United States Office of Acid Deposition, Environmental EPA/600/3-86/054a
PA"600/3-86-054a C. 1 Environmental Protection Monitoring and Quality Assurance January 1987
Agency Washington DC 20460
v-xEPA
Research and Development
Western Lake Survey
Phase I
Characteristics of
Lakes in the Western
United States: Volume I.
Population
Descriptions and
Physico-Chemical
Relationships
-------
o S
o k-
o o
Q- -Z.
o
O
-------
EPA/600/3-86/054a
January 1987
Western Lake Survey
Phase I
Characteristics of Lakes in the
Western United States
Volume I: Population Descriptions
and
Physico-Chemical Relationships
by
D. H. Landers, J. M. Eilers, D. F. Brakke, W. S. Overton, P. E. Kellar,
M. E. Silverstein, R. D. Schonbrod, R. E. Crowe, R. A. Linthurst, J. M. Omernik,
S. A. league, and E. P. Meier
A Contribution to the
National Acid Precipitation Assessment Program
U.S. Environmental Protection Agency
Office of Research and Development, Washington, DC 20460
Environmental Research Laboratory, Corvallis, Oregon 97333
Environmental Monitoring Systems Laboratory, Las Vegas, Nevada 89193
: T-onmental Protection Agency
Library (5PL-16)
••-uborn Street, Room 1670
,L 60604
-------
Notice
The information 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., Nos. 68-
02-7288 and 68-02-3994 to Radian Corporation, No. 68-03-3246 to Northrop
Services, Inc., and Interagency Agreement No. 40-1441-84 with the U.S.
Department of Energy. It has been subject to the Agency's peer and
administrative review, and it 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:
Landers, D. H.1, J. M. Eilers2, D. F. Brakke3, W. S. Overton4, P. E. Kellar5,
M. E. Silverstein6, R. D. Schonbrod7, R. E. Crowe7, R. A. Linthurst8, J. M.
Omernik9, S. A. league2, and E. P. Meier7. 1987. Characteristics of Lakes
in the Western United States. Volume I. Population Descriptions and Physico-
Chemical Relationships. EPA/600/3-86/054a, U.S. Environmental
Protection Agency, Washington, DC. 176 pp.
Eilers, J. M., P. Kanciruk10, R. A. McCord10, W. S. Overton, L Hook11, D. J.
Blick2, D. F. Brakke, P. E. Kellar, M. S. DeHaan2, M. E. Silverstein, and D. H.
Landers. 1987. Characteristics of Lakes in the Western United States. Volume
II. Data Compendium for Selected Physical and Chemical Variables. EPA/
600/3-86/054b, U.S. Environmental Protection Agency, Washington, DC.
425 pp.
Inquiries regarding the availability of the Western Lake Survey - Phase I data
base should be directed in writing to: Chief, Air Branch, USEPA
Environmental Research Laboratory, 200 S.W. 35th Street, Corvallis, Oregon
97333.
'USEPA, Environmental Research Laboratory, 200 S W 35th Street, Corvallis, Oregon 97333
2Northrop Services, Inc., Environmental Research Laboratory, 200 S W 35th Street, Corvallis, Oregon 97333
3Western Washington University, Institute for Watershed Studies, Environmental Sciences Building, Room
604, Belhngham, Washington 98225.
"Oregon State University, Department of Statistics, Kidder Hall No 8, Corvallis, Oregon 97331.
6Radian Corporation, 3200 E Chapel Hill Road, Research Triangle Park, North Carolina 27709
6Lockheed Engineering and Management Services Company, Inc. 1050 E Flamingo Road, Suite 120, Las
Vegas, Nevada 89119
'USEPA, Environmental Monitoring Systems Laboratory, 944 E Harmon Avenue, Las Vegas, Nevada 89193
HJSEPA, Office of Research and Development, 401 M Street, S.W, Washington DC 20460. Present Address-
Environmental Monitoring Systems Laboratory, Mail-Drop 39, Research Triangle Park, North Carolina 27711
'USEPA, Environmental Research Laboratory, 200 S W 35th Street, Corvallis, Oregon 97333
'"Environmental Sciences Division, Oak Ridge National Laboratory, Post Office Box X, Oak Ridge, Tennessee
37831. Operated by Martin Marietta Energy Systems, Inc. under Contract No DE-ACO5-84OR21400 for
the U S. Department of Energy
"Science Applications International Corporation, 800 Oak Ridge Turnpike, Oak Ridge, Tennessee 37831
-------
Preface
The National Surface Water Survey (NSWS) was begun in 1983 by the U.S.
Environmental Protection Agency (EPA) under the National Acid Precipitation
Assessment Program, a Federal Interagency Task Force mandated by Congress
in 1980. The goals of the NSWS are to: (1) determine the present chemical
conditions in surface waters in regions of the United States considered
potentially sensitive to the effects of acidic deposition, (2) evaluate the temporal
variability of chemistry and quantify the biological status of these waters,
and (3) develop and implement a project to monitor changes over time in
representative systems. The NSWS is one of several major projects in the
Acid Deposition Aquatic Effects Research Program. This program, one of many
research programs investigating acidic deposition, is administered in 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 Aquatic Effects Research Program addresses four primary assessment
questions:
1. How extensive is the damage to aquatic resources as a result of current
levels of acidic deposition?
2. What is the anticipated extent and rate of change to these resources
in the future?
3. What levels of damage to sensitive surface waters are associated with
various rates of acidic deposition?
4. What is the rate of change or recovery of affected systems, given
decreases in acidic deposition rates?
Five major research projects within the Aquatic Effects Research Program
specifically address these assessment questions from a regionalized
perspective. These projects and their goals are:
1. 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;
2. 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;
3. Watershed Manipulation Project (WMP): to verify that predictions of future
change are reasonably sound by manipulating watershed catchments
or system components;
4. 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; and
-------
5. Long-Term Monitoring Project (LTMP): 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 Aquatic Effects Research Program. Understanding the national-scale
effects of acidic deposition on aquatic resources requires that the present
chemical status of surface waters be understood on large geographical scales.
The Western Lake Survey - Phase I (WLS-I) was designed to describe
statistically the present surface water chemistry on a regional scale in the
western United States. Although the cause-and-effect relationship between
acidic deposition and surface water response cannot be determined on the
basis of the WLS-I data alone, continued analysis of these data will contribute
significantly to furthering our understanding of the extent to which western
lakes are at risk due to acidic deposition. Determining the relationships between
acidic deposition and surface water chemistry and biology is the goal of future
activities within the Aquatic Effects Research Program.
-------
Volume I
Contents
Section Page
Notice ii
Preface iii
Volume II Contents x
Figures xi
Tables xxi
Related Documents xxvi
Acknowledgments xxvii
Executive Summary xxxiii
1. Introduction 1
2. Methods 3
2.1 Design 3
2.1.1 Data Quality Objectives 3
2.1.2 Summary of the Statistical Design 3
2.1.3 Definition of Study Area 4
2.2 Lake Selection 4
2.2.1 Probability Sample 4
2.2.2 Identification of Non-target Lakes 5
2.2.3 Identification of Wilderness Lakes 8
2.2.4 Special Interest Lakes 8
2.2.5 Final Lake Lists and Maps 12
2.3 Design Applications and Restrictions 12
2.3.1 Extrapolation from Sample to Population 12
2.3.2 Estimating the Target Population Size and Attributes ... 12
2.3.3 Restrictions 15
2.4 Lake Characterization 15
2.4.1 Lake Area 15
2.4.2 Elevation 15
2.4.3 Hydrologic Lake Type 15
2.4.4 Watershed Area 16
2.4.5 Land Use/Land Cover 16
2.4.6 Slope 16
2.5 Sampling and Analytical Methods 16
2.5.1 Field Sampling Activities 17
2.5.1.1 Site Description and Location 17
2.5.1.2 In Situ Measurements and Sample
Collection 17
2.5.2 Field Laboratory Operations 19
2.5.3 Analytical Laboratory Operations 21
2.6 Calibration Study 21
-------
Volume I
Contents (continued)
Section Page
2.7 Data Base Management 22
3. Quality Assurance 26
3.1 Design of the Quality Assurance Program 26
3.1.1 Selection of Analytical Laboratories 26
3.1.2 Training of Sampling and Field Laboratory Crews 26
3.1.3 Communications 26
3.1.4 On-Site Inspections 26
3.1.5 Quality Assurance/Quality Control Data
Collection and Analysis 26
3.1.5.1 Quality Assurance Samples 27
3.1.5.2 Field Sampling and Field Laboratory
Quality Control Samples 29
3.1.5.3 Analytical Laboratory Quality Control
Samples 29
3.2 Data Verification 29
3.2.1 Review of Field Data Forms 29
3.2.2 Initial Review of Analytical Laboratory Sample
Data Package 29
3.2.3 Review of Quality Assurance/Quality
Control Data 29
3.3 Data Validation 30
3.4 Development of Final Data Set 31
4. Quality Assurance and Calibration Study Results 32
4.1 Site Confirmation 32
4.2 Detection Limits (Detectability) 32
4.3 Precision 33
4.4 Effects of Holding Time on Sample Concentration 37
4.5 Assessment of the Effect of Different Field Sampling
Protocols 40
4.5.1 Calibration of the Ground Sample Data 40
4.5.2 Relative Bias Between Analytical Laboratories 44
4.6 Assessment of Data Quality 47
5. Results of Population Estimates 52
5.1 Data Presentation and Considerations 52
5.1.1 Presentation 52
5.1.2 Design Considerations 54
5.1.2.1 Using Weights 54
5.1.2.2 Evaluation of Alkalinity Map Classes 55
5.2 Description of Target Population 55
5.2.1 Number of Lakes Sampled 55
-------
Volume I
Contents (continued)
Section Page
5.2.2 Treatment of Shallow, Thermally Stratified, and
Large Lakes 56
5.2.3 Description of Target Population and Sample 56
5.3 Physical Characteristics of Lake Populations 56
5.4 Population Estimates for Primary Variables 59
5.4.1 Acid Neutralizing Capacity 59
5.4.1.1 Reference Values 59
5.4.1.2 Subregional Estimates and Distribution of
Lakes Having Low Acid Neutralizing
Capacity 59
5.4.1.3 State Estimates 60
5.4.2 pH 60
5.4.2.1 Subregional Estimates 60
5.4.2.2 State Estimates 60
5.4.3 Sulfate 60
5.4.4 Calcium 66
5.4.5 Extractable Aluminum, Clearwater Lakes 67
5.4.6 Dissolved Organic Carbon 67
5.5 Statistics for Population Descriptions, Primary Variables 67
5.6 Statistics for Population Descriptions, Secondary Variables .... 68
5.6.1 Nutrients 68
5.6.2 True Color, Turbidity, and Secchi Disk Transparency. . . .71
5.6.3 Sodium, Potassium, and Magnesium 72
5.6.4 Iron, Manganese, and Total Aluminum 72
5.6.5 Other Secondary Variables 73
5.7 Statistics for Lakes in Wilderness Areas and National Parks ... .73
5.7.1 Wilderness Area and National Park Lakes 73
5.7.2 Subpopulations in Wilderness Areas and National
Parks 74
5.8 Sample Statistics for Special Interest Lakes 76
5.9 Statistics for Dilute Lakes 76
6. Associations Among Variables 85
6.1 pH and Acid Neutralizing Capacity 85
6.1.1 Comparison of pH Measurements 85
6.1.2 Relationship between pH and Acid Neutralizing
Capacity 85
6.2 Major Cations and Anions 86
6.2.1 Individual Example Lakes 87
6.2.2 Calcium and Magnesium 88
6.2.3 Calcium and Sulfate 88
6.2.4 Acid Neutralizing Capacity versus Base Cations 90
6.2.5 Relative Abundance of Major Ions 90
6.2.6 Relationships Among Major Ions 95
6.2.7 Dissolved Organic Carbon 100
6.2.7.1 Dissolved Organic Carbon and True Color ... 100
6.2.7.2 Anion Deficit 100
vii
-------
Volume I
Contents (continued)
Section Page
6.3 Geology 103
6.4 Physical Lake Characteristics and Water Chemistry 105
6.4.1 Lake Area 105
6.4.2 Lake Elevation 107
6.4.3 Hydrology 109
6.4.3.1 Hydrologic Lake Type 109
6.4.3.2 Hydraulic Residence Time 110
7. Subregional Characteristics 118
7.1 California (4A) 118
7.1.1 Sierra Nevada 118
7.1.2 Other Geomorphic Units 121
7.2 The Pacific Northwest (4B) 122
7.2.1 The North Cascades 124
7.2.2 The Oregon Cascades 124
7.2.3 Other Geomorphic Units 126
7.2.4 Sulfate in Geomorphic Units in the Pacific
Northwest 126
7.3 Northern Rocky Mountains (4C) 127
7.3.1 Idaho Batholith 127
7.3.2 Bitterroot Range 129
7.3.3 Other Geomorphic Units 129
7.4 Central Rocky Mountains (4D) 129
7.4.1 The Wind River Uplift 131
7.4.2 The Uinta Arch 131
7.4.3 Other Geomorphic Units 131
7.5 Southern Rocky Mountains (4E) 134
7.5.1 The Front Range 135
7.5.2 Other Geomorphic Units 135
7.6 Ionic Relationships for Lakes in the Major Geomorphic
Units 138
8. Comparison of Results from the Western and Eastern Lake
Surveys 140
8.1 Physical Characteristics 140
8.2 Primary Variables 140
8.3 Associations Among Variables 145
8.4 Discrete Subpopulations of Lakes in the East and West 146
9. Conclusions 149
10. References 150
11. Glossary 157
VIII
-------
Volume I
Contents (continued)
Section Page
11.1 Abbreviations 157
11.2 Definitions 159
Appendix 169
IX
-------
Volume II
Contents
Section Page
Notice ii
Preface iii
Related Documents vi
Figures vii
Tables xxv
1. Introduction 1
2. Population Estimates for Selected Physical and Chemical Variables . . .5
2.1 Applications of the Design 5
2.1.1 Extrapolation from Sample to Population 5
2.1.2 Estimating the Target Population Size and
Subpopulation 5
2.2 Data Presentation 8
3. USGS Maps Used to Define and Characterize the Lake
Population 190
3.1 Small-scale Maps 190
3.2 Large-scale Maps 190
4. Maps Showing Lake Locations by State 210
5. Lakes Sampled and Individual Lake Data 221
5.1 Probability Sample Lakes 221
5.1.1 List of Probability Sample Lakes Sorted by State and
Lake Name 221
5.1.2 List of Probability Sample Lakes Sorted by Lake ID .... 221
5.1.3 Data for Individual Probability Sample Lakes 222
5.2 Special Interest Lakes 223
5.2.1 List of Special Interest Lakes Sorted by State and
Lake Name 223
5.2.2 List of Special Interest Lakes Sorted by ID 223
5.2.3 Data for Individual Special Interest Lakes 223
6. References 424
-------
Figures
Number Page
2-1 Subregions and alkalinity map classes ( •: < 100 //eq L~1;
m\ 100-199 fjeq L"1; D: 200-400 /veq L~1), in study areas
of the Western Lake Survey - Phase I 6
2-2 Locations of Forest Service wilderness areas ( B ) and national
parks ( & ) containing lakes sampled in the California
(4A) and Pacific Northwest (4B) subregions, in study areas
of the Western Lake Survey - Phase I 9
2-3 Locations of Forest Service wilderness areas ( El) and national
parks ( ^ ) containing lakes sampled in the Rocky Mountain
subregions: the Northern Rockies (40), the Central Rockies
(4D), and the Southern Rockies (4E), in study areas of the
Western Lake Survey - Phase I. Lakes located in roadless areas
are also shown ( E3) 10
2-4 Procedures used to estimate the target population size, Western
Lake Survey - Phase I 13
2-5 Field sampling activities. Western Lake Survey - Phase I. Field
crews accessed most sampling sites in wilderness areas by
ground and in non-wilderness areas by helicopter 18
2-6 Field laboratory activities. Western Lake Survey - Phase I.
Samples collected by field crews were processed at mobile
field laboratories before shipment to analytical laboratories
for analysis 20
2-7 Allocation of samples to analytical laboratories for the calibration
study. Western Lake Survey - Phase I. This procedure was used
to assign routine (R), duplicate (D), and triplicate (T) samples
collected by ground crews (GC) and helicopter crews (HC) to one
of two analytical laboratories 23
2-8 Procedures used to develop Data Set 1 (raw). Data Set 2 (verified).
Data Set 3 (validated), and Data Set 4 (final) for the Western
Lake Survey - Phase I data base 24
3-1 Origin, use, and flow of quality assurance/quality control
samples, Western Lake Survey - Phase I. Results of the analyses
of these samples were used to quantify data quality and to ensure
that error was minimized at each level of sample handling and
processing 28
-------
Figures (continued)
Number Page
4-1 Results of ANC (/ueq L~1) measurements and pH determinations
on air-equilibrated duplicate samples collected by helicopter
and ground crews from 45 lakes for the calibration study
during the Western Lake Survey - Phase I. The dimensions of the
boxes indicate the magnitude of sampling and measurement
error; sampling bias is evident when the position of the boxes
consistently deviates from the 1:1 (diagonal) line 41
4-2 Results of sulfate (/ueq L"1) and calcium (/ueq L"1) analyses on
samples collected by helicopter and ground crews from 45 lakes
for the calibration study during the Western Lake Survey - Phase I.
The dimensions of the boxes indicate the magnitude of sampling
and measurement error; sampling bias is evident when the
position of the boxes consistently deviates from the 1:1
(diagonal) line 42
4-3 Results of extractable aluminum (/L/g L"1) and DOC (mg L~1)
measurements on duplicate samples collected by helicopter and
ground crews from 45 lakes for the calibration study during the
Western Lake Survey - Phase I. The dimensions of the boxes
indicate the magnitude of sampling and measurement error;
sampling bias is evident when the position of the boxes
consistently deviates from the 1:1 (diagonal) line 43
4-4 Results of ANC (/ueq L"1) and pH determinations on air-
equilibrated samples performed by Lab 1 and Lab 2. Analyses
were conducted on duplicate samples collected by helicopter
and ground crews from 45 lakes for the calibration study during
the Western Lake Survey - Phase I. The dimensions of the boxes
indicate the magnitude of sampling and measurement error;
laboratory bias is evident when the position of the boxes
consistently deviates from the 1:1 (diagonal) line 46
4-5 Results of sulfate (/ueq L""1) and calcium (/ueq L"1) measurements
by Lab 1 and Lab 2. Analyses were conducted on duplicate
samples collected by helicopter and ground crews from 45 lakes
for the calibration study during the Western Lake Survey -
Phase I. The dimensions of the boxes indicate the magnitude
of sampling and measurement error; laboratory bias is evident
when the position of the boxes consistently deviates from the
1:1 (diagonal) line 48
4-6 Results of extractable aluminum (/ug L"1) and DOC (mg L"1)
measurements performed by Lab 1 and Lab 2. Analyses were
conducted on duplicate samples collected by helicopter and
ground crews from 45 lakes for the calibration study during
the Western Lake Survey - Phase I. The dimensions of the boxes
indicate the magnitude of sampling and measurement error;
laboratory bias is evident when the position of the boxes
consistently deviates from the 1:1 (diagonal) line 49
5-1 F(x) and G(x) distributions ( ) and the 95 percent upper
confidence limits ( ) on these distributions for ANC (//eq L"1)
for the target population of lakes ( > 1 ha and < 2000 ha) sampled
in Region 4 (West), Western Lake Survey - Phase I. The population
-------
Figures (continued)
Number Page
statistics, quintiles (Qi-CU) and medians, and three reference values
(Xci, XC2, and Xca) for ANC are also shown. The proportions
(Pc or gc), and number or area (Nc or Ac) of lakes below the reference
values, and their 95% upper confidence bounds (Ncu or Acu) are
given below each distribution 53
5-2 Locations of lakes sampled in California (4A) with ANC (yueq L"1)
in one of three classes ( • : ANC < 50, + : ANC > 50-200, and
0 : ANC > 200), Western Lake Survey - Phase I. Geographic area
covered by the target population is bounded by a dashed line 61
5-3 Locations of lakes sampled in the Pacific Northwest (4B) with ANC
(/ueq L"1) in one of three classes ( • : ANC < 50, + : ANC > 50-200,
and 0 : ANC > 200), Western Lake Survey - Phase I. Geographic
area covered by the target population is bounded by a dashed
line 62
5-4 Locations of lakes sampled in the Northern Rockies (4C) with ANC
(fjeq L"1) in one of three classes ( • : ANC < 50, + : ANC > 50-200,
and 0 : ANC > 200), Western Lake Survey - Phase I. Geographic
area covered by the target population is bounded by a dashed
line 63
5-5 Locations of lakes sampled in the Central Rockies (4D) with ANC
(/ueq L"1) in one of three classes ( • : ANC < 50, + : ANC > 50-200,
and 0 : ANC > 200), Western Lake Survey - Phase I. Geographic
area covered by the target population is bounded by a dashed
line 64
5-6 Locations of lakes sampled in the Southern Rockies (4E) with ANC
(/ueq L"1) in one of three classes ( • : ANC < 50, + : ANC > 50-200,
and 0 : ANC > 200), Western Lake Survey - Phase I. Geographic
area covered by the target population is bounded by a dashed
line 65
5-7 Cumulative frequency distributions [F(x>] for ANC (fjeq L"1) and
pH in California ( ), the Pacific Northwest ( ), the
Northern Rockies ( ), the Central Rockies ( ), and the
Southern Rockies ( ), Western Lake Survey - Phase 1 69
5-8 Cumulative frequency distributions [F(x>] for calcium (/ueq L"1)
and inverse cumulative frequency distributions [1 -F(x)] for
sulfate (/ueq L"1) in California ( ), the Pacific Northwest
( ), the Northern Rockies ( ), the Central Rockies
( ), and the Southern Rockies ( ), Western Lake
Survey - Phase I 70
5-9 Cumulative frequency distributions [F(x)J for ANC (/ueq L"1) in
wilderness area ( ) and non-wilderness area ( ) lakes
in California (4A), the Pacific Northwest (4B), the Northern
Rockies (4C), the Central Rockies (4D), and the Southern Rockies
(4E), Western Lake Survey - Phase I 75
XIII
-------
Figures (continued)
Number Page
5-10 Cumulative frequency distributions [F(x)J for calcium (yueq L"1) in
wilderness area ( ) and non-wilderness area ( ) lakes
in California (4A), the Pacific Northwest (4B), the Northern
Rockies (4C), the Central Rockies (4D), and the Southern Rockies
(4E), Western Lake Survey - Phase I 75
5-11 Inverse cumulative frequency distributions [1 -F(x)] for sulfate
(yueq L"1) in wilderness area ( ) and non-wilderness area
( ) lakes in California (4A), the Pacific Northwest (4B),
the Northern Rockies (4C), the Central Rockies (4D), and the
Southern Rockies (4E), Western Lake Survey - Phase I 76
5-12 Concentrations (fjeq L"1) of ANC, calcium, and sulfate in lakes
in selected wilderness areas and national parks. Western Lake
Survey - Phase I. Numbers in parentheses after each area name
are those shown in Figures 2-2 and 2-3 and Table 2-4 in Section
2.2.3. First quintiles (Qi, 20th percentile), medians (M, 50th
percentile), and fourth quintiles (Ch, 80th percentile), are
given for each variable (O 1 a ) 77
Qi M Q4
5-13 Locations of dilute lakes (conductance < 10 pS crrT1) sampled
in California (4A), Western Lake Survey - Phase I. Geographic
area covered by the target population is bounded by a dashed
line 79
5-14 Locations of dilute lakes (conductance < 10 //S cm"1) sampled
in the Pacific Northwest (4B), Western Lake Survey - Phase I.
Geographic area covered by the target population is bounded by
a dashed line 80
5-15 Locations of dilute lakes (conductance < 10 jt/S cm"1) sampled
in the Northern Rockies (4C), Western Lake Survey - Phase I.
Geographic area covered by the target population is bounded by
a dashed line 81
5-16 Locations of dilute lakes (conductance < 10 pS cm"1) sampled
in the Central Rockies (4D), Western Lake Survey - Phase I.
Geographic area covered by the target population is bounded by
a dashed line 82
5-17 Locations of dilute lakes (conductance < 10 /uS cm"1) sampled
in the Southern Rockies (4E), Western Lake Survey - Phase I.
Geographic area covered by the target population is bounded
by a dashed line 83
6-1 Relationship between closed system pH and measured ANC for
lakes with pH < 9.0 and ANC < 1000 /ueq L"1 in all subregions,
Western Lake Survey - Phase I. This same relationship is shown
in the inset for air-equilibrated pH 86
6-2 Ionic composition of one lake selected from each of the five
western subregions surveyed during the Western Lake Survey -
Phase I. Each lake was selected on the basis of having concen-
XIV
-------
Figures (continued)
Number Page
trations of ANC (//eq L~1) and DOC (mg L~1) approximately equal to
the first quintile (Qi, 20th percentile) values estimated for the
subregional target population of lakes. Cations and anions are
expressed as percentages of the total measured ionic equivalents
(//eq L"1); ions with concentrations < 5 //eq L~1 are not shown.
Champion Lake (4C2-047) is shown on the USGS topographic
map as an unnamed lake and is listed as "No Name" in the data
base and lake lists, Volume II 87
6-3 Relationship between calcium ( < 500 //eq L~1) and magnesium
( < 250 //eq L~1), Western Lake Survey - Phase I 88
6-4 Median sulfate concentrations (//eq L~1) for lakes in three
measured ANC classes for California (4A), the Pacific Northwest
(4B), the Northern Rockies (4C), the Central Rockies (4D), and
the Southern Rockies (4E), Western Lake Survey - Phase I 90
6-5 Relationship between measured ANC ( < 200 //eq L~1) and the
sum of calcium and magnesium (//eq L"1) for lakes in California
(4A), Western Lake Survey - Phase I. The alkalinity map class
from which each lake was selected is shown ( Q < 100 //eq L"1;
A: 100-199 //eq L'1; X: 200-400//eq L"1; Omernik and
Griffith 1986) 91
6-6 Relationship between measured ANC ( < 200 //eq L"1) and the
sum of calcium and magnesium (//eq L"1) for lakes in the Pacific
Northwest (4B), Western Lake Survey - Phase I. The, alkalinity
map class from which each lake was selected is shown
(D: < 100//eq L"1; A: 100-199//eq L'1; X: 200-400//eq L"1;
Omernik and Griffith 1986) 91
6-7 Relationship between measured ANC ( < 200 //eq L"1) and the
sum of calcium and magnesium (//eq L~1) for lakes in the
Northern Rockies (4C), Western Lake Survey - Phase I. The
alkalinity map class from which each lake was selected is shown
(D:< 100//eq L"1; A: 100-199//eq L'1; X: 200-400//eq L"1;
Omernik and Griffith 1986) 92
6-8 Relationship between measured ANC ( < 200 //eq L"1) and the
sum of calcium and magnesium (//eq L~1) for lakes in the
Central Rockies (4D), Western Lake Survey - Phase I. The
alkalinity map class from which each lake was selected is shown
( D: < 100 //eq L"1; A: 100-1 99 //eq L"1; X: 200-400 //eq L"1;
Omernik and Griffith 1 986) 92
6-9 Relationship between measured ANC ( < 200 //eq L~1) and the
sum of calcium and magnesium (//eq L~1) for lakes in the
Southern Rockies (4E), Western Lake Survey - Phase I. The
alkalinity map class from which each lake was selected is shown
( A: < 100 //eq L"1; A: 100-1 99 //eq L~1; X: 200-400 //eq L"1;
Omernik and Griffith 1 986) 93
6-10 Concentration of chloride (//eq L~1) versus distance ( < 100 km)
from the Pacific Coast for lakes in the Klamath Mountains in
California (4A) and from Puget Sound or the Pacific Ocean
xv
-------
Figures (continued)
Number Page
(whichever is less) in five geomorphic units in the state of
Washington in the Pacific Northwest (4B), Western Lake
Survey - Phase I. Values in the Pacific Northwest are shown for
the Wenatchee Mountains ( • ), the Middle Washington
Cascades ( A ), the North Washington Cascades ( + ),
the Puget Lowlands ( X ), and the Olympic Mountains ( D ).
Distance from the coast was determined from a SAS file (SAS
Institute 1985) 94
6-11 Concentration of sodium (/ueq L~1) versus distance (< 100 km)
from Puget Sound or the Pacific Ocean (whichever is less) for
lakes in five geomorphic units in the state of Washington in
the Pacific Northwest (4B), Western Lake Survey - Phase I.
Values are shown for the Wenatchee Mountains (• ), the Middle
Washington Cascades ( A ), the North Washington Cascades
( + ), the Puget Lowlands ( X ), and the Olympic
Mountains ( D ). Distance from the coast was determined
from a SAS file (SAS Institute 1985) 96
6-12 Relationship between sodium ( < 100 fjeq L"1) and chloride
( < 50 fjeq L"1) for lakes in the Pacific Northwest (4B) and the
Northern Rockies (4C), Western Lake Survey - Phase I. The
solid line indicates the ratio of sodium to chloride in sea water 97
6-13 Trilinear diagram showing the relative abundance of major
anions and cations for lakes in California (4A), Western Lake
Survey - Phase I. Ratios are expressed as percent of total
ionic concentration, increasing from 0 to 100 percent along the
axes in the direction of the arrows. Lakes for which the total
ionic composition cannot be described as being dominated by
calcium bicarbonate are circled and their lake ID numbers are
given 98
6-14 Trilinear diagram showing the relative abundance of major
anions and cations for lakes in the Pacific Northwest (4B),
Western Lake Survey - Phase I. Ratios are expressed as percent
of total ionic concentration, increasing from 0 to 100 percent
along the axes in the direction of the arrows. Lakes for which
the total ionic composition cannot be described as being dominated
by calcium bicarbonate are circled and their lake ID numbers
are given 98
6-15 Trilinear diagram showing the relative abundance of major
anions and cations for lakes in the Northern Rockies (4C),
Western Lake Survey - Phase I. Ratios are expressed as percent
of total ionic concentration, increasing from 0 to 100 percent
along the axes in the direction of the arrows. Lakes for which
the total ionic composition cannot be described as being dominated
by calcium bicarbonate are circled and their lake ID numbers
are given 99
6-16 Trilinear diagram showing the relative abundance of major
anions and cations for lakes the Central Rockies (4D), Western
Lake Survey - Phase I. Ratios are expressed as percent of total
-------
Figures (continued)
Number Page
ionic concentration, increasing from 0 to 100 percent along the
axes in the direction of the arrows. Lakes for which the total
ionic composition cannot be described as being dominated by
calcium bicarbonate are circled and their lake ID numbers are
given 99
6-17 Trilinear diagram showing the relative abundance of major
anions and cations for lakes in the Southern Rockies (4E),
Western Lake Survey - Phase I. Ratios are expressed as percent
of total ionic concentration, increasing from 0 to 100 percent
along the axes in the direction of the arrows. Lakes for which
the total ionic composition cannot be described as being dominated
by calcium bicarbonate are circled and their lake ID numbers
are given 100
6-18 Relationship between DOC ( < 10 mg L"1) and true color ( < 50
PCU) for lakes in all subregions. Western Lake Survey - Phase I... 101
6-19 Relationship between the sum of anions ( < 200 /ueq L"1) and the
sum of cations ( < 200 /ueq L"1) for lakes in California ( • ), the
Pacific Northwest ( A ), the Northern Rockies ( + ), the Central
Rockies ( X ), and the Southern Rockies ( D ), Western Lake
Survey - Phase I. Points falling below the 1:1 (diagonal) line
indicate an anion deficit, and above the line, a cation deficit.
Inset shows the same relationship for anions < 2000 /ueq L"1
and cations < 2000 /ueq L'1 102
6-20 Relationship between DOC (mg L"1) and anion deficit (/ueq L"1) for
lakes in all subregions with total measured ionic equivalents
< 500 /ueq L"1, Western Lake Survey - Phase I. Anion deficit is
the difference (/ueq L"1) between the sum of measured anions
and the sum of measured cations 104
6-21 Percentage of low ANC ( < 50 /ueq L"1) lakes by bedrock class
(Norton et al. 1982) in California (4A), the Pacific Northwest (4B),
the Northern Rockies (4C), the Central Rockies (4D), and the
Southern Rockies (4E), Western Lake Survey - Phase I. Bedrock
sensitivity is highest in Class 1 and lowest in Class 3 106
6-22 Locations and names of geomorphic units used in data analysis for
specific subpopulations of lakes, Western Lake Survey - Phase I.
These units represent physiographic areas of common geologic
origin and generally coincide with major mountain ranges 107
6-23 Relationship between ANC ( < 1000 /ueq L"1) and lake area
( < 100 ha) for lakes in the Pacific Northwest (4B), Western Lake
Survey - Phase I. The alkalinity map class from which each lake
was selected is shown ( D: < 100 /ueq L"1; A: 100-1 99 /ueq L"1;
X: 200-400 /ueq L"1; Omernik and Griffith 1 986) 108
6-24 Estimated number of lakes with ANC < 50 /ueq L"1 or ANC > 50
/ueq L"1 by lake area class, Western Lake Survey - Phase I 109
6-25 Concentrations (/ueq L~1) of ANC, sulfate, and calcium in lakes
< 4 ha ( ) and > 4 ha ( ) in California (4A), the Pacific
XVII
-------
Figures (continued)
Number Page
Northwest (4B), the Northern Rockies (4C), the Central Rockies
(4D), the Southern Rockies (4E), and the West (4) in the Western
Lake Survey - Phase I. First quintiles (Qi, 20th percentile),
medians (M, 50th percentile), and fourth quintiles (Q4, 80th
percentile) are given for each variable (O 1 D) 110
Qi M Q4
6-26 Cumulative frequency distributions [F(x>] for ANC ( < 1000 //eq L"1)
and calcium (< 1000//eq L~1), and inverse cumulative frequency
distributions [1-F(x>] for sulfate ( < 100/ueq L~1) in lakes < 4 ha
( ) and > 4 ha ( ) in the Northern Rockies (4C),
Western Lake Survey - Phase I 111
6-27 Relationship between ANC ( < 1000 //eq L"1) and elevation (m)
for lakes in California (4A), Western Lake Survey - Phase I.
The alkalinity map class from which each lake was selected is
shown ( D: < 100 yueq L"1; A: 100-199 //eq L"1; X: 200-400
//eq L"1; Omernik and Griffith 1986) 112
6-28 Relationship between ANC ( < 1000 //eq L~1) and elevation (m)
for lakes in the Pacific Northwest (4B), Western Lake Survey -
Phase I. The alkalinity map class from which each lake was
selected is shown ( D: < 100 //eq L~1; A: 100-199 //eq L"1;
X: 200-400 //eq L"1; Omernik and Griffith 1986) 113
6-29 Relationship between ANC ( < 1000 //eq L"1 ) and elevation (m)
for the Northern Rockies (4C), Western Lake Survey - Phase I.
The alkalinity map class from which each lake was selected is
shown ( D: < 100 //eq L"1; A: 100-199 //eq L"1; X: 200-400
//eq L"1; Omernik and Griffith 1986) 113
6-30 Relationship between ANC ( < 1000 //eq L"1) and elevation (m)
for the Central Rockies (4D), Western Lake Survey - Phase I.
The alkalinity map class from which each lake was selected is
shown ( D: < 100 //eq L"1; A: 100-199 //eq L"1; X: 200-400
//eq L~1; Omernik and Griffith 1986) 114
6-31 Relationship between ANC ( < 1000 //eq L~1) and elevation (m)
for the Southern Rockies (4E), Western Lake Survey - Phase I.
The alkalinity map class from which each lake was selected is
shown ( D: < 100 //eq L"1; A: 100-199 /ueq L~1; X: 200-400
//eq L"1; Omernik and Griffith 1986) 114
6-32 Relationship between ANC ( < 500 //eq L"1) and elevation (m)
for lakes on the western and eastern slopes of the Sierra
Nevada in California (4A) and Oregon Cascades in the Pacific
Northwest (4B), Western Lake Survey - Phase I 116
7-1 Locations of sampled lakes within geomorphic units in California
(4A), Western Lake Survey - Phase I. Lakes located outside the
boundaries were not included in estimates for these
subpopulations 119
XVIII
-------
Figures (continued)
Number Page
7-2 Locations of sampled lakes within geomorphic units in the Pacific
Northwest (4B), Western Lake Survey - Phase I. Lakes located
outside the boundaries were not included in estimates for these
subpopulations 123
7-3 Locations of sampled lakes within geomorphic units in the
Northern Rockies (40), Western Lake Survey - Phase I. Lakes
located outside the boundaries were not included in estimates
for these subpopulations 128
7-4 Locations of sampled lakes within geomorphic units in the Central
Rockies (4D), Western Lake Survey - Phase I. Lakes located
outside the boundaries were not included in estimates for these
subpopulations 132
7-5 Locations of lakes within geomorphic units in the Southern
Rockies (4E), Western Lake Survey - Phase I. Lakes located
outside the boundaries were not included in estimates for
these subpopulations 136
7-6 Trilinear diagram showing the relative abundance of major
anions and cations for lakes in the largest geomorphic unit in
California (Sierra Nevada), the Pacific Northwest (Oregon
Cascades), the Northern Rockies (Bitterroot Mountains), the
Central Rockies (Wind River Range), and the Southern Rockies
(Front Range), Western Lake Survey - Phase I. Ratios are
expressed as percent of total ionic concentration, increasing
from 0 to 100 percent along the axes in the direction of the
arrows 139
8-1 Cumulative frequency distributions [F(x>] for ANC (/ueq L"1) and pH
for the Northeast ( ), the Upper Midwest ( ), the
Southern Blue Ridge ( ), and Florida ( ), Eastern Lake
Survey- Phase I (Linthurst et al. 1986), and the West ( ),
Western Lake Survey - Phase I 141
8-2 Concentrations of ANC (jueq L~1), pH, and concentrations of
sulfate (/yeq L~1) for lakes in the Northeast (Region 1), Upper
Midwest (Region 2), Southern Blue Ridge (Subregion 3A), and
Florida (Subregion 3B), Eastern Lake Survey - Phase I (Linthurst
et al. 1986), and West (Region 4), Western Lake Survey -
Phase I. First quintile (61, 20th percentile), median (M, 50th
percentile), and fourth quintile (Q4, 80th percentile) values are
shown (O 1 D) 142
Qi M 0.4
8-3 Population estimates for the percent of lakes with concentrations
of ANC < 0, < 50, and < 200 fjeq L'\ and pH < 6.0, for the
Northeast, Upper Midwest, Southern Blue Ridge, and Florida,
Eastern Lake Survey - Phase I (Linthurst et al. 1986), and West,
Western Lake Survey - Phase I 143
8-4 Inverse cumulative frequency distributions [1 -F(x)] for sulfate
L"1) and cumulative frequency distributions [F(x)] for calcium
L"1)for the Northeast ( ), Upper Midwest ( ),
xix
-------
Figures (continued)
Number Page
Southern Blue Ridge ( ), and Florida ( - ), Eastern
Lake Survey - Phase I (Linthurst et al. 1986), and West ( ),
Western Lake Survey - Phase I 144
8-5 Population estimates for the percent of lakes with concentrations
of calcium < 50 /ueq L"1, sulfate > 50 fjeq L'\ DOC < 2 mg L"1,
DOC > 6 mg L"1, and extractable aluminum > 50 //g L"1 for the
Northeast, Upper Midwest, Southern Blue Ridge, and Florida,
Eastern Lake Survey - Phase I (Linthurst et al. 1986) and West,
Western Lake Survey - Phase I 144
8-6 Concentrations of calcium (fjeq L"1), extractable aluminum
(fjg L"1) in clearwater lakes, and DOC (mg L"1) for lakes in the
Northeast (Region 1), Upper Midwest (Region 2), Southern Blue
Ridge (Subregion 3A), and Florida (Subregion 3B), Eastern Lake
Survey - Phase I (Linthurst et al. 1986), and West (Region 4),
Western Lake Survey - Phase I. First quintile (Q1f 20th
percentile), median (M, 50th percentile), and fourth quintile
(O.4, 80th percentile) values are shown (O 1 D) 145
Qi M Q4
8-7 Inverse cumulative frequency distributions [1 -F(x)] for extractable
aluminum (//g L~1) in clearwater lakes, and cumulative frequency
distributions [F(x)j for DOC (mg L"1) for the Northeast ( ),
Upper Midwest ( ), Southern Blue Ridge ( ), and
Florida ( - ), Eastern Lake Survey - Phase I (Linthurst et al.
1986) and West ( ), Western Lake Survey - Phase I 146
8-8 Concentrations of ANC (fjeq L~1) and pH in selected
subpopulations of lakes located in major geomorphic units in
the Northeast, Upper Midwest, and Florida, Eastern Lake
Survey - Phase I (Linthurst et al. 1986) and West, Western
Lake Survey - Phase I. First quintile (Qi, 20th percentile),
median (M, 50th percentile), and fourth quintile (Cu, 80th
percentile) values are shown (O 1 D) 147
Q! M Q<
8-9 Concentrations of sulfate (/jeq L"1), calcium (/ueq L"1), and DOC
(mg L"1) in selected subpopulations of lakes located in major
geomorphic units in the Northeast, Upper Midwest, and Florida,
Eastern Lake Survey - Phase I (Linthurst et al. 1986) and West,
Western Lake Survey - Phase I. First quintile (Q1f 20th
percentile), median (M, 50th percentile), and fourth quintile
Q4, 80th percentile) values are shown (O 1 D) 148
Q! M Q4
-------
Tables
Number Page
2-1 Data Quality Objectives for Detectability, Precision, and Accuracy,
Western Lake Survey - Phase I 4
2-2 Non-target Lakes and Lakes Not Visited or Not Sampled, Western
Lake Survey - Phase I 7
2-3 Population Estimates of Median Lake Area and Median Elevation
for Frozen Lakes and Sampled Lakes in Central Rockies (4D) and
Southern Rockies (4E), Western Lake Survey - Phase I 8
2-4 Forest Service Wilderness Areas and National Parks
Containing Probability Sample Lakes, Grouped by Subpopulations
Used in Data Analysis, Western Lake Survey - Phase 1 11
2-5 Number of Lakes Shown on Small-scale (1:100,000) and
Large-scale (1:62,500; 1:24,000) Maps, Western Lake
Survey - Phase I 16
2-6 Aliquots and Corresponding Measured Parameters, Western
Lake Survey - Phase I 22
2-7 Maximum Holding Times Specified for Samples, Western Lake
Survey - Phase I 22
3-1 Number of Samples from Probability Sample Lakes Analyzed
by the Analytical Laboratories, by Subregion, Western Lake
Survey - Phase I 27
3-2 Factors Used to Convert mg L~1 to /veq L~1 for Anion/Cation
Balance Check, Western Lake Survey - Phase I 30
3-3 Factors Used to Convert mg L"1 to //S cm"1 for Determining
Calculated Conductance, Western Lake Survey - Phase I 30
4-1 Detectability Based on Evaluation of Field Blank Data, Western
Lake Survey - Phase I 32
4-2 Overall Among-Batch Precision Estimates Calculated from Field
Synthetic Audit Samples, Western Lake Survey - Phase I 34
4-3 Overall Among-Batch Precision Estimates Calculated from Field
Natural Audit Samples, Western Lake Survey - Phase I 35
4-4 Overall Within-Batch Precision Estimates Calculated from Field
Duplicates, Western Lake Survey - Phase I 36
4-5 Overall Field Laboratory Within-Batch Precision Estimates
Calculated from Field Duplicate and Trailer Duplicate Data,
Western Lake Survey - Phase I 37
-------
Tables (continued)
Number Page
4-6 Overall Within-Batch Precision Estimates Calculated from
Analytical Laboratory Routine/Duplicate Sample Data, Western
Lake Survey - Phase I 38
4-7 Holding Time Samples Processed by Analytical Laboratories for
the Calibration Study, Western Lake Survey - Phase I 38
4-8 Regression Statistics for the Difference Between Routine and
Withheld Samples versus Holding Time by Laboratory, Western
Lake Survey - Phase I 39
4-9 Results of Regression of Values for Primary Variables Measured
on Samples Collected by Helicopter Crews (Dependent) Against
Samples Collected by Ground Crews for the Calibration Study
Lakes, Western Lake Survey - Phase 1 44
4-10 Estimates of Bias between Analytical Laboratories Based on
Synthetic and Natural Audit Samples (Audits), Paired Duplicate
Samples from the Calibration Study (Duplicates), and Split
Samples (Splits) with the Environmental Research Laboratory-
Corvallis. The Model Results are Shown Using Lab 2 as trie
Dependent Variable and Lab 1 as the Independent Variable 45
4-11 Comparison of Model Results of ANC on [Ca+2 + Mg+2] for the
Central and Southern Rockies Using Data with Calcium Adjusted
and Unadjusted for Possible Analytical Bias, Western Lake
Survey - Phase I 50
5-1 Use of Weights in Combined Strata Estimation, Western Lake
Survey -Phase I 54
5-2 Composition of the Alkalinity Map Classes: Numbers and
Percentages of Lakes Having Measured ANC in Those Same
Classes, Western Lake Survey - Phase I 55
5-3 Population Statistics for ANC by Alkalinity Map Class, Western
Lake Survey - Phase I 56
5-4 Population Statistics for ANC, Calcium, and Sulfate in Shallow,
Deep, Stratified, and Unstratified Lakes, Western Lake Survey -
Phase 1 57
5-5 Description of Target Population, Sample, and Weighting Factors
Excluding Lakes > 2000 ha. Western Lake Survey - Phase 1 57
5-6 Population Statistics for Physical Variables for Lakes in the
Target Population, Western Lake Survey - Phase 1 58
5-7 Population Estimates of the Percentage of Lakes in
Categories by Lake Type, Western Lake Survey - Phase 1 58
5-8 Population Estimates of the Percentage of Watersheds Dominated
by Land Use/Land Cover Categories, Western Lake Survey -
Phase 1 59
xxii
-------
Tables (continued)
Number Page
5-9 Population Estimates of Lakes with ANC < 50 or < 200 /ueq L"1,
Western Lake Survey - Phase I 60
5-10 Population Estimates and Population Statistics by State for ANC
and pH, Western Lake Survey - Phase I 66
5-11 Population Estimates of Lakes with pH < 6.0, Western Lake
Survey - Phase I 66
5-12 Population Estimates of Lakes with Sulfate > 50 fjeq L"1, Western
Lake Survey - Phase I 66
5-13 Population Estimates of Lakes with Calcium < 50 //eq L~1, Western
Lake Survey - Phase I 67
5-14 Population Estimates of Clearwater Lakes with Extractable
Aluminum > 50 //g L~1, Western Lake Survey - Phase I 67
5-15 Population Estimates of Lakes with Dissolved Organic Carbon
< 2 mg L"1 or > 6 mg L"1, Western Lake Survey - Phase I 68
5-16 Population Statistics for the Primary Variables, Western Lake
Survey - Phase I 71
5-17 Population Statistics for Secondary Variables (Nitrate, Ammonium,
and Total Phosphorus), Western Lake Survey - Phase I 71
5-18 Population Statistics for Secondary Variables (True Color,
Turbidity, and Secchi Disk Transparency), Western Lake
Survey - Phase I 72
5-19 Population Statistics for Secondary Variables (Sodium, Potassium,
and Magnesium), Western Lake Survey - Phase I 72
5-20 Population Statistics for Secondary Variables (Iron, Manganese,
and Total Aluminum), Western Lake Survey - Phase I 73
5-21 Population Statistics for Secondary Variables (Silica, Dissolved
Inorganic Carbon, Chloride, Conductance, and Bicarbonate),
Western Lake Survey - Phase I 73
5-22 Population Statistics for ANC, Calcium, and Sulfate in Wilderness
Area Lakes and Non-Wilderness Area Lakes, Western Lake
Survey - Phase I 74
5-23 Sample Statistics for Physical and Chemical Variables for Special
Interest Lakes, Western Lake Survey - Phase I 78
5-24 Population Statistics for Chemical and Physical Characteristics
of Dilute Lakes, Western Lake Survey - Phase I 84
6-1 Weighted Regression Statistics Comparing pH Measurements for
Lakes Sampled Using Helicopters, Western Lake Survey -
Phase I 85
-------
Tables (continued)
Number Page
6-2 Regression Statistics for Calcium (< 500 /ueq L~1, Dependent)
versus Magnesium ( <250/ueq L~1) by Subregion and Stratum,
Western Lake Survey - Phase I 89
6-3 Regression Statistics for Calcium (< 500 /ueq L"1, Dependent)
versus Sulfate (< 100/ueq L~1) by Subregion and Stratum,
Western Lake Survey - Phase I 89
6-4 Regression Statistics for ANC (<200 /ueq L~1, Dependent) versus
the Sum of Calcium and Magnesium (< 200 /ueq L"1) by
Subregion and Stratum, Western Lake Survey - Phase I 93
6-5 Relative Abundance of Major Cations and Anions Within
Subregions Based on Population Estimates of Concentrations at
the First Quintile (Qi) and Median (M) Values,
Western Lake Survey - Phase I 95
6-6 Regression Statistics for DOC (< 10 mg L"1, Dependent) versus
True Color (< 50 PCU) by Subregion and Stratum, Western Lake
Survey - Phase I 101
6-7 Regression Statistics for the Sum of Major Anions (< 2000 /ueq L~1,
Dependent) versus the Sum of Major Cations (< 2000 //eq L"1)
by Subregion and Stratum, Western Lake Survey - Phase I 103
6-8 Population Statistics for Anion Deficit, Western Lake Survey -
Phase I 103
6-9 Regression Statistics for DOC (< 10 mg L"1, Dependent) versus
Anion Deficit (< 100 /ueq L"1) by Subregion and Stratum,
Western Lake Survey - Phase I 104
6-10 Spearman Correlation Coefficients (r) for Lake Area versus ANC,
Calcium, and Sulfate, by Subregion and Stratum, Western Lake
Survey - Phase I 108
6-11 Weighted Correlation Coefficients for ANC versus Lake Elevation
by Geomorphic Unit, Western Lake Survey - Phase I 115
6-12 Population Statistics for ANC, Calcium, and Sulfate in High
(> 3000 m) and Low (<3000 m) Elevation Lakes in California
(4A), the Central Rockies (4D), and the Southern Rockies (4E),
Western Lake Survey - Phase I 115
6-13 Population Statistics for ANC, Calcium, and Sulfate in Lakes on
the Western and Eastern Slopes of the Sierra Nevada and
Oregon Cascades, Western Lake Survey - Phase I 116
6-14 Population Medians for Selected Parameters by Hydrologic Lake
Type, Western Lake Survey - Phase I 117
6-1 5 Estimated Hydraulic Residence Time (Years) for Drainage Lakes
and Reservoirs by Subregion, Western Lake Survey - Phase I 117
XXIV
-------
Tables (continued)
Number Page
7-1 Population Estimates and Statistics for Primary Variables by
Geomorphic Unit in California (4A), Western Lake Survey -
Phase I 120
7-2 Population Statistics for Selected Variables for Lakes in
Geomorphic Units in California (4A), Western Lake Survey -
Phase I 121
7-3 Population Estimates and Statistics for Primary Variables by
Geomorphic Unit in the Pacific Northwest (4B), Western Lake
Survey - Phase I 125
7-4 Population Statistics for Selected Variables for Lakes in
Geomorphic Units in the Pacific Northwest (4B), Western Lake
Survey - Phase I 1 25
7-5 Population Estimates and Statistics for Primary Variables by
Geomorphic Unit in the Northern Rockies (4C), Western Lake
Survey - Phase I 130
7-6 Population Statistics for Selected Variables for Lakes in
Geomorphic Units in the Northern Rockies (4C), Western Lake
Survey - Phase I 130
7-7 Population Estimates and Statistics for Primary Variables by
Geomorphic Unit in the Central Rockies (4D), Western Lake
Survey - Phase I 133
7-8 Population Statistics for Selected Variables for Lakes in
Geomorphic Units in the Central Rockies (4D), Western Lake
Survey - Phase I 133
7-9 Population Estimates and Statistics for Primary Variables by
Geomorphic Unit in the Southern Rockies (4E), Western Lake
Survey - Phase I 1 37
7-10 Population Statistics for Selected Variables for Lakes in
Geomorphic Units in the Southern Rockies (4E), Western Lake
Survey - Phase I 137
8-1 Physical Characteristics of Lakes Sampled During Phase I of the
Eastern and Western Lake Surveys 141
XXV
-------
Related Documents^
Anonymous. 1984. National Surface Water Survey, National Lake Survey -
Phase I, Research Plan. U.S. Environmental Protection Agency, Washington,
D.C. (internal document).
Bonoff, M.B. and A.W. Groeger. 1987. National Surface Water Survey, Western
Lake Survey - Phase I, Field Operations Report. Internal Report. EPA/600/
8-87/018. Environmental Monitoring Systems Laboratory, U.S. Environmental
Protection Agency, Las Vegas, Nevada. 26 pp.
Eilers, J.M., D.J. Blick, and M.S. DeHaan. 1987. National Surface Water Survey,
Western Lake Survey - Phase I, Validation of the Western Lake Survey - Phase
I Data Base. Environmental Research Laboratory, U.S. Environmental
Protection Agency, Corvallis, Oregon.
Kanciruk, P., M.J. Gentry, R.A. McCord, L.A. Hook, J.M. Eilers, and M.D. Best.
1987. National Surface Water Survey, Western Lake Survey - Phase I, Data
Base Dictionary. Oak Ridge National Laboratory, Technical Manual ORNL/
TM-10307, Oak Ridge, Tennessee.
Kerfoot, H.B. and M.L. Faber. 1987. National Surface Water Survey, Western
Lake Survey - Phase I, Analytical Methods Manual. EPA/600/8-87/038.
Environmental Monitoring Systems Laboratory, U.S. Environmental Protection
Agency, Las Vegas, Nevada.
Silverstein, M.E., S.K. Drous3, J.L. Engels, M.L. Faber, andT.E. Mitchell-Hall.
1987a. National Surface Water Survey, Western Lake Survey - Phase I, Quality
Assurance Plan. EPA/600/8-87/026. Environmental Monitoring Systems
Laboratory, U.S. Environmental Protection Agency, Las Vegas, Nevada.
Silverstein, M.E., M.L. Faber, S.K. Drouse, and T.E. Mitchell-Hall. 1987b.
National Surface Water Survey, Western Lake Survey - Phase I, Quality
Assurance Report. Environmental Monitoring Systems Laboratory, U.S.
Environmental Protection Agency, Las Vegas, Nevada.
'Many of the documents are in draft form at the time of this publication
XX vi
-------
A cknowledgments
The Western Lake Survey built upon the approach used in the Eastern Lake
Survey and benefited immensely from the experience gained in the previous
effort. Consequently, although many of the personnel involved in the Western
effort differed, a deep debt of gratitude is owed to all individuals involved
in the design, administration and execution of the Eastern Lake Survey. The
Western Lake Survey represented a challenge both administratively and
logistically. This challenge was met through the dedication and energy of
hundreds of individuals coordinated through the U. S. Environmental Protection
Agency and the USDA Forest Service. The short time period identified for
sampling in the West, the need to access nearly two-thirds of the lakes on
foot, and the rigors of sampling remote, high elevation lakes in pre-winter
conditions made the Western Lake Survey a great challenge. The authors
can acknowledge only a few of those who performed key roles in the Survey
but readily recognize the many others, without whom the Western Lake Survey
could not have succeeded.
L. Edwin Coate, then the Deputy Regional Administrator of EPA Region X,
was the inspirational and dedicated project leader who provided the patience,
experience, and vision necessary to guide this project through many uncertain
times. He also created a working management structure combining elements
from EPA Region X, EPA Office of Research and Development Laboratories,
and the Forest Service that allowed personnel to work together to complete
the Survey successfully. Without his leadership and belief in the critical need
for data on western lakes, this study may never have been completed.
Courtney Riordan, Director of the EPA Office of Acid Deposition, Environmental
Monitoring and Quality Assurance, Office of Research and Development, was
instrumental in early decisions to perform the Western Lake Survey and
supported the entire effort. Gary Foley, Director of the EPA Acid Deposition
and Atmospheric Research Division worked to insure that the Survey would
progress from the design stage to the implementation phase.
The close cooperation between the U.S. Environmental Protection Agency and
the Forest Service could not have been realized without the support and
encouragement of Lee Thomas, Administrator of the U.S. Environmental
Protection Agency and Maxwell Peterson, Chief of the Forest Service. Their
joint leadership and commitment to the successful completion of this regional
assessment are gratefully acknowledged.
Logistics, training and coordination were provided by several organizations
and many dedicated individuals. The following list identifies the major activity
area and lists the individuals involved and their affiliations (abbreviations
defined at end of this section) at the time of the Survey:
XXVII
-------
Logistics Coordination
Lee Marshall
Rick Ross
Richard Dearsley
Robert Cusimano
Stephen Pierett
Kenneth Asbury
John Baker
EPA, Region X
FS
FS
Northrop Services, Inc.
Lockheed-EMSCO
Lockheed-EMSCO
Lockheed-EMSCO
Base Station Operation
Carson City, Nevada:
Arnold Den
Robert Doty
Kenneth Greenberg
Wenatchee, Washington:
David Tetta
Carl Anderson
Ray Peterson
Missoula, Montana:
Rick Claggett
Michael Goggin
Douglas Fox
Bozeman, Montana:
C. E. "Cornie" Runas
Peter Stender
William Warner
Aspen. Colorado:
Lester Sprenger
Dennis Haddow
Larry Svoboda
EPA, Region X, Base Coordinator
FS, Field Manager
EPA, Region X, Duty Officer
EPA, Region X, Base Coordinator
FS, Field Manager
EPA, Region X, Duty Officer
EPA, Region VIII, Base Coordinator
FS, Field Manager
EPA, Region VIII, Duty Officer
EPA, Region VIM, Base Coordinator
FS, Field Manager
EPA, Region VIII, Duty Officer
EPA, Region VIII, Base Coordinator
FS, Field Manager
EPA, Region VIII, Duty Officer
Ken Asbury
Kevin Cabbie
John Henshaw
Franklin Morris
Training and Procurement
Lockheed-EMSCO (EMSL-LV):
John Baker
Linda Drewes
Cindy Mayer
David Peck
Barry Baldigo
Cindy Hagley
Molly Morison
James Wilson
Kevin Cabbie
Mel Knapp
Linda Drewes
C. Hunter Nolen
Field Laboratory Coordinators
Lockheed-EMSCO (EMSL-LV):
Gerald Filbin
Ky Ostergaard
Alan Groeger
Field Laboratory Supervisors
Lockheed-EMSCO (EMSL-LV):
John Henshaw Molly Morison
David Peck
XXVIII
-------
Michael Bonoff
Cindy Hagley
James Wilson
Helicopter Crews
Lockheed-EMSCO (EMSL-LV):
Karen Cougan
Cindy Mayer
Barry Gall
Mark Mitch
Allen Belenz
Betsy Dickes
Lesa Madison
Peggy Permann
Brad league
Field Laboratory Analysts
Lockheed-EMSCO (EMSL-LV):
James Crawford
Elizabeth Hill
Marty McKaig
Susan Piacentine
Gary Turner
David Grouse
David Janick
Katy Nedrow
Elizabeth Sweeney
Sandra Watford
Kathleen Cahall
Julie Segars
Ground Crews-Logistics
Lockheed-EMSCO (EMSL-LV):
Robert Danehy Thomas Loranger
James Williams
Steve Morris
Communications Center in Las Vegas
Lockheed-EMSCO (EMSL-LV):
Valerie Sheppe Brenda Whitfield
Byron Blasdell
Charles Shoemaker
Engineering Support
Lockheed-EMSCO (EMSL-LV):
Jeff Love Michael Reese
John Wengert
Daniel Allison
Bryant Hess
John Lau
Maria Maxwell
Technical and Clerical Support
Lockheed-EMSCO (EMSL-LV):
Ramon Denby
Daniel Hillman
Linda Marks
Peggy Oakes
Annalisa Hall
Henry Kerfoot
Richard Maul
Kathleen Hurley
Sharon Ziminski
Northrop Services, Inc. (ERL-C):
Regina O'Brien Roze Vanden Berg
John Hazard
Statistical Support
Forest Service:
Ronald McRoberts
Sandy Verry
Lockheed-EMSCO (EMSL-LV):
Martin Stapanian
XXIX
-------
Montana State University:
Daniel Goodman
James Blick
Northrop Services, Inc. (ERL-C):
Mark DeHaan
Thomas Permutt
Systems Applications, Inc.
Alison Pollack
Brooke Abruzzesse
Andrew Kinney
Geographic Support
Northrop Services, Inc. (ERL-C):
Sharon Clarke Colleen Johnson
Barbara Rosenbaum
Wesley Kinney
Quality Assurance and Quality Control
EPA, EMSL-LV:
Robert Schonbrod
Lockheed-EMSCO (EMSL-LV):
Karen Cougan
Marianne Faber
Tuijauna Mitchell-Hall
Lynn Creelman
Carol MacLeod
David Peck
Sevda Drouse
Dean Mericas
Sharon Chandler
Cindy Wear
Data Management and Data Entry
ORNL:
Paul Kanciruk Sheila Ladd
Merilyn Gentry
SAIC (ORNL):
Les Hook
Raymond McCord
Stephen Norton
Frank Sanders
John Turk
Robert Wetzel
Robert Wissmar
Report Reviewers
University of Maine, Orono, Maine
University of Wyoming, Laramie, Wyoming
U.S. Geological Survey, Denver, Colorado
University of Michigan, Ann Arbor, Michigan
University of Washington, Seattle, Washington
The Office of Aircraft Services, U.S. Department of the Interior, arranged for
helicopter support and provided flight training for all helicopter samplers.
Training for helicopter and ground crews was coordinated and performed by
John Baker of Lockheed-EMSCO (EMSL-LV). Kevin Cabbie, Lockheed-EMSCO,
was instrumental in selecting the locations of the base stations and ensuring
that required services were provided to the field laboratory and logistics room
at each base site.
xxx
-------
Several universities were involved in contributing to the Western Lake Survey:
Oregon State University-Corvallis, Oregon; State University of New York-
Oswego, New York; and Western Washington University-Bellingham,
Washington. State agencies also were important in providing information on
lakes in their respective jurisdictions and in providing special assistance in
acquiring permission for access. All are acknowledged for their contribution
to the survey.
Many individuals were provided by the Forest Service to coordinate and perform
the ground sampling of lakes in wilderness areas. The enthusiasm,
professionalism, and dedication of these people were a key component of
the success of this project and their stellar performance is gratefully
acknowledged.
Finally, this report was typed by Roze Royce of Northrop (ERL-C). This vital
task was completed accurately, timely, and cheerfully, and the authors would
like to express their gratitude for her important contribution to completing
this report. Regina O'Brien coordinated preparation of figures, and the graphic
artists were Linda Haygarth and Kris Daniels. The assistance of Avis Newell
of Northrop (ERL-C) and Thomas Loranger of Western Washington University
in finalizing the report is equally appreciated.
Abbreviation
Definition
EPA
FS
Lockheed-
EMSCO
ERL-C
EMSL-LV
ORNL
SAIC
SAI
United States Environmental Protection Agency
United States Department of Agriculture-Forest
Service
Lockheed Engineering and Management
Services Company, Incorporated
U.S. EPA Environmental Research Laboratory at
Corvallis, Oregon
U.S. EPA Environmental Monitoring Systems
Laboratory at
Las Vegas, Nevada
Oak Ridge National Laboratory
Science Applications International Corporation
Systems Applications, Inc.
-------
Executive Summary
Background
Phase I of the Western Lake Survey (WLS-I) was
conducted in the fall of 1985 by the U.S. Environ-
mental 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 that contain the majority of lakes
that are considered to be at risk as a result of acidic
deposition. The WLS-I was conducted as part of the
National Acid Precipitation Assessment Program
(NAPAP). It contributes directly to one of NAPAP's
principal objectives: the quantification of the extent,
location, and characteristics of sensitive and acidic
lakes and streams in the United States.
The WLS-I was conducted in five areas in the
western United States (see inside cover). The areas
surveyed included the major mountain ranges in the
West, containing the Sierra Nevada, the Cascade
Range, and the Rocky Mountains. Five subregions,
delineated on the basis of similar physiographic
characteristics, were identified: California (4A), the
Pacific Northwest (4B), the Northern Rockies (4C),
the Central Rockies (4D), and the Southern Rockies
(4E). An analogous survey of eleven subregions was
conducted in the fall of 1984 in three regions of
the eastern United States. These regions were the
Northeast, the Upper Midwest, and the Southeast.
The primary objectives of the WLS-I were to
determine in potentially sensitive areas of the
western United States: (1) the percentage (by number
and area) and location of lakes that are acidic; (2)
the percentage (by number and area) and location
of lakes that have low acid neutralizing capacity
(ANC); and (3) the chemical characteristics of lake
populations, providing a data base for selecting lakes
for future studies.
Because not all lakes in the western United States
could be sampled, a statistical procedure was
developed for selecting a subset of lakes as a
probability sample. Within each subregion, three
areas expected to be dominated by lakes with
alkalinity < 100 /ueq L~1, 100-199 /ueq L"1, or 200-
400 //eq L~1 were identified on maps. A systematic
procedure with a random starting point was used
to select lakes from each of the three areas. This
design ensured that each lake within each of the
three alkalinity map classes was equally likely to be
selected; thus, the physical and chemical charac-
teristics of the population of lakes could be estimated
from sample results with a known degree of
confidence. These lakes are referred to as probability
sample lakes. The number of lakes in the probability
sample upon which population estimates were based
was 719, representing an estimated 10,393 lakes
in the target population.
To maximize the number of lakes that could be
surveyed, a single sample was collected from each
lake. Fall was selected as the optimal period for
sampling because chemical variability within a single
lake is expected to be low in most temperate lakes
as a result of circulation of the water column. Data
from the small percentage of lakes that were
stratified did not alter the results. The results of the
WLS-I are applicable to the fall season of sampling;
representation of other periods during the year in
western lakes is unknown. For example, nitrate
concentrations in the lakes from the fall survey may
be much lower than values measured in the spring.
The prescribed method of sample collection for the
lake survey within the NSWS was by helicopter. Of
the 752 lakes identified for sampling in WLS-I, 455
were located in wilderness areas, to which motorized
access is not permitted. Thus, lakes located outside
wilderness areas were accessed by hel icopter crews,
and wilderness lakes were accessed by ground
crews. Ground sampling was conducted by person-
nel from the U.S. Department of Agriculture-Forest
Service. To evaluate whether there was a difference
between data collected by ground crews and by
helicopter crews, the Forest Service granted
permission to conduct a special study in which 45
lakes located in wilderness areas were sampled
using both methods. The results of this study
indicated that samples collected using the ground
method, even with longer holding times, yielded data
that were comparable to those obtained from
samples collected using helicopters.
Water samples were delivered to mobile field
laboratories for preliminary processing before they
were sent to one of two analytical laboratories for
chemical analysis. Extensive quality assurance and
XXXIII
-------
data management programs were implemented to
ensure that all sample collection activities, analytical
methods, and data reporting procedures were
performed within the limits of pre-established
criteria designed to meet specifically designed data
quality objectives. For nearly all variables, these
criteria were met; however, statistically significant
laboratory biases were observed for some variables.
The small magnitude of the bias had no perceptible
effect on the population estimates, but users of the
data should be cognizant of possible limitations in
the data.
This report, Chemical Characteristics of Lakes in the
Western United States, consists of two volumes.
Volume I, Population Descriptions and Physico-
chemical Relationships, provides details about the
design and implementation of the WLS-I, discusses
results, and presents conclusions based on these
results. Volume II, Data Compendium for Selected
Physical and Chemical Variables, contains additional
estimations of the characteristics of the populations
of lakes sampled within each area, descriptive
statistics for these lake populations, and a data
compendium of site characteristics and chemical
variables for each lake sampled.
