic/geomorphic
parameters collected from topographic maps and aerial photography.
5.7.2.1 TOPMODEL
5.7.2.1.1 Northeast -
5.7.2.1.1.1 Model description -
TOPMODEL (Beven and Kirkby, 1979; Beven, 1986; Wolock et al., in press) was chosen to estimate
an index of flowpath partitioning because the model requires readily available topographic and soils
5-204
-------
information, and it predicts internal hydrologic states that can be used to partition streamflow.
TOPMODEL characterizes flowpath partitioning by characterizing the spatial distribution of 1n(a/KbTanB)
where "a" is the area drained per unit contour, TanB" is the local slope, "K" is the hydraulic conductivity,
and "b" is the depth to bedrock. The critical topographic/soils information for a watershed as a whole
is the spatially aggregated distribution function of ln(a/KbTanB). The first three moments can be routinely
used to characterize the distribution (Wolock et al., 1989).
High values of ln(a/KbTanB) indicate areas in the catchment that are likely to produce surface
runoff. These would typically be characterized as topographically convergent, low transmissivity areas.
Conversely, low ln(a/KbTanB) 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) is the critical parameter
for characterizing an individual watershed.
5.7.2.1.1.2 Data sources -
5.7.2.1.1.2.1 Soil Conservation Service mapping
To characterize the distribution function of ln(a/KbTanB), TOPMODEL requires information on
depth to bedrock ("b") and hydraulic conductivity ("K"). Values of "b* and "K" were estimated based on
mapped information obtained from the DDRP Soil Survey (Lammers et al., 1987b; Lee et al., 1989a). To
obtain estimates of "K", soil texture classes were associated with the soil types based on SCS texture
classifications (Soil Survey Staff, 1981). Next, saturated hydraulic conductivity values ("K") were assigned
to the texture classes based on data available in Rawls et al. (1982) (Table 5-35). Values of "b" were
assigned by using a mid-point for each depth-to-bedrock class except the highest class (greater than or
equal to 30 m), in which case a value of 30 m was assigned (see Table 5-10).
5.7.2.1.1.2.2 Digital elevation models
Local slope (TanB") and the area drained per unit contour ("a") were computed using USGS
1:250,000-scafe digital elevation models (OEM) (Eiassal and Caruso, 1983). 1:250,000 OEM data comprise
5-205
-------
Table 5-35. DDRP texture classes and saturated hydraulic conductivity (K) for the NE study
systems. Estimates of (K) are based on data available from Rawls et al. (1982).
Soil Texture Class
Sand
Loamy sand
Sandy loam
Loam
Silt loam
Muck
Fine sandy loam
Mucky peat
Gravelly loam
Gravelly loamy sand
Channery silt loam
Variable
Mucky loamy fine sand
Channery loam
Complex
Very gravelly sandy loam
Peat
Channery very fine sandy loam
Coarse sand
Fibric
Gravelly sandy loam
Sandy clay loam
Clay loam
Silty clay loam
Sandy clay
Silt clay
Clay
Mucky loam
Hydraulic Conductivity fcm/hr)
21.00
6.11
2.59
1.32
.68
15.00
3.00
14.00
3.00
7.00
1.00
1.00
17.00
1.50
1.00
3.50
12.00
3.00
21.00
10.00
3.50
0.43
0.23
0.15
0.12
0.09
0.06
5.00
5-206
-------
a three arc-second elevation grid interpolated from USGS 1:250,000-scale topographic maps. Three arc-
seconds represented approximately 60 x 90 m in the NE.
5.7.2.1.1.3 Model calculations -
5.7.2.1.1.3.1 ln{a/KbTanB)
The spatial distribution of ln(a/KbTanB) was derived by combining estimates of "b" and "K" with
the topographic values of "a" and TanB" (Wolock et al., 1989). First, the appropriate DEM was overlaid
with DDRP Soil Survey soil and depth-to-bedrock maps for the individual watershed using the ERL-C
ARC/INFO GIS. Files containing the soils and topographic information were then output for subsequent
analysis. The elevational data were used to calculate the total area draining into each grid cell ("A"), as
well as the contour length ("C") and slope ("TanB") along which this area drained out of the cell (a=A/C).
