United States
Environmental Protection
Agency
Office of Acid Deposition, EPA/600/3-88/021 a
Environmental Monitoring and June 1988
Quality Assurance
Washington DC 20460
Research and Development
Chemical Characteristics of
Streams in the Mid-Atlantic
and Southeastern United
States (National Stream
Survey-Phase I)
Volume I: Population
Descriptions and Physico-
Chemical Relationships
S-*U «,'O- '%** '^ l^i* * '«' -^ V
^iJi.^W''*^'**™*'**"1
i1tf^^**ffe*?lr'-. ...- "^^^r-^ T*«^^ff
^^•.T-"; ," .'"'.-"' •--'' .'•"•'-.I-..-1-'8*-^ - •.'"'**:-" 'p^wfff""1-"-".",-";--'"'j
-------
UJ
CO
<
X
Q.
UJ
CO
UJ
CC
UJ
X
U-
o
CO
o
o
UJ
cc
m
CO
-------
EPA/600/3-88/021a
June 1988
Chemical Characteristics of Streams
in the Mid-Atlantic and
Southeastern United States
(Results of the National Stream Survey - Phase I)
Volume I: Population Descriptions
and
Physico-Chemicai Relationships
A Contribution to the
National Acid Precipitation Assessment Program
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
Environmental Research Laboratory - Corvallls, OR 97333
Environmental Monitoring Systems Laboratory - Las Vegas. NV 89119
-------
NOTICE
The research described in this document has been funded wholly or in part by the U.S.
Environmental Protection Agency under Contract No. 68-03-3249 to Lockheed Engineering and
Management Services Company, Inc., Contract Nos. 68-02-3889 and 68-02-3994 to Radian
Corporation, Contract Nos. 68-03-3246 and 68-C8-0006 to Northrop Services, Inc.; under
Interagency Agreement No. EPA#DW89931368; DOE# 1824-1557-A1 with the U.S. Department
of Energy (Contract No. DE-AC05-84OR21400 with Martin Marietta Energy Systems, Inc.); and
by cooperative agreements with Utah State University (CR812049) and Oregon State University
(CR813061). It has been subject to the Agency's peer and administrative review, and has been
approved for publication as an EPA document.
Mention of corporation names, trade names, or commercial products does not constitute
endorsement or recommendation for use.
Proper citation of this document is as follows:
Kaufmann, P.R.1, A.T. Herlihy1, J.W. Elwood2, M.E. Mitch1, W.S. Overton3, MJ. Sale2, JJ.
Messer4, K. A. Cougan5, D.V. Peck5, K.H. Reckhow6, A.J. Kinney7, SJ. Christie7, D.D.
Brown7, C.A. Hagley5, and H.I. Jager8. 1988. Chemical Characteristics of Streams in the
Mid-Atlantic and Southeastern United States. Volume I: Population Descriptions and
Physico-Chemical Relationships. EPA/600/3-88/02la. U.S. Environmental Protection
Agency, Washington, D.C.
Sale, MJ.2, P.R. Kaufmann1, H.I. Jager8, J.M. Coe2, K.A. Cougan5, A.J. Kinney7, M.E. Mitch1,
and W.S. Overton3. 1988. Chemical Characteristics of Streams in the Mid-Atlantic and
Southeastern United States. Volume II: Streams Sampled, Descriptive Statistics, and
Compendium of Physical and Chemical Data. EPA/600/3-88/021b. U.S. Environmental
Protection Agency, Washington, D.C.
Inquiries regarding the availability of the National Stream Survey - Phase I Mid-Atlantic
and Southeast data base should be directed in writing to:
Chief, Watershed Branch
U.S. Environmental Protection Agency
Environmental Research Laboratory
200 SW 35th Street
Corvallis, Oregon 97333
1 Utah State University, Utah Water Research Laboratory, Logan, Utah 84322. Present address: U.S. Environmental
Protection Agency, Environmental Research Laboratory, 200 SW 35th Street, Corvallis, Oregon 97333.
2 InVironmentll Sconces Division, Oak Ridge National Laboratory, P.O BoxX, Bldg. 1505, OakRidge Tennessee
37831. Operated by Martin Marietta Energy Systems, Inc., under Contract No. DE-AC05-84OR21400 for the U.S.
8 OwKonTtate^niversity, Department of Statistics, Kidder Hall No. 8, Corvallis, Oregon 97331.
* Utah StateUniversity, Utah Water Research Laboratory, Logan, Utah 84332. Present address: U.S. Environmental
Protection Agency, Environmental Monitoring Systems Laboratory, Mail Drop 39, Research Triangle Park, North
5 Lockheed Engineering and Management Services Company, Inc., Las Vegas, Nevada, 89119.
6 Duke University School of Forestry, 105 Biological Sciences Bldg., Durham, North Carolina 27706.
7 Northrop Services, Inc., U.S. Environmental Protection Agency, 200 SW 35th Street, Corvallis, Oregon 97333.
8 See AppHcatTons International Corporation, Environmental Science Division, Oak Ridge National Laboratory, P.O.
Box X, Bldg. 1505, Oak Ridge, Tennessee 37831.
-------
PREFACE
The National Surface Water Survey (NSWS) was initiated in 1983 by the U.S. Environmental
Protection Agency (EPA) under the National Acid Precipitation Assessment Program (NAPAP), a
federal interagency task force mandated by Congress in 1980. The NSWS is a phased, systematic
study designed to: (1) characterize the present chemistry of surface waters in the United States
and classify them for more intensive study, (2) describe chemical temporal variability and
biological resources in subsets of surface waters, and (3) provide a foundation for documenting
trends in surface water chemistry through long-term monitoring. The NSWS is one of several
major projects in the Acid Deposition Aquatic Effects Research Program (AERP). This program,
one of many research programs investigating acidic deposition, is administered by the Acid'
Deposition and Atmospheric Research Division: Office of Acid Deposition, Environmental
Monitoring and Quality Assurance, in the U.S. EPA Office of Research and Development.
The AERP addresses four primary assessment questions:
1.
2.
3.
4.
How extensive is the change in aquatic resources as a result of current levels of
acidic deposition?
What is the anticipated extent and rate of change to these resources in the future?
What levels of change in sensitive surface waters are associated with various rates of
acidic deposition?
What is the rate of change or recovery of affected systems, given decreases in acidic
deposition rates?
Five major research projects within the AERP specifically address these assessment
questions from a regional perspective. These projects and their goals are:
1.
2.
5.
National Surface Water Survey (NSWS): To determine the present chemistry,
characterize the chemical temporal variability, and determine the key biological
resources of lakes and streams in potentially sensitive regions of the United States.
Direct/Delayed Response Project (DDRP): To predict changes in these aquatic
resources at various levels of acidic deposition, considering the terrestrial and aquatic
variables that influence these changes.
Watershed Manipulation Project (WMP): To verify that predictions of future change
are reasonably sound by manipulating watershed catchments or system components.
Episodic Response Project (ERP): To evaluate the regional importance of short-term
acidification resulting from episodic hydrologic events and its effect on the quality of
the biological environment.
Temporally Integrated Monitoring of Ecosystems Project (TIME): To test the validity
of predicted changes through long-term monitoring of regionally characteristic lakes
and streams.
The NSWS, including surveys of lakes and streams, addresses the first goal of the AERP.
Understanding the national-scale effects of acidic deposition on aquatic resources requires that
111
-------
the present chemical status of surface waters be understood on large geographical scales. The
National Stream Survey - Phase I (NSS-I), conducted in the mid-Atlantic and southeastern
United States, was designed to describe, statistically, the present surface water chemistry on a
regional scale. Although cause-and-effect relationships between acidic deposition and surface
water response cannot be determined on the basis of NSS-I data alone, analysis of correlative
relationships within the NSS-I data base and other information do further our understanding
(and will continue to do so) of the extent to which mid-Atlantic and southeastern streams are at
risk due to acidic deposition. Determining the relationship between acidic deposition and surface
water chemistry is the goal of future activities within the AERP.
The National Stream Survey is funded and administered by the U.S. EPA, Office of Acid
Deposition, Environmental Monitoring, and Quality Assurance (OADEMQA) in Washington, D.C.
The Environmental Research Laboratory - Corvallis (ERL-C), with cooperating contracts with
Utah State University, Oregon State University, and Northrop Environmental Services, is
responsible for coordinating the activities of the Survey and for project design, data validation,
geographic analysis, and data interpretation. The Environmental Monitoring Systems Laboratory
- Las Vegas (EMSL-LV) is responsible for quality assurance and quality control, logistics, and
analytical support. Oak Ridge National Laboratory (ORNL) is responsible for developing and
maintaining the data base management system for the Survey. ORNL also participated in data
interpretation, provided statistical programming to implement the target population characteri-
zations, and did mapping and other geographic analysis.
IV
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
ENVIRONMENTAL RESEARCH LABORATORY
200 S.W. 35TH STREET
CORVALLIS, OREGON 97333
August 3, 1988
We are pleased to provide you with a complimentary copy of the enclosed report
entitled:
"Chemical Characteristics of Streams in the Mid-Atlantic and Southeastern United
States (Results of the National Stream Survey-Phase I)"
Volume I: Population Descriptions and Physico-Chemical Relationships
Volume II: Streams Sampled, Descriptive Statistics, and Compendium of Physical
and Chemical Data
This report is a product from our National Surface Water Survey (NSWS), first
implemented by the U.S. EPA in 1984 under the auspices of the National Acid
Precipitation Assessment Program with a survey of Eastern lakes. The enclosed
report describes the design and results of National Stream Survey field
activities in the Mid-Atlantic and Southeastern U.S. during the spring of 1986
and summarizes our current interpretation of these results. Data obtained during
the 1985 National Stream Survey pilot in the Southern Blue Ridge Mountains are
also included in this data analysis and interpretation.
We hope that you will find the report useful and look forward to receiving any
comments you may wish to share with us. The availability of the National Stream
Survey database will be announced in the AERP status. a publication of the
Aquatic Effects Research Program, in the fall of 1988. To obtain copies of the
AERP status. contact:
Wes Kinney
U.S. EPA Environ. Monit. Systems Lab.
P.O. Box 93478
Las Vegas, Nevada 89193-3478
(702)798-2358 (FTS 545-2358)
Sincerely,
Deb Chaloud
Lockheed Engr. and Sci. Co.
or 1050 E. Flamingo Rd. , Suite 209
Las Vegas, Nevada 89119
(702)734-3227
Dixon H. Landers
Research Director
National Surface Water Survey
-------
-------
ERRATA
August 29, 1988
Please note the following changes to the two-volume EPA report, "Chemical Characteristics of
streams in the Mid-Atlantic and Southeastern United States (National Stream Survey-Phase I)":
Volume I
1. On p. 31, Table 2-3, the subtotal of the number of site exclusions in subregion 2X should
be 90, NOT 91.
2. pp. 224, 308, 322, 330, and 388 are missing page numbers and should be labelled "THIS PAGE
INTENTIONALLY LEFT BLANK"
Volume II
1. On p. 61, section 5.1.2, Wjj is the inverse of the joint inclusion probability. The
equation for calculating Wij should read:
Wy = [2nWiWj - (Wi+Wj)] / 2(n-l).
2. An incorrect subregion prefix has been used in the stream identification code (STRM_ID) in
some of the data listings for the special interest stream North Fork of Ben's Creek. This
stream is located in subregion 2Cn, NOT 2Bn. The correct STRM_ID for the North Fork of
Ben's Creek is 2C035913L, as given in Tables 4-4 and 4-5 on pp. 56 and 57. It has been
INCORRECTLY labelled as STRM_ID 2B035913L in the special interest site physical and
chemical data listings in Tables 6-2, 7-5, 7-6, 7-7, and 7-8 (pp. 430, 586, 588, 590, and
592). The data in the tables is correct. The site map in Fig. 3-4 shows the correct
location of the North Fork of Ben's Creek (#35913) in subregion 2Cn. However, it should
have been put on the subregion 2Cn map page (Fig. 3-3, p. 14) rather than on the subregion
2Bn map page (fig. 3-4, p. 15).
3. pp. 22, 394, 432, and 594 are missing page numbers and should be labelled "THIS PAGE
INTENTIONALLY LEFT BLANK"
-------
-------
TABLE OF CONTENTS
Section
Page
Notice ii
Preface iii
List of Illustrations x
List of Tables xxi
Volume II Table of Contents xxiv
Related Documents xxv
Acknowledgements xxvi
EXECUTIVE SUMMARY xxxiii
SECTION 1 - INTRODUCTION . .1
1.1 Overview 1
1.2 National Surface Water Survey .1
1.3 National Stream Survey 4
1.3.1 Phase I Goals and Objectives 4
1.3.2 National Stream Survey Components 5
1.3.2.1 NSS-I Pilot Survey 7
1.3.2.2 Mid-Atlantic Phase I Survey 8
1.3.2.3 Southeast Screening 8
1.3.2.4 Mid-Atlantic Episodes Pilot 9
1.3.3 The NSS-I Data Report 9
SECTION 2 - SURVEY DESIGN 11
2.1 Overview 11
2.2 Project Design Criteria 12
2.3 Defining the Resource at Risk 13
2.3.1 Regions of Interest 13
2.3.2 Regional Prioritization 16
2.3.3 Identifying the Stream Resource of Interest 24
2.3.4 Alternative Methods for Identifying the Target Population 25
2.4 Statistical Sampling Design 26
2.4.1 The Stage I Sample 26
2.4.2 Site Inclusion Criteria (Site Rules) 29
2.4.3 The Stage I Data 29
2.4.4 Stage I Estimates 32
2.4.5 The Stage II Sample 34
2.4.6 Stage II Estimates 37
2.5 Index Sampling 40
2.6 Special Interest Sites 41
SECTION3-METHODS.. . / 43
3.1 Overview 43
3.2 Field Sampling Plan 43
-------
3.3 Field Methods 51
3.3.1 Site Characteristics 51
3.3.2 Sample Collection 51
3.3.3 In Situ Measurements 51
3.4 Sample Handling 52
3.5 Processing Laboratory Techniques 52
3.6 Analytical Laboratory Support 54
3.7 Quality Assurance and Quality Control Protocols 54
3.7.1 Quality Assurance Plan 54
3.7.2 Quality Assurance and Quality Control - Data Collection and Analysis . . 54
3.7.3 Training and Site Audits 59
3.8 Data Base Management 59
3.8.1 Raw Data (Data Set 1) 60
3.8.2 Verification (Data Set 2) 60
3.8.3 Validation (Data Set 3) 64
3.8.3.1 Episode Identification 68
3.8.3.2 Acid Mine Drainage 68
3.8.3.3 Flagging of Unusual Values or Sites 68
3.8.3.4 Geographic Data Verification and Validation . . . . : 69
3.8.4 Enhanced Data (Data Set 4) 69
3.9 Differences Between the NSS-I and the Pilot Survey 72
3.9.1 Sample Holding Times - 24 Hours (NSS-I) vs. 12 Hours (Pilot Survey) . . 72
3.9.2 Processing Laboratory Location 72
3.9.3 Field pH Measurement 72
3.9.4 Methods of Fractionation and Determination of Aluminum Species .... 72
3.9.5 Matrix Spike Quality Assurance Samples 74
3.9.6 Total Dissolved Phosphorus (NSS-I) vs. Total Phosphorus (Pilot) 74
3.9.7 Specific Conductance Measured in the Processing Laboratory for NSS-I . 74
SECTION 4 - DATA QUALITY ASSESSMENT. 75
4.1 Introduction 75
4.2 Completeness 75
4.3 Comparability 76
4.4 Representativeness 76
4.5 Detectability 77
4.5.1 Method Level Detectability 77
4.5.2 System Level Detectability 77
4.6 Accuracy. 79
4.6.1 Accuracy Within Laboratories 79
4.6.2 Pooled Accuracy Estimates 79
4.7 Laboratory Bias 81
4.8 Precision 82
4.9 Chemical Data Quality Overview 87
4.9.1 Charge Balance 87
4.9.2 Calculated vs. Measured Conductance 91
4.9.3 Calculated vs. Measured ANC 94
VI
-------
SECTION 5 - TARGET POPULATION—PHYSICAL CHARACTERISTICS ...... 97
5.1 Overview 97
5.2 Refinement of the Target Population ..97
5.3 Total Target Population Resource Estimates 100
5.4 Target Reach Physical Characteristics 101
5.5 Geographic Classification of Sample Sites 106
SECTION 6 - TARGET POPULATION REGIONAL CHEMISTRY . . 113
6.1 Interpreting NSS-I Population Distributions 113
6.1.1 Target Population of Interest 113
6.1.2 NSS-I Subregions 113
6.1.3 The Florida Subregion .113
6.1.4 The Southern Blue Ridge Subregion .114
6.1.5 Index Sample . ........ .114
6.1.6 Distributions Based on Upstream and Downstream Chemistry . ...... .114
6.1.7 Interpolated Length Distributions 116
6.2 Distributions of pH and ANC 116
6.2.1 Regional Overview 116
6.2.2 Mid-Atlantic Region 138
6.2.3 Interior Southeastern Region 147
6.2.4 The Florida Subregion 148
6.2.5 Interpolated Length Distributions 149
6.3 Distributions of Other Chemical Variables 153
6.3.1 Base Cations and Conductivity 153
6.3.2 Aluminum 155
6.3.3 Sulfate 157
6.3.4 Nitrate 172
6.3.5 Chloride 173
6.3.6 Dissolved Organic Carbon 177
6.3.7 Other Chemical Variables 179
6.4 Small Streams 179
6.5 Uncertainty in Regional Estimates .186
SECTION 7 - TEMPORAL VARIABILITY IN BASEFLOW CHEMISTRY 191
7.1 Overview 191
7.2 Among-Year Variability 191
7.3 Among-Season Variability - The Index Time Period . 198
7.4 Within-Season Variability .204
7.4.1 Within-Season Variability in Special Interest Sites 204
7.4.2 Within-Season Variability in NSS-I Mid-Atlantic Sites . 209
7.5 Summary 222
SECTION 8 - ION RELATIONSHIPS 225
8.1 Overview 225
8.2 Base Cations and Mineral Acid Anions . .225
8.3 Aluminum 246
8.4 Sources of Variation in ANC of Stream Water 247
Vll
-------
8.5 Organic Anions 254
8.6 Sources of ANC Variation in Low ANC (< 200 /zeq L'1) Streams 257
8.7 Effects of Carbonic Acid on pH - (CO2 Effects) 261
SECTION 9 - NATURE AND DISTRIBUTION OF ACIDIC AND LOW ANC
STREAMS 265
9.1 Overview. 265
9.2 Background 265
9.3 Classification and Location of Acidic and Low ANC Streams 267
9.3.1 Streams with Acid Mine Drainage and Substantial Watershed Sources
of Sulfate 267
9.3.1.1 Classification, 267
9.3.1.2 Location 270
9.3.1.3 Chemical Characteristics 270
9.3.2 Streams with Organic Dominance 273
9.3.2.1 Classificatioa 273
9.3.2.2 Location .273
9.3.2.3 Chemical Characteristics 273
9.3.3 High-Interest Subpopulatioa 273
9.3.3.1 Classificatioa 273
9.3.3.2 Location 277
9.3.3.3 Chemical Characteristics 277
9.3.4 Discussion 284
9.4 Population Estimates of the Probable Sources of Acidity 284
9.4.1 Interpolated Length Estimates 290
9.5 Geographic Distribution of Stream Chemical Groupings 293
9.5.1 Streams with Watershed Sulfate Sources 302
9.5.2 Streams with Chemistry Dominated by Organics 302
9.5.3 High-Interest Subpopulation (Inorganic and Organic-Influenced
Streams) 302
9.5.3.1 Allegheny Plateau, Northeast Mid-Atlantic, and
Valley and Ridge Geographic Sites 302
9.5.3.2 Florida, Arkansas/Oklahoma, Piedmont, and Southern
Appalachian Highland Geographic Sites 303
9.5.3.3 Coastal Plain Geographic Sites 303
9.6 Classification of Special Interest Sites 304
SECTION 10 - EVIDENCE OF ACIDIFICATION 309
10.1 Overview 309
10.2 Ion Balance and ANC Relationships 309
10.3 Atmospheric Sulfate Deposition and Stream Water Chemistry 314
SECTION 11 - SUMMARY AND CONCLUSIONS 323
11.1 Background 323
11.2 Objectives 324
11.3 Design 324
Vlll
-------
11.4 Selected Results 324
11.4.1 Regional Chemical Characteristics 324
11.4.2 Chemical Relationships 326
11.4.3 Sources of Acidity in Acidic and Low ANC Streams 327
SECTION 12 - REFERENCES 331
SECTION 13 - GLOSSARY. 345
13.1 Abbreviations and Symbols 345
13.1.1 Abbreviations 345
13.1.2 Symbols 347
13.2 Definitions 349
APPENDIX A - Data Quality Assessment Tables 361
APPENDIX B - Prediction of Spring Upper Reach Node Chemistry in the
Southern Blue Ridge Subregion (2As) Data Set (Pilot Survey) . . .381
APPENDIX C - Population Estimates and Standard Errors for ANC and
pH Reference Ranges 383
APPENDIX D - Standard Errors of the Geographic and Chemical
Classification Population Estimates 389
IX
-------
LIST OF ILLUSTRATIONS
Figure
1-1 Organization of the National Surface Water Survey, showing the two
major components, National Lake Survey and National Stream Survey,
and their relationship to later phases of study 3
1-2 Regions and subregions of the United States used to define target
populations of the National Surface Water Survey 6
2-1 Regions of the United States expected, on the basis of historical data,
to contain surface waters with alkalinity < 400 jueq L'1, shown with
National Surface Water Survey boundaries 14
2-2 NSS-I Mid-Atlantic and Southeast subregions, with atmospheric wet
sulfate deposition rates 15
2-3 Representation of the point frame sampling procedure for selecting
NSS-I Stage I reaches 28
2-4 NSS-I Stage II sample reaches and special interest sites in the
Mid-Atlantic subregiona 35
2-5 NSS-I Stage II sample reaches and special interest sites in the
Southeast subregions 36
2-6 Example of the type of cumulative distribution function plot used in the
NSS-I 39
3-1 Field sampling activities, NSS-I 50
3-2 Processing laboratory activities, NSS-I 53
3-3 QC sample flow for the NSS-I 58
3-4 NSS-I data base development 61
3-5 NSS-I data verification procedures 65
3-6 NSS-I data validation procedures 66
4-1 Charge balance plot for all NSS-I Mid-Atlantic and Southeast Screening
stream water samples 88
4-2 Relationship between anion deficit and DOC for all NSS-I stream
samples in Florida (3C) and the Mid-Atlantic Coastal Plain (3B) 90
-------
4-3 Charge balance plot for all NSS-I Mid-Atlantic and Southeast Screening
samples that include approximated equivalent concentrations for DOC,
Fe, Mn, and Al 92
4-4 Measured vs. calculated conductance of NSS-I Mid-Atlantic and
Southeast Screening stream water samples 93
4-5 Measured ANC vs. the calculated carbonate alkalinity for NSS-I stream
water samples 95
5-1 NSS-I target population refinement concept 98
5-2 Estimated percentages of refined target and noninterest stream reaches
in NSS-I subregions 99
5-3 Population frequency distributions for (a) drainage area (20th percentile,
median, 80th percentile) and (b) population frequency histograms,
Strahler Order (l:24,000-scale maps), at NSS-I target reach upper and
lower nodes 103
5-4 Population frequency distributions (20th percentile, median, 80th
percentile) for (a) depth and (b) width of NSS-I target reaches at upper
and lower nodes. 104
5-5 Population frequency distributions (20th percentile, median, 80th
percentile) for (a) upper and lower node elevation, and (b) gradient
between nodes of NSS-I target reaches 105
6-1 Conceptual representation of upper and lower reach node samples
and populations 115
6-2a Population distributions and comparison of index ANC at upper and
lower reach nodes in NSS-I subregion ID (Poconos/Catskills) 118
6-2b Population distributions and comparison of index ANC at upper and
lower reach nodes in NSS-I subregion 2Cn (Northern Appalachians) . . .119
6-2c Population distributions and comparison of index ANC at upper and
lower reach nodes in NSS-I subregion 2Bn (Valley and Ridge) 120
6-2d Population distributions and comparison of index ANC at upper and
lower reach nodes in NSS-I subregion 3B (Mid-Atlantic Coastal Plain) . .121
6-2e Population distributions and comparison of index ANC at upper and
lower reach nodes in NSS-I subregion 2As (Southern Blue Ridge) 122
6-2f Population distributions and comparison of index ANC at upper and
lower reach nodes in NSS-I subregion 3A (Piedmont) 123
xi
-------
6-2g Population distributions and comparison of index ANC at upper and
lower reach nodes in NSS-I subregion 2X (Southern Appalachians) . . . .124
6-2h Population distributions and comparison of index ANC at upper and
lower reach nodes in NSS-I subregion 2D (Ozarks/Ouachitas) 125
6-2i Population distributions and comparison of index ANC at upper and
lower reach nodes in NSS-I subregion 3C (Florida) 126
6-3a Population distributions and comparison of index pH at upper and
lower reach nodes in NSS-I subregion ID (Poconos/Catskills) 128
6-3b Population distributions and comparison of index pH at upper and
lower reach nodes in NSS-I subregion 2Cn (Northern Appalachians) . . .129
6-3c Population distributions and comparison of index pH at upper and
lower reach nodes in NSS-I subregion 2Bn (Valley and Ridge) 130
6-3d Population distributions and comparison of index pH at upper and
lower reach nodes in NSS-I subregion 3B (Mid-Atlantic Coastal Plain) . . 131
6-3e Population distributions and comparison of index pH at upper and
lower reach nodes in NSS-I subregion 2As (Southern Blue Ridge) 132
6-3f Population distributions and comparison of index pH at upper and
lower reach nodes in NSS-I subregion 3A (Piedmont). . 133
6-3g Population distributions and comparison of index pH at upper and
lower reach nodes in NSS-I subregion 2X (Southern Appalachians) . . . .134
6-3h Population distributions and comparison of index pH at upper and
lower reach nodes in NSS-I subregion 2D (Ozarks/Ouachitas) 135
6-3i Population distributions and comparison of index pH at upper and
lower reach nodes in NSS-I subregion 3C (Florida) 136
6-4 Population frequency distribution (Q20» median, and Qgo) of index
(a) ANC and (b) pH, at upper and lower reach nodes in the NSS-I
subregions 139
6-5 Geographic distribution of NSS-I lower node index ANC 143
6-6 Geographic distribution of NSS-I upper node index ANC 144
6-7 Geographic distribution of NSS-I lower node index pH 145
6-8 Geographic distribution of NSS-I upper node index pH 146
xn
-------
6-9
6-10
6-11
6-12
6-13a
6-13b
6-13c
6-13d
6-13e
6-13f
6-13g
Combined length of NSS-I target stream resources within specified
categories of (a) ANC and (b) pH, as estimated by linear interpolation
of index concentrations between upper and lower reach node sampling
locations
Population frequency distribution (Q20> median, and Qgo) of index
(a) conductivity and (b) base cation sum, at upper and lower reach
nodes in the NSS-I subregions.
150
154
Population frequency distributions (Q20> median, and Qgo) of index
(a) total aluminum and (b) total monomeric aluminum concentrations
(MIBK-extractable) at upper and lower reach nodes in the NSS-I
subregions .............................
Population frequency distribution (Q20» median, and Qgo) of index
(a) sulfate and (b) nitrate concentrations at upper and lower reach
nodes in the NSS-I subregions. ...................
Population distributions and comparison of index sulfate concentra-
tions at upper and lower reach nodes in NSS-I subregion ID
Poconos/Catskills), presented as inverse cumulative proportions . . .
156
159
160
Population distributions and comparison of index sulfate concentra-
tions at upper and lower reach nodes in NSS-I subregion 2Cn
(Northern Appalachians), presented as inverse cumulative proportions. . .161
Population distributions and comparison of index sulfate concentra-
tions at upper and lower reach nodes in NSS-I subregion 2Bn
(Valley and Ridge), presented as inverse cumulative proportions 162
Population distributions and comparison of index sulfate concentra-
tions at upper and lower reach nodes in NSS-I subregion 3B
(Mid-Atlantic Coastal Plain), presented as inverse cumulative
proportions
Population distributions and comparison of index sulfate concentra-
tions at upper and lower reach nodes in NSS-I subregion 2As
(Southern Blue Ridge), presented as inverse cumulative proportions
Population distributions and comparison of index sulfate concentra-
tions at upper and lower reach nodes in NSS-I subregion 3A
(Piedmont), presented as inverse cumulative proportions
163
. 164
165
Population distributions and comparison of index sulfate concentra-
tions at upper and lower reach nodes in NSS-I subregion 2X
(Southern Appalachians), presented as inverse cumulative
proportions
166
Xlll
-------
6-13h
6-131
6-14
6-15
6-16
6-17
6-18
6-19
6-20
6-21
6-22
6-23
7-1
7-2
Population distributions and comparison of index sulfate concentra-
tions at upper and lower reach nodes in NSS-I subregion 2D
(Ozarks/Ouachitas), presented as inverse cumulative proportions 167
Population distributions and comparison of index sulfate concentra-
tions at upper and lower reach nodes in NSS-I subregion 3C (Florida),
presented as inverse cumulative proportions 168
Geographic distribution of NSS-I lower node index sulfate
concentrations 169
Geographic distribution of NSS-I lower node index nitrate
concentrations 174
Population frequency distribution (Q2Q» median, and Qgo) of index
(a) chloride and (b) DOC concentrations at upper and lower reach
nodes in the NSS-I subregions. 175
Geographic distribution of NSS-I lower node index chloride
concentrations 176
NSS-I index chloride concentration versus distance from the ocean
in Mid-Atlantic sites within 200 km of the sea coast 178
Geographic distribution of NSS-I lower node index DOC
concentrations 180
NSS-I spring index ANC Qteq L"1) vs. watershed area (km2)
in headwater reaches in the Interior Mid-Atlantic 185
The population of small streams upstream of NSS-I blue-line
headwater reaches 187
Population distributions comparing ANC in the NSS-I target popula-
tion with ANC in a subpopulation including only headwater reaches:
(a) interior Mid-Atlantic subregions and (b) the Mid-Atlantic Coastal
Plain 188
Population distributions comparing ANC in the NSS-I target popula-
tion with ANC in a subpopulation including only headwater reaches:
(a) interior Southeast subregions and (b) Florida subregion 189
pH in Fernow Control during the spring index period from 1980
to 1987
193
ANC (peq L"1) in Fernow Control during the spring index period from
1980 to 1987 194
xiv
-------
7-3
7-4
7-5
7-6
7-7
7-8
7-9
7-10
7-11
7-12
Sulfate concentration (/zeq L'1) in Fernow Control during the spring index
period from 1980 to 1987 195
Sum of base cations (jueq IT1) in Fernow Control during the spring index
period from 1980 to 1987 196
The pH observed in 1987 during the acid stream reconnaissance vs.
the 1986 NSS-I spring index field pH value in 37 mid-Atlantic
stream nodes 197
Conductivity observed in 1987 during the acid stream recon-
naissance versus the 1986 NSS-I spring index in-situ conductivity
value in 37 Mid-Atlantic stream nodes .
199
The seasonal pattern in pH (mean plus 95% confidence interval) in
three streams in the Catskill Mountains of New York (Murdoch, 1986)
and three streams in the Laurel Hills of Pennsylvania (Witt and
Barker, 1986)
The seasonal pattern in ANC (mean plus 95% confidence interval) in
three streams in the Catskill Mountains of New York (Murdoch, 1986)
and three streams in the Laurel Hills of Pennsylvania (Witt and
Barker, 1986)
The seasonal pattern in sulfate concentration (mean plus 95%
confidence interval) in three streams in the Catskill Mountains
of New York (Murdoch, 1986) and three streams in the Laurel Hills
of Pennsylvania (Witt and Barker, 1986).
200
201
202
The seasonal pattern in the sum of base cations (mean plus 95%
confidence interval) in three streams in the Catskill Mountains
of New York (Murdoch, 1986) and three streams in the Laurel Hills
of Pennsylvania (Witt and Barker, 1986).
203
The pH range of NSS-I special interest site samples (1986) and
Long-Term Monitoring (LTM) Project samples (1985-1986) within
the spring index period (April 1 to May 15) in three streams in the
Catskill Mountains of New York (Murdoch, 1986) and three streams
in the Laurel Hills of Pennsylvania (Witt and Barker, 1986).....
205
ANC range of NSS-I special interest site samples (1986) and
Long-Term Monitoring (LTM) Project samples (1985-1986) within
the spring index period (April 1 to May 15) in three streams in the
Catskill Mountains of New York (Murdoch, 1986) and three streams
in the Laurel Hills of Pennsylvania (Witt and Barker, 1986)
206
xv
-------
7-13
7-14
7-15
7-16
7-17
7-18
7-19
7-20
7-21
7-22
Sulfate concentration range of NSS-I special interest site samples
(1986) and Long-Term Monitoring (LTM) Project samples (1985-1986)
within the spring index period (April 1 to May 15) in three streams
in the Catskill Mountains of New York (Murdoch, 1986) and three
streams in the Laurel Hills of Pennsylvania (Witt and Barker, 1986). .
The range in the sum of base cation concentrations in NSS-I
special interest site samples (1986) and Long-Term Monitoring (LTM)
Project samples (1985-1986) within the spring index period (April 1
to May 15) in three streams in the Catskill Mountains of New York
(Murdoch, 1986) and three streams in the Laurel Hills of Pennsyl-
vania (Witt and Barker, 1986)
207
ANC on first versus second visit to NSS-I sample reaches in the
Mid-Atlantic Region
pH on first versus second visit to NSS-I sample reaches in the
Mid-Atlantic Region
Median absolute changes in ANC (unweighted) between first and
second visits to NSS-I sample reaches in four subregions of the
Mid-Atlantic Region
Median absolute changes in pH (unweighted) between first and
second visits to NSS-I sample reaches in four subregions of the
Mid-Atlantic Region
Population distributions for ANC in the NSS-I Mid-Atlantic
Region based on first visit, second visit, and spring index
(average) measurements for (a) upper nodes and (b) lower nodes
Population distributions for pH in the NSS-I Mid-Atlantic
Region based on first visit, second visit, and spring index
(average) measurements for (a) upper nodes and (b) lower nodes
Population distributions for sum of base cations in the NSS-I
Mid-Atlantic Region based on first visit, second visit, and
spring index (average) measurements for (a) upper nodes and
(b) lower nodes
208
210
211
213
215
217
218
Population distributions for sulfate concentrations in the
NSS-I Mid-Atlantic Region based on first visit, second visit,
and spring index (average) measurements for (a) upper nodes and
(b) lower nodes
219
220
xvi
-------
8-la Trilinear plots of major cations and anions in stream water at
the upper and lower nodes combined in subregion ID
(Poconos/Catskills) 226
8-lb Trilinear plots of major cations and anions in stream water at
the upper and lower nodes combined in subregion 2Cn (Northern
Appalachians) 227
8-lc Trilinear plots of major cations and anions in stream water at
the upper and lower nodes combined in subregion 2Bn (Valley
and Ridge) .228
8-Id Trilinear plots of major cations and anions in stream water at
the upper and lower nodes combined in subregion 3B (Mid-Atlantic
Coastal Plain) 229
8-le Trilinear plots of major cations and anions in stream water at
the upper and lower nodes combined in subregion 2As (Southern
Blue Ridge) 230
8-If Trilinear plots of major cations and anions in stream water at
the upper and lower nodes combined in subregion 3A (Piedmont) 231
8-lg Trilinear plots of major cations and anions in stream water at
the upper and lower nodes combined in subregion 2X (Southern
Appalachians) 232
8-lh Trilinear plots of major cations and anions in stream water at
the upper and lower nodes combined in subregion 2D
(Ozarks/Ouachitas) 233
8-li Trilinear plots of major cations and anions in stream water at
the upper and lower nodes combined in subregion 3C (Florida) 234
8-2a Weighted average concentration of major anions and cations by
pH class of stream water at the upper and lower nodes in
subregion ID (Poconos/Catskills) 236
8-2b Weighted average concentration of major anions and cations by
pH class of stream water at the upper and lower nodes in
subregion 2Cn (Northern Appalachians) 237
8-2c Weighted average concentration of major anions and cations by
pH class of stream water at the upper and lower nodes in
subregion 2Bn (Valley and Ridge) 238
xvn
-------
8-2d
8-2e
8-2f
8-2g
8-2h
8-2i
8-3
8-4
8-5
8-6
8-7
Weighted average concentration of major anions and cations by
pH class of stream water at the upper and lower nodes in
subregion 3B (Mid-Atlantic Coastal Plain) ,
Weighted average concentration of major anions and cations by
pH class of stream water at the upper and lower nodes in
subregion 2As (Southern Blue Ridge) ,
Weighted average concentration of major anions and cations by
pH class of stream water at the upper and lower nodes in
subregion 3A (Piedmont) ,
Weighted average concentration of major anions and cations by
pH class of stream water at the upper and lower nodes in
subregion 2X (Southern Appalachians)
Weighted average concentration of major anions and cations by
pH class of stream water at the upper and lower nodes in
subregion 2D (Ozarks/Ouachitas)
Weighted average concentration of major anions and cations by
pH class of stream water at the upper and lower nodes in
subregion 3C (Florida)
Plots of (a) total aluminum vs. pH, (b) extractable (MIBK)
aluminum vs. pH, (c) labile or inorganic monomeric aluminum vs.
pH, and (d) organic aluminum vs. DOC in streams of the NSS-I .
The absolute value of the standardized linear regression
coefficient (bf) for each of the independent variables that
influence ANC in all streams in the population of interest in
the Poconos/Catskills (ID), Northern Appalachians (2Cn), and
Valley and Ridge (2Bn) subregions
239
240
241
242
243
244
248
250
The absolute value of the standardized linear regression
coefficient (bjO for each of the independent variables that
influence ANC in all streams in the population of interest in
the Mid-Atlantic Coastal Plain (3B), Southern Blue Ridge (2As),
and Piedmont (3A) subregions
251
The absolute value of the standardized linear regression
coefficient (bj') for each of the independent variables that
influence ANC in all streams in the population of interest in
the Southern Appalachians (2X), Ozarks/Ouachitas (2D), and
Florida (3C) subregions
252
The absolute value of the standardized linear regression
coefficient (bj') for each of the five independent variables
XVlll
-------
considered in streams in the Poconos/Catskills (ID), Northern
Appalachians (2Cn), and Valley and Ridge (2Bn) subregions with
ANC < 200 neq L'1 258
8-8 The absolute value of the standardized linear regression
coefficient (bj') for each of the five independent variables
considered in streams in the Mid-Atlantic Coastal Plain (3B),
Southern Blue Ridge (2As), and Piedmont (3A) subregions with
ANC < 200 /zeq L"1 259
8-9 The absolute value of the standardized linear regression
coefficient (bj') for each of the five independent variables
considered in streams in the Southern Appalachians (2X),
Ozarks/Ouachitas (2D), and Florida (3C) subregions, with
ANC < 200 fieci L"1 260
8-10 Plot of closed system minus air-equilibrated pH vs. air-
equilibrated pH of stream water in NSS-I subregions 262
8-11 Observed air-equilibrated pH vs. ANC in NSS-I subregions 264
9-1 Frequency distribution (number of nodes) of spring index sulfate
concentration in NSS-I stream nodes with index ANC values
< 200 neq L"1 269
9-2 Location of NSS-I study reaches with ANC concentrations
< 200 /zeq L"1 having evidence of acid mine drainage or sub-
stantial watershed sources of sulfate at either the upper
or lower node 271
9-3 Location of NSS-I lower nodes with ANC < 200 jueq L"1 classified
as being dominated by organic anions 274
9-4 Location of NSS-I upper nodes with ANC < 200 jueq L"1 classified
as being dominated by organic anions . . . . 275
9-5 Location of the high-interest subpopulation of acidic (ANC < 0
/zeq L"1) NSS-I lower nodes 278
9-6 Location of the high-interest subpopulation of acidic (ANC < 0
/ieq L'1) NSS-I upper nodes 279
9-7 Location of the high-interest subpopulation of low ANC NSS-I
lower nodes 280
9-8 Location of the high-interest subpopulation of very low and low
ANC NSS-I upper nodes 281
xix
-------
9-9
9-10
9-11
9-12
9-13
9-14
10-1
10-2
10-3
10-4
10-5
10-6
Chemical classification flowchart 285
Trilinear plots for SO4*, NOS", and organic anions showing the
different anionic makeup of inorganic, organic-influenced, and
organic-dominated NSS-I nodes in three different ANC classes 286
Scatterplot showing the relationship between stream water ANC
(/ieq L'1) and stream water Na+:Cl~ molar ratio in NSS-I nodes
within 50 km of the coast 287
Median, 20th percentile, and 80th percentile of inorganic mono-
meric aluminum concentrations (/*M) in inorganic, organic-
influenced, and organic-dominated chemical classes 288
Population estimates of the number of downstream reach nodes in
each NSS-I subregion found in inorganic, organic-influenced, and
organic-dominated chemical classes 291
Population estimates of the number of upstream reach nodes in
each NSS-I subregion found in inorganic, organic-influenced,
and organic-dominated chemical classes 292
ANC of stream water vs. the nonmarine base cations concentration
in stream water in NSS-I subregions, 311
ANC plus SO42" concentration in stream water vs. the nonmarine
base cation concentration in the nine NSS-I subregions 313
Map of reach average sulfate concentration (/*eq L"1) in streams
after excluding stream nodes with ANC > 200 /jeq L"1 and/or
SO42" > 400 /zeq L"1 in the eastern United States overlayed with
isopleths of the 1980-1984 annual average wet sulfate deposition
. . 316
Relationship between 1980 to 1984 median annual wet and dry
sulfate deposition and median stream water sulfate concentration
in stream reach upper nodes with ANC < 200 fieq L"1 and
SO42" < 400 A*eq L"1 in the nine NSS-I subregions
317
Relationship between median wet sulfate deposition (1980-1984)
and median sulfate concentrations in surface waters with SO42"
< 400 /zeq L"1 in National Lake Survey and NSS-I (upstream reach
ends) subregions
319
Map of upstream node spring index ANC concentration (/zeq L'1)
in streams with ANC < 200 /ieq L"1 in the eastern United States
overlayed with isopleths of the 1980 to 1984 annual average wet
sulfate deposition (g m"2 yr"1)
320
xx
-------
LIST OF TABLES
Table Page
2-1 Characteristics of the NSS-I Subregions 17
2-2 NSS-I Site Inclusion Criteria 30
2-3 NSS-I Stage I Sample Grid Point Disposition 31
2-4 Special Interest Sites 42
3-1 Chemical and Physical Variables Measured in NSS-I and Methods
Employed 44
3-2 Geographic Variables Measured in NSS-I 48
3-3 Maximum Holding Times Specified for NSS-I Samples 55
3-4 NSS-I QC Samples 56
3-5 Exception Generating Programs within the Verification System 62
3-6 Calculations Performed for Cation/Anion Balances, Conductance
Estimates, and Protolyte Analysis in Verification 63
3-7 Types of Validation Analyses Performed on NSS-I Raw Data 67
3-8 Numeric Changes Made in the NSS-I Data Base as a Result of
Verification and Validation 71
3-9 Differences Between NSS-I and the Pilot Survey 73
4-1 Estimates of SDLs for the Primary NSS-I Chemical Variables Based
on Analyses of Field Blank Samples Pooled Across Laboratories 78
4-2 Percent Accuracy Estimates for the Primary NSS-I Chemical
Variables Based on Synthetic Audit Samples Pooled Across
Analytical Laboratories 80
4-3 System Level and Among-Batch Precision Estimates for Primary
NSS-I Chemical Variables 84
5-1 Refined Target Population Total Stream Resource Estimates 102
5-2 Distribution of Refined Target Population Upper and Lower Reach
Nodes by Geographic Site Class in NSS-I Subregions 107
xxi
-------
5-3
6-1
6-2
6-3
6-4
6-5
6-6
6-7
6-8
7-1
7-2
7-3
Characteristics of NSS-I Geographic Site Classifications .108
Population Median ANC, pH, and Sulfate Concentration at Lower
(L) and Upper (U) Reach Nodes of NSS-I Subregions '. . .137
Population Estimates of the Percentage of Target Stream Lower (L)
and Upper (U) Reach Nodes with Spring Index ANC (/ieq L"1)
in Reference Ranges 140
Population Estimates of the Percentage of Target Stream Lower (L)
and Upper (U) Reach Nodes with Spring Index pH in Reference
Ranges 141
Interpolated Length (km) and Percentage Distribution for ANC
(/ieq L'1) in Specified Concentration Ranges, NSS-I Subregions 151
Interpolated Length (km) and Percentage Distribution for pH in
Specified Ranges, NSS-I Subregions 152
Population Estimates of the Percentage of Target Stream Lower (L)
and Upper (U) Reach Nodes with Spring Index Inorganic Monomeric
Aluminum Above Reference Concentrations. 158
Sulfate Concentration in Deposition and Best Estimates and Range
of Precipitation/Runoff Ratios and Predicted Stream Water Sulfate
Concentrations (Assuming Only Atmospheric Sulfate Sources) in
NSS-I Subregions 171
Population Medians for Chemical Variables at Upper (U) and Lower
(L) Reach Nodes in NSS-I Subregions 181
Distribution of Observed ANC Differences (unweighted) Between
First and Second Visits to NSS-I Mid-Atlantic Sample Reaches 214
Distribution of Observed pH Differences (unweighted) Between First
and Second Visits to NSS-I Mid-Atlantic Sample Reaches 216
Population Estimates of the Percentage of Acidic and Low ANC
Stream Reaches Calculated from First Visit, Second Visit, and
Spring Index (Average) ANC at Upper and Lower Reach Nodes in
NSS-I Mid-Atlantic Subregions 221
7-4
Population Estimates of the Percentage of Stream Reaches with
pH Less than Reference Values Calculated from First Visit, Second
Visit, and Spring Index (Average) pH at Upper and Lower Reach Nodes
in NSS-I Mid-Atlantic Subregions 223
xxn
-------
8-1 Estimates of the Strong Acid Organic Anion Equivalents per mg
Organic Carbon in Stream Water at the Upper (U) and Lower (L)
Sample Reach Nodes in NSS-I Subregions Based on Linear Regression
of the Calculated Anion Deficit and DOC 256
9-1 Weighted Mean and Standard Deviation (SD) of NSS-I Spring
Index Chemistry in Stream Nodes Classified as Impacted by
Acid Mine Drainage and Watershed Sources of Sulfate 272
9-2 Weighted Mean and Standard Deviation (SD) of NSS-I Spring
Index Chemistry in Organic-dominated Classified Stream Nodes 276
9-3 Weighted Mean and Standard Deviation (SD) of NSS-I Spring
. Index Chemistry in Inorganic Classified Stream Nodes 282
9-4 Weighted Mean and Standard Deviation (SD) of NSS-I Spring
Index Chemistry in Organic-Influenced Stream Nodes 283
9-5 Estimated Number and Length of Acidic Stream Reaches Classified
as Impacted by Acid Mine Drainage or as Having Substantial
Watershed Sources of Sulfate at Either the Upper or Lower Node 289
9-6 Interpolated Length (km) Estimates of Streams in the High-Interest
Subpopulation with ANC Less than Reference Values 294
9-7 Interpolated Length (km) Estimates of Streams in the High-Interest
Subpopulation with pH Less than Reference Values 295
9-8 Estimated Number of Target Reach Lower Nodes in Each
Geographic Site Class Present in Stream Chemical Classes 296
9-9 Estimated Number of Target Reach Upper Nodes in Each
Geographic Site Class Present in Stream Chemical Classes 298
9-10 Estimated Percentage of Target Stream Reach Lower Nodes in
Each Geographic Site Class in Acidic and Very Low ANC Chemical
Classes 300
9-11 Estimated Percentage of Target Stream Reach Upper Nodes in Each
Geographic Site Class in Acidic and Very Low ANC Chemical Classes . . .301
9-12 Chemical and Geographical Classification of Special Interest Site
Streams Sampled During NSS-I 305
10-1 Fraction (r2) of the Variation in the Estimated ANC Deficit in
Stream Water that Can Be Accounted for by Sulfate, Nitrate, Chloride,
and DOC, and the Partial Regression Coefficient (b) for Each
Relationship. 315
xxiii
-------
VOLUME II TABLE OF CONTENTS
Section
Notice ;**
Preface "j
List of Figures vi
List of Tables xxvii
Volume I Table of Contents xxviii
Related Documents xxxiii
Acknowledgements xxxiv
SECTION 1 - INTRODUCTION 1
SECTION 2 - DEFINITION OF VARIABLES 3
SECTION 3 - SUBREGION MAPS H
SECTION 4 - STREAM NAMES AND LOCATIONS 23
SECTION 5 - POPULATION ESTIMATES FOR SELECTED PHYSICAL AND
CHEMICAL VARIABLES ; 59
5.1 Application of the Survey Design 59
5.1.1 Sample Weights 59
5.1.2 Sample Stratification 61
5.1.3 Subpopulations 61
5.1.4 Calculation of Distribution Functions 62
5.1.5 Calculation of Confidence Bounds 63
5.2 Population Distribution Functions 63
5.3 Interpretation of NSS-I Distribution Functions 64
SECTION 6 - PHYSICAL DATA LISTINGS 395
SECTION 7 - CHEMICAL DATA LISTINGS 433
SECTION 8 - REFERENCES 595
XXIV
-------
RELATED DOCUMENTS
Supplemental information on the National Stream Survey - Phase I (NSS-I) can be found in
the series of ancillary manuals and reports. These publications include:
Draft Sampling Plan for Streams in the National Surface Water Survey. 1985. Technical
Report 114 (July 1986). Overton, W.S. Department of Statistics, Oregon State
University, Corvallis, Oregon, 97331.
Draft Research Plan, National Surface Water Survey: National Stream Survey, Mid-Atlantic
Phase I and Southeast Screening. 1985. U.S. Environmental Protection Agency, Office
of Research and Development, Washington, D.C., 20460.
National Surface Water Survey: National Stream Survey, Phase I - Pilot Survey. 1986.
Messer, J.J., C.W. Ariss, J.R.Baker, S.K. Drouse, K.N. Eshleman, P.R. Kaufmann, R.A.
Linthurst, J.M. Omernik, W.S. Overton, M.J. Sale, R.D. Schonbrod, S.M. Stambaugh, and
J.R. Tuschall, Jr. EPA/600/4-86/026. U.S. Environmental Protection Agency,
Washington, D.C.
Quality Assurance Plan for the National Surface Water Survey, Stream Survey (Middle
Atlantic Phase I, Southeast Screening, and Middle Atlantic Episodes Pilot). 1986.
Drouse, S.K., D.C. Hillman, L.W. Creelman, and S.J. Simon. EPA/600/4-86/044.
Lockheed Engineering and Management Services Company, Inc., Las Vegas, Nevada
89109.
Field Operations Report, National Surface Water Survey, National Stream Survey, Pilot
Survey. 1987. Knapp. C.H., C.L. Mayer, D.V. Peck, J.R. Baker, and G.J. Filbin.
EPA/600/8-87/019. Lockheed Engineering and Management Services Company, Inc., Las
Vegas, Nevada, 89109.
Evaluation of Quality Assurance and Quality Control Sample Data for the National Stream
Survey (Phase I - Pilot Survey). 1987. Drouse, S.K. EPA/600/8-87/057. Lockheed
Engineering and Management Services Company, Inc., Las Vegas, Nevada, 89109.
Analytical Methods Manual for the National Surface Water Survey, Stream Survey (Middle
Atlantic Phase I, Southeast Screening, and Middle Atlantic Episodes Pilot). 1987.
Hillman, D.C., S.H. Pia, and S.J. Simon. EPA/600/8-87/005. Lockheed Engineering and
Management Services Company, Inc., Las Vegas, Nevada, 89109.
A Sampling and Analysis Plan for Streams in the National Surface Water Survey. 1987.
Technical Report 117. Overton, W.S. Department of Statistics, Oregon State
University, Corvallis, Oregon, 97331.
Data Management and Analysis Procedures for the National Stream Survey. 1988. Sale,
M.J. (editor). ORNL/TM. Oak Ridge National Laboratory, Oak Ridge, Tennessee,
37831. (Draft)
National Surface Water Survey: National Stream Survey (Phase I, Southeast Screening, and
Episodes Pilot). Field Operations Report. 1988. Hagley, C.A., C.L. Mayer, and R.
Hoenicke. EPA/600/x-xx/xxx. U.S. Environmental Protection Agency, Las Vegas,
Nevada, 89109. (In press)
National Surface Water Survey: National Stream Survey (Middle Atlantic Phase I and
Southeast Screening Surveys). Quality Assurance Report. 1988. Cougan, K.A., D.W.
Sutton, D.V. Peck, V. Miller, J.E. Pollard, and J. Teborg. EPA/600/4-88/018. U.S.
Environmental Protection Agency, Las Vegas, Nevada, 89109.
XXV
-------
ACKNOWLEDGEMENTS
The National Stream Survey - Phase I (NSS-I) in the Mid-Atlantic and Southeast built upon
the approach used for the NSS Pilot in the Southern Blue Ridge and also that employed in the
National Lake Survey. We owe a debt of gratitude all individuals involved in the design,
administration, implementation, and interpretation of these previous survey efforts. The National
Stream Survey represented an administrative and logistical challenge. Broad regional scale
probability sampling of a stream network required an innovative sampling design. The large
number of sample streams, the short time frame allowed for sampling, and the relative inacces-
sibility of many streams made the National Stream Survey a great challenge. The vigor and
dedication of hundreds of individuals, coordinated through the U.S. Environmental Protection
Agency, made the NSS-I possible. We here acknowledge only a few individuals who performed
key roles in the Survey, but we readily recognize that there are many others not mentioned
without whom the Survey would not have succeeded.
We gratefully acknowledge the citizens of the United States through the U.S. Environ-
mental Protection Agency Office of Environmental Processes and Effects Research (Director,
Courtney Riordan) for providing funding for the National Stream Survey.
We are grateful to Tom Murphy, ERL-C Laboratory Director, to Robert Lackey and Spencer
Peterson, Branch Chiefs at different times during the NSS-I project, and to Dixon Landers,
Aquatic Team Leader, for administrative support of the NSS-I at the Environmental Research
Laboratory in Corvallis, Oregon. We particularly acknowledge the support and advice of Dr.
Landers throughout the process of writing and responding to reviewer comments. Special thanks
are extended to Bob Schonbrod, U.S. EPA, EMSL-Las Vegas, for overseeing all aspects of quality
assurance and NSS-I field operations.
We give special acknowledgement to Keith Eshleman for his important role in the NSS-I
research plan and in bringing that plan through its peer review. Dr. Eshleman also provided
critical review and comment on early drafts of the NSS-I report and significantly influenced our
analytical approach.
James Omernik deserves special thanks for his assistance in the geographic aspects of
regional survey design. Special thanks are also extended to Anastasia Allen for her assistance
in NSS-I site selection and geographic characterization, and to Charles Ariss for his role in site
selection and the early stages of data validation.
John Baker and Mel Knapp are gratefully acknowledged for their indispensable roles in
survey planning and implementation. Special thanks are due Sevda Drouse for a major role in
the planning of chemical data quality assurance activities in the National Stream Survey. We
are also grateful for the cooperative efforts of numerous U.S. Soil Conservation Service
personnel in helping to obtain access to stream sampling sites.
We thank David Helvey and the staff of the USFS Timber and Watershed Laboratory in
Parsons, West Virginia, for providing the stream data used to study temporal variability in NSS-I
special interest sites.
We thank Deborah Coffey, Joe Eilers, Al Groeger, Rich Holdren, and Paul Shaffer (all of
Northrop Services, Inc., at ERL-Corvallis) for important insights gained from their critical
review and comment on early drafts of this report. Similarly, we owe a debt of gratitude to
Penny Kellar, Paul Ringold, Courtney Riordan, and many other reviewers from state agencies and
EPA regional offices for insights gained from their thoughtful comments on an earlier draft.
We kindly thank Michael Bowman, Ken Lanf ear, Dennis Newbold, Abdel El-Shaarawi, and
David Schindler for providing comprehensive technical reviews. External reviews were requested
xxvi
-------
from these individuals because of their demonstrated and recognized knowledge and expertise in
various aspects of stream survey design and interpretation. Their comments were invaluable in
refining the analysis as well as the interpretation of NSS-I data.
We gratefully acknowledge Bill Fallon for his indispensible role over several years in
briefing EPA Headquarters personnel on the progress of NSS-I activities, answering public and
congressional inquiries, planning a communication strategy for NSS-I results, and many other
aspects of representing the NSS-I in Washington, D.C. on a day to day basis.
The tireless enthusiasm and constructive encouragement of Rick Linthurst has been a
source of inspiration to the authors that is gratefully acknowledged.
Logistics, training, coordination, data management, data analysis, and report production
were accomplished by several organizations and many dedicated individuals. The following list
identifies the major activities and the individuals involved (not including the authors) and their
affiliations at the time of the Survey. The authors gratefully acknowledge all who contributed
to the NSS-I, but may not have been named here. The success of the project reflects these
participants' contributions of ideas, efficiency, enthusiasm, and hard work.
xxvii
-------
ERL-C (Northrop Services, Oregon State University, and Utah State University)
Statistical and Technical Support
Charles Ariss
D. James Blick
Mark Dehaan
Keith Eshleman
Mary Kay Hernandez
Randy Hjort
Kathy Hurley
Robert McCleod
Avis Newell
Gina O'Brien
Geographical Support
Anastasia Allen Barrel Downs
Jeff Irish
Roze Royce
Shannon Stambaugh
George Weaver
Anthony Zagar
Sharon Ziminski
Suzanne Pierson
Oak Ridge National Laboratory (U.S. DOE and Martin-Marietta Energy Systems, Inc.)
Data Base Management and Analysis
Jan Coe Mary Alice Faulkner Barbara Jackson
Systems Applications, Inc.
Statistical Consulting
Steven Edland Stella Grosser Thomas Permutt
Alison Pollack
EMSL-LV (U.S. EPA and Lockheed-EMSCO)
Field Sampling and Logistical Support
David Anthony
John Baker
Barry Baldigo
Jeffrey Bielling
John Campbell
James Crawford
David Grouse
Betsy Dickes
Jacqueline DiMauro
Richard Easton
Elizabeth Floyd
Elizabeth Hill
Mark Hodgkins
Rainer Hoenicke
Paul Hook
David Horton
Michael Hoppus
Frederick Kirschner
C. Mel Knapp
Christopher LaFlash
Lou Lunte
Lesa Madison
Cindy Mayer
Celeste McCuish
Diane McDonald
Anne Neale
Donald Osborne
Mike Parker
Peggy Perman
Steve Pierett
Robert Rice
Gary Roecker
John Schutt
Monica Schwalbach
Karen Short
Eugene Smith
Brad Teague
Timothy Zebuske
XXVlll
-------
Processing Laboratory Support
Barney Akuna
John Alston
Lori Arent
Mary Balogh
Christina Borror
Deb Chaloud
Hal Coleman
Betsy Dickes
Robert Heine
Herb Herpolsheimer
Robert Hughes
Valerie Miller
Molly Morison
James Nitteraurer
Roxanne Parks
James Pendleton
Carla Schuman
Sally Snell
Carl Whitfield
Brenda Whitfield
Jeffrey Wolfe
Statistical and Technical Support
Byron Blasdell
Brian Cordova
Robert Corse
John Curtis
Ramon Denby
Jerry Dugas
Robert Enwall
Annalisa Hall
John Henshaw
Daniel Hillman
John Lau
Tim Lewis
Jeff Love
Linda Marks
Richard Maul
Richard Metcalf
Mohammed Miah
John Nicholson
Mick Reese
Valerie Sheppe
Paula Showers
Lynn Stanley
Martin Stapanian
Donna Sutton
Mark Sweeney
John Teberg
John Wengert
Quality Assurance and Quality Control
Sevda Drouse
Marianne Faber
Daniel Heggem
Wesley Kinney
Carol MacLeod
Dean Mericas
T. Mitchell-Hall
James Pollard
Robert Schonbrod
Donna Sutton
XXIX
-------
CONTRIBUTIONS OF NSS-I AUTHORS
The National Stream Survey - Phase I (NSS-I) and this document represent the efforts of
many individuals. The primary contributors to this report are noted here.
Section 1: Introduction
P.R. Kaufmann, Utah State University
J.J. Messer, Utah State University
Section 2: Survey Design
W.S. Overton, Oregon State University
J.J. Messer, Utah State University
A.J. Kinney, Northrop Services, Inc.
P.R. Kaufmann, Utah State University
Section 3: Methods
M.E. Mitch, Utah State University
K.A. Cougan, Lockheed-EMSCO, Inc.
C.A. Hagley, Lockheed-EMSCO, Inc.
Section 4: Data Quality Assessment
D.V. Peck, Lockheed-EMSCO, Inc.
A.T. Herlihy, Utah State University
J.W. Elwood, Oak Ridge National Laboratory
P.R. Kaufmann, Utah State University
Section 5: Target Population—Physical Characteristics
P.R. Kaufmann, Utah State University
MJ. Sale, Oak Ridge National Laboratory
M.E. Mitch, Utah State University
Section 6: Target Population Regional Chemistry
P.R. Kaufmann, Utah State University
W.S. Overton, Oregon State University
M.J. Sale, Oak Ridge National Laboratory
M.E. Mitch, Utah State University
H.I. Jager, Science Applications International, Inc.
J.J. Messer, Utah State University
Section 7: Temporal Variability in Baseflow Chemistry
A.T. Herlihy, Utah State University
P.R. Kaufmann, Utah State University
K.H. Reckhow, Duke University
Section 8: Ion Relationships
J.W. Elwood, Oak Ridge National Laboratory
H.I. Jager, Science Applications International, Inc.
XXX
-------
Section 9: Nature and Distribution of Acidic and Low ANC Streams
A.T. Herlihy, Utah State University
P.R. Kaufmann, Utah State University
D.D. Brown, Northrop Services, Inc.
Section 10: Evidence of Acidification
P.R. Kaufmann, Utah State University
J.W. Elwood, Oak Ridge National Laboratory
A.T. Herlihy, Utah State University
Section 11: Summary and Conclusions
P.R. Kaufmann, Utah State University
A.T. Herlihy, Utah State University
J.W. Elwood, Oak Ridge National Laboratory
Editing: S.J. Christie, Northrop Services, Inc.
NSS-I Technical Directors:
Design and Implementation - Jay J. Messer
Data Analysis and Interpretation - Philip R. Kaufmann
NSS-I Statistical Design: W. Scott Overton
xxxi
-------
THIS PAGE INTENTIONALLY LEFT BLANK
XXXll
-------
EXECUTIVE SUMMARY
Background
National Stream Survey - Phase I (NSS-I) field activities were conducted in the mid-
Atlantic and southeastern United States in the spring of 1986 by the U.S. Environmental
Protection Agency (EPA) as part of the National Surface Water Survey (NSWS). The first phase
of the NSWS was designed to determine the present chemical status of surface waters in regions
of the United States containing the majority of streams and lakes considered to be at risk as a
result of acidic deposition. The NSS-I was conducted as part of the National Acid Precipitation
Assessment Program (NAPAP). Like the previous EPA NSWS activities (Eastern and Western
Lake Surveys), it contributes directly to one of NAPAP's principal objectives: the quantification
of the extent, location, and characteristics of acidic and potentially sensitive streams and lakes
in the United States.
The NSS-I was conducted in four Mid-Atlantic and five Southeast subregions of the United
States (Figure E-l), identified on the basis of similar physiographic characteristics:
• Mid-Atlantic CMA") Region
- Interior Mid-Atlantic (IMA) Subregions
~ Poconos/Catskills (ID)
— Northern Appalachians (2Cn)
— Valley and Ridge (2Bn)
- Mid-Atlantic Coastal Plain Subregion (3B)
• Southeastern (SE) Region
- Interior Southeast (ISE) Subregions
— Southern Blue Ridge (2As) - Pilot Survey in 1985
— Piedmont (3A)
— Southern Appalachians (2X)
-- Ozarks/Ouachitas (2D)
- Florida Subregion (3C)
These subregions of the United States were expected, on the basis of geology, deposition
rates, and previous water quality data, to contain a substantial number of streams that have low
acid neutralizing capacity (ANC) or that are acidic (ANC < 0). Furthermore, NSS-I efforts
during 1986 were concentrated in areas where acidic deposition rates are relatively high, but
where lakes are not abundant. We do not have synoptic information from the National Lake
Survey (NLS) on surface water chemistry in these areas. A Pilot Survey was conducted in the
Southern Blue Ridge Subregion one year earlier than field activities in the remainder of the
NSS-I subregions. This NSS-I Pilot Survey was designed to assess the logistic and scientific
feasibility of the full-scale NSS-I Survey.
NSS-I field activities to date have not included areas of the Northeast, Upper Midwest, and
West. Though these regions are expected to contain low ANC or acidic streams potentially
XXXlll
-------
SUBREGIONS OF THE NATIONAL STREAM SURVEY-PHASE I
Northern
Appalachians (2Cn)
Valley and Ridge (2Bn)
Poconos/Catskills (ID)
Southern Blue Ridge (2As)
(Pilot Study)
Mid-Atlantic
Coastal Plain (3B)
Ozarks/Ouachitas (2D)
Southern Appalachians (2X)
Figure E-l. NSS-I subregions.
XXXIV
-------
sensitive to acidic deposition, they also contain numerous lakes that were sampled as part of
EPA's Eastern and Western Lake Surveys. Furthermore, NSS-I field activities thus far have not
included synoptic stream chemistry sampling in parts of the South Atlantic and Gulf Coastal
Plains expected to contain predominantly low ANC surface waters—but where deposition rates
are comparatively lower than in most of the Survey area, and where organic acidity is expected
to play an important role. Field activities in the Florida subregion test the utility of NSS-I
logistical and design protocols in lowland stream networks of the Southeast Coastal Plain.
Objectives
The objectives of the NSS-I in the Mid-Atlantic and Southeast were to:
• Determine the percentage, extent (number, length, and drainage area), location, and
chemical characteristics of streams in the Mid-Atlantic and Southeast that are
presently acidic, or that have low ANC and thus might become acidic in the future.
• Identify streams representative of important classes in each region that might be
selected for more intensive study or long-term monitoring.
Methods
Within the NSS-I subregions, the stream resource of interest (the target resource) was
identified as those streams that have drainage areas less than 155 square kilometers (60 square
miles), but that are large enough to be represented as blue lines on l:250,000-scale U.S.
Geological Survey (USGS) topographic maps. Reviewers accepted this size range as a reasonable
compromise that would include streams large enough to be important for fish habitat, yet still
small enough to be susceptible to the impacts of acidic deposition.
Unlike lakes, which can be counted and sampled as discreet entities, streams form a hier-
archical network in which small streams are tributaries to large streams. The NSS-I sampled
stream reaches defined as segments of the stream network, as represented by blue lines on the
l:250,000-scale maps. These segments, or reaches, were identified as mapped blue-line segments
between two tributary confluences. Sampling points on each of these reaches were just above
the downstream point of confluence (lower node) and just below the upstream point of con-
fluence (upper node). The upper node of each reach represented as a headwater was defined as
the farthest upstream extent of the mapped blue line representation.
Because not all stream reaches in the mid-Atlantic and southeastern United States could be
sampled, a statistical procedure was developed for selecting a subset of streams as a probability
sample from which the characteristics of the total reach population could be extrapolated. A
two-stage sampling procedure was used to obtain a randomized, systematic sample of approxi-
mately 500 reaches with good spatial distribution over each of the nine NSS-I subregions (50 to
80 reaches per subregion). Reaches were excluded if they were too large (drainage area > 155
km2), were located within metropolitan areas or tidal zones, or were affected by oil field brine,
acid mine drainage, or point-source pollution.
The NSS-I used index values to describe the chemical status of each stream sampled. This
index value was measured during baseflow of the spring season between snowmelt and leaf out
(approximately March 15 to May 15). The choice of the spring index sampling period involved a
trade-off between minimizing within-season and episodic chemical variability and maximizing the
probability of sampling chemical conditions potentially limiting for aquatic organisms.
xxxv
-------
As a result of Pilot Survey experience, two spring season samples were judged sufficient to
index chemical characteristics of streams in the Mid-Atlantic subregions. In the Southeast,
where acidic deposition effects were expected to be less probable, one spring sample was taken
at each site. To quantify and incorporate the variability between upstream and downstream ends
of reaches, chemical and physical variables were measured at both ends.
Chemical variables measured at each sampling site included those related to biological
effects (e.g., pH, extractable aluminum, and competing ligands such as fluoride and dissolved
organic carbon), other variables related to potential sensitivity and related geochemistry (ANC,
base cations, acid anions, and silica), and others indicative of anthropogenic disturbances or
nutrient status ( phosphorus, iron, ammonium, and turbidity). Samples were stabilized within 12
to 24 hours of collection and standardized quality control and quality assurance protocols were
followed during sample handling, analyses, data reporting, data storage and analysis. Population
frequency distributions (with 95% confidence bounds) were calculated for selected chemical and
physical variables.
Selected Results
Overview
The basic results of the NSS-I are population descriptions of the location, number, length,
and percentage of streams within referenced ranges of chemical concentration. The most
important of these descriptions are those concerning ANC and pH. Further data interpretation
includes an examination of regional patterns in the relationships among the chemical constituents
within stream waters in an effort to infer the possible geochemical factors and anthropogenic
impacts controlling stream chemistry. Lastly, a high-interest segment of the stream population
with low ANC was examined and classified according to probable sources of acidity.
Survey data such as that collected by the NSS-I cannot in itself be used to prove causal
relationships (e.g., the effect of acidic deposition on stream chemistry). However, the lines of
correlative evidence we present can be examined to determine whether or not they support
hypotheses concerning the true controls on streamwater chemistry.
Regional Chemical Characteristics
Target Population of Interest—
Physical and chemical characteristics of an estimated 57,000 stream reaches with a com-
bined length of approximately 200,000 km (124,000 mi) were extrapolated from a probability
sample of approximately 450 stream reaches in the stream population of interest within the nine
NSS-I subregions. (An additional 54 reaches were visited in the field, but were found to have
characteristics such as acid mine drainage or tidal effects that classified them as noninterest
sites for this particular assessment.) The population of streams targeted by the NSS-I is best
described as small to mid-sized streams in the low end of the size range typically managed by
state fishery agencies. Stream reaches sampled by the NSS-I are typically about 3 km long. At
their upstream ends, approximately half of these stream reaches are classified as headwater
reaches on l:24,000-scale topographic maps. At their downstream ends, most reaches (67%) are
of Strahler Order 2 and 3. Median drainage areas at the upstream and downstream ends of the
population of NSS-I target stream reaches are 1.5 km2 and 8.1 km2. Twenty percent of these
xxxvi
-------
reaches have drainage areas less than 0.18 km2 at their upstream ends. For perspective, the
majority of these streams have widths between 1 and 6 m and depths between 0.1 and 0.5 m.
Types of Population Descriptions—
All regional population descriptions of the NSS-I target population are based on spring
baseflow index chemistry (see Methods). The "index" is a representative value for each chemical
constituent based on spring baseflow samples. It is important to understand the limitations as
well as the strengths of the chemical index approach. For example, the number of stream
reaches that are acidic due to episodic pH depressions during storm runoff cannot be obtained
or inferred directly from the distribution of index chemistry provided by the NSS-I. The
number of streams experiencing acidic episodes may be much greater than the number estimated
to be acidic during spring baseflow. Further studies of short-term and seasonal variability are
expected to demonstrate a useful predictive relationship between index chemistry and chemistry
during other times of the season or year.
We present two types of population descriptions expressing the distribution of the popula-
tion of target streams over the range of ANC and pH. One type of description (Table E-l) is
based on enumerating reaches with upstream and downstream chemistry in reference ranges.
The population distribution estimates based on upstream and downstream sampling comprise two
spring baseflow "snapshots" focusing at different positions within the streamflow network
represented by blue lines on l:250,000-scale topographic maps.
Separate distributions of chemistry at the upstream and downstream ends of target stream
reaches provide a fairly complex picture of their status. It is perhaps of greater interest to
combine upstream and downstream chemical measurements by interpolation along the length of
reaches and to report the status of the stream resource in a second type of population estimate
expressed in terms of the combined length of streams within reference categories of pH and
ANC (Table E-2; Figures E-2 and E-3). We have focused our presentation on these length dis-
tributions because they may have more utility for fish habitat quality assessment. However, the
tables in this Executive Summary show, separately, the upstream and downstream estimates dis-
cussed above.
Reference Values for ANC and pH—
Population proportions are reported within ranges of ANC and pH defined by reference
values. ANC, a measure of the amount of acid that can be neutralized by a sample of water, is
probably the most important variable measured in the NSS-I. The value of ANC = 0 is of sig-
nificance because waters at or below this level have no capacity to neutralize acid inputs—they
are acidic by definition.
The 50 /aeq L"1 ANC reference value forms the upper bound of a range of aquatic systems
between 0 and 50 /xeq L"1 that may be considered to have very low ANC. Although there may
be no special significance to the 50 /zeq L'1 level, some scientists term surface waters with ANC
< 50 /zeq L'1 as "extremely acid sensitive"1. There is some evidence that streams and lakes with
spring baseflow ANC (streams) or fall turnover ANC (lakes) below this level may be prone to
acidic episodes during storm events or snowmelt. Low ANC is probably a necessary condition
predisposing acidic episodes, although certainly not a sufficient one. Streams or lakes that
Schindler, D.W. 1988. Effects of acid rain on freshwater ecosystems. Science 239:149-157.
XXXV11
-------
Table E-l. Population Estimates of the Percentage of NSS-I Target Stream Lower (L) and
Upper (U) Reach Nodes with Spring Baseflow Index ANC and pH Below Reference
Values
<0
<50
ANC (jteq L"1)
<200
Total
(No. of Reaches)
Region
U
U
U
U
Interior MA
MA Coastal
Interior SE
Florida
Total NSS-I
1
Plain 6
*
14
Area 2
.3
.8
.5
.3
5.4
11.8
0.6
39.2
6.1
8.6
19.5
5.8
57.5
11.2
20.4
30.2
7.5
70.0
19.6
40.6
41.0
46.4
81.2
43.7
49.1
55.8
51.6
77.6
52.1
25,715
11,287
18,720
1,555
57,277
24,945
11,284
18,686
1,727
56,642
< 5.0
pH
< 5.5 < 6.0
Total
(No. of Reaches)
Region
U
U
U
Interior MA 1.1 3.9
MA Coastal Plain 6.8 11.8
Interior SE * *
Florida 14.5 31.2
Total NSS-I Area 2.2 5.0 4.1 10.6 9.1 21.6 57,277
U
2.0
13.2
*
23.8
8.5
23.5
1.9
49.9
5.1
21.6
2.7
61.9
15.4
49.3
8.6
71.8
25,715
11,287
18,720
1,555
24,945
11,284
18,686
1,727
56,642
*No samples observed in this range; estimated percentage is less than
MA - Mid-Atlantic
SE « Southeast
XXXVlll
-------
Table E-2. Population Estimates of the Combined Length (km) and Percentage of NSS-I
Target Stream Reaches with Spring Baseflow ANC and pH less than Reference
Values
Region
Interior MA
MA Coastal Plain
Interior SE
Florida
Total NSS-I Area
-------
NSS-I INTERPOLATED LENGTH DISTRIBUTION - ANC
Interior Mid-Atlantic
• Poconos/Catskills (ID)
• Northern Appalachians (2Cn)
• Valley and Ridge (2Bn)
69,569 km
2Cn
Interior Southeast
• Southern Blue Ridge (2As)
• Southern Appalachians (2X)
• Piedmont (3A)
• Ozarks/Ouachitas (2D)
86,939 km
Mid-Atlantic Coastal Plain (3B)
40,296 km
Florida Subregion (3C)
3,848 km
ACID NEUTRALIZING
CAPACITY
B >0to £50
H >50 to £200
>200
Figure E-2. Population estimates for ANC in eastern United States streams. Pie diagram area
is proportional to the combined length of target streams in each region.
xl
-------
NSS-I INTERPOLATED LENGTH DISTRIBUTION - pH
Interior Mid-Atlantic
• Poconos/Catskills (ID),
• Northern Appalachians (2Cn)
• Valley and Ridge (2Bn)
69,569km
2Cn
Interior Southeast
• Southern Blue Ridge (2As)
• Southern Appalachians (2X)
• Piedmont (3A)
• Ozarks/Ouachitas (2D)
86,939 km
Mid-Atlantic Coastal Plain (3B)
40,296km
Florida Subregion (3C)
3,848 km
Figure E-3. Population estimates for pH in eastern United States streams. Pie diagram area
is proportional to the combined length of target streams in each region.
xli
-------
experience acidic episodes are also often subject to high atmospheric acid deposition rates, and
their watersheds may be more susceptible to episodic pH depressions due to their hydrology.
The 200 neq L'1 ANC reference value is of interest because it has been frequently cited in
acid deposition literature as a value above which aquatic systems are unlikely to be sensitive to
acidification. It should be kept in mind that sensitivity to acid deposition impacts is dependent
on other factors besides the ANC of the stream water itself. A stream is an integral part of its
watershed; as such, its sensitivity to acid deposition is dependent upon the ability of watershed
geology, soils, and vegetation to neutralize incoming acidity.
The variable pH is an inverse measure (negative logarithm) of hydrogen ion concentration.
A one-unit decrease in pH represents a ten-fold increase in hydrogen ion concentration.
Whereas ANC is a measure of the capacity of a water sample to neutralize acids, pH is a meas-
ure of the strength of the acidity.
The pH reference values of 5.0, 5.5, and 6.0 are defined for convenience in comparing
results with those of other studies (e.g., EPA's Eastern and Western Lake Surveys) that have
reported results referencing these values. The pH reference values bear some relationship to
critical values below which populations of different types of fish are not sustained in freshwater
ecosystems. Because of the influence of pH on the solubility and chemical form of aluminum, it
is difficult to separate the effects of low pH from those of aluminum toxicity. Interpretation'of
critical pH values are further confounded because of the effect of dissolved organic compounds
and calcium in reducing the toxicity of aluminum. However, several researchers have reported
that waters with pH chronically below 4.0 are virtually devoid of fish, and waters with pH
chronically below 4.5 contain few fish species. Estimates for critical values below which
salmonid fish populations (e.g., trout) are not sustained range from pH 4.7 to 5.5. Estimates for
smallmouth bass and blueback herring are between 5.0 and 5.7. Except for effects reported on
several species of dace, shiners, and minnows, pH levels above 6.0 have not been associated with
adverse impacts on fisheries. However, recent research on blueback herring, striped bass, and
yellow perch suggests that under certain conditions of high aluminum concentration, pH between
6.0 and 6.5 can lead to increased egg and larval mortality.
Population Distribution Estimates for ANC--
The estimated percentage of the total number of upstream and downstream reach ends
(nodes) with spring baseflow ANC less than 0, 50, and 200 /ieq IT1 are presented in Table E-l.
It is evident that acidic (ANC < 0) reaches are concentrated in the Interior Mid-Atlantic, the
Mid-Atlantic Coastal Plain, and Florida. It is also evident that the target population has
generally lower ANC at the upstream reach ends, reflecting a commonly observed pattern of
increasing buffering capacity in the downstream direction because of increased time in contact
with watershed rock and soils, plus the presence of more landuse impacts and weatherable soils
in lowland areas. To simplify discussion of the data, we will focus on the length distributions
that combine upstream and downstream chemistry to yield estimates of the combined length of
the stream resource within the ANC reference ranges (Table E-2 and Figure E-2). Table E-2
presents the estimated length and percentage of the total target reach length with ANC less
than or equal to each of the three reference values. In Figure E-2, the size of each pie
diagram is proportional to the total combined length of stream reaches in the target populations
within each of the four regions.
Tables E-3, E-4, E-5, and E-6 present detailed population estimates (and statistical con-
fidence bounds) for both ANC and pH in the nine NSS-I subregions and are provided as supple-
mentary reference tables at the end of this Executive Summary. Population estimates for the
xlii
-------
Florida subregion are not strictly comparable with those from other subregions, so these are
summarized separately from the other subregions in the following discussion.2
Acidic (ANC < 0) and very low ANC (0 to 50 /ieq L"1) baseflow conditions were observed
primarily in the Mid-Atlantic Coastal Plain and the Interior Mid-Atlantic Region (Table E-2,
Figure E-2). Acidic reaches (ANC < 0) made up 6%, or 2527 km, of the length of the target
stream resource in the Mid-Atlantic Coastal Plain, and were observed mostly in the New Jersey
Pine Barrens and on the Coastal Plain west of Chesapeake Bay. Approximately 3%, or 2,324 km,
of the target stream length in the Interior Mid-Atlantic Region was acidic during spring base-
flow, whereas acidic reaches were rare in the Interior Southeast Region (approximately 0.1% or
117 km). Acidic reaches in these interior subregions were observed largely in forested upland
watersheds with drainage areas less than 20 km2 and were most numerous in the Northern Appa-
lachian subregion of the Interior Mid-Atlantic, where they made up an estimated 7% (1,524 km)
of the stream resource.
The estimates of acidic stream reaches in the previous paragraph, and in the figures and
tables of this Executive Summary, do not include streams within the size range of the NSS-I
target reaches that were acidic and receive acid mine drainage. We estimate 3,500 km in
Pennsylvania and West Virginia, and 1,100 km in the Interior Southeastern subregions within this
acid mine drainage category. Acidic streams impacted by acid mine drainage comprised approxi-
mately 10% of the total number of reaches within the target stream size range in the Northern
Appalachian Plateau (subregion 2Cn).
Very low ANC reaches (ANC 0-50 /*eq L'1) make up a rather large fraction (17.6%, or
7,109 km) of the target stream resource in the Mid-Atlantic Coastal Plain. In the Interior Mid-
Atlantic, 7.3% (5,107 km) of stream reaches were estimated within this very low ANC range. In
the Interior Southeast, 4.5% (3,947 km) were within this range, where they were proportionately
most common in the Piedmont and Southern Blue Ridge subregions, making up 7% to 8% of the
resource. Stream reaches in the moderately low ANC range (50-200 peq L"1) were abundant in
the Interior Southeast, where nearly 38,000 km (43.5%) were within this range. In the
Mid-Atlantic Region, 23,804 km (34.2%) of the Interior, and 11,455 km (28.4%) of the Coastal
Plain streams were within this low ANC range.
Proportions of the target stream reach population with spring index ANC < 200 ^eq L"1
were similar in the Mid-Atlantic Region and the Interior Southeast Region (Table E-2, Figure
E-2). Note that the Interior Southeast Region excludes Florida, whereas the Mid-Atlantic
Region combines the Interior Mid-Atlantic and the Mid-Atlantic Coastal Plain. Approximately
half the combined reach length in the Mid-Atlantic and Interior Southeast regions had ANC
< 200 A*eq L"1.v
The Florida subregion stands out as a geographic area with a relatively high percentage of
acidic and very low ANC streams. It is also an area where there are many highly colored
streams with high concentrations of dissolved organic carbon (DOC) likely to be contributing to
the observed acidity. An estimated 12% (461 km) of streams in the portion of Florida surveyed
were acidic and nearly 50% (1,895 km) had ANC between 0 and 50 /ueq L'1. Whereas the acidic
and very low ANC reaches in the Florida peninsula were typically high in DOC and strongly
colored, many of those in the Florida panhandle were clear or slightly colored and had very low
ionic strength, with very low sulfate, nitrate, and DOC concentrations.
2The Florida sample was drawn from a more restrictive target population focused only on the portion of this state
with expected ANC less than 200 JUeq L"1, rather than 400 /Zeq IT1 (the criterion for all the other subregions). In
addition, the total size of the target population in Florida may be underestimated because many reaches visited were
dry or lacked flowing water at the time of sampling.
xliii
-------
Population Distribution Estimates for pH—
The estimated number and percentage of upstream and downstream reach ends (nodes)
within the four ranges of pH are presented in Table E-l. More detailed estimates, with
confidence bounds, are shown in Table E-4. Population distribution estimates for pH mirrored
those for ANC, which is not surprising in light of the relationship between ANC and hydrogen
ion concentration. As observed for ANC below a reference value of 0 /zeq L'1, reaches with
spring baseflow index pH less than 5.5 were concentrated largely in the Mid-Atlantic, the
Interior Mid-Atlantic, and Florida. Again, the general downstream increase in buffering was
reflected in the larger percentages of reaches with low pH at their upstream ends (compared
with their downstream ends). To simplify discussion of this upstream and downstream pH data,
we focus in this Executive Summary on population estimates that use both upstream and down-
stream data to yield estimates of the combined length of target reaches within the several pH
reference ranges (Table E-2 and Figure E-3). More detailed stream length estimates with
statistical confidence bounds are provided for reference in Table E-6. For reasons mentioned in
the previous section, Florida estimates are discussed separately from the other subregions.
Reaches with pH of 5.5 or less made up an estimated 24% (9,565 km) of the target stream
length in the Mid-Atlantic Coastal Plain (Table E-2). These low pH reaches were observed
primarily in the New Jersey Pine Barrens and on the Coastal Plain west of Chesapeake Bay.
Potential sources of acidity in streams of this and other regions are discussed in the following
sections of this Executive Summary. In the Interior Mid-Atlantic Region, an estimated 4,712 km
(6.8%) of target reach length had pH < 5.5. These low pH reaches were observed largely in
upland forested watersheds with drainage areas less than 30 km2 and were most common in the
Northern Appalachian subregion of the Interior Mid-Atlantic, where they made up an estimted
8.6% (1,870 km) of the combined length of target stream reaches. In contrast, the combined
reach length with pH < 5.5 in the Interior Southeast was proportionately small, with an esti-
mated 723 km (< 1%) of the target population in this category.
The greatest combined length of target reaches with baseflow pH between 5.5 and 6.0 was
estimated in the Mid-Atlantic Coastal Plain (9,142 km or 22.7%). The Interior Mid-Atlantic and
the Interior Southeast Regions had similar percentages of their length in this range (5 5% and
5.8%, respectively).
Reaches with spring baseflow index pH greater than 6.0 predominated in the Interior Mid-
Atlantic (88% of combined length) and the Interior Southeast (93% of combined length). In
contrast, reaches with pH greater than 6.0 made up a little more than half (54%) of the com-
bined length of target streams in the Mid-Atlantic Coastal Plain.
The Florida subregion stands out as an area with a relatively high estimated percentage of
its combined target reach length having spring baseflow index pH < 5.5 (1708 km, or 44%).
Another 1,120 km (29%) were estimated in the pH category between 5.5 and 6.0, leaving only
26.5% of the combined reach length in this subregion with pH greater than 6.0. Note the
previous discussion regarding the comparability of Florida results with those in other regions.
The potential sources of acidity in streams of Florida and the other areas surveyed by the NSS-I
are discussed in later sections of this Executive Summary.
Population Distribution Estimates for Sulfate—
Sulfate is an important chemical constituent for study in stream water because it is often
associated with hydrogen ion as sulfuric acid in streams acidified by mine drainage and
atmospheric deposition. Stream reaches within NSS-I subregions of the Mid-Atlantic Region had
xliv
-------
substantially higher spring baseflow index sulfate concentrations than did those in the Southeast.
Subregion median stream water sulfate concentrations ranged from 125 to 238 /ieq L"1 in the
four Mid-Atlantic subregions, whereas those of the five subregions in the southeastern United
States ranged from 10 to 71 jteq L"1. The potential sources of sulfate in stream water include
wet and dry atmospheric deposition (natural and anthropogenic sources), nonatmospheric anthro-
pogenic sources (e.g., agriculture), and the weathering of naturally occuring sulfur-bearing
minerals (which can be accelerated by mining activities). The potential role of sulfate from
several sources is discussed in the following paragraphs as it pertains to the observed pH and
ANC in stream reaches of the NSS-I target population.
Chemical Relationships
Regional variation in ANC among streams within NSS-I subregions was associated more
closely with differences in base cation concentrations than differences in concentration of
mineral acid or organic acid anions. This result suggests that watershed geochemical and
hydrologic characteristics controlling the supply of mineral weathering products are very
important in determining the regional patterns in stream water ANC and therefore buffering of
acid inputs.
The only direct way to conclusively demonstrate surface water acidification is through
historic water quality data. Because of a lack of such data on a regional scale for the streams
of interest in the NSS-I, a second approach for assessing whether streams in the Mid-Atlantic
and Southeast have been acidified involves an indirect method of inferring past changes in ANC
through an examination of present water chemistry. Previous researchers have used the ANC
deficit (nonmarine base cation concentration minus ANC) as a rough measure of surface water
acidification resulting from either natural or anthropogenic causes.
It is important to recognize that ANC deficits, in themselves, reflect simply a presence of
strong acid anions in water at the time of sampling. ANC deficits can be observed in stream
water for a number of reasons: (1) they may reflect a reduction in ANC caused by the
addition of a strong acid (like sulfuric acid) to the stream, as might occur with acid deposition
onto a poorly buffered watershed; (2) they may reflect an increase in base cation concentration
(with no change in ANC) in the stream resulting from cation exchange or the accelerated
weathering of watershed soils and rock (this can be due to acidic deposition); (3) they may
result from an addition of base cations in neutral salts (e.g., CaSO4), with no change in
ANC; or finally (4) they may result from a combination of all these mechanisms. The last
combination of factors is the most likely for a heterogeneous assemblage of streams subject to a
wide range of influences (e.g., the entire population of NSS-I target streams). However, the
interpretation of ANC deficits in a subpopulation of relatively pristine streams draining
watersheds underlain by metamorphic rock is not as uncertain. In these pristine streams, ANC
deficits were higher in areas that receive larger amounts of acidic deposition. The lack of
plausible sources of neutral salts in these streams makes it likely that the observed ANC deficits
in such upland, forested streams are due to titration by strong acids or increases in base cation
export, and that both mechanisms are probably associated with atmospheric acid deposition.
In order to interpret the observed ANC deficit as a measure of the amount of historic
change in ANC, one must first assume that the ANC deficits are due largely to titration by
strong acids and that the base cation concentrations in streams have remained nearly constant
over time. These assumptions are more likely to be correct in poorly buffered stream waters of
low ionic strength. There are indications, however, that base cations may not remain constant
xlv
-------
over .time in many streams. In addition, several conditions must also be met before the ANC
deficit in a given stream could be interpreted as a measure of historic acidification due to
atmospheric deposition: (1) an atmospheric source of strong acids can be demonstrated to be of
sufficient magnitude to cause the observed ANC deficit, (2) other potential sources of strong
acid anions can be quantified or ruled out as unlikely (e.g., sulfide weathering, organic acidity),
and (3) concentrations of acids not derived from atmospheric deposition (e.g., organics from
watershed sources) have remained relatively constant over time.
All NSS-I subregions had streams with an ANC deficit. The largest ANC deficits were
observed in streams of the four Mid-Atlantic subregions (Poconos/Catskills, Northern Appala-
chians, Valley and Ridge, and Mid-Atlantic Coastal Plain). The observation of smaller ANC
deficits in streams of most Southeast subregions, compared with those of the Mid-Atlantic, is
generally consistent with the lower atmospheric acid deposition rates and greater sulfate
adsorption capacities of soils in the southeastern United States. Sulfate was the dominant
strong acid anion in most acidic and low ANC streams and was observed to be in sufficient
concentration to account for the observed ANC deficits in the majority of stream reaches in all
subregions except Florida and the Mid-Atlantic Coastal Plain.
The distribution of sulfate concentration in streams of the eastern United States that are
not affected by large terrestrial sulfate sources corresponds well with sulfate wet deposition
rates3. A plot of population median stream sulfate concentrations versus rates of sulfate
deposition in the nine NSS-I subregions shows a strong positive linear relationship. When data
from the NSS-I and the National Lake Survey are combined, a positive near linear relationship is
clearly evident among northeastern, mid-Atlantic, western, and upper midwestern surface waters
(Figure E-4). It is apparent, however, that a group of southeastern streams and lakes have
lower sulfate concentrations than expected, given the sulfate deposition rates in their respective
subregions. These observations are consistent with other research showing substantial sulfate
retention in watersheds of some parts of the Southeast.
Sources of Acidity in Acidic and Low ANC streams
Potential sources of acidity were examined for the subpopulation of acidic and low ANC
(£ 200 neq L'1) streams in the NSS-I target population. Stream reaches affected by acid mine
drainage and organic acids were identified by site visits and examination of stream chemical
data. After elimination of categories of streams for which sources of acid anions other than
acid deposition largely mask the potential effects of acid deposition, a high-interest group of
streams remained. Acid deposition is likely to be a major source of acid ions in these streams.
Based on interpolations between upper and lower node chemistry, acidic reaches within the
high-interest subpopulation had an estimated combined length of 4,455 km. Most of these
reaches (4,272 km) were observed in the Mid-Atlantic subregions, where they comprised 4% of
the total length of target stream reaches. Somewhat less than one-third (1,212 km) of these
high-interest acidic reaches had chemistry influenced but not dominated by organic anions.
Most of these organic-influenced reaches are located in the New Jersey Pine Barrens and in the
low-gradient headwaters of reaches in the Glaciated Plateau of Pennsylvania and New York.
Estimates based upon the first NSS-I sample visit alone (rather than the average of the two
spring visits) were slightly higher than 4,455 km because of a seasonal trend of increasing
Sulfate wet deposition rates taken from Wampler, S.J., and A.R. Olsen. 1987. Spatial estimation of annual wet acid
deposition using supplemental precipitation data. Reprinted from the Preprint Volume: Tenth Conference on
Probability and Statistics. October 4 and 5, 1987. Edmonton, Alta., Canada. Published by the American
Meteorological Society, Boston, Massachusetts.
xlvi
-------
o 5
.s
UJ
:*;
9
v.
-Q
•a
co
•
<0
co
CO
c
o
•••
O
-------
spring baseflow ANC. Based on the earlier spring sample, a combined length of 7,481 km of
acidic stream reaches were within the high-interest subpopulation of streams for which acid
deposition impacts cannot be ruled out. Again, most of these reaches were observed in the
Mid-Atlantic subregions, where they comprised 6.6% of the total length of the target stream
resource. These reaches are located primarily in the forested upland drainages and coastal areas
of the Mid-Atlantic Region that experience high rates of atmospheric sulfate deposition.
Specifically, most reaches within the high-interest group of acidic streams are located:
• In upland forested drainages of the Allegheny Plateau, in ridges of the Valley and
Ridge physiographic province, and in the Glaciated Plateau in northeastern Pennsyl-
vania (included in NSS-I subregions ID, 2Bn, and 2Cn). An estimated 695 (9.7%) of
7190 reaches in ridges of the Valley and Ridge geographic area and 499 (11%) of 4548
reaches in forested drainages of the Allegheny Plateau were acidic at their upstream
nodes.
• In the Pine Barrens of New Jersey (within the Mid-Atlantic Coastal Plain, subregion
3B). Seventy-nine percent of upstream nodes and 93% of downstream nodes were in
this high-interest subgroup; approximately one-third of these had chemistry influenced
but not dominated by organic anions.
Historic data suggest that streams in the Pine Barrens may have been naturally acidic at
least since the early 1900's. There is no reason, however, to believe that most of the upland
forested reaches have always been acidic.
Identified within the NSS-I target population was a group of non-acidic stream reaches
with very low ANC (> 0 - < 50 /zeq L"1) for which the predominant sulfate source appears to be
acid deposition. Based on interpolation between upper and lower node chemistry, an estimated
15,590 km of target reach length is within this group. Approximately two-thirds of these
reaches were observed in the Mid-Atlantic subregions, where they made up approximately 9.6% of
the combined length of the target population. Calculations of these totals based on the earlier
spring sample did not differ from those based on the average from two spring visits. These
very low ANC streams are located primarily on upland, forested sites of the Allegheny Plateau,
Valley and Ridge Province, Blue Ridge Mountains, and Cumberland Plateau. In addition, a large
number of these reaches were found in Coastal Plain areas—Eastern and Western Mid-Atlantic
Coastal Plain, New Jersey Pine Barrens, and East and West Gulf Coastal Plain.
Summary and Conclusions Outline
I. Stream Length Population Estimates. (Uncertainties surrounding these estimates are
addressed in Tables E-5 and E-6.)
A. Acidic Stream Reaches (ANC < 0 fieq L'1 during spring baseflow)
1. 2.7% (5,429 km) of the combined length of streams in the NSS-I target population
were acidic.
xlviii
-------
2.
4.4% (4,851 km) of the combined length of streams in the NSS-I target population
in the Mid-Atlantic Region were acidic.
ANC < 0 jteq
Mid-Atlantic Subregions
Northern Appalachians
Poconos/Catskills
Valley and Ridge
Mid-Atlantic Coastal Plain
%
7.0
3.6
0.8
6.3
Length (km")
1,524
543
257
2,527
3.
0.6% (578 km) of the combined length of streams in the NSS-I target population
in the Southeast Region were acidic.
Southeast Subregions
Florida
Southern Appalachians
Southern Blue Ridge
Piedmont
Ozarks/Ouachitas
ANC < 0 /zeq L'1
% Length (km)
12.0 461
0.5 117
No acidic reaches observed
No acidic reaches observed
No acidic reaches observed
B. Stream Reaches with Low Spring Baseflow ANC
1. 11.7% (23,487 km) of the combined length of streams in the NSS-I target popula-
tion had ANC < 50 /jeq L"1, a value commonly used to indicate extreme sensi-
tivity to acidic deposition. 15.5% (17,067 km) of the estimated stream length in
the Mid-Atlantic Region and 7.1% (6,420 km) of the estimated stream length in
the Southeast Region had ANC < 50 #eq L"1.
NSS-I Subregions
Northern Appalachians
Poconos/Catskills
Valley and Ridge
Mid-Atlantic Coastal Plain
Southern Appalachians
Southern Blue Ridge
Piedmont
Ozarks/Ouachitas
Florida
ANC < 50 A*eq L'1
%_ Length (km)
17.1 3,713
10.6 1,606
6.5 2,111
23.9 9,636
3.5 763
7.8 706
7.1 2,390
0.9 205
61.2 2,356
23.4% (47,060 km) of the combined length of streams in the NSS-I target
population had ANC < 100 jueq L'1. 23.9% (26,274 km) of the estimated stream
length in the Mid-Atlantic Region and 22.9% (20,786 km) of the estimated stream
length in the Southeast Region had ANC < 100 /xeq L"1.
xlix
-------
NSS-I Subreeions
Northern Appalachians
Poconos/Catskills
Valley and Ridge
Mid-Atlantic Coastal Plain
Southern Appalachians
Southern Blue Ridge
Piedmont
Ozarks/Ouachitas
Florida
ANC < 100 #eq L'1
%_ Length (km)
31.9 6,927
23.8 3,600
14.8 4,852
27.0 10,895
12.7 2,778
44.1 3,987
21.0 7,044
19.2 4,307
69.4 2,670
3. 48.4% (97,125 km) of the combined length of streams in the NSS-I target popula-
tion had ANC < 200 /*eq L"1, a value commonly used to indicate potential sensi-
tivity to acidic deposition. 47.6% (52,327 km) of the estimated stream length in
the Mid-Atlantic Region and 49.3% (44,799 km) in the Southeast Region had ANC
< 200 /JCQ L"1. Within the nine NSS-I subregions, the percentage of stream
length with ANC < 200 /zeq L"1 ranged from 28.0% (6,130 km) in the Southern
Appalachians to 78.4% (7,084 km) in the Southern Blue Ridge.
4. Regional variation in stream water ANC was closely associated with concentra-
tions of base cations (e.g., sodium, potassium, magnesium, and calcium), indicating
that local geology is probably the primary factor controlling the sensitivity of
streams to acid inputs.
C. pH Distributions
1. Patterns in stream pH tended to follow the distribution of ANC. An estimated
2.9% (5,900 km) of the combined stream length in the NSS-I had pH < 5.0, 8.3%
(16,708 km) had pH < 5.5, and 17.8% (35,797 km) had pH < 6.0.
II. Probable Sources of Acidity
A. Atmospheric Deposition
1. The estimated 5,429 km of acidic NSS-I target stream length corresponds with an
estimated 3,476 upstream and 1,323 downstream reach ends (6.1% and 2.3% of the
target population, respectively). 63% of the acidic upstream ends and 55% of the
acidic downstream ends were classified into a high-interest subpopulation whose
major source of acidity is most likely to be acidic deposition. The approximate
length of this high-interest subpopulation is 4,455 km. Just over half of these
high-interest reaches drained small (< 20 km2) upland forested watersheds in the
Interior Mid-Atlantic (the Poconos/Catskills, Valley and Ridge, and Northern
Appalachians subregions), whereas most of the remainder occurred in lowland
drainages of the Mid-Atlantic Coastal Plain subregion, primarily in the New
Jersey Pine Barrens. However, most of the streams in the Pine Barrens are also
influenced by organic acidity and many are likely to have been acidic at least
since the early 1900's.
1
-------
An estimated 46% (11,505) of the upstream ends of stream reaches in the Interior
Mid-Atlantic Region were located in forested uplands. Of these forested upland
reaches, an estimated 11% (1,271) were acidic at their upstream ends and the
major source of their acidity is most likely acidic deposition. An estimated 34%
(3,857) of the upstream ends of these forested upland stream reaches had ANC
< 50 /zeq L"1 and the major source of their acid anions is most likely to be
atmospheric deposition.
Subregional median stream water sulfate concentrations (excluding streams with
substantial watershed sulfate sources) were highly correlated with subregional
median wet sulfate deposition rates. This finding strongly supports the
hypothesis that atmospheric inputs of sulfate are the major source of sulfate in
these streams. This correlation was also observed in lake populations in the
National Lake Survey.
B. Organic Acidity
1. Organic acidity, which characteristically results from the natural decay of
vegetation, is likely to be the major source of acidity in an estimated 79% of the
678 acidic upper reach nodes and more than 99% of the 225 acidic lower reach
nodes in the Florida subregion. Similarly, natural organic acids were estimated
to be the major source of acidity in 56% of the 1,334 acidic upper nodes and 47%
of the 772 acidic lower nodes in the Mid-Atlantic Coastal Plain subregion.