The purpose of this report is to describe, within the
context of the survey design, the chemical charac-
teristics of lakes in the western United States, the
sampling and analytical protocols, and the quality
of the data obtained. These aspects of the WLS-I,
described in detail in Sections 2 and 3, must be
understood before conclusions beyond the scope of
the original objectives can be drawn. The WLS-I data
cannot be used alone to establish a relationship
between the present chemical status of lakes and
the occurrence of acidic deposition. However, the
data can be used to determine associations among
variables. The data base also can be used to refine
our understanding of the response of lakes to acidic
deposition by permitting subpopulations of lakes to
be selected to test hypotheses regarding aquatic
response to acidic deposition.
Selected Results
To meet the three primary objectives of the survey,
the data from the WLS-I were used to examine the
characteristics of lakes in the West, both within and
among the five western subregions. Population
estimates were calculated for the target population
of lakes (10,393 lakes; > 1 ha and < 2000 ha) in
the West in the five subregions sampled. The
population estimates apply only to those geographic
areas identified for the study (see inside cover).
Estimates were also calculated for special popula-
tions of lakes including those located within various
geomorphic units generally coinciding with moun-
tain ranges, dilute lakes (those with conductance
< 10 /jS cm"1), and lakes within wilderness areas
and national parks. Associations among variables
also were examined through correlative analyses.
Extent and Location of Lakes with Low Acid
Neutralizing Capacity (ANC) and Low pH
Of the probability sample lakes, only one was acidic
(ANC < 0 Aieq L~1). Only one lake had pH < 5.0.
An estimated 16.8 percent of the target population
of lakes in the western United States had low ANC
(< 50 £ 6.0.
California (Subregion 4A) had the highest estimated
number and percentage of lakes with low ANC in
the West. Lakes with low ANC were located
throughout the sampling area in California, but
particularly in the Sierra Nevada.
Pacific Northwest (Subregion 4B) had the second
highest estimated number and percentage of low
ANC lakes in the West. These lakes were located
along the Cascade crest in Oregon and Washington.
The Northern Rockies (Subregion 4C) had the third
highest estimated number of lakes with low ANC.
Most of these lakes were located in the Bitterroot
Range along the Idaho-Montana border and in the
Sawtooth Mountains of Idaho.
The Central Rockies (Subregion 4D) had the fourth
highest percentage of lakes with low ANC. Low ANC
lakes in this Subregion were generally located in the
Wind River Range and the Big Horn Mountains in
Wyoming, the Absaroka-Beartooth Mountains in
Wyoming and Montana, and the Uinta Mountains
in Utah.
The Southern Rockies (Subregion 4E) had the fewest
lakes with low ANC. These were located primarily
in the San Juan Mountains and the Front Range
in Colorado.
Special Populations
Many of the lakes in the West had very low
concentrations of dissolved constituents. Concentra-
tions of most measured variables, such as base
cations (Ca+2, Mg+2, K+, Na+), major anions (HC03~,
SO/2, Cl~) and nutrients (total P, N03~, NH4+) were
very low in these dilute lakes (i.e., conductance <
10 /uS cm"1). Western lakes were nearly all clear-
water (true color < 30 PCU), and were very
transparent.
Lakes in the wilderness areas, representing approx-
imately two-thirds of the total target population in
the West, had substantially lower concentrations of
major anions and cations than lakes in non-
wilderness areas. Median ANC values for lakes in
the wilderness areas were 91.4 peq L"1 compared
-------
Table E-1. Population Estimates and Statistics for
Phase 1
Subregion
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
West
Subregion
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
West
n*** number of
ft estimated
PC estimated
ftc estimated
(4A)
(4B)
(4C)
(4D)
(4E)
(4)
(4A)
(4B)
(4C)
(4D)
(4E)
(4)
n***
149
159
143
129
139
719
n***
149
159
143
129
139
719
ft
2401
1706
2379
2299
1609
10393
ft
2401
1706
2379
2299
1609
10393
Acid Neutralizing Capacity and pH by Subregion, Western Lake Survey -
PC
0367
0195
0.127
0.069
0046
0.168
PC
0.013
0024
0000
0.013
0000
0010
lakes sampled.
target population size.
proportion of lakes with given criteria
number of lakes with given criteria.
ANC< 50A 50 jueq L"1 and with calcium < 50 /ueq L"1. Of
the few clearwater western lakes (16) with extrac-
table aluminum > 50 /ug L"1, most were located in
this subregion.
As was observed for lakes in other subregions,
concentrations of nitrate, ammonium and total
phosphorus were very low. The highest median
values for silica, manganese, and chloride were
observed in the Pacific Northwest. Concentrations
of other chemical constituents were similar to those
measured in the three Rocky Mountain subregions.
Northern Rockies (4C)--
In the Northern Rockies, an estimated 23.7 percent
of the target population of lakes had conductance
values < 10 //S cm"1, similar to the percentage of
dilute lakes estimated for the Pacific Northwest, but
much higher than the percentages estimated for the
Central and Southern Rockies. Relative to other
western subregions, lakes in the Northern Rockies
were intermediate in estimated median sulfate
concentration and percentage of lakes with sulfate
> 50 /ueq L"1. The median values for calcium and
extractable aluminum were the second highest for
any subregion surveyed, although all clearwater
lakes sampled had extractable aluminum <50//g L"1.
Dissolved organic carbon concentrations were also
low; only an estimated 8.2 percent of the lakes had
XXXV
-------
DOC > 6 mg L"1. Median values for most secondary
variables were intermediate relative to those
estimated for other subregions.
Central Rockies (4D)~
It was estimated that slightly more than 10 percent
of the lakes in the Central Rockies were dilute, which
was substantially less than the percentages
observed in California, the Pacific Northwest, and
the Northern Rockies. This subregion hadthe second
highest estimated median sulfate value, but the
lowest percentage and number of lakes in the West
with sulfate > 50 //eq L"1. Lakes in the Central
Rockies had the lowest median calcium concentra-
tion of the three Rocky Mountain subregions and
the second lowest number and percentage of lakes
in the West with calcium < 50 /ueq L"1. The estimated
median concentration of extractable aluminum for
clearwater lakes was the lowest among all subre-
gions, and no lakes were estimated to have
concentrations of extractable aluminum > 50 /ug L"1.
As in other western subregions, lakes in this
subregion had generally low DOC concentrations.
Southern Rockies (4E)~
Of the five western subregions, it was estimated-that
the Southern Rockies had the fewest dilute lakes
and the highest median calcium concentrations. An
estimated 33.7 percent of the lakes (543) had sulfate
concentrations > 50 /ueq L"1, nearly twice the
percentage and number of lakes in the Pacific
Northwest. The lakes with high sulfate concentra-
tions were generally associated with high ANC lakes.
Calcium concentrations exceeded 50/ueq L"1 in 96.6
percent of the lakes. All clearwater lakes sampled
had extractable aluminum concentrations < 50
/ug L"1. Dissolved organic carbon was generally
higher than in the other subregions.
Lakes in the Southern Rockies had the highest
estimated median values for nitrate, total phospho-
rus, sodium, potassium, magnesium, iron, dissolved
inorganic carbon, conductance, and bicarbonate.
These lakes also had the highest median true color
and turbidity, resulting in the lowest median
transparency of any western subregion.
Associations Among Variables
In most western lakes surveyed, the dominant cation
was calcium and the dominant anion was bicarbo-
nate. Sodium was the second most abundant cation
in California and the Pacific Northwest, probably due
to mineral weathering in California and the Oregon
Cascades and to mineral weathering and atmospher-
ically deposited sea salt in the North Cascades. In
the Rocky Mountains, magnesium was the second
most abundant cation, except in lakes with lower
total cation concentrations in the Northern Rockies,
where sodium was approximately equal to magne-
sium. Based on the relative abundance of major
anions and cations and on patterns observed in ratios
of major cations and major anions, most lakes
surveyed in the West can be classified as calcium
bicarbonate systems.
A positive relationship between calcium and sulfate
concentrations was observed in the lakes of the
Rocky Mountains, suggesting that some sulfate is
derived from mineral sources. An analysis of the
relationship between ANC and the sum of calcium
and magnesium in the western subregions, for which
the intercepts of the regressions were near zero and
the slopes were near one, suggests that wide-scale
acidification of lakes has not occurred in these areas.
The relationship between low ANC in lakes and
bedrock type indicates that geological characteristics
of watersheds in the western United States play a
major role in determining the ANC levels in surface
waters. Consequently, bedrock type is an important
factor determining the potential sensitivity, in terms
of static ANC measurements, of western lakes to
acidic deposition.
No relationship was found between lake area and
water chemistry for lakes < 30 ha, except that lakes
< 4 ha in the Northern Rockies were more dilute
than those lakes > 4 ha. Lakes < 30 ha had lower
ANC values than lakes > 30 ha. Elevation of lakes
was not always correlated with lake water chemistry.
Acid neutralizing capacity decreased with elevation
in the Wind River Range in Wyoming and in the
California subregion, but no relationship was
observed in the Pacific Northwest, the Northern
Rockies, or the Southern Rockies.
Differences in water chemistry among hydrologic
lake types were observed. Reservoirs in California
and the Southern Rockies had substantially higher
median ANC and calcium values than did drainage
or seepage lakes; in California and the Pacific
Northwest, drainage lakes had slightly higher
median ANC, calcium and sulfate concentrations and
lower median DOC concentrations than did seepage
lakes. In the Rocky Mountain subregion, similar
results were found for sulfate and DOC, but seepage
lakes had somewhat higher ANC than did drainage
lakes.
Comparison of Western Lakes to Eastern
Lakes
In contrast to lakes surveyed in the eastern United
States, no acidic (ANC < 0 peq L~1) lakes were
sampled in the West that were not associated with
hot springs. Of the four regions surveyed in the
xxxvi
-------
National Lake Survey, the highest percentage of
lakes with ANC < 200 //eq L"1 was observed in the
West. Only one lake with pH < 5.0 was sampled
in the West; in the East, 10 percent of the lakes
in the Adirondacks, 9.4 percent in the Upper
Peninsula of Michigan, and 12.4 percent in Florida
were estimated to have pH < 5.0. Only 1.0 percent
of western lakes had pH < 6.0, whereas in the East,
the percentage ranged from 9.6 percent in the Upper
Midwest to 32.7 percent in Florida.
Sulfate concentrations in most western lakes were
substantially lower than those in most eastern lakes.
The estimated median sulfate concentration in the
Northeast was approximately six times that for
western lakes as a whole. In the West, only 13.5
percent of the lakes had sulfate > 50 //eq L"1,
whereas 97.3 percent of the lakes in the Northeast,
61.1 percent in the Upper Midwest, and 68.1 percent
in Florida had sulfate > 50 //eq L"1.
Estimated regional calcium concentrations in
western lakes were also substantially lower than in
eastern lakes. The total estimated number of lakes
in the western target population with calcium < 50
//eq L"1 exceeded that for all eastern regions
combined. The number of clearwater lakes in the
West with extractable aluminum concentrations
> 50 //g L"1 was much smaller than that observed
in the East. Dissolved organic carbon concentrations
were also much lower in western lakes than in
eastern lakes with the exception of lakes in the
Southern Blue Ridge subregion; 69.7 percent of the
lakes in the West were estimated to have DOC < 2
mg L~1, whereas only 8.1 percent of the lakes in
the Northeast, 1.3 percent in the Upper Midwest,
and 11.5 percent in Florida had DOC < 2 mg L"1.
Conclusions
Within the areas of the western United States
sampled and covered by the design of the Western
Lake Survey - Phase I, it is concluded that:
• More than one-fourth (26.6 percent) of the
western lakes are dilute (conductance < 10
//S cm"1). The areas sampled in the western United
States contain a much higher percentage of lakes
with low concentrations of dissolved substances
than do areas sampled in three regions (Northeast,
Upper Midwest, and Southeast) in the eastern
United States. The lowest Qi value (20th percen-
tile) for conductance measured in the East was
20.7 pS cm"1 and the minimum value was 7.8
fjS cm"1. In contrast, 20 percent of the lakes in
the West had conductance values < 8.3 //S cm"1
and the minimum measured conductance in the
West was 1.6//S cm"1.
• Calcium concentrations in the West varied widely
among subregions; 56.5 percent of the lakes in
California had calcium <50//eq L 1, whereas only
3.4 percent of the lakes in the Southern Rockies
were below this criterion. Overall, the median
concentration of calcium for lakes in the West
was low (92.4 //eq L~1) compared to the Northeast
(177.4 //eq L"1), the Upper Midwest (238.2//eq L"1),
and Florida (238.3 //eq L"1). Lakes in the Southern
Blue Ridge had a comparable median value for
calcium (104.7 //eq L"1).
• Median sulfate concentrations were extremely
lowthroughoutthe West, ranging from 6.6//eq L~1
in California to 34.6 //eq L"1 in the Southern
Rockies. In comparison, median concentrations
for lakes in the East were 115.4 //eq L"1 for the
Northeast, 57.1 //eq L~1 for the Upper Midwest,
31.8 //eq L"1 for the Southern Blue Ridge, and
93.7 //eq L~1 for Florida.
• Extractable aluminum concentrations in clear-
water lakes were very low throughout the West.
Only 0.2 percent of the lakes in the West had
extractable aluminum concentrations exceeding
50 //g L"1 compared to 5.5 percent of the lakes
in the Northeast, and 7.4 percent of the lakes in
Florida. These results are, again, similar to those
for the Southern Blue Ridge, where no lakes
exceeded this criterion.
• Concentrations of dissolved organic carbon were
also low throughout the West; only 5.4 percent
of the lakes in the West had DOC values exceeding
6 mg L"1, whereas this criterion was exceeded
for 26.4 percent of the lakes in the Northeast,
62.9 percent in the Upper Midwest, and 68.9
percent in Florida. Again, the Southern Blue Ridge
shows a similarity to western lakes with 6.1
percent of the lakes with DOC > 6 mg L~1.
• pH values were not low in the West, where 99
percent of lakes had values greater than 6.0.
Median pH values for the subregions ranged from
6.94 in California to 7.60 in the Southern Rockies.
• Of fhe subregions sampled in the West, California
had the largest number (881) and percentage
(36.7%) of lakes with low ANC (< 50 //eq L"1);
followed by the Pacific Northwest (333 lakes;
19.5%). The Southern Rockies had the smallest
number and lowest percentage of lakes with low
ANC (74 lakes; 4.6 %).
• The lakes in wilderness areas had much lower
concentrations of ANC with a median value of
91.4 //eq L~1 compared to 282.7 //eq L"1 for non-
wilderness lakes.
• The Sierra Nevada (the area which contained the
majority of lakes sampled in California) had the
largest estimated number of low ANC lakes (834)
XXXVII
-------
and dilute lakes (1311) of any geomorphic unit
in the West.
No lakes sampled in the West had ANC
-------
Section 1
Introduction
Assessment of the effects of acidic deposition on
lakes in the western United States has received
considerable scientific attention relatively recently
(Lewis and Grant 1979; McColl 1981; Turk 1983;
Smith and Alexander 1983; Kling and Grant 1984;
Rothetal. 1985; Hidy and Young 1986). Few of these
studies reported decreases in acid neutralizing
capacity or pH in lakes as a result of acidic deposition.
Many of the lakes in the West have low alkalinity
(Omernik and Griffith 1986), and are, therefore,
considered to be potentially sensitive to acidic
deposition; however, their remote location and
difficult access have resulted in a limited under-
standing of their chemical characteristics. In
addition, regional patterns of acidic deposition in the
West have not been easily identifiable because of
an insufficient number of deposition monitoring sites
(Munger and Eisenreich 1983; Stensland 1984;
Stensland et al. 1986; Hidy and Young 1986). This
lack of monitoring sites increases the difficulty of
assessing the extent to which western lakes are at
risk as a result of acidic deposition. In contrast, in
the eastern United States where levels of acidic
deposition have been monitored much more exten-
sively, the relationships between acidic deposition
and surface water chemistry have been studied more
intensively (Galloway et al. 1976; Likens 1976;
Altshuller and Linthurst 1984).
To provide a high quality data base for use in
quantifying the present chemical status of western
lakes, the U.S. Environmental Protection Agency
(EPA) designed and implemented the Western Lake
Survey - Phase I (WLS-I) as part of the National
Surface Water Survey (NSWS). The NSWS is a three-
phase project with two components, the National
Lake Survey (NLS) and the National Stream Survey.
Phase I is designed to quantify the present chemical
status of U.S. surface waters. Phase II is designed
to assess the chemical variability and to quantify
biological resources in surface waters; and Phase
III will identify temporal trends in surface water
chemistry and biology.
Initially, the NLS was scheduled to be conducted
simultaneously in four regions of the United States
(the West, Upper Midwest, Northeast, and
Southeast) during the fall of 1984. As planning
progressed, it became apparent that logistical
considerations for the western effort were compli-
cated by rigorous terrain and weather and by
restrictions on motorized access to sampling sites.
Consequently, the NLS was divided into the Eastern
Lake Survey - Phase I (ELS-I) during which 1798
lakes were sampled in the fall of 1984 (Linthurst
et al. 1986) and the WLS-I during which 752 lakes
were sampled in the fall of 1985. Deferring the WLS-I
permitted the special conditions in the West to be
considered fully before field activities were imple-
mented.
Nearly two-thirds of the lakes sampled in the WLS-I
are located in federally designated wilderness areas
to which motorized access is prohibited (Wilderness
Act of 1964). Because the prescribed method of
accessing lakes in the NLS was by helicopter
(Linthurst et al. 1986), an alternative method of
sampling wilderness area lakes was proposed by the
Forest Service to EPA: personnel from the Forest
Service would serve as ground access sampling
teams and as field site managers. Additionally,
because of concerns that data collected by ground
access methods would not be comparable to those
collected by helicopter access methods, the Forest
Service granted permission to sample 50 wilderness
lakes by both methods (45 of which were actually
sampled). The results of this paired sampling were
used to determine whether or not the data collected
by ground crews needed to be calibrated against
those collected by helicopter crews.
The WLS-I was designed to provide information
to assess the chemical status of lakes in areas of
the western United States containing the majority
of lakes with low acid neutralizing capacity (Omernik
and Griffith 1986). Lakes were selected from five
subregions of the West (see inside front cover):
California (Subregion 4A), Pacific Northwest (Sub-
region 4B), Northern Rockies (Subregion 4C), Central
Rockies (Subregion 4D), and Southern Rockies
(Subregion 4E). The boundaries of these subregions
generally separated geomorphic provinces, and most
of the study areas were at high elevations. A number
of variables thought to influence or be influenced
by surface water acidification were measured using
standardized methods.
-------
The WLS-I data base may be used to investigate
correlative relationships among chemical variables
and to quantify the present chemical status of lakes
in areas of the western United States covered in
the survey. Additionally, the data from the WLS-I
can be used with the ELS-I data to obtain a national-
scale assessment of the extent to which lakes in
the United States are potentially at risk as a result
of present levels of acidic deposition.
The primary objectives of the WLS-I were to:
1. determine the percentage, by number and area,
and location of lakes that are acidic in
potentially sensitive regions of the western
United States;
2. determine the percentage, by number and area,
and location of lakes that have low acid
neutralizing capacity in potentially sensitive
regions of the western United States; and
3. determine the chemical characteristics of lake
populations in potentially sensitive regions of
the western United States and provide the data
base for selecting lakes for future studies.
The purpose of this report is to present results
obtained from the WLS-I. The data base generated
during the WLS-I will be used to frame subsequent
projects designed to examine short-term acidifica-
tion as a result of episodic events and long-term
acidification through an evaluation of trends in
aquatic ecosystems.
-------
Section 2
Methods
2.1 Design
2.1.1 Data Quality Objectives
The statistical design, sampling and analytical
methods, and quality assurance (QA) activities for
the Western Lake Survey - Phase I (WLS-I) were
structured to meet specific data quality objectives
(DQOs; Silverstein et al. 1987a) for reporting
population estimates and chemical variability. The
DQOs were used to evaluate sampling, field
laboratory, and analytical laboratory performance.
The primary DQOs for the WLS-I (Table 2-1) were
defined in terms of precision, accuracy, and
detectability. Precision was expressed as percent
relative standard deviation. Accuracy was expressed
as maximum absolute bias, in percent. Detectability
was expressed as an expected value range and a
required detection limit. Each parameter measured
at the lake sampling sites, in the field laboratories,
and in the analytical laboratories was evaluated with
respect to the DQOs, for which appropriate values
and ranges were established. The values and ranges
used for the WLS-I were based on experience gained
during the Eastern Lake Survey - Phase I (ELS-I;
Drouse et al. 1986) and on standard laboratory QA
requirements. Certain other DQOs were also
considered in the survey design. These included
completeness, comparability, and representative-
ness.
Completeness was defined as a measure of the
quantity of acceptable data actually collected relative
to the total quantity that was attempted. For the
WLS-I, completeness was targeted at or above 90
percent for all parameters. That is, of the probability
sample lakes selected (Section 2.2.1), 90 percent
were expected to yield samples that would meet the
QA criteria and that could be used in defining
characteristics of the target population.
Comparability of data is a measure by which
similarity within and among data sets can be
established confidently. Comparability can be
considered in the context of comparisons among data
sets of different origin and also within components
of a single data set. Comparability of the WLS-I data
was ensured by requiring that all sampling crews
and laboratory analysts use standardized proce-
dures. Comparability of the WLS-I data was
confirmed by checking the analytical results against
known standards and comparing the results to data
collected in previous studies on the same lakes
(Section 7).
Representativeness, another measure of data
quality, is the degree to which sample data accurately
and precisely reflect the characteristics of the
population of lakes. Representativeness of a subset
of sampled lakes can be established with complete
confidence only when the sample results can be
compared to the total population of lakes. Obviously,
representativeness of the WLS-I data cannot be
established unequivocally because complete survey
data for all lakes in the West are unavailable. In
the absence of this census information, the best
assurance that the subset of lakes sampled repre-
sents the target population is obtained through the
implementation of a sound, statistically based
sampling design (Section 2.2).
Additional confidence in the representativeness of
the sampled lakes is obtained by comparing results
from other statistically based surveys conducted in
the same areas. The results of one such survey by
Bradford et al. (1968) of 170 Sierra Nevada lakes,
and of a more recent survey of 124 of these same
170 lakes by Fox et al. (1982), are very similar to
those obtained in the WLS-I (Sections 5 and 7). This
agreement indicates, but does not necessarily
confirm, the comparability and the representative-
ness of the WLS-I data.
2.7.2 Summary of the Statistical Design
During the WLS-I, one sample per lake was collected
during the fall turnover period from the apparent
deepest part of the lake as an index to the essential
characteristics of each lake. The fall turnover period
was selected because lake water chemistry within
any single lake was expected to be the most
homogeneous during this season. Lakes were
selected using a stratified design with equal
allocation of sample lakes to strata. Lakes were
selected in each stratum by systematic sampling of
an ordered list following a random start. The choice
of a desired sample size of 50 target lakes per stratum
-------
was based on the judgment that this sample size
would yield adequate precision for population
estimates by stratum.
2.7.3 Definition of Study Area
The population of lakes to be sampled was defined
as lakes located in those areas of the western United
States expected to contain an abundance of lakes
with alkalinity < 400 /ueq L~1. The boundaries of the
western region (Figure 2-1) were derived from a
national map of surface water alkalinity (Omernik
and Powers 1983) and were considered to contain
from 95 to 99 percent of the lower alkalinity lakes
in the western United States. Unlike the subregions
sampled in the Northeast and Upper Midwest in the
ELS-I (Linthurst et al. 1986), no subregions in the
West were conterminous, because the low alkalinity
areas of the West generally coincide with major
mountain ranges.
2.2 Lake Selection
2.2.1 Probability Sample
The stratification factors used in lake selection were
region, subregion, and alkalinity map class. The West
was defined as one region (Region 4 in the National
Lake Survey) with five subregions (4A through 4E).
Descriptive names of subregions were assigned
(Figure 2-1). Lakes representing each of the three
alkalinity map classes (ANC < 100, 100-199, 200-
400 fjeq L"1 derived from Omernik and Griffith 1986,
Figure 2-1) were found within each of the five
subregions; thus, 15 strata were defined in the
WLS-I.
Lakes were identified and listed using 1:100,000-
scale U.S. Geological Survey (USGS) topographic
maps in contrast to the ELS-I, in which 1:250,000-
scale maps were used for lake selection. Use of
1:100,000-scale maps permitted lakes with a
Table 2-1. Data Quality Objectives for Detectability, Precision, and Accuracy, Western Lake Survey - Phase I
Detectability
Site3
2, 3
3
3
3
3
1, 3
2
2, 3
3
3
3
3
3
Variableb
Al,
extractable
Al, total
Acid Neutral-
izing Capacity
(ANC)
Ca
cr
Conductance
True Color
Dissolved
Inorganic
Carbon (DIG)
Dissolved
Organic
Carbon (DOC)
F~, total
dissolved
Fe
K
Mg
Method0
Complexation with 8-
hydroxyquinoline and
extraction into MIBK
followed by AAS
(furnace)
AAS (furnace)
Titration and Gran
analysis
AAS (flame) or
ICPAES
Ion chromatography
Conductivity cell
and meter
Comparison to
platinum-cobalt
color standards
Instrumental (acid-
ification, C02
generation, IR
detection)
Instrumental (UV-
promoted oxidation,
CO2 generation, IR
detection)
Ion-selective
electrode and
meter
AAS (flame) or
ICPAES
AAS (flame)
AAS (flame) or
ICPAES
Laboratory
Reporting
Units
mg L"1
mg L"1
peq L~1
mg L~1
mg L"1
^S cm"1
PCUh
mg L~1
mg L-1
mg L"1
mg L"1
mg L"1
mg L"1
Expected
Ranged
0.005-1.0
0.005-1
-100 -
0.5 -
0.2 -
10 -
0 -
0.05 -
0.1 -
0.01 -
0.01 -
0.1 -
0.1 -
.0
1000
20
10
1000
200
15
50
0.20
5.0
1.0
7.0
Required
Detection
Limit (RDL)
0.005
0.005
f
0.01
0.01
g
0
0.05
0.1
0.005
0.01
0.01
0.01
Intralaboratory
Precision
%RSD
Upper Limit6
10(AI>0.01 mg L~1)
20IAK0.01 mg L~1)
10(AI>0.01 mg L~1)
20IAK0.01 mg L~1)
10
5
5
2
±5'
10
5(DOC>5 mg L'1)
10(DOC<5 mg L~1)
5
10
5
5
Accuracy
Maximum
Absolute
Bias(%)
10
20
10
20
10
10
10
5
i
10
10
10
10
10
10
10
(continued)
-------
Table 2-1. (continued)
Detectability
Site8
3
3
3
3
3
1, 2
3
3
3
2
Variableb
Mn
Na
NH4 +
N03-
P, Total
PH
pH
Si02
so4-2
Turbidity
Method0
AAS (flame) or
ICPAES
AAS (flame)
Automated color-
imetry (phenate)
Ion chromatography
Automated color-
imetry (phospho-
molybdate)
pH electrode and
meter
pH electrode and
meter
Automated color-
imetry (molybdate
blue)
Ion chromatography
Instrument
(nephelometer)
Laboratory
Reporting Expected
Units Ranged
mg L~1
mg L~1
mg L~1
mg L~1
mg L~1
pH units
pH units
mgL-1
mg L~1
NTUk
0.01 -
0.5 -
0.01 -
0.01 -
0.005 -
3 -
3 -
0.2 -
1.0 -
2 -
5.0
7.0
2.0
5
0.070
8
B
250
20.0
15
Required
Detection
Limit (RDL)
0.01
0.01
0.01
0.005
0.002
j
i
0.05
0.05
2
Intralaboratory
Precision
%RSD
Upper Limit6
10
5
5
10
10(P>0.01 mg L"1)
20(P<0.01 mg L"1)
±0.1'
±0.05'
5
5
10
Accuracy
Maximum
Absolute
Bias(%)
10
10
10
10
10
20
±0.10'
±0.10'
10
10
10
a 1 = lake site, 2 = field laboratory, 3 = analytical laboratory.
bDissolved ions and metals were determined, except where noted.
c Methods descriptions are given in Volume II (Tables 2-1 and 5-3), Hillman et al. (1986) and Kerfoot and Faber (1987).
dRanges are for lake waters.
"Unless otherwise noted, this is the percent relative standard deviation (% RSD) at concentrations greater than 10 times the required
detection limit.
f Absolute blank value was required to be < 10 ^ieq L~1.
8The mean of six nonconsecutive blank measurements was required to be <0.9 /jS cm"1.
hAPHA platinum cobalt units.
' Absolute precision or accuracy goal in applicable units.
' Not applicable.
kNephelometric turbidity units.
minimum size of approximately 1 hectare (ha) to be
identified, as opposed to the approximately 4 ha used
in the ELS-I.
Strata boundaries were delineated, and all lakes on
the maps were numbered in spatial order within each
stratum. The final number of lakes identified in each
stratum was the total number of lakes in the map
population for the stratum. All population estimates
for physical and chemical attributes computed in this
study refer to the map population of lakes and do
not necessarily represent conditions in lakes outside
the area of coverage or in those not depicted on
the USGS topographic maps used. For example,
population estimates cannot be made for lakes
smaller than 1 ha.
Within each stratum, a systematic random sample
of 50 lakes was drawn. Lake numbers were entered
into a computer file in numerical order as labeled
on the maps. In each stratum, the first lake was
selected at random between lakes 1 and k (where
k is the size of the map population divided by the
desired sample size), and every kth lake was selected
thereafter. This sample is a true probability sample,
i.e., within each stratum, each lake had an equal
probability of inclusion.
2.2.2 Identification of Non-target Lakes
"Non-target" lakes (Table 2-2) are those bodies of
water that either were not the focus of the survey's
objectives or could not be sampled within the
constraints of a synoptic survey. Non-target lakes
(Table 2-2) were first identified in the probability
sample by the examination of large-scale (1:24,000
or 1:62,500) USGS topographic maps. Categories of
non-target lakes identified by examining these maps
include:
1. No lake present: lakes initially identified on
1:100,000-scale maps that did not appear on
more detailed, larger scale maps.
2. Flowing water: sites identified as lakes on
1:100,000-scale maps that appeared as points
on a stream on larger scale maps. However,
if the small-scale maps were more recent than
-------
Figure 2-1. Subregions and alkalinity map classes ( • .• < 100 fjeq i"1; Q : 100-199 yeq L 1; D . 200-400 fjeg L~\ in study
areas of the Western Lake Survey - Phase I.
Northern
Rockies (4C)
Alkalinity Map Classes
<100
100-199
200-400
Subregion
Boundary
Central
Rockies (4D)
Pacific
Northwest (4B)
California (4A)
Southern
Rockies (4E)
the large-scale maps and the lake in question
was known to be a new reservoir, the lake was
not eliminated.
3. Urban/Industrial/'Agricultural: lakes sur-
rounded by or adjacent to intense urban,
industrial, or agricultural land use including
tailing ponds, water treatment lagoons, andfish
hatcheries.
4. Wetlands: lakes identified on 1:100,000-scale
maps that appeared as wetlands on larger scale
maps.
5. Too small (< 1 ha): lakes identified on
1:100,000-scale maps that were smaller than
approximately 1 ha when measured on larger
scale maps. Because the resolution of most
1:100,000-scale maps was about 1 ha, this
limit was established for consistency.
Following the elimination of non-target lakes by map
examination, it was necessary to restore the number
of selected lakes per stratum to 50. Also, in the event
that field crews encountered non-target lakes,
additional lakes were selected. These additional
-------
lakes were chosen from the computer file of lakes
remaining after initial selection using the same
procedure as for the original set of lakes, and were
also evaluated on large-scale maps to eliminate non-
target lakes.
The lakes that met the selection criteria after the
map evaluation were provisionally designated as
"target" lakes. This designation was refined further
as a result of information obtained during or after
field sampling. The categories and numbers of non-
target lakes eliminated during or after sampling are
given in Table 2-2. The definitions of these categories
are as follows:
1. No lake present: sites visited that were found
to be dry.
2. Flowing water: sites visited that were found
to be streams.
3.
5.
6.
High conductance: lakes visited that were found
to have a measured conductance greater than
1500/uScm"1.
Urban/Industrial/Agricultural: lakes that were
surrounded by or were adjacent to intense
anthropogenic activities.
Too shallow: lakes that were too shallow
(generally, less than 0.75 m) to obtain a clean
(i.e., free of debris and sediment) sample.
Other: lakes that were inaccessible due to a
permanent feature of the lake that prevented
helicopters from landing safely (e.g., power
lines).
Lakes were classified as "not visited" (Table 2-2)
if the reason for not sampling was unrelated to a
permanent feature of the lake. This category of lakes
includes those that could not be visited because
permission to sample was not obtained or weather
Table 2-2. Non-Target Lakes and Lakes Not Visited or Not Sampled, Western Lake Survey - Phase I
A. Probability sample lakes determined to be non-target from large-scale map examination
California
Categories
No Lake Present
Flowing Water
Urban/lndustria I/Agricultural
Wetlands
Too Small « 1 ha)
4A
7
1
2
2
14
Pacific
Northwest
4B
0
0
1
4
24
Northern
Rockies
4C
1
0
0
2
7
Central
Rockies
4D
1
1
0
1
18
Southern
Rockies
4E
1
1
3
0
3
Total
10
3
6
9
66
Total
26
29
10
21
94
B. Non-target probability sample lakes determined during or after sampling
California
Categories
No Lake Present
Flowing Water
High Conductance
Urban/lndustnal/Agncultural
Too Shallow
Other
4A
7
0
0
1
10
0
Pacific
Northwest
4B
2
0
0
0
15
0
Northern
Rockies
4C
5
0
1
0
18
0
Central
Rockies
4D
6
0
0
0
14
2
Southern
Rockies
4E
2
0
1
0
13
1
Total
22
0
2
1
70
3
Total
18
17
24
22
17
98
C. Probability sample lakes that were not visited
Categories
No Access Permit
Bad Weather
Wrong Lake
Frozen
California
4A
2
0
0
1
Pacific
Northwest
4B
4
1
0
0
Northern
Rockies
4C
0
0
1
7
Central
Rockies
4D
0
1
0
21
Southern
Rockies
4E
2
0
0
20
Total
8
2
1
49
Total
22
22
60
-------
conditions prohibited access. Lakes that were visited
but were not sampled because they were frozen or
were sampled, but were found during data validation
(Section 3.3) to be the wrong lake, were also
classified as not visited. The target or non-target
status of this group of lakes could not be determined;
thus, they represent incompleteness in the sample.
For statistical analyses, it was assumed that
lakes not visited were a random subsample of the
original probability sample and thus had the same
proportion of non-target lakes as the lakes that were
visited. This assumption appears valid because no
systematic differences in lake area, watershed area,
or elevation were found between frozen lakes and
sampled lakes (Section 5.2.2).
However, some differences were noted in the Central
and Southern Rockies between the frozen lakes and
the sampled lakes (Table 2-3). Frozen lakes were
generally smaller and were located at higher
elevations than were the sampled lakes. It is also
probable that the frozen lakes were shallower than
the sampled lakes, although no measurements of
lake depth in the frozen lakes are available to confirm
this assumption. Analyses shown later in this report
(Section 6.4.2) indicate that ANC is inversely related
to elevation in the Central Rockies. That frozen lakes
in this subregion were generally located at elevations
higher than those estimated from the sampled lakes
for the target population suggests that the popu-
lation estimates for the Central Rockies are not based
on a random subsample. The degree to which the
population estimates underestimate the number of
low ANC lakes in the Central Rockies is difficult to
assess. However, comparison of WLS-I results with
previous surveys shows relatively close agreement
(Section 7.4).
The population estimates for the number of low ANC
lakes in the Southern Rockies are probably not
affected to the degree that may have occurred in
the Central Rockies. The differences in lake size
between frozen and sampled lakes is smaller and
there is no apparent relationship between lake ANC
and lake elevation in Subregion 4E (Section 6.4.2).
However, with such a large proportion of the selected
subsample unavailable for sampling in the Central
and Southern Rockies, it is not possible to determine
confidently the impact that frozen lakes had on the
characterizations of the lake populations.
2.2.3 Identification of Wilderness Lakes
All lakes in Forest Service wilderness areas and
some lakes in national parks could be sampled only
by ground crews because of legislation restricting
the use of motorized vehicles (Wilderness Act of
1964). To make plans for sampling access, all of the
selected lakes that were not eliminated as non-target
by map examination were evaluated to determine
if they were located within the boundaries of a Forest
Service wilderness area or a national park. These
evaluations were based on consultations with Forest
Service and National Park Service personnel and on
close examination of the most accurate maps
available. The wilderness lake classifications may
not be entirely correct, however, particularly for
newly created wilderness areas for which the
congressionally certified boundaries were not
available at the time wilderness lake identifications
were made. The names and locations of lakes
sampled in Forest Service wilderness areas and
national parks are given in Figures 2-2 and 2-3. The
number of lakes sampled in each area is given in
Table 2-4.
2.2.4 Special Interest Lakes
Other lakes, in addition to those chosen in the
probability sample, were included in the WLS-I.
Forty-two lakes that were not selected randomly and
are or were the subjects of relevant research
programs were selected as special interest lakes.
Of these lakes, samples were collected from 32. All
western lakes in the current EPA Long-Term
Monitoring Program were selected as special
interest lakes. Other special interest lakes were
included based on recommendations from state and
federal agencies. Data from these lakes were not
used in computing population estimates, but the
Table 2-3. Population Estimates of Median Lake Area and Median Elevation for Frozen Lakes and Sampled Lakes in Central
Rockies (4D) and Southern Rockies (4E), Western Lake Survey - Phase I
Stratum
Target Lakes Visited (n**»)
Lake Area (ha)
frozen
sampled
frozen
sampled
Elevation (m)
frozen
sampled
Central Rockies
4D1
4D2
4D3
Southern Rockies
4E1
4E2
4E3
11
7
3
14
5
1
43
47
39
46
52
41
3.5
2.4
6.1
2.6
2.6
1.7
5.6
5.5
3.5
3.0
3.3
3.5
3338
3042
2865
3394
3237
3188
3219
3042
2687
3307
3456
3147
-------
Figure 2-2. Locations of Forest Service wilderness areas ( S ) and national parks ( ® ) containing lakes sampled in the
California (4A) and Pacific Northwest (4B) subregions, in study areas of the Western Lake Survey - Phase I.
20 22 19
28
4A
I Wilderness Areas
| National Parks
—«^_*
Subregion Boundary
Pacific Northwest (4B)
19. Pasayten Wilderness
20. Mount Baker Wilderness
21. Noisy Diobsud Wilderness
22. North Cascades National Park
23. Lake Chelan-Sawtooth Wilderness
24. Boulder River Wilderness
25. Glacier Peak Wilderness
26. Henry M. Jackson Wilderness
27. Buckhorn Wilderness
28. Olympic National Park
29. Alpine Lakes Wilderness
30. William 0. Douglas Wilderness
31. Clearwater Wilderness
32. Mount Rainier National Park
33. Goat Rocks Wilderness
34. Indian Heaven Wilderness
35. Columbia Wilderness
36. Mount Hood Wilderness
37. Mount Jefferson Wilderness
38. Mount Washington Wilderness
39. Three Sisters Wilderness
40. Waldo Lake Wilderness
41. Diamond Peak Wilderness
42. Sky Lakes Wilderness
California (4AJ
1. Siskiyou Wilderness
2. Marble Mountain Wilderness
3. Trinity Alps Wilderness
4. Thousand Lakes Wilderness
5. Lassen Volcanic National Park
6. Caribou Wilderness
7. Bucks Lake Wilderness
8, Desolation Wilderness
9. Mokelumne Wilderness
10. Hoover Wilderness
11. Emigrant Wilderness
12. Yosemite National Park
13. Minarets Wilderness
14. John Muir Wilderness
15. Kaiser Wilderness
16. Dinkey Lakes Wilderness
17. Kings Canyon National Park
18. Sequoia National Park
-------
Figure 2-3. Locations of Forest Service wilderness areas ( Q ) and national parks ( @ ) containing lakes sampled in the
Rocky Mountain subregions: the Northern Rockies (4CJ. the Central Rockies (4D), and the Southern Rockies
(4E). in study areas of the Western Lake Survey - Phase I. Lakes located in roadless areas are also shown (\3).
Northern Rockies I4C)
43.
44.
45.
46.
47.
48.
Glacier National Park
Great Bear Wilderness
Cabinet Mountains Wilderness
Mission Mountains Wilderness
Rattlesnake Wilderness
Anaconda-Pintlar Wilderness
49. Selway-Bitterroot Wilderness
50. Gospel Hump Wilderness
51. River of No Return Wilderness
52. Hell's Canyon Wilderness
53. Eagle Cap Wilderness
54. Sawtooth Wilderness
I
Central Rockies (4D)
55. Lee Metcalf Wilderness
56. Absaroka-Beartooth Wilderness
57. Cloud Peak Wilderness
Yellowstone National Park
Teton Wilderness
Grand Teton National Park
61. Jedidiah Smith Wilderness
62. Gros Ventre Wilderness
63. Fitzpatrick Wilderness
64. Wind River Roadless Area
65. Popo Agie Wilderness
66. Bridger Wilderness
67. High Uintas Wilderness
68. Lone Peak Wilderness
58.
59.
60.
\
\
\
\
Southern Rockies (4E)
69. Mount Zirkel Wilderness
70. Comanche Peak Wilderness
71. Never Summer Wilderness
72. Rocky Mountain National Park
73. Indian Peaks Wilderness !""
74. Flat Tops Wilderness \
75. Eagles Nest Wilderness '
76. Mount Evans Wilderness I
77. Holy Cross Wilderness \
78. Hunter Frying Pan Wilderness
79. Collegiate Peaks Wilderness
80. Maroon Bells-Snowmass Wilderness
81. West Elk Wilderness
82. Weminuche Wilderness
83. South San Juan Wilderness
84. Wheeler Peak Wilderness
Wilderness Areas
National Parks
Roadless Area
Subregion Boundaries
10
-------
Table 2-4. Forest Service Wilderness Areas and National Parks Containing Probability Sample Lakes, Grouped by
Subpopulations Used in Data Analysis, Western Lake Survey - Phase 1
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
1 ^
I O.
14.
15.
16.
17.
18.
Total
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
Total
Subpopulation
California (4A)
Klamath Mountains (CA)
Siskiyou Wilderness
Marble Mountain Wilderness
Trinity Alps Wilderness
Southern Cascades (CA)
Thousand Lakes Wilderness
Lassen Volcanic National Park
Caribou Wilderness
Bucks Lake Wilderness
Tahoe Area (CA)
Desolation Wilderness
Mokelumne Wilderness
Yosemite Area (CA)
Hoover Wilderness
Emigrant Wilderness
Yosemite National Park
IV/T nm+ lA/iM norte?
Minarets vvnuerness
Southern Sierra Nevada (CA)
John Muir Wilderness
Kaiser Wilderness
Dinkey Lakes Wilderness
Kings Canyon National Park
Sequoia National Park
(4A)
Pacific Northwest (4B)
North Cascades Area (WA)
Pasayten Wilderness
Mount Baker Wilderness
Noisy Diobsud Wilderness
North Cascades National Park
Lake Chelan-Sawtooth Wilderness
Boulder River Wilderness
Glacier Peak Wilderness
Henry M. Jackson Wilderness
Olympic Mountains (WA)
Buckhorn Wilderness
Olympic National Park
Alpine Lakes (WA)
Alpine Lakes Wilderness
Washington Cascades (WA)
William O. Douglas Wilderness
Clearwater Wilderness
Mount Rainier National Park
Goat Rocks Wilderness
Indian Heaven Wilderness
Oregon Cascades (OR)
Columbia Wilderness
Mount Hood Wilderness
Mount Jefferson Wilderness
Mount Washington Wilderness
Three Sisters Wilderness
Waldo Lake Wilderness
Diamond Peak Wilderness
Sky Lakes Wilderness
(4B)
Lakes
Sampled
13
2
4
7
13
2
7
3
1
14
11
3
19
1
4
9
5
38
16
1
1
11
9
97
28
7
1
1
6
2
1
8
2
4
1
3
24
24
13
5
1
5
1
1
21
1
1
2
1
8
2
1
5
90
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
Total
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
Total
69.
70.
71.
72.
73.
74.
75.
Subpopulation
Northern Rockies (4C)
Northwest Montana (MT)
Glacier National Park
Great Bear Wilderness
Cabinet Mountains Wilderness
Mission Mountains Wilderness
Rattlesnake Wilderness
Bitterroot Area (MT/ID)
Anaconda-Pintlar Wilderness
Selway-Bitterroot Wilderness
Gospel Hump Wilderness
River of No Return Wilderness
Blue Mountains (ID/OR)
Hell's Canyon Wilderness
Eagle Cap Wilderness
Sawtooth Mountains (ID)
Sawtooth Wilderness
(4C)
Centra/ Rockies (4D)
Southern Montana (MT)
Lee Metcalf Wilderness
Absaroka-Beartooth Wilderness
Cloud Peak (WY)
Cloud Peak Wilderness
Northwest Wyoming (WY)
Yellowstone National Park
Teton Wilderness
Grand Teton National Park
Jedidiah Smith Wilderness
Gros Ventre Wilderness
'Wind River Mountains (WY)
Fitzpatrick Wilderness
Wind River Roadless Area3
Popo Agie Wilderness
Bridger Wilderness
Wasatch/Uinta Mountains (UT)
High Uintas Wilderness
Lone Peak Wilderness
(4D)
Southern Rockies (4E)
Mount Zirkel (CO)
Mount Zirkel Wilderness
Rocky Mountain Park Area (CO)
Comanche Peak Wilderness
Never Summer Wilderness
Rocky Mountain National Park
Indian Peaks Wilderness
Flat Tops (CO)
Flat Tops Wilderness
Central Colorado (CO)
Eagles Nest Wilderness
Lakes
Sampled
11
5
1
1
2
2
44
2
31
1
10
10
2
8
17
17
82
16
1
15
9
9
11
6
1
2
1
1
44
5
6
7
26
16
15
1
96
17
17
36
1
1
22
12
8
8
1 1
\lt
2
-------
Table 2-4. (continued)
Subpopulation
Lakes
Sampled
76. Mount Evans Wilderness
77. Holy Cross Wilderness
78. Hunter Frying Pan Wilderness
79. Collegiate Peaks Wilderness
80. Maroon Bells-Snowmass
Wilderness
81. West Elk Wilderness
S. Colorado/New Mexico
(CO.NM)
82. Weminuche Wilderness
83. South San Juan Wilderness
84. Wheeler Peak Wilderness
Total (4E)
2
2
2
1
2
1
17
14
2
1
90
Total in the West
442
'Located on an Indian Reservation and technically not a
wilderness area, but included here to complete the Wind River
Range (Figure 2-3).
results are summarized in Table 5-23 and individual
lake results are shown in Volume II (Eilers et al.
1987b).
2.2.5 Final Lake Lists and Maps
Lake names, identification (ID) numbers, geographi-
cal coordinates, wilderness area or park names, and
map names were entered into computer files and
were printed for field crews. If no name was printed
on the map for a given lake, the entry in the file
was "no name." Each lake was assigned a unique
ID number coded for the stratum in which it occurred
(e.g., 4A2-011 is the 11th lake selected in alkalinity
map class 2 of Subregion A in Region 4). The latitude
and longitude for each lake were measured with 11 -
point dividers to the nearest degree, minute, and
second and were checked by visual examination and
computer generated overlays. Additionally, ID codes
were printed on the topographic maps for use by
field crews in locating the lakes.
2.3 Design Applications and
Restrictions
2.3.1 Extrapolation from Sample to Population
For statistical analyses, the probability sample was
treated as a simple random sample within each
stratum. The ordering of the lakes and the systematic
selection process were designed to increase the
precision over that which would have been obtained
with a simple random sample; therefore, statements
of precision for population estimates are probably
conservative.
When population estimates from combined strata are
required (e.g., when making subregional or regional
estimates), expansion factors or weights (W) must
be used because the sampling intensity varied
among strata. These weights (the stratum target
population size divided by the stratum sample size)
vary considerably among strata; for example, lakes
sampled in one stratum may each represent three
lakes, whereas each lake sampled in another stratum
may represent 36 lakes.
This design permits strata to be combined when it
is meaningful to do so. The flexibility of the design
also allows estimates to be made for specific
subpopulations, or subsets of lakes; however, unless
the definition of the population of.interest is clearly
stated, conclusions based on the WLS-I data can
be misleading.
2.3.2 Estimating the Target Population Size and
Attributes
The first steps in the statistical analysis of the data
were to estimate the target population size in each
stratum ( N; Figure 2-4) and to determine the stratum-
specific weights needed for interstratum estimation.
The target population size in each stratum was
estimated in two units: number of lakes and lake
area (in hectares).
The statistical frame is defined by the list of lakes
identified by the map population. The map population
consists of both target and non-target populations.
Within each stratum, it is possible to estimate the
size of the target population by multiplying the
number of sampled lakes classified as target (n***)
by the stratum-specific weight (W). This weight is
the inverse of the inclusion probability (P) of a target
lake in the final sample, which is determined
according to the following equation:
P= 1/W = (n*/N*)q (1)
where n* = the size of the sample drawn from
the map population
N* = the size of the map population
q = the probability that a target lake in
the drawn sample is actually visited;
computed by dividing the actual
number of lakes visited (n* - nnb - n0)
by the number of lakes intended to
be visited (n* - nnb)
where nnb = the number of non-
target lakes in the origi-
nal sample, as deter-
mined from the maps,
and
n0 = the number of lakes not
visited.
12
-------
Figure 2-4. Procedures used to estimate the target population size. Western Lake Survey - Phase I.
Map Population
IN")
Large Scale Map Examination
Selected Lakes
(n*)
Estimated
Non-Target
Lakes in
Not Visited
Lakes
I
Lakes Scheduled for
Visitation (n* -n^)
Estimated
Non-Target
Lakes
Selected
I
Direct
Examination
I
Lakes Visited
(n* -ftnb- na)
N*
Multiply by —
/?*
1
Target Lakes
Visited
-------
A =W(IA)
The variances of N and A for single strata, were
estimated by:
V(N) = N* [(N* - n')/(n'-1)] [n** Vn'] [(n' - n***)/n'] (4)
(3) where summation is over the appropriate subset of
sample lakes in the appropriate strata, and where
the values of Ware assigned according to the stratum
in which the lake belongs.
V(A) = N*[(N* - n')/(n'-1)] [1 /n'] [IAZ - (ZA)Vn'] (5)
where n', the "effective sample size," is used in place
of n* because of incomplete visitation (i.e., n' = qn*).
The standard errors are calculated as the square
roots of the variances.
For estimates of populations covering multiple strata,
estimates and variances must be computed within
strata and added or else computed with equations
containing weights (see below). Any explicitly
defined subset of the total population of target lakes
in the West is a subpopulation. Subpopulations can
be defined over any combination of strata. For
example, for any given variable, each observed value
of a variable, X, defines a subpopulation of lakes
having a value x less than or equal to that value.
Subpopulation definitions also can be based on
geographic boundaries such as states or national
parks. This procedure was used in identifying specific
subpopulations in geomorphic units (Section 7).
Estimates for subpopulations that are defined within
single strata can be generated using formulae that
are modifications of the equations given above
(mathematically identical to the algorithms used in
generating all the statistics for the survey). To
generate single stratum subpopulation equations
from equations 2 through 5, each n*** is replaced
by nz and each I by Iz, where nz is the number
of sample lakes in the subpopulation z and Zz is the
summation over the sample lakes in the subpop-
ulation z.
For example, all 22 lakes (nz) sampled in Rocky
Mountain National Park are in Stratum 4E1, which
has W = 3.261 (Section 5.2.3, Table 5;5). Thus, the
estimated target population size (N) for Rocky
Mountain National Park is (3.261)(22) = 72 lakes.
The 22 lakes have a combined area of 106.8 ha (IZA);
thus, the estimated area (A) of the target population
in Rocky Mountain National Park is (3.261)(106.8
ha) = 348.3 ha.
A useful generalization, appropriate for any subpop-
ulation and any combination of strata, is that.
The use of equation 6 can be illustrated with the
data for population estimates of acid neutralizing
capacity (ANC) by state (Section 5.4.1.3, Table 5-10).
In Utah, 30 lakes were sampled. Eleven of these
were in Stratum 4D1 with W= 18.356, 13 in Stratum
4D2 with W = 19.744, and 6 in Stratum 4D3 with
W = 14.918 (Section 5.2.3, Table 5-5). The estimated
target population size (N) for Utah is calculated by
adding the product of nz and W for each stratum,
which is the same as 11(18.356) + 13(19.744) +
6(14.918) = 548 lakes. A subpopulation of the lakes
in Utah can be defined by a particular value for ANC.
Only one of the lakes sampled in Utah had an ANC
value < 50 /ueq L~1. This lake was in Stratum 4D2
with a weight of 19.744; thus, the estimated number
of lakes in Utah with ANC < 50 fjeq L"1 is 20, and
the estimated proportion is 20/548 = 0.04.
Afurther generalization, used in data analysis, leads
to a similar formula for the estimated variance of
any variable, X, over any subpopulation and
combination of strata:
Variance(X) = IWX2/IW - (IWX/IW)2 (7)
N = IW, and A= IWA,
14
where the set of sample lakes in the summation
defines the subpopulation of lakes for which the
variance is estimated.
The weighting factors are extremely important.
Estimating population parameters from sample data
without accounting for weights can lead to erroneous
calculations and incorrect interpretation. Examining
relationships among variables with the expectation
that these relationships are representative of the
population should only be done within strata or by
using weighting factors.
By a method equivalent to calculating subpopulation
estimates (number or area of lakes with concentra-
tion less than or equal to x) and their associated
upper confidence limits for all possible values of X,
cumulative frequency distributions [F(x>] and
cumulative areal distributions [G(x>] were calculated.
At any value x, these curves represent the estimated
number or area of lakes in the population having
a value for that variable less than or equal to x, with
the 95 percent upper confidence limit for that
number (Section 5.1). For some variables, interest
is in the number of lakes with concentrations above
a particular value (e.g., sulfate > 50 (jteq L'\ so the
(6) inverses of the cumulative frequency distributions
-------
[1-F(x)J and of the cumulative areal distributions
[1 -G(x)] were generated in a similar manner.
Quintiles and medians for these cumulative fre-
quency and areal distributions were also calculated.
The quintiles (Qi through Q4) reflect the estimated
values of x separating the distribution into five equal
parts (i.e., the 4th quintile is the 80th percentile).
The median is the estimated value of x such that
half the lakes in the population are characterized
by concentrations of the variable equal to or less
than the value of x.
2.3.3 Restrictions
The use and interpretation of any data set are
restricted by the design, the quality of the data
obtained, and the sampling protocols. The map and
target populations and the period of sampling are
the primary design considerations influencing the
proper interpretation of the WLS-I data.
Estimates of the number of lakes within an area are
strongly affected by the map scale used to define
the map population. Use of larger scale maps
provides greater resolution and allows smaller lakes
to be identified and included in the population under
consideration. The map scale used to define the map
population in the WLS-I was 1:100,000, which
identified lakes as small as approximately 1 ha. Lakes
identified from 1:24,000-scale maps can be as small
as 0.1 ha. Some estimates of the number of lakes
in portions of the West have been based on larger
scale maps such as 1:62,500 or 1:25,000 (Turk and
Adams 1983).
To illustrate the influence of map scale in estimating
the total number of lakes, the number of lakes on
1:100,000-scale maps within several areas of
interest were counted. The process was then
repeated using larger scale maps covering the same
area as the 1:100,000-scale maps. The results (Table
2-5) show that large-scale maps display more lakes.
Comparisons of the results of the WLS-I to those
from other lake surveys in the West must be done
with knowledge of the map scales used to prepare
the population estimates. No direct conclusions can
be drawn about the population of lakes less than
1 ha from WLS-I data. This restriction also applies
to other categories of non-target lakes defined in
the WLS-I.
The period of sampling restricts the conclusions of
the WLS-I to the fall of 1985. The accuracy of
extrapolating the fall index sample (Section 2.1.2)
to other times of the year or to other years is not
known.
2.4 Lake Characterization
Because of their effects on surface water chemistry,
several physical lake and watershed parameters are
important to consider when interpreting chemical
data from lakes. These parameters were measured
or determined from maps as described in the
following sections.
2.4.1 Lake Area
Lake area (in hectares) was measured in triplicate
with an electronic planimeter on 1:24,000- or
1:62,500-scale topographic maps. Large lakes or
reservoirs which occurred on more than one large-
scale map were measured on 1:100,000-scale maps.
Islands and wetlands within the lake were included
in the measurement of watershed area (Section
2.4.4) rather than lake area.
The following criteria were used in delineating
boundaries when a lake was connected to other
bodies of water, was divided between adjoining
maps, or did not appear the same at different scales
of resolution:
1. the boundary of a lake was defined to be at
the narrowest point of its inlet or outlet;
2. if part of a lake was not shown on an adjoining
map, the perimeter was estimated using
topographic contour lines; and
3. when lakes appeared as two distinct water
bodies on a 1:100,000-scale map, but as one
on larger scale maps, the entire water body
was measured as if it were one lake.
2.4.2 Elevation
The elevation of each lake was recorded in feet from
the largest scale map available and was subse-
quently converted to meters. When no lake elevation
was marked on the map, it was estimated by
interpolation of topographic contour lines.
2.4.3 Hydrologic Lake Type
Each lake was classified by hydrologic type (Wetzel
1983) through visual examination of its morphome-
try on the largest scale topographic maps available.
"Seepage" lakes were defined as those having no
inlet or outlet. "Closed" lakes were those with inlets
but no outlets. "Drainage" lakes were those with
outlets, regardless of the presence or absence of
inlets. "Reservoirs" were artificial lakes, as indicated
by a dam at an outlet. The reservoir class could
include dammed river valleys and natural lakes with
artificial outlet control structures.
15
-------
Table 2-5. Number of Lakes Shown on Small-scale (1:100,000) and Large-scale (1:62,500; 1:24,000) Maps.
Western Lake Survey - Phase I
Subregion
Northern Rockies
Central Rockies
Central Rockies
(4C)
(4D)
(4D)
Mountain
Range
Beartooth
Wind River
Uinta
1:100,000
Map Name
Alpine Lakes
Mt. Bonneville
Fremont Lake, N
Kings Peak
Number of Lakes
1:100,000
125
50
26
41
1:62,500
350
—
-
-
1:24,000
-
200
160
230
Ratio of
Large Scale to
Small Scale
2.8
4.0
6.1
5.6
2.4.4 Watershed Area
An electronic planimeter was used to measure
watershed area. Watershed areas greater than 10
to 15 mi2 (approximately 26 to 39 km2) were
measured on 1 -.100,000-scale topographic maps, if
available; otherwise, 1:250,000-scale topographic
maps were used. Smaller watersheds were mea-
sured on larger scale maps.
2.4.5 Land Use/Land Cover
The percent of watershed area in each of 11 land
use categories was determined from land use maps.
The following categories (Anderson et al. 1 976) were
used: urban or built-up land, agricultural land,
rangeland, deciduous forest land, evergreen forest
land, mixed forest land, surface water, wetland,
barren land, tundra, and perennial snow or ice. These
categories are represented on 1:250,000-scale
USGS Land Use and Land Cover overlays, which
were available for all areas except parts of the Rocky
Mountains (most of the Northern Rockies and four
lakes in the Central Rockies). In these areas, land
use information for publicly owned land was
obtained from the Forest Service, National Park
Service, Soil Conservation Service, and state
agencies. Watersheds on private land were char-
acterized by interpreting 1:250,000-scale USGS
topographic maps and potential natural vegetation
maps (U.S. Soil Conservation Service 1976; U.S.
Geological Survey 1970).
Watersheds were delineated on 1:250,000-scale
USGS topographic maps and then were traced onto
the land use maps. For most watersheds, an
equidistant dot grid appropriate to the size of the
watershed was placed over the land use map. The
sample lake was shaded so its area could be excluded
from the land use estimate. All the dots in each land
use class were counted; then the template was
moved and the dots were recounted. The total
number of dots for each class was then divided by
the total number of dots in the watershed to obtain
the percentage of each class in the watershed. If
the watershed contained few categories, the
percentage of each was estimated visually.
2.4.5 Slope
Slope was calculated as the median of percent slope
(vertical distance divided by horizontal distance times
100) for approximately ten equidistant points within
each lake's watershed. Slope measurements for
watersheds greater than about 10-15 mi2 (26-39
km2) were taken from 1:100,000- or 1:250,000-scale
USGS topographic maps; measurements for smaller
watersheds were taken from larger scale maps.
An equidistant dot grid appropriate to the map scale
was placed over the watershed. Percent slope
readings were taken from a slope calculator placed
perpendicularly to the contour lines at the points.
For watersheds divided between two or more maps,
the number of calculations per map corresponded
to the proportion of the watershed on that map.
2.5 Sampling and Analytical Methods
Sampling for the WLS-I was conducted from 10
September to 4 November 1985. Except where
permission to land a helicopter was denied, lakes
were accessed by helicopters with fixed floats and
were sampled by EPA and Lockheed Engineering and
Management Services Company, Inc. (Lockheed-
EMSCO) personnel. Most lakes in Forest Service
Wilderness Areas, some lakes in National Parks, and
one other lake were accessed by ground and were
sampled from inflatable boats by Forest Service and
Lockheed-EMSCO personnel. Water samples were
delivered for initial analyses, processing, and
preservation to mobile field laboratories located at
five field stations: Missoula and Bozeman, Montana;
Aspen, Colorado; Carson City, Nevada; and Wenat-
chee, Washington. Where necessary, remote base
sites were established to collect samples from lakes
located at distances greater than approximately 150
miles (250 km) from a field station. Regardless of
whether the helicopter crews were operating from
the field station or from a remote base site, helicopter
crew samples were delivered for field laboratory
analyses within 16 hours of collection. Sixty-two
percent of the samples collected in the Survey by
ground crews were also delivered to the field
laboratory on the same day they were sampled and
-------
were processed within 16 hours. Ninety percent of
all the ground crew samples were delivered to the
field laboratory no later than on the day following
sample collection. After processing, samples were
shipped by overnight courier to analytical laborato-
ries for completion of analysis.
2.5.1 Field Sampling Activities
Field sampling activities conducted by ground crews
and helicopter crews included site description and
location, as well as collection of in situ lake data
and water samples (Figure 2-5). Detailed field
sampling protocols are provided in Bonoff and
Groeger (1987). A comparison of data collected using
both protocols is provided in Section 4.5.
2.5.1.1 Site Description and Location—Because of
its capability for prolonged high-altitude flying, the
Aerospeciale Lama 315B helicopter was used to
sample the WLS-I lakes at all but one base site. In
Wenatchee, Washington, a Bell 206 Long Ranger
was used as in the ELS-I (Morris et al. 1986). Lakes
to be sampled from helicopters were identified from
the air using a Loran-C guidance system and
topographic maps. Aerial photographs of the lakes
were taken and watershed disturbances were
recorded on a standardized form. If a lake was
accessible from the air, the helicopter was landed
at what appeared to be the deepest part of the lake.
The deepest point near the landing area was located
with an electronic depth finder.
Ground sampling personnel used topographic maps
and their familiarity with the local area to locate lakes
(Peck et al. 1985). Site locations were verified using
a number of methods, including comparing observed
lake morphometry to that shown on the map,
identifying topographic features, determining
position by compass triangulation, and enlisting
assistance from local experts. The deepest point of
the lake was estimated using shoreline topography
and soundings taken with the calibrated anchor line.
2.5.1.2 In Situ Measurements and Sample Col-
lection—Standardized field data forms were used by
helicopter crews and by ground crews to record the
values of in situ measurements. Helicopter sampling
crews used the fixed floats as platforms from which
to take in situ measurements and to collect water
samples. Field measurements taken from the
helicopter were site depth, Secchi disk transparency,
temperature, pH, and conductance. Site depth was
determined with a weighted sounding line. Water
transparency was measured with a 20-cm-diameter,
black-and-white Secchi disk. Helicopter crews used
Hydrolab 4041 units retrofitted with glass pH
electrodes and Beckman Lazarin reference elec-
trodes to determine in situ values of temperature,
pH, and conductance. The Hydrolab units were
calibrated daily with a solution of 0.001 N KCI
(specific conductance = 147 /jS cm 1 at 25°C) and
National Bureau of Standards (NBS) traceable pH
buffer solutions (pH = 7.00 and 4.01). Each day before
and after sampling, temperature, pH, and conduc-
tance functions were also checked to ensure proper
instrument operation. Temperature was checked
against an NBS-traceable thermometer. Conduct-
ance and pH functions were checked using a COz-
saturated solution (pH = 3.91, specific conductance
= 50/yScm"1 at STP).
Hydrolab measurements were taken at the depth
from which the sample was collected, which was
usually 1.5 m. A second set of readings was taken
at 1.5 m above the lake bottom. If the temperature
difference between the two measurements taken at
the top and bottom was <4°C, the lake was classified
as nonstratified. If the temperature difference
between the two depths was > 4°C, a third set of
readings was taken at 60 percent of the site depth.
If the temperature difference between 1.5 m and
60 percent of the site depth was < 4°C, the lake
was classified as weakly stratified. If the difference
was > 4°C, the lake was classified as strongly
stratified, and vertical profiles of temperature and
conductance were determined.
Ground crews collected water samples and took in
situ measurements from inflatable boats. Because
of logistical constraints, ground crews did not use
Hydrolab units; therefore the protocol for taking in
situ measurements differed from that used by the
helicopter crews. A Yellow Springs Instruments (YSI)
model 425C telethermometer with a YSI 400 series
probe was used to determine water temperature and
to check the stratification status as described above.
The telethermometer was checked daily against an
NBS-traceable thermometer, and was recalibrated
if the temperature difference was more than ±0.5°C.
Site depth and Secchi disk transparency were
determined as described for the helicopter sampling
procedures.
Water samples were collected by helicopter and
ground crews with a 6.2-L Van Dorn acrylic plastic
sample bottle. Samples were collected from 1.5 m
if the site was deep enough to allow a clean sample
to be taken (i.e., a sample free from sediment, plants,
and other large particles associated with the lake
bottom). If a clean sample could not be obtained from
1.5 m, the sample was collected from 0.75 m. If
a clean sample could not be collected from 0.75 m,
the lake was designated as a shallow lake and was
classified as non-target.
The Van Dorn sampler was modified to accommodate
a nylon Luer-Lock fitting to permit collection of
sample aliquots in syringes without atmospheric
contact. Two 60-mL polyethylene syringes were
77
-------
Figure 2-5. Field sampling activities. Western Lake Survey - Phase I. Field crews accessed most sampling sites in wilderness
areas by ground and in non-wilderness areas by helicopter.
Hike/Pack
to Lake
Fly to
Lake
Verify Identity
Record Site Characteristics
Photograph Lake
Select Sampling Site
Flow Out and Anchor
Land on Lake
Determine
Stratification
Status
(Temperature Profile)
Other In Situ
Measurements
Secchi Disk Transparency
pH (Paper)
Record Data
in Logbook
pH (Probe)
Conductance
\ Return to Shore
Collect Water Samples
Field
Blank
Sample
(Deionized
water)
4-L
Container
125-mL
Bottle
Routine
Sample
(from 1.5 m)
4-L
Container
Two 60-mL
Syringes
125-mL
Bottle
Field
Duplicate
Sample
(from 1.5 mj
4-L
Container
Two 60-mL
Syringes
125-mL
Bottle
Preserve 125-mL
Aliquots with HgC/2
Complete Field
Data Form
Hike/Pack to
Trailhead
Fly to
Next Lake
Deliver Samples to
Field Laboratory
18
-------
attached, in turn, to the fitting, were filled with water,
and were sealed with locking syringe valves. These
syringes were used for pH and dissolved inorganic
carbon (DIC) analyses at the field laboratory. A bulk
water sample was collected by completely filling a
4-L polyethylene Cubitainer from the Van Dorn
sampler. The syringes and Cubitainers were placed
in coolers with frozen chemical refrigerant packs and
were delivered to the field laboratory. Although
helicopter crews and ground crews both followed
this protocol for sample collection, ground crews also
collected a 125-mLaliquot in an opaque polyethylene
bottle. This aliquot, which was preserved with 0.1
mL of 5 percent HgCl2 at the lake site, was used
as an independent check of the chemical stability
of nitrate and sulfate.