Given DX and DY (the X and dimensions of the cell), an initial value for 'A* of DX*DY was assigned to
each point. To perform the calculations for a given cell, the elevation of the cell was compared to that
of its four diagonal and four cardinal neighboring points. Values of ln(a/KbTanB) were then computed
as follows: (1) 'TanB" was calculated as the weighted average of all downhill direction slopes, (2) "C" was
determined as the cell boundary length with neighboring downhill cells, (3) estimates of °K" and "b" were
combined with a/TanB, and (4) ln(a/KbTanB) was calculated. The area that drained into the cell was
then partitioned to all its downslope neighbors in quantities proportional to TanB'and "C", and added to
the previous values of "A" for those downhill points. All calculations of ln(a/KbTanB) and subsequent
partitioning of area were performed in order of decreasing elevation. The estimated values of
ln(a/KbTanB) were then aggregated throughout the watershed; a shifted gamma distribution was fit; and
the first three moments of the distribution were estimated.
5.7.2.1.1.4 Model output -
The index ln(a/KbTanB) is used to characterize flowpath partitioning of the DDRP watersheds (see
Section 8.2.1.2.2.4.2). The index is a measure of the importance of quick flow mechanisms within a
watershed. Watersheds with high mean values of ln(a/KbTanB) tend to have a higher percentage of
storm runoff in quick flow (e.g., return flow). Conversely, watersheds that have low mean values of
5-207
-------
ln(a/KbTanB) tend to be dominated by slower hydrologic mechanisms (e.g., sub-surface storm flow).
Personnel and time constraints limited the SBRP analyses to ln(a/TanB) rather than ln(a/KbTanB).
5.7.2.1.2 Southern Blue Ridge Province -
5.7.2.1.2.1 Model description -
For the SBRP, we used TOPMODEL to estimate an index of flow/path partitioning by characterizing
the spatial distribution of ln{a/TanB), where "a" is the area drained per unit contour and TanB" is the
local slope. This is similar to analyses in the NE (see Section 5.7.2.1.1.1) except that only topographic
factors are used to partition streamflow. Personnel and time constraints limited the SBRP analyses to
ln(a/TanB) rather than In(a/Kb TanB).
5.7.2.1.2.2 Data sources -
Local slope (TanB*) and area drained per unit contour ("a") were computed using DEM data as
described in Section 5.7.2.1.1.2.2. Three arc-seconds represented approximately 75 x 90 m in the SBRP.
5.7.2.1.2.3 Model calculations ln(a/TanB) -
The spatial distribution of ln(a/TanB) was derived similarly to ln(a/KbTanB) (see Section
5.7.2.1.1.3.1) except that soils information ("K" and "b") was not included. The calculation of ln(a/TanB)
was completed as follows. First, a DEM was overlaid with the appropriate DDRP Soil Survey watershed
map using the ERL-C ARC/INFO GlS. Files containing the elevation for grid points within the watershed
were output for subsequent analysis. The elevational data were used to calculate the total area draining
into each grid cell ("A"), as well as the contour length ("C") and slope (TanB") along which this area
drained out of the cell (a=A/C). Given DX and DY (the X and Y dimensions of the cell), an initial value
for "A" of DX*DY was assigned to each point. To perform the calculations for a given cell, the elevation
of the cell was compared to that of its four diagonal and four cardinal neighboring points. Values of
tn(a/TanB) were then computed as follows: (1) TanB" was calculated as the weighted average of all
downhill direction slopes, (2) "C" was determined as the cell boundary length with neighboring downhill
cells, and (3) ln(a/TanB) was calculated. The area that drained into the cell was then partitioned to all
5-208
-------
its downslope neighbors in quantities proportional to TanB" and "C", and added to the previous values
of "A" for those downhill points. All calculations of ln(a/TanB) and subsequent partitioning of area were
performed in order of decreasing elevation. The estimated values of ln(a/TanB) were then aggregated
throughout the watershed, a shifted gamma distribution was fit, and the first three moments of the
distribution were estimated.