C. Acid Mine drainage
1. Unless specifically noted otherwise, all the estimates in this summary exclude
streams affected by acid mine drainage. However, acid mine drainage impacts
are common in some NSS-I subregions. In the Northern Appalachians subregion,
they make up roughly 10% of the stream population initially sampled in the
NSS-I. The NSS-I does not include all sizes of streams or geographic areas
affected by acid mine drainage; thus, a comprehensive estimate of acid mine
drainage impacts cannot be made. Nevertheless, in streams in the size range
sampled by the NSS-I, an estimated 1,377 reaches (4,594 km) in Pennsylvania,
West Virginia, and Tennessee were estimated to be acidic due to acid mine
drainage.
li
-------
p
oo
oo
oo
8 2
cr
ex
W
O
CM
VI
wo
V!
VI
O
oo e-
r- o\ ON —«
CO 0
-H t\
coooo\ooco— iin-^
irtOOO—•
mo* co>noot~* co
ts —< TT
* ts* ON
CO
* * * *
o C
a w DO
2
•a
S5
'•
D.
o\ o\ vo
vo oo O
OO
vo
oo
0\
VO
CN
01
CO
CM
CM
CO
-H CO
-^ CN
oo
-H 00
o\
VO
ON
(N -H
co TT
O\ —<
00 -H
cs TT
VO
ON
CN
VO
ON
ON
ON
ON
-H f~
«
CO
CN
*»
X
.=; a
3» a".
oQ
'
00 M HH
00
VO,
vo"
VO
CO
CO
CO
CO
VO
oo
s-i
11
s^
jj'S
-------
5
S
oo
a
oo
•a
o
|
CD
IH
CD
a
IH
I
§
£
CD
OO
CD
%
•*••
CD
IH
CD
PH
SCD
00
•J3
CO „
W Pi
S
A
VI
O
VI
I
O
w
^
PH
S
A A
OO*
* * * — <
* en* f-****
OO (N C-4
ON tN m
00 CM tN
•—i tN tN
VO C7\ en
OO (N — <
•^r oo v o
(M -^ CO cN
tN
O
V> OO
tN -H
CT\
O\
r-
ON
o
ON
%
vo
oo
en
ON
m oo
cn m
en
m
so m
— < * m — c
G - G -
« X « X
So
J'S
*^ •*-»
oo >S
'a *
.2 < I
•
tvo
o
U-l
ON
CM
tN
t/3
•a
y>
I
^s
« 8
rv •*•<
g-a
•5* Ul
rt ™
o
2 2
S
sl
a v co
CD * .2
,2? c8 2?
!§"
3 |5
. Qi -4-»
M es °
rP
tt -U
-x
CO CD
cu
-------
Table E-5. Population Estimates of the Combined Length (km) and Percentage of NSS-I
Target Stream Reaches with Spring Baseflow ANC Less than Reference Values
(Standard Errors in Parentheses).*
ANC<0
Subregion
Poconos/Catskills
N. Appalachians
Valley & Ridge
MA Coastal Plain
S. Blue Ridge
Piedmont
S. Appalachians
Ozarks/Ouachitas
Florida
SUBTOTALS
Interior MA
Interior SE
Mid-Atlantic (MA)
Southeast (SE)
Total NSS-I
Length
543
(270)
1,524
(750)
257
(210)
2,527
(1,200)
*
-
*
-
117
(120)
*
-
461
(160)
2,324
(1,000)
117
(120)
4,851
(1,600)
578
(210)
5,429
(1.500)
%
3.6
(1.8)
7.0
(3.5)
0.8
(0.6)
6.3
(2.9)
*
-
*
-
0.5
(0.5)
*
-
12.0
(4.1)
3.3
(1.4)
0.1
(0.1)
4.4
(1.4)
0.6
(0.2)
2.7
(0.8)
ANC
Length
1,606
(500)
3,713
(920)
2,111
(990)
9,636
(2,700)
706
(250)
2,390
(1,300)
763
(440)
205
(150)
2,356
(530)
7,431
(1,700)
4,064
(1,300)
17,067
(3,100)
6,420
(1,500)
23,487
(3.400)
< 50
O/o
10.6
(3.3)
17.1
(4.2)
6.5
(3.0)
23.9
(6.6)
7.8
(2.8)
7.1
(3.9)
3.5
(2.0)
0.9
(0.6)
61.2
(14)
10.7
(2.4)
4.7
(1.5)
15.5
(2.8)
7.1
(1.7)
11.7
(1.7)
ANC < 200 Total Length
Length
5,489
(1,100)
12,935
(2,200)
12,811
(3,400)
21,091
(4,400)
7,084
(940)
13,554
(2,900)
6,130
(1,700)
15,092
(2,500)
2,939
(590)
31,235
(4,300)
41,860
(4,300)
52,327
(6,200)
44,799
(4,400)
97,125
(7.700)
O/o
36.2
(7.3)
59.5
(10)
39.2
(10)
52.3
(11)
78.4
(10)
40.4
(8.5)
28.0
(8.0)
67.1
(11)
76.4
(15)
44.9
(6.2)
48.1
(4.9)
47.6
(5.6)
49.3
(4.8)
48.4
(3.8)
(km)
15,144
(1,912)
21,738
(2,746)
32,687
(4,492)
40,296
(5,799)
9,036
(960)
33,531
(4,402)
21,892
(2,807)
22,480
(2,507)
3,848
(678)
69,569
(5,601)
86,939
(5,871)
109,865
(8,063)
90,787
(5,910)
200,652
(9.996)
I using linear interpolation between upper and lower reach nodes. Standard errors
were approximated by an ad hoc procedure using the variances of separate length estimates
based on the upstream and downstream nodes.
* No samples observed below this reference value; estimated percentage is less than 1%.
NOTE: To calculate upper and lower one-sided 95% confidence bounds, multiply the standard
error by 1.645 and add or subtract that value from the length estimate. To calculate
the two-sided 95% confidence bounds, multiply the standard error by 1.96.
liv
-------
Table E-6. Population Estimates of the Combined Length (km) and Percentage of NSS-I
Target Stream Reaches with Spring Baseflow pH Less than Reference Values
(Standard Errors in Parentheses).*
Subregion
Poconos/Catskills
N. Appalachians
Valley & Ridge
MA Coastal Plain
S. Blue Ridge
Piedmont
S. Appalachians
Ozarks/Ouachitas
Florida
SUBTOTALS
Interior MA
Interior SE
Mid- Atlantic (MA)
Southeast (SE)
Total NSS-I
pH<
Length
550
(290)
1,424
(700)
257
(260)
3,147
(1,300)
*
-
*
-
*
-
*
-
522
(250)
2,231
(780)
*
-
5,378
(1,500)
522
(250)
5,900
(1.600)
; 5.0
%
3.6
(1.9)
6.6
(3.2)
0.79
(0.8)
7.8
(3.3)
*
-
*
-
*
-
*
-
13.6
(6.5)
3.2
(1.1)
*
-
4.9
(1.4)
0.57
(0.3)
2.9
(0.8)
pH<
Length
906
(420)
1,870
(710)
1,937
(1,300)
9,565
(3,000)
*
-
*
-
313
(310)
410
(290)
1,708
(440)
4,712
(1,700)
723
(430)
14,277
(3,400)
2,431
(800)
16,708
(3.400)
5.5
%
6.0
(2.8)
8.6
(3.2)
5.9
(4.0)
23.7
(7.5)
*
-
*
-
1.4
(1.4)
1.8
(1.3)
44.4
(12)
6.8
(2.4)
0.83
(0.5)
13.0
(3.1)
2.7
(0.9)
8.3
(1.7)
pH< I
Length
1,354
(520)
3,044
(900)
4,116
(1,900)
18,707
(4,300)
*
-
2,390
(1,200)
920
(540)
2,437
(990)
2,828
(620)
8,514
(2,300)
5,747
(1,700)
27,221
(4,800)
8,576
(1,900)
35,797
(5.200)
5.0 1
o/0
8.9
(3.4)
14.0
(4.2)
12.6
(5.9)
46.4
(11)
*
-
7.1
(3.7)
4.2
(2.5)
10.8
(4.4)
73.5
(16)
12.2
(3.4)
6.6
(1.9)
24.8
(4.4)
9.5
(2.1)
17.8
(2.6)
7otal Length
(km)
15,144
(1,912)
21,738
(2,746)
32,687
(4,492)
40,296
(5,799)
9,036
(960)
33,531
(4,402)
21,892
(2,807)
22,480
(2,507)
3,848
(678)
69,569
(5,601)
86,939
(5,871)
109,865
(8,063)
90,787
(5,910)
200,652
(9.996)
# Calculated using linear interpolation of [H+] between upper and lower reach nodes. Standard
errors were approximated by an ad hoc procedure using the variances of separate length
estimates based on the upstream and downstream nodes.
* No samples observed below this reference value; estimated percentage is less than 1%.
NOTE: To calculate upper and lower one-sided 95% confidence bounds, multiply the standard
error by 1.645 and add or subtract that value from the length estimate. To calculate
the two-sided 95% confidence bounds, multiply the standard error by 1.96.
Iv
-------
THIS PAGE INTENTIONALLY LEFT BLANK
Ivi
-------
SECTION 1
INTRODUCTION
1.1 OVERVIEW
The role of atmospheric deposition in the acidification of surface waters remains one of
the most important and controversial environmental issues of the 1980's. Studies on individual
water bodies and on regional lake and stream populations have produced data suggesting that pH
and acid neutralizing capacity (ANC) have declined over the past half century in some European
and North American surface waters (Beamish and Harvey, 1972; Beamish et al., 1975; Oden, 1976;
Wright and Gjessing, 1976; Schofield, 1982; Haines and Akielaszek, 1983; Smith and Alexander,
1983). Declines that have been observed or estimated are most commonly attributed to acidic
atmospheric deposition resulting from fossil fuel combustion (Drabljis and Tollan, 1980; National
Research Council, 1981,1983,1984; U.S. EPA, 1984a; Jeffries etal., 1985; Sullivan etal., 1988b).
However, alternative hypotheses for site specific and regional acidification have been discussed
and debated in recent literature (Rosenqvist, 1978; Krug and Frink, 1983; Hansen and Hidy, 1982;
Havas et al., 1984; Howells, 1984; Cogbill et al., 1984; Lefohn and Brocksen, 1984; Krug et al.,
1985; Pierson and Chang, 1986).
A prerequisite that is essential to understanding the environmental effects of acidic
deposition in the United States is the availability of quantitative estimates, with known
confidence, of the present status and extent of acidic surface waters over broad regional areas.
Such estimates cannot be made based on previously conducted independent surveys, monitoring
networks, and site-specific research, because these earlier studies lack one or more factors that
would allow quantitative regional extrapolation.
Current synoptic-scale studies being performed on monitoring networks cannot be used to
characterize surface water quality because they have inadequate statistical sampling designs or
regional coverage, inconsistent field or laboratory methods, or insufficient chemical measure-
ments. Also, these studies usually do not have adequate quality assurance data by which
potential bias among data collected from different studies can be evaluated. The lack of synop-
tic data, in turn, thwarts the regional extrapolation of site-specific research results. It is
almost impossible to quantitatively extrapolate the results from intensive process-oriented
research or long-term monitoring on a small number of watersheds to the larger lake or stream
population comprising the resource at risk in a given geographic region. Such regional extrapo-
lation requires a classification of surface waters based on, or linked with, a statistically rigorous
description of the physical and chemical characteristics of regional surface waters. Quantitative
information is seldom available on whether research sites are broadly typical of the majority of
systems in a region, are representative of a relatively small but perhaps important class of
surface waters within a subregion, or are in fact unique systems. Research sites are often rela-
tively pristine, probably because of some unique or uncommon characteristic.
1.2 NATIONAL SURFACE WATER SURVEY
In response to the need for knowledge regarding the present extent of acidic or potentially
susceptible aquatic resources and their associated biota, the U.S. Environmental Protection
Agency and cooperating scientists were asked in 1983 to design a program that would satisfy
five major goals:
-------
1. Characterize the chemistry of surface waters (both lakes and streams) in regions of
the United States presently believed to be potentially susceptible to change as a
result of acid deposition.
2. Examine associations among chemical constituents and define important factors that
may affect surface water chemistry.
3. Determine the biological resources within these systems.
4. Evaluate correlations among surface water chemistry and the corresponding biological
resources.
5. Quantify any regional trends in surface water chemistry and biological resources.
The program designed to meet these goals was designated the National Surface Water
Survey (NSWS). The NSWS became an integral part of the National Acid Precipitation
Assessment Program (NAPAP), an interagency research, monitoring, and assessment effort
mandated by Congress in 1980. NAPAP provides policy makers with technical information
concerning the extent and severity of the effects of acid deposition on human, terrestrial,
aquatic, and material resources.
The NSWS design (Figure 1-1) incorporates two parallel components, the National Lake
Survey (NLS) and the National Stream Survey (NSS), in order to satisfy the five major research
goals. In both components, early project phases contribute to the design and interpretation of
subsequent phases. The synoptic surveys of lake and stream chemistry performed in the early
phases of the NLS and NSS contribute substantially to the design of subsequent project phases
and are essential to the regional extrapolation of their results.
The NSWS design grew out of the recognition that while it is clearly not feasible to
perform intensive, process-oriented studies or monitoring programs on all surface waters within
the United States, it is equally inappropriate to study a few systems that later may be found to
have atypical biological and chemical characteristics. Therefore, each component of the NSWS
begins with Phase I, a synoptic survey designed to characterize and quantify the chemistry of
lakes and streams throughout the United States, focusing on the areas expected to contain the
majority of low alkalinity waters.
The NSWS was designed to overcome the obstacles to assessment discussed in Section 1.1.
Lakes and streams were sampled on a regional basis using a statistically rigorous survey design,
appropriate standardized field and analytical techniques, a relatively complete set of chemical
and physical measurements, and an adequate quality assurance/quality control (QA/QC) program
to maximize confidence in the resulting data. The initial survey component (Phase I) provides a
snapshot of the present condition of surface waters in regions of the United States most likely
to be affected by acid deposition. The Phase I data also serves as a basis for classification of
lakes and streams. Such classifications will allow the regional extrapolation, with known
confidence, of results from past and future intensive studies on both high-interest aquatic
subpopulations and individual lakes and streams.
The impact of acid deposition on biological resources is a major concern, but rather than
beginning a study of the problem with a biotic survey of all surface waters in a region, regard-
less of water quality, it is more effecient to first characterize surface waters in terms of the
physico-chemical factors that are expected to impact biota. The present study design, based on
the Phase I chemical classification, can be used not only to quantify the present status of
-------
NATIONAL SURFACE WATER SURVEY (NSWS)
NATIONAL LAKE SURVEY (NLS)
SYNOPTIC CHEMISTRY
Eastern Lake Survey-I (1984)*
Western Lake Survey-1 (1985)*
EASTERN LAKE SURVEY-IE
Temporal Variability in
Northeast (1986-87)
Biological Resources in
Upper Midwest (1986)
NATIONAL STREAM SURVEY (NSS)
SYNOPTIC CHEMISTRY
Phi Pilot (S. Blue Ridge) (1985)*
Ph I Mid-Atlantic and
Southeast (1986)*
Episodic Effects (1988)
Biological Resources (1988)
Long-Term Monitoring (Planned)
(TIME Project)
Figure 1-1. Organization of the National Surface Water Survey, showing the two major
components, National Lake Survey and National Stream Survey, and their
relationship to later phases of study. Dates in parentheses are year of field data
collection. Final reports are completed for those marked with an asterisk.
-------
aquatic resources, but also to examine correlative relationships within relatively homogeneous
lake and stream types. It also allows the selection of geochemically representative sites for
more studies of intensive biological characteristics, temporal variability in water chemistry, and
long-term changes.
In the second phase of the NLS, the Eastern Lake Survey - Phase II (ELS-II), and in the
Episodic Response Project (ERP), short-term (seasonal, weekly, or episodic) variability in water
chemistry is quantified within and among representative lakes and streams in each geographical
region. The definition of representativeness is based on Phase I water chemistry and associated
hydrology, aquatic organisms, regional acid deposition inputs, land use, physiography, and other
basin characteristics. Some regionally representative sites will later become the foundation for a
long-term monitoring program, the Temporal Integrated Monitoring of Ecosystems (TIME), pres-
ently being designed (Thornton et al., 1987a) to detect and quantify future chemical and bio-
logical changes in potentially susceptible lakes and streams in critical subregions. Many lakes
and streams that have been the focus of intensive or long-term studies in the past are included
in the Phase I lake and stream surveys as special interest sites. Such sites found to be repre-
sentative of important classes of aquatic systems in their respective regions could serve as
nuclei for long-term monitoring efforts.
Phase I of the NLS has been completed. A total of 1,798 lakes in the eastern United
States were sampled in the fall of 1984 (Linthurst et al., 1986), and 752 lakes were sampled in
selected areas of the western United States in the fall of 1985 (Landers et al., 1987). Phase II
field work measuring seasonal variability in lakes took place in 1986 and 1987. The ERP will
begin field work in 1988, concentrating on short-term chemical variation in streams and the
effects of these changes on fish populations. The NSS has been implemented in two regions, the
Mid-Atlantic and a large area of the Southeast. The status of the NSS is discussed further in
Section 1.3.
1.3 NATIONAL STREAM SURVEY
1.3.1 Phase I Goals and Objectives
Planning for Phase I of the National Stream Survey (NSS-I) began in mid-1984 and resulted
in a Draft Research Plan (U.S. EPA, 1984b). NSS-I was designed to chemically and physically
characterize a target population of streams existing within any relatively homogeneous physio-
graphic region, based on a probability sample of those streams. It has the joint major goals of
describing and classifying streams in the target population. The specific primary objectives of
NSS-I are:
1. To determine the percentage, extent (number, length, and drainage area), location, and
chemical characteristics of streams in the United States that are presently acidic, or
that have low ANC and thus might become acidic in the future.
2. To identify streams representative of important classes in each region that might be
selected for more intensive study or long-term monitoring.
The NSS-I was designed to achieve these objectives within known confidence limits. It was
also designed to allow these objectives to be met for any chemical or physical variable
measured. For example, the NSS-I design allows an estimation of the percentage of the stream
population within a given region having sulfate, nitrate, aluminum, or calcium concentrations
-------
above or below any criterion value of interest. Should some other characteristic of target
streams, such as frequency or duration of biologically deleterious episodes, or sensitivity to
chronic acidification, be acceptably defined in the future, based on one or more of the variables
measured on sample streams, the NSS-I design will also permit the regional extrapolation of such
information. The sampling design lends itself to many comparative evaluations. For example,
other questions that could be investigated by means of the design include:
1. Do stream ANC and pH decrease as elevation increases?
2. What is the relationship between stream ANC (or pH) and watershed area?
3. Are acidic streams found primarily in small watersheds?
4. Are most acidic streams found in geographic areas with high atmospheric acid
deposition?
5. Are sulfate and base cation concentrations in streams of different regions of the
United States correlated primarily with regional deposition chemistry or with
watershed soils and geology?
6. How may NSS-I chemical data be used to refine existing alkalinity maps?
7. What associations exist among water. chemistry, land use, vegetation type, and
geographic data?
8. To what extent can sources of acidity be inferred from chemical, physical, and
geographical data collected by the NSS?
Except for question number 6, dealing with the refinement of alkalinity maps, all these questions
are addressed in this report, at least at a broad regional level. More detailed analyses addres-
sing many of these questions are underway and will be reported in the scientific literature.
1.3.2 National Stream Survey Components
The highest priority for NSS field activities was placed on low alkalinity subregions that
have high atmospheric acid deposition rates, but do not contain abundant lakes. Unlike sub-
regions covered by the Eastern and Western Lake Surveys (Figure 1-2), no synoptic data (exist-
ing lake or stream data) were available for these subregions from which rough inferences about
regional stream chemistry could be made. Such low alkalinity subregions included the New
Jersey Pine Barrens and Chesapeake lowlands, the northern portion of the Valley and Ridge
Province, and the Northern Appalachians in the Mid-Atlantic United States.
The Poconos/Catskills in the southern part of the glaciated Northeast also were given high
priority because data in this subregion (like data from the Southern Blue Ridge) could be com-
pared with ELS data (Linthurst et al., 1986) for the same land area, allowing us to evaluate the
feasibility of making regional predictions of stream chemistry on the basis of ELS results.
In the southeastern subregions shown in Figure 1-2 (Northern and Central Blue Ridge,
Southern Appalachians, Southern Valley and Ridge, the Piedmont, and the Ozarks/Ouachitas),
where atmospheric deposition rates are lower and where soils are expected to have higher
sulfate adsorption capacities, a less intensive screening survey was undertaken in the spring of
1986. This Southeast Screening survey also included portions of Florida, in order to test the
NSS-I design and logistic protocols in lowland stream networks.
Concurrent with the Mid-Atlantic and Southeast Screening components of the NSS-I, an
episodes pilot study was undertaken in the Mid-Atlantic Region to provide a preliminary assess-
ment of the frequency, duration, and causes of short-term episodic changes in chemistry during
storm events.
-------
Eicri 3K —V1UOA *** =» 3C«>—!U
|_j^5 5 I i* = a — b* «/» oaeoeaci-.
o
I
a
o
a
s
•8
•o
s >•
5H
s w
2s
0 3
S^
o «
to TJ
I I
$3
-
-------
1.3.2.1 NSS-I Pilot Survey—
At an initial NSS-I design workshop held in Washington, D.C., in December 1984, peer
reviewers recommended that the full-scale survey be preceded by a pilot study, the findings of
which might optimize the design of a full-scale effort. The NSS-I Pilot Survey was conducted
in the Southern Blue Ridge province during the spring and summer of 1985, and is described in
detail by Messer et al. (1986), Drouse et al. (1987), and Knapp et al. (1987). A probability
sample of 54 stream reaches was drawn from a target population of blue-line streams on
l:250,000-scale topographic maps, draining catchments less than 155 km2 (60 mi2) and satisfying
specific site inclusion criteria. Downstream ends of these reaches were visited three times
during the spring season and once during the summer. Upstream ends of the reaches were
visited during the spring (incomplete sample) and during the summer. The objectives of the
pilot survey were as follows:
1. Test the ability of the NSS-I sampling design to meet Phase I objectives, based on
analysis of data collected during the pilot survey.
2. Evaluate the Phase I logistics plan (including safety aspects and uncertainties
concerning legal and physical site access) and alternative methods of collection,
handling, and chemical analysis of samples.
3. Develop and test a data analysis plan for Phase I, using actual data collected in the
pilot survey.
The NSS-I Pilot Survey demonstrated that a regional-scale synoptic survey of streams is
logistically feasible and that it will produce population estimates, with known confidence, for
important variables such as pH and ANC. These population estimates appear to be robust and
are not particularly sensitive to small chemical changes that take place during the spring
sampling season (Messer et al., 1986, 1988). In addition, Messer et al. (1986) concluded that
spring temporal variability does not preclude chemical classification of target systems in the
Southern Blue Ridge, providing that data collected during storm episodes are not used in such
analyses.
The draft research plan (U.S. EPA, 1985), implementation, and data analysis of the full-
scale Phase I stream survey incorporated a number of changes as a result of pilot survey
experience. The more important changes, peer reviewed in January 1986, included:
1. Making minor alterations in the stream reach inclusion criteria and the method of
selecting sample reaches to better address the assessment objectives of the survey and
to increase sampling efficiency.
2. Reducing sampling to two spring visits prior to leaf-out, to satisfy the classification
objective, or to one visit, to satisfy the objective of population estimation.
3. Sampling reaches at both their upstream and downstream ends on each visit to
characterize spatial variability within reaches.
4. Increasing sample holding time protocols to 24 hours to allow central placement of
mobile analytical laboratories, thereby greatly increasing the logistical efficiency of
the survey.
-------
5. Changing a number of field measurement techniques to improve accuracy and
reliability.
6. Altering QA/QC and data management techniques to increase efficiency.
7. Developing new data analysis techniques to aid interpretation in an assessment
context.
1.3.2.2 Mid-Atlantic Phase I Survey—
The focus of the full-scale NSS-I field activities in 1986 was on the Mid-Atlantic region
and the survey was designed to fully meet the NSWS primary and secondary objectives outlined
in subsection 1.3.1. In the Mid-Atlantic component of the NSS-I, more than 1,000 water chem-
istry samples were collected from approximately 270 stream reaches in an area bounded by the
Pocono/Catskill Mountains to the north, the North Carolina-Virginia state line to the south, the
approximate western boundaries of Pennsylvania and West Virginia to the west, and the Atlantic
Ocean to the east (Figure 1-2). This region was subdivided into four physiographic subregions
on the basis of climate, geology, soils, land use and vegetation. The Mid-Atlantic region is
known from previous water quality data (Omernik and Powers, 1983; Omernik and Kinney, 1985;
Lynch and Dise, 1985; Witt and Barker, 1986; Murdoch, 1986) to contain waters of low alkalinity.
The region generally has high acid deposition rates (NADP, 1987) and many upland areas are
known to be overlain with thin soils with a relatively low capacity to neutralize acidity. Field
sampling was more intensive in the Mid-Atlantic region than in the Southeast. Each reach was
sampled twice during the spring baseflow period, March 15 to May 15, at each of the upstream
and downstream reach ends.
1.3.2.3 Southeast Screening—
Previously existing stream water quality databases are not adequate for making quantitative
regional chemical distribution estimates in the southeastern United States. The Southeast
Screening component of the NSS-I was originally designed to provide the minimal synoptic
chemical data required to decide whether Phase I stream sampling efforts should be extended
beyond those high priority areas targeted in the Mid-Atlantic. National Lake Survey results
(Linthurst et al., 1986; Landers et al., 1987) are useful in targeting potential areas in the
Northeast, Midwest, and West that were not sampled in the NSS-I but where substantial numbers
of acidic and low ANC streams may exist. However, virtually no data were available for the
southeastern states other than the historical data depicted on regional alkalinity maps (Omernik
and Kinney, 1985; Omernik and Griffith, 1986a,b) and synoptic surveys in relatively small geo-
graphic areas such as the Southern Blue Ridge (Messer et al., 1986; Winger et al., 1987).
The sampling/statistical design of the Southeast Screening component allows regional
characterization (e.g., population distribution estimates with known confidence of acidic and low
ANC streams), as did the Mid-Atlantic Phase I Survey component. However, each stream in the
Southeast Screening was sampled only once at each reach end, thereby providing less stable
chemical index values with no estimate of temporal variance during the spring sampling period.
As in the Mid-Atlantic, stream water quality sampling was avoided during extreme hydrologic
events (episodes). The single spring base flow sample was not originally expected to provide
enough information for the robust classification of sample streams that would identify regionally
representative sites or important classes of streams for more intensive study of episodes, water-
shed manipulation, or long-term temporal changes. However, results from the NSS-I Pilot Survey
8
-------
(Messer et al., 1986,1988) indicated that chemical classifications of Southern Blue Ridge streams
based on single spring base flow samples from each stream were virtually the same as those
based on the average of three such samples, suggesting that classifications based on Southeast
Screening data are probably robust. Without additional data, however, the stability of these
classifications cannot be quantitatively evaluated.
The Southeast Screening includes subregions of the United States expected, on the basis of
alkalinity maps (Omernik and Powers, 1983), to contain low alkalinity surface waters of widely
different geochemical types experiencing a broad range of acid deposition rates. The region
includes the Southern Piedmont, the Ozarks/Ouachitas, the Florida panhandle and peninsula, and
parts of the Southern Appalachians that were not sampled as part of the NSS-I Mid-Atlantic
Survey or the NSS-I Pilot Survey in the Southern Blue Ridge. The Southern Appalachians and
the Ozarks/Ouachitas represent two types of upland watersheds separated by hundreds of miles
that experience different deposition rates. The Piedmont is characterized by lower elevation,
predominantly clearwater streams. Florida contains both colored and clearwater lakes and
streams at low elevations. The target areas studied in the Southeast Screening also provide an
indication of the types and occurrence of low ANC waters to be found in large areas of the
South Atlantic and Gulf coastal plains. The field effort in the Florida subregion provides a test
of the utility of the NSS-I design and logistical protocols in lowland stream networks. Approxi-
mately 400 samples were collected from 200 stream reaches in the Southeast Screening area
between March 15 and May 15,1986, one sample from each upstream and downstream end of the
sample reaches.
1.3.2.4 Mid-Atlantic Episodes Pilot--
The primary objective of the Episodic Response Project (ERP) is to predict the magnitude,
frequency, duration, and causes of episodic events in streams and lakes (Thornton et al., 1987b).
These events cause marked shifts in pH and associated chemistry from baseflow conditions. In
areas dominated by snowpack, such episodes result from snowmelt and are often exacerbated by
warm days or rainfall. In warmer regions, episodes are usually associated with rainfall. The
Mid-Atlantic Episodes Pilot, a precursor to the ERP, was undertaken in parallel with the NSS-I
Mid-Atlantic field effort during the spring of 1986. This pilot effort was aimed at providing a
preliminary assessment of the frequency, duration, and causes of such episodes in the Mid-
Atlantic. It was also aimed at testing a regional synoptic design for episodes sampling. Messer
and Eshleman (1987) summarized the results of the Mid-Atlantic Episodes Pilot and concluded
tnat a synoptic survey of episodes was not cost effective. Rather, a model-based study would
be needed to satisfy data needs concerning episodic effects. The results of the Episodes Pilot
were used, however, to aid in the design of a cost-effective, regional study of episodes, the
ERP, to be implemented in 1988 (Thornton et al., 1987b). The results of the Mid-Atlantic
Episodes Pilot are summarized by Messer and Eshleman (1987) and will not be discussed further
in this NSS-I data report.
1.3.3 The NSS-I Data Report
This report documents the design and methods of the NSS-I Mid-Atlantic and Southeast
Screening surveys, reports the results of these studies, and discusses current interpretations.
Section 2, Survey Design, defines the target stream population of interest and describes
how representative samples were drawn from it in time and space. It then outlines the pro-
cedure by which regional population estimates of stream chemistry were made.
-------
Section 3, Methods, describes procedures for field sample collection, handling, and chemical
analysis. It also describes the QA/QC protocols, including data management procedures such as
verification, validation, and the construction of the database used for analysis.
Section 4, Data Quality Assessment, discusses aspects of data completeness, chemical
detectability, analytical accuracy, and sampling/analytical precision in light of the results of
blank, duplicate, and audit sample analyses.
Section 5, Target Population—Physical Characteristics, describes the physical characteristics
of the stream resource sampled. The section presents total resource estimates for the NSS-I
subregions, along with distributions of such characteristics as drainage area, Strahler order, and
elevation. Section 5 also describes the general categories of streams included ,in the sampled
population, such as those affected by acid mine drainage. By removing several types of streams
from the general categories, a more restricted target population of interest for assessing acid
deposition effects is delineated. This is the population for which population distribution
estimates are made in Section 6.
Sections 6 through 9 report and discuss the major results of the NSS-I, including those
previously published for the NSS-I Pilot Survey in the Southern Blue Ridge (Messer et al., 1986).
Section 6 summarizes the regional distribution estimates of stream chemistry, which are the pri-
mary outputs of the NSS-I. Section 7 discusses the uncertainties regarding these estimates in
terms of temporal variability in baseflow chemistry. Section 8 presents a geochemical inter-
pretation of the relationships observed among measured chemical constituents in stream water.
Section 9 describes the chemical characteristics and geographic distribution of a selected group
of target streams of highest assessment interest—those that are acidic or that have low ANC.
The discussion covers the potential sources of acidity in these streams. Those for which
substantial acidification by atmospheric deposition is very unlikely are systematically eliminated
from consideration. Section 9 then examines the chemistry of the remaining streams (those for
which atmospheric deposition remains a possible cause of acidification).
Section 10 examines evidence for acidification of NSS-I sample streams and summarizes our
current regional understanding of the potential for acid deposition impacts on streams in the
Mid-Atlantic and Southeastern United States.
Finally, Section 11 summarizes the important findings of the NSS-I in the Mid-Atlantic and
Southeast, Section 12 lists the references, and Section 13 provides a glossary of abbreviations
and symbols, and defines terms used throughout the report.
Volume II, the data compendium, is a companion document to this report. It contains data
listings and additional statistical analyses of the NSS-I physical and chemical data.
10
-------
SECTION 2
SURVEY DESIGN
2.1 OVERVIEW
Phase I of the National Stream Survey (NSS-I) employed a randomized, systematic technique
for selecting a probability sample of stream reaches (Overton, 1986, 1987; Messer et al., 1986)
within areas of the United States expected, on the basis of previous chemical data (Omernik and
Powers, 1983; Omernik and Kinney, 1985), to contain waters of low acid neutralizing capacity
(ANC). Field crews then visited these stream sampling locations, made field physical and
chemical measurements, and collected water samples for later analysis (Section 3). Identification
of the target population for the NSS-I involved two separate activities: (1) identification and
prioritization of geographical regions expected to contain significant numbers of low pH or low
ANC streams, and (2) identification of the types of streams within these regions that were of
interest from the standpoint of assessing the present extent of effects due to acid deposition.
The highest priority for NSS-I field activities was placed on low alkalinity subregions that
have high atmospheric acid deposition rates, but that do not contain abundant lake resources.
Subregions of high interest that do contain such lake resources (the Northeast, Upper Midwest,
and West) were sampled by the Eastern and Western Lake Surveys (ELS and WLS) (Linthurst et
al., 1986; Landers et al., 1987).
Within the NSS-I subregions, target stream reaches were identified as those that have
drainage areas less than 60 square miles (155 km2) at their lower ends, but that are large
enough to be represented as blue lines on l:250,000-scale U.S. Geological Survey (USGS) topo-
graphic maps. The development and justification for this definition is discussed in Section 2.3.
The targeted size range was accepted by reviewers as a reasonable compromise that would
include streams large enough to be recreationally and economically important for fish habitat,
yet still small enough to be susceptible to change as a result of acidic deposition.
The NSS-I relies, as do other components of Phase I of the National Surface Water Survey
(NSWS), on samples taken from a number of water bodies during an appropriate season to pro-
vide an index of the chemical characteristics of the water bodies in the region. The reasons for
choosing a spring index sampling period for the NSS-I are discussed and justified in Section 2.5.
The choice involved a trade-off between minimizing within-season and episodic chemical variabil-
ity and collecting samples during the season most likely to exhibit conditions potentially limiting
for aquatic organisms. As a result of NSS-I Pilot Survey experience (Messer et al., 1986), the
average of two spring samples taken between snowmelt and leaf-out was deemed sufficient for
establishing a chemical index for streams in the Mid-Atlantic subregions. In the Southeast,
where it is less probable that substantial amounts of present surface water acidification (not
necessarily watershed acidification) have been caused by acidic deposition, and where the Survey
was aimed at broad-scale regional screening for potential effects, one spring sample was taken
at each sampling location. To quantify and incorporate the variability between upstream and
downstream ends of reaches, chemical and physical variables were measured at both ends.
The chemical variables measured included (1) those related to biological effects—pH,
extractable aluminum, and competing ligands such as fluoride and dissolved organic carbon
(DOC), (2) others related to potential sensitivity and related geochemistry—ANC, base cations,
sulfate and other acid anions, and silica, and (3) others indicative of pollution or other relevant
factors—phosphorus, iron, ammonium, and turbidity. Samples were stabilized within 24 hours of
collection and rigorous quality assurance (QA) and quality control (QC) protocols were followed
during sample handling, chemical analysis, and data reporting (Section 3). The data were stored
11
-------
in several readily accessible data bases, again following rigorous QA/QC protocols, prior to
analysis and interpretation.
An important strength of the NSS-I design, and that of the NSWS in general, is that it
employs a regionally well-distributed probability sample of streams. This statistically rigorous
sampling frame allows quantitative regional extrapolations to be made from survey data or from
any additional information about NSS-I sample streams obtained at a later date. Furthermore,
regional extrapolations can be made using data from other streams that fit within specific
classes based on NSS-I sample data. The combination of the NSS-I sampling frame with this
stream classification allows an accurate appraisal of the regional significance of past studies and
the focusing of future detailed, process-oriented research efforts or long-term monitoring on
streams that represent important stream types or ones that are regionally characteristic.
2.2 PROJECT DESIGN CRITERIA
The design criteria of the NSS-I were intended to overcome some of the historical prob-
lems that occurred in past data sets as noted in Section 1. Not all of the design criteria in a
project like the NSS-I (aimed at description and classification) can be so narrowly and quanti-
tatively defined that data not in conformance with the criteria are useless. Instead, some of
the design criteria represent ideal targets. However, even qualitative specification of the design
criteria in the planning process resulted in significant improvements to project designs and
protocols. The following design criteria were established:
1. The target population will explicitly and accurately represent the set of streams of
interest in this investigation (those constituting the most important resource at risk
from acid deposition).
2. The NSS-I sample population will be a probability sample from the target population.
3. The suite of variables measured will be complete enough to provide information on (1)
the suitability of streams for key fish species and (2) the geochemical parameters that
can be used to classify the streams and formulate hypotheses about past and future
acidification. Phase I data may be supplemented from other data sources during later
phases to meet some of these objectives.
4. The data must be of high quality, with low and quantifiable analytical error, and with
known precision, representative of the state-of-the-art attainable in high volume
contract analytical laboratories. Data quality objectives (DQOs) for chemical data are
discussed in Section 4 and Appendix A.
5. Sample sizes will be large enough to provide precise population estimates and robust
stream classifications, to the extent that natural classes exist in the target popula-
tions.
For certain aspects of the project design criteria, evaluation of how well the criteria were
met is necessarily subjective. For example, the streams of interest to this investigation were
explicitly represented and the probability sample was extracted from these streams, but there is
room for questioning the degree to which the streams that were sampled constitute the most
important resource at risk from acid deposition. The degree to which design criteria 3 (vari-
12
-------
ables measured) and 5 (sample size) were met is based on the precision and utility of results.
The project goals for analytical detection, bias, precision, and data completeness, specified
through DQOs, are discussed in Section 4 and in Cougan et al. (1988), along with the degree to
which these were met and the implications of the results for data interpretation. Aspects of
temporal and spatial variance not quantified by the spatial sampling design or QA program are
evaluated in Section 7.
2.3 DEFINING THE RESOURCE AT RISK
2.3.1 Regions of Interest
Regions of interest for the NSS-I were defined as all areas of the conterminous United
States whose surface water alkalinity was expected to be less than 400 /zeq L'1, and which also
are subject to higher-than-background levels of atmospheric acid deposition. These criteria
should encompass those areas with surface waters most likely to exhibit chemical and biological
effects resulting from acid deposition. Although the NSS-I is potentially a nationwide effort, we
know that many regions of the United States are not likely to contain a significant number of
low ANC or low pH waters. These areas were initially delineated by Omernik and Powers (1983)
for the whole nation, and more recently, at greater resolution, for individual regions (Omernik
and Kinney, 1985; Omernik and Griffith, 1985, 1986a,b). Figure 2-1 shows the areas of the
United States where ANC was expected to be < 400 /ieq L"1. Nonsensitive areas include vast
portions of the Great Plains, as well as low-elevation areas in central and coastal states where
soils or bedrock geology provide ample buffering capacity for neutralizing acid inputs from
either natural or anthropogenic sources. These areas were excluded from consideration to allow
resources to be directed at areas in which low surface water alkalinity is more likely to be
found. In the conterminous United States, streams with low alkalinity are found in the follow-
ing areas:
1. Continental glaciated regions of the Northeast
2. Continental glaciated regions of the Upper Midwest
3. High elevation glaciated (alpine) portions of the mountainous west
4. Appalachians and Ozark Plateau region
5. Mid-Atlantic Coastal Plain, including the New Jersey Pine Barrens
6. Southern Piedmont, Southeast Coastal Plain, and Gulf Coastal Plain
For the NSS-I, two regions, the Mid-Atlantic and Southeast (Figure 2-2), were judged to
contain most of the land area not included in the National Lake Survey in which clearwater (low
in organic acids) streams of low alkalinity were likely to be found in the conterminous United
States. The boundaries of these areas were based on a national map of surface water alkalinity
(Omernik and Powers, 1983), individual regional maps of surface water alkalinity (Omernik and
Kinney, 1985; Omernik and Griffith, 1986a,b), land surface form (Hammond, 1970), and physical
divisions (Fenneman, 1946). The criteria used to draw geographical boundaries of the regions
differed somewhat between the NSS-I and Phase I of the National Lake Survey (NLS), because of
the different focus of the two surveys, the different logistical and statistical design constraints,
and the different factors understood to control surface water quality in each area. The differ-
ences in the region numbering systems between the NLS and the NSS-I are arbitrary, but have
become sufficiently ingrained in the NSWS literature and data bases to make change inadvisable.
13
-------
00
00 '
g
oo
S
14
-------
NSS-I Subregions with 1980-1984 Annual let Sulfate
Ion Deposition in Precipitation (g m~2 yr-i )
2.00
3B
line of Approximately Equal
Solfate Ion Deposition (| m-2
NSS-I Mid-Atlantic Snbregions
NSS-I Pilot Snbregioa (2As)
NSS-I SE Screening Snbregions
Figure 2-2. NSS-I Mid-Atlantic and Southeast subregions, with atmospheric wet sulfate
deposition rates. Wet sulfate deposition isopleths were taken from Wampler and
Olsen (1987).
15
-------
To ensure that representative samples were drawn from each distinct portion of a region,
the regions were divided into subregions (Figure 2-2 and inside front cover). Subregions were
defined by assessing the spatial associations between pre-existing water chemistry data and
terrestrial characteristics important to the understanding of surface water sensitivity to acidi-
fication. These characteristics included land surface form, land use patterns, and potential
natural vegetation. It was intended that by delineating subregions as sampling strata for the
NSS-I, inter-regional sampling variance could be reduced. Characteristics of the subregions
sampled (subsection 2.3.2) are summarized in Table 2-1.
It was desirable to focus physical sampling on areas most likely to have potentially
sensitive streams. Not all areas delineated within the subregions shown in Figure 2-1 are likely
to be sensitive. The areas sampled within each subregion were delineated on the basis of the
400 peq L"1 isopleth on the regional alkalinity maps. At the NSS-I sampling plan workshop,
however, researchers decided that where internal valleys or basins and external boundary
intrusions with alkalinity > 400 /zeq L"1 were expected to occur, it would be preferable to
include the valleys and basins and smooth the boundaries (Figure 2-1). Conditions within the
corresponding subregions could thus be described without relying on unproven assumptions to
interpret the historical data and construct the alkalinity map. Expanding the system boundaries
could reduce the estimated percentage of streams below any particular ANC value, if the maps
are correct, but the procedure should increase the estimate of the total number of stream miles
in the area in the same ANC category. In Florida, the subregion boundaries were drawn on the
basis of the 200 /teq IT1 isopleth on the alkalinity maps. This approach was undertaken in
Florida because the NSS-I field activities in this subregion were viewed as a way of providing
supplemental information on streams in an area already surveyed by the ELS, while at the same
time testing the feasibility of the NSS-I sampling design for application to other low elevation,
low relief areas of the Southeastern and Gulf coastal plains.
2.3.2 Regional Prioritization
Choice of the regions and subregions for study in this first regional implementation of the
NSS-I was somewhat subjective. The overall plan was to provide some level of synoptic infor-
mation (Phase I or Screening) for as broad an area of the eastern United States as possible
before providing increased precision in any one region. For initial assessment purposes, it was
more valuable to know whether acidic waters exist in areas previously not known or expected to
contain them than to supplement the NLS estimates of lake chemistry by conducting a Phase I
stream survey in the same areas.
Four subregions were identified for the Phase I survey in the mid-Atlantic states (Table
2-1, Figure 2-1, and inside front cover):
1. The Poconos and Catskills (ID)
2. The Mid-Atlantic Coastal Plain (3B), including the New Jersey Pine Barrens and the
Chesapeake Bay area
3. The northern portion of the Valley and Ridge Province (2Bn)
4. The Northern Appalachians (2Cn)
16
-------
Table 2-1. Characteristics of the NSS-I Subregions
SUBREGION ID - POCONOS/CATSKILLS
Geographic Extent
- Covers the Pocono Mountains, the Catskill Mountains, the glaciated highlands of
Southern New York and Northeastern Pennsylvania, and the Valley and Ridge
Province of Northeastern Pennsylvania and Southeastern New York.
Terrestrial Characteristics
- Tablelands with considerable relief, open high hills, and low mountains.
- Elevations 500 to 3,000 feet.
- Vegetation cover is northern hardwoods and Appalachian oak forests.
- Land use is generally woodland and forest with some cropland pasture.
Precipitation/Hydrological Characteristics
- Annual precipitation 40-48 inches, with most rainfall occurring during May-August.
- Annual surface water runoff 20-30 inches.
- Approximate annual evapotranspiration 28-32 inches.
- Low stream flow, August-September; high flow, March.
Aquatic Characteristics
- Clearwater streams.
- Low alkalinity streams generally drain watersheds < 50 mi2, usually above 1,500
feet elevation.
- Alkalinity values are generally low in spring.
SUBREGION 2As - SOUTHERN BLUE RIDGE
Geographic Extent
- Includes the Southern Blue Ridge Mountains and the French Broad River Valley of
North Carolina and Tennessee.
Terrestrial Characteristics
- Uniformly dissected low mountains.
- Elevations 1,800 to 5,200 feet.
- Vegetation cover is Appalachian oak forest with pockets of northern hardwoods.
- Land use is forest and woodland, mostly ungrazed, mixed with cropland and
pasture.
Precipitation/Hydrological Characteristics
- Annual precipitation 45-80 inches, with most rainfall occurring during January-
April.
- Annual surface water runoff 15-40 inches.
17
-------
Table 2-1. Characteristics of the NSS-I Subregions (Continued)
- Approximate annual evapotranspiration 32-40 inches.
- Low stream flow, September-October; high flow February-March.
Aquatic Characteristics
- Clearwater streams.
- Low alkalinity streams generally drain watersheds < 200 mi2; lowest alkalinity
streams generally drain watersheds < 25 mi2, usually above 2,500 feet elevation.
- Alkalinity values are generally low in spring.
SUBREGION 2Bn - VALLEY AND RIDGE
Geographic Extent
- Centered over the Valley and Ridge Province of Pennsylvania, Maryland, West
Virginia, and Virginia. Includes western section of the Piedmont in Virginia and
Maryland.
Terrestrial Characteristics
- Open hills, plains with low mountains, and open low mountains.
- Elevations 600 to 3,700 feet.
- Vegetation cover is Appalachian oak and oak-hickory-pine forests.
- Land use is forest and woodland, mostly ungrazed on the ridges, with cropland and
pasture in the valleys.
Precipitation/Hydrological Characteristics
- Annual precipitation 40-50 inches, with most rainfall occurring during July-August.
- Annual surface water runoff 20-30 inches.
- Approximate annual evapotranspiration 32-40 inches.
- Low stream flow, September-October; high flow February-March.
Aquatic Characteristics
- Clearwater streams.
- Lowest alkalinity streams generally drain watersheds < 20 mi2, usually above 2,000
feet.
- Alkalinity values are generally low in spring.
- Subregion characterized by trellised drainage patterns.
SUBREGION 2Cn - NORTHERN APPALACHIANS
Geographic Extent
- Covers the northern Appalachian Plateau of Western New York, Pennsylvania,
Western Maryland, and West Virginia. Includes western sections of the Valley and
Ridge Province in West Virginia and Maryland.
18
-------
Table 2-1. Characteristics of the NSS-I Subregions (Continued)
Terrestrial Characteristics
- Open high hills, high hills, and low mountains.
- Elevations 600 to 3,000 feet.
- Vegetation cover is Appalachian oak and mixed mesophytic forests.
- Land use of low alkalinity streams is generally forest and woodland, mostly
ungrazed, with some cropland and pasture.
Precipitation/Hydrological Characteristics
- Annual precipitation 40-55 inches, with most rainfall occurring during May-July.
- Annual surface water runoff 20-30 inches.
- Approximate annual evapotranspiration 28-36 inches.
- Low stream flow, September-October; high flow, March.
Aquatic characteristics
- Mostly clearwater streams with some colored water streams.
- Low alkalinity streams generally drain watersheds < 100 mi2; lowest alkalinity
streams generally drain watersheds < 10 mi2.
- Alkalinity values are generally low in spring.
SUBREGION 2D - OZARKS/OUACHITAS
Geographic Extent
- Includes the Boston Mountains, the Ouachita Mountains, and the Arkansas River
Valley in Arkansas and Eastern Oklahoma, as well as northern portions of the Gulf
Coastal Plain.
Terrestrial Characteristics
- High hills and open low mountains.
- Elevations 500 to 2,500 feet.
- Vegetation cover is oak-hickory-pine and oak-hickory forests.
- Land use of low alkalinity streams is generally forest and grazed woodland.
Precipitation/Hydrological Characteristics
- Annual precipitation 48-56 inches, with most rainfall occurring during April-June.
- Annual surface water runoff 10-20 inches.
- Approximate annual evapotranspiration 42-50 inches.
- Low stream flow, August; high flow, April.
19
-------
Table 2-1. Characteristics of the NSS-I Subregions (Continued)
Aquatic Characteristics
- Clearwater streams.
- Low alkalinity streams generally drain watersheds < 200 mi2.
- Alkalinity values are generally low in spring.
SUBREGION 2X - SOUTHERN APPALACHIANS
Geographic Extent
- Covers southern portions of the Appalachian Plateau and the Valley and Ridge
Province in Tennessee, Georgia, and Alabama. Central protions of the Blue Ridge
Mountains in North Carolina and Virginia are also included.
Terrestrial/Hydrological/Aquatic Characteristics
- This subregion includes characteristics of subregions 2As, 2Bn, and 2Cn.
SUBREGION 3A - PIEDMONT
Geographic Extent
- Covers the western margin of the Piedmont in Virginia, North Carolina, South
Carolina, and Georgia. Includes northern portions of Gulf Coastal Plain in Alabama
and Mississippi.
Terrestrial Characteristics
- Tablelands, plains with high hills, and open high hills.
- Elevations 300 to 1,500 feet.
- Vegetation cover is predominantly oak-hickory-pine forest.
- Land use is cropland with pasture, woodland, and grazed forest.
Precipitation/Hydrological Characteristics
- Annual precipitation 40-55 inches, with most rainfall occurring during January-
April.
- Annual surface water runoff 15-20 inches.
- Approximate annual evapotranspiration 34-42 inches.
- Low stream flow, October; high flow March.
Aquatic Characteristics
- Clearwater streams.
- Low alkalinity streams generally drain watersheds < 300 mi2; some streams draining
watersheds as large as 1,000 mi2 may be relatively low.
- Alkalinity values are generally low in spring.
20
-------
Table 2-1. Characteristics of the NSS-I Subregions (Continued)
SUBREGION 3B - MID-ATLANTIC COASTAL PLAIN
Geographic Extent
- A geographically diverse subregion that includes the following:
The Piedmont of Southern Pennsylvania, Maryland, and Virginia.
The Pine Barrens of Southern New Jersey.
The Coastal Plain covering Delaware, Southern New Jersey and the eastern
sections of Maryland, Virginia, and Northern North Carolina.
Terrestrial Characteristics
- Flat plains and irregular plains.
- Elevations 0 to 300 feet.
- Vegetation cover is oak-hickory-pine and northeastern oak-pine forests.
- Land use is cropland with pasture, woodland, and grazed forest and pockets of
forest and woodland (mostly ungrazed).
- Soils are very sandy and nutrient poor.
Precipitation/Hydrological Characteristics
- Annual precipitation 40 inches, with most rainfall occurring July-August.
- Annual surface water runoff 15-20 inches.
- Approximate annual evapotranspiration 34-38 inches.
- Low stream flow, September-October; high flow, February-March.
Aquatic Characteristics
- Mostly colored streams with some clearwater streams.
- Low alkalinity streams generally drain watersheds < 500 mi2; lowest alkalinity
streams generally drain watersheds within the Pine Barrens < 50 mi2.
- Alkalinity values are generally low from late winter to early summer.
SUBREGION 3C - FLORIDA
Geographic Extent
- Covers the Florida panhandle and the northcentral portion of the Florida peninsula.
Terrestrial Characteristics
- Flat marine plain with sand hills, swamps, sinks, and lakes.
- Elevations 0 to 100 feet.
- Vegetative cover is mostly southern mixed forest with pockets of sand pine scrub
and wetlands.
- Soils are sandy and nutrient poor; subregion contains areas of karst topography.
21
-------
Table 2-1. Characteristics of the NSS-I Subregions (Continued)
Precipitation/Hydrological Characteristics
- Annual precipitation 56 inches, with most rainfall occurring during July-August
(panhandle); most rainfall occurring during June-September (peninsula).
- Annual surface water runoff 10-30 inches.
- Approximate annual evapotranspiration 46 inches.
- Low stream flow, May (panhandle); high flow, August.
- Low stream flow, November (peninsula); high flow, March (peninsula).
Aquatic Characteristics
- Subregion contains mostly colored and some clearwater streams.
- Low alkalinity streams that are clear generally drain the sand hills and have
watersheds < 50 mi2; colored water, low alkalinity streams generally drain
watersheds < 200 mi2; however, due to the lack of relief and drainage patterns,
topographic watersheds are poorly defined.
- Alkalinity values are generally lowest in winter and spring.
- Streams are generally slow moving and often have imperceptible flow in swamps.
22
-------
Four additional subregions were targeted for the Southeast Screening Survey (Table 2-1,
Figure 2-2, and inside front cover):
1. The Piedmont (3A)
2. The Ozarks and Ouachitas (2D)
3. Florida (3C)
4. The Southern Appalachians (2X), which includes the southern portions of the
Appalachians (2Cs) and the Valley and Ridge Province (2Bs) not sampled as part of
the NSS-I Mid-Atlantic Survey, and a northern portion of the Blue Ridge Mountains
(2An) not sampled as part of the NSS-I Pilot Survey. These areas have been
combined into the heterogeneous subregion labeled 2X.
In many of the areas cited above, there is no synoptic lake chemistry data from the NLS,
and the general state of the surface water resource from the standpoint of the NSWS objectives
was unknown. The regions chosen for the Mid-Atlantic Phase I Survey are particularly notable
for their high acid deposition rates (Figure 2-2) relative to other regions of the country. The
concern over declines in salmonid fisheries in mountainous areas of the eastern United States
(Schofield, 1982; Magnuson et al., 1984), and over the decline of the blue herring, American
chad, striped bass, yellow perch, and other warmwater fisheries in the Chesapeake Bay (Hall et
al., 1985, 1987; Hall, 1987; Klauda and Palmer, 1987; Bowman, 1988 [pers. comm.]), any or all of
which may be related to acid deposition, also contributed to the prioritization decision. Only
the Poconos/Catskills subregion (ID) in the northeastern part of the Mid-Atlantic was sampled
during the ELS-I. A comparison of lake versus stream chemistry in that subregion, together
with data collected during the ELS Phase II Episodes Pilot, may be useful in determining the
criteria for prioritizing future NSS field work. The four Mid-Atlantic subregions are contiguous
and extend their coverage over a large area, facilitating the logistical and sampling designs.
The screening areas in the southeastern United States were included because they exemplify
areas expected to contain low ANC surface waters of widely different geochemical types. The
Southern Appalachians and the Ozark Plateau represent two types of upland watersheds that are
separated by hundreds of miles and that have different acid deposition rates (Figure 2-2). The
Piedmont is characterized by lower elevations, but still predominantly clearwater streams. Based
on the ELS-I results, Florida was known to contain both colored and clearwater acidic lakes,
and the target areas studied should provide a good example of the types and occurrence of
colored or clearwater low ANC waters typical of the broad expanses of the Coastal Plain that
were not surveyed in the 1986 field implementation of the NSS-I. The Florida subregion also
tests the utility of NSS-I logistical and design protocols in lowland stream networks. The
implementation of NSS-I activities in 1985 (Pilot Survey) and 1986 (full-scale survey) did not
include areas of the Northeast, Upper Midwest, and West. Whereas these regions are expected
to contain acidic streams or low ANC streams potentially sensitive to acidic deposition, they also
contain numerous lakes that were sampled as part of the ELS-I in 1984 (Linthurst et al., 1986)
and the WLS in 1985 (Landers et al., 1987). In addition, NSS-I field activities to date have not
included synoptic stream chemistry sampling in parts of the South Atlantic and Gulf coastal
plains expected to contain predominantly low ANC surface waters (Figure 2-1).
23
-------
2.3.3 Identifying the Stream Resource of Interest
A sampling design depends on the identification of a target population, that is, a collection
of entities about which we want to make estimates, and ultimately management decisions. Only
when such a population of interest is explicitly defined can representative samples be drawn
from it in order to make statistical inferences about the properties of that population. In the
case of the NSS-I, the first design criterion indicates that the population studied should best
represent the resource at risk. As a first approximation, this population was assumed to be
located in areas of historically low ANC surface waters that receive acid deposition.
Furthermore, the primary resource of interest within the regions studied is presumed to be
water quality for sport fisheries, and the size of streams of interest should reflect the portion
of the stream continuum that provides the majority of fish habitat for critical life stages. This
first regional implementation of the NSS-I was limited largely to areas not sampled by the NLS.
Consequently, NSS-I field activities in 1985 and 1986 were undertaken in potentially sensitive
regions in which streams, as opposed to lakes, are the predominant surface water resource. Full
implementation of the NSS-I in lowland coastal regions with lower rates of acidic deposition in
the Southeast has not yet been accomplished.
Identification of the target population of streams (design criteria 1) required consideration
of the characteristics of large versus small streams with respect to the aquatic resource potenti-
ally at risk from acid deposition. Provided that differences among fish species are ignored,
larger streams provide considerably more fish habitat per unit stream length than do very small
streams, and thus are relatively more important from a fishery resource standpoint. However, in
most regions of the United States, very large streams or rivers generally do not experience low
pH conditions, because natural and anthropogenic buffering sources (e.g., agricultural liming or
discharge of treated wastewater) tend to buffer any atmospherically derived acidity, once a river
has descended into populated valley bottoms.
At the other end of the size spectrum, low-order, high-elevation streams within a given
basin are expected to exhibit lower pH and ANC than their downstream counterparts and are
therefore more likely to serve as early warning indicators of acid deposition impacts. These
smaller streams, however, offer quantitatively less fish habitat, and for that reason may not best
represent the biological resource at risk. While the impacts of acidification on spawning and
detritus processing in very small headwater and intermittent streams should not be discounted, it
would be very difficult, given the present state of the science, to relate headwater acidification
to fish productivity further downstream. Given this uncertainty, together with the frequently
observed pattern of maximum productivity and species diversity of fish and invertebrates in mid-
order reaches (Platts, 1979; Vannote et al., 1980; Minshall et al., 1983), the small to medium-size
stream category appeared to be the best target for Phase I sampling from a biological resource
standpoint.
Rivers and streams at opposite ends of the size spectrum also present special logistic and
sampling design difficulties. Larger rivers require substantially different physical sampling
(measurement) techniques and equipment than those used on smaller streams. The geographical
point sampling frame that was used for the statistical sampling design (Section 2.4) also works
less effectively on reach populations containing watersheds of drastically different sizes
(Overton, 1987). On the other hand, populations of very small streams are poorly represented
on maps and often very difficult to access physically, and their flow may dry up entirely in
some years or seasons.
These general criteria define a conceptual population of interest that is not easily defined
in quantitative, explicit terms that facilitate rigorous statistical sampling. Ideally, there may be
24
-------
a single size range of streams that satisfies these abstract criteria. In practice, however, even
if habitat characteristics in streams of interest could be quantitatively defined, it would be
difficult to arrive at an explicit definition of this population so that it might be targeted for
sampling. There are two aspects to the problem of explicitly and quantitatively defining the
conceptual population of interest. The first deals with actual stream locations and character-
istics, and the second with the correspondence between these "on-the-ground" characteristics
and abstract representations of them. A simple solution to the first aspect of the problem is
confounded by the regional, temporal, species, and life stage differences in fish habitat require-
ments (not to mention different ways of defining stream size), which make a precise definition
of the size of streams in such a population impossible. Once we satisfactorily define the size of
streams of interest, we must then contend with the bias and imprecision associated with our
abstract representation of the assemblage of streams on maps, lists, and remote imagery.
Given all these considerations, the most expedient approach for the NSS-I was to explicitly
define a target population in terms of blue-line representation of streams on l:250,000-scale
topographic maps, modified by certain site inclusion criteria. The resulting target population
was our best attempt to define this conceptual population of interest. Its precise definition
(subsection 2.4.1) was influenced by the expertise of local fishery biologists in a number of
regions and was tempered with our understanding of watershed response to acid deposition. The
decision was also constrained by logistical considerations that influenced the number of sites
that could be sampled. The physical characteristics of target streams reaches are examined in
Section 5 along with an evaluation of whether that target population is a reasonable representa-
tion of the population of interest.
2.3.4 Alternative Methods for Identifying the Target Population
During the planning stages, several different approaches were evaluated before the final
decision was made to use a sampling frame based on l:250,000-scale map representations of
streams. Subsection 2.3.3 identifies the conceptual population of interest (the resource at risk)
as all reaches that are not grossly polluted, that drain watersheds of intermediate size, and that
occur within certain relatively homogeneous physiographic areas expected to contain surface
water with ANC predominantly < 400 /zeq L"1. Several alternative approaches were evaluated for
identifying or listing sample sites from which a probability sample of streams could be selected
for field visitation. These included computerized data files originally constructed for other
purposes, remote imagery collected by satellite or aircraft, and blue-line representations of
streams on several scales of USGS topographic maps.
The use of existing computerized lists of streams was rejected, because they tend to
describe large streams and rivers. For example, the U.S. EPA REACH data file (Olsen et al.,
1981) comprises only those reaches large enough to appear on l:500,000-scale topographic maps.
At this scale, the stream resources of interest (subsection 2.3.3) are not adequately represented.
A large number of these smaller streams that could be potentially sensitive to acidic deposition
but that are still large enough to offer abundant fish habitat are not included (SFI, 1984).
Generally, streams that appear on medium- (1:250,000) and large-scale (1:24,000) maps are too
small to be of interest to water supply managers and therefore have not historically been
represented in computerized water resources data bases. Remote imagery was rejected as being
too costly and time-consuming for constructing a frame of thousands of reaches. As a result,
the most expedient approach for the NSS-I was to define a target population in terms of blue-
line representations of streams on topographic maps.
25
-------
Of the two applicable map scales (1:250,000 and 1:24,000), the former was chosen because it
was judged to best represent the conceptual population of interest for the NSS-I. An earlier
survey (TIE, 1981) indicated that historical fishery and aquatic resources values are more closely
associated with blue-line streams on medium-scale (1:250,000) maps. Although the relationship
varies from map to map, l:250,000-scale maps generally exclude the smaller streams (streams
with drainage areas < 0.5 km2 at their first confluence) that appear on the corresponding
l:24,000-scale maps for the eastern United States.
Although streams and stream reaches to be sampled in the survey are identified on
l:250,000-scale maps, measurements of physical characteristics, including length of stream reach,
total watershed area, and direct watershed area, as well as latitude and longitude for sample
sites were calculated from l:24,000-scale topographic maps. Field sampling crews were provided
with l:24,000-scale maps to aid them in locating the sampling sites.
2.4 STATISTICAL SAMPLING DESIGN
The statistical sampling design for the NSS-I was developed by Overton (1986, 1987;
Overton and Stehman, 1987). In statistical terminology, the stream survey is a double sample,
with a third sampling level on (within) a sample reach. The term stage refers to the two levels
of the double sample (Stage I and Stage II) and also to the sampling within a reach (Stage III).
The major stratification structure is provided by subregions. Secondary stratification is provided
by expected ANC class (greater than 50 /zeq L'1 and less than 50 /feq L"1) as represented on
maps of expected alkalinity prepared by Omernik and Kinney (1985) and Omernik and Powers
(1983).
2.4.1 The Stage I Sample
The target population for the NSS-I has been defined as that set of blue line segments
appearing on l:250,000-scale USGS topographic maps, within the designated regions of interest,
and conforming to a specified set of criteria designed to eliminate those reaches not in the
designated resource of interest. The maps on which the population is defined are archived as a
part of the permanent record. It is important to recognize that it would be possible to
enumerate this set of reaches for any region to produce a list containing each reach in the
population of that region. Such a list would provide a list frame that could be the basis of a
sampling design. The explicit nature of the population is established by the possibility of
generating this list frame; the population that is being described is precisely this set of reaches.
A reach is a blue-line segment on the identifying maps. A blue-line segment is identified
at each end either by the end of a blue line or by the confluence of two blue lines. The upper
or lower node of a reach can also be defined by artificial or natural structures that interfere
with stream patterns, like reservoirs or lakes. A target reach must have at least half its length
within the subregion and must not be excluded by one or more of the site rules (discussed in
detail in subsection 2.4.2). A reach is excluded from the target population if it has an attribute
identifying it as noninterest; for example, a reach may drain a total watershed area greater than
155 km2 (60 mi2) or it may drain a watershed that is predominantly urban.
The site rule criteria are applied at the time sample reaches are identified from maps.
Other criteria identifying nontarget reaches are not available from maps, and are applied to
sample reaches at the time of field visitation. Any condition that excludes a reach from
consideration in the resource at risk, and that will be apparent from visiting the reach, can be
applied at that time. For example, reaches may be considered nontarget because of gross pollu-
26
-------
tion, as by mine drainage, oil field drainage, or tidal influence. This process reduces the field
sample below the number selected for the Stage II sample, but does not affect the validity of
that sample. The subpopulation of the modified set of target reaches is then characterized from
the subsample.
The conceptual frame contains all reaches considered to be target reaches by virtue of
inspection of the map materials, and this entire set is the frame population. The stream survey
generated a sample of this population without generating the list, by the simple device of an
area/point frame and a rule of association between a sample on this frame and the population of
reaches. Specifically, a square dot-grid transparent overlay was constructed, with a scale that
projected 64 mi2 of the l:250,000-scale maps per point on the overlay. Thus, the distance
between the dots on the overlay projects to 8 mi on the maps. To select a sample, the overlay
is randomly placed on the map, with the result that the points are randomly, but not indepen-
dently, placed on the study area.
The Stage I sample of reaches is then obtained by applying a rule of association between a
point identified on the map and the network of blue-line streams. From a point, a line is
traced perpendicular to map contours, down the fall line, until either a blue line is encountered,
or some other structure that is not a stream is encountered. If a blue line is encountered, then
the stream segment from the nearest upstream confluence, or beginning of the blue line, to the
nearest downstream confluence, is the selected reach. This reach is examined to determine if it
qualifies as a target reach, according to the site rules. All of the target reaches, so selected
by the dot-grid overlay, constitute the Stage I sample of target reaches. Each sample point
within the study area leads to: (1) no reach, (2) a nontarget reach, or (3) a target reach. The
effective sample size associated with this selected sample is given by the number of grid-dots
that fall within the boundary of the study area. Figure 2-3 illustrates the selection of reaches
by the point frame. Point 98 in the figure selects a nonheadwater reach, whereas point 99 sel-
ects a headwater reach. The topographic direct drainage area of sample reach downstream nodes
is designated as av The size of ax is an important variable measured on maps in the NSS-I,
because the probability of selecting any given reach is proportional to this area. Area a2 in
Figure 2-3 represents the area draining into the upstream node of the nonheadwater reach.
The sample selected by the grid-point frame is a probability sample of the population.
Every member of the population has the potential of being selected. The probability (TTJ) that
any member is selected (included in the sample) is given by a known function of a measurable
property of that member. Specifically, JTJ = a^/64, where ax is defined in the next paragraph,
and 64 is the area (square miles) per grid dot. Reach order is determined from the 1:250,000-
scale maps, being the number of headwater reaches in the full watershed of the target reach.
The next step requires transfer of the selected reach to l:24,000-scale maps for identifica-
tion of the direct watershed and measurement of the physical attributes. Reach length, /, in
miles, and area of direct watershed, al5 in sq mi, were obtained by planimeter (on l:24,000-scale
maps) on the full Stage I sample. In particular, it is important that aj be determined very
precisely, because the weights used in estimation will be inversely related to the aj's, and errors
in measurement of ax cause biases in the estimates. Additionally, the area draining into the
upstream node of the selected target reach is determined from planimetering. The direct drain-
age area contributing to the upstream node of headwater reaches is included in area a^ water-
shed areas for reaches of order 2 or greater, determined from the l:250,000-scale maps, are
designated as a2 and measured on the finer scale maps (1:24,000).
27
-------
Non-Headwater Reach
Headwater Reach
Figure 2-3. Representation of the point frame sampling procedure for selecting NSS-I Stage
I reaches. Area aj represents the direct drainage area to the lower node of
nonheadwater reaches, or the total drainage area to the lower node of headwater
reaches. Area a2 is the total drainage area to the upper node of nonheadwater
reaches.
28
-------
The frame population is thus explicitly defined, but not explicitly identified. It is
represented by a Stage I probability sample. It is reasonable to consider the frame population
as the initial target population, but it must be remembered that additional sample reaches are
eliminated from target status at the field sampling stage. Thus the target population is a subset
of the frame population. Selection of the Stage II sample visited in the field is treated in a
later section.
2.4.2 Site Inclusion Criteria (Site Rules)
The reaches identified by the grid points are categorized into various target or nontarget
categories according to the criteria discussed in the following paragraphs. The target population
thus defined is identical to that which might have been defined by a list frame. Unlike the
exhaustive population defined by a list frame, however, the point frame (with inclusion criteria)
defines a probability sample of that population. The inclusion criteria that were used for
drawing the first stage sample of NSS-I sites are shown in Table 2-2. Specific decision proto-
cols provided by the site inclusion criteria were used by the project geographers to identify the
resource at risk, as addressed in general by Project Design Criterion 1 (Section 2.2).
Each grid dot may lead to a nontarget reach, a target reach, or no reach at all. Table 2-3
is a breakdown of the disposition of all Stage I grid points in the NSS-I subregions; the table
also includes the Stage II sample numbers discussed in subection 2.4.5. A grid dot identifies no
reach if the topographic fall line identifies a reach wholly outside the study area boundary, or
if the dot identifies something other than a reach (e.g., a lake, reservoir, swamp, or closed
basin). Nontarget reaches are excluded because some characteristic puts them in a noninterest
category. Reaches with highly urbanized drainages, for example, were excluded by the NSS-I.
Reaches contacting the NSS-I subregion boundaries may penetrate far enough into suspected
high ANC regions outside the study area to be unlikely to have low alkalinity over much of
their length. Boundary reaches were excluded if more than 50% of their length was outside the
study area.
2.4.3 The Stage I Data
The first stage data base includes a listing for each grid point, including:
1. Site identification code (e.g., 3B041016L). An eight digit code containing three fields
that indicate the NSS-I subregion code (3B), the l:250,000-scale map ID (041), special
interest site designation (0 = Routine site, 9 = Special Interest Site), the grid-dot
identification (16), and the node identification (L = Lower, U = Upper).
2. Stream name. This is recorded from l:250,000-scale or l:24,000-scale maps.
3. Site inclusion criteria applicable to the grid point (Section 2.4.2). For target reaches,
certain additional information is obtained as described in 4 to 7 below.
4. County and state in which the reach is located.
5. State in which the associated watershed is located.
6. Administrative jurisdiction, if sites lie within national or state parks or on military
reservations.
7. Miscellaneous comments.
29
-------
Table 2-2. NSS-I Site Inclusion Criteria
I. NON-TARGET GRID DOTS. Grid dots that do not identify a stream reach.
NO1. Reach Out. The topographic fall line leading from the grid dot leads out of
the study area before intersecting a reach.
NO2. Reservoir. The grid dot falls in a reservoir, or the topographic fall line
leads to a reservoir shoreline.
NO3. Lake. The grid dot falls in a lake, or the topographic fall line leads to a
lake shoreline.
NO4. Closed Basin. The topographic fall line leading from a grid dot leads into a
closed basin that does not contain a reach.
NO5. Swamp. The grid dot falls in an area with insufficient topographic relief to
identify a reach.
II. NON-INTEREST REACHES. Grid dots identify a reach that is not in the target
population.
Nl. Boundary Reach. The reach identified by a grid dot crosses an external
study area boundary, such that > 50% of its length lies outside the study
area.
N13. River Reach. The total drainage area above the downstream node of a
nonheadwater reach is > 60 mi2.
N14. Metro Reach. The grid dot falls in a metro area, or the dot identifies a
reach with > 20% of its total drainage area occurring within a metro area, as
indicated on either a 1:250,000- or l:24,000-scale topographic map.
III. TARGET REACHES. Grid dots identify a target reach.
N2. Target Reach (general). The indicated reach has a total drainage area < 60
mi2 and > 50% of its length within the study area boundary.
N21. Wetland Reach. The indicated reach has a topographically definable
watershed, but contains cross channels that may interconnect it with
adjacent watersheds under some hydrologic regimes.
N23. Large Headwater Reach. The indicated reach has a direct drainage area > 60
mi2 but is a headwater reach.
N28. Tailwater Reach. The indicated reach drains a lake or reservoir, but the
total drainage area is < 60 mi2.
30
-------
Table 2-3. NSS-I Stage I Sample Grid Point Disposition
SUBREGION*
ID 2BN 2CN 2D 2X 3A 3B 3C 2As
Total Sites (Number of Grid Points)
302
501
305
360
333
607
564
165
168
Sites Excluded:
NO1 Reach Out of Study Area
NO2 Reservoir
NO3 Lake
NO4 Closed Basin
NO5 Swamp
Nl Boundary Reach
N13 River Reach
N14 Metro Reach
Subtotal (Exclusions):
4
5
20
-
-
5
69
6
109
—
3
7
4
-
-
98
6
118
2
9
2
1
-
1
65
-
80
2
4
15
-
-
8
65
-
94
2
9
10
-
1
5
62
1
91
4
19
18
1
-
11
128
3
184
77
4
10
-
17
11
73
24
216
3
2
50
-
1
3
22
-
81
2
8
3
-
-
2
17
-
32
Target Reaches:
N2 Target Reach (general)
N21 Wetland Reach
N23 Large Headwater Reach
N28 Tailwater Reach
Subtotal (Target Reaches)
Stage II Sample Size
Effective Sample Size (Stage II)#
188
-
1
4
193
61
91
380
-
1
2
383
53
70
219
-
-
6
225
74
100
245
8
-
13
266
50
68
240
-
-
3
243
50
66
413
-
-
10
423
50
72
316
32
-
-
348
62
100
81
3
-
-
84
50
94
134
-
-
2
136
54
84
* ID = Poconos/Catskills
2Bn = Valley and Ridge
2Cn = Northern Appalachians
2D = Ozarks/Ouachitas
2X = Southern Appalachians
3A = Piedmont
3B = Mid-Atlantic Coastal Plain
3C = Florida
2As = Southern Blue Ridge
Effective sample size (ESS) is the number of grid points examined to obtain the Stage II
sample reaches to be visited in the field. ESS is stratum specific. Exact values for each
ANC stratum and its application in variance calculations are discussed and detailed in Sale
(1988).
31
-------
The information listed in the seven items on page 29 was critical in the reconnaissance
procedures described in Section 3. In addition, data were recorded for certain quantitative
geographic variables, including:
8. The area of direct drainage, aA. This is the portion of the watershed that drains
directly into the chosen reach (Figure 2-3), and also the area within which a grid
point will select the same reach. This variable is very important, as it is a measure
of the probability of selecting the reach, and is used in making all population
estimates. It was measured as accurately as possible on the l:24,000-scale maps.
9. Reach order, R. The number of reach regions (headwaters) in the watershed above
and including the selected reach, as identified on the l:250,000-scale maps. This
reach ordering system is basically that of Shreve (1966), and has certain topological
advantages over the more familiar Strahler or Horton ordering systems.
10. For reaches of order higher than 1, the area, a2, of the upstream watershed. The
variable a2 is the area of the entire watershed that produces the streamflow that
enters the selected reach at the upstream node, determined from surface topography
on l:24,000-scale maps. (This value is zero for first order reaches).
11. Reach length, /, is the length of the selected reach. Locations of the reach ends
were determined on l:250,000-scale maps, but measurement of / was made on 1:24,000-
scale maps.
12. Headwater drainage area, a3, being the area draining into upper node of each head-
water stream (identical to a2 for reaches with R > 1).
2.4.4 Stage I Estimates
The sampling design described in the preceding subsections produced a sample of target
reaches from each of the nine subregions, with sample sizes ranging from 84 to 423 (Table 2-3).
These samples were subsampled (subsection 2.4.5) to yield the Stage II samples that provided the
water chemistry data. The full set of Stage I samples were used to estimate the physical attri-
butes of the initial target populations in the nine subregions. However, because field derived
information was subsequently used to eliminate some sites, designating them as not representa-
tive of the refined target population, the physical characteristics reported in Section 5 are
calculated from the Stage II sample.
Details of estimation, and the statistical foundation of the methods, are provided elsewhere
(Overton, 1986, 1987; Blick et al., 1987; Overton and Stehman, 1987; Stehman and Overton,
1987a, 1987b). The general form of estimator is provided by the Horvitz-Thompson Theorem:
tv = 2y/jr = Swy
32
-------
where:
Ty is the estimate of the total of any attribute, y, over the population.
y is any reach attribute of interest, known over the sample, S.
S indicates summation over the appropriate sample of target reaches, whether the
S full sample, a subsample, or mixed subregion sample.
aj is the area of direct watershed of a reach.
TT = aj/64 is the inclusion probability of a reach, where 64 is the map area per grid-dot.
w = I/TT is the weight assigned to a reach in making population estimates.
By assigning different definitions to y, and by summing over different sets of sample units, S,
the various population attributes can be estimated from this one equation. Specifically, in the
NSS-I, several attributes are identified as parameters of the resource at risk:
1. Total number of target reaches (ft), y = 1: fr = Sw.
2. Total Area of Direct Watershed (A), y = a^ A = Swaj
3. Total Length of Target Reach (t), ^ y = I: t = Ew/
4. Total Discharge Index (b), y = (ax + a2) = d: b = Swd
5. Total Stream Surface Area (§), y = s: § = Sws
Other attributes are best estimated by ratios of these estimates, as for example Stream Density,
Stream Density = L/A .
Additionally, a number of subpopulations are apparent. Several examples will serve to
identify the nature of subpopulation definition:
1. The subpopulation is defined as the set of all headwater reaches; the sample is the
set of headwater reaches in the Phase I sample.
2. The subpopulation is defined as the set of reaches having total watershed area (a1 +
a2) < 5.0; the sample is the set of reaches in the sample having this property.
Similarly, combinations of subregion populations are assessed by combining the samples from
those subregions, and in general it is only necessary for the sample subset to be defined in
exactly the same way as the estimated population. The possibilities are virtually unlimited.
Stage I variance estimates, leading to estimated standard errors (SE) of the estimated
attributes, are obtained by the Horvitz-Thompson variance estimator, with second order inclusion
probability specified for this design (Overton, 1986; Stehman and Overton, 1987a). Several of
the studies cited earlier addressed the basis for and validity of this variance estimator in the
context of the stream survey. The formula for estimated variance, incorporating the features
explicit for the established treatment, is:
33
-------
V (ty) = E y2w (w - 1) + S
S ieS jeS
where n is the effective sample size, and v^ = [(wj + Wj)/2 - WjWj]/(n - 1), if i and j are from
the same stratum; if i and j are from different strata, then v^ = 0.
This same formula will be used for Stage II estimates. The only change is that the effective
sample size will be defined differently.
2.4.5 The Stage II Sample
The Stage I sample generated a sample point for each 64 square miles of each subregion.
A substantial fraction of these sample points resulted in sample target reaches. There were too
many to visit in the field for assessment of water chemistry, so the sample size had to be
reduced for the field sampling stage of the survey. A subsample size of 50 to 74 target reaches
was selected from each subregion in a manner that provided good spatial distribution (Overton,
1986, 1987). The subsample was a probability sample of the Stage I sample, resulting in a
double sample allowing general estimation methodology to be used. Figure 2-4 maps the loca-
tions of Stage II sample sites in the Mid-Atlantic subregions; those in the Southeast and the
Pilot survey subregions are shown in Figure 2-5. Spatial sampling density was considerably
greater in the Pilot Survey subregion, because that survey was used to provide information
concerning the desirable density for the full-scale Phase I Survey (Messer et al., 1986, 1988).
Within several of the subregions, there are areas where ANC was expected to be very low
(< 50 peq L'1) on the basis of historical alkalinity data. It was desirable for all Stage I sample
reaches in these areas to be field sampled, so they were all included and identified in a separate
stratum. Then subsampling occurred only in the stratum of higher ANC. The low ANC stratum
sites are separately identified in Figure 2-4, as are special interest sites that were not part of
the probability sampling frame (Section 2.6).
In the stratum of higher ANC, the Stage II sampling design provided for variable probabil-
ity selection that neutralized, so far as possible, the variable probability selection that occurred
at Stage I. Specifically, the Stage I inclusion probabilities were proportional to direct watershed
size (aj). Then the Stage II sample was selected from the Stage I sample with inclusion proba-
bilities inversely proportional to a^ A slight complication arose, since some of the Stage I
sample reaches had such small a^s that they came into the Stage II sample with certainty (i.e.,
with conditional inclusion probability =1). The resultant Stage II sample for field visitation in
a particular subregion had the following structure:
1. Stratum 1 (most reaches) inclusion probability = n /N
2. Stratum 1 (a few reaches) inclusion probability = aj/64
3. Stratum 2 inclusion probability = aj/64,
where n* is the number of Stage II sample reaches in that subset of Stratum 1 and N is the
Stage I estimate of number of reaches in that subset of Stratum 1. Only the first of these
three categories represents a subsample of the Stage I sample; the last two contain all the Stage
I sample reaches in these categories.
34
-------
NSS-I SECOND STAGE SAMPLE SITES
(MID-ATLANTIC SUBREGIONS)
+ Stage II Site
* Low ANC Site
* Special Interest Site
ID - POCONOS/CATSKILLS
2Bn - VALLEY AND RIDGE
2Cn - NORTHERN APPALACHIANS
3B - MID-ATLANTIC COASTAL PLAIN
Figure 2-4. NSS-I Stage II sample reaches and special interest sites in the Mid-Atlantic
subregions.
35
-------
NSS-I SECOND STAGE SAMPLE SITES
(SOUTHEASTERN SUBREGIONS)
- Stage
2 Slie
2As - NSS-I SOUTHERN BLUE RIDGE (PILOT)
20 - OZARKS/OUACHITAS
2X - SOUTHERN APPALACHIAN PLATEAU
3A - PIEDMONT
3C - FLORIDA
Figure 2-5. NSS-I Stage II sample reaches and special interest sites in the Southeast
subregions.
36
-------
The weights used in estimation and analysis are inverses of the inclusion probabilities, and
uniform within subregions in the first of the three categories. The second part of Stratum 1
and Stratum 2 still have variable weights, but these are minor parts of the samples, and create
minor complexity in estimation and analysis.
In order to obtain good spatial distribution of the Stage II sample illustrated in Figures 2-4
and 2-5, a modification of the standard variable probability systematic sample was used. Specif-
ically, spatially compact clusters of Stage I points were identified on the maps, and a list was
constructed with these clusters blocked in the list. This ensured that some sample elements
were taken from each cluster. An additional element to the design was randomization of the
points within the clusters, so that zero pairwise inclusion probabilities were eliminated from the
second stage selection.
2.4.6 Stage II Estimates
Estimation of population attributes and standard errors from the Stage II sample followed
the same pattern as that applied to the Stage I sample. The physical attributes of the initial
target population were best estimated from the full Stage I sample. However, the identity of
the target population was refined further based on information obtained from field visitation, so
the attributes of this most refined target population had to be estimated from the Stage II
sample. To estimate the physical attributes of the population, the values of the chemical
variables were ignored, unless those values were invoked in defining the target population.
These estimates were used as denominators of the distributions described in the next paragraph.
The primary outputs of the Stage II analyses are descriptions of the various distributions
of the chemical variables. Distributions of chemistry within any subset of the target population
were analyzed in the same way. That is, an estimate was made of the number of reaches in the
subpopulation having a value of the chemical equal to or less than a particular value:
= Sw.
X(x) = I! w/, estimating total length of target reaches having X < x,
X(x) = S w(a! + a2), estimating total discharge index of target reaches having X < x.
X
-------
F(x) =
Fu(x) =
and SE(fr(x))]
where ((x)) is calculated in exactly the manner of
restricted by X < x.
, with summation further
Such distributions can be generated for any quantity generated by the general estimators.
For example, the distribution, F(x), described above and illustrated in Figure 2-6 is the fre-
quency distribution, interpreted as the proportion of numbers of target reaches in the particular
subpopulation having the attribute, X < x. To read this figure, pick a value, x, of the attribute
X, along the horizontal axis (ANC in this example) and read the y-axis value of the two curves,
F(x) and Fu(x) at this value. Then F(x) is the estimated proportion of reaches in the population
with a value of the attribute equal to or less than X. Fu(x) is the upper confidence bound on
this proportion, to be read as: one is 95% confident that the true proportion is less than this
bound. Some distributions plot the declining function, 1 - F(x). For these, read the distribu-
tion as the estimated proportion of reaches having a value of the attribute equal to or greater
than X. Confidence bounds are as before: an upper confidence bound on the true proportion.
A lower one-sided confidence bound on F(x) can be generated, if needed, by measuring the
distance between the two curves, and projecting the identical distance below F(x). This curve is
not presented because it could be confused with two-sided confidence bounds, which are of dif-
ferent width. The bound provided is the one usually considered appropriate for expressing the
status of the resource.
By generating distribution functions for the other characteristics, such as direct watershed
area, length, or total drainage area, other types of distributions can be obtained. G(x) is the
proportion of direct watershed area associated with reaches having attribute, X < x. H(x) is the
proportion of reach miles (or km) having reach attribute, X < x. D(x) is the proportion of
reach total drainage area (a crude index of discharge) having reach attribute, X < x. In all
cases, these distributions and the represented confidence bounds are dimensionless, and a shift
from miles to kilometers or square miles to the metric equivalent has no effect on F(x) or
Fu(x). Any other distribution, such as stream surface area, or reach density, can be analyzed in
exactly the same way, requiring only an appropriate reach attribute that sums to the population
attribute.
Other population statistics of interest can be generated from the distributions. Each
distribution has identified quantiles, the median and the four quintiles, Qj, Q2—.Q4- The median
of F is the value of x such that F(x) = 1/2. Qj of F is the value of x such that F(x) = i/5.
These statistics are defined for all distributions.
Additionally, from the frequency distribution, F(x), the mean and standard deviation of the
variable x on the population is estimated.
Mean(x) = Swx/Sw,
SD(x) « V Swx2/Sw - [Swx/Sw]2.
38
-------
CO
C/5
.s
-d
^o
"H.
|
o
I
a
o
a
•«_»
CO
•3
>
I
o
CH
O
(U
5
O
"a
C3
W
vo
i
£
g,
E
3AiivnniAino
39
-------
2.5 INDEX SAMPLING
The collection of water samples and associated observations on reaches in the Stage II sub-
sample constitutes Stage III of the statistical design of the NSS-I. Sampling at this third stage
was not conducted as a probability sample, but rather in a way that would best characterize or
index the properties of a sample reach. 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 integra-
tive 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 poten-
tially 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 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 (Olem, 1986), and the
Ouachita 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 etal., 1975;MunizandLeivestadt, 1980;KelsoandGunn, 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,
40
-------
relative to other fish life stages (Schofield, 1976; Raines, 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 leaf out (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 stormf low 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 nonlittoral lakewater vol-
ume, a sample taken at a single point on a stream reach would not adequately describe chemis-
try 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.
2.6 SPECIAL INTEREST SITES
In addition to the reaches selected in the probability sample, the NSS-I sampled 36 special
interest reaches. These reaches were selected as sites similar in drainage area to those in the
NSS-I target population, but where intensive process-oriented studies or long-term monitoring
projects are underway. The special interest sites included sites selected for long-term study in
the Long-Term Monitoring (LTM) Project, part of NAPAP Task Group E, at which storm events
were also monitored. The special interest reaches (listed in Table 2-4) were sampled in the
same manner as those included in the probability sample. However, data from these sites cannot
be analyzed as part of that probability sample. Classification of special interest sites (Section 9
and future work) will become increasingly useful for placing the intensive study results from
those sites into a regional perspective.
41
-------
Table 2-4. Special Interest Sites
Reach ID
1D030910
1D030911
1D030912
1D030924
1D030925
1D030926
2A07891
2A07892
2A07893
2A07894
2A07895
2A07896
2A08891
2B036903
2B036904
2B036905
2B042922
2B047916
2B047917
2B047918
2B047919
2B047920
2B047929
2C035913
2C035915
2C041901
2C041902
2C041914
2C077923
2D083927
2D083928
3B042907
3B042908
3B042921
3B043909
3B048906
Stream Name
Biscuit Brook
High Falls Creek
E. Branch Neversink
Van Campens Brook
Green Brook
Tillman Brook
Cosby Creek
Twenty-mile Creek
Jarrett Creek
Shope Fork/Gauge 8
Moses Creek
Pinnacle Branch
Chester Creek
Detweiler Run
Leading Ridge #1
Laurel Run
Hunting Creek
White Oak Run
Deep Run
N. Fork Dry Run
Mill Run
Old Rag Run
Mill Run Trib
N. Fork Bens Creek
N. Br. Quemahoning
Fernow Control
Elk Lick Run
Cole Run
Camp Branch
N. Fork Saline
E. Fork Saline
Magothy River
Bacon Ridge
Pequea Creek
McDonalds Branch
Lyons Creek
State
NY
NY
NY
NJ
NJ
NJ
TN-NC
NC
NC
NC
NC
NC
GA
PA
PA
PA
MD
VA
VA
VA
VA
VA
VA
PA
PA
WV
WV
PA
TN
AR
AR
MD
MD
PA
NJ
MD
Cooperator
Pete Murdoch
Pete Murdoch
Pete Murdoch
Bill Stansley
Bill Stansley
Bill Stansley
Harvey Olem
Harvey Olem
Harvey Olem
Jack Waide
Harvey Olem
Jack Waide
Harvey Olem
Jim Lynch
Jim Lynch
Jim Lynch
Owen Bricker
Jim Galloway
Jim Galloway
Jim Galloway
Owen Bricker
Owen Bricker
Owen Bricker
Jim Barker
Jim Barker
David Helvey
David Helvey
Jim Barker
Mike Kelly
Joe Nix
Joe Nix
Tony Janicki
Tony Janicki
Owen Bricker
Debbie Lord
Ron Klauda
Elevation
Org. (Ft)
USGS
USGS
USGS
Bur. of
Freshwater
Fisheries
TVA
TVA
TVA
USFS
TVA
USFS
TVA
Penn State
Penn State
Penn State
USGS
U. of Va.
U. of Va.
U. of Va.
USGS
USGS
USGS
USGS
USGS
USFS
USFS
USGS
TVA
Ouachita
Baptist U.
MDDNR
MDDNR
USGS
USGS
MDDNR
2090
1910
2140
660
960
710
2457
1426
3052
2259
2298
3091
2515
995
870
990
1045
1480
1360
1600
1175
1440
1180
1539
1940
2440
1680
1915
1705
1210
1080
10
10
600
115
17
Watershed
Area
(mi2)
3.59
2.75
9.04
4.66
0.65
1.47
11.5
8.1
5.5
3.6
9.1
0.4
2.8
3.29
0.45
10.74
3.87
1.93
1.32
0.94
1.15
1.09
1.18
3.11
2.63
0.14
5.84
1.32
0.37
0.84
2.58
5.40
6.97
0.22
2.10
14.54
42
-------
SECTION 3
METHODS
3.1 OVERVIEW
This section discusses the methods employed in the National Stream Survey - Phase I
(NSS-I) for sample collection, handling, and analysis. It also describes the quality assurance and
quality control (QA/QC) protocols that were implemented, including the procedures for data
management (verification, validation, and construction of the final data bases used for analysis).
Much of the NSS-I methodology was shared with the Eastern and Western Lake Survey
components of the National Lake Survey (NLS). The QA/QC sample collection design and the
data management protocol for the NSS-I were similar to the procedures implemented in the NLS.
The protocols used by the sample processing laboratories during the NLS also applied to the
NSS-I. The NLS procedures for sample collection and in situ measurements, however, were not
applicable to the NSS-I. All NSS-I field techniques were developed or refined during the NSS-I
Pilot Survey, conducted in the Southern Blue Ridge in 1985 (Messer et al., 1986; Knapp et al.,
1987).
An Episodes Pilot Survey was conducted concurrently with NSS-I sampling in order to
determine the feasibility of sampling acidic episodes in streams in the Mid-Atlantic sampling
area. This study evaluated the use of a probability-based sampling design to assess the extent,
duration, and frequency of acidic episodes on a regional scale. The Episodes Pilot Survey also
attempted to test proposed physical and chemical sampling protocols. The design, methods, and
results of the Episodes Pilot Survey are summarized by Messer et al., 1986.
The NSS-I data acquisition process began with the collection of water samples and
descriptive site information about the stream sample points. The 35 chemical and physical
characteristics of stream water that were selected for in situ or laboratory measurement for the
NSS-I effort are listed in Table 3-1. A brief description of each parameter is given in the
NSS-I analytical methods manual (Hillman et al., 1987). Once collected, water samples were
transported via express courier to a central processing laboratory where they were preserved
and split into aliquots within 24 hours after sample collection. In addition to preparing the
samples for shipment to the contract analytical laboratory, the processing laboratory measured
pH, dissolved inorganic carbon (DIG), color, turbidity, conductance, and certain aluminum
species. At the analytical laboratory, 24 chemical variables were measured (Table 3-1). Data
from the analytical laboratory were entered into a data base, which then underwent a series of
stringent QA checks.
In addition to the chemical data, geographic data were compiled for all NSS-I sample
streams. This information represents an important component of the data base, as the weighting
factors used in making population estimates are related to the direct drainage of each NSS-I
stream (subsection 2.4.1). The geographic data compiled for each stream, the U.S. Geological
Survey (USGS) map scale used to calculate the data, and the NSS-I data base variables are
listed in Table 3-2.
3.2 FIELD SAMPLING PLAN
The NSS-I logistics plan specified that field samples were to be collected during a spring
index sampling period between snowmelt and leafout (mid-March to mid-May) at 533 streams
(1,037 sampling points) in a geographic area totaling more than 200,000 mi2. Based on hydrology
and phenology, this plan was designed to provide sampling consistency among areas and to
43
-------
Table 3-1. Chemical and Physical Variables Measured in NSS-I and Methods Employed
Parameters
Instrument or
analytical methods
Reference
laboratory
methods*
FIELD SITE
pit, in situ
Specific conductance
and temperature,
in situ
Dissolved oxygen
in situ
Current velocity
Portable meter (Beckman pHI-21);
glass combination electrode
(Orion Ross Model 8104)
Portable conductivity meter
with probe (YSI 33 S-C-T)
Portable dissolved oxygen meter
(YSI Model 54A); pressure-
compensating oxygen-temperature
probe (YSI 5739)
Electromagnetic current meter
(Marsh-McBirney, Model 201-D)
Hagley et al. (1988)
Hagley et al. (1988)
Hagley et al. (1988)
Hagley et al. (1988)
PROCESSING LABORATORY
Aluminum
Total monomeric
Nonexchangeable
monomeric
Specific conductance
pH, closed system
Colorimetry (pyrocatechol violet,
automated flow injection
analyzer)
Colorimetry as with total mono-
meric (after passing through
strong cation-exchange column)
YSI conductivity meter (Model 32);
YSI probe (YSI 3417)
pH meter (Orion Model 611);
glass combination electrode
(Orion Model 8104)
Hillman et al. (1987)
Hillman et al. (1987)
EPA 120.1
EPA 150.1
* EPA methods are taken from U.S. EPA (1983); USGS methods are from Skougstad et al. (1979).
44
-------
Table 3-1. Chemical and Physical Variables Measured in NSS-I and Methods Employed
(Continued)
Parameters
Instrument or
analytical methods
Reference
laboratory
methods*
Dissolved inorganic carbon, Infrared spectrophotometry
closed system (Dohrmann DC-80 carbon
analyzer)
EPA 415.2 (modified)
True color Comparator (Hach Model CO-1) EPA 110.2 (modified)
Turbidity Nephelometer (Monitek Model 21) EPA 180.1
CONTRACT ANALYTICAL LABORATORY
Acid neutralizing
capacity (ANC)
Aluminum
Extractable
Total
Ammonium
Base neutralizing
capacity (BNC)
Calcium
Chloride
Acidimetric tritration,
modified Gran analysis
Atomic absorption spectroscopy
(furnace) on methyl-isobutyl-
ketone extract
Atomic absorption spectroscopy
(furnace)
Hillman et al. (1987);
Kramer (1984)
Hillman et al. (1987);
EPA 202.2
EPA 202.2
Colorimetry (phenate, automated) EPA 350.1
Alkalimetric titration, modified
Gran analysis
Atomic absorption spectroscopy
(flame)
Ion chromatography
Hillman et al. (1987);
Kramer (1984)
EPA 215.1
ASTM (1984); O'Dell
et al. (1984)
* EPA methods are taken from U.S. EPA (1983); USGS methods are from Skougstad et al. (1979).
45
-------
Table 3-1. Chemical and Physical Variables Measured in NSS-I and Methods Employed
(Continued)
Parameters
Instrument or
analytical methods
Reference
laboratory
methods*
Dissolved inorganic
carbon (DIG)
Initial
Air equilibrated
Dissolved organic
carbon (DOC)
Fluoride, total dissolved
Iron
Magnesium
Manganese
Nitrate
PH
Air equilibrated
Initial ANC
Initial BNC
Infrared spectrophotometry
Infrared spectrophotometry,
after bubbling with 300
ppm CO2 air for 20 minutes
Infrared spectrophotometry,
after acidification and
sparging to remove DIG
Ion-specific electrode
Atomic absorption spectroscopy
(flame)
Atomic absorption spectroscopy
(name)
Atomic absorption spectroscopy
(flame)
Ion chromatography
pH electrode and meter, after
bubbling with 300 ppm CO2
air for 20 minutes
pH electrode and meter, at
start of ANC titration
pH electrode and meter, at
start of BNC titration
EPA 415.2 (modified)
EPA 415.2 (modified)
EPA 415.2
EPA 340.2 (modified)
EPA 236.1
EPA 242.1
EPA 243.1
ASTM (1984); O'Dell
et al. (1984)
EPA 150.1
EPA 150.1
EPA 150.1
* EPA methods are taken from U.S. EPA (1983); USGS methods are from Skougstad et al. (1979).
46
-------
Table 3-1. Chemical and Physical Variables Measured in NSS-I and Methods Employed
(Continued)
Parameters
Instrument or
analytical methods
Reference
laboratory
methods*
Phosphorus, total
dissolved
Potassium
Silica
Sodium
Specific conductance
Sulfate
Colorimetry (phosphomolybdate
automated), after acid-
persulfate digestion
Atomic absorption spectroscopy
(flame)
Colorimetry (silicomolybdate,
automated)
Atomic absorption spectroscopy
Conductivity cell and meter
Ion chromatography
USGS 1-4600-78
(modified)
EPA 258.1
USGS 1-2700-78
EPA 273.1
EPA 120.1
ASTM (1984);
O'Dell et al. (1984)
* EPA methods are taken from U.S. EPA (1983); USGS methods are from Skougstad et al. (1979).
47
-------
Table 3-2. Geographic Variables Measured in NSS-I
Geographic Attribute
USGS Map Scale
NSS-I Data Base
Variable Name
Watershed area above the site (km2) 1:24,000
Direct drainage area (mi2) 1:24,000
Watershed area above lower site (km2) 1:24,000
Site elevation (m) 1:24,000
Stream gradient (%) 1:24,000
Site latitude (decimal form) 1:24,000
Site longitude (decimal form) 1:24,000
Reach length between upper and lower sample sites (km) 1:24,000
Number of headwater reaches 1:250,000
Reach length between stream confluences (km) 1:250,000
Shreve order 1:24,000
ANC stratum (1 or 2) (section 2.4)
Strahler order 1:24,000
Reach identification code
Stream identification code
State (2-character code) 1:250,000
Stream name 1:24,000
Subregion identification code
A_WS
Al
A4
ELEV
GRADE
LATJSTD
LON_STD
L2
RCH_HW
RCHJLN
SHREV75
STRATUM
STRA75
RCH_ID
STRM_ID
STATE1
STRMNAM
SUB ID
48
-------
minimize the influence of external factors affecting stream chemistry during the season of
maximal plant growth. Florida was an exception to the phenological scheduling because leaf-
out was essentially completed by the time the survey began.
Sampling followed a pattern of early to late leafout (in general, early to late seasonal
warming). In order to visit streams during the index period, the sampling crews moved
through the Mid-Atlantic areas in such a way that they collected the second set of samples
from a given stream at least two weeks after the first visit, but before leafout.
Sampling crews visited streams between March 17 and May 15, 1986, accessing stream
sampling sites by foot and by means of two- or four-wheel drive vehicles. Water samples were
collected from the center of each stream at mid-depth by means of a peristaltic pump, which
pumped water through Tygon tubing held in position by a sampling boom. The NSS-I field
sampling activities are shown in Figure 3-1.
Thirty-nine sites served as temporary base stations for field operations, based on proximity
to NSS-I stream sites. Two-member sampling teams collected samples from one or two streams
each day at both upstream and downstream locations.
In order to avoid the bias that could be caused by diel fluctuations in chemistry resulting
from stream biota, sampling crews did not visit downstream and upstream nodes (sample
locations) at the same time each day. Half of the streams were selected randomly for sampling
at the upstream location first; the other half was sampled at the lower location first. In order
to determine sampling order, crews treated multiple visits to one stream as independent visits to
separate streams.
Although nontarget reaches were originally screened out by means of mapped information,
(subsection 2.3.1), on-site data was used to further refine the target population. The following
criteria were established to assist in identifying noninterest reaches:
1. Intermittent reaches that were at least 90% dry or stagnant.
2. High conductance: reaches having an in situ specific conductivity > 500 /iS cm"1
(e.g., reaches contaminated by oil well brine, industrial pollution).
3. Low pH: reaches having a field pH value of 3.3 or less (e.g., reaches impacted by
acid mine drainage).
4. Tidal influence: coastal reaches with water chemistry influenced by tides (e.g.,
specific conductivity greater than 250 0S cm"1).
5. Reservoir: reach inundated by water project.
6. Episode (other than spring baseflow conditions): reach chemistry influenced by a
precipitation event at the time of sampling (e.g., high turbidity and high flow). The
identification of precipitation episodes is discussed in subsection 3.8.3.1.
7. No channel, dry: no sample could be collected because of a lack of water (dry) or
explicit stream channel (swamp).