Two types of QA samples (Section 3.1.5.1), field
blanks and field duplicates, were collected by the
sampling crews. Field blank samples were "col-
lected" by filling the Van Dorn sampler with
deionized water (obtained from the field laboratory)
which was then used to fill a 4-L Cubitainer. Because
deionized water is especially prone to CO2 contam-
ination as a result of atmospheric contact, syringe
samples for pH and DIC analyses were not collected
from field blanks. Field duplicate samples, consisting
of two syringes and one 4-L Cubitainer, were
collected by obtaining a second Van Dorn sample
from a lake, at the same location and depth as the
routine sample.
2.5.2 Field Laboratory Operations
Five field laboratory trailers, one at each field station,
provided facilities for sample receipt, analysis,
aliquot preparation, preservation, and shipping. The
field laboratory also facilitated distribution of water
to the field crews for blank samples and of
conductivity standards for Hydrolab calibration. Field
laboratory personnel included a laboratory coordi-
nator, a laboratory supervisor, and three analysts.
Laboratory activities (Figure 2-6) were conducted in
accordance with established protocols for laboratory
safety and maintenance and included preparation
of reagents, blanks, and samples; instrument
calibration; data recording; and supply inventory.
Morris et al. (1986) give a detailed discussion of field
laboratory protocols.
The field laboratory coordinator, who had overall
responsibility for administration of the field labor-
atory, provided the helicopter crews with water for
blank samples and with conductivity standards.
Supplies for the ground crews were provided through
the field laboratory coordinator by the logistics
coordinator. The laboratory coordinator received lake
samples, lake data forms, and chain-of-custody
forms (from the ground crews through the Forest
Service field manager) and received field audit
samples from a designated supplier (Radian Corpo-
ration, Austin, Texas). The laboratory coordinator
organized the lake samples, blanks, and audit
samples into "batches," i.e., all samples processed
at one field laboratory on a single day. Samples were
then assigned batch and sample ID numbers to
facilitate sample tracking. After the sample batches
were organized, the laboratory coordinator refriger-
ated the samples at 4°C until they were removed
for analysis or processing. The laboratory coordinator
also was responsible for maintaining and procuring
all supplies.
The laboratory supervisor, who was responsible for
overseeing daily laboratory operations, performed
DIC and pH determinations. These two measure-
ments were conducted on the syringe samples, and
are termed "closed system" measurements because
the samples did not come into contact with the
atmosphere during collection or analysis. The
supervisor also designated one sample in each batch
as the trailer duplicate (Section 3.1.5.2). This sample
was analyzed in duplicate for DIC, pH, turbidity, and
true color.
Three laboratory analysts shared sample processing
and sample analysis duties. One analyst performed
aluminum extractions, the second performed sample
filtration, and the third prepared and preserved
aliquots and split samples (Table 2-6). Split samples
were prepared from routine lake samples both in
the field laboratory and by field sampling crews for
special studies conducted at the EPA Environmental
Research Laboratory in Corvallis, Oregon (ERL-
Corvallis) and the EPA Environmental Monitoring
Systems Laboratory in Las Vegas, Nevada (EMSL-
Las Vegas). The third analyst also performed turbidity
and true color determinations. All laboratory
personnel were responsiblefor maintaining logbooks
of their respective tasks; these records were checked
daily by the laboratory supervisor.
After a set of aliquots was processed, each container
was sealed with electrical tape and was placed in
an individual plastic bag which was then sealed.
Next, each set of aliquot containers was placed in
a one-gallon Ziploc bag, which was also sealed.
Containers were refrigerated at 4°C until shipment.
For transport to the analytical laboratories, each set
of aliquots and the corresponding shipping forms
were placed in an insulated shipping box that was
maintained at approximately 4°C. The boxes were
shipped to the analytical laboratories via overnight
courier service.
All data and shipping forms that were completed
at the lake site and at the field laboratory were
reviewed by the laboratory coordinator. Copies of all
forms were then sent to EMSL-Las Vegas, where
19
-------
Figure 2-6. Field laboratory activities. Western Lake Survey - Phase I. Samples collected by field crews were processed
at mobile field laboratories before shipment to analytical laboratories for analysis.
Sample Transfer Supply Requests from
*
Laboratory
1
DIC
Determination
,
'
Laboratory
Coordinator
,
1
'
Check in Audit Samples
and Organize Batch
Ana
3
,
i
lyst Analyst Analyst Logit
i
tics
nator
,111
Aliquot
Preparation
and
Preservation
pH
Determination
L
Turbidity
etermination
True Color
Determination
Data
Transfer
Sample Aluminum Track In
Filtration Extractions for Cr
Samolin
Shipping
Preparation
|
ventory
ound
g Crews
Field Supplies
Prepare
Reagents for
Field Crews
*
Pack Aliquots
Shipping Form ___
* uata horrns
and Supplies
Ship Aliquots
and Data Forms
Transfer Supplies
to Sampling Crews
20
-------
they were reviewed for data consistency (Section
3.2.1). Copies of data forms were also sent to the
data base manager at Oak Ridge National Laboratory
(ORNL) in Oak Ridge, Tennessee, for use in data entry
(Section 2.7).
2.5.3 Analytical Laboratory Operations
Analytical laboratory personnel were responsible for
receiving the samples shipped to them by the field
laboratory, inspecting the samples for damage,
logging in the sample batches, analyzing the samples
according to procedures described in the statement
of work and the WLS-I analytical methods manual
(Kerfoot and Faber 1987), and preparing and
distributing data packages on the analyses per-
formed. Analytical methods are described in
Volume II.
Each sample received by the analytical laboratory
consisted of the seven aliquots prepared at the field
laboratory. Each aliquot was processed in a different
manner depending on the analytes to be analyzed
(Table 2-6). In addition, each analyte had to be
measured within a specified holding time (Table 2-7).
For each batch of samples, analytical laboratory
personnel completed a data package. The data
package consisted of analytical results and support-
ing information from the field and analytical
laboratories (Silverstein et al. 1987a). On the basis
of the analytical results reported for internal and
external QA/QC samples (Section 3.1.5) the QA
staff, with the approval of the QA manager, could
direct the analytical laboratory to confirm reported
values or to reanalyze selected samples or sample
batches. In such cases, sample reanalysis was
usually requested for a given parameter on a per-
batch basis.
2.6 Calibration Study
The standardized protocol for the National Lake
Survey prescribed sampling by helicopter; however,
more than half the lakes selected for sampling in
the WLS-I (Section 2.2.3) were located in Forest
Service wilderness areas, where access by moto-
rized vehicles is prohibited (Wilderness Act of 1 964).
Access to these sampling sites by ground was
expected to be more difficult than was access by
helicopter. Access to the sampling site, identification
of the lake, and location of the correct site on the
lake for sampling are more readily accomplished by
helicopter. Additionally, adherence to established
sample handling protocols, particularly delivering
samples to the field laboratory within prescribed
holding times, was expected to be more difficult for
ground crews than for helicopter crews.
To evaluate the differences between the two
sampling methods, the Forest Service granted
permission to perform a "calibration" study on a
subset of the lakes located in wilderness areas. Each
calibration lake was sampled by a helicopter crew
and a ground crew. Results of analyses of samples
collected using these methods were compared to
determine whether or not calibration of the chemical
variables was necessary. Specifically, the data were
compared to evaluate the practical and statistical
significance of potential differences between the
sampling protocols and the effect of holding samples
for different lengths of time before processing,
preservation, and analysis. Additionally, samples
collected as part of the calibration study were used
to evaluate potential bias between the two analytical
laboratories. The results of these analyses are given
in Section 4.3 and in the Appendix.
The 50 calibration study lakes were selected as a
systematic random sample from a list of the Forest
Service wilderness area lakes ordered by lake ID
(Section 2.2.5). This procedure permitted the
calibration study lakes to be distributed proportion-
ately among subregions and wilderness areas. Five
lakes originally determined notto be in Forest Service
wilderness areas (and thus not included in the pool
from which the calibration study sample was drawn)
were later found to be in Forest Service wilderness
areas and were sampled by ground crews. In
addition, six lakes in national parks and another lake,
to which access by helicopter was denied, were
sampled by ground access; these lakes were not
included in the original pool from which the
calibration study sample was drawn.
Both crews (helicopter and ground) collected
samples from approximately the same location on
the lake. When logistically possible, the ground crew
sampled the lake first, and the helicopter crew
sampled the lake as soon as possible thereafter. Of
the 45 lakes sampled in the calibration study, 30
lakes were sampled by both crews on the same day.
Twenty-three of the 30 lakes were sampled by the
ground crews first and seven of the 30 lakes were
sampled by the helicopter crews first. The period
of time between sampling for these lakes never
exceeded five hours. The remaining 15 lakes were
sampled on different days, in most cases as a result
of weather-related delays. The ground crew collected
a routine and a duplicate sample; the helicopter crew
collected a routine, a duplicate, and a triplicate
sample.
To ensure that each ground crew's sampling
procedure would not be biased by knowledge of the
calibration lakes, the ground crews were not told
which lakes were part of the study. Because the
helicopter sample collection procedure had been a
27
-------
Table 2-6. Aliquots and Corresponding Measured Parameters, Western Lake Survey - Phase I
Aliquot'
Container
Volume (mL)
Preservative and
Description
Variable
Field Laboratory Aliquots
Syringe A
Syringe B
Cubitainer
Cubitamer
Analytical Laboratory Aliquots
1
2
3
4
5
6
7
Split Sample Aliquots
EMSL-Las Vegas Splits
ERL-Corvallis Splits
60
60
25
50
250
10
250
125
500
125
125
125
125
Unfiltered
Unfiltered
Centrifuged
Unfiltered
Filtered, pH < 2 with HN03
Filtered, MIBK-HO" extract
Filtered
Filtered, pH < 2 with H2SO4
Unfiltered
Unfiltered, pH < 2 with H2SO4
Unfiltered, pH < 2 with HN03
Unfiltered, 0.1 mL 5% HgCI2
Filtered, pH < 2 with HN03
pH, closed system
DIG, closed system
True color
Turbidity
Ca, Fe, K, Mg, Mn, Na
Extractable At
cr, r, NO3", Sio2, so4~2
DOC, NH/
ANC, DIG, pH, conductance
Total P
Total Al
NO3~, S04"2
Metals, SCU'2, Si02
"All ahquots and the EMSL-Las Vegas split samples were stored at 4°C in darkness.
ERL-Corvallis split samples were stored at room temperature.
"Methyl-isobutyl ketone, hydroxyquinolme.
Table 2-7. Maximum Holding Times Specified for Samples,
Western Lake Survey - Phase I
Holding Time
Parameter
7 days
14 days
28 days
6 months3
N03 , air-equilibrated pH, extractable Al
ANC, conductance, DIG, DOC
Total P, NH4 + , Cl~, SO4~2, total dissolved
F~, Si02
Ca, Fe, K, Mg, Mn, Na, total Al
a Although holding time has been established at 6 months (U.S.
EPA 1983), samples were required to be analyzed within 28
days for these parameters due to survey data reporting
restrictions (Linthurst et al. 1986).
tested and proven protocol in the ELS-I, there was
no need for the helicopter crews to be unaware of
the identity of calibration lakes.
Concern that the ground sampling protocol might
result in delayed delivery of samples to the field
laboratories for processing led to the development
of an experiment using three preservation proce-
dures. Both of the ground crew's samples and two
of the helicopter crew's samples were preserved at
the field laboratory on the day of collection. Of the
three samples collected by the helicopter crew, one
was selected randomly and was held in the dark
at 4°C at the field laboratory for a specified period
of time before it was processed and preserved. This
holding time depended on which of the three possible
processing procedures applied (Section 4.4). Further
discussion of the processing and preservation-time
procedures for samples from the calibration lakes
appears in the WLS-I QA plan (Silverstein et al.
1987a).
To evaluate interlaboratory bias, all samples
collected for the calibration study were assigned to
the two analytical laboratories for analyses. A die
roll was used to determine the assignment of the
samples collected by the helicopter crews, and a coin
toss was used to determine the assignment of the
sample collected by the ground crews. This proce-
dure (Figure 2-7) ensured that one sample collected
by each field crew (helicopter and ground) would
be sent to each analytical laboratory.
2.7 Data Base Management
The objectives of the data management component
of the WLS-I were to enter the data, prepare a
validated data set, and perform statistical analysis
and evaluation of the data. The Environmental
Sciences Division of ORNL designed and imple-
mented data base management activities using the
Statistical Analysis System (SAS) software (SAS
Institute 1982) on tandem IBM 3033 mainframe
computers.
The validated data set consists of individual records
for each lake sampled and contains site descriptors,
field observations, laboratory measurements, and
calculated variables for the lakes sampled. The
definitions of all variables measured, their formats,
units of measure, and comments are presented in
a data base dictionary (Kanciruk et al. 1986b).
Detailed discussions of the QA measures developed
for the data base are given in Rosen and Kanciruk
(1985) and Kanciruk et al. (1986c).
The lake ID assigned before field work began
(Section 2.2.5) was used as the key site identifier.
22
-------
Figure 2-7. Allocation of samples to analytical laboratories for the calibration study. Western Lake Survey - Phase I. This
procedure was used to assign routine (R), duplicate (D), and triplicate (T) samples collected by ground crews
(GC) and helicopter crews (HC) to one of two analytical laboratories.
Ground Samples
(Forest Service)
Routine
(1 st Sample
Taken)
Duplicate
(2nd Sample
Taken)
Helicopter Samples
{EPA, Lockheed-EMSCO)
V
RGC
DGC
V
.
Routine
(1st Sample
Taken)
. RHC
Duplicate
(2nd Sample
Taken)
DHC
Triplicate
(3rd Sample
Taken)
THC
y
V
Laboratory
1
Randomly
Selected
Sample
Shipment
\ X
X
Withheld
Helicopter
Sample
/
/
\
Laboratory
2
The lake ID numbers were cross-referenced to lake
names, state and county Federal Information
Processing Standards codes, latitude and longitude,
and other identifiers within the site descriptor file.
Data base development is summarized in Figure 2-8.
The working data base consisted of three main
data sets: raw (Data Set 1), verified (Data Set 2),
and validated (Data Set 3). Each contained numerous
SAS relational (tabular) data files (SAS Institute
1982). The verified and validated data sets were
developed from the raw data set as quality assurance
procedures were implemented (Section 3). The final
reported data set (Data Set 4) was developed from
23
-------
Figure 2-8. Procedures used to develop Data Set 1 (raw). Data Set 2 (verified). Data Set 3 (validated), and Data Set 4
(final) for the Western Lake Survey - Phase I data base.
Visual Form Check
I
Data Entry 1
I
Data Entry 2
Error and
Range Check
Raw Data Set
(Data Set 1)
Substitution
and
Replacement
Final Data Set
(Data Set 4)
Batch Reports
Verified Data Set
/Data Set 2)
Verification
Data Editing
and Flagging
Site
Reports
Maps
Validation
Data Editing
and Flagging
1
Validated Data Set
(Data Set 3)
24
-------
the validated data set by removing erroneous data
and substituting for missing values when appropriate
(Section 3.4). Protocols used for these changes are
discussed in Eilers et al. (1987). Copies of all data
sets have been maintained on tape as a permanent
record.
Background information (lake names, lake IDs, and
physical characteristics) was transmitted from ERL-
Corvallis to ORNL, where it was entered into a data
file. Data forms that contained the information
obtained at field sites and analytical laboratories
were sent to ORNL and to EMSL-Las Vegas. The
information was entered into the data base at ORNL.
After entry, the data were sent to the EPA IBM
computer at Research Triangle Park, North Carolina,
for access and review by EMSL-Las Vegas personnel.
Corrections and flags were returned to ORNL as the
verified data set. After verification, the data were
validated jointly by ERL-Corvallis, EMSL-Las Vegas,
and ORNL staff.
25
-------
Section 3
Quality Assurance
3.1 Design of the Quality Assurance
Program
The WLS-I quality assurance (QA) and quality control
(QC) activities summarized in this section are
described in detail in Silverstein et al. (1987a). Most
of the QA/QC procedures were previously used in
the ELS-I (Drouse et al. 1986).
The QA/QC activities included (1) selecting analyt-
ical laboratories, (2) training field sampling and field
laboratory crews, (3) maintaining communication
with management, sampling, and analytical person-
nel, (4) conducting on-site field and laboratory
inspections, and (5) collecting and analyzing a variety
of QA/QC data to quantify data quality. Quality
assurance protocols were developed for collection,
preparation, preservation, shipment, and analysis of
samples, as well as for reporting, verification, and
validation of analytical results.
3.1.1 Selection of Analytical Laboratories
The objective of the selection process for analytical
laboratories was to choose the minimum number
of qualified laboratories needed to analyze WLS-I
samples. The Contract Laboratory Program estab-
lished to support EPA's hazardous waste monitoring
activities was the most efficient mechanism for
accomplishing this objective.
The selection procedure was initiated by preparing
a statement of work that defined the analytical and
QA/QC requirements in a contractual format. Next,
bids were solicited from analytical laboratories. On
the basis of pre-award sample analyses and on-site
evaluations. Environmental Monitoring and Ser-
vices, Inc. (EMSI) in Thousand Oaks, California, and
Versar, Inc., in Springfield, Virginia, were selected
from among the responding laboratories. During the
WLS-I, one laboratory processed about 40 percent
of the samples, primarily from the Central (4D) and
Southern (4E) Rockies, and the other laboratory
processed the remaining 60 percent (Table 3-1).
3.1.2 Training of Sampling and Field Laboratory
Crews
Because data quality is affected by the ability of
personnel to collect, process, and analyze samples
properly, training of the WLS-I personnel was
essential to ensure consistent application of
sampling and analytical protocols. Before field
sampling activities began, field sampling and field
laboratory personnel from Lockheed-EMSCO were
trained in sampling and field laboratory procedures,
respectively. These personnel, most of whom had
also participated in the ELS-I activities (Linthurst et
al. 1986), then were sent to the field stations where
they trained EPA and Forest Service personnel. This
ensured that field activities were performed consist-
ently and according to approved procedures.
3.1.3 Communications
Monitoring of QA activities required continuous
communication among the many groups responsible
for data collection. During the sampling and
analytical phases, these communications were
centralized through the QA staff at EMSL-Las Vegas
who made daily calls to the field stations and to the
analytical laboratories. These calls were used (1) to
determine whether QA/QC guidelines were being
followed, (2) to check that samples were being
processed and analyzed properly, (3) to obtain current
sample and QA/QC data, and (4) to discuss sampling,
processing, and analysis issues related to logistics,
methods, and QA/QC so that problems could be
resolved quickly and efficiently. All communications
were logged on appropriate field communications
forms and in bound notebooks.
3.1.4 On-Site Inspections
On-site inspections of field stations, remote sites,
and analytical laboratories were conducted during
the WLS-I to ensure that sampling and analytical
activities were being performed according to
established protocols. The results of these evalua-
tions were documented in site-evaluation reports
prepared by the QA staff and presented to EPA.
3.1.5 Quality Assurance/Quality Control Data
Collection and Analysis
Specified QA/QC procedures and samples were
used in the field, at the field laboratories, and at
the analytical laboratories (Silverstein et al. 1987a)
to maintain the quality of the data and to ensure
that data quality could be characterized accurately.
Rigid requirements for instrument calibration helped
26
-------
Table 3-1. Number of Samples from Probability Sample Lakes Analyzed by the Analytical Laboratories, by Subregion,
Western Lake Survey - Phase I
Subregion
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
Non-Calibration Lakes
4A
4B
4C
4D
4E
Lab 1
6
0
0
115
130
Lab 2
135
149
133
7
0
Non-
Calibration
141
149
133
122
130
Total
Calibration3
9
10
10
7
9
Total
150
159
143
129
139
West
251
424
675
45
720
a Analyzed by both laboratories.
to provide reliable measurements. Data derived from
the QA and QC samples were compared to the
expected values and established ranges (Section
2.1.1, Table 2-1).
Several types of QA and QC samples (Figure 3-1)
were used to ensure that sampling and analytical
methods were performed according to specifications.
The results of QA sample analysis were used to
evaluate the performance of field sampling methods
and field and analytical laboratory procedures and
to evaluate overall data quality for the survey.
Analysis of the QC samples allowed field samplers
and laboratory personnel (in field and analytical
laboratories) to identify and correct immediately local
problems such as poor instrument performance or
reagent contamination before and during routine
sample analyses.
3.1.5.1 Quality Assurance Samples— Quality
assurance samples, introduced into batches of lake
samples at the field station, were analyzed at the
field laboratory and at the analytical laboratory.
Analytical laboratory personnel were not told the
origin, identity, or chemical composition of the
samples. Thus, the QA samples were analyzed as
if they were routine lake samples. The QA samples
used were field blanks, trailer blanks, field
duplicates, and field audits (Figure 3-1).
Field blanks were prepared at the field laboratory
from American Society for Testing and Materials
(ASTM) Type 1 reagent-grade deionized water. The
sampling crew transported the blank water in
Cubitainers to the sampling site and processed the
blank through the Van Dorn sampler as if the blank
were a lake sample. Each helicopter crew collected
one field blank on each operating day; each ground
crew collected two field blanks during the course
of the survey.
Field blanks were added to the sample batches that
were sent to the analytical laboratories. These QA
samples were used to identify potential
contamination resulting from sampling and
analytical processes and to provide data to estimate
the system decision and detection limits,
quantitation limit, and background concentrations
(Sections 4.2 and 4.3) for each variable. Measured
sample values at or below these system decision
and detection limits were considered to be indis-
tinguishable from blanks.
Occasionally, due to logistical constraints, field
blanks were not scheduled to be processed at any
site on a particular sampling day. In these instances
a deionized water sample, prepared in the field
laboratory without processing through the Van
Dorn sampler, was substituted for the field blank.
This "trailer blank" was then added to the sample
batch.
Field duplicates were collected immediately after
routine samples were collected. The same procedure
was used to collect both types of samples. For each
base site, one helicopter crew collected one field
duplicate on each sampling day. Each ground crew
collected two field duplicates during the course of
their sampling. Field duplicates were processed by
the field laboratory and were added to the sample
batches sent to the analytical laboratories. Field
routine/duplicate pairs were used to determine the
homogeneity of the lake sample and the precision
of the field sampling techniques, field laboratory
procedures, and analytical laboratory procedures
(Section 4.3).
Two types of field audit samples (natural and low-
concentration synthetic) were prepared by Radian
Corporation in Austin, Texas, and were used to
evaluate overall field and analytical laboratory
performance (Section 4.3). A third type of audit
sample, a laboratory audit, was used during the ELS-I
(Linthurst et al. 1986) but was not used during
the WLS-I. An evaluation of data collected during
the ELS-I indicated that field audit results were no
more variable than laboratory audit results;
therefore, the laboratory audits were deleted.
27
-------
Figure 3-1. Origin, use, and flow of quality assurance/quality control samples. Western Lake Survey - Phase I. Results
of the analyses of these samples were used to quantify data quality and to ensure that error was minimized
at each level of sample handling and processing.
Field
Samplers
I
Ftf\M P/rtnlr
Field
Laboratory
-" I
Analytical
Laboratory
ta. rVn/r/n/ln<
Trailer Blank
(in Lieu of Field Blank}
Trailer Blank
Field Duplicate
Field Duplicate
Field Duplicate
Trailer Duplicate
(Split of a Randomly
Selected Routine
Lake Sample)
QCCS
Hydrolab pH. Cond.
Natural
Audits
Lake Superior, MN
(FN3)
Big Moose Lake, NY
IFN4)
BagleyLake, WA
(FNS; FN6)
Radian Corp.
Synthetic Audits
(FL11;FL12)
Prepared FNs
Field Audits
QCCS
pH, DIC, Turbidity
Calibration Blank
DIC
Field Audits
QCCS
Calibration/
Reagent Blank
Matrix Spike
(on Field Sample)
Laboratory Duplicate
(Split of Field Sample)
Field natural audit samples were composed of
natural lake water. Field synthetic audit samples,
which were prepared to simulate natural lake water,
included a matrix of analytes at specified theoretical
concentrations. The mean concentrations of the
audit samples are given in Section 4.3. Field audit
samples were used (1) to determine relative bias
between analytical laboratories so that
measurements made by the two laboratories could
be compared and (2) to indicate precision of those
measurements through repeated analysis of the
same sample type.
Both types of field audit samples, received at the
field laboratory in 2-L aliquots, were filtered and
processed into aliquots as if they were routine lake
samples. The samples then were shipped to the
analytical laboratory as if they were routine samples.
i.e., their identities and concentrations were
unknown. Four field natural samples (designated
FNS, FN4, FN5, and FN6) and one low-concentration
field synthetic sample (two lots, designated FL11 and
FL12) were used.
Three of the natural samples (FN4 from Big Moose
Lake in the Adirondack Mountains of New York, and
two separate samples, FN5 and FNS, from Bagley
Lake inthe North Cascade Mountains of Washington)
represented two types of low ANC and low ionic
strength surface waters expected to be encountered
during the survey. The fourth natural sample (FNS
from Lake Superior) had relatively high ANC and high
ionic strength.
Following collection, each natural audit sample lot
was filtered in bulk and divided into 2-L aliquots
28
-------
(as individual audit samples), and the aliquots were
stored at 4°C until use. Field synthetic audit samples
were prepared as lot concentrates and diluted just
before they were sent in 2-L bottles to the field
laboratory.
3.1.5.2 Field Sampling and Field Laboratory
Quality Control Samples—Quality control check
samples (QCCSs) were used by the helicopter crews
to check the calibration of the Hydrolab pH and
conductance functions before sampling and to check
for instrument drift after sampling. Daily QC checks
were made before and after sampling (Section
2.5.1.2).
Three types of QC samples were used by the field
laboratory staff to ensure that instruments and data
collection were within specified control limits
(Silverstein et al. 1987a). A calibration blank,
analyzed before samples, was used to check for
baseline drift of the carbon analyzer and to check
for contamination. Calibration and drift of the carbon
analyzer and of the instruments used to measure
pH and turbidity were checked with QCCSs. A "trailer
duplicate" sample was used to evaluate the precision
of field laboratory measurements. The field
laboratory supervisor randomly selected one lake
sample per operating day as the trailer duplicate,
which was analyzed in duplicate for pH, DIG, true
color, and turbidity.
3.1.5.3 Analytical Laboratory Quality Control
Samples—Six types of QC samples were used by
the analytical laboratories to ensure that instruments
and data collection were within control limits. These
six QC sample types were (1) calibration blanks, (2)
reagent blanks, (3) detection limit QCCSs, (4) low-
concentration and high-concentration QCCSs, (5)
matrix spikes, and (6) laboratory duplicates. The use
of these QC samples in evaluating data quality is
described in detail in Silverstein et al. (1987a).
3.2 Data Verification
The objectives of the data verification process were
to identify, correct, and "flag" raw data of question-
able and unacceptable quality, and to identify data
that may have had to be eliminated during or after
validation (Section 3.3). These objectives were
accomplished by reviewing QA/QC data recorded
at the sampling site, at the field laboratory, and at
the analytical laboratory. Computer programs were
used to automate the verification procedure as much
as possible. For each data package (representing one
sample batch) the audit team performed a sample-
by-sample evaluation. The audit team reviewed
comments and questions associated with the sample
batch; performed QA checks for data consistency and
reasonableness; reviewed QA sample data; obtained
confirmation, correction, and reanalysis data from
the analytical laboratories; and provided a verified
data set to ORNL. For each batch, the audit team
prepared a summary of the reporting errors and the
required confirmation and reanalyses.
3.2.1 Review of Field Data Forms
Verification began with the receipt of the data forms
from the field. The auditor reviewed the field forms
for the following items: lake ID, instrument calibra-
tion data, pH and DIG sample and QCCS data, site
depth and Secchi disk transparency, Loran-C and
map coordinates, and tags (data qualifiers assigned
by field sampling crews and field laboratory
personnel). Data anomalies were reported to the field
laboratory coordinator for corrective action, and data
reporting errors were corrected before the data were
entered into the raw data set.
3.2.2 Initial Review of Analytical Laboratory
Sample Data Package
The sample data packages received from the
analytical laboratories were reviewed for complete-
ness, internal QC compliance, and appropriate use
of data qualifiers. A checklist was used by the auditor
to ensure consistency in the review of all data
packages. Problems were reported to the analytical
laboratory manager for corrective action. Comments
that the laboratory had submitted with the data
package cover letter also were reviewed to determine
their impact on data quality and to determine any
need for follow-up action by the laboratory.
3.2.3 Review of Quality Assurance/Quality
Control Data
After the data were entered into the raw data set,
a magnetic tape containing the data was sent to the
National Computer Center in Research Triangle Park,
North Carolina. The QA personnel then were able
to access the data by telecommunication. A series
of computer programs that comprise the Aquarius
QA/QC system (Fountain et al. 1986) was used
during verification. These programs were used to
identify or flag results that were classified as
"exceptions" (i.e., results that did not meet the
expected QA/QC limits). The auditor used the output
from these programs (with original raw data and field
notebooks) to complete the verification report form
specified in the QA plan (Silverstein et al. 1987a).
The verification report was a worksheet designed
to guide the auditor systematically through the
verification process by explaining howto(1)flag data,
(2) track data resubmissions and requests for
reanalysis and confirmation, (3) list the steps that
lead to identification of QA exceptions, and (4)
summarize modifications to the raw data set (i.e.,
to prepare records of numeric and flag changes).
29
-------
Each sample was verified individually and by
analytical batch. The sample had to meet checks for
anion/cation percent ion balance difference and for
percent difference between calculated and mea-
sured conductance in order not to be flagged, unless
the discrepancy could be explained by either the
presence of organic species (as indicated by a
protolyte analysis program) or by an obvious and
correctable reporting error. The ion balance was
determined by converting measured values in mg L~1
to fjeq L~\ and then comparing the sum of measured
anions to the sum of measured cations.
Conversion factors are given in Table 3-2. Calculated
conductance was determined by using the conver-
sion factors in Table 3-3. Additional data qualifiers
were added to a given variable when the QA samples
(field blanks, field duplicates, or audit samples)
within the same analytical batch did not meet the
acceptance criteria. Data were also qualified if
internal QC checks such as matrix spike recovery,
calibration and reagent blank analyses, internal
duplicate precision, required instrumental detection
limit, QCCS percent recovery, and required holding
times were not met. The protolyte analysis program
flagged field laboratory and analytical laboratory
measurements of pH, DIG, ANC, and DOC (Drouse
et al. 1986). In all cases, each flag generated by
the Aquarius system was evaluated by the auditor
for reasonableness and consistency before it was
entered into the verified data set. These programs
Table 3-2.
Ion
Factors Used to Convert mg L~1 to ^ieq L~1 for
Anion/Cation Balance Check8, Western Lake
Survey - Phase I
Factor
Ion
Factor
Ca + 2
cr
Mg + 2
N03-
F-
49.9
28.2
82.3
16.1
52.6
NH4+
K +
Na +
so4-2
55.4
25.6
43.5
20.8
aHillman et al. (1986).
Table 3-3. Factors Used to Convert mg L 1 to pS cm 1 for
Determining Calculated Conductance"-6, Western
Lake Survey - Phase I
Ion
Conductance Factors0
Ion
Conductance Factors0
Ca+2
cr
co3-2
H +
HC03-
Mgt2
2.60
2.14
2.82
3.5 x 10B
0.715
3.82
Na+
NH4 +
SO4~2
N03-
K +
OH' 1
2.13
4.13
1.54
1.15
1.84
.92 x 105 (mole
L")
aHillmanetal. (1986).
b Factors for H+ and OH~ are used to convert mole L'1 to
'1
/iS cm
cConductance
strength.
cm"1) factors at 25°C for 0.001 ionic
also identified outlier data based on QA and QC
samples. The outlier data were the basis for
requesting confirmation of data from the analytical
laboratories and for reanalysis of suspicious
samples. Data verification and data validation
(Section 3.3) were performed in tandem; however,
the verification process had to be completed before
the validation process could be completed (i.e., Data
Set 2 had to be in place before final conclusions
could be drawn from Data Set 3).
3.3 Data Validation
The data validation process identified possible
systematic and random errors in chemical analyses
that were not always revealed by verification
procedures (Eilers et al. 1987a). The validation
process closely followed that used in the ELS-I
(Linthurst et al. 1986). The validity of nonchemical
measurements (e.g., lake area, watershed area, lake
depth and elevation) was also evaluated.
Potential random errors were identified by gener-
ating a list of outliers, i.e., observations that were
not typical of other sample values. Outliers were
detected using a variety of univariate, bivariate, and
multivariate approaches. Initially, each variable was
considered individually to identify values that were
outliers with respect to the sample distribution.
Box plots (Tukey 1977) for each variable were
prepared using SAS statistical procedures (SAS
Institute 1982). These plots summarized the data for
a variable by using the difference between the upper
and lower quartile (interquartile range). For this
procedure, outliers were defined arbitrarily as those
values greater than the absolute value of three times
the interquartile range. Certain pairs of variables
were expected to exhibit a linear relationship.
Outliers in these relationships were identified by
examination of scatter plots and least-squares linear
regression analyses. Standardized residuals were
calculated as follows:
actual value - predicted value | /
residual standard deviation I
Values for which the standardized residuals were
greater than three were identified arbitrarily as
outliers. Because least-squares analysis can be
biased strongly by certain types of outliers, the
residuals from resistant line fits (i.e., lines fit through
the medians of partitions of data, Velleman and
Hoaglin 1981) were examined for selected variables
measured in the field and field laboratory. Analytical
variables were examined using a process of linear
regression with three iterative steps. Sequentially
removing outliers is a helpful technique for locating
additional outliers that would not have been
identified without previously removing the major
30
-------
aberrant values. Outliers among related groups of
variables were detected by using the SAS FASTCLUS
cluster analysis procedure (SAS Institute 1982) and
principal components analysis.
Statistical outliers are useful for highlighting suspect
data, but considerable caution must be exercised in
the treatment of these outliers. It cannot be
presumed that all outliers are errors. Once the
outliers were identified, it was necessary to
determine the reason for the aberrant data.
Outliers were initially screened by comparing them
to other related variables that might explain their
high residual standard deviation. For example, a
sample with a suspiciously high concentration of
nitrate might appear reasonable if the sample also
exhibits high concentrations of other nutrients.
Those outliers remaining after screening and
confirmation of the reported data were flagged in
the validated data set. In a few cases, sufficient
evidence was available to indicate that the reported
value was erroneous. These values were flagged in
the validated set and were marked for substitution
in the final data set (Section 3.4).
Analytical variables having possible systematic
errors were detected by comparing values from the
WLS-I to those from other lake survey data sources.
These data sets were selected on the basis of
geographic location, accessibility of the data, and
documentation of sampling and analytical methods.
Comparisons were made using scatter plots and
linear regression procedures to identify values
requiring additional scrutiny. This approach failed
to identify systematic differences that were suffi-
ciently large to indicate major analytical errors or
differences of a magnitude that would affect data
analysis and interpretation. The details of the
validation process are described in Eilers et al.
(1987a).
3.4 Development of Final Data Set
Calculating population estimates (Section 2.3) is
difficult if values are missing from the data set. A
final data set (Data Set 4) was prepared to resolve
problems relative to data interpretation because of
missing values in the validated data set. Data Set
4 was modified by averaging field duplicate values
and substituting for analytical values determined to
be in error during validation.
Several substitution methods were used. Values
from duplicate samples were used when available.
Redundant analyses were performed for pH, DIG, and
conductance (Section 2.5). Redundant measure-
ments on split samples (Section 2.5.2) were
performed for metals and other elements. If a
duplicate measurement was not available, a
comparable measurement (e.g., in situ conductance
or analytical laboratory conductance) was chosen
and was substituted for the missing value using a
linear regression routine. If redundant measure-
ments were not available or acceptable, a substi-
tution value was calculated from the available data
using observed relationships with other variables
(e.g., calcium and ANC). The last option used for
identifying a substitution value was to use the
stratum mean. All substitution values were exam-
ined a second time for acceptability before including
them in the final data set. Substituted values were
flagged as such in the final data set. Only 56 values
out of approximately 20,000 chemical measure-
ments examined during the validation process were
substituted in the final data set. Of these 56 values,
only 2 were missing from the data collection process.
Two other changes were made in the final data set.
If duplicate data met QA precision criteria, the
average value of the duplicates was used in the final
data set. Negative values for parameters other than
acid neutralizing capacity that resulted from
analytical calibration bias were set equal to zero.
The bias in the estimate of variance due to this
adjustment did not affect data analyses. All values
modified in the final data set were flagged to indicate
that they did not represent original measurement
values.
31
-------
Section 4
Quality Assurance and Calibration Study Results
4.1 Site Confirmation
Confirmation that the lakes sampled by helicopter
were the lakes intended to be sampled in the WLS-I
was accomplished by comparing aerial photographs
taken during the survey to lakes as they appeared
on the topographic maps. Lakes accessed by ground
crews were checked by comparing comments and
observed lake outlines noted on the field forms with
topographic maps. Two lakes were mistakenly
sampled. The data from these lakes were deleted
from the final data set and the lakes targeted for
sampling were classified as "not visited" (Section
2.2.2). Three lakes were determined after sampling
to be non-target; these data were also removed from
the data base. For two of these lakes, conductance
values exceeded the maximum criterion of 1500
/jS cm"1; the third lake was determined to be a
livestock watering pond.
4.2 Detection Limits (Detectability)
Helicopter crews and ground crews collected 236
field blanks during the WLS-I that were used for
statistical analyses. Data from the analysis of these
blanks were used to estimate the overall background
contamination that occurred during sampling and
analysis. Two criteria (Table 4-1) were used to
evaluate analytical detectability: the system decision
limit (SDL) and the required detection limit (RDL).
The SDL for each variable is based on a nonpara-
metric statistical evaluation of the field blanks
(Section 3.1.5.1), and is the concentration limit
(background) above which an analyte can be detected
with a known degree of confidence. This limit is
defined as the 95th percentile (P95) of the distribution
of field blank measurements. The SDL provides an
estimate of the level of an analyte that potentially
can be introduced during sample collection, han-
dling, processing, or analysis. For measured values
below the SDL, it cannot be known with certainty
whether the analyte was present in the lake or was
introduced at some stage of handling. The SDL is
compared tothe RDL, i.e., the level of detection which
the analytical laboratory was required by contract
to meet when analyzing laboratory blanks. Because
field blanks were subject to more handling than were
Table 4-1. Detectability Based on Evaluation of Field
Data8, Western Lake Survey - Phase 1
Variable13
Al, extractable
Al, total
ANC (Meq L"1)
Ca
cr
Conductance (^S cm"1)
DIG, air equilibrated
DIG, initial
DOC
F~, total dissolved
Fe
K
Mg
Mn
Na
IMH4 +
N03-
P, total
SiO2
so4-2
SDLC
0.004
0.019
3.9
0.07
0.04
1.6
0.33
0.43
0.3
0.003
0.01
0.01
0.01
0.01
0.01
0.01
0.071
0.006
0.18
0.07
Blank
RDLd
0.005
0.005
e
0.01
0.01
f
0.05
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
apH was not determined for field blanks and thus decision and
detection limits do not apply to pH measurements.
bAII units are reported as mg L~1 unless otherwise indicated.
cSystem decision limit; the 95th percentile of 236 field blank
measurements.
d Required detection limit.
e Absolute blank values were required to be < 10 /jeq I/1
f This value, the mean of six nonconsecutive blank analyses,
was required not to exceed 0.9 j^S cm"1.
laboratory blanks, background concentrations infield
blanks were expected to be higher than levels in
laboratory blanks.
The quality assurance results are presented in the
units reported by the analytical laboratories (gener-
ally, in mg L"1). Throughout the remainder of the
report, the results are presented in units more
convenient for direct comparison among variables
(generally pieq L"1). Conversion factors are given in
Section 3.2.3, Tables 3-2 and 3-3.
The SDL was below or near the RDL for most
variables. For total Al, Cl~, DIG (initial and air
equilibrated), and DOC, the SDLs were notably
higher than the RDLs. However, the SDLs for these
parameters are comparable to field blank data
collected in the ELS-I. Field blank data are discussed
in detail in Silverstein et al. (1987b). Other analytes
32
-------
(Ca, NOa , and Si02) for which the SDLs were outside
the preferred range are discussed below.
The SDL for N0a~ was 0.071 mg L"1. The concen-
trations of NOs" in WLS-I lake samples were
generally very low. Therefore, nitrate was not a major
contributor of anion equivalents in the samples
collected from WLS-I lakes in the fall. As with N03~,
more than 5 percent of the field blanks collected
had concentrations of Ca and Si02 that caused the
SDLs to be substantial. The SDL for Ca was 0.07
mg L"1, and the SDL for Si02 was 0.18 mg L"1.
Therefore, there is a 5 percent, or greater, chance
of introducing contamination at levels above these
limits for Ca and Si02.
These results indicate that there was no evidence
for systematic contamination for any analyte.
However, random contamination as a result of
sampling, processing, or analytical error may have
caused the SDLto exceed the RDLfor some analytes
(Section 4.6).
4.3 Precision
Sampling and analytical variance, apart from
seasonal variations in lake chemistry, can arise from
three major sources: (1) a field component associated
with short-term variability in lake chemistry, (2) an
analytical component associated with aliquot
preparation or with variation in instrument response
within an analytical batch, and (3) an analytical
component associated with batch-to-batch variation
in instrument calibration and response. The relative
importance of these sources of variation was
assessed by comparative statistical evaluations of
field audits, field duplicates, trailer duplicates, and
laboratory duplicates.
Precision can be calculated in two ways. One method
is by determining overall (among-batch) precision
using field audit (natural and synthetic) samples to
determine the precision across batches and over
time. The other method is to determine precision
within batches by using duplicate samples (field,
trailer, and analytical laboratory) to estimate
precision associated with the different stages of
sample handling (i.e., sampling, processing, and
analysis).
Overall among-batch precision for the WLS-I field
audits was calculated as the standard deviation (SD)
and as the percent relative standard deviation
(%RSD) except for pH, which was calculated as the
SD only. The %RSD is expressed as 100 times the
ratio of the standard deviation to the mean (x):
%RSD= 100^
where x = the mean of all sample concentrations.
Tables 4-2 and 4-3 show the mean concentrations,
%RSDs, and SDs for synthetic and natural audits,
respectively. Data from the two field synthetic lots
(FL11 and FL1 2) were combined, or pooled, because
the theoretical concentrations of these two lots are
the same. The overall precision for most variables
in the six lots of field audits (two pooled synthetic
and four natural) was reasonable, considering that
they were analyzed among many batches, except for
those variables that had mean concentrations so
close to the instrument detection limit that the
precision estimates have little meaning. Data users
can compare these mean concentrations to the SDLs
(Section 4.2 and Tables 4-2 and 4-3) to gauge
whether the calculated precision is within acceptable
limits for a particular application. Among all the field
audits, no precision estimates were consistently high
for any one variable, and no specific trend of
imprecision was observed. A detailed, audit-by-audit
discussion is not within the scope of this report
because there are different concentrations for every
variable in each of the five audit types (one synthetic
formula and four natural systems). Silverstein et al.
(1987b) provide this detailed discussion.
For the duplicate samples (field, trailer, and
laboratory), within-batch precision is expressed as
the root-mean-square (RMS) of the %RSDs of all
sample pairs, except for pH, for which within-batch
precision is expressed as the RMS of the SDs of
the sample pairs. To determine if duplicate sample
pair concentrations are sufficiently above the
detection limit to estimate precision reliably, a
quantitation limit is calculated for all variables except
pH. The quantitation limit was calculated as 10 times
the SD (10 SB) of the concentrations of the
corresponding blanks (field, trailer, or laboratory).
Precision estimates can be calculated from all
sample pairs (pairs with x > 0), some of which are
greatly affected by background (pairs with x near
0) or from those pairs minimally affected by
background (only pairs with x > 10 SB). Tables 4-4,
4-5, and 4-6 compare the values for all pairs with
x > 0 to only those pairs with values above the
quantitation limit (x > 10 SB).
Overall within-batch precision estimates calculated
from field duplicate pairs are shown in Table 4-4.
Like field blank SDLs, which were expected to be
higher than RDLs (Section 4.2), field duplicate pairs
were expected to be less precise than laboratory
duplicate pairs. Field duplicates were subjected to
more handling in less controlled environments (e.g.,
the lake sites) than were laboratory duplicates. In
addition, field duplicates were truly separate
samples, whereas laboratory duplicates were split
samples of a particular aliquot that was created
33
-------
Table 4-2.
Overall Among-Batch Precision Estimates Calculated from Field Synthetic Audit Samples, Western Lake Survey -
Phase I
Variable3
Al, extractable
Al, total
ANC (jtfsq L~1)
Ca
cr
Conductance (/^S cm"1)
DIG, air equilibrated
DIG, initial
DOC
F~, total dissolved
Fe
K
Mg
Mn
Na
NH4 +
NO3-
P, total
pH, acidity (pH units)
pH, alkalinity (pH units)
pH, air equilibrated (pH units)
Si02
SO4~2
Theoretical
Concentration
0.020
0.020
e
0.19
0.34
e
e
0.96
1.0
0.04
0.06
0.20
0.45
0.10
2.74
0.17
0.464
0.027
e
e
e
1.07
2.28
SDLb
0.004
0.019
3.9
0.07
0.04
1.6
0.33
0.43
0.3
0.003
0.01
0.01
0.01
0.01
0.01
0.01
0.07
0.006
g
g
g
0.18
0.07
Mean
Concentration0
0.005
0.027
111.0
0.22
0.36
19.7
1.44
1.54
1.0f
0.043
0.10
0.21
0.45
0.10
2.77
0.1 6f
0.483
0.025
6.94
6.95
7.24
1.10
2.30
Precision
% RSD
48. Od
29.3
6.1
16.1
5.6
4.4
18.3
20.4
25.0
7.1
153.0d
8.7
3.8
14.0
8.5
16.2
6.5
22.0
h
h
h
10.5
5.4
SD
0.002
0.008
6.76
0.03
0.02
0.87
0.26
0.31
0.26
0.003
0.009
0.02
0.02
0.01
0.24
0.03
0.03
0.005
013
0.11
0.14
0.12
0.12
aAII units are reported as mg L 1 unless otherwise indicated.
bSystem decision limit; the 95th percentile of 236 field blank measurements.
cMean concentration of pooled field synthetic lot 11 (FL11, n = 30) and field synthetic lot 12 (FL12, n = 17).
dPoor precision may be the result of sample instability, sample mixing error, or both (Best et al. 1986).
eAlthough theoretical values for ANC, DIC, and pH can be calculated, the theoretical value is dependent upon the concentrations of
chemicals added to the synthetic audit sample.
f n = 45.
9Decision and detection limits are not applicable to pH measurements.
hOnly standard deviation values are calculated for pH measurements.
within a controlled environment in the analytical
laboratory. Therefore, the DQOs for intralaboratory
precision that were used to check analytical
laboratory precision should not be applied rigidly to
field duplicate precision. These DQOs, however,
were useful as a gauge to assess field duplicate
precision estimates.
For field duplicates, eight variables met intralabor-
atory precision goals. These variables were total Al,
ANC, Ca, DIC (air-equilibrated and initial), K, N03",
and S04~2. Seven variables did not meet the
intralaboratory precision goals, but the precision
estimates were only slightly higher than the
intralaboratory precision goals. These variables were
CI", Mg, Na, NH4+, pH (alkalinity and acidity), and
S\0z. Therefore, 15 of the 23 variables presented
in Table 4-4 can be considered to have reasonable
to excellent precision. For one variable (Mn),
precision above the quantitation limit could not be
calculated because there was only one duplicate pair
where the mean of both sample concentrations was
greater than the quantitation limit.
Seven variables had poor field duplicate precision.
These were extractable Al, conductance, DOC, Fe,
total dissolved F", air-equilibrated pH, and total P.
Extractable Al had a quantitation limit of 0.021 mg L 1
and a small number of duplicate pairs (n = 11) with
concentrations above that limit, which would be
expected based on the low number of acidic lakes
sampled and the correspondingly low extractable Al
concentrations. Most of the field duplicate pairs had
conductance values that were low, because many
dilute lakes were sampled in the WLS-I (Section 5.9).
This factor may account for much of the variability
in the precision of conductance measurements. For
Fe and total P, few field duplicate pairs had mean
concentrations that were above the quantitation
limit, indicating that the majority of the lakes in the
WLS-I had low concentrations for these variables.
Total dissolved F" had a quantitation limit of 0.016
mg L"1 and a substantial number (n = 90) of field
duplicate pairs with concentrations above that limit.
The variability of total dissolved F" seems to be most
apparent within one analytical laboratory (Silverstein
et al. 1987b). The calculated precision for air-
equilibrated pH was 0.17 pH unit compared to an
intralaboratory precision goal of 0.05 pH unit. This
variability also may be related to the large number
of dilute lakes sampled in the WLS-I because
samples from such lakes are more sensitive to pH
changes due to the partial pressure of C02 (Norton
and Henriksen 1983). The precision estimate for DOC
34
-------
CD
CO
CO
a.
>•
*
2
3
U)
CD
CO
—I
2
in
i
« *
(0
CD
Q.
E
W
—
3
4
"to
3
a
Z
2
CD
il
E
g
•D
2
CO
3
_O
o
^
to
2
CO
E
'•^
M
UJ
c
o
2
'5
£
a.
£
u
CO
QQ
cn
c
o
E
^
15
k
CD
O
CO
J)
p
CO
z
in
z
^
z
u.
CO
Z
u_
a
C CO
o
to
0
ID Q
0_ CO
cc
&v
c
0
^
c fo
^ Q]
a g
c
o
0
Q
c CO
0
o
CD Q
£ to
cc
SR
c
o
'JZ
c g
CD *-
CD C
^ O
C
o
U
Q
c w
0
'to
CD Q
£ CO
or
as
c
o
C 'D
ID t
CD C
5 CD
^ O
o
0
O
C CO
o
!A
'o
CD ._
»- LJ
Q. (O
ff
ss
c
o
Mean
icentratic
o
U
i
Q
CO
Variable"
O-J CO CO CN CT> 00 00 O O5 CO
§o to ^~ o o) O) o CN o o o o o CN o ^* o oo r*^ 10 oo co
o en o o <* o •- •- o o o o o o o o o o o «- co o
OO.-OOOOOOOOOOOOOOOOO OOO
•-cococQp~incnocoeneocnoo*-eocMcD CMIO
01 oo . CM m co in r^ C5 05 in CM CM . CM o ^ . *- «- •- i^ ^
COT- inp~ cocMcocn
CM CO CM »-
co in «- co *- CD CM
§*— oco cop^ CM»— cnp^o*— o*~OP^cn incoco
OCOin*-CM*tO
OOinOO'-OOOOOOOOOOOOOO OOO
^»— incMco^cococMOooinop^cM»— «— CM coco
ocococo»— co»— coocop^incM*— incocop*-*. ^ ^- OOP^
CN^I- i- i- ^ r» r^ in OCNOT
«- «- CO «- •-
§co «— p^ ^
co ^ o co ^ ^ o O CM en p^ p^
COCMOOO--OOO cMcoen
OOCO'— OP^*— »— OOOOOO'— OOOP^P*- P^*— O
^ *~ "~
en p^ in oo co
OO'tOO'-CM^^OOOOOOO^OOO O^co
OOCMOO'-OOOOOOOOOOOOOO OOO
^ CM CM co T- co en •* •- P^ CM p> co en p~ ^J- oo •* o in
T— COOCOCDCOOOCMT— ocM^-cncoo^'T- T— in
COi-r- COCO«-T- P»^-~~ ~ *-
£ ""
1^
in CM ^j- oo O «- CM
cnin o*t CM-- p^p~oocor--^tomococo omoo
OO^-CMOcMOOOOOOOOOOOCMof'l' •* ^ «O
CM CO
1 1
CM CO 00 P^ 00 CM
§o COO>-COCOOOOOCMOOOCOCMP^OOOP> encoi-
O^-COt-OOO3OCNOOOOOOOOOOO O^tcM
0 O CM O O «- 0 ^ O O 0 O O O O O O O 0 0 OOO
TJ *>
«— ooooocncnp^p^comoo^'cocMincMP^p^ co^*
co»— in^coi— cnooT~in^cMencM»— ^ co*- ••- ~ oo co
T— in *— CM CM in cocMen *~
«- r- Tj- CM CM
CMCM mm CMOOOO
o-- ^tco oco coocMOocoT-i-ocoin coi-t
• -Q £— —
*-« c CT ~~ .2 >• ? ^ +2
CJCT CD(Dn3"O .t-rCT —
2-0) « " '-e - "DTO®^
^(CS =; •— r- fO — •— Si_D
i§o "° " o S +,'„ 1 « "'5X «™,
_- T- z | i x o - x" i i — ? o
CO
c
CD
£
CD CO
^. 3 C J2
O CO Q) C
CM ^,^_ jg E CD
II 00 P* r- CD £
^ c CO CO t 1- y,
00 • — ' II II -Q -^ CO 3
UZ-E-E2.2 CUID
r- - -f < S -° E *
-* 1 1 ^ 2 I E
O JO - - C J> Q. I
CDCD-J'-^eDcO °--
citocoigcop, go
3O § CN 2
^5 £ v o> "5 gl
•Xoi^wCD ^CD
(0 — COCOQ— t^.~^
"i-^^wc co^i
EEEEiScD *-ID
ooooJJcj oo
H^ 'f"" -|~ -^~ C (JJ C Q)
"^ ^-^ ^-^ ^— ^ 3 Q_ m J—
co^inco^- r-Q*-™
^^^^'-gco10"
oooo — ^moct/)CD
— i — i — i — i ojo^.t: 3
e - - - E to °0 E 5
"OTJ'OTJ «£^ —
3333CO"*-c5
CO ID CD ID T;, if -t o .2
tOCDCDID**t(]jj: a; c
LU LJ. U_ LJ- < U5 1— Qn
« ja o TJ a »>^
-------
Table 4-4. Overall Within-Batch Precision Estimates Calculated from Field Duplicates8, Western Lake Survey - Phase I
Variable13
Al, extractable
Al, total
ANC (^eq L~1)
Ca
cr
Conductance (fjS cm"1)
DIC, air equilibrated
DIC, initial
DOC
F~, total dissolved
Fe
K
Mg
Mn
Na
NH4 +
NO3"
P, total
pH, acidity (pH units)
pH, alkalinity (pH units)
pH, air equilibrated
(pH units)
Si02
so4-2
n
189
215
215
215
215
215
215
215
215
213
197
215
215
137
215
83'
205
210
215
215
214
215
215
(x > 0)°
Achieved
Precision (%)e
299.7
33.4
4.5
2.3
15.7
6.5
6.9
7.7
16.7
24.7
112.5
9.6
8.1
139.2
7.2
324.7
61.3
77.4
0.08'
0.08'
0.1 7J
16.1
14.5
Quantitation
Limit
(10SB)
0.021
0.085
24.9
0.30
0.21
6.1
0.71
1.32
2.0
0.016
0.07
0.08
0.03
0.10
0.17
0.13
0.342
0.043
k
k
k
2.07
0.56
(x>10SB)C|d
n
11
11
204
208
65
192
173
103
49
90
16
199
212
1
206
3
8
9
k
k
k
119
124
Achieved
Precision (%)f
44.2
10.2
4.1
2.3
8.6
6.4
6.9
2.7
12.7
26.2
24.6
4.5
8.2
h
7.3
7.9
1.1
37.5
k
k
k
7.0
4.3
Intralaboratory
Precision
Goals
1 0/209
1 0/209
10
5
5
2
10
10
5/1 0s
5
10
5
5
10
5
5
10
1 0/209
0.05 (pH
0.05 (pH
0.05 (pH
5
5
unit)
unit)
unit)
a Sampling methods (helicopter and ground crew) and laboratories pooled.
bAII units are reported as mg L~1 unless otherwise indicated.
cx is the mean value of a routine and its duplicate.
dSg is the standard deviation of field blank measurements.
eAchieved precision = root-mean-square of the percent relative standard deviation for all pairs with x > 0.
' Achieved precision = root-mean-square of the percent relative standard deviation for all pairs with x > 10 SB
9 Precision goal based on analyte concentration,
hAchieved precision cannot be calculated because x of only one pair of observations exceeds 10SB.
1 Number of observations is smaller due to low NH4+ concentrations in most WLS-I samples and due to instrumental drift (i.e., con-
centrations of NH4+ < 0).
1 Precision calculated as the root-mean-square of the standard deviation values.
k Quantitation limits are not calculated for pH measurements.
(12.7%) was based on a large number of sample pairs
(n = 49) above the quantitation limit (2.0 mg L~1).
The high variability may be due to the low DOC
concentrations present in the majority of lakes
sampled in the WLS-I (Section 6.2.7.1). Silverstein
et al. (1987b) calculated a precision estimate of 2.4
percent using only duplicate pairs having concen-
trations greater than 5 mg L"1 (n = 9), while precision
calculated using pairs with concentrations between
the quantitation limit and 5 mg L"1 (n = 40) was
14.0 percent.
Overall within-batch precision estimates calculated
from field duplicate pairs indicate how consistently
the field samplers collected the sample, how
consistently the field laboratory personnel processed
the sample, and how consistently the analytical
laboratory personnel analyzed the sample. Determin-
ing within-batch precision on the basis of analyses
performed at the five WLS-I field laboratories
provides a measure of performance for this step in
the sample flow. This precision was calculated using
results from field duplicate pairs and from field
laboratory (trailer) duplicates, which are field
laboratory splits of routine samples. Table 4-5 shows
the precision estimates for closed system pH, closed
system DIG, turbidity, and true color based on all
sample pairs with mean values greater than zero
(x > 0). Quantitation limits could not be calculated
for pH and DIG because no field or trailer blanks
were analyzed for these variables (Section 2.5.1.2).
The pH and DIG measurements were within desired
precision goals except for the pH of field duplicates
which slightly exceeded criteria.
Precision goals for turbidity and true color were met
when precision was calculated using mean values
above the quantitation limit (calculated using trailer
blanks) for trailer duplicates. When values below the
quantitation limit were included in the calculation
(i.e., x > 0), precision goals were not met. Because
most WLS-I lakes sampled had very low values for
turbidity and true color, poor precision was not
unexpected. Based on the results of the field
36
-------
Table 4-5. Overall Field Laboratory Within-Batch Precision Estimates Calculated from Field Duplicate and Trailer Duplicate Data,
Western Lake Survey - Phase I
Precision (Field Duplicates)
Variable
PH
DIC
True Color
Turbidity
Unit
PH
mg L 1
PCU1
NTU'
Intralaboratory
Precision
Goal
(%RSD)a
±0.19
10
±59
10
Achieved Quantitation Limit Achieved
nd Precision (%)e
208 0.129
208 6.8
165 61.3
206 24.4
(10 SB)C nd Precision (%)f
h h h
h h h
32 6 4.7
0.8 37 16.5
Precision (Trailer Duplicates)
Variable
pH
DIC
True Color
Turbidity
Unit
PH
mg L 1
PCU1
NTU'
(x > 0)b
Achieved
nd Precision(%)e
132 0.039
134 3.4
99 25
134 11.8
Quantitation '* '
Limit
(10SB)C nd
h h
h h
15 25
0.35 65
>10SB)b'c
Achieved
Precision (%)f
h
h
3.8
7.8
a%RSD = percent relative standard deviation.
bx = Mean values of a routine and. its duplicate.
°10 SB= 10 times the standard deviation of trailer blanks for trailer duplicates and field blanks for field duplicates
dn = number of duplicate pairs.
eAchieved precision = root-mean-square of the percent relative standard deviation for pairs with x > 0 (for pH, root-mean-
square of the standard deviation).
'Achieved precision = root-mean-square of the percent relative deviation for pairs with x >10 SB.
8 Absolute precision goals in terms of applicable units.
hQuantitation limits cannot be calculated, trailer blanks not analyzed.
| PCU = platinum-cobalt units.
' NTU = nephelometric turbidity unit.
laboratory duplicate data, the overall within-batch
precision across the five WLS-I base sites met the
DQOs.
Intralaboratory within-batch precision (Table 4-6)
was determined using data from analytical laboratory
duplicates. The precision goals for laboratory
duplicates were met for every analyte except Mn.
Concentrations of Mn in WLS-I routine samples were
generally low, and the DQOs do not apply at these
low levels. Therefore, it is unlikely that a sample
with a concentration of Mn high enough to yield
a precision estimate that would meet the DQOs
would be selected for duplicate analysis. Based on
the estimated analytical laboratory precision, the two
analytical laboratories showed excellent
reproducibility.
In conclusion, precision estimates based on field
audit and duplicate sample data indicate that
samplers, sample handlers, and analysts performed
their tasks consistently. No systemmatic precision
problem for any variables across batches, labora-
tories, or methods was observed.
4.4 Effects of Holding Time on Sample
Concentration
As part of the calibration study, one of the three
samples collected by helicopter crews was selected
randomly to be stored at the field laboratory (in the
dark, at 4°C) for as long as 4 days before it was
processed and preserved (Section 2.6). The length
of time that the withheld sample was kept at the
field laboratory depended on the amount of time
required to transport the corresponding ground
samples from the lake site to the field laboratory
(Sections 2.5 and 2.6). This element of the sampling
design was intended to evaluate the impact of
delayed sample processing and preservation that
could result if delivery of the samples collected by
ground crews was delayed. The Forest Service
ground crews, however, were extremely efficient in
providing same-day sample delivery, even from lakes
that were difficult to access. Consequently, the
number of samples for which delivery was delayed
(and thus the number of corresponding withheld
samples) was much smaller than had been antic-
ipated. In fact, about one-half of the holding times
for withheld samples was contrived in order to have
a sufficient number of samples to perform statistical
analyses on the results of the holding time exper-
iment. The number of withheld samples assigned
to each holding time and the analytical laboratories
to which the samples were sent are summarized
in Table 4-7. A more detailed discussion of the
holding time study appears in Silverstem et al.
(1987b).
37
-------
Table 4-6. Overall Within-Batch Precision Estimates Calculated from Analytical Laboratory Routine/Duplicate Sample Data,
Western Lake Survey - Phase I
(x > 0)b
Variable3
Al, extractable
Al, total
ANC (^eq L"1)
Ca
cr
Conductance (^S cm"1)
DIC, air equilibrated
DIC, initial
DOC
F~, total dissolved
Fe
K
Mg
Mn
Na
NH4 +
N03"
P, total
pH, acidity (pH units)
pH, alkalinity (pH units)
pH, air equilibrated
(pH units)
SiO2
so4-2
n
140
148
139
149
149
149
149
149
146
148
144
149
149
120
149
59h
148
137
149
149
149
148
149
Achieved
Precision1*
13.3
3.3
121.6
0.7
2.3
1.1
6.4
4.3
11.1
2.7
56.1
1.4
0.6
105.1
1.1
95.3
8.8
41.7
0.03'
0.03'
0.02'
25.9
1.8
Quantitation
Limit
(10SB)C
0.014
0.029
g
0.04
0.06
3.4
0.33
0.26
0.6
0.008
0.04
0.08
0.02
0.02
0.13
0.07
0.052
0.010
j
j
l
0.28
0.12
(x>10SB)b'c
n
22
64
g
148
145
147
126
131
119
148
63
147
148
45
147
6
118
46
j
i
l
125
143
Achieved
Precision6
4.3
3.3
g
0.7
2.3
1.1
5.8
2.7
4.1
2.7
4.9
1.4
0.6
16.6
1.0
0.9
2.8
7.2
)
i
i
2.9
1.7
Intralaborato
Precision
Goal
1 0/20f
10/20f
10
5
5
2
10
10
5/1 Of
5
10
5
5
10
5
5
10
10/20f
0.05 (pH
0.05 (pH
0.05 (pH
5
5
ry
unit)
unit)
unit)
aAII units are reported as mg L 1 unless otherwise indicated.
bx = mean concentration of a routine and its duplicate.
c10Se =10 times the standard deviation of calibration/reagent blanks.
dAchieved precision = root-mean-square of percent relative standard deviation for pairs with x > 0.
6 Achieved precision = root-mean-square of percent relative standard deviation for pairs with x >
f Precision goal based on analyte concentration.
9Analytical laboratories were not required to analyze calibration/reagent blanks for ANC.
hn is low due to low NH4+ concentrations (i.e., NH4+ < 0) and instrumental drift.
1 Standard deviation values were calculated for pH measurements.
1 Quantitation limits are not calculated for pH measurements.
Table 4-7. Holding Time Samples Processed by Analytical
Laboratories for the Calibration Study,
Western Lake Survey - Phase I
Holding Time
(days)
Oa
1
2
3
4
Lab 1
n
19
1
4
1
3
Lab 2
n
6
2
6
3
0
Total
n
25
3
10
4
3
Total
28
17
45
aZero holding time indicates that samples were processed on
the date the lake was sampled (i.e., within 16 hours of
collection).
For each variable, the difference between the
concentrations of two of the three samples collected
by the helicopter crew was calculated. The two
samples used in this calculation were the withheld
sample and the sample analyzed by the same
laboratory that analyzed the withheld sample. The
third sample in each set was sent to the second
analytical laboratory and was not used in the holding
time calculation. For each variable, the variances of
the differences among holding times were equal (i.e.,
they were homoscedastic). The differences were
analyzed as a function of holding time by using
standard linear regression. The number of days that
the withheld sample was stored before it was
processed and preserved was the independent
variable in each regression. The dependent variable
was the difference between the concentration of the
withheld sample and the concentration of the other
sample analyzed at the same analytical laboratory.
The null hypothesis was that the difference between
samples would not change over time. If holding time
had no significant effect, the slope and intercept of
the regression equation would be zero.
The results of the regression analyses are presented
in Table 4-8. A significant effect of holding time was
demonstrated in only three cases: extractable Al in
samples analyzed at Lab 2 (p < 0.032), air-
equilibrated DIG in samples analyzed at Lab 2 (p <
0.010), and air-equilibrated pH in samples analyzed
38
-------
Table 4-8. Regression Statistics for the Difference Between Routine and Withheld Samples versus Holding Time by Laboratory,
Western Lake Survey - Phase I
Lab 2
Variable (units)
Al, extractable (/ug L"1)
Al.totaKAigL'1)
ANC (fjeq I/1)
Ca+2 (//eq L"1)
CHA/eqL-1)
Conductance (/jS cm"1)
DIG, air equilibrated (mg L' ')
DIG, initial (mg L"1)
DOC(mgL"1)
F~, total dissolved (//eq L"1)
Fe (/jg I'1)
r
0.243
0.033
0.062
0.054
0.085
0.051
0.330
0.082
0.045
0.003
0.005
0.093
0.010
0.001
0.059
0.013
0.055
0.083
0003
0.005
0.005
0.0001
0.090
P(b)
0.032
0.451
0.304
0.335
0.226
0.350
0.010
0.235
0.385
0.819
0.770
0.205
0.685
0.878
0.318
0.642
0.332
0.231
0.830
0.781
0.779
0.964
0.211
Slope
-0.0008
0.0001
0.169
-0.01 1
-0.007
-0.092
-0.016
0.013
-0.009
0.002
0.0004
-0.0004
-0.004
0.0002
-0.005
0.003
0.012
0.001
-0.010
-0.008
-0.044
0.070
-0.053
Lab
Intercept
1.920
-1.528
-1.180
0.050
0.395
0.010
-0.001
-0.01 1
0.053
-0.421
-2.213
0.026
-0.247
-0.309
0.522
-0.039
0.354
-1.436
0.010
-1 .003
0.024
-0.115
0.187
1
r2O
0.102
0.0004
0.002
0.070
0.046
0.106
0.036
0.024
0.013
0.027
0.013
0.001
0.133
0.041
0.069
0.011
0.004
0.008
0.039
0.024
0.349
0.039
0.113
P(b)
0.103
0.912
0.823
0.182
0.281
0.098
0.340
0.443
0.576
0.417
0.566
0.871
0.062
0.312
0.186
0.610
0.745
0.667
0.326
0.440
0.001
0.321
0.086
ar = Proportion of total variance explained by the linear model.
bp = Statistical probability of occurrence using standard linear regression procedure. Values less than 0.05 are considered significant.
at Lab 1 (p < 0.001). For extractable Al, the effect
was most likely due to one exceptionally low
concentration from one sample held for three days.