5.7.2.1.2.4 Model output -
The index ln(a/TanB) was used for SBRP analyses (see Section 5.7.2.1.2.3) rather than ln(a/KbTan-
B). Model interpretation is similar, however, and is more fully described in Section 5.7.2.1.1.4.
5.7.2.2 Soil Contact (Darcy's Law)
The amount of contact precipitation has with the soils component of a watershed is one factor
determining the chemistry of the resultant runoff. The potential for contact is a function of soil depth,
permeability, and slope. One approach to estimating a potential for soil contact is to use Darcy's Law
to calculate a theoretical maximum soil contact time and an index of potential contact. Darcy's Law can
be defined as:
Q=KAS (Equation 5-2)
where Q equals lateral soil flow, K is an estimate of the saturate hydraulic conductivity, A is the cross-
sectional area of flow and S is the hydraulic gradient. Q is then normalized by watershed area and
related to annual runoff to estimate an index of potential soil contact (Peters and Murdoch, 1985).
Peters and Murdoch (1985) working with the ILWAS study systems (Murdoch et al., 1984) used
Darcy's Law to develop an index of potential soil contact for Woods Lake and Panther Lake watersheds.
They found the high pH lake system (Panther Lake) had a high potential for soil contact based on
Darcy's Law and the low pH system (Woods Lake) had a low potential for contact. They found that the
high contact system characteristically had deeper soils than the low potential contact system. We have
applied the Darcy's Law technique described by Peters and Murdoch (1985) to the DDRP study sites to
5-209
-------
calculate (1) an estimate of potential lateral flow and (2) an index of the maximum potential for soil
contact (see Section 8.2.1.2.2.4.1).
5.7.2.2.1 Data used and sources -
5.7.2.2.1.1 Soil mapping data -
5.7.2.2.1.1.1 Hydraulic conductivity (K)
A weighted-average hydraulic conductivity (K) for each watershed was determined by estimating
K using soil mapping texture delineations for each of the soil components mapped in the DDRP Soil
Survey as described earlier (Section 5.7.2.1.1.2.1). Estimates of K based on soil texture were obtained
using values presented in Rawts et al. (1982). The DDRP Soil Survey provides an estimate of the
percentage of watershed occupied by each soil component (see Section 5.2). We used these
percentages to calculate a weighted-average K per watershed.
5.7.2.2.1.1.2 Cross-sectional area (A)
The cross-sectional area "A* was determined by multiplying the perimeter of the lake by the average
depth of permeable material. By using lake perimeter we were able to determine the average area at the
point of contact between the soil matrix and the surface water system. Lake perimeter was measured
from watershed maps prepared by the DDRP Soil Survey and digitized into the ERL-C ARC/INFO GIS
(Campbell et al., in press). As pan of the DDRP Soil Survey, depth-to-bedrock classes were mapped
for each of the watersheds (see Table 5-10). We determined an average depth for each watershed by
calculating a weighted-average depth to bedrock based on the proportion of the watershed occupied by
each class. For the calculation we used the mid-point of each Class (see Table 5-10) except for Classes
I and VI; we used 0.5 m for Class I and 30 m for Class VI.
5.7.2.2.1,1.3 Slope (S)
An average slope for each watershed was calculated based on the slope estimates associated with
the DDRP Soil Survey map units (Lee et al., 19893). Each map unit has a slope class designation
indicating associated slope. Table 5-36 shows the SCS slope classifications associated with the map
5-210
-------
units. We calculated an area weighted-average slope based on the area of each map unit within a
watershed proportional to the total watershed area using the mid-point of each class presented in Table
5-36.
5.7.2.2.1.2 Runoff-
An estimate of the average annual runoff for each site was determined from the Krug et al. (in
press) runoff contour map described in Section 5.7.1. Runoff Interpolation methods are discussed in
Section 5.7.1.3.