8. Other: reach inaccessible because access permission had not been given or because of
a permanent feature of the watershed (e.g., active military area).
49
-------
Arrive at
Stream Site
Photograph Sample Site
and Record Watershed
Characterists
Collect Water
Samples
Make In Situ
Measurements
Mean Site Depth
Dissolved Oxygen
pH
Conductance
Temperature
Field Blank
Sample
(Deionized Water)
Two 60 mL
Syringes
4 liter
Container
Routine Sample
Four 60ml_
Syringes
4 liter
Container
1
Field Duplicate
Sample
Four 60mL
Syringes
4 liter
Container
Record Hydrology Data
Stage Height
Stream Width
Velocity
Cross-sectional Depth Profile
Store at 4°C
Complete Field
Data Form
Travel to Next
Sample Site or
Return to Base
Figure 3-1. Field sampling activities, NSS-I (summarized from Hagley et al.f 1988).
50
-------
3.3 FIELD METHODS
3.3.1 Site Characteristics
On the first visit to a stream sampling location (node), photographs were taken of the site.
Watershed disturbances, land use,' bank vegetative cover, and stream substrate were recorded.
At downstream nodes, sampling crews collected hydrologic data, including stream channel and
flow measurements. In Mid-Atlantic subregions, these included stream stage, discharge, width,
and depth. On the first visit to each downstream site, nonrecording staff gauges were installed
and read. Mean depth and width were estimated. On the second visit, discharge was measured
using the velocity-area integration method. Flow velocities (0.6 depth from surface) were
measured at 8 to 15 intervals across the stream with an electromagnetic current velocity meter;
the calibration was checked each morning and again on site before each set of measurements.
At the downstream sites of the Southeast Screening areas, flow was estimated by timing the
movement of a floating object through a measured length of stream. Width and mean depth
were estimated.
3.3.2 Sample Collection
Water samples were collected using the techniques developed for the NSS-I Pilot Survey.
Routine samples were collected from each stream by pumping water through 1/4-inch Tygon tub-
ing held in the center of the stream at mid-depth with a 6-foot sampling boom. Water samples
were pumped into a 4-liter polyethylene Cubitainer using portable, battery-driven peristaltic
pumps. In addition, four gas-tight 60-mL polypropylene syringe samples were collected without
exposing the samples to the atmosphere in order to minimize changes in the water sample prior
to analysis. These syringes were used for the analysis of pH, dissolved inorganic carbon (DIG),
and total monomeric and nonexchangeable aluminum performed in the processing laboratory.
Detailed discussions of these techniques appear in Knapp et al., 1987, and Hagley et al., 1988.
Two types of QC samples were collected. Each day two teams collected a field blank
sample at the first site visited using reagent grade water from the processing laboratory. This
water was carried to the sample site and pumped through Tygon tubing into clean sample con-
tainers using techniques identical to those used to collect routine samples. In addition, two
teams each day collected a field duplicate. A second set of sample containers (Cubitainer and
syringes) was filled with stream water immediately after the routine sample was collected, using
identical techniques.
3.3.3 In Situ Measurements
Streamside and in-situ chemistry measurements included pH, conductivity, dissolved
oxygen, and temperature. Measurement and calibration techniques are described in detail in
Hagley et al. (1988). All field pH determinations were made using portable meters with com-
bination electrodes and temperature compensators. Field pH meters were calibrated each
morning using commercially available high ionic strength buffer solutions (pH 4 and 7). A
quality control check solution (QCCS) was used to check the calibration of the meter before
leaving for the field and again before and after in-situ measurements. Measurements of pH were
made in beakers of stream water, collected with the peristaltic pump. Replicate readings were
taken on new sample aliquots until two successive readings agreed within 0.03 pH units. The
final pH and temperature were the values reported.
51
-------
Conductivity was measured in situ using a portable meter with the probe immersed in the
stream at mid-depth. Three different concentration solutions were used to check the factory
calibration of the meters before each day's sampling activities. The meter calibration was
rechecked before and after each stream reading.
Dissolved oxygen concentration in a stream was measured using a dissolved oxygen meter
combined with a pressure compensating oxygen-temperature probe. The instrument was air cali-
brated each morning, at each stream site, and at the base site at the conclusion of sampling.
The calibration was checked using air-saturated water after morning and evening calibrations.
In-situ measurements were taken by immersing the probe at mid-depth in the streamflow.
3.4 SAMPLE HANDLING
Water samples were transported from the sample site in coolers containing chemical
refrigerant packs that maintained a temperature of approximately 4°C until they arrived at the
processing laboratory. Samples were shipped by overnight courier to ensure their arrival at the
processing laboratory in Las Vegas, Nevada, on the morning after collection. Upon arrival at
the processing laboratory, samples were organized into a batch for processing. A sample batch
consisted of a group of routine stream samples and related QA samples. In almost all cases,
processing laboratory analyses were completed and samples were preserved and split into aliquots
within 24-36 hours after sampling.
Within each batch, each sample was assigned a unique identification number to distinguish
it from all other samples in the survey. After the batch of samples was preserved and split into
aliquots, it was shipped by overnight courier to a contracted analytical laboratory for chemical
analysis.
3.5 PROCESSING LABORATORY TECHNIQUES
The processing laboratory provided a controlled environment for processing and preserving
water samples and performing certain chemical measurements that needed to be completed as
soon as possible after sample collection. Figure 3-2 illustrates the processing laboratory
activities. Chemical parameters that tend to become unstable over time (i.e., pH, DIG, and the
aluminum species) were measured in the processing laboratory.
Each 4-liter Cubitainer stream sample was split into seven aliquots and prepared for ship-
ping to analytical laboratories for additional analyses. Subsamples were also taken from each
Cubitainer for measuring turbidity, conductance, and true color.
Field crews capped the syringe samples with air-tight valves to prevent air equilibration
from occurring before analysis. The syringe samples were used to measure pH, DIC, and total
and nonexchangeable monomeric aluminum species and to prepare the total extractable aluminum
aliquot. Processing laboratory measurements of these variables were essential for providing
quality data within holding time requirements. Processing laboratory analytical methods are
described in Hillman et al. (1987).
Figure 3-2 depicts the seven aliquots prepared from each Cubitainer. The aliquots were
stabilized by filtration (0.45 urn filter), acid preservation, or refrigeration, or some combination
of these procedures. Filtration removed suspended material in order to reduce biological activity
and to eliminate surfaces that could adsorb or release dissolved chemical species. Acid was
added to some aliquots to prevent loss of dissolved analytes through precipitation, chemical
reaction, or biological reaction. All aliquots were stored and shipped at 4°C to reduce biological
activity and, for total extractable aluminum aliquots, to reduce volatilization of solvent.
52
-------
-IOC
is
SZ3
O
>0
O
I
cr
w.
ID
f
.O
rfz
CO
CO
O>
ll
CO
w»-
"5.J5
E
CO O
$
G)
QJ
_
J3
CO
0)
C O) c
I E g I
c u ^ c
o c o 'E
Q
£
53
-------
Table 3-1 lists the instruments and analytical methods used in the processing laboratory.
The processing laboratory also prepared and shipped reagents and supplies to field base sites.
Arent et al. (1988) give a detailed discussion of processing laboratory protocols for the NSS-I.
3.6 ANALYTICAL LABORATORY SUPPORT
Analytical contract laboratories were solicited by bid and selected through the analysis of
performance evaluation samples and on-site inspections (Cougan et al., 1988). Two analytical
laboratories were used to analyze NSS-I water samples. Table 3-1 lists the analytical instru-
ments and methods used by the analytical laboratories. A brief description of each analytical
method is given in the NSS-I analytical methods manual (Hillman et al., 1987). The maximum
allowable holding time for each analyte before analysis is given in Table 3-3.
3.7 QUALITY ASSURANCE AND QUALITY CONTROL PROTOCOLS
An important design criterion of the NSS-I is that the data collected must be of known
quality. An extensive data QA program was established to meet this requirement. The QA pro-
gram has two separate but integrated components that include both operational and data man-
agement activities. In the operations component, the QA procedures ensured that all samples
were collected and analyzed in a consistent manner. QC samples were collected to provide data
from which the accuracy and precision of the reported values could be estimated with a known
degree of confidence. The data management component established a program that stored and
tracked the data, identified and corrected entry, reporting, and analytical errors, and kept a
record of such changes. These procedures produced documented files of data of known quality
that were accessible to project scientists and extramural users.
3.7.1. Quality Assurance Plan
The NSS-I QA plan (Drouse et al., 1986) defines the activities required to ensure that
procedures are performed consistently and that the quality of the data generated can be deter-
mined. QA and QC activities identified in the plan include (1) training field sampling and
processing laboratory crews, (2) maintaining communication with management, sampling, and
analytical personnel, (3) conducting on-site field and laboratory inspections, and (4) collecting
and analyzing QA and QC data in order to quantify data quality. Quality assurance protocols
(Drouse et al., 1986) were developed for collecting, preparing, preserving, shipping, and analyzing
samples, as well as for reporting, verifying, and validating analytical results. Both the
processing laboratory and the analytical laboratories analyzed QC samples within each batch.
3.7.2 Quality Assurance and Quality Control - Data Collection and Analysis
The NSS-I QA program used several types of QC samples (Table 3-4 and Figure 3-3) to
ensure that sampling and analytical methods were performed according to specifications. The
results from QC sample analyses were used to evaluate the performance of field sampling
methods, field procedures, and laboratory analyses. The interpretations of QC data were also
used to evaluate overall data quality for the survey. Analyses of the QC samples allowed field
samplers and laboratory personnel (in both the processing and analytical laboratories) to identify
and immediately correct specific problems such as poor instrument performance or reagent
contamination before and during routine sample analyses.
54
-------
Table 3-3. Maximum Holding Times Specified for NSS-I Samples
Variable
Holding Time
NO3'&; extractable aluminum
ANC; BNC; specific conductance; DIG; DOC; pH*
P; NH4+; CT; SO42-; F-; SiO2
Ca; Fe; K; Mg; Mn; Na; total aluminum
7 days
14 days
28 days
28 days*
& Although the EPA (U.S. EPA, 1983) recommends that nitrate in unpreserved samples (unacidi-
fied) be determined within 48 hours of collection, evidence exists (Peden, 1981; APHA, 1985)
that nitrate is stable for 2 to 4 weeks if the sample is stored in the dark at 4°C.
* Although the EPA (U.S. EPA, 1983) recommends that pH be measured immediately after
sample collection, evidence exists (McQuaker et al., 1983) that it is stable for as long as 15
days if the sample is stored at 4°C and sealed from the atmosphere. The pH was also
measured in a sealed sample at the processing laboratory within 24-36 hours of sample
collection.
$ Although the EPA (U.S. EPA, 1983) recommends a 6-month holding time for these metals, this
study required that all the metals be determined within 28 days, which ensured that
significant changes would not occur and that data would be obtained in a timely manner.
55
-------
Table 3-4. NSS-I QC Samples
Sample type
Description
Function
Frequency of use
Field blank
Processing
Laboratory blank
Field duplicate
Field
Performance
Audit
Laboratory
Performance
Audit
Reagent-grade deionized
water* subjected to
sample collection,
processing, and
analysis
Reagent-grade deionized
water* subjected to
sample processing and
analysis
Duplicate stream
sample
Synthetic or natural
lake sample; proc-
essed at processing
laboratory
Calibration blank
Synthetic or natural
lake sample; proc-
essed at analytical
laboratory
Reagent-grade deionized
water*
Estimate system One per batch
decision and
detection limits;
identify sample
contamination
Estimate background In lieu of field
due to sample proc- blank when
essing and analysis logistical con-
straints pre-
vented its
collection
Estimate field and
system variability
Estimate analytical
precision of proc-
essing and analyti-
laboratory measure-
ments; estimate
accuracy and inter-
laboratory bias
Estimate analytical
precision of anal-
lytical laboratory
measurements; esti-
mate accuracy and
interlaboratory bias
Identify signal
drift
One per batch
As scheduled
As scheduled
One per batch
*ASTM (1984)
56
-------
Table 3-4. NSS-I QC Samples (Continued)
Sample type
Description
Function
Frequency of use
Reagent blank
Quality control
check solution
(QCCS)
Detection limit
QCCS
Processing
laboratory
duplicate
Analytical
laboratory
duplicate
Reagent-grade deionized
water* plus reagents
for total Al, SiC>2
analyses
Standard solution from
source other than cali-
bration standard
Standard solution at 2
to 3 times the required
detection limit
Split of stream sample
Split of sample aliquot
Identify contam-
inants present
in reagents
Determine accuracy
and consistency
of instrument
calibration
One per batch
Before the first
measurement
and as speci-
fied
Determine accuracy One per batch
and precision at
lower end of linear
dynamic range of
measurement method
Determine precision
of processing lab-
oratory measure-
ments
Determine precision
of analytical lab-
oratory measure-
ments
One per batch
One per batch
*ASTM (1984).
57
-------
S • v
DATA QUALITY
ESTIMATE
DETERMINED
L
ANALYTICAL
LABORATORY
/Al \
DATA QUALITY
ESTIMATE
DETERMINED
x_ J
PROCESSING
LABORATORY
(PL)
ius
<*~^
X- J
Q i.i
T
c
c
C
1-
u
/ V
1*.
J
c
•*
»
•
SYSTEM LEVEL
^ DETECTABILITY J
T
z
a
UI
C
1 i
je
•z
_i
o
5
u
u.
1
a
_j
n
£
f
c.
:
c
a
s
c.
<
tc
^v.
3
:
«
c
c
i
*
*
a
a
CONTAMINATION J
r
e
:
D
j
L
i
i
' SYSTEM LEVEL "\
PRECISION V
T
[ FIELD DUPLICATE
FIELD DUPLICATE
l\
UI
\
0.
a
o
_i
UI
u.
< METHOD LEVEL ^
PRECISION j
T
u
a
a.
i
|
1
/!\
& JL
ui Z
•z <
P CO
AMONG BATCH PRECISION
V_
i
a
C
J
o
3
UI
u.
j
T
x—
/ 'INSTRUME
T
8£
0 0.
d"
V
i £
i|
P
r
LABORATORY AUDITS
>
LABORATORY AUDITS
^
L
c
a
/
t^^
C
I
w
.£>
i
^
<
— ^
\
INSTRUMENT
DRIFT
T
CO
u
u
O
./* INSTRUMENT ^
V OR|FT >
T
o •
i *S >
8— u.t:
< iS
o j-oja
o|s|
JL =
•
i
<
METHOD LEVEL
DETECTABILITY
Y
CALIBRATION AND
REAGENT BLANKS
f METHOD LEVEL >
V^ DETECTABILITY J
t
*
n J S
s '. -3
25 i
° i •§"
33 s
S
UI
_i.
o
o
1
i
LABORATORY
UI
_J
a
1
I-
5
Uj
UI
CO
i:
m :
PRECISION
J
n
\
I-
8
a
V.
-
I
^2
t
o
a
ro
I
CO
58
-------
Additional QC samples, introduced into batches of stream samples at the stream site or at
the processing laboratory, were analyzed at the processing and analytical laboratories. These
samples were used to judge the overall performance of NSS-I sampling and analytical activities
and to establish data quality. Analytical laboratories were not told the origin, identity, or
chemical composition of audit samples. Thus, the audit samples were analyzed as if they were
routine stream samples.
3.7.3 Training and Site Audits
Field personnel participated in a training program that covered NSS-I objectives,
stream sample collection and measurement techniques, orienteering, outdoor skills, and safety.
Processing laboratory personnel training covered all technical aspects of laboratory
operations, including analytical methods, instrument calibration, sample handling, QA, and
safety procedures. On-site evaluations were conducted to ensure that sampling and analysis
activities were being performed according to the QA plan and the statement of work. The
field base sites, the processing laboratory, and one contract analytical laboratory were evaluated
during the NSS-I. Budget constraints precluded an on-site evaluation of the other contract
laboratory (Cougan et al., 1988). Quality control sample data were reviewed and used in
conjunction with on-site inspections to identify the need for changes in sampling or analytical
protocols.
n
3.8 DATA BASE MANAGEMENT
NSS-I data management and analysis were patterned after procedures developed for the NLS
(Kanciruk et al., 1986) and are described in detail in Sale (1988). All NSS-I data sets were
maintained at Oak Ridge National Laboratory (ORNL) on tandem IBM 3033 mainframe
computers, using the Statistical Analysis System (SAS) software package (SAS Institute, Inc.,
1985). When the data sets were complete, they were transferred to the National Computer
Center (NCC) at Research Triangle Park, North Carolina, via magnetic tape, where they could be
accessed by NSS scientists at the Las Vegas and Corvallis laboratories.
Ultimately, the success of the NSS-I will be based on the ability of the data to produce
robust population estimates regarding the target population of streams as defined in this report.
The NSS-I data base used in making population estimates has been subjected to four levels of
stringent QA evaluation to ensure that the data collected during NSS-I is representative of the
physical and chemical characteristics of the streams at the time of sampling. These checks were
also useful in identifying reaches that are unusual with respect to others in the target
population and that may represent polluted sites, or episode samples.
An important tool in the development of the NSS-I data base is the use of data qualifiers
to mark an individual value or even an entire stream as having a particular feature that may be
useful in data interpretation. Two types of data qualifiers, flags and tags, are used to document
notes made about particular data records or values in the NSS-I data base. Tags identify
specific notes made in the field, at the processing laboratory, or at the analytical laboratory.
Flags identify observations or notes made during the QA process (verification and validation)
involved in creating the enhanced data base. These qualifiers provide a method for identifying
nonrepresentative and questionable data with appropriate flags and tags.
The completion of each level of QA produced a new working data set of increased refine-
ment. These working data bases are defined as: raw (Data Set 1), verified (Data Set 2),
validated (Data Set 3), and enhanced (Data Set 4). The final product of this refinement process,
59
-------
the enhanced data set (Data Set 4), incorporates data substitution and replacement of missing
values. This is the data set that is used for calculating NSS-I population estimates. Figure 3-4
summarizes the development of these working data bases. The data bases are described further
in the following subsections.
3.8.1 Raw Data (Data Set 1)
The collective data from all components of the sampling and analysis made up the raw data
set. Field, processing laboratory, and analytical laboratory personnel sent the original data
forms to the QA staff at the Environmental Monitoring Systems Laboratory (EMSL) in Las Vegas
for review, in order to ensure that data were correct and consistent. Completed forms were
then forwarded to the Oak Ridge National Laboratory (ORNL), where the data were entered into
the data base. To ensure accurate data transfer from the data forms to computer files, the
information was double-entered into computer files and subjected to automated checking
procedures (Rosen and Kanciruk, 1985). The raw data set was used to screen the data for
problems, perform exploratory analysis, and evaluate the need for any adjustments in the data
analysis plan.
3.8.2 Verification (Data Set 2)
The objectives of the data verification process were to identify, correct, and flag raw data
of questionable or unacceptable quality and to identify data that might need to be corrected
during or after data validation. These objectives were met by reviewing the QC data measured
and recorded at the sampling site, at the processing laboratory, and at the analytical labora-
tories. The verification process was automated as much as possible through applicable computer
programs. Data verification took place in two parts. First, the required numeric changes were
identified, and then final numeric changes and flag setting were accomplished.
Verification began with receipt of the data forms from the field and processing laboratory.
An NSS-I auditor reviewed the forms for completeness, agreement between field and laboratory
forms, and proper assignment of sample identification codes and data qualifier tags. Data
anomalies were reported to the field base site and processing laboratory coordinators for
corrective action. Data reporting errors were usually corrected on the data forms before the
data were entered into the raw data set. During verification, each sample was evaluated
individually and by analytical batch. Individual values that were identified as exceptions (as a
result of cation/anion balances, conductance balance, or protolyte analysis) or that did not meet
the acceptance criteria, were flagged in the data base. Suspect values were also identified by
examining QC data (blank, duplicate, and audit samples) measured and recorded at the processing
and analytical laboratories. In addition, data qualifier flags were added when QC samples did
not meet acceptance criteria, or when sample analysis holding time requirements or instrument
detection limits were not met. Table 3-5 lists the computerized verification checks that were
used to identify exceptions in the chemistry data. Table 3-6 shows the calculations performed
for anion/cation balances, conductance estimates, and protolyte analysis.
The output from these checks, along with original data and field notebooks, was used to
evaluate analytical results. Based on the evaluation of the analytical results reported for QC
samples, analytical laboratories were directed to confirm reported values or to reanalyze selected
samples. If a value was identified as an exception to expected results, a flag was placed in the
data base to relay information about the disposition of this individual observation or value
(Drouse et al., 1987).
60
-------
rield Sampling)
land Processing
.Laboratory
Analytical
Laboratories
I
Visual Form Check
Data Entry I
Data Entry 2
Error and
Range Check
RAW DATA SET
(Data Set I)
Batch Reports
Verification
Data Editing
and Flagging
^ I
VERIFIED DATA SET
(Data Set 2)
Site
Reports
Maps
Validation
Data Editing
and Flagging
I
Substitution
and
Replacement
VALIDATED DATA SET
(Data Set 3) |J_
ENHANCED DATA SET
(Data Set 4)
Figure 3-4. NSS-I data base development.
61
-------
Table 3-5. Exception Generating Programs within the Verification System*
Verification Process
Data Evaluated
Audit Sample Evaluation
Lab/Field Blank Evaluation
Field Duplicate Precision Evaluation
Instrumental Detection Limit Evaluation
Holding Time Evaluation
Conductance Check Calculations
Anion/Cation Balance Calculations
Protolyte Analysis
Reagent/Calibration Blanks and QCCS
Field synthetic audit
Laboratory synthetic audit
Field natural audit
Laboratory natural audit
Field blank
Laboratory blank
Routine field duplicate pairs
All species
All species
All samples
All samples
All samples
All batches
Summarized from Cougan et al., 1988.
62
-------
Table 3-6. Calculations Performed as Part of Cation/Anion Balances, Conductance Estimates,
and Protolyte Analysis in Verification
Verification Checks
Anion/cation Balance
%Ion Balance Difference =
ANC +• Sum of Anions-Sum of Cations
Total ion strength
* 100
Sum of Anions= [Cl'] + [F~] + [NO3-] + [SO42-]
Sum of Cations= [Na+] + [K+] + [Ca2+] + [Mg2+] + [NH4+j
Total ion strength = ANC + sum of anions + sum of cations + 2[H+]
ANC = [HCO3-] + 2[COS2-] + [OH"] + [titrated organic anions] - [H+]
Conductance Balance
% Conductance Difference =
(calculated conductance - measured conductance)
* 100
measured conductance
Protolyte Analysis
Comparison of:
Measured ANC vs. carbonate alkalinity + organic anion
Measured BNC vs. CO2 acidity + organic acid anions
Where:
Measured ANC and BNC were from Gran titrations
Carbonate alkalinity was calculated from DIG and pH variables
CO2 acidity was calculated from DIG and pH variables
Organic anion was calculated from DOC using Oliver Model (Hillman et al., 1987; Oliver et al.,
1983)
63
-------
Figure 3-5 shows the verification process flow. Detailed information about verification
procedures is found in Drouse et al. (1987) and Cougan et al. (1988).
3.8.3 Validation (Data Set 3)
Whereas the purpose of verification procedures was to evaluate data at the sample and
batch level, validation was intended to compare samples within the population of units (all
samples collected within a subregion). The validation process identified unusual data that would
need special attention when used for statistical analysis, particularly for regional estimates
concerning the target population of streams. The two main components of the validation process
were (1) the identification of statistical outliers from subregional distributions of chemistry and
(2) the evaluation of possible systematic errors in the measurement process. This approach is
empirical in nature, as opposed to the geochemical and charge balance orientation of verifica-
tion. Because NSS-I subregional boundaries generally group streams of similar geochemistry
together, these empirical relationships reflect the geochemical trends of each subregion. Valida-
tion outliers may result from a number of conditions at the reach or from data base problems.
Conditions that may cause outliers include:
1. The determination that the water sample was collected during a time that might be
considered a precipitation episode event for a given reach.
2. The determination that a given reach is being influenced by factors beyond normal
geochemical processes (e.g., pollution or watershed disturbance, including acid mine
drainage, brine, or other nonpoint sources).
3. Unusual geochemical properties within a given subregion.
4. Impossible datum, clearly erroneous when reviewing chemistry for that reach.
5. Observation or identification inconsistencies (e.g., mislabelling of stream or batch ID).
Observations that surface as atypical during review of data at a subregional level are con-
sidered outliers from the rest of the data. Outliers in the data may result from the natural
variability of streams in the set of sample reaches, from anthropogenic disturbances in the
natural environment, and from errors in the sampling design, as well as from sampling and
analytical errors. Although outliers may represent unusual data in comparison with other data,
such values are not necessarily inaccurate in their representation of a sample stream reach. The
validation process is, therefore, not meant to be a stringent pass/fail test, but rather a way to
search for observations that may represent entry or analytical errors, or perhaps unusual water
chemistry. Data outliers are not typical or common in the sample data sets. The unusual
observations become apparent when the data are viewed as a set of information using univariate,
bivariate, and multivariate analyses.
Figure 3-6 shows a flow diagram of the validation process. The statistical approaches and
specific techniques summarized in Table 3-7 were applied to the validation of NSS-I data. Dur-
ing validation, a matrix for each subregion was constructed that depicted the results of vali-
dation checks on each individual water sample and datum. Using this matrix, outliers were
identified and sent to the QA/QC group at EMSL-LV to be checked for possible entry errors.
By means of principal component analysis, cluster analysis, and multivariate regression analysis,
64
-------
RAW DATA
QA Review and
Exception Generating
Programs
Review Exception
Records
Reanatysts or
Confirmation
Required ?
Contact Datum Source
Request Confirmation
or Reanalysis
NO
Update Database
V
Exception Generating
Programs
VERIFIED DATA
Figure 3-5. NSS-I data verification procedures.
65
-------
VERIFIED DATA
UNIVARIATE STATISTICS
Box Plots
Probability Plots
MULTIVARIATE STATISTICS
Principal Component Analysis
Cluster Analysis
Trilinear Diagrams
Multivariate Regressions
Outlier Evaluation,
Unusual Site Identification
BIVARIATE STATISTICS
Scatter Plots
Regressions
Figure 3-6. NSS-I data validation procedures.
66
-------
Table 3-7. Types of Validation Analyses Performed on NSS-I Raw Data
UNIVARIATE ANALYSIS
Univariate statistics: Summary statistics, stem and leaf diagrams, probability plots, box
plots, and quartile estimates, applying techniques of Velleman and
Hoaglin (1981).
Soatial and Temporal Analysis
Comparison of downstream vs. upstream differences and visit 1 vs. visit 2 differences.
Difference ratios treated as univariates and subjected to standard univariate analyses.
MULTIVARIATE ANALYSIS
Principal Component
Analysis:
Regression Analysis:
Cluster Analysis:
Statistical relationships among data were identified using corre-
lation coefficients and sets of eigenvectors. Within in each
eigenvector, eigenvalues for each variable were calculated. From
these eigenvector and eigen values, relationships between vari-
ables were identified and used to help explain data variability.
Multi-linear regressions were performed from cross-correlation
matrices, using statistical relationships among variables. A pre-
dicted versus a reported value for each variable observation was
assessed and compared to a predicted value based on the rela-
tionship expected from the correlation matrices. Least squares
regressions were performed that identif yed outliers at a particular
residual threshold—standardized residuals (SAS, 1985). Removing
outliers sequentially was useful in identifying outliers that might
not have been apparent without removing aberrant values.
Attention was given to leverage points, ordinate extremes, and
abscissa extremes.
Agglomerative disjointed cluster analysis (SAS, PROC FASTCLUS)
was performed using Euclidean distance with an iterative
algorithm for minimizing the sum of squared distances from the
cluster means.
67
-------
sites having atypical chemistry compared with other sites in a subregion (unique multivariate
relationships) were identified and evaluated as unusual sites (sites affected by acid mine drain-
age, agricultural impact, tidal influence, etc.). Data from unusual sites generally appeared as
outliers in several analyses (regression, univariate statistics, etc.).
3.8.3.1 Episode Identification—•
Besides identifying outliers, validation served to identify stream chemistry that might have
been sampled during a precipitation or snowmelt episode (or influenced by it) and therefore
would not provide an acceptable index of base flow chemistry. Although great care was taken
not to collect samples during an episode, such conditions might have not been apparent to a
sampling team. Episodes samples were identified according to the following criteria:
1. Site is identified as a validation outlier.
2.
3.
4.
Change in stream gauge height of 7.5 cm or more, between site visits, supporting
evidence of flood stage (except for Southeast Screening sites where no second visits
to a site were made that would provide comparative stage height information).
Field comments indicated precipitation or snow melt activity (within 24 hours before
sampling at one or other node).
Data for turbidity, total aluminum, manganese, and iron indicate possible episode at
time of sampling (i.e., high turbidity and/or total aluminum relative to other visits at
the same site or its corresponding upstream or downstream node).
3.8.3.2 Acid Mine Drainage-
Sites impacted by acid mine drainage represent a category of streams that must be identi-
fied in order to distinguish them from those impacted by acidic deposition (i.e., those with low
alkalinity and pH and high sulfate). These sites were considered noninterest sites for calcu-
lation of NSS-I population estimates and were identified as satisfying all the following criteria,
the rationale for which are discussed in subsection 9.3.1:
1.
2.
3.
4.
5.
6.
ANC < 0 /«;q IT1
Sum of base cations > 400 peq L"1
Sulfate/sum of anions > 75%
DOC < 5 mg L'1
[SO42-] > 300 neq L"1 in the Mid-Atlantic; [SO42-] > 200 fieq L'1 in the Southeast.
Mining activity confirmed by means of maps, field visits, or aerial photographs.
3.8.3.3 Flagging of Unusual Values or Sites--
Values identified as unusual as a result of validation analyses were flagged in the validated
data set. When data for an entire reach was considered unacceptable for the intended use of
the data—to make population estimates, for example—a site flag code was placed in the data set.
Examples include data from sites affected by acid mine drainage, pollution, tidal influence, and
watershed disturbance.
68
-------
3.8.3.4 Geographic Data Verification and Validation-
Concurrent with the verification and validation of chemical data, the NSS-I geographic data
underwent a separate verification and validation process that examined data associated with the
Stage I and II selection process, as well as data from reaches as they were sampled (e.g., eleva-
tion of actual sampling location versus intended location). The verification and validation of the
geographic data included:
1. Comparison of total drainage area to sum of als a2, and as
2. Comparison of Stage I reach order to Shreve order
3. Comparison of Shreve versus Strahler orders
4. Examination of headwater reach length, watershed area, and order
5. Comparison of upper and lower node elevations
6. Comparison of Stage I and II geographic data versus geographic data of sites as
sampled
3.8.4 Enhanced Data (Data Set 4)
An important aim of the NSS-I data base development process was the creation of a data
base that could be used to calculate population estimates. This was accomplished by using a
single index observation for each sample node (upstream and downstream). In cases where there
were multiple visits, as in the mid-Atlantic, the index value was calculated as the average of the
values reported from the separate visits. The mean of multiple pH values was calculated as the
mean of hydrogen ion and converted into pH.
The calculation of population estimates and their confidence bounds is difficult if there are
inconsistencies in the data (e.g., missing values). A final data set was prepared to resolve prob-
lems with erroneous data and missing values in the validated data set. When it was necessary,
substitutions were performed according to the following criteria:
1. Values from duplicate samples were used whenever possible.
2. If a duplicate measurement was not available, a value from an alternate visit to the
site was used.
3. If a duplicate measurement or a measurement from an alternate visit was not avail-
able, a substitution value was calculated by means of a linear regression model or ion
balance estimate. This was done by calculating a predicted value based on observed
relationships with other chemical variables or predicting a value based on relationships
between upstream and downstream observations of the same chemical variable.
All substitute values were examined for acceptability before they were included in the
final data set. In addition to the substitute values that were calculated, negative values for
parameters other then ANC and base neutralizing capacity (BNC) were set equal to zero. All
values modified in the final data set were flagged. Streams considered to be noninterest were
flagged in a manner by which they could be excluded when using the enhanced data set for
making regional estimates of the target population.
69
-------
The NSS-I data base development process focused on providing data for calculating esti-
mates of characteristics of the target stream population. The enhanced data set was structured
in such a way to allow the easy inclusion or exclusion of data from noninterest (or nontarget)
sites (e.g., tidal influence or acid mine drainage) in statistical analyses (Section 5.2). This is
done by using a drop code variable (DRPCDE) as a criterion to subset data for making extensive
estimates of NSS-I target population of streams. In addition, estimates can be made for the
target population plus streams that did not meet exclusion criteria (i.e., those sites identified as
noninterest). The drop codes are encoded as follows:
Drop Code Exclusion Criteria Description
5
4
3
2
1
NSS-I special interest sites (Section 2.6).
Pilot Survey nonindex data (subsection 1.3.2.1).
Sites impacted by acid mine drainage (subsection 9.3.1).
Noninterest sites (Section 3.2).
Alternate node of noninterest site (discussion follows).
NSS-I target observation for both the upper and lower node.
The extensive estimates presented in Volume II exclude observations with exclusion criteria
of "2" or greater. The data base and drop code variable allowed calculation of extensive esti-
mates of upper and lower stream nodes as separate populations. However, in order to perform
analyses requiring matched upper and lower node (e.g., interpolated length estimates), the data
must be subset to remove observations that have a DRPCDE value of "1" or greater. The
resulting data subset contains only data for sites that have matching upper and lower node
observations.
Table 3-8 summarizes the total number of numeric changes to chemical variables during the
NSS-I verification, validation, and enhancement process. The verification data include all
samples—routine, duplicate, processing laboratory duplicate, audit, and blank. There was a total
of 56,019 values. As a result of verification, 9,052 changes (16.2% of the data) were made. The
changes were associated primarily with the following:
1. All alkalinity and acidity were recalculated after a refinement in the Gran titration
regression curve fitting program.
2. Processing laboratory conductivity was recalculated for temperature compensation.
3. Chemical reanalyses were performed by the analytical laboratories.
4. Corrections were made to blank data reported from analytical laboratories.
Only data from routine and duplicate samples were examined during the validation process.
Of these 44,975 values, 83 (less than 0.5% of the data base) were flagged. Only 34 values were
actually substituted (enhanced) in data set 4. Of the 34 replaced values, 13 were replaced with
data from an alternate visit. The remaining 14 values were replaced using multiple linear
regression and ion balances. An additional step in creating data set 4 was the scoring of
impossible negative values to zero for variables other than ANC and BNC. A total of 225 values
(0.5%) of the data base) across 9 variables (extractable aluminum, total aluminum, DOC, iron,
manganese, ammonium, nitrate, total dissolved phosphorus, and silica) had values scored to zero,
with total dissolved phosphorus having the most scoring changes, at 99 values. The bias due to
this adjustment did not affect the population estimates presented in this report.
70
-------
Table 3-8. Numeric Changes Made in the NSS-I Data Base as a Result of Verification and
Validation
No. of Numeric
Changes from
Variable Verification
ACCO11 BNC
ALDS16 Monomeric Al (PCV)
ALEX16 Extractable Al (MIBK)
ALKA11 ANC
ALOR16 Nonexch. Monomeric Al (PCV)
ALTL16 Total Al
CA16 Calcium
CL16 Chloride
COLVAL True Color
CONDI 1 Conductivity, Analytical Lab
CONIS Conductivity, in situ
CONVAL Conductivity, Processing Lab
DICE 11 DIG, air equilibrated
DICI1 1 DIG, initial
DICVAL DIG, closed system
DO_IS Dissolved Oxygen
DOC 11 DOC
FE16 Iron
FTL16 Fluoride
K16 Potassium
MG16 Magnesium
MN16 Manganese
NA16 Sodium
NH416 Ammonium
NO316 Nitrate
PH_R pH, in situ
PHAC11 pH, initial BNC
PHAL11 pH, initial ANC
PHEQ11 pH, air equilibrated
PHSTVL pH, closed system
PTD16 Total Dissolved Phosphorus
SIO216 Silica
SO416 Sulfate
TURVAL Turbidity
1642
7
4
1626
11
1298
594
10
3
19
1
1358
17
9
5
0
0
8
6
92
65
37
4
369
17
0
1073
1072
1
21
72
904
28
9
No. of Numeric
Changes from
Validation
0
0
7
0
2
0
0
0
0
1
2
2
13
0
0
0
0
0
0
0
0
0
0
45
0
1
0
0
0
0
0
0
0
1
No. of Numeric
Changes from
Enhancement
0
1
0
1
1
1
1
1
0
2
1
0
2
1
2
6
1
1
0
0
0
0
1
2
2
0
0
0
2
0
0
1
2
2
71
-------
3.9 DIFFERENCES BETWEEN THE NSS-I AND THE PILOT SURVEY
A number of changes in methods and procedures were made for the NSS-I as a result of
the Pilot Survey. Table 3-9 summarizes these changes.
3.9.1 Sample Holding Times - 24 Hours (NSS-I) vs. 12 Hours (Pilot Survey)
Holding times for Cubitainer and syringe samples (before aliquots were prepared and
samples were analyzed upon arrival at the processing laboratories) were increased from 12 hours
in the Pilot Survey to 24 hours in the NSS-I. The decision to increase sample holding times
was based on the results of two experiments (Messer et al., 1986): (1) a laboratory study of
CO2 impermeability in syringe sample containers (Burke and Hillman, 1987) and (2) a field study
of Cubitainer sample holding times (Stapanian et al., 1987; Knapp et al., 1987). Syringe
experiments determined that holding times for DIG and pH held in syringes at 4°C, could be
increased to 24 hours without significant change in these variables. Since pH changes and CO2
degassing appear to be the most significant causes of changes in aluminum speciation, it was
assumed that syringes for aluminum extraction can also be held for 24 hours. These Cubitainer
experiments also demonstrated that increasing holding times to as much as 24 hours would not
have a significant effect on determinations of the chemical characteristics involved.
3.9.2 Processing Laboratory Location
Because of the increased holding time criteria for syringes and Cubitainers during NSS-I,
sample processing (aliquot preparation, preservation, and preliminary analyses) was centralized at
the U.S. EPA Environmental Monitoring Systems Laboratory in Las Vegas, Nevada. The reloca-
tion of the processing laboratory to a centralized location provided savings in resources and
better QC, compared to the decentralized deployment of the laboratories at locations in the
NSS-I subregions.
3.9.3 Field oH Measurement
Comparisons of closed- and open-system measurements made during the NSS-I Pilot Survey
(Messer et al., 1986) indicated high comparability between the two field pH measurements and
the processing laboratory pH measurements. The logistically simpler and less time-consuming
open-system measurement was chosen to determine field pH for NSS-I.
3.9.4 Methods of Fractionation and Determination of Aluminum Species
A semi-automated colorimetric method for fractionation and determination of aluminum
species (complexation with pyrocatechol violet) was used during the NSS-I because it was
expected to be less expensive, less time consuming, and more reproducible than the 8-oxine
method used during the NSS-I Pilot Survey (Messer et al., 1986). Automation of this method
reduced the variability among analysts and eliminated the problems of reproducibility and precise
timing inherent in the manual 8-oxine method. However, since the large-scale application of
this method was being developed at EMSL-Las Vegas, total extractable aluminum measurements
using the MIBK method were also continued throughout the NSS-I to permit comparison.
72
-------
Table 3-9. Differences Between NSS-I and the Pilot Survey
Technique
Pilot
Phase I
Sample holding times
Processing laboratory
location
Field pH
Methods of fraction-
ation and determination
of aluminum species
Matrix spike QC samples
Phosphorus measurement
Specific conductance
in processing
laboratory
12 hours
Within subregion
Closed system
8-oxine method
Used
Unfiltered total
phosphorus
Not measured
24 hours
Centralized at
EMSL - Las Vegas
Open system
Colormetric (pyrocatechol
violet)
Not used
Filtered dissolved phosphorus
Measured
73
-------
3.9.5 Matrix Spike Quality Assurance Samples
The purpose of the matrix spike samples was to establish a concentration matrix, similar to
that of the samples collected, which could be used to verify the accuracy of analysis. An
aliquot of a sample was spiked with a known quantity of analyte prior to analysis. The per-
centage of spiked analyte recovered (percent recovery) determined whether or not there was a
matrix effect on the original sample. During the NSS-I Pilot Survey, the limits for spike
recovery were met for every batch and no matrix interferences were observed (Drouse et al.,
1987). In addition, matrix spike analyses from the Eastern and Western Lake Surveys showed no
matrix interference. Because these QC samples did not appear to provide any additional infor-
mation about the quality of the data, matrix spike samples were not used in the NSS-I full
survey.
3.9.6 Total Dissolved Phosphorus (NSS-I) vs. Total Phosphorus (Pilot)
Recent studies (e.g., Young et al., 1985), have shown that particulate-bound phosphorus
tends to have a wide range of bioavailability, depending on its source. Soluble (filterable) phos-
phorus, rather than total (unfiltered) phosphorus was selected for the NSS-I in order to provide
a better measure of biologically active phosphorus.
3.9.7 Specific Conductance Measured in the Processing Laboratory for NSS-I
During both the Pilot Survey and the NSS-I, specific conductance was measured in the field
and at the analytical laboratories. Since the equipment for the specific conductance measure-
ment was already available in the processing laboratory, an additional conductivity measurement
was made during the NSS-I to provide another comparison for the data user.
74
-------
SECTION 4
DATA QUALITY ASSESSMENT
4.1 INTRODUCTION
An important objective of the National Stream Survey - Phase I (NSS-I) is that the data be
of high quality, have low and quantifiable analytical error, have known precision, and be repre-
sentative of the state-of-the-art attainable in high-volume contract analytical laboratories. The
quality assurance and quality control (QA/QC) program of the NSS-I was designed to maximize
the utility of the collected data and to minimize the likelihood of erroneous chemical data. This
was accomplished through the establishment of data quality objectives (DQOs), which were used
as guidelines in maintaining a high level of data integrity during the sampling, analysis, and data
collection and recording.
This section describes the analytical approach, results, and conclusions for each of the five
aspects of data quality (completeness, comparability, representativeness, accuracy, and precision),
plus detectability, applicable to NSS-I data. Detectability was also analyzed because the low
ionic strength of many of the NSS-I samples necessitated an evaluation of the background levels
of analytes. For detectability, accuracy, and precision, data were evaluated at two levels. The
analytical (method) level, for which specific target values or DQOs were identified for each
analyte, evaluates performance of the measurement system under a high degree of process
control. The second, or system level, for which target values were not specifically defined, was
intended to evaluate the additional effects that the sampling design, sample collection, and
sample handling and processing had on the analytical results.
Representativeness, completeness, and comparability were considered important data quality
goals in the development of the statistical sampling design of the NSS-I and the QA plan. They
were affected by uncontrollable events that influenced the number of samples actually collected
and analyzed by the proper protocols during the course of the survey. Detectability, accuracy,
and precision were quantitatively assessed by using the analytical results from QC samples.
Analysis of the QC samples (blanks, performance audits, and duplicates) described in Sec-
tion 3.3 provided two kinds of information for the assessment of data quality. Sampling and
laboratory performance could be gauged against the DQOs established for precision, accuracy,
and detectability. In addition, unforeseen effects of the collection and measurement process on
analytical results could be quantified and their impact on data interpretation discerned. For
example, the addition of background levels of an analyte during sample collection and subsequent
handling hinders the comparison and interpretation of data from streams having naturally low
levels of that analyte. Interlaboratory bias also can confound statistical comparisons of data,
because true differences may not be distinguishable from differences resulting from systematic
measurement errors at the different laboratories.
4.2 COMPLETENESS
Of the 504 stream reaches initially selected for sampling, 500 (99%) were actually visited in
the field. Four reaches were not sampled because of access permission difficulties or physical
inaccessibility (two in the Southern Appalachians and two in Florida). Water samples were not
collected from 39 upstream reach nodes and 39 downstream reach nodes because they were class-
ified as noninterest sites at the time they were visited (subsections 3.2, 5.2; Hagley et al., 1988).
In the context of the NSS-I statistical design, sampling is accomplished by visiting the stream
and determining its target (interest) status, regardless of whether water samples were collected.
75
-------
Thus, samples were collected from the required number of reaches for each sampling stratum in
all subregions.
Of the 1,651 water samples collected during the NSS-I, 1,613 (97.7%) were analyzed for all
physico-chemical variables. Missing or unacceptable values were replaced during data enhance-
ment (subsection 3.8.4). Of the 31,185 chemical numeric values in the data base, only 34 were
replaced in the enhanced data base (Table 3-8). The NSS-I data base has been shown to be
sufficiently complete (both in terms of spatial representation and chemical completeness) to
provide representative spring baseflow chemical indices to estimate the chemical status of
streams as described in Sections 6 through 10.
4.3 COMPARABILITY
All the NSS-I field crews and laboratory personnel used standardized protocols (Hagley et
al., 1988; Hillman et al., 1987), which maximized internal comparability within given streams or
subregions. Similarly, the use of these standardized methods, combined with the quantitative
results of QC sample analysis (Cougan et al., 1988), facilitated comparison with data from other
studies, such as the National Lake Survey (NLS) (Linthurst et al., 1986; Landers et al., 1987),
and detailed process oriented studies on NSS-I special interest site streams.
4.4 REPRESENTATIVENESS
Representativeness can be viewed in a hierarchical manner from specific analyses to the
gereral representativeness of the group of sample streams. At the lowest level (analytical),
representativeness refers to how well the chemical and physical analyses reflect chemical
conditions in the stream at the instant of sampling. Analytical representativeness is largely
defined by how well the analytical measurements reflect chemical conditions in the sample of
water taken from the stream. This aspect of data quality is discussed under the subheadings of
accuracy, bias, and precision (Sections 4.6, 4.7, and 4.8). An additional aspect of analytical
representativeness is the degree to which the chosen set of chemical analyses describes the
water in question. To assess acid deposition effects, we must be concerned primarily with major
constituents of water chemistry that affect geochemical interpretation and toxicity to aquatic
biota at low pH. Representativeness with regard to geochemical interpretation can be evaluated
by examining the completeness of the analyses, that is, how well the anions and cations balance
(subsection 4.9.1), how well the calculated conductivity agrees with that actually measured (sub-
section 4.9.2), and how well the calculated carbonate alkalinity agrees with that measured by
Gran titration (subsection 4.9.3).
At a higher level (system level), the concept of representativeness refers to how well a
water sample characterizes a specific reach of a particular stream. Representativeness in this
context is influenced by the location of the sampling site within the reach, the specific location
(or microhabitat) from which the sample was collected, and the local conditions at the time of
collection. Standardized collection protocols directed sampling crews to collect the samples as
near as possible to the upper and lower nodes of a reach, while avoiding areas influenced by
tributary inflow or estuaries. Samples were collected in mid-stream from areas of flowing water
to maximize sample representativeness. Sampling crews were instructed not to collect samples
during precipitation events. Water samples are intended, therefore, to be representative of
spring season base flow.
At the population level, representativeness refers to whether the NSS-I sample reaches
were representative of the target population as a whole. The probability sampling design
76
-------
employed by the NSS-I minimizes the likelihood of substantial bias that might result from
seriously undersampling any particular geographic area or class of streams. The temporal and
spatial aspects of the NSS-I sampling design are discussed in Section 2; the design and its
implementation are believed to have produced data representative of spring season baseflow
conditions in the target population.
4.5 DETECTABILITY
4.5.1 Method Level Detectabilitv
Indices of detectability were calculated for the NSS-I at both the method level and the
system level. Estimates of method level detectability were calculated from the results of cali-
bration or reagent blank analyses. Limits of detection were estimated following the approach
advocated by the American Chemical Society (Keith et al., 1983), based on the analyses of blank
samples. The limit of detection was estimated as three times the standard deviation of blank
sample measurements.
The NSS-I QA report (Cougan et al., 1988) evaluates method level detectability at each
laboratory. Detectability for all variables except total aluminum and silica measured at one
analytical laboratory were within the DQOs. Appendix A (Table A-l) contains the DQOs for
each analyte. For reference, Appendix A (Table A-2) presents data from laboratory blank sample
analyses and calculated method detection limits for all NSS-I chemical variables. Method level
detectability primarily evaluates the degree to which analytical methods limit the overall ability
of the sampling analysis system to detect analytes. System level detectablility includes the
effects of data collection and processing, in addition to the limitation of the analytical methods,
and thus is of greater interpretive value in relation to the distribution of chemistry in the
regional stream populations.
4.5.2 System Level Detectabilitv
Background levels of analyte added during the collection or handling of samples were
estimated by computing a system decision limit (SDL). The SDL was estimated as the 95*^
percentile of the distribution of field blank measurements. The SDL represents the lowest
measured quantity of analyte that can be distinguished (with 95% confidence) from the dis-
tribution of field blank measurements.
DQOs were not defined for system level detectability. SDLs (Table 4-1) were initially
compared to method level decision limits to determine if the handling and processing of field
blanks had resulted in the addition of analyte to the blank samples. SDLs for almost all NSS-I
chemical variables were less than three times the method level limits of detection (Appendix A,
Table A-2), indicating that background contamination was not a serious problem. Problems are
not apparent with either detection or contamination in the analysis of the major cations (Ca2+,
Mg2+, Na+, K+) and anions (SO42", NO3', Cl"). The SDLs (95th percentile of field blanks) for
these ions were all < 1 /jeq L"1 (Table 4-1). SDLs and median field blank values for all NSS-I
chemical variables are listed in Appendix A (Table A-3).
For extractable aluminum, DOC, and nitrate, SDLs were between three and four times the
method limit of detection. In the case of extractable aluminum, the SDL (0.30 pM) is near the
DQO for detectability (0.19 fiM), and no data quality problem is indicated. For DOC, the back-
ground level (SDL = 0.45 mg L"1) is low enough not to seriously impact data interpretation.
77
_
-------
Table 4-1. Estimates of SDLs for the Primary NSS-I Chemical Variables Based on Analyses
of Field Blank Samples Pooled Across Laboratories.
Variable
n
Median Field
Blank Concentration
System Decision*
Limit (SDL)
Aluminum (pM)
Total
Extractable
61
61
0.37
0.04
1.0
0.30
Monomeric Aluminum (/zM)
Total
Nonexchangeable
Ca2+ (Meq L'1)
Cl- (/zeq L'1)
Conductivity (pS cm"1)
DOC (mg L"1)
K+(/;eq L"1)
Mg2+ (peq L-1)
Na+ (/xeq L'1)
NOS- (/wq L'1)
SO42- (peq L'1)
61
61
62
63
63
63
62
62
62
62
63
0.37
0.52
<0.50
<0.28
0.98
0.17
<0.26
<0.82
<0.44
0.08
0.21
0.63
0.93
1.0
0.85
1.2
0.45
0.41
< 0.82
0.52
0.56
1.0
*The system decision limit was estimated as the 95th percentile of field blank measurements.
Data for all NSS-I chemical variables is given in Appendix A (Table A-3).
78
-------
The SDL for nitrate (0.56 jzeq L"1) was apparently affected by occasional contamination at the
processing laboratory (Cougan et al., 1988). However the SDL is still very low and should not
seriously affect data interpretation.
System level decision limits should be considered in the interpretation of chemical distri-
butions in NSS-I regional stream populations. Population estimates of stream resources with
concentrations at or below the SDL should be interpreted with caution. There can be little
confidence that values reported at these very low analyte levels are significantly different from
zero. It is likely, in fact, that variations in the distributions of observed values less than the
SDL are artifacts of sample collection, handling, and analysis. For the same reason, groups of
streams characterized by analyte levels below system decision limits should not be compared.
4.6 ACCURACY
4.6.1 Accuracy Within Laboratories
Accuracy within each of the laboratories was evaluated using the results from performance
audit samples. Although not certified standard solutions, the performance audit samples provided
a good means of understanding possible systematic errors. Data should not be corrected on the
basis of the accuracy estimates presented here, but data interpretation should be conducted
considering the potential for systematic errors in cases where estimates of accuracy are poor.
Accuracy was estimated as percent accuracy for all analytes except pH by:
Accuracy = [(x - R)/R] x 100
where x equals the mean of the measured values in an audit sample, and R equals the theoreti-
cal value, or an index value based on measured values at the laboratory where audit samples
were prepared.
For pH, accuracy was expressed as the difference between the measured value and the
index value. Negative estimates of accuracy imply that measured values were less than index
values; positive estimates of accuracy imply that measured values were greater than index values.
Within-laboratory accuracy estimates for all chemical variables are detailed in Cougan et al.
(1988), and are listed for each analytical laboratory in Appendix A (Tables A-4 and A-5).
Accuracy estimates for specific conductance and calcium measured at laboratory #1 were not
within the stated DQOs but the observed differences from the theoretical values were small
(approximately 2 /zS cm"1 for conductivity, 1.5 ^eq L"1 for calcium). At the other analytical
laboratory (#2), the accuracy estimates for DOC were not within the stated DQOs. The
observed difference from the theoretical value for DOC was 0.3 mg L"1 and 0.1 mg L'1 in the
two audit lots (Appendix A, Table A-5). Accuracy estimates were within the stated DQOs for all
of the other primary NSS-I analytes.
4.6.2 Pooled Accuracy Estimates
Estimates of percent accuracy (pooled across lot and laboratory) for the primary NSS-I
chemical variables (Table 4-2) showed that accuracy was good for the major cations (Ca2+,
Mg2+, Na+, K+) and anions (SO42~, NO3", Cl~). All of these ions had pooled percent accuracies
less than 5%. All the synthetic audit samples for the aluminum species had significant devia-
tions from the theoretical value. The theoretical value (0.74 /iM), however, is close to the
79
-------
Table 4-2. Percent Accuracy Estimates for the Primary NSS-I Chemical Variables Based on
Synthetic Audit Samples Pooled Across Analytical Laboratories
Variable
Theoretical*
Value
Measured Audit Value
Data
Standard Percent Quality
Mean Deviation n Accuracy Objective
Aluminum
Total - Lot 14
Total - Lot 15
Extractable - Lot 14
Extractable - Lot 15
ANC Ozeq L'1)
Ca2+ (peq L'1)
Cr (peq L"1)
Spec. Cond.-Lab (/zS cm"1)
DIC-closed sys. (mg L"1)
DOC (mg L-1)
K+ faeq L-1)
Mg2+ (/zeq L-1)
Na+ (peq L'1)
NO3- (/zeq L'1)
L'1)
0.74
0.74
0.74
0.74
1.26
1.11
0.889
0.556
106 *
9.68
9.67
17.5
0.959
1.00
5.19
36.8
120
7.53
47.5
114
9.48
9.31
18.5
1.40
1.1
5.12
35.4
119
7.56
46.8
0.823
0.267
0.393
0.074
8.75
1.55
0.818
1.58
0.046
0.27
0.20
2.63
4.31
0.645
2.73
19
31
9
26
56
56
56
56
14
56
56
56
56
55
56
70.3
50.0
20.1
-24.9
7.55
-2.07
-3.72
5.71
46.0
10.0
-1.35
-3.80
-0.83
0.40
-1.47
10,20a
10,20a
10,20a
10,20a
10
10
10
5
10
5,10b
10
10
10
10
10
Accuracy evaluated
as pH difference
pH (air equilibrated)
pH (closed system)
7.25$
7.25$
7.23
6.94
0.179
0.085
56
14
-0.02
-0.31
0.1
0.1
theoretical value is the expected value of the synthetic audit sample assuming no
preparation error and no external effect.
$ The theoretical value was determined by measurements made at the support laboratory that
supplied the audit samples.
a Lower limit represents the accuracy DQO for [Al] > 0.37 pM while the upper limit is for [Al]
< 0.37 uM.
b Lower limit represents the accuracy DQO for DOC > 5.0 mg L"1 while the upper limit is for
DOC < 5.0 mg L'1.
80
-------
SDL for these species (0.3 - 1.0 /xM). The amount of aluminum in the audit samples, therefore,
is approximately the same concentration as that just discernable from blank values (with 95%
confidence). Also, synthetic audit samples sent to the processing laboratory were subject to loss
of aluminum due to adsorption or precipitation between the time of preparation and the" time the
sample was filtered and preserved (Cougan et al., 1988).
There was also a large deviation from the theoretical value for audit sample measurements
of closed-system pH (-0.31 pH units) and DIG (0.441 mg L'1; Table 4-2). It is likely that the
DIG overestimation in the closed-system sample resulted in underestimation of closed-system pH
audit values. The synthetic audit samples were very sensitive to changes in CO2 concentration,
indicating that the pH and DIG values would not necessarily be stable in the synthetic audit
samples. Data from circumneutral Bagley Lake and acidic Big Moose Lake indicated no serious
inaccuracy problem with pH measurements at either the analytical laboratories or the processing
laboratory. Thus, the most probable source of the pH deviation is the introduction of CO2 into
the audit samples between measurement in the support laboratory and measurement in the proc-
essing laboratory. This explanation agrees with the observation that there was almost no pH
deviation (-0.02 pH unit) in the air-equilibrated pH measurement (Table 4-2). Furthermore, the
relationship between field site pH measurements and closed-system processing laboratory pH in
the NSS-I data base was very good: r2 = 0.991, n = 1290; the mean pH difference (processing
lab pH - field pH) was -0.047; 71.3% of the pH differences were within 0.1 pH units and 93.6%
were within 0.2 pH units. This comparison indicates that transport conditions and processing did
not cause a large pH shift. The NSS-I method of using sealed syringes appears to have
succeeded in preventing pH change due to CO2 degassing. The failure of the support laboratory
to use syringe sampling and transporting containers for the audit samples, as were used in the
NSS-I, makes it likely that inaccuracies observed in the audit samples are not applicable to the
routine NSS-I samples.
4.7 LABORATORY BIAS
Seven of the eight different groups of audit sample types analyzed during NSS-I were used
to test interlaboratory bias. These audit samples included two lots of low concentration
synthetic audit samples and two types of audit samples from natural lake systems (Bagley Lake
in Washington and Big Moose Lake in New York). Each lot or type of audit sample consisted of
both field and laboratory samples (Figure 3-3). One of the low synthetic audits was analyzed at
only one of the contract laboratories, thus it could not be used for evaluating interlaboratory
bias.
Edland et al. (1987) have developed model-based approaches, based on the results of NSS-I
audit sample analyses, to evaluate and possibly correct for biases among the laboratories. The
authors of that report, however, do not recommend indiscriminate application of their techniques
to the NSS-I data. Any correction applied to the NSS-I data, based upon audit sample analyses,
will have an unknown effect on data quality because there is no information on bias over the
entire range of analyte levels in the NSS-I data. Because accuracy estimates for most variables
were within or near their DQOs (Cougan et al., 1988), interlaboratory differences should not
seriously affect most data interpretation activities. In cases where accuracy estimates were
outside the DQOs, the potential bias was most important at low concentrations, and thus may be
eclipsed by problems of analytical precision at concentration ranges close to the detection limit
for certain variables (e.g., total aluminum and extractable aluminum). The estimates of among-
batch precision presented in Section 4.8 include the effects of interlaboratory bias.
81
-------
4.8 PRECISION
The precision associated with various components of the NSS-I collection and measurement
system was evaluated using the analytical results of performance audit samples, analytical and
processing laboratory duplicate pairs, and field duplicate sample pairs (Cougan et al., 1988). No
QC sample type provided all the information necessary to calculate a completely satisfactory
estimate of the total measurement error associated with the collection, processing, and analysis
of samples during the NSS-I.
Analysis of laboratory duplicate sample pairs was used to estimate the measurement pre-
cision of a particular analytical methodology within a sample batch (operationally defined as
method level precision). Analysis of field duplicate pairs provided an estimate of overall
precision within a batch, including the effects of sample collection and processing (operationally
defined as system level precision). Performance audit samples (both synthetic and natural)
provided estimates of among-batch precision within a laboratory, and when pooled across the
two laboratories, this included the effects of inter laboratory bias (operationally defined as
among-batch precision). A batch refers to a batch of samples grouped together at the proces-
sing laboratory and kept together throughout the analytical process with the same batch ID.
Because of the dilute nature of the waters sampled during the NSS-I, a single estimate of
pooled precision for the entire range of analyte values may be misleading because of the effects
of concentration on precision. Model based approaches to estimating precision as a function of
concentration (e.g., Mericas and Schonbrod, 1987) were not used because the range of concentra-
tions represented by duplicate sample pairs was not evenly distributed over the entire range of
measured values, and models may be unduly influenced by outlying measurements.
To minimize the effect of concentration on precision estimates, and to provide data users
with estimates of precision relevant to particular ranges of interest, pooled precision estimates
were calculated for subset ranges over the total range in analyte concentration. For some
variables, the range subsets were defined in terms of biologically or chemically relevant limits,
while for other variables the ranges are arbitrary. Scatterplots of the standard deviations of
sample pairs versus mean concentration of a pair indicated that there were no trends in the
relationship between concentration and variance within each range subset,
Data from both analytical laboratories were combined before calculating pooled precision
estimates. A total of 68 laboratory duplicate pairs and 65 field duplicate pairs were analyzed
for most analytes. Field or laboratory duplicate pairs having values identified as unacceptable
during the verification process were not included in calculating precision estimates. For each
range subset, precision was estimated as a pooled standard deviation based on the means and
variances of the sample pairs included in that subset. The individual variances of each sample
pair were summed to calculate a pooled variance, and a pooled standard deviation was calculated
as the square root of the pooled variance divided by the sample size. The percent relative
standard deviation (%RSD) was then calculated as:
pooled standard deviation
%RSD = * 100
pooled mean.
Method level estimates of precision were compared to the DQOs. For all variables, pooled
method level precision estimates were within the DQOs for all range subsets greater than the
SDL (Appendix A, Table A-6).
82
-------
Pooled (across laboratory) system level and among-batch precision estimates and associated
summary statistics for the primary NSS-I chemical analytes are presented in Table 4-3. Data for
all NSS-I analytes is given in Appendix A (Table A-7). System level precision was estimated for
all field duplicate measurements. The distribution of field duplicate pairs among concentration
range subsets was not uniform because duplicate sample pairs were not purposefully selected to
cover the concentration range of samples. For the among-batch estimates, when more than one
audit sample type was present within a concentration subset, the sample type that had the
largest %RSD is presented (Table 4-3).
At concentrations above the SDL, the %RSDs for the system level precision of the major
cations and anions were all less than 5%, except for the lowest subranges where they were all
less than 10% (Table 4-3). The %RSD tended to be higher in the lowest concentration range
subset. At concentrations below the SDL, precision was usually poor (%RSD = 6% - 56%), as
would be expected for samples at the detection limit. Values of %RSD were not calculated for
pH and ANC because ANC can have a value of zero and pH values are logarithmic. Instead,
pooled standard deviation values were used to evaluate precision. ANC had a system level
precision of < 5 /ueq L"1 for values less than zero and a precision of 6-8 /zeq L'1 for values
between 0 and 200 /*eq L"1. The system level pooled standard deviations for pH values below
6.0 were less than 0.1 pH unit, whereas at pH values greater than 6.0, they were less than 0.2
pH unit. Aluminum species and DOC had rather high %RSDs (15% - 36%) in concentration sub-
sets just above their SDLs, indicating a lack of precision at these low concentrations. Total
aluminum had %RSD greater than 22% in all concentration subsets; population estimates for total
aluminum should be interpreted with this in mind.
Among-batch precision was estimated for each of the three different types of audit samples
by pooling the measurements from both analytical laboratories. The among-batch precision was
expressed as the standard deviation of the pooled measurements. Among-batch precision within
each laboratory is discussed more fully in Cougan et al. (1988). In almost all concentration
range subsets, for all variables, the among-batch variance estimates (pooled standard deviations)
were larger than the corresponding system level (within batch) precision estimates. This finding
suggests that variation among batches within a laboratory, or among laboratories, contributed
more to overall measurement error than sample collection or processing. These differences
result from differences in calibrations from batch to batch (day to day), either within a
laboratory or among laboratories. Aluminum species and DOC had relatively high among-batch
%RSD (> 20%) in the lower concentration range subsets (Table 4-3). The poor among-batch
precision for DOC and aluminum is in a large part due to their poor within-batch precision.
Above the SDL, all of the among-batch %RSDs for the major cations and anions were less
than 10% (Table 4-3), except for the lowest concentration range subset for calcium, chloride,
conductivity, and closed system DIG (%RSD = 10% to 16%). A small interlaboratory bias (or
between-batch bias within each laboratory) may exist at low concentrations for these analytes
(1-2 /*eq L'1 for Ca2+ and Cl~; 2 /zS cm'1 for conductivity; 0.06 mg L'1 for DIG). In general,
the magnitude of interlaboratory bias does not appear to be an important problem in the NSS-I
data base. For interpreting the NSS-I data, the most conservative estimate of precision (among-
batch precision) should be used as the best estimate of measurement error, because it is the
largest variance and it takes into account biases between laboratories.
83
_
-------
Table 4-3. System Level and Among-Batch Precision Estimates for Primary NSS-I Chemical
Variables
System Level Precision*
Variable
Total
Aluminum
(pM)
Extract.
aluminum
G*M)
Total
monomeric
Alum. (/*M)
Nonexchang.
monomeric
Alum. (/*M)
ANC
0*eq L'1)
Ca2*
C*eq L'1)
cr
(Meq L-1)
Concentration Number Pooled
Range Subset of pairs SD
< 1.0 (SDL)
1.0 to 5.0
5.0 to 10.0
> 10.0
< 0.30 (SDL)
0.30 to 2.0
> 2.0
< 0.63 (SDL)
0.63 to 2.0
>2.0
< 0.93 (SDL)
>0.93
<0
0 to 50
50 to 200
>200
<50
50 to 250
250 to 500
> 500
< 30
30 to 55
55 to 150 .
150 to 275
> 275
5
26
14
20
34
20
11
35
19
11
51
14
7
9
20
29
4
35
19
7
8
18
24
6
9
0.400
0.900
1.38
5.75
0.0590
0.237
0.523
0.0801
0.143
0.178
0.124
0.263
2.94
7.83
6.10
13.7
0.749
2.84
4.49
80.0
0.508
0.931
1.21
2.06
20.7
%RSD
56.4
29.4
22.0
31.5
43.5
35.7
6.9
19.6
16.9
2.5
20.1
14.9
__
—
—
—
2.7
2.2
1.3
5.6
2.6
2.2
1.4
1.0
4.0
Among-Batch Precision*
Number of
samples SD
38
50
—
37
38
35
37
13
__
24
13
24
39
—
38
—
56
39
__
—
39
—
__
__
—
1.48
0.550
__
0.945
0.126
0.209
1.22
0.271
— _
0.434
0.311
0.471
4.88
—
11.73
1.55
3.84
__
—
1.44
__
__
__
—
%RSD
148
47.0
9.3
48.6
32.6
15.3
73.2
6.0
55.9
21.9
__
16.4
4.1
11.9
Sample
type*
BL
S
BM
BL
S
BM
S
BM
S
BM
BM
BL
S
BM
BM
SDL ^ System decision limit.
%RSD = Percent relative standard deviation.
* Overall within-batch precision calculated from field routine duplicate sample pairs.
&Among-batch precision calculated from performance audit samples.
* Audit sample types from which precision estimate was derived. S = Synthetic audit sample;
BL « Bagley Lake natural audit sample; BM « Big Moose Lake natural audit sample.
84
-------
Table 4-3. System Level and Among-Batch Precision Estimates for Primary NSS-I Chemical
Variables (Continued)
System Level Precision*
Concentration Number Pooled
Variable Range Subset of pairs SD
Spec. Cond
Lab Value
GuS cm'1)
DIG
Closed
System
(mg L'1)
DOC
(mg L-1)
K+
(Meq L'1)
Mg2+
(Meq L'1)
Na+
(Meq L'1)
< 25
25 to 50
50 to 100
> 100
< 1.0
1.0 to 2.0
2.0 to 5.0
5.0 to 10.0
> 10.0
< 0.45 (SDL)
0.45 to 2.0
2.0 to 5.0
5.0 to 10.0
< 10
10 to 25
25 to 50
> 50
< 80
80 to 160
160 to 400
> 400
< 20
20 to 42
42 to 90
90 to 225
> 225
6
24
24
11
7
9
31
13
5
2
41
15
7
8
20
28
9
19
23
18
5
4
4
24
25
8
0.87
0.27
0.42
1.22
0.020
0.063
0.092
0.184
0.365
0.02
0.24
0.44
0.58
0.0824
0.346
0.773
3.60
1.07
1.15
4.77
19.9
0.0435
1.30
1.22
1.61
4.22
Among-Batch Precision*
Number of Sample
%RSD samples SD %RSD type*
4.2
0.7
0.6
0.6
4.3
4.4
2.7
2.8
1.5
6.1
20.8
14.3
8.1
1.3
2.2
2.2
4.4
2.0
0.9
1.9
2.6
0.5
3.6
1.9
1.1
1.0
38
39
—
— —
27
27
—
—
--
39
56
39
—
39
39
—
—
56
—
—
—
—
39
—
56
—
1.91 14.6
0.89 3.6
—
— —
0.062 11.7
0.061 3.6
—
—
—
0.42 105
0.27 24.5
0.28 7.2
—
0.409 5.3
0.563 5.1
—
—
2.63 7.4
—
—
—
—
1.83 4.9
4.31 3.6
—
BL
BM
BM
BL
BL
S
BM
BL
BM
S
BL
S
SDL = System decision limit.
%RSD = Percent relative standard deviation.
* Overall within-batch precision calculated from field routine duplicate sample pairs,
&Among-batch precision calculated from performance audit samples.
$ Audit sample types from which precision estimate was derived. S = Synthetic audit sample;
BL = Bagley Lake natural audit sample; BM = Big Moose Lake natural audit sample.
85
-------
Table 4-3. System Level and Among-Batch Precision Estimates for Primary NSS-I Chemical
Variables (Continued)
System Level Precision*
Among-Batch Precision*5
Concentration Number Pooled Number of Sample
Variable Range Subset of pairs SD %RSD samples SD %RSD type$
NOs'
(peq L'1)
< 0.56 (SDL)
0.56 to 50
> 50
8
45
12
0.0660
0.739
5.22
24.0
5.7
2.2
38
55
—
0.305
0.645
—
158
8.5
BL
S
PH
Proc. Lab
Closed
System
4.00 to 5.00
5.00 to 6.00
6.00 to 7.00
7.00 to 8.00
>8.00
pH 4.00 to 5.00
Anal. Lab 5.00 to 6.00
Air 6.00 to 7.00
Equil. 7.00 to 8.00
> 8.00
6
13
24
19
3
6
4
11
40
4
0.032
0.036
0.036
0.036
0.026
0.017
0.041
0.113
0.171
0.083
27
14
27
0.059
0.085
0.049
39 0.071
56 0.179
BM
S
BL
BM
S
so/-
(peq L"1)
< 50
50 to 100
100 to 275
> 275
18
10
24
13
1.10
1.69
2.66
5.06
3.6
2.1
1.5
1.1
56
__
39
—
2.73
— _
3.48
—
5.8
2.6
S
BM
SDL s System decision limit.
%RSD - Percent relative standard deviation.
* Overall within-batch precision calculated from field routine duplicate sample pairs.
&Among-batch precision calculated from performance audit samples.
* Audit sample types from which precision estimate was derived. S = Synthetic audit sample;
BL = Bagley Lake natural audit sample; BM = Big Moose Lake natural audit sample.
86
-------
4.9 CHEMICAL DATA QUALITY OVERVIEW
Another way to test the quality of the NSS-I chemical data is to examine the ionic rela-
tionships between the analytes in the data base. By examining the charge balance between
anions and cations, measured versus calculated conductivity, and measured ANC versus calculated
carbonate alkalinity (from DIG and pH), one can get an overall impression of the quality of the
chemical data. The enhanced data set (data set 4, subsection 3.8.4) was used to make these
comparisons, because it is the data set from which all of the results presented in this report are
derived.
4.9.1 Charge Balance
Because of the electroneutrality constraint, the sum of positively charged ions (i.e., cations)
in water must equal the sum of those with a negative charge (i.e., anions). Major discrepancies
between the sum of measured anions and cations thus reflect analytical errors, failure to
measure all ions with significant concentrations, or a combination of both. Although charge
balances alone cannot identify the cause of a charge imbalance, they can serve as a QC check
on the completeness and accuracy of the ion chemistry data.
To assess the quality of the data for the ionic species in water, ion charge balances,
involving a comparison of the sum of cations and anions, were determined for all stream samples
with chemical data. The sum of the cations is defined as the equivalent sum of Ca2+ + Mg2+ +
Na+ + K* + NH4+ + H+, and the sum of anions is SO42- + NO3" + Cl' + F~ + (ANC + H+). ANC
+ H+ includes the carbonate protolytes (HCO3' + CO32'), hydroxide (OH'), plus any noncarbonate
protolytes (e.g., weak organic anions, metal oxides) that react with hydrogen ions during the
titration of ANC.
A charge balance plot for all NSS-I Mid-Atlantic and Southeast Screening stream samples is
shown in Figure 4-1. Some deviation from the one-to-one line (i.e., y = x) is expected because
the analytical errors associated with the measurement of the individual anions and cations are
additive in the calculation of the sum of cations and anions. Although random analytical errors
would tend to cancel in calculating the sum of anions or cations, the analytical accuracy and
precision differ for each of the ions measured. Hence some charge imbalance may occur because
of the differences in the analytical precision and" accuracy between methods.
In most cases, the discrepancies in the charge balance are relatively small and are within
the combined limits of precision of the analytical methods used to measure the respective ions.
Percent ion balances ([cation sum - anion sum]/[cation sum + anion sum] * 100) were calculated
for all of the NSS-I routine stream samples. Of the 1,342 samples, 86.4% had ion balance differ-
ences less than 5% and 94.7% had ion balance differences less than 10%. The charge balance
plot, however, does show some bias, with more of the samples having an anion deficit (i.e., sum
of cations > sum of anions) than a cation deficit. This bias suggests a small systematic ana-
lytical error (e.g., an underestimate of one or more anions or an overestimate of cations),
failure to include one or more anions in the charge balance, or a combination of both.
Although the calculated charge balances do not include all ions that could potentially con-
tribute to the sum of the cations and anions, those included generally contribute most of the
total ionic strength in freshwater environments. Inorganic ions not included, such as phosphorus
and trace metals, are generally present in low concentrations in most streams, and hence are
not significant contributors to the total ionic concentration. Silica is not included in the charge
balance, but this should not make a serious difference because in most natural waters, silica
exists predominantly as o-silicic acid (H4Si(OH)4; Stumm and Morgan, 1981) and is not charged.
87
-------
2000-f
cr
-------
Aluminum, which becomes more soluble with decreasing pH, may be a major contributor to
the cation sum in acidic environments. However, failure to include aluminum in the charge
balance would most likely cause a cation deficit rather than an anion deficit because cationic
species of aluminum [A13+, A1(OH)2+, A1(OH)2+] would be favored at the acidic pH where alum-
inum solubility is maximum. Thus, it is unlikely that failure to include silica, phosphorus, and
trace metals in the cation and anion sums is the reason for the observed anion deficits. Alum-
inum, however, may account for some of the observed cation deficits.
Another potential cause of the observed anion deficits is unmeasured strong organic anions
that do not react with hydrogen ions during the ANC titration. This deficit is particularly large
(> 150 /xeq L"1) for some of the streams in Florida and in the Mid-Atlantic Coastal Plain (Figure
4-2). As discussed in Section 8.6, many of the streams in these two subregions have relatively
high concentrations of DOC (median concentration of DOC at the upper and lower node of 8 and
4 mg L"1, and 5 and 4 mg L"1 in Florida and the Mid-Atlantic Coastal Plain; respectively)
compared to those in streams of the other subregions. Naturally occurring organic solutes can
exhibit a wide range of dissociation constants (pKa) and the extent to which they affect the
charge balance of surface waters depends on their concentration in water, the pKa of the
organic acids, and the pH of the water. For example, naturally occurring organic acids that are
strongly acidic (i.e., pKa < 4.0) will behave like strong mineral acids, in that the protons are
completely dissociated, resulting in a depression in the ANC. Because there is no direct measure
of the anions associated with these strong organic acids, they are not included in the anion
sum. Hence, if they are present in significant concentrations, anion deficits would exist.
Weak (nonprotolytic) organic acids (pKa > 5.0) may also influence the ANC and the ionic
strength of water, depending on their concentration and pKa, and on the pH of the stream
water. However, the anions of the weak organic acids that are titrated in the measurement of
ANC would be included in the charge balance as noncarbonate ANC, and hence would not con-
tribute to the observed anion deficits.
To assess whether anions of strong organic acids could account for the anion deficit in the
streams in Florida and the Mid-Atlantic Coastal Plain, the calculated deficit was regressed
against the measured DOC concentration in these two subregions (Figure 4-2). The latter was
used as a surrogate variable for strong organic anions because there was no direct measurement
of strong acid organic anions in NSS-I samples. The results for individual subregion/node
regression showed that variation in the concentration of DOC accounts for a large percentage of
the variation in the anion deficit among streams in Florida (r2 = 0.54 and 0.87) and a smaller
percentage in the Mid-Atlantic Coastal Plain (r2 = 0.33 and 0.03 at the lower and upper nodes,
respectively). This result is consistent with the conclusion that unmeasured organic anions that
behave as strong mineral acids contribute to the observed anion deficit in many streams in these
subregions. The contribution of strong and weak organic acids to the ANC and BNC in streams
is discussed in more detail in Section 8.6.
In higher conductivity waters, not all of the ions that contribute to ionic strength were
included in the calculated anion and cation sums. For example, acid mine drainage waters con-
tain significant amounts of heavy metals (Fe, Mn, Cu, Pb, etc.) that may be responsible for some
of the cation deficits observed in Figure 4-1 at high ionic strength. The exact speciation of
iron, manganese, aluminum, and DOC was not measured in the NSS-I, so exact charge balances
cannot be calculated'using these species. Their equivalent strength, however, can be approxi-
mated. The charge balances were reanalyzed using approximations for iron, manganese, alum-
inum, and DOC. Organic anion equivalents were calculated from DOC using the Oliver model
89
-------
600 -\
cr
0
O
Ld
Q
O
500-
400-
300-
200-
100-
0-
0 10 20 30 40 50 60 70 80 90 100
DISSOLVED ORGANIC CARBON (mg L~ j)
Figure 4-2. Relationship between anion deficit and DOC for all NSS-I stream samples in
Florida (3C) and the Mid-Atlantic Coastal Plain (3B).
90
-------
(Oliver et al., 1983). Iron and manganese equivalents were approximated from total dissolved
iron and manganese measurements using a charge of +2. Aluminum equivalents were calculated
from total aluminum measurements assuming an average charge of 0.87 per total aluminum ion
(Driscoll et al., 1984). The (ANC + H+) term was removed from the anion sum and replaced with
measured values of HCO3', CO32', and OH~ (calculated from pH and DIG measurements), because
the organic component of ANC is now included in the Oliver model approximation.
In the new charge balance plot shown in Figure 4-3, which includes the approximations for
DOC, Fe, Mn, and Al, deviations from the equality line were smaller at anion concentrations less
than 1000 /jeq L"1 than those present in the first charge balance plot (Figure 4-1). There were,
however, larger deviations from equality at anion concentrations greater than 1000 /ieq L"1. The
overall ion balance differences were roughly the same (95% of the samples had ion balance dif-
ferences less than 10%) for both figures. The fact that the revised charge balance was better
in dilute systems indicates that in these systems some of the charge balance discrepancies
observed in Figure 4-1 were caused by not measuring all of the ions that contribute to ionic
strength. It is likely that at high ionic strengths the approximations used in Figure 4-3 are less
valid, causing the observed increase in charge balance deviations.
4.9.2 Calculated vs. Measured Conductance
The presence of ionic species in water increases the electrical conductance of that aquatic
solution. Conductance, therefore, provides an indication of ion concentration. Further, since
the relationship between ion concentration and specific conductance is known for most ionic
species, the measured conductance of water samples can be used as an internal check on both
the accuracy and the completeness of measurement of ionic species by comparing the measured
and expected conductance. The expected conductance is calculated as the sum of the product of
the ionic concentration and the molar conductivity (mho cm"2 mol"1) of each of the measured
ions in water. Because the molar conductivity constants are valid only in dilute (infinite
dilution) solutions, they were corrected for concentration effects using the Debye-Hyckel-
Onsager equation (Atkins, 1978):
Mc - Mc° - (A
BMC°)
where Mc is the corrected molar conductivity, Mc° is the molar conductivity at infinite dilution,
C is concentration, and A and B are constants. In water at 25°C, A equals 60.2 mho cm2
morV(mol L'1)* and B equals 0.229 (mol L'1)*.
A plot of the calculated vs. measured conductances for all NSS-I Mid-Atlantic and
Southeast Screening stream samples is shown in Figure 4-4. Agreement between the expected
(calculated) and observed conductance is excellent. The coefficient of variation (r2) for the
relationship was 0.987. The slope of the regression line was 0.981 (SE = 0.00314) and the y
intercept was 4.87 (SE = 0.522). Differences between the observed and expected conductances
can be attributed to analytical errors in the measurement of the ions or of the conductance (or
both), the inclusion of electrostatically neutral species in the calculated conductance, or the
failure to measure ions that may contribute to the total conductivity.
91
-------
2000 -f
Ul
o 1000
i—
o
b_
O
ID
00
0-
0
1000
SUM OFANIONS (jj,eq L~1)
2000
Figure 4-3. Charge balance plot for all NSS-I Mid-Atlantic and Southeast Screening samples
that include approximated equivalent concentrations for DOC, Fe, Mn, and Al.
The line is y = x.
92
-------
u
00
o
z>
Q
-z.
a
o
Q
UJ
5.00-1
400-
300-
200-
100-
0-
0
100
200
300
400
1
500
CALCULATED CONDUCTIVITY (pS cm ')
Figure 4-4. Measured vs. calculated conductance of NSS-I Mid-Atlantic and Southeast
Screening stream water samples. The line is y = x.
93
-------
4.9.3 Calculated vs Measured ANC
A third method for assessing the quality of the stream survey data involves comparisons of
the calculated carbonate alkalinity with the measured ANC. The carbonate alkalinity is the
protolyte (i.e., reacts with hydrogen ions) portion of the carbonate system only and is defined as
[Alk]c - [HC03-] + 2 [C032-] + [OH'] - [H+].
Carbonate alkalinity ([Alkc]) was calculated from the initial pH and DIG concentration measured
at the analytical laboratory (Hillman et al., 1987). ANC is a measure of the sum of protolytes
for a defined pH titration range that includes the bicarbonate system. Thus [ANC] must be
greater than or equal to [Alkc] because [Alkc] is a portion of the titrated [ANC]. Any
discrepancy between the calculated carbonate alkalinity and the measured ANC must be due to
analytical errors in the measurement of ANC, pH, or DIG, the presence of noncarbonate
protolytes in the water, or a combination of both. Comparisons of the calculated alkalinity and
measured ANC thus can serve as a QC check on the measured pH, DIG, and ANC.
Comparisons of the calculated carbonate alkalinity and the measured ANC for all NSS-I
stream samples are shown in Figure 4-5. In general, the measured ANC is greater than or equal
to the calculated carbonate alkalinity as indicated by the predominance of the data plotting on
or below the one-to-one line. Since the ANC must account for at least the carbonate protolytes
in water, the fact that few of the data are above this line indicates that the pH, DIG, and ANC
data for most of the streams are reliable.
The ANC in a few of the samples is less than the carbonate alkalinity, indicating some
analytical error in the measured ANC and/or in the DIG and pH, both of which were used to
compute the carbonate alkalinity. At most sites, however, the ANC equals or exceeds the
carbonate alkalinity, suggesting the presence of noncarbonate protolytes, a systematic
overestimate of the ANC or underestimate of the carbonate alkalinity, or some combination of
the above. Assuming there is no systematic error in the estimates of ANC and bicarbonate
alkalinity, there appear to be few cases in which the analytical error exceeds the effect on ANC
of noncarbonate protolytes.
Potential noncarbonate protolytes include dissolved and particulate organic carbon,
aluminum and other proton-reacting metal complexes, and particulate aluminosilicates and metal
oxides. The potential contribution of weak organic anions to the ANC of stream water is
discussed in Section 8.6 of this report.
94
-------
600-
CF
0)
UJ
a
co
o
Q
UJ
ID
O
O
500-
400-
300-
200-
100
0
0
T 1 1 1 1 1 1 T
100 200 300
400
500 600
i
MEASURED ANC (>eq L ')
Figure 4-5. Measured ANC vs. the calculated carbonate alkalinity for NSS-I stream water
samples. Data below the line of equality (y = x) indicate the presence of
noncarbonate species that react with hydrogen ions, whereas data above the line
indicate anlaytical error.
95
-------
THIS PAGE INTENTIONALLY LEFT BLANK
96
-------
SECTION 5
TARGET POPULATION—PHYSICAL CHARACTERISTICS
5.1 OVERVIEW
Section 5 describes how the National Stream Survey - Phase I (NSS-I) target population
was refined from the population initially identified from map criteria to the refined target
population whose chemical characteristics are described in Sections 6 through 10. This section
presents the total numbers and the combined length of target stream reaches, and provides
statistics that describe the distribution of physical characteristics of the reaches (e.g., drainage
area, elevation, channel width). Population estimates are included for those categories within
the initial target population not of interest in an assessment of acid deposition effects.
5.2 REFINEMENT OF THE TARGET POPULATION
The set of stream reaches intended for field visitation was selected in two stages, as
discussed in Section 2.4. Site inclusion criteria were applied at the first stage (the examination
of reaches selected by grid points). These criteria excluded lakes, reservoirs, wetlands, and
other nonflowing water selected by map grid points, as well as reaches too large, too urban, or
too greatly affected by ocean tides to be of interest for this survey. Because map-based
criteria may not adequately exclude some types of noninterest streams, further refinement of the
target population was accomplished by means of site or sample exclusion criteria applied in the
field (Section 3.2). These criteria specified that sampling was not to be undertaken during
storm episodes. The criteria also enabled samplers to identify and avoid reaches severely
affected by acid mine and oil field brine drainage, those influenced by ocean tides, those with
no defined channel, and those in which flow was insufficient for clean sampling.
The target population was further refined after field sampling through examination of
stream chemistry data and field comments. Reaches with severe point-source pollution were
identified, along with those with acid mine drainage, tidal, or storm episode effects not readily
apparent during field sampling. Stream reaches where acid mine drainage was suspected, but
where evidence was ambiguous, were revisited the following spring (1987). The analytical cri-
teria used for these after-the-fact refinements of the target population are described in
subection 3.8.3 and the basis for the criteria in subsection 9.3.1. Figure 5-1 illustrates the
process of target population refinement.
During the first stage of NSS-I sample reach selection, all 3,305 sites identified by the grid
point sampling frame were evaluated with respect to inclusion in the target population. This
evaluation according to map criteria was the first attempt to refine the sample population iden-
tified by the grid point sampling frame. From 51 to 76 percent of the stream reaches defined
by the grid points (64 mi2 or 166 km2 per grid point) were identified, depending on the sub-
region, as having characteristics that tentatively placed them in the category of target reaches
(Section 2, Table 2-3). These 2,301 reaches comprised a sample of the initial target population
(selected solely on mapped information); a subsample of 500 of these were visited during the
spring baseflow period of 1986 (spring and summer 1985 for the Pilot Survey in subregion 2As).
As a result of field visits and chemical data analysis, additional stream reaches or sampling
locations (upstream or downstream nodes) were subsequently classified as representative of non-
interest subpopulations (Section 3.2). A total of 445 upper, and 446 lower, reach nodes were
retained as representative of the refined target population. Population estimates of target and
noninterest reaches (Figure 5-2) show that 80% to 100% of the initial target population in all
97
-------
Frame Population
(Expected ANC <400 peq L"1)
Target
Population
Noninterest
Sites
Stage I
Sample
Stage H
Sample
field sampling
exclusions
Extrapolation to
Refined Target
Population
Refined Target
Sample Sites
exclusions after
examining chemical
characteristics
Figure 5-1. NSS-I target population refinement concept.
98
-------
UJ
Q
O
O
UJ
oc
CD
V)
— ;
*j
;X
x"
x*.
•!•*•
:£
lv
X;
X-
|
-J
i
~
-4
I
£
(
I
I
<
<
— J
^
•:•:•
.".*.
•X'
:•:•:
X;
X-
S
;X
*-*-
%
D
/}
2
n
5
•x
/>
3
J
..
:•:•
•x
X*
•:•:
X;
:•:•
•x
x*
:j:-
X*
x':
1
-J
{
(
J
(
~
<
4
..
•X
X;
g
•X
•x
X;
•v
x*
:•:•
X;
:D
£
5
•_
j
3
5
Q.
Q.
t
?f
rar
^
X
Iv
X-
•X
X;
:•:•
j:j:
I-!;
X"
:*'
:§
I*!*
_j
(
i
T
0
1
I
(
(
j.
i —
•
\y
/
V"
S^
x)
|x
•X
•x
*v
X;
xj
jj:
ix
*x
X;
=>
o
3>
3
E
3
3
>-,
J
3
>
•— ^
s
::::
•x
;X
v-
;X
§:
x'
X;
;X
•:•:
_J
i
'i
C
~t
••
c
<
<
C
•4
<
•;
^!
\
•X
X;
ix
r>
5
L
3
J5
3
D
J
I
3
X;
*X
X;
X;
:•:•
x«
X;
jx
;X
-1
<
J
1
0
<
c
c
^=7
¥:
%
^.;
:•:•
X;
;X
X*
S
•x
=J
1>
3D
3
E
D
D
Q
f)
H
5
•:•:
iji:
X;
:>:
v*
:•:•
S
_J
H
(
f
I
C
C
H
j
TX
x|
X*
X;
X;
;X
-*.^
":•:
^
1>
L
^
/
^
r
-6
vl
x"
X*
•:•:
:S
X;
-i
<
j
it
<
j
i
i
<
0
37
/
^
?
^
^
X;
:•:•
;.;•
X;
X;
|:j:
x]
x*
D
o
3
3
3
3
DL
I
0
7-
^
x':
;X
§:
|x
:•:•
•x
X;
_J
t
1
J
<
(
(
«
"»
1
C
J7-
xl
jiij
TX
;X
X*
|:|:
ix
r>
o
3
3
3
•s
n
£.
I
•j
C
7
r
4
/
.
r
4
,
r
/
,
r
^
-A
lv
j:j:
:;:;
_J
c
T
*!
<
D
-p
5
/
j
r
j
^
^
.
r
-6
3
3
3
O
- o
*
_ o
O) ^,
•*•
-0 '1
00 :
Ld "^
O c
_ 0 < <-
f^ h- -^
_ LJ ^
-° 0
to 2 v
UJ >
_o °- <
10 u K
o ~~
- ^. t-
— 1
_0 D
D _c
O *
0 -i
CM £
c
Q
_o \
1^
T:
On
"
«f
a
Q
i— i
8
z
•^ >3
I en
: -g
1 ^E"a5 S
•! a) c t->
5 •»- E fi
?'Eg g
" Q) +J
?-^o ">
— ^ j^zTu j^
-1-12: «
2
™f— nrro O
" ^K 'i
" "jD §
] Jy "§
at
CO
2?
S
•Q
CO
5 fl>
1 (M
'« °
3 D 00
".E CO S
- n O d
?> £ 8
5 co— »
- c1*- Q. .
• — C in
3^1—1 ^O C
:T3 D g 'ob
:
-------
subregions except Florida was retained in the refined target population for which population
estimates are given in Section 6.
In Florida, 61% of upstream and 54% of downstream reach ends were retained. The major
cause for exclusion of reaches after field visitation in Florida and most other subregions was
insufficient flow or imperceptible flow with no defined channel. Drought conditions in parts of
the survey area during 1986 may have caused streams that usually flow during the spring to run
dry. Another major reason for excluding streams was the extremely high conductivity or low pH
resulting from accelerated weathering of alkaline or acid generating minerals exposed by mining
activities. Chemical data from eight visits to five separate stream reaches were identified
after-the-fact as having been influenced by storm episodes (subsection 3.8.3.1); however, substi-
tute data from alternate site visits were available to provide baseflow chemistry at those sites.
No population estimates of the chemical impacts of episodes can be provided, as the project
design specified that index chemical samples from the target population exclude storm episodes.
Substantial portions (32% and 38%) of upstream and downstream ends of reaches in the
Florida subregion (3C) were excluded from the refined target population because of insufficient
depth or flow velocity for sampling during spring 1986. An estimated 15% of upstream and
downstream reach ends in the Southern Appalachians (2X) were dry or too shallow for sampling.
Only 3% to 6% of reaches in the Piedmont (3A), Ozark/Ouachita (2D), and Valley and Ridge
(2Bn) subregions were similarly excluded.
An estimated 8% of the upstream and 10% of the downstream ends of the initial target
reach population in the Northern Appalachian subregion (2Cn) were acidic and were affected by
acid mine drainage. The rationale for this classification is discussed in subsection 9.3.1.
Smaller proportions (2% or less) of the initial target population were excluded from the refined
target population because of acid mine drainage in the Valley and Ridge (2Bn), the Southern
Appalachians (2X), and the Poconos/Catskills (ID). It has been estimated that 4,594 km, or 2%,
of the combined length of stream reaches in the initial target population of the combined NSS-I
subregions were acidic at either node because of mine drainage (Section 9.4).
Seven to ten percent of upstream and downstream reach ends in subregions ID and 2Bn had
temperature corrected field conductivities higher than 500 pS cm"1 but were not acidic. Of the
sample streams in this category for which chemistry was measured, most had acid neutralizing
capacity (ANC) > 1000 /*eq L"1. These cases may have occurred because of oil field brine
drainage or strip mining in geologic formations containing carbonate minerals.
In the Mid-Atlantic Coastal Plain (3B), approximately 7% of upper and lower reach ends in
the initial target population were eliminated because of tidal influence. Other stream reaches
were excluded from the refined target population because sampling points represented wetland
conditions where no defined channel could be identified. In still other cases, stream channels
were inundated by impoundments, or there was no evidence that water had ever flowed in the
sample location.
5.3 TOTAL TARGET POPULATION RESOURCE ESTIMATES
Estimates of the total number of upper and lower reach ends and the combined length of
reaches in the refined target population are presented in Table 5-1. Both the target population
chemical distributions based on reach numbers described in Section 6, and those based on num-
ber and length presented in Volume II of this report use these total resource estimates.
Length-based population estimates employing interpolation of upstream and downstream reach
chemistry (Section 6) are based upon that portion of the total stream length that includes both
the upstream and downstream reach ends in the refined target population.
100
-------
The NSS-I sample reaches are representative of a total population of 55,918 reaches for
which both the upstream and downstream nodes are included in the refined target population
(Table 5-1). These reaches have a combined length of 200,652 km. The four Mid-Atlantic sub-
regions contain 36,045 of these reaches (109,865 km), whereas the five Southeast subregions
contain 19,873 reaches with a combined length of 90,787 km.
5.4 TARGET REACH PHYSICAL CHARACTERISTICS
Population frequency distributions of target reach drainage areas and Strahler Order
(Strahler, 1957) (Figures 5-3a and 5-3b) confirm that NSS-I reaches comprise a population of
streams best described as small to mid-sized. At the upstream ends, 46% of these reaches are
classified as headwater reaches (Strahler Order = 1) on l:24,000-scale maps. At the downstream
ends, most reaches (67%) are of Strahler Order 2 and 3. Streams managed by fishery agencies
are typically of Strahler Order 1 to 4 (TIE, 1981; Higbee, 1986; Johnson, 1986). Median drainage
areas at the upstream ends of the target reaches range from 0.44 to 5.0 km2 in the nine NSS-I
subregions, with lower 20th percentiles between 0.10 and 1.1 km2. Downstream medians range
from 5 to 16 km2 with 80th percentile values between 19 and 55 km2.
Stream width and depth distributions (Figures 5-4a and 5-4b) further support the descrip-
tion of NSS-I target streams as small to mid-size, which is the size range of interest for
management of sport fisheries. The nine subregion median widths range from 1.6 to 3.0 m at
the upper reach nodes, and from 2.4 to 4.8 m at the lower nodes. Depending on the subregion,
at least 20% of the reaches are narrower than 0.9 m at their upper ends and 2.7 m at the
downstream ends. Fishery management agencies (e.g., Pennsylvania Fish Commission) typically
manage streams between 4 and 20 m in width (Higbee, 1986; Johnson, 1986).
Median depths of NSS-I target stream reaches are between 0.09 and 0.27 m at the upstream
ends, and between 0.16 and 0.31 m at the downstream ends. Figure 5-4a illustrates that,
depending on the subregion, at least 20% of the target stream reaches are shallower than 0.12 m
at their upstream node. Similarly, depending on the subregion, no more than 20% of the stream
reaches are deeper than 0.76 m at the lower node. The shallowest upstream and downstream
medians are in subregions such as the Northern Appalachians (2Cn), where many target streams
are of relatively high gradient. The median gradient is 3% in subregion 2Cn, for example.
Conversely, the deepest upstream medians are in the Florida (3C) and Mid-Atlantic Coastal Plain
(3B) subregions, where median gradients are considerably less than 0.5%. Subregions with the
greatest median depths at the downstream ends of their target reaches are in the low-elevation
subregions (e.g., 3C and 3B) and also in the Pocono/Catskill (ID) and Ozark/Ouachita (2D)
subregions. These latter two subregions have particularly large median downstream node
drainage areas when compared with the other subregions (Figure 5-3a).
Population frequency distributions of target reach elevation (Figure 5-5a) show considerably
greater variation among and within subregions than between upstream and downstream ends of
reaches. This result reflects the broad range of target reach drainage areas and the broad
geographical extent of the NSS-I subregions (compared with smaller differences between up-
stream and downstream drainage areas on single reaches). The 20th to 80th percentile ranges
indicate, for each subregion, the elevation range containing the majority (60%) of target reaches.
More than 60% of the reaches in the Florida (3C) and Mid-Atlantic Coastal Plain (3B) subregions
are found at low elevations between 4 and 76 m. In the Valley and Ridge (2Bn), Ozark/Ouachita
(2D), and Piedmont (3A) subregions, target reaches are of moderately low elevation, with the
majority found between 96 and 390 m. The Southern Appalachian (2X) and Pocono/Catskill (ID)
101
-------
Table 5-1. Refined Target Population Total Stream Resource Estimates&
NSS-I Subregion
ID (Poconos/Catskills)
2Cn (N. Appalachians)
2Bn (Valley and Ridge)
3B (Mid-Atlantic Coastal Plain)
2As (Southern Blue Ridge)
3A (Piedmont)
2X (Southern Appalachians)
2D (Ozarks/Ouachitas)
3C (Florida)
Subtotal Mid-Atlantic
Subtotal Southeast
Reach Nodes*
Downstream
N (n)
3,235 (56)
8,488 (61)
13,992 (47)
11,287 (58)
2,031 (54)
7,515 (47)
5,057 (40)
4,117 (48)
1,555 (34)
37,002 (222)
20,275 (223)
Upstream
N (n)
3,244 (58)
8,663 (67)
13,038 (44)
11,284 (57)
2,031 (54)
7,515 (47)
4,936 (39)
4,204 (49)
1,727 (31)
t
36,229 (226)
20,413 (220)
Length*
(km)
t
15,144
21,738
32,687
40,296
9,036
33,531
21,892
22,480
3,848
109,865
90,787
- 15,270
- 22,373
- 36,405
- 40,344
- 9,036
- 33,531
- 23,015
- 22,753
- 5,284
-114,392
- 93,619
Total NSS-I
57,277(445) 56,642(446) 200,652-208,011
&
Prefers to the population estimate of the total number of reaches; n is the sample size;
standard errors for the population estimates are tabulated in Appendix C.
The number of upstream and downstream reach ends may not agree because differing
numbers of upstream and downstream ends may be assigned to noninterest subsets of the
target population.
The smaller number represents all target reaches for which both ends met the site
inclusion criteria. The larger number includes additional reaches for which one end was a
noninterest point.
102
-------
Poc/Catskills "
NAnnolooh U
Valley & Rdg. "
MA Coastal PI. "
S. Blue Rdg. ^
Piedmont ^
S. Appalaoh. "
Ozark/Ouach. ^
Florida ^
1
[•
1" f"
r--;:;::;:.:
i r
t
i &;::
f
i
E „,
1 F
j;
! 1
! t
! 1
;i t—.
! |
'
13
'"*
i
J
:•
•1
:' '
..'• 1
t. '
, t
1
-1
....,
]
"J
n
•• i
J]
•xw:
« -.•-•::.!.:
O 10 2O 30 40 50 60 70 80
I I Second Order t:i!::i;::i::| Fourth Order
Fifth Order
VJ Sixth Order or higher
Poc/CatsWHs
N. Appalach. j|
Valley & Rdg. "
MA Coastal PI.
S. Blue Rdg.
Piedmont
S. Appalach. ^
Ozark/Ouach.
Florida
40 60
% of Reaches
100
(b) Blue Line Strahler Order (1:24,000 Maps)
Figure 5-3. Population frequency distributions for (a) drainage area (20th percentile, median,
80th percentile) and (b) population frequency histograms, Strahler Order
(l:24,000-scale maps) at NSS-I target reach upper and lower nodes.
103
-------
Poc / Catskills
N. Appalach.
Valley & Rdg.
MA Coastal PI. "
S. Blue Rdg.
Piedmont
S. Appalach.
Ozark/Ouach. u
Florida
u
1
1
u
1
L
1
U
L
U
L
u
L
U
L
0
£
| '
L,
1
c
L.
CJ
!
I
; i
! 1
.0
! I | |
lx-:jx-x-x-xvx-x*'X.x! j j
1 ; B'X¥S¥X'X-x¥x-:-3l !
i • '• '•
till
1 1 1 I
Kx^X'XWx'x^x^x-xV:! » i «
t'x-*t:*:*x::W:*:>:x::^ { J
iiii
iiii
i i : :
i i r i
• i i i
K'l'X'X'X'X-X'XJi i i
i i i i
_._._,,.,..,,. iiit
l:::X:x;xx:>x:x-x=1 '. i ',
! ! ! i
[:;X:X
LJ j FX-:-:-:-;-X';:;:;: frvfflvtt' x-x-;fo>W'i''':*:*: .
1111
1 ' I !
1 1 t:*x:::x:x££l ! !