Because values for all sample pairs were near the
detection limit, no conclusions can be drawn about
the effect of holding time on extractable Al
concentrations. For air-equilibrated DIG, the effect
was likely due to one exceptionally low value for
this variable analyzed in one sample held for three
days. Removal of the results of the analysis on this
sample, and the fact that the initial DIG regression
showed no statistical significance, indicate that
holding time did not affect air-equilibrated DIG
concentrations. For air-equilibrated pH, the effect
was due to relatively large differences in this variable
in three samples held for four days. Two of these
sample pairs had differences of 0.1 pH unit. Because
a difference of 0.1 pH unit is acceptable even for
field duplicate pairs processed at the same time, it
has no practical significance. The third sample pair
showed a difference of 0.5 pH unit; of 45 samples,
this was the only one that showed such a large
difference. This difference probably is attributable
to random error. Therefore, a practical difference in
holding time for this variable cannot be concluded
from calibration study data.
The two chemical variables not included in Table
4-8 are closed-system pH and closed-system DIG,
analyzed at the field laboratories. These analyses
were not performed in the analytical laboratories and
the number of samples from calibration lakes
analyzed in each field laboratory with actual holding
times (i.e., those samples not analyzed on the day
they were collected) was insufficient to perform
reliable statistical analyses. Therefore, a regression
analysis was done on each of these variables in the
same manner as for the analytical values except that
all the calibration lake sample pairs were pooled.
For closed-system DIG, the slope was 0.005, the
intercept was 0.024, the value of r2 was 0.003, and
the value of p was 0.236. For closed-system pH, the
slope was -0.020, the intercept was 0.032, the value
of r2 was 0.032, and the value of p was 0.236. The
results of this analysis indicate that there was no
significant effect of holding time on these field
laboratory analyses.
In general, analyses of the calibration study samples
showed no significant effects of sample holding time
on analyte concentration. Possible effects of holding
time, however, were not adequately tested for
samples with low analyte concentration (i.e.,
extractable Al and NH4+) because reported values
for these samples were near the detection limits.
39
-------
4.5 Assessment of the Effect of
Different Field Sampling Protocols
Use of differing protocols involves the risk of bias,
with the result that data generated under one
protocol may not be comparable to data generated
under another. For this reason, the field laboratory
protocols of the survey were rigidly standardized,
with holding times and handling procedures
narrowly prescribed. The necessity to alter the field
protocol for wilderness areas surveyed in the WLS-I,
by use of ground crews to collect samples, caused
some concern about the effect of the associated
changes in protocol, particularly sample handling
and holding times. Differences in the protocols for
ground and helicopter crews are discussed in Section
2.5.1.
A study was designed to assess the differences in
the two protocols and to minimize any resultant
effects of the change in protocol. This study required
a random subset of the wilderness sample lakes to
be visited by both ground and helicopter crews. The
purposes of the study were to identify (1) those
variables for which the values obtained by the
alternate ground crew method could be used
interchangeably with values obtained by the
standard helicopter crew method, (2) those variables
for which the helicopter values could be predicted
accurately from the ground values, and (3) those
variables for which the ground crew method
produced results that were unacceptable for use in
data analysis. The second aspect of the approach
involved calibrating the ground crew results with
those from the helicopter crews. Used in this
context,calibration is simply adjustment of the
ground crew values using a linear regression model
to make them comparable to data collected by the
helicopter crews.
Of the 50 lakes selected for this calibration study,
45 were actually sampled in duplicate by both ground
crews and helicopter crews. Visits were scheduled
for the same day, but in order to ensure data
comparability, ground crews were not told which of
the lakes also were to be sampled by helicopter
crews. The identity of the calibration lakes inadvert-
ently was made known to the field crews in one
subregion before sampling began. The sample for
this subregion was redrawn, and, in order not to
indicate the newly selected lakes, duplicates were
not reassigned for the ground crews. For this reason,
several sample lakes do not have duplicate ground
sample values. Additionally, some data were
removed from the analysis as a result of data
validation (Section 3.3).
In addition to the use of two methods of field sample
collection, the survey also employed two analytical
laboratories. The duplicate samples collected for the
calibration study provided evidence of the repeat-
ability of the analytical measurements, which is
expressed as measurement error. One sample of
each duplicate pair was sent to each of the two
analytical laboratories contracted for the survey.
Additionally, information regarding laboratory bias
was provided by audit samples (Section 4.3) and field
laboratory splits.
4.5.1 Calibration of the Ground Sample Data
The objective of the calibration study was to identify
those variables for which (1) the two field sampling
protocols generate interchangeable results, (2) those
for which the helicopter values are different, but are
predictable from the ground crew values, and (3)
those for which such prediction involves unaccepta-
ble error, so that the ground crew data must be
considered unusable in generating population
estimates. Details of this analysis are provided in
the Appendix.
Analysis began by plotting the helicopter value (y)
againstthegroundvalue(x; Figures4-1 to4-3). Lakes
with duplicates for each variable are represented by
a box, with the four possible pairings of ground data
with helicopter data forming each corner. Missing
values result in missing sides of some boxes. When
the two protocols yield identical values for the
variables, the results appear as points on the one-
to-one, diagonal line.
Based on inspection of these plots, a linear model
of the form y = bx + e was used to compare the
sampling methods. The model results (Table 4-9)
indicated that the differences between samples
collected by helicopter and by ground crews were
small, a conclusion that was also supported by the
plots of the paired data.
The slopes of the regression between helicopter and
ground samples for the primary variables were close
to one, although statistically significant differences
were observed for S04 2, pH, and DOC. However,
calibration of the ground data using the helicopter
data provided little improvement for SO«~2 and DOC.
This finding also was evident when the relative
variances (total variance/population variance) of
helicopter data were compared to these for the
calibrated data (Table 4-9).
For extractable Al and pH, differences were apparent
between the samples collected by ground and by
helicopter crews. However, calibration of these
variables, although statistically defensible, is not
warranted considering the nature of the data. For
pH, maximum bias caused by use of ground crew
samples is decreased only 0.001 unit (from 0.007
to 0.006) by use of the calibration model. This change
in hydrogen ion concentration would be small in
40
-------
Figure 4-1. Results of ANC (fjeq L"1) measurements and pH determination on air-equilibrated samples on duplicate samples
collected by helicopter and ground crews from 45 lakes for the calibration study during the Western Lake
Survey - Phase I. The dimensions of the boxes indicate the magnitude of sampling and measurement error;
sampling bias is evident when the boxes consistently deviate from the 1:1 (diagonal) line.
650
-j
o-
.
-------
Figure 4-2. Results of sulfate (fieq £"V and calcium (/jeq L'^) analyses on samples collected by helicopter and ground crews
from 45 lakes for the calibration study during the Western Lake Survey - Phase I. The dimensions of the boxes
indicate the magnitude of sampling and measurement error; sampling bias is evident when the boxes consistently
deviate from the 1:1 (diagonal) line.
140
0)
!
70
140
Ground Sample (fjeq L ^j
600
-S
% 300
&
Ca
300
Ground Sample faeq i"
600
42
-------
Figure 4-3. Results of extractable aluminum ffjg £~V and DOC (mg L V measurements on duplicate samples collected by
helicopter and ground crews from 45 lakes for the calibration study during the Western Lake Survey - Phase I.
The dimensions of the boxes indicate the magnitude of sampling and measurement error; sampling bias is evident
when the boxes consistently deviate from the 1:1 (diagonal) line.
I
0)
I
10
*
01
I
.o
1
20
10 '
10 20 30
Ground Sample (/jg L~
6.5
I
5
325
DOC
3.25
Ground Sample (mg L'
6.5
43
-------
Table 4-9. Results of Regression of Values for Primary Variables Measured on Samples Collected by Helicopter Crews
(Dependent) Against Samples Collected by Ground Crews for the Calibration Study Lakes, Western Lake Survey -
Phase I"
Relative Variance15
Primary
Variable
ANC
pH (closed system)
Ca + 2
so4-2
DOC
Al, extractable
Slope
0.99961
0.99313d
1.00261
1.01314d
0.97454d
1.05926
r2
0.9991
0.9997
0.9982
0.9979
0.9912
0.5889
Helicopter
Protocol
1.00108
1.00976
1.00156
1.00290
1.02431
1.09462
Calibrated
Ground
Protocol
1.00123
1.06918
1.00111
1 .00390
1 .02909
1.46587
Direct Use of
Ground Protocol0
e
1.07856
e
1.00395
1.03010
e
a Regressions, based on a sample size of 45, were forced through zero and thus no intercept term is shown.
bRelative variance is defined as total variance/population variance where total variance is the sum of population variance, measure-
ment error variance, and prediction variance.
0Shown only for those variables where the relative variance resulting from calibration is less than direct use of ground protocol.
d Significantly different from 1.00 (at p<0.05).
eNot possible to compute.
acidic lakes, and would be unnoticeable at the high
pH values encountered in the western lakes.
Although the slope for extractable Al showed no
significant bias between the two protocols, extract-
able Al based on results from the ground crew
samples were much more variable than those based
on helicopter crew samples, with relative variances
of 1.46587 and 1.09462, respectively. This increased
variance between the two protocols would normally
indicate the need for calibration. However, most of
the extractable Al concentrations were below the
detection limit, and those concentrations above the
detection limit were far below any level of concern
for biological effects associated with aluminum
toxicity.
Consequently, samples collected by ground crews
were judged acceptable for all variables and no
calibration or replacement was conducted. Based on
the result of this analysis, the results generated from
the ground crew sampling protocol were combined
with those generated from the helicopter sampling
protocol. All population estimates presented in this
report are based on the combined data.
4.5.2 Relative Bias Between Analytical
Laboratories
The constraints of processing a large number of
samples within the prescribed holding times for the
measured chemical variables necessitated the use
of two analytical laboratories in the WLS-I. To
evaluate systematic differences in performance
between laboratories, three groups of samples were
collected: (1) synthetic and natural audits, (2)
helicopter and ground duplicate samples from the
calibration study lakes, and (3) field laboratory splits
of all samples analyzed by EPA's Environmental
Research Laboratory at Corvallis. These three types
of samples are referred to as audits, duplicates, and
splits, respectively. Methods of sample collection and
processing are described in Sections 2.5 and 2.6.
Analysis of each type of these QA samples provided
the opportunity to examine relative bias between the
two analytical laboratories (Lab 1 and Lab 2).
Results of analyses on audits were analyzed
differently from those derived from analyses on
duplicates and splits because the means of six audit
lots constitute the "sample." Estimates of variance
of each datum provide the basis for weighting each
of these samples, using a maximum likelihood
approach (Permutt and Moezzi 1986). Data for the
duplicates and splits were analyzed as individual
records because these data are not amenable to
maximum likelihood analysis. The duplicate and split
samples also differ from one another because the
duplicates constitute pairs of samples from the same
lakes sent to both laboratories; the split samples also
were sent to a third analytical laboratory so that EMSI
and Versar do not appear together in the same
sample pair.
Despite the differences among the three types of
QA samples, relative bias between the analytical
laboratories was assessed using a weighted least
squares model, with some modification. First, the
conventional regression analysis (shown as Model
2) can be modified as shown in Models 0 and 1
where:
Model 0: y = a + x + e (b=1)
Model 1: y = bx + e (a = 0)
Model 2: y = a + bx + e
These three models were used to describe the
differences between laboratories where Model 1 was
the model of first choice if more than one model
satisfactorily described the relationship. The
parameter estimates for the audit samples using the
least squares approach were very similar to those
44
-------
Table 4-10. Estimates of Bias between Analytical Laboratories Based on Synthetic and Natural Audit Samples (Audits),
Paired Duplicate Samples from the Calibration Study (Duplicates), and Split Samples (Splits) from the
Environmental Research Laboratory-Corvallis. The Model Results are Shown Using Lab 2 as the Dependent
Variable and Lab 1 as the Independent Variable
Variable
Primary
ANC
Ca+2
so4-2
DOC
Al, extractable
Secondary
Mg + 2
Na +
K +
NH/
N03-
cr
F-
DIC, initial
DIG, air equil.
pH, air equil.
pH, acidity
pH, alkalinity
Al, total
Fe
Mn
P, total
Si02
Conductance
Unit
#eq L~1
Aieq L~1
/ueq L'1
mgL"1
/ugL'1
/ueq L-'
/ueq I"1
,ueq L"1
yueqL'1
yueq L"1
/ueq L"1
Veq L'1
mgL-1
mgL"1
pH units
pH units
pH units
//gL-1
MJL'1
«JL-1
A/gL'1
mgL-1
,uS cm'1
Audits'1
NS
(0)
(0)
0.17
-1.9
(0)
NS
0.67
0.55
(0)
NS
(0)
NS
NS
(0)
(0)
NS
5.9
-8.3
NS
-1.1
(0)
(0)
Intercept3
Duplicates0
-5.9
(0)
(0)
0.21
NS
-0.46
NS
NS
-0.414
0.36
NS
(0)
(0)
-0.11
1.34
-0.1317
-0.90
NS
-3.54
3.74
2.39
NS
NS
Splits0
NA
(0)
(0)
NA
NA
NS
NS
NS
NA
NA
NA
NA
NA
NA
NA
NA
NA
7.12
NS
1.72
NA
NS
NA
Audits'
NS
1.054
0.945
0.948
0.68
1.016
NS
0.925
0.16
0.959
NS
1.048
NS
NS
0.9952
1.0123
NS
0
(1.0)
1.041
NS
1 .0984
0.989
Slopeb
Duplicates0
1.055
1.073
0.970
0.853
1.272
1.056
NS
NS
(1.0)
0.788
NS
0.913
1.050
1.085
0.807
(1.0)
(1.0)
0.721
0.840
0.133
0.589
NS
NS
Splits0
NA
1.077
NS
NA
NA
NS
NS
0.949
NA
NA
NS
NA
NA
NA
NA
NA
NA
0.904
NS
0.850
NA
NS
NA
a Intercepts shown as (0) indicate that Model 1 was used in the analysis.
bSlopes shown as (1.0) indicate that Model 0 was used in the analysis.
CNA = not analyzed, NS = not significant.
derived from the maximum likelihood method in
Permutt and Moezzi (1986).
The results of the analysis of analytical laboratory
bias using the audits, duplicates, and splits are
shown in Table 4-10. Cases evaluated using Model
0 are indicated with slopes (b) set to 1.0. Those cases
analyzed with Model 1 are designated with the
intercept (a) set to 0. Those variables with both
intercept and slope significantly different from 0 and
1, respectively were fitted with Model 2. All analytical
variables, except sodium and chloride, exhibited at
least one significant regression parameter for the
QA comparisons between laboratories. Estimates of
bias in the intercept term are generally inconsistent
among the three types of QA samples. Also, the
magnitude of the estimates of intercept bias are
relatively small. Several exceptions are DOC, air-
equilibrated DIC, pH (acidity) and Fe. In all of these
cases, the magnitude of the bias term for the
intercept, although statistically significant, is
unimportant with respect to potential effects on data
interpretation. In contrast, the estimate of bias for
air-equilibrated pH of 1.34 units based on analysis
of the duplicates is large. However, the analysis of
the audit samples indicate no problems with bias
for this variable and removing two duplicate samples
with aberrantly high deviation (Figure 4-4) produces
results consistent with the analysis of the audits.
Estimates of slope show consistent bias for Ca+2,
S04~2, DOC, N03", DIC and air-equilibrated pH.
Differences in ANC are also evident between the
laboratories (Figure 4-4), but at concentrations less
than 100 /ueq L"1, these differences were very small.
At higher values of ANC, the differences observed
between laboratories were generally smaller than
the difference in split samples measured within a
single laboratory.
The maximum difference observed between the two
laboratories for air-equilibrated pH measurements
was approximately 0.4 pH unit. Possible sources of
error in this pH measurement include: (1) incomplete
equilibration with 300 ppm COa in air, (2) electrode
errors, or (3) differences in measurement procedures
between the laboratories. Many of the lakes sampled
in the calibration study were very dilute, and the
pH is sensitive to changes in partial pressure of COz.
Small differences in the COa gas used at each
laboratory in equilibration, the equilibration times,
or measurement procedure, coupled with electrode
45
-------
Figure 4-4. Results of ANC f/jeq L'^) and pH determinations on air-equilibrated samples performed by Lab 1 and Lab 2.
Analyses were conducted on duplicate samples collected by helicopter and ground crews from 45 lakes for
the calibration study during the Western Lake Survey - Phase I. The dimensions of the boxes indicate the magnitude
of sampling and measurement error; laboratory bias is evident when the boxes consistently deviate from the
1:1 (diagonal) line.
700
560
•8
420
250
140
740
2SO
420
560
700
5.5
6.1
6.7 7.3
Air-Equilibrated pH - Lab 2
7.9
8.5
46
-------
errors, could produce the observed differences in the
air-equilibrated pH values. The air-equilibrated pH
measurement was not used in deriving population
estimates, and thus no corrective action was taken.
For Ca+2, the three parameter estimates of slope
show that Lab 1 measurements are between 4 and
8 percent greater than those from Lab 2 (Figure 4-5).
If Model 1 is used for Ca+2 on the audit samples,
the estimate for bias in the slope approaches that
observed for the duplicates and splits. Differences
in calcium can be attributed to different analytical
protocols that were allowed under the EPA contract;
Lab 2 used atomic absorption spectroscopy and Lab
1 used inductively coupled plasma emission
spectroscopy.
Differences between the results of analyses for the
following variables cannot be attributed to differen-
ces in instrumentation. For S04 2, the Lab 1
measurements appear to be from 2 to 5.5 percent
less than those from Lab 2. Nitrate analyzed by Lab
1 also appears to be somewhat lower than Lab 2
measurements (Table 4-10). The analytical labora-
tory measurements of DIG show that Lab 1 values
are greater than those from Lab 2 by approximately
5to 13 percent. Extractable aluminum results (Figure
4-6) illustrate the high degree of variability typically
observed for variables with most of the values near
the detection limit. Dissolved organic carbon, in
contrast, shows a much higher degree of precision
that allows differences between laboratories to be
detected (Figure 4-6).
Given the presence of statistically significant
differences between analytical laboratories, it would
be desirable to adjust for measured differences.
However, deriving an appropriate adjustment for
analytical bias involves a high degree of uncertainty.
The results from the three types of QA samples show
considerable lack of agreement in the estimated
magnitude of the relative bias. An additional problem
in adjusting for analytical bias is that at least nine
variables show a significant lack of fit with the linear
models, as illustrated in the analysis of the audits
(Permutt and Moezzi 1986). As shown in the
calibration study (Appendix), correction for
differences in protocol also introduces an additional
source of error.
The audit, duplicate, and split sample results
presented here to evaluate relative bias between
laboratories cannot be used to judge how well the
survey results met the DQOs for absolute accuracy.
Determination of absolute accuracy requires that the
results be compared to certified standards of known
concentrations, e.g., from the National Bureau of
Standards. Because the use of certified standards
was beyond the scope of the QA program for the
WLS-I, such a comparison is not possible. Never-
theless, it is possible to show that the relative bias
between laboratories for most variables is either not
statistically significant, or not meaningful in the
context of the survey objectives. However, because
it is difficult to establish absolute accuracy for an
analyte, correcting for the observed relative bias in
this survey requires more information than is
currently available. Consequently, the data pre-
sented in this report have not been adjusted for
relative bias.
4.6 Assessment of Data Quality
Data quality is a measure or description of the types
and magnitude of error associated with a data set.
The quality of a data set can be expressed in terms
of five characteristics: precision, accuracy, represen-
tativeness, completeness, and comparability.
The QA data results presented in Section 4 allow
the data user to evaluate how well the characteristics
of each sample are represented by the data
associated with the lake sample collected in the
WLS-I. To make this determination, the QA data are
compared to the data quality objectives (DQO)
established before the survey was initiated (Section
2.1.1).
Of the issues that can be addressed by comparing
WLS-I data to the initial DQOs, precision may provide
the most extensive information. Precision was
calculated by using two types of field audit samples
and by using duplicate samples (Section 3.5.2.2). The
use of the two types of audit samples is an effective
means of determining analytical precision. Water
samples collected from lakes (natural audits) can be
used to estimate how well the variables within a
natural matrix are repeatedly analyzed by the same
laboratory and by different laboratories. Water
samples prepared from a matrix of known analytes
and deionized water by a reference laboratory can
be used to estimate the precision with which the
analytical laboratory measures the analytes known
to be present in the sample. The overall among-batch
precision estimated on the basis of more than 200
of these field audits was excellent both within and
between the WLS-I analytical laboratories.
Overall within-batch precision estimated from
duplicate samples is another means of determining
confidence in sample values. For the WLS-I,
duplicate sample values were compared from the
time the lake was sampled until the analyte
concentrations were reported by the analytical
laboratory. For the WLS-I duplicate samples, when
the quantitation limit is used as a criterion to
eliminate pairs that are of low concentration, within-
batch precision can be shown to be well within the
acceptable limits established for sampling, process-
47
-------
Figure 4-5. Results of suit ate (fjeq L V and calcium (fjeq I"1 measurements by Lab 1 and Lab 2. Analyses were conducted
on duplicate samples collected by helicopter and ground crews from 45 lakes for the calibration study during
the Western Lake Survey - Phase I. The dimensions of the boxes indicate the magnitude of sampling and
measurement error; laboratory bias is evident when the boxes consistently deviate from the 1:1 {diagonal) line.
•8
1
-------
Figure 4-6. Results of extractab/e aluminum (fjg i"V and DOC (mg L 1^ measurements performed by Lab 1 and Lab 2. Analyses
were conducted on duplicate samples collected by helicopter and ground crews from 45 lakes for the calibration
study during the Western Lake Survey - Phase I. The dimensions of the boxes indicate the magnitude of sampling
and measurement error; laboratory bias is evident when the boxes consistently deviate from the 1:1 (diagonal)
line.
50
40
/O 20 30
AI-Ext. (fjg L-') - Lab 2
40
50
6.5
•Q
10
-J
-J
I
O
Q
325 -•
0.0
3.25
DOC (mg L'') - Lab 2
6.5
49
-------
ing, and analytical error. For field duplicates, most
analytes showed precision within or near the desired
limits. Those analytes with precision slightly outside
the acceptable limits should not have a large effect
on overall population estimates, but their impact on
interpretation must be considered by the data user.
Values for all other analytes, based on trailer
duplicates and laboratory duplicates, indicate
consistency in all steps from sample collection at
the lake site to preparation of the verified and
validated data sets.
Accuracy is sometimes defined as the combined
components of precision and bias (Kirchmer 1983);
but, for the purpose of assessing data quality it is
used here synonymously with the term "bias." Bias
can be introduced through systematic contamination
of the samples, through differences in sample
collection, or through differences in treatment and
analysis of the samples.
Except for random error, the samples analyzed during
the WLS-I were minimally influenced by contam-
ination. Most of the field blank values were
sufficiently near the detection limits to conclude that
no systematic contamination in field sampling,
preparation of sample aliquots, sample processing
and transfer, or sample analysis occurred. Analytical
results from trailer blanks, laboratory calibration and
reagent blanks also indicate that there was no
systematic contamination in the field laboratory or
in the analytical laboratory. As a result, population
estimates are not influenced by contamination of the
samples.
However, bias was introduced in the data through
differences in helicopter and ground access to the
lakes and through the use of two analytical
laboratories. Bias resulting from differing field
sampling protocols was found to be statistically
insignificant for most variables, and marginally
significant for the remaining variables. Bias intro-
duced by the laboratories was observed for most
variables, but absence of an unambiguous standard
prohibited calibrating the data to minimize analytical
laboratory bias. However, the accuracy of the
measurements of cations and anions can be checked
by comparing the calculated equivalent conductance
of the sums of ions with the measured conductance.
The measured conductance values between the two
laboratories agreed very well. The only significant
difference observed among the three types of QA
samples (Table 4-10) was a slope of 0.989 for the
audit samples. However, using a least-squares
regression through the origin, the calculated
conductance values from Lab 1 on those from Lab
2 yields a significant slope of 1.049. This slope is
almost identical to that obtained when regressing
the sum of cations (1.055) or the sum of anions
(1.048) for Lab 1 on Lab 2. This indicates that Lab 1
is consistently overestimating anions and cations by
approximately 5 percent (within the range of samples
in the calibration study).
A sensitivity analysis was conducted to determine
the impact of adjusting calcium for bias, the
parameter showing the greatest degree of bias.
Population estimates for the percent of lakes in each
subregion with calcium < 50 /ueq L"1 changed less
than one percent when a bias correction was applied
to the data by subtracting 10 percent from Lab 1
values. Thus, the impact of analytical bias on an
objective of the survey, namely characterizing the
chemistry of lakes, appears minimal.
Other analyses, such as those presented in Section
6, are potentially more sensitive to misinterpretation
caused by analytical bias. Again, a "worst-case"
sensitivity analysis was conducted by subtracting 10
percent from the calcium values from Lab 1, the
laboratory that analyzed the samples from the
Central (4D) and Southern Rockies (4E). A linear
regression of ANC on the sum of [Ca+2 + Mg+2]
showed the parameter estimates for these two
subregions to exhibit minor changes in intercept
between the adjusted and unadjusted data, but more
appreciable differences in the slopes (Table 4-11).
Although these differences in slope represent a
measurable change in the relationship between ANC
and [Ca+2 + Mg ], they have little impact on the
interpretation of these particular analyses. Users of
the data should be cognizant of the possible
limitations in the data attributed to differences
between laboratories.
Table 4-11. Comparison of Model Results of ANC on
[Ca+2 + Mg+2] for the Central and Southern
Rockies Using Data with Calcium Adjusted
and Unadjusted for Possible Analytical Bias9,
Western Lake Survey - Phase I
Subregion
Central Rockies (4D)
bias adjusted
unadjusted
Southern Rockies (4E)
bias adjusted
unadjusted
Intercept
-8.9
-9.9
-16.0
-11.4
Slope
0.979
0.935
1.085
1.007
aCa + 2 was adjusted by reducing the reported values from Lab 1
by 10 percent. A linear regression model weighted for the
inverse sampling probabilities was used
Representativeness is the degree to which sample
data accurately and precisely reflect the character-
istics of a population, i.e., how well the WLS-I
samples characterize the lakes in the region (Section
2.1.1). Representativeness has several aspects
including:
50
-------
1. how well a single water sample collected at
1.5 m over the deep portion characterizes a
lake (spatial representativeness);
2. the degree to which a sample on one day
characterizes the nature of that lake (temporal
representativeness); and
3. the degree to which the sample of lakes
represents the population of lakes (sampling
representativeness).
The assumption of spatial representativeness is that
the water mass in a lake is chemically homogeneous,
a condition most likely to exist when lakes are non-
stratified, or mixed. Nearly all lakes were confirmed
to be nonstratified through the measurement of
temperature profiles.
Sampling representativeness cannot be known
exactly without conducting a complete census of the
target population. However, the sampling scheme
was designed to maximize the representativeness.
A systematic random sample drawn within strata
ensured good geographical coverage without bias.
Completeness is a necessary part of evaluating
population estimates because valid statistical
evaluations cannot be performed unless a sufficient
number of samples are collected and analyzed. A
goal of the survey was to sample approximately 50
lakes per stratum to allow population estimates to
be computed with a high degree of precision. Each
of the five subregions had three strata for a desired
total target of approximately 750 lakes for the
probability sample. However, only 720 probability
sample lakes were sampled; five strata where the
achieved sample size was less than 90 percent of
the sampling goal were 4A3, 4C3, 4D1, 4D3, and
4E3 (Table 5-5). Only in Subregion 4D were two of
the three strata inadequately sampled; this lack of
completeness caused concern regarding the popu-
lation estimates in Subregion 4D.
the calibration study and other QA samples failed
to identify any serious problems in comparability
among field crews or field laboratories. The subject
of comparability between laboratories was discussed
earlier in Section 4.6.
The evaluation of the quality of the WLS-I data is
dependent upon its intended use; therefore, the
quality was assessed from the perspective of the
objectives stated in Section 1. For the purposes of
preparing population estimates and characterizing
the nature of these lakes, the quality of the data
is judged to be acceptable. Other users of the data
may wish to reevaluate the quality of the data
consistent with its intended use and their objectives.
Comparability refers to the similarity of data from
different sources included in a single data set.
Different sources used to collect and generate the
data for the survey include 68 field crews used to
collect the samples and make in situ measurements,
five field trailers used to make field laboratory
measurements, and two analytical laboratories used
to analyze the samples (not including split samples
sent to ERL-Corvallis). Because of the large number
of field crews used to collect the samples, it was
not practical to provide a sufficient number of QA
samples to quantify differences among crews.
However, the common training and equipment
provided to all crew members served to minimize
this source of variability. Available evidence from
51
-------
Section 5
Results of Population Estimates
5.1 Data Presentation and
Considerations
5.1.1 Presentation
In the interest of condensing the results, most tables
show data for regions and subregions, omitting the
results of the alkalinity map class strata (Section
2.2.1). Summaries of subregional population
estimates are provided for California (4A), the Pacific
Northwest (4B), the Northern Rockies (4C), the
Central Rockies (4D), and the Southern Rockies (4E).
Only the primary variables, acid neutralizing
capacity, pH, sulfate, calcium, extractable aluminum,
and dissolved organic carbon, are summarized by
extensive tables in this section. Of the variables
measured, these were selected because of their
direct relevance to issues of acidic deposition effects.
Nitrate may also be important, but it was present
in low concentrations. Lake population statistics are
also provided for estimates of the physical attributes
of the lakes sampled, as well as for secondary
variables of interest. Summary statistics for all
measured variables are presented in Volume II of
this report.
The descriptions of each population were summar-
ized on a single page of output for each variable.
The cumulative distributions, F(x) and G(x), supple-
mented by descriptive statistics, serve as the primary
data outputs and were used to generate the tabular
information presented in this report. An example of
these outputs is shown in Figure 5-1 and Volume
II includes these outputs for other variables and
subpopulations.
Each page of data output (Figure 5-1) is identified
at the top by the name of the variable (chemical or
physical parameter) and by the definition of the
subset of lakes being examined. A subset can be
defined on the basis of any physical or chemical
variable or design stratification factor (region,
subregion or alkalinity map class), either singly or
in combination.
The basic population estimates, N and A (denoted
only as N and A on the output), and the standard
errors of these estimates are presented (Figure 5-1)
in the heading of the output. N is defined as the
estimated total number of lakes in the population
being described, and A is the total amount of lake
area (ha) for all lakes in that population. The indicated
sample size (n***. Section 2.3.2) is the number of
sampled lakes in the target population being
described. The identity of reference values (Xc1, Xca,
XC3> chosen for comparative analyses of that variable
completes the heading material on each page of
output.
The distribution curves, F(x) and G(x), are similar in
interpretation, representing estimated frequency
and areal distributions, respectively. For any value
x of variable X, F(x) is interpreted as the estimated
proportion of lakes having^ a value of X < x. F(x) is
calculated as the ratio of Nx, the estimated number
of lakes having a value of X < x, to N. The upper
95 percent ^confidence limit on NX, also scaled by
dividing by N, is shown as a dashed line (Figure 5-1).
Although only a one-sided upper 95 percent
confidence limit is shown, a two-sided 90 percent
limit can be made by using the existing limit and
one equidistant below the curve. This procedure
would also provide a lower 95 percent one-sided
limit.
The function G(x) is the estimated proportion of lake
area in the population having a value of X < x, for
any variable X. This cumulative areal curve and its
confidence curve are determined analogously to the
F(x) curves.
An example of interpreting these distributions is
derived from Figure 5-1. At the value of ANC = 200
/ueq L~1, F(200) = 0.67 (calculated and shown on the
figure as pca = 0.666); the upper confidence limit
is read from the dashed line as 0.70. These values
are multiplied by the estimated total number of lakes
in the population, 10393, to yield 6859 (calculated
and shown on the figure as Nc3 - 6926) as the
estimated number of lakes in the population having
ANC < 200 //eq L~1, and an upper confidence limit
for that number, 7275 (calculated and shown on the
figure as NCU3 = 7276). The calculated values and
those derived from the figure differ due to the
imprecision of reading the proportions (0.66 and
0.70) from the curves. Reading values from the
52
-------
Figure 5-1. F(x) and Gfx) distributions f •
•) and the 95 percent upper confidence limits (
) on these distributions
for ANC (fjeq L ) for the target population of lakes f > 1 ha and < 2000 ha) sampled in Region 4 (West).
Western Lake Survey - Phase I. The population statistics, quintiles (Q, - Q« ) and medians, and three reference
values (Xci. Xca, and XC3 ) for ANC are also shown. The proportions (pc or
-------
curves is suitable only for obtaining rough estimates,
due to the potential error that can be introduced.
Precise computation requires the use of the data
base and the appropriate algorithms.
Plots of F(x) and G(x) can be used only to estimate
confidence limits on the number or area of lakes
below a reference value of X. When the reference
value for variable X might best be discussed in terms
of the population greater than x, the plots are
presented as complementary distributions, 1-F(x)
and 1 -G(x), which can be computed for any variable
of interest. Examples of these distributions are found
in Section 5.4.3 for sulfate and in Volume II.
Associated with each distribution shown on the
output are the following descriptive statistics:
• Mean and standard deviation, appropriately
weighted to reflect the estimated distribution, F(x).
There are no comparable statistics for G(x).
• Min and Max are the minimum and maximum
values of variable X observed in samples for the
population or subpopulation being examined.
• The median is the estimated value of variable X
such that half the number of lakes, or half the
lake area, in the population is at or below this
value.
• The quintiles (Qi, Q2, Q3, Q4, Qs) are the values
of X that partition the distribution into five equal
parts (e.g., the 4th quintile is the 80th percentile;
Q5 is the maximum value). These values are useful
when comparing differences in F(x) or G(x)
distributions among regions, subregions, or
subpopulations. The interquintile difference (Q Xc).
54
5.1.2 Design Considerations
5.1.2.1 Using Weights—The design of the WLS-I
requires that the results be presented as population
and/or subpopulation estimates whenever conclu-
sions on combined strata are to be drawn (Section
2.3). Expansion factors or weights (W) must be used
when making combined strata estimates of attributes
for the populations of lakes (Linthurst et al. 1986).
These weights are defined, and the estimating
equations are given, in Section 2.3.
Using Strata 4A1 and 4A2 illustrates the require-
ment that all unweighted estimates be made within
strata and that means or other statistics involving
more than one stratum be calculated with the
appropriate stratum weights (Table 5-1). The correct
way to estimate the total number of lakes in two
strata below a reference value (in this example ANC
< 50 //eq L"1) is to determine first the total number
of lakes in the sample below the reference value
in each stratum (nc). The next step is to determine
the proportion of lakes in the sample below the
reference value for each stratum (nc/n***: 23/54
= 0.426 and 12/53 = 0.226). Next, multiply the
proportion of sample lakes below the reference value
in the stratum by the estimated number of lakes in
the stratum population (N), which results in Nc, the
estimated number of lakes in the stratum population
below the reference value. Adding the Nc for each
stratum (735.5 + 101.1) yields the combined stratum
Nc (836.6). The same answer can be obtained by
multiplying nc by W for each stratum and summing
the results.
The most accurate estimate for the overall proportion
of lakes in the designated population below the
reference value, therefore, is 836.6/2173 = 0.384
(Table 5-1). If the overall proportion of lakes below
Table 5-1. Use of Weights in Combined Strata Estimation,
Western Lake Survey - Phase I
ANC < 50 Meq L"1
Stratum
4A1
4A2
Combined
N
1727
446
2173
n***
54
53
107
W
31.978
8.422
nc
23
12
35
Pc
0.426
0.226
0.384
Nc
735.5
101.1
836.6
N = estimated number of lakes within an alkalinity map
class stratum.
n*** = number of lakes from which samples were obtained.
W = weighting or expansion factor.
nc = number of lakes in the probability sample with ANC
< 50 ^eq L~1, the reference value.
pc = estimated proportion of lakes in the sample (for a
stratum) or population (for combined strata) which has
ANC < 50 /jeq L~1 (nc/n***).
Nc = estimated number of lakes in the population which has
ANC < 50 f^eq L~1, the reference value.
-------
the reference value were computed as 35/107 =
0.327 (nc/n*** for the sum of nc and n*** for both
strata), the answer would be biased. For example,
there is an estimated total of 2173 lakes in Strata
4A1 and 4A2. Using the correct value of 0.384 as
PC, the estimated number of lakes with ANC < 50
/jeq LT1 would be 837. Using the incorrect pc value
of 0.327 (based on the combined nc/n***), the
estimated number of lakes with ANC < 50 peq L"1
would be 711. Therefore, the number of lakes
estimated to have ANC < 50 /ueq L"1 in both strata
would be underestimated by 126 (837-711).
A less clear issue associated with the design and
weighting is related to examining relationships
among variables. Unweighted analyses such as
regressions or correlations should not be used unless
the relationships between the variables are the same
across strata. Unless the relationships are independ-
ent of alkalinity map class (and any factor associated
with the alkalinity map class strata) unweighted
estimates can be biased, just as unweighted means
or medians and total numbers can be.
In this report, the estimated statistics for regression
and correlation analyses and their associated
standard errors are presented by stratum; thus, they
are unweighted. Analyses which combine strata
(e.g., on a subregional level) are weighted and, as
for strata, the regression statistics are unbiased.
However, the standard errors associated with these
combined estimates are biased and, therefore, are
not presented.
5.1.2.2 Evaluation of Alkalinity Map Classes—The
third level of stratification for the design was
alkalinity map class (Section 2.2.1). In order to
evaluate the effectiveness of the stratification based
on alkalinity map class, measured ANC values were
compared to the ranges of alkalinity for each map
class. For the design to be most efficient, the largest
percentage of lakes with ANC < 100 /ueq L"1 should
be observed in Map Class 1, the largest percentage
with ANC from 100-199 fjeq L~1 should be observed
in Map Class 2 and the largest percentage having
ANC > 200 fjeq L"1 should be observed in Map Class
3. In general, the map classes used in lake selection
were good estimates of the measured ANC (Table
5-2), but the intermediate class (100-199 /ueq L~1)
was less effectively classified by the maps than were
the other two classes.
The maps also can be evaluated by examining the
distribution of ANC within map classes. Table 5-3
gives the estimated statistics Qi, median (M), and
Cu for the distribution, F(x). In all cases, Map Class
1 had lower quintile values than either Map Class
2 or 3. Map Class 3 had the highest quintile values
of ANC. These results indicate that the map classes
Table 5-2. Composition of the Alkalinity Map Classes8:
Numbers and Percentages of Lakes Having
Measured ANC in Those Same Classes, Western
Lake Survey - Phase I
Estimated Percentage:
Measured ANC (/jeq L~1)
< 100 100-199 200-400
California (4A)
Map Class 1
Map Class 2
Map Class 3
75.9
66.0
33.3
16.7
17.0
14.3
7.4
17.0
52.4
1727
446
227
Pacific Northwest (4B)
Map Class 1 62.7 27.1 10.2 586
Map Class 2 47.2 34.0 18.9 567
Map Class 3 8.5 31.9 59.6 553
Northern Rockies (4C)
Map Class 1 79.2 18.9 1.9 331
Map Class 2 36.0 34.0 30.0 573
Map Class 3 22.5 10.0 67.5 1475
Central Rockies (4D)
Map Class 1 69.8 30.2 0 789
Map Class 2 40.4 44.7 14.9 928
Map Class 3 23.1 12.8 64.1 582
Southern Rockies (4E)
Map Class 1 69.6 23.9 6.5 150
Map Class 2 26.9 30.8 42.3 235
Map Class 3 9.8 19.5 70.7 1223
West (4)
Map Class 1
Map Class 2
Map Class 3
72.4
43.9
17.4
21.9
34.6
16.5
5.7
21.6
66.1
3583
2750
4060
aMap Class 1: ANC < 100 ueq L~1 (Omernik and Griffith 1986).
Map Class 2: ANC 100-199 //eq L"1 (Omernik and Griffith
1986).
Map Class 3: ANC 200-400 ^eq L'1 (Omernik and Griffith
1986).
bN = Estimated number of lakes in the target population in each
map class.
used in selection led to increased efficiency in the
design.
5.2 Description of Target Population
5.2.7 Number of Lakes Sampled
A total of 973 probability sample lakes was selected
from the map population. Of those, 95 were classified
as non-target by examination of large-scale maps,
98 were classified as non-target when visited, and
60 were not visited (Section 2.2.2). Data from water
samples collected from 720 lakes were subsequently
considered for use in making population estimates.
One lake which was larger than 2000 ha was
excluded from population estimates (Section 5.2.2);
thus, the number of lakes upon which population
estimates are based is 719.
Of the 42 special interest lakes selected, 32 were
sampled. The data collected from special interest
55
-------
Table 5-3. Population Statistics for ANC by Alkalinity Map
Class8, Western Lake Survey - Phase I
ANC (neq L~1)
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
Stratum
4A1
4A2
4A3
4B1
4B2
4B3
4C1
4C2
4C3
4D1
4D2
4D3
4E1
4E2
4E3
O.!b
32.4
44.1
56.9
30.9
33.9
128.6
35.8
69.7
91.7
60.3
62.5
92.3
40.6
63.4
127.4
Mc
54.4
76.7
200.2
82.7
108.8
234.8
50.9
119.8
425.5
87.9
113.9
270.0
74.6
175.7
440.7
Q4d
106.2
165.6
597.0
158.0
195.9
475.3
98.1
406.5
1230.7
108.2
171.3
1239.7
117.6
422.6
1120.5
"Map Class 1: ANC < 100 neq L'1 (Omernik and Griffith 1986).
Map Class 2: ANC 100-199 A/eq L"1 (Omernik and Griffith
1986).
Map Class 3: ANC 200-400 neq L"1 (Omernik and Griffith
1986).
bQ, = first quintile (20th percentile).
CM = median (50th percentile).
dO-4 = fourth quintile (80th percentile).
lakes are presented in Volume II and summarized
in Section 5.8. Because these lakes were not part
of the random selection process, weighting factors
do not apply in this case, and the representativeness
of these lakes with respect to the chemical
characteristics of the lake population as a whole is
uncertain.
5.2.2 Treatment of Shallow, Thermally
Stratified, and Large Lakes
In the WLS-I, as in the ELS-I, the intent of the design
was to sample lakes in a sufficiently deep (> 1.5
m) location when they were isothermal (i.e., mixed).
This sampling objective was not achieved in all cases;
consequently, it was necessary to evaluate the
potential influence of samples collected from shallow
and stratified lakes on the population estimates. The
number of stratified lakes sampled was small (Table
5-4) and thus their impact on the overall population
estimates is minimal.
Although there are apparent differences between
shallow lakes and deep lakes with respect to
chemistry (Table 5-4), the population estimates for
the shallow lakes are based on an insufficient
number of sample lakes (n*** = 7) from which to
draw any inferences. The small number of shallow
lakes had little influence on these population
estimates (Table 5-4) so shallow lakes were not
excluded from population estimates.
Concentrations of ANC, calcium, and sulfate (with
the exception of the median value) are higher in
stratified lakes than in mixed lakes. However, most
of the stratified lakes in the West were low elevation
reservoirs. Thus, it is inappropriate to compare high
elevation, mixed lakes with low elevation reservoirs
to assess whether stratification influences lake
chemistry. For example, a comparison of unstratified
lakes and reservoirs to unstratified reservoirs
indicates much larger differences in chemistry than
observed between stratified and unstratified lakes.
Therefore, there appears to be little basis for
designating stratified lakes as a separate class in
the data presentations.
Lakes > 2000 ha in surface area were treated as
a separate subpopulation in the ELS-I data analysis
(Linthurst et al. 1986) to minimize the influence of
a very few large lakes on the cumulative area!
distributions [G(x)]. To maintain parallel data
presentations with the ELS-I, the population
estimates for the WLS-I also exclude lakes > 2000
ha. However, only one lake > 2000 ha was sampled
in the WLS-I (Lake Almanor, California, ID number
= 4A3-017), so eliminating it had little influence on
the results for F(x) or G(x). Data for Lake Almanor
are retained in the data base and are shown in
Volume II. Lakes < 4 ha were retained as part of
the target population. The relationship between lake
area and water chemistry is discussed in Section
6.4.1.
5.2.3 Description of Target Population and
Sample
Table 5-5 gives the components of the population
and sample by strata, subregions, and region. All
estimates were made with the equations in Section
2.3. The weights provided in Table 5-5 are approp-
riate for all analyses in which weighting is necessary.
5.3 Physical Characteristics of Lake
Populations
Marked differences in lake elevation among subre-
gions were observed (Table 5-6). The lowest
elevation lakes were in the Pacific Northwest (4B),
where the Q4 value for lake elevation (1710m) was
lower than the Qi values for all the other western
subregions except the Northern Rockies (4C, Qi =
1660 m). The Southern Rockies (4E) had the highest
median and Q4 lake elevations (3264 and 3552 m,
respectively), although these values were not
markedly higher than those in the Central Rockies
(4D) or California (4A). The Central Rockies had the
least variation in lake elevation (as indicated by Qd
= OU - Qi), and California had the most.
Differences in watershed slope among subregions
were not as marked as elevational differences and
56
-------
Table 5-4. Population Statistics for ANC, Calcium, and Sulfate in Shallow, Deep, Stratified, and Unstratified Lakes, Western Lake
Survey - Phase I
ANC (MCI
Subpopulations of Lakes3 n'** N
All lakes 719 10393
Shallow « 1.5m) 7 79
Deep {> 1.5m) 712 10314
Stratified 36 459
Unstratified 683 9934
Unstratified Reservoirs 40 640
QI
54.5
70.3
54.5
87.0
54.1
279.4
M
:jL-1>
119.4
188.4
117.9
235.4
115.7
631.7
Q4
425.6
212.3
427.9
567.5
420.6
1315.5
Ca+2 (^eq L~1)
QI
42.9
28.4
42.9
78.5
42.2
176.6
M
92.4
49.4
92.5
137.4
89.7
428.0
Q4
354.9
139.2
355.9
398.6
351.4
1045.8
S04~
o,
6.2
7.8
6.2
8.0
6.2
20.9
2 (^eq L !)
M
18
11
19
18
18
48
.9
.2
.2
.2
.9
.6
Q4
38.7
13.6
38.7
49.7
38.0
132.4
n"* = number of lakes from which samples were obtained.
N = estimated number of lakes in the subpopulation.
Q! = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
aLakes > 2000 ha in area are not included.
Table 5-5. Description of the Target Population9
Survey - Phase I
Sample, and Weighting Factors Excluding Lakes > 2000 ha. Western Lake
STR
n*
W
SE(N)
SE(A)
4A1
4A2
4A3a
4B1
4B2
4B3
4C1
4C2
4C3
4D1
4D2
4D3
4E1
4E2
4E3
1885
538
383
695
724
781
343
675
2317
885
1024
1061
150
261
1784
60
65
72
70
70
70
60
60
65
60
60
75
60
63
63
54
53
42
59
53
47
53
50
40
43
47
39
46
52
41
31.978
8.422
5.416
9.929
10.700
11.766
6.246
11.455
36.875
18.356
19.744
14.918
3.261
4.526
29.834
1726.81
446.37
227.47
585.81
567.10
553.00
331.04
572.75
1475.00
789.31
927.97
581.80
150.01
235.35
1223.19
67.58
23.94
20.34
28.87
34.78
42.01
7.85
30.38
139.79
38.89
40.78
60.90
0.00
9.11
106.19
14256
10822
10931
7676
9185
51177
1831
4136
30097
9242
7726
12818
711
1119
10123
2409
3355
3296
1566
3469
23022
196
794
11284
1877
1121
5324
65
136
2979
Subregion
4Aa
4B
4C
4D
4E
Region
4a
STR =
N*
n* =
n*** =
W
N
SE(N) =
A
SE(A) =
2806
2200
3335
2970
2195
13506
stratum.
map population.
number of lakes
number of lakes
weight.
estimated target
standard error of
197
210
185
195
186
973
149
159
143
129
139
719
2400.65
1705.91
2378.79
2299.08
1608.55
10392.98
74.52
61.71
143.27
82.98
106.58
219.39
36009
68038
36065
29786
11953
181851
5284
23335
11314
5756
2983
27248
in the probability sample.
sampled.
population
N.
size.
estimated area of target population.
Standard error of
A.
aLake Almanor (ID number = 4A3-017) is not included in these estimates (Section 5.2.2). The estimates for the target population
which included Lake Almanor are:
STR 4A3: N = 232.89, SE(N) = 20.22, A = (65148), SE(A) = 48927.
Subregion 4A: N = 2406.07, SE(N) = 74.52, A = 90227, SE(A) = 49101.
Region 4: N = 10,398.40, SE(N) = 219.38, A = 236069, SE(A) = 55906.
57
-------
Table 5-6. Population Statistics for Physical Variables for Lakes in the Target Population.
Lake Elevation (m)
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
West
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
West
4A
4B
4C
4D
4E
4
4A
4B
4C
4D
4E
4
QI
2136
889
1660
2804
2659
1746
QI
2.3
2.3
2.2
3.0
1.5
2.2
M
3008
1437
2064
3040
3264
2613
Lake Area (ha)
M
4.6
4.5
4.6
5.4
3.5
4.6
Q4
3330
1710
2454
3255
3552
3237
Q4
12.3
18.3
8.9
15.7
7.1
11.7
Watershed Area (ha)
QI
47
36
43
67
35
45
QI
4.9
3.3
3.5
4.8
2.8
3.8
M
145
98
95
178
75
118
Site Depth
M
8.8
7.6
8.8
8.9
5.1
7.8
Q4
440
381
214
618
403
436
(m)
Q4
20.0
21.9
17.3
19.0
10.4
17.7
Western Lake Survey -
Phase 1
Watershed Area to
Lake Area Ratio
QI
11.8
9.6
10.8
13.1
11.4
11.0
M
22.9
19.8
23.2
30.2
23.5
23.7
Q4
66.7
42.7
48.3
72.3
73.3
58.5
Watershed Slope (%)
QI
30.1
14.3
23.5
17.7
15.5
19.3
M
51.6
37.1
46.3
30.7
31.1
39.9
Q4
62.7
62.9
61.5
49.9
56.9
61.4
Q, = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
did not show the same pattern (Table 5-6). The
subregions with the highest median (M) elevations,
the Central and Southern Rockies, had the lowest
median watershed slopes (30.7% and 31.1%,
respectively). California had the highest median
watershed slope (51.6%) and the Pacific Northwest
had the greatest variability in watershed slope (Qd
= 48.6%).
The largest variation in watershed size and the
largest watersheds (at Qi, M, and Q4) were found
in the Central Rockies (M = 178 ha. Table 5-6). The
median watershed area in California (145 ha) was
also considerably larger than in other subregions.
The Southern Rockies had the smallest median
watershed area (75 ha) which varied considerably
(Qd = 368 ha). Watershed area in the Northern
Rockies was considerably less variable than in other
subregions (Qd = 171 ha).
The smallest watershed area to lake area ratio, as
well as the least variability (Qd = 33.1) in this
parameter, occurred in the Pacific Northwest (M =
19.8, Table 5-6). Variability was also low in the
Northern Rockies (Qd = 37.5). The Southern Rockies
had the greatest variability in this ratio. The Central
Rockies had the highest median watershed area to
lake area ratio (30.2) and a variability similar to that
in California.
The Southern Rockies had the smallest lakes (M =.
3.5 ha) and the least variation (Qd) in lake area (Table
5-6). Median and Qi values for lake size were similar
among the other subregions, but Q4 values were
much higher in the Pacific Northwest and the Central
Rockies than in California and the Northern Rockies.
The shallowest lakes occurred in the Southern
Rockies (M = 5.1 m, Table 5-6). Site depth was similar
among the other subregions (median values from
7.6 to 8.9 m).
Drainage lakes were predominant in all subregions,
ranging from 58.8 percent in the Southern Rockies
to 83.8 percent in California (Table 5-7). The
percentage of seepage lakes ranged from 24.0
percent in the Central Rockies to 11.8 percent in
California. Seepage lakes were second in importance
in all subregions except in the Southern Rockies
where reservoirs (21.0%) were slightly more
common. Reservoirs comprised 3.8 to 4.7 percent
of the lakes in each of the other subregions. Closed
lakes comprised 0.3 to 3.4 percent of the lakes in
each subregion.
In the analysis of land use types, the categories of
Anderson et al. (1976) were used (Section 2.4.5).
Table 5-7. Population Estimates of the Percentage of Lakes
in Categories by Lake Type. Western Lake
Survey - Phase I
Hydrologic
Drainage
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
West
4A
4B
4C
4D
4E
4
83.
71.
80,
69.
58,
74.
,8
8
,5
,3
,8
.0
Seepage
11.
22.
12,
.8
2
,3
24.0
20.0
17.6
Lake
Type1
Reservoir
4.
4.
3.
4,
21.
6,
.0
,7
.8
,7
.0
.9
a
Closed
0.
1.
3,
2
0
1
.4
,3
.4
.1
.3
.6
aWetzel (1983).
58
-------
Evergreen forest and tundra were the predominant
land cover types in the watersheds of the WLS-I
(Table 5-8). Evergreen forests were most important
in the Pacific Northwest and in the Northern Rockies,
where they dominated 77.8 percent and 52.3
percent, respectively, of the watersheds. Tundra was
the predominant land cover in 57.1 percent of the
watersheds in the Central Rockies and 54.0 percent
of the watersheds in both California and the
Southern Rockies. Barren land was the only other
type of land cover to dominate over 5 percent of
the watersheds in any subregion. Barren land was
predominant in 25.5 percent of the watersheds in
the Northern Rockies, although this unusually high
percentage may be an artifact of using different land
use information sources for most of this subregion
because USGS maps were not available.
5.4 Population Estimates
for Primary Variables
Population estimates for six key variables are
presented below. These were selected for discussion
because of their direct relevance to the potential
effects of acidic deposition. All values cited are the
estimated population statistics, but are not always
indicated as estimates in the text. The reference
values are consistent with those in the ELS-I report
(Linthurst et al. 1986) and were not based on
inspection of WLS-I data. These reference values
were selected to facilitate comparisons between
subregions and are not meant to imply critical or
background conditions. Comparisons between
subregions in the Eastern and Western Lake Surveys
can also be made using the estimates for lakes given
the reference values.
5.4.1 Acid Neutralizing Capacity
5.4.1.1 Reference values—Reference values for
ANC were selected to discuss differences among
subregions. Lakes with ANC <0/ueq L~1 were defined
as acidic. A reference value of 50 /ueq L~1 was
selected to represent "low" acid neutralizing
capacity. Lakes with ANC < 50 /ueq L 1, while having
some buffering capacity, have been demonstrated
to have decreases in ANC during snowmelt (Alt-
shuller and Linthurst 1984), although snowmelt
runoff has been reported to have also caused loss
of ANC in lakes > 50 /ueq L"1. An ANC of < 200
fjeq L~1 has been used frequently as a criterion to
separate lakes that may be "sensitive" to acidic
deposition from lakes that may not be (Swedish
Ministry of Agriculture 1982, Altshuller and
Linthurst 1984).
It is important to note that the estimates for ANC
< 200/ueq L"1 in Table 5-9 are cumulative and include
lakes with ANC < 50 /ueq L~1. Only one lake (Fern
Lake, Wyoming, ID number = 4D3-017) in the West
had ANC < 0 /ueq L~1; thus, no summary table is
provided for this reference value, whereas Linthurst
et al. (1986) used this value to present results for
lakes in the eastern United States. Cumulative
frequency and areal distributions for ANC are given
in Section 5.5 and Volume II.
5.4.1.2 Subregional Estimates and Distribution of
Lakes Having Low Acid Neutralizing Capacity—
California (4A) contained the highest estimated
percentage and number of lakes (36.7%, 800) in the
West with ANC < 50 /ueq L"1 (Table 5-9). The next
highest percentage (19.5%) occurred in the Pacific
Northwest (4B), followed by 12.7 percent in the
Northern Rockies (4C). Less than seven percent of
the lakes in the Central Rockies (4D) and Southern
Rockies (4E) had ANC < 50 /ueq L"1.
California also had the highest percentage of lakes
(86.6%) with ANC < 200 yueq L"1 (Table 5-9). The
Central Rockies had the next highest percentage
(77.8%), followed by the Pacific Northwest with 71.6
percent. Slightly more than one-half of the lakes in
the Northern Rockies and 39.4 percent of the lakes
in the Southern Rockies had ANC < 200 /ueq L"1.
Lakes with ANC < 50 /ueq L"1 were found throughout
California, particularly in the southern Sierra Nevada
(Figure 5-2). In the Pacific Northwest, lakes with ANC
< 50 //eq L"1 were located along the Cascade Crest
Table 5-8. Population Estimates of the Percentage of Watersheds Dominated8 by Land Use/Land Cover Categories,
Western Lake Survey - Phase I
Land Use/Land Cover Categories'3
Agriculture Range Deciduous Evergreen Mixed Barren Tundra Snow
West
0.1
1.9
0.6
46.9
0.6
6.7
38.4
0.4
Total
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
4A
4B
4C
4D
4E
0
0.7
0
0
0
0.6
0
1.6
3.5
4.0
0
0
0
0
3.7
41.1
77.8
52.3
32.5
35.4
0
3.4
0
0
0
2.6
0
25.5
1.4
0
54.0
10.0
14.3
57.1
54.0
0
2.5
0
0
0
98.3
94.4
93.7
94.5
91.7
95.6
3 More than 50 percent of the watershed in the category.
bAnderson et al. (1976).
59
-------
Table 5-9. Population Estimates of Lakes with ANC S 50 or
< 200 f/eq L~1, Western Lake Survey - Phase I
ANC <50 peq L'1
PC
9c
California 4A 0.367
Pacific Northwest 4B 0.195
Northern Rockies 4C 0.127
Central Rockies 4D 0.069
Southern Rockies 4E 0.046
880 1081 0.234 842312544
332 415 0.043 2916 4197
301 413 0.037 1335 1792
158 240 0.049 1454 2584
74 96 0.031 374 509
West
4 0.168 1745 20040.0801450118988
ANC <200^eq L~
fl.
9c
California 4A 0.866 2078 22330.6122203427706
Pacific Northwest4B 0.716 1222 13330.2161472119624
Northern Rockies 4C 0.507 1205 1411 0.185 6669 8351
Central Rockies 4D 0.778 1788 19290.6591962724327
Southern Rockies 4E 0.394 634 7870.213 2540 3232
West
4 0.666 6926 7276 0.361 65591 74624
pc = estimated proportion of lakes with ANC S 50 pieq L 1 or
<200 neq L~1.
Nc = estimated number of lakes with ANC < 50 ^eq L~1 or
< 200 f^eq L~1.
Ncu = 95 percent upper confidence limit for Nc.
gc = estimated proportion of lake area with ANC < 50 f*eq
L~1 or < 200 ^eq L~1.
Ac = estimated area of lakes with ANC < 50 /jeq L~1 or
<200neq L~1.
Acu = 95 percent upper confidence limit for Ac.
refer only to the areas covered by the Western Lake
Survey - Phase I and conditions in lakes outside the
area sampled cannot be inferred. The state of
California contained the largest number of lakes
(880) with ANC < 50 yueq L"1. An estimated 219
lakes in Washington and 189 lakes in Idaho had ANC
< 50 //eq L~1. The estimated numbers of lakes with
ANC < 50 /ueq L~1 were lowest for Colorado (70)
and Utah (20). Of the eight states for which estimates
were calculated, all had several hundred lakes with
ANC < 200 fjeq L~1. These estimates ranged from
461 lakes in Oregon to 2078 lakes in California.
5.4.2 pH
5.4.2.1 Subregional Estimates—The two refer-
ence values chosen for pH in the ELS-I were 5.0
and 6.0. Because only one lake (Fern Lake, Wyoming,
ID number = 4D3-017) had pH < 5.0, population
estimates are presented here only for lakes with pH
< 6.0. Cumulative frequency and areal distributions
for pH are given in Section 5.5 and Volume II.
Only one percent of the lakes in the West were
estimated to have pH < 6.0 (Table 5-11). In the
Northern (4C) and Southern (4E) Rockies, no lakes
with pH < 6.0 were sampled. The Pacific Northwest
(4B) had the highest percentage (2.4%) of lakes with
pH < 6.0. In California (4A) and the Central Rockies
(4D), an estimated 1.3 percent of the lakes had pH
<6.0.
in Oregon, throughout the Cascades in Washington,
and on the Kitsap Peninsula (Figure 5-3). In the
Northern Rockies, lakes with ANC < 50 /ueq L~1 were
located principally in the Bitterroot Range of western
Montana and northern Idaho and in the Sawtooth
Mountains in Idaho (Figure 5-4). Some low ANC lakes
were also found in the Salmon River Mountains, the
Clearwater Mountains, and the Selkirk Mountains
in Idaho, and in the Lewis Range in Montana. Lakes
in the Central Rockies with ANC < 50 /ueq L"1
occurred in the Wind River Range and the Big Horn
Mountains in Wyoming, the Beartooth Mountains
and the Yellowstone Plateau in Wyoming and
Montana, and the Uinta Mountains in Utah (Figure
5-5). In the Southern Rockies, lakes with ANC <
50 /ueq L"1 were found primarily in the San Juan
Mountains and in the Front Range. Some low ANC
lakes were also found in the Park Range in Colorado
and the Medicine Bow Range in Wyoming (Figure
5-6).
5.4.1.3 State Estimates—Population estimates for
lakes with ANC < 50 or < 200 A 50 //eq L~1 Table 5-12). The
highest percentage and number (33.7%, 543) of lakes
with SOT2 > 50 jueq L"1 occurred in the Southern
Rockies. The Central Rockies was estimated to
60
-------
Figure 5-2. Locations of lakes sampled in California (4A) with ANC (/jeq i"V in one of three classes ( • .' ANC < 50.
-\- : ANC > 50-200. and O : ANC > 200). Western Lake Survey - Phase I. Geographic area covered by
the target population is bounded by a dashed line.
Acid Neutralizing
Capacity (/jeq L'^)
• <50
+ > 50-200
O >200
61
-------
Figure 5-3. Locations of lakes sampled in the Pacific Northwest (4B) with ANC (pet/ L'1) in one of three classes ( • :
ANC < 50. -+- : ANC > 50-200. and O : ANC > 200). Western Lake Survey - Phase I. Geographic area
covered by the target population is bounded by a dashed line.
CT--
\ /
«&?!?
W°
vt 1 •> +
^V «NV I ' •
^Kr °:
W^if , o
IJ\I \ + °r>
•//* W f +
fo£?K
^/fr*H '
v4f _y/ /
WV/ /o
1 / •
/ +
•
/' "' °
v \
/
1--
t
/
V t'
\ '' + ••
\ '*
1 »
I t
1 >• __
V "-ix^^
— ^-^r
/*+*
,- ?
t
,< +
/'•f
/
/ •»• +
/ 0 *
1 -t-
< I--
1 •
-t-
' _
' „ +
x- o T^
/ +
|
+*°
. T
»*
/ ^«1
"
I fr /
•> /
"^
/
i' ^J
^ \
1 •*- '
/ + '
\o \
x y
\ /•
' • — T T :
o • °? o i
o- 0° -. ',
+ . V > l
+ B. + Ov -
4 + + o o X^
' ^^ +\
+ + 4 )
o. + + «'
• ••• ^
j.T /
_i. f 9 '
•'Wf O '
**• + v
\+ 0 ^
y-«., i
\
X
N
^
0 'l
+ t'
"^ \
o o /
(*/ WX>
0 '-,
** ^ ^
"~ \
I "— — - >.
1 _S*~*'~~~^
/ **•
-i-~^ ^^^^^
1 >«y^-.X^
(
i Off
/
/
1
1
1
j
/
;
\
\
1
/
, Acid Neutralizing
Capacity f/ieq L'^)
m <50
+ > 50-200
0 >200
\
\
1
— \
/
/
/
/
(
V
^
62
-------
Figure 5-4. Locations of lakes sampled in the Northern Rockies (4C) with ANC (peq O in one of three classes ( • : ANC
< BO, -\- : ANC > 50-200, and O -' ANC > 200). Western Lake Survey - Phase I, Geographic area covered
by the target population is bounded by a dashed line.
4C
/ o
I
\ o
v"v
s ^ 'v
^ (
1
/
/
\
V _
^
0
0
o \
o ° \
^.
{ "*». o \
0 V ' + V
<'•*)' x
1 ^**' ^ O X
\ ^ o x
\ / \
r, ' \
\ * " ' '
i \ ^ i '
1 \ > » 0 '
' J . ' ' O ' «*T
WA
Acid Neutralizin
Capacity djeqL
<50
o/?
+ > 50-200
°>200
63
-------
Figure 5-5. Locations of lakes sampled in the Central Rockies (4D) with ANC f/jeg L V /" one of three classes I • / ANC
< 50. -f- -' ANC > 50-200. and O : ANC > 200). Western Lake Survey - Phase I. Geographic area covered
by the target population is bounded by a dashed line.
4D
\++1
MT
WY
Acid Neutralizing
Capacity (tteq L~*)
• <50
-|_ > 50-200
>200
64
-------
Figure 5-6. Locations of lakes sampled in the Southern Rockies (4E) with ANC (peg L ^ in one of three classes ( • : ANC
< 50. -\- : ANC > 50-200. and O -- ANC > 200). Western Lake Survey - Phase I. Geographic area covered
by the target population is bounded by a dashed line.
/ /
WY
4E
'~
CO
I O
I O
..J^S ° f/
S;"~" ° v>/ -
O, * r* x-' >
-1 + . \ i
o o
\^
° C"iN.
"] \ ' „ \ \ ""•«, Acid Neutralizing
\ ^J t~- \ °\ ' \ Capacity (/jeqL~'}
l' ' ^^ ,' . ^Co
V^'' 10 i Vf" " ^°
^0 - \_ + > 50-200
•x Aft O . ^^ ! .. \
f» ++ "> ^o,"» i O >20°
x T + > \
^x 0 ^ / 1 ,
^x + , it
x +l / /
N T \ ^j
"• '^ '.
NM / , p -N \ ,
/ ^ » ; 'V
_ j
_
65
-------
Table 5-10. Population Estimates and Population Statistics by State for ANC and pH, Western Lake Survey - Phase I. These
Estimates and Statistics Refer Only to Those Lakes Within the Study Area8
State"
Estimated Number of Lakes (UCL)C
Population Statistics
ANC (/jeq
(N)
ANC < 50
ANC< 200
-1
M
California 147 2390 880(1081)
Colorado 132 1476 70 (90)
Idaho 72 972 189 (284)
Montana 80 1597 160 (240)
Oregon 55 551 113 (165)
Utah 30 548 20 (51)
Washington 117 1338 219 (290)
Wyoming 83 1480 94 (158)
Stateb n*** (N) pH < 6.0
California 147 2390 32(84)
Colorado 132 1476 O(-)
Idaho 72 972 O(-)
Montana 80 1597 O(-)
Oregon 55 551 10(25)
Utah 30 548 O(-)
Washington 117 1338 31(59)
Wyoming 83 1480 30(63)
n*" = number of lakes from which samples were obtained.
N = estimated number of lakes in the area of the state covered by
Q, = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
(-) = undefined.
a Refer to Figure 2-1 (Section 2.1.3) for study area.
bOnly one lake was sampled in New Mexico and two in Nevada; thus.