5.7.2.2.2 Model calculations -
We applied the Darcy's Law calculation to non-seepage lakes In DDRP NE watersheds. Figure 5-
29 diagrams the algorithm used to apply the Darcy's Law equation to the watersheds. The final outputs
from these calculations are an estimate of the potential lateral flow and an Index of soil contact.
5.7.2.3 Mapped Hydrologic Indices
Previous investigators (Hewlett and Hibbert, 1967; Dingman, 1981; Woodruff and Hewlett, 1970;
Carlston, 1963; Lull and Sopper, 1966; Vorst and Bell, 1977) have attempted to relate hydrologic
characteristics to mapped watershed geomorphic/hydrologic parameters for forested watersheds. In
general, most of the previously reported research is at the event level or covers short time periods (i.e.,
days or weeks). The DDRP is primarily interested in longer time frames (e.g., annual time steps). For
use In the Level I Analyses, we have developed a hydrologic indices database for the NE and SBRP
study systems. The primary goal is to be able to link these geomorphic/hydrologic parameters to the
current surface water chemistry of the study systems (NSWS chemistry).
5.7.2.3.1 Data sources -
The geomorphic/hydrologic parameters (hydrologic indices) were determined from three data
sources. First, all area measurements came from maps prepared as part of the DDRP Soil Survey
(Section 5.2). The second major source of information is 7.5' and 15' topographic maps. Topographic
5-211
-------
Table 5-36. SCS slope classifications.
Class Slope(%) midpoint
A CT~3Tl>
B 3-8 5.5
C 8-15 11.5
D 15-25 20
E 25-45 35
F 45-70 57.5
5-212
-------
( Enter 1-
Calculate Q - Lateral Soil Row (m3 /d)
Q=K*A*S
Where:
K = saturated hydraulic conductivity estimated
based on soil texture (m/d)
A = x-sectional area for flow; lake perimeter *
soil depth estimated from soil survey (m2}
S = average slope based on soil survey data
Normalize Q
NormQ = Q/WA (m/yr)
WA - watershed area (m 2 )
Compare to Runoff and calculate Index (I)
I« NormQ/R
R« runoff (m)
Assume greater potential
for quick response, flashy system
-i/-
Assume greater potential
tor high soil contact, slow response
Figure 5-29. Flow chart of Darcy's Law soil contact calculation as applied to the DORP study sites.
5-213
-------
maps were used for elevational and length measurements, for stream delineations and for sub-basin
determination. Whenever possible, 7.5' maps were used. For 70 of the DORP sites, however, only 15'
maps were available. The last source of mapped information was obtained from aerial photography taken
as part of the DDRP land use survey (see Sections 5.4.1.6 and 5.4.2.7; Liegel et al., in review). The aerial
photos were used to check stream delineations and other specific information obtained from the
topographic maps.
5.7.2.3.2 Data collection procedures (Northeast) -
Geomorphic parameters were defined from map measurements taken from 7.5' topographic maps
(when available) or from 15' topographic maps from the USGS topographic map series. Ail map
Information was digitized and entered directly into a computer database via an interactive program (K.
Nash, personal communication). Table 5-37 describes all measured or computed parameters. The
majority of the measures identified in Table 5-37 were selected from geomorphic/hydrologic parameters
listed by the U.S. Department of Interior (1977). Additionally, we have included parameters that are
specifically descriptive of lake watersheds. These are watershed area-to-lake area ratio (WS_LA),
watershed perimeter-to-lake perimeter ratio (PERIMRAT), and percent area in open water bodies, including
the primary lake of the watershed (H2O_WS).
Three additional measures included in the geomorphic/hydrologic database are average annual
runoff (R), retention time (TR) for the primary lake, and lake volume (V). "R" was interpolated (Rochelle
et al., in pressb) from a runoff contour map (Krug et al., in press) of average annual runoff for 1951-80.
TR and V were estimated by the NSWS (see Kanciruk et al., 1986a).