1 |x:::X:::::X:X:^^ j |
i i i i
0.2 0.4 0
6
•X'X-X'Xi
WSWSftSI
o
1
B
i
1.(
(a) Stream Depth (m)
Poc/Catskills "
N. Appalach. "
Valley & Rdg. ^
MA Coastal PI. JJ
S. Blue Rdg. "
Piedmont "
S. Appalach. ^
Ozark/Ouach. "
Florida u
^
i
c;
: i
1 L
i j
i
c
1 1 1
i |:-XvX-:-:<;:;:;:::>>:::X:y
i i
k:>x-x-x-:-x-:^x-X':vX:X::*::::X::i j j
i i :
P:o:Wx-:¥x-:<:::x-x-x-:-x-
i i
i i
I * ' l*XvX:.:.:-o
; ;
• i
i i
i i
X'X-X-XrX^ i
i i
x-x-i-x-x-li !
i i
i i
.sff&J I !
xWSfSrw^x^X'X-Xv'HX'XvX^:::::
i i
i i
T i i
1 ! K:x:x:::::::::x;R:xS$^::Xi!:X:::::X:X:::::S:XJ
! i
i i
p^ft^xix-fK'KSSft^^ysftWs::!! •
I 1 ! 1:::::=:::
; ;
^x:>:o:-:'Xv:::::x:::x:::::l !
i i
i^xSxSx&xWrc^
i i :
i i
i i
! !
; tvx-xwx^ta i i
I ! t'Si:WX;xX:::>x:Xx:xW*:X:X:::::::X:>::|
i i
1 1
....I.... 1 ....
'i
•:-x-:-x-x«
....
'
'
1
10
(b) Stream Width (m)
Figure 5-4. Population frequency distributions (20th percentile, median, 80th percentile) for
(a) depth and (b) width of NSS-I target reaches at upper and lower nodes.
104
-------
Poc / Catskills ^
N. Appalach. "
Valley & Rdg u
MA Coastal PI. "
S Blue Rdg u
Piedmont u
S. Appalach. u
KK L
Ozark/Ouach. u
L
Florida "
<
1
on
m
) )0
1 1
1 1
1 ! Jx-x-x*:*:*:*
1 ! t::*x*:5*x:x|
i i
i i
: : d
• ', i
1 ! myJx:*:::*::***:!
1 J KWSfS:W:f SSWSftia
1 j
1 1
1 1
1 1
t 1
1 1
1 |
1 1
1 1
1 1
| ' |;XyX-XvXvXt|
1 ""l-'>':::'::::x:::::j '
t i
! L- ! .._E
: i i ii""":::::
i i
i i
1 £:x«x:x:x:¥x$xx:i:x:3
IWxWsis&l '
i i
i i
i i
i i
i i
t i
i i
i i
i i
0 20O 300 40
•X-Xl
1
I
0 SO
1 L^"
' — "^
1
0 60
*x*x:|
3
0 70
fffi
i
0 80
0 800
(a) Elevation (m)
Poc / Catskills "
N. Appalach. ^
Valley & Rdg u
MA Coastal PI u
Spli ,e\ D<-l<-« U
. Blue nag.
Piedmont u
S. Appalach. u
Ozark / Ouach u
Florida ^
i
!_
1
i
i
[
! 1
! 1
! 1 Kfffffffi
i I |*:*x*:V:!
1 j
! 1 t-:*x*x:
! i — mm?
i | ||xX
1 t*x
1 fx*
1 b*xl
> 1
I
, \XX*^
~^~^^^—^^~~^~
:*x*:J
x*:*:l
>:*x*x*:*:*Xr:-J
*x*x*:*x:x*fl
*:*:*!
:-X*x|
2
, fey::::Sff:x::ff.
:W:-xx:"x%%Wx
4
-•.•.•.•.'.•.•.•.•.•.•.•.•.*.•.•.-.
3
s
f::::m*?:::l
e
7
(b) Stream Gradient (%)
Figure 5-5. Population frequency distributions (20th percentile, median, 80th percentile) for
(a) upper and lower node elevation, and (b) gradient between nodes of NSS-I
target reaches.
105
_
-------
target reaches are of moderately high elevation, with most between 130 and 685 m. The North-
ern Appalachian (2Cn) and Southern Blue Ridge (2As) subregions have target reaches at rela-
tively high elevation, with most reaches between 334 and 861 m.
Target population gradient distributions (Figure 5-5b) reflect the elevation distributions,
with the lowest gradients (medians less than 0.5%) found in subregions 3C, 3B, and 3A. Those
subregions with the highest elevations (2Cn, 2As, 2X, and ID) have higher median gradients
(0.8% to 3.0%).
5.5 GEOGRAPHIC CLASSIFICATION OF SAMPLE SITES
Examination of regional patterns and associations between stream chemistry and site char-
acteristics revealed that in many cases the NSS-I subregions encompass a large amount of geo-
graphic variability. To improve the interpretation of the influence of watershed characteristics
on stream chemistry, it was necessary to reclassify NSS-I sites on the basis of geographic char-
acteristics alone. The NSS-I subregion boundaries (Section 1, Figure 1 -2) were drawn with con-
sideration for topography, geology, soils, climate, vegetation, land use, and presence or absence
of lake resources, but also with respect to the rate of atmospheric acid deposition. The sub-
region names refer to the dominant geographic or physiographic feature or unit within each
subregion. However, subregion boundaries do not always correspond with the recognized boun-
daries of the physiographic units for which these subregions are named. For example, NSS-I
subregion ID, the Pocono/Catskill subregion, contains glaciated portions of the Northern Appa-
lachian Plateau and the extreme northern portion of the Valley and Ridge Province (Fenneman,
1946), in addition to the Catskill and Pocono Mountains (Table 5-2). The Mid-Atlantic Coastal
Plain subregion (3B) includes a number of sites from the northern Piedmont physiographic prov-
ince, in addition to those on the Coastal Plain itself. Similarly, the Blue Ridge Mountains
physiographic province (Fenneman, 1946) is largely contained within NSS-I subregion 2As, the
Southern Blue Ridge, but the northern Blue Ridge sample sites are contained within NSS-I sub-
region 2X, the Southern Appalachians. A few sites of the Southern Blue Ridge physiographic
province are contained in NSS-I subregion 3A, the Piedmont.
NSS-I sites were reclassified solely upon geographic considerations—largely according to
the physiographic regions of Fenneman (1946). Individual sites within these larger regions were
subclassified according to geology, local elevation, vegetation cover, and land use (Tables 5-2
and 5-3). The geographic classes (Table 5-2) are properly treated as substratifications within
the NSS-I subregions when calculating population distribution estimates and their variance
(Section 9.5; Appendix D). Throughout this report, population estimates of the distributions of
target reach physical or chemical characteristics (e.g., Section 6) refer to the NSS-I subregion
strata unless explicitly stated otherwise (e.g., Section 9.5).
106
-------
Table 5-2.
Distribution of Refined Target Population Upper and Lower Reach Nodes by
Geographic Site Class in NSS-I Subregions
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plat.
Northeast Mid-Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plat.
Glaciated Forested Plat.
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid-Atlantic
Western Mid-Atlantic
New Jersey Pine Barrens
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
Total
SUBREGION ID
ID 2Cn 2Bn 3B 2As 3A 2X 2D 3C Total
- 13 ....... ,3
- 56 ....... 56
- 47 .... 7 - - 54
16 - 16
31 - - - 31
13 ........ 13
9 6 32 .--- 9 - - 56
45 6 40 - 6 - 27 - - 124
20 - 20
- 30 - 30
29 - 29
9 18 - 68 4 - - 99
- - 10 7 17
- 94 4 14 112
----- 2 18 - - 20
- - - - 8 - - - - 8
- - - 18 ----- 18
- - - 50 - - - _ . 50
- - - 22 ----- 22
20 - - - 20
- 18 - 18
- 36 36
- 9 9
9 9
- - - 11 11
114 128 91 115 108 94 79 97 65 891
107
-------
Table 5-3. Characteristics of NSS-I Geographic Site Classifications
ALLEGHENY PLATEAU
High Plateaus
Land Use: Forest
Geology: Sandstone
Relief: Shallow to steep
Soils: Sandy loam, deep, well-drained
Forested Plateau
Land Use: Forest with several strip mines throughout the region
Geology: Sandstone, shales, and siltstones
Relief: Moderate
Soils: Silt loam to sandy loam, good drainage
Mixed Forest and Agriculture Plateau
Land Use: Mixed forest and agriculture
Geology: Sandstones and shales
Relief: Moderate
Soils: Silt loam to sandy loam, good to well drained
NORTHEASTERN MID-ATLANTIC
Pocono/Catskill Mountains
Land Use: Forest or woodland
Geology: Sandstone, shales, quartz conglomerates
Relief: Moderate to Steep
Soils: Silt loam, shallow to deep
Glaciated Agricultural Plateau
Land Use: Agriculture, extensive transportation development throughout the region
Geology: Sandstone, siltstone, and shales
Relief: Moderate
Soils: Silt loam, loam, poor to good drainage
Glaciated Forested Plateau
Land Use: Forest or woodland
Geology: Sandstones, shales, siltstones, and coal
Relief: Moderate
Soils: Silt loam, loam, good to well drained
108
-------
Table 5-3. Characteristics of NSS-I Geographic Site Classifications
(Continued)
VALLEY AND RIDGE PROVINCE
Ridges
Land Use: Forest
Geology: Sandstone, shales, and siltstones
Relief: Moderate to steep
Soils: Silt loam to sandy loam, good to well drained
Valleys
Land Use: Agriculture and urban
Geology: Limestones, calcareous shales, dolomites, glacial till
Relief: Shallow in lower watershed, moderate in upper watershed
Soils: Silt loam to sandy loam, good to well drained
ARKANSAS/OKLAHOMA
Boston Mountains
Land Use: Forested with < 10% of the watershed in agriculture
Geology: Shales and sandstones
Relief: Steep
Ouachita Mountains
Land Use: Forested with < 5% of the watershed in agriculture
Geology: Sandstone and shales
Relief: Moderate in the West, steep in the East
Arkansas River Valley
Land Use: Mixed forest and agriculture
Geology: Shales and sandstones
Relief: Shallow
PIEDMONT
Piedmont
Land Use: Mixed forest and agriculture
Geology: Schist, gneiss, small number of watersheds with granite or phyllite
Relief: Shallow to moderate
Soils: Clays of clayey loam, moderate well drained, deep to shallow
Piedmont Lowlands
Land Use: Agriculture with patches of forest
Geology: Limestone, schist, and gneiss
Relief: Little or none
Soils: Silty loam, sandy loam, clays
109
-------
Table 5-3. Characteristics of NSS-I Geographic Site Classifications
(Continued)
SOUTHERN APPALACHIAN HIGHLANDS
Blue Ridge Mountains
Land Use: Forest with some agriculture isolated in valley bottoms
Geology: Phyllite, quartzite, graywacke, conglomerates
Relief: Shallow to steep
Soils: Loam, good drainage
Cumberland Plateau
Land Use: Forested with some mixed forest and agriculture
Geology: Sandstones and shales
Relief: Shallow
Soils: Loamy soils, well drained, strongly acid
French Broad River Valley
Land Use: Agricultural
Geology: Phyllite, quartzite, graywacke, conglomerates
Relief: Shallow
Soils: Deep loam, good drainage
COASTAL PLAIN
Eastern Mid-Atlantic
Land Use: Agriculture, urban, small patches of forest
Geology: Sands, mix of sands, gravels, clays, and silts
Relief: Little or none
Soils: Sandy, increasing amounts of clay to the south
Western Mid-Atlantic
Land Use: Mixed forest and agriculture in north, increasing forest cover in the
southern portions of the region
Geology: Sands in the northern portion, fossilerous clays and sands in the southern
portion
Relief: Shallow to moderate (swampy in the 3B070 sites)
Soils: Fine loamy with a small percentage of sites with coarse loamy or fine
silty, well drained
New Jersey Pine Barrens
Land Use: Forested with some urbanization, cranberry bogs
Geology: Sandstone, gravel and sandstone
Relief: Very shallow
Soils: Quartz sands
110
-------
Table 5-3. Characteristics of NSS-I Geographic Site Classifications
(Continued)
Eastern Gulf of Mexico
Land Use: Mainly forest with some agriculture along stream reach
Geology: Sandstone, shales, clays, gravels, lignite, and coal
Relief: Shallow
Soils: Loamy, well drained, deep to moderately deep
Western Gulf of Mexico
Land Use: Mixed forest and agriculture
Geology: North - shales and sand; south - alluvium
Relief: Shallow
FLORIDA
Panhandle Upland
Land Use: Forest or mixed forest and agriculture
Geology: Gravel, sand, micaceous kaolinitic clay
Relief: Shallow
Soils: Sandy, well drained
Peninsula Upland
Land Use: Mixed forest and agriculture
Geology: Coastal sands, sand and clayey fine sands
Relief: Shallow
Soils: Sandy, well drained
Panhandle and Peninsula Swamp Sites
Land Use: Forest
Geology: Sands over clayey sands
Relief: Shallow or none in most watersheds
Soils: Sands to sandy loam, poor to well drained
111
-------
THIS PAGE INTENTIONALLY LEFT BLANK
112
-------
SECTION 6
TARGET POPULATION REGIONAL CHEMISTRY
Section 6 addresses the National Stream Survey - Phase I (NSS-I) primary objectives of
estimating the regional extent of acidic and low acid neutralizing capacity (ANC) streams and
describing their chemical characteristics. Further interpretation of the NSS-I data is contained
in the four subsequent sections. Section 7 addresses uncertainty in the population estimates due
to temporal variability. Section 8 discusses the chemical characteristics of the target population
in greater detail, examining the relationships among chemical constituents. Section 9 classifies
acidic and low ANC streams according to probable sources of acid anions, and Section 10
examines evidence of acidification in streams of the target population.
6.1 INTERPRETING NSS-I POPULATION DISTRIBUTIONS
The subjects discussed in subsections 6.1.1 through 6.1.6 are relevant to interpretation of
the NSS-I regional population distribution data presented throughout Section 6.
6.1.1 Target Population of Interest
The target population as defined in the discussion of the statistical sampling frame (Section
2.4) has been refined to encompass a slightly more restrictive population in most of the NSS-I
subregions. In Florida (3C), where a substantial proportion of intended sample reaches were dry,
the difference between the original and refined target populations is greater. This refined sub-
population of interest within the target population of the nine NSS-I subregions excludes sites
affected by acid mine drainage but not previously excluded on the basis of low field pH (< 3.3)
or high field conductivity (> 500 fiS cm'1). Stream reaches that were without discernable flow
or that were too shallow to sample were also excluded. Section 5.2 discusses the process of
excluding sample reaches in more detail and estimates the population proportions of these types
of noninterest categories in the nine NSS-I subregions.
6.1.2 NSS-I Subregions
Throughout Section 6 and the remainder of this report, population estimates for distribu-
tions of target reach physical or chemical characteristics refer to the NSS-I subregion strata
shown in Figure 2-2 and the inside of the front cover, not the geographic classification of
target streams, unless explicitly stated otherwise. Except in Section 9, the geographic site
classes within NSS-I subregions are used in a descriptive way to aid interpretation of the
observed geographic patterns in chemistry. In Section 9, population estimates show the
distribution of target reaches according to geochemical types within each geographic class.
6.1.3 The Florida Subregion
The Florida subregion (3C) was surveyed as a feasibility study to evaluate the point-frame
sample reach selection procedure and site inclusion criteria in a coastal lowland setting.
Sampling was concentrated on a geographic region expected to contain surface waters with ANC
less than 200 fieq L"1, rather than 400 jueq L"1, as was the case in the other NSS-I and Eastern
113
-------
Lake Survey (ELS) subregions. NSS-I results for Florida are discussed separately in Section
6.2.4 because results are not strictly comparable with those from other subregions.
6.1.4 The Southern Blue Ridge Subregion
A Pilot Survey was undertaken in the Southern Blue Ridge in 1985, a year before field
activities in the remaining NSS-I subregions. Population estimates given in this section based on
the average downstream chemistry during spring baseflow are as reported and discussed by
Messer et al. (1986, 1988), with very small changes resulting from a minor modification in the
procedure used to calculate ANC from titration curves. Spring season samples were collected
only at the upstream nodes on 20 of the total 54 sample reaches in this subregion. However,
the full set of downstream nodes was sampled in both spring and summer, and the full set of
upstream nodes was sampled in the summer. To provide population distribution estimates for the
Southern Blue Ridge that are comparable to spring upstream estimates for other subregions,
highly predictive empirical relationships have been employed using available data to synthesize
upstream spring values for the 34 missing sites (Appendix B). These synthesized values are not
included in the enhanced data base, but can be provided in a separate small data set.
6.1.5 Index Sample
The spring baseflow chemical index concept and the rationale for its selection and use are
discussed in Section 2.5. All population distributions are based on spring baseflow index
chemistry. In the Southeast, measurements from single visits to each upstream and downstream
sample site comprise the spring index for each node of the sample reaches. In the mid-Atlantic,
the spring index at each reach node is calculated as the average from two separate visits.
Indices for pH were calculated by averaging hydrogen ion concentrations from the separate
visits. Because a trend of decreasing acidity between visits was observed in some subregions,
population distribution estimates based on the separate values of pH and ANC observed on each
visit are provided in subsection 7.4.2.
It is important to understand the limitations as well as the strengths of the chemical index
approach. For example, the number of stream reaches that are acidic due to episodic pH
depressions during storm runoff, or to seasonal variability at other times of the year, cannot be
obtained or inferred directly from the distribution of index chemistry provided by the NSS-I.
The number of streams experiencing acidic episodes may be much greater than the number esti-
mated to be acidic during spring baseflow. Further studies of short-term and seasonal variabil-
ity are expected to demonstrate a useful predictive relationship between index chemistry and
chemistry during other times of the season or year, as has been shown for the Adirondack Lakes
(Driscoll, 1986) and other eastern Lakes (Newell, 1987), in the United States.
6.1.6 Distributions Based on Upstream and Downstream Chemistry
This report provides separate regional chemical distributions calculated from chemical data
collected at upstream and downstream ends of NSS-I sample reaches. The population distribution
estimates based on upstream and downstream sampling comprise two spring baseflow "snapshots"
focusing at different positions within the streamflow network represented by blue lines on
l:250,000-scale topographic maps (Figure 6-1). Each upstream reach end (upper node) is paired
114
-------
Blue Line Reach
Sampled Reach
Upper Node
Lower Node
Figure 6-1. Conceptual representation of upper and lower reach node samples and
populations.
115
-------
with a corresponding lower node at the downstream end of the reach. Because roughly two-
thirds of the NSS-I target population is made up of headwater reaches (on l:250,000-scale maps),
the population of upper reach nodes is located largely upstream of the lower node population
(i.e., for two-thirds of the population, there is no overlap). However, the other one-third of
the target reaches that are not headwaters have upstream nodes farther downstream in the net-
work than lower nodes of reaches higher up in the drainage network.
One can think of the downstream sampling points as representing stream water at a given
moment in time as it exits the population of target reaches that form the total blue-line
network. (This is not the same thing as the water quality leaving the whole network, however,
because the population of target reaches is a hierarchical network in which smaller target
reaches are nested within the network as tributaries to larger reaches). Downstream ends of
the target reach population are typically of Strahler Order 2 and 3 (sensu Strahler, 1957, based
on l:24,000-scale maps) and drain land areas primarily in the range from 1.2 to 56 km2.
Upstream ends are located, on the average, 3 km upstream from lower ends of target population
reaches. Water quality samples taken at the upstream ends of the sample of target reaches are
therefore representative of water draining a range of watershed areas that overlaps the range of
the lower ends, but is somewhat smaller. Upstream ends of target reaches are typically of
Strahler Order 1 and 2 with drainage areas primarily in the range from 0.1 to 47 km2.
6.1.7 Interpolated Length Distributions
Separate distributions and scatter plots of chemistry at the upstream and downstream ends
of target stream reaches provide a fairly complex picture of their status. Similar graphs
depicting chemical distributions based on the length and other attributes of the stream resource
are provided in the data compendium comprising Volume II of this report. The length and total
drainage area (discharge index) plots in Volume II are based on the chemistry at one reach node
(lower node). It is perhaps of greater interest to combine upstream and downstream chemical
measurements and to report the status of stream resources in terms of the combined length of
streams within reference categories of pH and ANC. Such estimates may have more utility for
fish habitat quality assessment. It is possible now to present population estimates of this
nature, but the statistical confidence bounds provided for them are somewhat crude at this time.
These interpolated length estimates are based on a simplistic model that assumes that chemical
concentrations change in a linear fashion from the upstream sampling point to the lower one.
Concentrations of ANC and hydrogen ion for positions on stream reaches between sampling
locations have been linearly interpolated between sampling points to produce the population
distributions of ANC and pH reported in the bar charts presented in subsection 6.2.5.
6.2 DISTRIBUTIONS OF pH AND ANC
6.2.1 Regional Overview
Population proportions are reported within ranges defined by reference values. The value
of ANC - 0 is of significance because waters at or below this level have no capacity to neutral-
ize acid inputs—they are acidic by definition.
The 50 /teq L"1 reference value forms an upper bound of the range of aquatic systems
between 0 and 50 fieq L"1 that may be considered to have very low ANC. Although there may
be no special significance to the 50 peq L"1 level, some scientists term surface waters with ANC
< 50 /Jeq L"1 as "extremely acid sensitive" (Schindler, 1988). There is some evidence that
116
-------
certain groups of streams and lakes with spring baseflow ANC (streams) or fall turnover ANC
(lakes) below this level may be prone to acidic episodes during storm events or snowmelt
(Driscoll, 1986; Eshleman, in press). Low ANC is probably a necessary condition predisposing
acidic episodes, but certainly not a sufficient one. Streams or lakes experiencing acidic episodes
are also often subject to high atmospheric acid deposition rates, and may have watersheds that
are more susceptible to episodic pH depressions due to their hydrology (Marmorek, et al., 1986).
The 200 /zeq L"1 ANC reference value is of interest because it has been frequently cited in
acid deposition literature as a value below which streams or lakes are often defined or hypothe-
sized to be sensitive to acidification (Schindler, 1988). Some researchers appear to define all
waters with ANC < 200 /zeq L"1 as sensitive (Johnson, 1986), whereas most consider 200 jteq IT1
as a threshold above which aquatic systems are unlikely to be sensitive. It should be kept in
mind that "sensitivity" to acid deposition impacts depends on factors other than the ANC of the
stream water itself. A stream is an integral part of its watershed; as such, its sensitivity to
acid deposition depends on the ability of watershed geology, soils, and vegetation to neutralize
incoming acidity.
Before proceeding with an overview of ANC population distributions in the NSS-I regions,
it is appropriate to keep in mind that the system level precision of ANC measurements was 3 to
8 Ateq L'1 (pooled SD of field duplicate pairs) for ANC ranges < 200 0eq L"1 (Table 4-3).
Precision estimates that take into account laboratory bias and day-to-day variation in procedures
had a %RSD of approximately 11% in the ANC range of 50 to 200 /teq L"1 (Table 4-3).
The proportions of the target stream reach population with spring index ANC values of 200
/zeq L"1 or less were similar in the Mid-Atlantic (MA) and Southeast (SE) regions surveyed by
the NSS-I. Excluding Florida, 51% (18,542) of MA reaches and 52% (9,642) of SE reaches had
ANC < 200 jteq L"1 at their upstream ends. After flowing an average of 3 km to their down-
stream ends, an estimated 41% (15,060) of MA reaches and 47% (8,682) of SE reaches were at or
below this index ANC value. In the Mid-Atlantic Region, 7.4% of reaches (2,677) were acidic
(ANC < 0 /zeq L'1) at their upstream ends and 3% (1,098) were acidic at their downstream ends.
In the NSS-I Interior Southeast subregions, however, reaches that were actually acidic during
spring baseflow were rare. (Note that Florida is not included and will be discussed separately.)
An estimated 0.6% of SE reaches (121 reaches in the Cumberland Plateau of the Southern Appa-
lachians) had acidic upstream ends. No reaches with ANC less than zero (upstream or down-
stream) were observed in the NSS-I in the remaining SE subregions (Southern Blue Ridge,
Piedmont, and the Ozarks/Ouachitas). Cumulative frequency distributions for ANC in the
upstream and downstream reach node populations of individual subregions are presented in
Figures 6-2 (a - i).
Discussion of the pH distribution in the regions surveyed by the NSS-I refers to several
reference pH values (5.0, 5.5, and 6.0). These are defined partially for convenience, and
partially for comparison with other studies (e.g., Linthurst et al., 1986; Landers et al., 1987;
Messer et al., 1986) that have reported results referencing such values. These reference values
bear some relationship to critical values below which populations of different types of fish in
freshwater ecosystems are not sustained. Because of the influence of pH on the solubility and
speciation of aluminum (Stumm and Morgan, 1981), it is difficult to separate the effects of low
pH and aluminum toxicity. However, it has been reported that waters with pH chronically below
4.0 are virtually devoid of fish (Harvey, 1980; Leuven et al., 1984; Wendelaar-Bonga and Dederen,
1986). Similarly, waters with pH chronically below 4.5 are reported to contain few fish species
(Cooper and Wagner, 1973). Cooper and Wagner (1973) conclude that pH values less than 5.0
may limit the distribution of brook trout (Salvelinus fontinalis). one of the most tolerant species
(Haines, 1981; Magnuson et al., 1984; Johnson et al., 1987).
117
-------
ANC
POCONOS / CATSKILLS
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
20*
40%
MEDIAN
60*
80*
MAX
MEAN
STO
UPPER LOWER
NODE NODE
-76.4
41 .3
173.7
206.9
245.0
635.0
4580.8
431
1 .5
1 14.4
230.2
257.4
357. 1
655.5
4871.4
526
697.9 747.6
SAMPLE(n) 58 56
POP EST(N) 3244 3235
SECN) 347 347
aoo *oo «oo
LOWER NODE ANC
UPSTREAM POPULATION
NUMBER OF REACHES
JIPP£ft*5K O.
PROPORTION <•> X
_2OO —TOO
1OO 2OO 300
ANC (peq L"1)
4OO SOO
DOWNSTREAM POPULATION
NUMBER OF REACHES
UPPER95ECL
PgOPORT10(l<= X
Q.o
-200
BOO
ANC
Figure 6-2a. Population distributions and comparison of index ANC at upper and lower reach
nodes in NSS-I subregion ID (Poconos/Catskills).
118
-------
ANC
NORTHERN APPALACHIANS
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
205!
4-05!
MED IAN
60%
80^
MAX
MEAN
STD
UPPER LOWER
NODE NODE
-48.0
19.6
64.4
1 10.5
155. 1
404.3
2702.3
296.4
519.9
-26.0
43.5
85.9
127.6
171 .7
445.4
215O.3
312.5
452.5
SAMPLE(n) 67 61
POP EST(N) 8663 8488
SE(N) 807 814
200 4-QO 000
LOWER NODE ANC
S
UPSTREAM POPULATION
NUMBER OF REACHES
UPPER93ZCL
: PROPORTION <= X
500
ANC
I
§
r
1OO 2OO 3OO
ANC (tteq L'1)
DOWNSTREAM POPULATION
NUMBER OF REACHES
UPPER8SBCL
PROPORTION <= X
SOO
Figure 6-2b. Population distributions and comparison of index ANC at upper and lower reach
nodes in NSS-I subregion 2Cn (Northern Appalachians).
119
-------
ANC
VALLEY AND RIDGE
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
20*
4-OK
MEDIAN
605!
80*
MAX
MEAN
STD
SAMPLE(n)
POP EST(N)
SE
-------
ANC
MID-ATLANTIC COASTAL PLAIN
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
20%
40%
MEDIAN
60%
80%
MAX
MEAN
STD
SAMPLE(n)
POP EST(N)
SE(N)
UPPER
NODE
-442.3
6.5
114.2
150.5
240.2
431 .2
1117.1
258.8
282.9
57
1 1284
1078
LOWER
NODE
-181 .5
54.4
188.6
308.7
387.6
657.6
2196 . 1
452.4
506 .0
58
1 1287
1078
ZOO +OO 000 BOD 1000
COWER NODE ANC
UPSTREAM POPULATION
NUMBER Of REACHES
UPPER95ZCL
PROPORTION <= X
SOO
ANC
DOWNSTREAM POPULATION
NUMBER OF REACHES
UPPEBSSK CL
PROPORTION <= X
SOO
ANC (peg L"1)
Figure 6-2d. Population distributions and comparison of index ANC at upper and lower reach
nodes in NSS-I subregion 3B (Mid-Atlantic Coastal Plain).
121
-------
ANC
SOUTHERN BLUE RIDGE
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
20*
40s
MEDIAN
60*
80*
MAX
MEAN
STD
UPPER
NODE
20.9
67.7
88.5
99.6
1 14.3
157. 1
2227.9
241 .3
451 .7
LOWER
NODE
16.5
88.4
104.4
121 .6
137.5
206.7
1730.4
257.2
403. 1
SAMPLECn) 54 54
POP EST(N) 2031 2031
SE(N) 326 326
E
8:
LOWER NODE ANC
UPSTREAM POPULATION
NUMBER OF REACHES
UPPERS3XCL
PROPORTION <-X
1OO 2OO
ANC
sou
OOWNSTRCAM POPULATION
NUMBER OF REACHES
UPPER 05XCL
PROPORTION <= X
60O
ANC (|ieq L~')
Figure 6-2e. Population distributions ana comparison of index ANC at upper and lower reach
nodes in NSS-I subregion 2As (Southern Blue Ridge).
122
-------
ANC
PIEDMONT
STATISTIC UPPER
(ESTIMATE) NODE
UPPER VS LOWER SAMPLES
WIN
20*
4-05!
MEDIAN
60%
805!
MAX
MEAN
STD
29.7
86.0
173.0
201 .4
256. 1
372. 1
736. 1
256.2
179.6
SAMPLE(n) 47
POP EST(N) 7515
SE(N) 650
LOWER
NODE
33.6
101 .9
188.4
259.3
335.3
426.2
799.9
285.3
183.5
47
7515
650
ZOO +QO 6OQ
LOWER NODE ANC
BOO 1000
UPSTREAM POPULATION
NUMBER OF REACHES
UPPER 45% CL
PROPORTION <= X
ANC
SOO
1OO 2OO
ANC
DOWNSTREAM POPULATION
NUMBER OF RUCHES
UPPERS5ECL
PROPORTION <= X
SOO
Figure 6-2f. Population distributions and comparison of index ANC at upper and lower reach
nodes in NSS-I subregion 3A (Piedmont).
123
-------
ANC
SOUTHERN APPALACHIANS
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
Ml N
20*
40*
MEDIAN
BOx
80*
MAX
MEAN
STO
SAMPLECn)
POP EST(N)
SE
-------
ANC
OZARKS / OUACHITAS
STATISTIC UPPER LOWER
(ESTIMATE) NODE NODE
UPPER VS LOWER SAMPLES
M1N
20%
40%
MEDIAN
60%
80*
MAX
MEAN
STD
SAMPLE(n) 49 48
POP EST(N) 4204 4116
SE(N) 406 410
15
85
1 18
122
151
254
2089
245
336
2
2
1
6
9
5
6
4
63
120
151
161
180
327
2987
303
453
2
2
8
9
1
1
8
2
0
o
2OQ *QO «OO
LOWER NODE ANC
UPSTREAM POPULATION
NUMBER OF REACHES
UPPERS3S CL
PROPORTION <=X
2OO 300
ANC (peq LT1)
soo
DOWNSTRCAU POPULATION
NUMBER OF REACHES
UPPEHSSXa
PROPORTION <= X
SOO
ANC
Figure 6-2h. Population distributions and comparison of index ANC at upper and lower reach
nodes in NSS-I subregion 2D (Ozarks/Ouachitas).
125
-------
ANC
FLORIDA
STATISTIC
(ESTIMATE)
MIN
20*
405!
MEDIAN
60s
80*
MAX
MEAN
STD
UPPER VS LOWER SAMPLES
UPPER LOWER
NODE NODE
-645.2 -138.3
-545.0
-0.4
7.6
25.3
207.0
7.0
18. 1
33.2
60.5
174.1
368.0 1801.1
-69.3 156.6
313.4 323.4
SAMPLE(n) 31 34
POP EST(N) 1727 1555
SE(N) 437 306
-oo o aoo 400 aoo «oo 1000
UPSTREAM POPULATION
NUMBER Of REACHES
WPERtSXCL
PSOTOHTKM) <- X
ANC
DOWNSTRCAM POPULATION
NUMBER Of REACHES
UPPER85XCL
moromm<=x
soo
Figure 6-2i. Population distributions and comparison of index ANC at upper and lower reach
nodes in NSS-I subregion 3C (Florida).
126
-------
Blueback herring (Alosa aestiyalis), an important anadromous fish that spawns in small
tributaries entering Chesapeake Bay, like many other warm water fish species, is apparently
more sensitive to low pH and elevated aluminum concentrations than are brook trout and brown
trout (Klauda and Palmer, 1987; Klauda et al., 1987). Blueback herring eggs and larvae
experienced high mortality when exposed to acid pulses or prolonged exposure at pH 5.0 to 5.7
and calculated total monomeric aluminum concentrations between 0.03 and 0.09 mg L'1.
Klauda and Palmer (1987) inferred, from a review of field and laboratory studies, a general
ranking of the relative sensitivity of several fishes to pH and aluminum. They report, without
attempting a life-stage-specific comparison, that percids (e.g., darters), cyprinids, flagfish
(Jordanella floridae), desert pupfish (Cvorinodon n. nevadensis). and striped bass appear to be
more sensitive than blueback herring. On the other hand, some percids (e.g., yellow perch Perca
flavescens and walleye Stizostedion vitreum. centrarchids, esocids, catostomids, and salmonids)
are apparently less sensitive to low pH and high aluminum than are blueback herring.
Haines and Baker (1986) have summarized a large volume of literature concerning pH-toxic-
ity bioassays and the occurrence of various fish species in lakes and streams over a range of
pH. Estimates for critical values below which salmonid fish populations are not sustained range
from pH 4.7 to 5.5 (Haines and Baker, 1986). Estimates for Smallmouth Bass are between pH 5.2
and 5.5. Except for effects reported on several species of dace, shiners, and minnows, pH levels
above 6.0 have not been associated with adverse impacts on fish populations. However, the
recent blueback herring bioassays by Klauda and Palmer (1987) and Klauda et al. (1987) and
others (Hall et al., 1985, 1987; Hall, 1987) employing striped bass (Morone saxatilis) further
suggest that, under certain conditions of aluminum concentration, pH between 6.0 and 6.5 can
lead to increased egg and larval mortality.
Several types of pH measurements were made in the NSS-I (Section 3). The measurements
that most accurately reflect pH in the stream at the time of sampling are the closed-headspace
measurements made in the processing laboratory. These measurements are recorded in the NSS-I
data base as the variable PHSTVL. Before proceeding with an overview of population distribu-
tions of pH in the NSS-I regions, it is helpful to briefly review some quality assurance
information from Section 4 regarding the precision and accuracy of pH measurements. System
level precision estimates were 0.026 to 0.036 pH unit (Table 4-3), depending on the pH range of
samples. Precision estimates incorporating day-to-day variability in the processing laboratory
were less than 0.1 unit.
In the Mid-Atlantic Region, an estimated 13% of the reaches (4,770) had spring base flow
pH < 5.5 at their upstream ends. At their downstream ends, 5.4% (1,991) were at or below this
pH level. In the portions of the Southeast Region surveyed during NSS-I (excluding Florida), 2%
(347) had pH < 5.5 at their upstream ends and less than 1% had pH at or below 5.5 at their
downstream ends. The Florida subregion (3C) stands out as a geographic area with relatively
high percentages of acidic and low ANC streams. It is also an area, like the Mid-Atlantic
Coastal Plain subregion (3B), in which there are many highly colored streams with high concen-
trations of dissolved organic material that may ameliorate to some extent the adverse effects of
aluminum on fish in these streams (by complexing aluminum ion). Cumulative frequency distribu-
tions for pH (closed headspace, measured in the processing laboratory) in the upstream and
downstream reach node populations of the NSS-I subregions are presented in Figures 6-3 (a - i).
Table 6-1 compares population median pH, ANC, and SO42- based on both upstream and
downstream sampling locations across the nine subregions sampled by the NSS-I and the Pilot
Survey. It is evident that there is a consistent pattern of higher ANC and pH at downstream
ends of target stream reaches, supporting a hypothesis that streamflow tends to acquire chemical
weathering products in the downstream direction due to the increased time and opportunity for
127
-------
PH
POCONOS / CATSKILLS
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
20*
40*
MEDIAN
80*
MAX
MEAN
STD
SAMPLE(n)
POP EST
Figure 6-3a. Population distributions and comparison of index pH at upper and lower reach
nodes in NSS-I subregion ID (Poconos/Catskills).
128
-------
PH
NORTHERN APPALACHIANS
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
Ml N
20*
4-0%
MEDIAN
60%
BO*
MAX
MEAN
STD
SAMPLE(n)
POP EST(N)
SE(N)
UPPER
NODE
4.40
5.68
6.49
6.60
6.72
7. 14
8.00
5.66
5. 19
67
8663
807
LOWER
NODE
4.74
6.42
6.66
6.91
7 .07
7.37
8.25
6.05
5.55
61
8488
814
LOWER NODE PH
1
UPSTREAM POPULATION
NUMBER OF REACHES
UPPER BS% CL
PROPORTION OX
PH
-------
PH
VALLEY AND RIDGE
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
20*
40*
MEDIAN
605:
80*
MAX
MEAN
STD
SAMPLECn)
POP EST(N)
SE(N)
UPPER
NODE
4.68
6.57
6.96
7.05
7. 17
7.67
8.62
6.02
5.46
44
13038
1249
LOWER
NODE
5.59
6.94
7.31
7.41
7.52
7.99
8.88
7. 17
6.80
47
13992
1213
LOWEFJ NODE PH
UPSIREXU POPULATION
NUMBER OF REACHES
UPPER MX CL
PROPORTION <=X
r
j.
PH
DOWNSTREAM ropuunoH
NUMBER OF REACHES
UPPER 05X CL
PROPORTION <=X
PH CPH UN I TS)
Figure 6-3c. Population distributions and comparison of index pH at upper and lower reach
nodes in NSS-I subregion 2Bn (Valley and Ridge).
130
-------
PH
MID-ATLANTIC COASTAL PLAIN
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
20!!
40*
MEDIAN
60%
80*
MAX
MEAN
STD
SAMPLE(n)
POP EST(N)
SE(N)
UPPER
NODE
3
5,
5
5,
51
27
81
98
LOWER
NODE
3 83
6.40
7,
7,
5,
07
51
38
4.87
5
6
6
6
7
7
5
5
77
43
71
90
38
98
62
00
o
z
a:
s.
57
1 1284
1078
58
1 1287
1078
LOWER NODE PH
UPSTREAW POPULATION
NUMBER OF REACHES
UPPER 051CL
PROPORTION <=X
PH
-------
PH
SOUTHERN BLUE RIDGE
UPPER VS LOWER SAMPLES
STAT 1 ST 1 C
(ESTIMATE)
MIN
20*
4-Dx
MEDIAN
60x
80 r.
MAX
MEAN
STD
SAMPLECn)
POP EST(N)
SE(N)
UPPER
NODE
6.31
6.79
6.92
6.99
6.99
7.09
8.38
6.92
7.10
54-
2031
326
LOWER
NODE
6.38
6.86
6.97
7.03
7 .07
7.21
8.4-3
6.99
7. 17
54
2031
326
LOWER NODE PH
unutui roruuTBN
HUUSEKOFKACHES
OfPE«tSXO.
PH
-------
PH
PIEDMONT
STATISTIC
(ESTIMATE)
WIN
20*
40*
MEDIAN
60%
BO'S
MAX
MEAN
STD
SAMPLE(n)
POP EST(N)
SE(N)
UPPER
NODE
5.64
6.20
6.64
6.80
6.84
7.00
7.40
6.46
6.36
47
7515
650
LOWER
NODE
5.90
6.54
6.89
6.95
7.03
7. 15
7.36
6.64
6.55
47
7515
650
UPPER VS LOWER SAMPLES
LOWER NODE PH
UPSTREAM POPULATION
NUMBER OF REACHES
UPPERVSXO.
PROPORTION <^
f
PH CPH UNITS}
ft o.e •
I
S
50.*
DOWNSTREAM POPULATION
NUMBER OF REACHES
UPPERMXCI.
PROPORTION <- X
PH
-------
PH
SOUTHERN APPALACHIANS
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
20*
40*
MEDIAN
SOr.
&Q%
MAX
MEAN
STD
SAMPLE(n)
POP EST(N)
SE(N)
UPPER
NODE
5
6
7
7
7
7
a
6
5
39
4936
529
. 10
.63
.06
.33
.44
.80
.59
.40
.89
LOWER
NODE
5.88
6.92
7.29
7.36
7.48
7.82
8.40
6.91
6.63
40
5057
526
LOWER NODE PH
UPSTREAM POPULATION
NUMBER OF REACHES
UPPER WXCL
PROPORTION <=X
PH
Figure 6-3g. Population distributions and comparison of index pH at upper and lower reach
nodes in NSS-I subregion 2X (Southern Appalachians).
134
-------
PH
OZARKS / OUACHITAS
UPPER VS LOWER SAUPLES
STATISTIC
(ESTIMATE)
MIN
20*
405!
MEDIAN
60%
80*
MAX
MEAN
STD
SAMPLE(n)
POPNEST(N) 4204
UPPER
NODE
5.33
6.20
6.59
6.62
6.72
7.00
8.05
6.29
6.05
49
04
06
LOWER
NODE
5.80
6.65
6.79
6.82
6.93
7.21
8. 17
6.73
6.63
48
41 16
410
LOWER NODE PH
UPSTREAM POPUU70M
NUMBER OF REACHES
UPPER MX CL
PROPORTION <=X
s e
PH
-------
PH
FLORIDA
STATISTIC
(ESTIMATE)
MIN
20"
4-0%
MEDIAN
BOx
80*
MAX
MEAN
STD
SAMPLE(n)
POP EST(N)
SE(N)
UPPER
NODE
3.44
3.52
5. 16
5.48
5.70
6.25
6.92
4.08
3.85
31
1727
437
LOWER
NODE
4. 10
5.37
5.60
5.62
5.98
6.30
7.57
5. 16
4.80
34
1555
306
UPPER VS LOWER SAMPLES
LOWER NODE PH
UP3TREXU POPULATION
KISSER OF REACHES
UPPERB3XCL
PROPORTION <» X
PH CPH UNITS}
5 6
PH CPH UNITS)
Figure 6-3i. Population distributions and comparison of index pH at upper and lower reach
nodes in NSS-I subregion 3C (Florida).
136
-------
Table 6-1.
Population Median ANC, pH, and Sulf ate Concentration at Lower (L) and Upper
(U) Reach Nodes of NSS-I Subregions
Subregion
ID (Poconos/Catskills)
2Cn (N. Appalachians)
2Bn (Valley and Ridge)
3B (MA Coastal Plain)
2As (S. Blue Ridge)
3A (Piedmont)
2X (S. Appalachians)
2D (Ozarks/Ouachitas)
3C (Florida)5
ANC
L
257
128
396
309
122
259
383
162
33
G*eq L"1)
U
207
110
223
151
100*
201
320
123
8
L
7.3
6.9
7.4
6.7
7.0
7.0
7.4
6.8
5.6
PH
U
7.0
6.6
7.1
6.0
7.0*
6.8
7.3
6.6
5.5
so42-
L
238
220
185
151
23.2
38.2
70.6
64.5
16.1
Ozeq L"1)
U
214
209
182
125
20.7*
37.5
61.0
699
10.0
See text: many S. Blue Ridge upper node chemistry values were calculated (Appendix B).
See text: Florida medians are based on a more restricted sample of low ANC streams. Do
not compare directly with other subregions.
137
-------
contact with watershed soil, rock, and vegetation. These upstream-downstream differences were
generally overshadowed by the more pronounced differences that occurred among different
streams within a subregion, or among subregions, as is perhaps more clearly illustrated in Figure
6-4 (a and b). The lowest subregional median ANC was observed in Florida (3C), with upstream
and downstream median ANC of 7.6 and 33 /*eq L'1, respectively. Florida was also the only sub-
region with median pH < 6.0. Of the remaining subregions, only upstream ends of target reaches
in the Mid-Atlantic Coastal Plain subregion (3B) had median pH < 6.5. Population median values
for spring index ANC at the downstream ends of target reaches in the Mid-Atlantic subregions
ranged from 128 A*eq L'1 in the Northern Applachians (2Cn) to 396 jteq L'1 in the Valley and
Ridge (2Bn) subregion. Subregion median ANC at the upstream reach ends in the Mid-Atlantic
were lower, ranging from 110 /zeq L'1 in subregion 2Cn to 223 ^eq L'1 in subregion 2Bn. Popu-
lation median ANC values < 200 /ieq L'1 (upstream or downstream) were observed in the Florida
(3C), Southern Blue Ridge (2As), Northern Appalachian (2Cn), Mid-Atlantic Coastal Plain (3B),
and Ozark/Ouachita (2D) subregions.
6.2.2 Mid-Atlantic Region
Population distributions for ANC and pH in the Mid-Atlantic subregions (3B, ID, 2Bn, and
2Cn) are illustrated in Figures 6-2 (a - d) and 6-3 (a - d). Among the subregions of the Mid-
Atlantic, the NSS-I revealed the greatest number of acidic stream reaches in the Mid-Atlantic
Coastal Plain (3B), where 12% of the target reaches (1334) were acidic (ANC < 0 /zeq L"1) or
had pH ^ 5.0 at their upper ends (Tables 6-2 and 6-3; standard errors tabulated in Appendix C).
Another 12% of these upper reach ends had pH between 5.0 and 5.5. After flowing typically
2 to 3 km to their downstream ends, 7% (772 reaches) remained acidic (ANC < 0) and had pH
< 5.5. Of particular interest in subregion 3B, because of the presence of sensitive warm water
fisheries (subsection 6.2.1), is the relatively large percentage of target reaches that have spring
baseflow pH < 6.0 (49% at the upstream node, 22% at the downstream node). This pattern is in
marked contrast with the other Mid-Atlantic and Southeast subregions, excluding Florida, where
an estimated 0 to 23% of upstream nodes and 0 to 14% of downstream nodes had pH < 6.0. The
general pattern of increasing ANC downstream in subregion 3B (scatter plot in Figure 6-2d) is
reflected in the comparatively large difference between upstream and downstream population
medians in this subregion (151 and 309 jieq L'1).
Among the other Mid-Atlantic subregions--the Poconos/Catskills (ID), the Valley and Ridge
(2Bn), and the Northern Appalachians (2Cn)~relatively few of the reaches observed were acidic
at their downstream ends. (Acid mine drainage effects are not included, but are discussed in
Section 9). Of these three subregions, only 2Cn is estimated to contain reaches acidic at their
downstream nodes that were unaffected by acid mine drainage (4% or 326 reaches). However, at
the upstream ends, between 5% and 6% (209, 636, and 499 reaches, respectively) were acidic in
subregions ID, 2Bn, and 2Cn. The generally higher downstream ANC in streams of these sub-
regions is again evident in the scatter plots, as well as in the separate upstream and down-
stream population distributions of ANC (Figures 6-2 [a, b, and c]). The more frequent occur-
rence of lowland drainages, often with limestone geology, in subregions ID and 2Bn is reflected
in their relatively high median ANC values (Table 6-1) and by the large percentage (70%) of
reaches in these subregions with downstream ANC greater than 200 /xeq L"1, compared to 38%
for subregion 2Cn.
138
-------
Poc/Catskills
N. Appalach.
Valley & Rdg.
MA Coastal PI.
S. .Blue Rdg.
Piedmont
S. Appalach.
Ozark / Ouach.
Florida
L
L
L
L
U
U
L
U
L
L
U
L
I
I
1
i ! i
l ! i-r-a^xyxaoi
• ! !
L_Ji*tf:S:;:;*} j
III ,il ,
i i i
1 i k-XvX-k-x-x-x-jx-x-x-:
I ' '
R-ix-x-x-xf) !
1 ' iy:':fcx'x':":fc'i
i i i
rpi : : i
uia i i i
LJ ItsvXv.-.v.'.".—.;.^
j ! 1
IreSx-i J |
•X'X-"-*j I j
i^i i i
i * i
i i i
*:•:•:•?:
X'X*X*I
:yx|
j
1
i
i
-W -200 200 600 1000 1400 1800 ' 2200 ' 260
(a) ANC (|ieq L'1)
Poc / Catskills ,u
. Appalach. u
L
Valley & Rdg. ~
MA Coastal PI. u
L
S. Blue Rdg. u
Piedmont u
. Appalach. u
Ozark / Ouach. u
Florida u
L
!
I
1
I
!
•
(
4
i i
i
i
i
i
i
i
t
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
t
i
i
t
i
i
i
t
!
1
i
i
5
1
i
I '
i
i
mwvx*
t-XvX-X-X
6
1
£
X-X-X'X-X-X-
1
1
B&a
•:rx-x*l
i
i
K*X-:*>X*::X
l j (x-x
i
l-xixWiyfey
t*i':*x-:-x*Sf
i J
i
1 ! fXrX-x.x-S
'i t
1
!x-:-X'X->:'X'xx:x:>i
i
r~fe
t
fc-x-x|
' It-x)
|
1 | F:*X
I! m
t
fcX:X*X-xJ
( C;::xf}x::;:::if
i
i
i
i
t
i
7
«X;3 '
XtXl i
•i-x-x-x-x-x-:-
ssssa
w>xg
i
!
-•.'!'
a
Figure 6-4.
(b) pH
Population frequency distribution (Q20, median, and Q80) of index (a) ANC and
(b) pH, at upper and lower reach nodes in the NSS-I subregions.
139
-------
.H
o
§
8
cs
01
VI
VI
VI
a>
oo t--
t-- ON ON -*
Trvocoooco—icoOol
01 vo o.cx%o\-c4%r-^
5
w
&
CO
OOOOONOlOCN^TON
mooo—^Trcsovo^
CNCOCOOlt~COOlVOOl
.-« 01 -H <^
ino* co
01 —i
* OJ* ON
r-****
CJ g co
ololcoolcoololc*-
ON
VO
ON
oo
vo
01
-H -H 01
01
o)
ON
oo
01
01
- oi
VO
CO
IN)
OO
-H 00
VO
o\
ON
ON
01
*8
||
Is
00
VO
CO
CO
CO
ON
vo
O
ed
5?
PQ
o
o
cs
r>
X!
P3
CO
m
CJ
01
X
01
I:2
o «
.e «
61
5,-H
CD
.S o
&s
to a
•88
-------
a
o.
X
!
00
•a
a
I
JS
a
a
D
•a
CB
i*
co
I
•4—»
co
«
8?
CM
O
CO
oo
C3
•4«l
CO
CO
p.
CO
CO
W co
SCO
oo
'-C C
co C8
woi
« co
2 w
*3 C3
c3 <1>
"3 S
Q, CM
O "U
AH (^
CO
I
VO
CO
g.s
.« o
CO C3
rt
CO
.8
jS^O
ffi
a
o
vd
A
O
VI
VI
2
o
A
O
>X
VI
g
I
co
MOO* OO* TftNtNOO
X-lVOtStN* * * * —i
— 1 tf)
8 -2
ON ON vo O
vo oo tN tN
*•* •—< cN tN
VO 00
>r> VO
s. s
rr" oo"
tN —i
tN co
Ci CJ-
ON to
tN '—i
vo" o"
CO tN
*/"> O tN
« tN O
t~~ r- o
**•»•»
>r> oo r-
tN -^ to
oo
CT\
SO
oo
CO
t- r~ co oo
to co
* CO -<
* vo co
-^ * CO
s
Q 8 Q
. a
*e o
»-
o
s
a>
oo
VO
01
tN
4>
2
a
x.S
09 S
X! >*
11
C8
2
g -b
15
CO Cg
O o
x 13.
» la
*o **
.S »
Jco
-
« « 2
«>
etf
w o
3JJ
I- 4fc 6^
141
-------
Population distributions of pH (Figures 6-3 [a, b, c, and d] and Table 6-2) showed a
pattern similar to that of ANC with 6% of reaches (494) in the Northern Appalachians (2Cn)
with pH £ 5.5 at the lower end. Less than 1% of target reaches had downstream pH < 5.5 in
the Poconos/Catskills (ID) and Valley and Ridge (2Bn). Considering the upper reach ends,
however, 6% to 7% of reaches in subregions ID and 2Bn had pH < 5.5. In 2Cn, 13% of the
reaches (1,133) had pH of 5.5 or less.
The geographic distributions of ANC and pH in NSS-I Mid-Atlantic sample sites are shown
in Figures 6-5 through 6-8. Acidic stream reaches were concentrated in parts of the Mid-
Atlantic Coastal Plain (3B), Northern Appalachian (2Cn), and Pocono/Catskill (ID) subregions.
Upper nodes were more commonly acidic than were lower nodes. Because various stream reaches
had different statistical sampling inclusion probabilities, and therefore different sample
weightings, caution should be used in interpreting these and other maps presented in this report.
For example, spatial sampling density was increased in areas such as the New Jersey Pine
Barrens, where a large number of acidic or low ANC sites was expected. Sample sites in such
areas have relatively lower weightings for calculating statistical population distributions.
Displaying all these sites on maps does allow greater resolution of spatial patterns in the
distribution of these low ANC streams, however, and such low ANC sites are of greatest interest
in the assessment of acidic deposition effects.
Within subregion 3B, stream reaches with ANC < 0 /*eq L'1 were found primarily in the
New Jersey Pine Barrens and the coastal plain west of Chesapeake Bay. The Pine Barrens is a
geographic area underlain by quartz sands and gravels (Forman, 1979; Lewis and Kummel, 1950).
Of 12 sample reaches visited in the Pine Barrens, 11 were acidic at one or both ends. Although
some of these streams can be classified as clearwater acidic streams (Section 9), many of them
had colored waters high in dissolved organic carbon (DOC > 4 mg L'1), suggesting that some of
their acidity may result from organic acids. About half of the acidic streams in subregion 3B,
plus many others with ANC between 0 and 15 /ieq L'1 (and pH < 5.5), were found along the
coastal plain west of Chesapeake Bay, a relatively swampy area underlain with noncalcareous
sands and clays (Cleaves et al., 1968). Most acidic sample streams in this part of subregion 3B
had high DOC (> 4 mg L'1); the potential sources of acidity in these streams are discussed in
Section 9. The highest ANC in the Mid-Atlantic Coastal Plain (3B) was found in reaches
draining agricultural basins of the Piedmont lowlands in Maryland and Virginia—areas with
highly weathered soils derived from limestone, schist, and gneiss (USDA, 1979; Commonwealth of
Virginia, 1963; Cleaves et al., 1968). ANC in these areas ranged from 200 to as high as 2,200
jieq L"1 in drainages where limestone was mapped.
In NSS-I subregions ID, 2Bn, and 2Cn (Figures 6-5 through 6-8), acidic (ANC < 0 /ieq L'1
and low ANC (< 200 jieq L'1) stream reaches were observed to be primarily upper reach nodes in
forested upland areas. In subregion ID, these included the glaciated plateaus of northern
Pennsylvania, the Catskills, the Poconos, and forested ridges of the Valley and Ridge physio-
graphic unit (Fenneman 1946) that extend into central and eastern Pennsylvania. These forested
uplands are largely underlain by sandstones, shales, siltstones, and quartz conglomerates (Cline
and Marshall, 1976; Commonwealth of Pennsylvania, 1980; Cunningham et al., 1977; New York
State Education Department, 1970). Coal is mined in many of these uplands. High ANC reaches
in subregion ID were found on glaciated plateaus with agricultural land use (largely between 200
and 500 peq L'1) and in lowland agricultural valleys (ANC ranging from 200 to > 4,000 /ieq L'1).
These valleys are part of the Valley and Ridge physiographic unit (Fenneman, 1946) and are used
intensively for both agriculture and urbanization. They contain alluvium and glacial till derived
from limestone, calcareous shales, and dolomite.
142
-------
ACID NEUTRALIZING CAPACITY
NSS PHASE I-LOWER NODE
.» 0 to 50
<$> 50 to 200
X 200 to 400
>400
Figure 6-5. Geographic distribution of NSS-I lower node index ANC.
143
-------
ACID NEUTRALIZING CAPACITY
NSS PHASE I - UPPER NODE
ANC (peqL
• £0
8 0 to 50
50 to 200
X 20O to 400
•f >400
Figure 6-6. Geographic distribution of NSS-I upper node index ANC.
144
-------
STREAM pH
NSS PHASE I - LOWER NODE
PH
£5.0
5.0 to 5.5
5.5 to 6.O
X 6.0 to 7.0
>7.0
Figure 6-7. Geographic distribution of NSS-I lower node index pH.
145
-------
STREAM pH
NSS PHASE I - UPPER NODE
• £5.0
B 5.0 to 5.5
O 5.5 to 6.0
X 6.0 to 7.0
>7.0
Figure 6-8. Geographic distribution of NSS-I upper node index pH.
146
-------
In 2Bn, the NSS-I subregion dominated by the Valley and Ridge physiographic province
(Fenneman, 1946), few acidic reaches were observed. Those that were found, as well as most of
the reaches with ANC < 200 peq L"1, were located on forested ridges. Underlying geology on
these ridges is primarily sandstone, shale, and siltstone. More highly buffered reaches, with
ANC up to 6,300 jueq L'1, were found in the valley bottoms alternating between these ridges.
These valleys, as discussed in the last paragraph, often contain soils derived from calcareous
parent material.
In the Northern Appalachian subregion (2Cn), acidic (ANC < 0 /zeq L'1) stream reaches and
low ANC (< 200 jueq L"1) stream reaches were located in forested upland drainages of the
Allegheny High Plateau in northwestern Pennsylvania and at higher elevation portions of the
Northern Appalachian Plateau, extending from Pennsylvania southward into the eastern half of
West Virginia. These upland drainages have deep, well-drained soils derived from sandstones,
shales, and siltstones. Many of these drainages have been strip-mined for coal (see maps and
discussion in Section 9). Those receiving acid mine drainage have been removed in calculations
of regional population distribution estimates and in maps presented in this section, which deals
only with the target population of primary interest for assessment of acid deposition effects.
Higher ANC (mostly 300 to 900 /zeq L"1) was found in Northern Appalachian Plateau drainages of
lesser relief that are able to support agricultural land use.
6.2.3 Interior Southeastern Region
The NSS-I Southeast Screening included sample reaches in the Southern Appalachians (2X),
the Piedmont (3A) and the Ozarks/Ouachitas (2D). Results from the NSS-I Pilot Survey in the
Southern Blue Ridge (Messer et al., 1986) are included here as results for subregion 2As of the
Southeast Region, although sampling took place in 1985, one year earlier than the remainder of
NSS-I sampling. The NSS-I included only a portion of the southeastern coastal plain (parts of
the Florida panhandle and peninsula) in its Southeast Screening. Results for Florida are
discussed in subsection 6.2.4~those from the remainder of the Southeast have been grouped into
the Interior Southeast Region discussed in the following paragraphs.
Population distribution estimates for ANC in Interior Southeast subregions 2X, 2As, 3A, and
2D are illustrated in Figure 6-2 (e, f, g, and h) and summarized in Table 6-2. Except for the
Southern Appalachians (2X), where an estimated 121 reaches (2%) were acidic (ANC < 0 /zeq L"1)
at their upstream ends, no acidic reaches (not affected by acid mine drainage) were observed in
the remaining Interior Southeast subregions. Though acidic stream reaches were rare in the
Interior Southeast, two Southeast subregions, the Southern Blue Ridge (2As) and the Ozarks/
Ouachitas (2D), had among the highest proportion of streams with relatively low ANC. In
subregion 2As, 84% of the reaches (1,703) had ANC < 200 /jeq L"1 at their upper ends; in sub-
region 2D, the proportion was 68% (2,850 reaches).
Population distribution estimates of pH in the Southeast are illustrated in Figures 6-3 (e, f,
g, and h) and summarized in Table 6-3. No sample stream reaches with spring index pH < 5.5 at
their downstream ends were observed in the Interior Southeast subregions of the NSS-I. Except
for the Piedmont (3A) subregion, where an estimated 316 reaches (4%) had pH < 6.0 at their
downstream ends, 2% or fewer (0 to 121) reaches in subregions 2As, 2D, and 2X had spring
index pH that low. In subregion 2D, 5% of the upstream reach ends (225) had pH < 5.5. In the
remaining Southeast subregions, 2% or fewer (0 to 121 reaches) had upstream pH that low.
From 5% to 11% of reaches in the 2D, 2X, and 3A subregions had pH between 5.5 and 6.0. No
upstream or downstream sampling points in subregion 2As streams had pH < 6.0 (during spring or
summer sampling in 1985).
147
-------
The geographical distributions of spring index ANC in the Southeast subregions of the
NSS-I are mapped in Figures 6-5 (downstream) and 6-6 (upstream). The majority of Interior
Southeast sites with upstream or downstream ANC < 50 /jeq L"1 were sample reaches in the East
Gulf Coastal Plain (western end of subregion 3A), upstream ends of sample reaches in the
Southern Blue Ridge Mountains (including the Great Smoky Mountains National Park in subregion
2As), and the Cumberland Plateau (a portion of subregion 2X located mostly in Tennessee). The
East Gulf Coastal Plain is a generally forested area of lower elevation and relief than the
Piedmont or Blue Ridge. Deep to moderately deep, well-drained, loamy soils overlie sandstone,
shales, clays, gravels, and lignite in the East Gulf Coastal Plain portion of the NSS-I Piedmont
subregion. The Southern Blue Ridge is largely underlain by schists, feldspar, and quartzite,
which provide little ANC upon weathering (Velbel, 1985). The Cumberland Plateau is largely
underlain by noncalcareous shales and sandstones (Hardeman, 1966a,b; Springer and Elder, 1980).
Stream reaches with ANC < 50 /zeq L"1 at their upstream ends were also observed in the O/ark
and Ouachita Mountains of Arkansas and Eastern Oklahoma. Spring index ANC values in the
Piedmont and in lowland and agricultural sites of the other Southeast subregions were primarily
in the ANC range between 200 and 400 peq L'1; stream reaches with ANC above this range were
not commonly observed. Those with ANC > 400 /*eq L'1 were observed mainly in agricultural
valleys of the Piedmont, Southern Appalachian Plateau, Arkansas River basin, and the southern
part of the Valley and Ridge Province contained in subregion 2X.
Figures 6-7 and 6-8 map the geographic distribution of spring index pH at upstream and
downstream sampling locations of reaches in the Southeast subregions. Excluding Florida (3C),
it is apparent that pH in target stream reaches over most of the region is circumneutral (pH 6.5
to 7.5). Three upstream sites with pH < 5.5 were observed, one on the Cumberland Plateau and
two in the Ozark and Ouachita Mountains. Index pH in the range between 5.5 and 6.5 was
observed mainly at upstream ends of sample reaches in the Great Smokies, the Cumberland
Plateau, the Ozarks and Ouachita Mountains, and the East Gulf Coastal Plain.
6.2.4 The Florida Subregion
The Florida subregion (3C), like parts of the Mid-Atlantic Coastal Plain (3B), stands out as
a geographic area with a relatively large percentage of acidic and low ANC streams. Like the
New Jersey Pine Barrens of subregion 3B, it also is an area with soils largely derived from
quartz sands. Again like the Pine Barrens, Florida contains many highly colored streams with
high concentrations of dissolved organic material. The potential sources of acidity in 3B, 3C,
and other subregions are discussed in Section 9.
Population percentage estimates for subregion 3C are not strictly comparable with those
from other subregions. The Florida sample was drawn from a more restrictive target population
focused only on the portion of Florida with expected ANC < 200 j*eq L'1, eliminating reaches
with expected ANC between 200 and 400 /zeq L"1 that were sampled in all the other subregions.
However, estimates of the number or combined length of stream reaches within the boundaries
of the subregion characterized by given index chemistry are comparable. Other problems with
the Florida estimates stem from the fact that the index time period, mid-March to mid-May, is
well into the dry season in Florida, relative to the other NSS-I subregions. An estimated 30%
to 40% of the target population consisted of streams that were dry, too shallow to obtain clean
water samples, or too swampy and stagnant to be considered streams in the normal sense. For
these reasons, regional extrapolations (population estimates) based on the Florida sample data
should be used with caution, though inferences regarding chemical characteristics of the sample
streams themselves are not similarly hampered.
148
-------
Population distribution estimates for ANC in subregion 3C are shown in Figure 6-2i and
Table 6-2. At the upstream ends, an estimated 678 (39%) of the Florida reaches were acidic
(ANC < 0 A*eq L'1); another 531 (31%) had ANC between 0 and 50 jueq L'1. A smaller number of
reaches were acidic at their downstream ends (225 or 14%), but a large number (670 or 43%) had
downstream ANC between 0 and 50 jueq L"1. Only 388 (22%) reaches had ANC > 200 /*eq L"1 at
the upstream ends; 292 (19%) had ANC greater than this value at the downstream ends.
Population distribution estimates for spring index pH in subregion 3C are shown in Figure
6-3i and Table 6-3. At the upstream ends, an estimated 539 reaches (31%) had pH < 5.0; another
324 (19%) were between pH 5.0 and 5.5. Index pH was generally higher at the downstream ends
of reaches, where an estimated 225 reaches (14%) had pH < 5.0 and another 146 (9%) were
between pH 5.0 and 5.5.
Figures 6-5 through 6-8 map the distribution of ANC and pH in Florida and other NSS-I
subregions. Acidic reaches were observed in both the Florida panhandle and the peninsula.
However, highly acidic reaches with ANC between -40 and -500 jueq L"1 were primarily highly
colored streams on the Florida peninsula. Acidic reaches in the Florida panhandle tended to be
of two types (see Section 9). One group contained reaches having high DOC (6 to 24 mg L'1)
and color (> 60 PCU); these were similar to the colored-water acidic reaches in the peninsula.
The second group was at relatively higher elevation and had less acidity (ANC between 0 and
-10 A*eq L'1). Water in these reaches was very dilute (conductivity < 40 juS cm'1) and clear to
weakly colored (color 10 to 60 PCU), with DOC concentrations < 2 mg L'1.
6.2.5 Interpolated Length Distributions
Subsection 6.1.7 discusses the procedure of linear interpolation by which the combined
length of the target stream resource within specified ranges of ANC and pH was estimated
(detailed in Sale [1988]). For some applications and interpretations, it is more useful to know
the distribution of chemistry on the basis of length, rather than the number of reaches. Esti-
mates based on length could also be calculated using only one or the other reach node—or they
could be calculated by averaging the upper and lower node chemistries. Distributions based upon
the average reach chemistry tend to mask or underestimate the extremes of the distributions.
Because the low end of the ANC and pH distributions have considerable importance in this
assessment, it would be misleading to present length estimates based on the average of reach
nodes. In contrast, length estimates based on one or the other node of sample reaches, as
presented in the data compendium to this report (Volume II), tend to exaggerate the extremes of
the chemical distribution if there is (as has been observed) a consistent pattern of increasing
ANC or pH in the downstream direction. The interpolated length estimates presented in this
section were based upon linear interpolation of ANC and hydrogen ion (for pH). In Figure 6-9,
the shaded fractions of each stacked bar in the diagram show the proportion of the total length
in the population of target stream reaches that is estimated to fall into specified classes of ANC
and pH. The interpolated length estimates provided (Figure 6-9, Tables 6-4 and 6-5) are not a
perfect representation of the chemical distribution, but without detailed knowledge of the
patterns of chemical change between the upstream and downstream sampling points, they are the
most reasonable estimates that can be made with the NSS-I data. The standard errors of these
interpolated length estimates (Appendix C) are approximate and were calculated from the vari-
ance of length-based (f^xj) distributions.
Population distribution estimates based on interpolated length (Figure 6-9 [a and b])
followed regional patterns similar to those described based on numbers of upper or lower reach
nodes. Because many lower and upper sampling locations could not be sampled in Florida (see
149
-------
(a) NSS-I ANC frieq l_-1) DISTRIBUTION
2Cn 2Bn 3Bn 2As 3A
SUBREGION
2X 2D 3C
(b) NSS -1 pH DISTRIBUTION
ID
2Cn 2Bn 3Bn 2As 3A
SUBREGION
2X 2D 3C
Figure 6-9.
Combined length of NSS-I target stream resources within specified categories of
(a) ANC, and (b) pH, as estimated by linear interpolation of index concentrations
between upper and lower reach node sampling locations.
150
-------
O —icNOONCO—itNcO
•-i CN co rr co CN fN
00
oo r- vo vo —< ON vo
or-Oirit^-voooo
oo^ oo^ tN^ ON^ ON^ t-^ co^ ON
' oo o\ a\ -^ o\ v) t^
VO
tNCNOOOcOTj-VOX"*
TfcotNC-cOtNvo-H
CO CN
00 CN
oo cN
tn r~- vo so oo oo
cooNO—Tvo"-H>>X'Tr'
0
—I s~^ VO s-*,.
CN o
, tN
ON
fNONTONVOOVO>r>>r)
vOooiOOOON^OON
—< CN —i r-
vq CD oo co
tN
*-*. c>
o ^1
CO
rr
—< tN
* *
•iA a .y E 5
yj •*•< ^> MH (-•
vo oo
oo^ r~
oCo"
o ON
—< tN O
v> >r>
ON oo
co oo
tN fN
CO l^-
o
vo
" oo
CO CO
CO >r> ^ TT
t-* ^ I-H vc5
VO tN
—i -*r
CN
J2-3, 3,2,
rr r-- —< oo
tN -H in r-
co —i oo v)
o |UJ)
•S co
g^
r-J co
<: S
si
ss
fN
V>
o"
O
CN
v
r--
tN
CO
o
VO
CO
ON
CO
VO,
co"
r-
0
ON
tN
ON
CN
v-
3
2
a
£s
•S ^**
n> LJ
a «
a m
J2
3
o
o
03
3 co
f U
• P O
I—H ^« ***
c-0 °
-------
DO
Jo-
oo
CO
0)
00
C9
oo
a
t-<
0>
vo
oo
VI
V
VI
U-l
VI
00
n
CN
oo
ON
o\
ON O CN O
CO vo O CN
~ CN ".
^^s^^x^* ON
"! -! -1 vi
CN/CN-lCxO'
vo vo o oo '
»— 1 VO
v v oo o
CN
CN
COOVO
v
ir> CN
ir> rr cN — <
— i co
* * * *
CN
CN
O
v c *o .
- ^
O
o «
M -a -ff 28
•o s .y ^-, o s
c «a M « ~ O
o
o < ^J
CT\ ON
vo co
vo oo
10 CN
Wi c7\
O —i
vo oo
O CN
oo O
-H CO
OO CN
CN
CN
CN
W
•< CD
oo
C3 ej «
—tcNCNcOCNcOCNcNco
VO 00
oc
o ON
>r> o
VO^CN^
-H CN" CN"
CO OO
vo
vq oq -H —i
co O oo CN
00 CN
r-~ CN
o pa
'£ 00
CN
W-l
VO
CN
CN
CN
VO
o\_
v> oo
•
-------
Section 5.2), an accurate estimate of the length of acidic streams in subregion 3C cannot be
provided. However, an incomplete estimate based on a reduced sample of matched upstream-
downstream reach node pairs indicates that approximately 12% to 14% of the combined length of
such reaches may be acidic (ANC < 0 /zeq L'1) or may have pH < 5.0. Another 49% may have
ANC between 0 and 50 /xeq L"1. It should be kept in mind that the length of the total resource
represented by the matched pairs in this analysis is only about three-fourths of that in the
initial target population identified on maps in subregion 3C.
The greatest combined length of target stream resources with ANC < 0 jueq L"1 was esti-
mated for the Mid-Atlantic Coastal Plain (3B) (2,527 km or 6.3%) and the Northern Appalachians
(2Cn) (1,524 km or 7.0%), as summarized in Table 6-4. Respectively, 7.8% and 6.6% of the total
target stream length in these two subregions had pH < 5.0 (Table 6-5). The total stream length
with ANC between 0 and 50 jueq L'1 was considerable in these two subregions, with 7,109 km
(18%) in subregion 3B and 2,189 km (10%) in subregion 2Cn.
Of the remaining subregions, only in the Poconos/Catskills (ID) did more than 1% (543 to
550 km or approximately 3.6%) of the target stream length have ANC < 0 /zeq L"1 or pH < 5.0.
Acidic (and pH < 5.0) water was estimated in less than 1% of the combined length of target
stream resources in the Valley and Ridge (2Bn), Southern Blue Ridge (2As), Southern Appalachian
(2X), Piedmont (3A), and Ozark/Ouachita (2D) subregions. Between 6% and 8% (706 to 2,390 km)
of the combined length of target reaches in subregions ID, 2Bn, 2As, and 3A had ANC between
0 and 50 jueq L'1, whereas smaller proportions of the combined reach length in subregions 2X
(2.9%) and 2D (0.9%) were in this very low ANC category.
The highest percentages of stream length in the 50 to 200 ^eq L"1 ANC category were
estimated in two subregions where no acidic stream reaches were observed, the Southern Blue
Ridge (2As) (71% or 6,378 km) and the Ozarks/Ouachitas (2D) (66% or 14,887 km). Florida had
the least stream length in this ANC range (15% or 584 km), although the estimates do not apply
to the entire state. Considering the remaining NSS-I subregions, the percentage of stream
length in this ANC category ranged between 25% (5,367 km) in subregion 2X to 42% (9,222 km)
in subregion 2Cn, with between 9,000 and 12,000 km in most of the remaining subregions.
6.3 DISTRIBUTIONS OF OTHER CHEMICAL VARIABLES
6.3.1 Base Cations and Conductivity
There were no data quality problems that would affect data interpretation of any of the
constituents making up the sum of base cations (SBC). System decision limits (SDL) were all
< 1 jueq L'1, indicating no background contamination problems (Table 4-1). Percent accuracy
estimates for all base cations were < 4% (Table 4-2). Among-batch precision estimates, the most
conservative (largest) measure of this attribute of data quality, were < 5 /zeq L"1 for all base
cations (Table 4-3).
Examination of distributions of base cations and conductivity in the NSS-I subregions
suggests that weathering rates in the Mid-Atlantic Region and the Southern Appalachians sub-
region are substantially higher than those in the remaining subregions of the southeastern
United States (Figure 6-10). The SBC is calculated by adding the concentrations of calcium,
magnesium, sodium, and potassium (equivalent basis). Conductivity values measured in the
analytical laboratories were used to make the population estimations presented in this report.
Downstream medians of both variables were higher than upstream medians in all subregions,
indicating that water at downstream locations has had more time in contact with watershed
rock, soils, and vegetation.
153
_
-------
Poc/Catskills
N. Appalach.
Valley & Rdg.
MA Coastal PI. H
S. Blue Rdg.
Piedmont
S. Appalach.
Ozark/Ouach. H
Florida
Poc/Catskills ?
N. Appalach.
Valley & Rdg.
MA Coastal PI. "
S. Blue Rdg.
Piedmont
S. Appalach.
Ozark/Ouach. H
Florida
u
L
U
L
U
L
U
L
U
L
U
L
U
L
U
L
U
L
PEal
C3
C
1
L
1 — i
I
i !
:i
: t&s
I !
t
1
i :
! 1
1
I
1
3 !
i
f
foX*! ;3
i
« fx-
1 k
1
1
lx#&x<1
|x':'x'x'&':3
i
1 — fSS£fxl
i
i
tiii
! | ! i
fcX;X:tXx::X:X:X;{v:x:x-::^
i,i i i
w.:.x*x.x5-X'X<*X'Xv.-.-.w.v.'.-.v.v,y.y,v^; .
i i t i
i i i i
i i i i
t*3x-X'X;X;X;i;;X'| ! i
i f:XxX;:;:i::::Xx::%x:;>:X:X;^
i i i i
i i i i
i i i i
i i t i
i t i i
ill!
i ! i i
i i i i
i i t i
i i i i
i i i i
SCWK-WCWC^MC-I-W-CSSf^WiS^^iSSft^X::^
1 1 " - - - J-- - - - *
1 1 1 1
1 1 1 1
1 1 1 1
1 1 1 1
1 1 1 1
1 1 1 1
tiii
till
1 t 1 1
1 1 1 1
.
x.x-x-x-x-
-~-^.^^_
Tl
1
1
1
..... ,-.——.)
3 SO 10O 150 200 250 301
(a) Lab Conductivity (n-Scrrr1)
u
L
u
L
u
L
u
L
u
L
u
L
u
L
U
L
U
L
CE
D
i
l
! (
| l
i
t
: i
• tsv*xy
1 !
;
* l
i
3 i
m :
i
i
tvfX-Xl
1
1 1
1 !
i
i
fcrX*X:X*X'3
1 f&xjxwx*]
l
l!X!>«X*>Xv|
E&&&&&&I
i
i i
J ! j
i i
i i
r , •
x-:-x-x-x£*:-xi ;
i i
i i
i i
i i i i i
i i i i i
!^x•x•x•^K*xX-XvX%v
i i
l i
i i
l l
i i
l i
i i
l l
l l
i i
1 1
f'X*X*.v>.v.*.*.'.v.*.^v.;.X'X*X
: :
i i
i i
i i
i i
i t
i i
i i
i i
* i
i i
fe&vi ! i
• i
i i
i i
i i
i i
l |
! !
X*X*>X-X'X'XvX*>X*X-M*M:
S'fftvX-x-X'X-X'X-xt'X*:*:*:
i i
: : :
: :
1 ! ! .
i ! !
i i i
i : :
t i i
i
Iv.vX'.v
1
.V.V.'.V.*.
1 . . . 1
t
1
j
0 400 BOO 1200 1600 2000 2400 280
(b) Sum of Base Cations (neqL"1)
Figure 6-10. Population frequency distribution (Q2Q, median, and Qgo) of index (a)
conductivity and (b) base cation sum, at upper and lower reach nodes in the
NSS-I subregions.
154
-------
SBC and conductivity were highly correlated (r = +0.95 to +0.99 in seven subregions), which
is not surprising because both are good indicators of total ionic strength. As a result, the two
variables also exhibited very similar regional patterns (compare box plots in Figures 6-10 (a and
b). High subregional median SBC concentrations (445 to 981 jueq L"1) with large variation (80th
percentile - 20th percentile ranges largely > 1000 jueq L'1) were found in all Mid-Atlantic
subregions and in the Southern Appalachians (2X) in the southeastern United States. All other
Southeast subregions had medians < 425 /teq L"1 with lower variation (subregion 80th percentile
- 20th percentile ranges largely < 400 jueq L'1). SBC concentrations were generally the lowest
in the Southern Blue Ridge (upstream and downstream medians approximately 200 jueq L"1). The
Florida (3C) and the Ozark/Ouachita (2D) subregions also had low median SBC concentrations
(approximately 300 /zeq L'1).
Only the Southern Blue Ridge (2As) had median conductivities < 25 /zS cm'*, a value similar
to that of rainfall. Subregion medians in the Mid-Atlantic (ID, 2An, 2Cn, and 3B) and the
Southern Appalachians (2X) fell between 54 and 107 i*S cm'1. Median conductivities in the
remainder of Southeast subregions were all < 40 /zS cm'1.
6.3.2 Aluminum
The highest median total (unf iltered) aluminum concentrations were found in the Ozark and
Ouachita Mountains (11 to 12 /iM), Florida (7.3 to 7.7 /iM), the Piedmont (6.8 to 7.8 /iM), and
the upper ends of reaches in the Mid-Atlantic Coastal Plain subregion (3B) (6.7 pM) (Figure
6-11 a). All other upstream and downstream medians were < 5 /*M. More than 80% of the
upstream and downstream ends of reaches in all subregions except the Valley and Ridge (2Bn)
had total aluminum concentrations above the system decision limit (SDL) of 0.027 mg L"1 (1.00
/*M), the value above which field sample concentrations are statistically distinguishable (p < 0.05)
from blanks. At concentrations less than the SDL, measurements of aluminum variables were
very imprecise (%RSD of 50% to 100%), as shown in Table 4-3. Even at concentrations greater
than the SDL, precision for the various aluminum components was not good (%RSD of 10% to
30%). This imprecision should be kept in mind when interpreting the NSS-I aluminum data.
The MIBK-extractable component of total aluminum is generally interpreted to be a measure
of the sum of organic and inorganic monomeric aluminum (Driscoll, 1984). MIBK-extractable
aluminum concentrations were generally low in target reaches of all subregions except Florida
(3C) and the Mid-Atlantic Coastal Plain (3B) (Figure 6-lib). In the other seven subregions,
roughly 60% of the reaches had concentrations below the SDL of 0.30 /zM, meaning the total
monomeric aluminum concentration in these samples was statistically indistinguishable from that
of deionized water (p < 0.05). In the Southern Blue Ridge (2As), more than 80% of the stream
reaches were below this concentration. In contrast, more than 90% of the stream reaches in
subregion 3C, and 75% of the stream reaches in subregion 3B had extractable aluminum concen-
trations > 0.30 i*M. These concentrations were markedly higher at the upstream reach ends in
these two subregions, an exaggeration of the pattern observed in pH and ANC. Subregional
median concentrations at the upstream ends of reaches in subregions 3C and 3B were respec-
tively, 2.3 and 1.5 /zM, which is 5 to 20 times those observed in the other subregions. Approx-
imately 20% of the reaches in these subregions had extractable aluminum concentrations > 6 fjM.
Comparison of total aluminum concentrations with measurements of pH, MIBK-extractable
aluminum, DOC, and turbidity suggests a possible interpretation of these results. Total aluminum
was positively correlated with turbidity (r = 0.97) and color (r = 0.90), but not with pH or DOC
in the Ozark/Ouachita (2D) and Piedmont (3A) subregions. Also, MIBK-extractable aluminum
concentrations were low in these subregions. It appears likely that high total aluminum in these
155
-------
Poc/Catskills v
N. Appalach.
Valley & Rdg.
MA Coastal PI. "
S. Blue Rdg. "
Piedmont "
S. Appalach. "
Ozark/Ouach. u
Florida • u
1
rm i
i
1 |x<::-::x*x-:-:-:-:-3
1 fX'*x*x*x-xi
i
i
1 t&xWxtox-x
r i
risa!
i i
; i • h-x-x-s
! 1 i ix-x*
! !
i I H*x*g*:is}
I 1 l*:*:-:*:*:*t
I
1
1
1
sasj
x*x*:-3
(>X-X*X*.*.M.v:
fcx*:*:-
t-X-X'X'.v.
x-x-x-X'X'X*
•x-x-x-x-1
£3
:•»:•:•>:•>:•»:•
tt»M«av-v
•X-X-Xxl
•:->X'X;X:X:X
•:%-'.':««-i
•:*:*M
.
__ — —.,...,.
1
i
i -....i
0 5 10 15 20 25 30 35 40 45 50
(a) Total Aluminum
Poc/Catskills ^
N. Appalach. "
Valley & Rdg. "
MA Coastal PI. "
S. Blue Rdg. "
Piedmont J|
S. Appalach. J^
Ozark/Ouach. "
Florida }|
l
i
i Eftixa
i
i Ix^j
i
1
cm\
i
1 !
i : is
i
ie !
03 j
r^i
c^ i
Km \
i i
i i r-
1 f— -
i i
SSS&ftx
^X'X-X-X-
fc-:¥:*x
•:-x-x-x-:
KvT
x-x-x-x-
•x-x-x-x-
XvXrX:::?
¥:-x-x-:::-
X-XvX-X
;XvXx->
>x-r*.-.-.-.-
^
.".".V.V.V.tt
1
-v-.-^^_
.*.".•.*.•.•.%•.
rXSTXrXS
'
:-X:X-X-X
;:v::::.J
1 . .1 . .
1
1
1
1
.,. . -.1- 1
O' i 234567
(b) Extractable Aluminum (|iM)
Figure 6-11. Population frequency distributions (Q20» median, and Qgo) of index (a) total
aluminum and (b) total monomeric aluminum (MIBK-extractable) concentrations
at upper and lower reach nodes in the NSS-I subregions.
156
-------
areas is associated with the suspension of particulate and colloidal aluminum derived from kaolin
and other aluminum-rich clays or silts contained in the soils of these regions. In contrast, the
MIBK-extractable aluminum concentration was relatively high, and made up a greater fraction of
the total aluminum concentration in subregions 3C and 3B. In addition, the total aluminum
concentration was positively correlated with DOC and negatively correlated with pH in these two
subregions, suggesting that aluminum is largely in the form of dissolved and colloidal inorganic
aluminum and organo-aluminum complexes.
Inorganic monomeric aluminum concentration was defined as the difference in aluminum
concentration (measured by the pyrocatechol violet technique) before and after passing the
sample through a strong cation exchange column. It is this inorganic monomeric component of
total aluminum concentration that is of greatest concern because of its toxicity to fish (Driscoll
et al., 1980; Baker and Schofield, 1982; Gagen and Sharpe, 1987) and aquatic invertebrates
(Witters et al., 1984; Havas, 1985). Free aluminum ions at concentration thresholds between 0.05
and 0.2 mg L'1 (1.9 and 7.4 /zM) have been demonstrated to reduce growth and survival of vari-
ous species of fishes (Muniz and Leivestadt, 1980; Baker and Schofield, 1982; Muniz et al., 1987).
Spring index concentrations of inorganic monomeric aluminum (Al/im\) > 0.20 mg L'1 in the
NSS-I were largely restricted to upper reach nodes of the Mid-Atlantic subregions (3B, 2Cn,
2Bn, and ID) and lower reach nodes in the Florida (3C) subregion, where between 2% and 5% of
the reaches had concentrations above this reference value (Table 6-6). Concentrations greater
than 0.10 mg L"1 were not observed in any of the Southeast subregions except Florida. The
Mid-Atlantic Coastal Plain (3B) had the largest percentage of both upper (13%) and lower (6%)
reach nodes with Al/im\ concentrations above 0.10 mg L"1. Two percent or more of the upper
reach nodes in all subregions except the Southern Blue Ridge (2As) and Piedmont (3A) had
Al(im) concentrations above the lowest threshold for adverse effects on fisheries (0.05 mg L'1).
Mid-Atlantic upper nodes above this concentration ranged from a low of 2% in the Valley and
Ridge subregion (2Bn) to a high of 30% in subregion 3B. Subregion 3B also had a high per-
centage of lower nodes (15%) above the 0.05 mg L~^ reference value. Florida had a greater
number of lower (18%) than upper nodes (7%) above the 0.05 mg L~l reference value.
6.3.3 Sulfate
There were no data quality problems in the measurement of sulfate that would affect
interpretation at the concentration levels observed in the NSS-I. The SDL was 1 /*eq L"1,
indicating no substantial background or detection problems (Table 4-1). Likewise, precision
presented no problem for data interpretation, as all types of precision measurement had %RSD
< 6% (Tables 4-2 and 4-3).
Comparing subregional population distributions (Figures 6-12 [a] and 6-13 [a - i]), and
medians (Table 6-1), and examining the mapped geographic distribution of sulfate in Figure 6-14,
it is readily apparent that the four Mid-Atlantic subregions have substantially higher (and more
widely ranging) concentrations than those in the Southeast subregions. Mid-Atlantic subregion
medians ranged from a high of 238 jueq L"1 for the downstream ends of reaches in subregion ID,
the Poconos/Catskills, to a low of 125 /*eq L"1 at the upstream ends of target reaches in the
Mid-Atlantic Coastal Plain (3B). Medians in the Southeast ranged from 10 jteq L"1 at the
upstream ends of Florida (3C) reaches to 71 /xeq L'1 at the downstream ends of Southern
Appalachian (2X) reaches.
157
-------
Table 6-6. Population Estimates of the Percentage of Target Stream Lower (L) and Upper
(U) Reach Nodes with Spring Index Inorganic Monomeric Aluminum Above
Reference Concentrations
INORGANIC MONOMERIC ALUMINUM CONCENTRATIONS*
> 0.05 mg L'1
SUBREGION
ID
2Cn
2Bn
3B
2As
3A
2X
2D
3C
(Poconos/Catskills)
(N. Appalachians)
(Valley and Ridge)
(MA Coastal Plain)
(S. Blue Ridge)
(Piedmont)
(S. Appalachians)
(Ozarks/Ouachitas)
(Florida)
(> 1.85
L
*
4
*
15
*
*
*
2
18
>0.10
mg L'1 > 0.20 mg L'1
MM) (> 3.71 MM) (> 7.41 MM)
U
7
8
2
30
*
*
2
2
7
L
*
4
*
6
*
*
*
*
5
U L U
3*2
6 24
2
13
*
*
*
*
2
4
*
*
*
*
4 5 *
Less than 1%.
Based on Pyrocatechol violet technique and use of a cation exchange column for all
Subregions except 2As, where it was based on the MIBK technique and use of a cation
exchange column.
158
-------
Poc / Catskills "
N. Appalach
Valley & Rdg. "
MA Coastal PI.
S. Blue Rdg.
Piedmont
S. Appalach.
Ozark/Ouach. "
Florida
u
L
U
L
U
L
U
L
U
L
U
L
U
L
U
L
U
L
CE5x1
1 ff
cn
1
1
i i
i i
t i
i i
i
i c,
1
t
li
l-i •
t
1
i i m*
1 i
t
i
i
l
3 :
-1 t
i
3 '•
eJ t
n t
a .
i
i
t:::x:x->x*>1
te&&&fx*x3
i
i
t
i
i
o 100
: i :
1 ! |::x:::x:::jx:::x:::::x:xf;x::;x|
1 ! IXfcXxXxWxlXxl
i i i
i i i
Px:<::x¥x:x:::>*-x-:-x-x<-xj:-:-XvX-x-:
i i i
K-X*X-X-X?*X-X-X'| i ,
X:XK:xX;XiX:::X:::::X:X|X:::.x:x-x-x>::X'X::i
I j j
1 1 1
t t 1
1 1 1
! ! I
i i <
i i i
i i t
iiii
|lll
| 1 t 1
lilt
frXvj ' * '
ill*
lit'
i i * l
' * i
l l * J
iiii
iii:
Wxi
1 ...-,.
2OO 300 10O 500 60
0 100 200 300 400 500 60C
(a) Sulfate (ujsqL-1)
Poc / Catskills {^
N. Appalach. j-1
Valley & Rdg. "
MA Coastal PI. u
S. Blue Rdg. ^
Piedmont ^
S. Appalach. "
Ozark/Ouach. "
Florida u
t * i * ' '
t::::!X:Kx:x:xx?:::f:::^ ; ; j ;
i i i i i i
1 ! tswi ill!'.
i ma i i i i i
1 i E'" ] I \ ', :
txyx:fx:::x:x:::x:!::yx::yx.s^^
rwm j i j i | j
j i i j j i
i i i ' ' J
!| |ix^x'x;xyx3; J J 1 !
i , frK-KW^Hfr^ • » i i i
! ! ! I i ! !
m i i i i
El 1 1 1 ' ' 1
E3 i 1 1 1 1 l
1 1 1 J | |
l i i i ! i i
! I ! I i j
.
,
0 so 100 150 200 250 3O
(b) Nitrate (u.eql_-1)
Figure 6-12. Population frequency distribution (Q20> Median, and Qgo) of index (a) sulfate,
and (b) nitrate concentrations at upper and lower reach nodes in the NSS-I
subregions.
159
-------
SULFATE
POCONOS / CATSKILLS
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
80*
60*
MEDIAN
40*
20r.
MAX
MEAN
STD
SAMPLE(n)
POP EST(N)
SE= X
DOWNSTREAM POPUIA1TON
MUUBCR OF REACHES
UPPER«S% CL
PROPQBT10H >= X
ISO 20O 2SO
SULFATE (peq L'1)
Figure 6-13a. Population distributions and comparison of index sulfate concentrations at upper
and lower reach nodes in NSS-I subregion ID (Poconos/Catskills), presented as
inverse cumulative proportions.
160
-------
SULFATE
NORTHERN APPALACHIANS
STATISTIC UPPER LOWER
(ESTIMATE) NODE NODE
UPPER VS LOWER SAMPLES
WIN
BO'.
60%
MEDIAN
4-0%
205!
MAX
MEAN
STD
72.6
135.0
187.8
208.9
242.4
377.4
2217.3
404.5
520.3
74. 1
138.8
181 .0
220.0
270.6
552.2
2383.9
471 .6
572.9
SAMPLE(n) 67 61
POP EST(N) 8663 8488
SE(N) 807 814
a
g
o
3OO «O6 «OO
LOWER NODE SULFATE
UPSTREAM POPULATION
NUMBER Of REACHES
UPPER »5*CL
PROPORTION >= X
ISO 2OO 2SO
SULFATE (petit.-1)
DOWNSTREAM POPULAT1OH
NUUBtRCr REACHES
WPER»S%0-
PROPORTION >= X
1SO ZOO
SULFATE
Figure 6-13b. Population distributions and comparison of index sulfate concentrations at upper
and lower reach nodes in NSS-I subregion 2Cn (Northern Appalachians), pre-
sented as inverse cumulative proportions.
161
_
-------
SULFATE
VALLEY AND RIDGE
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
805!
SOr.
MED I AN
40K
20*
MAX
MEAN
STD
SAMPLE(n)
POP EST(N)
SE(N)
UPPER
NODE
28.6
92.7
149.3
182.4
213.7
389.0
809.5
246.9
195.6
44
13038
1249
LOWER
NODE
18.0
91.4
145.2
185. 1
222.7
395.4
1262.3
278.0
266.4
47
13992
1213
a
o
e «•>
200 *oo AQQ aoo taaa
LOWER NODE SULFATE
UPSTREAM POPULATION
NUMBER Of REACHES
UPPER »5ZCL
PROPORTION >= X
SULFATE (peqLT1)
DOWNSTREAM POPULATION
NUMBER OF REACHES
UPPER9SSCL
PDOPORTION >= X
1SO 200
SULFATE
Figure 6-13c. Population distributions and comparison of index sulfate concentrations at upper
and lower reach nodes in NSS-I subregion 2Bn (Valley and Ridge), presented as
inverse cumulative proportions.
162
-------
SULFATE
MID-ATLANTIC COASTAL PLAIN
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
80*
60*
MEDIAN
4055
20%
MAX
MEAN
STO
SAMPLE(n)
POP EST(N)
SE(N)
UPPER
NODE
24.
76.
1 16.
125.
150.
237.
883.
173.
137.
57
1 1284
1078
7
3
3
1
5
9
9
0
2
LOWER
NODE
14.8
74.3
139. 1
150.5
198. 1
328.8
760. 1
217.0
171.9
58
1 1287
1078
LOWER NODE SULFATE
UPSTREAM POPULATION
NUMBER OF REACHES
LPPER»5ZCL
PROPORTION >= X
150 2OO
SULFATE
DOWNSTREAM POPULATION
NUMBER OF REACHES
UPPER95XCL
PSOPORTIOM >= X
SULFATE (ueqL'1)
Figure 6-13d. Population distributions and comparison of index sulfate concentrations at upper
and lower reach nodes in NSS-I subregion 3B (Mid-Atlantic Coastal Plain),
presented as inverse cumulative proportions.
163
-------
SULFATE
SOUTHERN BLUE RIDGE
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
60*
60x
MEDIAN
40*
20*
MAX
MEAN
STO
SAMPLECn)
POP EST(N)
SE(N)
UPPER
NODE
6.
10.
15.
20.
22.
44.
150.
28.
26.
54
2031
326
2
2
4
7
8
3
3
5
1
LOWER
NODE
12.9
18.6
22.4
22.9
27.6
59.4
184.4
39. 1
34.2
54
2031
326
o
S «= X
SULFATE ((jeqLT1)
Figure 6.1-13e.
DOWNSTREAM POPULATION
NUMBER OF REACHES
UPPER 95% CL
PROPORTION >= X
SULFATE (ueq L'1)
Population distributions and comparison of index sulfate concentrations at
upper and lower reach nodes in NSS-I subregion 2As (Southern Blue Ridge),
presented as inverse cumulative proportions.
164
-------
SULFATE
PIEDMONT
STATISTIC
(ESTIMATE)
WIN
805!
60%
MEDIAN
4-0%
205!
MAX
MEAN
STD
UPPER VS LOVER SAMPLES
UPPER
NODE
12
19
29
37
41
57
158
44
32
SAMPLE(n) 47
POP EST(N) 7515
SE(N) 650
LOWER
NODE
16.9
21 .9
34.7
38.3
46.3
61 .7
257.5
53.3
46.8
47
7515
650
s
9
I *°°
aoo *oo coo BOO
LOWER NODE SULFATE
UPSTREAM POPULATION
NUMBER OF REACHES
UPPER 05% CL
PROPORTION >=X
ISO 2QO
SULFATE
ISO 2OO
SULFATE
DOWNSTRUM POPULATION
NUMBER OF REACHES
WPER95XCL
PROPORTION >= X
Figure 6-13f. Population distributions and comparison of index sulfate concentrations at upper
and lower reach nodes in NSS-I subregion 3A (Piedmont), presented as inverse
cumulative proportions.
165
-------
SULFATE
SOUTHERN APPALACHIANS
UPPER VS LOWER SAMPLES
STATISTIC
(ESTIMATE)
MIN
80*
60x
MEDIAN
40*
20*
MAX
MEAN
STD
UPPER LOWER
NODE NODE
8.5
37.4
54.8
61 .0
68.3
116.4
19.2
43.6
63.3
70.6
81 .0
173.4
2644.1 2852.3
155.9 244.0
408.4 519.4
SAMPLE(n) 39 40
POP EST(N) 4936 5057
SE(N) 529 526
2OO *O6 GOO BOO
LOWER NODE SULFATE
'\
UPSTREAM POPULATION
NUMBER OF REACHES
UPPER 95ZCL
PROPORTION >= X
SULFATE (ueq L'1)
DOWNSTREAM POPULATION
NUMBER OF REACHES
UPPER 95% CL
PROPORTION >= X
SO 1OO ISO 2OO 2SO 3OO 3SO -+OO
SULFATE (peqLr1)
Figure 6-13g. Population distributions and comparison of index sulfate concentrations at upper
and lower reach nodes in NSS-I subregion 2X (Southern Appalachians), presented
as inverse cumulative proportions.
166
-------
SULFATE
OZARKS / OUACHITAS
STATISTIC UPPER LOWER
(ESTIMATE) NODE NODE
UPPER VS LOWER SAMPLES
MIN
80*
60%
MEDIAN
4055
205!
MAX
MEAN
STD
20.6
49.0
62.2
69.9
81.4
123.6
473.7
102. O
89.5
45.0
52.9
60.5
64.5
78. 1
122.7
801 .6
127. 1
160.4
SAMPLE(n) 49 48
POP EST(N) 4204 4116
SE(N) 406 410
j
a
960 *OOC AOO BOO
LOWER NODE SULFATE
1
UPSTREAM POPULATION
NUMBER OF REACHES
UPPER 95X O.
PROPORTION >= X
1SO , 2OQ
SULFATE
DOWNSTREAM POPULATION
NUMBER OF REACHES
UPPER KK d.
PROPORTION >= X
SULFATE (ueqC1)
Figure 6-13h. Population distributions and comparison of index sulf ate concentrations at upper
and lower reach nodes in NSS-I subregion 2D (Ozarks/Ouachitas), presented as
inverse cumulative proportions.
167
-------
SULFATE
FLORIDA
STATISTIC
(ESTIMATE)
MIN
UPPER VS LOWER SAMPLES
605!
MEDIAN
40*
20*
MAX
MEAN
STD
SAMPLE(n)
POP EST(N)
SE= X
100 150 200
SULFATE
A
SULFATE
DOWNSTREAM POPULATION
NUMBER OF REACHES
UPPER 95* CL
PBOTOKTION >= X
Figure 6-13i. Population distributions and comparison of index sulfate concentrations at upper
and lower reach nodes in NSS-I subregion 3C (Florida), presented as inverse
cumulative proportions.
168
-------
SULFATE CONCENTRATION
NSS PHASE I - LOWER NODE
Figure 6-14. Geographic distribution of NSS-I lower node index sulfate concentrations.
169
-------
The potential sources of sulfate in stream water are wet and dry atmospheric deposition
(natural and anthropogenic sources), nonatmospheric anthropogenic sources (e.g., agriculture),
and the dissolution of sulfur-bearing minerals. Crude approximations of steady state sulfate
concentrations for streams with no internal sulfate sources can be calculated using approximate
concentrations of sulfate in precipitation within the various subregions and approximate regional
values for the ratio of precipitation to runoff to account for evapoconcentration (Table 6-7).
The figures in parentheses in column 3 represent a reasonable range of stream sulfate concen-
trations that could be attributed to atmospheric deposition. They were calculated using the
range of available regional precipitation, runoff, and sulfate deposition data. The method of
calculation and sources of information are cited in the table. The middle figure is a best esti-
mate of what a typical (perhaps a median) stream water concentration might be for each sub-
region, assuming no internal sources or sinks (i.e., sulfate is acting conservatively; mass input
equals output and sulfate is simply concentrated by evapotranspiration). Streams with index
sulfate concentrations less than the lower end of the predicted sulfate range almost certainly
have substantial sinks for sulfate. Those with sulfate greater than the high end of this range
almost certainly have substantial internal sources of sulfate. If the median in a subregion is
above the best estimate, it does not imply that 50% of the stream reaches have substantial
internal sulfate sources. An elevated median sulfate concentration does suggest, however, that
internal sources are likely in some streams, particularly if the distribution of measured stream
water sulfate includes concentrations above the high value estimate for predicted stream water
sulfate.
The most striking result to be gleaned from comparing Table 6-7 with the cumulative fre-
quency distributions of stream water sulfate in the nine NSS-I subregions is the large difference
between most Mid-Atlantic and most Southeast subregions with regard to apparent retention or
export of sulfate. The overall pattern appears to be that of sulfate retention in watersheds of
some Southeast subregions (Southern Blue Ridge, Piedmont, and Florida) and concentrations near
steady state in many streams within the Interior Mid-Atlantic subregions. These results, except
for Florida, support the hypothesis advanced by Rochelle and Church (1987) that the deeper,
more highly weathered soils in unglaciated areas of the southeastern United States are more
highly retentive of sulfate from atmospheric deposition than are those in the mid-Atlantic and
northeast, many of which are generally younger and shallower, and most of which have been
subjected to high rates of atmospheric sulfate deposition for longer periods of time.