2078 (2233)
591 (739)
599 (727)
824(1035)
461 (558)
484 (620)
822 (935)
1068(1234)
WLS-I.
estimates for these
34.4
114.8
49.8
79.0
44.7
59.9
55.3
75.0
QI
6.6
7.1
6.8
7.0
6.6
6.8
6.7
6.9
states were not
62.5
305.1
108.3
191.9
104.0
85.8
156.9
128.4
pH
M
6.9
7.6
7.0
7.4
7.1
7.0
7.0
7.2
computed.
136.6
981.8
614.4
977.6
198.3
114.0
461.8
241.6
0-4
7.4
8.4
8.0
8.0
7.3
7.3
7.5
7.5
c The 95 percent upper confidence limit (UCL = Nco) is shown in parentheses
Table 5-11. Population Estimates of Lakes with pH < 6.0,
Western Lake Survey - Phase I
Table 5-12. Population Estimates of Lakes with Sulfate
>50 fjeq \-~\ Western Lake Survey - Phase I
PC
"cu
PC
9c
California 4A
Pacific Northwest 4B
Northern Rockies 4C
Central Rockies 4D
Southern Rockies 4E
0.013
0.024
0.000
0.013
0.000
32
41
0
30
0
84
73
—
63
-
0.002
0.002
0.000
0.022
0.000
64
106
0
647
0
168
193
—
1572
-
California 4A
Pacific Northwest 4B
Northern Rockies 4C
Central Rockies 4D
Southern Rockies 4E
0.078
0.171
0.107
0.056
0.337
187
292
256
128
543
300 0.093
374 0.138
399 0.414
198 0.056
711 0.569
3363 6793
9389 17408
14914 30241
1667 3123
6805 11737
West
4 0.010 103 172 0.004 817 1752
pc = estimated proportion of lakes with pH 5= 6.0.
Nc = estimated number of lakes with pH 2 6.0.
Ncu = 95 percent upper confidence limit for Nc.
gc = estimated proportion of lake area with pH < 6.0.
Ac = estimated area of lakes with pH ^ 6.0.
Acu = 95 percent upper confidence limit for Ac.
(-) = undefined.
contain the fewest lakes (128) and lowest percentage
(5.6%) of lakes with S04"2 > 50 //eq L"1.
5.4.4 Calcium
A reference value for calcium of 50 //eq L"1 selected
in the ELS-I was also used to compare the five
western subregions. Cumulative frequency and areal
distributions for Ca+2 are given in Section 5.5 and
Volume II.
West
4 0.135 14051676 0.1993613854507
pc = estimated proportion of lakes with sulfate ^ 50 ^eq L 1.
Nc = estimated number of lakes with sulfate ^ 50 ^eq L~1.
Ncu = 95 percent upper confidence limit for Nc.
gc = estimated proportion of lake area with sulfate > 50 ^eq
In the West, an estimated 24.6 percent of the lakes
had Ca+z < 50 peq L"1 (Table 5-13). The percentages
of lakes with low levels of Ca+2 varied markedly
among subregions. More than one-half (56.5%) of
the lakes in California were estimated to have Ca+2
< 50 /jeq L"1 (Table 5-13). Of the target populations
of lakes in the Central Rockies and in the Southern
Rockies, 6.9 and 3.4 percent, respectively, had Ca+2
66
-------
5.4.5 Extractable Aluminum, Clearwater Lakes
Population estimates for extractable aluminum are
given for clearwater lakes only, which comprised
96.1 percent of the western target population. The
reference value used was 50 /JQ L~1 based on ELS-I
results. Clearwater lakes were defined as having true
color < 30 PCU. Darkwater lakes (> 30 PCD) were
assumed to have adequate concentrations of
dissolved organic material to produce organic
complexes of aluminum, which can ameliorate the
toxic effects of aluminum (Driscoll et al. 1980; Baker
and Schofield 1 982). Population estimates are given
for extractable Al > 50 //g L~1. Inverse cumulative
frequency and areal distributions are given in
Volume II.
Less than one percent of the clearwater lakes in
the West had extractable Al > 50 jjg L~1 (Table 5-14).
Eleven clearwater lakes in the Pacific Northwest and
five clearwater lakes in California had extractable
Al > 50 fjg L~1. No clearwater lakes with extractable
Al > 50 //g L~1 were sampled in the Rockies. These
results sharply contrast those for the northeastern
United States (Section 8).
5.4.6 Dissolved Organic Carbon
Two values were used to compare population
estimates for dissolved organic carbon (DOC) in the
ELS-I, DOC < 2 mg L~1 (low) and DOC > 6 mg L~1
(high). The latter value was selected because, in lakes
with high DOC > 6 mg L~1, organic anions can
contribute appreciably to the ion balance (Oliver et
al. 1983). Cumulative frequency and areal distribu-
tions for DOC are given in Volume II.
Most of the lakes in the West had low DOC (69.7%,
Table 5-15). The percentage of lakes with low DOC
ranged from 56.5 percent in the Southern Rockies
to 80.7 percent in California. Very few lakes in the
West had high DOC. The highest estimated number
of lakes with DOC > 6 mg L~1 was observed in the
Northern Rockies (196) followed by the Southern
Rockies (158).
5.5 Statistics for Population
Descriptions, Primary Variables
Direct qualitative comparisons of lake chemistry
among subregions can be made by superimposing
cumulative frequency [F(x>] curves. This approach
can serve to highlight major differences in the
distributions of physical or chemical characteristics
of lakes among subregions. Superimposed curves
can also show subtle differences among subregional
distributions which are not always evident from an
analysis of population statistics, such as quintiles
or medians.
Table 5-13. Population Estimates of Lakes with Calcium
<50 neq L~1, Western Lake Survey - Phase I
PC
9c
California 4A 0.565 1357 15660.3971428719191
Pacific Northwest 4B 0.284 485 578 0.058 3954 5380
Northern Rockies 4C 0.213 506 6490.073 2617 3615
Central Rockies 4D 0.069 158 2400.034 1000 1696
Southern Rockies 4E 0.034 55 740.023 274 398
West
4 0.246 2562 28440.1222213227383
pc = estimated proportion of lakes with calcium < 50 ^eq L 1.
Nc = estimated number of lakes with calcium < 50 ^eq L"1.
Ncu= 95 percent upper confidence limit for Nc.
gc = estimated proportion of lake area with calcium < 50^ieqL~1.
Ac = estimated area of lakes with calcium < 50 neq L"1.
Acu= 95 percent upper confidence limit for Ac.
Table 5-14. Population Estimates of Clearwater Lakes with
Extractable Aluminum >50 yueq L 1, Western
Lake Survey - Phase I
PC
California 4A 0.002
Pacific Northwest 4B 0.006
Northern Rockies 4C 0.000
Central Rockies 4D 0.000
Southern Rockies 4E 0.000
West 4 0.002
*c
5
11
0
0
0
16
NCU
14
28
—
—
—
35
9c
0.000
0.000
0.000
0.000
0.000
0.000
*c
10
33
0
0
0
43
Acu
24
85
_
—
—
97
pc = estimated proportion of lakes with extractable aluminum
>50yugL"1.
Nc = estimated number of lakes with extractable aluminum
>50yugL"1
NCU = 95 percent upper confidence limit for Nc.
gc = estimated proportion of lake area with extractable
aluminum >50/ug L'1.
Ac = estimated area of lakes with extractable aluminum
>50yug L"1
Acu = 95 percent upper confidence limit for Ac.
(-) = undefined.
The median, Qi and Q4 values also can be used to
compare the distributions among subregions. The
Qi value is useful in comparing variables for which
low concentrations are of interest, such as ANC or
pH. Similarly, Q4 values are useful in comparing
variables for which high concentrations are of
primary interest, such as S04 2 or extractable Al.
The superimposed cumulative frequency distribu-
tions for ANC for the five western subregions
indicates that ANC concentrations in lakes in
California were generally lower than in other
subregions (Figure 5-7). The distribution for ANC in
the Pacific Northwest was similar to that in the
Central Rockies. The distributions indicate that
generally, the ANC values for lakes in the Northern
Rockies and in the Southern Rockies were substan-
tially higher than in the other subregions.
Comparisons of the distribution curves for pH (Figure
5-7) yield results similar to those observed for ANC
67
-------
Table 5-15. Population Estimates of Lakes with Dissolved
Organic Carbon < 2 mg L~1 or >6 /veq L~1,
Western Lake Survey - Phase I
Dissolved Organic Carbon <2 mg L~1
California 4A 0.807 1937 2105 0.937 33728 42462
Pacific Northwest 4B 0.681 1161 12770.880 59873 98049
Northern Rockies 4C 0.708 1685 1929 0.511 18427 29879
Central Rockies 4D 0.673 1 548 1717 0.772 23005 32426
Southern Rockies 4E 0.565 90810880.525 6276 10417
West
4 0.697 7239 7642 0.777 141309 183389
Dissolved Organic Carbon >6 mg L
PC
9c
A A
"c "cu
California 4A
Pacific Northwest 4B
Northern Rockies 4C
Central Rockies 4D
Southern Rockies 4E
0,
0
0,
0
0
,030
.027
.082
.040
.098
71
45
196
93
158
129
80
326
152
263
0
0
0
0.
0,
.004
.003
.039
.034
,131
156
214
1423
1025
1566
276
400
2481
2065
3660
West
4 0.054 563 753 0.024 4385 6961
pc = estimated proportion of lakes with dissolved organic
carbon <2 mg L~1 or >6 mg L~1.
Nc = estimated number of lakes with dissolved organic
carbon <2 mg L~' or >6 mg L~1.
Ncu = 95 percent upper confidence limit for Nc.
gc = estimated proportion of lake area with dissolved organic
carbon < 2 mg L~1 or 2- 6 mg L'1.
Ac = estimated area of lakes with dissolved organic carbon
<2 mg L"1 or >6 mg L"1.
Acu = 95 percent upper confidence limit for Ac.
for the three subregions in the Rocky Mountains.
In contrast to ANC, however, the distributions for
pH in California and the Pacific Northwest were
virtually indistinguishable.
Sulfate data are presented as the inverse distribu-
tion, 1 -F(x) (Figure 5-8). Higher values of SO4~2 were
observed in lakes in the Southern Rockies than in
other subregions. Lakes in California were generally
characterized by very low SO*~2 concentrations. The
distributions were very similar for the Pacific
Northwest and the Northern Rockies, particularly at
low concentrations. The distribution in the Central
Rockies was shifted toward higher concentrations,
which approached those observed in the Southern
Rockies.
In most lakes Ca+2 is the dominant base cation; thus
the distributions for Ca+2 would be expected to be
similar to those for ANC. For lakes sampled in the
West, the patterns for Ca+2 distributions closely
follow those for ANC distributions (Figure 5-8). The
relationship between ANC and Ca+2 is discussed in
more detail in Section 6.2.
The distribution curves for the other two primary
variables, DOC and extractable Al, are not presented
as superimposed F(x) curves because concentrations
were very low and the differences among subre-
gional curves are indistinguishable. Distributions for
these variables are given in Volume II.
California had the lowest values of ANC at Qi,
median and Q4 levels (Table 5-16). The Pacific
Northwest had the second lowest Qi value, whereas
the Central Rockies had the second lowest median
value of ANC. The Southern Rockies had the highest
estimated quintile values.
The lowest Qi value for S04~2 was estimated for
the Pacific Northwest, whereas California had the
lowest estimated median value. Sulfate was highest
at the Qi level in the Central Rockies and the highest
median value was estimated for the Southern
Rockies.
The distribution for Ca+2 paralleled that for ANC,
because calcium is the primary base cation (Section
6.2). Calcium was lowest in California and highest
in the Southern Rockies. The Pacific Northwest had
the second lowest Qi value for Ca*2, whereas the
Central Rockies had a slightly lower median than
did the Pacific Northwest.
Little interquintile difference was observed in DOC
values. Dissolved organic carbon was generally
lowest in California and highest in the Southern
Rockies. Extractable Al values were all low and
median values were near or below the system
decision limit (4 //g L"\ Section 4.2).
5.6 Statistics for Population
Descriptions, Secondary Variables
Many variables measured in the WLS-I, in addition
to the primary variables discussed above, are useful
in characterizing lakes. In this section, related sets
of variables are grouped: nutrients (nitrate, ammo-
nium and total phosphorus), those related to
transparency (true color, turbidity and Secchi disk
transparency), major cations (potassium, sodium and
magnesium), metals (iron, manganese and total
aluminum), and others (silica, dissolved inorganic
carbon, chloride, conductance and bicarbonate).
5.5.7 Nutrients
Nutrient concentrations in western lakes were
generally low. Estimated Qi values for total
phosphorus were less than 2 fig L"1 for California,
the Pacific Northwest, and the Northern Rockies, and
were less than 4 /ug L~1 for all subregions (Table
5-17). Median values of total phosphorus were less
than 4 /ug L~1 and Q4 values were less than 10 //g L'1
for lakes in California, the Pacific Northwest, and
the Northern Rockies. The Southern Rockies had the
68
-------
Figure 5-7. Cumulative frequency distributions [F(x)J for ANC (peg L V and pH in California ( 1, the Pacific Northwest
( ), the Northern Rockies ( ;, the Central Rockies ( ). and the Southern Rockies ( ). Western
Lake Survey - Phase I.
FM
1.0
0.8
0.6
0.4
0.2
0.0
California (4A)
Pacific Northwest (4B)
Northern Rockies (4CJ
Central Rockies (4D)
Southern Rockies I4E)
1.0
0.8
0.6
0.4
0.2
0.0
4.0
600
1
soo
;ooo
ANC (fjeq L"1)
California (4A)
Pacific Northwest (4B)
Northern Rockies (4C)
Central Rockies (4D)
Southern Rockies (4E!
PH
69
-------
Figure 5-8.
Cumulative frequency distributions [F(x)J for calcium lueq L'^) and inverse cumulative frequency distributions
[1-F(x)J for sulfate (fteq £"V '" California ( ;, the Pacific Northwest ( ). the Northern Rockies f ).
the Central Rockies ( ), and the Southern Rockies ( ). Western Lake Survey - Phase I.
1.0
o.s -
0.6
0.4 -
0.2 ~
0.0
California I4A)
Pacific Northwest /4B>
Northern Rockies I4C)
Central Rockies (4D)
Southern Rockies f4E)
200
600
800
1000
Calcium (ueq L'
California (4A)
— Pacific Northwest (4BJ
Northern Rockies (4C)
Central Rockies (4D)
Southern Rockies (4EI
0.0
40 60
Sulfate (ueq L~')
80
100
70
-------
Table 5-16. Population Statistics for the Primary Variables, Western Lake Survey - Phase I
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
West
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
4A
4B
4C
4D
4E
4
4A
4B
4C
4D
4E
0-1
34.5
51.7
69.1
64.7
113.9
54.5
QI
22.8
37.3
48.8
65.4
79.2
ANC (fiec
M
62.6
139.8
195.2
104.9
317.0
119.4
Ca+2 (jieq I
M
43.2
95.8
154.2
92.7
233.1
|L-1)
a
136
293
1011,
207
1039
425
--')
Q<
106
206
730.
169,
724,
i
.9
.1
.4
.1
.4
.6
t
.1
.0
,0
,3
.7
QI
6.59
6.63
6.90
6.91
7.10
6.81
QI
1.2
2.1
1.1
0.0
0.1
pH
M
6.94
7.00
7.35
7.18
7.60
7.16
Ext Al (Mg L'1)a
M
2.7
5.6
3.3
1.1
2.5
Q4
7.38
7.34
7.99
7.47
8.27
7.69
Q4
4.6
11.1
7.2
3.8
6.7
QI
3.6
2.3
6.3
19.1
16.6
6.2
QI
0.46
0.52
0.75
0.68
0.82
S04 2 (jieq L
M
6.6
14.7
15.6
24.4
34.6
18.9
DOC (mg L"1)
M
0.97
1.30
1.20
1.25
1.48
')
Q4
13.5
43.3
32.3
35.0
115.5
38.7
Q4
1.93
2.57
2.63
3.23
3.72
West
42.9
92.4
354.9
0.6
2.8
6.6
0.64
1.21
2.76
Q, = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
aClearwater (true color < 30 PCU) lakes only.
Table 5-17. Population Statistics for Secondary Variables (Nitrate, Ammonium and Total Phosphorus), Western Lake Survey -
Phase I
California
Pacific Northwest
Northern Rockes
Central Rockies
Southern Rockies
West
4A
4B
4C
4D
4E
4
Q,
0.1
0.0
0.1
0.2
0.1
0.1
N03~
(fjeq L'1)
M
0.4
0.1
0.4
0.4
0.5
0.4
CU
2.5
1.4
2.1
3.1
2.5
2.2
0.1
0.0
0.0
0.0
0.0
0.0
0.0
NlV
(fjeq L'1)
M
0.0
0.0
0.0
0.0
0.0
0.0
0.4
0.6
0.6
0.4
0.4
0.6
0.5
Qi
0.7
1.6
1.9
3.0
3.6
1.7
Total P
G«g L"1)
M
2.9
3.5
3.7
6.9
8.1
4.7
O.4
7.3
7.3
9.7
12.6
26.6
10.9
Qi = first quintile (20th percentile).
M = median (50th percentile).
Qt = fourth quintile (80th percentile).
highest concentrations of total phosphorus (Q4 = 26. 6
fjQ L"1), although the median value was only 8.1
Nitrate and ammonium concentrations were also low
for lakes in all subregions. The Qi values for nitrate
were less than or equal to 0.2 /ueq L"1 and median
values were less than or equal to 0.5 /ueq L"1.
Ammonium concentrations at the Q4 value did not
exceed 0.6 /ueq L"1 for lakes in any subregion.
Nitrate may be an important anion during periods
of snowmelt runoff. For example, it has been
regarded as a major contributor to reductions in acid
neutralizing capacity during snowmelt episodes in
northeastern North America (Galloway and Dillon
1 983). Nitrate levels in western lakes were very low.
The lakes were sampled during the fall, well after
the snowmelt period. Biological uptake of nitrate has
been documented for lakes in the North Cascades
during open water periods, which could contribute
to the low levels of nitrate found in western lakes
(Loranger and Brakke 1986). The potential contri-
bution of nitrate to lakewater acidity during the
spring or other periods cannot be inferred from the
fall sampling.
5.6.2 True Color, Turbidity, and Secchi Disk
Transparency
Lakes sampled during the WLS-I were generally
clearwater with low DOC (Section 5.4.6) and turbidity
(Table 5-18). Consequently, Secchi disk transparency
was very high. Lakes in California had very low true
color (Q4 = 3.5 PCU) and very low turbidity (Q4 =
0.4 NTU). Median Secchi disk transparency was
7.4 m and Qi (which corresponds to the Cu value
77
-------
for color and turbidity) was 4.2 m. True color was
similar among the three subregions in the Rockies,
with a somewhat higher Q4 value in the Southern
Rockies (21.6 PCU). Lakes in the Pacific Northwest
had lower true color values and, at the Q4 level,
higher Secchi disk transparency than the Rockies.
Southern Rockies (Qd = 234.5 /ueq L~1). Magnesium
concentrations in California were low (Q4 = 17.7
/ueq L"1). Lakes in the Pacific Northwest and Central
Rockies had intermediate levels of magnesium
relative to other western subregions (M = 25.8 and
34.2 /ueq L"1 and Q4 = 73.4 and 55.2 /ueq L"1
respectively).
5.6.3 Sodium, Potassium, and Magnesium
Values of sodium, potassium and magnesium were
generally quite low (Table 5-19). Although Qi values
for sodium were similar among all subregions, Q4
values ranged widely (33.6 to 91.7 /ueq L~1). The
lowest sodium concentrations were observed for
lakes in California. The highest median (33.8 /ueq L"1)
and Q4 (91.7 /ueq L"1) values were estimated for the
Southern Rockies.
Potassium concentrations varied little and were
lowest in lakes in California and the Pacific
Northwest. Slightly higher concentrations of potas-
sium were observed in the Southern Rockies where
the Q4 value was 21.8 /ueq L~1.
Magnesium concentrations were markedly variable
in the Northern Rockies (Qd = 234.1 /ueq L~1) and
5.6.4 Iron. Manganese, and Total Aluminum
Concentrations of iron, manganese and total
aluminum were generally low for most lakes and
did not vary markedly among subregions (Table
5-20). Estimated median values for iron were less
than 18 /ug L"1 for all subregions. The lowest median
value occurred in the Northern Rockies (9.7 /ug L"1).
The maximum estimated Q4 value for iron (66.9
/ug L"1) was observed in the Southern Rockies.
Manganese concentrations were also low; the Q4
value ranged from 5.3 /ug L"1 in the Central Rockies
to 17.4/ugL~1 in the Pacific Northwest. Median values
for total aluminum were estimated to be less than
or equal to 30 /ug L"1 for all subregions. The lowest
Qi and median values for total aluminum were
observed in the Central and Southern Rockies.
Table 5-18. Population Statistics for Secondary Variables (True Color, Turbidity and Secchi Disk Transparency), Western
Lake Survey - Phase I
California
Pacific Northwest
Northern Rockes
Central Rockies
Southern Rockies
West
4A
4B
4C
4D
4E
4
Qi
0.0
1.1
40
3.6
3.7
0.0
True Color
(PCU)
M
0.0 3.5
4.2
8.0
8.4
11.3
5.0
CU
0.1
10.1
12.9
17.9
21.6
13.4
Qi
0.2
0.2
0.2
0.3
0.4
0.2
Turbidity
(NTU)
M
0.4
0.3
0.3
0.6
0.7
0.4
Secchi Disk
Transparency
(m)
Q4
4.2
0.7
0.5
1.4
2.0
0.9
Qi
7.4
26
2.4
2.7
1.7
2.6
M
11.1
4.9
4.9
4.2
3.4
4.8
Cu
10.0
7.9
6.9
5.4
8.2
Qi = first quintile (20th percentile).
M = median (50th percentile).
Cu = fourth quintile (80th percentile).
Table 5-19. Population Statistics for Secondary Variables (Sodium, Potassium, and Magnesium), Western Lake Survey -
Phase I
1) Mg*2(Aieq L'1)
M
Cu
Qi
M
Q,
M
California
Pacific Northwest
Northern Rockes
Central Rockies
Southern Rockies
West
4A
4B
4C
4D
4E
4
10.1
17.2
11.7
12.4
15.9
12.4
18.7
31.0
24.8
20.8
33.8
23.9
33.6
57.5
50.2
38.6
91 7
49.2
2.7
2.6
2.7
5.2
3.5
29
3.7
4.8
5.2
7.5
9.0
5.6
7.6
10.6
14.8
12.4
21.8
12.8
3.5
11.1
9.3
18.5
21.4
8.9
7.2
25.8
36.8
34.2
91.5
26.4
17.7
73.4
243.4
55.2
255.9
96.7
Qi = first quintile (20th percentile).
M = median (50th percentile).
Cu = fourth quintile (80th percentile).
72
-------
5.5.5 Other Secondary Variables
Estimated concentrations of silica were lowest for
the Central Rockies, where the median concentration
was 1.37 mg L"1 (Table 5-21). Median Si02
concentrations in all subregions were less than or
equal to 3.14 mg L~1.
Very low concentrations of dissolved inorganic
carbon (DIG) were observed for lakes in California
and in the Central Rockies (Q4= 1.88 and 2.80 mg L"1,
respectively). Median values for DIG in the Northern
and Southern Rockies were only slightly higher than
those in the other subregions, but Q4 values were
much higher (1 2.73 and 1 2.04 mg L"1, respectively).
Chloride concentrations were lowest in California
and in the Northern Rockies. The highest Q4 values
were observed in the Pacific Northwest, which may
reflect the proximity of some lakes to the coast
(Sections 6.2.5 and 7.2). The lowest conductance
values were estimated for California (M = 8.8
/jS cm"1) and the highest for the Northern and
Southern Rockies (M = 25.1 and 37.1 fjS cm"1,
respectively).
Calculated bicarbonate concentrations were lowest
in California and highest in the Southern Rockies
(Table 5-21). The Pacific Northwest had the second
lowest estimated Qi value, whereas the Central
Rockies had the second lowest median (96.8 /ueq
5.7 Statistics for Lakes in Wilderness
Areas and National Parks
5. 7. 1 Wilderness Area and National Park Lakes
The estimated population statistics for ANC, Ca+2 and
S04~2 were examined for wilderness area lakes by
subregion and region (Table 5-22). Wilderness area
lakes were defined as lakes in Forest Service
Wilderness Areas, National Parks, and the Wind
River Roadless Area (Table 2-4). The same relative
differences between subregions described in Section
5.5 for the total target population of lakes were
observed in the data from wilderness area lakes.
Wilderness area lakes in all subregions had lower
ANC values at the median and the first and fourth
quintiles than did non-wilderness lakes. The
medians and quintiles for Ca+2 were also generally
lower in wilderness area lakes but the differences
Table 5-20. Population Statistics for Secondary Variables (Iron, Manganese, and Total Aluminum), Western Lake Survey
Phase I
California
Pacific Northwest
Northern Rockes
Central Rockies
Southern Rockies
West
4A
4B
4C
4D
4E
4
Qi
5.4
5.0
4.6
1.1
5.5
42
Fe
M
14.5
15.7
97
10.8
17.8
137
Cu
37.1
36.1
283
54.9
66.9
431
Qi
0.0
0.0
0.0
0.0
00
0.0
Mn
M
0.0
25
0.0
0.9
1 7
0.8
CU
11.0
17.4
6.6
53
11.4
9.9
Qi
14.4
18.4
16.5
7.5
8.1
13.1
Total Al
(A
-------
Table 5-22. Population Statistics for ANC, Calcium, and Sulfate in Wilderness Area" Lakes and Non-Wilderness Area Lakes,
Western Lake Survey - Phase I
Wilderness Area Lakes
California 4A
Pacific Northwest 4B
Northern Rockies 4C
Central Rockies 4D
Southern Rockies 4E
n
97
90
82
96
90
XV
N
2017
939
1061
1734
807
/
QI
30.6
34.8
54.8
59.1
74.4
<\NC (/jeq L
M
54.5
104.2
94.6
97.5
157.5
QA
106.3
207.8
418.4
173.3
445.3
(
QI
21.1
29.0
38.3
63.8
69.0
Za+z (peq I
M
40.6
74.3
65.4
90.1
121.0
--1)
Q4
82.2
206.9
351.8
141.0
330.4
QI
4.1
3.5
5.8
19.1
14.0
so4-2 (
M
6.9
12.8
10.0
24.2
24.1
:^eq L-1)
Q4
13.7
41.6
21.5
33.2
42.1
West
455
6559
46.1
91.4
198.8
35.3 71.7
169.7
5.8 15.4 32.2
Non-Wilderness Area Lakes
ANC (fieq L~1)
n"" N
California 4A 52 383
Pacific Northwest 4B 69 766
Northern Rockies 4C 61 1318
Central Rockies 4D 33 565
Southern Rockies 4E 49 801
West 4 264 3834
QI
71.8
68.9
106.3
82.4
303.6
94.4
M
163.0
171.5
387.3
140.6
668.2
282.7
n*** = number of wilderness lakes from which samples were
Q4
365.
351.
1227.
552.
1403.
1108.
obtained
9
5
4
5
5
2
Ca + 2 (fjeq L 1)
QI
36.6
50.0
63.4
72.6
220.1
66.6
M
91.4
109.2
336.5
112.3
590.9
201.4
Q4
200.1
205.1
807.5
387.5
1156.3
740.4
Q
1.
2
7.
16
28,
7
,
.4
.0
.9
.5
,0
.7
so4-2 1,
M
3.7
16.2
21.7
27.9
56.2
25.5
.eqL-1)
°4
11.1
47.7
39.1
38.6
198.6
68.8
ft = estimated number of wilderness lakes.
Q, = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
'Includes Forest Service wilderness areas, national parks, and the Wind River Roadless Area (Table 2-4)
were less than those for ANC. Sulfate values in
wilderness area lakes were lower than in non-
wilderness area lakes at the median and fourth
quintile in all subregions except California. Sulfate
was lower in non-wilderness area lakes at Qi values
in California, the Pacific Northwest, and the Central
Rockies. These observations indicate that wilderness
area lakes generally contain solute concentrations,
particularly ANC, in the lower end of the range
observed for all lakes in the West.
Another way to examine differences between
wilderness area and non-wilderness area lakes is
to superimpose cumulative frequency distributions
(Figures 5-9 through 5-11). Acid neutralizing
capacity and Ca+2 are consistently lower in wilder-
ness area than in non-wilderness area lakes except
in the Pacific Northwest. The differences are
particularly striking in the Rocky Mountain subre-
gions. Sulfate is consistently lower in wilderness
area than in non-wilderness area lakes only in the
Southern Rockies. In the Central Rocky Mountains
and the Pacific Northwest, little difference in S04~2
exists between wilderness area and non-wilderness
area lakes with low concentrations of S04 2- Sulfate
is consistently higher in wilderness area than in non-
wilderness area lakes in California, although the
differences are not great.
5.7.2 Subpopulations in Wilderness Areas and
National Parks
Acid neutralizing capacity, Ca+2and SO4~2data were
examined for 21 Subpopulations of lakes in
Wilderness Areas and National Parks (Figure 5-12).
The lowest Qi values for ANC were estimated for
the Tahoe area in California and the Washington
Middle Cascades in the Pacific Northwest (17.6 and
18.0 /ueq L"1, respectively). Other areas having QT
values for ANC < 30 /ueq L"1 were the Yosemite area
and the Southern Cascades in California, and the
Oregon Middle Cascades and Alpine Lakes
Wilderness in Washington. As indicated by the
interquintile difference, some areas were very
homogeneous, particularly the Tahoe and Yosemite
areas and the Sawtooth and Wasatch/Uinta
Mountains. Other areas, such as northwest Montana
and northwest Wyoming had large Qd values for
ANC. Similar results were observed for Ca+2. Sulfate
concentrations were highest and, in general, most
variable in areas of the Central and Southern
Rockies. In the Southern Cascades in California and
the Oregon Middle Cascades in the Pacific
Northwest, S04 2 concentrations were very low and
uniform. All areas of the Cascades in Washington
had higher and much more variable SO4~2
concentrations than did areas of the Cascades in
other states.
74
-------
Figure 5-9. Cumulative frequency distributions [F(x)J for
ANC (fjeq i"V in wilderness area ( ) and
non-wilderness area (• ) lakes in California
(4A), the Pacific Northwest (4B). the Northern
Rockies (4C). the Central Rockies (4D). and
the Southern Rockies (4E). Western Lake
Survey - Phase I.
Figure 5-10. Cumulative frequency distributions [F(x)J for
calcium (fjeq i~V in wilderness area ( )
and non-wilderness area (• ) lakes in
California (4A). the Pacific Northwest (4B),
the Northern Rockies (4C), the Central
Rockies (4D), and the Southern Rockies (4E),
Western Lake Survey - Phase I.
California (4Aj
— Wilderness
•••• Non-wilderness
Pacific Northwest (4B)
Wilderness
Non - wilderness
Northern Rockies (4C)
Wilderness
Non-wilderness
Central Rockies I4D)
Wilderness
Non - wilderness
Southern Rockies (4E)
Wilderness
Non-wilderness
200
400
600
7000
California (4A)
— Wilderness
•••• Non-wilderness
Pacific Northwest (4B)
Wilderness
Non - wilderness
Northern Rockies (4C)
Wilderness
Non-wilderness
Central Rockies (4D)
Wilderness
Non-wilderness
Southern Rockies (4E)
— Wilderness
Non - wilderness
200 400 600
Calcium (fjeg L~1)
800
1000
75
-------
Figure 5-11. Inverse cumulative frequency distributions
[1-F(x)] forsulfate (peg i"V /'" wilderness area
(• ) and non-wilderness area (• )
lakes in California (4A). the Pacific Northwest
(4B). the Northern Rockies (4C), the Central
Rockies I4D). and the Southern Rockies (4E),
Western Lake Survey - Phase I.
1.0
0.8 -
0.6 -
1-FM
0.4 -
0.2 -
1.0
0.8 -
0.6 -
1-FM
0.4
0.2 -
1.0
0.8 -
0.6 -
1-FM
0.4 -
0.2-
1.0
0.8 -
0.6 -
1-FM
0.4
0.2 -
1.0
0.8 -\
0.6 -
1-FM
0.4 -
0.2 -
0.0
California (4A>
— Wilderness
Non - wilderness
Pacific Northwest (4B)
Wilderness
Non-wilderness
Northern Rockies (4C)
Wilderness
Non-wilderness
Central Rockies (4D)
Wilderness
A/on - wilderness
Southern Rockies (4E)
Wilderness
Non-wilderness
—i—
20
—i—
40
—I—
60
—i—
80
100
Sulfate (peq L'
5.8 Sample Statistics for Special
Interest Lakes
Thirty-two special interest lakes, selected on the
basis of available historical data, were sampled
during the WLS-I using protocols identical to those
used for the probability sample lakes. Data from
these lakes were used in the data validation process
(Section 3.3). Population estimates using data from
these lakes were not calculated because the lakes
were not selected as part of the probability sample.
The data presented here are, therefore, sample
statistics (Table 5-23). The data for individual lakes
are listed in Volume II.
Special interest lakes were located at high elevations
(M = 2903 m). As a group, these lakes markedly varied
in watershed area (23 to 26571 ha) and lake area
(1.0 to 5397 ha). The largest lake was Crater Lake,
Oregon (ID number = 4B3-071), and the largest
watershed area was estimated for Fremont Lake,
Wyoming (4D3-076). Although the maximum site
depth (109.7 m) was recorded for Fallen Leaf Lake,
California (4A3-047), site depths for Crater Lake,
Oregon and Fremont Lake, Wyoming were too great
to be measured with the sampling equipment.
None of the special interest lakes were acidic
(minimum ANC = 23.2 /ueq L"1). The maximum ANC
(1100/ueq L'1) was measured in Lake Catherine, Utah
(4D3-079). All special interest lakes had pH
exceeding 6.3. Sulfate was generally low for most
lakes except Crater Lake, Oregon, where the value
was 221.9 fjeq L~\ Most special interest lakes had
low values for DOC, turbidity, and true color, and
consequently had high Secchi disk transparencies.
5.9 Statistics for Dilute Lakes
Throughout the West, many of the lakes sampled
had very low ionic strength. Because such low ionic
strengths can be indicative of low acid neutralizing
capacity, it is interesting to examine the physical
and chemical characteristics of these lakes. A
subpopulation of "dilute" lakes was defined as those
lakes in the target population having conductance
< 10 fjS cm"1. An estimated total of 2766 lakes in
the West had conductance < 10 /uS crrf1 (Figures
5-13 to 5-17). In contrast, the 20 percentile value
for conductance measured in the ELS-I was 20.7
/uS cm~1 (Linthurst et al. 1986). California contained
the highest percentage of dilute lakes (58.2%),
whereas only 7.2 percent of the lakes in the Southern
Rockies had conductance < 10 /uS cm"1.
Dilute lakes were more transparent than the target
population of lakes in each subregion, with the
exception of California (Table 5-24). Median Secchi
76
-------
Figure 5-12. Concentrations (ueq L~1J of ANC, calcium, and sulfate in lakes in selected wilderness areas and national parks.
Western Lake Survey - Phase I. Numbers in parentheses after each area name are those shown in Figures
2-2 and 2-3 and Table 2-4 in Section 2.2.3. First quintiles (Qt. 20th percentile). medians (M. 50th percentile).
and fourth quintiles fQ4, 80th percentile), are given for each variable ( Q 1 D ).
Qi M Qt
California (4A)
Klamath Mountains (1 -3)
Southern Cascades (4-7}
Tahoe Area (8-9)
YosemiteArea(10-13)
Southern Sierra Nevada (.14-18) -
Pacific Northwest (4B)
North Cascades Area (19-26)
Alpine Lakes (29)
Washington Cascades (30-34)
Oregon Cascades (35-42)
Northern Rockies (4C)
Northwest Montana (43-47)
Bitterroot Area (48-51)
Blue Mountains (52-53)
Sawtooth Mountains (54)
Central Rockies (4D)
Southern Montana (55-56)
Northwest Wyoming (58-62)
Wind River Range (63-66)
Wasatch/Uinta Mtns. (67-68)
Southern Rockies (4E)
Mount Zirkel (69)
Rocky Mountain Park (70-73)
Central Colorado (75-81)
S. Colorado/New Mexico (82-84) -
1 1 _T . _J
H-d
Ki
CH-Q
X-D
cH — n
0|Q £7
o .!..„_. . .__n
1 1 1 1 I
- T 1 1
M a
HH3
)-K3
n 1 n
— h- a
687
o-|-a
0 1 a 547
o+a
1 1 1
100 200 300 400
ANCfueqL'')
10 20 30 40 50
Sulfate (ueq L'^)
100 200 300
Calcium foeq L''1)
disk transparencies for dilute lakes in the Pacific
Northwest, and in the Rockies were more than one
meter greater than for the target population (Section
5.6.2). Dissolved organic carbon and true color were
also low, factors which also contributed to the high
transparencies.
Acid neutralizing capacity and Ca+z were lower in
the subpopulation of dilute lakes than in the total
population (Section 5.4.1 and 5.4.4), particularly in
the Northern and Southern Rockies. Median S04~2
concentrations in dilute lakes in California, the
Pacific Northwest and the Northern Rockies were
6.2//eqL"1orless, and were approximately 16//eqL~1
in the Central and Southern Rockies. For dilute lakes
in the West as a whole, the CU value for SO,T2 was
13.0 peq L"1. Concentrations of extractable Al were
very low in dilute lakes in the West (M = 3.3 /JQ L"1).
Dilute lakes were associated with somewhat higher
elevations than the target population lakes as a
whole. Dilute lakes in the Pacific Northwest occurred
at relatively lower elevations than in the other areas
(Table 5-24). Median elevations for dilute lakes in
California and in the Central and Southern Rockies
were greater than 2900 m. Watershed areas were
slightly smaller than those estimated for the entire
target population (Section 5.3), although lake areas
for the populations were comparable; consequently,
watershed to lake area ratios were less for dilute
lakes.
77
-------
Table 5-23. Sample Statistics for Physical and Chemical
Variables for Special Interest Lakes8,
Western Lake Survey - Phase I
Variable Units
Lake Elevation m
Watershed Area ha
Lake Area ha
Site Depth m
ANC neq L~1
PH
c\f*L —2 i —1
b(J* ueq L
i~ +2 i-1
La /yeq L
Al, ext. /^gL^1
DOC mg L"1
NOs" yueqL"1
NH4 jueq L ~
P, total pgL~1
True Color PCU
Turbidity NTU
Secchi disk
transparency m
Na+ f*eq L
K + /ueq L -1
Mg+2 /jeqL-1
Fe /jg L~1
Mn /ug L~1
Al, total /jg L~1
SiO2 mg L ~
DIG mg L~1
Cr /"eqL~1
F", total dissolved fjeq L~1
Conductance /jS cm"1
MINb
1424
23
1.0
2.4
23.2
6.36
1.4
10.5
0.0
0.24
0
0
0
0
0.1
1.6
5.6
1.5
4.9
0
0
3.5
0.16
0.46
0.8
0.2
3.4
Mc
2903
220
6.6
7.8
103.8
7.18
18.3
81.1
3.0
1.28
0.4
0
7.6
10
0.35
4.9
20.1
6.2
20.8
13.25
2.5
17.0
1.51
1.39
3.4
0.7
13.2
MAXd
3913
26571
5396.7
109.76
1100
8.73
221.9
675.8
9.4
4.28
21.4
7.1
41.4
25
5.85
15.2f
445.7
50.1
424.0
162.5
20
74.8
19.04
12.00
302.6
4.9
112.0
aSample size = 32.
bMIN = minimum (Oth percentile).
CM = median (50th percentile)
dMAX = maximum (100th percentile).
e Depth was undefined for Crater Lake, Oregon (4B3-071) and
Fremont Lake, Wyoming (4D3-076).
f Secchi disk transparency was undefined for Crater Lake,
Oregon.
-------
Figure 5-13. Locations of dilute lakes (conductance < lOftS cm'^ samples in California (4A), Western Lake Survey - Phase I.
Geographic area covered by the target population is bounded by a dashed line.
-------
Figure 5-14. Locations of dilute lakes (conductance < 10 /uS cm V sampled in the Pacific Northwest (4B). Western Lake
Survey - Phase I. Geographic area covered by the target population is bounded by a dashed line.
80
-------
Figure 5-15. Locations of dilute lakes (conductance < 10 (iS cm'*) sampled in the Northern Rockies (4C). Western Lake
Survey - Phase I. Geographic area covered by the target population is bounded by a dashed line.
81
-------
Figure 5-16. Locations of dilute lakes (conductance < 10 /jS cm V sampled in the Central Rockies (4D), Western Lake
Survey - Phase I. Geographic area covered by the target population is bounded by a dashed line.
82
-------
Figure 5-17. Locations of dilute lakes (conductance < 10 /jS cm'*) sampled in the Southern Rockies (4E), Western Lake
Survey - Phase I. Geographic area covered by the target population is bounded by a dashed line.
WY
*-\\ .' '' /
V I \
-'UU
83
-------
Table 5-24. Population Statistics for Chemical and Physical Characteristics of Dilute Lakes8, Western Lake Survey - Phase
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
West
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
West
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
West
4A
4B
4C
4D
4E
4
4A
4B
4C
4D
4E
4
4A
4B
4C
4D
4E
4
N
1398
452
564
235
116
2766
N
1398
452
564
235
116
2766
N
1398
452
564
235
116
2766
QI
26.2
19.4
34.4
35.6
30.1
26.9
QI
1.0
2.8
1.9
0
0.2
1.0
QI
2.2
2.1
2.1
3.3
2.0
2.2
ANC (neq L~1)
M
45.1
33.0
46.6
52.3
45.8
45.8
Ext. Al (^g L~1)b
M
3.0
6.5
4.1
1.0
1.6
3.3
Lake Area (ha)
M
4.7
3.4
3.5
5.2
4.6
4.1
C.«0«qL-i) S04-*0-qL-')
Q4
61.9
58.8
68.2
58.6
60.9
63.7
Q4
5.1
11.4
8.6
2.6
3.9
6.9
Q4
13.3
10.9
6.3
12.4
8.0
11.8
QI
18.7
10.6
25.9
34.7
38.6
19.8
QI
0.58
0.33
0.75
0.50
0.49
0.54
QI
45
28
21
44
29
36
M
32.0
22.9
32.2
44.8
50.6
33.2
DOC (mg L"1)
M
0.85
0.97
1.15
0.73
0.66
0.92
Watershed Area
M
110
91
49
112
95
94
Q4
43.0
44.7
48.8
54.5
67.4
47.9
Q4
1.76
1.75
1.61
1.05
0.91
1.66
(ha)
Q4
337
217
101
476
114
247
QI
3.9
1.8
5.2
12.7
11.2
4.1
Secchi
QI
4.5
3.8
3.6
3.6
3.4
4.2
QI
2350
1276
1989
2919
3272
1828
M
5.2
5.4
6.2
16.2
16.0
6.8
Q4
9.0
15.8
8.1
23.1
18.8
13.0
Disk Transparency (m)
M
7.6
6.0
7.0
5.5
6.2
7.0
Elevation (m)
M
2999
1600
2197
3131
3531
2676
Q4
11.1
11.0
9.8
7.3
6.2
10.4
Q4
3295
1763
2599
3376
3761
3241
N = estimated number of dilute lakes.
QI = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
aConductance < 10 ^S cm"1.
84
-------
Section 6
Associations Among Variables
6.1 pH and Acid Neutralizing Capacity
6.1.1 Comparison of pH Measurements
For each lake sampled, pH was measured five times
on three separate aliquots (Section 2.5). Because
pH is a non-conservative parameter, in situ mea-
surement is preferred, but even under optimal
conditions, in situ measurements are difficult to
perform with a high degree of accuracy. In situ
measurements were conducted by the helicopter
crews during the WLS-I, but portable pH meters were
not available to make in situ measurements on the
wilderness lakes sampled by the ground crews
(Section 2.5.1.2). Thus, the surrogate measurement
for in situ pH was the "closed system" pH made
on the syringe aliquot that was unexposed to the
atmosphere (Section 2.5.1.2). Results of the
comparison of all pH measurements for lakes
sampled by helicopter crews are shown in Table 6-1;
comparison between helicopter and ground sampled
lakes was discussed in Section 4.
Differences in the intercepts and slopes for regres-
sions of all pairs of measurements made in the field
and analytical laboratories were minor. As indicated
by the r2 values, the two open system measurements
of pH made by the analytical laboratories are
constant. The regression for air-equilibrated pH and
closed system pH had a relatively low r2 value and
a slope of 1.00, indicating that, upon equilibration
with the atmosphere, some lake samples gained CO2
while others lost COa-
All laboratory measurements showed relatively poor
agreement with the in situ pH measurements.
Although in situ measurements generally agreed
with closed system pH, the agreement among all
laboratory measurements was lower than that
observed in ELS-I (Linthurst et al. 1986). The
agreement among pH measurements (as indicated
by r2 values) was lower in the West compared to
that observed in the East and is probably attributable
to several factors. The range of pH measurements
in the West was much less than that observed in
the East. Thus, the regression equations for the East
and West represent different populations of lakes.
A direct comparison of pH measurements with
eastern and western lakes should be restricted to
the same range. The pH values commonly encoun-
tered in the West (usually greater than 6) also
showed greater variability among pH values in the
East. Additionally, the low conductance of western
lakes likely contributed to instability in the measure-
ment of pH (Brezinski 1983; Metcalf 1984).
6.1.2 Relationship Between pH and A cid
Neutralizing Capacity
At a given pH, ANC is a linear function of the sum
of carbonate species (Cj) where:
CT = [C03~2] + [HC03~ ] + [C02] + [H2CO3]
If CT is fixed by holding the partial pressure of CO2
constant, a unique value of ANC is defined for each
Table 6-1. Weighted Regression Statistics Comparing pH Measurements for Lakes Sampled Using Helicopters,
Western Lake Survey - Phase I
pH Pair3
Dependent/Independent Variable
Closed system/acidity titration
Closed system/alkalinity titration
Closed system/air equilibrated
Closed system/in situ
Acidity titration/alkalinity titration
Acidity titration/air equilibrated
Acidity titration/in situ
Alkalinity titration/air equilibrated
Alkalinity titration/in situ
Air equilibrated/in situ
"n = 41 1, except for in situ pairs for which n = 406;
Intercept
-0.21
-0.19
-0.14
1.64
0.02
-0.14
2.60
-0.17
2.63
3.89
variables are defined in Volume I
Slope
1.053
1.050
1.004
0.773
0.996
0.981
0.619
0.986
0.615
0.481
II, Tables 2-1 and 5-1.
r2
0.866
0.862
0.671
0.798
0.994
0.820
0.653
0.828
0.644
0.467
85
-------
pH. This condition can be achieved by equilibrating
the samples with a standard concentration of COa
in the atmosphere (referred to as air-equilibrated pH;
Section 2). The comparison of air-equilibrated pH
to ANC (Figure 6-1) shows that measurements of
these variables agree closely to the partial pressure
of C02 predicted on the basis of the standard air
used (10~35 atm). However, natural lake waters vary
widely in their partial pressure of COa resulting in
considerable scatter between closed system pH and
ANC (Figure 6-1). It is also likely that organic acids
contribute to deviations from the expected
relationship between air-equilibrated pH and ANC.
Most of the western lakes exhibit higher partial
pressures of COa (i.e., oversaturation) than predicted
on the basis of COa in the atmosphere (1CT35 atm),
as indicated by observations shown below the curve.
The lakes that were undersaturated plot above the
curve. The large variation in COa partial pressures
for western lakes makes it difficult to generalize on
patterns in the relationship between pH and ANC.
6.2 Major Cations and Anions
Although bicarbonate and Ca+2 are the principal ions
in most western lakes, a more thorough understand-
ing of lake chemistry can be achieved by evaluating
all major anions and cations collectively. Comparison
of major anions and cations is a common technique
for analyzing lake chemistry data. An assumption
in the use of these comparisons is that all major
anions and cations have been measured.
Figure 6-1. Relationship between closed system pH and measured ANC for lakes with pH < 9.0 and ANC < 1000 fjeq i"1
in all subregions. Western Lake Survey - Phase I. This same relationship is shown in the inset for air-equilibrated
pH.
8 -
Hi
2.
to
7 -
600 800 7000
ANC(ueqL^)
-200
200
400
ANC (fieq L"1)
I
600
I
800
1000
86
-------
6.2.1 Individual Example Lakes
The strength and focus of the Survey design is that
it allows specific subpopulations of lakes to be
described. However, differences in lake chemistry
among subregions also can be evaluated by
examining the chemistry of individual lakes that are
"typical" of a population. Although no lake can be
representative of other lakes with respect to all
chemical attributes, the relative concentrations of
anions and cations observed among individual lakes
can be used to highlight similar features among
lakes.
Associations among anions and cations can be
examined directly with ion bar charts. Ion bar charts
were developed for a single lake selected from each
of the five subregions (Figure 6-2). To standardize
the presentation, lakes were selected on the basis
of ANC and DOC values. Each lake selected had ANC
and DOC concentrations approximately equal to the
first quintile (20th percentile) values for the
respective subregion (Section 5.5, Table 5-16). The
concentrations of cations and anions in each lake
are expressed as percentages of the total measured
ionic concentration, which are shown adjacent to
the bar chart for each lake. Ions for which the
measured concentrations were generally less than
5 //eq L"1 (Fe+3, Mn+2, Al+n, NH4+, N03~) are not
included in this presentation.
All lakes shown exhibit a small anion deficit (defined
as the sum of cations - sum of anions), presumably
attributable to the presence of unmeasured organic
anions. When organic anions are estimated from
DOC and pH using the model by Oliver et al. (1 983),
much of the deficit is explained. The absolute
magnitude of this deficit is small (generally about
10 //eq L"1 in these lakes), but because the lake
waters are so dilute, particularly in California (4A)
and the Pacific Northwest (4B), the percentage of
the total anion concentration represented by organic
anions is moderate. An alternate explanation for the
anion deficit, i.e., analytical error, is discussed in
Section 6.2.7.
Bicarbonate is the dominant anion in the five
example lakes, followed by SO4~2 and then Cl~.
Despite a four-fold difference in the concentrations
of total ions among these example lakes, the
proportions of anions are similar, with the exception
of Champion Lake, Idaho (ID number = 4C2-047).
Unlike the other lakes shown here, in which F~
concentrations are low, this lake exhibits a high
concentration of F". Although F~ concentrations were
generally low in western lakes (95.9% of western
lakes had F~ < 10 //eq L"1), moderately high
concentrations were observed in the Northern (4C)
and Southern Rockies (4E) where 7.2% and 17.1%,
Figure 6-2. Ionic composition of one lake selected from
each of the five western subregions surveyed
during the Western Lake Survey - Phase I. Each
lake was selected on the basis of having
concentrations of ANC (ueq i"V and DOC
(mg L ~V approximately equal to the first quintile
(Ct\, 20th percentile) values estimated for the
subregional target population of lakes. Cations
and anions are expressed as percentages of the
total measured ionic equivalents (ueq L~^); ions
with concentrations < 5 ueq i~1 are not shown.
Champion Lake (4C2-047) is shown on the
USGS topographic map as an unnamed lake
and is listed as "No Name" in the data base
and lake lists. Volume II.
Total
Ionic
0 10 20 30 40 50 60 Equivalents
(ueq Z."V
Swamp Lake, CA
(4 A1-037)
Airplane Lake, WA +
(4B1-053)
Champion Lake, ID +
(4C2-047)
Weyman Lake, UT +
f4D 1-046)
Tobacco Lake, CO +
(4E3, 049)
67.7
108.3
144.5
181.8
280.2
Cations (+)
I Calcium
0 JO 20 30 40 50 60
Percent Ionic Equivalents
n inn nyi
j Magnesium Ba Potassium Kj Sodium
Anions (-)
In Bicarbonate
Sullate
| Chloride E3 Fluoride
respectively, had F > 10 yueq L \ No lakes with F
> 10 fjeq L"1 were sampled in California, and F'
exceeded this concentration in only one lake in the
Pacific Northwest.
Calcium comprises approximately one-half of the
total concentration of cations in the lakes shown
here. In contrast to lakes in the East, Mg+2 is a
relatively minor cation in these lakes, except in
Weyman Lake (ID number = 4D1 -046), Utah in which
Mg+2 is the second most abundant cation. In the other
four lakes, Na+ is the second most abundant cation.
The excess of Na+ relative to Cl~ suggests that
87
-------
watersheds, and not marine influences, are the
major source of Na+ to these lakes. Concentrations
of K+ were moderately high in Tobacco Lake,
Colorado (ID number = 4E3-049), exceeding even the
concentration of Mg+2.
6.2.2 Calcium and Magnesium
The relationship between Ca+2 and Mg+2 concentra-
tions in western lakes was highly variable (Figure
6-3). The regression coefficients (r , Table 6-2), were
similar to those observed for the Northeast. In all
subregions except the Northern Rockies, the lowest
r2 values are associated with the strata representing
the lowest alkalinity map class. This is consistent
with the low concentrations of Ca+2 and Mg+2
observed in the lakes of these strata. The slopes for
Ca+2 and Mg+2 are also similar to those observed
for the Northeast and Upper Midwest, except for
California where the slope approaches one. How-
ever, the large differences in slopes and intercepts
apparent among strata within the same subregion,
illustrate the difficulty in developing a model that
adequately describes a single relationship between
two variables applicable for the West as a whole.
6.2.3 Calcium and Sulfate
A weak positive relationship between Ca+2 and S04~2
was observed in the West (Table 6-3), suggesting
that some of the S04~2 is derived from mineral
sources. The sources of S04~2 could be weathering
of minerals such as sulfides or gypsum (Stumm and
Morgan 1981) or the deposition of wind-borne soils
containing gypsum. Gypsum could also be a source
of Ca+2 in addition to that derived from carbonates.
Hidy and Young (1986) cite the high correlation
coefficient between Ca+2 and S04~2 in wet deposition
(r = 0.92) as evidence that most of the S04~2 is derived
from wind-borne soils. Because Ca+2andSO4~2occur
in equivalent amounts in gypsum (CaS04-2H20) and
anhydrite (CaS04), the slope for this relationship in
precipitation should be one. However, SO4~2 occurs
in greater amounts than Ca+2, casting some doubt
on wind-borne gypsum as the primary source of
S04~2 to some of these lakes (Oppenheimer et al.
1985). Nevertheless, S04~2 concentrations show a
positive association with ANC (Figure 6-4) suggest-
ing that watershed sources of sulfate are important
to some lakes in the West. In all subregions, the
highest median S04~2 concentrations are observed
Figure 6-3. Relationship between calcium 500
(< 500 iieq L'1) and magnesium
(< 250 peg L''1). Western Lake
Survey - Phase I.
400 -
300 -
-------
Table 6-2. Regression Statistics for Calcium (< 500 /jeq L~1, Dependent) versus Magnesium
and Stratum, Western Lake Survey - Phase 1
Subregion/Stratum
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
4A1
4A2
4A3
4A
4B1
482
4B3
4B
4C1
4C2
4C3
4C
4D1
4D2
4D3
4D
4E1
4E2
4E3
4E
n
54
50
34
138
56
52
43
151
53
43
25
121
43
47
29
119
46
45
22
113
Intercept
53.0
33.3
36.6
51.0
40.4
55.7
89.4
48.4
21.4
53.8
91.5
68.3
62.5
14.7
52.9
35.7
49.5
75.5
55.4
56.8
SEa
12.9
8.7
19.3
12.2
16.2
31.6
5.4
12.0
33.2
9.0
16.4
25.9
11.1
21.8
25.4
Slope
1.870
1.335
1.462
1.424
1.585
1.370
1.675
1.897
3.277
2.313
1.333
1.557
0.773
2.673
1.975
2.088
1.761
1.705
2.057
2.004
(< 250 fjoq L"
SEa
0.706
0.282
0.292
0.529
0.338
0.415
0.397
0.291
0.350
0.317
0.421
0.345
0.167
0.355
0.333
0.344
1) by Subregion
r2
0.119
0.318
0.440
0.166
0.143
0.247
0.284
0.395
0.572
0.607
0.386
0.489
0.127
0.473
0.548
0.572
0.358
0.378
0.642
0.590
3SE = standard error.
Table 6-3. Regression Statistics for Calcium (< 500 fjeq L 1, Dependent) versus Sulfate (< 100 jueq L 1) by Subregion and
Stratum, Western Lake Survey - Phase I
Subregion/Stratum n Intercept SEa Slope SEa r2
California 4A1 54 38.8 5.6 1.339 0.091 0.807
4A2 50 38.3 9.8 3.781 1.108 0.195
4A3 37 77.4 16.6 3.333 0.698 0.395
4A 141 46.6 1.345 0.638
Pacific Northwest 4B1 57 47.4 9.8 1.529 0.416 0.197
4B2 52 58.7 15.8 2.067 0.520 0.240
4B3 43 77.3 17.0 3.016 0.323 0.680
4B 152 49.3 2.884 0.523
Northern Rockies 4C1 53 27.6 7.6 2.596 0.580 0.282
4C2 43 57.7 17.7 4.327 0.986 0.320
4C3 25 106.5 39.0 3.228 1.243 0.227
4C 121 68.6 3.894 0.305
Central Rockies 4D1 43 54.8 14.2 0.999 0.501 0.088
4D2 46 42.5 15.9 2.579 0.534 0.346
4D3 29 169.0 22.7 0.125 0.146 0.026
4D 118 106.1 0.242 0.050
Southern Rockies 4E1 46 25.9 9.5 2.289 0.268 0.624
4E2 45 120.0 20.9 1.113 0.362 0.180
4E3 22 134.7 22.5 1.023 0.280 0.400
4E 113 118.2 1.104 0.347
aSE = standard error.
-------
Figure 6-4. Median sulfate concentrations
(fjeq t"V for lakes in three
measured ANC classes for Cali-
fornia (4A), the Pacific North-
west (4B). the Northern Rock-
ies (4C), the Central Rockies
(4D), and the Southern Rockies
(4E), Western Lake Survey -
Phase I.
50
40
30
I
to
I 20
I
10
-
-
1
^
§
1
'//////////////////////////////A
r
Acid Neutralizing
Capacity (fjeq L)
D - 50
H] > 50-200
g > 200
\
—
7T.
|
$
^«
^
I
Y//////////////////////////////A
—
—
/
«•
^
//
V
//
^/
^.*
/£
°v
in the lakes in the highest ANC class. Only in the
Central Rockies, the subregion containing Yellow-
stone National Park, is there little difference in SC<4 '
concentrations among the ANC classes. Additional
discussion of SCV2 in the Pacific Northwest is
included in Section 7.2.
6.2.4 Acid Neutralizing Capacity Versus Base
Cations
In most surface waters, Ca+2 and Mg+z are the major
base cations and HCOs" is the dominant anion (Rodhe
1949). In many watersheds, Ca+2 and Mg+2 are the 6.2.5 Relative Abundance of Major Ions
approaching a slope of one and a zero intercept (Table
6-4). Lakes in the Central and Southern Rockies
showed a small apparent deficit of ANC relative to
the sum of Ca+2 and Mg+2 However, examination
of the standard errors from the strata for these two
subregions indicates that only the intercept for
stratum 4E1 is significantly less than zero. An
adjustment of the data for possible analytical bias
may eliminate the apparent ANC deficit suggested
in this analysis (Section 4.5.2; 4.6).
primary weathering products (Drever 1982), because
the weathering of calcite or dolomite results in the
production of alkalinity in proportion to Ca+2 and
Mg+2. Small amounts of calcite in watersheds may
determine weathering products and influence the
production of alkalinity (Drever and Hurcomb 1986).
The relationship between ANC and the sum of Ca+2
and Mg+2 is plotted in Figures 6-5 to 6-9. This
regression has been used to infer acidification based
on departures from a near zero to negative intercept
and a slope reduced from one (Aimer et al. 1978).
This relationship is indicative of carbonic acid
weathering and also forms the basis of the predictor
nomograph of Henriksen (1980), which is an
empirical model of acidification.
Lakes in the West generally showed a relationship
between ANC and the sum of Ca+2 and Mg+2
The relative importance of major cations and anions
among subregions can be examined by ranking the
ions in order of decreasing concentration (Table 6-5).
The relative abundances presented here are based
on population estimates for the first quintile (20th
percentile) value and the median concentration.
Calcium is the dominant cation in all subregions at
both the first quintile and median values. However,
the relative abundance of Na+ and Mg+2 in California
and the Pacific Northwest differs from that in the
Central and Southern Rockies.
The higher Na+ concentrations in the Pacific Coast
subregions cannot be attributed solely to marine
contributions of sea spray. Examining the relation-
ship between lake CI" concentration and distance
from the coast (Figure 6-10) suggests that the
distance in the West over which sea spray influences
90
-------
Figure 6-5. Relationship between mea-
suredANC (< 200 ueq L~') and
the sum of calcium and mag-
nesium (ueq L~1) for lakes in
California (4A). Western Lake
Survey - Phase I. The alkalinity
map class from which each
lake was selected is shown (O:
< 100 ueq i"1; A. 100-199
/jeq i~1; X. 200-400 ueq L~\-
Omernik and Griffith 1986).
200
160 -
120
-j
o-
80
40
Alkalinity Map
Class (ueq L'^)
D < 100
A 100 - 199
X 200- 400
40 80 120 160
Calcium + Magnesium (ueq L~^)
200
200
160 -
120-
80
40-
Alkalinity Map
Class (ueq L'^)
D <100
A 100 - 199
X 200-400
i i I l I I I
40 80 120
Calcium + Magnesium (ueq L'^)
Figure 6-6. Relationship between mea-
sured ANC (< 200 ueq L"1) and
the sum of calcium and mag-
nesium (ueq L' V for lakes in the
Pacific Northwest (4B),
Western Lake Survey - Phase /.
The alkalinity map class from
which each lake was selected
is shown fO: < 100 ueq L'1;
A. 100-199 ueq L~\- X. 200-
400 ueq L~\- Omernik and
Griffith 1986).
160
200
91
-------
Figure 6-7. Relationship between mea-
sured ANC (< 200 ueq O and
the sum of calcium and mag-
nesium (ueqL~1) forlakesin the
Northern Rockies (4C), West-
ern Lake Survey - Phase I. The
alkalinity map class from
which each lake was selected
is shown fC\: < 100 ueq L'\-
A. 100-199 ueq L'\- X: 200-
400 fjeg L'\- Omernik and
Griffith 1986).
200
160 -
120-
-j
I
40-
A Ika Unity Map
Class (ueq L'^J
n <'oo
A 700 - 199
X 200-400
40
i
80
120
160
200
Calcium + Magnesium (ueq L'1)
200
160 -
120 -
t>
I
80
40
Alkalinity Map
Class (ueq L''1)
D <700
A 700-733
X 200-400
Figure 6-8. Relationship between mea-
sured ANC (< 200 ueq L~*) and
the sum of calcium and mag-
nesium (ueq L' V for lakes in the
Central Rockies (4D). Western
Lake Survey - Phase I. The
alkalinity map class from
which each lake was selected
is shown (O: < 700 ueq i~V
A: 100-199 ueq i"1; X: 200-
400 ueq L~\' Omernik and
Griffith 1986).
I 1 I f I l
40 80 120
Calcium + Magnesium (ueq L''1)
92
160
200
-------
Figure 6-9. Relationship between mea- 200
suredANC (< 200 fjeq L~^) and
the sum of calcium and mag-
nesium (fieq L ' V for lakes in the
Southern Rockies (4E). West-
ern Lake Survey - Phase I. The
alkalinity map class from 160 -
which each lake was selected
is shown (O: < 100 iieq L~\-
A.- 100-199 iieq L~\- X: 200-
400 neq I"1; Omernik and
Griffith 1986).
120 -
t>
1
80
40
Alkalinity Map
Class (ueqL'1)
D <100
A 100 - 199
X 200 - 400
40 80 120
Calcium + Magnesium (ueqL'^)
160
200
Table 6-4. Regression Statistics for ANC (< 200 fjeq L~', Dependent) versus the Sum of Calcium and Magnesium (< 200
//eq L~1) by Subregion and Stratum, Western Lake Survey - Phase I
Subregion/Stratum
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
4A1
4A2
4A3
4A
4B1
4B2
4B3
4B
4C1
4C2
4C3
4C
4D1
4D2
4D3
4D
4E1
4E2
4E3
4E
n
49
44
20
113
53
41
16
110
52
34
12
98
41
38
13
92
40
26
10
76
Intercept
8.7
7.1
0.9
8.0
6.2
2.0
-2.7
3.8
9.3
11.3
-6.4
4.5
-3.2
-17.8
3.7
-9.9
-22.9
21.5
6.8
-11.4
SEa
4.5
4.2
4.0
5.2
9.0
19.1
3.4
6.1
9.1
9.3
10.6
15.2
6.4
12.8
18.3
Slope
0.968
1.131
1.339
1.020
0.978
0.997
1.054
0.995
0.943
1.046
1.109
1.046
0.858
0.993
0.870
0.935
1.089
1.102
0.869
1.007
SEa
0.068
0.060
0.059
0.058
0.088
0.150
0.047
0.068
0.090
0.089
0.082
0.127
0.069
0.108
0.147
r2
0.813
0.895
0.966
0.827
0.847
0.769
0.780
0.815
0.888
0.881
0.939
0.905
0.705
0.804
0.811
0.790
0.867
0.812
0.814
0.835
aSE = standard error.
93
-------
Table 6-5. Relative Abundance of Major Cations and Anions Within Subregions Based on Population Estimates of
Concentrations at the First Quintile (Qi) and Median (M) Values, Western Lake Survey - Phase I
Cations8
Anions8
Subregion
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
4D
4B
4C
4D
4E
Qi
Ca > Na > Mg ~ K
Ca > Na > Mg > K
Ca > Na ~ Mg > K
Ca > Mg > Na > K
Ca > Mg > Na > K
M
Ca>Na>Mg~K
Ca >Na>Mg>K
Ca>Mg>Na>K
Ca>Mg>Na>K
Ca >Mg >Na> K
Qi
HC03>S04~CI
HC03>CI ~SC>4
HC03>SO4~CI
HC03>SO4>CI
HCO3>S04>CI
M
HCO3>SO4~CI
HCO3>SO4~CI
HC03>S04>CI
HCO3>SO4>CI
HCO3>S04>CI
West
Ca > Na ~ Mg > K
Ca>Mg ~Na >K
HC03 > SC>4 ~ Cl
HCO3>S04>CI
'—indicates difference between estimates is <5/jeq L
lake chemistry is minimal beyond 50 km inland.
Brakke and Waddell (1985) found that beyond 50
km inland from Puget Sound, lake water Cl"
concentrations in the Pacific Northwest showed no
further decline with distance. The terrestrial
contribution of Na+to lakes in the Pacific Northwest
is evident from the plot of Na+ versus distance from
the coast (Figure 6-11). Sodium concentrations in
some lakes over 70 km from the coast exceed 50
//eq L"1. The source of Na+ in some western lakes
not influenced by sea spray has been attributed to
weathering products from plagioclase (Garrels and
Mackenzie 1967; Nelson 1985). Further evidence for
mineral sources of sodium to some lakes is presented
in the relationship between Na* and Cl" in the Pacific
Northwest. The observed ratio departs from that
expected on the basis of the ratio of these ions in
seawater (Figure 6-12).
In the Northern Rockies, Na+ is more abundant than
Mg+2 at the first quintile, which is also consistent
with weathering of plagioclase minerals. Most of the
low ionic strength lakes in the Northern Rockies are
located over the Idaho Batholith, the granitic
composition of which is similar to the batholith
underlying the Sierra Nevada (Hunt 1974). For lakes
with high total ionic concentration, Mg+2 is more
abundant than Na+ because the lakes appear to be
located in areas underlain, in part, by ultramafic
rocks (Hunt 1974).
The relative abundance of major anions is the same
in all subregions except in the Pacific Northwest,
where Cl" was more abundant than SO4'2. In
Northern Washington, the dominant source of Cl"
is sea salt, whereas in the Oregon Cascades
relatively high amounts of Cl" are derived from
weathering (Nelson 1985). The uniform relative
abundance of major anions and cations observed in
the West differs considerably from that observed in
the East where marked differences within, and
among, regions were observed (Linthurst et al.