To ensure consistency, all map measurements were made by the same individual according to
pre-established methods. Also, we conducted a quality check to ensure that the data were accurate
and measurements were consistent. First, a 10 percent subset of 144 watersheds was re-digitized and
compared against the original measurements. The differences were compared to published interpretation
errors (USDI, 1977).
5-214
-------
Table 5-37. Mapped and calculated geomorphic parameters collected for the NE study
sites.
Parameter
Description
Units
Measured
B_CENT
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 km
distance from basin outlet to farest
upper point in basin
The length of the line which defines km
the surface divide of the drainage
basin
Area of all open water bodies in drainage km2
basin
Total length of intermittent streams km
as defined from USGS topographic maps of
aerial photos
Area of the primary take km2
Primary lake centroid expressed as
X,Y coordinates
Perimeter of primary basin lake km
Elevation at approx. highest point m
Elevation of primary lake m
Total perennial stream length as defined km
from USGS topographic maps and aerial photos
Area of each sub-catchments in the km2
drainage basin
Maximum stream Order (Horton) of streams
in the watershed (aerial photos used to aid
in reducing cooling problems between 7.5 and
15 minute maps)
TOTSTRM
AW
Total stream length; combination of
perennial and intermittent
Total watershed area
km
km2
continued
5-215
-------
Table 5-37 (continued)
Parameter
Calculated
B_SHAPE
B_WIDTH
COMPACT
DDENSITY
ELONG
H20_WS
MAX_REL
M_PATH
PER_DD
PERIMRAT
REL_RAT
ROTUND
WM_PATH
WS LA
Description
Units
Basin Shape ratio;
B_LEN */WS_AREA
Average basin width;
WS_AREA/B_LEN
Compactness Ratio; ratio of perimeter
of basin to the perimeter of a circle
with equal area;
(PERIM)/(2*( *AW)'5)
Drainage Density;
TOTSTRM/WS_AREA
Elongation Ratio;
(4 * WS_AREA)/L_BEN
Ratio of open water bodies area to
total watershed area
H2O_AREA/WS_AREA
Maximum relief;
MAX_ELEV - MJN_ELEV
Estimate of mean flow path;
Drainage density calculated from
perennial streams only
PERIN/WS_AREA
Ratio of the lake perimeter
to the watershed perimeter;
Lake Perimeter/B_PERIM
Relief Ratio;
(MAX_ELEV-MI N_ELEV)/B_LEN
Rotundity Ratio;
(B_LEN)2/(4 * WS_AREA)
Estimate of weighted mean flow
path;
Ratio of the total watershed area to
the area of the primary lake
km
m
m
m
continued
5-216
-------
Table 5-37 (continued)
Parameter Description Units
Additional
Tfk Lake retention time yr
V* Volume of the primary lake I06m3
R Average annual runoff; interpolated cm
to each site from Krug et al. (1988)
runoff map
5-217
-------
Second, we conducted internal database checks to determine gross data entry errors and to identify
obvious errant values. The internal data verification checks were (1) all areas and lengths were greater
&
than zero, (2) maximum elevation was greater than minimum elevation, (3) sub-basin areas were less than
watershed area, (4) open water areas were less than watershed area, and (5) the lake perimeter was less
than the basin perimeter.
Third, watershed elevations and areas were also checked against data in a separate database
constructed as part of the DORP Soil Survey (Lee et al., 1989a). The elevation data in the databases
differed by less than 10 percent. Total areas in both databases were within 10 percent for 136
watersheds. We checked the 8 watersheds individually to determine and resolve remaining discrepancies.
5.7.2.3.3 Data collection procedures (Southern Blue Ridge Province) -
Data collection procedures were the same as those used for the NE study watersheds except that
some parameters not appropriate for stream systems were not derived for the SBRP dataset. Table 5-
38 lists parameters included in the SBRP hydrologic indices database.
As with the NE, data collection was mainly from 7.5' and 15' topographic maps. All map
measurements for the SBRP were made in-house at ERL-C, and area measurement data were obtained
using the GIS. Stream channels were estimated based on USQS perennial stream delineations and field
checks by SCS soils mappers (see Section 5.4.2.8.3.2).