The relatively wide range of sulfate concentrations found in many subregions reflects the
heterogeneity of watershed characteristics controlling watershed sulfate dynamics (hydrology,
sources, and sinks), as well as local differences in sulfate deposition and precipitation. This
heterogeneity is particularly evident in the four Mid-Atlantic subregions plus the Southern
Appalachians (2X) and the Ozarks/Ouachitas (2D). Even though many streams in the Mid-
Atlantic subregions appear to have sulfate concentrations roughly approximating steady-state
concentrations (mass input equals output, as very crudely approximated by comparing spring
baseflow with annual average precipitation sulfate concentrations), there is evidence for both
sulfate retention and internal sources of sulfate in many other streams within these same sub-
regions. Substantial sinks for sulfate are apparent in about 1% to 2% of the Poconos/Catskills
(ID) streams, 8% to 11% of the streams in the Northern Appalachians (2Cn), and about 20% of
the streams in the Valley and Ridge (2Bn) subregions. Similarly, substantial amounts of sulfate
appear to be retained by approximately 30% of the stream reaches in the Ozarks/Ouachitas (2D)
and by 40% to 57% of the streams in the Southern Appalachians (2X) and the Mid-Atlantic
Coastal Plain (3B).
170
-------
Table 6-7. Sulfate Concentration in Deposition and Best Estimates and Range of Precipitation/
Runoff Ratios and Predicted Stream Water Sulfate Concentrations (Assuming Only
Subreeion
ID (Poconos/Catskills)
2Cn (N. Appalachians)
2Bn (Valley and Ridge)
3B (MA Coastal Plain)
2As (S. Blue Ridge)
3A (Piedmont)
2X (S. Appalachians)
2D (Ozarks/Ouachitas)
3C (Florida)
1986 Volume*
Wgt. Wet
Deposition Precipitation®
[SO42-] /Runoff
(uea L'1} Ratio
57-67
62-73
62-73
52-62
36-46
31-41
36-46
21-31
16-26
1.70
(1.03-3.46)
1.84
(1.24-2.89)
2.11
(1.10-3.77)
2.56
(1.90-3.85)
2.30
(1.00-6.38)
2.71
(1.08-5.33)
2.36
(1.19-4.07)
2.45
(1.79-4.08)
3.18
(1.56-7.00)
Predicted*
Streamwater
[S042-]
(uea L'1}
158
(88-348)
186
(115-316)
214
(102-413)
219
(148-358)
141
(54-440)
146
(50-328)
145
(64-281)
96
(56-190)
100
(37-273}
Percent of Reach
Nodes Above
High Estimate
Upper Lower
14%
27%
18%
9%
<1%
<1%
7%
13%
2%
14%
34%
19%
20%
<1%
<1%
14%
13%
2%
# Range in precipitation volume weighted sulfate concentration in each subregion. Data taken
from the Eastern United States sulfate concentration isopleth map in the 1986 NADP/NTN
annual data summary (National Atmospheric Deposition Program, 1987).
@ Estimates of the precipitation/runoff ratio in each subregion, from Geragnty et al. (1973).
C»
Predicted stream water sulfate concentration, assuming only atmospheric sulfate sources; calcu-
lated by multiplying the midpoint of the subregional range in precipitation sulfate concentration
by 1.5 (to account for dry deposition) and then multiplying by the precipitation/runoff ratio (to
account for evaporation). Low range estimate was calculated using the low estimate of
precipitation sulfate concentration and precipitation/runoff ratio. High range estimate was
calculated using the high estimates.
171
-------
Substantial sulfate sinks are apparent in several of the Southeast subregions (Southern Blue
Ridge, Piedmont, and Florida). A large majority of target stream reach upper and lower nodes
(80% to 95%) in the Southern Blue Ridge (2As) and Florida (3C) were estimated to have concen-
trations less than the low stream water sulfate predictions for each subregion. In the Piedmont
(3A), more than 60% had sulfate lower than this value. Sulfate retention by adsorption in the
deep, highly weathered iron- and aluminum-rich soils of the Southern Blue Ridge and in other
areas of the Southeast has been hypothesized by Galloway et al. (1983a) and by Rochelle and
Church (1987) on the basis of sulfate input-output modelling incorporating deposition, stream
water, and soil chemistry data. The low sulfate concentrations in Florida streams are somewhat
anomolous. Several plausible speculations are: (1) precipitation sulfate concentrations for this
area are overestimated, or perhaps more likely, the use of annual average precipitation data
obscures the actual relationship between stream and precipitation sulfate concentrations (e.g.,
spring 1986 runoff and sulfate deposition were substantially different from annual averages),
(2) sulfate reduction in soils or sediments of slow moving, organic-rich streams results in a net
loss of sulfur, or (3) sulfate adsorption may be occurring in watershed soils (though these sandy
soils are likely to have low sulfate adsorption capacity, they probably have not been subjected
to a long history of high sulfate deposition rates, so present rates of adsorption may be high in
clay layers within the soil mantle).
Substantial watershed contributions of sulfate appear likely at 9% to 27% of the upper
nodes and 14% to 34% of the lower nodes of target stream reaches in the four Mid-Atlantic
subregions. In the Southeast Region, 7% to 14% of Southern Applachian and Ozark/Ouachita
reaches had sulfate concentrations high enough to necessitate considering substantial sources in
addition to atmospheric deposition. Section 9 discusses the chemical characteristics and probable
sources of acidity in the subpopulation of target streams with ANC < 200 /zeq L"1. An impor-
tant point discussed in Section 9 is the finding that, after excluding reaches affected by acid
mine drainage, the majority of streams with sulfate concentrations > 250 /ieq L'1 are streams
within the target population that have relatively high ANC (> 200 /xeq L"1). These findings
suggest that high sulfate in many of these high ANC mid-Atlantic streams does not contribute to
acidity, but enters in the form of calcium and magnesium sulfates. On the other hand, in
stream reaches with ANC < 200 jteq L"1, sulfate concentrations tend not to be extremely high
(most are near or less than the best estimates for predicted stream water sulfate) and correlate
highly with ANC deficits (Section 10), suggesting that sulfate is likely to be associated with
hydrogen ion as a strong acid. Section 10 contains a more thorough discussion of the potential
role of sulfate from various sources in the acidification of NSS-I target streams.
6.3.4 Nitrate
Population distributions of nitrate at the upper and lower ends of NSS-I target reaches are
summarized in Figure 6-12b. There were no detectability or accuracy problems in the measure-
ment of nitrate (Section 4). At concentrations above the SDL, all precision estimates were < 9%
(Table 4-3). All subregion medians were lower than or about the same as the 1986 annual
volume-weighted mean concentrations of nitrate in precipitation (as reported by NADP, 1987).
Nitrate in precipitation ranged from 10 to 30 /*eq L"1 within the region covered by the NSS-I,
whereas subregion median values ranged from 0.4 to 37 /*eq L'1. There was no consistent pat-
tern of difference between medians for upstream and downstream populations.
The relative differences in nitrate concentrations among NSS-I subregions showed some
similarity with those observed for sulfate, but sulfate median concentrations were generally 5 to
15 times as high (compare box plots in Figures 6-12a and b). As observed for sulfate, nitrate
172
-------
medians were relatively high in the Mid-Atlantic subregions (11 to 37 /zeq L"1) and in subregion
2X, the Southern Appalachians (11 to 16 /zeq L"1), and relatively low in the remainder of the
Southeast (0.4 to 8.7 /zeq L"1). As observed for sulfate, Florida (3C) had the lowest median
nitrate concentrations (0.4 and 1.2 /zeq L"1, respectively, at the upper and lower reach ends).
The Ozark/Ouachita subregion (2D) also had very low nitrate; upstream and downstream medians
in this subregion were, respectively, 1.0 and 1.5 /zeq L"1. Comparison of geographic distributions
of nitrate (Figure 6-15) with those of sulfate in NSS-I sample sites (Figure 6-14) again illus-
trates the similarity in these regional distribution patterns.
Observed nitrate concentrations in all subregions were substantially less than the published
drinking water standard of 10 mg L"1 nitrate as N, or 714 /zeq L"1. Substantial fractions (more
than 40%) of the stream reaches in the Northern and Southern Appalachians (2Cn and 2X), the
Valley and Ridge (2Bn), and the Mid-Atlantic Coastal Plain (3B) had nitrate concentrations above
21 /zeq L"1 (0.03 mg L-1 nitrate as N), a level reported by Vollenweider (1971), Golterman
(1975), and Wetzel (1975) as a threshold concentration for contributing to nuisance algal blooms
in lakes and reservoirs, assuming sufficient phosphate is present. These high nitrate concentra-
tions were found primarily in reaches draining agricultural areas, and probably result from
fertilizer nitrogen inputs.
6.3.5 Chloride
Population distributions of chloride at the upstream and downstream ends of target reaches
in the NSS-I subregions are summarized in Figure 6-16a. There were no detectability or accu-
racy problems in the measurement of chloride (Section 4). The %RSD for field duplicates was
< 5% for all concentration ranges. An obvious pattern is that chloride concentrations were
substantially higher in the Mid-Atlantic subregions and Florida (3C) than in all the other NSS-I
subregions. In all subregions except 3C, the median concentration at lower reach ends was
higher than that for upper ends. Although these comparisons of medians are not conclusive
(comparison of medians is only a rough indicator), a trend of increasing chloride with distance
downstream suggests either a weathering source or the influence of agriculture, other anthro-
pogenic activities, road salt, or evaporative concentration of chloride from deposition.
Chloride concentrations were generally highest in the Mid-Atlantic Coastal Plain subregion
(3B), where the upstream and downstream medians were 264 and 318 /zeq L"1, more than ten
times those for the Southern Blue Ridge (2As), where chloride concentrations were lowest.
Regional variability was also high in subregion 3B, with inter-quintile ranges (80th - 20th
percentiles) both approximately 350 /teq L'1, again much higher (more than 10 times) than those
observed in subregion 2As. The Poconos/Catskills (ID) and Florida (3C) had upstream and down-
stream medians ranging between 110 /zeq L"1 and 173 /zeq L"1. The median of downstream reach
ends (101 /zeq L"1) in the Valley and Ridge subregion was intermediate. All the other sub-
regions, which are primarily upland, inland areas, had medians ranging from 32 to 67 /zeq L"1.
A map of chloride concentrations at NSS-I sample sites (Figure 6-17) shows that reaches
with chloride concentrations > 200 /zeq L"1 were found primarily within 200 km of the Atlantic
Ocean. On the basis of studies of relatively pristine New England lakes (Sullivan et al., 1988a),
we would not expect a significant sea salt contribution from precipitation to New England lakes
or streams more than 100 km from the coast. The empirical relationship reported by Sullivan
et al. (1988a) predicts sea water contributions of 69 /zeq L'1 and 38 /zeq L'1 for distances 20
and 50 km from the coast. The chloride concentrations in excess of 50 /zeq L"1 observed in the
173
-------
NITRATE CONCENTRATION
NSS PHASE I - LOWER NODE
N03(MeqL-')
H- < 10
X 10 to 20
<> 20 to 40
40 to 80
>80
Figure 6-15. Geographic distribution of NSS-I lower node index nitrate concentrations.
174
-------
Poc / Catskills ^
N. Appalach. "
Valley & Rdg. "
MA Coastal PI. "
S. Blue Rdg. ^
Piedmont
S. Appalach.
Ozark/Ouach. ^
Florida
u
L
U
L
U
L
U
L
U
L
U
L
U
L
U
L
U
L
Q
1 Vff&
1 1
c
CS
03
C
c
j d
! C
I II
! C
1
1
1
1
1
•x-x:xl
•x-x-x-x-X'X
I'X-x-x*
E88SI
x-x-X'X*x-:
$&$&$£•$&
frx-X'X-x
l*X*Xv
1
1
!
•x-x-x-x-x*
X*X-X-X*X*
i
r
x*x-3
I
SSal
;X;X-xl
tfx-x-x-x
feXiX*
•X'X*x-x*x
•:•»:»»:•»:•>
|x-x*x*
^xSxSXv
;:;:;:%w:-:-:-:':':
:>x>x*x*x-X'4
X'X'X-X-X-X-
X*X*X-X'X'
>XvX-X-X*X
x-x-x1!
•warn
*X^rXv.:-:-:.!.;.
"X"X<*X:.*X:I:
•»»:•:•»:•»
IM
3
-:.:-:-:-:-:->:1
IXwX-X-
X-X-X-J
"
•X-X*X*X'X*>.
3 !
wzm :
1 !
0 SO 100 ISO 200 250 300 35O 4OO 450 SO
Poc / Catskills
N. Appalach.
Valley & Rdg.
MA Coastal PI. "
S. Blue Rdg.
Piedmont
S. Appalach.
Ozark/Ouach. H
Florida
(a) Chloride (jieqL'1)
u
L
U
L
U
L
U
L
U
L
U
l_
U
L
U
L
U
L
m
m
B
IB
i
B
1 fc*a-SSS
1 ESS?
1
e
m
m
s
B
is
on
: i
1 i ess
1
0 1
•.*.•••• W,*.V»V.
{^
0 2
, yX'X-X-X*: -L
0 3
ww&m
0 4
.
«x*x-x-x-x-x*
0 S
•X'X-X'X-XvI
0 6
x-x-x-x-x*.-
0 7
,*.•.-.•.*.-.*.•.•.•.•.•
1 , ,. ,-..,..
o e
*4
1 ,
O 9
j... i
0 10
(b) DOC (mgL'1)
Figure 6-16. Population frequency distribution (Q20> median, and Qgo) °f index (a) chloride
and (b) DOC concentrations at upper and lower reach nodes in the NSS-I
subregions.
175
-------
CHLORIDE CONCENTRATION
NSS PHASE I - LOWER NODE
Figure 6-17. Geographic distribution of NSS-I lower node index chloride concentrations.
176
-------
majority of NSS-I Mid-Atlantic sample streams 60 to 200 km from the Atlantic coast appear,
therefore, to be substantially higher than expected. Though in NSS-I sites there was not a
close relationship between chloride concentrations and distance from the coast (r = 0.22), the
lowest chloride concentrations observed within given ranges of distance decreased logarithmically
as distance increased (Figure 6-18). Estimations of sea salt chloride contributions were based
upon a curve of concentration versus distance that envelopes these low chloride sites:
LOG[C1] = 3.61 - 1.08(LOG Distance in km)
The rationale underlying this approach is that the lowest chloride values along the curve
correspond to those streams with the least input of excess chloride from nonmarine sources.
This relationship predicts sea water chloride contributions of approximately 160 jueq L'1 at
20 km from the coast, decreasing to 60 jueq L'1 at 50 km from the coast, 28 ^eq L'1 at 100 km,
and 13 /zeq L'1 at 200 km. These estimates are about twice those predicted on the basis of the
pristine New England lakes relationship of Sullivan et al. (1988a). Though both approaches must
be considered crude, higher evapotranspiration rates in the Mid-Atlantic should result in higher
stream water chloride concentrations. In addition, it is possible that the more gentle slope of
the Mid-Atlantic Coastal Plain presents less impedance to weather systems and sea spray.
The high chloride concentrations observed by the NSS-I in parts of the Mid-Atlantic sub-
regions may be derived from roadsalt, agricultural sources, or ground water flowing through salt-
bearing geologic formations. Some high chloride concentrations (> 100 /teq L'1) were observed
in the Ozark/Ouachita subregion (2D); these may be associated with salt-bearing shales of
marine origin (Arkansas Geological Survey, 1929) or with agricultural activity in drainages within
the Arkansas River Valley.
In Florida (3C), chloride concentrations in peninsula sites were mostly > 100 /zeq L"1.
Concentrations in panhandle streams, on the other hand, were largely between 50 and 100
/zeq L"1. Chloride concentrations in these Florida streams were not highly correlated with
distance from the coast, though a reasonable amount of evaporative concentration (3.19; Table
6-7) of precipitation at 12.4 jueq L"1 chloride at the Quincy, Florida, NADP site in the panhandle
(from NADP, 1987) yields a concentration of 39 peq L"1. The concentration of chloride in pre-
cipitation appears sufficient to account for most of the chloride observed in panhandle streams.
The higher chloride concentrations in peninsula streams are probably augmented by another
source.
6.3.6 Dissolved Organic Carbon
Population distributions of DOC at the upstream and downstream ends of NSS-I target
stream reaches are summarized in Figure 6-16b. The SDL for DOC was 0.45 mg L'1 (Table 4-1).
Concentrations below this value are not statistically different from field blanks with a
concentration of zero. DOC concentrations < 2 mg L"1 were not very precise (%RSD = 25%),
whereas concentrations > 2 mg L'1 had %RSD between 7% and 14%. Population median DOC
concentrations exceeded 4 mg L'1 only in reaches in the Florida (3C) subregion (6.18 mg L"1 at
lower, 9.67 mg L"1 at upper) and at the lower end of reaches in the Mid-Atlantic Coastal Plain
(3B) subregion (4.45 mg L'1 at lower, 3.48 mg L"1 at upper). All other subregions had medians
< 2 mg L'1. Medians in the Southern Blue Ridge (2As) (upstream and downstream ends) and at
the upstream ends of the target reaches in the Northern Appalachians (2Cn) were very low (< 1
mg L'1). Twenty percent of the Florida reaches had DOC > 80 mg L'1 at their upper nodes.
Except for the Poconos/Catskills (ID) (upper and lower 80th percentiles = 3.2 and 4.3 mg L'1),
177
-------
O
O
O
4-
3-
1 -I
1
LOG C\ = 3.61 - 1.08 LOG(DIST)
i i i I i i i i i i i i I r
.0 1.5 2.0 2.3
LOG DISTANCE FROM COAST (km)
The NSS-I chloride distance envelope method fits a line through the lowest chloride values (see above figure). The
following eteps describe the process used to calculate this line.
1. The X-axis was divided into 10 equal intervals and data were subset to include the 5 lowest values along
the Y-axis within each interval.
2. A least squares regression line was fit to these points using a modification of trimmed least squares
(Ruppert and Carroll, 1980). Values were iteratively removed based on a residual threshold (studentieed
residual > 4). Studentized residual = residual/standard error of residual (SAS, 1985).
Figure 6-18. NSS-I index chloride concentration versus distance from the ocean in Mid-
Atlantic sites within 200 km of the sea coast. The line is an envelope
describing the lower bounds of the data.
178
-------
and the Ozarks/Ouachitas (2D) (upper and lower 80th percentiles = 3.3 mg L"1), all other
subregions had 80th percentiles < 2.5 mg L"1. Thus the vast majority of streams in the NSS-I
target population had low DOC.
The geographical distribution of DOC at NSS-I sampling sites is mapped in Figure 6-19.
High DOC concentrations were observed primarily in subregions 3C and 3B, as illustrated by
population distributions in the preceding paragraph. In these two subregions, DOC was posi-
tively correlated with color (r = 0.88) and hydrogen ion activity (r = 0.61) and negatively
correlated with ANC (r = -0.20). The relationship between DOC and acidity is discussed in
Sections 8 and 9, as is the location of acidic streams whose acid-base chemistry is likely to be
dominated by organic acids. High DOC stream reaches in subregions 3C and 3B were primarily
swampy, lowland coastal streams. These were distributed throughout the Florida (3C) subregion.
In the Mid-Atlantic Coastal Plain (3B), however, they were concentrated in the New Jersey Pine
Barrens and the lowlands west of Chesapeake Bay.
The elevated DOC concentrations in the Poconos/Catskills subregion (ID) were found in low
gradient, headwater reaches flowing through swampy meadows on plateaus subjected to past gla-
ciation. In other inland subregions, high DOC was usually associated with agricultural land use.
6.3.7 Other Chemical Variables
Population medians of other chemical variables at upstream and downstream ends of NSS-I
target stream reaches are listed in Table 6-8. Where appropriate, sample concentrations or
weighted statistics for these other chemical variables are incorporated into discussions in
Sections 8 and 9.
6.4 SMALL STREAMS
ANC tends to decrease with distance upstream within most stream drainages (scatter plots
in Figures 6-2 [a - i]), and vulnerability to acid deposition impacts therefore tends to increase.
Similarly, pH tends to decrease as one proceeds upstream in a given drainage (scatter plots in
Figures 6-3 [a - i]). A plot of ANC versus drainage area for a subset of headwater reaches in
the Interior Mid-Atlantic Region (Figure 6-20) shows that acidic (ANC < 0) reaches are largely
restricted to drainages of < 20 km2. (Organic-dominated reaches and reaches impacted by acid
mine drainage were excluded from this analysis.) The converse was not true, however, as most
reaches with drainages < 20 km2 were not acidic. The most acidic reaches, nevertheless, were
in the smallest drainages. A similar relationship was observed between pH and drainage area
(same data set), with all clearwater sample reaches having pH < 5.5 found in drainages < 30
km2. As with ANC, the lowest index pH values were found in the reaches with the smallest
drainage areas. Many reaches with small drainage areas, however, had relatively higher pH,
ranging as high as pH 7.
Examination of a subset of small streams sampled by the NSS-I allows some inference about
the regional distribution of chemistry in streams smaller than those sampled during the survey.
Two-thirds of the stream reaches sampled in the NSS-I were blue-line headwater reaches (Shreve
Order = 1) on l:250,000-scale maps. When taken together, these reaches are representative of a
population of streams smaller in watershed area than the total target population. These were
almost universally first order (sensu Strahler, 1957, on l:24,000-scale maps) with subregion
median drainage areas of 1.5 to 9.0 km2 at their downstream nodes and 0.18 to 1.2 km2 at their
upstream nodes. Population distributions of ANC at the upstream sampling nodes of these small
headwater streams give some indication of the acid/base status at the downstream ends of a
179
-------
DISSOLVED ORGANIC CARBON
NSS PHASE I - LOWER NODE
DOC (mg L'1)
H- < 1.0
X 1.0 to 2.0
<> 2.0 to 4.0
4.0 to 8.0
>8.0
Figure 6-19. Geographic distribution of NSS-I lower node index DOC concentrations.
180
-------
S
.2
'oo
CD
(4
45
CO
I— (
1
CO
z
.s
1
Q
43
0
at
CD
(2
p*^
jj.
u.
1
HH
73
C
CO
g
1
a
c3
CO
1
t>
8
S"*
Je
U
IM
O
tN tN
O CO
VO vo
O O
VO vo
O O
o r-
tN -H
o o
VO OO
f-; — <
oi co
tJ D
CO
CO
O
O
o
&
Q
O tN
00 O
•-* Vk
CO tN
CO -H
CO OO
v> ON
"* Tr
oo vo
f- O
tN -H
tN OO
vo v>
O 0
CO vo
VO vo
O O
t-» co
tN tN
O O
V> Tf
ON -H
co •t'
i-j ID
CA
Appalachian
"t;
£
-H
CJ
tN
ON oo
O vo
CO O
1—1 1—1
CO CO
t- o
\f) f^
co -^-
O ON
r~ co
W> CO
ON tN
co tN
O vo
VO vo
0 0
tN -sr
o o
v> co
tN CO
o o
OO Tf
•— < v>
tN 01
i-J |3
'o
2
rt
53
C4
{-
tN
oo t-~
r- vo
CO -H
CO CO
VO V)
-H* tN
vo ON
tN CO
r- v>
00 O
o >/•»
CO — 1
O 00
oo O
0 -H
CO O
OO CO
O -H'
VO •*»•
00 V>
o -«
CO tN
CO t—
Tfr vd
hJ £>
!s
JS
s
CO
1
^
%
03
CO
ON "3-
VN vo
— i CO
CO CO
£*
oo
CO
tN T
vo vo
tN ON
* *
* *
tN O
*— i 1—1
O 0
— H
CO tt
CO
1-4 t3
"aT
OO
•o
2
3
S
CO,
CO
tN
V> O
V} VI
tN O
v» ON
•— ' ON
CO CO
if) O
tN tN
O 00
0 0
ON f-
CO CO
O O
tN "3-
tN tN
O O
tN ON
OO f-
c-' vd
i-J t3
1
T3
vo
^3" 'T
CO vo
CO V>
VO vo
CO CO
VO V>
tN ON
00 -*
CO CO
VO vo
0 O
f- 1^-
>r> vo
o o
z:^
o o
tN vo
Tf Tt
hJ t)
^
ppalachians
<;
CO,
X
tN
vo •t
•^- tN
— i O
ON tN
CO C--
o -*r
ON -H
oo vo
-H tN
VO tN
1-^ i-H
ON r-
TT Tt
o o
o o
o o
CO VO
tN co
o o
tN V>
d ci
i-j £3
CO
a
•ks/Ouachit
*H
o
Q
tN
ON ON
co r-
>r> ON
-H CO
CM ON
r~ v>
— < vo
co r-
co
00 -H
tN ON
-H —4
>r> CO
ON tN
-H' CO
CO vo
VO CM
— ' tN
— i O
r- to
r-' r-'
i-3 *3
S
J-
E,
o
CO
CD
OO
73
2
S
|
I
.S
CO
1
CD
a
a
00
•a «
S oo
o
•0 C S
3-al
0.2 3
g S S
CD CD CD
"3 *g S
53 «^
!«§«
.all
2 2 r
ca •*
eo *O *T3
•SJ CD g
'a .2 's
3 « C
-1^
5
CD
a
1
o
0
§
CD
'•3 S
G ft
CD CD
Q, to
S *j
X
X -4!
^
S «
JS a
? &
"
•8
II
0
00 TJ
•o a>
S 8.1
3 « oo
w «
43=0 S
§ 1|
CO rt -S
*
<3 C J3
C8 O tS
§ E &
1
z.
181
-------
oo
43
i— i
.S
t/5
8
1
«M
43
o
en
o>
^
d
V
O
J
•s
CO
g
1
0.
o
*•*
o
s
s
.2
w
J>
^^
C3
1
8
i_
<2
i
"•3
e>
S^
a *c
ll
•S-l
a§
SB
00
VO
e>
1
H
It.
FM
. ?
fTt M
x_^
/^N
8 "^
Q ^
H
^•^
*•*
S i
w oo
*—'
<<_^
CO ,— .
ul"j
all
_
1 8
a S
i
o
+
^
w
8
Z
c
*o
a
4J
V.
r- co
CM CM
VO CO
in in
O 0
—i vo
ON ON
»-H f— *
c- -H
vo -*
o o
Tf VO
co C"*
CO CM
0 0
m o
CM W>
vo oo
CM in
r-- CM
ON •f
CM CO
O Tf
•n •—"
•^ to
JO
is
1
&
a
Q
CO
O
a
o
o
o
v~'
Q
00 t-
«-< —1
CM — <
O O
CO OO
O 00
•-* o
ON f-
ON ON
-« ON
O\ ON
.-* — i
§0
0
O 0
CO O
r- CM
T CO
O oo
CO O
•—< ON
— < o\
CO -*
JO
CO
.1
O
.2
*ra
n
Q.
<
Z
o
CM
VO -«
O ON
CM -H
•
00
."2
"§
C3
' |
>
C3
CM
oo in
oo vo
CM CM
CO CM
CO VO
TT CO
in oo
T CO
O "3-
ON CO
ON ON
CO vo
CO CO
in in
O 0
•n o
CM 0
^* ffi
•*r co
CO T
OO TT
-i vo
CO CM
f- O
CO «-<
ss
CO CO
JO
^
•J
2
1
s
03
CO
CM VO
T CO
*""* ^^
ON VO
TT CM
O 0
CM OO
r-~ in
O 0
•n
o\
vo
^^
CO
to *
in vo
rr vo
CO O
CM CM
CM O
VO TT
JO
/«>
1
i
52-
CO
CM
in oo
— i OO
CM -H
oo CM
oo vo
ON vo
in co
0 0
CM CM
ON ON
oo — <
OO CO
CO CO
88
O m
CO CM
•n -*
in o
oo in
in in
co vo
ON CM
-H OO
to o
~H — <
JO
^
•o
a>
S
n •<«•
80
0
•n o
>n — <
t-~ ON
oo O
in in
CM •<*
CM CM
CO -H
to co
JO
CO
1
1
a
<
52
X
tS
m r-
vo in
"
— < CM
in -«
CM ON
CO CO
O vo
r- CM
ON ON
-H CO
•<3- O
CM CO
O 0
0 0
O O
CM CN|
in in
in o
VO CO
vo vo
Tf VO
OO CO
t— -H
— c ON
•""
JO
^^
CO
I
o
1
S
o
CM
— < O
•n o
•-H I— 1
vo ON
ON CM
CM rr
oo r-
-H vo
VO ON
ON vo
— i vo
t— vo
00 VO
i— 1 CO
—i O
r- o
co in
OO CM
p— 4
ON to
— < CM
»«M •— (
in vo
ON vo
ON t-
VO CO
JO
1
fe
o
CO
co
00
r°
2
m
S
1
a
a
oo
a-o
co 9
9> 55
x «
« 00
9
o
8J3
•s W
•ss^
W -g 43
.2 -3 ^
S 73 g
° .y 9
« s %
tf\ VN TO
111
111
«a 5 43
I
182
-------
i
o
'So
2^
-0 |
3 c^
oo o
*"? 2^
eg
cj
• »H
CO Li?
^j 3G
t^ *T
\J ~f
O ^
j-
^
(2 +rt
iZj
*^^
ci
IH
^
o .|
-§
g
5" +
"•^ fN
IH OO
| ^
-4_»
ctf
«
3 +
.2 fc4
^
*« ,
1 8
O "
IH
c2 W
2 Q
•2 g
*s
^^ ^*s.
6-8. Population
(Continued
;ion
oo
CD CU
3 3
£ £
o o
O CO
ON fN
»— t •— i
ON vo
rr >o
-H — I
fN OO
c*"* r*"*
-H f-
rr '•H
— < -H
rr ON
fN rr
0 O
ZH o
rr r-
VO CO
ON ON
ON 0
o r-
fN — •
r- rr
CO X->
ON —i
ON CO
i-l: £>
'
(Poconos/Catskills
Q
gg
CO ON
tN fN
rr ON
-H ON
-H 0
OO vo
0 0
ON fN
>f* rr
fN ON
rr -H
o o
fN co
rr rr
t— rr
V> CO
r-- ON
fN fN
-H O
CN oo
ON t-
oo vo
J £
(N. Appalachians)
a
u
•0 CO
O CO
-H OS
i-H
rr ON
o o
-H fN
r- rr
CO OO
fN — <
r- vo
ON ON
fN fN
t-- oo
VO fN
VO — <
VO tN
CO fN
HH £
^
00
i
I
13
c
rr oo
rr O
CO *"H
vo r*-
r- rr
-H -<
ON co
r- fN
(N CO
V) O
fN tN
o ON
ON ON
o o
r- tN
rr fN
fN O
ON O
ON
rr co
vo oo
OO CO
If) •*$•
r- rr
.•J S
(— ^
'!
i
CO
<
«
«•> ON
r- vo
vo if*
oo oo
0 0
OO -H
O V>
ON O
«•» vo
~ 0
0 0
ON r-
ON >rt
r- tN
S2
VO -H
-H VO
-H rr
ON r-
^ ^
I
CO
ON rN
VO vo
oo oo
r- ON
CO CO
-H -H
\O vo
t-~ fN
VO co
fN fN
f-H 1-H
co rr
r- o
-H *
ON rr
rr co
O t-
VO vo
rr co
OO vo
fN fN
VO O
O ON
CO CO
rr oo
fN -H
JD
-4-»
1
•o
£
CO
ON oo
O v>
CO CO
O CO
fN -
oo r~
O —i
oo c—
o t-
>0 CO
o o
vo o
fN ON
VO fN
rr fN
rr co
r- fN
fN fN
co O
r- v>
co vo
fN ON
CO fN
>-)£>
(S. Appalachians)
X
rN
0 —i
«•» o
*-H —4
VO CN
O O
-H -C
CO ON
CO fN
>/•> CO
ON oo
O ON
-< 0
0 0
rr co
VO ON
O ON
O 00
1-H
vo oo
ON fN
ON ON
-H f-
r- oo
O vo
ir> fN
1-H «-H
-)£
(Ozarks/Ouachitas
Q
rN
WrN
fN rr
—i O
O —H
fN O
-H _
to rr
CO O
ON >/•>
CO
0 *--
1-H 1-H
0 0
VO fN
rr rr
CO If)
rM co
C-- ON
rr CN
r— ,00
W"> O
C"*" vo
,JS
E
CO
DO
T3
K
§
s
a
Wt
^
1
4)
,£3
*^
a
•*H
"O
o
a
IH
o.
o,
3
00
.s ^
Q.T3
« 3
« I/I
t' o
Is
ll
*c3 *^
t) 5
•g n
•$ ; g
.2^2
'•3 „ o
•S ^
* *ci "*
1 'S -s
•S«2
o.y S
co G 2
CO ^ P3
*S •* fi
a o c
CO **
7 J I
a % ^
.ll s
'S t> '3
3-^S
>
1
••3
o
a
rt
• .^
«" g
.X CO
*o €
C c3
Q, J3
O. co
^.y
« <
« 1
CO
•o °
« _a>
^ "o,
Jl
13 ">
""S
II
S£
o «
*H Q
*:
O
oo -o
*o o>
• 1«1 «
ill
pq « g,
eg|
•S "2 OT
(3 'C ,g
w
H
i
183
-------
1
•I
o
co
•g
ea
*O "^
is
CN
a
i
O
W
Q
eo
CO
•* ON
en en
vo
in vo
OO 00
vo o
OO O
p- vo
o o\
-^ CN
•-« -H O O
en en
ON r-;
CN CN
CN
ON
^.
IN
"*• CN
en O
CN ON
en vq
en CN
r4 a\
o en
en —< CN en
eN en vq vo
od ON -H CN
O CM
OO O
CN en
ON CN
CN —<
vooo
t- o
CN VO
O rr
O ON
CN —i
0\ O
oo en
ONOO
O oo
CM 0)
o a\
en oo
w-i o
cs
ON a\
o r-
r- vo
vo
tN en
en en
en CN
ON
CN O
oo en
en -H
CN CN
en vo
vo o
O\ oo
IT)
CN CN -^
CN O\
t- CN
CN C-
CN
oo O
CN O
CN CN
CN CN
oo r-
en vo
ON
VO O
vo vo •—i —i
CN en —i
vo
vo r~-
00 CN
ON ON
en v->
O Tf
vo 'sr
oo CN
ir vo
VO CN
ON oo
CN ON
O ON
O -H
oo vo
ON
ON oo -H -H
en vo
CN
ON
CN
en O
CN r-
•*t CN
en ON
vo
oo O
CN en
oo CN —<
v en
VO CN
VO irt
O C-
CN -H
oo ON —< ON
vo vo —< ON
ON ON
r- ON
O O —i -H
en
TT ON
CN —i
rr -HO
-H cN CN
«n >/•> vo o
ON ON —(en
O O
O O
O oo *-i ••*•
en O\
ON v
O O
vo
O O
* #
en -H en ON
r- O O ON
—i ON
O O
ON oo
VO ON
O CN
r- -H
—1 —< -H -H CN CN
o o
vo en
o o
oo v
O O
CN OO
VO TT
v
v
v
v v
00
o
-------
I
_J
cr
0
3
o
z
2004
150-
100-
50-
0
-50-
0
10
20
30
40
WATERSHED AREA (km2)
50
Figure 6-20. NSS-I spring index ANC 0*eq L'1) vs. watershed area (km2) in headwater reaches
in the Interior Mid-Atlantic.
185
-------
population of small streams with substantially smaller watershed areas than those of the NSS-I
target population as a whole (Figure 6-21). Distributions of chemistry for this population can be
interpreted as a description of water quality measured where stream water enters the blue-line
network of streams on l:250,000-scale maps.
The influence of map scale on the interpretation of headwater distribution estimates
warrants consideration. On l:250,000-scale maps, the mainstem blue lines extend upstream nearly
to the limits of perennial flow, whereas lateral tributaries equal in size to the upstream nodes
of blue-line headwaters are generally not shown. Thus the upstream nodes of headwater streams
very nearly represent the smallest streams that flow perennially, but sample reaches drawn from
the l:250,000-scale maps probably greatly, underestimate the total number of these very small
streams (on the ground or on l:24,000-scale maps) within the total population of all streams.
However, because mainstem headwaters are likely to be at least as acidic as lateral tributaries,
population distribution estimates describing chemical characteristics within this population of
smaller streams probably do not underestimate the status and extent of acidic and low ANC
conditions on a proportionate basis (e.g., "x percent of these reaches are acidic"). But because
of the many equal-sized lateral tributaries that may not be represented as blue lines on
l:250,000-scale maps, the actual number or combined length of these small streams may be
underestimated from the sample and may be very large.
For the reasons stated in the preceding paragraph, only percentage distributions for the
headwater stream population have been presented. Figures 6-22 and 6-23 compare the population
distribution estimates for ANC at the upstream and downstream reach nodes of blue-line head-
water stream reaches with the upstream and downstream nodes of the NSS-I target population as
a whole. In all NSS-I subregions containing acidic reaches, except the Mid-Atlantic Coastal
Plain (3B), the percentages of acidic reaches in the headwater reach subpopulation were greater
than those for the NSS-I target population as a whole (Figure 6-23).
Stream reaches upstream of these smallest streams sampled during the NSS-I (i.e., upstream
of blue-line representation on 1:250,000-scale maps) probably do not themselves contain signifi-
cant amounts of economically or recreationally important fish habitat. However, the impact of
acidification of these small headwaters on downstream fish habitat, through changes in rates of
headwater detritus processing (Allard and Moreau, 1986;Chamier, 1987;Mulhollandetal., 1987),
nutrient cycling (Elwood and Mulholland, in review) and contribution of drifting aquatic macro-
invertebrates has not been extensively studied and is largely unknown. Further study is under-
way, aimed at quantifying the number and length of the population of lateral tributary reaches
not represented as blue lines on l:250,000-scale maps.
6.5 UNCERTAINTY IN REGIONAL ESTIMATES
Several areas of uncertainty regarding NSS-I population estimates are quantified or
explored in this report. The primary areas of uncertainty, along with the report sections where
they are addressed, are as follows:
1. The variance resulting from the spatial sampling procedure, including the number of
sample reaches. (Section 2.4, confidence bounds on CDF figures, and standard errors
in tables in Appendices C and D.)
2. Uncertainty associated with the definition and identification of the target resource.
What are the physical characteristics of NSS-I target reaches, and are these streams
the appropriate size and type for study? (Sections 2.3, 5.2, 5.4, and 6.4.)
186
-------
Blue-line network on l:250,000-scale map. The large stream
(heavy line) is excluded from the population of interest because
its drainage area exceeds 155 km2.
Enlargement: Broken lines represent small streams not sampled
by NSS-I, but drawn on l:24,000-scale maps, including a subset
flowing into extreme upstream ends of blue-line target streams.
Figure 6-21. The population of small streams upstream of NSS-I blue-line headwater reaches.
187
-------
(a) Interior Mid-Atlantic
NUMBER OF REACHES <= X
" UPSTREAM NODES OF HEADWATERS
- AU. NSS-I UPSTREAM NODES
DOWNSTREAM NODES OF HEADWATERS
ALL NSS-I DOWNSTREAM NODES
(b) Mid-Atlantic Coastal Plain
NUMBER OF REACHES <» X
UPSTREAM NODES OF HEADWATERS
AU. NSS-I UPSTREAM NODES
- DOWNSTREAM NODES OF HEADWATERS
AU. NSS-I DOWNSTREAM NODES
"I ' T
ISO 200
ANC (/zeq L'1)
Figure 6-22.
Population distributions comparing ANC in the NSS-I target population with ANC
in a subpopulation including only headwater reaches: (a) Interior Mid-Atlantic
subregions and (b) the Mid-Atlantic Coastal Plain.
188
-------
o
&
a:
O
1.0-
0.9-
0.8-
0.7-
0.5-
0.4 -
0.3-
0.2-
0.1
0.0-1
-50
(a) Interior Southeast
NUMBER OF REACHES <= X
UPSTREAM NODES OF HEADWATERS
AU.NSS-1 UPSTREAM NODES
DOWNSTREAM NODES OF HEADWATERS
AU.NSS-1 DOWNSTREAM NODES
50
too
ANC
150
200
l
250
300
350
400
(b) Florida
cc.
o
QL
O
QL
Q.
<
_i
z>
s
o
1.0-
0.9-
0.8-
0.7-
0.6-
0.5-
0.4-
0.3-
0.2
0.1 -|
0.0
NUMBER OF REACHES <- X
UPSTREAM NODES OF HEADWATERS
AU.NSS-1 UPSTREAM NODES
DOWNSTREAM NODES OF HEADWATERS
AU.NSS-1 DOWNSTREAM NODES
-50
50
100
ANC
150
L'1)
200
250
300
350
400
Figure 6-23. Population distributions comparing ANC in the NSS-I target population with ANC
in a subpopulation including only headwater reaches: (a) Interior Southeast sub-
regions and (b) Florida subregion.
189
_
-------
3. Uncertainty regarding the chemical status of streams smaller than those targetted by
the NSS-I. Are reaches upstream of the NSS-I target population more acidic than
those of the target population? (Section 6.4.)
4. Uncertainty regarding the factors that control stream chemistry. In particular, what
are the probable sources of acidity in target streams? (Section 8 and 9.)
5. Applicability of the Chemical Index for Streams.
a. The survey year 1986 was a dry year, particularly in the Southeast. Would
subregional estimates of the number of reaches that are acidic or have low ANC
during spring baseflow change substantially, had they been measured in a more
typical water year? (Section 7.2.)
b. Are there times of the year when non-episodic pH or ANC would be substantially
or consistently lower than in the spring baseflow period? (Section 7.3.)
c. Does the variability of baseflow chemistry within the spring index period
preclude calculation of stable population distribution estimates and site
classifications. (Section 7.4.)
d. Do estimates of sampling system detection limit, analytical accuracy, and
sampling/analytical precision significantly affect confidence bounds on NSS-I
regional population estimates? (Sections 4 and 6.)
190
-------
SECTION 7
TEMPORAL VARIABILITY IN BASEFLOW CHEMISTRY
7.1 OVERVIEW
The NSS-I population estimates in this report and in its data compendium (Volume II) are
based on one or two samples taken during spring baseflow conditions in 1986, except for the
Southern Blue Ridge, where baseflow chemistry was indexed by three spring samples in 1985. In
order to assess the uncertainty in the population estimates caused by this limited temporal
sampling, we need to understand the temporal variability in NSS-I streams. The concept of
temporal variability can be broken down into three components in order to answer three major
questions about the accuracy and robustness of the NSS-I population estimates:
1. Among-year variability. Was 1986 an atypical year? Would subregional estimates of
the number of acidic or low ANC reaches during spring baseflow differ substantially
had measurements been made in a more typical year?
2. Among-season variability. Are there times of the year when non-episodic pH or ANC
would be substantially or consistently lower than in the spring baseflow period? Was
spring the right season in which to conduct the NSS-I?
3. Within-season variability. Does the variability of baseflow chemistry within the spring
index period preclude calculation of stable population distribution estimates and site
classifications?
These questions are addressed in the next three parts of this section, drawing on NSS-I data
(special interest site [Section 2.6] and regular samples) and intensive long-term monitoring data
from several of these streams provided by other researchers.
The NSS-I was not designed to quantify temporal variability in streams, but to provide
relatively stable index chemical descriptions that can be related to other characteristics of these
streams (e.g., basin geochemistry, episodic variability). Studies of seasonal trends and spring
seasonal variability in long-term and intensive data sets collected at special interest sites within
the NSS-I subregions will be an important aspect of additional research. An eventual goal is to
define a stochastic model of seasonal and episodic chemical change in these streams. Using the
NSS-I index chemistry, precipitation data, mapped hydrologic data, and classification of synoptic
and intensive study sites, researchers hope to be able to predict the temporal patterns in con-
centrations of important chemical constituents in baseflow (and possibly episodes). For the
present, however, it is important to use available information on temporal variability to evaluate
potential inaccuracies in the population estimates provided in this report.
7.2 AMONG-YEAR VARIABILITY
The NSS-I was designed to make population estimates of streamwater baseflow chemistry
during the spring of 1986 (1985 for the Southern Blue Ridge). It was not designed to predict or
examine long-term trends in chemical data. Nevertheless, it is important to know whether
stream chemistry during the 1986 spring index period was atypical in comparison to other years,
191
-------
because nearly all of the estimates in this report are based on 1986 spring index data. Of
potentially major concern was the fact that 1986 was a relatively dry year in most of the NSS-I
subregions.
In order to address the question of annual variability, NSS-I special interest site data
(Section 2.6) were compared to long-term data gathered by other researchers at the same sites.
Chemical data from one such site, Fernow Control (Section 2, Figure 2-1 and Table 2-4; Section
9, Table 9-12), a stream in the forested Allegheny Plateau (subregion 2Cn, the Northern Appala-
chians) of West Virginia, is used to illustrate the annual changes in spring index chemistry. The
data from Fernow Control were collected and kindly provided by the USFS Timber and Water-
shed Laboratory in Parsons, West Virginia (Dr. David Helvey, principal investigator). Annual
spring index period chemical statistics (20^ percentile, median, and 80"1 percentile) in Fernow
Control were calculated for data collected between April 1 and May 15 (the spring index period
in this part of West Virginia) of each year. The data for each year's index period includes from
3 to 12 separate samples. No attempt was made to remove hydrologic episodes from the data.
The spring index values in Fernow Control since 1980 for pH (Figure 7-1), alkalinity
(Figure 7-2), sulfate (Figure 7-3), and sum of base cations (Figure 7-4) were plotted, along with
the two NSS-I spring index samples collected in 1986. In all four chemical variables, spring
index period data in 1986 were similar to spring index period data collected in other years in
the 1980s. The data exhibit a cyclical pattern, but the 1986 values do not seem atypical. Using
the Timber and Watershed Laboratory data, there was no significant difference between the
mean alkalinity during the spring index periods of 1986 and the mean alkalinity during the com-
posite spring index periods of 1980-1985 and 1987 (t = -1.26; p = 0.22). Similarly, there was no
statistically significant difference in mean H+ concentrations (t = 0.92; p - 0.36) and in sum of
base cations (t « 1.00; p = 0.32) between 1986 and the other years. There was a significant dif-
ference in mean sulfate concentrations between 1986 and the other years (t = 3.22; p = 0.003).
The 1986 mean value was 10 /*eq L"1 higher than the mean for the other years.
Sulfate, base cation, and pH data collected in the NSS-I were in good agreement with data
collected by the Timber and Watershed Laboratory. One of the NSS-I alkalinity data points,
however, was approximately 10 /zeq L"1 higher than 80% of the Fernow Control spring index
period alkalinity data. Note, first, that the standard deviation of duplicate ANC measurements
by the NSS-I within this ANC range is 7.8 jueq L'1 (Section 4, Table 4-3), and second, that two
different methods were used (Gran's titrations in NSS-I and double endpoint titrations by the
Timber and Watershed Laboratory). While the NSS-I index pH (6.05) was larger than the 1986
Timber and Watershed Laboratory pH measurements (median = 5.99; 80*" percentile = 6.01), these
pH differences are well within the precision of the method, especially in light of the fact that
the measurements were made with different equipment and transport conditions.
In order to estimate between-year chemical changes on a broad regional scale, 1986 NSS-I
index data were compared to data collected in an acid streams reconnaissance, conducted by EPA
from April to June in 1987 to revisit stream reaches where acidic nodes had been observed in
1986. Longitudinal transects of pH and conductivity were collected for 26 stream reaches in the
Mid-Atlantic Region. By comparing the 1986 spring index pH and conductivity to the 1987 pH
and conductivity, an estimate of among-year variability on a regional scale can be made.
Resampled streams with evidence of acid mine drainage (subsection 9.3.1) were not included in
this comparison because chemical concentrations in acid mine drainage streams were highly
variable due to a strong dependence on stream discharge. A good relationship for pH (r2 =
0.836) exists between streams sampled in 1987 and 1986 (Figure 7-5). The slope of the regres-
sion line was 0.914 (SE = 0.071) indicating a one-to-one correspondence in pH between the two
years. The root mean square error of the relationship was 0.47. There was also a strong
192
-------
Spring Index Period- Fernow Control
6.5 r-
6.0
T t
5.5
1
1
1
J.
I960 1981 1982 1983 1984 1985 1986 1987
80%
Median USFS value
20%
1986 NSS-I
Figure 7-1. pH in Fernow Control during the spring index period from 1980 to 1987. Data
were provided by the USFS, Timber and Watershed Laboratory in Parsons, West
Virginia. The pH range between the two NSS-I observations was smaller than
the data symbol.
193
-------
Spring Index Period - Fernow Control
25 r-
20
LJ
cr 15
O 10
0
t
1
1
1
1
1980 1981 1982 1983 1984 1985 1986 1987
80%
Median USFS value
20%
1986 NSS-I
Figure 7-2. ANC (peq L"1) in Fernow Control during the spring index period from 1980 to
1987. Data were provided by the USFS Timber and Watershed Laboratory in
Parsons, West Virginia. ANC was determined using a double endpoint titration
by the USFS and a Gran's titration in the NSS-I.
194
-------
Spring Index Period - Fernow Control
150 i-
100
cr
-------
Spring Index Period - Fernow Control
25O
U
O; 200
O
O
-------
oo
CD
8-
7-
6-
5-
4-
3-
1987 SAMPLING
TIME
O SPRING
• JUNE
I
3
\
4
6
1986 pH
i
7
8
Figure 7-5. The pH observed in 1987 during the acid stream reconnaissance versus the 1986
NSS-I spring index field pH value in 37 mid-Atlantic stream nodes. The line is y
= x. The best-fit regression line was 1987 pH = 0.914 (1986 pH) + 0.616, (r2 =
0.836).
197
-------
relationship (r2 = 0.881; root mean square error = 25.9 ^S cm'1) between conductivity measured
in 1986 and 1987 (Figure 7-6). Except for two high conductivity outliers, there was a strong
one-to-one correspondence between the conductivities measured in 1986 and 1987.
As we would expect, there are differences in spring index period water chemistry between
years. Although there is a certain amount of annual fluctuation, there does not appear to be
any reason, based on the Mid-Atlantic acid streams reconnaissance data and Fernow Control
data, to assume that the index chemistry in 1986 was atypical compared to recent years, despite
the drought conditions. Thus in the Mid-Atlantic, in terms of annual variation, the NSS-I index
data appear to be representative of stream water chemistry. Because the 1986 drought was more
severe in the southeastern United States, further assessment of Southeast special interest site
data needs to be done.
7.3 AMONG-SEASON VARIABILITY - THE INDEX TIME PERIOD
The sampling window chosen for indexing chemical characteristics in NSS-I sample reaches
was the spring period between March 15 and May 15. The reasons for sampling baseflow chem-
istry during this season are discussed in detail in Section 2.5. The choice was made after
consultation with hydrologists, biochemists, and fishery experts from Pennsylvania, Virginia,
North Carolina, Florida, and Arkansas (U.S. EPA, 1985) and was supported by the results of
recent studies of seasonal and short-term chemical variability in streams (Murdoch, 1986; Witt
and Barker, 1986; Olem, 1986; Nix et al., 1986). These studies showed that minimum flow-
weighted pH, ANC, and base cation concentrations occured during the spring at almost all sites
(Ford et al., 1986). Sites with minima at other times of the year had winter minima and spring
values nearly as low.
To illustrate the seasonal changes in streamwater baseflow chemistry, data were obtained
from six streams studied during the EPA's Long Term Monitoring (LTM) Project (Newell et al.,
1987) that were also NSS-I special interest sites (Section 2, Table 2-4 and Figure 2-1; Section 9,
Table 9-12). Data in these streams were collected from fall 1984 to winter 1987. Three of
these streams are in the Catskill Mountains of southeastern New York (Biscuit Brook, East
Branch Neversink River, and High Falls Brook; Murdoch, 1986) and the other three streams are
in the Laurel Hills of southwestern Pennsylvania (North Fork Bens Creek, North Branch
Quemahoning, and Cole Run; Witt and Barker, 1986). Before using the data, samples suspected
of having been influenced by hydrologic episodes were removed. Some of the LTM data were
collected during precipitation events so that episodes could be studied. Thus when more than
one sample was collected in a day, only the first sample was used in the present analysis
because the remaining samples comprised sample sets taken during episodes. After this initial
episode screening, the remaining data were examined by plotting discharge versus time. Two
samples were removed because their discharges were an order of magnitude greater than the
remaining discharges in the same streams. The remaining data appeared to be representative of
baseflow conditions. The data were divided into seasons using the following criteria: summer
(May 16 to September 15), fall (September 16 to December 15), winter (December 16 to March
31), and spring (April 1 to May 15). The spring season was selected to correspond with the
spring index period used in the NSS-I for these streams (snowmelt to leafout).
Within-season chemical variation was roughly the same size as the among-season variation
(Figures 7-7 to 7-10). The variability within spring and summer seasons was lower than that
within fall and winter samples. The among-season data were analyzed using a two-way analysis
of variance model with stream and season as the main effects. The season "variable" was
divided into two groups, spring samples and nonspring samples (lumping the other three seasons
198
-------
200 -\
£
o
CO
Ld
O
O
ID
O
~z.
o
o
r--
OO
0)
150-
100-
50-
0-
1987 SAMPLING
TIME
O SPRING
• JUNE
0
50
100
150
-1
200
1986 CONDUCTANCE (^S cm ')
Figure 7-6. Conductivity observed in 1987 during the acid stream reconnaissance versus the
1986 NSS-I spring index in-situ conductivity value in 31 mid-Atlantic stream
nodes. All conductivity data were corrected to 25°C for comparison. The line is
y = x.
199
-------
9-r
X
0.
5-
D E. BR. NEVERSINK
A BISCUIT BROOK
O HIGH FALLS BROOK
m
D N. BR. QUEMAHONING
A N. FORK BENS CREEK
O COLE RUN
WINT SPRG SUMM FALL
NEW YORK
- 1 - 1 - 1
WINT SPRG SUMM
PENNSYLVANIA
FALL
Figure 7-7. The seasonal pattern in pH (mean plus 95% confidence interval) in three streams
in the Catskill Mountains of New York (Murdoch, 1986) and three streams in the
Laurel Hills of Pennsylvania (Witt and Barker, 1986). The spring season was
defined as April 1 to May 15 (spring index period).
200
-------
300-
!_, 200
cr
^^
O
100-1
0-
D E. BR. NEVERSINK
A BISCUIT BROOK
O HIGH FALLS BROOK
D N. BR. QUEMAHONING
A N. FORK BENS CREEK
O COLE RUN
WINT SPRG SUMM FALL
NEW YORK
WINT SPRG SUMM FALL
PENNSYLVANIA
Figure 7-8. The seasonal pattern in ANC (mean plus 95% confidence interval) in three
streams in the Catskill Mountains of New York (Murdoch, 1986) and three
streams in the Laurel Hills of Pennsylvania (Witt and Barker, 1986). The spring
season was defined as April 1 to May 15 (spring index period).
201
-------
350 H
cr 250-j
LU
H-
U_
^ 150
50
D E. BR. NEVERSINK
A BISCUIT BROOK
O HICH FALLS BROOK
T 1 1 i 1
WINT SPRG SUMM FALL
NEW YORK
D N. BR. OUEMAHONING
A N. FORK BENS CREEK
O COLE RUN
1 1 1 1
WINT SPRG SUMM FALL
PENNSYLVANIA
Figure 7-9. The seasonal pattern in sulfate concentration (mean plus 95% confidence interval
in three streams in the Catskill Mountains of New York (Murdoch, 1986) and
three streams in the Laurel Hills of Pennsylvania (Witt and Barker, 1986). The
spring season was defined as April 1 to May 15 (spring index period).
202
-------
450
350-
co
z
o
1-
<
o
Ul
00
CD
u_
o
250-
150-
50-
D E. BR. NEVERSINK
A BISCUIT BROOK
O HIGH FALLS BROOK
1 I I I
WINT SPRG SUMM FALL
NEW YORK
O N. BR. QUEMAHONING
A N. FORK BENS CREEK
O COLE RUN
I I 1 1 T
WINT SPRG SUMM FALL
PENNSYLVANIA '
Figure 7-10. The seasonal pattern in the sum of base cations concentration (mean plus 95%
' - - confidence interval) in three streams in the Catskill Mountains of New York
(Murdoch, 1986) and three streams in the Laurel Hills of Pennsylvania (Witt and
Barker, 1986). The spring season was defined as April 1 to May 15 (spring index
period).
203
-------
together), to test the hypothesis that the spring samples were different than the samples in the
other seasons. In all analyses, the stream main effect was always significant (p < 0.01).
pH values in the spring (Figure 7-7) were significantly lower than the pH values in other
seasons (F - 6.4; p < 0.05). In general, mean ANC values were highest in the fall (Figure 7-8)
and lowest in the spring. The spring ANC values, however, were not significantly lower than
the values in the other seasons (F = 3.01; p = 0.08). Spring values were, however, significantly
lower than fall values (F = 13.24 p < 0.01). Sulfate concentrations showed relatively small
changes both within and between seasons (Figure 7-9) and sulfate concentrations in the spring
were not significantly different from those in the other seasons (F = 0.56; p *> 0.45). The
seasonal mean concentration of the sum of base cations (Figure 7-10) was lower in the spring
than in the other seasons combined (F = 8.7; p < 0.01).
These findings were essentially in agreement with those of Ford et al. (1986) and Messer et
al. (1986,1988) showing that spring and winter are typically seasons of low ANC and pH in Mid-
Atlantic and Southeast streams. In combination with information citing the spring season as a
time of the year when acid-sensitive fish life history stages are present (section 2.5; Bowman,
pers. comm.), this data supports the selection of spring baseflow as the appropriate time for
sampling streams in the NSS-I.
7.4 WITHIN-SEASON VARIABILITY
A potential source of uncertainty in the NSS-I population estimates presented in this report
is the chemical variability within the spring index sampling period. The NSS-I Pilot Survey
demonstrated that chemical variability among the three sample visits to Southern Blue Ridge
streams between March 15 and May 15,1985 had an insignificant effect on resultant population
distributions of pH, ANC, sulfate, and a number of other major cations and anions (Messer et al,
1986, 1988). These results, however, do not necessarily apply outside the Southern Blue Ridge
subregion. Within-season variability was examined in two ways: (1) through use of the LTM
and special interest site data and (2) through comparison of the two sample visits to 226
upstream and 222 downstream reach sampling sites in the Mid-Atlantic during the NSS-I.
7.4.1 Within-Season Variability in Special Interest Sites
Analysis of the LTM project and special interest site data set of three Catskill Mountain
and three Laurel Hill streams (discussed in Section 7.3) provided a finer resolution assessment of
the variability within the spring index period than did the comparison of paired visits to numer-
ous NSS-I sample reaches. The spring index period was defined as April 1 to May 15 to corre-
spond to post-snowmelt/pre-leafout conditions in these streams. The LTM data presented here
were collected in 1985 and 1986, whereas the NSS-I Mid-Atlantic data were collected only in
1986. In general, there was good agreement between the data collected by the NSS-I and that
of the LTM researchers (Figures 7-11 to 7-14). There was more variability in the LTM data
than in the NSS-I data, as would be expected, since the LTM data were collected for two dif-
ferent spring index periods and there were more samples.
The variability in spring index pH in the Pennsylvania streams (Figure 7-11) was fairly low
(less than 0.1 pH unit change in Cole Run and the Quemahoning and 0.4 pH unit range in Bens
Creek). In the New York Catskills streams, the pH variability was larger (0.7 to 1.1 pH unit
ranges). Variability in spring index ANC (Figure 7-12) showed a similar but somewhat less vari-
able pattern than pH. The largest ranges in ANC were observed in Biscuit Brook (57 /ieq L"1)
204
-------
SPRING INDEX PERIOD (APR1 - MAY 15)
NSS-I (X) vs. LTM (0)
Ox
X
8
E. BRANCH
NEVERSINK
X
8X
8
o
o
BISCUIT
BROOK
Ox
O
O
HIGH
FALLS BR.
°X
N. BRANCH
QUEMAHON 1 NG
O
O
N.FORK
BENS CR.
X
ox
0
COLE
RUN
6-
5-
4-
NEW YORK
PENNSYLVANIA
Figure 7-11. The pH range of NSS-I special interest site samples (1986) and Long-Term
Monitoring (LTM) Project samples (1985-1986) within the spring index period
(April 1 to May 15) in three streams in the Catskill Mountains of New York
(Murdoch, 1986) and three streams in the Laurel Hills of Pennsylvania (Witt and
Barker, 1986).
205
-------
SPRING INDEX PERIOD (APR1 - MAY 15)
NSS-I (X) vs. LTM (0)
-------
SPRING INDEX PERIOD (APR1 - MAY 15)
NSS-I (X) vs. LTM (0)
350.00
300.00-
l 250.CO-
O-
ID
UJ
200.00-
150.00-1
100.00-
50.00-
E. BRANCH
NEVERSINK
0
03
O
0
i
BISCUIT
BROOK
•
1
0
1
I
HIGH
FALLS BR.
O
I
X
X
0
O
N. BRANCH
QUEMAHONING
,
•
:
:
1
©X
N.FORK
BENSCR.
1
O
COLE
RUN
i
1
NEW YORK
PENNSYLVANIA
Figure 7-13. Sulfate concentration range of NSS-I special interest site samples (1986) and
Long-Term Monitoring (LTM) Project samples (1985-1986) within the spring index
period (April 1 to May 15) in three streams in the Catskill Mountains of New
York (Murdoch, 1986) and three streams in the Laurel Hills of Pennsylvania (Witt
and Barker, 1986).
207
-------
SPRING INDEX PERIOD (APR1 - MAY 15)
NSS-I (X) vs. LTM (0)
400.00-
r* 350.00-
_i ;
| 300.00-
o 250.00-
200.00-
150.00-
o
LU
03
U.
O
In 100.00-1
50.00-
O
§
E. BRANCH
NEVERSINK
X
OX
O
O
0
BISCUIT
BROOK
O
X
X
O
O
HIGH
FALLS BR.
X
X
O
O
N. BRANCH
QUEMAHONING
X
X
O
N.FORK
BENS CR.
X
O
• O
COLE
RUN
NEW YORK
PENNSYLVANIA
Figure 7-14. The range in the sum of base cation concentrations in NSS-I special interest
site samples (1986) and Long-Term Monitoring (LTM) Project samples (1985-1986)
within the spring index period (April 1 to May 15) in three streams in the
Catskill Mountains of New York (Murdoch, 1986) and three streams in the Laurel
Hills of Pennsylvania (Witt and Barker, 1986).
208
-------
and High Falls Brook (80 jueq L"1). Alkalinity variations in the other four streams were less
than 30 /xeq L"1. In streams that had negative ANC, as measured by the NSS-I, the agreement
between NSS-I and LTM project ANC data was not good. The LTM researchers used different
procedures to deal with negative ANC samples and did not report them as negative values. Thus
the LTM and NSS-I ANC data are not directly comparable for samples with negative ANC.
In general, the variability in spring index sulfate concentrations was lower than that of the
other ions (Figure 7-13). The largest range in sulfate concentration was 50 /zeq L"1 in Biscuit
Brook. In comparison, the variability in the sum of base cations (Figure 7-14) was rather large
(over 100 #eq L"1 in Biscuit Brook and High Falls Brook), indicating that base cations have a
major terrestrial source component. Sulfate appears to behave as a more conservative ion. This
suggests that there is not a major terrestrial source of sulfate and that the dominant sulfate
source is probably atmospheric. The concentrations of sulfate and base cations measured in the
North Branch of the Quemahoning during the NSS-I were twice those observed in the LTM
project. A field visit to the Quemahoning during the Acid Streams Reconnaissance showed that
it is affected by acid mine drainage along sections of its reach. It is possible that the LTM
project and the NSS-I sampled slightly different locations along the stream. If the NSS-I
sampling location was slightly downstream of the LTM sampling site, it could receive sulfate and
base cations from mine drainage that was not present at the upstream LTM site.
7.4.2 Within-Season Variability in NSS-I Mid-Atlantic Sites
During the NSS-I field activities of 1986, Mid-Atlantic sites were visited twice during the
spring, and Southeast sites were visited only once. The two sampling visits in the Mid-Atlantic
subregion streams were typically two weeks apart. The NSS-I design intended them to be repli-
cates, though their timing was not randomly set. The population distribution estimates presented
for Mid-Atlantic subregions are based on the spring index chemistry, calculated by averaging
values from the two spring visits—thereby obtaining more stable values with which to charac-
terize each reach. This section evaluates, where possible, the effects of spring season baseflow
chemical changes on estimates of the regional distribution of chemical characteristics. Of
particular interest is the effect that such changes have on estimates of the number or fraction
of acidic and low ANC stream reaches.
Figures 7-15 and 7-16 show that, for upper and lower reach sampling nodes in the Mid-
Atlantic, the second spring sample ANC and pH are excellent predictors of values from the first
sample, but there is evidence of a time trend of increasing ANC and pH between spring sampling
visits. Linear regressions of the first visit versus the second visit are both highly significant
(p < 0.0001) with r2 values of 0.93. However, the second visit ANC and, to a lesser extent, pH,
are consistently higher, as expressed by regression slopes of +0.82 and +0.97. This pattern is
likely to result from a general decrease in streamflow over the spring season. During periods of
decreasing baseflow, there is a general increase in the concentration of many chemical species
in stream water. In particular, concentrations of base cations and other weathering products
increase as the streamflow proportion contributed by older groundwater increases in relation to
quickflow contributions from soil macropores and shallow soil horizons. The increase in alka-
linity associated with base cations at lower base flows tends to neutralize acids derived from
organic soil horizons and atmospheric deposition. (Note that the pattern observed with
decreasing flow stage in reaches impacted by acid mine drainage was an increase in base cations
and sulfate, with a reduction in ANC and pH.)
The magnitude of change observed in ANC and pH between the two NSS-I spring sampling
visits was very dependent upon the concentration level on the first visit. With ANC, for
209
-------
>
» LJ
S" a!
* i
~ IS
o
10004
750-
500-
250-
0-
-250-
-500-
-500 -250
o
SECOND SAMPLE VISIT
i • i ' r ' r
250 500 750 1000
Figure 7-15. ANC on first visit versus second visit to NSS-I sample reaches in the Mid-
Atlantic Region.
210
-------
X
OL
CO
>
<
CO
CO
u_
10-
9-
8-
7-
6-
. 5 •
4
3
-i—r
5
T T
7
45 67 8
SECOND SAMPLE VISIT pH
T
9
10
Figure 7-16. pH on first visit versus second visit to NSS-I sample reaches in the Mid-Atlantic
Region.
211
-------
example, the median absolute value of changes was 9.7 /teq L'1 in the range beween -50 and +50
jieq L"1 (Figure 7-17). For reaches with ANC between 50 and 100 /*eq L"1 on the first visit,
the median change was 29 /zeq L"1. Forty-nine percent of stream sample sites with ANC < 50
/ieq L'1 changed less than 10 /zeq L'1 (Table 7-1). Sixty-seven percent of those with first visit
ANC between 50 and 200 fieq L'1 changed by less than 50 //eq L"1. The largest changes
between visits were observed at sample sites with ANC > 400 /^eq L"1 on the first visit.
Seventy-one percent of sample reaches in this group experienced ANC changes (mostly increases)
of greater than 100 peq L'1. The Northern Appalachians (2Cn) and the Poconos/Catskills (ID)
tended to have changes greater than the regional median change in the higher ANC ranges
(Figure 7-17). Within the ANC ranges of greatest interest (those below 100 jjeq L'1), the
smallest median ANC changes were observed in the Valley and Ridge (2Bn) and the largest in
subregion ID (Figure 7-17). ANC changes observed within the lowest range are of the magni-
tude expected due to field sample collection, transport, and analysis (pooled standard deviation
of field duplicate ANC measurements was 2.9 to 7.8 ^eq L"1 in this range, Table 4-3).
A related pattern in the magnitude of change between sample visits was observed in pH
(Figure 7-18 and Table 7-2). Stream water pH in reach sampling sites with pH lower than 5.5
tended to be less variable during baseflow than those with with pH above this value. However,
the logarithmic nature of the pH variable must be taken into account. The change in hydronium
ion concentration corresponding to a 0.25 unit pH change near pH 5 is 10 times greater than
that associated with the same magnitude of pH change near pH 6. The median pH change (abso-
lute value) for sample reaches with a first visit pH < 5.5 was 0.08 (Figure 7-18), about twice the
standard deviation of field duplicate pH measurements in this pH range (0.036). All sample
reaches with pH < 5.5 showed pH changes of less than 0.50 pH unit; 80% changed less than 0.25
unit. Considering the full pH range of streams, 94% experienced pH changes of less than 0.5
unit, whereas 75% showed smaller pH changes of less than 0.25 unit. The Poconos/Catskills (ID)
showed the largest median between- visit pH changes in all ranges of pH (Figure 7-18), whereas
the Valley and Ridge subregion (2Bn) showed the smallest median changes in the lower pH
ranges of greatest interest.
The variability of stream baseflow chemistry between sample visits (or the apparent trend,
for some variables and subregions) within the spring index sampling period has bearing on
questions such as: "What is the number or fraction of the reach population that is acidic during
the index period?" or "How many streams in the region have spring baseflow pH less than 5.0?"
Figures 7-19 to 7-22 compare, in the combined Mid- Atlantic subregions, the population cumula-
tive distribution functions (CDFs) for ANC, pH, sum of base cations (SBC) and sulfate, based on
first visit, second visit, and spring index (average) measurements. Differences between first and
second visits are greatest for ANC and minimal for the other three variables. The effect of
greater within-season ANC increases in streams with ANC > 50 /zeq L"1 is apparent in these
plots, as is the greater apparent stability of baseflow (non-episodic) ANC and the other variables
at upper reach nodes in comparison with lower nodes.
Table 7-3 summarizes the differences in population distributions of ANC based on the two
different visits and the chemical index made by averaging the two. Considering the estimates
for acidic (ANC < 0 peq L'1) lower reach nodes in the target population, only the estimate for
subregion 3B is different from that based on the chemistry averaged over the two visits (visit 1
estimates 10% are acidic, whereas the index chemistry estimates 7%). For the reach upper
nodes, the population percentage of acidic reaches based on the chemical index is virtually the
same for subregions ID and 2Bn, but the index underestimates (by 3 percentage points) the
fraction of acidic reaches during the first visit in subregions 2Cn and 3B. Similarly, the
index-based CDFs made slightly lower estimates of reaches with pH < 5.0 at the lower nodes in
212
-------
ANC CHANGES BETWEEN VISIT I AND 2
NSS-I Mid-Atlantic
200-
UJ
e>
X 150-
UJT.
-J 400
CONCENTRATION RANGE
L-')
ALL
Figure 7-17. Median absolute changes in ANC (unweighted) between first and second visits to
NSS-I sample reaches in four subregions of the Mid-Atlantic Region. MA indi-
cates the median value for all Mid-Atlantic subregions.
213
-------
Table 7-1. Distribution of Observed ANC Differences (unweighted) Between First and Second
Visits to NSS-I Mid-Atlantic Sample Reaches
Index ANC
<5
Absolute Change in ANC (/ieq L"1)
>5 - <10 >10 - <25 >10 - <50 50 - <100 >100
ANC £0 13
ANC > 0 - <: 50 19
ANC > 50 - < 200 8
ANC > 200 - < 400 3
ANC > 400 2
8
10
6
6
1
15
23
23
7
3
4
6
46
9
11
3
2
32
25
20
9
24
90
214
-------
Ul
o
0.24-
0.20-
o
X 0.16-
o.
Ul
ID
_J
O
CO
CD
0.12-
0.08-
0.04^
pH CHANGES BETWEEN VISIT I AND 2
NSS-I Mid-Atlantic
ID
2Cn
MA
3B
.0
2^
3B
2Bn
2Bn
ID
I
<5.5
5.5-6.5 >6.5
pH RANGE
I
ALL
Figure 7-18. Median absolute changes in pH (unweighted) between first and second visits to
NSS-I sample reaches in four subregions of the Mid-Atlantic Region. MA indi-
cates the median value for all Mid-Atlantic subregions.
215
-------
Table 7-2. Distribution of Observed pH Differences (unweighted) Between First and Second
Visits to Mid-Atlantic Sample Reaches
Index pH
Absolute Change in pH
< 0.25 >0.25 - <0.5 >0.5 - <1.0
1.0 - <2.0
pH < 5.0
pH > 5.0 - < 5.5
pH > 5.5 - < 6.0
pH > 6.0
34
13
25
249
6
6
9
59
3
17
216
-------
(a) Upper Nodes
OL
O
o;
UJ
•i.o-
0.9-
0.8-
0.7-
0.5-
0.4-
0.3
0.2
0.1
0.0
T
100
—I—
-50
INDEX VALUE
FIRST VISIT
SECOND VISIT
-1 ' 1 ' 1 ' 1 ' 1 i • . • .
50 too 150 200 250 300 350 400
ANC (jueq L"')
O
o:
O
1.0-
0.9 -
0.8-
0.7-
0.6-
0.5-
0.4-
0.3-
0.2-
0.1 -
0.0
(b) Lower Nodes
T ' 1—
-100 -50
—T~
50
—T—
100
—I—
150
—I—
200
T
INDEX VALUE
FIRST VISIT
SECOND VISIT
1
I
250 300 350 400
ANC Qteq L"1)
Figure 7-19. Population distributions for ANC in the NSS-I Mid-Atlantic Region based on first
visit, second visit, and spring index (average) measurements for (a) upper nodes
and (b) lower nodes.
217
-------
(a) Upper Nodes
o
UJ
1.0
0.9
0.8
0.7
0.5-
O.-t-
0.2-
0.1-
0.0-
"
4
pH
(b) Lower Nodes
NUMBER OF REACHES
WDEXVALUE
pH
Figure 7-20. Population distributions for pH in the NSS-I Mid-Atlantic Region based on first
visit, second visit, and spring index (average) measurements for (a) upper nodes
and (b) lower nodes.
218
-------
o
1.0
0.9
0.8
0.7
0.4-
0.2
0.1 -
0.0-
(a) Upper Nodes
NUMBER OF REACHES
INDEX VALUE
FIRST VISIT
SECOND VISIT
400
800
1200
1600
SUM OF BASE CATIONS (//eq L"1)
(b) Lower Nodes
400 800 1200
SUM OF BASE CATIONS (/ieq L"1)
1600
Figure 7-21. Population distributions for sum of base cations in the NSS-I Mid-Atlantic Region
based on first visit, second visit, and spring index (average) measurements for
(a) upper nodes and (b) lower nodes.
219
-------
(a) Upper Nodes
o
cg
O
UJ
1.0-
0.9-
0.8-
0.7-
0.6-
0.5-
0.4-
3 0.3-
0 0.2-
0.1
0.0
50 100 150 200
SULFATE (/«sq L'1)
NUMBER OF REACHES
INDEX VALUE
FIRST VISIT
SECOND VISIT
250
300
o:
1.0-
0.9-
0.8-
0.7-
0.6-
0.5-
0.3-
O
0.2-
0.1 •
0.0-
(b) Lower Nodes
50
100
T1"
150
NUMBER OF REACHES
INDEX VALUE
FIRST VISIT
SECOND VISIT
T'-
200
250
300
SULFATE (/teq L'1)
Figure 7-22. Inverse population distributions for sulfate concentrations in the NSS-I Mid-
Atlantic Region based on first visit, second visit, and spring index (average)
measurements for (a) upper nodes and (b) lower nodes.
220
-------
Table 7-3. Population Estimates of the Percentage of Acidic and Low ANC Stream Reaches
Calculated from First Visit, Second Visit, and Spring Index (Average) ANC at
Upper and Lower Reach Nodes in NSS-I Mid-Atlantic Subregions
ANC < 0 /teq L'1
visit 1
Lower Node
visit 2 Index (UCB)
visit 1
Upper Node
visit 2 Index (UCB)
ID (Poconos/Catskills)
2Bn (Valley and Ridge)
2Cn (N. Appalachians)
3B (MA Coastal Plain)
*
*
4
10
*
*
4
7
*
*
4(8)
7(12)
6
5
9
15
7
5
6
12
6(12)
5(10)
6 (11)
12 (19)
ANC < 50
L'1
visit 1
Lower Node
visit 2 Index (UCB)
Upper Node
visit 1 visit 2 Index (UCB)
ID (Poconos/Catskills)
2Bn (Valley and Ridge)
2Cn (N. Appalachians)
3B (MA Coastal Plain)
11
3
30
20
5
*
22
16
5(9)
*(D
23 (36)
20 (29)
25
11
41
29
17
11
34
29
23 (34)
11 (19)
34 (48)
30 (42)
(UCB) = Upper 95% Confidence Bound.
* - less than 1%.
221
-------
two subregions (2Cn and 3B) than did those based on pH during the first visit (Figure 7-20). In
subregions ID and 3B, index-based estimates for reaches with pH < 5.0 at their upstream nodes
were lower than those based on the first visit (Table 7-4). In all cases, the single visit
estimates of the number (or percentage) of acidic reaches are within the 95% confidence bounds
of the figures based on the chemistry averaged over the two visits. These confidence bounds,
based entirely on the spatial sampling frame (i.e., they do not incorporate within-season
chemical variance) suggest that the variance resulting from the number and spatial distribution
of sample sites is greater than that due to observed within-season chemical variability. How-
ever, the observation of a trend of increasing ANC over the portions of the index sampling
period leaves open the possibility that, had all streams been sampled at the beginning of the
index period, estimates of the number of acidic reaches might have been greater than reported.
As described earlier in this section, the quantitative study of intensive data from special
interest sites may allow extrapolation from index chemistry to the expected annual minimum
baseflow ANC.
7.5 SUMMARY
In this section, we have examined the three major components of temporal variability
(among-year, among-season, and within-season) that could affect the NSS-I population estimates.
The limited data available for among-year variability suggest that: (1) the NSS-I year (1986)
was not highly unusual with regard to several important chemical constituents, and (2) among-
year variability probably would not cause population estimates to change substantially from year
to year (in the short term). Chemical measurements (pH, ANC, sulfate, and sum of base cations)
taken in 1986 from Fernow Control in West Virginia were very similar to those observed in 1980
- 1985 and 1987. On a regional scale, in 26 acidic stream reaches in the Mid-Atlantic, devia-
tions between 1986 and 1987 spring pH and conductivity measurements were fairly small.
Data collected from six Mid-Atlantic special interest sites and LTM project streams showed
that spring index samples had lower (or equally low) pH, ANC, and sum of base cation concen-
trations than samples collected in other seasons. Sulfate concentrations showed little seasonal
change. Given the seasonal variation, and in light of the fact that populations of many fish
species are most sensitive to acidic conditions in the spring when acid-sensitive life history
stages are present, spring appears to be have been the appropriate season to conduct the NSS-I.
There was a fair amount of chemical variability within the spring index period, although it
was not great enough to preclude the estimation of relatively robust spring baseflow population
descriptions and classifications based on spring index values. However, quantitative models of
stream baseflow concentrations incorporating seasonal change and flow variation may be neces-
sary tools for assessing small long-term changes in stream chemistry—particularly if the moni-
toring programs incorporate baseflow indices calculated from one or several spring samples.
Data collected from the first sample visit in NSS-I Mid-Atlantic streams were excellent pre-
dictors of ANC and pH in second visit samples. There was, however, a systematic trend of
decreasing ANC and increasing pH from the first to the second sample visit in most streams.
The median absolute change in ANC was around 50 #eq L"1, and the median absolute change in
pH was 0.13. Median changes in ANC were smaller (9.8 fieq L"1) in streams within the ANC
range of critical interest (-50 to +50 /zeq L'1). NSS-I population estimates made using only first
visit samples were very similar to estimates made using only second visit samples or the spring
index value (average of visits 1 and 2). Almost all work so far has been done on streams in the
Mid-Atlantic and Southern Blue Ridge. Future work on temporal variability will incorporate data
collected from special interest site streams throughout the southeastern United States.
222
-------
Table 7-4. Population Estimates of the Percentage of Stream Reaches with pH Less than
Reference Values Calculated from First Visit, Second Visit, and Spring Index
(Average) pH at Upper and Lower Reach Nodes in NSS-I Mid-Atlantic Subregions
pH < 5.0
visit 1
Lower Node
visit 2 Index (UCB)
Upper Node
visit 1 visit 2 Index (UCB)
ID (Poconos/Catskills)
2Bn (Valley and Ridge)
2Cn (N. Appalachians)
3B (MA Coastal Plain)
*
*
4
8
*
*
3
7
*
*
3(7)
7 (12)
5
2
6
15
3
2
6
10
5(9)
2(6)
6(11)
12 (19)
pH < 5.5
Lower Node
visit 1 visit 2 Index (UCB)
Upper Node
visit 1 visit 2 Index (UCB)
ID (Poconos/Catskills)
2Bn (Valley and Ridge)
2Cn (N. Appalachians)
3B (MA Coastal Plain)
*
*
6
12
*
*
4
9
*d)
*
6 (11)
13 (21)
7
6
16
25
7
6
13
24
7 (12)
6(12)
13 (22)
24 (34)
(UCB) = Upper 95% Confidence Bound.
* - less than 1%.
223
-------
-------
SECTION 8
ION RELATIONSHIPS
8.1 OVERVIEW
Both natural and anthropogenic factors can influence the chemistry of streams and contrib-
ute to the intra- and inter-regional variation in their acid neutralizing capacity (ANC). These
factors include differences in (1) the input of strong bases generated through weathering,
(2) the input of strong mineral acids from natural (e.g., nitrification, sulfide oxidation) and
anthropogenic (acidic deposition) sources of such acids, and (3) the input of naturally occurrring
strong and weak organic acids leached from soils and wetland watersheds.
Because the stream survey data are for only a single point in time, and detailed informa-
tion on the geology, soils, and hydrology of each watershed is not yet available, it is difficult
to discern the sources of variation in acid/base chemistry. Such variation in stream water
chemistry may result from generic differences in watershed geochemistry (e.g., weatherable and
exchangeable pools of base cations) and hydrology (e.g., precipitation and evapotranspiration), as
well as from differences in anthropogenic and natural sources of acidity deposited on, or
generated within, the watersheds. Such survey data on the ion composition of stream water
can, however, provide inferential evidence concerning the role of different factors in causing
the observed patterns in ANC and pH within and among streams in the various subregions.
In this section, the relationships among various ions in stream water are examined to
assess the factors associated with the observed differences in the acid/base chemistry within and
among streams in the nine subregions sampled in the NSS-I. This is done through an examina-
tion of some bivariate and multivariate relationships that may be important in understanding the
role of certain variables in stream acidification.
8.2 BASE CATIONS AND MINERAL ACID ANIONS
Information on the concentration of total and individual base cations is useful in assessing
the role of various factors that potentially contribute to the between-stream differences in the
acid/base chemistry of water. Trilinear plots of anion and cation distributions for each sub-
region (Figures 8-1 [a - i]) are particularly useful for characterizing intra-regional patterns in
ion distribution and for assessing inter-regional differences and similarities in the composition of
stream water. The ion distribution diagrams show the relative contribution of the respective
cations and anions, expressed as a fraction of the total of those ions considered. For example,
streams in which calcium is the dominant base cation would be clustered in the lower left
portion of the base cation triangle (lower left), while streams dominated by sodium and/or
potassium would be clustered in the lower right portion of the triangle. The diamond in the
center of the trilinear plots illustrates the fractional contribution of the anions and cations
combined. Thus, streams in which calcium and/or magnesium are the dominant cations are
plotted on the left and upper portions of the diamond, whereas streams in which sodium and/or
potassium are the dominant cations are plotted on the right and lower portions. Similarly,
streams in which bicarbonate is the dominant anion are plotted on the left and lower portions
of the diamond, whereas those in which sulfate and/or chloride are the dominant anions are
plotted on the upper and right portions of the diamond.
These trilinear plots show that the cation composition varies from region to region but is
generally dominated by a mixture of calcium and magnesium or calcium, magnesium, sodium, and
potassium (Figures 8-1 [a - i]). Exceptions to this are in Florida (3C) (Figure 8-li) and in the
225
-------
Poconos/Catskills (1D)
\
\
\
CAT I ONS
AN I ONS
Figure 8-la. Trilinear plots of major cations and anions in stream water at the upper and
lower nodes combined in subregipn ID (Poconos/Catskills).
226
-------
Northern Appalachians (2Cn)
/
CAT I ON
CL >
AN I ONS
Figure 8-lb. Trilinear plots of major cations and anions in stream water at the upper and
lower nodes combined in subregion 2Cn (Northern Appalachians).
227
-------
Valley and Ridge (2Bn)
< CA
CAT I ONS
AN I ONS
Figure 8-lc. Trilinear plots of major cations and anions in stream water at the upper and
lower nodes combined in subregion 2Bn (Valley and Ridge).
228
-------
Mid-Atlantic Coastal Plain (3B)
\
CAT I ONS
CL >
AN I ONS
Figure 8-Id. Trilinear plots of major cations and anions in stream water at the upper and
lower nodes combined in subregion 3B (Mid-Atlantic Coastal Plain).
229
-------
Southern Blue Ridge (2As)
\
\
CAT I ONS
CL >
AN I ONS
Figure 8-le. Trilinear plots of major cations and anions in stream water at the upper and
lower nodes combined in subregion 2As (Southern Blue Ridge).
230
-------
Piedmont (3A)
\
\
\
CAT I ONS
CL >
AN I ONS
Figure 8-If. Trilinear plots of major cations and anions in stream water at the upper and
lower nodes combined in subregion 3A (Piedmont).
231
-------
Southern Appalachians (2X)
CAT I ONS
CL >
AN I ON
Figure 8-lg. Trilinear plots of major cations and anions in stream water at the upper and
lower nodes combined in subregion 2X (Southern Appalachians).
232
-------
Ozarks/Ouachitas (2D)
CAT I ONS
AN 1 ONS
Figure 8-lh. Trilinear plots of major cations and anions in stream water at the upper and
lower nodes combined in subregion 2D (Ozarks/Ouachitas).
233
-------
Florida (3C)
\
\
\
< CA
CAT I ON
CL >
AN I ONS
Figure 8-li. Trilinear plots of major cations and anions in stream water at the upper and
lower nodes combined in subregion 3C (Florida).
234
-------
Mid-Atlantic Coastal Plain (3B) (Figure 8-ld), where sodium and/or potassium are the dominant
cations in many of the streams. Sodium is a major fraction of the cation equivalents in both of
these coastal subregions. The fact that chloride is also an important contributor to the anionic
equivalence in both subregions suggests a sea salt contribution to the sodium and chloride
content of stream water in both regions, although anthropogenic (e.g., road salt), and geological
sources of chloride (groundwater flowing through salt-bearing formations) cannot be ruled out.
However, given the proximity of both subregions to the coast, it is likely that sea salt
deposition is the primary source of the higher sodium and chloride in stream water in both
subregions.
The trilinear plots also show that sodium and/or potassium are important in some streams
in the Ozark/Ouachitas (2D) (Figure 8-lh) and the Valley and Ridge (2Bn) (Figure 8-lc).
Because chloride is also an important contributing anion in some streams in these subregions and
the greater distance of the streams in subregion 2D from the coast compared to those in sub-
regions 3C and 3B, it seems likely that geological and/or anthropogenic sources of salt, rather
than sea salt deposition, are the primary causes of the higher contribution of sodium and
chloride to the ion equivalents in these streams.
The fractional composition of anions is quite variable among the subregions. In most of
the subregions, ANC, chloride, and sulfate dominate the anionic equivalents, although organic
anions and nitrate are important in some streams in the Mid-Atlantic Coastal Plain (3B) (Figure
8-2d) and the Southern Appalachians (2X) (Figure 8-2g). Since nitrate is generally found in low
concentrations in most pristine surface waters (Hem, 1985), these high nitrate levels suggest
either agricultural (fertilizer) runoff, runoff from mature forests that are losing nitrate perhaps
because of reduced nitrogen demand by the forest, or industrial/municipal inputs of wastes con-
taining nitrate. Sulfate is the dominant anion in most of the streams in the Northern Appala-
chians (2Cn) and the Poconos/Catskills (ID) (Figures 8-la,b; 8-2a,b]). Sulfate also contributes a
major fraction of the anion equivalents in several sample streams in the Valley and Ridge (2Bn)
and the Mid-Atlantic Coastal Plain (3B). In contrast, bicarbonate or bicarbonate plus chloride
are the dominant anions in most streams in the Southern Appalachians (2X), the Piedmont (3A),
the Ozarks/Ouachitas (2D), and the Southern Blue Ridge (2As) Figures 8-1 [e - h]). Bicarbonate
is also the major anion in over half of the sample stream reaches in subregion 2Bn (Figure
8-lc). These plots thus show major differences in the ion composition of streams among the
subregions, with sulfate being a major anion in many or most of the streams in the more nor-
thern subregions (2Cn and ID). Streams in the Interior Southeast subregions (2X, 2D, 3A, and
2As), however, are dominated by bicarbonate or bicarbonate plus chloride. Streams in Florida
(3C) are dominated by chloride, reflecting the apparent effect of sea salt deposition. Streams in
subregions ID and 2Bn exhibit no clear pattern, with some streams dominated by bicarbonate or
sulfate and many of the streams containing almost equal fractions of bicarbonate, sulfate, and
chloride (Figures 8-1 [a and c]).
The total ion content as well as the distribution of the major inorganic cations and
inorganic and organic anions in stream water can also be examined as a function of pH or ANC
classes by means of so-called ion fingerprint plots (Figures 8-2 [a - i]). These figures illustrate
the population-weighted average distribution of ions in absolute concentrations in streams of
different pH. The plots thus show not only the average ion composition of streams in each pH
class in the various subregions but also the changes in average ion composition over the range
of pH. In these plots, organic anions were assumed to be equal to the anion deficit.
The ionic strength of stream water in most of the regions exhibits a general pattern of
decline with decreasing pH, although it does not decline monotonically in all subregions. The
ionic content of streams in the lowest pH class in subregion 3C, for example, increases sharply,
235
-------
POCONOS /CATSKILLS
LOWER UPPER
>7.0
6.5-7.0
6.0-6.5
5.5-6.0
5.0-5.5
<5.0
(0
CO
1
70.96
24.20
1.92
2.63
0.30
51.70
19.92
15.53
5.95
2.06
4.83
sof-
NO5
CI"
HCOs
RCOO'
0 500 1000 1500 0 500 1000 1500 ANIONS
ANION EQUIVALENTS
Mg
2*
K+
H +
0 500 1000 1500 0 500 1000 1500 CATIONS
CATION EQUIVALENTS
-------
NORTHERN APPALACHIANS
LOWER UPPER
(0
CO
_j
o
'X
a
<5.0
32.72
26.83
17.53
9.85
7.31
5.76
0 500 1000 1500 0 500 1000 1500
ANION EQUIVALENTS (p«q L~1)
5.0-5.5
<5.0
0 500 1000 1500 0 500 1000 1500
CATION EQUIVALENTS (}ieq L~1)
SOj"
NO;
cr
HCOJ
RCOQ-
ANIONS
CATIONS
Figure 8-2b. Weighted average concentration of major anions and cations by pH class of
stream water at the upper and lower nodes in subregion 2Cn (Northern
Appalachians). The numbers next to the anion bars give the estimated
percentage of streams in each of the pH classes.
237
-------
VALLEY AND RIDGE
LOWER UPPER
>7.0
6.5-7.0
6.0-6.5
5.5-6.0
5.0-5.5
<5.0
CO
CO
_i
O
X
O.
76.93
22.73
0.34
1
1
54.72
31.70
2.51
5.24
3.40
2.44
1000 2000 0 1000 2000
ANION EQUIVALENTS (|i«q L >
>7.0
6.5-7.0
6.0-6.5
5.5-6.0
5.0-5.5
<5.0
1000 2000 0 1000 2000
CATION EQUIVALENTS
-------
MID-ATLANTIC COASTAL PLAIN
LOWER UPPER
>7.0
6.5-7.0
6.0-6.5
5.5-6.0
5.0-5.5
<5.0
37.15
21.88
19.33
8.46
6.35
6.84
CO
U
1000 2000 0
ANION EQUIVALENTS (|J0q L )
<5.0
1000 2000 0 1000 2000
CATION EQUIVALENTS (}ieq L~1)
soj
cr
HCO;
RCOO'
1000 2000 ANIONS
-1,
Na+
K+
CATIONS
Figure 8-2d. Weighted average concentration of major anions and cations by pH class of
stream water at the upper and lower nodes in subregion 3B (Mid-Atlantic Coastal
Plain). The numbers next to the anion bars give the estimated percentage of
streams in each of the pH classes.
239
-------
SOUTHERN BLUE RIDGE
LOWER UPPER
>7.0
6.5-7.0
6.0-6.5
5.5-6.0
5.0-5.5
<5.0
to
(0
<
O
I
%
1
I
K
3 59.40'
1 39.28
i
1.32
^
y\
£ \
; 1
; |
5
1
h
5 39.36
I 57.48
3.17
>7.0
6.5-7.0
soj
HCOs
RCOO"
0 200 400 600 0 200 400 600 ANIONS
ANION EQUIVALENTS (|i«q L~1)
NH4
0 200 400 600 0 200 400 600 CATIONS
CATION EQUIVALENTS (peq L~1)
Figure 8-2e. Weighted average concentration of major anions and cations by pH class of
stream water at the upper and lower nodes in subregion 2As (Southern Blue
Ridge). The numbers next to the anion bars give the estimated percentage of
streams in each of the pH classes.
240
-------
PIEDMONT
LOWER UPPER
>7.0
6.5-7.0
6.0-6.5
5.5-6.0
5.0-5.5
<5.0
49.55
33.63
12.61
4.20
20.12
54.65
1
14.71
10.51
tf>
200 400 600 0 200 400 600
ANION EQUIVALENTS (peq L~1)
<5.0
0 200 400 600 0 200 400 600
CATION EQUIVALENTS
L-1)
SO}"
NOB
cr
HCOa
RCOO'
ANIONS
Na+
CATIONS
Figure 8-2f. Weighted average concentration of major anions and cations by pH class of
stream water at the upper and lower nodes in subregion 3A (Piedmont). The
numbers next to the anion bars give the estimated percentage of streams in each
of the pH classes.
241
-------
SOUTHERN APPALACHIANS
LOWER UPPER
>7.0
6.5-7.0
6.0-6.5
5.5-6.0
5.0-5.5
<5.0
CO
W n
1 k
441 8 )
0 76.00
1 12. 00
9.60
2.40
• 1 i 1 • 1 •
•inn A «>nr
!
i
j
?
i
I
in n
;;' O
- 8
| 65.57
L 17<22
9.84
B 4-92
2.46
i | i i . i
•innn on
o
X
Q.
>7.0
<5.0
0 1000 2000 0 1000 2000
CATION EQUIVALENTS (}ieq L~1)
sof-
NO;
cr
HCOa
RCOO"
2000 ANIONS
ANION EQUIVALENTS (peq L~1)
K+
CATIONS
Figure 8-2g. Weighted average concentration of major anions and cations by pH class of
stream water at the upper and lower nodes in subregion 2X (Southern Appala-
chians). The numbers next to the anion bars give the estimated percentage of
streams in each of the pH classes.
242
-------
OZ ARKS /OUACHITAS
LOWER UPPER
>7.0
6.5-7.0
6.0-6.5
5.5-6.0
5.0-5.5
<5.0
i
tn
CO
-i
o
a
38.10
49.14
10.94
1.82
19.75
48.06
21.48
7.14
3.57
400 800 0 400 800
ANION EQUIVALENTS (peq L~1)
>7.0
400 800
soj-
HC03
RCOO
ANIONS
CATION EQUIVALENTS ((ieq L )
Figure 8-2h. Weighted average concentration of major anions and cations by pH class of
stream water at the upper and lower nodes in subregion 2D (Ozarks/Ouachitas).
The numbers next to the anion bars give the estimated percentage of streams in
each of the pH classes.
243
-------
FLORIDA
LOWER UPPER
>7.0
6.5-7.0
6.0-6.5
5.5-6.0
5.0-5.5
<5.0
6.18
8.29
23.68
38.01
9.36
14.48
5.56
22.60
21.93
18.74
31.17
SO*'
NO5
cr
HCOs
RCOO'
V)
O
1000 2000 0
ANION EQUIVALENTS (|ieq L~1)
1000 2000 ANIONS
>7.0
6.5-7.0
6.0-6.5
5.5-6.0
5.0-5.5
<5.0
Mg:
Na
NH;
0 1000 2000 0 1000 2000 CATIONS
CATION EQUIVALENTS (jieq L~1)
Figure 8-2i. Weighted average concentration of major anions and cations by pH class of
stream water at the upper and lower nodes in subregion 3C (Florida). The
numbers next to the anion bars give the estimated percentage of streams in each
of the pH classes.
244
-------
due primarily to increases in sodium, chloride, and organic anions (Figure 8-2 [a - i]). Because
of the proximity of Florida streams to the coast, this suggests a sea salt influence on the ion
composition of these streams.
In all of the other subregions, ionic strength either decreases monotonically with pH across
the pH classes, or exhibits a biomodal distribution. Because streams of low ionic strength also
have low concentrations of base cations, they also have a lower ANC and hence include those
streams considered most sensitive to acidification. The association of lower pH and ANC with
lower base cation concentrations is not in itself evidence of anthropogenic acidification, how-
ever, because the decline to a level approximately equal to the equivalence point for carbonic
acid may simply reflect generic differences among watersheds in the production and export of
alkalinity. Such differences could be due to variation among watersheds in the export of bases
caused by geologic and hydrologic factors that control weathering and exchange of base cations
associated with soils and bedrock.
The monotonic decline in ionic strength with decreasing pH is most obvious at both the
upper and lower nodes in the Poconos/Catskills (ID) (Figure 8-2a), the Ozarks/Ouachitas (2D)
(Figure 8-2h), and the Piedmont (3A) (Figure 8-2f), and in the lower nodes in the Valley and
Ridge (2Bn) (Figure 8-2c), the Southern Appalachians (2X) (Figure 8-2g), and the Southern Blue
Ridge (2As) (Figure 8-2e). This pattern is an indication that the differences in pH among
streams with pH > 5.5, is due in part to generic differences among the watersheds in the
generation of alkalinity through the weathering of base cations by carbonic acid—not solely due
to differences in the extent of acidification resulting from inputs of strong acids. The ionic
strength of stream water would be expected to remain the same, or to increase with declining
pH, if the decrease in pH was due entirely to titration in which bicarbonate is replaced by
strong acid anions, eventually resulting in an increase in acidic cations. For streams with pH
far enough below the equivalence point for carbonic acid (« 5.5), differences in carbonic acid
weathering among watersheds cannot explain the low pH, because at this level, strong acids
predominate (ANC < 0). It should be emphasized however, that use of the average ion composi-
tion of streams among pH classes may obscure major differences in the ion composition among
streams within a given pH class. Hence, the presence of a monotonic decline in ionic strength
with declining pH in a subregion should not be taken as evidence that no acidification has
occurred.
If the differences in pH and ANC between streams in the different pH classes were due
mainly to differences in recent, ongoing inputs of strong acids from the watersheds, and base
cation export rates remained constant, the ionic content of stream water would be expected to
remain the same or to increase with decreasing pH. In this case, as the carbonate alkalinity is
replaced by strong acid anions (SO42~, NO3~, Cl~) and the concentration of acidic cations (H+,
Aln+) increases, ionic strength increases or remains constant. The fact that the observed ionic
strength tends to decrease across stream groupings of decreasing pH suggests that ongoing titra-
tion of the alkalinity is not occurring in the subregions where this pattern exists, unless base
cation depletion is occurring.
The inherent or pre-existing differences in alkalinity generation among watersheds could be
due entirely to natural differences in the geological, soil, and hydrological characteristics of
each watershed, for example, differences in pool sizes of readily weatherable and exchangeable
base cations, differences in hydrologic flow paths of water through the watershed, or differences
in the production of natural organic acids. However, they could also be a result of anthropo-
genically caused differences: long-term deposition of strong acids and a resulting leaching of
base cations from the soil. In areas where the base saturation of soils is low, a decline in the
export of base cations could result from the reduction in the pools of readily weatherable or
245
-------
exchangeable base cations due to soil acidification (i.e., reduction in base exchange capacity)
(Reuss and Johnson, 1985).
The ion composition of stream water in those regions where ionic strength does not decline
monotonically with decreasing pH suggests major geochemical or hydrologic differences between
watersheds, major differences in atmospheric deposition of materials on the watershed, or both.
Evidence of such intra-regional generic differences is illustrated by the data on cation and anion
composition of stream water in the Valley and Ridge (2Bn) (Figure 8-2c) and the Northern Appa-
lachians (2Cn) (Figure 8-2b). At the upper node in subregion 2Bn, streams with a pH > 7 have
much higher calcium and magnesium concentrations than those with lower pH. The second peak
in ionic strength, however, is due primarily to increases in sodium and chloride, suggesting an
internal source of salt (e.g., road salt) in some of the watersheds in these intermediate pH
classes. This pattern for calcium and magnesium content at the upper node indicates that
streams with a pH > 7 drain watersheds with some carbonate or dolomitic bedrock, whereas
those with a lower pH drain watersheds with little'or no carbonate bedrock.
The pattern of increasing sodium and chloride content along the gradient between the pH >
7 group and the pH 5.5 - 6.0 group in subregion 2Bn (Figure 8-2c) suggests an internal source
of chloride in these watersheds. Although the exact source is unknown, runoff of road salt and
inputs of groundwater that has flowed through salt-bearing geological formations are suspected
causes. The Valley and Ridge Province is known to contain marine shales that have a relatively
high content of teachable sodium and chloride (Von Damm, 1987) and some of the watersheds
have roads that may have been de-iced with road salt.
Both the upper and lower nodes of streams sampled in the Northern Appalachians (2Cn)
(Figure 8-2b) and Florida (3C) (Figure 8-2i) and the upper nodes of streams in the Mid-Atlantic
Coastal Plain (3B) (Figure 8-2d) and the Southern Appalachians (2X) (Figure 8-2g) exhibit
bimodal distributions in ionic strength. In subregion 2Cn, the bimodal distribution of cations is
due primarily to changes in calcium and magnesium content, while the trend in the anionic sol-
ute distribution is caused primarily by changes in sulfate and bicarbonate. This pattern could be
caused by several factors, among them a high deposition rate of sulfate acidity or an internal
source of sulfate acidity (e.g., pyrite oxidation), which enhanced the export of calcium and
magnesium from the watersheds following increased weathering and/or cation exchange, an
anthropogenic source of calcium that buffered some of the acidity (liming streams), or some
combination of both.
In the Florida streams, the bimodal distribution in ionic content is associated primarily
with changes in the calcium, sodium, organic anions, and chloride concentration (Figure 8-2i).