1986).
6.2.6 Relationships Among Major ions
The ratios of major anions and cations are presented
as trilinear plots (ternary diagrams) in Figures 6-13
to 6-17. Like the ion bar charts (Figure 6-2), these
plots include only those ions for which the measured
concentrations were typically 5 /ueq L"1 or greater.
The axes on the plots represent the ion or pair of
ions expressed as percent of total ionic concentra-
tion. The values increase in the direction of the
arrows from 0 to 100 percent. Points representing
the percent ionic composition are plotted for cations
on one triangle, for anions on the second triangle,
and for the pairs of cations and anions on the
parallelogram. Thus, a lake for which all of the total
ionic concentration is due to dissolved CaCOa would
appear in the extreme lower left portion of each
triangle and would appear in the left corner of the
parallelogram. A lake acidified by sulfuric acid would
be expected to appear near the apex of the
parallelogram. Although these trilinear plots do not
show differences in concentrations or include other
ions of possible importance, they are useful for
displaying patterns among the major ions.
The ion ratios were relatively uniform among the
five western subregions. In nearly all cases, the lakes
can be described as calcium bicarbonate systems
(i.e., Ca+2 and HCOs" comprise over 50% of the
cations and anions, respectively). Exceptions to this
pattern, indicated as circled points on the plots, were
noted for each subregion and are described below.
These outliers were subjectively identified based on
separation from the predominant pattern for a given
subregion.
The selected outliers for lakes in California display
an interesting geographic split: the outlier lakes on
the cation plot were located in either the California
Cascades (4A3-011 and 4A3-013 in Lassen National
Park, and 4A2-005 and 4A2-060 in Caribou
Wilderness Area) or in the Klamath Mountains (4A3-
94
-------
'ۥ?
«i|l
•S o ^ *
*i^
sis-
,<0
ll.ll!
lilt!
l» 1
\ \ \
8
^§
-8!
i
I
40
nc
_ o
8 ••1
EteSl
O J *4
"••S€*
t8|^
•* .g < Q.
a o "5 o '«
1
-------
Figure 6-11.
Concentration of sodium
f/jeq L~^) versus distance
(< 100 km) from Puget
Sound or the Pacific Ocean
(whichever is less) for lakes
in five geomorphic units in
the state of Washington in
the Pacific Northwest (4B).
Western Lake Survey -
Phase I. Values are shown for
the Wenatchee Mountains
(9). the Middle Washington
Cascades ( &.) Cascades( +),
the Puget Lowlands fX), and
the Olympic Mountains (O).
Distance from the coast was
determined from a SAS file
(SAS Institute 1985).
100
90
80
70
60
^ 50 ~
b-
I: 40 -
|
I 30 ~
20 -
10
Pacific Northwest (4B)
B Wenatchee Mountains
A Middle Washington Cascades
+ North Washington Cascades
X Puget Lowlands
Q Olympic Mountains
\i "
•. a •
)
10
]
20
]
30
\
40
SO
60
\
70
\
80
90 100
Distance from Coast (km)
007, 4A3-058 and 4A3-060). These lakes contain
a relatively high proportion of Mg+2, which is
abundant in the ultramafic rocks found in these
areas. The two outliers on the anion plot, attributable
to a high proportion of SO4~2, were both found near
Yosemite National Park. Kidney Lake (4A1-014) had
a high concentration of SO4~2 (93 jueq L"1). Lake 4A1 -
005 (No Name) was extremely dilute (conductance
= 4.0 pS cm"1) and, with a SC>4~2 concentration of
only 10 fjeq L~1, SO4~2 was the dominant anion.
The outlier lakes in the Pacific Northwest had
different chemistries than the outlier lakes in
California. Trout Lake (4B1 -031) has high ANC (579
/jeq L"1) and is located in the Middle Washington
Cascades. The high magnesium concentration (401
,ueq L"1) was attributable to the watershed bedrock
containing dunite. The high [Na+ + K+] ratios in five
lakes located in the Oregon Cascades cannot be
explained by sea spray. Four of these lakes (4B1-
004, 4B1-005, 4B2-004, and 4B2-005), located in
or near the Diamond Peak Wilderness Area, were
very dilute (conductance = 2.2 to 4.5 //S cm"1) with
only moderate concentrations of Na+ (10 to 20
yueq L"1). Only Hosmer Lake (4B1-011) located on
the eastern slope of the Oregon Cascades near
Waldo Lake Wilderness Area, contained high Na+
(219/jeqL"1).
The Northern Rockies contained three outliers, all
located in the Selway-Bitterroot Wilderness Area.
The three lakes were dilute (conductance = 5.5 to
9.5 fjS cm"1) and had relatively low proportions of
Na+ (Jeanette Lake, 4C1 -015) or SO4"2(Holloway and
Little Carlton Lakes, 4C1 -010 and 4C1 -011).
Four lakes in the Central Rockies had atypical ion
ratios. Three of these (4D3-016, 4D3-017, and 4D3-
052) are located in Yellowstone National Park. Lake
4D1-013 was a dilute lake (conductance = 7.2
(iS cm"1) located in the Bridger Wilderness Area in
the Wind River Range. The three lakes in Yellowstone
identified as outliers on the trilinear plot were very
different from one another. Fern Lake (4D3-017) was
the only acidic lake (ANC = -24 ^eq L"1) sampled
in the West. It was affected by hot springs, and the
S04~2 concentration was 818 yueq L~1. Nymph Lake
(4D3-016) also had a high S04~2concentration (2909
jueq L"1), but had an ANC value of 1341 /jeq L~\
In contrast, Goose Lake (4D3-052) was a sodium
bicarbonate lake with low S04 2(30^eq L"1).
Several lakes in the Southern Rockies were outliers
on the anion plot. Black Lake #2 (4E3-022), located
south of the Eagles Nest Wilderness Area, was very
alkaline (pH = 9.13, ANC = 682 yueq L"1). The
remaining outlier lakes, located in the San Juan
96
-------
<0
I
O
OC
1
I
i
-
« S!
€.5
1
10
<*
10
10
t\l
O-
01
I
g
8
o
O)
o
03
o Q
10 >
1 barl) uinipos
3.
0-8
«
45*
II
•c .u
w-5
O K-
o ^
yjl
I
1
-c
I
•
«0
CM
I
_ o
to
s
I
o o
00 IS
o o
-------
Figure 6-13. Trilinear diagram showing the
relative abundance of major
anions and cations for lakes
in California (4A), Western
Lake Survey - Phase I. Ratios
are expressed as percent of
total ionic concentration,
increasing from 0 to 100
percent along the axes in the
direction of the arrows. Lakes
for which the total ionic com-
position cannot be described
as being dominated by cal-
cium bicarbonate are circled
and their lake ID numbers are
given.
4A2-005
4A3-060
4A3-011
4A3-013
4A3-058—I
4A1-014
4A1 -005
//\\
4B'1-031
4B'1-005
Figure 6-14. Trilinear diagram showing the
relative abundance of major
anions and cations for lakes
in the Pacific Northwest (4B).
Western Lake Survey -
Phase I. Ratios are expressed
as percent of total ionic con-
centration, increasing from 0
to 100 percent along the axes
in the direction of the arrows.
Lakes for which the total ionic
composition cannot be des-
cribed as being dominated by
calcium bicarbonate are
circled and their lake ID
numbers are given.
\4B1-004
4B2-004
\ 4B2-005
4B1-011
98
-------
Figure 6-15. Trilinear diagram showing the
relative abundance of major
anions and cations for lakes
in the Northern Rockies (4C).
Western Lake Survey -
Phase I. Ratios are expressed
as percent of total ionic con-
centration, increasing from 0
to 100 percent along the axes
in the direction of the arrows.
Lakes for which the total ionic
composition cannot be de-
scribed as being dominated by
calcium bicarbonate are
circled and their lake ID
numbers are given.
//\\
4C1-010
4C1-011
4C1-01S
Figure 6-16.
//\\
4D3-046
4D3-052
'403-016
4D3-017
' 4D3-OS2
Trilinear diagram showing the
relative abundance of major
anions and cations for lakes
in the Central Rockies (4D).
Western Lake Survey -
Phase I. Ratios are expressed
as percent of total ionic con-
centration, increasing from 0
to 100 percent along the axes
in the direction of the arrows.
Lakes for which the total ionic
composition cannot be de-
scribed as being dominated by
calcium bicarbonate are
circled and their lake ID
numbers are given.
99
-------
453-053
4E2-040
4E2-059
Figure 6-17.
4E3-053
4E3-040
4E2-059
4E1-040
4E1-026
4E1-022
Tfilinear diagram showing the
relative abundance of major
anions and cations for lakes
in the Southern Rockies (4E).
Western Lake Survey -
Phase I. Ratios are expressed
as percent of total ionic con-
centration, increasing from 0
to 100 percent along the axes
in the direction of the arrows.
Lakes for which the total ionic
composition cannot be de-
scribed as being dominated by
calcium bicarbonate are
circled and their lake ID
numbers are given.
Ca
Cl
Mountains, had high relative concentrations of
sulfate. With the exception of Columbine Lake, (4E2-
040, ANC = 34 /ueq L"1) these lakes had relatively
high ANC (> 138 //eg L"1). The high S04~2 concen-
trations may have been derived from the Tertiary
volcanic bedrock or sulfide deposits in the area
(Section 7.5).
6.2.7 Dissolved Organic Carbon
6.2.7.1 Dissolved Organic Carbon and True
Color—Water passing through plant materials
leaches complex organic compounds that are
resistant to decomposition. These organic com-
pounds often contain functional groups that impart
a brown color to water (Gjessing 1976). Because
of this property, color has often been used as a
surrogate for dissolved organic carbon (DOC).
Regression equations were used to compare the
relationships between dissolved organic carbon and
true color for lakes in each stratum and subregion
(Table 6-6). As expected, the variance explained by
the regression equations is low due to the small
range of DOC and color in western lakes (Figure
6-18).
6.2.7.2 Anton Deficit—The principle of electro-
neutrality states that the sum of cation equivalents
should equal the sum of anion equivalents, assuming
all contributing ions are measured accurately. In
addition to serving as an analytical tool for quality
assurance checks (Sections 3 and 4), evaluation of
ion balances provides additional insight into regional
lake chemistry.
There is a close agreement between the sum of
anions and the sum of cations among all subregions
of the West (Figure 6-19) with a minimum r2 of 0.983
in the Southern Rockies (Table 6-7). In most
subregions the intercept was different from zero and
the slope was significantly less than one. Although
attenuation in the slope, caused by errors in the X
axis, would contribute to a small reduction in the
slope, the magnitude of the deviation from 1.0
suggests that the anion deficit may be related to
factors other than organic anions.
Many of the lakes sampled had anion deficits, i.e.,
the sum of measured anions was less than the sum
of measured cations. Assuming no analytical bias,
anion deficit reflects the presence of unmeasured
organic anions. Table 6-8 shows population esti-
mates of the first and fourth quintiles (Qi and Q4)
and the median for anion deficit by subregion. As
expected, due to the low color and high transparency
in western lakes, the magnitude of the anion deficit
was very low. The lowest anion deficits were
observed for California, whereas the Southern
Rockies showed the greatest anion deficit. The
magnitude of the anion deficit among subregions
was roughly proportional to the total ionic concen-
100
-------
Table 6-6.
Regression Statistics for DOC (< 10 mg L'\ Dependent) versus True Color (< 50 PCU) by Subregion and Stratum,
Western Lake Survey - Phase I
Subregion/Stratum
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
4A1
4A2
4A3
4A
4B1
4B2
4B3
4B
4C1
4C2
4C3
4C
4D1
4D2
4D3
4D
4E1
4E2
4E3
4E
n
54
51
41
146
58
53
46
157
53
50
38
141
43
47
35
125
46
51
40
137
Intercept
0.8
1.5
1.6
1.0
1.2
1.1
1.0
0.9
0.1
0.2
0.2
0.119
0.2
0.4
0.5
0.3
0.2
0.3
0.2
0.2
SEa
0.1
0.2
0.2
0.2
0.2
0.3
0.2
0.3
0.5
0.2
0.2
0.3
0.1
0.2
0.4
Slope
0.143
0.193
0.062
0.148
-0.021
0.091
0.120
0.103
0.125
0.147
0.192
0.181
0.123
0.141
0.120
0.138
0.103
0.124
0.158
0.152
SEa
0.018
0.022
0.017
0.031
0.020
0.021
0.016
0.029
0.042
0.022
0.016
0.020
0.010
0.019
0.023
r2
0.558
0.611
0.250
0.506
0.008
0.290
0.406
0.290
0.552
0.345
0.362
0.360
0.432
0.618
0.593
0.613
0.698
0.465
0.559
0.560
aSE = standard error.
Figure 6-18. Relationship between DOC
(< 10 mg L~') and true color
(< 50 PCU) for lakes in all
subregions. Western Lake
Survey - Phase I.
10
8~
'~j 5
2~
• i
-
10
20 30
Color (PCU)
1
40
50
101
-------
Figure 6-19. Relationship between the sum of anions (< ZOO peg i~V and the sum of cations (< 200 fjeq L ^ for lakes
in California ( • ). the Pacific Northwest ( A ). the Northern Rockies ( + ), the Central Rockies ( X ). and
the Southern Rockies fO), Western Lake Survey - Phase I. Points falling below the 1:1 (diagonal) line indicate
an anion deficit, and above the line, a cation deficit. Inset shows the same relationship for anions ^ 2000
fjeq i'1 and cations < 2000 peq L~\
200
-j
o-
<
o
180 _
160 _
140
120 _
100 _
80 _
60 _
40 -]
20 _
2000
1000
• California (4A)
A Pacific Northwest (4B)
4- Northern Rockies (4C)
^ Central Rockies (4DI
E Southern Rockies I4E)
20
\
40
\
60
80
Sum of Anions (fjeq i"
I I I
100 120 140
160
180 200
102
-------
Table 6-7. Regression Statistics for the Sum of Major Anions (< 2000 /Jeq L~1, Dependent) versus the Sum of Major Cations'
(< 2000 A»q L~1) by Subregion and Stratum, Western Lake Survey - Phase I
Subregion/Stratum
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
4A1
4A2
4A3
4A
4B1
4B2
4B3
4B
4C1
4C2
4C3
4C
4D1
4D2
4D3
4D
4E1
4E2
4E3
4E
n
54
53
41
148
59
53
45
157
53
50
38
141
43
46
33
122
46
51
35
132
Intercept
-9.4
-9.3
-8.0
-5.0
-13.6
-4.6
3.4
-3.4
-2.2
-6.5
-2.9
-3.8
8.5
-18.1
8.2
7.7
-5.8
0.6
-8.9
-7.34
SEb
2.5
2.2
4.3
2.6
2.7
6.0
1.8
3.2
14.4
6.3
7.1
18.5
5.5
11.7
22.4
Slope
0.989
0.912
0.908
0.922
0.982
0.885
0.882
0.895
0.879
0.914
0.902
0.904
0.848
0.964
0.828
0.832
0.925
0.872
0.906
0.904
SEb
0.017
0.009
0.008
0.013
0.011
0.011
0.018
0.008
0.017
0.045
0.036
0.028
0.033
0.030
0.028
r2
0.985
0.995
0.997
0.992
0.990
0.992
0.993
0.993
0.980
0.996
0.988
0.991
0.897
0.942
0.966
0.972
0.946
0.947
0.970
0.973
aOnly anion and cations with individual concentrations > 5 j*eq L 1 are included in the sums.
bSE = standard error; not calculated for subregions (Section 5.1.2.1).
Table 6-8. Population Statistics for Anion Deficit,
Western Lake Survey - Phase I
Anion Deficit (^eq L""1)
Subregion
M
California 4A
Pacific Northwest 4B
Northern Rockies 4C
Central Rockies 4D
Southern Rockies 4E
4.5
7.0
9.6
5.4
10.2
10.2
22.0
32.4
19.6
38.6
22.2
40.5
82.4
38.7
132.3
West
6.9
19.6
51.1
Q, = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
tration. This observation is consistent with the
possibility that some of the anion deficit can be
attributed to analytical error (Sections 4.5 and 4.6).
It can be inferred that the anion deficit results from
the presence of unmeasured organic anions if a
relationship between anion deficit and DOC can be
established. To evaluate further the presence of
unmeasured anions among subregions, anion deficit
was plotted and regressed on DOC (Figure 6-20,
Table 6-9). The regression results indicate a poor
relationship between DOC and anion deficit in the
West. This poor fit can be attributed, in part, to the
low concentrations of DOC and total ions observed
in the West. However, it also suggests that the anion
deficit in the West may be attributable to an analytical
error. Substituting ANC for HCOa , computed using
measured DIG, in the computation of the sum of
anions reduces the anion deficit by approximately
70 percent. Other studies of low ANC lakes in the
West report close agreement (i.e., approximately 5%
difference) between the sums of anions and cations
(Baron 1983; Brakke and Loranger 1986; Nelson and
Delwiche 1983; Duncan 1985; Meglen et al. 1985;
Melack et al. 1985). The results of these studies
further support the contention that the relatively
small anion deficit observed here is an artifact of
a small analytical error in the measurement of a
major cation (e.g., Ca+2) or anion (e.g., HC03" which
is computed from DIG).
6.3 Geology
Early investigations of the sensitivity of aquatic
resources to acidic deposition implied a relationship
between the composition of bedrock and the ANC
of surface water (Galloway and Cowling 1978). This
relationship can be affected by surficial geology
(Shilts 1981), hydrologic flow paths (Chen et al.
1984), soil type (McFee 1980), or vegetation (Krug
and Frink 1983). These factors contribute to the lack
of a good relationship between bedrock geology and
water chemistry in eastern North America. These
factors would be expected to exert less influence
on lakes in the mountainous western United States,
because soil development is generally poor, glacial
till is often absent, and bedrock exposures are
common.
103
-------
Figure 6-20. Relationship between DOC
(mg L~^) and anion deficit 10
l/jeq L'^j for lakes in all sub-
regions with total measured
ionic equivalents < 500 9
fjeq L~\ Western Lake
Survey -Phase I. Anion deficit
is the difference (fjeq i"V 8
between the sum of mea-
sured anions and the sum of
measured cations. 7
I
3 ~
2 -
1 -
t
.%'•
I .
v **.•:**•*"
»* *
.
10
20
I
40
l
50
30
Anion Deficit {peq L''1
i
60
70
80 90 100
Table 6-9. Regression Statistics for DOC (< 10 mg L~\ Dependent) versus Anion Deficit !
and Stratum, Western Lake Survey - Phase I
Subregion/Stratum
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
4A1
4A2
4A3
4A
4B1
4B2
4B3
4B
4C1
4C2
4C3
4C
4D1
4D2
4D3
4D
4E1
4E2
4E3
4E
n
54
51
37
142
58
53
44
155
53
47
31
131
43
47
27
117
46
48
26
120
Intercept
0.5
1.4
1.6
0.7
0.6
1.3
1.9
1.1
0.6
1.1
1.6
1.3
0.8
1.8
1.9
1.7
1.0
1.2
2.0
1.9
SEa
0.1
0.4
0.3
0.2
0.3
0.5
0.2
0.2
0.6
0.2
0.4
0.3
0.2
0.2
0.2
Slope
0.052
0.042
0.008
0.044
0.031
0.015
0.008
0.021
0.044
0.013
0.013
0.017
0.029
0.019
0.001
0.002
0.023
0.007
-0.009
-0.008
[< 100 yueq L'1
SEa
0.009
0.014
0.007
0.008
0.008
0.011
0.010
0.006
0.013
0.009
0.012
0.002
0.008
0.006
0.003
) by Subregion
r2
0.416
0.164
0.039
0.298
0.191
0.060
0.011
0.099
0.267
0.104
0.031
0.063
0.193
0.058
0.012
0.005
0.158
0.032
0.210
0.147
°SE = standard error; not calculated for subregions (Section 5.1.2.1)
104
-------
The relationship between bedrock geology and
lakewater chemistry was evaluated by comparing
WLS-I lake chemistry to bedrock class (Norton et al.
1982). The bedrock classification scheme used by
Norton et al. (1982) is as follows:
Class 1—Low to no acid neutralizing capacity
(granitic gneiss, quartz sandstones or
equivalent).
Class 2—Medium to low acid neutralizing capac-
ity (sandstones, shales, conglomerates,
or their metamorphic equivalents, high-
grade metamorphic felsic to interme-
diate volcanic rocks, intermediate
igneous rocks, calc-silicate gneisses
with no free carbonate phases).
Class 3—High to medium acid neutralizing
capacity (slightly calcareous rocks, low-
grade intermediate to mafic volcanic
rocks, ultramafic rocks, glassy volcanic
rocks).
Class 4—"Infinite" acid neutralizing capacity
(highly fossiliferous sediments or meta-
morphic equivalents, limestones or
dolostones).
Class 5—Covered by glacial debris or Quaternary
alluvial material which obscures the
bedrock.
These bedrock classifications were based on recent
state geologic maps generally of 1:250,000-scale or
1:500,000-scale. The bedrock classifications were
developed by Norton et al. (1982) to evaluate
regional-scale sensitivity to acidification. For this
comparison, a class was assigned to each lake based
on the dominant bedrock class in which the lake
basin was located. Presumably, lake chemistry
represents the integrated contribution of all rock
types within a watershed, and it is possible that rock
types (e.g., calcites or dolomites) which may
represent a relatively small proportion of the total
watershed bedrock could have a dominant influence
on the lake chemistry. Thus, a formal test of the
utility of these maps would require that the class
assigned to a lake be based on the synthesis of all
bedrock classes within the total watershed, not just
within the lake basin as presented here. Neverthe-
less, the bedrock classifications for several subre-
gions are sufficiently homogenous to be acceptable
for this application.
Population estimates for the proportion of low ANC
(<50//eq L~1) lakes by bedrock class show reasonably
good agreement between the potential sensitivity of
the lake as indicated by ANC and bedrock class
(Figure 6-21). In most subregionsthe lakes in bedrock
Class 1 (low to no acid neutralizing capacity) have
the highest proportion of low ANC lakes, whereas
lakes in the less sensitive bedrock classes showed
low proportions of low ANC lakes. This pattern
showed inconsistencies in California and the Central
Rockies. This may be an artifact of the criterion used
here to define low ANC lakes. Nearly all of the Wind
River Range was in bedrock Class 1, yet a small
percentage of lakes in this area (6.2%) had ANC <
50 //eq L"1. However, over 89 percent of the lakes
in the Wind Rivers had ANC < 200 //eq L"1. Using
a different criterion to define sensitive bedrock types
(such as ANC < 200 //eq L~1) would result in a
consistent ANC-bedrock pattern in the Central
Rockies. The pattern for Ca+2 (< 50 //eq L"1) was
similar to that presented for ANC, whereas little
relationship to the bedrock classes was observed for
conductance and SiOa.
The importance of evaluating the influence of
bedrock type within subregions is apparent from the
large differences in the proportion of low ANC lakes
observed among subregions within the same bedrock
class. The higher proportion of low ANC lakes in
California and the Pacific Northwest, compared to
the Central and Southern Rockies on similar bedrock
classes, likely reflects additional factors, such as
precipitation and evaporation (Section 6.6).
6.4 Physical Lake Characteristics and
Water Chemistry
To examine more closely the relationship between
physical and chemical lake characteristics in the
West, the subregions were divided into geomorphic
units. These units represent physiographic areas of
common geologic origin, the boundaries of which
generally coincide with the mountain ranges in the
West (Figure 6-22). Evidence for common geologic
history (Hunt 1974; Chronic 1986) and topographic
features were used to delineate geomorphic units,
which generally follow those in Snead (1980).
Although population estimates were computed for
some geomorphic units in which 10 or more lakes
were sampled, extensive descriptions are given only
for those in which the sample size was at least 25.
Qualitative descriptions are presented for those
subpopulations based on a sample size of 10 to 25.
Population estimates for selected geomorphic units
are presented in Section 6.4; the complete results
for geomorphic units are discussed in Section 7.
6.4.1 Lake Area
The major distinction between the target population
of lakes defined for the WLS-I and that defined for
the ELS-I is that the population of western lakes
included those < 4 ha (but > 1 ha) in size (Section
2.2.1), whereas the ELS-I sample lakes were
105
-------
Figure 6-2 J. Percentage of low ANC (< 50
fieq i~V lakes by bedrock class
{Norton et al. 1982) in Cali-
fornia (4A). the Pacific
Northwest (4B), the Northern
Rockies (4C), the Central
Rockies (4D), and the South-
ern Rockies (4E). Western
Lake Survey - Phase I. Bed-
rock sensitivity is highest in
Class 1 and lowest in Class 3.
50
7 40
of
0
10
VI
o
«; 30
^
1
in
•5 20
1
10
0
—
-
-
— I
-
__
^
N
s
1
^
1
I
W////Z7////7///77////////?.
\
, :
Y/&/&////Z&//A
Bedrock Class
D (Highest
i Sensitivity)
D-
H! 3
g^i
J
I
-
"/////////////////////////A
-
\ \ \ \ \ \
California Pacific Northern Central Southern West
Northwest Rockies Rockies Rockies
restricted to those > 4 ha (Linthurst et a\. 1986).
Small lakes (those > "\ and < 4 ha) comprise a
significant percentage of the western lake resource,
representing 43.8 percent of the target population.
The relationship between lake area and water
chemistry in western lakes was evaluated to examine
the physical and chemical characteristics of low ANC
lakes in the West and to determine if population
estimates in the WLS-I needed to be restricted to
lakes > 4 ha for later comparison to eastern lakes.
In this analysis, large lakes are defined as > 30 ha
and small lakes as < 30 ha.
Lake area and ANC, SO*'2, and Ca+2 showed little
relationship, as indicated by the correlation coeffi-
cients (Table 6-10). The lack of a strong relationship
between ANC and lake area is apparent for the Pacific
Northwest (Figure 6-23). However, because there are
relatively few large lakes (> 30 ha) in the Pacific
Northwest, no pattern can confidently be deter-
mined. For lakes < 30 ha, which represent 92.4
percent of the target population for the West, no
relationship between lake chemistry and lake area
was observed. However, western lakes > 30 ha had
higher concentrations of ANC compared to those
lakes < 30 ha (Figure 6-24).
The impact of including lakes < 4 ha in the population
estimates for the WLS-I was evaluated by comparing
quintiles (Qi and Q4) and median values for ANC,
Ca+2, and S04~2 for lakes > 4 ha and lakes < 4 ha
(Figure 6-25). Only in the Northern Rockies did lakes
< 4 ha have substantially lower concentrations of
ANC, Ca+2 and S04~2 than lakes > 4 ha. Cumulative
frequency distributions for ANC in four subregions
showed little difference between lakes > 4 ha and
<4 ha; only in the Northern Rockies was there strong
indication that the chemistry of these two groups
of lakes differ (Figure 6-26). Within the Northern
Rockies, a high percentage of small lakes are located
in the geomorphic units overlying the Idaho Batholith
(Bitterroot, Sawtooth, Clearwater, and Salmon River
Mountains). This area also contains most of the low
ANC lakes within the Northern Rockies. Thus, the
apparent positive relationship between ANC and lake
area within the Northern Rockies is found to be
spurious in this case when the analysis is focused
on the geomorphic unit. Because the small lakes
in the Northern Rockies represent only 10 percent
of the estimated number of lakes in the western
target population, their impact on the regional
population estimates was minimal.
In summary, lakes with areas > 1 and < 4 ha and
lakes > 4 ha in four subregions of the West showed
no appreciable differences in ANC, Ca+2, and S04~2.
A substantial chemical difference between large and
small lakes was observed only in the Northern
706
-------
Figure 6-22.
Locations and names of geomorphic units used in data analysis for specific subpopulations of lakes. Western
Lake Survey - Phase I, These units represent physiographic areas of common geologic origin and generally
coincide with major mountain ranges.
Olympic
Uplift
North
Washington
Cascades
4C
Selkirk Cabinet
Mountains Mountains
Puget
Lowlands
Klamath
Mountains
Middle
Washington
Cascades
Wenatchee
Mountains
South I
Bitterroot
Range \
Anaconda-Pintlar
Mountains Beartooth
Washington ,,lAwater
Cascades __,w_A_^tns-
Salmon River
Mountains
Sawtooth
Mountains
Medicine Bow
Range
4E
Front
Range
Sawatch Uplift
Sangre De Cristo
Uplift
Rockies, where the difference was apparently caused
by a spatial separation of large and small lakes into
areas with contrasting bedrock characteristics.
Finally, the number of lakes with area < 4 ha in
the Northern Rockies is insufficient to seriously
influence the regional population estimates for the
West. On the basis of this analysis, there appears
to be little reason to eliminate the small lakes from
the western population estimates when comparing
them to eastern lakes, despite the difference in the
sampling frames for the two surveys.
6.4.2 Lake Elevation
An obvious distinction between lakes in the East and
lakes in the West is that the elevation of western
lakes was generally high and exhibited substantial
subregional variation. Based on previously collected
data, it was assumed that most low ANC lakes in
the West would be located at high elevation sites
(Omernik and Griffith 1986; Turk 1983; Gibson et
al. 1983). In the West, an inverse relationship
between ANC and elevation was found in the Central
107
-------
Table 6-10. Spearman Correlation Coefficients (ra) for Lake
Area versus ANC, Calcium, and Sulfate by
Subregion and Stratum, Western Lake
Survey - Phase I
Correlation Coefficient
Subregion
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
Stratum
4A1
4A2
4A3
4B1
4B2
4B3
4C1
4C2
4C3
4D1
4D2
4D3
4E1
4E2
4E3
n
54
53
42
59
53
47
53
50
40
43
47
39
46
52
41
ANC
-0.011
0.060
0.216
0.016
0.216
0.023
0.051
0.230
0.330
0.001
0.081
-0.080
-0.338
0.041
0.021
Ca+2
0.019
0.133
0.221
0.065
0.162
-0.129
0.000
0.256
0.342
0.108
0.060
-0.054
-0.292
0.056
0.009
so4-2
-0.019
0.376
0.199
0.206
0.132
-0.022
0.033
0.321
0.246
-0.307
-0.087
-0.041
-0.068
0.091
-0.075
3 Regression was performed using weighted values of variables,
i.e., the values are population estimates (Section 5.1.2.1).
Rockies and in California, but the relationship was
very weak or absent in the Pacific Northwest, the
Northern Rockies, and the Southern Rockies (Figures
6-27 to 6-31). The absence of a strong relationship
between ANC and lake elevation in the Northern
Cascades also has been reported by Brakke and
Loranger (1986).
Rank correlation coefficients within geomorphic
units indicate a somewhat stronger degree of
association between ANC and elevation than is
observed within subregions (Table 6-11), although
the relationship is still weak. In the Park Range and
the Uinta Mountains, the presence of one outlier
in each unit greatly influenced the correlation
coefficient. In the Middle Washington Cascades, the
positive slope and the correlation coefficient suggest
that lake ANC increases with elevation. This
relationship again is observed to be spurious because
of the presence of two high ANC lakes at high
elevation. Only in the Wind River Range was the
decreasing relationship between ANC and elevation
substantially demonstrated.
The relationship between ANC and lake elevation
was explored further by examining high elevation
Figure 6-23. Relationship between ANC
{< 1000 peg L~V and lake area
(< 100 ha) for lakes in the
Pacific Northwest (4B),
Western Lake Survey - Phase I.
The alkalinity map class from
which each lake was selected
is shown fD: < 100 fjeg L'\-
A: 100-199 tieq L~\ X 200-
400 [Jeq i~1; Omernik and
Griffith 1986).
^
'fj
Cr
O
900-
800 _
700-
600-
500-
400.
300-
200-
1OO -
0
Alkalinity Map
Class foeq L~1J
n
-------
Figure 6-24. Estimated number of lakes with ANC < 50
lieq i'1 or ANC > 50 peg i"' by lake area
class. Western Lake Survey - Phase I.
25
15
vi
6000
5000
* 4000
-J
^ 3000
"2 2000 -
I
<2 rooo
Acid Neutralizing
Capacity (peq L'^j
n >5°
<50
0 5 10 15 20 25 30
Lake Area Class (ha)
lakes, defined as those at elevations exceeding 3000
m. An elevation of 3000 m generally corresponds
to timberline, although timberline varies from
approximately 1800 m in the Olympic Mountains of
Washington to over 3600 m in the Sangre de Cristo
Mountains of Colorado because of temperature,
wind, and other factors (Arno and Hammerly 1984).
No lakes > 3000 m were sampled in the Pacific
Northwest or in the Northern Rockies. In California
and the Central Rockies, high elevation lakes have
slightly lower concentrations of ANC and Ca*2 than
lakes at low elevations (Table 6-12). In the Southern
Rockies, however, high elevation lakes have far
lower concentrations than those at low elevations.
The minimum elevation of lakes in the Southern
Rockies is 2100 m; thus, it is surprising that an
elevational gradient of 900 m would result in such
large differences in ANC, Ca+2, and S04'2.
A previous example (Section 6.4) showed that the
apparent relationship between lake area and water
chemistry in the Northern Rockies was attributed,
in part, to spatial separation of large and small lakes
into geomorphic units with different bedrock types.
To illustrate confounding orographic effects in
evaluating the relationship between ANC and
elevation, the lakes in the Sierra Nevada and Oregon
Cascades were grouped as those located east and
west of the topographic divides for the respective
ranges. Both mountain ranges have similar bedrock
types across the width of the units. Population
estimates for ANC, Ca+2, and S04"2 concentrations
for lakes in the eastern Sierra Nevada were much
higher than for lakes on the western slope (Table
6-13). In contrast, the chemistry of the lakes across
the Oregon Cascades is homogeneous, as illustrated
for ANC (Figure 6-32). These observations are
consistent with the pronounced "rain shadow" effect
on the eastern slope of the Sierra Nevada indicated
by a 3000-m crest throughout much of the mountain
range. The crest of the Oregon Cascades is
approximately 1500 m less than that of the Sierra
Nevada, with only isolated volcanic peaks approach-
ing the elevation of the Sierra Nevada crest.
Consequently, the orographic effects on ANC in lakes
in the Sierra Nevada are striking, whereas little effect
is observed in the Oregon Cascades.
6.4.3 Hydrology
6.4.3.1 Hydrologic Lake Type—A qualitative
assessment of the influence of hydrology on the
chemistry of lakes was performed by categorizing
lakes on the basis of the presence or absence of
permanent inlets and outlets. The objective of this
classification was to distinguish lakes that receive
substantial hydrologic contributions from
watersheds from those that do not. Greater contact
time with neutralizing processes in watersheds can
increase the ANC of inflowing water. Thus, in lakes
for which the hydrologic contribution is primarily
through the watershed, rather than as a direct
surface input, lake ANC concentrations would be
expected to be higher. For the same reason, higher
ANC concentrations would also be expected in lakes
for which the watershed area to lake area ratios are
large. The presence of permanent inlets and outlets
was determined by examining large-scale USGS
topographic maps. Lakes were classified (Wetzel
1983) as seepage (no inlets, no outlets), drainage
(outlet), closed (inlets, no outlets), and reservoir
(outlet control structure present).
For the western United States, the chemistry of
drainage lakes and seepage lakes was similar (Table
6-14). In California and the Pacific Northwest,
drainage lakes had higher estimated concentrations
of ANC and Ca+2 than did seepage lakes. In the three
Rocky Mountain subregions, the inverse was true.
Sulfate and DOC concentrations in both drainage
and seepage lakes were low, but drainage lakes had
slightly higher S04~2 and lower DOC concentrations
109
-------
Figure 6-25. Concentrations f/jeq L'^) of ANC. sulfate and calcium in lakes <4 ha ( ) and > 4 ha ( •
) in California
7200 -
7000 -
500 -
600 -
400 -
200 -
(4A). the Pacific Northwest (48). the Northern Rockies (4C). the Central Rockies (4DJ, the Southern Rockies
(4£), ^and the West (4) in the Western Lake Survey - Phase I. First quintiles (Qi. 20th percentile) medians
(M, 50th percentile). and fourth quintiles (Q4. 80th percentile) are given for each variable (.'
-
-
.
-
.
•
_
c
9
_ 1
1
99 ;•
"xi 2
w ,
c
9
] .
1
i
9
0
«t"
9
1
1 LI
', | i
.
-
.
-
fl
, fo JJ ~° i,
— ii ii <
i i i
1
»
720
700
80
&
CD
^ 60
u
1
40
20
n
~ <4ha '
' Q, M Q4 (p
>4/J8
o | a
' Q, M Q4
-
-
-
9 o
1C
-
"~ n
p
?J- T,
YY A
p
9
1
1
1
!
*
iD
i I
•M
1 I
.1 .
Q*1^ O
p _
-
^
-
—
-
-
~
9n-
-
• > —
•f"
6
» A* i°
soo
700
600
f" 500
o-
f 400
.§
(0
i
1 — l-l
]
ft
9
T
T
TjT J^
i AO ^^
1 M 1-
p
9
c
. T
^ 1
6<
-
-
-
«.
-
M
P
—
-
—
1 "
4A 4B 4C 4D 4E 4
4A 4B 4C 4D 4E 4
4A 4B 4C 4D 4£ 4
than did seepage lakes. Lack of major chemical
distinctions between drainage and seepage lakes in
the West contrasts with the large chemical differ-
ences between these lake types observed in the East
(Linthurst et al. 1986). This may result from the
highly seasonal nature of surface input into western
lakes; consequently, intermittent runoff, although
high in total volume from the deep snowpack, would
not be identified as a permanent inlet on a map.
Also, meltwater passing through highly porous talus
slopes would not be identified as an inlet. Thus, using
the presence of stream inlets from mapped infor-
mation may have limited utility in the West. Except
in the Pacific Northwest, reservoirs had relatively
high concentrations of DOC and SOT2. These
elevated concentrations in reservoirs are consistent
with large contributions from the watersheds.
The Pacific Coast subregions receive far greater
amounts of precipitation and experience lower rates
of evaporation than do the Rocky Mountain
subregions (U.S. Department of Commerce 1974).
For lakes with equal watershed areas and similar
lake morphometries, higher rates of precipitation
result in greater dilution. Thus, the combined effects
of greater precipitation and lower rates of evapo-
transpiration might account for the lower concen-
trations of chemical constituents observed in the
Pacific Coast subregions. Differences in precipitation
and evaporation also occur on the slopes of the Sierra
Nevada, which are reflected in higher values of ANC
on the eastern slopes (Section 6.5).
6.4.3.2 Hydraulic Residence Time—Use of
hydrologic lake types is a qualitative approach for
evaluating the relationship between lake chemistry
and hydrology. A more quantitative approach is to
compute hydraulic residence time. Hydraulic
residence time is defined as the time required to
exchange the total volume of water in a lake:
Hydraulic Residence Time =
Lake Volume/Total Volume of Inflow per Year
A large, deep lake with a relatively small watershed
would be expected to have a long (e.g., > 3 yr)
hydraulic residence time, whereas a shallow lake
with a large watershed would be expected to have
a short (e.g., < 0.5 yr) hydraulic residence time.
110
-------
Figure 6-26. Cumulative frequency distributions [F(x>] for ANC (< 1000 peg L V *nd calcium (< 1000 fjeq L V- and inverse
cumulative frequency distributions [1-F(x)] for sulfate (< 100 fjeq L~^) in lakes < 4 ha ( ) and > 4 ha ( }
in the Northern Rockies (4C), Western Lake Survey - Phase I. Other subregions showed no substantial differences.
FM
1.0
0.8-
FM 0.6 -
0.4-
0.2-
0.0
1.0
0.8 -
0.4 -
0.2 _
0.0
200
I
400
I
600
1
800
1000
ANC (fjeq
<4ha
^4 ha
200
\
400
I
500
500
WOO
Calcium (fjeq L'^)
1.0 .
0.8 _
1-FM 0.6 _
0.4 _
0.2 _
0.0
20
I i
40 60
Sulfate (fjeq L'')
<4ha
> 4 ha
I
80
100
111
-------
Hydraulic residence time was approximated using
the following expression:
RT = LA (0.464 ZS)/[RO (WA - LA) +
LA (PRECIP)]
where:
RT = hydraulic residence time (yr)
LA = lake area (ha), measured from
USGS topographic maps
Zs = site depth (m), measured in the
lake
RO = runoff (m/yr), interpolated from
national runoff maps
WA = watershed area (ha), measured
from USGS topographic maps
PRECIP = precipitation (m/yr), interpo-
lated from National Weather
Service data.
Lake volume was computed by multiplying lake area
by mean depth, the latter of which was approximated
by multiplying measured site depth by 0.464 (Wetzel
1983). This constant (0.464) for relating mean depth
to maximum depth varies with the shape of a lake
basin and may deviate considerably for individual
lakes. Runoff was estimated for each lake's
catchment using linear interpolation from mapped
contours of mean annual runoff (Busby 1966).
Precipitation at each lake was interpolated from
precipitation maps (Environmental Data Service
1983). Some of the assumptions and potential errors
in estimating hydraulic residence time with this
method are discussed in Linthurst et al. (1986). An
additional source of error in computing hydraulic
residence times, especially for western lakes, is the
assumption that surface runoff completely mixes
with water present in the lake. This assumption is
violated where snowmelt runoff flows through the
lake to the outlet without appreciably mixing with
the denser lake waters. This might be expected to
occur more commonly in deep lakes with relatively
small surface areas. Lakes responding in this manner
will have a longer hydraulic residence time than
indicated by the estimate calculated here. Another
problem with use of mapped values for estimating
hydraulic residence time is that orographic effects
can cause high spatial variability in precipitation
amounts.
The population estimates for hydraulic residence
time among subregions (Table 6-15) are similar,
except for the Pacific Northwest. In this subregion,
which experiences high total precipitation and
relatively low evapotranspiration rates, hydraulic
residence time is approximately 50 percent of those
in other subregions. Although the estimated
hydraulic residence times for California and the
Figure 6-27. Relationship between ANC
(< 1000 iteq L"1! and eleva-
tion (m) for lakes in California
(4A). Western Lake Survey -
Phase I. The alkalinity map
class from which each lake
was selected is shown ( D:
< 100 peg L'\- A; 100-199
Vieq L"; X200-400 fjeq r1;
Omernik and Griffith 1986).
300.
800-
700
600
500 _
300 _
200-
700 _
0
Alkalinity Map
Class ffjeq L'^j
n < 100
A 100 - 199
X 200-400
i i i i \ \ i
500 1000 1500 2000 2500 3000 3500 4000
Elevation (m)
112
-------
Figure 6-28. Relationship between ANC
(< 1000 fjeq i~V *nd eleva-
tion (m) for lakes in the Pacific
Northwest (4B). Western
Lake Survey - Phase I. The
alkalinity map class from
which each lake was selected
is shown ( D: < 100 fjeq L'1;
A. 100-199 fjeqL~\ X: 200-
400 peq i~1; Omernik and
Griffith 1986).
^
ANC (fjeq L'
1000
900 -
800-
700-
600 _
500-
400 _
300 _
200 _
700 _
0
Alkalinity Map
Class (fjeq L''1)
D < 700
A 700 - 799
X 200-400
„
a n
o
x y
a D
X
ax x
A
4x i a
& °x& C? fi * ^! a
* x D *x
°°° ^rf,^" Dc
i i 1 t i 1 1
0 500 7000 7500 2000 2500 3000 3500 4000
Elevation (m)
1000
900 _
SOO _
700 _
600
T-J 500
o-
O 400
300 _
200 _
700
0
Alkalinity Map
Class (fjeq L'^)
D <700
A 700-799
X 200 - 400
P*°£D D
]Bn «^*
Figure 6-29. Relationship between ANC
(< 1000 iieq O and eleva-
tion (m) for the Northern
Rockies <4C). Western Lake
Survey - Phase I. The alkalin-
ity map class from which each
lake was selected is shown
(O: < 700 fieq L~\- A: 700-
199 fjeq L'1; X. 200-400
fjeq i"1; Omernik and Griffith
1986).
0 500 1000 1500 2000 2500 3000 3500 4000
Elevation (m)
113
-------
Figure 6-30. Relationship between ANC
(< 1000 /jeq i"V and eleva-
tion (m) for the Central Rock-
ies <4D). Western Lake
Survey - Phase I. The alkalin-
ity map class from which each
lake was selected is shown
(O: < 100 neq L~'; A: 100-
199 fjeg L'\- X: 200-400
fieq i"1; Omernik and Griffith
1986).
1000
900.
800-
700.
600-
500.
400_
200-,
100
0
Alkalinity Map
Class (ueqL^)
D <100
A 100 - 199
X 200 - 400
i i i i i i r
500 7000 7500 2000 2500 3000 3500 4000
Elevation (m)
'-J
0-
J
900-
800-
700-
600-
50O-
400-
300.
200
100
0
Alkalinity Map 5
Class tueq O
D <700
A 100 - 199
X 200-400
\ '
X
x a
fl a
^ A
D x° fl
&
A X 6 y
X6
jp D a ^a x
" ° D^ »
i i i i i i i
Figure 6-31.
Relationship between ANC
(< 1000 neq L'^> and eleva-
tion (m) for the Southern
Rockies (4E). Western Lake
Survey - Phase I. The alkalin-
ity map class from which each
lake was selected is shown
( Q < 100 peq L~\- A: 100-
199 veq L'\- X: 200-400
peg L~\- Omernik and Griffith
1986).
500 1000 1500 2000 2500 3000 3500 4000
Elevation (m)
114
-------
Table 6-11. Weighted Correlation Coefficients for ANC versus Lake Elevation by Geomorphic Unit, Western Lake Survey
Phase I
Subregion
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
Geomorphic Unit
4A Sierra Nevada
4B Oregon Cascades
South Washington Cascades
North Cascades
Wenatchee Mountains
Middle Washington Cascades
North Washington Cascades
4C Idaho Batholith
Salmon River Mountains
Sawtooth Mountains
Bitterroot Range
Clearwater Mountains
4D Wind River Range
Uinta Arch
4E Park Range
Front Range
San Juan Mountains
n***
114
42
20
68
32
24
12
88
22
20
37
g
47
27
21
49
21
Correlation
Coefficient
(r)
-0.232
-0.298
-0.053
-0.104
-0.242
0.469
-0.280
0.156
0.302
0.159
-0.395
0.230
-0.582
-0.448
-0.922
-0.429
-0.586
Table 6-12. Population Statistics for ANC, Calcium, and Sulfate in High (> 3000 m) and Low (< 3000 m) Elevation Lakes
in California (4A). the Central Rockies (4D), and the Southern Rockies (4E), Western Lake Survey -
Phase I
ANC (^eq L~1)
Ca
+2
Elevation Category
M
M
cx»
M
California
> 3000 m
< 3000 m
(4A)
Central Rockies (4D)
> 3000 m
< 3000 m
Southern Rockies (4E)
38
111
73
56
1215
1186
1363
936
35.9
29.6
59.7
88.5
54.5
65.6
92.5
166.9
106.8
204.5
128.8
464.5
29.4
19.9
61.4
89.1
43.0
43.2
83.3
140.4
85.1
126.6
112.0
333.9
4.9
2.3
20.1
15.7
7.3
4.6
23.9
26.3
26.7
10.5
30.9
38.7
> 3000 m 118 1110
< 3000m 21 499
n*"*
4V
N
QI
M
Q4
= number of lakes from which samples
= estimated number of lakes.
= first quintile (20th percentile).
= median (50th percentile).
= fourth quintile (80th percentile).
76.5 174.0 567.0 70.0 141.2 410.7 14.0 27.3 47.4
449.0 1116.4 1673.8 386.3 810.6 1235.4 29.9 78.6 206.3
were obtained.
115
-------
Table 6-13. Population Statistics for ANC, Calcium, and Sulfate in Lakes on the Western and Eastern Slopes of the Sierra Nevada
and Oregon Cascades, Western Lake Survey - Phase I
nrionta+irtrt
Mountain Range of Slope n*** N Q,
Sierra Nevada
West 88 1697 28
Oregon
n*** =
N
Q-i =
M =
Q4 =
Figure
East 26 422 86
Cascades
West 22 231 38
East 14 148 30
number of lakes from which samples were
estimated number of lakes.
first quintile (20th percentile).
median (50th percentile).
fourth quintile (80th percentile).
6-32. Relationship between ANC
(< 500 iieq i~1J and elevation
(m) for lakes on the western
and eastern slopes of the
Sierra Nevada in California
(4A) and Oregon Cascades in
the Pacific Northwest (4B),
Western Lake Survey -
Phase 1.
ANC (/jeq L""1) Ca+2 (jieq L~1) SO4~2 (j*eq L~1)
M Q4 a, M 0.4 Q, M Q4
9 52.2 94.4 20.5 36.0 64.5 4.0 6.4 10
9 138.4 289.6 66.2 116.9 378.0 5.3 26.0 91
9 91.5 166.0 17.2 48.9 85.5 0.9 1.8 4
4 66.3 209.4 8.0 30.4 88.5 0.4 1.8 3
obtained.
^nrt Sierra Nevada
West Slope fast Slope
400 . .
,- 300 ' .
-j
I " "
•2: 200 •
* ' . . •
<* \-
100 •,•.•••/•
' ' \ .\'-':4r: • ••' ' •
o •«••••
7000 2000 3000 4000 7000 2000 3000 4000
Elevation (m)
Oregon Cascades
hOO
West Slope East Slope
^400
J1300
^~ •
^ • .
^200 .
. • *
n
500 7500 2500 500 7500 2500
Elevation (m)
116
-------
Table 6-14. Population Medians for Selected Parameters by Hydrologic Lake Type, Western Lake Survey Phase - I
Subregion
California (4A)
Pacific Northwest (4B)
Northern Rockies (4C)
Central Rockies (4D)
Southern Rockies (4E)
West (4)
Lake Type
Drainage
Seepage
Reservoir
Closed
Drainage
Seepage
Reservoir
Closed
Drainage
Seepage
Reservoir
Closed
Drainage
Seepage
Reservoir
Closed
Drainage
Seepage
Reservoir
Closed
Drainage
Seepage
Reservoir
Closed
n**
107
27
14
1
115
35
7
2
118
18
4
3
88
31
7
3
104
21
13
1
532
132
45
10
N
2013
282
97
8
1224
380
80
22
1916
292
92
80
1592
551
108
48
945
322
337
4
7691
1826
714
163
WA:LAa
23.1
15.2
24.1
_ b
20.9
10.8
—
—
20.4
26.0
—
—
32.6
20.0
—
—
24.1
13.2
91.2
—
24.0
16.6
75.7
24.2
Site Depth
(m)
9.1
6.5
12.6
—
8.7
4.0
—
—
9.4
6.1
—
—
8.6
7.8
—
—
5.3
4.3
5.0
—
8.7
5.8
7.7
10.4
Lake Area
(ha)
4.7
2.7
37.3
-
4.6
3.2
—
-
4.6
3.1
—
-
5.5
3.4
—
-
3.1
4.1
5.1
-
4.6
3.4
15.6
9.0
ANC
61.8
54.1
142.9
-
148.5
79.6
—
-
178.5
271.0
—
-
98.4
128.9
—
-
180.9
201.6
1051.6
-
108.1
113.2
554.1
196.0
DOC
(mg L"1)
1.82
1.74
1.28
-
1.13
2.09
—
-
1.15
1.12
—
-
1.09
1.60
—
-
1.06
2.19
3.55
-
1.06
1.84
2.53
1.85
Ca+2l
43.4
36.4
86.6
-
103.9
41.0
—
-
146.2
209.1
—
-
89.7
111.4
—
-
178.2
141.2
722.0
-
88.3
70.4
422.0
97.4
so4-2
6.7
4.8
6.3
17.1
5.6
—
-
13.5
12.3
—
-
24.3
21.8
—
-
28.7
24.5
77.8
-
17.8
16.6
49.6
17.2
n*** = number of lakes from which samples were obtained.
N = estimated number of lakes.
"WA:LA = Ratio of watershed area to lake area
b(—) Insufficient sample size, n*** < 10.
Table 6-15. Estimated Hydraulic Residence Time (Years)
for Drainage Lakes and Reservoirs by Subre-
gion, Western Lake Survey - Phase I
Hydraulic
Residence Time (yr)
Subregion
California
Pacific Northwest
Northern Rockies
Central Rockies
Southern Rockies
4A
4B
4C
4D
4E
n***
121
122
122
95
117
N
2110
1304
2007
1700
1282
Qi
0.10
0.05
0.13
0.06
0.06
M
0.29
0.16
0.42
0.29
0.34
Qi
0.97
0.57
1.33
1 16
1.02
West
577 8403 0.08 0.29 1 12
n*** = number of lakes from which samples were obtained
N = estimated number of lakes.
Qi = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile)
Southern Rockies lakes are similar, the greater depth
of the California lakes compensates for the higher
precipitation received compared to lakes in the
Southern Rockies.
The relatively short residence times for lakes in the
West compared to those in the East (Linthurst et
al. 1986) have important implications relative to
potential impacts from atmospheric deposition.
Lakes in the West likely receive most of their annual
inflow during snowmelt. Replacement of the lake
volume in a short period of time would minimize
the opportunity for the runoff to be neutralized by
contact with the catchment or through in-lake
processes. Thus, a high volume of snowmelt inflow
over bedrock that is highly resistant to weathering
would explain why such a high proportion of western
lakes are so dilute.
7/7
-------
Section 7
Subregional Characteristics
7.1 California (4A)
California (Subregion 4A) contained the highest
percentage (36.7%) and the highest number (880)
of western lakes with ANC < 50 //eq L~1 (Table 5-9).
Lakes in California generally had the lowest
concentrations of Ca+2, SO.T2 and dissolved organic
carbon of any subregion (Table 5-16). Secondary
variables for which California lakes had the lowest
median (M) values were total phosphorus, true color,
turbidity, sodium, potassium, magnesium, chloride,
conductance (Tables 5-17 to 5-19, Table 5-21), and
fluoride (Section 6.2.1). These very dilute waters in
California were the clearest lakes in the West, with
a median Secchi disk transparency of 7.4 m (Table
5-18).
In California, 58.2 percent of the lakes had
conductance values < 1OfjS cm"1, representing more
lakes in this class (Nc = 1398) than in all other
subregions combined (Table 5-24). The bedrock is
similar to that in some of the areas in the Rockies
containing batholiths, but it is distinct from that in
the Pacific Northwest (4B). In addition, California
receives slightly less total precipitation than the
Pacific Northwest (U.S. Department of Commerce
1974), primarily as snow. The excess precipitation
(precipitation - evaporation) is much greater in
California than in the Rockies, thus contributing to
less concentrated solutes in lakes in California.
The lakes in California were the most dilute and had
the lowest ANC concentrations, and also were the
largest and among the deepest of any sampled in
the West. This observation contrasts with that for
the East where the lakes with low ANC were
generally much smaller than those with high ANC
(Linthurstetal. 1986).
Over one-half of the lakes in California were found
at elevations > 3000 m (Table 6-12). This proportion
was less than that for the Central Rockies (4D) and
the Southern Rockies (4E), but was much greater
than in the Pacific Northwest and the Northern
Rockies (4C) where no lakes > 3000 m were sampled.
The estimated percentage of lakes in wilderness
areas in California (44.6%) was comparable to that
for the other subregions, except for the Central
Rockies where the percentage of wilderness lakes
was 65.4 percent. Thus, the high percentage of low
ANC lakes in California cannot be attributed to an
association of low ANC lakes with wilderness areas.
The geomorphic units (Section 6.3.2) for California
(Figure 7-1) were delineated to exclude lakes located
in the foothills. These lower elevation lakes are
associated with different vegetation, soils and
landuse factors that are atypical of lakes located in
the mountainous areas. Over 91 percent of the target
population of lakes in California was represented by
three geomorphic units, the Sierra Nevada, the
Klamath Mountains and the California Cascades. The
minimum elevation of lakes selected in the geomor-
phic unit identified as the Sierra Nevada was 1219m
(4000feet). The minimum elevation of lakes selected
in the California Cascades and in the Klamath
Mountains was 914 m (3000 feet).
7.1.1 Sierra Nevada
The Sierra Nevada geomorphic unit extends approx-
imately 700 km and is about 100 km wide. The range
is dominated by a large batholith composed of
granites and granodiorite, although the northern and
eastern portions contain some metavolcanic and
metasedimentary rocks (Chronic 1986). The individ-
ual plutons comprising the Sierra Nevada batholith
contribute to slight differences in rock type; locally
dramatic differences in rock composition can be
attributed to the presence of dikes. Extensive faulting
resulted in the uplifting of the eastern portion of
the Sierra Nevada. Consequently, the range is highly
asymmetrical with the western slope relatively
gentle, in contrast to the extremely steep eastern
escarpment. Most of the lakes in the Sierra Nevada
are located on the broad western slope and annually
receive up to 400 cm of precipitation. The amount
of precipitation on the western slope is highly
dependent on elevation. Average annual precipita-
tion increases by approximately one cm for every
30-m increase in elevation (Hunt 1974). On the
eastern slope, precipitation decreases dramatically
with decreasing elevation. The eastern slope
generally receives less than 50 percent of the
amount of precipitation received on the western
slope (Hunt 1974). The relationship between
elevation and ANC for the eastern and western
slopes of the Sierra Nevada is shown in Figure 6-29.
118
-------
Figure 7-1. Locations of sampled lakes within geomorphic units in California (4A), Western Lake Survey - Phase I. Lakes
located outside the boundaries were not included in estimates for these subpopulations.
OR
Klamath
Mountains
\
119
-------
An estimated 2119 lakes comprised the subpopu-
lation of lakes in the Sierra Nevada geomorphic unit.
Concentrations of all measured chemical constitu-
ents were extremely low in Sierra Nevada lakes
(Tables 7-1 and 7-2). The Sierra Nevada contained
an estimated 834 lakes (39.3%) with ANC < 50
//eq L"1. This large number, in part, reflects the large
surface area of the Sierra Nevada that was covered
during the WLS-I. However, the California, Oregon,
and Washington Cascades, a range of comparable
surface area, contained fewer lakes with ANC < 50
//eq L"1 (351 or 24%, Tables 7-1 and 7-3). Over 88
percent of the lakes in the Sierra Nevada had ANC
values < 200 //eq L~\ The median ANC (59.8 //eq L"1)
was the lowest among lakes in all geomorphic units
examined in the WLS-I. A study of 26 high elevation
lakes (5000-9000 feet) on the western slope of the
Sierra Nevada in 1980-81 showed a much higher
median alkalinity concentration, exceeding 100//eq
L"1 (Tonnessen 1983).
Concentrations of Ca+2 were similar to those
measured for ANC. The median Ca*2 concentration
was 42.5 //eq L'1 and 61.0 percent of the lakes had
Ca+2<50//eqL"1.
Most lakes in the Sierra Nevada had extremely low
concentrations of DOC; 82.9 percent of the lakes
had DOC < 2 mg L"1 (Table 7-1), and only 3.1 percent
had DOC > 6 mg l_"1 (Volume II). Similarly, very low
concentrations of S04~a were found in lakes in the
Sierra Nevada. The median SO*"2 concentration was
7.1 //eq L~1, and only 8.6 percent of the lakes had
SO4 2 concentrations > 50 //eq L"1 (Table 7-1).
The total ionic concentration of Sierra Nevada lakes,
as reflected in conductance values, shows that these
lakes as a group are the most dilute sampled to date
in the United States. The total ionic concentration
of lakes in the Sierra Nevada is comparable to that
measured in very low ionic strength lakes in Norway
and Sweden (Henriksen and Brakke 1986). The
estimated median conductance of 8.2 //S cm"1
compares closely with that reported by Fox et al.
(1982) for 124 lakes in Sequoia, Kings Canyon, and
Yosemite National Parks in October, 1980. In sharp
contrast, the minimum conductance measured in the
eastern United States was 7.8 //S cm~1 and the
median value for regions in the East ranged from
38.5 //S cm"1 in the Southern Blue Ridge to 90.8
//S cm"1 in Florida (Linthurst et al. 1986).
The median CI" concentration (2.0 //eq L"1) in Sierra
Nevada lakes was considerably less than that
measured in deposition at the National Acid
Deposition Program (NADP) monitoring sites in
Yosemite and Sequoia National Parks (NADP 1984).
The annual CI" concentration, weighted for precip-
itation, measured at Yosemite was 4.5 //eq L"1 and
at Sequoia, 8.2 //eq L"1. Assuming constant CI" in
precipitation between years, the fact that the lake
CI" concentrations do not reflect the composition of
the precipitation at these sites indicates that the
precipitation measured at deposition monitoring
sites is not representative of what the lakes are
actually receiving, or that Cl~ is not conservative in
these watersheds. The elevation of the Yosemite site
is 1408 m and of the Sequoia site, 1856 m; both
elevations are considerably lower than those of the
Table 7-1. Population Estimates and Statistics for Primary Variables by Geomorphic Unit in California (4A), Western Lake
Survey - I
ANC < 50 fieq L"1
Geomorphic Unit n" N pc Nc Ncu
Sierra Nevada 114 2119 0.393 834 1033
California Cascades 16 102 0.402 41 64
Klamath Mountains 14 153 0.035 5 13
S04-*>50,eqL-'
Geomorphic Unit n*** N pc Nc Ncu
Sierra Nevada 114 2119 0.086 182 294
California Cascades 16 102 0.000 0 -
Klamath Mountains 14 153 0.000 0 -
n*'* = number of lakes from which samples were obtained.
N = estimated number of lakes.
ANC
QI
33.6
27.0
57.4
so4-
QI
4.1
0.7
2.5
(Meq
M
59.8
64.5
76.3
M
7.1
1.9
6.1
L"1)
Q4
126.2
176.2
286.1
1L-1)
Q4
13.7
3.2
10.1
Ca+2 < 50 /jeq L"1
Pc
0.610
0.592
0.035
DOC
Pc
0.829
0.485
0.744
NC Ncu
1292 1499
60 88
5 13
< 2 mg LT1
NC cu
1757 1933
49 76
114 192
Ca + 2
QI
22.4
9.0
46.7
DOC
QI
0.44
1.39
0.83
(Meq
M
42.5
22.8
59.5
(mg
M
0.83
2.04
1.30
L-1)
Q4
91.0
77.1
188.2
L-1)
Q4
1.80
2.57
2.08
pc = estimated proportion of lakes meeting the given criteria.
Nc = estimated number of lakes meeting the given criteria.
Ncu = 95 percent upper confidence limit for Nc.
Q! = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
( — ) = undefined.
720
-------
Table 7-2. Population Statistics for Selected Variables for Lakes in Geomorphic Units in California (4A),
Western Lake Survey - Phase I
Conductance
Mg+2
(/jeq L-1)
Ma +
K+ cr
SiO2
(mg L'1)
Fe
(H9 L-1)
Geomorphic Unit
M
Q,
M
M
Q,
M
M
Q,
M
M
Sierra Nevada
California Cascades
Klamath Mountains
4.
3.
7.
9
8
9
8.2
8.2
11.0
Site Depth
(m)
Geomorphic Unit
Sierra Nevada
California Cascades
Klamath Mountains
Q
4.
2.
5.
1
9
6
2
M
8.9
5.4
9.7
3.4
15.8
5.4
6.3
30.2
11.3
Lake Area
(ha)
QI
2.2
2.0
2.8
M
4.6
4.0
4.1
10
6
9
.2
.4
.3
WA:
Q
12
9
10
1
.0
.2
.1
18.7
16.2
14.8
;LA3
M
22.8
15.4
23.7
2.6
2.6
3.3
3.6 1.4
3.5 3.1
4.5 3.3
2.0 1.24 1.99 4.8 13.8
3.6 0.29 1.06 14.3 19.5
5.0 1.26 1.89 17.0 34.0
Secchi Disk
Transparency
(m)
QI
4.2
3.4
5.4
M
7.5
5.8
6.8
Q-i = first quintile (20th percentile).
M = median (50th percentile).
aWA:LA = watershed area:lake area.
Table 7-3. Population Estimates and Statistics for Primary Variables by Geomorphic Unit in the Pacific Northwest (4B),
Western Lake Survey - I
ANC < 50 /jeq L"
Geomorphic Unit n*** N pc
Oregon Cascades 42 443 0.255
S. Washington Cascades 20 218 0.345
North Cascades 68 710 0.171
Wenatchee Mountains3 32 329 0.181
M.Washington Cascades3 24 248 0.169
N.Washington Cascades3 12 133 0.150
Puget Lowlands 22 252 0.089
SO.T2 >
Geomorphic Unit n*** N pc
Oregon Cascades 42 443 0.024
S. Washington Cascades 20 218 0.157
North Cascades 68 710 0.090
Wenatchee Mountains3 32 329 0.000
M.Washington Cascadesa 24 248 0.163
N.Washington Cascades3 12 133 0.177
Puget Lowlands 22 252 0.538
Nc
113
75
122
60
42
20
22
50 Aieq L
Nc
11
34
64
0
40
24
136
i
NCU
165
119
174
96
74
42
47
-1
NCU
28
65
105
—
72
50
195
ANC lf
-------
Figure 7-2. Locations of sampled lakes within geomorphic units in the Pacific Northwest (4B), Western Lake Survey - Phase I.
Lakes located outside the boundaries were not included in estimates for these subpopulations.
North
Washington
Cascades
Olympic
Uplift
Puget
Lowlands
4B
South
Washington
Cascades
Middle
Washington
Cascades
Wenatchee
Mountains
WA
i
i
i
OR
122
-------
lower concentrations of ANC, Ca+2, and Mg+2;
differences in CT and S04~2 may, in part, be
attributable to differences in analytical techniques
for these anions.
7.1.2 Other Geomorphic Units
As noted in the Survey design (Section 2), the desired
number of probability lakes to be sampled per
stratum was 50. Small departures from this desired
number do not substantially reduce confidence in
the population estimates. However, substantial
reduction in the achieved sample size (e.g., n***
< 25) greatly increases the uncertainty of the
population estimates. Subsequent to completion of
the Survey, it was determined that stratifying the
data by geomorphic unit was a useful approach to
interpreting patterns in lake chemistry. Some of
these geomorphic units were of sufficient area (e.g.,
the Sierra Nevada) or contained a sufficient density
of lakes (e.g., the Wind River) such that the probability
sample size within these units approached or
exceeded the goal of 50 lakes per stratum. In many
geomorphic units discussed here, the sample size
was small and the confidence bounds for the
population estimates representing these units are
therefore greater.
The Klamath Mountains (Figure 7-1) include the
Siskiyou Range, the Trinity Alps, and the Salmon
Mountains. This area contained relatively few lakes
(N = 153) and too few were sampled (n*** = 14)
to have high statistical confidence in population
estimates. The formation and composition of bedrock
in the Klamath Mountains is similar to that in the
Sierra Nevada. Older metamorphic Paleozoic and
early Mesozoic formations were displaced by
intrusive granitic rocks in the Mesozoic era; these
formations were deformed by uplifting during the
Tertiary and Quaternary periods (Hunt 1974). The
Klamath Mountains differ from the Sierra Nevada
in that the former contains sizable areas of ultramaf ic
rocks. Of the lakes in the Klamath Mountains, 96.5
percent had ANC values > 50 fjeq L~1. Nevertheless,
these lakes had a relatively low estimated median
ANC concentration (76.3 /ueq L"1). As was observed
throughout the West, Ca+2 concentrations closely
paralleled those for ANC. This suggests that either
the ultramafic rocks were not present in these
watersheds or that they exerted little influence on
the ANC or Ca+2 concentrations.
Dissolved organic carbon (DOC) concentrations in
lakes intheKlamath Mountains weresimilar to those
in the Sierra Nevada; all lakes sampled had DOC
< 6 mg L~1 (Volume II). Although the lakes in this
range are relatively close to the coast (the minimum
distance from the Pacific Ocean is about 75 km),
S04~2 and Cl~ concentrations were very low. All lakes
sampled had S04~2 concentrations < 50 fjeq L 1 and
the median concentration was 6.1 /ueq L~1.