To ensure consistency, all SBRP measurements were made by the same individual. We conducted
an independent check on all map measurements and digitized information. Any measurement or data
entry discrepancies were checked and corrected as appropriate.
5-218
-------
Table 5-38. Mapped and calculated geomorphic parameters collected for the SBRP
study sites.
Parameter
Description
Calculated
AVG_EL
B_SHAPE
B_WIDTH
COMPACT
DDENSITY
Average elevation;
(MAX_ELEV + MIN_ELEV)/2
Basin Shape ratio;
B_LEN 2/WS_AREA
Average basin width;
WS_AREA/B_LEN
Compactness Ratio; ratio of perimeter
of basin to the perimeter of a circle
with equal area;
(PERIM)/(2*( *AW)'5)
Drainage Density;
TOTSTRM/WS_AREA
Units
Measured
B_CENT Drainage basin centroid expressed as
an X,Y coordinate
BJJEN Length of drainage basin; air-line km
distance from basin outlet to farest
upper point in basin
B_PER1M The length of the line which defines km
the surface divide of the drainage
basin
MAX_EL Elevation at approx. highest point m
MIN_EL Elevation at watershed outlet m
SUB_BAS(n) Area of each sub-catchments in the km2
~ drainage basin
STRMORDER Maximum stream Order (Horton) of streams
in the watershed (aerial photos used to aid
in reducing cooling problems between 7.5 and
15 minute maps)
TOTSTRM Total stream length; perennial km
WS AREA Total watershed area km2
m
km
continued
5-219
-------
Table 5-38.
study sites.
Parameter
Mapped and calculated geomorphic parameters collected for the SBRP
(cont)
ELONG
MAX_REL
M_PATH
REL_RAT
ROTUND
TOT_DD
WM_PATH
Additional
R
Description
Elongation Ratio;
(4 * WS_AREA)/L_BEN
Maximum relief;
MAX_ELEV - MIN_ELEV
Estimate of mean flow path;
Relief Ratio;
(MAX_ELEV-MIN_ELEV)/B_LEN
Rotundity Ratio;
{B_LEN)V(4 * WS_AREA)
Estimated drainage density based on
crenulations identified on topo map
Estimate of weighted mean flow
path;
Average annual runoff; interpolated
to each site from Krug et al. (1988)
runoff map
Units
m
m
m
cm
5-220
-------
SECTION 6
REGIONAL POPULATION ESTIMATION
6.1 INTRODUCTION
The purpose of this section is to describe the procedures used to extrapolate analyses on individual
watersheds to the target populations in the study regions. This process of extrapolation is called
population estimation.
6.2 PROCEDURE
6.2.1 Use of Variable Probability Samples
Probability samples were selected for lake watersheds in the Northeast and stream watersheds in
the Southern Blue Ridge Province (SBRP). Any quantity that can be defined for a sample unit (i.e., for
each watershed) can be extended to a corresponding population quantity through the probabilistic
structure of the sample. The quantity can be a measured variable or a model-based estimate. It can
be a number, a vector, or a function. In the Eastern Lake Survey (ELS), most quantities were measured
values, and the measurement error tended to be small relative to the sampling variation. In contrast to
the ELS, many of the quantities produced in the DDRP are model outputs believed to have significant
uncertainty associated with them. The population estimation techniques provided below apply to any
probability sample with defined inclusion probabilities. Thus, they are applicable to any identifiable subset
of the DDRP sample. Explicit provision is made for including uncertainty associated with the quantity
that is extended to the regional population.
In the ELS and, hence, the DDRP, the size of the target population is not precisely known. The
sampling frame for the ELS consisted of designated lakes on USQS maps. In some cases during field
sampling in the ELS, a field visit to the sample lakes selected from this frame indicated that some water
bodies designated as lakes on the map actually were not lakes, but rather marshes or old beaver ponds,
for example. When these "non-lakes" were subsequently excluded from the sample, a similar proportion
6-1
-------
of lakes also had to be excluded from the target population, effectively reducing its size. Thus, the size
of the target population is estimated from the sample size. This presents no particular difficulty as long
as each unit in the sample has a known inclusion probability.