The decreasing gradient in ionic strength between the pH > 7 group and the pH 5.5 - 6.0 group
appears to be associated primarily with decreases in calcium, whereas the increasing ionic
strength in the pH < 5.0 group is due primarily to increases in sodium, chloride, and organic
anions. Because of the proximity of the Florida streams to the ocean, this pattern for sodium
and chloride suggests that sea salt is probably the source of these ions. Whether these sources
of sodium and chloride are related to wet and dry deposition, land use, or geologic sources is
unknown. It is also possible that the low pH (< 5.0) is due to organic acidity as evidence by
the large amount of organic anions in this pH group. The potential sources of acidity in Florida
(3C) streams are discussed in more detail later in this section and in Section 9.
8.3 ALUMINUM
Cronan and Schofield (1979) hypothesized that mineral acids from atmospheric deposition
have altered the natural process of soil weathering and development by augmenting the export
246
-------
of aluminum from soil to surface waters. The mechanism responsible for this increased export
of aluminum may involve cation exchange in soils with low base saturation, dissolution of
primary and secondary mineral phases of aluminosilicate minerals, or some combination of both.
Supporting evidence for the effect of acidic deposition on the mobilization and "export of
aluminum from watersheds is that in acid-sensitive regions receiving high atmospheric loadings
of strong mineral acids, high concentrations of both H+ and aluminum are observed in surface
waters when the ratio of acidic anion equivalents (sulfate and nitrate) to basic cations exceeds
one (Driscoll and Newton, 1985). Because the change in this ratio from < 1 to > 1 reflects a
shift from carbonic acid weathering to mineral acid weathering, dissolution at low pH of
aluminum in clay minerals by strong mineral acids is generally invoked to explain the aluminum
increase in surface waters.
Because of the dependency of aluminum solubility on pH, extractable aluminum concentra-
tion is expected to increase with decreasing pH, particularly when the pH declines to < 5.5.
Plots of both extractable (MIBK) aluminum and an estimate of inorganic monomeric aluminum
(subsection 6.3.2) versus pH of streams in the NSS-I show that, in general, this relationship
holds, with streams having a pH > 5.5 typically containing low and relatively uniform concentra-
tion of extractable and inorganic monomeric aluminum, whereas those with a pH < 5.5 exhibit
higher concentrations of aluminum (Figure 8-3b,c). In contrast, total aluminum concentrations
showed no relationship with pH, indicating that most of the aluminum in stream water is in a
colloidal or particulate form (Figure 8-3a).
Aluminum appears to be toxic to fish at concentrations above the range of 100 to 200
jig L"1, levels commonly observed in acidic surface waters. The extent of toxicity, however,
appears to be dependent on the form of aqueous aluminum. Driscoll et al. (1980) and Baker and
Schofield (1982) reported that aluminum toxicity to brook trout fry was greatly reduced when
aluminum was complexed to organic matter. They also reported that fish survival was highly
correlated with inorganic monomeric aluminum concentrations and pH. Because the labile or
inorganic monomeric aluminum is thought to be the most toxic fraction, its concentration in
water is of particular concern. At pH < 5.5, inorganic monomeric aluminum concentration is
related to pH (Figure 8-3). However, only in the Mid-Atlantic Coastal Plain (3B) do streams
with inorganic monomeric aluminum concentrations in excess of 100 /zg L"1 (3.7 fiM) comprise
more than 10% of the target population (Table 6-6). Population estimates for the various
aluminum fractions are presented in subsection 6.3.2.
The concentration of nonlabile or organic aluminum (nonexchangeable, PCV technique) in
streams sampled in the NSS-I is highly correlated with DOC concentration (Figure 8-3), a
pattern consistent with that observed by Driscoll et al. (1984) for surface waters in the
Adirondacks and by Turner et al. (1985) for streams in the New Jersey Pine Barrens. The
nonlabile aluminum in stream water could be caused by leaching of organically complexed
aluminum from plant litter and organic debris, complexation by dissolved organic matter of
inorganic aluminum mobilized from soil minerals at low pH, or some combination of both.
8.4 SOURCES OF VARIATION IN ANC OF STREAM WATER
Although sulfate, nitrate, and chloride may not make up a major fraction of the anion
equivalents in streams with pH > 7.0 (e.g., many Southeast streams), strong mineral acids may
still contribute to the differences in pH and ANC among these streams. To determine the
relative importance of mineral acid anions, organic anions, and base cations to the variation in
ANC among streams, standardized multiple regression analysis (SAS, 1985) was conducted.
Because NSS-I subregions are not extremely heterogeneous and because weighting factors do not
247
-------
30-
~25-
«~
i
o 20-
1.
N^X
1 15"
C
|to-
•
I 5.
.
0-
. ' (a)
" •" * * • .
.*. •• •
• , %
— * .
.. /
• • •
• •' '.' '. . .-*•' ' .
. ; t^ . .
I . .. • A'* ' •••
... :. .*v. -
•* I *'":'../• V:" •
'•I * •' .r'..v. *\'*l*' .*••"'• •',' '
' •' ^V^fefv'
••\ •^"t^iS^&i'a'w" -"
* *• y ™ ***• Sls^i'^tP^' * " *
• • • : * •«s*;r.7v. '•
1 ' -1 • 1 ' . 1 ' 1 ' 1
456 7 8 9
15-
T
0
£
3 10-
£ :
3
C
£ :
3
I 5:
S
o
o ;
^ '.
u
0-
(b)
* • •
•
-
•
• *f *
•
• •
• * * • *
• , •
."••.. %
* •** * • * • •
* * '*???* •<**?* v" ». \ '.
• 4? ^f! ^^TfM^il* ^V-1^- vCijf * *
*••• -..iaswiSwSi^'.'-'.
i • i • i • i • i • i
4 5 67 89
pH
PH
15 i
^
Li
3. ;
| j
c
1 5^
.-= ;
o ;
n -
u
(c)
•
*
• • '
• .
• * • "
* •
/ '•'*'.-- '
• * * • •
* ' •"$<• >' *
" ' • .••\'>«&v&O!tLf!yi:tt*2>i '
1 ' 1 ' 1 ' 1 ' 1 ' 1
456789
10-
pH
o
£
|5H
O
"E
O
0-
i
(d)
r * r
0 10 20 30
Dissolved Organic Carbon (mg L~ )
\
10
I
40
\
50
Figure 8-3. Plots of (a) total aluminum vs. pH, (b) extractable (MIBK) aluminum vs. pH, (c)
labile or inorganic monomeric aluminum vs. pH, and (d) organic aluminum vs.
DOC in streams of the NSS-I.
24*
-------
vary substantially within each NSS-I subregion, regressions employing unweighted sample data
will not differ substantially from descriptions of population characteristics within NSS-I
subregions. When weighting factors differ substantially, as they do among sample streams from
different NSS-I subregions, multivariate models based on unweighted sample data may incorrectly
describe a relationship for the regional stream population. This section describes the standard-
ized multivariate regression analyses conducted within separate NSS-I subregions, removing the
absolute necessity for using population weighting.
The standardized regression coefficient (bjO is defined as:
where bj is the partial regression coefficient for the independent variable i, and sxj and Sy are
the standard deviations of the independent variable xj and the dependent variable y, respec-
tively.
Because the standard deviations of the dependent and independent variables are generally
not the same, and because not all of the independent variables are measured in the same units
(e.g., acid anions are measured in /neq L"1, DOC is in mg L"1), the partial regression coefficient
does not provide a relative measure of the influence of each independent variable (sulfate,
nitrate, chloride, base cations) on the dependent variable, ANC. By multiplying the regression
coefficient (bj) by the ratio of the standard deviations of xj to y, the problems of unequal
variation and different units of measurement are eliminated because the standardized regression
coefficient has a unitless dimension that is normalized for differences in variance between the
dependent and independent variable.
Comparison of the standarized regression coefficients thus provides a means to rank and
compare the relative importance of each independent variable in terms of its ability to explain
the variation in ANC among streams in each subregion. If the partial regression coefficient for
sulfate is twice as great as that for nitrate, then sulfate can be considered to be twice as
important as nitrate in terms of explaining the variation in ANC. The direction of this
influence (i.e., positive or negative) is indicated by the sign of the slope. The results of these
standardized multiple regression analyses describe associations among chemical species in the
sample streams. They cannot be used to infer the cause of the observed ANC variation.
The relative influence on ANC of base cations, sulfate, nitrate, chloride, and organic
anions, as reflected by DOC, was assessed using this approach. DOC was used as a surrogate
for organic anions because the only estimates of the strong and weak acid organic anions are
based on the calculated anion deficit and the difference between the measured ANC and the cal-
culated carbonate alkalinity, respectively. The estimated anion deficit includes the other acid
anions and base cations in the calculation and hence is not independent of the other indepedent
variables used in the regression model. Further, the estimated strong acid organic anion con-
centration, as computed from the charge balance, is affected by the cumulative analytical errors
associated with the measurement of the individual anions and cations used in the charge balance.
Thus, for this purpose, and assuming that the strong organic acid anions are a constant propor-
tion of DOC among the streams, strong organic acidity is probably more accurately estimated by
DOC than by the charge imbalance (anion deficit). Similarly, the weak organic anions (estimated
from the difference between the measured ANC and the calculated carbonate alkalinity) are not
independent of ANC. Hence, they were not used in the regression analysis.
The absolute values of the standardized regression coefficient were plotted for the upper
and lower nodes in each subregion (Figures 8-4, 8-5, and 8-6). In all nine subregions, the five
independent variables together account for over 90% of the variation in ANC between streams,
249
-------
Pocono/Catskills
BC
so?-
ci-
NOJ
DOC
ta Upper Node (R2> 0.99)
• Lower Node (R2 > 0.99)
b=1.58, n.s.
b = -19.08
0.0 0.5 1.0
Northern Appalachians
1.5
BC
= -2.26,n.s.
= 21.56,p<0.05
tH Upper Node (R2 = 0.99)
• Lower Node (R2 = 0.99)
0 1
Valley and Ridge
BC
Cl
El Upper Node (R2 > 0.99)
• Lower Node (R2 > 0.99)
DOC
b = 7.26, p< 0.05
= 5.12,n.s.
0 1 2
STANDARDIZED REGRESSION COEFFICIENT
Figure 8-4. The absolute value of the standardized linear regression coefficient (bj') for each
of the independent variables that influence ANC in all streams in the population
of interest in the Poconos/Catskills (ID), Northern Appalachians (2Cn), and
Valley and Ridge (2Bn) subregions. Also shown is the estimated regression
coefficient (b) (i.e., slope) for each independent variable. Unless otherwise indi-
cated, the standardized regression coefficients were highly significant (bj' ^ 0, p
< 0.001) for the nodes plotted.
250
-------
Mid-Atlantic Coastal Plain
BC
C!
NOJ
DOC
| b = -0.91
= -1.07
Upper Node (R2 = 0.94)
Lower Node (R2 = 0.99)
0 1
Southern Blue Ridge
BC
Upper Node (R2 > 0.99)
Lower Node (R2 > 0.99)
DOC
= 8.10n. s.
b = 25.32, p < 0.05
0.0 0.2 0.4
Piedmont
0.6 0.8 1.0
1.2
BC
= -1.00
= -1.06
b = -0.82, n.s.
b = -0.64, p < 0.01
b =-0.94, p< 0.10
b = -1.04, p<0.01
E Upper Node (R2 = 0.92)
• Lower Node (R2 = 0.99)
DOC
b = 8.98, n. s.
b = -2.49, n. s.
0.0 0.2 0.4 0.6 0.8 1.0 1.2
STANDARDIZED REGRESSION COEFFICIENT
Figure 8-5. The absolute value of the standardized linear regression coefficient (bjO for each
of the independent variables that influence ANC in all streams in the population
of interest in the Mid-Atlantic Coastal Plain (3B), Southern Blue Ridge (2As),
and Piedmont (3A) subregions. Also shown is the estimated regression coeffic-
ient (b) (i.e., slope) for each independent variable. Unless otherwise indicated,
the standardized regression coefficients were highly significant (b/ ^ 0, p <
0.001) for the nodes plotted.
251
-------
Southern Appalachians
BC
b =-17.41, p< 0.05
b =-15.84, n. s.
H Upper Node (R2 > 0.99)
• Lower Node (R2 > 0.99)
DOC
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Ozarks/Ouachitas
BC
Upper Node (R2 > 0.99)
Lower Node (R2 = 0.99)
b = -1.01
b =-1.00, p< 0.005
= -7.74
b = -5.25, n. s.
0 Upper Node (R2 = 0.95)
• Lower Node (R2 = 0.97)
DOC
b = -4.14
b =-4.95, p < 0.01
0.0 0.5 1.0 1.5
STANDARDIZED REGRESSION COEFFICIENT
Figure 8-6.
The absolute value of the standardized linear regression coefficient (bjO for each
of the independent variables that influence ANC in all streams in the population
of interest in the Southern Appalachians (2X), Ozarks/Ouachitas (2D), and
Florida (3C). Also shown is the estimated regression coefficient (i.e., slope) for
each independent variable. Unless otherwise indicated, the standardized regres-
sion coefficients were highly significant (b/ ^ 0, p < 0.001) for the nodes
plotted.
252
-------
and in most cases account for > 99% of the variation. The sum of base cations, however, ranks
as the variable of most importance in explaining the observed differences in ANC among streams.
The only subregion where a strong mineral acid anion is of comparable importance to base
cations is the Northern Appalachians (2Cn) (Figure 8-4), where sulfate is only slightly less
important than base cations.
These results indicate that when all of the streams in the subregional populations of
interest are considered together, sulfate is of major importance in explaining the variation in
ANC among streams only in subregion 2Cn. In all other subregions, the contribution of sulfate
as a predictor of variation in ANC among streams is substantially less compared to that for the
base cations.
The important role of base cations in explaining differences in ANC among streams is to be
expected when all streams in the target population are considered together, because of the wide
range of base cations relative to the range of ANC. That differences in ANC among streams are
associated primarily with differences in base cation concentration suggests that differences in
weathering and exchange rates in soils and bedrocks among watersheds are the primary mechan-
ism regulating ANC. Such differences could be caused by kinetic limitations associated with the
extent to which drainage waters are in contact with weatherable and exchangeable pools of base
cations in the soil and groundwater. These differences could also be due, in part, to a capacity
limitation associated with differences in the pools of weatherable and exchangeable bases in the
soil and bedrock (e.g., base-poor versus base-rich soils).- Factors associated with the variation
in the more acid sensitive streams (ANC < 200 neq L"1) are examined later in this section.
The apparent importance of base cations as a source of the observed variation in ANC
among streams does not mean that mineral and organic acid anions do not contribute to that
variation. In fact, the standardized regression analysis shows a highly significant negative
relationship (p < 0.0001) between ANC and sulfate in all subregions, with the regression
coefficient (b) being equal to or close to the theoretically expected value of -1.0.
Chloride also shows a significant negative effect on alkalinity in all the subregions except
at the upper node in the Southern Appalachians (2X) and the upper node in the Piedmont (3A).
Further, the slope is close to the theoretically expected value (-1.0). In the Mid-Atlantic
Coastal Plain (3B), Pocono/Catskills (ID), Valley and Ridge (2Bn), and Florida (3C), the
importance of chloride is comparable to or greater than that of sulfate in explaining the
variation in ANC among streams.
Similarly, nitrate appears to exhibit a significant negative effect on ANC in all the
subregions except at the upper nodes in subregion 3A and at the lower nodes in subregion 2X.
Nitrate, however, is of less importance than base cations, sulfate, and chloride in most of the
subregions. Only in subregion 3B is nitrate of comparable or greater importance than sulfate.
The estimated slope for nitrate is also close to the theoretical value of -1.0.
In summary, analysis of the effect of mineral acid anions on the between-stream variation
in ANC indicates that mineral acids play a less important role than base cations in explaining
the overall variation in ANC among streams in all regions except for the Northern Appalachians
(2Cn). In terms of the relative importance of the mineral acid anions, sulfate is at least twice
as important as nitrate and chloride in subregions 2Cn, 2X, and 2D, but is of comparable or
lesser importance than chloride or nitrate in subregions 2Bn, ID, 3B, and 3C. The importance of
organic acids in explaining the among-stream variation in ANC appears to be relatively minor in
all subregions except at the upper node in 3C, where DOC is more important than any of the
mineral acid anions and is within a factor of 1.7 of being comparable to base cations in
importance. The role of organic acid anions in stream acidification is discussed further in the
next section.
253
-------
8.5 ORGANIC ANIONS
Surface water may contain varying amounts of strong and weak acids originating within the
watershed. The weak acids are organic compounds and aluminum, whereas the strong acids may
include organic compounds that behave like strong acids or mineral acids from natural sources
(e.g., nitrification, sulfide oxidation). To assess the role of organic acids in the acidification of
surface waters requires an analysis of the acid/base characteristics of dissolved organic matter
(i.e., organic anions) to determine the amounts and relative contribution of the weak and strong
organic acid anions. Weak organic anions are protolytic (react with H+) and, depending on the
pKa of the weak acids, may behave like bicarbonate in terms of their reaction with protons.
Inputs of weak acid organic anions (WAOA) thus can cause an increase in the ANC. In con-
trast, strong organic anions are nonprotolytic (do not react with H+ for pH > 4.0) and behave
like SO^2" and NO3" in terms of their proton dissociation properties. Because inputs of strong
acid organic anions (SAOA) cause a decline in ANC, the primary concern is in assessing whether
the effect of inputs of organic anions is acidifying (decrease in ANC) or alkalinizing (increase in
ANC).
As discussed in Section 8.4, DOC is either a minor or nonsignificant contributor to the
overall variation in ANC in all the subregions except Florida (3C), based on standardized
regression analysis of the relationship between ANC and DOC. Although, this is only circum-
stantial evidence of the role of strong organic acids, it suggests that, except for Florida,
organic acids are not a major source of the subregional variation in ANC among streams in any
of the subregions. This does not preclude the possibility that individual streams or groups of
streams (e.g., high organic, acidic streams) may be acidic because of organic acids. Because
DOC was used as the independent variable and DOC can include both strong and weak acid
organic anions, the effect of such anions on the ANC measured by Gran titration could be posi-
tive, negative, or neutral, depending on the proportion of strong and weak organic anions, the
dissociation constant of these acids, and the pH of stream water. If the proportion of DOC
made up of strong and weak acids varies among streams within a subregion, it is possible that
the effect of these organic acids on the ANC within a subregion is masked when data for all
streams are combined in the regression analysis.
The net effect of organic anions on the ANC of stream water is indicated by the results of
the standardized regression analysis. This analysis shows whether there is a statistically
significant relationship between the concentration of DOC and the variation in ANC among
streams in each of the subregions. It also shows whether the relationship is negative (b < 0) or
positive (b > 0), the former indicating that the DOC consists primarily of strong acid organic
anions which depress the ANC, the latter indicating that the DOC consists primarily of weak
organic protolytes that contribute to the noncarbonate component of the ANC.
At the upper nodes in Florida (3C), there is a significant negative relationship between
DOC and ANC, suggesting that strong organic acids may be partly responsible for the low ANC
in streams in this subregion (Figure 8-6). The regression coefficient between ANC and DOC is
-4.14, indicating that the ANC declines by 4.14 jueq for every mg increase in DOC in stream
water. The regression analysis suggests that DOC is more important than sulfate, chloride, and
nitrate in explaining the variation in ANC among streams at the upper nodes in Florida. At the
lower nodes, the effect of DOC on ANC is also negative; however, DOC is slightly less
important than chloride and sulfate in explaining the variation in ANC among streams. This
suggests that there are major changes in the organic acid/ANC relationship between the
upstream and downstream nodes in subregion 3C streams.
254
-------
Given the relatively high concentrations of DOC in many streams in the Mid-Atlantic
Coastal Plain (3B), it is surprising that DOC ranks lowest in importance of the variables
considered in explaining the variation in ANC among streams. Further, DOC accounts for a
significant fraction of the variation in ANC in subregion 3C streams only in the lower reach
nodes. The negligible effect of DOC on ANC in this subregion may be due to the use of bulk
DOC as the independent variable in the regression model. If the equivalents of strong and weak
acid organic anions per mole of DOC is quite variable among streams, use of bulk DOC as a
measure of organic anions would tend to obscure the effect of such differences on the ANC.
An additional reason why DOC may not explain the subregional variance in ANC in the
coastal plain is that the variation in DOC is small relative to the variation in ANC. Comparison
of the partial regression coefficients and the standardized regression coefficients at the upper
and lower nodes in subregion 3B indicates that this is the case. This observation points out a
potential pitfall in the interpretation of these standardized regression analyses. If, for example,
ANC in all streams in a subregion is depressed the same amount by equal concentrations of a
strong acid anion, the variation in ANC is likely to be best explained by among-stream base
cation concentration (buffering) or by a different strong acid anion that shows greater variation.
The partial regression coefficients show a high potential for DOC to influence ANC as indicated
by the negative slopes of -6.22 and -1.57 for the lower and upper nodes, respectively. The
partial regression coefficient is not significantly different from zero at the upper node,
suggesting that the variation in the concentration of DOC in this subregion is small relative to
that for ANC. As such, DOC would not be expected to account for a significant fraction of the
variation in ANC among streams. This analysis, however, assumes that the equivalents of strong
acid organic anions per mg of DOC are the same among all streams. The variation among the
regions in both the sign and magnitude of the partial regression coefficients for DOC indicates
that this is not a valid assumption. Thus, it may be tenuous to assume that the DOC in all
streams within a subregion is the same in terms of the fraction that behaves as a strong acid.
Strong acid organic anions (SAOA) alone were also estimated from the difference in the
charge balance (measured cations and anions):
SAOA = [sum of measured cations] - [sum of measured anions],
where [sum of measured cations] = Ca2+ + Mg2+ + Na+ + K+ + NH4+ + H+, and [sum of measured
anions] = SO42" + NO3" + Cl~ + F" + (ANC + H+). ANC + H+ includes both carbonate and non-
carbonate protolytes (ANC = HCO3" + CO32" + OH~ + weak noncarbonate protolytes - H+). The
anion deficit was then regressed against DOC and the results of this analysis are shown in Table
8-1. Only regressions with slopes significantly different from zero (p > 0.05) are shown. The
slope of these regressions is a measure of the strong organic acid equivalents per mg of DOC.
In all cases, positive slopes were observed, indicating a significant positive relationship
between anion deficit and the concentration of DOC in stream water. The slopes, however, vary
from region to region and from the upper to the lower nodes within a subregion, ranging from
4.0 at the upper nodes in the Ozark/Ouachitas (2D) to 12.4 /zeq mg'1 DOC at the upper nodes in
the Southern Blue Ridge Province (2As).
. This three-fold range in the regression coefficient supports the results of the standardized
regression analysis which showed that the equivalents of strong acid organic anions per mg of
DOC varied not only from region to region, but also between the upper and lower nodes within
a region. The high slope for the Southern Blue Ridge (2As) implies that for every mg increase
in DOC in stream water, the ANC will decline by approximately 12.4 jueq. However, the DOC
255
-------
Table 8-1. Estimates of the Strong Acid Organic Anion Equivalents per mg Organic Carbon
in Stream Water at the Upper (U) and Lower (L) Sample Reach Nodes in NSS-I
Subregions Based on Linear Regression of the Calculated Anion Deficit and DOC
SUBREGION/NODE$
Poconos/Catskills (L)
Mid-Atlantic Coastal Plain (U)
Mid-Atlantic Coastal Plain (L)
Southern Blue Ridge (U)
Ozarks/Ouachitas (U)
Ozarks/Ouachitas (L)
Florida (U)
Florida (L)
SLOPE
(/zeq rag"1)
8.0
6.2
7.9
12.4
4.0
5.5
6.2
7.0
r2
0.48
0.54
0.52
0.61
0.31
0.88
0.59
0.93
P
< 0.01
< 0.001
< 0.001
< 0.001
< 0.001
< 0.001
< 0.001
< 0.001
For those nodes and subregions not listed, the slope of the regression was not significantly
different from zero (p > 0.05).
256
-------
levels in streams in subregion 2As (median < 1 mg L"1 DOC) are typically lower than in most of
the other subregions. Thus, despite the apparently higher equivalents of strong acid organic
anions per mg of DOC in stream water in this subregion, the watersheds export very little DOC
during baseflow, with the result that organic acids do not play a major role in contributing to
the variation in ANC between streams under baseflow conditions.
DOC also had a significant positive relationship with ANC in the lower nodes in some
subregions (e.g., the Northern Appalachians [2Cn] and the Southern Blue Ridge [2As], and in the
upper nodes in the Valley and Ridge [3B]), indicating that the DOC in these subregions and
nodes consisted of mainly weak organic protolytes. Because the concentration of weak organic
anions was low in these subregions, DOC was not a major source of the variation in the ANC
among streams. In all of the remaining nodes and subregions, the relationship between DOC and
ANC was not statistically significant (p > 0.10), suggesting that variation in DOC among streams
in these subregions and nodes was also not a contributing source of the variation in the ANC.
In those subregions with a significant negative relationship between DOC and ANC, DOC
appears to be of major importance in explaining the variation in ANC among streams only in
Florida (3C) because of the high concentrations of DOC. In this case, DOC was comparable to
base cations and chloride in importance as a source of variation in ANC among streams in both
the upper and lower nodes. In Florida streams with an ANC < 200 fieq L"1, however, DOC was
important only at the upper nodes. In the other subregions (Mid-Atlantic Coastal Plain [3B],
Pocono/Catskills [ID], Southern Appalachians [2X], and Ozark/Ouachitas [ID]), DOC ranks
substantially lower in importance than the sum of base cations and mineral acid anions as a
source of the variation in ANC among streams.
8.6 SOURCES OF ANC VARIATION IN LOW ANC (< 200 0eq L'1) STREAMS
The previous analysis of factors that account for the among-stream differences in ANC
included all streams in the population of interest within a subregion. Because of the concern
about the acidification of streams having the lowest buffering capacity and the possibility that
inclusion of streams with a high ANC would obscure the role of mineral acids as a source of
variation, the data were reanalyzed using only those sites with an ANC < 200 ^eq L"1.
Variation in ANC between streams was analyzed for each subregion using the standardized
linear regression analysis described in Section 8.4. The importance of variation in base cations,
sulfate, nitrate, chloride, and DOC in explaining the variation in ANC was determined by regres-
sing these variables on ANC. The absolute values of the standardized regression coefficients are
plotted for the upper and lower nodes in each subregion (Figures 8-7, 8-8, and 8-9).
These plots show that for all the subregions except the Northern Appalachians (2Cn)
(Figure 8-7) and Florida (3C) (Figure 8-9), the sum of base cations is again the most important
variable in explaining the variation in ANC among the more acid-sensitive streams. In subregion
2Cn, sulfate is of comparable importance to base cations, indicating that sulfate is as important
as base cations in explaining the variation in ANC among streams in this subregion. Differences
in ANC among streams within most of the remaining subregions are still largely associated with
base cation differences, despite the statistical significance of the association between ANC and
strong acid anions in these subregions.
In Florida streams (Figure 8-9), however, chloride is of comparable importance to base
cations at the lower nodes, although at the upper nodes, base cations, chloride, and DOC are of
approximately equal importance in explaining the variation in ANC among streams. Sulfate is
not associated with a significant portion of the ANC variation at either the upper or lower
nodes in subregion 3C stream reaches. Surprisingly, DOC does not contribute significantly to an
257
-------
Pocono/Catskills
BC
b = -0.73
b =-0.96
NOJ
DOC
jb = -0.96
b= -1.01
3 =-6.09, p< 0.05
b = -8.23, p < 0.05
0 Upper Node (R2 = 0.95)
• Lower Node (R2 = 0.98)
01234
Northern Appalachians
BC
Upper Node (R2 = 0.84)
Lower Node (R2 = 0.92)
DOC
| b = -8.75, p < 0.05
b = -3.33, n. s.
024
Valley and Ridge
BC
b = -0.89
= -1.07
DOC
b = -5.89, n. s.
b = 7.52, p > 0.05
Upper Node (R2 = 0.97)
Lower Node (R2 = 0.99)
0123456
STANDARDIZED REGRESSION COEFFICIENT
Figure 8-7. The absolute value of the standardized linear regression coefficient (bjO for each
of the five independent variables considered in streams in the Poconos/Catskills
(ID), Northern Appalachians (2Cn), and Valley and Ridge (2Bn) subregions, with
ANC < 200 /*eq L"1. Also shown is the estimated regression coefficient (b) for
each independent variable. The standardized coefficients are highly significant
(p < 0.001) unless otherwise indicated.
258
-------
Mid-Atlantic Coastal Plain
BC
DOC
Upper Node (R2= 0.93)
Lower Node (R2 = 0.98)
0 1 2
Southern Blue Ridge
BC
0 Upper Node (R2 = 0.96)
• Lower Node (R2 = 0.97)
DOC
b = 14.02, p> 0.05
b = 3.94, n. s.
0.0 0.5
Piedmont
1.0
1.5
BC
DOC
I b = -0.53, p > 0.05
b = 0.02, n. s.
| b = 2.14, n. s.
b = 0.69, n. s.
a Upper Node (R2 = 0.96)
• Lower Node (R2 = 0.90)
012
STANDARDIZED REGRESSION COEFFICIENT
Figure 8-8.
The absolute value of the standardized linear regression coefficient (bjO for each
of the five independent variables considered in streams in the Mid-Atlantic
Coastal Plain (3B), Southern Blue Ridge (2As), and Piedmont (3A) subregions,
with ANC < 200 /zeq L'1. Also shown is the estimated regression coefficient (b)
for each independent variable. The standardized coefficients are highly
significant (p < 0.001) unless otherwise indicated.
259
-------
Southern Appalachians
BC
b = 1.00
b = 0.92
= -1.05
b = -0.88, p < 0.01
0 Upper Mode (R2 = 0.99)
b = -1.01 • Lower Mode (R2 = 0.98)
b = -1.05,p>0.05
DOC
I b = -6.93, p < 0.01
b = 2.15, n. s.
0 1 2
Ozarks/Ouachitas
BC
0 Upper Node (R2 = 0.91)
b = -1.01 • Lower Node (R2 = 0.94)
b = -1.95, n. s.
b = -5.28, p < 0.01
CI
N°3 y—,b=-°-62'n-s-
b = 0.60, p > 0.05
_ Upper Node (R2 = 0.93)
| b = -3.95 • Lower Node (R2 = 0.79)
b = -0.67, n.s.
DOC
0123
STANDARDIZED REGRESSION COEFFICIENT
Figure 8-9.
The absolute value of the standardized linear regression coefficient (bjO for each
of the five independent variables considered in streams in the Southern Appala-
chians (2X), Ozarks/Ouachitas (2D), and Florida (3C) subregions, with ANC < 200
peq L- . Also shown is the estimated regression coefficient (b) for each
independent variable. The standardized coefficients are highly significant (p <
0.001) unless otherwise indicated.
260
-------
explanation of the variation in ANC at the lower nodes in Florida streams, suggesting that (1)
the variation in DOC is small relative to that observed in ANC or (2) the acidic functional
groups are either lost from water or are so diluted during transport from upstream to
downstream that they no longer contribute to the variation in ANC. Both the median concen-
tration and the range of DOC decreases from the upper to the lower nodes, indicating either a
loss or dilution of the soluble organic matter.
8.7 EFFECTS OF CARBONIC ACID ON pH - (CO2 EFFECTS)
Carbon dioxide (CO2) is not an inert gas. When dissolved in water, CO2 becomes hydrated
to form carbonic acid (H2CO3), as shown in the following reaction:
C0
H20 = H2C03.
Depending on the pH of the water in which the CO2 is dissolved, carbonic acid can dissociate
and depress the pH. At pH < « 5.5, CO2 should have little or no effect on pH because essen-
tially all of the carbonic acid present will be undissociated (i.e., exist as H2CO3).
In contrast to other strong or weak acids, changes in carbonic acid content will have no
effect on the ANC unless there are phase changes in the carbonate system (e.g, carbonate pre-
cipitation). This lack of an effect of CO2 on alkalinity is because the increase in HCO3" is
balanced by an increase in H+ according to the reaction:
CO2 + H20 = H2CO3 - H+
HCO3-.
Hence, there is no net change in alkalinity as CO2 increases or decreases.
The effects of CO2 on the pH of stream water was assessed by comparing the closed-
system pH measured in the processing laboratory with that measured in the analytical laboratory
after equilibration with atmospheric CO2 (pCO2 = 10"3-5 atm). The latter was determined by
bubbling 300 ppm CO2 air through the stream water samples for 20 minutes. If the air-
equilibrated pH is greater than the closed-system pH, the difference is assumed to be due to
CO2 supersaturation. Similarly, a decrease in pH indicates that the sample was undersaturated
with respect to the pCO2 in the atmosphere.
Plots of the difference between closed-system and air-equilibrated pH versus the air-
equilibrated pH suggest that almost all of the streams in every subregion are supersaturated with
CO2 (Figure 8-10). The small differences between the air-equilibrated pH and the closed-system
pH at pH levels < 5.5 are consistent with the decrease in the dissociation of carbonic acid with
declining pH, and suggest that systematic errors in the measurement of pH are not the cause for
the difference in pH between air-equilibrated and closed-system samples.
The increase in pH when the samples are air equilibrated (i.e., excess CO2 is degassed)
ranges from approxiately 0.25 units to one full pH unit, indicating that carbonic acid alone has
an important effect on the pH of stream water in almost all subregions, particularly Florida
(3C), the Piedmont (3A), the Mid-Atlantic Coastal Plain (3B), and the Ozarks/Ouachitas (2D).
This suggests that streams in these subregions are more supersaturated with CO2 than those in
the other subregions. Possible reasons for these apparent regional differences in CO2 super-
saturation may be related to the gradient of the streams and the resulting differences in the
outgassing rate of excess CO2 that enters the stream channel in groundwater and soil water (in
equilibrium with higher pCO2), regional differences in biological activity that generates CO2 in
soil solutions, or some combination of both.
261
-------
o
u
U
4]
•U
-U
•
U
u
45
-U
-U
u
45
-U
-H
POCONOS/CATSKILLS
• *
'"•"•. •'.••'."^: •
•• *ffi.
* • • »
5 « 7 1
MD-ATLANTIC COASTAL PLAIN
...
• •-. -v.-
• •:• " '.
-„• . t
; l I 7 t
SOUTHERN APPALACHIANS
."?•':
• • *
•
u
u
4.5
-U
-U
'
u
•u
-u
-u
u
M
45
-U
-U
NORTHERN APPALACHIANS
"*' "" "" ".* " • " *
* (» »»^ «»a
* * *™ • " •
• • I "
, •
51711
SOUTHERN BLUE RIDGE
• •• % * •
*•" •
• V
'•
1 5 1 7 1
OZARKS/OUACHITAS
•:V '»
•».-.'
." * •
1.9
1.1
-
-1.1
-1.5
1
I.Si
1.1
45
-1.1
-1.5
1
15
U
4j
-U
VALLEY AND RIDGE
••• *. ' *
* ^ "-*
.""T *' •"•
5." * •
• * *
5 1 7 1 1
PIEDMONT
•"• * *
« •' \V-
': •. • %"
' ' '.
S f 7 t 1
FLORIDA
/ .
. • •' ; '
* » •
4 i I 7 I I t i I 7 I I 4 5 9 7 I I
AIR-EQUILIBRATED pH
Figure 8-10. Plot of closed-system minus air-equilibrated pH vs. air-equilibrated pH of stream
water in NSS-I subregions. Points plotted below the line indicate the samples
were supersaturated with CO2; points above the line indicate undersaturation.
262
-------
To assess whether the air-equilibrated pH was close to the theoretical pH of water with
carbonate species in equilibrium with atmospheric CO2, the measured air-equilibrated pH was
compared to the theoretical air-equilibrated pH for different ANC values (Figure 8-11). The
theoretical air-equilibrated pH curve was calculated assuming that the ANC was made up entirely
of carbonate species. The plot shows that in almost all subregions, many of the streams with
air-eqilibrated pH > 6.0 plot below the theoretical value based on carbonate ANC alone. The
exception is the Southern Blue Ridge (2As), where the measured air-equilibrated pH of almost all
streams was close to the theoretical value.
Reasons for the deviations from the theoretical curve include errors in the measurement of
air-equilibrated pH, failure to equilibrate the samples with atmospheric CO2 before measurement
of air-equilibrated pH, the presence of weak organic acids that dissociate near the measured air-
equilibrated pH, and the presence of other weak acids (e.g., aluminum). Since aluminum is
present in low concentrations at pH > 6.0, where the air-equilibrated pH difference in most
subregions is the greatest, aluminum is unlikely to be the cause for the differences between the
theoretical and air-equilibrated pH values in these cases. Aluminum, however, will contribute to
the pH difference in the acidic streams with elevated aluminum levels.
Organic anions will cause the air-equilibrated pH to be lower than expected because these
weak organic anions have a lower pKa than that for bicarbonate. Subregions with the highest
DOC levels—Florida (3C) and the Mid-Atlantic Coastal Plain (3B)~also have large differences in
air-equilibrated pH, suggesting that weak organic acids contribute to this difference.
Differences in the response of pH electrodes resulting from DOC interferences could also
account for the apparent air-equilibrated pH differences. Herczeg and Hesslein (1984) reported
that high levels of DOC in lake waters interfered with the response of pH electrodes, resulting
in air-equilibrated pH values lower than expected. pH bias due to DOC interference could
account for the observed differences in the air-equilibrated pH of streams only if the bias were
not linear over the entire pH range. Since there was little or no difference between the
expected and observed pH of most streams with air-equilibrated pH < 6.0, this suggests either
that the air-equilibrated pH difference is not due to DOC interference with the pH measurement
or that DOC interference is not constant over the entire air-equilibrated pH range.
263
-------
POCONOS/CATSKILLS
NORTHERN APPALACHIANS
VALLEY AND RIDGE
-M 50 til 250 350 «0 551 -50 58
LU
5 ,
DC
CQ
MID-ATLANTIC
COASTAL PLAIN
SOUTHERN BLUE RIDGE
350 455 550 -5J 50 ISO JH J50 «0 550
5
PIEDMONT
4
•« 58 I5t » 350 450 551 -50 55 150 J50 350 (S3 550 -50 50 150 » J50 (50 550
SOUTHERN APPALACHIANS
OZARKS/OUACHITAS
X
FLORIDA
-X M 151 S iD 450 551 -50 M 150 258 SO (58 550 -5! 30 150
J50 CO
ACID NEUTRALIZING CAPACITY (Lieq L"1)
Figure 8-11. Observed air-equilibrated pH vs. ANC in NSS-I subregions. The solid line
represents the theoretical pH calculated from ANC, assuming all of the ANC is
due to carbonate protolytes in equilibrium with atmospheric CO2 at 15°C.
264
-------
SECTION 9
NATURE AND DISTRIBUTION OF ACIDIC AND LOW ANC STREAMS
9.1 OVERVIEW
In order to determine the status and extent of acidic and low ANC streams that may have
been affected by acid deposition, stream reaches acidified by other sources (e.g., acid mine
drainage, organic acidity, neutral salt hydrogen ion substitution) must first be identified. In this
section, all NSS-I stream reaches with spring index ANC values < 200 ^eq L'1; a number
commonly used to indicate sensitivity to acid deposition, were classified according to probable
sources of acid anions. Streams classified as being affected by acid anion sources other than
acid deposition, and those affected by significant watershed sources of sulfate that would mask
atmospheric contributions, were removed from the list of streams possibly impacted by acid
deposition. These include streams affected by acid mine drainage or natural sulfide mineral
weathering, non-acidic sulfate inputs (e.g., sulfate mineral weathering or fertilizers), organic
acidity, and displacement of hydrogen ion by sea salt. The remaining streams constitute a high-
interest subpopulation of streams with anion balances dominated by strong inorganic acid anions
at concentrations that could reasonably be produced by evaporative concentration of anions in
precipitation. Acidic streams in this high-interest group are likely to have been affected by
acid deposition. Although streams with ANC between 0 and 200 /zeq L"1 may have naturally low
ANC, it is also possible that acid deposition could cause ANC depressions in these streams.
Within each stream chemical classification, population estimates of the numbers of acidic (ANC <
0 ^eq L'1), very low ANC (0 < ANC < 50 /ieq L'1), and low ANC (50 < ANC < 200 /zeq L'1)
stream reaches were made. The distribution of streams by chemical groupings in the NSS-I
subregions and geographic site classes are also presented.
9,2 BACKGROUND
In addition to acid deposition, other mechanisms, both natural and anthropogenic, can cause
acidic conditions in streams. One of the most severe acid generating mechanisms in streams is
acid mine drainage, which has acidified streams (damaging or eliminating their fish populations)
in many mining regions of the United States (Barnes and Romberger, 1968; Ahmad, 1972). When
mine spoils containing sulfide (e.g., pyrite in coal tailing piles) are exposed to air and water, the
reduced iron and sulfur are oxidized by a series of microbial and chemical processes according
to the following overall reaction (Stumm and Morgan, 1981; Dugan, 1985):
4 FeS2 + 15 02 + 14 H2O —> 4 Fe(OH)3 + 16 H+
8 SO2'
The products of this reaction can be hydrologically transported into surface waters and cause
their acidification. Acidity is generated from both sulfide oxidation and iron hydrolysis.
Streams affected by acid mine drainage are characterized by high iron, manganese, and sulfate
concentrations, high specific conductance, and low pH. Typical pH values in these streams range
from 2.0 to 4.0; sulfate concentrations range from 1000 to 25,000 /zeq L'1, whereas iron
concentrations range from 1 to 100 mg L'1 (Mills, 1985).
Weathering of sulfide minerals in undisturbed watersheds (e.g., those in the Anakeesta
formation in parts of the Southern Blue Ridge) may also generate acidity as shown in the above
reaction (Huckabee et al., 1975; King et al., 1968; Hadley and Nelson, 1971). The rates are
limited, however, because the minerals are underground and are only slowly exposed to oxygen
265
-------
and water through natural weathering processes. It is only when the sulfide minerals are
disturbed (as is done in mining activities) that accelerated weathering and hydrologic flow cause
the severe acidification typical of streams affected by acid mine drainage (Dugan, 1985; Dick et
al., 1986). Chemical conditions similar to acid mine drainage may occur when slumps or road
cuts expose sulfide containing material to air and water.
Waters with high concentrations of dissolved organic matter can be naturally acidic (pH 3.0
- 5.0) due to the leaching of organic acids from the natural decay of biotic material (Gorham et
al., 1984). Not all organic carbon, however, contributes to organic acidity in aqueous systems.
In natural waters, dissolved organic carbon (DOC) is typically composed of roughly 50% humic
substances (10% humic acids, 40% fulvic acids), 30% hydrophillic acids, 5% - 10% carbohydrates,
5% - 8% carboxylic acids and phenols, and 4% amino acids (Thurman, 1985). The analytical
procedures that measure any one of these groups of organic compounds are quite complex, and
in most cases are operational definitions (rather than analyses of specific compounds). In the
NSS-I, therefore, DOC concentration was used as a composite measure of the organic contribu-
tion to'stream water chemistry. Unfortunately, DOC determinations measure all of these organic
carbon forms and represent a group of compounds that vary greatly in their acid/base character-
istics. DOC is, therefore, only a crude measure of organic acidity in surface water.
In order to quantify the contribution of organic acidity to stream acidity, it is necessary
to know the acid dissociation constants (Ka) of the organic acids in the stream. Like carbonic
acid (pKa - 6.3), organic acids with relatively high pKa (5.0 - 7.0) can be considered weak acids
and their dissociated anions will contribute to ANC in circumneutral waters. However, organic
acids with low pKa (< 4) will behave like strong, inorganic acids (e.g., H2SO4, HNO3) and
depress stream pH. Measuring stream water organic Ka is a complicated task and was imprac-
tical for a synoptic survey like the NSS-I. It is possible, however, to estimate a composite
stream water organic acid pKa and thus the organic anion concentration by using the Oliver
model (Oliver et al., 1983). The model is based on two empirical relationships. Oliver found
that the pKa of organic acids is related to pH and that the amount of carboxyl acidity averages
about 10 /jeq mg"1 DOC. The organic anion concentration of a sample can, therefore, be calcu-
lated from measurements of DOC and pH. The relationship between organic anions and inorganic
anions in a stream water sample can be used to evaluate the effect of organic acidity on the
acid/base chemistry of the stream.
Estimation of the organic anion concentration in NSS-I streams was necessary in order to
classify streams that were affected by organic acidity. Organic anion concentrations were
calculated from pH and DOC using the Oliver model (Oliver et al., 1983) as described above. In
streams with ANC < 200 /ieq L"1 and DOC concentrations > 2 mg L"1, the organic anion concen-
tration calculated by the Oliver model was positively correlated with the anion deficit (base
cations + H+ + NH4+ - SO42- - NOS- - Cl~ - F~ - HCO3- - CO32- - OH") (r2 = 0.64, slope =
0.89, intercept = 23; also see Sections 4.9 and 8.6). Thus, calculation of organic amons by the
Oliver model appears to yield reasonable results in low ANC streams.
Sea salt deposition has been proposed as a significant acidification mechanism in coastal
surface waters. When a solution high in neutral salt (such as NaCl) passes through an acidic
soil, Na+ ions can be exchanged for H+ ions, rendering the emerging leachate richer in H and
thus more acidic than the solution added to the soil (Krug and Frink, 1983). This hypothesis
can be examined by plotting ANC versus Na+rCl" and looking for a trend of low ANC streams
with Na+ depletions. .
Sea salt in wet and dry deposition can contribute a significant amount of ions to coastal
streams that do not contribute acidity and may confound a classification of streams according to
probable sources of acidity. Therefore, for classification purposes, all streams within 200 km of
266
-------
the coast were corrected for sea salt additions of SO42', CT, Na+, Ca2+, K+ and Mg2+.
Seawater Cl" contributions to mid-Atlantic streams were estimated using the equation relating
stream water Cl~ to distance from the ocean, as described in subsection 6.3.5 (Figure 6-18). The
other sea water ions were corrected based on their molar ratios in sea water with respect to
Cl~. In Florida (3C), there was no relation between stream Cl" concentration and distance from
the ocean. The Cl~ concentration contributed by sea water was estimated from the 1986 volume
weighted concentration of Cl~ in precipitation (12.5 jueq L"1) observed at the NADP/NTN site in
Quincy, Florida (NADP, 1987), about 70 km from the ocean in the panhandle area of the sub-
region. The Cl~ concentration in precipitation was then multiplied by the evaporative concen-
tration factor (precipitation/runoff) for subregion 3C (Geragnty et al., 1973) to determine the
Cl" concentration contributed by sea water in Florida streams. The NSS-I sites in Florida were
on the average 58.5 km (SD = 24.6 km) from the ocean. The other ions affected by sea salt
were corrected based on their molar ratios in sea water with respect to Cl~.
9.3 CLASSIFICATION AND LOCATION OF ACIDIC AND LOW ANC STREAMS
In this section of the report, all streams with ANC < 200 /zeq L'1 have been classified
according to probable source of acidity. Classification was done using the spring index data on
a node-by-node basis, treating the upper and lower nodes as separate samples. The goal of the
classification was to identify and remove from consideration stream nodes with substantial acid
anion concentrations derived from sources other than acid deposition (mine drainage, organic
acids). The acidic streams (ANC < 0) that remain, after removing streams with other sources of
acidity, constitute a high-interest subpopulation of streams whose most likely source of acidity
is acid deposition. The source of acidity in the nonacidic high-interest streams (ANC 0 - 200
/zeq L"1) cannot be conclusively inferred from survey data. It is possible that the streams in
this group have naturally low ANC. However, it is also possible that acid deposition has caused
ANC reductions, because an atmospheric source can explain the strong acid anion concentrations
in these high-interest low ANC streams, and other sources of acidity are unlikely. Streams
affected by other sources of acidity could also be affected by acid deposition. Their chemistry,
however, is likely to mask evidence of acid deposition effects. Population estimates of the
number and length of reaches in each chemical classification group are presented in Sections 9.4
and 9.5.
9-3.1 Streams with Acid Mine Drainage and Substantial Watershed Sources of Sulfate
9.3.1.1 Classification-
Streams were classified as being impacted by acid mine drainage based on field observa-
tions, mapped information, and chemical screening. All 26 sites at which water samples were
not collected due to high conductivity (> 500 jtS cm"1) were checked for the possibility of acid
mine drainage. Seven of these high-conductivity nodes had pH less than 3.3 and examination of
aerial photographs and topographic maps indicated the presence of strip mining activity in their
watersheds. Because these nodes were not sampled for complete water chemistry, checking for
other acid mine drainage chemical signatures was not possible. The available evidence does
indicate that these seven nodes were being impacted by acid mine drainage and they were
classified as such.
All stream nodes with index ANC values < 0 peq L"1 were screened for the presence of
acid mine drainage. Acidic streams with sulfate concentrations > 300 /ieq L"1 in the Mid-
Atlantic and 200 fieq L'1 in the Southeast were considered possible acid mine drainage sites.
267
-------
These cutoff values were chosen after examination of all site data in Table 6-7, which lists
predicted stream water sulfate concentrations assuming only atmospheric sources of acidity.
From this data, the best estimate of stream water sulfate concentration in the Mid-Atlantic is
158-219 /zeq L'1 with maximum estimates of 316-413 /zeq L'1. A value of 300 /zeq L'1 was
chosen as a cutoff to separate streams whose sulfate comes mainly from the atmosphere from
streams whose sulfate source is primarily internal (from watershed sources). Mid-Atlantic
streams whose sulfate concentration is < 300 /zeq L"1 most likely receive a majority of their
sulfate, and possibly all their sulfate, from the atmosphere. Those streams with sulfate
concentration > 300 /zeq L'1 most likely receive a substantial proportion from watershed sources.
However, they may also be subject to locally high sulfate deposition. A similar line of
reasoning applies to the selection of 200 /zeq L"1 as a cutoff value in Southeast streams.
Predicted stream water sulfate concentrations (Table 6-7) in the Southeast (96-146 /zeq L"1 with
maximum estimates of 190-440 /zeq L"1) are lower than in the Mid-Atlantic, so a lower cutoff
was established.
The use of these sulfate cutoff values was corroborated through examination of field
comments, maps, aerial photographs, and site visits to most of the acidic reaches during the
Mid-Atlantic Acid Streams Reconnaisance in the spring of 1987. In general, streams with sulfate
concentrations over the cutoff value tended to have field or map evidence or chemical signatures
indicating mine drainage or some other point-source perturbance (tree farm, gas well, orchard,
cleared watershed), whereas streams below the cutoff value did not.
The frequency distribution of stream water sulfate concentration in nodes with ANC < 200
/zeq L'1 shows that most reach nodes had sulfate concentrations < 300 /zeq L"1 or > 500 /zeq L'1
(Figure 9-1). Of the 512 acidic, very low and low ANC nodes, only 13 had sulfate concentra-
tions between 300 and 400 /zeq L'1, and 8 had values between 400 and 500 /zeq L"1. Another 14
nodes had sulfate concentrations between 250 and 300 /zeq L"1. Only 6 of these 35 nodes with
intermediate sulfate concentration (250-500 /zeq L"1) had ANC < 0, and an additional 6 of these
nodes had ANC between 0 and 50 /zeq IT1. The vast majority (92%) of the acidic sample nodes
(ANC ^ 0 /zeq L'1) have sulfate concentrations > 500 /zeq L'1 (acid mine drainage) or < 250
/zeq L"1. Because of the small number of nodes with intermediate sulfate concentrations, the
picking of an arbitrary cutoff value did not have a great effect on the estimated population
distribution of acidic and very low ANC streams likely to be affected by atmospheric deposition.
Twenty stream nodes were found to have ANC levels < 0 /zeq L'1 and SO42' concentrations
> 300 /zeq L'1 in the Mid-Atlantic and > 200 /zeq L"1 in the Southeast. In 17 of these sites,
there was evidence of acid mine drainage impact, such as the presence of strip mines (noted in
aerial photographs, in topographic maps, or in site comments made during sampling visits). In
most cases, the chemical evidence was overwhelmingly in support of acid mine drainage classi-
fication ([SO42-] > 1000 /zeq IT1, pH < 4, [Fe] > 100 /zM, and specific conductivity > 100 /zS).
Stream sites with weaker chemical signatures of acid mine drainage were revisited in the spring
of 1987 during the Acid Streams Reconnaissance that confirmed field evidence of acid mine
drainage. Therefore, these 17 stream nodes, plus the 7 nodes not sampled due to high conduc-
tivity (with low pH, as discussed above), were classified as impacted by acid mine drainage. The
Other three acidic high-sulfate nodes all had high DOC concentrations (17-32 mg L'1) and color
values (110-163 PCU) and were located in the Mid-Atlantic Coastal Plain (3B). These three
nodes are likely to be acidic due to organic acids, but their chemistry is dominated by high
sulfate concentrations and they were classified as receiving substantial watershed sources of
sulfate.
268
-------
300
to
UJ
o
o
Q_
<
a:
UJ
CD
200
ANC CLASS
ANC < 0
ANC 0-50
ANC 50-200
0- 100- 200- 300- 400- 500- 600- 700- 800- 900- >1000
100 200 300 400 500 600 700 800 900 1000
SULFATE CONCENTRATION (MeqL~1)
Figure 9-1. Frequency distribution (number of nodes) of spring index sulf ate concentration in
NSS-I stream nodes with index ANC values < 200 peq L"1.
269
-------
Streams with ANC between 0 and 200 peq L"1 and SO42" concentrations > 300 /zeq L'1 in
the Mid-Atlantic and > 200 /xeq L"1 in the Southeast were classified as having substantial
watershed sources of sulfate. These streams were not classified as acid mine drainage streams
because they are not acidic. The ANC values are low. enough to warrant concern about acid
deposition, but since they receive significant amounts of sulfate from their watersheds, the
potential role of acidic sulfate deposition cannot be estimated. The elevated concentrations of
sulfate in these streams could be derived from mine drainage that has been neutralized by
dilution with nonacidic streamflow, or by contact with alkaline weathering products from sources
such as limestone (either natural or artificially added). Sulfate in these streams may also be
derived from natural sulfur mineral weathering (e.g., gypsum or sulfides), or may be contributed
as a result of agricultural activity or marine influences.
9.3A.2 Location--
Reaches impacted by acid mine drainage at either their upper or lower nodes were
concentrated in the Allegheny Plateau in Pennsylvania and West Virginia (Figure 9-2). One
additional reach affected by acid mine drainage was observed in Tennessee. The majority of the
streams with significant watershed sources of sulfate were observed in the same area as the acid
mine drainage sites. The source of sulfate in these reaches is most likely sulfide or sulfate
mineral weathering or neutralization of upstream acid mine drainage. The source of sulfate in
the other high-sulf ate reaches, mostly located in the Mid-Atlantic Coastal Plain (3B), is probably
either agricultural activity or sulfate derived from weathering of marine sedimentary deposits.
The maps are provided to show the geographic location of the sites in each chemical group.
No quantitative estimate of population distributions or density of streams within a geographic
area should be made directly from these maps because the population weighting and site
selection factors have not been taken into account. Population estimates for each chemical
group are presented in Section 9.4.
9.3.1.3 Chemical Characteristics—
The weighted mean pH and ANC (4.3 and -306 peq L'1) of the stream reaches classified as
being affected by acid mine drainage (Table 9-1) illustrate the severe acidity associated with
this type of pollution. The mean acidity would almost certainly have been higher if chemistry
had been measured at the seven sites with conductivity > 500 /*S cm"1. These streams have high
concentrations of base cations and metals (Fe, Mn), reflecting the high rates of weathering of
metallic sulfides and other minerals exposed by mining. The release of hydrogen ions during
sulfide weathering further enhances the weathering of other exposed minerals, dissolving or
desorbing base cations. The acid mine drainage streams contain very low concentrations of
DOC, and sulfate is by far the most dominant anion (> 95% of total anions). The chemistry of
the group of streams with substantial watershed sources of sulfate is similar to that in streams
affected by acid mine drainage, in that sulfate is the dominant anion (> 50% of the total
anions). The mean pH, however, is much higher (6.18) in the class of streams with large water-
shed sources of sulfate. Mean nitrate concentrations were high in this group (131 /*eq L"1),
possibly as a result of agricultural activity. Concentrations of base cations were also high in
these high sulfate streams.
270
-------
ACID MINE DRAINAGE
• Acid Mine Drainage Indications
A Watershed Sulfate Sources
Figure 9-2. Location of NSS-I study reaches with ANC concentrations < 200 /*eq L'1 having
evidence of acid mine drainage or substantial watershed sources of sulfate at
either the upper or lower node.
271
-------
Table 9-1. Weighted Mean and Standard Deviation (SD) of NSS-I Spring Index Chemistry in
Stream Nodes Classified as Impacted by Acid Mine Drainage and Watershed
Sources of Sulfate.5
Acid Mine Drainage
Substantial Watershed
Sources of Sulfate
Chemical
Species
S04*
NOS-
Cl*
HC03-
Organic Anions
DOC (mg L"1)
% Organic Anions
Color (PCU)
Sum Base Cations*
SO42':SBC ratio
SO42":ANC ratio
ANC
pH (closed system)
Altot (MM)
Al t (pM)
Al!*m} (/zM)
Fe (/*M)
Mn (fJM)
Conductivity (p,S cm'1)
n -
Mean
3272.46
20.34
64.07
0.29
5.28
0.94
0.40
15.71
2718.88
1.21
-17.59
-305.81
4.34
99.19
100.29
72.42
102.03
43.68
377.00
17*
SD
1835.35
26.43
69.18
0.26
2.22
0.33
0.48
13.39
1581.82
0.27
8.10
341.43
0.38
63.24
62.59
59.19
132.57
32.36
198.58
n =
Mean
688.48
130.79
233.71
86.86
18.80
2.19
2.17
15.01
1192.42
0.56
13.31
112.43
6.35
5.99
1.62
0.80
2.21
1.39
140.61
39
SD
620.91
217.18
300.06
45.73
25.20
3.73
2.93
24.29
729.24
0.20
41.08
66.10
0.60
6.62
3.64
2.19
6.80
2.19
79.35
5 All concentrations are expressed in /teq L"1, except as noted.
* Concentrations were sea salt corrected as discussed in text.
#24 nodes were classified as acid mine drainage, but 7 nodes were not sampled for water
chemistry due to field exclusion criteria (pH < 3.3, conductivity > 500 /iS cm'1).
n « number of nodes classified in each group.
% Organic Anions - percentage of total anions that are organic.
Al^t » MIBK-extractable aluminum (total monomeric).
Aljm » inorganic monomeric aluminum (PCV method).
272
-------
9.3.2 Streams with Organic Dominance
9.3.2.1 Classification--
To classify the streams that are likely to be acidic due to organic acids, the concentration of
organic anions was compared to the concentration of sulfate and nitrate. Streams in which the
organic anion concentration was greater than the sum of the nitrate and sea salt corrected
sulfate concentrations were classified as organic-dominated streams. Chloride was not included
in the ratio since it is not normally a major contributor to acid/base chemistry in natural
waters and it is not usually associated with acid deposition. Nitrate and sulfate were compared
to organic anions because they are the anions associated with acid deposition. If organic anion
concentrations are larger than the concentrations of sulfate and nitrate, the stream water
acidity is controlled largely by organic acids, not by additions of sulfate and nitrate associated
acid deposition. For classification purposes, the class of organic-dominated streams was
subdivided into three groups, acidic (ANC < 0 /ieq L"1), very low ANC (0 < ANC < 50 /jeq L"1),
and low ANC (50 < ANC < 200 /zeq L"1).
9.3.2.2 Location--
Reaches that were organic dominated at their downstream end were found only in Florida
(3C) and the Mid-Atlantic Coastal Plain (3B) (Figure 9-3). Only one acidic organic-dominated
lower node was observed outside Florida. Acidic reaches that were organic dominated at the
upstream end (Figure 9-4) were found mainly in subregions 3C and 3B, although there was one
upstream site with very low ANC in the Poconos/Catskills (ID) and one downstream site with
low ANC in the Ozarks/Ouachitas (2D).
9.3.2.3 Chemical Characteristics—
The organic-dominated streams, as expected, had high weighted mean concentrations of
DOC (10-50 mg IT1) (Table 9-2). They also were highly colored (> 100 PCU). The acidic sub-
group had low pH (4.5) and large negative values of ANC (mean = -159 /zeq L"1). Sulfate and
nitrate concentrations were very low, and Cl~ concentrations were approximately equal to the
organic anion concentrations. The high sea salt corrected Cl~ concentrations probably occurred
because these sites were near the ocean; chloride may have been contributed by surface water
or groundwater in contact with marine sedimentary deposits or soils inundated by sea water at
some time in the past. The SO42":base cation ratio was uniformly low in all three ANC groups,
indicating that weathering due to sulfate acidity was not an important process in these streams.
9.3.3 High-Interest Subpopulation
9.3.3.1 Classification-
After removal of organic-dominated streams, streams impacted by acid mine drainage, and
streams having large watershed sulfate contributions, we are left, by process of elimination, with
a high-interest subpopulation of acidic and low ANC streams. The acidic (ANC < 0) and very
low ANC (>0 - 50 /zeq L"1) stream reaches within this high-interest subpopulation are likely to
have been affected by acid deposition. If strong acid anions are present in stream reaches
within this group that have ANC between 50 and 200 /zeq L"1, the likely source of these anions
is atmospheric deposition. Within the high-interest subpopulation, there is a group of streams
that, while not dominated by organic anions, had organic anion contents high enough to have an
important influence on stream acid/base chemistry. Therefore, those in which organic anions
have some influence and those in which organic anions are virtually absent were identified.
273
-------
ORGANIC ANION DOMINANCE
NSS PHASE I SITES - LOWER NODE
Organic Dominated Streams
• ANC < 0 fjeq L"
Figure 9-3. Location of NSS-I lower nodes with ANC < 200 jueq L"1 classified as being
dominated by organic anions.
274
-------
ORGANIC ANION DOMINANCE
NSS PHASE I SITES-UPPER NODE
Organic Dominated Streams
0
-------
Table 9-2. Weighted Mean and Standard Deviation (SD) of NSS-I Spring Index Chemistry in
Organic-dominated Classified Stream Nodes$
ORGANIC DOMINATED STREAMS
ANC<0
Chemical
Species
SO4*
NOS-
Cl*
HCOjf
Organic Anions
DOC (mg L"1)
% Organic Anions
Color (PCU)
Sum Base Cations*
SO42':SBC ratio
SO^'cANC ratio
ANC
PH
Altot 0*M)
Al«t G*M)
Al(fa») 0*M)
Fe G«M)
Mn friM)
Conductivity (ftS cm
n =
Mean
39.58
7.31
168.34
1.92
246.89
49.96
48.12
331.07
290.36
0.15
-2.76
-159.28
4.46
19.65
8.87
4.20
14.15
0.35
-1) 76.86
16
SD
39.21
15.75
108.51
1.63
236.55
61.24
15.95
293.98
162.45
0.11
2.36
247.43
0.58
9.56
5.22
3.39
7.85
0.38
67.65
0 < ANC < 50
n =
Mean
32.68
3.31
95.88
12.38
89.93
10.24
36.48
117.96
203.99
0.15
2.02
25.37
5.58
12.97
3.89
1.66
6.85
0.84
28.42
16
SD
19.10
4.29
54.99
4.99
64.16
7.39
8.66
47.20
99.23
0.08
1.76
12.01
0.19
8.46
3.70
1.72
5.68
0.75
9.93
50 < ANC < 200
n =
Mean
65.28
1.53
190.76
111.75
137.80
14.63
27.63
140^53
474.51
0.13
0.62
147.87
6.10
29.29
5.04
1.70
16.76
2.45
54.72
14
SD
53.75
2.71
99.78
36.43
58.20
6.26
7.26
49.11
203.10
0.07
0.44
35.13
0.21
16.78
4.10
1.22
12.33
2.87
22.33
* All concentrations are expressed in peq L"1 except as noted.
* Concentrations were sea salt corrected as discussed in text.
n « number of nodes classified in each group.
% Organic Anions « percentage of total anions that are organic.
Al^j « MIBK-extractable aluminum (total monomeric).
inorganic monomeric aluminum (PCV method).
276
-------
The organic-influenced group was defined as those streams in which organic anions comprised
more than 10% of the sea salt corrected total anion sum. The value of 10% was chosen as a
cutoff after examination of DOC, color, and site comment data to find streams with an organic
character (moderate DOC, colored, slow moving water). The organic-influenced group represents
streams likely to be affected by both acid deposition and organic acidity.
The second subgroup within the high-interest subpopulation comprises those streams in
which less than 10% of the total anions were composed of organic anions. These acidic or low
ANC inorganic streams represent the group of streams in which no other source of strong acid
anions besides acid deposition was likely. Acidic streams in this class are very likely to have
been affected by acid precipitation. For classification purposes, both the inorganic and organic-
influenced class of streams were subdivided into three groups, acidic (ANC < 0 /teq L'1), very
low ANC (0 < ANC < 50 /zeq L"1), and low ANC (50 < ANC < 200 jueq L"1).
Streams in the acidic subgroup represent streams that are already acidic, whereas those in
the very low ANC subgroup represent streams that are most sensitive to acid deposition and
that, if located in geographic areas with high atmospheric acid deposition rates, may become
acidic during episodes (Eshleman, in press). Streams sensitive to acid deposition are found
within the low ANC (50-200 peq L"1) subgroup. Not all streams with low ANC should be con-
sidered sensitive, however, as stream water ANC is not of itself a complete indicator. Sensi-
tivity to acid deposition effects depends also on the acid neutralizing capability of the stream
watershed, which is, in turn, dependent upon such processes as sulfate adsorption and cation
exchange in watershed soils.
9.3.3.2 Location—
Three of the four inorganic, acidic lower reach nodes were located in the Northern Appala-
chians (2Cn) (Figure 9-5). All of the acidic, organic-influenced lower reaches were found in the
Mid-Atlantic Coastal Plain (3B), primarily in the New Jersey Pine Barrens. There were more
acidic upper nodes (Figure 9-6) than acidic lower nodes. Inorganic, acidic upper nodes were
found in the Northern Appalachians (2Cn), the Valley and Ridge (2Bn), Florida (3C), the
Poconos/Catskills (ID), and the Mid-Atlantic Coastal Plain (3B). Organic-influenced, acidic upper
nodes were found in subregions ID and 3B, with one isolated site in the Southern Appalachians
(2X) and two in the Florida panhandle. There were many acidic sample reaches in the New
Jersey Pine Barrens, which is discussed in subsection 9.5.3.
Streams with very low ANC (between 0 and 50 jueq L'1) and low ANC (between 50 and 200
jieq L'1) at lower (Figure 9-7) and upper reach nodes (Figure 9-8) were found in all NSS-I sub-
regions. All subregions except 2Bn contained at least one organic-influenced sample reach, but
most were found in subregions 3C and 3B. There were more very low ANC sample reaches
within the high-interest class in the Mid-Atlantic than in the Southeast. No stream reaches
were observed in the Florida peninsula with ANC < 200 /*eq L'1 that were not organic
dominated.
9.3.3.3 Chemical Characteristics—
Sulfate was the most abundant anion within the high-interest subpopulation of streams not
influenced by organic anions (inorganic class) (Table 9-3). In the acidic subgroup, more than
75% of the mean total anion concentration was made up of sulfate. DOC concentrations were
low (< 2-3 mg L'1) and the color value was almost always less than 30. The sum of base
cations increased with increasing ANC subgroup while the SO42":base cation ratio decreased.
The organic-influenced streams within the high-interest subpopulation were moderately
colored (29-70) and had mean DOC concentrations of 3-7 mg L'1 (Table 9-4), intermediate
277
-------
HIGH INTEREST ACIDIC SITES
NSS PHASE I SITES - LOWER NODE
• Inorganic
O Organic Influenced
Figure 9-5. Location of the high-interest subpopulation of acidic (ANC < 0 /xeq L"1) NSS-I
lower nodes. All nodes classified as acid mine drainage and organic-dominated
have been removed from the high-interest subpopulation.
278
-------
HIGH INTEREST ACIDIC SITES
NSS PHASE I SITES-UPPER NODE
Inorganic
O Organic Influenced
Figure 9-6. Location of the high-interest subpopulation of acidic (ANC < 0 /zeq L'1) NSS-I
upper nodes. All nodes classified as acid mine drainage and organic-dominated
have been removed from the high-interest subpoplation.
279
-------
HIGH INTEREST LOW ANC SITES
NSS PHASE I SITES - LOWER NODE
• Inorganic; ANC 0-50peqtT
© Inorganic; ANC 50-200 ueqL
Organic Influenced; ANC 0-50 ueqL'1
4- Organic Influenced; ANC 50-200 ueqll1
Figure 9-7.
Location of the high-interest subpopulation of very low and low ANC NSS-I
lower nodes. All nodes classified as having substantial watershed sources of
sulfate and organic dominance have been removed from the high-interest
subpopulation.
280
-------
HIGH INTEREST LOW ANC SITES
NSS PHASE I SITES - UPPER NODE
• Inorganic; ANC 0-50ueqL~'
O Inorganic; ANC 50-200 ueq L~'
Organic Influenced; ANC 0-50ueql_~'
Organic Influenced; ANC 50-200ueqL
Figure 9-8. Location of the high-interest subpopulation of very low and low ANC NSS-I
upper nodes. All nodes classified as having substantial watershed sources of
sulfate and organic dominance have been removed from the high-interest
subpopulation.
281
-------
Table 9-3. Weighted Mean and Standard Deviation (SD) of NSS-I Spring Index Chemistry in
Inorganic Classified Stream Nodes
INORGANIC STREAMS
Chemical
Species
SO4*
N03-
Cl*
HCOS-
Organic Anions
DOC (mg L'1)
% Organic Anions
Color (PCU)
Sum Base Cations*
SO/'iSBC ratio
SO42-:ANC ratio
ANC
PH
Altot (pM)
AW (A*M)
Al(im) (/xM)
Fe (MM)
Mn 0*M)
Conductivity (/zS cm'1)
ANC < 0
n = 21
Mean SD
162.95
6.90
29.82
0.96
14.70
2.05
6.62
16.11
162.43
0.96
-15.68
-18.16
4.82
11.58
8.85
6.65
1.09
1.30
34.20
66.31
10.19
25.36
0.86
9.22
1.33
2.25
9.08
90.96
0.35
15.78
11.68
0.22
5.12
4.75
4.83
1.49
0.67
13.34
0 < ANC < 50
n = 69
Mean SD
121.40
49.01
78.07
19.46
12.84
1.45
4.72
14.02
270.91
0.52
12.95
23.83
5.88
4.82
1.54
0.76
0.78
0.65
35.76
65.46
101.78
134.35
15.81
13.28
1.59
2.34
13.67
222.44
0.20
26.16
16.04
0.46
4.27
1.85
1.24
1.21
0.74
29.95
50 < ANC < 200
n = 255
Mean SD
109.18
33.82
71.69
102.27
12.96
1.34
4.13
18.22
340.99
0.31
1.06
122.26
6.74
6.12
0.31
0.06
1.22
0.56
38.66
71.90
80.55
75.04
43.64
8.12
0.85
1.82
11.09
187.38
0.16
0.87
42.54
0.35
9.48
0.35
0.12
2.94
1.44
23.95
* All concentrations are expressed in fteq L"1 except as noted.
* Concentrations were sea salt corrected as discussed in text.
n « number of nodes classified in each group.
% Organic Anions = percentage of total anions that are organic.
Al^t « MIBK-extractable aluminum (total monomeric).
Al(Jm) » inorganic monomeric aluminum (PCV method).
282
-------
Table 9-4. Weighted Mean and Standard Deviation (SD) of NSS-I Spring Index Chemistry in
Organic-influenced Stream Nodes
Chemical
Species
SO4*
N03-
Cl*
HCO3-
Organic Anions
DOC (mg L"1)
% Organic Anions
Color (PCU)
Sum Base Cations*
SO42':SBC ratio
SO42-:ANC ratio
ANC
PH
Altot GiM)
Alext 0*M)
Al(im) (MM)
Fe (MM)
Mn (/zM)
Conductivity (/iS cm'1)
ANC < 0
n = 21
Mean SD
105.33
6.72
16.09
1.50
24.01
4.09
14.55
28.76
73.54
0.84
-4.50
-54.09
4.59
8.05
6.60
4.25
4.61
1.27
40.47
58.90
13.84
26.60
1.44
18.28
3.69
4.22
32.55
42.71
0.42
9.26
54.05
0.47
6.20
6.48
3.81
4.69
2.57
25.25
ORGANIC
0 < ANC
n = 21
Mean
74.63
10.02
48.66
21.75
29.35
3.26
15.80
36.04
165.24
0.37
2.92
24.51
5.66
11.01
2.37
0.81
2.73
1.01
24.76
INFLUENCED STREAMS
<50
SD
49.79
15.03
38.46
13.23
19.27
2.11
5.08
19.73
92.14
0.16
7.74
15.58
0.36
9.65
2.33
0.89
2.35
0.79
10.74
50 < ANC < 200
n = 23
Mean SD
92.02
40.46
106.08
103.77
59.63
6.32
14.57
70.08
379.19
0.25
1.33
121.33
6.25
17.52
1.78
0.69
9.56
2.51
47.13
49.55
78.37
98.84
40.21
36.84
4.00
4.71
39.60
218.95
0.11
2.09
40.61
0.40
19.25
1.75
0.99
14.89
2.50
28.88
* All concentrations are expressed in /zeq L"1 except as noted.
* Concentrations were sea salt corrected as discusssed in text.
n = number of nodes classified in each group.
% Organic Anions = percentage of total anions that are organic.
Alext = MIBK-extractable aluminum (total monomeric).
Al/im) = inorganic monomeric aluminum (PCV method).
283
-------
between the organic-dominated and inorganic streams (Tables 9-2 and 9-3). Sulfate was the
major anion in the acidic and very low ANC subgroups, whereas chloride and bicarbonate
concentrations were higher than sulfate concentrations in the low ANC subgroup.
9.3.4 Discussion
The classification process described in this subsection can best be summarized with a flow-
chart (Figure 9-9). The end result of the classification is a group of high-interest streams for
an assessment of acidic deposition effects. The criteria used to remove streams with probable
acid sources other than acid deposition are summarized on the flowchart.
The differences in relative composition of anions among the stream chemical classes can be
seen in trilinear plots of sulfate, nitrate, and organic anion (Figure 9-10). In each of the ANC
classes, the organic-dominated stream nodes are clustered near the high organic anion tip of the
triangle. Inorganic stream nodes are concentrated near the high-sulfate end of the triangle with
some dispersion along the nitrate axis, all with low organic anion values. The organic-
influenced stream nodes are intermediate between the organic-dominated and inorganic stream
nodes.
In order to determine if neutral salt displacement of hydrogen ions was occurring in
coastal areas, stream water ANC was plotted against the stream water Na+:Cl~ ratio. If Na+ is
replacing H+ in these watersheds, the more acidic streams (lower ANC) should have lower
Na+:Cr ratios, because the sodium would be removed while the chloride would remain constant
(assuming chloride is conservative). In stream nodes within 50 km of the ocean, there was no
relationship (r2 = 0.017) between ANC and the Na+:Cl~ ratio (Figure 9-11). The most acidic
stream nodes (ANC < 0), had Na+:Cl~ ratios greater than or equal to that found in sea water
(0.86). Similar plots of ANC in coastal streams versus the ratios of the other base cations to
chloride also showed little relationship (Ca2+:Cr r2 = 0.297, Mg2+:Cl~ r2 = 0.077, K+:C1" r2 =
0.018). Thus on the regional level, neutral salt hydrogen ion displacement does not appear to
play a major role in stream acidification. There were many streams with Na+:Cl~ greater than
0.86, indicating internal watershed sources of sodium.
Fish populations are very susceptible to high concentrations of aluminum, particularly
inorganic monomeric aluminum (Al(imp (subsection 6.3.2). The Al(im) concentrations were much
higher in NSS-I streams with ANC < 0 /ieq L"1 compared to low ANC streams (Figure 9-12) due
to aluminum mobilization in the acidic water. In streams with ANC < 200 jteq L"1, all of the
inorganic, 85% of the organic-influenced, and roughly 50% of the organic-dominated streams had
AI(imj concentrations below 0.05 mg L"1, a level generally considered a threshold concentration
above which various fish species experience reductions in growth and survival (Figure 9-12).
The inorganic, acidic stream group, those most likely to have been affected by acid deposition,
had high Al(imj concentrations. About 75% of the streams in this group had Abim) concentra-
tions within the range of concentrations thought to cause damage to fish populations. Acidic
streams with organic influence and dominance had lower Al/im) concentrations than the inor-
ganic streams but the median concentrations were still within the range of toxic concentrations.
9.4 POPULATION ESTIMATES OF THE PROBABLE SOURCES OF ACIDITY
The NSS-I statistical design permits an estimation of the number of stream reaches in each
subregion within each chemical classification group (Section 2.4; Overton, 1987). The population
estimates for stream reaches classified as having acid mine drainage and substantial watershed
sources of sulfate are presented in Table 9-5. The majority of these high-sulfate streams are
284
-------
CHEMICAL CLASSIFICATION OF NSS-I STREAMS
NSS-I Sample Streams
CRITERIA
1. ANC > 200 ueq L'1
CHEMICAL CLASS
LOW ANC SUBPOPULATION
/Substantial
[ Watershed
I Sources of
Sulfate
2. [SO42'] > 300 |Lieq L"1
In Mid-Atlantic
[SO42-] > 200 |jeq L'1
. In Southeast
3. -ANC 400 ueq L"1
- DOC < 5 mg L'1
- SO42YAnion Sum
>0.75
4. [Organic Anlons] >
[S04* + N03-J
Org. Dominated \
- Acidic 1
-Very Low ANC I
- Low ANC /
5. [Organic Anion]
[Total Anions]
HIGH INTEREST SUBPOPULATION
NO
\
f
Inorganic Class
~^
s~
Org. Influenced
- Acidic
- Very Low ANC
-Low ANC
Inorganic
- Acidic
- Very Low ANC
- Low ANC
ANC CLASSES (ueq L'1): Acidic (< 0), Very Low (0 to < 50), Low (50 to < 200)
Figure 9-9. Chemical classification flowchart.
285
-------
ANC
-1
ANC
0-50
LEGEND
• INORGANIC
O ORGANIC
INFLUENCED
A ORGANIC
DOMINATED
Figure 9-10. Trilinear plots for 804*, NO3~, and organic anions showing the different anionic
makeup of inorganic, organic-influenced, and organic-dominated NSS-I nodes in
three different ANC classes.
286
-------
200-f
00-
0-
o
z
-100-
-200-
i—
0.0
• . .
SEAWATER-
—i 1 1—
0.5 1.0
Na + :CI~ Ratio
1.5
Figure 9-11. Scatterplot showing the relationship between stream water ANC (fieq L"1) and
stream water Na+:Cl~ molar ratio in NSS-I nodes within 50 km of the coast. The
vertical line at 0.86 represents the sea water Na+:Cl~ ratio.
287
-------
j.o
<
0)
E
o
c
o
c
a
o>
o
c
0
BOth Percentile
Median
2Oth Percentile
REF
REF
ANC ANC ANC
<0 0-50 50-200
Inorganic
ANC
-------
Table 9-5. Estimated Number and Length of Stream Reaches Classified as Impacted by Acid
Mine Drainage or as Having Substantial Watershed Sources of Sulfate at Either
the Upper or Lower Node
ACID MINE DRAINAGE IMPACTED STREAMS
GROUP
Estimated
Number of
Reaches
Estimated
Length
(km)
SUBREGION
Poconos/Catskills (ID)
Valley and Ridge (2Bn)
N. Appalachians (2Cn)
S. Appalachians (2X)
TOTAL
9
318
929
121
1377
126
849
2496
1123
4594
GEOGRAPHIC SITE CLASS
Allegheny High Plateau
Allegheny Forested Plateau
Allegheny Forested/Agric. Plateau
Glaciated Forested Plateau
Valleys of the Valley & Ridge Province
TOTAL
151
296
603
9
318
1377
627
780
2212
126
849
4594
STREAMS WITH SUBSTANTIAL WATERSHED SOURCES OF SULFATE
Poconos/Catskills (ID)
N. Appalachians (2Cn)
Valley and Ridge (2Bn)
Coastal Plain (3B)
Piedmont (3A)
S. Appalachians (2X)
Ozarks/Ouchitas (2D)
Total
Estimated
Number of
Reaches
123
1372
1078
1139
158
225
225
4216
Estimated
Length
(km)
180
3818
4992
3409
564
1568
1568
15145
The distribution of streams with substantial watershed sources of sulfate among the different
geographic site classes is given in Tables 9-8 and 9-9.
289
_
-------
found in the Northern Appalachians (2Cn). Of the estimated 9,417 reaches in the initial target
population in this subregion, 929 (9.9%) were impacted by acid mine drainage at one or both
nodes; 1,372 (15%) reaches had substantial watershed sources of sulfate at one or both nodes.
Estimations indicate that 1,377 stream reaches, or over 4,500 km of streams in the entire NSS-I
sampling area, were impacted by acid mine drainage at either upper or lower nodes, or both.
Almost all the sites impacted by acid mine drainage were found in the Allegheny Plateau
geographic site class. A higher number of stream reaches (4,216) have ANC < 200 /zeq L"1 and
receive substantial contributions of sulfate from watershed sources.
It is important to keep in mind that the NSS-I subregion locations and target stream
resource definitions were delineated for the purpose of assessing the status and extent of acid
deposition effects. Therefore, the population estimates and maps presented in this section apply
only to the streams affected by acid mine drainage within these areas and within the target
reach size limitation and other inclusion criteria. The total stream resource impacted by acid
mine drainage may be considerably larger.
The number of acidic, very low ANC and low ANC stream reaches, divided into inorganic,
organic-influenced and organic-dominated groups for each subregion, are presented for lower
nodes in Figure 9-13 and for upper nodes in Figure 9-14. No acidic lower nodes were observed
in the Poconos/Catskills (ID), Valley and Ridge (2Bn), Ozarks/Ouachitas (2D), Southern Blue
Ridge (2As), Southern Appalachians (2X), or Piedmont (3A). The acidic lower reaches in Florida
(3C) were all organic dominated, whereas those in the Mid-Atlantic Coastal Plain (3B) were
found in roughly equal amounts as inorganic, organic influenced, and organic dominated. The
acidic lower reaches in subregion 2Cn were all inorganic. No acidic upper nodes were observed
in subregion 2D, 2As, or 3A. The vast majority of the acidic upper reaches observed in the
Interior Mid-Atlantic subregions (ID, 2Bn, 2Cn) were classified as inorganic.
Almost all the reaches in Florida (3C) estimated to have low and very low ANC were
affected by organic acidity. Very few reaches affected by organic acidity were estimated
outside of subregions 3A, 3B, and 3C. In the Ozarks/Ouachitas (2D), very few reaches had ANC
^ 50 peq L"1, but almost 3,000 upper and lower reaches were estimated to have ANC between 50
and 200 peq L'1. The pattern observed in lower reach nodes in subregions 2Bn and 2As was
similar to that in 2D.
In the combined NSS-I subregions, an estimated 565 downstream nodes (1.0% of the target
population) and more than three times as many upstream nodes (1,812, or 3.2%) were acidic and
classified as inorganic, indicating that the most probable source of acidity was acid deposition.
Another 168 (0.3%) lower and 375 (0.7%) upper nodes were organic-influenced acidic streams,
with at least a portion of their acidity potentially derived from organic acids. There were 4,159
downstream nodes (7.3%) with very low ANC estimated in the high-interest subpopulation; 3,070
were inorganic and 1,089 were organic influenced. In contrast, 5,975 upstream nodes (10.5%) had
very low ANC (4,815 inorganic, 1,160 organic influenced). Over 13,000 upstream and downstream
reach nodes (23-25%) were classified as inorganic, with low ANC. A smaller number of low ANC
reaches (1,305 lower nodes, 1,704 upper nodes) were classified as organic influenced.
9.4.1 Interpolated Length Estimates
Chemistry at upper and lower ends of stream reaches with one or both nodes in the high-
interest subpopulation was linearly interpolated to estimate the stream length having less than
specified pH and ANC reference values. The NSS-I population expansion algorithm and the
interpolated reach lengths were used to estimate the length of streams having less than
reference values for the target population. Hydrogen ion concentration was used to linearly
290
-------
NSS-I LOWER NODES
SUBREGION
POCONO/CATSKILLS
N. APPALACHIANS
VALLEY AND RIDGE
MA COASTAL PLAIN
S. BLUE RIDGE
PIEDMONT
S. APPALACHIANS
OZARKS/OUACHITAS
FLORIDA
1000 2000 3000
ESTIMATED NUMBER OF REACHES
4000
Figure 9-13. Population estimates of the number of downstream reach nodes in each NSS-I
subregion found in inorganic, organic-influenced, and organic-dominated chemical
classes.
291
-------
NSS-I UPPER NODES
SUBREGION (jjeq L~')
POCONO/CATSKILLS
N. APPALACHIANS
VALLEY AND RIDGE
MA COASTAL PLAIN
S. BLUE FUDGE
PIEDMONT
S. APPALACHIANS
OZARKS/OUACHITAS
FLORIDA
1OOO 200O 3000
ESTIMATED NUMBER OF REACHES
400O
Figure 9-14. Population estimates of the number of upstream reach nodes in each NSS-I
subregion found in inorganic, organic-influenced, and organic-dominated chemical
classes.
292
-------
interpolate pH. In cases where the alternate node of an inorganic or organic-influenced node
was noninterest (affected by acid mine drainage, tidal influences, etc.), impacted by watershed
sulfate sources, or dominated by organic anions, one-half of the reach was assumed to have the
high-interest chemistry. This assumption was necessary because the ANC and pH of the alter-
nate node could not be used in the interpolation, since its chemistry did not place it in the
high-interest subpopulation. If one node was classified as inorganic and the other as organic
influenced, and the chemistry of both nodes was below the reference value, then the reach
length was classified as half inorganic and half organic influenced. If one node was above the
reference value and the other below the reference value, the entire interpolated reach length
was classified the same as the node below the reference value.
The estimated combined length of acidic (ANC < 0) stream reaches within the high-interest
subpopulation was 4,455 km (Table 9-6). Of this length, 73% was classified in the inorganic
group. These inorganic acid streams were found primarily in the interior subregions of the Mid-
Atlantic Region (ID, 2Bn, and 2Cn), in the New Jersey Pine Barrens in subregion 3B, and in a
small length in Florida (3C). An estimated 19,945 km of streams had ANC < 50 /zeq L'1 and
were classified in the high-interest subpopulation. Seventy-one percent of these streams were
classified as inorganic.
The estimated combined length of stream reaches with pH < 5.0 within the high-interest
subpopulation was 4,145 km (Table 9-7). No sites with pH < 5.0 classified as being in the high-
interest subpopulation were observed in the southeastern United States. An estimated 13,642 km
of streams had pH < 5.5. A greater length of streams had pH < 5.5 in the Mid-Atlantic than
the Southeast.
Estimates of acidic stream length (pH < 5.0, ANC < 0) based on the first NSS-I sample (in
the Mid-Atlantic) were somewhat higher than estimates based on the index value (Table 9-6),
due to a trend of increasing spring baseflow ANC. For ANC < 50 /*eq L"1, ANC < 200 (teq L"1,
pH < 5.5, and pH < 6.0, the estimated stream lengths based on first visit and index values were
very similar.
9.5 GEOGRAPHIC DISTRIBUTION OF STREAM CHEMICAL GROUPINGS
In order to examine the relationships between probable sources of stream water acidity and
geographic characteristics, stream node chemical classifications were categorized by geographic
site classification (Section 5.5). Using the NSS-I population expansion algorithm, the number of
stream reaches in each chemical class was estimated for each geographic site class. These esti-
mated numbers are reported for both downstream (Table 9-8) and upstream (Table 9-9) reach
nodes. The percentage of acidic (ANC < 0 0eq L"1) and very low ANC (0-50 /*eq L"1) down-
stream (Table 9-10) and upstream reach nodes (Table 9-11) within the refined target population
of each geographic site class were also estimated. The geographic distribution of acid mine
drainage sites, which are not in the refined target population, was presented in Table 9-5.
It should be emphasized that the number of reaches reported in Tables 9-8 and 9-9 are
estimates, based on roughly 450 stream reach samples and calculated using the NSS-I probability
sampling framework (Section 2.4). In some cases, the population estimate is based on only one
sampled reach or a very few sampled reaches belonging to a particular chemical and geographic
site class combination. While these estimates should be interpreted with caution, they represent
a statistical best estimate of the number of stream reaches within specific combinations of geo-
graphical and chemical class. Standard errors associated with these population estimates are
detailed in Appendix D. The number of streams (n) actually sampled in the refined target
population within each geographic site class is listed in Tables 9-10 (lower nodes) and 9-11
293
-------
Table 9-6. Interpolated Length (km) Estimates of Streams in the High-Interest Subpopulation
with ANC Less than Reference Values
Region
Interior Mid- Atlantic
Mid-Atlantic Coastal Plain
Interior Southeast
Florida
Total
Inorganic
1984 (2607)
1246 (3292)
13
3243 (5912)
ANC < 0 fieq L'1
Organic Influenced
382 (390)
660 (1009)
121
49
1212 (1569)
Total
2366 (2997)
1906 (4301)
121
62
4455 (7481)
Region
Inorganic
ANC < 50 Ateq L'1
Organic Influenced
Total
Interior Mid-Atlantic
Mid-Atlantic Coastal Plain
Interior Southeast
Florida
Total
Region
6925 (8686)
5173 (5218)
3000
790
15,888 (17,694)
Inorganic
454 (433)
2217 (2336)
1070
416
4157 (4255)
ANC < 200 /zeq L
Organic Influenced
7379 (9119)
7390 (7554)
4070
1206
20,045 (21,949)
-i
Total
Interior Mid-Atlantic
Mid-Atlantic Coastal Plain
Interior Southeast
Florida
Total
26,042 (27447)
7938 (7972)
33,706
499
68,185 (69,624)
509 (558)
5782 (5902)
6298
936
13,525 (13,694)
26,551 (28,005)
13,720 (13,874)
40,004
1435
81,710 (83,318)
Values in parentheses give the estimated length using data from the first sampling visit rather
than the spring index value. Sites in the Southeast were only visited once.
294
-------
Table 9-7. Interpolated Length (km) Estimates of Streams in the High-Interest Subpopulation
with pH Less than Reference Values
Region
Interior Mid- Atlantic
Mid- Atlantic Coastal Plain
Interior Southeast
Florida
Total
Inorganic
1782 (1933)
1241 (2401)
3023 (4334)
pH < 5.0
Organic Influenced
426 (434)
696 (1009)
1122 (1443)
Total
2208 (2367)
1937 (3410)
4145 (5777)
Region
Inorganic
pH < 5.5
Organic Influenced
Total
Interior Mid-Atlantic
Mid-Atlantic Coastal Plain
Interior Southeast
Florida
Total
3961 (4093)
6154 (6319)
411
314
10,840 (11,137)
495 (514)
1439 (1414)
317
551
2802 (2796)
4456 (4607)
7593 (7733)
728
865
13,642 (13,933)
Region
Inorganic
pH < 6.0
Organic Influenced
Total
Interior Mid-Atlantic
Mid-Atlantic Coastal Plain
Interior Southeast
Florida
Total
6394 (6732)
8057 (7766)
3601
491
18,543 (18,590)
780 (726)
5806 (5201)
1917
790
9293 (8634)
7174 (7458)
13,863 (12,967)
5518
1281
27,836 (27,224)
Values in parentheses give the estimated length using data from the first sampling visit instead
of the spring index value. Sites in the Southeast were only visited once.
295
_
-------
Table 9-8. Estimated Number of Target Reach Lower Nodes in Each Geographic Site Class
Present in Stream Chemical Classes
INORGANIC
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plat.
NE Mid- Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plat.
Glaciated Forested Plat.
N*
218
4,178
3,547
351
866
209
ANC ANC
< 0 0-50
36 25
291 1320
-
2
-
80
ANC
50-200
158
760
833
229
119
119
ORGANIC INFLUENCED
ANC ANC ANC
< 0 0-50 50-200
- - -
- - -
- - -
- - -
_
- - -
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid-Atlantic
Western Mid-Atlantic
4,359
11,809
873
1,126
1,365
9,258
2,297
2,962
1,307
64
2,036
6,133
New Jersey Pine Barrens 432
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I TOTALS
1,580
753
687
442
264
163
57,277
427
-
-
-
_
-
-
67
243
-
239
478
239
158
-
32
-
_
-
565 3,070
1,012
2,370
723
375
773
1,108
-
2,053
821
-
239
478
32
474
453
32
-
32
-
13,194
_
16
..
150
-
397
7
28
121
_
239
239 239
161
474 158
75
227 32
-
- - -
168 1,089 1,305
# N represents the estimated number of reaches in each geographic site class.
See Figure 9-9 for a flow chart depicting the chemical classification procedure.
296
-------
Table 9-8. Estimated Number of Target Reach Lower Nodes in Each Geographic Site Class
Present in Stream Chemical Classes (Continued)
Geographic Site Class
ORGANIC DOMINATED Watershed
ANC ANC ANC sulfate ANC
< 0 0-50 50-200 sources > 200
Allegheny Plateau
High Plateau 218
Forested Plateau 4,178
Forested Agric. Plat. 3,547
767
600
1,040
2,114
NE Mid-Atlantic
Pocono/Catskill Mts. 351
Glaciated Agric. Plat. 866
Glaciated Forested Plat. 209
119
746
10
Valley and Ridge Province
Ridges 4,359
Valleys 11,809
636
318
2,284
9,106
Arkansas/Oklahoma
Boston Mts. 873
Ouachita Mts. 1,126
Arkansas R. Valley 1,365
75
150
525
591
Piedmont Province
Piedmont 9,258
Piedmont Lowlands 2,297
7,754
2,290
S. Appalachian Highlands
Blue Ridge Mts. 2,962
Cumberland Plateau 1,307
French Broad R. Valley 64
121
814
64
Coastal Plain
Eastern Mid-Atlantic 2,036
Western Mid-Atlantic 6,133
New Jersey Pine Barrens 432
239
478
478
365
239
955
3,267
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I TOTALS
1,580
753
687
442
264
163
57,277
-
-
-
-
129
96
464
-
-
203
208
-
-
889
158
75
32
170
35
35
983
158
-
w
_
_
3,277
_
150
129
64
67
32
32,272
* N represents the estimated number of reaches in each geographic site class.
See Figure 9-9 for a flow chart depicting the chemical classification procedure.
297
-------
Table 9-9. Estimated Number of Target Reach Upper Nodes in Each Geographic Site Class
Present in Stream Chemical Classes
INORGANIC
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plat.
NE Mid-Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plat.
Glaciated Forested Plat.
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid-Atlantic
Western Mid-Atlantic
ANC
N* <0
221 63
4,327 436
3,448
351
818
266 77
7,190 695
8,342
873
1,126
1,452
8,940
2,290 239
2,962
1,307
64
2,036
6,133
New Jersey Pine Barrens 436 239
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I TOTALS
1,580
753
623 64
492
117
495
56,642 1,812
ANC
0-50
-
1,464
291
140
-
60
1,044
60
-
75
-
-
-
231
121
-
239
478
32
474
75
32
-
-
-
4,815
ANC
50-200
158
589
955
209
60
60
2,119
1,970
648
525
773
1,739
-
2,134
364
-
716
239
-
790
302
-
-
-
-
14,353
ORGANIC INFLUENCED
ANC ANC ANC
< 0 0-50 50-200
- - -
29
168
2
_
69
60
— — —
75
150
- - -
397
™ — —
17
121 - 279
- - -
- - -
716 478
107
158 158
75
75 71 32
— - —
— - —
_ — —
375 1,160 1,704
# N represents the estimated number of reaches in each geographic site class.
See Figure 9-9 for a flow chart depicting the chemical classification procedure.
298
-------
Table 9-9. Estimated Number of Target Reach Upper Nodes in Each Geographic Site Class
Present in Stream Chemical Classes (Continued)
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plat.
NE Mid-Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plat.
Glaciated Forested Plat.
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid-Atlantic
Western Mid- Atlantic
New Jersey Pine Barrens
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I TOTALS 5
ORGANIC DOMINATED
ANC ANC ANC
N* < 0 0-50 50-200
221 -
4,327 - - -
3,448 - - -
351 - -
818 -
266 - -
7,190 15
8,342 - -
873 -
1,126 - -
1,452 - - -
8,940
2,290 - - -
2,962 - -
1,307 - -
64
2,036 - - 239
6,133 716 239 478
436 -
1,580 - - -
753 - - -
623 - 221 64
492 - 208
117 78 -
495 460 - 35
'6,642 1,255 683 815
Watershed
sulfate
sources
— -
769
167
mm
_
-
443
123
_
75
75
_
-
_
_
-
365
478
58
_
-
_
_
_
-
2,551
ANC
> 200
_
1,040
1,867
758
-
2,813
6,189
150
300
604
6,804
2,051
579
421
64
478
2,312
—
—
300
64
284
39
- -
27,118
* N represents the estimated number of reaches in each geographic site class.
See Figure 9-9 for a flow chart depicting the chemical classification procedure.
299
-------
Table 9-10. Estimated Percentage of Target Stream Reach Lower Nodes in Each Geographic
Site Class in Acidic and Very Low ANC Chemical Classes
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plat.
NE Mid- Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plat.
Glaciated Forested Plat.
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid- Atlantic
Western Mid- Atlantic
New Jersey Pine Barrens
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I Totals
n$
6
27
26
8
15
6
19
72
10
15
14
50
9
56
10
4
9
25
11
10
9
19
4
6
5
445
N*
218
4,178
3,547
351
866
209
4,359
11,809
873
1,126
1,365
9,258
2,297
2,962
1,307
64
2,036
6,133
432
1,580
753
687
442
264
163
57.277
Inorganic
ANC
< 0 0-50
16.3 11.4
7.0 31.6
-
0.7
-
38.1
9.8
-
-
-
-
-
-
2.3
18.6
-
11.7
7.8
55.3
10.0
-
4.7
-
.
-
1.0 5.4
Organic Organic
Influenced Dominated
ANC ANC
< 0 0-50 < 0 0-50
_ _ - -
- - - -
- - - -
_
- - - -
- - - -
-
- - - -
- - - -
_
_
0.3 -
1.0
9.3
- - - -
3.9 3.9 7.8
37.3 -
30.0
- - - -
33.0 - 29.6
47.0
49.0
59.0
0.3 1.9 0.8 1.6
is the estimated number of target reaches within each geographic site class.
is the number of target stream nodes sampled within each geographic site class.
300
-------
Table 9-11. Estimated Percentage of Target Stream Reach Upper Nodes in Each Geographic
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plat.
NE Mid-Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plat.
Glaciated Forested Plat.
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid- Atlantic
Western Mid-Atlantic
New Jersey Pine Barrens
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I Totals
n*
7
29
28
8
16
7
37
52
10
15
15
49
8
56
10
4
9
25
11
10
9
17
5
3
6
446
N*
221
4,327
3,448
351
818
266
7,190
8,342
873
1,126
1,452
8,940
2,290
2,962
1,307
64
2,036
6,133
436
1,580
753
623
492
117
495
56.642
Organic
Inorganic Influenced
< 0 0-50 < 0 0-50
28.4 - -
10.1 33.8 - 0.7
8.4 - 4.9
39.9 0.7
- - - -
28.9 22.5 26.1
9.7 14.5
0.7
— — — -
6.7
- - -
— — _ _
10.4 - -
7.8 - 0.6
9.3 9.3
- _
11.7
7.8 - 11.7
54.8 7.4 24.5
30.0 - 10.0
10.0
10.3 5.1 12.1 11.3
- - _ _
- - _ _
- _
3.2 8.5 0.7 2.0
Organic
Dominated
< 0 0-50
_ _
_ _
-
_ —
_ _
-
0.2
-
— —
— _
-
— —
-
— _
_ _
-
_ _
11.7 3.9
_ _
_ _
-
35.5
42.2
66.5
93.0
22 12
N* is the estimated number of target reaches in each geographic region.
n* is the number of target stream nodes sampled within each geographic site class.
301
-------
(upper nodes). It would be erroneous to ignore weighted estimates in Tables 9-8 and 9-9 by
inferring population characteristics based on the distribution of stream sample numbers without
weighting, because the stream target population was not uniformly sampled. The weighting
factor takes into account the difference in sampling inclusion probabilities (including ANC
stratification) among the stream samples.
9.5.1 Streams with Watershed Sulfate Sources
Streams with ANC < 200 /zeq L'1 and substantial watershed sources of sulfate were found
mainly in the Allegheny Plateau, the Ouachita Mountains, the Valley and Ridge Province, and the
agricultural regions in the Coastal Plain. These high-sulfate sites made up 5.7% of the lower
node target population and 4.5% of the upper node target population.
9.5.2 Streams with Chemistry Dominated bv Organics
Except for an estimated 15 upstream reach nodes in the ridges of the Valley and Ridge
Province, all organic-dominated streams were found in coastal areas—Florida, and the Mid-
Atlantic and Gulf Coastal Plains. Organic acidic (ANC < 0 /zeq L'1) stream reaches were found
primarily along the western Mid-Atlantic Coastal Plain and in large numbers at lowland swampy
sites in both the Florida panhandle and peninsula. It was estimated that over 50% of the lower
nodes and 88% of the upper nodes of these swampy reaches in Florida were organic acidic.
Very low and low ANC (< 200 /zeq L"1), organic-dominated reaches were found in the relatively
higher elevation Florida upland and Mid-Atlantic and Gulf Coastal Plain sites.
9.5.3 High-Interest Suboopulation (Inorganic and Organic-Influenced Streams')
In the high-interest subpopulation, acidic lower nodes (an estimated 733 reaches, 1.3% of
the target population) were observed only in the Allegheny Plateau, the New Jersey Pine Bar-
rens, and the Piedmont Lowlands. Acidic high-interest upper nodes in the target population
(2,187 reaches, 3.9%) were observed in the Allegheny Plateau, the Pine Barrens, the forested
Glacial Plateau in northeast Pennsylvania, the Poconos/Catskills, the ridges of the Valley and
Ridge Province, the uplands of the Florida panhandle, the Cumberland Plateau, and the Piedmont
Lowlands.
9.5.3.1 Allegheny Plateau, Northeast Mid-Atlantic, and Valley and Ridge Geographic Sites—
In addition to a large number of acid mine drainage streams, the Allegheny Plateau was
estimated to have a large number of inorganic, acidic reaches (327 lower nodes (4.1%), and 499
upper nodes (6.2%)). In the Allegheny Plateau, acidic high-interest sites were found only on the
High Plateau and the Forested Plateau. On the Forested Plateau, 1,611 (39%) lower nodes and
1,929 (45%) upper nodes had ANC values < 50 /zeq L'1. No acidic sites were found on the
Plateau in areas of mixed forest and agriculture.