Lakes in the Klamath Mountains are notably deep,
with a median site depth of 9.7 m, and have a high
watershed area to lake area ratio (M = 23.7). The
somewhat higher concentrations of dissolved
material in these Klamath Mountain lakes may be
related to their greater watershed size relative to
those in the Sierra Nevada and California Cascades.
A survey of 78 lakes in the Klamath Mountains by
Wilson and Wood (1984) in summer and fall, 1983
showed a median conductance of 1 2 ^S cm'1 which
is close to the estimated value of 11.0 /vS cm~1
reported in this study. The median concentrations
of Ca*2 (37. 9 fjeq L"1) and Mg+2 ( 1 3.2 jueq L"1) reported
by Wilson and Wood (1984) also are close to the
concentrations reported here.
The California Cascades lie at the extreme southern
end of the Cascade Range which extends northward
from northern California through Oregon and
Washington. The extrusive bedrock of the California
Cascades contrasts to the intrusive coarse grained
granitic bedrock of the Sierra Nevada. Despite this
major difference in bedrock type, the lakes of the
California Cascades are very similar chemically to
those of the Sierra Nevada. The California Cascades
contain a higher percentage (40.2%) of lakes with
ANC < 50 fjeq L"1 than does the Sierra Nevada.
However, this percentage represents a small number
of lakes because relatively^ few lakes are present in
the California Cascades (N = 102, Table 7-1). Only
16 lakes were sampled in this geomorphic unit, so
the statistical confidence in the population estimates
is not high. The Qi concentration of ANC was 27.0
peq L~1, almost as low as that for lakes in the
Southern Washington Cascades (21.3 fjeq L'\
Section 7.2.3). The Qi and median Ca+2 concentra-
tions for lakes in the California Cascades were the
lowest among all geomorphic units (9.0 and 22.8
L~1, respectively).
The concentrations of DOC in the California
Cascades (M = 2.04 mg L'1), although low relative
to those for eastern lakes (Linthurst et al. 1986),
were approximately three times those observed in
lakes in the Sierra Nevada. However, the maximum
DOC concentration observed in lakes in this
geomorphic unit was only 3.8 mg L"1. Sulfate
concentrations in the California Cascades were
extremely low, with a Qi value of 0.7 fjeq L"1 and
a median of 1 .9 fjeq L~1. Although numerous natural
sources of sulfur exist in this area (for example, hot
springs in Lassen National Park), they apparently
have little impact on the chemistry of nearby lakes.
The S04~2 concentrations in these lakes and in the
nearby Oregon Cascade lakes are believed to be
among the lowest in the world (Armstrong and
123
-------
Schindler 1971). The lakes in the California Cascades
differ from those in the Sierra Nevada in that the
California Cascades have much higher Mg+2 con-
centrations (indicative of the presence of mafic rocks
in the watershed), have lower SiO2 concentrations,
and are relatively shallow.
7.2 The Pacific Northwest
The size of the target population of lakes in the Pacific
Northwest (Subregion 4B) was estimated to be 1706
lakes. Among the subregions sampled in the West,
the Pacific Northwest contained the highest percen-
tage of lakes with pH < 6.0 (2.4%, Table 5-11) or
extractable aluminum > 50/ug L"1 (0.6%, Table 5-14),
and the lowest percentage of lakes with DOC > 6
mg L"1 (2.7%, Table 5-15). The Pacific Northwest
was second only to California in the percentage of
lakes with ANC < 50 pteq L"1 (19.5%, Table 5-9) or
Ca+2 < 50 /ueq L"1 (28.4%, Table 5-13). Concentra-
tions of ANC and Ca+2 tend to decrease from north
to south in this subregion.
The median Secchi disk transparency (4.9 m, Table
5-18) was considerably less than that in California
(M = 7.4 m), but was greater than those in the Central
and Southern Rocky Mountain subregions. The
Pacific Northwest had the second highest percentage
of dilute (conductance < 10 fjS cm"1) lakes (26.7%)
and the second lowest Qi value (7.8 /uS cm"1. Table
5-21) for conductance of the western subregions.
Although the Pacific Northwest contained fewer
dilute and low ANC lakes than did California, many
of these lakes are comparable to the most dilute
waters sampled anywhere in the West.
The Pacific Northwest had the lowest watershed area
to lake area ratios of any subregion. Precipitation
is greater than in other subregions. Adeepsnowpack
is accumulated, which melts slowly in some areas.
Lakes in the Pacific Northwest had the highest
subregional median value for CI" (Table 5-21) and
the second highest median for Na+ (Table 5-1 9). The
relative abundance for the major cations in the
Pacific Northwest was Ca > Na > Mg > K (Table
6-5). This order has also been reported by Wissmar
et al. (1982) for subalpine lakes in the Lake
Washington basin and can be observed in data from
Nelson and Delwiche (1983). The concentrations of
Mg+2 in the Pacific Northwest were relatively low
compared to other subregions. Some lakes, however,
can have high Mg+2, such as those located on dunite
in the North Washington Cascades (Brakke and
Loranger 1986). The second highest subregional
percentage of lakes with S04"2 > 50 /ueq L"1 (17.1 %)
occurred in the Pacific Northwest (Table 5-12).
The six geomorphic units for the Pacific Northwest
(Figure 7-2) were delineated on the basis of elevation
and major mountain passes. The Puget Lowlands
included lakes less than 610m (2000 ft) in elevation
in Washington. The North Washington Cascades
included lakes from the Canadian border south to
Rainy Pass. The Middle Washington Cascades
contain lakes between Rainy Pass and Stevens Pass.
Lakes between Stevens Pass and Snoqualmie Pass
were included in the Wenatchee Mountains. The
Southern Washington Cascades covered lakes
between Snoqualmie Pass and the Columbia River
(Oregon border). Lakes at elevations above 914 m
(3000 ft) in Oregon were included in the Oregon
Cascades. These five geomorphic units represented
93 percent of the estimated target population of lakes
in the Pacific Northwest. The Olympic Uplift was also
designated as a geomorphic unit, but the insufficient
sample size greatly reduced the certainty in
population estimates.
7.2.1 The North Cascades
The North Washington Cascades, Middle Washing-
ton Cascades, and Wenatchee Mountains are
comprised largely of Paleozoic sedimentary rocks,
which have been folded and metamorphosed,
although some younger metamorphic rocks and
intrusions of granitic rocks are present. Unlike the
North Cascades, the bedrock of the South Washing-
ton Cascades and Oregon Cascades is largely
volcanic.
The North Cascades cover an area of about
20,000 km2 which is almost entirely forested except
for the highest peaks (Franklin and Dyrness 1973).
Above 1800 m permanent snow fields comprise 13
percent of the area (Rasmussen and Tangborn 1976).
Pleistocene glaciers carved deep U-shaped valleys,
and many active glaciers are still present. Many small
lakes occur (Qi for lake area = 1.9 ha, Table 7-4).
Bedrock types are very diverse, especially in the
North Washington Cascades, where metamorphosed
Paleozoic granitic and volcanic rocks occur with
Quaternary volcanic. Cretaceous sedimentary, and
pre-Jurassic metamorphic rocks. This geological
diversity is reflected in the variability of the chemistry
of lakes.
In the Wenatchee Mountains, the median ANC was
104.8 /ueq L"1 and an estimated 86.8 percent of the
lakes had ANC < 200 /ueq L"1. This area contained
the second highest estimated number of lakes with
ANC < 50 /ueq L"1 in Washington. These findings
agree with those of Welch et al. (1986) for the Alpine
Lakes Wilderness which comprises most of the
Wenatchee Mountains.
An estimated 1 5.0 percent of the lakes in the North
Washington Cascades had ANC < 50 /ueq L"1 (Table
7-3); 49.0 percent had ANC < 200 //eq L"1. These
percentages compare favorably with results obtained
124
-------
Table 7-4. Population Statistics for Selected Variables for Lakes in Geomorphic Units in the Pacific Northwest (4B),
Western Lake Survey - Phase I
Conductance
(l*S cm"1)
Mg + 2
(Meq L 1)
K +
cr
Si02
(mg L~1)
Fe
Geomorphic Unit
Q-|
M
Q,
M
0-1
M
M
M
M
Q, = first quintile (20th percentile).
M = median (50th percentile).*
3These three geomorphic units together form the North Cascades geomorphic unit.
bWatershed area:lake area.
M
Oregon Cascades
S. Washington Cascades
North Cascades
Wenatchee Mountains3
M. Washington Cascades3
N. Washington Cascades3
Puget Lowlands
Geomorphic Unit
Oregon Cascades
S. Washington Cascades
North Cascades
Wenatchee Mountains3
M. Washington Cascades3
N. Washington Cascades3
Puget Lowlands
4.4 11.3
7.2 17.6
7.8 14.8
6.5 13.9
9.0 13.6
8.1 27.1
22.3 34.1
Site Depth
(m)
Q! M
2.5 4.8
3.4 7.5
3.9 10.6
3.5 11.1
3.1 8.2
5.4 10.4
2.2 5.5
9.2 23.9
10.9 23.5
8.4 16.7
7.6 13.6
9.7 16.7
6.8 28.7
46.5 79.2
Lake Area
(ha)
Q, M
2.4 3.4
2.0 4.8
1.9 4.2
2.0 4.5
1.8 3.2
1.5 5.9
3.2 6.7
17.6
23.4
13.7
15.6
12.8
11.5
47.6
WA:
QI
5.3
6.7
11.1
11.3
11.1
7.2
9.8
32.6
34.5
19.3
21.4
18.5
19.9
67.4
:LAb
M
21.4
12.6
19.5
20.0
19.7
12.0
22.8
2.6 4.6
2.0 3.8
1.9 4.5
1.6 4.1
2.9 8.7
1.5 3.4
4.3 6.2
Secchi Disk
Transparency
(m)
Q! M
2.3 4.7
2.7 4.6
3.5 7.5
3.5 8.1
2.7 5.8
4.6 7.9
1.8 2.5
6.0
11.6
2.5
3.7
2.4
1.7
24.4
8.2
14.5
4.7
9.1
3.5
2.4
38.2
1.8
1.1
1.6
1.6
1.3
1.6
1.7
4.5
3.0
2.7
2.7
2.7
2.3
5.0
1.9
8.0
3.8
5.1
0.0
2.6
21.6
10.6
20.8
12.4
10.2
13.5
13.4
58.5
by Brakke and Loranger (1986) from a survey of 57
lakes in the same area. They calculated that 15.8
percent of the lakes had ANC < 50 (jeq L~1 and 66.7
percent had ANC < 200 /jeq L"1. Although some
extremely low ANC values (3-5 /jeq L~1) have been
observed in the North Washington Cascades, ANC
has been reported to be variable and strongly
associated with bedrock geology (Brakke and
Waddell 1985; Brakke and Loranger 1986). The
interquintile difference (Qd) for ANC in the present
study was 320 peq L"1 for the North Washington
Cascades. Lakes in the Middle Washington Cascades
and Wenatchee Mountains showed much less
variability in ANC (Qd = 128 and 107 (jeq L'1,
respectively).
As in the case of the West in general, Ca+2
concentrations in North Cascade lakes were similar
to ANC concentrations. In watersheds within the
Cascades, weathering rates are high for some lakes,
particularly those most recently deglaciated (Rey-
nolds and Johnson 1972). Relatively small areal
exposures of calcite can generate significant
alkalinity and Ca+2 (Drever and Hurcomb 1986) as
a result of weathering. However, when calcite
materials are absent, very dilute lakes with low ANC
can result.
Sulfate concentrations in the Middle and North
Washington Cascades were the highest found in the
Cascade range (M = 23.8/jeq L"1 and M = 21.6 (jeq L"1,
respectively). The lowest values of SCu 2 in Washing-
ton were found in the Wenatchee Mountains (Qi
= 5.4 fjeq L"1; M = 12.6 yueq L"1, Table 7-3).
Dissolved organic carbon concentrations were very
low in the North Cascades. The estimated Q4 value
was 1.43 mg L"1 and 90.4 percent of the lakes had
DOC < 2 mg L"1 (Table 7-3). Similarly, conductance
was low, with 20 percent of the lakes having values
< 7.8 fjS cm~1 (Table 7-4).
7.2.2 The Oregon Cascades
The Oregon Cascades are more uniform geologically
than the Washington Cascades, although two
separate units can be recognized. These are the High
Cascades on the East and the West Cascades,
running parallel from north to south. The West
Cascades are older and lower in elevation than the
High Cascades (Baldwin 1981). However, interme-
diate volcanic rocks occur in both areas and because
of the uniformity of chemistry in lakes, we have not
subdivided the Oregon Cascades.
The Oregon Cascades have the second lowest
median ANC (92.1 //eq L'1) and the second highest
percentage of lakes with low ANC (25.5% with ANC
< 50 fjeq L"1) in the Pacific Northwest (Table 7-3).
These estimates agree with the median ANC (83.9
/jeq L"1) for 63 Oregon Cascade lakes sampled by
Nelson and Delwiche (1983). Calcium concentra-
725
-------
tions in the Oregon lakes are among the lowest of
all geomorphic units, with a median of 48.6 //eq L~1
and 55.8 percent of the lakes having Ca+2 < 50
//eq L"1 (Table 7-3). Similar results were obtained
by Nelson and Delwiche (1983); the mean Ca+2 was
57.9 //eq L~1. Lakes in the Oregon Cascades showed
relatively little interquintile difference (Qd) in Ca+2
andAIMC.
Lakewater S04~2 concentrations in the Oregon
Cascades and in the adjacent California Cascades
were the lowest in the West, and are among the
lowest observed in the Northern Hemisphere
(Armstrong and Schindler 1971). Lakes in the
Oregon Cascades had the lowest median (1.6//eq L"1)
and the second lowest 04 (4.0 //eq L~1) value for
SO*"2 of any geomorphic unit (Table 7-3). These
results are comparable to those of Nelson and
Delwiche (1983), who found a mean SO4~2 concen-
tration of 5.82 //eq L"1 with a minimum concentration
of 0.62//eq L"1.
Lakes in the Oregon Cascades are generally shallow
(median site depth = 4.8 m) and dilute (median
conductance = 11.3 //S cm"1. Table 7-4). It appears
that Na+ and CI" are generated from watersheds in
both the Oregon Cascades and the South Washing-
ton Cascades. Even though more distant from marine
waters, the Q, and median values of Na+ and CI"
were much greater than in lakes in the North
Cascades (Table 7-4). Similarly, Nelson and Delwiche
(1983) observed moderate concentrations of Na+ and
CI" in the Oregon Cascades. Nelson (1 985) analyzed
data from Oregon Cascade lakes and concluded that
lakewaters were in equilibrium with kaolinite and
gibbsite. The slow dissolution kinetics of these
minerals were reflected in very dilute lakes. The
weathering products of feldspars include Na+ and
K+, which contribute to lakewater concentrations of
these cations.
7.2.3 Other Geomorphic Units
In contrast to the North Cascades, the South
Washington Cascades contains few areas with
exposed sedimentary or metamorphic rocks. The
South Washington Cascade Range is comprised
mainly of andesite and basalt and is heavily forested.
This geomorphic unit is much more uniform than
the North Cascades, even though it includes a series
of recent Quaternary deposits related to the volcanic
activity of Mount Rainier, Mount Adams, and Mount
St. Helens.
The lowest Qi for ANC (21.3 //eq L"1) of any
geomorphic unit (Table 7-3) was estimated for the
South Washington Cascades. In the Pacific Northw-
est, the highest percentages of lakes with low Ca+2
and low ANC occurred in the South Washington
Cascades, but lakes in this geomorphic unit also
showed the most variability in these variables (Qd
for ANC = 355 //eq L"1, Qd for Ca+2 = 246 //eq L"1).
Relatively little sampling has been conducted in the
South Washington Cascades, although some low
ANC lakes are known to occur (Duncan and Ausserer
1984; Wissmar et al. 1982; Bortelson et al. 1977).
Concentrations of DOC were slightly higher in the
South Washington Cascades than in the North
Cascades, although none of the lakes sampled had
DOO6 mg L"1.
Concentrations of all measured constituents were
relatively high in lakes in the Puget Lowlands (Tables
7-3 and 7-4). A low percentage (8.9%) of lakes in
the Puget Lowlands had ANC < 50 //eq L"1 (Table
7-3). As expected from the proximity of these lakes
to coastal waters, the highest concentrations of Na+
and CI" of any geomorphic unit occurred in the lakes
of the Puget Lowlands. The Puget Lowlands had the
highest percentage (53.8%) of lakes with S04~2 > 50
//eq L"1 and the highest median value for S04~2 (50.2
//eq L~1) of any geomorphic unit (Table 7-3).
Dissolved organic carbon concentrations in the Puget
Lowlands were very high for the West, with the Qi
value (2.26 mg L"1) exceeding the highest Q4 value
in any geomorphic unit of the Cascade Range (Table
7-3). These high DOC concentrations may be
attributable to greater production of organic matter
in lakes related to higher concentrations of total
phosphorus (Sumioka and Dion 1985; Gilliom and
Bortelson 1983).
7.2.4 Sulfate in Geomorphic Units in the Pacific
Northwest
Sulfate concentrations in lakes increased north of
the Columbia River, and were markedly higher in
Washington than in Oregon. The estimated Qi value
for SO4~2 (10.8 //eq L"1) in the South Washington
Cascades is more than twice the Q4 value for S04~2
in the Oregon Cascades (4.0 //eq L"1). Along the
Cascades, the concentrations of SO4~2 are low in
California and highest in the North Cascades. Within
the North Cascades, lakewater S04~2 concentrations
are highest in the North Washington Cascades and
lowest in the Wenatchee Mountains.
Sulfate can be derived from the weathering of sulfur-
containing minerals within watersheds. In the North
Cascades, Loranger (1986) reported that the
watershed source of S04~2 exceeds the atmospheric
contribution in some lakes, indicating weathering
from bedrock as the likely source. Geological sources
of SO4~2 are also indicated for some lakes in the
Alpine Lakes Wilderness (Welch et al. 1 986). In these
cases, S04"2 can be a major anion resulting in a
126
-------
small deficit of ANC relative to Ca^ and Mg+2
(Loranger 1986; Welch et al. 1 986).
Some of the SO/2 could also be derived from sea
salts contained in precipitation. To examine the
contribution of sea salt to lakewater S04 2 concen-
trations, SO/2 was corrected for sea salt using the
formula in Gorham et al. (1985). The results, which
are not shown, indicated that very little of the lake
SO/2 is derived from sea salt.
Sulfur contained in emissions from volcanic
eruptions can serve as sources of SO/2 to lakes.
The recent eruption of Mount St. Helens temporarily
may have increased SO/2 concentrations in lakes
in the surrounding area; Wissmar et al. (1982)
reported changes in lakewater chemistry near the
blast zone. However, if the lakes in this area have
short hydraulic residence times, as calculated on the
basis of map information (Section 6.4.3.1), and hence
rapid flushing rates, it is likely that the Mount St.
Helens eruption had only a short-term effect on
lakewater SO/2 concentrations for most lakes in the
area.
7.3 Northern Rocky Mountains (4C)
The Northern Rocky Mountains (Subregion 4C) is
geologically distinct from the Central and Southern
Rockies. The chemical characteristics of the lakes
in this subregion are also distinct from those of lakes
in the Central (4D) and Southern (4E) Rocky
Mountains. Geological and, perhaps, meteorological
differences likely account for this variability. Lakes
in the Northern Rockies were lower in elevation (M
= 2064 m. Table 5-6) than in the Central Rockies
(M = 3040 m) and in the Southern Rockies (M =
3264 m). Only lakes in the Pacific Northwest (4B)
were lower in elevation (M = 1437) than those in
the Northern Rocky Mountains.
About one-half (50.7%) of the estimated number of
lakes in the target population in the Northern Rockies
(N = 2379) were estimated to have ANC < 200 fjeq L~1
(Table 5-9); in the West, only the Southern Rockies
had a lower percentage of lakes in this category
(39.4%). The median ANC values in these two
subregions were the highest in the West at 195.2
fjeq L~1 for the Northern Rockies and 317.0 fjeq L~1
for the Southern Rockies (Table 5-16).
As in other western subregions, pH values for lakes
in the Northern Rockies were generally circumneu-
tral. The Qi value for pH, for example, was 6.90 and
for other subregions ranged from 6.59 in California
to 7.10 in the Southern Rockies. The median pH
(7.35) in the Northern Rocky Mountain lakes was
the second highest observed in the West behind the
Southern Rockies (7.60, Table 5-16).
Sulfate concentrations in lakes of the Northern Rocky
Mountains were generally low (M = 15.6 fjeq L"1,
Table 5-16) and were intermediate when compared
to the other western subregions. Two other subre-
gions had greater SO/2 medians—the Central and
Southern Rockies with concentrations of 24.4 and
34.6 fjeq L"1, respectively. However, at the Q4 level,
SO/2 was 32.3 fjeq L"1 in the Northern Rockies and
only California had a lower Q4 value for SO/2 (13.5
fjeq I/1).
Lakes in the Northern Rockies had the second
highest median conductance in the West (25.1
/jS cm"1) and the highest Q4 value (108.9 fjS crrf1;
Table 5-21). Calcium values for the lakes in this
subregion were also high (M = 154.2 fjeq L~1 and
Q4 = 730.0 fjeq L"1) (Table 5-16). Dissolved organic
carbon was generally low in western lakes and the
Northern Rockies had the second lowest population
estimates for this variable; 70.8 percent of the lakes
in this subregion had DOC < 2.0 mg L"1. There were
few lakes with large surface area in the Northern
Rockies (M = 4.6 ha and Q4 = 8.9 ha). Only the
Southern Rockies had smaller lakes.
Nine geomorphic units in the Northern Rockies were
delineated based on mountain ranges (Figure 7-3).
The Clearwater Mountains, Salmon River Moun-
tains, Sawtooth Mountains, and Bitterroot Range are
collectively identified and discussed in this section
as the Idaho Batholith. A total of 88 lakes were
sampled in the Idaho Batholith, 37 of which are in
the Bitterroot Mountains. Because many lakes were
sampled in the Bitterroots, this geomorphic unit is
also discussed separately. Population estimates
were also calculated for the Lewis Range, the Blue
River Uplift, and the Anaconda-Pintlar Mountains,
each of which includes several distinct mountain
ranges. However, because relatively few lakes were
sampled in these latter geomorphic units, population
estimates have large upper confidence limits (Ncu,
Section 7.1.2). The geomorphic units for which
population estimates were calculated represent 72
percent of the target population in the Northern
Rockies. The Selkirk Mountains and the Cabinet
Mountains were also designated as geomorphic
units, but due to the small number of lakes sampled,
population estimates are not presented.
7.3.1 Idaho Batholith
The Northern Rockies is largely formed of granitic
intrusions of Cretaceous age. This region is
dominated by the Idaho Batholith, a roughly
rectangular area 140 km wide and 390 km long
which ranges in elevation from about 900 to 2100 m
(Hunt 1974). Approximately 85 percent of the
batholith, which is wholly north of the Snake River
in Idaho, is composed of granitic rocks; related
727
-------
Figure 7-3. Locations of sampled lakes within geomorphic units in the Northern Rockies (4C), Western Lake Survey - Phase I.
Lakes located outside the boundaries were not included in estimates for these subpopulations.
Selkirk
Mountains
4C
Cabinet
Mountains
Bitterroot
Range
Lewis Range
Anaconda-Pintlar
Mountains
r
Sawtooth
Mountains
ID
-1
128
-------
portions of the batholith are located in Montana,
Oregon, Washington and British Columbia (Larson
and Schmidt 1958).
Median ANC for the Idaho Batholith is 98.3 /ueq L~1
(Table 7-5). The Qi and Q4 values for ANC in this
geomorphic unit were 43.8 and 396.0 /ueq L"1,
respectively. In comparison, lakes of the Sierra
Nevada had a median ANC of 59.8 /ueq L~1 and Qi
and Q4 of 33.6 and 126.2 peq L"1, respectively (Table
7-1). Conductance was higher in the Idaho Batholith
(M = 11.5 /uS crrf1, Table 7-6) than that observed
in the Sierra Nevada (M = 8.2 jjS crrf1. Table 7-2).
7.3.2 Bitterroot Range
The Bitterroot Mountains lie on the northeastern
edge of the Idaho Batholith, an area dominated by
igneous rocks of Mesozoic age. An estimated 77.6
percent (Nc = 220) of the lakes in this mountain range
had ANC < 200 peq L"1 and 33.1 percent had ANC
< 50 fjeq L"1 (Table 7-5). The Bitterroot Mountains
contained the third lowest median ANC value of any
geomorphic unit in the West (70.4 /ueq L"1). Only
lakes in the Sierra Nevada and in the Sawtooth
Mountains had lower median ANC values (59.8
/ueq L~1 and 67.8 /ueq L"1, respectively) (Table 7-1).
Lakes in the Bitterroots also had the second lowest
median Ca+2 value (48.3 /ueq L"1). Conductance for
lakes in the Bitterroots was very low (M = 9.8 fiS cm"1,
Table 7-6); only in the Sierra Nevada was conduc-
tance lower (M = 8.2 pS cm"1. Table 7-2). Sulfate
was also low in the Bitterroot Mountains (M = 9.0
/aeq L~1) and it was estimated that 80 percent of the
lakes had SO4~2 concentrations less than 17.1 /ueq
L"1 (Table 7-5).
7.3.3 Other Geomorphic Units
Of the geomorphic units in the Northern Rockies,
the Lewis Range had the largest estimated number
of lakes in the target population (N = 553). This area
contains Glacier National Park and is dominated by
Precambrian sedimentary rock which has been
metamorphosed, largely, into quartzite (Alt and
Hyndman 1978). Only 15 lakes were sampled in this
geomorphic unit; therefore, the upper confidence
limit (Ncu) on statistical estimates is large. Recog-
nizing this limitation, lakes in this area were
estimated to have high ANC relative to lakes in other
geomorphic units (M = 392.2 /ueq L"1; Table 7-5).
The percentage of lakes with ANC < 200 fieq L"1
in this area (33.3%) was the lowest of all geomorphic
units analyzed in the Northern Rockies.
Only 12 and 10 lakes, respectively, were sampled
in the Anaconda-Pintlar Mountains and the Blue
River Uplift. As in for the Lewis Range, population
estimates for these geomorphic units have large
upper confidence limits (Section 7.1.2). No lakes with
ANC < 50 /ueq L"1 were sampled in either of these
areas, but their median ANC values were less than
that estimated for the Lewis Range (Table 7-5). The
Anaconda-Pintlar Mountain lakes had the highest
median S04"2 concentration in the Northern Rockies
(32.0 /ueq L"1).
7.4 Central Rocky Mountains (4D)
Although less than seven percent of the lakes in
the Central Rockies (Subregion 4D) had ANC < 50
/ueq L"1, 77.8 percent had ANC < 200 /ueq L"1 (Table
5-9), which was the second highest percentage in
the West next to California. The Central Rockies had
the second lowest median ANC (104.9 //eq L~1, Table
5-16). Lakes in the Central Rockies also had the
second highest Qi values, but the second lowest M
and Q4 values for Ca+2 and conductance (Tables 5-16
and 5-21). Extractable aluminum concentrations
were the lowest among all subregions at the Qi,
M, and Q4 levels.
The relative abundance of major cations at both Qi
and median values was Ca > Mg > Na > K (Table
6-5). The median Na+ concentration was the second
lowest among western subregions. Nearly all of the
lakes in the Central Rockies were clearwater (true
color < 30 PCU) with Ca+2 and HC03" as the dominant
ions (Section 6.2.6). However, in some lakes Na+ was
the dominant cation and HCOs" or S04"2 was the
dominant anion. The Central Rockies had the lowest
percentage of lakes with S04"2 > 50 /ueq L"1 among
western subregions.
An estimated 235 lakes (10%) in the Central Rockies
had conductance < 10 /uS cm"1, the second lowest
percentage of dilute lakes among western subre-
gions. Although evaporation rates in this subregion
are comparable to those observed for lakes in
California and in the Central Rockies (Kohler et al.
1955), total precipitation is less in the Central
Rockies (U.S. Department of Commerce 1974). As
a consequence, higher concentrations of solutes
were observed for lakes in the Central Rockies. Lakes
in this subregion also may receive atmospheric
contributions of materials in the form of airborne
dust containing gypsum or calcium carbonate from
the surrounding arid basins. This influence is
indicated by higher Ca+2 concentrations in precip-
itation measured at NADP/NTN sites in Colorado and
Wyoming (15 /ueq L"1) than those measured at sites
in California, Oregon, and Washington (3 /ueq L"1,
Hidy and Young 1986). Hidy and Young (1986) also
found that Ca+2 explained 85 percent of the variance
in S04"2 concentrations, which supports the
hypothesis that wind-borne alkaline dust comprised
of CaCOs and CaS04 influences lake chemistry in
this subregion.
129
-------
Table 7-5. Population Estimates and Statistics for Primary Variables by Geomorphic Unit in the Northern Rockies (4C),
Western Lake Survey - Phase I
ANC;
Geomorphic Unit n*" N pc
Lewis Range 15 553 0.067
Anaconda-Pmtlar Mtns. 12 239 0.000
Blue River Uplift 10 73 0.000
Idaho Batholith 88 937 0.236
Bitterroot Range3 37 283 0.331
Salmon River Mountains3 22 379 0.127
Sawtooth Mountains3 20 192 0.228
SO4"2
Geomorphic Unit n*** N pc
Lewis Range 15 553 0.067
Anaconda-Pintlar Mtns. 12 239 0.356
Blue River Uplift 10 73 0.157
Idaho Batholith 88 937 0.000
Bitterroot Range3 37 283 0.000
Salmon River Mountains3 22 379 0.000
Sawtooth Mountains3 20 192 0.000
~ 50 (jeq
NC
37
0
0
221
94
48
44
> 50 /jeq
Nc
37
85
12
0
0
0
0
L-1
NCU
97
—
—
297
125
111
67
L"
NCU
97
171
30
—
—
—
-
ANC
QI
77.2
143.7
95.9
43.8
38.9
71.9
44.2
S04-2
QI
8.0
18.5
4.3
5.5
5.4
6.3
5.9
(MeqL
M
392.2
255.0
123.4
98.3
70.4
177.4
67.8
(/^eq L
M
10.3
32.0
5.7
8.8
9.0
10.7
7.5
-1) Ca*2 < 50 yueq L~1
Q4
1208.3
469.3
178.5
396.0
285.8
421.3
884.0
-1)
Q4
32.2
58.6
23.0
17.4
17.1
15.7
20.8
PC
0.200
0.048
0.171
0.344
0.500
0.188
0.260
DOC <
PC
0.733
0.596
0.843
0.781
0.809
0.715
0.808
Nc
111
11
12
322
142
71
50
2 mg
Nc
406
142
61
731
229
271
155
NCU
213
30
26
406
184
138
75
I'1
NCU
587
236
92
859
288
389
209
Ca4
Q,
49.0
148.7
57.6
31.2
26.8
50.7
40.8
DOC
QI
0.35
0.92
0.37
0.81
0.79
0.99
0.38
-2 / i — 1 \
M
328.0
202.0
87.2
63.7
48.3
72.0
63.5
(mg L~
M
0.78
1.67
0.70
1.22
1.23
1.54
0.87
Q4
832.6
326.0
131.3
360.2
216.4
368.8
805.6
1,
Q4
2.67
2.28
1.59
2.08
1.93
2.30
1.82
n*" = number of lakes from which samples were obtained.
N = estimated number of lakes.
pc = estimated proportion of lakes meeting the given
Nc = estimated number of lakes meeting the
Ncu = 95 percent upper confidence limit for N
Q, = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
( — ) = undefined.
criteria
given criteria.
C"
aThese three geomorphic units, together with the Clearwater Mountains (not presented separately), form the Idaho Batholith
geomorphic unit.
Table 7-6. Population Statistics for Selected Variables for Lakes in Geomorphic Units in the Northern Rockies (4C),
Western Lake Survey - Phase I
Conductance Mg + 2 NE
(^iS cm~1) (Mecl L ) (^eP
Geomorphic Unit
Lewis Range
Anaconda-Pmtlar Mtns.
Blue River Uplift
Idaho Batholith
Bitterroot Range3
Salmon River Mountains3
Sawtooth Mountains3
Q!
9.0
18.4
11.2
6.6
5.4
8.9
7.7
Site
M
39.1
32.6
13.1
11.5
9.8
18.4
10.5
Depth
(m)
Geomorphic Unit
Lewis Range
Anaconda-Pmtlar Mtns.
Blue River Uplift
Idaho Batholith
Bitterroot Range3
Salmon River Mountains3
Sawtooth Mountains3
QI
3.1
3.3
1.4
4.6
5.4
4.8
4.8
M
11.0
5.4
3.4
10.1
11.8
8.6
9.9
QI
30.4
15.6
7.9
6.6
5.5
9.1
5.4
Lake
M
110.6
22.5
9.2
10.4
13.2
17.4
8.4
Area
(ha)
QI
2.3
4.2
2.1
2.2
2.6
2.1
2.4
M
4.9
4.8
2.9
3.7
4.8
2.5
5.4
QI
4.0
28.5
20.8
13.5
11.9
21.9
16.7
WA;
Q,
6.0
10.6
20.4
9.9
10.3
10.5
5.7
L-1)
M
9.7
43.1
26.0
23.1
15.7
38.8
22.6
:LAb
M
13.6
23.8
23.0
18.5
16.0
20.8
20.4
K +
Q, M
2.1 2.8
3.2 6.0
9.1 10.4
2.9 4.7
2.6 4.1
4.4 5.8
2.6 3.0
Secchi Disk
Transparency
(m)
Q, M
2.0 4.8
1.5 4.0
1.4 3.5
3.7 6.3
4.1 6.4
2.4 6.3
4.2 5.8
Cl
(^eq I
QI
1.6
2.7
1.4
1.7
1.8
1.7
1.7
SiO2
."'I (mg L-1)
M Q,
2.1 0.8
3.5 3.0
1.9 2.2
2.4 2.1
2.3 1.7
2.4 2.6
2.4 2.7
M
1.5
4.2
3.0
3.3
2.2
3.8
3.9
Fe
(/"9 I-'1)
Q, M
3.0 7.8
9.7 15.8
5.3 14.0
3.7 9.7
5.6 11.5
4.1 7.4
0.0 8.0
QT = first quintile (20th percentile).
M = median (50th percentile).
"These three geomorphic units, together with the Clearwater Mountains (not presented
separately), form the Idaho Batholith geomorphic unit.
bWatershed area:lake area.
730
-------
The median lake elevation (3040 m) in the Central
Rockies was the second highest among western
subregions. Median values for watershed area (178
ha), lake area (5.4 ha), and watershed area to lake
area ratio (30.2) were the highest among western
subregions (Table 5-6); however, watershed area in
the Central Rockies was also the most variable (Qd
551 ha).
The eight geomorphic units in the Central Rockies
(Figure 7-4) were delineated on the basis of major
mountain ranges. Population estimates and statistics
were calculated for the Beartooth Uplift, the
Yellowstone Plateau, the Bighorn Uplift, the Wind
River Uplift, and the Uinta Arch. These five
geomorphic units represented 90 percent of the
target lake population in the Central Rockies (N =
2299). The Teton Uplift, the Gros Ventre Uplift, and
the Wasatch Mountains were also designated as
geomorphic units; however, fewer than 10 lakes
were sampled in each of these areas, so population
estimates are not presented.
7.4.1 The Wind River Uplift
The Wind River Range is an uplift of Precambrian
granites and gneisses. Glacial cirques are numerous
and many ponds less than 1 ha in area are present.
The lakes in the estimated target population (N =
884) were larger than those in any other geomorphic
unit; they were also relatively deep (Table 7-8).
Several active glaciers remain in the alpine tundra
above 3350 m, particularly on the east side of the
Continental Divide (Arno 1969). The southern and
eastern portions receive only about 50 cm of
precipitation, whereas the northwestern portion
receives more precipitation (Steele et al. 1 983).
The median ANC value for lakes in the Wind River
Range was 109.3 /c/eq L~1 (Table 7-7). This was
slightly higher than the means of 90 /ueq L"1 and
60 /ueq L"1 measured in lakes on the western and
eastern slopes, respectively, in a Forest Service
survey of 92 lakes in the Wind River (Stuart 1984,
Galbraith 1984).
The Qi value for SO-f2 in Wind River lakes (21.7
yueq L"1) was among the highest for all geomorphic
units; the median value was 24.3 fjeq L~1 (Table 7-7).
There was little interquintile difference in S04"2, with
a Q4 value of 31.3 /ueq L~1. These values are very
similar to those measured in a study by Galbraith
(1985) in the Wind Rivers in which the reported range
was 20-35 //eq L"1.
The Qi value for conductance was 12.4 /jS cm'1,
which was close to the median value of 15.0/uS cm"1
(Table 7-8). As in the other geomorphic units in the
Central Rockies, relatively few lakes had conduc-
tance < 10/uS cm , but many lakes had conductance
between 10-1 5 /uS cm"1.
7.4.2 The Uinta Arch
The Uinta Arch is a large uplift running east to west.
Glacial cirques and lakes are numerous at higher
elevations. The mountains consist mainly of
Precambrian sedimentary and metamorphic rocks,
particularly quartzites. At lower elevations, lime-
stones, sandstones and shales are common.
Timberline occurs at about 3350 m (Arno and
Hammerly 1984). Lakes in the Uintas are smaller
and shallower than lakes in the rest of the Central
Rockies (Table 7-8).
Similar to the Bighorn and Wind River areas, the
Uintas have a high percentage of lakes with ANC
< 200 /jeq L"1 (93.1 %) but a low percentage of lakes
with ANC < 50 /ueq L"1 (3.9%). Results for Ca+2 were
similar (Table 7-7). The Qi and median values for
conductance were 11.5 and 15.1 juS cm"1, respec-
tively, which were nearly identical to those for lakes
in the Wind River Range. Chloride and iron
concentrations were higher in lakes in the Uintas
than in other geomorphic units in the Central
Rockies. The median SiO2 concentration in the
Uintas (1.0 mg L"1) was the lowest estimated for
any geomorphic unit (Table 7-8).
7.4.3 Other Geomorphic Units
The Beartooth Uplift contains many lakes, with about
80 percent at elevations > 2700 m. The Beartooth
Uplift is a large (5200 km2), rectangular block of
Precambrian igneous and metamorphic rocks
uplifted along faults. The area was originally covered
by sedimentary rocks that subsequently have been
removed by glaciation and other erosive forces (Alt
and Hyndman 1 972). The Beartooth Uplift was once
covered almost completely by a glacier. Lakes in the
Beartooths are deep (M = 11.8 m) and relatively large
(M = 6.3 ha. Table 7-8).
All lakes sampled in the Beartooth Uplift had ANC
< 200 /yeq L"1, and 15.7 percent (the highest among
geomorphic units in this subregion) had ANC < 50
/ueq L"1 (Table 7-7). However, because only 12 lakes
were sampled in this geomorphic unit, statistical
confidence in these population estimates is low. In
contrast, the Qi value for ANC (61.3 /ueq L~1) in this
area was the second highest estimated for geomor-
phic units in the Central Rockies. The median Ca+2
value in the Beartooths (69.7/ueq L~1) was the second
lowest among the geomorphic units in the Central
Rockies. Lakes in the Beartooth Uplift had the lowest
Q4 value for DOC (0.84 mg L"1) among all geomorphic
units. All the lakes sampled had DOC < 2.0 mg L"1,
and the median DOC value (0.60 mg L'1) in the
Beartooths was as low as the Qi values in other
737
-------
Figure 7-4. Locations of sampled lakes within geomorphic units in the Central Rockies (4D), Western Lake Survey - Phase I.
Lakes located outside the boundaries were not included in estimates for these subpopulations.
4D
ID
Teton
Uplift
Beartooth
Uplift
Wasatch
Mountains
MT
Bighorn
Uplift
Cros
Ventre
Uplift
Uinta
Arch
...J
UT
132
-------
Table 7-7. Population Estimates and
Statistics for Primary Variables by Geomorphic
Unit in the Central Rockies (4D).
Western Lake Survey - Phase 1
Geomorphic Unit n"* N
Beartooth Uplift 12 212
Yellowstone Plateau 1 5 233
Bighorn Uplift 1 1 203
Wind River Range 47 884
Uinta Arch 27 503
Geomorphic Unit n*** N
Beartooth Uplift 12 212
Yellowstone Plateau 15 233
Bighorn Uplift 1 1 203
Wind River Range 47 884
Uinta Arch 27 503
ANC<5(
PC
0.157
0.128
0.097
0.062
0.039
SO/2 >
PC
0.000
0.192
0.000
0.000
0.036
0 fjeq L"1
Nc Ncu
33 71
30 63
20 51
55 105
20 51
50 ,ueq L~1
Nc Ncu
0 -
45 85
0
0
18 48
ANC (neq
0-1
61.3
58.5
52.7
74.0
59.3
S04
QI
13.7
8.2
18.1
21.7
19.5
M
92.3
175.1
82.4
109.3
83.4
^ (Meq
M
20.8
12.6
26.2
24.3
23.7
I'1)
Q4
155.6
705.7
145.4
167.6
112.8
U-1,
Q4
33.6
33.3
35.4
31.3
32.9
Ca+2
PC
0.157
0.212
0.097
0.062
0.000
DOC
PC
1.000
0.617
0.755
0.646
0.524
<50/ueq
Nc
33
50
20
55
0
< 2 mg L
Nc
212
144
153
571
264
LM Ca+2 (Meq
NCU
71
96
51
105
-
-1
Ncu
308
218
235
693
369
QI
48.8
43.8
54.6
70.5
65.3
DOC
QI
0.50
0.80
0.51
0.58
0.79
M
69.7
137.0
67.7
93.8
81.6
(mg L
M
0.60
1.63
0.94
1.17
1.79
L-')
Q4
117.6
426.0
115.3
140.7
93.8
.-,,
Q4
0.84
5.42
2.48
3.19
3.37
n*" = number of lakes from which samples were obtained.
N = estimated number of lakes.
pc = estimated proportion of lakes meeting the
given criteria
f\lc = estimated number of lakes meeting the given criteria.
Ncu = 95 percent upper confidence limit for Nc.
QT = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
( — ) = undefined.
Table 7-8. Population Statistics for Selected Variables for Lakes in Geomorphic Units in the Central Rockies (4D), Western
Lake Survey - Phase I
Geomorphic Unit
Beartooth Uplift
Yellowstone Plateau
Bighorn Uplift
Wind River Range
Uinta Arch
Geomorphic Unit
Beartooth Uplift
Yellowstone Plateau
Bighorn Uplift
Wind River Range
Uinta Arch
Conductance
(^S cm"1)
Q! M
8.9 12.0
9.2 23.4
8.3 11.5
12.4 15.0
11.5 15.1
Site Depth
(m)
Q! M
6.3 11.8
4.2 6.1
5.5 8.6
6.6 10.9
3.2 5.7
Mg+2 Na+
Q, M
20.4 37.0
18.3 50.1
14.1 27.5
15.0 22.2
26.1 34.7
Lake Area
(ha)
Q! M
2.9 6.3
2.8 4.3
2.7 4.7
4.0 6.4
2.3 3.4
0-1
8.3
15.3
8.7
15.4
11.2
WA:
QI
28.6
8.0
17.4
13.1
11.7
M
12.0
31.4
16.5
22.9
18.7
:LAa
M
38.6
23.5
41.6
22.5
30.1
K+ Cr SiO2
-------
The lakes with the maximum (4948.6 /jeq L 1) and
the minimum (-24.0 (jeq L"1) ANC measured in the
WLS-I occurred in the Yellowstone Plateau. Of all
lakes sampled in the geomorphic units in the Central
Rockies, those in the Yellowstone Plateau had the
highest median and Q4 values for ANC, pH, and Ca+2.
However, this area was the only one containing lakes
with pH < 5.0, and it had the highest percentage
of lakes with Ca+2 < 50 fjeq L"1 (21.2%, Table 7-7)
in the Central Rockies. The Yellowstone Plateau also
had the second highest percentage of lakes with ANC
< 50 /jeq L"1 (12.8%, Table 7-7). This observed
variability in the population estimates for ANC, pH
and Ca+2 can be explained in part by the variability
in geology and the variety of rock types subject to
mineral weathering (Alt and Hyndman 1972).
First quintile and median S04"2 concentrations in
Yellowstone lakes were the lowest in the Central
Rockies (8.2 and 12.6 /jeq L~1, respectively. Table
7-7); however, the Yellowstone Plateau contained
the highest percentage of lakes with S04~2 > 50
/jeq L"1. Some of the lakes on the Yellowstone
Plateau are probably influenced by natural emissions
of sulfur or contact with sulfate-rich groundwater
as evidenced by the extremely high S04~2 concen-
trations in lakes adjacent to active hot springs. The
Yellowstone Plateau had the highest percentage of
lakes (19.2%) with DOC > 6 mg L"1 of any geomorphic
unit in the WLS-I.
The Bighorns are anticlinal uplifted, with Precam-
brian bedrock that rises between the arid Bighorn
and Powder River basins. The granitic core of the
uplifted dome, which forms a broad plateau, is
surficially exposed. The mountains are relatively dry,
receiving less than 63.5 cm of precipitation annually
(Hoffman and Alexander 1976). Several areas are
above timberline, which is located at approximately
2900 m (Arno and Hammerly 1984). Eleven lakes
were sampled in the Bighorns. Within the Central
Rockies, lakes in the Bighorns had the lowest Qi
value for ANC (52.7 fjeq L"1), but only 9.7 percent
of the lakes in the Bighorns had ANC < 50 //eq L"1
(Table 7-7). Sulfate concentrations were slightly
higher in lakes in the Bighorns than in other
geomorphic units in the Central Rockies. No lakes
sampled in the Bighorn Uplift had DOC > 6 mg L~1.
7.5 Southern Rocky Mountains (4E)
The Southern Rocky Mountains (Subregion 4E) is
a continuous high-mountain barrier extending for
about 725 km from Laramie, Wyoming in the north
to Santa Fe, New Mexico in the south. These
mountains are the highest in the Rocky Mountains.
All mountain ranges here have peaks exceeding
3650 m and 55 peaks exceed 4260 m (Arno and
Hammerly 1984). The estimated target population
size (N) in the Southern Rockies was 1609 (Table
5-5). The median lake elevation (3264 m) in the
Southern Rockies was the highest of the five western
subregions surveyed. This subregion included the
lake with the highest elevation (ID number = 4E3-
053, 3883 m). Twenty percent of the lake population
exceeded 3552 m in elevation. Lakes in the Southern
Rockies were also the smallest, with a median area
of 3.5 ha and a Q4 area of 7.1 ha. They were also
the shallowest (Q-, = 2.8 m, M = 5.1 m, and Q4 =
10.4 m) (Table 5-6).
The highest ANC values for any subregion in the
WLS-I were observed in the Southern Rockies. Only
4.6 percent of the lakes were estimated to have ANC
< 50 /jeq L"1 and only 39.4 percent of the lakes
had ANC < 200 //eq L"1 (Table 5-9). The median
ANC value (317.0 fjeq L"1) for this subregion was
the highest in the West. Similarly, the Qi for ANC
(113.9 (Jeq L"1) was the highest of any WLS-I
subregion (Table 5-16).
The Southern Rockies was one of two subregions
in the West (the other was the Northern Rockies,
4C) in which no sampled lakes had pH < 6.0; the
minimum pH observed in this subregion was 6.02
(Volume II). The median pH was 7.60, the highest
estimated for any western subregion (Table 5-1 6).
Sulfate concentrations were also higher in the
Southern Rockies than in any other subregion (M
= 34.6 peq L"1). This concentration was approxi-
mately five times greater than the median S04~2 in
California (6.6 //eq L"1), which was the lowest
estimated for the West (Table 5-16). Wind-blown
particles and dissociation products of calcium
carbonate and gypsum from the Great Basin may
be the source of the Ca+2 and S04~2 in precipitation
samples collected in the Southern Rocky Mountains
(Reddy and Claassen 1 985; Creager 1 985). Although
the existing NADP/NTN precipitation monitoring
stations in the Southern Rocky Mountains are not
located at high elevations, Hidy and Young (1986)
have calculated that for the existing low elevation
sites, Ca+2 in precipitation is strongly associated with
SO4~2 (r = 0.92). The strength of this association
supports the assumption that some of the lake water
S04"2 in the Southern Rocky Mountains is related
to atmospheric deposition of gypsum, especially in
alpine areas. Lewis et al. (1984) report that there
is evidence that strong acid anion deposition occurs
in Colorado and that strong acid precursors are
transported greater distances and to greater
elevation than are carbonates and cations such as
Ca+2 and Mg+2. However, Lewis et al. (1984)
evaluated data from 42 deposition collectors, only
one of which was above 3000 m. Clearly, more
deposition data from higher elevations, more closely
related to high elevation lakes, is required to
134
-------
understand this issue. Calcium concentrations in the
lakes of the Southern Rockies were the highest jn
the West with a median value of 233.1 /ueq L~1,
almost five times the median Ca+2 value of 43.2
//eq L"1 estimated for California (Table 5-16).
In the Southern Rockies, as elsewhere in the West,
there was a strong relationship between ANC and
[Ca+2 + Mg+2] (Table 6-4). Median concentrations for
K+ (9.0 A
-------
Figure 7-5. Locations of sampled lakes within geomorphic units in the Southern Rockies (4E), Western Lake Survey - Phase I.
Lakes located outside the boundaries were not included in estimates for these subpopulations.
WY
Medicine Bow
Range
Park
Range
CO
Front
Range
4E
S'a watch Uplift
Sangre de Cnsto
Uplift
San Juan
Mountains
136
-------
Table 7-9. Population Estimates and Statistics for Primary Variables by Geomorphic Unit in the Southern Rockies (4E),
Western Lake Survey - Phase I
ANC < 50 /jeq L'1 ANC
Geomorphic Unit n*** N pc Nc Ncu QI
Park Range 21 134 0.058 8 16 82.0
Front Range 51 311 0.126 39 53 74.4
White River Uplift 10 172 0.000 0 - 130.7
Sawatch Uplift 11 252 0.000 0 - 103.2
San Juan Mountains 21 247 0.092 23 37 72.2
SO/2 > 50 /ueq L'1 SO4~
Geomorphic Unit n*** N pc Nc Ncu Q,
Park Range 21 134 0.222 30 78 13.3
Front Range 51 311 0.338 105 188 20.5
White River Uplift 10 172 0.174 30 78 8.2
Sawatch Uplift 11 252 0,391 99 181 35.6
San Juan Mountains 21 247 0.194 48 98 13.9
(pieq L~1) Ca+2 < 50/veq L 1 Ca+2 (jieq L 1)
M Q4 pc Nc Ncu Q, M Q4
345.11621.8 0.067 9 18 70.5 303.01486.1
337.5 564.3 0.105 33 45 63.9 222.6 478.1
509.7 1068.8 0.000 0 - 79.2 314.5 672.6
327.91216.9 0.000 0 - 100.3 306.6 977.9
124.3 233.3 0.037 9 18 67.5 129.3 369.8
2 (^eq L'1) DOC<2mgL"1 DOC (mg L'1)
M Q4 pc Nc Ncu Q, M Q4
24.6 264.9 0.290 39 56 1.44 3.40 5.59
35.2 90.2 0.647 201 271 0.84 1.51 3.37
10.3 22.4 0.426 73 142 0.87 2.12 3.90
42.6 118.6 0.645 163 268 0.66 1.37 2.29
21.9 44.8 0.740 183 279 0.68 1.09 1.88
n*** = number of lakes from which samples were obtained.
N = estimated number of l.akes.
gc = estimated proportion of lakes meeting the given criteria.
Nc = estimated number of lakes meeting the given criteria.
Ncu = 95 percent upper confidence limit for Nc.
Q! = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
( — ) = undefined.
Table 7-10. Population Statistics for Selected Variables for Lakes in Geomorphic Units in the Southern Rockies (4E),
Western Lake Survey - Phase I
Conductance
Mg + 2
(Meq L-1)
Na +
(jieq L^1)
K +
cr
SiO2
Fe
Geomorphic Unit
M
0^
M
Q,
M
Q,
M
M
M
M
Park Range
Front Range
White River Uplift
Sawatch Uplift
San Juan Mountains
Geomorphic Unit
Park Range
Front Range
White River Uplift
Sawatch Uplift
San Juan Mountains
13.0 38.7
11.9 37.5
14.2 45.7
15.8 42.2
11.5 20.2
Site Depth
(m)
Q, M
2.8 4.2
3.0 6.7
1.4 2.6
3.4 4.8
3.9 4.7
20.3 76.0
15.7 91.6
61.9 146.6
23.3 73.8
14.4 21.4
Lake Area
(ha)
Q, M
2.4 3.2
1.7 3.7
0.6 1.2
1.5 2.9
1.6 4.7
20.7
17.3
21.4
16.7
10.0
42.3
33.1
44.2
25.4
31.5
WA:LAa
QI
11.4
11.4
14.6
9.6
6.4
M
18.9
23.8
65.4
13.1
15.8
2.9 16.6
3.5 5.2
4.0 14.3
3.5 8.1
3.6 7.1
Secchi Disk
Transparency
(m)
Q! M
0.7 2.8
2.0 2.4
1.4 2.6
1.2 4.3
1.9 3.9
2.2
2.1
2.3
1.4
2.0
4.0
3.6
5.7
3.5
2.5
1.6
1.3
0.4
0.6
1.5
2.0
2.3
2.9
1.7
3.5
11.6
3.7
7.7
0.7
7.7
24.5
42.5
25.7
11.3
15.4
QI = first quintile (20th percentile).
M = median (50th percentile).
a Watershed area:lake area.
White River Uplift with ANC < 200 fjeq L"1 was 142
ha. Turk and Adams' (1983) study included lakes
less than 1 ha in area (mean = 0.4 ha) and these
smaller lakes are included in their estimate of 370
lakes with alkalinity < 200 /ueq L"1. Based on the
WLS-I, the estimated number of lakes greater than
or equal to 1 ha with ANC < 200 fjeq L"1 in the
White River Uplift is 40. Excluding lakes smaller than
1 ha from the Western Lake Survey - Phase I target
population may affect population estimates if lakes
< 1 ha are prevalent. In the White River Uplift,
however, excluding small lakes had little effect on
the estimated lake area with ANC < 200 yueq L"1.
The lakes in the White River Uplift were smaller and
shallower than in any other geomorphic unit.
In the Park Range, the bedrock consists of Cretaceous
sedimentary and Precambrian metamorphic mate-
137
-------
rial. This area had the fourth highest median ANC
in the West. The Park Range also had the second
highest median DOC concentration (3.40 mg L"1)and
the highest Q4 values for Ca+2 (1486.1 fjeq L"1) and
S04~2 (264.9 /jeq L"1) of any geomorphic unit in the
WLS-I (Table 7-9).
Lakes of the Sawatch Uplift had the second highest
0.4 value for SO.T2 of any geomorphic unit (118.6
fjeq L"1). Acid neutralizing capacity and Ca+2
concentrations were relatively high in lakes in this
area. Of the 11 lakes sampled in this geomorphic
unit, none had ANC or Ca+2 < 50 fjeq L"1 (Table 7-9).
7.6 Ionic Relationships for Lakes in the
Major Geomorphic Units
Ternary diagrams (Section 6.2.6) were developed for
the major geomorphic units in each subregion.
Examples of these are given for the Sierra Nevada
(from 4A), the Oregon Cascades (from 4B), the
Bitterroots (from 4C), the Wind River Range (from
4D), and the Front Range (from 4E; Figure 7-6). These
five geomorphic units contain the largest number
of dilute lakes estimated for any geomorphic unit
in each of the subregions. The Sierra Nevada is
underlain by a granitic batholith. The Bitterroots are
on the edge of the Idaho Batholith, whereas the
Oregon Cascades are extrusive andesite. The Wind
River Range is an uplift of Precambrian granites and
gneisses and the Front Range is composed of
Precambrian granitic and metamorphic rocks.
Lakes in the Sierra Nevada, the Bitterroot Mountains,
and the Front Range are similar in composition with
respect to proportions of major ions (Figure 7-6), and
are largely dominated by Ca+2 and HC03~, although
lakes in all three areas have moderate proportions
of Na+and K+. The proportion of Cl~ tends to be higher
in the Sierra Nevada than in the Bitterroot Moun-
tains, and the proportion of Cl~ in the Front Range
is extremely low. Lakes in the Wind River Range
have notably uniform ionic composition.
The Oregon Cascades also contain many lakes with
low ionic strength. These lakes have slightly higher
Mg+2, Na+, and Cl~ concentrations and proportions
than do lakes in the Sierra Nevada and the
Bitterroots. The proportion and concentration of
S04~2 in the Oregon Cascades are especially low.
Other areas of the Cascades, such as the Wenatchee
Mountains, have lower relative proportions of Mg+2,
Na+, and Cl~ and higher relative proportions of Ca+2
and S04~2. Lakes in the Wind River Range and the
Front Range, in the Central and Southern Rockies,
have similar ionic composition (Figure 7-6).
138
-------
Figure 7-6. Trilinear diagrams showing the relative abundance of major anions and cations for lakes in the largest geomorphic
units in California (Sierra Nevada), the Pacific Northwest (Oregon Cascades), the Northern Rockies (Bitterroot
Mountains), the Central Rockies (Wind River Range), and the Southern Rockies (Front Range), Western Lake
Survey - Phase I. Ratios are expressed as percent of total ionic concentration, increasing from 0 to 100 percent
along the axes in the direction of the arrows.
.Ca Cl
Bitterroot Range
Ca Cl
Wind River Range
139
-------
Section 8
Comparison of Results from the Western and Eastern Lake Surveys
Phase I of the Eastern and Western Lake Surveys
(ELS-I and WLS-I) collectively comprise the first
phase of the National Lake Survey, which was
designed to provide a data base that could be used
as a basis for determining the present chemical
status and location of lakes in the United States that
are at risk due to acidic deposition. The results
discussed below relate to only the areas covered by
sampling in each survey.
Data from 1592 lakes were used to generate
population estimates for eleven subregions in three
regions in the eastern United States - the Northeast,
Upper Midwest and Southeast. Because the two
subregions in the Southeast - the Southern Blue
Ridge and Florida - contained lakes with substantially
differing characteristics, a regional data summary
for the Southeast was not presented (Linthurst et
al. 1986; Overton et al. 1986; Kanciruk et al. 1986a).
The number of lakes used to generate population
estimates in the West was 719, distributed among
five subregions. Based on the number of lakes
sampled and the estimated number of lakes
comprising the map population in each region
(Linthurst et al. 1986; Section 2.3.2), the size of the
characterized lake population was 7157 in the
Northeast, 8575 in the Upper Midwest, 286 in the
Southern Blue Ridge, 2138 in Florida, and 10,393
in the West. Because of the scales of maps available
on which map populations were identified, the target
population of lakes in the ELS-I was restricted to
lakes > 4 ha, whereas the target population of the
WLS-I included lakes > 1 ha in area. Approximately
44 percent of the target population of lakes in the
West were 1-4 ha, but there was little difference
in lake chemistry between these lakes and those
> 4 ha (Section 6.4.1). Consequently, the compar-
isons between the lakes of the East and West
discussed here include those lakes in the West
between 1 and 4 ha. The summary that follows
highlights the results of the two surveys with respect
to physical and chemical characteristics of regional
lake populations.
8.1 Physical Characteristics
Lakes in the western United States were much
higher in elevation than lakes in the eastern United
States; the Qi value for elevation in the West (Table
8-1) was higher than all Q4 values in the East
(Linthurst et al. 1986). Watershed areas for western
lakes were smaller than for lakes in the Northeast,
Upper Midwest, and Southern Blue Ridge, but were
slightly larger than for lakes in Florida; lake area
was smaller in the West than in all eastern regions.
Watershed area to lake area ratios were much
greater in the West than in the Northeast, the Upper
Midwest, and Florida, but less than in the Southern
Blue Ridge. Despite their smaller size, western lakes
were deeper than eastern lakes. In the West, seepage
lakes were more common and reservoirs and closed
lakes were less common than in the Northeast.
Compared to the Upper Midwest and Florida,
seepage lakes and closed lakes were less common
and drainage lakes and reservoirs were more
common in the West (Table 8-1).
Land use/land cover differed markedly between
watersheds in the East and West (unpublished data,
EPA Environmental Research Laboratory, Corvallis;
Section 5.3). Urban and agricultural land use was
much less prevalent in western watersheds. No
western watersheds were dominated by wetlands
whereas wetlands dominated in one to six percent
of the watersheds in six eastern subregions.
Rangeland was more common in the West than in
the East. Barren land (exposed rock), tundra, and
perennial snow or ice collectively dominated 12-58
percent of the watersheds in each western subre-
gion; in contrast only one to four percent of the
watersheds in three eastern subregions were
dominated by barren land and no watersheds
contained tundra or perennial snow or ice. Total
forest cover was similar in eastern and western
watersheds; however, forests were almost exclu-
sively evergreen in the West, whereas eastern
forests were more likely to be mixed coniferous-
deciduous or deciduous.
8.2 Primary Variables
The population distribution for acid neutralizing
capacity (ANC) for lakes in the western United States
was most similar to that for lakes in the Northeast
(Figure 8-1). Florida, however, had a higher
percentage of lakes with low ANC and a higher
percentage of lakes with high ANC than did the West,
140
-------
Table 8-1. Physical Characteristics of Lakes Sampled During Phase I of the Eastern and Western Lake Surveys8
Lake Elevation (m)
Q, M CL
Lake Area (ha)
, M C
Watershed Area (ha) Watershed Area:Lake Area
M
M
Northeast 1
Upper Midwest 2
Southern Blue Ridge 3A
Florida 3B
West 4
110
339
221
15
1746
307
408
265
21
2613
491
487
454
31
3237
8.0
7.6
5.2
6.5
2.2
16.7
14.8
10.8
17.3
4.6
62.7
46.1
64.3
44.1
11.7
90
56
182
35
45
271
177
682
115
118
1289
1023
5836
352
436
6.2
4.3
20.7
3.5
11.0
13.0
10.3
44.2
5.5
23.7
38.6
30.3
127.9
12.0
58.5
Site Depth (m)
, M C
Hydrologic Lake Type (Percent of Population)
Drainage
Seepage
Reservoir
Closed
Northeast
Upper Midwest
Southern Blue Ridge
Florida
West
1
2
3A
3B
4
1.6
2.2
2.3
1.5
3.8
4.2
5.6
4.8
2.7
7.8
8.3
10.9
12.5
5.2
17.7
71
50
7
21
74
7
43
0
66
18
17
2
90
0
7
5
5
3
13
2
QI = first quintile (20th percentile).
M = median (50th percentile).
Q4 = fourth quintile (80th percentile).
aData for Regions 1 and 2 and Subregions 3A and 3B are from Linthurst et al. (1986).
Figure 8-1. Cumulative frequency distribu-
tions [F(x>] for ANC (/jeq L~")
and pH for the Northeast
( ), the Upper Midwest
( ), the Southern Blue
Ridge f ). and Florida
( ). Eastern Lake Survey -
Phase I (Linthurst et al. 1986),
and the West ( ). Western
Lake Survey - Phase I.
1.0
0.8-
0.6-
FM
0.4-
0.2-
0.0
Northeast (1)
Upper Midwest (2)
Southern Blue Ridge (3A)
Florida (3B)
West (4)
-100
200
I
400
I
500
500
WOO
ANC (ueq O
Southern Blue Ridge (3A)
Florida (3B)
West(4) -'I I
r I
141
-------
as reflected by the Qi and median ANC values (Figure
8-2).
With the exception of one lake associated with hot
springs, there were no acidic lakes (ANC < 0 //eq L~1)
sampled in the WLS-I (Figure 8-3). In contrast, 4.6
percent of the lakes in the Northeast, including 10.7
percent in the Adirondacks, were acidic. Of the lakes
in the Upper Midwest, 1.7 percent had ANC < 0
yueq L~1, ranging from none sampled in Northeastern
Minnesota to 9.8 percent in the Upper Peninsula
of Michigan (Linthurst et al. 1 986). None of the lakes
in the Southern Blue Ridge was estimated to have
ANC < 0 yueq L"1, but 22 percent of the lakes in
Florida were estimated to be acidic.
The percentage of lakes estimated to have ANC < 50
/ueq L~1 in the West was similar to that estimated
for the Northeast, but was roughly one-half that for
Florida (Figure 8-3). The percentage of lakes in the
West with ANC < 200 /ueq L"1 (66.6%) was the
highest estimated for any region. Estimates of lakes
in this ANC class were 60 percent for the Northeast,
55.1 percent for Florida, 41.4 percent for the Upper
Midwest and 34.3 percent for the Southern Blue
Ridge. The percentages of lakes with ANC < 200
fjeq L"1 in California, the Pacific Northwest and the
Central Rockies were greater than for any subregion
sampled in the eastern United States. In the East,
lakes with low ANC were generally much smaller
than those with high ANC (Linthurst et al. 1986).
In contrast, lakes in California were the most dilute
and had the lowest values of ANC. California lakes
were generally larger and were among the deepest
sampled in the West.
Most lakes surveyed in the western and eastern
United States had pH > 5.0 (Section 5.4.2; Linthurst
et al. 1986). Values for pH were generally higher
in the West than in other areas, but, as was observed
for ANC, considerable variation in the population
distributions existed (Figure 8-1). The most marked
differences were observed at low pH values. First
quintile values for the West and the Southern Blue
Ridge were similar and were higher than in any other
area (Figure 8-2). Approximately one percent of the
lakes in the West and 0.4 percent in the Southern
Blue Ridge had pH < 6.0 (Figure 8-3). In contrast,
12.9 percent of the lakes in the Northeast, 9.6
percent of the lakes in the Upper Midwest and 32.7
percent of the lakes in Florida had pH < 6.0.
Figure 8-2.
Concentrations of ANC (neq L'''), pH. and concentrations of sulfate (/jeq L 1) for lakes in the Northeast (Region
1). Upper Midwest (Region 2). Southern Blue Ridge (Subregion 3A), and Florida (Subregion 3B), Eastern Lake
Survey - Phase I (Linthurst, et al. 1986). and West (Region 4). Western Lake Survey - Phase I. First quintile
fQi. 20th percentile), median (M, 50th percentile). and fourth quintile (Qt, 80th percentile) values are shown
o-
to
7200
7000
800
600
400
200
0
-100
_ O
O
-H
H—h
2 3A 3B 4
142
ej.
8.0
7.5
7.0
6.5
6.0
5.5
5.0
1
04
M
O _
H — I — I — I — H
1 2 3A 3B 4
280
240
200
160
120
80
40
O -•
H—I—I—h
1 2 3A 3B 4
-------
Figure 8-3. Population estimates for the
percent of lakes with concen-
trations ofANC <0,< 50. and
< 200 ueq L~\ and pH < 6.0,
for the Northeast, Upper Mid-
west, Southern Blue Ridge,
and Florida, Eastern Lake Sur-
vey - Phase I (Linthurst et al.
1986). and West, Western Lake
Survey - Phase I.
T
ANC<0
ueq L'\
ANC < 50
ANC < 200 fjeq L
pH < 6.0
20
B Northeast (1)
• Upper Midwest (2)
D Southern Blue Ridge (3A>
Q Florida (3B)
West (4)
T
40 60
Percent
80
100
A striking contrast among the distributions for
primary variables between lakes in the western and
eastern United States was observed f or S04~2 (Figure
8-4). The distributions for S04~2 in the West closely
paralleled that observed for the Southern Blue Ridge
and, in both these areas, S04~2 concentrations were
much lower than in the Northeast, Upper Midwest
and Florida. The Q4 value for S04~2 in lakes in the
West was 38.7 yueq L~1, less than one-half the Qi
value in the Northeast (Figure 8-2). In the Northeast,
97.3 percent of the lakes had S04"2 > 50 ,ueq L~1,
while only 13.5 percent of the lakes in the West
had S04"2 > 50 /ueq L"1 (Figure 8-5).