The design of the surface water surveys and the DORP also permits arbitrary subsetting of the
sample. In some cases, the subsetting would correspond to a redefinition of the target population (e.g.,
the exclusion of seepage lakes). In such cases, the Inclusion probabilities for the remaining sample units
do not change, which, as can be seen from Equation 6-1 below, implies a smaller target population.
In other cases, the subset should be viewed as a subsample. In these cases, a smaller sample is being
used to make an inference about the same target population, and the inclusion probabilities do change.
This might occur if a selected lake could not be sampled or simulated for some reason. Inferences can
still be made about the same target populations, but the inclusion probabilities would change.
6.2.2 Estimation Procedures for Population Means
The structure of the DORP sample is almost identical to the structure of the ELS Phase II sample.
The differences are primarily in the conditional probability of inclusion in the second phase of the sample:
the DDRP sample was reduced by exclusion of lakes with large watersheds and the Phase II sample was
reduced at random. The estimation procedures are parallel to those detailed in the ELS Phase II Data
Analysis Plan (Overton,1987). Let n be the size of the sample selected from the target population, let
PI be the probability that sample unit I was included in the sample, and let Py be the joint inclusion
probability of units t and j. For sample unit i, let yf be the "true" quantity, and let z; be the observed
quantity, i.e., the unknown true value with an associated error e,. The error may be an observation error
or a measurement error; it could also be a prediction error. In each case we assume that the
characteristics of the error distribution are known, and that the uncertainty in the observed values is
characterized by that error distribution. The basic estimation procedures will follow the Horvttz-Thompson
estimator (Cochran, 1977) for variable probability samples; some details, however, will depend on
assumptions made about the observation error. Several distinct error models are treated below.
6-2
-------
In one case, the uncertainty is due to an additive error term, so that the magnitude of the
uncertainty is constant over the range of the response. The observation is related to the true value
through the equation Z| = y, + e s. Two distributions were available to handle this case: the error term
was assumed to have either a normal distribution with mean 0 and variance a2 or a uniform distribution
over the interval (-a,a). For this uniform distribution, the mean is 0 and a2 = a2/3.
In a second case, the magnitude of the uncertainty depends on the magnitude of the response.
This can be modelled with a multiplicative error term, where the uncertainty is proportional to the
response, so that z( = y; e;. We assumed that the uncertainty followed a log-normal distribution with
a mean value of 1 and a variance a2 = RSD2, where RSD was the relative standard deviation.
An implication of the above multiplicative model is that the uncertainty goes to 0 along with the
response. In some instances, however, there was appreciable uncertainty even when the response was
0. For these cases, we assumed that the uncertainty was proportional to the sum of the response plus
an offset (h), so that the observation equation was z, = y, + (y, + h)e j = yf (e( + 1) + he,. The mean
value of the error term was 0, and the a2 = RSD2. As above, a log-normal distribution was used for
this case.
The error structure affects only the variance of the population total, the variance of the population
mean, and the estimator of the cumulative distribution function and its associated variance. The estimator
of the target population size and population total take the same form under all of the above error
structures.
/v
Estimator of population total, T :
T = S Zj/p, (Equation 6-1)
6-3
-------
Estimator of the size of the target population, N.
N = "s 1/pr (Equation 6-2)
Estimator of population average, 7:
? = T/N. (Equation 6-3)
Both t and N are random variables, and both are unbiased estimators of the respective population
quantities. However, ?, similar to most ratio estimators, is a slightly biased estimator of the population
average.
6.2.3 Estimators of Variance
/\
For all three error models, the estimator of the variance of T has the form
VarfT> • f .0 ' Pj)2! -t- 2&V (Pii ' PjPj)zizi) + g(e,z) (Equation 6-4)
""""
where g(e,z) is a function that depends on the error model and the sample data. For the additive model,
g(e,2) =* a2 N ; for the multiplicative model, g(e,z) = az £z,2/Pj. and for the multiplicative model with
offset, g(e,z) = CT2 s(z( + h)2/D|, where h is the offset.