In the Valley and Ridge Province geographic site class (within subregions ID, 2Bn, 2Cn,
2As and 2X), only upper nodes on the ridges were acidic. Of the 7,190 estimated upper node
target sites on ridges, an estimated 695 (9.7%) reaches were acidic, 1,044 (15%) reaches had
spring index ANC concentrations between 0 and 50 /zeq L'1, and another 2,179 (30%) reaches had
ANC between 50 and 200 /zeq L"1. Valley sites had almost no sites with ANC < 50 /zeq L'1
(only 60 upper reach nodes), whereas there were 2,386 (20%) lower nodes and 1,970 (24%) upper
nodes at Valley sites with ANC between 50 and 200 /zeq L'1.
302
-------
In the Glacial Plateau, as in the Allegheny Plateau, there were more acidic and low ANC
reaches in forested drainages than in those with mixed forest and agricultural land use. No
reaches with ANC < 50 /jeq L'1 were observed in the agricultural part of the Glacial Plateau
and only an estimated 119 lower nodes (14%) and 60 upper nodes (7%) had ANC between 50 and
200 /*eq L-1. All of the reaches (except 10 downstream reach nodes) in the forested section of
the Glacial Plateau had ANC < 200 /zeq L"1. Over half of the estimated 266 upper nodes in this
geographic site class were acidic (77 nodes were inorganic and 69 were organic influenced). All
of the estimated 351 upper nodes and 231 of the lower nodes (65.8%) in the Pocono/Catskill
Mountains had ANC < 200 /zeq L"1. However, very few reaches had ANC < 50 jueq L"1 (an esti-
mated 142 upper nodes and 2 lower nodes).
Very few of the estimated 25,537 lower nodes and 24,963 upper nodes in the interior
section of the Mid-Atlantic (Allegheny Plateau, Glaciated Plateau, Valley and Ridge, and
Pocono/Catskill Mountains) were organic influenced (16 lower nodes and 328 upper nodes). The
organic-influenced stream reaches present were usually located in valleys or in low gradient
headwaters atop plateaus.
9.5.3.2 Florida, Arkansas/Oklahoma, Piedmont, and Southern Appalachian Highland
Geographic Sites—
The only Southeast geographic site class in which acidic streams not influenced by organics
were observed was in Florida. These estimated 64 inorganic acidic reaches were located in
upstream nodes of upland sites in the Florida panhandle. Between 20% and 30% of upstream and
downstream nodes in upland drainages in the Florida panhandle were classed as organic-
influenced reaches with ANC below 50 /zeq L"1.
In Arkansas and Oklahoma, no acidic sites were observed and only 75 upper nodes were
estimated to have ANC between 0 and 50 /zeq L"1. All three geographic site classes in this
area, however, were dominated by streams in the low ANC class (50-200 /zeq L"1). Of the 3,364
lower nodes and 3,451 upper nodes in this area, 63% of the lower nodes and 60% of the upper
nodes were in the low ANC class, mostly in the inorganic group.
There were no reaches with ANC < 50 peq L'1 observed in the Piedmont, but there were
an estimated 1,505 (16%) lower nodes and 2,136 (24%) upper nodes with ANC between 50 and 200
jieq L"1. The Piedmont geographic site class extends from southeastern Pennsylvania in the
north to Mississippi in the south and is not entirely within the Southeast Screening area.
Except for 121 inorganic, acidic upper reach nodes and 7 organic-influenced lower nodes, there
were no reaches with ANC < 200 #eq L"1 in the Piedmont Lowlands. The estimate of 121 acidic
reaches resulted from 1 upper node sampled in a forested ridge in southeastern Pennsylvania.
The land use pattern for this node was atypical for the Piedmont Lowlands. The estimate of 7
organic-influenced acidic downstream reaches resulted from 1 lower node sampled that was just
outside the New Jersey Pine Barrens. The upper node of the reach was in the Pine Barrens.
The majority of the estimated 4,269 upstream and downstream nodes in the Blue Ridge
Mountains and Cumberland Plateau were in the 50-200 jteq L"1 ANC class (2,874 lower nodes and
2,777 upper nodes). A small number of streams in this area (149 lower nodes and 417 upper
nodes) were organic influenced. All of the estimated 64 reaches in the French Broad River
Valley of North Carolina had ANC > 200 peq L"1.
9.5.3.3 Coastal Plain Geographic Sites--
In the New Jersey Pine Barrens, 93% of the estimated 432 lower nodes and 79% of the 436
upper nodes were acidic. Of these acidic nodes, 239 (55% of target population) upper and lower
reach nodes were inorganic, whereas 107 (25%) of the upper reach nodes and 161 (37%) of the
303
-------
lower reach nodes were classified as organic-influenced, but not organic-dominated, acidic
reaches. The fact that there was a greater number of acidic lower reach nodes than upper
nodes is probably due to organic acid influence. The water at the lower nodes has been in
contact with organic sources longer and thus could have leached out more organic acids. The
remaining nodes in the Pine Barrens all had ANC < 200 /zeq L"1. Contrary to expectations, no
reaches were observed in the Pine Barrens where organic anions clearly dominated sulfate and
nitrate, even though Pine Barrens reaches had high DOC concentrations (mean = 7.96 mg L"1,
SD = 6.88) and color (mean - 54, SD - 43). This is in marked contrast to findings in Florida,
where a clear majority of reaches in the target population were dominated by acidic and very
low ANC reaches with high DOC concentrations and very low concentrations of sulfate and
nitrate. These findings are to be expected, given the considerably higher atmospheric sulfate
deposition rates in the Pine Barrens (2.5-3.0 g m'2 yr"1) compared to Florida (< 1.8 g m'2 yr"1)
(Section 2, Figure 2-2).
Johnson (1979a,b) has noted an annual drop of 0.2-0.5 pH units in the long-term record of
two streams in the Pine Barrens and suggested that it was due to an observed decrease in
precipitation pH. Morgan (1984) disagreed with Johnson's assessment and concluded that any
long-term trend in these data could be divided into several short-term trends, each explainable
by natural causes, and none of which was entirely consistent with an acid deposition hypothesis.
In addition, aquatic flora and fauna communities characteristic of very low pH waters have long
existed in the Pine Barrens (since at least 1900), as evidenced by historic descriptions of flora
and fauna characteristic of acidic environments (Forman, 1979; Morgan, 1984). Almost all the
historic pH data (1950 - present) from the Pine Barrens have shown that the normal pH of the
waters in this area is between 4.0 and 5.5. Although it appears likely that much of the acidity
in the streams in the Pine Barrens is of natural origin, it is possible that acid deposition may
currently be causing decreases in already low stream ANC and may have acidified some stream
systems.
No inorganic, acidic reaches were observed in the Mid-Atlantic and Gulf Coastal Plain sites
(an estimated 10,502 upper and lower nodes). However, a large number of reaches influenced,
but not dominated by organic acids (1,424 lower and 1,585 upper reach nodes) were estimated in
this area. One-fourth to one-third of the stream reaches in the Coastal Plain (excluding the
Pine Barrens) were inorganic, with ANC < 200 #eq IT1 (2,519 lower nodes, 3,313 upper nodes).
9.6 CLASSIFICATION OF SPECIAL INTEREST SITES
The 36 NSS-I special interest sites (Section 2.6, Table 2-4) were classified geographically
and chemically as described in Section 9.3 using spring index data (see flowchart in Figure 9-9).
The vast majority of the special interest streams (Table 9-12) had ANC below 200 /*eq L'1 (31 of
the 36 streams). Most of the streams (21) were classified as inorganic. Almost all of these
inorganic streams were located in forested, mountainous, or ridge top areas.
The special interest site in the New Jersey Pine Barrens (McDonalds Branch) was classified
as organic-influenced acidic, as were many of the NSS-I streams sampled there. All of the
organic-influenced streams with ANC between 0 and 200 p,eq L"1 were found in the Southeast,
mainly in the Blue Ridge Mountains. In spite of the fact that the DOC concentration in all of
these streams was less than 1 mg L"1, the concentration of the inorganic ions was so low
(sulfate « 14-40 /*eq L"1) that the streams were classified as organic influenced. It should be
remembered, however, that the system decision limit for DOC samples was 0.45 mg L"1 (Table
4-1). Also, DOC measurements at concentrations below 2 mg L'1 had a %RSD of 20% - 25%
(Table 4-3). Thus, it is possible that there is almost no DOC present in these streams and
304
-------
Table 9-12. Chemical and Geographical Classification of Special Interest Site Streams Sampled
during NSS-I. (Index chemistry is given in units of /zeq L"1 except for DOC, which is
given in mg L"1.)
Stream Name
INORGANIC CLASS,
E. Branch Neversink
Mill Run
Cole Run
INORGANIC CLASS,
Biscuit Brook
Tillman Brook
Cosby Creek
Detweiler Run
White Oak Run
Deep Run
N. Fork Bens Creek
Fernow Control
N. Fork Saline R.
INORGANIC CLASS,
High Falls Creek
Van Campens Brook
Green Brook
Leading Ridge #1
Laurel Run
N. Fork Dry Run
Old Rag Run
E. Fork Saline R.
Pequea Creek
Reach ID
ANC < 0 /zeq
1D030912
2B047919
2C041914
Geographic Class
L-1
Catskill/Pocono Mts
Ridges of Valley &
Ridge
Forested Allegheny Plateau
ANC
-7.3
-20.7
-30.4
Index Chemistry
SO42" pH DOC
124.3
233.2
186.6
5.0
4.7
4.8
1.88
2.39
1.00
ANC 0-50 /zeq L"1
1D030910
1D030926
2A07891
2B036903
2B047916
2B047917
2C035913
2C041901
2D083927
ANC 50-200
1D030911
1D030924
1D030925
2B036904
2B036905
2B047918
2B047920
2D083928
3B042921
ORGANIC INFLUENCED CLASS,
McDonalds Branch
3B043909
ORGANIC INFLUENCED CLASS,
Pinnacle Branch
Chester Creek
Camp Branch
2A07896
2A08891
2C077923
Catskill/Pocono Mts
Ridges of Valley &
Blue Ridge Mts.
Ridges of Valley &
Blue Ridge Mts.
Blue Ridge Mts.
Forested Allegheny
Forested Allegheny
Ouachita Mts.
/zeq L'1
Catskill/Pocono Mts
Ridges of Valley &
.
Ridge
Ridge
Plateau
Plateau
Ridge
Glaciated Forested Plateau
Ridges of Valley &
Ridges of Valley &
Blue Ridge Mts.
Blue Ridge Mts.
Ouachita Mts.
Piedmont
ANC < 0 /ieq L'1
NJ Pine Barrens
ANC 0-50 /ieq L'1
Blue Ridge Mts.
Blue Ridge Mts.
Ridge
Ridge
Cumberland Plateau
31.6
41.0
37.0
1.7
18.9
3.0
7.1
18.2
16.0
101.8
129.8
124.0
171.0
51.4
51.3
61.1
57.5
97.4
-101.0
47.1
41.2
14.9
141.9
221.7
50.7
149.8
81.5
99.8
195.2
92.6
59.8
141.5
238.8
247.0
177.5
160.6
101.0
37.0
55.9
191.1
176.0
22.3
15.2
39.6
6.4
6.5
6.7
5.5
6.2
5.7
5.8
6.0
6.2
6.9
7.1
6.7
6.9
6.5
6.6
6.6
6.6
6.7
4.0
6.8
6.7
5.2
1.51
2.21
0.38
0.80
0.55
0.88
0.49
0.60
0.45
1.76
1.15
2.04
1.04
0.87
0.52
0.63
0.00
0.85
6.22
0.44
0.42
0.98
305
-------
Table 9-12. Chemical and Geographical Classification of Special Interest Site Streams Sampled
during NSS-I. Index chemistry is given in units of /zeq L"1 except for DOC, which is
given in mg L'1. (Continued)
Stream Name
Reach ID Geographic Class
ANC
Index Chemistry
SO42- pH DOC
ORGANIC INFLUENCED CLASS, ANC 50-200 fjieq L'1
Twenty-mile Creek
Jarrett Creek
Shope Fork/Gauge 8
Moses Creek
2A07892
2A07893
2A07894
2A07895
Blue Ridge Mts.
Blue Ridge Mts.
Blue Ridge Mts.
Blue Ridge Mts.
61.6
86.7
76.3
134.0
19.5
13.8
14.0
21.1
6.8
6.9
6.9
7.2
0.81
0.55
0.55
0.77
ACID MINE DRAINAGE IMPACTED
N. Br. Quemahoning 2C035915 Forested Allegheny Plateau -51.4
298.4 4.5 0.72
SUBSTANTIAL WATERSHED SOURCES OF SULFATE CLASS
Bacon Ridge 3B042908 West MA Coastal Plain 183.4 425.7
ANC > 200 peq L'1 CLASS
6.5 1.92
Hunting Creek
Mill Run Trib
Elk Lick Run
Magothy River
Lyons Creek
2B042922
2B047929
2C041902
3B042907
3B048906
Blue Ridge Mts.
Ridges of Valley & Ridge
Forested Allegheny Plateau
West MA Coastal Plain
West MA Coastal Plain
332.6
468.1
325.1
224.6
255.3
122.2
147.5
178.7
430.2
589.8
7.4
7.4
7.5
6.6
6.9
1.86
1.18
0.69
3.08
2.55
MA « Mid-Atlantic
306
-------
they should be in the inorganic class. It should be emphasized that streams in this organic-
influenced group are still in the high-interest subpopulation of streams. The organic anion
proportion is high enough, however, to potentially have some effect on acid/base chemistry.
The North Branch of the Quemahoning was classified as impacted by acid mine drainage,
even though its index SO42' concentration (298 /zeq L"1) was just below the 300 /*eq L"1 cutoff.
This classification was made after evidence of acid mine drainage was found on a field visit
during the 1987 Acid Stream Reconnaisance. The acid mine drainage, however, was not severe.
The headwaters of this stream are fairly pristine. It is possible that the NSS-I sample was
taken downstream of a site where other research is taking place and picked up acid mine
drainage effects.
Examination of the special interest site data is in its beginning phase. It is of special
interest because some of these streams may be selected for future long-term monitoring of
trends and effects of acid deposition. Thus it is important to find out how these special
interest site streams compare with the streams selected through a randomized systematic sample.
307
-------
-------
SECTION 10
EVIDENCE OF ACIDIFICATION
10.1 OVERVIEW
The National Stream Survey - Phase I (NSS-I) was designed primarily to determine the
status and extent (number, length) of streams in subregions of the mid-Atlantic and southeastern
United States that are presently acidic or that may be at risk as a result of acidic deposition.
Other important issues about the effects of acidic deposition, however, include the extent to
which the chemical status of streams has already been altered by acidic deposition and the
relationship between current deposition rates and surface water quality among the subregions.
The only direct way to conclusively demonstrate surface water acidification is through
historical data on water quality. Because of a lack of such data for the target population of
interest, a second approach for assessing whether surface waters have been acidified involves
indirect methods using chemical models to infer changes in water quality. In this section, we
use this approach to assess whether streams in the various subregions may have been acidified.
In addition, we have examined the relationship between sulfate deposition and surface water
chemistry in streams in the different subregions to assess whether observations are consistent
with a hypothesis that atmospheric deposition has influenced stream water chemistry. It is
important to note that the results presented (e.g., ANC deficits and relationships between
surface water sulfate and deposition rates) are not, in themselves, conclusive evidence for
acidification. They are to be compared with other information that may also support or refute a
hypothesis of surface water acidification by atmospheric deposition.
10.2 ION BALANCE AND ANC RELATIONSHIPS
Henriksen (1979, 1980, 1982) has argued that the acidification of surface waters by acidic
deposition can be described as a continuous process analogous to the titration of a bicarbonate
solution with sulfuric acid. The bicarbonate is produced through the weathering of primary
minerals in the watershed by carbonic acid generated in soil respiration, and the atmosphere is a
source of the acid through precipitation and dry deposition.
Except where the ANC has been completely titrated (ANC < 0), a determination of whether
streams have been acidified (decline in ANC due to titration) is constrained by the lack of
historical information on ANC under pristine conditions (low deposition rates of strong mineral
acids). Aimer et al. (1978), Dickson (1980), and Henriksen (1982), however, have shown that
oligotrophic surface waters in areas with low acidic deposition rates, and with no geological
source of sulfate contain approximately equivalent amounts of alkalinity and nonmarine calcium
plus magnesium. This one-to-one ratio of bicarbonate alkalinity to nonmarine divalent base
cations is a result of the fact that calcium, magnesium, and bicarbonate typically make up most
of the total ion content in bicarbonate dominated waters. It also implies that approximately one
equivalent of divalent base cation is released to solution for every equivalent of alkalinity
produced in the process of carbonic acid weathering.
Based on this relationship, Henriksen (1979, 1980) argued that acidification of surface
waters due to inputs of strong mineral acids could be estimated by examining the relationship
between the nonmarine calcium plus magnesium and the alkalinity or ANC. In areas where
sulfate was the major acid anion in precipitation and where little sulfur was supplied by
weathering, Henriksen and others have shown that the difference (in equivalents) between the
309
-------
divalent base cations and alkalinity was directly proportional to the nonmarine sulfate con-
centration. The so-called ANC deficit, defined as the difference between the nonmarine base
cation equivalents and the alkalinity or ANC, was assumed to provide a quantititive measure of
acidification. This assertion, however, was based on the assumption that the ANC deficit is
caused entirely by a titration of the ANC and that the base cation concentration has remained
constant over time. Although this assumption may be nearly correct in some waters (e.g., in
poorly buffered streams of low ionic strength), there are indications that base cation concen-
trations in some streams may not remain constant over time (Reuss and Johnson, 1986; Cresser
and Edwards, 1987).
As pointed out by Kramer and Tessier (1982) and Reuss et al. (1986), these relationships
among base cations, sulfate, and ANC support the Henricksen model only in the sense that they
demonstrate the validity of the charge balance principle; that is, the present ion composition
would occur regardless of whether surface waters reached their present chemical condition
because of titration by a strong mineral acid, increased leaching of base cations, or some
combination of both. Reuss et al. (1986) concluded that it did not seem possible using this
approach to assess whether the ANC deficit was due to acidification (reduced ANC) or increased
base cation export.
To separate the effect of titration from that of increased base cation leaching on the ANC
deficit, Henriksen (1982) introduced the so-called F-factor, which is defined as the fraction of
nonmarine sulfate that is accompanied by change in base cations, leaving ANC unaffected. The
remainder of the sulfate presumably goes to titrate the ANC. F is thus defined as:
A [sum of base cations!
_ ^ .j
A [S042-]
Thus, when F - 0, all the sulfate goes to titrate the alkalinity, and when F = 1, all the sulfate
goes to increase the export of base cations through increased weathering and cation exchange;
the ANC remains constant.
The relationship between nonmarine base cations [Ca2+ + Mg2+ + Na+ + K+] and ANC was
examined for each of the NSS-I subregions to determine if there was evidence of acidification or
increased base cation leaching. An F-factor of zero was applied, so that the deficits represent
the maximum change in ANC resulting from a hypothesized strong acid titration. As Figure 10-1
shows, all the subregions have some streams with an ANC deficit (ANC less than the sum of
nonmarine base cations), as demonstrated by the data points plotting below the line, implying
stream acidification (natural or anthropogenic), increased base cation export, the presence of
neutral salts (e.g., NaCl, CaSO4), or some combination of these. Base cations have been reduced
by an amount proportional to their chloride ratio in sea water, based on their stream water
[Cl~], in an attempt to adjust for neutral salt additions from sea salt deposition or road salting.
Subregions in which streams have the largest ANC deficit are the Northern Appalachians
(2Cn), Poconos/Catskills (ID), Mid-Atlantic Coastal Plain (3B), and the Valley and Ridge (2Bn).
Almost all the streams in the target population in these subregions plot below the one-to-one
line, indicating that an ANC deficit exists in almost the entire stream population of interest in
these subregions (Figure 10-1).
In contrast, most of the streams in the Southern Blue Ridge (2As), Ozarks/Ouachitas (2D),
Piedmont (3A)} Southern Appalachians (2X), and Florida (3C) fall on or closer to the line of
equality, indicating a lower ANC deficit, and therefore, either less stream acidification or lower
increases in base cation export in these subregions. The lower ANC deficits in streams in these
subregions in the southeastern United States is generally consistent with the greater sulfate
310
-------
200
POCONOS/CATSKIULS
too
«0
0
NORTHERN APPALACHIANS
no «o w m o no «t
cr
o
O
<
a
<
o
a
N
DC
1U
Z
a
§
MD-ATLANTIC
COASTAL PLAIN
100
200
SOUTHERN BLUE RIDGE
si us o
too
SOUTHERN APPALACHIANS/
«0 KS
o no m
m
VALLEY AND RIDGE
«o
209
PIEDMONT
m
m
FLORIDA
I Hi IBB
it! NO
SUM OF NON-MARINE BASE CATIONS ftieq L ~1)
Figure 10-1. ANC of stream water vs. the nonmarine base cation concentration in stream
water in NSS-I subregions.
311
-------
adsorption capacity of soils in the southeastern United States. Because the acidic anion must be
mobile in soils for surface waters to be acidified by acidic deposition or for sulfate to cause an
increase in base cation leaching, adsorption of atmospherically deposited sulfate by soils effec-
tively precludes these processes from occurring, even though the streams may have a low ANC.
Because a sea salt correction was used in estimating the nonmarine component of base
cations in stream water (by assuming that all of the chloride present is from sea salt deposi-
tion), there is some uncertainty associated with the estimated contribution of weathering to base
cations in stream water. If some of the chloride present is from internal (i.e., watershed)
sources, the contribution of weathering will be underestimated. However, this portion of base
cations is associated neither with bicarbonate nor with an acid source of anions, so the deficit
generated by including these base cations would have overestimated the ANC deficit potentially
reflecting stream acidification. A conservative approach for sea salt correction was used in this
analysis in order not to overestimate the ANC deficit. The chloride-distance envelope approach
for sea salt correction (used in subsections 6.3.5, Figure 6-18) was not used because it took the
perspective of not overestimating the sea salt correction factors (minimum estimate of sea salt
correction).
As previously stated, an ANC deficit alone does not prove that surface water acidification
has occurred. It does, however, suggest either that some of the ANC has been titrated or that
the export of base cations has increased without a corresponding increase in the ANC. If the
latter occurs, it reflects soil acidification resulting from the exchange of protons for strong base
cations in the soil, a source of neutral salts (e.g., CaSO4) not accounted for by sea salt correc-
tion, and/or increased weathering of base cations.
The exact causes of anion deficits are unknown, but they include natural or anthropogenic
sources of mineral acids that increase the leaching of base cations or depress the ANC, strong
organic acids which have the same effect, or some combination of both. To determine whether
sulfate alone could account for the apparent ANC deficit in the various subregions, ANC plus
nonmarine sulfate was regressed against the sum of nonmarine base cations (Figure 10-2). This
figure shows that with the exception of the Mid-Atlantic Coastal Plain (3B) and Florida (3C),
streams in most of the subregions plotted on or near to the one-to-one line when sulfate was
added to the ANC. This indicates that sulfate is the major anion that contributes to the ANC
deficit in these subregions. To assess whether other mineral acid anions and organic acids could
account for the calculated ANC deficit of streams in each subregion, the deficit was regressed
against sulfate, nitrate, chloride, and DOC, the latter variable serving as a surrogate of strong
organic acid anions. Only SO42" was corrected for sea salt inputs.
Results of these step-wise linear regressions are shown in Table 10-1. The high partial
coefficient of determination (r2) for sulfate in most subregions shows that sulfate alone can
account for most of the variation in the anion deficit among streams in these subregions. In
the Mid-Atlantic region, for example, sulfate accounts for > 75% of the variation in the ANC
deficit in almost all of the subregions. The only exception is the Mid-Atlantic Coastal Plain
(3B) where nitrate is associated with more of the variation in the ANC deficit than is sulfate.
This result is consistent with results of the standardized regression analysis showing that nitrate
was associated with a portion of the variation in ANC among streams in the Mid-Atlantic
Coastal Plain. Whether the effect of nitrate on the ANC deficits in this subregion are due to
nitric acid inputs from the atmosphere or to inputs of nitrate as either a neutral salt or
ammonia that was applied as fertilizer and was subsequently nitrified is unknown.
312
-------
NORTHERN APPALACHIANS
VALLEY AND RIDGE
POCONOS/CATSKLLS
HO an i an «s w «M
M K9 0 81
SOUTHERN BLUE RDGE
(SUMMER)
IN Ml HO SO I
MMO mWHOEOOl 2MWRIHO
SUM OF NON-MARINE BASE CATIONS (peq L ~1)
Figure 10-2. ANC plus SO42" concentration versus the nonmarine base cation concentration in
stream water in NSS-I subregions.
313
-------
In the Southeast subregions, sulfate also explains a large percentage of the variation in the
ANC deficit among streams. The only exceptions are at the upper nodes of reaches in the Pied-
mont (3A) and in Florida (3C). In both of these subregions, sulfate alone explains < 15% of the
variation in the ANC deficit. The results for Florida are consistent with the results of the
standardized regression analysis, which showed that dissolved organic carbon was associated with
a major portion of the variation in ANC among the upper reach nodes in this subregion. In the
upper reach nodes of the Piedmont (3A), sulfate and nitrate are the only two anions that are
associated with a significant fraction of the variation in the ANC deficit. However, together
they account for only 20% of the variation. The reason for the large residual error in the
regression model is unknown but may be due to analytical errors in the measurement of ANC,
base cations, and sulfate. Because the ANC deficit in most of the streams in this subregion is
relatively small, analytical errors could account for the large residual error in the regression
model.
In summary, analysis from the charge balance/ANC deficit model show that many streams in
the target population have sea water corrected base cation concentrations greater than their
ANC (i.e., an ANC deficit). Such deficits can indicate acidification of stream water, increased
leaching of base cations from the surrounding watershed due to inputs of strong acids, or
sources of strong mineral acid anions in association with base cations as neutral salts. Sulfate
alone explains most (> 50%) of the variation in the ANC deficit among streams in all of the sub-
regions except at the upper reach nodes in subregions 3C and 3A and at the upper and lower
reach nodes in subregion 3B. In Florida (3C), organic acids appear to account for most (81%) of
the variation in the deficit among streams, whereas in the Mid-Atlantic Coastal Plain (3B),
nitrate accounts for most of the deficit (70 and 73% at the lower and upper nodes, respectively).
Additional analysis of these data is warranted to refine our quantitative understanding of
potential watershed contributions of neutral salts and the effect of various acid anions on the
ANC deficit when different F-factors are assumed.
10.3 ATMOSPHERIC SULFATE DEPOSITION AND STREAM WATER CHEMISTRY
To examine the relationship between stream water sulfate concentration and sulfate
deposition in streams potentially sensitive to acid deposition, a representative sulfate
concentration was calculated for each reach (average of upper and lower node index value) after
deleting nodes that had either ANC > 200 /zeq L"1 or SO42' > 400 /zeq L"1, or both. Reaches
with the highest sulfate concentrations were found in the areas of highest wet sulfate deposition
(Figure 10-3). A large fraction of the sample streams that had wet sulfate deposition rates
> 3 g m"2 yr'1 had stream water sulfate concentrations > 200 /zeq L'1. With few exceptions,
the rest of the sample streams in this area had sulfate concentrations > 100 /zeq L"1. In areas
that had deposition rates < 2.5 g m"2 yr'1, most of the streams had sulfate concentrations < 100
jueq L"1. The streams in this area with sulfate concentrations > 100 /zeq L"1 were mainly found
near the coast, where they may have been influenced by agriculture, sulfate from groundwater,
or weathering of old marine deposits. To further reduce the possible influence of watershed
sulfate sources on this analysis, we confined the analysis to upstream nodes. There was a
strong positive relationship (r = +0.88) between the median upstream reach node sulfate con-
centration and the median annual average total (wet and dry) sulfate deposition from 1980 to
1984 (deposition rates taken from Wampler and Olsen, 1987, and Dennis and Seilkop, 1987) in
each subregion (Figure 10-4). Although a large range of stream water sulfate concentrations
was observed within most subregions, the regional pattern observed in subregion medians is
314
-------
Table 10-1. Fraction (r2) of the Variation in the Estimated ANC Deficit in Stream Water that
Can be Accounted for by Sulfate, Nitrate, Chloride, and DOC, and the Partial
Regression Coefficient (b) for Each Relationship
Subregion (Node)
Pocono/Catskills (L)
Pocono/Catskills (U)
N. Appalachians (L)
N. Appalachians (U)
Valley and Ridge (L)
Valley and Ridge (U)
Coastal Plain (L)
Coastal Plain (U)
Southern Blue R. (L)
Southern Blue R. (U)
Piedmont (L)
Piedmont (U)
S. Appalachians (L)
S. Appalachians (U)
Ozarks/Ouachitas (L)
Ozarks/Ouachitas (U)
Florida (L)
Florida (U)
so42-
r2 b
0.81
0.77
0.99
0.84
0.84
0.86
0.23
0.21
0.94
0.53
0.84
0.15
0.99
0.98
0.97
0.76
0.61
0.10
1.12
1.19
1.02
1.09
1.06
1.40
1.34
1.04
0.97
0.84
0.98
0.69
1.02
1.05
1.00
0.92
1.14
0.56
N03-
r2 b
0.13
0.19
0.002
0.005
0.14
0.12
0.70
0.73
0.03
0.16
0.04
0.05
0.003
0.012
0.007
0.15
0.02
1.08
1.15
1.36
0.90
0.94
0.81
1.34
1.19
0.62
0.78
0.97
0.71
1.19
1.13
0.92
0.97
0.39
ns
cr
r2 b
ns*
ns
0.003 1.02
ns
0.001 0.04
ns
0.003 0.05
ns
0.002 -0.04
0.11 -0.18
ns
ns
ns
ns
0.002 -0.08
0.007 -0.15
0.27 0.81
0.01 0.24
Overall
DOC Model$
r2 b R2
0.005 11.05
ns
ns
ns
ns
0.001 -6.82
0.03 6.59
ns
ns
ns
ns
ns
ns
0.001 7.65
ns
0.04 5.23
0.02 3.05
0.81 4.42
0.95
0.96
>0.99
0.99
0.98
0.99
0.96
0.94
0.97
0.79
0.88
0.20
0.99
>0.99
0.98
0.95
0.93
0.93
* ns = partial regression coefficient (b) not significantly different from zero (P > 0.15).
* R2 is the coefficient of multiple determination when all four independent variables are
included in the regression model.
315
-------
Slreamwater Sulfate with Atmospheric Sulfate let Deposition
(Average of NSS-I Upstream and Downstream Sample Nodes)
2.30
a
A
Line of Approiiraale]j Equal
Sulfile Ion Deposition (g m"2?:"1)
200 < 804 S 400 jieq L"1
100 < S04 & 200 jieo, L-1
50 < S04 £ 100 jieq L"1
S04 s 50 peq L"1
Figure 10-3.
Map of reach average sulfate concentration (/jeq L'1) in streams after excluding
stream nodes with ANC > 200 /zeq L'1 and/or SO42' > 400 jueq L'1 in the east-
ern United States overlayed with isopleths of the 1980-1984 annual average wet
sulfate deposition. Sulfate wet deposition rates taken from Wampler and Olsen,
(1987).
316
-------
NSS-I UPPER NODE
cr
-------
consistent with the hypothesis that atmospheric sulfate deposition is an important factor
controlling sulfate concentration in the subpopulation of low ANC streams.
Sullivan et al. (1988) reported that lake water sulfate concentrations, when expressed as
subregion median values, were highly correlated with median estimated deposition rates in a
probability sample of eastern and western lakes from the National Lake Survey (Linthurst et al.,
1986; Landers et al., 1987). Sullivan and his colleagues observed that lakes in the Southern Blue
Ridge had lower sulfate concentrations than expected, supporting observations by Galloway et al.
(1983) and Rochelle and Church (1987) that sulfur retention (principally sulfate adsorption in
soils) is high in many southeastern watersheds. To compare stream results with those of
Sullivan et al. (1988), we have combined NSS-I data with data from the National Lake Survey
(Linthurst et al., 1986; Landers et al., 1987) to produce a plot of median surface water sulfate
versus wet sulfate deposition in low ANC regions of the United States (Figure 10-4). Streams in
the Mid-Atlantic subregions (ID, 2Cn, 2Bn, 3B) and the Ozarks/Ouachitas (2D) fall closely in
line with eastern and upper midwest lakes in a positive, approximately linear trend. As also
observed by Sullivan et al. (1988) for Southern Blue Ridge lakes, Figure 10-5 shows that streams
in this area had lower than expected SO42" in relation to deposition. In addition, streams in
the Piedmont (3A) and Southern Appalachian (2X) subregions also fell below those predicted by
the roughly linear trend. These observations further support those of Galloway et al. (1983) and
Rochelle and Church (1987) regarding sulfur retention in southeastern watersheds.
Florida streams (NSS-I Subregion 3C) had median sulfate concentrations considerably lower
than those reported for Florida lakes (NLS-I Subregion 3B). Linthurst et al. (1986) observed a
large amount of heterogeneity in Florida lakes, and reported that extremely high sulfate con-
centrations in some Florida lakes appear to be associated with an influx of sulfur-rich ground-
water from the Floridan Aquifer. These lakes are typically found in parts of the central Florida
peninsula not contained within the NSS-I Florida Subregion 3C (Eilers, pers. comm. 1987).
There was no obvious relationship between median stream water ANC and the median esti-
mated sulfate deposition rates for the nine NSS-I subregions. This lack of a clear relationship
was observed also for eastern and western lakes by Sullivan et al. (1988) and supports con-
clusions given in Section 8 that most of the regional variation in stream water ANC is probably
due to pre-existing differences in base cation supply from weathering sources.
To further examine the relationship between atmospheric sulfate deposition and stream
water acidity, the locations of low ANC (< 200 peq L"1) upstream reach nodes were plotted on a
map of the eastern United States and then overlayed with isopleths of atomospheric wet sulfate
deposition rates (Figure 10-6). Sample sites affected by acid mine drainage, substantial
watershed sources of sulfate, and organic anion dominance were excluded from the plot. Most
of the acidic and low ANC upper reach nodes observed in the NSS-I were in areas of high
sulfate deposition (> 3.0 g m"2 yr'1). Very few of the remaining acidic stream reaches were
observed in areas with deposition rates less than 2.5 g m~2 yr"1.
It would be a mistake, however, on the basis of these findings, to interpret 2.5 g m"2 yr"1
as an acceptable target loading simply because of the absence of inorganic, acidic streams.
Many of the NSS-I areas with deposition rates below 2.5 g m"2 yr"1 are known to be areas
where deep, well-weathered soils are highly retentive of sulfate (Galloway et al., 1983; Rochelle
and Church, 1987). In such areas, stream water sulfate and ANC would not be expected to be
in steady state with regard to acid deposition. Although more common in the Mid-Atlantic
(where there is higher deposition), low and very low ANC sites were also observed in the
Interior Southeast.
Causal inferences cannot be made based on these survey data alone. Whereas the data are
consistent with a hypothesis of an atmospheric source of acidification, other processes could also
318
-------
CO
c
co
co
CO
<4
c
o
0)
V.
-Q
a
«0
-J
%
o
LxJ
es
LU
—i
LU
i£ — .
-o
O LULU
— >ce
i— or CD
:D
z toco
o
to
z
o
0
LU
ce
CO
^
to
1
to
to
2
.
O
m
CM
-t-j-T
O
o
CM
SO
so
30
$0
so
o
o
CD
LO
i i
CD
CM
I
- CM
- o
d ^2 fi
« fl O
'•S .1 3
a> £ co
S ca o
^^ cx
* a-8
03
- 5
2J o1 «a
o ^"i
o
_o
'5b
«sfl
"•* «
S S
03 eS
Q a
t= •§ -s
oo
o
Q_
LU
O
LU
ID
00
LU
c
f s
o5 o
O ni
S
I
O
O
§
2
3
oo
I —
barf) 3-LVJinS
30VJdHS
319
-------
Streamirater ANC with Atmospheric Sulfate let Deposition
(NSS-1 Upstream" Nodes)
2.08 Line of Approiimatell Equal
SuUale loa Deposition (| m-Z
• ANC S 0 neq I"1
D 0 < ANC S 56 ]ieq IT1
A 50 < ANC * 100 peq L"1
* 100 < ANC s 200 |ieq L^
Figure 10-6. Map of upstream node spring index ANC concentration (pieq L'1) in streams with
ANC < 200 /zeq L"1 in the eastern United States overlayed with isopleths of the
1980 to 1984 annual average wet sulfate deposition (g m"2 yr"1). All streams
classified in the acid mine drainage, substantial watershed sources of sulfate, or
organic-dominated groups were excluded from the map. Sulfate wet deposition
rates taken from Wampler and Olsen (1987).
320
-------
cause the observed patterns. For example, the same pattern in surface water sulfate could be
produced by regional differences in export of sulfate derived from geologic sources in the
absence of sulfate deposition. This explanation is unlikely in data that exclude streams and
lakes with [SO42~] greater than 400 /zeq L"1. Furthermore, one would have to evoke unsupport-
able assumptions of high deposition sulfate retention in northeastern and mid-Atlantic watersheds
to account for the sulfate received by these sites in atmospheric deposition. An alternative
explanation is that the observed sulfate pattern results entirely from differences in sulfate
retention by watersheds. This explanation, however, is also consistent with a hypothesis of
regional surface water or watershed soil acidification. The geographic patterns in sulfur
retention and their relationship to watershed characteristics is an important part of ongoing EPA
studies in the Direct/Delayed Response Project (DDRP)
321
-------
-------
SECTION 11
SUMMARY AND CONCLUSIONS
11.1 BACKGROUND
National Stream Survey - Phase I (NSS-I) field activities were conducted in the mid-
Atlantic and southeastern United States in the spring of 1986 by the U.S. Environmental
Protection Agency (EPA) as part of the National Surface Water Survey (NSWS). The first phase
of the NSWS was designed to determine the present chemical status of surface waters in regions
of the United States containing a majority of streams and lakes considered to be at risk as a
result of acidic deposition. The NSS-I was conducted as part of the National Acid Precipitation
Assessment Program (NAPAP). Like the previous EPA NSWS activities (Eastern and Western
Lake Surveys), it contributes directly to one of NAPAP's principal objectives: the quantification
of the extent, location, and characteristics of acidic and potentially sensitive streams and lakes
in the United States.
The NSS-I was conducted in four Mid-Atlantic and five Southeast subregions of the United
States, identified on the basis of similar physiographic characteristics:
• Mid-Atlantic (MA) Region
- Interior Mid-Atlantic (IMA) Subregions
— Poconos/Catskills (ID)
— Northern Appalachians (2Cn)
— Valley and Ridge (2Bn)
- Mid-Atlantic Coastal Plain Subregion (3B)
• Southeast (SE) Region
- Interior Southeast (ISE) Subregions
— Southern Blue Ridge (2As) - Pilot Survey in 1985
~ Piedmont (3A)
~ Southern Appalachians (2X)
— Ozarks/Ouachitas (2D)
- Florida Subregion (3C)
NSS-I field activities to date have not included areas of the Northeast, Upper Midwest, and
West. Though these regions are expected to contain low ANC or acidic streams potentially
sensitive to acidic deposition, they also contain numerous lakes that were sampled as part of
EPA's Eastern and Western Lake Surveys. Furthermore, NSS-I field activities thus far have not
included synoptic stream chemistry sampling in parts of the South Atlantic and Gulf Coastal
Plains expected to contain predominantly low ANC surface waters—but where deposition rates
are comparatively lower than in most of the Survey area, and where organic acidity is expected
to play an important role. Field activities in the Florida Subregion test the utility of NSS-I
logistical and design protocols in lowland stream networks of the Southeastern Coastal Plain.
323
-------
11.2 OBJECTIVES
The objectives of the NSS-I in the Mid-Atlantic and Southeast were to:
• Determine the percentage, extent (number, length, and drainage area), location, and
chemical characteristics of streams in the Mid-Atlantic and Southeast regions that are
presently acidic, or that have low acid neutralizing capacity (ANC) and thus might
become acidic in the future.
• Identify streams representative of important classes in each region that might be
selected for more intensive study or long-term monitoring.
11.3 DESIGN
Within the NSS-I subregions, the stream resource of interest (the "target" resource) was
identified as those streams that have drainage areas less than 155 square kilometers (60 square
miles), but that are large enough to be represented as blue lines on l:250,000-scale U.S. Geo-
logical Survey (USGS) topographic maps.
The NSS-I sampled stream reaches defined as segments of the stream network, as repre-
sented by blue lines on the l:250,000-scale maps. These segments, or reaches, were identified as
the mapped blue-line segments between two tributary confluences. Sampling points on each of
these reaches were just above the downstream point of confluence (lower node) and just below
the upstream point of confluence (upper node). The upper node of a reach represented as a
headwater was defined as the farthest upstream extent of the mapped blue line representation.
A two-stage sampling procedure was used to obtain a randomized, systematic sample of
approximately 500 reaches with good spatial distribution over each of the nine NSS-I subregions
(50 to 80 reaches per subregion). Reaches were excluded if they were too large (drainage area
> 155 km2) or were located within metropolitan areas. The target population was further
refined as a result of field visits and data analysis to exclude tidal influences and such
perturberances as acid mine drainage.
The NSS-I used index values based on one or the average of two or three baseflow
measurements to describe the chemical status of each stream sampled. This index value was
measured during the spring season between snowmelt and leaf out (approximately March 15 to
May 15) and excluded storm episodes. The choice of the spring index sampling period involved a
trade-off between minimizing within-season and episodic chemical variability and maximizing
the probability of sampling chemical conditions potentially limiting for aquatic organisms.
11.4 SELECTED RESULTS
11.4.1 Regional Chemical Characteristics
Proportions of the target stream reach population with spring index ANC < 200 peq L"1
were similar in the Mid-Atlantic (MA) and Southeast (SE) regions surveyed by the NSS-I. Many
published works have used 200 peq L"1 ANC as a reference level below which surface waters
may be considered potentially sensitive to acidification. Excluding Florida, 51% (18,542) MA
reaches and 52% (9,642) of SE reaches had ANC < 200 peq L'1 at their upstream ends. After
flowing an average of 3 km to their downstream ends, an estimated 41% (15,060) of MA reaches
and 47% (8,682) of SE reaches were at or below this index value.
324
-------
In the Mid-Atlantic Region, 7.4% (2,677) of the reaches were acidic (ANC < 0 /zeq L"1) at
their upstream ends and 3% (1,098) were acidic at their downstream ends. Estimates based on
the first NSS-I sample visit alone (rather than the average of the two spring visits) were
slightly higher because of a seasonal trend of increasing spring basef low ANC. Based upon the
earlier spring sample, 8.9% (3,123) of the reaches were acidic at their upper ends and 3.8%
(1,339) were acidic at their downstream ends. These figures do not include reaches acidic due
to acid mine drainage.
In the Interior Southeast (ISE) subregions (this excludes Florida), we estimate that less
than 1% (120) of target reaches were acidic at either their upstream or downstream end. As in
the Mid-Atlantic, these estimates do not include reaches acidic due to acid mine drainage.
In the Mid-Atlantic Region, an estimated 13% (4,770) of the reaches had spring basef low pH
< 5.5 at their upstream ends and 5.4% (1,991) had pH at or below this level at their downstream
ends. An estimated 26% (9,406) of the reaches had pH < 6.0 at their upstream nodes; 10%
(3,760) had pH below this reference value at their lower nodes. Again, these figures do not
include reaches acidic because of acid mine drainage.
In the portions of the Interior Southeast Region surveyed by the NSS-I (this excludes
Florida), 2% (347) of the reaches had pH < 5.5 at their upstream ends and less than 1% had pH
at or below this value at their downstream ends. An estimated 8.6% (1,605) had pH below a
reference level of 6.0 at their upper nodes; 2.7% (512) had index pH levels less than this value
at their lower nodes.
Among the subregions of the Mid-Atlantic Region, the NSS-I found the greatest number of
acidic stream reaches in the Mid-Atlantic Coastal Plain, where 12% of the target reaches (1,334)
were acidic or had pH < 5.0 at their upper ends. Another 12% of these upper reach ends had
pH between 5.0 and 5.5. After flowing typically 2 to 3 km to their downstream ends, 7% (772)
of the reaches remained acidic (ANC < 0) and had pH < 5.5. In the other Mid-Atlantic sub-
regions (Poconos/Catskills, Valley and Ridge, and Northern Appalachians), relatively few reaches
were acidic at their downstream ends. Of these three subregions, only the Northern Appala-
chians were estimated to have reaches acidic at their downstream nodes (4%, or 326 reaches).
However, between 5% and 6% (209, 636, and 499 reaches, respectively) were acidic at their
upstream nodes in the Pocono/Catskill, Valley and Ridge, and Northern Appalachian subregions.
Among the Interior Southeast subregions (this excludes Florida), acidic reaches were
observed only in the Southern Appalachians, where an estimated 120 reaches (2%) were acidic at
their upstream ends. Though acidic stream reaches were rare in the Southeast, two Southeast
subregions, the Southern Blue Ridge and the Ozarks/Ouachitas, had among the highest propor-
tions of streams with relatively low ANC. In the Southern Blue Ridge, 84% (1,703) of target
reaches had ANC < 200 /zeq L'1 at their upstream ends; in the Ozarks/Ouachitas, the proportion
was 68% (2,850).
The Florida subregion stands out as a geographic area with a relatively high percentage of
acidic, low ANC, and low pH streams. It is also an area where there are many highly colored
streams with high concentrations of dissolved organic carbon (DOC). Population percentage
estimates for the Florida subregion are not strictly comparable with those from other subregions,
because the Florida sample was drawn from a more restrictive target population focused only on
the portion of this state with expected ANC < 200 #eq L"1, rather than 400 jueq L'1 (as was the
criterion for all the other subregions). At the upstream ends, however, an estimated 678 (39%)
of the Florida reaches were acidic (ANC < 0 //eq L"1); another 531 (31%) had ANC between 0
and 50 /^eq L"1. Smaller numbers of reaches were still acidic at their downstream ends (225, or
14%), but there remained a large number (670, or 43%) with downstream ANC betweem 0 and 50
/zeq L"1. Whereas the acidic reaches in the Florida peninsula were primarily high DOC, colored
325
-------
water systems, some of those in the panhandle were clear or slightly colored acidic streams with
very low ionic strength and very low sulfate, nitrate, and DOC concentrations. As expected, pH
distributions were similar to those for ANC. At the upstream ends, an estimated 539 Florida
reaches (31%) had pH < 5.0; another 324 (19%) were between pH 5.0 and 5.5. Index pH was
generally higher at the downstream ends of reaches, where an estimated 225 reaches (14%) had
pH ^ 5.0, and another 146 (9%) were between pH 5.0 and 5.5.
Stream water sulfate concentrations were markedly higher in the four Mid- Atlantic sub-
regions (medians from 125 to 238 /zeq L'1) than in the five Southeast subregions (medians from
10 to 71 0eq L"1).
11.4.2 Chemical Relationships
Standardized multiple regression analyses showed that regional variation in ANC among
streams within NSS-I subregions was associated more closely with differences in base cation
concentrations than with differences in concentrations of mineral acid or organic acid anions.
This result suggests that watershed geochemical and hydrologic characteristics controlling the
supply of mineral weathering products are very important in determining the regional patterns in
stream water ANC and therefore buffering of acid inputs.
Examination of the difference between air-equilibrated and non-equilibrated pH measure-
ments indicates that almost all NSS-I sample streams with pH > 6.0 were super-saturated with
CO2. The pH increases upon air equilibration (excess CO2 is degassed) ranged from approxi-
mately 0.25 units to one full pH unit, indicating that carbonic acid alone has an important effect
on the pH of stream waters with pH > 6.0. In all NSS-I subregions except Florida and the
upper reach nodes in the Mid-Atlantic Coastal Plain, this is the majority of streams.
The only direct way to conclusively demonstrate surface water acidification is through
historic data on water quality. Because of a lack of such data on a regional scale for the
streams of interest in the NSS-I, a second approach for assessing whether streams in the Mid-
Atlantic and Southeast have been acidified involves an indirect method of inferring past changes
in ANC through an examination of present water chemistry. Previous researchers have used the
ANC deficit (nonmarine base cation concentration minus ANC) as a rough measure of surface
water acidification resulting from either natural or anthropogenic causes. It is important to
recognize that ANC deficits, in themselves, reflect simply a presence of strong acid anions in
water at the time of sampling. ANC deficits can be observed in stream water for a number of
reasons: (1) they may reflect a reduction in ANC caused by the addition of a strong acid (like
sulfuric acid) to the stream, as might occur with acid deposition onto a poorly buffered water-
shed; (2) they may reflect an increase in base cation concentration (with no change in ANC) in
the stream resulting from cation exchange or accelerated weathering of watershed soils and rock
(this can be due to acidic deposition); (3) they may result from an addition of base cations in
neutral salts (e.g., CaSO4), with no change in ANC; or finally (4) they may result from a com-
bination of all these mechanisms. The last combination of factors is the most likely for a
heterogeneous assemblage of streams subject to a wide range of influences. However, the inter-
pretation of ANC deficits in relatively pristine streams draining watersheds underlain by meta-
morphic rock is not as uncertain. In these pristine streams, ANC deficits were higher in areas
that receive larger amounts of acidic deposition. The lack of plausible sources of neutral salts
in these streams makes it likely that the observed ANC deficits in such upland, forested streams
are due to titration by strong acids or increases in base cation export, and that both
mechanisms are probably associated with atmospheric acidic deposition.
326
-------
In order to interpret the observed ANC deficit as a measure of the amount of historic
change in ANC, one must first assume that the ANC deficits are due largely to titration by
strong acids and that the base cation concentrations in streams have remained nearly constant
over time. These assumptions are more likely to be correct in poorly buffered stream waters of
low ionic strength. There are indications, however, that base cations may not remain constant
over time in many streams. In addition, several conditions must also be met before the ANC
deficit in a given stream could be interpreted as a measure of historic change in ANC due to
acid deposition: (1) an atmospheric source of strong acids can be demonstrated to be of suf-
ficient magnitude to cause the observed ANC deficit, (2) other potential sources of strong acid
anions can be ruled out as unlikely (e.g., sulfide weathering, organic acidity), and (3) concen-
trations of acids not derived from atmospheric deposition (e.g., organics from watershed sources)
have remained relatively constant over time.
Most of the NSS-I subregions have some streams with a deficit of ANC in relation to non-
marine base cations. The four Mid-Atlantic subregions (Poconos/Catskills, Northern Appala-
chians, Valley and Ridge, and Mid-Atlantic Coastal Plain) had the largest number of streams
with a deficit of ANC in relation to base cations. However, some streams in all the NSS-I
subregions show a deficit.
The observation of smaller ANC deficits in most Southeast streams, compared with those in
Mid-Atlantic subregions, is generally consistent with the lower atmospheric acid deposition rates
and greater sulfate adsorption capacities of soils in many parts of the southeastern United
States. These results do not preclude the possibility of soil acidification leading to future
stream acidification in any of the NSS-I subregions. Sulfate was the dominant strong acid anion
in most acidic and low ANC streams and was in sufficient concentration to account for the
observed ANC deficits in the majority of stream reaches in all subregions except Florida and the
Mid-Atlantic Coastal Plain.
11.4.3 Sources of Acidity in Acidic and Low ANC streams
Potential sources of acidity were examined in a subpopulation of acidic and low ANC (< 200
/zeq L'1) streams within the NSS-I target population. Stream reaches affected by acid mine
drainage, substantial watershed sources of sulfate, and organic acids were identified and
excluded from this group by site visits and examination of stream chemical data. After the
exclusions, a group of acidic (ANC < 0) and low ANC (0 - 200 /zeq L'1) stream reaches remain,
for which sulfate is the dominant strong acid anion. Acid deposition cannot be ruled out as the
major source of acidity in this group of streams. Stream water sulfate concentrations can
readily be accounted for by evaporative concentration of sulfate in wet plus dry deposition.
Within the subpopulation of acidic streams in the NSS-I target population, a group of acidic
stream reaches has been identified for which atmospheric deposition is the most probable source
of acidity and effects due to acid deposition cannot be ruled out. Based on interpolations
between upper and lower node chemistry, this high-interest subpopulation within the group of
acidic reaches had an estimated combined length of 4,455 km. Most (4,272 km) of these reaches
are in the Mid-Atlantic subregions, where they comprise 4% of the total length of target stream
reaches. Estimates based upon the first NSS-I sample visit alone (rather than the average of
the two spring visits) were somewhat higher than 4,455 km because of a seasonal trend of
increasing spring baseflow ANC. Based on the earlier spring sample, a combined length of 7,481
km of acidic stream reaches were within the high-interest subpopulation of streams for which
acid deposition impacts cannot be ruled out. Again, most of these reaches are in the Mid-
Atlantic subregions, where they comprise 6.6% of the total length of target stream resource.
327
-------
These reaches are located primarily in the forested upland drainages and coastal areas of the
Mid-Atlantic Region that experience high rates of atmospheric sulfate deposition. Specifically,
most stream reaches within this group are located:
• In upland forested drainages of the Allegheny Plateau, in the ridges of the Valley and
Ridge physiographic province, and in the Glaciated Plateau in northeast Pennsylvania
(included in NSS-I subregions ID, 2Bn and 2Cn). An estimated 695 (9.7%) of 7,190
reaches on ridges of the Valley and Ridge geographic area and 499 (11%) of 4,548
reaches in forested drainages of the Allegheny and Glaciated Plateaus were acidic at
their upstream nodes.
• In the New Jersey Pine Barrens (within subregion 3B). Seventy-nine percent of
upstream nodes and 93% of downstream nodes were in this high-interest subgroup;
approximately one-third of these had chemistry influenced (> 10% of the anion sum) but
not dominated by organic anions (organic anion < SO42" + NO3").
Although streams in the New Jersey Pine Barrens appear to have elevated concentrations of
sulfate derived from atmospheric deposition, historic data suggest that some Pine Barrens
streams have probably been acidic at least since the early 1900's. There is no reason, however,
to believe that most of the upland forested reaches in the Mid-Atlantic have always been acidic.
A high-interest group of streams was identified within the subpopulation of non-acidic
stream reaches with very low ANC (> 0 - < 50 peq L'1) in the NSS-I target population. For
these streams, organic acidity and substantial terrestrial sulfate sources are unlikely; the
predominant sulfate source appears to be atmospheric deposition. Based on interpolation between
upper and lower node chemistry, an estimated 15,590 km of target reach length is within this
group. Approximately two-thirds of these reaches were in the Mid-Atlantic subregions, where
they made up about 9.6% of the combined length of the target population. Calculations of these
totals based on the earlier spring sample did not differ substantially from those based on the
average from two spring visits. These very low ANC streams are located primarily on upland,
forested sites of the Allegheny Plateau, Valley and Ridge Province, Blue Ridge Mountains, and
Cumberland Plateau. In addition, a large number of these reaches were found in Coastal Plain
areas (East and West Mid-Atlantic, New Jersey Pine Barrens, East and West Gulf Coast).
A third group of high-interest, low ANC (50-200 /ieq L'1) stream reaches was identified
that contained sulfate as the dominant strong acid anion. As in the acidic and very low ANC
groups above, reaches with likely geologic or other non-atmospheric sulfate sources, and those
dominated by organic anions, were excluded. Based on upper node chemistry, an estimated
61,665 km, or 31%, of the combined length of reaches in the target population within the NSS-I
survey area were in this category. Stream reaches in this group were found in all the NSS-I
subregions and in virtually all geographic site classes. Approximately half the combined reach
length in this category was located in the Valley and Ridge, Piedmont, and Blue Ridge geo-
graphic site classes. This comprised an estimated 35,934 km of reaches in this category in the
Interior Southeast subregions (41% of the total resource). In the Interior Mid-Atlantic, 19,172
km of reaches in this category made up 28% of the target stream length.
An estimated 88% (538) of the stream reaches in a subpopulation of low elevation, swampy
reaches in the panhandle and peninsula of Florida were acidic at their upstream nodes, appar-
ently due to organic acids. In the Glaciated Plateau and Allegheny Plateau, and on ridges of
the Valley and Ridge Province, stream reaches were more commonly acidic at their upstream
ends than at their downstream ends.
328
-------
NSS-I estimates of number or length of acidic stream resources do not include an addi-
tional estimated 1,300 reaches in Pennsylvania and West Virginia (combined length 3,500 km) and
an additional 120 reaches (1,100 km in length) in the Southeast that were acidic at one or both
ends due to acid mine drainage. Within the subregions surveyed by the NSS-I, acid mine drain-
age impacts were concentrated in the Allegheny Plateau (western Pennsylvania and West Vir-
ginia). Of the estimated 9,417 stream reaches in the initial target population (before refine-
ment) in the Northern Appalachian subregion that were within the NSS-I target stream size
range, 10% (929 reaches with a combined length of 2,276 km) were acidic due to acid mine
drainage.
The distribution of sulfate concentrations in eastern United States streams with baseflow
sulfate concentrations not dominated by terrestrial sulfate sources corresponds well with sulfate
deposition rates. A plot of population median stream sulfate concentration in the nine NSS-I
subregions versus rates of sulfate deposition shows a strong positive linear relationship.
However, when data from the National Lake Survey (NLS) are added to the plot, a group of
Southeast lakes and streams appears to have lower sulfate concentrations than expected, given
the sulfate deposition rates in their respective subregions. These observations are consistent
with other research showing substantial amounts of sulfate retention, by mechanisms such as
sulfate adsorption, in watersheds of the Southeast.
329
-------
-------
SECTION 12
REFERENCES
Ahmad, M.V. 1972. Coal mining and its effect on water quality. In: B. Gate (ed). Ground
Water Pollution. Directory Press.
Allard, M., and G. Moreau. 1986. Leaf decomposition in an experimentally acidified stream
channel. Hydrobiologia. 139:109-117.
Aimer, B., W. Dickson, C. Ekstrom, and E. Hornstrom. 1978. Sulfur pollution and the aquatic
ecosystem, pp. 271-311. In: J. O. Nriagu (ed). Sulfur in the Environment, Part II,
Ecological Impacts. John Wiley and Sons, New York.
American Public Health Association (APHA), American Water Works Association, and Water
Pollution Control Federation. 1985. Standard Methods for the Examination of Water and
Wastewater. 16th ed. APHA, Washington, D.C.
American Society for Testing and Materials. 1984. Standard specification of reagent water.
D1193-77 (reapproved 1983). In: Annual Book of ASTM Standards. Vol. 11.01.
Philadelphia, Pennsylvania.
Arent, L.J., M.O. Morrison, and C.S. Soong. 1988. National Surface Water Survey: Eastern
Lake Survey - Phase II. National Stream Survey - Phase I, Laboratory Operations Report.
U.S. Environmental Protection Agency, Las Vegas, Nevada. (In preparation).
Arkansas Geological Survey. 1929. Geologic Map of Arkansas. Map (scale 1:500,000). U.S.
Geological Survey.
Atkins, P.W. 1978. Physical Chemistry. W.H. Freeman and Co., San Francisco.
Baker, J.P., and C.L. Schofield. 1982. Aluminum toxicity to fish in acidic waters. Water Air
Soil Pollut. 18:289-309.
Barnes, H.L., and S.B. Romberger. 1968. Chemical aspects of acid mine drainage. Water Pollut.
Control Fed. J. 40:371-384.
Beamish, R.J., and H.H. Harvey. 1972. Acidification of the La Cloche mountain lakes, Ontario,
and resulting fish mortalities. J. Fish. Res. Board Can. 29:1131-1143.
Beamish, R.J., W.L. Lockhart, J.C. Van Loon, and H.H. Harvey. 1975. Long-term acidification of
a lake and resulting effects on fishes. Ambio 4:98-102.
Blick, D.J., JJ. Messer, D.H. Landers, and W.S. Overton. 1987. Statistical basis for the design
and interpretation of the National Surface Water Survey, Phase I: Lakes and Streams. Lake
Reserv. Manage. 3:470-475.
331
-------
Bowman, M.L. Administrator, Acid Deposition, Maryland Dept. of Natural Resources Energy
Administration, Power Plant Research Program. Tawes State Office Building, Annapolis,
Maryland, 21401. Personal communication.
Burke, E.M., and D.C.Hillman. 1987. Syringe Holding Time Study. Appendix A in Knapp,C.M.,
C.L. Mayer, D.V. Peck, J.R. Baker, and G.J. Filbin.
Chamier, A.C. 1987. Effect of pH on microbial degradation of leaf litter in seven streams of
the English Lake District. Oecologia (Berlin) 71:490-500.
Cleaves, E.T., J. Edwards, Jr., and J.D. Glaser. 1968. Geologic Map of Maryland. Map (scale
1:250,000). Maryland Geological Survey, Baltimore, Maryland.
Cline, M.G., and R.L. Marshall. 1976. Soils of New York Landscapes. Map (scale 1:750,000).
Information Bulletin 119. New York State College of Agriculture and Life Sciences. 61 pp.
Cogbill, C.V., G.E. Likens, and T.A. Butler. 1984. Uncertainties in historical aspects of acid
precipitation: getting it straight (including comments by Hansen and Hidy). Atmos.
Environ. 18:2261-2270.
Commonwealth of Pennsylvania. 1980. Geologic Map of Pennsylvania. Map (scale 1:250,000).
Department of Environmental Resources, Bureau of Topographic and Geologic Survey.
Commonwealth of Virginia. 1963. Geologic Map of Virginia. Map (scale 1:500,000). Department
of Conservation and Economic Development, division of Mineral Resources, Charlottesville,
Virginia.
Cooper, E.L., and C.C. Wagner. 1973. The effects of acid mine drainage on fish populations.
In: Fish and food organisms in acid mine waters of Pennsylvania. EPA-R-3-73-032.
Cougan, K.A., D.W. Sutton, D.V. Peck, VJ. Miller, and J.E. Pollard. 1988. National Stream
Survey - Phase I: Quality Assurance Report. EPA/600/4-88/018. U.S. Environmental
Protection Agency, Las Vegas, Nevada.
Grosser, M., and A. Edwards. 1987. Acidification of Fresh Waters. Cambridge University Press,
London.
Cronan, C., and C.L. Schofield. 1979. Aluminum leaching response to acid precipitation:
Effects on high-elevation watersheds in the Northeast. Science. 204:304-306.
Cunningham, R.L., E.J. Ciolkosz, G.W. Petersen, R.P. Matelski, R. Pennock, Jr., R.F. Shipp, A.D.
Kuhl, I.W. Ratcliff, Jr., G.H. Lipscomb, W.N. Paechatka, and A. Schadel. 1977. Soils of
Pennsylvania. Map (scale 1:750,000). Progress Report 365. Pennsylvania State University
College of Agriculture, Agricultural Experiment Station, University Park, Pennsylvania. 24
PP.
332
-------
Dennis, R.L., and S.K. Seilkop. 1987. Interim Estimates of Annual Dry Sulfur Deposition for the
Eastern United States for the Aquatics Research Program. Internal Report. U.S.
Environmental Protection Agency, Washington, D.C.
Dick, W.A., J.V. Bonta, and F. Haghiri. 1986. Chemical quality of suspended sediment from
watersheds subjected to surface coalmining. J. Environ. Qual. 15:289-292.
Dickson, W. 1980. Properties of acidified waters, pp. 75-83. In: D. Drablos and A. Tollan
(eds). Ecological Impact of Acid Precipitation. SNSF Project.
Drablffe, D., and A. Tollan, eds. 1980. Ecological Impact of Acid Precipitation. Proceedings of
an international conference, Sandefjord, Norway, March 11-14, 1980.
Driscoll, C.T. 1984. A procedure for the fractionation of aqueous aluminum in dilute acidic
waters. Intern. J. Environ. Anal. Chem. 16:267-284.
Driscoll, C.T. 1986. A program to evaluate long-term changes in the chemistry of Adirondack
lakes. Annual Report to the Environmental Research Laboratory, Corvallis, Oregon.
Project Code: El-25. 47 pp.
Driscoll, C.T., and R.M. Newton. 1985. Chemical characteristics of acid-sensitive lakes in the
Adirondack region of New York. Environ. Sci. Technol. 19:1018-1024.
Driscoll, C.T., J.P. Baker, JJ. Bisogni, and C.L. Schofield. 1980. Effect of aluminum speciation
on fish in dilute acidified waters. Nature 284:161-164.
Driscoll, C.T., J.P. Baker, J.J. Bisogni, and C.L. Schofield. 1984. Aluminum speciation in dilute
acidified surface waters of the Adirondack region of New York State, pp. 55-75. In: O.R.
Bricker (ed). Acid Precipitation: Geological Aspects. Butterworth Publ., New York.
Drouse, S.K., D.C. Hillman, L.W. Creelman, and S.J. Simon. 1986. National Surface Water
Survey: National Stream Survey (Pilot, Middle-Atlantic Phase I, Southeast Screening, and
Episodes Pilot). Quality Assurance Plan. EPA/600/4-86/044. U.S. Environmental Protection
Agency, Las Vegas, Nevada.
Drouse, S.K. 1987. Evaluation of Quality Assurance and Quality Control Sample Data for the
National Stream Survey (Phase I- Pilot Survey). EPA/600/8-87/057. Lockheed Engineering
and Management Services Company, Inc., Las Vegas, Nevada, 89109.
Dugan, P.R. 1985. Biochemical Ecology of Water Pollution. Plenum Press, New York.
Edland, S.D., T.J. Permutt, and T.S. Stocking. 1987. Estimating Relative Interlaboratory Bias for
the National Streams Survey. Appendix D in Cougan, K.A., D.W. Sutton, D.V. Peck, V.J.
Miller, and I.E. Pollard.
Eilers, J.M. Senior Scientist. Northrop Services, Inc. U.S. Environmental Protection Agency
Environmental Research Laboratory, 200 SW 35th Street, Corvallis, Oregon, 97333. Personal
communication.
333
-------
Elwood, J.W., and PJ. Mulholland. 1987. Effects of Acid Precipitation on Stream Ecosystems.
In: A. Johnson and D. Adriano (eds). Acid Precipitation: Biological and Ecological
Effects, Volume IV. Springer-Verlag, New York. (In review)
Eshleman, K.N. 1988. Predicting regional episodic acidification of surface waters using
empirical techniques. Water Resour. Res. (In press)
Eshleman, K.N., and P.R. Kaufmann. 1988. Assessing the regional effects of sulfur deposition
on surface water chemistry: the Southern Blue Ridge. Environ. Sci. Technol. (In press).
Eshleman, K.N., D.J. Blick, P.R. Kaufmann, M.E. Mitch, S.M. Stambaugh, and J. J. Messer. 1987.
Acid-base status of surface waters in the Southern Blue Ridge: a comparison of results
from the National Surface Water Survey. Lake Reserv. Manage. 3:476-481.
Fenneman, N.M. 1946. Physical divisions of the United States. Map (Scale 1:7,000,000). U.S.
Department of Interior, Geological Survey. Reston, Virginia.
Ford, D.E., K.W. Thornton, J.F. Nix, J.T. Malcolm, and F.E. Payne. 1986. Acidic episodes and
surface water chemistry: a comparison of Northeast and Southeast study sites. Report
submitted to EPA Environmental Research Laboratory, Corvallis, Oregon. Arkadelphia,
Arkansas: Ouachita Baptist University, and Little Rock, Arkansas: FTN Associates, Ltd.
Forman, R.T.T. 1979. Pine Barrens: Ecosystem and Landscape. Academic Press, New York.
Gagen, C.J., and W.E. Sharpe. 1987. Net sodium loss and mortality of three salmonid species
exposed to a stream acidified by atmospheric deposition. Bull. Environ. Contam. Toxicol.
39:7-14.
Galloway, J.N., S.A. Norton, and M.R. Church. 1983. Freshwater acidification from atmospheric
deposition of sulfuric acid: a conceptual model. Environ. Sci. Technol. 17(ll):541-545.
Geragnty, J.J., D.W. Miller, F. Van Der Leeden, and F.L. Troise. 1973. Water Atlas of the
United States. Water Information Center, Inc., Manhasset Isle, Port Washington, New York.
200 pp.
Golterman, H.L. 1975. Physiological limnology, an approach to the physiology of lake
ecosystems. Elsevier, New York. 489 pp.
Gorham, E., S.E. Bayley, and D.W. Schindler. 1984. Ecological effects of acid deposition upon
peatlands: a neglected field in "acid rain" research. Can. J. Fish. Aquat. Sci. 41:1256-1268.
Gunn, J.M., and W. Keller. 1984. Spawning site water chemistry and lake trout (Salvelinus
namavcush) sac fry survival during spring snowmelt. Can. J. Fish. Aquat. Sci. 41:319-329.
Hadley, J.B., and A.E. Nelson. 1971. Geologic map of the Knoxville quadrangle, North Carolina,
Tennessee, and South Carolina. USGS Misc. Geol. Investigations map 1-654, scale 1:250,000.
334
-------
Hagley, C.A., C.L. Mayer, and R. Hoenicke. 1988. National Surface Water Survey: National
Stream Survey (Phase I, Southeast Screening, and Episodes Pilot). Field Operations Report.
EPA/600/x-xx/xxx. U.S. Environmental Protection Agency, Las Vegas, Nevada. (In press)
Haines, T.A. 1981. Acidic precipitation and its consequences for aquatic ecosystems: a review.
Trans. Am. Fish. Soc. 110:669-707.
Haines, T.A., and J.J. Akielaszek. 1983. A regional survey of the chemistry of headwater lakes
and streams in New England: vulnerability to acidification. In: Air Pollution and Acid
Rain. Report No. 15, U.S. Dept. Interior, Fish and Wilflife Service, FWS/OBS-80/40/15.
141 pp.
Haines, T.A., and J.P. Baker. 1986. Evidence of fish population responses to acidification in the
eastern United States. Water Air Soil Pollut. 31:605-629.
Hall, L.W. 1987. Acidification effects on larval striped bass, Morone saxatilis. in Chesapeake
Bay tributaries: A review. Water Air Soil Pollut. 35:87-96.
Hall, L.W., A.E. Pinkney, and L.O. Horseman. 1985. Mortality of striped bass larvae in relation
to contaminants and water quality in a Chesapeake Bay tributary. Trans. Amer. Fish. Soc.
114:861-868.
Hall, L.W., A.E. Pinkney, R.L. Herman, and S.L. Finger. 1987. Survival of striped bass larvae
and yearlings in relation to contaminants and water quality in the Upper Chesapeake Bay.
Arch. Environ. Contam. Toxicol. 16:391-400.
Hammond, E.H. 1970. Classes of land-surf ace form. Map (scale 1:7,500,000). Plates 62-63. In:
The National Atlas of the United States of America. U.S. Geological Survey, Washington
D.C.
Hansen, D.A., and G.M. Hidy. 1982. Review of questions regarding rain acidity data. Atmos.
Environ. 16:2107-2126.
Hardeman, W.D. 1966a. Geologic Map of Tennessee, East-Central sheet. Map (scale 1:250,000).
Tennessee Department of Conservation, Division of Geology.
Hardeman, W.D. 1966b. Geologic Map of Tennessee, East sheet. Map (scale 1:250,000).
Tennessee Department of Conservation, Division of Geology.
Harvey, H.H. 1980. In Ecological Impact of acid precipitation, SNSF Project, Oslo-A. pp.93-
98.
Havas, M. 1985. Aluminum bioaccumulation and toxicity to Daohnia magna in soft wate at low
pH. Can. J. Fish. Aquat. Sci. 42:1741-1748.
Havas, M., T.C. Hutchinson, and G.E. Likens. 1984. Red herrings in acid rain research.
Environ. Sci. Technol. 18:176A-186A.
335
-------
Henricksen, A. 1979. A simple approach for identifying and measuring acidification of fresh
waters. Nature 278:542-545.
Henricksen, A. 1980. Acidification of freshwater—a large-scale titration. pp. 68-74. In: D.
DrabtySs and A. Tollan (eds). Ecological Impact of Acid Precipitation. SNSF Project.
Henricksen, A. 1982. Susceptibility of surface waters to acidification, pp. 103-121. In: R.E.
Johnson (ed). Acid Rain/Fisheries. American Fisheries Society, Bethesda, Maryland.
Herczeg, A.L. and R.H. Hesslein. 1984. Determination of hydrogen ion concentration in
softwater lakes using carbon dioxide equilibria. Geochim. Cosmochim. Acta 48:837-845.
Higbee, H.W. 1986. Stream map of Pennsylvania, showing stocked trout streams and their
tributaries vulnerable to further impacts of acid deposition. Pennsylvania Fish Commission.
Hillman, D.C., S.L. Pia, and S.J. Simon. 1987. National Surface Water Survey: National Stream
Survey (Pilot, Middle-Atlantic Phase I, Southeast Screening, and Middle-Atlantic Episodes
Pilot). Analytical Methods Manual. EPA/600/8-87/005. U.S. Environmental Protection
Agency, Las Vegas, Nevada.
Howells, G.D. 1984. Fishery decline: mechanisms and predictions. Phil. Trans. R. Soc. Lond.
B305:529-547. Lockheed-EMSCO, Las Vegas, Nevada.
Huckabee, J.W., C.P. Goodyear, and R.D. Jones. 1975. Acid rock in the Great Smokies:
unanticipated impact on aquatic biota of road construction in regions of sulfide
mineralization. Trans. Am. Fish. Soc. 104:677-684.
Jeffries, D.S., B. LaZerte, and R.A. Linthurst. 1985. Effects of acid deposition on the chemistry
of aquatic ecosystems in eastern North America. Intern. Symp. on Acid Precip., Muskoka
Conference, September 1985 (Abstract).
Johnson, A.H. 1979a. Evidence of acidification of headwater streams in the New Jersey
Pinelands. Science 206:834-836.
Johnson, A.H. 1979b. Acidification of headwater streams in the New Jersey Pine Barrens. J.
Environ. Qual. 8:383-386.
Johnson, F. 1986. Acid precipitation. Publication Number PFC 1003 7/86 10M. Pennsylvania
Fish Commission, Harrisburg, Pennsylvania.
Johnson, D.W., H.A. Simonin, J.R. Colquhoun, and P.M. Flack. 1987. In-site toxicity tests of
fishes in acid waters. Biogeochemistry 3:181-208.
Kanciruk, P., R.J. Olson, and R.A. McCord. 1986. Quality control in research data bases: the
U.S. Environmental Protection Agency National Surface Water Survey experience. In: W.K.
Michener (ed). Research Data Management in the Ecological Sciences, University of South
Carolina Press, Columbia, South Carolina, pp. 193-207.
336
-------
Keith, L.H., W. Cummett, J. Deegan, Jr., R.A. Libby, J.K. Taylor, and G. Wentler.
Principles of environmental analysis. Analytical Chemistry. 55:2210-2218.
1983.
Kelso, J.R.M., and J.M. Gunn. 1982. Responses of fish communities to acid waters in Ontario.
In: G.R. Hendrey (ed). Chronic stress of acidification on aquatic biota: pH 5 to 6. Am.
Chem. Soc. Symp. Ser., Ann Arbor Science Publishers, Ann Arbor, Michigan.
King, P.B., R.B. Newman, and J.B. Hadley. 1968. Geology of the Great Smoky Mountains Nation
Park, Tennessee and North Carolina. USGS Professional Paper #587, U.S. Government
Printing Office, Washington, D.C. 23 pp.
Klauda, R.J., and R.E. Palmer. 1987. Responses of Blueback Herring of Eggs and Larvae to
Pulses of Acid and Aluminum. Trans. Am. Fish. Soc. 116:561-569.
Klauda, R.J., R.E. Palmer, and M.J. Lenkevich. 1987. Sensitivity of Early Life Stages of
Blueback Herring to Moderate Acidity and Aluminum in Soft Freshwater. Estuaries
10(l):44-53.
Knapp, C.M., C.L. Mayer, D.V. Peck, J.R. Baker, and G.J. Filbin. 1987. National Surface Water
Survey: National Stream Survey (Pilot Survey). Field Operations Report. EPA/600/8-
87/019. U.S. Environmental Protection Agency, Las Vegas, Nevada.
Kramer, J.R. 1984. Modified Gran analysis for acid and base titrations. Environmental Geo-
chemistry Report No. 1984-2. McMaster University, Hamilton, Ontario, Canada.
Kramer, J.R., and A. Tessier. 1978. Acidification of aquatic systems: a critique of chemical
approaches. Environ. Sci. Technol. 16:606-615.
Krug, E.G., and C.R. Frink. 1983. Acid rain on acid soil: a new perspective. Science
221:520-525.
Krug, E.G., P.J. Isaacson, and C.R. Frink. 1985. Appraisal of some current hypotheses
describing acidification of watersheds. JAPCA 35:109-114.
Landers, D.H., J.M. Eilers, D.F. Brakke, W.S. Overton, P.E. Kellar, M.E. Silverstein, R.D.
Schonbrod, R.E. Crowe, R.A. Linthurst, J.M. Omernik, S.A. Teague, and E.P. Meier. 1987.
Characteristics of Lakes in the Western United States. Volume I. Population Descriptions
and Physico-Chemical Relationships. EPA/600/3-86/054a. U.S. Environmental Protection
Agency, Washington, D.C. 176 pp.
Lefohn, A.S., and R.W. Brocksen. 1984. Acid rain effects research—a status report. JAPCA
34:1005-1013.
Leuven, R.S.E.W., C. den Hartog, M.M.C. Christiaans, and F.G.F. Oyen. 1984. Proc. Acid. Precip.
Symp. 's-Hertogenbosch. Pudoc, Wageningen.
337
-------
Lewis, J.V., and H.B. Kummel. 1950. Geologic Map of New Jersey, Atlas sheet No. 40. Map
(scale 1:250,000). New Jersey Department of Conservation and Economic Development.
Linthurst, R.A., D.H. Landers, J.M. Eilers, D.F. Brakke, W.S. Overton, E.P. Meier, and R.E.
Crowe. 1986. Characteristics of Lakes in the Eastern United States. Volume I.
Population Descriptions and Physico-Chemical Relationships. EPA/600/4-86/007a. U.S.
Environmental Protection Agency, Washington, D.C. 136 pp.
Lynch, D.D., and N.B. Dise. 1985. Sensitivity of Stream Basins in Shenandoah National Park to
Acid Deposition. USGS Water Resources Investigations Report 85-4115. U.S. Geological
Survey, Washington D.C., and Dept. of Environmental Sciences, University of Virginia,
Richmond, Virginia.
Magnuson, J.J., J.P. Baker, and EJ. Rahel. 1984. A critical assessment of the effects of
acidification on fisheries in North America. Philosophical Transactions of the Royal Society
of London, B. Biological Sciences 305:501-516.
Marmorek, D.R., K.W. Thornton, J.P. Baker, D.P. Bernard, MX. Jones, and B.S. Reuber. 1986.
Acidic Episodes in Surface Waters: The State of Science. Final Report from ESSA
Environmental and Social Systems Analysts Ltd., Vancouver, British Columbia.
McQuaker, N.R., P.D. Kluckner, and D.K. Sandberg. 1983. Chemical analysis of acid
precipiation: pH and acidity determinations. Environ. Sci. Technol. 17(7):431-435.
Mericas, C.E., and R.D. Schonbrod. 1987. Measurement uncertainty in the National Surface
Water Survey. Lake and Reserv. Manag. 3:488-497.
Messer, J.J., and K.N. Eshleman. 1987. The Feasibility of Quantifying the Regional Extent,
Magnitude, Duration, and Frequency of Episodes: Results of the National Surface Water
Survey Pilot Studies. Presentation at NAPAP Aquatic Effects Task Group VI Peer Review,
May 17-23, New Orleans, Louisiana.
Messer, J. J., C.W. Ariss, J.R. Baker, S.K. Drouse, K.N. Eshleman, P.R. Kaufmann, R.A. Linthurst,
J.M. Omernik, W.S. Overton, M. J. Sale, R.D. Schonbrod, S.M. Stambaugh, and J.R. Tuschall,
Jr. 1986. National Stream Survey Phase I - Pilot Survey. EPA/600/4-86/026. U.S.
Environmental Protection Agency, Washington, D.C. 179 pp.
Messer, J. J., C.W. Ariss, J.R. Baker, S.K. Drouse, K.N. Eshleman, A. J. Kinney, W.S. Overton, M. J.
Sale, and R.D. Schonbrod. 1988. Stream chemistry in the Southern Blue Ridge: feasibility
of a regional synoptic sampling approach. Water Resour. Bull. (In press)
Messer, J.J., D.H. Landers, R.A. Linthurst, and W.S. Overton. 1987. Critical design and inter-
pretive aspects of the National Surface Water Survey. Lake Reserv. Manage. 3:463-469.
Mills, A.L. 1985. Acid mine waste drainage: microbial impact on the recovery of soil and water
ecosystems. In: D. Klein and R.L. Tate (eds). Soil Reclamation Processes. Marcel
Dekker, Inc., New York.
338
-------
Minshall, G.W., R.C. Peterson, K.W. Cummins, T.L. Bott, J.R. Sedell, C.E. Gushing, and R.L.
Vannote. 1983. Interbiome comparison of stream ecosystem dynamics. Ecol. Monogr.
53:1-25.
Morgan, M.D. 1984. Acidification of headwater streams in the New Jersey Pinelands: a re-
evaluation. Limnol. Oceanogr. 29:1259-1266.
Mulholland, P.J., A.V. Palumbo, J.W. Elwood, and A.D. Rosemond. 1987. Effects of acidification
on leaf decomposition in streams. J. N. Am. Benthol. Soc. 6(3):147-158.
Muniz, I.P., and H. Leivestad. 1980. Toxic effects of aluminum on the brown trout, Salmo
trutta L. In: D. Drablos and A. Tollan (eds). Ecological Impacts of Acid Precipitation. In:
Proc. Int. Conf., Sandefjord, Norway. SNSF Project 1432 AS-NLW. pp. 320-321.
Muniz, I.P., R. Andersen, and TJ. Sullivan. 1987. Physiological response of brown trout (Salmo
trutta) spawners and postspawners to acidic aluminum-rich stream water. Water Air Soil
Pollut. 36:371-379.
Murdoch, P.S. 1986. Chemical input-output budgets and stream chemistry dynamics during a
two-year period in Biscuit Brook, Catskill Mountains, New York. U.S. Geological Survey
Water Resources Investigations Report 86 (Draft). Prepared in cooperation with U.S.
Environmental Protection Agency.
National Acid Deposition Program (NADP). 1987. NADP/NTN Annual Data Summary: Precipita-
tion Chemistry in the United States. 1986. Natural Resource Ecology Laboratory, Colorado
State University, Fort Collins, Colorado. 363 pp.
National Research Council. 1981. Atmosphere-Biosphere Interactions: Toward a Better
Understanding of the Ecological Consequences of Fossil Fuel Combustion. National
Academy Press, Washington, D.C. 263 pp.
National Research Council. 1983. Acid Deposition: Atmospheric Processes in Eastern North
America—a Review of Current Scientific Understanding. National Academy Press,
Washington, D.C. 375 pp.
National Research Council. 1984. Acid Deposition: Processes of Lake Acidification. National
Academy Press, Washington, D.C.
Newell, A.D. 1987. Predicting Spring Lake Chemistry from Fall Samples. In: Proceedings of
the International Conference on Acid Rain. September 1-3, 1987, Lisbon, Portugal.
Newell, A.D., C.F. Powers, and S.J. Christie. 1987. Analysis of Data from Long-Term
Monitoring of Lakes. EPA/600/4-87/014. U.S. Environmental Protection Agency,
Washington, D.C. 150 pp.
New York State Education Department, University of the State of New York. 1970. Geologic
Map of New York, Lower Hudson sheet. Map (scale 1:250,000). Map and Chart Series No.
15. New York State Museum and Science Service.
339
-------
Nix, J.F., K.W. Thornton, D.E. Ford, and J.T. Malcolm. 1986. Storm event sampling of
southwestern Arkansas streams susceptible to acid precipitation. EPA Cooperative
Agreement 811863-01-0 (Draft).
O'Dell, J.W., J.D. Pfaff, M.E. Gales, and G.D. McKee. 1984. Technical Addition to Methods for
the Chemical Analysis of Water and Wastes. Method 300.0, The Determination of Inorganic
Anions in Water by Ion Chromatography. EPA/600/4-85/017. U.S. Environmental Protection
Agency, Cincinnati, Ohio.
Oden, S. 1976. The acidity problem—an outline of concepts, pp. 1-36. In: Proc. of the First
Intern. Symp. on Acid Precip. and the Forest Ecosystem. L.S. Dochinger and T.A. Seliga
(eds). U.S. Dept. Agric. For. Serv. Gen. Tech. Rep. NE-23.
Olem, H. 1986. Episodic changes in stream water quality in five watersheds in the Southern
Blue Ridge Province (Draft). Interagency Agreement No. DW64930283-01, TV-61968A.
Oliver, E.G., E.M. Thurman, and R.K. Malcolm. 1983. The contribution of humic substances to
the acidity of colored natural waters. Geochim. Cosmochim. Acta 47:2031-2035.
Olsen, I.E., E.A. Whippo, and G.C. Horak. 1981. REACH file Phase II report: A standardized
method for classifying status and type of fisheries. U.S. Fish Wildl. Serv. FWS/OBS-81/31.
56 pp.
Omernik, J.M. 1987. Ecoregions of the conterminous United States. Annals Assoc. Amer.
Geographers. 77(1):118-125.
Omernik, J.M., and G.E. Griffith. 1985. Total alkalinity of surface waters: a map of the Upper
Midwest Region. EPA/600/D-85/043. U.S. Environmental Protection Agency, Corvallis,
Oregon.
Omernik, J.M., and G.E. Griffith. 1986a. Total alkalinity of surface waters of the United
States: a map of the Upper Midwest Region. Environ. Manage. 10:829-839.
Omernik, J.M., and G.E. Griffith. 1986b. Total alkalinity of surface waters of the United
States: a map of the Western Region. Environ. Manage. 41:374-378.
Omernik, J.M., and AJ. Kinney. 1985. Total alkalinity of surface waters: a map of the New
England and New York Region. EPA/600/D-84/216. U.S. Environmental Protection Agency,
Environmental Research Laboratory, Corvallis, Oregon.
Omernik, J.M., and C.F. Powers. 1983. Total alkalinity of surface waters—a national map.
Annals Assoc. Amer. Geographers. 73(1):133-136.
Overton, W.S. 1986. A Sampling Plan for Streams in the National Stream Survey. Technical
Report 114. Department of Statistics, Oregon State University, Corvallis, Oregon.
340
-------
Overton, W.S. 1987. A Sampling and Analysis Plan for Streams in the National Surface Water
Survey. Technical Report 117. Department of Statistics, Oregon State University,
Corvallis, Oregon.
Overton, W.S., and S.V. Stehman. 1987. An Empirical Investigation of Sampling and Other
Errors in the National Stream Survey; Analysis of a Replicated Sample of Streams.
Technical Report 119. Department of Statistics, Oregon State University, Corvallis, Oregon.
Peden, M.E. 1981. Sampling Analytical and Quality Assurance Protocols for the National
Atmospheric Deposition Program. ASTM D-22 Symposium and Workshop on Sampling and
Analysis of Rain. ASTM, Philadelphia, Pennsylvania.
Pierson, W.R., and T.Y. Chang. 1986. Acid rain in western Europe and northeastern United
States—a technical appraisal. CRC Critical Reviews in Environmental Control 16:167-192.
Platts, W.S. 1979. Relationships among stream order, fish populations, and aquatic geo-
morphology in an Idaho river drainage. Fisheries 4:5-9.
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 Water.
Springer-Verlag, New York.
Reuss, J.O., N. Christophersen, and H.M. Seip. 1986. A critique of models for freshwater and
soil acidification. Water Air Soil Pollut. 30:909-930.
Rochelle, B.P., and M.R. Church. 1987. Regional patterns of sulfur retention in watersheds of
the Eastern U.S. Water Air Soil Pollut. 36:61-73.
Rosen, A.E., and P. Kanciruk. 1985. A Generic Data Entry Quality Assurance Tool. Proceedings
of the Tenth Annual Interntional SAS (Statistical Analysis System, Inc.) Users Group
Conference, March 10-13, Reno, Nevada.
Rosenqvist, I. 1978. Acid precipitation and other possible sources for acidification of rivers
and lakes. Sci. Total. Environ. 10:271-272.
Ruppert, D., and R.J. Carroll. 1980. Trimmed least squares estimation in the linear model. J.
Am. Stat. Assoc. 75:828-838.
Sale, M.J. (ed.). 1988. Data Management and Analysis Procedures for the National Stream
Survey. Oak Ridge National Laboratory, Oak Ridge, Tennessee.
SAS Institute, Inc. 1985. SAS User's Guide: Statistics, Version 5 Edition. SAS Institute, Inc.,
Gary North Carolina. 956 pp.
Schindler, D.W. 1988. Effects of Acid Rain on Freshwater Ecosystems. Science 239:149-157.
341
-------
Schofield, C.L. 1976. Acid precipitation: effects on fish. Ambio 5:338-340.
Schofield, C.L. 1982. Historical fisheries changes in the United States related to decreases in
surface water pH. p. 57-67. T.A. Haines and R.E. Johnson (eds). Proc. Intl. Symp. Acidic
Precip. and Fishery Impacts in Northeastern North America. Cornell University, Ithica,
New York. Published by American Fisheries Society, Bethesda, Maryland.
Sharpe, W.E., D.R. DeWalle, R.T. Leibfried, R.S. Dinocola, W.G. Kimmel, and L.S. Sherwin. 1984.
Causes of acidification of four streams on Laurel Hill, southwestern Pennsylvania. J.
Environ. Qual. 13(4):619-631.
Shreve, R.L. 1966. Statistical law of stream numbers. Journal of Geology 74:17-37.
Skougstad, M.W., J. Fishman, L.C. Friedman, D.E. Erdmary, and S.S. Duncan (eds). 1979.
Methods for Determination of Inorganic Substances in Water and Fluvial Sediments:
Techniques of Water Resources Investigations of the United States Geological Survey, Book
5, Chapter Al. U.S. Government Printing Office, Washington, D.C.
Smith, R.A., and R.B. Alexander. 1983. Evidence for acid precipitation induced trends in stream
chemistry at hydrologic benchmark stations. U.S. Geological Survey Circular 910. U.S.
Dept. of Interior. 12 pp.
Sports Fishing Institute (SFI) Bulletin. 1984. Nov./Dec. No. 360.
Springer, M.E., and J.A. Elder. 1980. Soils of Tennessee. Map (scale 1:750,000). Bulletin 596.
University of Tennessee, Agricultural Experiment Station. Knoxville, Tennessee. 66 pp.
Stapanian,M.A.,A.K. Pollack, and B.C. Hess. 1987. Container Holding Time Study. Appendix B
in Knapp, C.M., C.L. Mayer, D.V. Peck, J.R. Baker, and G.L. Filbin.
Stehman, S.V., and W?SL Overton, 1987a, Estimating the variance of the Horvitz-Thompson
estimator for variable probability, systematic samples. Proceedings of the Survey Research
Section. 1987 ASA Annual Meetings, San Francisco, California.
Stehman, S.V., and W.S. Overton. 1987b. An Empirical Investigation of the Variance Estimation
Methodology Prescribed for the National Stream Survey: Simulated Sampling from Stream
Data Sets. Technical Report 118. Department of Statistics, Oregon State University,
Corvallis, Oregon.
Strahler, A.N. 1957. Quantitative analysis of watershed geomorphology. Trans. Am. Geophys.
Union 38(6):913-920.
Stumm, W., and J.J. Morgan. 1981. Aquatic Chemistry: An Introduction Emphasizing Chemical
Equilibria in Natural Waters, 2nd edition. Wiley-Interscience, New York.
Sullivan, T.J., C.T. Driscoll, J.M. Eilers, and D.H. Landers. 1988a. Evaluation of the role of sea
salt inputs in the long-term acidification of coastal New England lakes. Environ. Sci.
Technol. 22(2): 185-190.
342
-------
Sullivan, T.J., J.M. Eilers, M.R. Church, DJ. Blick, K.N. Eshleman, D.H. Landers, and M.S.
DeHaan. 1988b. Atmospheric wet sulfate deposition and lakewater chemistry. Nature.
331(6157):607-609.
Thornton, K.W., F.E. Payne, J. Ford, and D.H. Landers. 1987a. The Concept of TIME. U.S.
Environmental Protection Agency, Environmental Research Laboratory, Corvallis, Oregon.
91 pp.
Thornton, K.W., J.P. Baker, D.R. Marmorek, D.P. Bernard, M.L. Jones, P.J. McNamee, C. Wedeles,
and K.N. Eshleman. 1987b. Episodic Response Project Draft Research Plan. EPA/600/x-
xx/xxx. U.S. Environmental Protection Agency, Washington, D.C.
Thurman, E.M. 1985. Organic geochemistry of natural waters. Dr. W. Junk Publ., Dordrecht.
TIE Staff. 1981. Regional assessment of aquatic resources at risk from acid deposition. Inst.
of Ecology, Indianapolis, Indiana.
Turner, R.S., A.H. Johnson, and D. Wang. 1985. Biogeochemistry of aluminum in McDonalds
Branch Watershed, New Jersey Pine Barrens. J. Environ. Qual. 14:314-323.
USDA Soil Conservation Service. 1979. General Soil Map of Virginia. Map (scale 1:750,000).
U.S. Environmental Protection Agency. 1983. Methods for Chemical Analysis of Water and
Wastes. EPA/600/4-79/020. U.S. Environmental Protection Agency, Cincinnati, Ohio.
U.S. Environmental Protection Agency. 1984a. The Acidic Deposition Phenomenon and Its
Effects: Critical Assessment Review Papers. EPA-600/8-83/016AFand/016BF. Volume 1:
Atmospheric Sciences; Volume 2: Effects Sciences. ORD: Washington, D.C.
U.S. Environmental Protection Agency. 1984b. National Surface.Water Survey: National Stream
Survey Draft 1985 Research Plan. Environmental Research Laboratory, Corvallis, Oregon.
134 pp.
U.S. Environmental Protection Agency. 1985. National Surface Water Survey, National Stream
Survey; Mid-Atlantic Phase I and Southern Screening Draft Research Plan. ORD,
Washington, D.C.
U.S. Environmental Protection Agency. 1987. Handbook of methods for acid deposition studies:
Laboratory analyses for surface water chemistry. EPA/600/4-87/026. U.S. Environmental
Protection Agency, Office of Research and Development, Washington, D.C. 342 pp.
Vannote, R.L., G.W. Minshall, K.W. Cummins, J.R. Sedell, and C.E. Gushing. 1980. A river
continuum concept. Can. J. Fish. Aquat. Sci. 37:130-137.
Velbel, M.A. 1985. Hydrogeochemical Constraints on Mass Balances in Forested Watersheds of
the Southern Appalachians, pp. 231-247. In: J.I. Drever (ed). The Chemistry of
Weathering. D. Reidel Publ. Co.
343
-------
Velleman, P.P., and D.C. Hoaglin. 1981. Applications, Basics, and Computing of Exploratory
Data Analysis. Duxbury Press, Boston, Massachusetts. 354 pp.
Vollenweider, R.A. 1971. Scientific fundamentals of the eutrophication of lakes and flowing
waters, with particular reference to nitrogen and phosphorus as factors in eutrophication.
Org. for Econ. Co-op, and Dev., Paris. 257 pp.
Von Damm, K.L. 1987. Geochemistry of Shale Groundwaters: Survey of Available Data and
Postulated Mineralogic Controls on Composition. ORNL/TM-10488. Oak Ridge National
Laboratory, Oak Ridge, Tennessee.
Wampler, S.J., and A.R. Olsen. 1987. Spatial estimation of annual wet acid deposition using
supplemental precipitation data. Reprinted from the Preprint Volume: Tenth Conference on
Probability and Statistics. October 4 and 5,1987. Edmonton, Alta., Canada. Published by
the American Meteorological Society, Boston, Massachusetts.
Wendelaar-Bonga,S.E.,andL.H.T.Dederen. 1986. Effects of acidified water on fish. Endeavor,
New Series 10(4):198-202.
Wetzel, R.G. 1975. Limnology. W.B. Saunders Company, Philadelphia, Pennsylvania. 743 pp.
Winger, P.V., P.J. Lasier, M. Hudy, D. Fowler, and MJ. Van Den Avyle. 1987. Sensitivity of
high-elevation streams in the Southern Blue Ridge Province to acidic deposition. Water
Resour. Bull. 23(3):379-386.
Witt, E.G., and J.L. Barker. 1986. Stream chemistry response during episodes of acidic rainfall
and snowmelt runoff on Laurel Hill, Somerset County, Pennsylvania, November 1983 to July
1985. U.S. Geological Survey Open File Report 86 (Draft). Prepared in cooperation with
U.S. Environmental Protection Agency.
Witters, H., J.H.D. Vangenechten, S. Van Puymbroeck, and O.LJ. Vanderborght. 1984. Inter-
ference of aluminum and pH on the Na-influx in an aquatic insect Corixa punctata (Illig.)
Bull. Environ. Contain. Toxicol. 32:575-579.
Wright, R.F., and E.T. Gjessing. 1976. Acid precipitation: changes in the chemical composition
of lakes. Ambio 5:219-223.
Young, T.C., J.V. Depinto, S.C. Martin, and J.S. Bonner. 1985. Algal available particulate
phosphorus in the Great Lakes Basin. J. Great Lakes Research. ll(4):434-446.
344
-------
SECTION 13
GLOSSARY
13.1 ABBREVIATIONS AND SYMBOLS
13.1.1 Abbreviations
AERP
AL
AMD
ANC
ANOVA
ASTM
BC
BNC
C.I.
CDF
CL
CRDL
DDRP
DIG
DOC
DQO
ELS-I
ELS-II
EMSL-LV
EPA
ERL-C
ERP
ESS
IBM
ID
IMA
ISE
Lockheed-EMSCO
LTM
MA
MD DNR
MIBK
Aquatic Effects Research Program
analytical laboratory
acid mine drainage
acid neutralizing capacity
analysis of variance
American Society for Testing and Materials
base cations
base neutralizing capacity
confidence interval
cumulative distribution function
Confidence Limit
contract required detection limit
Direct/Delayed Response Project
dissolved inorganic carbon
dissolved organic carbon
data quality objective
Eastern Lake Survey - Phase I
Eastern Lake Survey - Phase II
Environmental Monitoring Systems Laboratory - Las Vegas
U.S. Environmental Protection Agency
Environmental Research Laboratory - Corvallis
Episodic Response Project
effective sample size
International Business Machines
identification number
Interior Mid-Atlantic subregions (ID, 2Cn, and 2Bn)
Interior Southeast subregions (2As, 3A, 2X, and 2D)
stream reach lower node or liter
Lockheed Engineering and Management Services Company, Inc.
Long-term Monitoring (Project)
Mid-Atlantic Region
State of Maryland, Department of Natural Resources
methyl-isobutyl-ketone
345
-------
NADP/NTN — National Acid Deposition Program/National Trends Network
NAPAP — National Acid Precipitation Assessment Program
NBS — National Bureau of Standards
NCC — National Computer Center
NLS — National Lake Survey
NRDC — Natural Resources Defense Council
NSS-I -- National Stream Survey - Phase I
NSWS -- National Surface Water Survey
NTU — nephelometric turbidity unit
ORNL ~ Oak Ridge National Laboratory
PCA — Principal Component Analysis
PCU — platinum cobalt units (for water color measurement)
PCV ~ Pyrocatechol violet
PL — processing laboratory
QA — quality assurance
QC — quality control
QCCS — quality control check solution
RSD -- relative standard deviation
SAOA — strong acid organic anions
SAS — Statistical Analysis System
SBC — sum of base cations
SBR — Southern Blue Ridge
SD — standard deviation
SDL — system decision limit
SE — Southeast Region or standard error
TIME — Temporal Integrated Monitoring of Ecosystems (Project)
TVA — Tennessee Valley Authority
U — stream reach upper node
UCB — upper 95% confidence bound
USDA — U.S. Department of Agriculture
USFS -- U.S. Forest Service
USGS -- U.S. Geological Survey
USU -- Utah State University
WAOA — weak acid organic anions
YSI — Yellow Springs Instruments
346
-------
13.1.2 Symbols
Al
Alk
bi
V
Ca2+
Cl-
em
C02
co32-
b
Fe
Fe(OH)3
FeS2
F(x)
H+
ha
HCOjf
HNOS
H2C03
H20
H2S04
kg
km
— total area of direct watershed
—- direct drainage area
— drainage area of upstream node for nonhead water reaches
— headwater drainage area of upstream node
— aluminum
— MIBK-extractable aluminum (total monomeric)
— inorganic monomeric aluminum
— carbonate alkalinity
— regression coefficient
— standardized regression coefficient
— calcium
— chloride
-- centimeter
— carbon dioxide
— carbonate
— total discharge index
— fluoride
— iron
— ferric hydroxide
— pyrite (ferrous sulfide)
— cumulative distribution function
— hydrogen ion
— hectare
— bicarbonate
— nitric acid
— carbonic acid
— water
— sulfuric acid
— acid dissociation constant
— potassium
— kilogram
— kilometer
m
MAX
mg
mg L'1
— total length of target reaches
— meter or slope of regression line
— maximum value of x
— molar conductivity
— milligram
— miligrams per liter, unit of concentration
347
-------
Mg2+
mi
MIN
mL
Mn
n
ft
Na+
NaCl
NOS-
n.s.