The variance of N is estimated by
Var(N) = 3 JLlPj) + $£ ( (Pii ' PiPj) ) + g(e,z) (Equation 6-5)
6-4
-------
The joint inclusion probabilities PJ, are determined by the structure of the DDRP sample. They are
computed according to the algorithm in the ELS Phase II Analysis Plan (Overton, 1987).
Finally, the variance of the estimator of the population average was obtained from a first-order variance
propagation using Equations 6-4 and 6-5:
Var(Y) = Var(T)/N2 + T^VarM/N4 - TCov(f,N)/N2, (Equation 6-6)
where
Cov{T,N) = X zftpfl - pdXl/p, - l/pj)(z,/p? - z,/pf)
i i>j
Confidence intervals will be derived from the usual normal theory, e.g., a 95 percent Cl on the population
average is given by
Y ± 1.967 Var(Y),
6.2.4 Estimator of Cumulative Distribution Function
Let N(y) be the total number in the population with the value of Y less than or equal to y, so that
the cumulative distribution function of Y is F Y (y) = N(y)/N, An estimator of N(y) is
Nz(y)
where
6-5
-------
An estimator of the cumulative distribution function of Y is
FY(y) = N(y)/N
A
The van'ance of FY has both a sampling component and a component due to measurement uncertainty.
The variance of the N(y) and covariance of N(y) and N are needed to calculate the sampling variance
of Fy. These are given by
Var(N(y» = FY(y)(1-FY(y))2 1/pf + FY(y)Var(N)
and
Cov(N,N
-------
In the DDRP, several techniques have been used to propagate uncertainty through a functional
relationship (which could be a complex simulation model as well as an explicit function). Let f(xt ,x2,...,
xn) be a function of the variables x^x,,,... ^ with uncertainties et e2 en, respectively. The
probability distributions (or at the least the variances) of the uncertainties are presumed known. If the
functional relationship is such that partial derivatives can be easily obtained, then the variance of
functional values can be estimated using a first-order linear approximation to the functional relationship:
Var(f) = 2(af/ax,)2a?
In the case of a simulation model, the function is the model itself, and the partial derivatives cannot
be calculated explicitly. An approximation to the partials can be obtained by perturbing the Xj's in turn.
If a suitably small perturbation is chosen, then the ratio of the change in output to the perturbation is an
estimate of the partial derivative. These estimates can then'be used in a first-order propagation as above.
A disadvantage of both of the above techniques is that they ignore possible correlations among
the uncertainties. One way to account for such correlations is to propagate not only variances but also
covariance terms. The "first-order, second-moment* technique used in the Enhanced Trickle Down
uncertainty analysis Is a means of doing exactly that. A first-order approximation is made to the model,
and Kalman filtering techniques are used to build up an estimate of the state variable variance-covariance
matrix. A final method that was used in uncertainty assessment was Monte Carlo. The Monte Carlo
method is applied by repeatedly calculating the value of f, each time perturbing the value of each x, by
a random quantity drawn from the respective uncertainty distribution. Monte Carlo is most easily applied
when uncertainties are statistically independent, but can also be applied when correlations exist. A variant
of Monte Carlo, called "fuzzy optimization", was used in the uncertainty analyses for the Model of
Acidification of Groundwater in Catchments.
6-7
-------
6.4 APPLICABILITY
I
This section discusses the procedures for the Level I, II, and III population estimation approaches
for DORP, including the statistical formulas that will be used to estimate population means, variances, and
cumulative frequency distributions. The population estimation procedures are generic and do not depend
on the level of analysis. The specific target populations for inference, however, do depend on the
analyses performed. Not all DDRP watersheds were used at each level of analysis so the target
population will vary. The explicit target populations being considered in the analysis are discussed In
Sections 8, 9, and 10. The generic uncertainty estimation procedures introduced In this section also
are more explicitly discussed for each of the individual analyses in Sections 8, 9, and 10.
6-8
------- |