OH-
02
P
pCO2
PH
Qao
Qso
r
RCOO-
r2
R2
§
SiO2
so42-
— magnesium
— mile
— minimum value of x
— milliliter
— manganese
— number of samples
— total number of target reaches
— sodium
— sodium chloride
— ammonium
— nitrate
— not significant
— hydroxide
— oxygen
— significance level
— partial pressure of CO2
— the negative logarithm of the hydrogen ion molar concentration
— negative logarithm (base 10) of acid dissociation constant
— population 20th percentile, 1st quintile
— population 80th percentile, 4th quintile
— Pearson's correlation coefficient
— organic anion
— coefficient of determination
— coefficient of multiple determination
— total stream surface area
— silicon dioxide
— sulfate
— estimated variance
w
"C
A
7T
L'1
cm
"1
ID
2As
— weight or expansion factor
-- degrees Celsius
— delta: denotes an increment of change
— reach inclusion probability
— microequivalents per liter, unit of concentration
— micrograms per liter, unit of concentration
— micro-Siemens per centimeter, unit of electrical conductance
— micromole per liter, unit of concentration
— Poconos/Catskills subregion
— - Southern Blue Ridge subregion (NSS Pilot Survey)
348
-------
2Bn
2Cn
2D
2X
3A
3B
3C
13.2 DEFINITIONS
— Valley and Ridge subregion
— Northern Appalachians subregion
— Ozarks/Ouachitas subregion
— Southern .Appalachians subregion
— Piedmont subregion
— Mid-Atlantic Coastal Plain subregion
— Florida subregion
ACID MINE DRAINAGE - runoff with high concentrations of metals, sulfate and acidity resulting
from the oxidation of sulfide minerals that have been exposed to air and water (usually
from mining activities). Considered a NONINTEREST attribute for calculating NSS-I
population estimates.
ACID NEUTRALIZING CAPACITY - the total acid-combining capacity of a water sample deter-
mined by titration with a strong acid. Acid neutralizing capacity includes ALKALINITY
(carbonate species) as well as other basic species (e.g., borates, dissociated ORGANIC
ACIDS, alumino-hydroxy complexes).
ACIDIC DEPOSITION - rain, snow, or dry fallout containing high concentrations of sulf uric acid,
nitric acid, or hydrochloric acid, usually produced by atmospheric transformation of the
by-products of fossil fuel combustion (power plants, smelters, autos, etc.); precipitation
with a pH less than 5.0 is generally considered to be unnaturally acidic.
ACIDIC STREAM - for this report, a stream with ACID NEUTRALIZING CAPACITY less than
or equal to 0 /*eq L'1.
ACIDIFICATION - any temporary or permanent loss of ACID NEUTRALIZING CAPACITY.
ACCURACY - the closeness of a measured value to the true value of an ANALYTE.
AIR-EQUILIBRATED - a sample ALIQUOT that has been brought to equilibrium with standard
air (300 ppm CO2) before analysis; used with some pH and DISSOLVED INORGANIC
CARBON measurements.
ALIQUOT - a subsample of a 4-L water sample processed in a specific manner for chemical
analysis at a later time.
ALKALINITY - the titratable base of a sample consisting of hydroxide, carbonate, and
bicarbonate ions, i.e., the equivalents of acid required to neutralize the basic carbonate
components.
ALKALINITY MAP CLASS - a geographic area defined by the expected ALKALINITY of a
majority of SURFACE WATERS (does not necessarily reflect measured alkalinity); used in
the NSWS probability sampling design. (See LOW ANC STRATUM.)
349
-------
AMONG-BATCH PRECISION - the estimate of PRECISION that includes the effects of different
laboratories and day-to-day (BATCH-to-BATCH) differences.
ANALYTICAL LABORATORY - in this report, a laboratory under contract with the U.S.
Environmental Protection Agency to analyze water samples shipped from the PROCESSING
LABORATORIES.
ANALYTE - a chemical species that is measured in a water sample.
ANC DEFICIT - the difference in concentration between the sum of the nonmarine base cations
and ANC.
ANION - a negatively charged ion.
ANION-CATION BALANCE - an assessment of the total charge of positive ions (CATIONS)
compared to the total charge of negative ions (ANIONS).
ANION DEFICIT - the concentration in neq L'1 of measured CATIONS minus measured ANIONS.
AUDIT - an on-site evaluation of field sampling or laboratory activities to verify that
standardized protocols are being followed.
AUDIT SAMPLE - a standardized water sample submitted to laboratories to check overall
performance in sample analysis. See FIELD AUDIT SAMPLE and LABORATORY AUDIT
SAMPLE.
BASE CATION - a nonprotoly tic cation that does not affect ACID NEUTRALIZING CAPACITY;
in the NSS-I, consists of calcium, magnesium, sodium, and potassium.
BATCH - A group of samples (including QUALITY CONTROL samples) processed and analyzed
together at the PROCESSING LABORATORY; associated with a unique batch ID code.
BEDROCK - solid rock exposed at or underlying the land surface.
BIAS - the systematic difference between the measurements or estimations and the true values.
BLANK - a sample of REAGENT GRADE WATER analyzed as a QUALITY ASSURANCE/
QUALITY CONTROL sample in the NSS-I. See FIELD BLANK SAMPLE, LABORATORY
BLANK SAMPLE, and PROCESSING LABORATORY BLANK.
BLUE LINE REACH - a REACH appearing as a BLUE LINE on a l:250,000-scale USGS
topographic map.
BUFFERING CAPACITY - the quantity of ACID or BASE that can be added to a water sample
with little change in pH.
350
-------
CALCULATED CONDUCTANCE - the sum of the theoretical specific conductances of all
measured ions in a sample.
CALIBRATION BLANK - a solution used in standardizing or checking the calibration of analytical
instruments; also used to determine instrument detection limits.
CATCHMENT - see WATERSHED.
CATION - a positively charged ion.
CATION EXCHANGE - a reversible process occurring in soil in which acidic CATIONS (e.g.,
hydrogen ions) are adsorbed and BASE CATIONS are released.
CHELATOR - a class of compounds, organic and inorganic, that can bind metal ions and change
their biological availability.
CIRCUMNEUTRAL - close to neutrality with respect to pH (pH = 7).
CLOSED SYSTEM - method of measurement in which a water sample is collected and analyzed
for pH and DISSOLVED INORGANIC CARBON without exposure to the atmosphere.
CLUSTER ANALYSIS - a multivariate classification technique for identifying similar (or
dissimilar) groups of observations.
COMPARABILITY - a measure of data quality that assesses the similarity within and among data
sets.
COMPLETENESS - a measure of data quality that is the quantity of acceptable data actually
collected relative to the total quantity for which collection was attempted.
CONDUCTANCE - a measure of the electrical conductance (the reciprocal of the electrical
resistance) of a water sample; proportional to the total ionic concentration.
CONTRACT REQUIRED DETECTION LIMIT - for each chemical variable, the highest INSTRU-
MENT DETECTION LIMIT allowable in the ANALYTICAL LABORATORY contract.
CUBITAINER - a 3.8 L container made of semirigid polyethylene used to transport field samples
(routine, duplicate, blank) from the stream site to the processing laboratory.
CUMULATIVE DISTRIBUTION FUNCTION - a function, F(x), such that for any reference value
X, F(x) is the estimated proportion of streams in the population having a value x < X.
DATA QUALIFIER - see FLAG and TAG.
DATA QUALITY OBJECTIVES - ACCURACY, DETECTABILITY, and PRECISION limits
desired by data users, established prior to the beginning of a sampling effort. Also
includes goals for COMPLETENESS, COMPARABILITY, and REPRESENTATIVENESS.
351
-------
DATA SET 1 - set of files containing RAW data.
DATA SET 2 - set of files containing VERIFIED data.
DATA SET 3 - set of files containing VALIDATED data.
DATA SET 4 - set of files containing final, enhanced data; missing values or errors in the
VALIDATED data set were replaced by substitution values, duplicate values were averaged,
and negative measurements (except ACID NEUTRALIZING CAPACITY) were set equal to
zero.
DETECTABILITY - the capacity of an instrument or method to determine a measured value for
an ANALYTE above background levels. See INSTRUMENT DETECTION LIMIT and
SYSTEM DECISION LIMIT.
DISSOLVED INORGANIC CARBON - a measure of the dissolved carbon dioxide, carbonic acid,
bicarbonate and carbonate anions in a water sample; comprises the major part of ACID
NEUTRALIZING CAPACITY in a stream.
DISSOLVED ORGANIC CARBON - a measure of the organic (noninorganic) fraction of carbon in
a water sample that is dissolved or unfilterable (0.45 ftm pore size in the NSS-I).
DOWNSTREAM REACH NODE - see LOWER NODE.
DUPLICATE STREAM SAMPLE - see FIELD DUPLICATE SAMPLE.
EFFECTIVE SAMPLE SIZE (n') - the number of grid points examined within each STRATUM to
obtain the Stage II sample reaches to be visited in the field.
ELECTRONEUTRALITY - having no electric charge.
EPISODE - A short-term change in stream pH and ANC during storm flows or snowmelt runoff;
considered a NONINTEREST attribute for calculating NSS-I population estimates.
EQUIVALENT - unit of ionic concentration; the quantity of a substance that either gains or
loses one mole of protons or electrons.
EXTRACTABLE ALUMINUM - an operationally defined aluminum fraction determined using a
MIBK extraction.
FIELD AUDIT SAMPLE - a standardized water sample submitted to PROCESSING LABORA-
TORIES to check overall performance in sample analysis by both PROCESSING and
ANALYTICAL LABORATORIES.
FIELD BLANK SAMPLE - a sample of REAGENT GRADE WATER prepared at the stream site by
field crews; these samples were analyzed by both PROCESSING and ANALYTICAL
LABORATORIES.
352
-------
FIELD DUPLICATE SAMPLE - an additional water sample collected at stream side immediately
after a ROUTINE SAMPLE in accordance with standardized protocols.
FLAG - qualifier of a data value; used to identify specific notes or observations made during the
VERIFICATION and VALIDATION procedures.
FRAME - a structural representation of a population providing a sampling capability.
FRAME POPULATION - in the NSS-I, the set of BLUE LINE STREAM segments appearing on a
l:250,000-scale USGS topographic map within a designated region of interest.
GEOGRAPHIC SITE CLASS - a post sampling stratification of the NSS-I TARGET POPULA-
TION according to a combination of physio-geographic attributes and land use patterns that
is not constricted by SUBREGION boundaries.
GRADIENT - the ratio between the elevation difference and length distance between the UPPER
and LOWER NODE of a REACH.
GRAN ANALYSIS - a mathematical procedure used to determine the equivalence points of a
titration curve for ACID and BASE NEUTRALIZING CAPACITY.
HEADWATER REACH - map scale dependent, the uppermost reach of a stream drainage network
(i.e., a reach with a STRAHLER ORDER and SHREVE ORDER equal to 1).
HOLDING TIME - the time elapsed between sample collection and analysis.
HYDROLOGIC FLOW PATHS - the distribution and circulation of water on the surface of the
land, in the soil, and underlying rocks within a watershed.
IN SITU - referring to measurements collected within the water column at a stream.
INCLUSION PROBABILITY - the chance of a stream being selected for sampling.
INDEX PERIOD - see SPRING BASEFLOW INDEX PERIOD.
INDEX VALUE - the average of one to three samples collected at a stream NODE during the
SPRING BASEFLOW INDEX PERIOD, used to represent chemical conditions in the stream.
INITIAL DIC - a measurement of DISSOLVED INORGANIC CARBON made on an ALIQUOT
immediately before it is titrated for ACID NEUTRALIZING CAPACITY.
INORGANIC MONOMERIC ALUMINUM - an operationally defined aluminum fraction deter-
mined in the NSS-I as the difference in concentration between TOTAL MONOMERIC
ALUMINUM and ORGANIC ALUMINUM (PCV technique).
INSTRUMENT DETECTION LIMIT - a value calculated from LABORATORY BLANK SAMPLES
that indicates the minimum concentration (with 95% confidence) that can be distinguished
from blank samples by the instrument(s) used.
353
-------
INTERLABORATORYBIAS - systematic differences in performance between laboratories esti-
mated from analysis of the same type of AUDIT SAMPLES.
INVERSE DISTRIBUTION FUNCTION - a function, 1 -F(x), such that for any reference value X,
l-F(x) is the estimated number of streams in the population having a value x > X.
ION BALANCE - see ANION-CATION BALANCE.
IONIC STRENGTH - a measure of the interionic effect resulting from the electrical attraction
and repulsion between various ions; can be calculated from the measured concentrations of
ANIONS and CATIONS.
LABILE ALUMINUM - see INORGANIC MONOMERIC ALUMINUM.
LABORATORY BLANK SAMPLE- a sample of REAGENT GRADE WATER prepared and anal-
yzed by ANALYTICAL LABORATORIES.
LABORATORY DUPLICATE SAMPLE - a split sample prepared and analyzed at the ANA-
LYTICAL LABORATORIES.
LONG-TERM MONITORING PROGRAM - an ongoing program sponsored by the U.S. Environ-
mental Protection Agency that monitors the chemistry of selected lakes and streams at
least three times per year, designed to detect long-term trends in chemistry.
LOW ANC STRATUM - a subset of sample reaches within the STAGE I SAMPLE representing
those having an ANC less than 50 neq L"1 based on expected alkalinity maps.
LOWER NODE - the downstream NODE of a stream REACH.
MAP POPULATION - in the NSS-I, the blue-line representation of streams depicted on
l:250,000-scale U.S. Geological Survey TOPOGRAPHIC MAPS.
MEAN SQUARE ERROR - a measure of accuracy, usually variance plus square of BIAS.
MEDIAN (M) - the value of x such that F(x) or G(x) = 0.5; the 50th percentile.
METHOD LEVEL PRECISION - PRECISION associated with the analysis of PROCESSING and
ANALYTICAL LABORATORY DUPLICATE SAMPLES.
MINERAL WEATHERING - dissolution of rocks and minerals by erosive forces.
NATURAL AUDIT SAMPLE - an AUDIT SAMPLE collected from a surface waters having known
ion concentrations similar to those expected in sample streams.
NODE - the points identifying either an upstream or downstream end of a REACH. In the
NSS-I, The point above or below the confluence of two streams.
354
-------
NONINTEREST REACHES - a set of REACHES excluded during the refinement of the TARGET
POPULATION after field visitation due to some external influence (tidal effects, acid mine
drainage, dry stream channel, etc.). Data from these reaches were excluded in the
calculation of NSS-I population estimates.
NONTARGET REACHES - a set of REACHES not in the focus of the NSS-I objectives (urban
areas, swamps, large rivers, etc.); excluded from the NSS-I TARGET POPULATION before
field visitation.
NTU - nephelometric turbidity unit; a measure of light scattered by a solution of suspended
materials.
OLIVER MODEL - a set of empirical equations used to calculate ORGANIC ANION concentra-
tions from DOC and pH.
OPEN SYSTEM - a measurement of pH or DISSOLVED INORGANIC CARBON obtained from a
sample that was exposed to the atmosphere during collection and/or measurement.
ORGANIC ACIDS - organic compounds possessing an acidic functional group; includes fulvic and
humic acids.
ORGANIC ALUMINUM - an operationally defined aluminum fraction determined by colorimetry
(PCV technique) after passing the sample through a strong cation exchange column.
ORGANIC ANION - an organic molecule with a negative net ionic charge.
OUTLIERS - observations not typical of the population from which the sample is drawn.
PERCENT ION BALANCE - percent difference between the anion deficit and the total ionic
equivalents [(sum cations - sum anions)/(sum of cations + sum of anions) * 100)].
PERCENT RELATIVE STANDARD DEVIATION (% RSD) - the STANDARD DEVIATION
divided by the mean, then multiplied by 100.
pH - the negative logarithm of the hydrogen ion activity. The pH scale runs from one (most
acidic) to fourteen (most alkaline); the difference of one pH unit indicates a ten fold
change in hydrogen ion activity.
PLATINUM COBALT UNIT - measure of the color of a water sample defined by a potassium
hexachloroplatinate and cobalt chloride standard color series.
POPULATION ESTIMATE - a statistical estimate of the numbers (or length, watershed area) of
stream reaches within a specified population or subpopulation of interest.
PRECISION - a measure of the capacity of a method to provide reproducible measurements of a
particular ANALYTE (often represented by variance).
355
-------
PROBABILITY SAMPLE - a sample in which each unit (stream reach, in this case) has a known
probability of being selected.
PROCESSING LABORATORY - laboratory in which sample processing (preservation, separation
into ALIQUOTS, and analysis of selected chemical variables) was performed before shipping
of water samples to the ANALYTICAL LABORATORY.
PROCESSING LABORATORY BLANK - a REAGENT GRADE WATER sample prepared at the
PROCESSING LABORATORY, but analyzed at an ANALYTICAL LABORATORY.
PROCESSING LABORATORY DUPLICATE - a SPLIT SAMPLE prepared in the PROCESSING
LABORATORY.
PROTOLYTE - an analyte that reacts with either H+ or OH" in solution.
QUALITY ASSURANCE - steps taken to ensure that a study is adequately planned and
implemented to provide data of the highest quality, and that adequate information is
provided to determine the quality of the data base resulting from the study.
QUALITY CONTROL - steps taken during sample collection and analysis to ensure that the data
quality meets the minimum standards established by the QUALITY ASSURANCE plan.
QUALITY CONTROL CHECK SOLUTION - a sample of known concentration used to verify
continued calibration of an instrument.
QUINTILE - any of the four values (Qa, Q2, Qs, QJ that divide a population distribution into
five equal classes, each representing 20% of the population; used to provide additional
values to compare among populations of streams.
RAW DATA SET - the initial data set (DATA SET 1) that has received a cursory review to
confirm that data are provided in proper format and are complete and legible.
REACH - segments of the stream network represented as blue lines on l:250,000-scale USGS
maps. Each REACH (segment) is defined as the length of stream between two blue-line
confluences. In the NSS-I, stream reaches were the sampling unit. See HEADWATER
REACH.
RE AGENT BLANK - a LABORATORY BLANK SAMPLE that contained all the reagents required
to prepare a sample for analysis.
REAGENT GRADE WATER - deionized water used for BLANK samples that had measured
CONDUCTANCE less than 1 /*S cm"1.
REFERENCE VALUE (Xc) - a concentration of interest for a given chemical variable.
REFINED TARGET POPULATION - a subset of the TARGET POPULATION that remains after
removing NONINTEREST REACHES (post-sampling field and chemical exclusions).
356
-------
REGION - for this report, a major area of the conterminous United States where a substantial
number of streams with ALKALINITY less than 400 jieq L"1 can be found.
REPRESENTATIVENESS - a measure of data quality; the degree to which sample data accurately
and precisely reflect the characteristics of a population.
ROOT MEAN SQUARE ERROR - the square root of the MEAN SQUARE ERROR.
ROUTINE SAMPLE - a water sample collected using standardized protocols at a stream sample
site and analyzed for all NSS-I chemical and physical variables.
SAMPLE ID - each water sample collected during a visit to NSS-I REACH is identified by a
unique sample identification code, a number used to distinguish every sample from another.
SAS - Statistical Analysis System, Inc. (Gary, NC). A statistical data file manipulation and data
analysis software package.
SHREVE ORDER - a map scale dependent system devised by Shreve (1966), for ranking streams
according to their positions within the larger stream drainage network. The Shreve order
of a stream is equal to the number of fingertip tributaries (headwaters) within the drainage
area upstream from the point of reference on the stream.
SITE INCLUSION CRITERIA - see SITE RULES.
SITE RULES - the set of rules used to categorize REACHES or other features identified by grid
points into various TARGET or NONTARGET categories.
SPECIAL INTEREST SITE - in this Survey, a non-randomly selected stream that is not part of
the PROBABILITY SAMPLE; selection was based on quality and amount of data available or
involvement in other research programs.
SPLIT SAMPLE - a subsample of a field BATCH sample that was sent to a laboratory other
than an ANALYTICAL LABORATORY for analysis.
SPRING BASEFLOW INDEX PERIOD - a period of the year when streams are expected to exhibit
chemical characteristics most closely linked to acidic deposition. The time period between
snowmelt and leaf out (March 15 to May 15 in the NSS-I) when NSS-I stream REACHES
were visited coinciding with expected periods of highest geochemical and assessment
interest (i.e., low seasonal pH and sensitive life stages of biota).
STAGE I SAMPLE - the set of TARGET POPULATION, BLUE LINE STREAM REACHES
selected from 1:250,000 USGS topographic maps using a rectangular dot grid sampling frame
(1 point per 64 mi2) and which met the SITE INCLUSION CRITERIA.
STAGE II SAMPLE - the set of stream reaches visited in the field, selected (using a randomized
systematic sampling procedure) from the STAGE I SAMPLE.
357
-------
STAGE HI SAMPLING - the procedures by which field observations and chemical measure-
ments were taken on STAGE II sample reaches.
STANDARD ERROR - the square root of the variance of a statistic.
STRAHLER ORDER - a map scale dependent system devised by Strahler (1957) for ranking
streams on the basis of their position within the stream drainage network. Within this
ordering system, tributaries at the head of the drainage network are designated as first-
order stream segments. Two first-order segments join to form a second-order stream
segment; two second-order streams join to form a third-order segment, and so on.
STRATIFICATION FACTORS - factors used to define STRATA before stream selection; the
factors used in the NSS-I were SUBREGION and ANC STRATUM. See STRATUM.
STRATIFIED DESIGN - a statistical design in which the population is divided into strata, and a
sample selected from each STRATUM.
STRATUM - in the NSS-I, sampling strata were designated as region and expected alkalinity map
class (greater than or less than 50 neq L"1).
STREAM ID - a unique nine-character identification code given to each stream sampling site
(NODE). It contains four fields indicating the NSS-I subregion code, the l:250,000-scale
map ID, the grid point identification code, and the node identifier. Stream ID 3B041016L
designates a lower node of the 16th grid point on the l:250,000-scale map 041 in subregion
3B.
SUBPOPULATION - any defined subset of the TARGET POPULATION.
SUBREGIONS - areas within the NSS-I study area that are similar in water quality,
PHYSIOGRAPHY, vegetation, climate, and soil; used as a STRATIFICATION FACTOR in
NSS-I design.
SURFACE WATER - for this report, streams and lakes.
SYNOPTIC - relating to or displaying conditions as they exist over a broad area.
SYNTHETIC AUDIT SAMPLE - an AUDIT SAMPLE prepared by diluting solutions of known
chemical composition.
SYSTEM DECISION LIMIT - a value calculated from FIELD BLANK data that indicates the
minimum concentration (with 95% confidence) that can be distinguished from background
levels.
SYSTEM LEVEL PRECISION - PRECISION associated with the sample collection, transport,
processing, preservation, shipment, analysis, and data reporting of FIELD DUPLICATE
SAMPLES.
358
-------
SYSTEMATIC ERROR - a consistent error introduced in the measuring process. Such error
commonly results in biased estimations.
SYSTEMATIC RANDOM SAMPLING - a sampling technique in which the units in the population
are ordered and the sample selected in a systematic probabilistic manner (see Section 2.4).
TAG - code on a data point that is added at the time of collection or analysis to qualify the
data.
TARGET POPULATION - a subset of the FRAME POPULATION conforming to the specified
criteria established for conduct of the NSS-I (see SITE INCLUSION CRITERIA).
TOTAL ALUMINUM - in the NSS-I, the total aluminum (measured by atomic absorption
spectroscopy) in an unfiltered, acidified sample.
TOTAL MONOMERIC ALUMINUM - an operationally defined aluminum fraction determined by
colorimetry (PCV technique).
TRILINEAR DIAGRAM - a diagram showing three components; for this report, these diagrams
were used to examine the ratios of major CATIONS and ANIONS (synonymous with ternary
diagrams).
TRUE COLOR - the color of water that has been filtered or centrifuged to remove particles
that may impart an apparent color; true color ranges from clear (0 PCU) to blackish-brown
(> 500 PCU).
TURBIDITY - a measure of light scattering by suspended particles in an unfiltered water sample.
UPPER CONFIDENCE BOUND (95%) - a value such that one is 95% confident that the true value
is below this bound.
UPPER NODE - the upstream NODE of a stream REACH.
UPSTREAM REACH NODE - see UPPER NODE.
VALIDATION - process by which data are evaluated for quality with reference to the intended
data use; includes identification of OUTLIERS and evaluation of potential SYSTEMATIC
ERROR.
VERIFICATION - process of ascertaining the quality of the data in accordance with the minimum
standards established by the QUALITY ASSURANCE plan.
WATERSHED - the land area contributing to the flow of a stream at a given point.
Topographically defined in the NSS-I as the drainage area contained within a drainage
divide above a specified point on a stream.
359
-------
WEATHERING - see MINERAL WEATHERING.
WEIGHT - the inverse of a sample stream's INCLUSION PROBABILITY; each sample stream
represents this number of streams in the target population.
360
-------
APPENDIX A
DATA QUALITY ASSESSMENT TABLES
The following eight data tables present additional QA data that was not presented in
Section 4. These tables contain information on the data quality objectives, instrument detection
limits, system decision limits, and accuracy estimates for each of the two analytical laboratories,
along with method, system, and among-batch precision estimates for all of the NSS-I chemical
variables. The last table (Table A-8) summarizes the QA results with respect to detectability,
accuracy, and precision for all NSS-I chemical variables. These tables were taken from the
NSS-I Quality Assurance Report (Cougan et al., 1988) and interested readers should refer to that
document for additional information, explanations, and interpretation.
361
-------
Table A-1. Analytical Data Quality Objectives for Detectability, Precision, and Accuracy for
the NSS-I
Variable (units)
FIELD SITE
pH, field
Detection
limit
objective
(units)
__
Within-
laboratory
precision
(%RSD)a
—
Within-
laboratory
accuracy (%)
±0.1 b
Specific conductance
cm-1)
Dissolved oxygen
(mg L-1)
PROCESSING LABORATORY
Aluminum (mg L'1)
Total monomeric 0.01
(PCV)
Nonexchangeable 0.01
monomeric
(PCV)
Specific conductance
(pS cm'1) c
pH, closed system
Dissolved inorganic
carbon, closed system
(mg L'1) 0.05
10 (>0.01 mg L"1)
20 (<0.01 mg L'1)
10 (>0.01 mg L'1)
20 (<0.01 mg L'1)
3
O.lb
10
10 (>0.01 mg L-1)
20 (<0.01 mg L'1)
10 (>0.01 mg L'1)
20 (<0.01 mg L"1)
5
±0.1b
10
True color (PCU)
Turbidity (NTU)
0
2
10
10
362
-------
Table A-l. (continued)
Variable (units)
Detection
limit
objective
(units)
Within-
laboratory
precision
(%RSD)a
Within-
laboratory
accuracy (%)
ANALYTICAL LABORATORY
Acid neutralizing
capacity (/zeq L"1) d
Aluminum (mg L"1)
Total extractable
(MIBK)
Total
Ammonium (mg L"1)
Base-neutralizing
capacity (/zeq L"1)
Calcium (mg L"1)
Chloride (mg L"1)
Specific conductance
cm"1)
Dissolved inorganic
carbon (mg L"1)
Initial
Air equilibrated
Dissolved organic
carbon (mg L"1)
Fluoride, total
dissolved (mg L'1)
Iron (mg L"1)
0.005
0.005
0.01
10
10
10 (>0.01 mg L'1)
20 (<0.01 mg L'1)
10 (>0.01 mg L'1) 10 (>0.01 mg L"1)
20 (<0.01 mg L'1) 20 (<0.01 mg L'1)
10 (>0.01 mg L'1)
20 (<0.01 mg L'1)
10
d
0.01
0.01
10
5
5
10
10
10
0.05
0.05
0.1
0.005
0.01
10
10
5 (>5.0 mg L'1)
10 (<5.0 mg L"1)
5
10
10
10
10
10
10
363
-------
Table A-l. (continued)
Detection
limit
objective
Variable (units) (units)
Magnesium (mg L"1) 0.01
Manganese (mg L"1) 0.01
Nitrate (mg L'1) 0.005
PH
Air equilibrated
Initial ANC
Initial BNC
Within-
laboratory
precision
(%RSD)a
5
10
10
0.05b
0.05b
0.05b
Within-
laboratory
accuracy (%)
10
10
10
±0.1b
±0.1b
+O.lb
Phosphorus, total 0.002
dissolved (mg L"1)
Potassium (mg L"1) 0.01
Silica (mg L'1) 0.05
Sodium (mg L'1) 0.01
Sulfate (mg L'1) 0.05
10 (>0.010 mg L'1) 10 (>0.010 mg L'1)
20 (<0.010 mg L'1) 20 (<0.010 mg L'1)
5
5
5
5
10
10
10
10
a %RSD =* percent relative standard deviation. Unless otherwise noted, this is the precision goal
at concentrations greater than or equal to 10 times the required detection limit.
b Precision of accuracy goal in terms of applicable units.
c The mean of six nonconsecutive blank measurements must not exceed 0.9 pS cm"1.
d The absolute value of each laboratory calibration blank measurement was required to be less
than or equal to 10 peq L"1.
364
-------
Table A-2. Estimates of Limits of Detection Based on Analyses of Laboratory Blank Samples,
NSS-I
Analytical laboratoriesa
Variable Units
Detection
Limit
Objective n
Laboratory 1
Estimated
Limit of
Mean SDb Detection0
Laboratory 2
Estimated
Limit of
Mean SD^ Detection0
Al-ext
Al-total
Ca2+
cr
Cond-AL
DICf
DOC
F-
Fe
K+
Mg2+
Mn
Na+
NH4+
NOjf
P
Si02
so42-
mg
mg
mg
mg
us
mg
mg
mg
mg
mg
mg
mg
mg
mg
mg
mg
mg
mg
L'1
L-1
L-1
L-1
cm^X
L-1
L-1
L'1
L'1
L"1
L'1
L-1
L-1
L-1
L-1
L-1
L-1
L'1
0.005
0.005
0.01
0.01
<0.9e
0.05
0.1
0.005
0.01
0.01
0.01
0.01
0.01
0.01
0.005
0.002
0.05
0.05
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
<0.001
0.002
<0.01
0.00d
0.00d
0.00d
<0.1
0.002
<0.01
<0.01
<0.01
<0.01
0.00d
<0.01
o.oood
<0.001
<0.01
<0.01
_
0.0017
-
0.00
0.00
0.000
-
0.0007
-
-
-
-
0.000
-
0.000
-
-
-
<0.001
0.005
0.01
o.ood
_e
o.ood
<0.1
0.002
0.003
<0.01
<0.01
<0.01
o.ood
<0.01
o.oood
0.001
<0.01
<0.01
42
42
42
42
42
42
42
42
42
42
42
42
42
42
42
42
38
42
0.001
0.010
<0.01
<0.01
0.6
<0.01
<0.1
0.000§
<0.01
<0.01
<0.01
<0.01
<0.01
0.00
<0.001
<0.01
0.04
<0.01
0.0007
0.0057
-
-
0.20
-
-
0.0000
-
-
-
-
-
0.001
-
-
0.022
-
0.002
0.017
0.02
0.01
_e
0.07
0.1
0.0108
<0.01
0.02
0.01
<0.01
0.01
0.003
0.009
0.001
0.07
0.04
Processing laboratory3-
n Mean
Estimated
limit of
detection0
Al-mono
Al-Nonexch
Cond-PLh
Cond-PL1
DIC-closed
mg
.mg
fjS
fiS
mg
L'1
cm"1
cm'1
L'1
X
X
0.010
0.010
<0.9
<0.9
0.05
65
65
7
39
58
0.009
0.014
5.6
0.6
0.02
0.0047
0.0064
1.48
0.27
0.011
0.012
0.019
e
e
0.03
a
b
c
d
e
f
g
h
i
n = number of samples; SD = standard deviation.
Dashes indicate that the standard deviation is nearly zero.
Estimated limit of detection = 3SD.
All measurements reported as zero by laboratory.
Detection limit objective expressed as the mean of the blank sample measurements
DIG detection limits apply to both equilibrated and initial measurements.
Laboratory reported all concentrations less than 0.010 mg L"1 as zero. Limit of detection set
at 0.010.
Values for blanks measured in the first 28 batches.
Values for blanks measured in the remaining 40 batches.
365
-------
Table A-3. Estimates of System Decision Limits Based on Analyses of Field Blank Samples
Pooled Across Laboratories, NSS-I
Parametric
System
decision
Variable Units n Mean
Al-ext mg L'1 61 0.002
Al-total mg L"1 61 0.011
Al-mono mg L'1 61 0.010
Al-Nonexch mg L-1 61 0.014
ANC /zeqL'1 58 1.8
BNC /zeq L"1 63 20.8
Ca2+ mgL-1 62 0.01
Cr mg L'1 63 0.01
Cond-PLd /zScro.-1 26 3.5
Cond-PLe juScm-1 33 1.5
Cond-AL /zS cm"1 63 0.6
DIC-air eq mgL-1 63 0.11
DIC-init mg L'1 63 0.13
DOC mg L'1 63 0.2
F- rngL'1 63 NC
Fe mg L'1 63 <0.01
K+ mg L'1 62 <0.01
Mg2+ mgL-1 62 <0.01
Mn mg L"1 63 <0.01
Na+ mg L"1 62 <0.01
NH4+ mgL'1 63 <0.01
NOS- mg L'1 62 0.008
P mg L'1 63 0.001
pH-ANC 58 5.73
pH-BNC 58 5.77
pH-air eq 63 5.96
SiO2 mg L'1 63 0.03
SO42' mg L'1 63 0.01
True color PCU 63 6
Turbidity NTU 63 0.3
a SDLp » mean + 1.65SD.
b Dashes indicate that the standard
SDb limit (SDLp)a
0.0028
0.0081
0.0038
0.0061
4.10
7.99
0.006
0.010
2.59
1.08
0.49
0.075
0.085
0.25
NC
—
—
—
—
—
0.008
0.0112
0.0029
0.155
0.131
0.233
0.043
0.014
3.7
1.36
deviation
c SDLnp = 95th percentile of distribution of
0.007
0.024
0.016
0.024
8.6
34.0
0.020
0.026
7.8
3.2
1.4
0.236
0.276
0.6
0.010f
0.01
0.01
<0.01
0.010
0.01
0.02
0.026
0.006
NC
NC
NC
0.10
0.033
12.1
2.5
(SD) is nearly
n
61
61
61
61
58
63
62
63
26
33
63
63
63
63
63
63
62
62
63
62
63
62
63
58
58
63
63
63
63
63
zero
Nonparametric
System
decision
Median limit (SDLnp)c
0.001
0.010
0.010
0.015
1.5
19.3
<0.01
<0.01
2.0
L2
0.9
0.10
0.13
0.2
NC
<0.01
<0.01
<0.01
<0.01
<0.01
0.01
0.005
<0.001
5.69
5.74
5.89
0.02
0.01
5
0.1
f
0.007
0.027
0.015
0.023
8.7
34.9
0.02
0.03
8.0
2.2
1.2
0.23
0.23
0.5
0.010f
0.02
0.02
<0.01
0.012
0.012
0.32
0.02
0.006
NC
NC
NC
0.09
0.05
10
0.2
field blank measurements.
d Measurements for first half of survey.
e Measurements for second half of
* Laboratorv 2 reDorted all measur
survey.
fid enneen
trations IP.SS tli
inn ft
ftlft mo T.-l a
c -TArn Q^rofom
decision limit estimated as 0.010 mg L"1.
NC « Not calculated.
366
-------
Table A-4. Percent Accuracy Estimates for Laboratory 1 Measurements of Synthetic Audit Samples,
NSS-I
Lot 14a
Variable
Al-ext
Al-total
Ca2+
cr
Cond-AL
DOC
F~
Fe
K+
Mg2+
Mn
Na+
NH4+
N03-
P
Si02
so42-
Accuracy
objec-
Units tive (%)
mg
mg
mg
mg
L"1 10
L"1 10
L"1 10
L'1 10
Theo-
retical
valueb
0.020
0.020
0.19
0.34
/iScm'1 5 17.5
mg
mg
mg
mg
mg
mg
mg
mg
mg
mg
mg
mg
L"1 10
L'1 10
L'1 10
L"1 10
L'1 10
L'1 10
L"1 10
L'1 10
L'1 10
L'1 10
L'1 10
L'1 10
1.0
0.042
0.06
0.20
0.45
0.10
2.75
0.17
0.467
0.027
1.07
2.28
n
4
5
14
13
14
13
13
9
13
14
14
14
13
13
13
14
14
Mean
0.027
0.018
0.22
0.33
16.3
1.0
0.040
0.03
0.20
0.43
0.09
2.82
0.17
0.477
0.021
1.20
2.21
Lot 15a
Percent
accu-
CI racv n Mean CI
0.0254
35ns 3 0.015
0.0063 -10™ 3 0.029
0.026
0.006
0.18
0.06
0.0013
0.011
0.001
0.002
0.004
0.023
0.003
0.0243
0.0025
0.015
0.029
16 4 0.20
afe
-3 4 0.33
*
-7 4 16.0
0 4 1.0
j.
-5 4 0.039
-50* 4 0.03
0 4 0.19
-4 4 0.43
*
-10 4 0.09
2* 4 2.78
0 4 0.16
2 3 0.487
jfc
-22 3 0.015
12ns 4 1.19
-3 4 2.25
Lot 14a
0.0086
0.0226
0.047
0.012
0.20
0.42
0.0043
0.041
0.004
0.008
0.007
0.036
0.011
0.0027
0.0087
0.036
0.028
Lot 15a
Percent
accu-
racy
-25ns
-21ns
5
-3
*
-9
0
-7
-50ns
*
-5
-4
*
-10
1
-6
4*
*
-44
10ns
-1
Accuracy
objective Index6
ANC
BNC
DIC-eq
DIC-init
pH-ANC
pH-BNC
oH-ea
a n =
CI =
ns =
*
(%)
10
10
10
10
O.lc
O.lc
O.lc
number
valuefCn
101.7(3.04)
32.0(4.58)
n
14
13
Mean CI
104.7
22.6
1.14(0.050) 14
1.14(0.050) 14
7.22(0.044) 13
7.22(0.044) 13
7.22C0.044-) 14
of measurements.
1.56
2.10
1.33 0.049
1.41 0.030
7.04 0.035
7.07 0.035
7.17 0.159
Percent
Index®
accuracy value(CI)
3
-29*
17*
24*
-0.18
-0.15
-0.05
109.2(3.07)
22.8(4.04)
1.26(0.050)
1.26(0.050)
*>d 7.29(0.044)
*>d 7.29(0.044)
d 1.29(0.044)
Percent
n Mean
3 104.2
4 18.8
4 1.23
4 1.39
4 7.06
4 7.08
4 7.31
CI accuracy
1.54
2.42
0.081
0.099
0.083
0.078
0.166
-5*
-18ns
_2
10
-0.23*'d
ate J
-0.21 >d
0.02d
one-sided 95% confidence intervals.
not significantly different from the
theoretical or index value at p
= 0.05.
sienificantlv different from the theoretical or index value at p < 0.05.
b The theoretical value is the expected value of the synthetic audit sample assuming no preparation error
and no external effects.
c Objective expressed in pH units.
d Accuracy expressed as the difference between the index value and the mean analytical laboratory value.
e Measured mean values from the support laboratory for Lots 14 and 15.
367
-------
Table A-5. Percent Accuracy Estimates for Laboratory 2 Measurements of Synthetic Audit Samples,
NSS-I
Lot 14a
Lot 15a
Variable
Al-ext
Al-total
Ca2+
cr
Cond-AL
DOC
F-
Fe
K+
Mg2*
Mn
Na+
NH4+
NOS-
P
SiO2
8042"
Accuracy Theo-
objec- retical
Units tive (%) valueb n
mg L-1
mg L'1
mg L'1
mg L'1
ftS cm'1
mg L"1
mg L"1
mg L'1
mg L"1
mg L'1
mg LT1
mg L"1
mgL'1
mg L"1
mg L"1
mg L"1
mg L"1
10
10
10
10
5
10
10
10
10
10
10
10
10
10
10
10
10
0.020
0.020
0.19
0.34
17.5
1.0
0.042
0.06
0.20
0.45
0.10
2.75
0.17
0.467
0.027
1.07
2.28
5
4
9
8
9
9
9
5
9
9
9
8
9
8
9
9
9
Mean
0.021
0.035
0.19
0.32
19.5
1.3
0.043
0.04
0.20
0.44
0.11
2.76
0.19
0.465
0.022
0.92
2.15
Percent
accu-
CI racv n
0.0024
0.0103
0.017
0.018
0.21
0.20
0.0012
0.003
0.006
0.016
0.002
0.062
0.010
0.0140
0.0026
0.044
0.157
5
75*
0*
-6*
11*
30*
2
-33*
0
-2
10
0
12*
0
-18
-10*
-6
23
23
29
28
28
28
28
23
29
28
29
29
28
28
28
28
29
Mean
0.015
0.031
0.18
0.33
19.6
1.1
0.043
0.02
0.20
0.44
0.10
2.70
0.17
0.473
0.023
0.92
2.30
Percent
accu-
CI racv
0.0008
0.0032
0.004
0.011
0.07
0.08
0.0004
0.006
0.004
0.004
0.001
0.038
0.005
0.0132
0.0012
0.024
0.046
-25*
55*
-5*
-3
12*
10*
2*
-67*
0
-2*
0
-2*
0
1
15*
-14*
1
Lot 14a
Lot 15a
Accuracy
objective Index0
ANC
BNC
DIC-eq
DIG-init
(%)
10
10
10
10
pH-ANC O.ld
pH-BNC
oH-ea
a n -
ns a
* 3.
O.ld
O.ld
number
value(CI)
101.7(3.04)
32.0(4.58)
1.14(0.050)
1.14(0.050)
7.22(0.044)
7.22(0.044)
7,22(0,044)
n
9
8
9
9
9
8
9
Mean
114.4
49.1
1.37
1.62
6.73
6.71
7.25
Percent
Indexc
CI accuracy value(CI)
5.73
2.59
0.202
0.074
0.049
0.033
0,069
7*
53*
20ns
42*
0.49*
0.51*
0.03e
of measurements.
not significantly different
mean va
iue significant^
f d
from the
if ferent fi
109.2(3.07)
22.8(4.04)
1.26(0.050)
1.26(0.050)
e 7.29(0.044)
e 7.29(0.044)
7.29(0.044)
theoretical or index value at p
rom the
i index '
value at n < 0.05
Percent
n
29
29
29
29
29
29
29
Mean
119.0
58.9
1.32
1.51
6.69
6.69
7.23
CI accuracy
2.71
4.74
0.082
0.102
0.048
0.049
0.056
10*
158*
5
20*
-0.6*e
-0.6*e
0.06e
= 0.05.
i.
c
d
e
The theoretical value is the expected value of the synthetic audit sample assuming no preparation
error and no external effect.
Measured mean values from the support laboratory for Lot 14 and Lot 15.
Objective expressed in pH units.
Accuracy expressed as the difference between the index value and mean analytical laboratory values.
J68
-------
Table A-6. Method-Level and System-Level Precision Estimates by Concentration Ranges of
Variables (Laboratories Pooled"). NSS-I
(E
la
Variable (units)
and measurement 1
range oi
Al-ext (MIBK) (mg
<0.007
0.007 to 0.050
0.050 to 0.100
>0.100
All data
Al-total (mg L'1)
<0.027
0.027 to 0.100
0.100 to 0.500
0.500 to 1.000
> 1.000
All data
Al-mono (PCV) (mg
<0.015
0.015 to 0.100
0.100 to 0.500
0.500 to 1.000
> 1.000
All data
Al-Nonexch. (PCV)
<0.023
0.023 to 0.100
0.100 to 0.500
All data
ANC (/zeq L"1)
<0
0 to 50
>50
All data
BNC Oieq L"1)
0 to 50
>50
All data
Method-level precision
>rocessing and analytical
.boratory duplicate pairs)
Slumber Grand
f pairs mean
L-1)
3
35
15
15
68
9
6
51
1
1
68
L-1)
33
24
7
2
1
67
(mg
52
12
3
67
2
4
62
68
24
44
68
0.004
0.019
0.071
0.269
0.085
0.014
0.056
0.204
0.791
1.235
0.189
I
0.010
0.026
0.273
0.534
1.908
0.087
L'1)
0.014
0.043
0.225
0.029
-6.8
21.9
401.2
366.9
32.6
93.4
71.9
Pooled
SD
0.0002
0.0009
0.0026
0.0055
0.0029
0.0018
0.0034
0.0042
0.0304
0.0573
0.0087
0.0011
0.0017
0.0033
0.0045
0.0030
0.0019
0.0022
0.0033
0.0139
0.0039
0.70
1.05
4.42
4.23
2.68
6.72
5.64
%RSDpa
5.3
4.6
3.6
2.0
3.4
13.1
6.1
2.0
3.8
4.6
4.6
11.0
6.4
1.2
0.8
0.2
2.1
5.8
7.6
6.2
3.4
—
—
—
—
—
—
—
System-level precision
(field routine-duplicate pairs)
Number Grand
of pairs mean
29
25
4
7
65
4
16
38
7
0
65
30
29
6
0
0
65
47
18
0
65
7
9
49
65
28
37
65
0.003
0.016
0.072
0.278
0.042
0.019
0.056
0.214
0.767
—
0.222
0.010
0.031
0.283
—
—
0.045
0.016
0.042
—
0.023
-42.4
27.9
425.0
319.7
36.2
111.1
78.8
Pooled
SD
0.0016
0.0059
0.0021
0.0166
0.0067
0.0035
0.0079
0.0807
0.1953
—
0.0905
0.0022
0.0038
0.0049
__
__
0.0033
0.0031
0.0068
—
0.0045
2.94
7.83
11.22
10.22
6.38
12.25
10.14
%RSDpa
51.5
37.2
2.9
6.0
16.1
18.7
14.1
37.8
25.5
—
40.7
21.4
12.4
1.7
__
—
7.4
19.2
16.2
—
19.2
--
__
-_
—
—
—
—
369
-------
Table A-6 (Continued)
ft
la
Variable (units)
and measurement 1
range ol
Ca2+ (mg L'1)
0,02 to 1.00
1.00 to 5.00
5.00 to 10.00
>10.00
All data
Cr (mg L-1)
<0.03
0.03 to 1.00
1.00 to 2.00
2.00 to 5.00
5.00 to 10.00
>10.00
All data
Cond-PL (/zS cm"1)
<25.0
25.0 to 50
50.0 to 100.0
>100.0
All data
Cond-AL (/iS cm-1)
<25.0
25.0 to 50.0
50.0 to 100.0
>100.0
All data
DIC-closed-sys (mg
<1.00
1.00 to 2.00
2.00 to 5.00
5.00 to 10.00
>10.00
All data
Method-level precision
>rocessing and analytical
.boratory duplicate pairs)
Slumber Grand
F pairs mean
6
35
12
15
68
2
13
13
23
11
6
68
5
22
21
19
67
14
20
17
17
68
L"1)
9
8
26
18
7
68
0.59
2.37
7.06
25.59
8.16
0.00
0.57
1.49
3.18
6.73
5.89
3.96
20.1
37.8
75.0
244.3
106.7
19.9
35.6
69.2
206.9
83.6
0.60
1.56
3.42
6.50
31.79
6.56
Pooled
SD
0.009
0.041
0.034
0.529
0.251
0.000
0.006
0.022
0.045
0.077
0.336
0.108
1.11
1.94
5.27
16.99
9.59
0.12
0.23
0.29
0.49
0.31
0.025
0.040
0.068
0.168
0.236
0.123
System-level precision
(field routine-duplicate pairs)
Number Grand Pooled
%RSDpa of pairs mean SD
1.5
1.7
0.5
2.1
3.1
—
1.1
1.5
1.4
1.1
2.1
2.7
5.5
5.1
7.0
7.0
9.0
0.6
0.6
0.4
0.2
0.4
4.2
2.5
2.0
2.6
0.7
1.9
4
35
19
7
65
0
8
18
24
6
9
65
5
24
24
12
65
6
24
24
11
65
7
9
31
13
5
65
0.56
2.55
6.82
28.90
6.51
—
0.69
1.53
3.05
7.46
18.21
4.85
20.4
37.7
70.1
192.0
76.8
20.6
36.9
69.9
197.4
74.7
0.46
1.43
3.37
6.65
24.47
5.07
0.015
0.057
0.090
1.604
0.530
—
0.018
0.033
0.043
0.073
0.734
0.276
0.26
1.26
1.32
3.12
1.75
0.87
0.27
0.42
1.22
0.64
0.020
0.063
0.092
0.184
0.365
0.147
%RSDpa
2.7
2.2
1.3
5.6
8.1
—
2.6
2.2
1.4
1.0
4.0
5.7
1.3
3.3
1.9
1.6
2.3
4.2
0.7
0.6
0.6
0.9
4.3
4.4
2.7
2.8
1.5
2.9
370
-------
Variable (units)
and measurement
range
DIC-air eq (mg L"
<0.23
0.23 to 1.00
1.00 to 2.00
2.00 to 5.00
5.00 to 10.00
> 10.00
All data
Method-level precision
(processing and analytical
laboratory duplicate pairs)
Number
of pairs
i)
1
4
8
24
18
13
68
Grand
mean
0.07
0.74
1.69
3.60
7.27
19.21
7.11
Pooled
SD
0.002
0.029
0.031
0.067
0.086
0.639
0.286
%RSDpa
2.9
4.0
1.8
1.9
1.2
3.3
4.0
System-level precision
(field routine-duplicate pairs)
Number
of pairs
10
12
12
22
5
4
65
Grand
mean
0.12
0.55
1.50
3.49
6.90
26.22
3.72
Pooled
SD
0.054
0.099
0.162
0.261
0.184
1.050
0.317
%RSD a
45.3
18.1
10.8
7.5
2.7
4.0
8.5
DIC-initial AL (mg L"1)
<0.23
0.23 to 1.00
1.00 to 2.00
2.00 to 5.00
5.00 to 10.00
> 10.00
All data
DOC (mg L'1)
<0.5
0.5 to 2.0
2.0 to 5.0
5.0 to 10.0
>10.0
All data
F- (mg L-1)
0.010 to 0.050
>0.050
All data
Fe (mg L'1)
<0.02
0.02 to 0.05
0.05 to 0.10
0.10 to 0.50
0.50 to 1.00
>1.00
All data
0
1
17
18
18
14
68
4
27
16
14
7
68
29
39
68
2
3
11
36
3
12
67
—
0.24
1.60
3.42
7.22
18.82
7.09
0.3
1.2
3.3
7.2
17.2
4.5
0.037
0.121
0.085
0.01
0.02
0.08
0.19
0.72
6.85
1.38
—
0.007
0.036
0.111
0.139
0.780
0.366
0.02
0.05
0.11
0.15
0.15
0.10
0.0000
0.0021
0.0016
0.001
0.001
0.002
0.004
0.008
0.114
0.048
—
2.9
2.3
3.2
1.9
4.1
5.2
6.5
4.0
3.3
2.1
0.9
2.3
0.0
1.7
1.9
7.4
2.7
3.0
2.2
1.2
1.7
3.5
3
10
15
26
6
5
65
2
41
15
7
0
65
51
14
65
14
18
10
18
3
2
65
0.17
0.56
1.50
3.39
6.17
24.46
4.24
0.3
1.2
3.1
7.2
__
2.23
0.033
0.082
0.044
0.01
0.03
0.08
0.22
0.64
1.55
0.16
0.014
0.091
0.211
0.250
0.050
1.009
0.339
0.02
0.24
0.44
0.58
_«
0.34
0.0011
0.0029
0.0017
0.003
0.009
0.029
0.117
0.163
0.377
0.098
8.2
16.3
14.0
7.4
0.8
4.1
8.0
6.1
20.8
14.3
8.1
__
15.4
3.4
3.5
3.8
66.0
34.0
34.3
53.3
25.6
24.4
61.4
371
-------
Variable (units)
and measurement
range
K+ (mg L"1)
<0.15
0.15 to 0.35
0.35 to 0.45
>0.45
All data
Mg2+ (mg L'1)
<1.00
1.00 to 2.00
2.00 to 5.00
>5.00
All data
Mn (mg L'1)
<0.01
0.01 to 0.05
0.05 to 0.10
>0.10
All data
Na+ (mg L-1)
<0.50
0.50 to 1.00
1.00 to 2.00
2.00 to 5.00
>5.00
All data
NH4+ (mg L'1)
<0.02
0.02 to 0.05
0.05 to 0.10
>0.10
All data
NOS- (mg L-1)
<3.000
>3.000
All data
Method-level precision
(processing and analytical
laboratory duplicate pairs)
Number
of pairs
1
10
2
55
68
26
21
15
6
68
2
9
12
44
67
4
8
11
33
12
68
12
20
14
22
68
55
13
68
Grand
mean
0.11
0.26
0.39
1.47
1.24
0.54
1.47
3.42
18.15
3.02
<0.01
0.03
0.08
1.00
0.68
0.23
0.73
1.47
3.40
10.27
3.80
0.02
0.03
0.07
0.35
0.14
0.722
8.182
2.148
Pooled
SD
0.001
0.006
0.002
0.015
0.014
0.004
0.013
0.033
0.158
0.050
0.001
0.001
0.003
0.017
0.014
0.001
0.027
0.017
0.028
0.174
0.076
0.000
0.001
0.002
0.004
0.003
0.0162
0.1212
0.0549
%RSDpa
1.3
2.5
0.6
1.0
1.1
0.8
0.9
1.0
0.9
1.7
11.8
2.1
3.5
1.7
2.0
0.6
3.7
1.1
0.8
1.7
2.0
0.0
3.2
3.1
1.2
1.9
2.2
1.5
2.6
System-level precision
(field routine-duplicate pairs)
Number
of pairs
1
5
6
53
65
19
23
18
5
65
19
19
7
20
65
4
4
24
25
8
65
33
21
7
4
65
53
12
65
Grand
mean
0.03
0.26
0.40
1.45
1.24
0.68
1.48
3.04
9.39
2.29
<0.01
0.02
0.07
0.24
0.09
0.28
0.84
1.42
3.52
9.66
3.14
0.01
0.03
0.07
0.17
0.03
0.686
14.885
3.307
Pooled
SD
0.003
0.003
0.005
0.063
0.056
0.013
0.014
0.058
0.242
0.075
—
0.015
0.002
0.023
0.015
0.001
0.030
0.028
0.037
0.097
0.045
0.006
0.007
0.014
0.011
0.008
0.0422
0.3239
0.1443
%RSDpa
8.3
1.0
1.3
4.3
4.5
2.0
0.9
1.9
2.6
3.3
56.6
65.0
2.7
9.5
16.8
0.5
3.6
1.9
1.1
1.0
1.4
42.0
23.4
19.7
6.6
22.8
6.2
2.2
4.4
372
-------
Variable (units)
and measuremei
range
P-tot dissolved
<0.001
Method-level precision
(processing and analytical
laboratory duplicate pairs)
nt
Number
of pairs
(mg L-1)
2
0.001 to 0.005 24
0.005 to 0.015 20
>0.015
All data
pH-closed sys
<4.00
4.00 to 5.00
5.00 to 6.00
6.00 to 7.00
7.00 to 8.00
>8.00
All data
pH-initial ANC
4.00 to 5.00
5.00 to 6.00
6.00 to 7.00
7.00 to 8.00
>8.00
All data
pH-initial BNC
4.00 to 5.00
5.00 to 6.00
6.00 to 7.00
7.00 to 8.00
>8.00
All data
pH-air eq
<4.00
4.00 to 5.00
5.00 to 6.00
6.00 to 7.00
7.00 to 8.00
>8.00
All data
22
68
4
4
7
25
26
2
68
(pH units)
1
5
34
28
0
68
0
6
34
28
0
68
2
7
4
3
34
18
68
Grand
mean
-0.001
0.003
0.008
0.105
0.037
3.75
4.51
5.61
6.63
7.43
8.26
6.59
4.98
5.73
6.69
7.25
—
6.82
—
5.61
6.70
7.25
—
6.83
3.73
4.55
5.55
6.86
7.55
8.45
7.22
Pooled
SD %RSDDa
0.0003 30.5
0.0003 8.6
0.0004 4.6
0.0018 1.7
0.0010 2.8
0.000
0.012
0.041
0.012
0.028
0.010
0.023
0.014
0.021
0.053
0.039
—
0.046
— —
0.034
0.042
0.040
—
0.041
0.005
0.009
0.000
0.004
0.019
0.028
0.020
System-level precision
(field routine-duplicate pairs)
Number
of pairs
7
31
20
7
65
0
6
13
24
19
3
65
6
8
27
22
2
65
6
8
26
23
2
65
0
6
4
11
40
4
65
Grand
mean
-0.001
0.003
0.008
0.025
0.006
.._
4.55
5.69
6.66
7.26
8.45
6.53
4.49
5.70
6.63
7.31
8.10
6.59
4.50
5.69
6.63
7.31
8.12
6.60
__
4.53
5.73
6.73
7.48
8.37
7.03
Pooled
SD %RSDpa
0.0009 62.9
0.0016 53.7
0.0034 43.8
0.0040 16.1
0.0026 40.5
— — __
0.032
0.036
0.036
0.036
0.026
0.035
0.022
0.104
0.073
0.091
0.034
0.080
0.035
0.092
0.065
0.086
0.062
0.075
•
0.017
0.041
0.113
0.171
0,083
0.144
373
-------
Table A-6 (Continued)
Variable (units)
Method-level precision
(processing and analytical
laboratory duplicate pairs)
System-level precision
(field routine-duplicate pairs)
and measurement
range
SiO2 (mg IT1)
<0.50
0.50 to 1.50
1.50 to 5.50
5.50 to 10.50
>10.50
All data
SO^2- (mg L'1)
<0.05
0.05 to 2.50
2.50 to 5.00
5.00 to 12.50
> 12.50
All data
True color (PCU)
<30
>30
All data
Turbidity (NTU)
<2.0
2.0 to 20.0
20.0 to 100.0
All data
Number
of pairs
1
1
29
19
18
68
2
13
18
26
9
68
50
17
67
28
38
1
67
Grand
mean
0.44
1.03
3.71
7.73
16.38
8.10
<0.01
1.43
3.60
7.85
22.56
7.21
13
12
43
1.1
8.1
33.5
5.6
Pooled
SD
0.007
0.035
0.052
0.141
0.195
0.129
0.002
0.019
0.048
0.117
0.239
0.116
0.9
98.7
4.4
0.07
0.26
0.71
0.22
%RSDpa
1.6
3.4
1.4
1.8
1.2
1.6
66.7
1.3
1.3
1.5
1.1
1.6
6.5
6.8
10.5
6.5
3.2
2.1
3.9
Number
of pairs
0
1
25
21
18
65
0
18
10
24
13
65
44
21
65
27
36
2
65
Grand
mean
—
0.57
3.31
7.70
14.51
7.79
—
1.49
3.86
8.38
21.28
8.36
13
50
25
1.0
7.7
24.0
5.4
Pooled
SD
—
0.064
0.071
0.122
0.730
0.393
--
0.054
0.081
0.128
0.243
0.140
2.6
1.9
2.4
0.08
0.39
1.41
0.38
%RSDpa
—
11.3
2.2
1.6
5.0
5.0
-—
3.6
2.1
1.5
1.1
1.7
19.6
3.9
9.6
8.6
5.0
5.9
7.1
a %RSD0 » relative pooled standard deviation, calculated as (pooled SD :-grand mean) x 100.
374
-------
Table A-7. Comparison of Method-Level, System-Level, and Among-Batch Precision Estimates
NSS-I
Within-batch
Variable (units)
and measurement
range
Al-ext (MIBK) (mg
<0.007
0.007 to 0.050
0.050 to 0.100
>0.100
Al-total (mg L'1)
<0.027
0.027 to 0.100
0.100 to 0.500
0.500 to 1.000
>1.000
Al-mono (PCV) (mg
<0.015
0.015 to 0.10
0.10 to 0.50
0.50 to 1.00
>1.00
Method-levela
Number
of pairs
L'1)
3
35
15
15
9
6
51
1
1
L'1)
33
24
7
2
1
Pooled
SD
0.0002
0.0009
0.0026
0.0055
0.0018
0.0034
0.0042
0.0304
0.0573
0.0011
0.0017
0.0033
0.0045
0.0030
System-level1*
Number Pooled
of pairs SD
29
25
4
7
4
16
38
7
0
30
29
6
0
0
0.0016
0.0059
0.0021
0.0166
0.0035
0.0079
0.0807
0.1953
—
0.0022
0.0038
0.0049
__
—
Among-batchc
Number
samples
38
9
« w
37
38
19
37
__
—
13
24
—
of
SD
0.0034
0.0106
0.0329
0.0399
0.0222
0.0255
__
—
0.0074
0.0117
—
Sample
tvped
BL
s
BM
BL
S
BM
S
BM
Al-Nonexch (PCV) (mg L'1)
<0.023
0.023 to 0.10
0.10 to 0.50
ANC (neq L'1)
<0
Oto 50
>50
BNC (A*eq L'1)
0 to 50
>50
Ca2+ (mg L'1)
0.02 to 1.00
1.00 to 5.00
5.00 to 10.00
>10.00
52
12
3
2
4
62
24
44
6
35
12
15
0.0022
0.0033
0.0139
0.70
1.05
4.42
2.68
6.72
0.009
0.041
0.034
0.529
47
18
0
7
9
49
28
37
4
35
19
7
0.0031
0.0068
—
2.94
7.83
11.22
6.38
12.25
0.015
0.057
0.090
1.604
13
24
—
39
__
38
38
39
56
39
__
__
0.0080
0.0127
4.88
__
11.73
23.55
11.34
0.031
0.077
__
S
BM
BM
BL
BL
BM
S
BM
375
-------
Table A-7 (continued1)
Within-batch
Variable (units)
and measurement I
ranse of
Cl- (mg L'1)
<0.03
0.03 to 1.00
1.00 to 2.00
2.00 to 5.00
5.00 to 10.00
>10.00
Cond-PL (pS cm'1)
<25.0
25.0 to 50.0
50.0 to 100.0
>100.0
Cond-AL 0*S cm-1)
<25.0
25.0 to 50.0
50.0 to 100.0
>100.0
Method-levela
dumber Pooled
r pairs SD
2
13
13
23
11
6
5
22
21
19
14
20
17
17
0.000
0.006
0.022
0.045
0.077
0.336
1.11
1.94
5.27
16.99
0.12
0.23
0.29
0.49
System-level13
Number Pooled
of pairs SD
0
8
18
24
6
9
5
24
24
12
6
24
24
11
—
0.018
0.033
0.043
0.073
0.734
0.26
1.26
1.32
3.12
0.87
0.27
0.42
1.22
Among-batchc
Number of Sample
samples SD tvped
--
39
—
—
—
— —
12
26
—
— —
38
39
—
—
— —
0.051
--
—
—
— —
8.35
1.76
—
--
1.91
0.89
—
--
BM
S
BM
BL
BM
DIC-closed sys (mg L"1)
<1.00
1.00 to 2.00
2.00 to 5.00
5.00 to 10.00
> 10.00
DIC-air eq (mg L'1)
<0.23
0.23 to 1.00
1.00 to 2.00
2.00 to 5.00
5.00 to 10.00
>10.00
DIC-initial AL (mg L
<0.23
0.23 to 1.00
1.00 to 2.00
2.00 to 5.00
5.00 to 10.00
>10.00
9
8
26
18
7
1
4
8
24
18
13
,-1)
0
1
17
18
18
14
0.025
0.040
0.068
0.168
0.236
0.002
0.029
0.031
0.067
0.086
0.639
—
0.007
0.036
0.111
0.139
0.780
7
9
31
13
5
10
12
12
22
5
4
3
10
15
26
6
5
0.020
0.063
0.092
0.184
0.365
0.054
0.099
0.162
0.261
0.184
1.050
0.014
0.091
0.211
0.250
0.050
1.009
27
27
—
—
-—
39
—
38
—
—
— —
—
39
38
—
—
—
0.062
0.061
—
—
— —
0.078
—
0.203
—
—
— —
--
0.070
0.214
—
—
— —
BM
BL
BM
BL
BM
BL
376
-------
Within-batch
Variable (units)
and measurement
ranee
DOC (mg L'1)
<0.5
0.5 to 2.0
2.0 to 5.0
5.0 to 10.0
>10.0
F- (mg L"1)
0.010 to 0.050
>0.050
Fe (mg L'1)
<0.02
0.02 to 0.05
0.05 to 0.10
0.10 to 0.50
0.50 to 1.00
>1.00
K+ (mg L'1)
<0.15
0.15 to 0.35
0.35 to 0.45
>0.45
Mg2+ (mg L"1)
<1.00
1.00 to 2.00
2.00 to 5.00
>5.00
Mn (mg L"1)
<0.01
0.01 to 0.05
0.05 to 0.10
>0.10
Na+ (mg L'1)
<0.50
0.50 to 1.00
1.00 to 2.00
2.00 to 5.00
>5.00
Method-level3
Number
of pairs
4
27
16
14
7
29
39
2
3
11
36
3
12
1
10
2
55
26
21
15
6
2
9
12
44
4
8
11
33
12
Pooled
SD
0.02
0.05
0.11
0.15
0.15
0.0000
0.0021
0.001
0.001
0.002
0.004
0.008
0.114
0.001
0.006
0.002
0.015
0.004
0.013
0.033
0.158
0.001
0.001
0.003
0.017
0.001
0.027
0.017
0.028
0.174
mmucm
System-level5
Number
of oairs
2
41
15
7
0
51
14
14
18
10
18
3
2
1
5
6
53
19
23
18
5
19
19
7
20
4
4
24
25
8
Pooled
SD
0.02
0.24
0.44
0.58
—
0.0011
0.0029
0.003
0.009
0.029
0.117
0.163
0.377
0.003
0.003
0.005
0.063
0.013
0.014
0.058
0.242
__
0.015
0.002
0.023
0.001
0.030
0.028
0.037
0.097
Among-batchc
Number
samples
39
56
39
«
—
39
39
39
42
39
__
—
__
39
39
—
56
• w
^^
--
39
w w
39
56
w«
39
mf ^
56
_ —
of
SD
0.42
0.27
0.28
,
--
0.0033
0.0024
0.017
0.018
0.017
--
0.016
0.022
0.032
__
—
0.004
0.020
0.009
__
0.042
0.099
Sample
typed
BL
s
BM
BL
BM
BL
S
BM
BL
BM
S
BL
BM
S
BL
S
377
-------
Within-batch
Variable (units)
and measurement
ranee <
NH4+ (mg L'1)
T* *
<0.02
0.02 to 0.05
0.05 to 0.10
>0.10
NO." (mg L'1)
w >
<3.000
>3.000
P-tot dissolved (mg ]
<0.001
0.001 to 0.005
0.005 to 0.015
>0.015
pH-closed sys
<4.00
4.00 to 5.00
5.00 to 6.00
6.00 to 7.00
7.00 to 8.00
>8.00
pH-initial ANC
4.00 to 5.00
5.00 to 6.00
6.00 to 7.00
7.00 to 8.00
>8.00
pH-initial BNC
4.00 to 5.00
5.00 to 6.00
6.00 to 7.00
7.00 to 8.00
>8.00
Method-levela
Number Pooled
of oairs SD
12
20
14
22
55
13
L'1)
2
24
20
22
4
4
7
25
26
2
1
5
34
28
0
0
6
34
28
0
0.000
0.001
0.002
0.004
0.0162
0.1212
0.0003
0.0003
0.0004
0.0018
0.000
0.012
0.041
0.012
0.028
0.010
0.014
0.021
0.053
0.039
—
—
0.034
0.042
0.040
—
System-level5
Number Pooled
of oairs SD
33
21
7
4
53
12
7
31
20
7
0
6
13
24
19
3
6
8
27
22
2
6
8
26
23
2
0.006
0.007
0.014
0.011
0.0422
0.3239
0.0009
0.0016
0.0034
0.0040
—
0.032
0.036
0.036
0.036
0.026
0.022
0.104
0.073
0.091
0.034
0.035
0.092
0.065
0.086
0.062
Among-batchc
Number of Sample
samoles SD tvoed
39
—
39
56
39
—
39
39
—
53
—
—
27
14
27
—
—
39
56
38
—
39
56
38
—
0.010
—
0.009
0.015
0.0553
-—
0.0014
0.0022
—
0.0039
— —
—
0.059
0.085
0.049
— —
--
0.159
0.187
0.100
—
—
0.161
0.200
0.122
—
BL
BM
S
BM
BL
BM
S
BM
S
BL
BM
S
BL
BM
S
BL
378
-------
Table A-7 (continued1)
Within-batch
Variable (units)
and measurement
ranee
pH-air eq
<4.00
4.00 to 5.00
5.00 to 6.00
6.00 to 7.00
7.00 to 8.00
>8.00
Si02 (mg L-1)
<0.50
0.50 to 1.50
1.50 to 5.50
5.50 to 10.50
>10.50
Method-level1
Number
of oairs
2
7
4
3
34
18
1
1
29
19
18
Pooled
SD
0.005
0.009
0.000
0.004
0.019
0.028
0.007
0.035
0.052
0.141
0.195
System- level'5
Number
of oairs
0
6
4
11
40
4
0
1
25
21
18
Pooled
SD
__
0.017
0.041
0.113
0.171
0.083
— _
0.064
0.071
0.122
0.730
Among-batchc
Number
samples
__
39
.._
56
—
__
56
39
39
__
of Sample
SD tvppd
0.071 BM
0.179
0.143
S
S
0.419 BM
0.537 BL
S042- (mg L-1)
<0.05 2 0.002
0.05 to 2.50 13 0.019
2.50 to 5.00 18 0.048
5.00 to 12.50 26 0.117
>12.50 9 0.239
True color (PCU)
<30 50 0.9
>30 17 8.7
0
18
10
24
13
44
21
0.054
0.081
0.128
0.243
2.6
1.9
56 0.131 S
39 0.167 BM
25 4.6 BM
turbidity (NTU)
<2.0
2.0 to 20.0
20.0 to 100.0
28
38
1
0.07
0.26
0.71
27
36
2
0.08
0.39
1.41
—
— — ...
a
b
c
d
SD
Method-level = within-batch precision calculated from laboratory routine-duplicate sample pairs.
System-level - overall within-batch precision calculated from field routine-duplicate sample pairs.
Among-batch = among-batch precision (across laboratories) calculated from audit sample measurements.
Audit sample types from which precision estimate was derived: S = synthetic audit sample, BL = Bagley
Lake sample, and BM = Big Moose Lake sample.
standard deviation.
379
-------
Table A-8. Summary of Data Quality Assessment for Chemical Variables with Respect to
Detectability, Accuracy, and Precision, NSS-Ia.
Variable
Detectability
Accuracy
Precision
Al-ext (MIBK) +
Al-total +
Al-mono (PCV) +
Al-Nonexch (PCV) -b
ANC +
BNC -b
Ca2+ +
cr +
Cond-PL -d
Cond-AL +
DIC-closed sys +
DIC-air eq +
DIC-initial AL +
DOC +
Fe +
K+ +
Mg2+ +
Mn +
NH/ +
NOS- +
O
P-tot dissolved +
pH-closed sys +
pH-initial ANC +
pH-initial BNC +
pH-air eq +
SiO2 +
True color NE
Turbidity NE
+ _b
_b _c
+ -b
+ _b,e
+ +
_b _b
+ +
+ -
_d,e
_b +
+ +
+ +
+ +
_b _b,c
_b _b,c
+ +
+ +
I !b,c
+• +
_e _b,c
+ +
+ +
+ +
+ +
-b +
+ +
NE +
a + s acceptable in terms of data quality objective or primary project objectives.
- «• estimate not near data quality objective.
NE « not evaluated.
k Possible limitations at low concentrations.
c Overall within-batch precision estimate is larger than among-batch precision estimate.
d Estimates not of acceptable quality.
e Possible limitations at high concentrations.
380
-------
APPENDIX B
PREDICTION OF SPRING UPPER REACH NODE CHEMISTRY
IN THE SOUTHERN BLUE RIDGE SUBREGION (2As) DATA SET (PILOT SURVEY)
The Pilot Survey component of the NSS-I was implemented in the Southern Blue Ridge in
1985, a year before the full NSS-I survey, in order to test the logistics and sampling design of
the NSS-I before initiating the full-scale survey effort (Messer et al., 1986). The Pilot Survey
was originally designed to index stream chemistry using only lower reach node measurements. In
the middle of the Pilot Survey spring sampling period, it became apparent that at least two
points along a reach had to be sampled to characterize its chemistry. In addition, it was
obvious that the upper reach nodes represented an important subpopulation for the determination
of acid deposition impacts. For these reasons, upper reach node sampling was initiated. The
end result, however, was an incomplete spring data set (20 of 54 sites) for the upper reach
nodes. A full set of both upper and lower reach node chemical data was collected in the
summer season. For this report, it was desirable to compare population estimates and medians
among the nine different NSS-I subregions using data collected in the same season (e.g., not
comparing Southern Blue Ridge summer values to Piedmont spring values). Thus, predictive
relationships were used to calculate chemical data for the missing 34 spring upper nodes in the
Southern Blue Ridge.
The missing spring upper reach node values for 22 chemical variables were calculated using
multiple regression techniques. Using data from the 20 sites with measured spring upper reach
node chemistry, a multiple regression model was constructed for each chemical variable, relating
the observed spring upper reach node chemistry to the observed summer (upper and lower reach
nodes) and spring lower reach node chemistry. The regression model was then used to predict
the missing spring upper reach node chemical values, based on the measured summer upper,
summer lower, and spring upper reach node values. For each chemical variable, all possible
combinations of regression models (spring lower reach nodes and summer upper and lower reach
nodes) were compared. The final model for each variable included those seasons and nodes that
contributed significantly to the explained variation (r2), without causing multi-colinearity among
the predictors.
One of the 20 upper reach nodes with measured spring upper chemistry was excluded from
the regression as an outlier (stream 2A07803) for ANC, Ca2+, Mg2+, specific conductivity, DIG,
and HCO3~ because it had concentrations of these species roughly an order of magnitude higher
than those present in the other 19 samples (ANC = 2,200 /zeq L'1, Ca2+ = 1,840 jteq L"1, specific
conductivity = 224 pS cm'1), and that one point tended to dominate the regression. The regres-
sion based on the remaining 19 data points (Table B-l) was used to calculate all of the missing
34 spring upper reach node data points except for those of two streams (2A08801 and 2A07702).
These two streams had high ion concentrations, so their spring upper reach node data points
were calculated based on the regression equations that included the high concentration outlier
that had been removed to predict the other 32 upper nodes.
The regression equations used to calculate the spring upper node chemical variables (after
removing the one high ionic concentration outlier) are presented in Table B-l. In general, the
summer upper reach node chemistry was the best predictor of spring upper reach node chemical
conditions. The best model was usually obtained by considering summer upper and spring lower
reach node chemistry. In most cases the summer lower reach node data was not added to the
model because it had a high degree of colinearity with the other predictors.
381
-------
Table B-1. Regression Coefficients Used to Calculate the Missing Spring Upper Reach Node
Chemical Data in the Southern Blue Ridge Subregion (Pilot Survey)
REGRESSION COEFFICIENTS
Variable
Al-ext (MIBK)
Summer
Upper
0.3325
Al-org (PCV) -1.5767
ANC
Ca2+
cr
Specific Cond.
DIC-initial lab
DOC
Fe
F"
H+
HCOS-
K+
Mg2+
Mn
Na+
NH4
NOjf
o
pH
P-total
Si02
so42-
Regression Equations
ANC
Ca2+
Specific conductivity
DIC-initial lab
HCOS-
Mg2+
0.1347
0.1917
0.6171
0.4902
0.1660
0.0852
0.4139
0.5682
0.1806
0.1199
0.1401
0.3266
0.1068
0.1909
0.1835
0.4753
0.5000*
0.1540
0.1977
0.5236
Including
0.4208
0.3742
0.7621
0.3588
0.3032
0.6482
Spring
Lower
_.
3.9489
0.6002
0.5488
—
0.1705
0.6395
0.2826
—
0.2087
0.2830
0.6524
0.5395
0.3572
0.1481
0.0322
-.2563
0.5153
0.5000*
—
0.7458
0.3817
Summer
Lower
___
0.5959
—
—
—
—
—
0.1878
—
—
0.4050
—
—
—
—
0.3011
-.1119
0.1322
—
—
—
—
Model
Intercept
0.0657
-1.316
7.7087
-1.3859
7.3103
3.0204
0.0320
0.1273
0.1227
0.1706
0.0477
2.5846
3.6857
4.5925
0.0166
27.6296
1.0634
-3.2143
0*
0.5421
9.4919
-5.18
R2
0.702
0.835
0.952
0.949
0.944
0.939
0.950
0.692
0.722
0.986
0.468
0.960
0.949
0.971
0.671
0.752
0.339
0.948
#
0.459
0.807
0.858
RMSE&
0.05
0.25
26.96
20.56
13.21
4.60
0.35
0.21
0.18
0.13
0.07
23.41
2.24
6.59
0.05
28.11
0.58
3.09
#
<0.01
27.67
12.71
Outlier Stream 2A07803$
0.5206
0.5758
—
0.7450
0.7966
0.0816
—
—
—
—
—
—
-37.34
-27.91
0.0556
-0.6812
-48.63
0.0085
Regression Model: Spring upper value - b^summer upper value) + b2(spring lower value) +
bs(summer lower value) + intercept.
— Indicates the best regression model did not include this predictor.
* Due to colinearity problems, the spring upper reach node pH was predicted as the mean of the
summer upper and spring lower pH value.
& RMSE is the model root mean square error (square root of the mean squared residual).
$ These equations were used only to predict spring upper reach node stream chemistry in two
streams, 2A08801 and 2A07702; see text for explanation.
382
-------
APPENDIX C
POPULATION ESTIMATES AND STANDARD ERRORS
FOR ANC AND pH REFERENCE RANGES
Tables C-l and C-2 present the population estimates and standard errors of the number of
upstream and downstream reach nodes within specified ranges of ANC (Table C-l) and pH (Table
C-2). Tables C-3 and C-4 present the population estimates and standard errors of the combined
length and percentage of NSS-I target stream reaches with spring basef low ANC (Table C-3) and
pH (Table C-4) less than reference values. Standard errors were calculated as the square root
of the variance associated with each population estimate. The equations and theories involved
in the calculations are discussed in Section 2.4 and are described in detail by Overton (1987).
383
-------
Table C-1. Population Estimates of the Number (and standard Error) of Target Stream Lower
(L) and Upper (U) Reach Nodes with Spring Index ANC (0eq L"1) in Reference
Ranges
SUBREGION
Poc./Cats.
(ID)
N. Appal.
(2Cn)
Val. & Ridge
(2Bn)
MA Coast. PL
(3B)
S. Blue Ridge
(2As)
Piedmont
(3A)
S. Appal.
(2X)
Oz./Ouach.
(2D)
Florida
(3C)
<0
L U
* 209
(-) (102)
326 499
(206) (250)
* 636
(-) (445)
772 1334
(362) (524)
* *
* *
* 121
(-) (121)
* *
225 678
(107) (344)
>0 -
L
157
(85)
1669
(612)
47
(47)
1433
(565)
95
(44)
632
(308)
364
(206)
*
670
(240)
- 50
U
528
(215)
2421
(712)
808
(462)
2069
(652)
127
(47)
632
(308)
364
(206)
150
(105)
531
(233)
ANC Range
>50 - 200
L
802
(239)
3244
(685)
4190
(1033)
2420
(711)
1494
(328)
2369
(546)
1027
(346)
2700
(376)
368
(185)
U
831
(217)
2764
(614)
3554
(969)
2890
(767)
1576
(328)
3001
(593)
971
(323)
2700
(376)
131
(63)
>200
L
2276
(331)
3249
(631)
9755
(1283)
6662
(1041)
441
(130)
4514
(672)
3666
(531)
1416
(375)
292
(92)
U
1676
(311)
2979
(616)
8040
(1260)
4991
(961)
328
(120)
3882
(653)
3480
(538)
1354
(373)
387
(189)
TOTALS
L
3235
(347)
8488
(814)
13,992
(1213)
11,287
(1078)
2031
(326)
7515
(650)
5057
(526)
4116
(410)
1555
(306)
U
3244
(347)
8663
(807)
13,038
(1249)
11,284
(1078)
2031
(326)
7515
(650)
4936
(529)
4204
(406)
1727
(437)
*No samples were observed within this range.
384
-------
Table C-2. Population Estimates of the Number (and Standard Error) of Target Stream Lower
(L) and Upper (U) Reach Nodes with Spring Index pH in Reference Ranges
SUBREGION
Poc./Cats.
(ID)
N. Appal.
(2Cn)
Val. & Ridge
(2Bn)
MA Coast. PI.
(3B)
S. Blue Ridge
(2As)
Piedmont
(3A)
S. Appal.
(2X)
Oz./Ouach.
(2D)
Florida
(3C)
< 5.0
L U
* 157
(-) (85)
291 499
(203) (250)
* 318
(-) (317)
772 1334
(362) (524)
* *
* *
* *
* *
225 539
(107) (340)
pH Range
5.0 - <5.5 5.5 - 6.0
L U
10 67
(9) (60)
203 634
(150) (385)
* 443
(-) (339)
716 1320
(408) (531)
* *
* *
* 121
(-) (121)
* 225
(-) (127)
146 324
(71) (103)
L
85
(62)
681
(402)
47
(47)
955
(468)
*
316
(221)
121
(121)
75
(75)
591
(236)
U
193
(145)
853
(425)
683
(448)
2907
(767)
*
790
(342)
243
(170)
225
(127)
379
(224)
> 6.0
L
3140
(349)
7313
(814)
13,945
(1214)
8844
(1090)
2031
(326)
7199
(661)
4936
(529)
4041
(414)
593
(197)
U
2827
(342)
6,677
(792)
11,594
(1289)
5723
(1000)
2031
(326)
6725
(674)
4572
(537)
3754
(429)
485
(195)
TOTALS
L
3235
(347)
8488
(814)
13,992
(1213)
11,287
(1078)
2031
(326)
7515
(650)
5057
(526)
4116
(410)
1555
(306)
U
3244
(347)
8663
(807)
13,038
(1249)
11,284
(1078)
2031
(326)
7515
(650)
4936
(529)
4204
(406)
1727
(437)
*No samples were observed within this range.
385
-------
Table C-3. Population Estimates of the Combined Length (km) and Percentage of NSS-I
Target Stream Reaches with Spring Baseflow ANC Less than Reference Values
(Standard Errors in Parentheses).*
ANC <0
Subregion
Poconos/Catskills
N. Appalachians
Valley & Ridge
MA Coastal Plain
S. Blue Ridge
Piedmont
S. Appalachians
Ozarks/Ouachitas
Florida
SUBTOTALS
Interior MA
Interior SE
Mid-Atlantic (MA)
Southeast (SE)
Total NSS-I
Length
543
(270)
1,524
(750)
257
(210)
2,527
(1,200)
*
-
*
-
117
(120)
*
-
461
(160)
2,324
(1,000)
117
(120)
4,851
(1,600)
578
(210)
5,429
(1.500)
%
3.6
(1.8)
7.0
(3.5)
0.8
(0.6)
6.3
(2.9)
*
-
*
-
(0.5)
(0.5)
*
-
12.0
(4.1)
3.3
(1.4)
0.1
(0.1)
4.4
(1.4)
0.6
(0.2)
2.7
(0.8)
ANC
Length
1,606
(500)
3,713
(920)
2,111
(990)
9,636
(2,700)
706
(250)
2,390
(1,300)
763
(440)
205
(150)
2,356
(530)
7,431
(1,700)
4,064
(1,300)
17,067
(3,100)
6,420
(1,500)
23,487
(3.400)
<50
%
10.6
(3.3)
17.1
(4.2)
6.5
(3.0)
23.9
(6.6)
7.8
(2.8)
7.1
(3.9)
3.5
(2.0)
0.9
(0.6)
61.2
(14)
10.7
(2.4)
4.7
(1.5)
15.5
(2.8)
7.1
(1.7)
11.7
(1.7)
ANC < 200 Total Length
Length
5,489
(1,100)
12,935
(2,200)
12,811
(3,400)
21,091
(4,400)
7,084
(940)
13,554
(2,900)
6,130
(1,700)
15,092
(2,500)
2,939
(590)
31,235
(4,300)
41,860
(4,300)
52,327
(6,200)
44,799
(4,400)
97,125
(7.700)
%
36.2
(7.3)
59.5
(10)
39.2
(10)
52.3
(11)
78.4
(10)
40.4
(8.5)
28.0
(8.0)
67.1
(11)
76.4
(15)
44.9
(6.2)
48.1
(4.9)
47.6
(5.6)
49.3
(4.8)
48.4
(3.8)
(km)
15,144
(1,912)
21,738
(2,746)
32,687
(4,492)
40,296
(5,799)
9,036
(960)
33,531
(4,402)
21,892
(2,807)
22,480
(2,507)
3,848
(678)
69,569
(5,601)
86,939
(5,871)
109,865
(8,063)
90,787
(5,910)
200,652
(9.996)
[ using linear interpolation between upper and lower reach nodes. Standard errors
were approximated by an ad hoc procedure using the variances of separate length estimates
based on the upstream and downstream nodes.
* No samples observed below this reference value; estimated percentage is less than 1%.
NOTE: To calculate upper and lower one-sided 95% confidence bounds, multiply the standard
error by 1.645 and add or subtract that value from the length estimate. To calculate
the two-sided 95% confidence bounds, multiply the standard error by 1.96.
386
-------
Table C-4. Population -Estimates of the Combined Length (km) and Percentage of NSS-I
Target Stream Reaches with Spring Baseflow pH Less than Reference Values
Subregion
Poconos/Catskills
N. Appalachians
Valley & Ridge
MA Coastal Plain
S. Blue Ridge
Piedmont
S. Appalachians
Ozarks/Ouachitas
Florida
SUBTOTALS
Interior MA
Interior SE
Mid- Atlantic (MA)
Southeast (SE)
Total NSS-I
pH<
Length
550
(290)
1,424
(700)
257
(260)
3,147
(1,300)
*
-
*
-
*
-
*
-
522
(250)
2,231
(780)
*
—
5,378
(1,500)
522
(250)
5,900
(i.eoo)
:5.0
%
3.6
(1.9)
6.6
(3.2)
0.79
(0.8)
7.8
(3.3)
*
-
*
-
*
-
*
-
13.6
(6.5)
3.2
(1.1)
*
-
4.9
(1.4)
0.57
(0.3)
2.9
(0.8)
pH<
Length
906
(420)
1,870
(710)
1,937
(1,300)
9,565
(3,000)
*
-
*
-
313
(310)
410
(290)
1,708
(440)
4,712
(1,700)
723
(430)
14,277
(3,400)
2,431
(800)
16,708
(3.400)
5.5
%
6.0
(2.8)
8.6
(3.2)
5.9
(4.0)
23.7
(7.5)
*
-
*
-
1.4
(1.4)
1.8
(1-3)
44.4
(12)
6.8
(2.4)
0.83
(0.5)
13.0
(3.1)
2.7
(0.9)
8.3
(1.7)
pH<
Length
1,354
(520)
3,044
(900)
4,116
(1,900)
18,707
(4,300)
*
-
2,390
(1,200)
920
(540)
2,437
(990)
2,828
(620)
8,514
(2,300)
5,747
(1,700)
27,221
(4,800)
8,576
(1,900)
35,797
(5.200)
6.0 1
%
8.9
(3.4)
14.0
(4.2)
12.6
(5.9)
46.4
(11)
*
-
7.1
(3.7)
4.2
(2.5)
10.8
(4.4)
73.5
(16)
12.2
(3.4)
6.6
(1.9)
24.8
(4.4)
9.5
(2.1)
17.8
(2.6)
rotal Length
(km)
15,144
(1,912)
21,738
(2,746)
32,687
(4,492)
40,296
(5,799)
9,036
(960)
33,531
(4,402)
21,892
(2,807)
22,480
(2,507)
3,848
(678)
69,569
(5,601)
86,939
(5,871)
109,865
(8,063)
90,787
(5,910)
200,652
(9.996)
* Calculated using linear interpolation of [H+] between upper and lower reach nodes. Standard
errors were approximated by an ad hoc procedure using the variances of separate length
estimates based on the upstream and downstream nodes.
* No samples observed below this reference value; estimated percentage is less than 1%.
NOTE: To calculate upper and lower one-sided 95% confidence bounds, multiply the standard
error by 1.645 and add or subtract that value from the length estimate. To calculate
the two-sided 95% confidence bounds, multiply the standard error by 1.96.
387
-------
-------
APPENDIX D
STANDARD ERRORS OF THE GEOGRAPHIC AND CHEMICAL CLASSIFICATION
POPULATION ESTIMATES
The following eight tables (D-l through D-8) present the standard errors associated with
the population estimates of the number of stream reaches present in the different NSS-I
geographic site classes within chemical classes. Standard errors were calculated as the square
root of the variance associated with each population estimate. The equation and theory involved
in the variance calculation are discussed in subsection 2.4.4 and are described in detail by
Overton (1987).
389
-------
Table D-l. Population Estimates of the Number of Target Reach Lower Nodes ± Standard
Error in Each Geographic Site Class Present in Inorganic Classified Streams
Geographic Site Class
N
ANC < 0
INORGANIC
ANC > 0-50 ANC > 50-200
Allegheny Plateau
High Plateau \ 218 ± 150 36 ±35 25 ± 17 158 ±145
Forested Plateau 4178 ± 823 291 ± 203 1320 ± 541 760 ± 399
Forested Agric. Plateau 3547 ± 635 833 ± 368
NE Mid-Atlantic
Pocono/Catskill Mts. 351 ± 164 2 ± 2 229 ± 144
Glaciated Agric. Plateau 866 ± 208 119 ± 83
Glaciated Forested Plateau 209 ± 102 80 ± 61 119 ± 83
Valley and Ridge Province
Ridges 4359 ± 1035 427 ± 314 1012 ± 544
Valleys 11809 ± 1397 2370 ± 778
Arkansas/Oklahoma
Boston Mts. 873 ± 267 723 ± 252
OuachitaMts. 1126 ± 257 375 + 162
Arkansas R. Valley 1365 ± 395 773 ± 258
Piedmont Province
Piedmont 9258 ± 1173 1108 ± 467
Piedmont Lowlands 2297 ± 793
S. Appalachian Highlands
Blue Ridge Mts. 2962 ± 496 67 ± 34 2053 ± 430
Cumberland Plateau 1307 ± 394 243 ± 170 821 ± 331
French Broad R. Valley 64 ± 37
Coastal pfain
Eastern Mid-Atlantic 2036 ± 651 239 ± 238 239 ± 238
Western Mid-Atlantic 6133 ± 1030 . 478 ± 335 478 + 335
New Jersey Pine Barrens 432 ± 249 239 ± 238 32 ± 32
Eastern Gulf 1580 ± 465 158 ± 157 474 ± 269
Western Gulf 753 ± 245 453 ± 207
-j-™
Panhandle Upland 689 ± 145 32 ±32 32 ± 32
Peninsula Upland 442 ± 269
Panhandle Swamp 264 ± 113 32 ± 32
Peninsula Swamp 163 ± 70
NSS-I TOTALS
57,277 ± 2118 565 ± 315 3070 ± 778 13194 ± 1412
& represents the estimated number of reaches ± standard error within each geographic site
class.
390
-------
Table D-2. Population Estimates of the Number of Target Reach Lower Nodes ± Standard
Error in Each Geographic Site Class Present in Organic-Influenced Classified
Streams
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plateau
NE Mid-Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plateau
Glaciated Forested Plateau
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid-Atlantic
Western Mid- Atlantic
New Jersey Pine Barrens
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I TOTALS
ft
218 ± 150
4178 ± 823
3547 ± 635
351 ± 164
866 ± 208
209 ± 102
4359 ± 1035
11809 ± 1397
873 ± 267
1126 ± 257
1365 ± 395
9258 ±1173
2297 ± 793
2962 ± 496
1307 ± 394
64 ± 37
2036 ± 651
6133 ± 1030
432 ± 249
1580 ± 465
753 ± 245
689 ± 145
442 ± 269
264 ± 113
163 ± 70
57.277 ± 2118
ORGANIC INFLUENCED
ANC < 0 ANC > 0-50 ANC > 50-200
16 ± 15
150 ± 105
397 ± 286
7 ± 7
28 ± 28
121 ± 121
239 ± 238
239 ± 238 239 ± 238
161 ± 69
474 ± 269 158 ± 157
75 ± 75
227 ±90 32 ± 32
168 ± 69 1089 ±391 1305 ± 482
N represents the estimated number of reaches ± standard error within each geographic site
class.
391
-------
Table D-3. Population Estimates of the Number of Target Stream Reach Lower Nodes ±
Standard Error in Each Geographic Site Class Present in Organic-Dominated
Classified Streams
ORGANIC DOMINATED
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plateau
NE Mid- Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plateau
Glaciated Forested Plateau
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid- Atlantic
Western Mid-Atlantic
New Jersey Pine Barrens
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I TOTALS f
N ANC < 0
218 ± 150
4178 ± 823
3547 ± 635
351 ± 164
866 ± 208
209 ± 102
4359 ± 1035
11809 ± 1397
873 ± 267
1126 ± 257
1365 ± 395
9258 ± 1173
2297 ± 793
2962 ± 496
1307 ± 394
64 ± 37
2036 ± 651
6133 ± 1030 239 ± 238
432 ± 249
1580 ± 465
753 ± 245
689 ± 145
442 ± 269
2641113 129 ± 94
163 ±70 96 ± 54
57.277 ±21 18 464 ± 261
ANC > 0-50 ANC > 50-200
478 ± 335 478 ± 335
158 ± 157
75 ± 75
203 ±93 32 ± 32
208 ± 207 170 ± 170
35 ± 35
•a* + -iA
j*f — j*t
889 ± 404 983 ± 417
N represents the estimated number of reaches ± standard error within each geographic site
class.
392
-------
Table D-4. Population Estimates of the Number of Target Stream Reach Lower Nodes ±
Standard Error in Each Geographic Site Class Present in Watershed Source of
Sulfate and ANC > 200 Classified Streams
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plateau
NE Mid-Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plateau
Glaciated Forested Plateau
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid-Atlantic
Western Mid-Atlantic
New Jersey Pine Barrens
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I TOTALS
N
218 ± 150
4178 ± 823
3547 ± 635
351 ± 164
866 ± 208
i 209 ± 102
4359 ± 1035
11809 ± 1397
873 ± 267
1126 ±257
1365 ± 395
9258 ±1173
2297 ± 793
2962 ± 496
1307 ± 394
64 ± 37
2036 ± 651
6133 ± 1030
432 ± 249
1580 ± 465
753 ± 245
689 ± 145
442 ± 269
264 ± 113
163 ± 70
57.277 ± 2118
Substantial
watershed
sources of sulfate
767 ± 403
600 ± 284
636 ± 445
318 ± 317
75 ± 75
121 ± 121
365 ± 268
239 ± 238
158 ± 157
3277 ± 834
ANC > 200
1040 ± 445
2114 ± 511
119 ± 83
746 ± 197
10 ± 9
2284 ± 774
9106 ± 1268
150 ± 105
525 ± 188
591 ± 321
7754 ±1106
2290 ± 793
814 ± 274
64 ± 37
955 ± 468
3267 ± 841
150 ± 105
129 ± 62
64 ± 44
67 ± 47
32 ± 32
32.272 ± 2034
N represents the estimated number of reaches ± standard error within each geographic site
class.
393
-------
Table D-5. Population Estimates of the Number of Target Reach Upper Nodes ± Standard
Error in Each Geographic Site Class Present in Inorganic Classified Streams
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plateau
NE Mid-Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plateau
Glaciated Forested Plateau
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid- Atlantic
Western Mid- Atlantic
New Jersey Pine Barrens
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I TOTALS
INORGANIC
N ANC < 0 ANC > 0-50 ANC > 50-200
221 ± 150 63 ± 38
4327 ± 826 436 ± 247
3448 ± 624
351 ± 164
818 ± 203
266 ± 117 77 ± 60
7190 ± 1217 695 ± 449
8342 ± 1249
873 ± 267
1126 ± 257
1452 ± 400
8940 ± 1134
2290 ± 793 239 ± 238
2962 ± 496
1307 ± 394
64 ± 37
2036 ± 651
6133 ± 1030
436 ± 249 239 ± 238
1580 ± 465
753 ± 245
623 ± 141 64 ± 44
492 ± 273
117 ± 67
495 ± 337
56.642 ± 2158 1812 ± 618
1464 ± 606
291 ± 203
140 ± 130
60 ± 59
1044 ± 484
60 ± 59
75 ± 75
231 ± 129
121 ± 121
239 ± 238
478 ± 335
32 ± 32
474 ± 269
75 ± 75
32 ± 32
4815 ± 959
158 ± 145
589 ± 284
955 ± 383
209 ± 103
60 ± 59
60 ± 59
2119 ± 731
1970 ± 707
648 ± 244
525 ± 188
773 ± 258
1739 ± 546
2134 ± 430
364 ± 206
716 ± 408
239 ± 238
790 ± 342
302 ± 183
14. 353 ± 1448
N represents the estimated number of reaches ± standard error within each geographic site
class.
394
-------
Table D-6. Population Estimates of the Number of Target Reach Upper Nodes ± Standard
Error in Each Geographic Site Class Present in Organic-Influenced Classified
Streams
Geographic Site Class
N
ANC <0
ORGANIC INFLUENCED
ANC > 0-50 ANC > 50-200
Allegheny Plateau
High Plateau 221 ± 150
Forested Plateau 4327 ±826 29 ± 29
Forested Agric. Plateau 3448 + 624 168 ± 168
NE Mid-Atlantic
Pocono/Catskill Mts. 351 ±164 2 ± 2
Glaciated Agric. Plateau 818 ± 203
Glaciated Forested Plateau 266 ± 117 69 ± 60
Valley and Ridge Province
Ridges 7190 ± 1217 60 ± 59
Valleys 8342 ± 1249
Arkansas/Oklahoma
Boston Mts. 873 ± 267 75 ± 75
Ouachita Mts. 1126 ± 257 150 ± 105
Arkansas R. Valley 1452 ± 400
Piedmont Province
Piedmont 8940 + 1134 397 ± 286
Piedmont Lowlands 2290 ± 793 —-
S. Appalachian Highlands
Blue Ridge Mts. 2962 ± 496 17 ± 17
Cumberland Plateau 1307 ± 394 121 ± 121 279 ± 199
French Broad R. Valley 64 ± 37 -—
Coastal Plain
Eastern Mid-Atlantic 2036 ± 651
Western Mid-Atlantic 6133 ± 1030 716 + 408 478 ± 335
New Jersey Pine Barrens 436 ± 249 107 ± 67
Eastern Gulf 1580 ± 465 158 ± 157 158 ± 157
Western Gulf 753 ± 245 75 ± 75
Florida
Panhandle Upland 623 + 141 75 ± 52 71+49 32 + 32
Peninsula Upland 492 + 273
Panhandle Swamp 117 + 67
Peninsula Swamp 495 ± 337
NSS-I TOTALS
56.642 + 2158 375 ± 159
1160 ± 472 1704 ± 528
N represents the estimated number of reaches ± standard error within each geographic site
class.
395
-------
Table D-7. Population Estimates of the Number of Target Reach Upper Nodes ± Standard
Error in Each Geographic Site Class Present in Organic-Dominated Classified
Streams
Geographic Site Class
ORGANIC DOMINATED
N ANC < 0 ANC > 0-50 ANC > 50-200
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plateau
NE Mid-Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plateau
Glaciated Forested Plateau
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid-Atlantic
Western Mid- Atlantic
New Jersey Pine Barrens
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I TOTALS
221 ± 150
4327 ± 826
3448 ± 624
351 ± 164
818 ± 203
266 ± 117
7190 ± 1217
8342 ± 1249
873 ± 267
1126 ± 257
1452 ± 400
8940 ± 1134
2290 ± 793
2962 ± 496
1307 ± 394
64 ± 37
2036 ± 651
6133 ± 1030 716 ± 408
436 ± 249
1580 ± 465
753 ± 245
623 ± 141
492 ± 273
117 ±67 78 ± 55
495 ± 337 460 ± 336
56.642 ±2158 1255 ± 531
15 ± 15
239 ± 238
239 ± 238 478 ± 335
221 ± 100 64 ± 44
208 ± 207
35 ± 34
683 ± 330 815 ±411
N represents the estimated number of reaches ± standard error within each geographic site
class.
396
-------
Table D-8. Population Estimates of the Number of Target Stream Reach Upper Nodes ±
Standard Error in Each Geographic Site Class Present in Watershed Source of
Sulfate and ANC > 200 Classified Streams
Geographic Site Class
Allegheny Plateau
High Plateau
Forested Plateau
Forested Agric. Plateau
NE Mid- Atlantic
Pocono/Catskill Mts.
Glaciated Agric. Plateau
Glaciated Forested Plateau
Valley and Ridge Province
Ridges
Valleys
Arkansas/Oklahoma
Boston Mts.
Ouachita Mts.
Arkansas R. Valley
Piedmont Province
Piedmont
Piedmont Lowlands
S. Appalachian Highlands
Blue Ridge Mts.
Cumberland Plateau
French Broad R. Valley
Coastal Plain
Eastern Mid-Atlantic
Western Mid-Atlantic
New Jersey Pine Barrens
Eastern Gulf
Western Gulf
Florida
Panhandle Upland
Peninsula Upland
Panhandle Swamp
Peninsula Swamp
NSS-I TOTALS
N
221 ± 150
4327 ± 826
3448 ± 624
351 ± 164
818 ± 203
266 ± 117
7190 ± 1217
8342 ± 1249
873 ± 267
1126 ± 257
1452 ± 400
8940 ±1134
2290 ± 793
2962 ± 496
1307 ± 394
64 ± 37
2036 ± 651
6133 ± 1030
436 ± 249
1580 ± 465
753 ± 245
623 ± 141
492 ± 273
117 ± 67
495 ± 337
56.642 ± 2158
Substantial
watershed
sources of sulfate
769 ± 403
167 ± 146
443 ± 339
123 ± 123
75 ± 75
75 ± 75
365 ± 268
478 ± 335
58 ± 30
2551 ± 709
ANC > 200
1040 ± 445
1867 + 483
758 ± 197
2813 ± 820
6189 ± 1109
150 ± 105
300 ± 146
604 ± 324
6804 ±1051
2051 ± 759
579 ± 243
421 ± 243
64 ± 37
478 ± 335
2312 ± 737
300 ± 146
64 ± 44
284 ± 181
39 ± 39
27.118 ± 1972
N represents the estimated number of reaches ± standard error within each geographic site
class.
397
U. S.GOVERNMENT PRINTING OF F I CE I I 9 8 8-5 2 Z-.3 5 5 / 00 3 4 8
-------
-------
-------
-------
UJ
CO
<
X
Q-
UJ
CO
LJ
(T
CO
LJ
X
H
U_
o
CO
z
o
o
UJ
DC
GO
z>
CO
-------
'
Is
CO
O
0) =|
05
oo
§-~ £!
2 S
OJ
oo
§00
• _. "> ™ =
183
T1 «r 1-1
£?
S o
8S
5 2.
V)
Q
------- |