The population distributions for Ca+2 in western and
eastern lakes generally paralleled those for ANC
(Figure 8-4). At lower Ca+2 concentrations, the
distribution in the West was most similar to that
in the Southern Blue Ridge, but, at higher concen-
trations, the distribution in the West was practically
indistinguishable from that for the Northeast. This
comparison is also reflected in the median and Qi
values for the West and the Southern Blue Ridge
and in the Q4 value for the West and the Northeast
(Figure 8-6). More lakes in the West had Ca+2 < 50
/ueq L"1 than in any area in the eastern United States
(Figure 8-5).
Extractable Al in clearwater lakes (true color < 30
PCU) was low among most areas surveyed. The
distribution for extractable Al for western lakes was
intermediate between that for the Southern Blue
Ridge and those for the Northeast, Upper Midwest
and Florida; the distributions in the latter three areas
were indistinguishable (Figure 8-7). The lowest Qi
value was observed for clearwater lakes in the Upper
Midwest, while the highest occurred in the Northeast
(Figure 8-6); the median value in the West was
similar to that in the Upper Midwest. Less than one
percent of the clearwater lakes in the West had
extractable Al > 50 /JQ L~1 and no lakes in this
category were sampled in the Southern Blue Ridge
(Figure 8-5).
Dissolved organic carbon differed markedly among
eastern and western lakes. The distribution for DOC
in the West most closely paralleled that in the
Southern Blue Ridge (Figure 8-7). Concentrations of
DOC in these areas were markedly less than in other
areas surveyed (Figure 8-6). Nearly 70 percent of
the lakes in the West had DOC < 2 mg L~1, whereas
less than ten percent of the lakes in the Northeast
and Upper Midwest had DOC < 2 mg L"1 (Figure
8-5). The West also contained very few systems with
high DOC (> 6 mg L~1), similar to the Southern Blue
Ridge, but unlike the Upper Midwest and Florida
where more than 60 percent of the lakes had DOC
> 6 mg L"1 (Figure 8-5). The West contained many
more clearwater lakes with low DOC than did areas
in the eastern United States, and Secchi disk
transparencies were correspondingly higher (Sec-
tion 5.6.2; Linthurst et al. 1986).
Many lakes in the West had low ionic concentration
(Section 5.9). The lowest Qi value (20th percentile)
for conductance in the East was 20.7 fjS cm"1
(Linthurst et al. 1986). In contrast, 20 percent of
143
-------
Figure 8-4. Inverse cumulative frequency
distributions [1 -F(x)] for sulfate
f/jeq L~^t and cumulative fre-
quency distributions [F(x)J for
calcium (/jeq L~^) for the North-
east ( ), Upper Midwest
( ), Southern Blue Ridge
I ), and Florida f ). 1-nxc>
Eastern Lake Survey - Phase I
(Linthurst et al. 1986), and
West ( ). Western Lake
Survey - Phase I. 0.2 -
Northeast (1)
Upper Midwest (2)
Southern Blue Ridge <3A)
Florida (3B)
West 14)
1.0-
F(x)
0.8-
0.6-
0.4-
0.2-
0.0-
720 180
Sulfate ffjeq L'^)
240
300
Northeast (1)
Upper Midwest (2)
Southern Blue Ridge (3A)
Florida (3B)
West (4)
200 400 600
Calcium ffjeq L~^)
800
WOO
Figure 8-5. Population estimates for the
percent of lakes with concen-
trations of calcium < 50
Lieq L~\ sulfate > 50 iieq L~\
DOC <2 mg L~\ DOC > 6 mg
L~\ and extractable aluminum
> 50 fjg i~1 for the Northeast.
Upper Midwest, Southern Blue
Ridge, and Florida. Eastern
Lake Survey - Phase I (Linthurst
et al. 1986) and West, Western
Lake Survey - Phase I.
I ' T ' I
53 Northeast (1)
• Upper Midwest (2)
0 Southern Blue Ridge (3A)
Florida (3B)
D West (4)
DOC > 6 mg L
Ext.A/>50/jgL~-
100
144
-------
Figure 8-6. Concentrations of calcium (/MO, L 1A extractable aluminum ffjg L'^) in clearwater lakes, and DOC (mg L'^) for
lakes in the Northeast (Region 1), Upper Midwest (Region 2), Southern Blue Ridge (Subregion 3A), and Florida
(Subregion 3B). Eastern Lake Survey - Phase I (Linthurst et at. 1986), and West (Region 4). Western Lake Survey
- Phase I. First quintile (Q,. 20th percentile), median (M. 50th percentile), and fourth quintile (Q*. 80th percentile)
values are shown (O—I—D ).
Q, M Q4
1000
800 -
600 -
a
400 -
200 -
C
-
~ 0
'_
<
•> <
p
C
' <
1 — I — I
[
p
> ,
p
-
C
0 C
p
'_
. -
>
I
I
§
I
1 2 3A 3B 4
20
18
16
14
12
10
g
6
4
2
0
r
-
-
~ C
-
—
-
—
(
p
C
}
p
J
1 J
p
-
-
—
-
D
—
-
_
} .
k 7 , O
1 2 3A 3B 4
14
12
10
~J
I
O
O
• • Q
1 2 3A 3B 4
the lakes in the West had conductance^ 8.3//S cm"1,
and the same percentage in California had
conductance < 4.9 fjS cm"1 (Section 5.6.5). These
dilute western lakes had very low concentrations of
dissolved constituents, such as calcium, magnesium,
sulfate, and nutrients (Section 5.9).
8.3 Associations Among Variables
Many lakes in the West occurred at relatively high
elevations, particularly in the Sierra Nevada in
California, the Central Rockies, and the Southern
Rockies (Section 5.3). There was a moderately strong
inverse relationship between elevation and ANC in
the Sierra Nevada and Central Rockies, but there
was little or no relationship in the other areas
(Section 6.4.2). The only area in the East where a
strong positive relationship between elevation and
ANC was found was the Adirondacks, although few
eastern areas had appreciable elevational gradients.
Seepage lakes had the lowest values of Ca+2 and
ANC in the Upper Midwest and in the Northeast;
drainage lakes had the lowest ANC in the Northeast
(Linthurst et al. 1986). In California and the Pacific
Northwest, drainage lakes had higher ANC and Ca+2
than did seepage lakes, but the reverse was true
for the subregions in the Rockies (Section 6.4.3).
The sums of Ca+2 and Mg+2 for lakes in the western
subregions were much lower than for lakes in the
eastern subregions (Section 6.2.2; Linthurst et al.
1986). Lakes in several areas of the West had
concentrations of base cations that were comparable
to some of the most dilute lakes in the world
(Armstrong and Schindler 1971).
Most lakes in the West were dominated by calcium
bicarbonate (Section 6.2.5). In some lakes, Na+ was
the dominant cation and HC03" or CI" was the
dominant anion. A few lakes were sampled that had
Ca+2 and S04~2 as the dominant ions. In contrast.
145
-------
Figure 8-7. Inverse cumulative frequency
distributions [1-F(x)] for
extractable aluminum fog L~^)
in clearwater lakes, and cumu-
lative frequency distributions
[Ffxj] for DOC (mg i"V for the
Northeast ( ). Upper
1.0
0.8-
0.6-
Midwest ( ). Southern l-F(x)
Blue Flidge ( ), and Florida
( ), Eastern Lake Survey -
Phase I (Linthurst et al. 1986)
and West f ). Western
Lake Survey - Phase I.
0.4-
0.0
Northeast (1)
- — Upper Midwest (2J
Southern Blue Ridge (3Aj
Florida (3B)
West (4)
10
Extractable Aluminum (ug L~
Southern Blue Ridge (3A)
Florida (3B)
West(4)
DOCfmgL"1)
S04"2 was the dominant anion in 20 percent of the
lakes in three of five subregions in the Northeast,
in the Upper Peninsula of Michigan, and in
Northcentral Wisconsin (Linthurst et al. 1986).
Sulfate was also high in Florida lakes.Many more
lakes in the East had Na+ and Cl~ as major ions.
In general, lakes in the West had lower concentra-
tions of Ca+2, Mg+2, S04"2, and DOC than did lakes
in the East. Many very dilute lakes were found,
particularly in the Sierra Nevada, the Cascades and
the Wind River Range. These lakes have very low
concentrations of dissolved constituents and are
highly transparent.
8.4 Discrete Subpopulations of Lakes
in the East and West
By combining lake populations from entire regions,
important differences among groups containing the
most low ANC lakes are sometimes obscured. Four
subpopulations containing the highest proportion of
acidic lakes were selected from the East for
comparison with five previously described geomor-
phic units in the West (Sections 7.1-7.5). The
Adirondack Park represents the estimated lake
population within the park boundary. The eastern
half of the Upper Peninsula of Michigan describes
those lakes in the Peninsula east of longitude 87°.
Northcentral Wisconsin is Subregion 2C as shown
in Linthurst et al. (1 986). Northcentral Florida is also
described in Linthurst et al. (1986) and closely
coincides with the area known as the Trail Ridge.
The population statistics for pH clearly demonstrate
the major differences between the eastern and
western subpopulations, with Qi values differing by
more than one pH unit (Figure 8-8). However, when
compared on the basis of ANC, the eastern and
western subpopulations are similar (Figure 8-8). The
large interquintile differences in ANC for lakes in
Michigan, Wisconsin, and Florida compares closely
with those in the Bitterroot Mountains and the Front
Range. Small interquintile differences for ANC were
observed in the Adirondack Park, the Sierra Nevada,
Oregon Cascades, and Wind River Range. Such low
variability suggests a relatively high degree of
chemical and physical homogeneity within these
geomorphic units. Differences in S04 2 concentra-
tions in the eastern and western subpopulations
146
-------
Figure 8-8. Concentrations of ANC f/jeg L'^) and pH in selected subpopulations of lakes located in major geomorphic units
in the Northeast, Upper Midwest, and Florida. Eastern Lake Survey - Phase I (Linthurst et al. 1986) and West,
Western Lake Survey - Phase I. First quintile (Q,. 20th percentile), median (M, 50th percentile). and fourth
quintile (Qt. 80th percentile) values are shown (O—t—O)
Qi M Q4
Adirondack Park,
New York
Eastern Upper Penin-
sula of Michigan
Northcentral
Wisconsin
Northcentral
Florida
Sierra Nevada,
California
Oregon Cascades
Bitter root Mtns.,
Idaho/Montana
Wind River Range,
Wyoming
Front Range,
Colorado
„*•« jv
62 54U
loo J4BU
50 963 ~
114 2119-
42 443 ~
J/ 283
47 884 -
(
i | i | i | i | i | i
OH D
OH D
. 1 , 1 1 1 1 1 I 1 1 1
3 200 400 600 SOO 1000 1200 I
I I I
o — ho
o-H-a
I I I
5 6 7 8
ANC(ueqL")
PH
were also striking (Figure 8-9). Only in the Front
Range in Colorado did concentrations approach
those observed in the East.
Large differences in DOC concentrations between
lakes in the East and West were also observed (Figure
8-9). Dissolved organic carbon can be used to
estimate organic acid concentration, which may be
much more important in the chemistry of eastern
lakes than in western lakes. As expected, the pattern
observed for Ca+2 concentrations in these subpop-
ulations (Figure 8-9) closely mirrors that for ANC.
147
-------
Figure 8-9. Concentrations of sulfate (fieq L'1). calcium (/jeq L '), and DOC (mg £~V '" selected subpopulations of lakes
located in major geomorphic units in the Northeast, Upper Midwest, and Florida. Eastern Lake Survey - Phase I
(Linthurst et al. 1986) and West, Western Lake Survey - Phase I. First quintile (Q,, 20th percentile). median
(M, 50th percentile), and fourth quintile (Qt, 80th percentile) values are shown ( Q 1—D )
Q, M Q<
/i"* N
Adirondack Park. 127 1091-
New York
Eastern Upper Penin- 62 540 .
sula of Michigan
Northcentral
Wisconsin
Northcentral
Florida
Sierra Nevada,
California
153 1480-
50 963 -
114 2779-lK)
Oregon Cascades 42 443-D
Bitterroot Mtns ,
Idaho/Montana
Wind River flange,
Wyoming
Front Range,
Colorado
37 283-
47 884-
51 311-
o-f-Q
0-H3
419
I i I
1 1 1 1 1 1 1
1020
876
H-Q
H-a
0+-D
1 1 I 1 1 1 I 1
' 1 ' 1 ' 1 *
H-o
o-K3
o4-n
o-| a
i 1 i 1 i 1 i
0 100 200 300
Sulfate (yeqL"11
0 200 400 600 800 0
Calcium (fieq i"V
4 8 12
DOC(mgL^)
16
148
-------
Section 9
Conclusions
Within the areas of the western United States
sampled and covered by the design of the Western
Lake Survey it is concluded that:
• More than one-fourth (26.6 percent) of the
western lakes are dilute (conductance < 10
fjS cm"1). The areas sampled in the western United
States contain a much higher percentage of lakes
with low concentrations of dissolved substances
than do areas sampled in three regions (Northeast,
Upper Midwest, and Southeast) in the eastern
United States. The lowest Qi value (20th percen-
tile) for conductance measured in the East was
20.7 fjS cm"1. In contrast, 20 percent of the lakes
in the West had conductance values < 8.3 fjS cm"1
and the minimum measured conductance in the
West was 1.6 /uS cm"1.
• Calcium concentrations in the West varied widely
among subregions; 56.5 percent of the lakes in
California had calcium < 50 /jeq L"1 whereas only
3.4 percent of the lakes in the Southern Rockies
were below this criterion. Overall, the median
concentration of calcium for lakes in the West
was low (92.4 fjeq L"1) compared to the Northeast
(177.4 fjeq L"1), the Upper Midwest (238.2 fjeq L"1),
and Florida (238.3 fjeq L"1). Lakes in the Southern
Blue Ridge had a comparable median value for
calcium (104.7 //eq L"1).
• Median sulfate concentrations were extremely
low throughout the West, ranging from 6.6 yueq L"1
in California to 34.6 fjeq L"1 in the Southern
Rockies. In comparison, median concentrations
for lakes in the East were 1 15.4 fjeq L"1 for the
Northeast, 57.1 fjeq L"1 for the Upper Midwest,
31.8 fjeq L"1 for the Southern Blue Ridge, and
93.7 fjeq L"1 for Florida.
• Extractable aluminum concentrations in clear-
water lakes were very low throughout the West.
Only 0.2 percent of the lakes in the West had
extractable aluminum concentrations exceeding
50 //g L"1 compared to 5.5 percent of the lakes
in the Northeast, and 7.4 percent of the lakes in
Florida. These results are, again, similar to those
for the Southern Blue Ridge where no lakes
exceeded this criterion.
• Concentrations of dissolved organic carbon (DOC)
were also low throughout the West; only 5.4
percent of the lakes in the West had DOC values
exceeding 6 mg L"1, whereas this criterion was
exceeded for 26.4 percent of the lakes in the
Northeast, 62.9 percent in the Upper Midwest,
and 68.9 percent in Florida. Again, the Southern
Blue Ridge shows a similarity to western lakes
with 6.1 percent of the lakes with DOC > 6 mg L"1.
• pH values were not low in the West, where 99
percent of lakes had values greater than 6.0.
Median pH values for the subregions ranged from
6.94 in California to 7.60 in the Southern Rockies.
• Of the subregions sampled in the West, California
had the largest number (881) and percentage
(36.7%) of lakes with low ANC (< 50 fjeq L"1);
followed by the Pacific Northwest (333 lakes;
19.5%). The Southern Rockies had the smallest
number and the lowest percentage of lakes with
low ANC (74 lakes; 4.6%).
• The lakes in wilderness areas had much lower
concentrations of ANC with a median value of
91.4 fjeq L"1 compared to 282.7 fjeq L"1 for non-
wild.erness lakes.
• The Sierra Nevada (the area which contained the
majority of lakes sampled in California) had the
largest estimated number of low ANC lakes (834)
and dilute lakes (1311) of any geomorphic unit
in the West.
• No lakes sampled in the West had ANC <0//eq L"1,
with the exception of one lake associated with
a hot spring.
• Evaluation of ion ratios, the relationship of base
cations to ANC, and the positive relationship of
calcium to sulfate suggest that chronic regional
lake acidification was not evident in the moun-
tainous areas of the West.
149
-------
Section 10
References
1. Aimer, B., W. Dickson, C. Ekstrom, and E.
Hornstrom. 1978. Sulphur Pollution and the
Aquatic Ecosystem, pp. 271-311. In: Sulphur
in the Environment: Part II, Ecological
Impacts, J.O. Nriagu, ed. John Wiley and
Sons, New York, New York.
2. Alt, D. D. and D. W. Hyndman. 1972.
Roadside Geology of the Northern Rockies.
Mountain Press, Missoula, Montana. 280 pp.
3. Alt, D. D. and D. W. Hyndman. 1978. Rocks,
Ice and Water. The Geology of Waterton
Glacier Park. Mountain Press, Missoula,
Montana. 104 pp.
4. Altshuller, A. P. and R. A. Linthurst, eds.
1984. The Acidic Deposition Phenomenon
and Its Effects. Critical Assessment Review
Papers, Volume II: Effects Sciences. EPA-
600/8-83-016 BF, U.S. Environmental
Protection Agency, Washington, D.C.
5. Anderson, J. R., E. E. Hardy, J. T. Roach,
and R. E. Whitman. 1976. A Land Use and
Land Cover Classification System for Use
with Remote Sensor Data. U.S. Geological
Survey Professional Paper 964. 28 pp.
6. Armstrong, F. A. J. and D. W. Schindler.
1971. Preliminary Chemical Characteriza-
tion of Waters in the Experimental Lakes
Area, Northwestern Ontario. J. Fish. Res. Bd.
Canada 28:171-187.
7. Arno, S. F. and R. P. Hammerly. 1984.
Timberline. Mountain and Arctic Forest
Frontiers. The Mountaineers, Seattle,
Washington. 304 pp.
8. Arno, S. F. 1969. Glaciers in the American
West. Natural History 78:84-89.
9. Baker, J. P. and C. L Schofield. 1982.
Aluminum Toxicity to Fish in Acidic Waters.
Water, Air, and Soil Poll. 18:289-309.
10. Baldwin, E. M. 1981. Geology of Oregon, 3rd
Edition. Kendall/Hunt Publishing Company,
Dubuque, Iowa. 170 pp.
11. Baron, J. 1983. Comparative Water Chem-
istry of Four Lakes in Rocky Mountain
National Park. Water Resources Bulletin
19:897-902.
12. Bonoff, M. B. and A. W. Groeger. 1987.
National Surface Water Survey, Western
Lake Survey - Phase I, Field Operations
Report. Internal Report. EPA/600/8-87/
018. Environmental Monitoring Systems
Laboratory, U.S. Environmental Protection
Agency, Las Vegas, Nevada, 26 pp.
13. Bortelson, G. C., R. T. Wilson, B. L. Fox-
worthy. 1977. Water quality effects on Baker
Lake of recent volcanic activity at Mt. Baker,
WA. U.S. Geological Survey Professional
Paper 1022-B. 30pp.
14. Bradford, G. R., F. L. Bair, and V. Hunsacker.
1968. Trace and Major Element Content of
170 High Sierra Lakes in California. Limnol.
Oceanogr. 13:526-530.
15. Brakke, D. F. and T. J. Loranger. 1986. Acid
Neutralizing Capacity of Lakes in the North
Cascades Area of Washington State. Water,
Air, and Soil Poll. 30:1045-1053.
16. Brakke, D. F. and D. A. Waddell. 1985. Water
Chemistry in the North Cascades of
Washington, pp. 79-101. In: Acid Precipita-
tion in the Pacific Northwest, J. H. Baldwin,
ed.. University of Oregon, Eugene, Oregon.
17. Brezinski, D. P. 1983. Kinetic, Static, and
Stirring Errors of Liquid Junction Reference
Electrodes. The Analyst 108:425-442.
18. Busby, M. W. 1966. Annual Runoff in the
Conterminous United States. Atlas HA-212
Hydrologic Investigations. U.S. Department
of the Interior. U.S. Geological Survey.
19. Chen, C. W., S. A. Gherini, N. E. Peters, P.
S. Murdock, R. M. Newton, and R. A.
Goldstein. 1984. Hydrologic Analysis of
Acidic and Alkaline Lakes. Water Resources
Research 20:1875-1882.
750
-------
20. Chronic, H. 1986. Pages of Stone. Geology
of Western National Parks and Monuments.
2: Sierra Nevada, Cascades, and Pacific
Coast. The Mountaineers, Seattle, Washing-
ton. 182pp.
21. Creager, C. S. 1985. Characterization of the
Influence of Soil Particulateson Precipitation
Chemistry at Five Sites in the Western United
States. Report No. 0785-224-F, Kilkelly
Environmental Associates, Raleigh, North
Carolina. 111 pp.
22. Drever, J. I. 1982. The Geochemistry of
Natural Waters. Prentice Hall, Englewood
Cliffs, New Jersey. 388 pp.
23. Drever, J. I. and D. R. Hurcomb. 1986.
Neutralization of Atmospheric Acidity by
Chemical Weathering in an Alpine Drainage
Basin in the North Cascade Mountains.
Geology 14:221-224.
24. Driscoll, C. T., J. P. Baker, J. J. Bisogni, and
C. L. Schofield. 1980. Effect of Aluminum
Speciation on Fish in Dilute Acidified Waters.
Nature 284:161-164.
25. Drouse, S. K., D. C. Hillman, L. W. Creelman,
and S. J. Simon. 1986. National Surface
Water Survey, Eastern Lake Survey - Phase
I, Quality Assurance Plan. EPA/600/4-86/
008. Environmental Monitoring Systems
Laboratory. U.S. Environmental Protection
Agency, Las Vegas, Nevada. 211 pp.
26. Duncan, L. C. and W. Ausserer. 1984. Acid
Precipitation in the Washington Cascades,
Phase II. Final Report Contract A-119-
WASH, Washington Water Research Center.
Pullman, Washington.
27. Duncan, L. C. 1985. The Status (with Respect
to Acidification or Potential Acidification) of
Twenty-six Lakes Lying in the Cascades
South of Chinook Pass. Final report to the
Washington State Department of Ecology.
28. Eilers, J. M., D. J. Blick, Jr., and M. S.
DeHaan. 1987a. Validation of the Western
Lake Survey - Phase I Data Base. Corvallis
Environmental Research Laboratory, U.S.
Environmental Protection Agency, Corvallis,
Oregon.
29. Eilers, J. M., P. Kanciruk, R. A. McCord, W.
S. Overton, L. Hook, D. J. Blick, D. F. Brakke,
P. E. Kellar, M. S. DeHaan, M. E. Silverstein,
and D. H. Landers. 1987b. Characteristics
of Lakes in the Western United States.
Volume II: Data Compendium for Selected
Physical and Chemical Variables. EPA/600/
3-86/054b, U.S. Environmental Protection
Agency, Washington, D.C.
30. Environmental Data Service. 1983. Clima-
tography of United States #81 (by State),
Monthly Normals of Temperature, Precipita-
tion, and Heating and Cooling Degree Days,
1951-1980. National Oceanic and Atmos-
pheric Administration, U.S. Department of
Commerce.
31. Fountain, J., D. T. Hoff, and C. C. MacLeod.
1986. Aquarius Programmers and Users
Guide. Volumes I and II. Environmental
Monitoring Systems Laboratory, U.S. Envir-
onmental Protection Agency, Las Vegas,
Nevada. 323 pp.
32. Fox, C. A., G. R. Bradford, A. L. Page, and
A. A. Elseewi. 1982. Evaluation of Acidity
and Chemical Composition of High Elevation
Lakes in California. Presented at Acid Rain:
a North American Challenge - Soil Conser-
vation Society of America Symposium.
October 25-28, 1985, Burlington, Vermont.
33. Franklin, J. F. and C. T. Dyrness. 1973.
Natural Vegetation of Oregon and Washing-
ton. USDA Forest Service Technical Report
PNW-8. 417pp.
34. Galbraith, A. F. 1984. The Acid Deposition
Potential for the Streams and Lakes of the
Wind River Mountains, pp. 275-272. In:
Workshop Proceedings - Air Quality and Acid
Deposition in the Bridger and Fitzpatrick
Wildernesses. USDA Forest Service, Inter-
mountain Region, Ogden, Utah. 292 pp.
35. Galbraith, A. F. 1 985. Acid Deposition in the
Wind River Mountains. In: Proceedings of
Watershed Management in the Eighties -
New Directions, Amer. Soc. Civil Engrs. April
30-May 1, 1 985, Denver, Colorado. 18 pp.
36. Galloway, J. N., G. E. Likens, and E. S.
Edgerton. 1976. Acid Precipitation in the
Northeastern United States: pH and Acidity.
Science 194:722-724.
37. Galloway, J. N. and E. B. Cowling. 1978. The
Effects of Precipitation on Aquatic and
Terrestrial Ecosystems-a Proposed Precipi-
tation Chemistry Network. J. Air Poll. Cont.
Assoc. 28:229-235.
151
-------
38. Galloway, J. N. and P. J. Dillon. 1983. Effects
of Acidic Deposition: the Role of Nitrogen,
pp. 145-160. In: Ecological Effects of Acidic
Deposition. Report PM 1636. National
Swedish Environment Protection Board.
39. Garrets, R. M. and F. T. Mackenzie. 1967.
Origin of the Chemical Compositions of
Some Springs and Lakes. In: Equilibrium
Concepts in Natural Water Systems, R. M.
Garrels, ed. Amer. Chem. Soc. Adv. Chem.
Ser. 67:222-242.
40. Gibson, J. H., J. N. Galloway, C. Schofield,
W. McFee, R. Johnson, S. McCarley, N. Dise,
and D. Herzog. 1983. Rocky Mountain
Acidification Study. U.S. Department of
Interior, Fish and Wildlife Service, FWS/
OBS-80/40.17, Air Pollution and Acid Rain
Report No. 17. 137 pp.
41. Gilliom, R. J. and G. C. Bortelson. 1983.
Relationships between Water Quality and
Phosphorus Concentrations for Lakes of the
Puget Sound Region, Washington. U.S.
Geological Survey Open File Report 83-255.
26pp.
42. Gjessing, E. T. 1976. Physical and Chemical
Characteristics of Aquatic Humus. Ann Arbor
Science Publishers, Ann Arbor, Michigan.
120 pp.
43. Gorham, E., S. J. Eisenreich, J. Ford, and
M. V. Santelmann. 1985. The Chemistry of
Bog Waters, pp. 339-363. In: Chemical
Processes in Lakes, W. Stumm., ed., John
Wiley and Sons, New York, New York.
44. Henriksen, A. 1980. Acidification of Fresh-
waters—a Large-Scale Titration, pp. 68-74.
In: Ecological Impacts of Acid Precipitation,
D. Drablos and A. Tollan, eds. SNSF Project,
Oslo, Norway.
45. Henriksen, A. and D. F. Brakke. 1986.
Estimates of Critical Loads for Sulfur to
Surface Waters. Environ. Sci. Technol. (in
review).
46. Hidy, G. M. and J. R. Young. 1986. Acid
Deposition and the West, a Scientific
Assessment. ERT Document No. P-D572-
503, ERT Inc., Newbury Park, California. 187
pp.
47. Hillman, D. C., J. F. Potter, and S. J. Simon.
1986. National Surface Water Survey,
Eastern Lake Survey - Phase I, Analytical
Methods Manual. Environmental Monitoring
Systems Laboratory, U.S. Environmental
Protection Agency, Las Vegas, Nevada.
48. Hoffman, G. R. and R. Alexander. 1976.
Forest Vegetation of the Bighorn Mountains,
Wyoming: a Habitat Classification. USDA
Forest Service Research Paper RM-170. 38
pp.
49. Hunt, C. B. 1974. Natural Regions of the
United States and Canada. W.H. Freeman,
San Francisco, California. 725 pp.
50. Kanciruk, P., J. M. Eilers, R. A. McCord, D.
H. Landers, D. F. Brakke, and R. A. Linthurst.
1986a. Characteristics of Lakes in the
Eastern United States. Volume III: Data
Compendium of Site Characteristics and
Chemical Variables. EPA/600/4-86/007c,
U.S. Environmental Protection Agency,
Washington, D.C. 374 pp.
51. Kanciruk, P., M. J. Gentry, R. A. McCord, L.
A. Hook, J. M. Eilers, and M. D. Best. 1986b.
National Surface Water Survey, Eastern Lake
Survey - Phase I, Data Base Dictionary. Oak
Ridge National Laboratory, TM-10153, Oak
Ridge National Laboratory, Oak Ridge,
Tennessee. 85 pp.
52. Kanciruk, P., R. J. Olson, and R. A. McCord.
1986c. Data Quality Assurance in a
Research Data Base: The 1984 National
Surface Water Survey, pp. 193-207. In:
Research Data Management in the Ecolog-
ical Sciences, Belle W. Baruch Library of
Marine Science No. 16, W. K. Michener, ed.
University of South Carolina Press, Colum-
bia, South Carolina.
53. Kerfoot, H. B. and M. L. Faber. 1987. National
Surface Water Survey, Western Lake
Survey - Phase I, Analytical Methods Man-
ual. Environmental Monitoring Systems
Laboratory, U.S. Environmental Protection
Agency, Las Vegas, Nevada.
54. Kirchmer, C. J. 1983. Quality Control in
Water Analyses. Env. Sci. Tech. 17:174A-
181A.
55. Kling, G. W. and M. C. Grant. 1984. Acid
Precipitation in the Colorado Front Range:
An Overview with Time Predictions for
Significant Effects. Arctic and Alpine
Research 16:321-329.
56. Kohler, M. A., T. J. Nordenson, and D. R.
Baker. 1955. Evaporation Maps for the
152
-------
United States. U.S. Weather Bureau
Research Paper 38. 21 pp.
57. Krug, E. C. and C. R. Frink. 1983. Acid Rain
on Acid Soil: A New Perspective. Science
221:520-525.
58. Laird, L. B., H. E. Taylor, and V. C. Kennedy.
1986. Snow Chemistry of the Cascade-
Sierra Nevada Mountains. Env. Sci. Technol.
20:275-290.
59. Larson, E. S. and R. G. Schmidt. 1958. A
Reconnaissance of the Idaho Batholith and
Comparison with the Southern California
Batholith. U.S. Geological Survey Bulletin
1070-A. 33pp.
60. Lewis, W. M., Jr. 1982. Changes in pH and
Buffering Capacity of Lakes in the Colorado
Rockies. Limnol. Oceanogr. 27:167-172.
61. Lewis, W. M., Jr. and M. C. Grant. 1979.
Acid Precipitation in the Western United
States. Science 207:176-177.
62. Lewis, W. M., M. C. Grant, J. F. Saunders.
1984. Chemical Patterns of Bulk Atmos-
pheric Deposition in the State of Colorado.
Water Resources Research 20:1691-1704.
63. Likens, G. E. 1 976. Acid Precipitation. Chem.
Eng. News 54:29-44.
64. 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.
65. Loranger, T. J. 1986. Temporal Variability of
Water and Snowpack Chemistry in the North
Cascade Mountains. M. S. Thesis, Western
Washington University, Bellingham,
Washington. 140 pp.
66. Loranger, T. J. and D. F. Brakke. 1986.
Temporal Variability of Lake Waters in the
North Cascades Mountains (Washington,
USA). Water, Air, and Soil Poll. 30:123-129.
67. McColl, J. G. 1981. Increasing Hydrogen Ion
Activity of Water in Two Reservoirs Supp-
lying the San Francisco Bay Area, California.
Water Resources Research 17:1510-1516.
68. McFee, W. W. 1980. Sensitivity of Soil
Regions to Acid Precipitation. EPA/600/3-
80/013. Corvallis Environmental Research
Laboratory, U.S. Environmental Protection
Agency, Corvallis, Oregon. 178 pp.
69. Meglen, R. R., W. R. Chappell, R. B. McNelly,
R. J. Sistko, G. A. Swanson, L. A. Taylor,
and R. W. Klusman. 1985. Chemical Clas-
sification of Colorado Lakes, Part I. Center
for Environmental Sciences. University of
Colorado at Denver, Colorado. 224 pp.
70. Melack, J. M., J. L. Stoddard, and D. R.
Dawson. 1982. Acid Precipitation and Buffer
Capacity of Lakes in the Sierra Nevada,
California, pp 35-41. In: International
Symposium on Hydrometerology, American
Water Resources Association. June 13-17,
1982, Denver, Colorado.
71. Melack, J. M., J. L. Stoddard, D. R. Dawson,
and C. A. Ochs. 1 983. Assessment of Acidity
of Lakes and Precipitation in the Sierra
Nevada. Technical Completion Report, Pro-
ject UCAL- WRC-W-590. California Water
Resources Center, University of California,
Davis, California. 36 pp.
72. Melack, J. M., J. L. Stoddard, and C. A. Ochs.
1985. Major Ion Chemistry and Sensitivity
to Acid Precipitation of Sierra Nevada Lakes.
Water Resources Research 21:27-32.
73. Metcalf, R. C. 1984. Field pH Determination
in Glacial Melt Waters. J. Glaciology 30:106-
111.
74. Morris, F. A., D. V. Peck, D. C. Hillman, K.
J. Cabbie, and S. L. Pierett. 1986. National
Surface Water Survey, Western Lake Survey
- Phase I, Field Training and Operations
Manual. Internal Report. Environmental
Monitoring Systems Laboratory, U.S. Envir-
onmental Protection Agency, Las Vegas,
Nevada. 177 pp.
75. Munger, J. W. and F. G. Eisenreich. 1983.
Continental Scale Variations in Precipitation
Chemistry. Env. Sci. Tech. 17:32A-42A.
76. National Acid Deposition Program (NADP).
1984. Unpublished annual summaries.
Natural Resource Ecology Lab, Ft. Collins,
Colorado.
77. Nelson, P. O. 1985. Acid Rain Buffering
Potential in Oregon Cascade Lakes: Secon-
dary Mineral Solubility Control of Solution
Ionic Composition. Water Resources
Research Institute, Oregon State University,
Corvallis. Oregon. 16 pp.
153
-------
78. Nelson, P. 0. and G. K. Delwiche. 1983.
Sensitivity of Oregon's Cascade Lakes to
Acid Precipitation. Project Completion Report
A-061-ORE. Water Resources Research
Institute, Oregon State University, Corvallis,
Oregon. 53 pp.
79. Norton, S. A., J. J. Akielaszek, T. A. Haines,
K. L. Stromborg, and J. R. Langcone. 1982.
Bedrock Geologic Control of Sensitivity of
Aquatic Ecosystems in the United States to
Acidic Deposition. 11 pp. Explanatory Text
for National Atmospheric Deposition Pro-
gram. Distribution of Surface Waters Sen-
sitive to Acidic Precipitation: A State-Level
Atlas. University of Maine, Geological
Sciences, Orono, Maine. 66 pp.
80. Norton, S. A. and A. Henriksen. 1983. The
importance of C02 in Evaluation of Effects
of Acidic Deposition. Vatten 39:346-354.
81. Ogden, J. G., III. 1982. Seasonal Mass
Balance of Major Ions in Three Small
Watersheds in a Maritime Environment.
Water, Air, and Soil Poll. 17:119-1 30.
82. Oliver, B. G., E. M. Thurman, and R. L.
Malcolm. 1983. The Contribution of Humic
Substances to the Acidity of Colored Natural
Waters. Geochim. Cosmochim. Acta
47:2031-2035.
83. Omernik, J. M. and G. E. Griffith. 1986. Total
Alkalinity of Surface Waters: a Map of the
Western Region. EPA/600/D-85/219. Cor-
vallis Environmental Research Laboratory,
U.S. Environmental Protection Agency,
Corvallis, Oregon. 52 pp.
84. Omernik, J. M. and C. F. Powers. 1983. Total
Alkalinity of Surface Waters—a National
Map. Annals of the Association of American
Geographers 73:133-1 36.
85. Oppenheimer, M., C. B. Epstein, and R. E.
Yuhnke. 1985. Acid Deposition, Smelter
Emissions, and the Linearity Issue in the
Western United States. Science 229:859-
861.
86. Overton, W. S. and D. J. Blick. 1986. Effects
of Measurement and Other Extraneous
Errors on Estimated Cumulative Distribu-
tions in the National Lake Survey. Presented
at the Western Regional Meeting of the
Institute of Mathematical Statistics, July
1986. (Unpublished).
87. Overton, W. S., P. Kanciruk, L. A. Hook, J.
M. Eilers, D. H. Landers, D. F. Brakke, D. J.
Blick, Jr., R. A. Linthurst, and M. D. DeHaan.
1986. Characteristics of Lakes in the Eastern
United States. Volume II: Lakes Sampled and
Descriptive Statistics for Physical and
Chemical Variables. EPA/600/4-86/007b,
U.S. Environmental Protection Agency,
Washington, D.C. 369 pp.
88. Peck, D. V., R. F. Cusimano, and W. L. Kinney.
1985. National Surface Water Survey,
Western Lake Survey - Phase I, Ground
Sampling Training and Operations Manual.
Internal draft. Environmental Monitoring
Systems Laboratory. U.S. Environmental
Protection Agency, Las Vegas, Nevada. 37
pp.
89. Pennak, R. W. 1963. Rocky Mountain States,
pp. 349-369. In: Limnology in North America,
D. G. Frey, ed. University of Wisconsin Press,
Madison, Wisconsin.
90. Permutt, T. and M. Moezzi. 1986. Relative
Interlaboratory Bias in the Western Lake
Survey. Systems Applications, Inc. San
Rafael, California. 43 pp.
91. Rasmussen, L. A. and W. V. Tangborn. 1976.
Hydrology of the North Cascades Region,
Washington. I. Runoff, Precipitation and
Storage Characteristics. Water Resources
Research 12:187-202.
92. Reddy, M. M. and H. C. Claassen. 1985.
Estimates of Average Major Ion Concentra-
tions in Bulk Precipitation at Two High-
Altitude Sites Near the Continental Divide
in Southwestern Colorado. Atmospheric
Environment 19:1199-1203.
93. Reynolds, R. C. and N. M. Johnson. 1972.
Chemical Weathering in the Temperate
Glacial Environment of the Northern Cas-
cade Mountains. Geochim. Cosmochim.
Acta. 36:537-554.
94. Robertson, C. E. and J. D. Hem. 1969.
Solubility of Aluminum in the Presence of
Hydroxide, Fluoride and Sulfate. U.S. Geo-
logical Survey Water-Supply Paper 1827-C.
37pp.
95. Rodhe, W. 1949. The Ionic Composition of
Lake Waters. Verh. Int. Verein. Limnol.
13:121-141.
96. Rosen, A. E. and P. Kanciruk. 1985. A
Generic Data Entry Quality Assurance Tool.
154
-------
Proceedings of the Tenth Annual Interna-
tional SAS (Statistical Analysis System, Inc.)
Users Group Conference, March 10-13,
1985. Reno, Nevada.
97. Roth, P., C. Blanchard, J. Harte, H. Michaels,
and M. T. EI-Ashry. 1985. The American
West's Acid Rain Test. Research Report #1,
World Resources Institute, Washington, D.C.
50pp.
98. SAS Institute. 1982. The Statistical Analysis
System, SAS Institute, Gary, North Carolina.
99. SAS Institute. 1985. SAS/GRAPH User's
Guide, Version 5 Edition, Appendix 2, SAS
Institute, Gary, North Carolina. 596 pp.
100. Shilts, W. W. 1981. Sensitivity of Bedrock
to Acid Precipitation; Modification by Glacial
Processes. Geol. Surv. Can. Paper 81-14.
101. Silverstein, M. E., S. K. Drouse, J. L. Engels,
M. L. Faber, and T. E. Mitchell-Hall. 1987a.
National Surface Water Survey, Western
Lake Survey - Phase I, Quality Assurance
Plan. EPA/600/8-87/026. Environmental
Monitoring Systems Laboratory, U.S. Envir-
onmental Protection Agency, Las Vegas,
Nevada.
102. Silverstein, M. E., M. L. Faber, S. K. Drouse,
and T. E. Mitchell-Hall. 1987b. National
Surface Water Survey, Western Lake
Survey - Phase I, Quality Assurance Report.
Environmental Monitoring Systems Labora-
tory, U.S. Environmental Protection Agency,
Las Vegas, Nevada.
103. Smith, R. A. and R. B. Alexander. 1983.
Evidence for Acid Precipitation Induced
Trends in Stream Chemistry at Hydrologic
Bench-Mark Stations. U.S. Geological Sur-
vey Circular 910. 12 pp.
104. Snead, R. E. 1980. World Atlas of Geomor-
phic Features. Robert E. Krieger, Inc.,
Huntington, New York, New York. 301 pp.
105. Steele, R., S. Cooper, D. Ondov, D. Roberts,
and R. Pfister. 1983. Forest Habitat Types
of Eastern Idaho and Western Wyoming.
USDA Forest Service General Technical
Report. INT-144. Intermountain Forest and
Range Experiment Station, Ogden, Utah. 1 22
PP-
106. Stensland, G. J. 1984. Chapter 8: Deposition
Monitoring, pp. 8-32. In: The Acidic Depo-
sition Phenomenon and Its Effects. Critical
Assessment Review Papers, Volume I:
Atmospheric Sciences, A. P. Altshuller and
R. A. Linthurst eds., EPA/600/8-83/016 AF,
U.S. Environmental Protection Agency,
Washington, D.C.
107. Stensland, G. J., D. M. Whelpdale, and G.
Oehlert. 1986. Chapter 5. Precipitation
Chemistry, pp. 128-199. In: Acid Deposition,
Long Term Trends. National Academy Press,
Washington, DC. 506 pp.
108. Stuart, S. A. 1984. Hydrology and Aquatic
Chemistry of Monitor Lake Watersheds in the
Wind River Mountains, Wyoming, pp. 130-
173. In: Workshop Proceedings - Air Quality
and Acid Deposition in the Bridger and
Fitzpatrick Wildernesses. USDA Forest
Service, Intermountain Region, Ogden, Utah.
292 pp.
109. Stumm, W. and J. J. Morgan. 1 981. Aquatic
Chemistry: An Introduction Emphasizing
Chemical Equilibria in Natural Waters. 2nd
Edition. John Wiley & Sons, New York, New
York. 779 pp.
110. Sumioka, S. S. and N. P. Dion. 1 985. Trophic
Classification of Washington Lakes Using
Reconnaissance Data. Washington State
Department of Ecology, Water Supply Bul-
letin 57. 320pp.
111. Swedish Ministry of Agriculture. 1982.
Acidification Today and Tomorrow. Ministry
of Agriculture, Environment Committee,
Stockholm, Sweden. 232 pp.
112. Tonnessen, K. A. 1983. Potential for Aquatic
Ecosystem Acidification in the Sierra Nev-
ada, 147-169 pp. In: Acid Precipitation:
Aquatic Effects, G. Hendrey, ed., Ann Arbor
Science, Woodburn, Masachusetts. 173 pp.
113. Tukey, J. W. 1977. Exploratory Data Anal-
ysis. Addison-Wesley, Reading, Massachu-
setts. 688 pp.
114. Turk, J. T. 1983. An Evaluation of Trends
in the Acidity of Precipitation and the Related
Acidification of Surface Waters in North
America. U.S. Geological Survey Water-
Supply Paper 2249. 18 pp.
115. Turk, J. T. and D. B. Adams. 1983. Sensitivity
to Acidification of Lakes in the Flat Tops
Wilderness Area, Colorado. Water Resour-
ces Research 19:346-350.
755
-------
116. U.S. Department of Agriculture-Soil Conser-
vation Service. 1976. Scale 1:1,000,000.
Climax Vegetation. Montana.
117. U.S. Geological Survey. 1970. Potential
Natural Vegetation. In: National Atlas of the
United States of America. Washington, D.C.
417pp.
118. U.S. Department of Commerce. 1974. Cli-
matic Atlas of the United States. 80 pp.
119. U.S. Environmental Protection Agency.
1983. Methods for Chemical Analysis of
Water and Wastes. EPA/600/4-79/020,
Environmental Monitoring and Support
Laboratory, Cincinnati, Ohio.
120. Van Alstine, R. E. and E. W. looker. 1979.
Preliminary Map of Fluorine Provinces in the
Conterminous United States. In: Atlas of
Metal and Nonmetal Mineral Provinces of the
Conterminous United States. United States
Geological Survey Open-File Report 79-
576G.
121. Velleman, P. F. and D. C. Hoaglin. 1981.
Applications, Basics and Computing of
Exploratory Data Analysis. Duxbury Press,
Boston, Massachusetts.
122. Welch, E. B., D. E. Spyridakis, and T. Smayda.
1986. Acidification Potential in High Moun-
tain Lakes in Washington: 3. Final Report,
Washington State Department of Ecology. 24
PP.
123. Wetzel, R. G. 1983. Limnology. 2nd Edition.
Saunders College Publishing, Philadelphia,
Pennsylvania. 767 pp.
124. Wilderness Act of 1964. Public Law 88-577.
125. Wilson, D. and R. Wood. 1984. Mountain
Lake Water Quality Survey. 1983. California
Department of Fish and Game, Region 1,
Data Report. 7 pp.
126. Wissmar, R. G., J. E. Richey, A. H. Devol,
and D. M. Eggers. 1982. Lake Ecosystems
of the Lake Washington Drainage Basin, pp.
333-385. In: Analysis of Coniferous Forest
Ecosystems in the Western United States,
R. L. Edmonds, ed. Hutchison Ross Publish-
ing Company, Stroudsburg, Pennsylvania.
419pp.
156
-------
Section 11
Glossary
11.1
A
AAS
Ac
ANC
APHA
ASTM
Abbreviations
—lake surface area contained in a
population of target lakes
—estimated total lake surface area
in a target population
—atomic absorption spectroscopy
—estimated lake surface area
having a value for a variable
below Xc
—95 percent upper confidence limit
on Ac
—acid neutralizing capacity
—American Public Health
Association
—American Society for Testing and
Materials
CaCOa —calcium carbonate or calcite
CaMg (C03)2 —dolomite
CaSC>4 —anhydrite or anhydrous calcium
sulfate
CaSO4-2H2O —gypsum
cm —centimeter
DDRP —Direct/Delayed Response Project
DGC —duplicate (ground crew)
DHC —duplicate (helicopter crew)
DIG —dissolved inorganic carbon
DOC —dissolved organic carbon
DQO —data quality objective
ELS-I —Eastern Lake Survey - Phase I
EMSI —Environmental Monitoring and
Services, Inc.
EMSL-
Las Vegas —Environmental Monitoring
Systems Laboratory - Las Vegas
EPA —Environmental Protection Agency
ERL-Corvallis —Environmental Research
Laboratory - Corvallis
ERP —Episodic Response Project
ExtAI —extractable aluminum
FL —field low synthetic audit
FN —field natural audit
ft —feet
F(x) —cumulative frequency distribution
gc —reference proportion (lake area)
G(x) —cumulative areal distribution
ha —hectare
IBM —International Business Machines
ICPAES —inductively coupled plasma
atomic emission spectroscopy
ID —identification number
IR —infrared
km2 —square kilometers
L —liter
LA —lake area
Lockheed- —Lockheed Engineering and
EMSCO Management Services Company,
Inc.
m —meter
M —median value of an F(x) or G(x)
distribution
MAX —maximum value of variable X
mg L"1 —milligrams per liter
mi2 —square miles
MIBK-HQ —methyl isobutyl ketone-
hydroxyquinoline
MIN
mL
mm
N
N
-minimum value of variable X
-miHiliter
-millimeter
-normal (normality), as in
equivalents per liter
-number of lakes in a target
population
757
-------
N
N*
Nc
Nc
Ncu
Nn
n*
n*
n'
nc
nna
n0
nz
NADP/NTN
NAPAP
NBS
NLS
NSWS
NTU
ORNL
P
PC
PCU
ppm
PRECIP
—estimated number of lakes in a
target population
—number of lakes in map
population
—number of lakes below Xc
—estimated number of lakes below
Xc
—95 percent upper confidence limit
on No
—estimated nontarget population
size
—number of lakes selected in
probability sample
—number of lakes visited
—number of target lakes sampled
—effective sample size
—number of lakes in the probability
sample below Xc
—number of lakes in the probability
sample determined after visiting
the lake
—number of non-target lakes in the
probability sample determined
from maps before visitation
—number of lakes not visited
—number of sample lakes in
subpopulation z
—National Atmospheric Deposition
Program/National Trends
Network
—National Acid Precipitation
Assessment Program
—National Bureau of Standards
—National Lake Survey
—National Surface Water Survey
—nephelometric turbidity unit
—Oak Ridge National Laboratory
—inclusion probability
—reference proportion (number of
lakes)
—platinum cobalt unit
—parts per million
—variable name for precipitation
—proportion of lakes visited
—first quintile of F(x) or G(x); 20th
percentile
Q3
CU
QA
QC
QCCS
Qd
RDL
RGC
RHC
RMS
RO
%RSD
RT
SAS
SB
SO
SDL
SE
STP
SIR
THC
UCL
fjtm
fjS cm"
USGS
UV
V
W
WA
WLS-I
WMP
x
—second quintile of F(x) or G(x);
40th percentile
—third quintile of F(x) or G(x); 60th
percentile
—fourth quintile of F(x) or G(x); 80th
percentile
—fifth quintile of F(x) or G(x);
maximum
—quality assurance
—quality control
—quality control check sample
—interquintile difference
—required detection limit
—routine (ground crew)
—routine (helicopter crew)
—root mean square
—runoff
—percent relative standard
deviation
—residence time
—Statistical Analysis System
—standard deviation of field blank
measurements
—standard deviation
—system decision limit
—standard error
—standard temperature and
pressure
—stratum
—triplicate (helicopter crew)
—95 percent upper confidence limit
—microequivalents per liter
—micrograms per liter
—micrometers
—microsiemens per centimeter
—United States Geological Survey
—ultraviolet
—estimated variance
—weight or expansion factor
—watershed area
—Western Lake Survey - Phase I
—Watershed Manipulation Project
—measured value of variable
158
-------
X
x
Xc
YSI
z
Zs
—variable
—mean value for a variable x
—reference value of the variable X
—Yellow Springs Instruments
—subpopulation of lakes
—site depth
11.2 Definitions
ABSOLUTE BIAS—the expected difference between
a measured value and the true value.
ACID NEUTRALIZING CAPACITY—the total acid-
combining capacity of a water sample determined
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 sulfuric acid,
nitric acid or hydrochloric acid, usually produced
by atmospheric transformation of the by-products
of fossil fuel combustion (power plants, smelters,
autos, etc.); precipitation with a pH less than 5.0
is generally considered to be unnaturally acidic.
ACIDIC LAKE—for this report, a lake with ACID
NEUTRALIZING CAPACITY less than or equal to
0/ueq 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 C02> before analysis; used with some pH and
DISSOLVED INORGANIC CARBON measure-
ments.
ALGORITHM—a step-by-step mathematical proce-
dure for solving a problem.
ALIQUOT—portion of a sample prepared for the
analysis of particular constituents, sent in a
separate container to the ANALYTICAL LABOR-
ATORY.
ALKALINITY—the titratable base of a sample
containing hydroxide, carbonate, and bicarbonate
ions, i.e., the equivalents of acid required to
neutralize the basic carbonate components.
ALKALINITY MAP CLASS—a geographic area
defined by the expected ALKALINITY of a majority
of SURFACE WATERS (does not necessarily reflect
measured alkalinity); used as a STRATIFICATION
FACTOR in the lake selection design. Three
alkalinity map classes were used: < 100, 100-199,
and 200-400 ,ueq L"1.
ALLUVIAL—deposited by rivers and streams.
AMONG-BATCH PRECISION—the estimate of PRE-
CISION that includes effects of different labora-
tories and day-to-day difference within a single
laboratory.
ANALYTICAL LABORATORY—in this report, a
laboratory under contract with the U.S. Environ-
mental Protection Agency to analyze water
samples shipped from the FIELD LABORATORIES.
ANALYTE—a chemical species that is measured in
a water sample.
ANDESITE—dark gray volcanic rock composed
largely of FELDSPAR, often having large feldspar
crystals in a finer matrix.
ANION—a negatively charged ion.
ANION-CATION BALANCE—in an electrically neu-
tral solution, such as water, the total charge of
positive ions (CATIONS) equals the total charge
of negative ions (ANIONS).
ANION DEFICIT—the concentration in /ueq L"1 of
measured ANIONS less than measured CATIONS.
ANTICLINE—an arching fold in layered rocks.
ANTHROPOGENIC—from a human source.
ASTM TYPE I REAGENT GRADE WATER—deionized
water used for BLANK samples and REAGENT
preparation that had a measured CONDUCTANCE
less than 1 pS cirT1.
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 perfor-
mance in sample analysis. Natural audit samples
were lake water; synthetic audit samples were
prepared by diluting concentrates of known
chemical composition. See FIELD AUDIT SAMPLE
and LABORATORY AUDIT SAMPLE.
BASALT—dark gray to black volcanic rock which is
poor in silica and rich in iron and magnesium
minerals.
755
-------
BASE CATION—a non-protolytic cation that does not
affect ACID NEUTRALIZING CAPACITY; usually
calcium or magnesium. See PROTOLYTE.
BATCH—a group of samples processed and analyzed
together. A field batch of samples is defined as
all samples (including QUALITY ASSURANCE
samples) processed at one FIELD LABORATORY
in one day. A laboratory batch is defined as all
samples processed and analyzed at one ANALYT-
ICAL LABORATORY, associated with one set of
laboratory QUALITY CONTROL samples.
BATHOLITH—a large mass of INTRUSIVE rock,
usually granite, more than 100 square kilometers
(40 square miles) in surface exposure.
BEDROCK—solid rock exposed at or near the surface.
BIAS—the systematic difference between the
measured and the true values. See ABSOLUTE
BIAS and RELATIVE BIAS.
BLANK—a sample of ASTM TYPE I REAGENT GRADE
WATER analyzed as a QUALITY ASSURANCE/
QUALITY CONTROL sample in the WLS-I. See
FIELD BLANK SAMPLE, LABORATORY BLANK
SAMPLE, and TRAILER BLANK.
BUFFERING CAPACITY—the quantity of acid or base
that can be added to a water sample with little
change in pH.
CALCIUM BICARBONATE SYSTEM—see CARBO-
NATE SYSTEM.
CALCULATED CONDUCTANCE—the sum of the
theoretical specific conductances of all measured
ions in a sample.
CALCAREOUS—made of or containing calcium
carbonate. See CALCITE.
CALCITE—calcium carbonate (CaCOs); a common
rock-forming mineral, the principal mineral in
limestone, marble, and chalk.
CALCIUM—a silver-white metallic element of the
alkaline-earth group occurring only in compound
forms.
CALC-SILICATE GNEISS—a METAMORPHIC rock
consisting of mainly calcium-bearing silicate
minerals.
CALDERA—a broad, basin-shaped volcanic depres-
sion formed by the explosion or collapse of a
magma chamber.
CALIBRATION BLANK—a solution used in calibrat-
ing or standardizing analytical instruments.
CALIBRATION STUDY—a study conducted during
the WLS-I to determine if the method of sample
collection (helicopter crew vs. ground crew)
affected the chemistry of the water samples;
samples collected during this study were also used
to evaluate analytical laboratory bias.
CARBONATE SYSTEM—a lake in which the major
part of ACID NEUTRALIZING CAPACITY is com-
posed of carbonate; organic or other weak ahions
contribute less than 10 percent to the total ANION
charge.
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).
CIRQUE—a steep-walled, usually semicircular, basin
excavated by the head of a glacier.
CLEARWATER LAKE—for this report, .a lake having
TRUE COLOR less than or equal to 30 PLATINUM
COBALT UNITS.
CLOSED LAKE—a lake with a surface water inlet
but no surface water outlet.
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
allows the similarity within and among data sets
to be established confidently.
COMPLETENESS— a measure of data quality that
is the quantity of acceptable data actually collected
relative to the total quantity that was attempted
to be collected.
CONDUCTANCE—a measure of the electrical
conductance (the reciprocal of the electrical
resistance) or total IONIC STRENGTH of a water
sample.
760
-------
CONGLOMERATES—rocks composed of rounded,
waterworn fragments of older rocks, cemented
together by another mineral substance.
CRETACEOUS—the third and latest of the periods
included in the MESOZOIC Era; also the system
of strata deposited in the Cretaceous Period.
CUMULATIVE FREQUENCY DISTRIBUTION—a func-
tion, F(x), such that for any reference value X, F(x)
is the estimated proportion of lakes in the
population having a value x < X.
CUMULATIVE AREAL DISTRIBUTION—a function,
G(x), such that for any REFERENCE VALUE X, G(x)
is the estimated proportion of lake area in the
population having a value x < X.
DARKWATER LAKE—for this report, a lake with
TRUE COLOR greater than 30 PLATINUM COBALT
UNITS.
DATA BASE—all computerized results of the survey,
which include DATA SETS 1, 2, 3, and 4, as well
as back-up and historical data sets.
DATA FILE—a subset of information, often in tabular
form (e.g., the lake location file contains informa-
tion on lake identification and location).
DATA PACKAGE—a report, generated by an ANA-
LYTICAL LABORATORY for each BATCH of
samples analyzed, that included analytical results,
ACID NEUTRALIZING CAPACITY titration data, and
QUALITY CONTROL CHECK SAMPLE results.
DATA QUALIFIER—see FLAG and TAG.
DATA QUALITY OBJECTIVES—ACCURACY, DETEC-
TABILITY, and PRECISION limits desired by data
users, established prior to the beginning of a
sampling effort.
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 INSTRU-
MENT DETECTION LIMIT and SYSTEM DECISION
LIMIT.
DIKE—a tabular body of IGNEOUS rock that cuts
across the structure of adjacent rocks or cuts
massive rocks. Most dikes result from the intrusion
of magma; some are the result of mineral
replacement.
DILUTE LAKE—for this report, a lake with a
CONDUCTANCE less than 10 /uS cm"1.
DISSOLVED INORGANIC CARBON—a measure of
the dissolved carbon dioxide, carbonic acid,
bicarbonate and carbonate anions which comprise
the major part of ACID NEUTRALIZING CAPACITY
in a lake.
DISSOLVED ORGANIC CARBON—the organic
fraction of carbon in a water sample that is
dissolved or unfilterable (for this report, 0.45 ^m
pore size).
DOLOMITE—a common rock-forming mineral, CaMg
(C03>2, often with some iron replacing magnesium;
those rocks that approximate the mineral dolomite
in composition.
DOLOSTONE—SEDIMENTARY rock composed of
fragmental, aggregated, or precipitated DOLOMITE
of organic or inorganic origin.
DRAINAGE BASIN—see WATERSHED.
DRAINAGE LAKE—a lake with surface water
outlet(s) or with both inlets and outlets.
DUNITE—an ULTRAMAFIC rock consisting primarily
of olivine, (Mg, Fe>2 Si04, a common rock-forming
mineral of basic and low-silica IGNEOUS rocks.
DUPLICATE LAKE SAMPLE—see FIELD DUPLICATE
SAMPLE.
EFFECTIVE SAMPLE SIZE (n')—within each STRA-
TUM, the total sample size modified because of
incomplete visitation.
ELECTRONEUTRALITY—having no electric charge.
EMPIRICAL DISTRIBUTION—observed distribution
of results from the Survey.
EMPIRICAL RELATIONSHIP—observed relationship
between variables based on survey results.
EQUIVALENT—unit of ionic charge; the quantity of
a substance that either gains or loses one mole
of protons or electrons.
EVAPOTRANSPIRATION—loss of water from the soil
by evaporation and by transpiration from the plants
growing thereon.
161
-------
EXPANSION FACTOR—see WEIGHT.
EXTRACTABLE ALUMINUM—operationally defined
aluminum fraction that is extracted by the
procedure used in the WLS-I; this measurement
is intended to provide an indication of the
concentration of aluminum that may be available
in a form toxic to fish.
EXTRUSIVE ROCK—rock formed of magma which
reaches the surface and solidifies (also called
volcanic rock).
FALL CIRCULATION—see TURNOVER.
FELDSPAR—a group of common, light-colored, rock-
forming minerals containing aluminum oxides and
silica. Feldspars constitute 60 percent of the
earth's crust.
FELSIC—a mnemonic term derived from (fe) for
feldspar, (I) for lenads or feldspathoids, and (s) for
silica and applied to light-colored rocks containing
an abundance of one or all of these constituents.
FELDSPAR and quartz are examples of felsic
minerals.
FIELD AUDIT SAMPLE—a standardized water
sample submitted to FIELD LABORATORIES to
check overall performance in sample analysis by
both FIELD and ANALYTICAL LABORATORIES.
Natural field audit samples were lake water;
synthetic field audit samples were prepared by
diluting concentrates of known chemical compo-
sition.
FIELD BLANK SAMPLE—a sample of ASTM TYPE
I REAGENT GRADE WATER prepared at the lake
site by field crews; these samples were analyzed
by both FIELD and ANALYTICAL LABORATORIES.
FLUORITE—a clear to translucent mineral, CaF2; the
ore of fluorine.
FLUORAPATITE—a mineral of the apatite group,
Cas(P04)sF; common accessory mineral in IGNE-
OUS rocks.
FLUVIAL—of, relating to, or living in a stream or river;
produced by stream action.
FLUVIAL DEPOSITS—deposits produced by the
action of a stream or river.
FRAME—a structural representation of a population
providing a sampling capability.
GEOMORPHIC—of or relating to the form of the solid
surface features of the earth.
GEOMORPHIC UNIT—an area with similar physio-
graphic or geologic characteristics. See PHYSIO-
GRAPHY.
GIBBSITE—a mineral, Al (OH)3; a principal constit-
uent of many bauxites (aluminum ores).
GLACIAL CIRQUE—see CIRQUE.
GLACIAL TILL—glacially deposited sediment, which
consists of debris of all sizes mixed indiscrimi-
nately together.
GLASSY VOLCANIC ROCK (VOLCANIC GLASS)—
natural glass produced by the cooling of molten
lava, or some liquid fraction of molten lava, too
rapidly to permit crystallization, and forming
materials such as obsidian and pitchstone.
GNEISS—banded METAMORPHIC rock thought to
form from granite (which it commonly resembles),
sandstone, and other continental rocks.
FIELD DUPLICATE SAMPLE—a second sample of
lake water collected by the same crew at the same
lake site immediately after the ROUTINE SAMPLE
in accordance with standardized protocols.
FIELD LABORATORY—mobile laboratory located at
each FIELD STATION in which sample processing
and measurement of selected variables were
performed.
FIELD STATION—a location providing support for
helicopters, sampling personnel, and FIELD
LABORATORIES during field sampling operations.
FINAL DATA SET—see DATA SET 4.
FLAG—qualifier of a data value assigned during the
VERIFICATION and VALIDATION procedures.
GRAN ANALYSIS—a mathematical procedure to
determine the equivalence points of a titration
curve for ACID NEUTRALIZING CAPACITY.
GRANODIORITE—coarse-grained INTRUSIVE rock
with less quartz and more FELDSPAR than granite.
GROUNDWATER—water in the part of the ground
that is completely saturated.
GYPSUM—a common mineral (CaSOv2H20) con-
sisting of hydrous calcium sulfate. Alabaster is an
example.
HALITE—a mineral form of NaCI; rock salt.
HECTARE—a measure of surface area equal to ten
thousand square meters or 2.47 acres.
762
-------
HETEROSCEDASTIC—inequality of variances
among groups.
HIGH DOC LAKE—for this report, a lake with a
DISSOLVED ORGANIC CARBON (DOC) concentra-
tion > 6 mg L~1.
HIGH ELEVATION LAKE—lakes with surface eleva-
tion > 3000 m above mean sea level.
HOMOSCEDASTIC—equality in variances among
groups.
HYDRAULIC RESIDENCE TIME—the time required
to exchange the total volume of water in a lake.
HYDROLOGIC FLOW PATHS—the distribution and
circulation of water on the surface of the land,
in the soil and underlying rocks within a
watershed.
IGNEOUS—rock formed from molten magma.
IN SITU—referring to measurements collected within
the water column at a lake.
INCLUSION PROBABILITY—the chance of a lake
being selected for sampling.
INDEX SAMPLE—in this report, one sample per lake,
used to represent chemical conditions in that lake.
INITIAL DIC—a measurement of DISSOLVED INOR-
GANIC CARBON made on an ALIQUOT imme-
diately before it is titrated for ACID NEUTRALIZING
CAPACITY.
INSTRUMENTAL DETECTION LIMIT—for each chem-
ical variable, a value calculated from LABORA-
TORY BLANK SAMPLES that indicates the min-
imum concentration reliably detectable by the
instrument(s) used.
INTERLABORATORY BIAS—systematic differences
in performance between laboratories estimated
from analysis of the same type of AUDIT SAMPLES.
INTERMEDIATE VOLCANIC ROCKS—IGNEOUS rock
containing between 52 and 66 percent silica (SiOa).
INTERQUARTILE RANGE—the difference between
the upper and lower QUARTILE values in a
population frequency distribution. See QUARTILE.
INTERQUINTILE DIFFERENCE—the difference
between the fourth QUINTILE (80th percentile) and
first QUINTILE (20th percentile) values in a
population frequency distribution.
INTRALABORATORY PRECISION GOAL—a preci-
sion goal based on the DATA QUALITY OBJEC-
TIVES for the analysis of laboratory duplicates
within a single laboratory.
INTRUSIVE—describing magma (naturally occurring
molten rock material, generated within the earth)
that has been injected into a pre-existing rock.
INVERSE CUMULATIVE AREAL DISTRIBUTION—a
function, 1-G(x), such that for any reference value
X, 1-G(x) is the estimated proportion of lake area
in the population having a value x > X.
INVERSE CUMULATIVE FREQUENCY DISTRIBU-
TION—a function, 1-F(x), such that for any
reference value X, 1-F(x) is the estimated number
of lakes 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. In very dilute
solutions, ions behave independently of each other
and the ionic strength can be calculated from the
measured concentrations of ANIONS and
CATIONS present in the solution.
ISOTHERMAL—defined as a temperature difference
in a lake of less than 4°C between the reading
at 1.5 meters below the surface and at 1.5 meters
above lake bottom (synonymous with NONSTRAT-
IFIED or mixed).
JURASSIC—the middle of three periods comprising
the MESOZOIC Era. The system of strata deposited
during that period.
KAOLINITE—a clay mineral commonly used in
making porcelain and fire bricks.
LABORATORY AUDIT SAMPLE—During the ELS-I,
a standardized water sample submitted blind to
ANALYTICAL LABORATORIES to check overall
performance in sample analysis. All laboratory
audit samples were synthetic, prepared by diluting
concentrates of known chemical composition. This
sample was not used during the WLS-I.
LABORATORY BLANK SAMPLE—a sample of ASTM
TYPE I REAGENT GRADE WATER prepared and
analyzed by ANALYTICAL LABORATORIES.
LABORATORY DUPLICATE SAMPLE—a split sample
prepared at the ANALYTICAL LABORATORIES.
LAKE ID—a unique, seven-character identification
(ID) code given to each lake in the Survey
designating the REGION, SUBREGION, and
ALKALINITY MAP CLASS to which the lake
163
-------
belongs; e.g., lake ID4A2-011 designates the 11th
lake in alkalinity map class 2 of Subregion A in
Region 4.
LAKE TYPE—a classification of lakes based on the
presence or absence of inlets, outlets, and dams
as represented on LARGE-SCALE MAPS.
LARGE-SCALE MAP— 1:24,000-, or 1:62,500-scale
U.S. Geological Survey topographic maps.
LIMESTONE—a SEDIMENTARY rock consisting
largely of calcium carbonate.
LONG-TERM ACIDIFICATION—a partial or complete
permanent loss of ACID NEUTRALIZING CAPACITY
from a lake.
LONG-TERM MONITORING PROGRAM—an ongo-
ing program sponsored by the U.S. Environmental
Protection Agency that monitors the chemistry of
selected lakes and streams at least three times
per year, which is designed to detect long-term
trends in chemistry.
LORAN-C GUIDANCE SYSTEM—a navigational
system based upon the time-delay of received radio
signals. The system provides latitude and longitude
at any point on the earth's surface.
LOW ANC LAKE—for this report, a lake with an ACID
NEUTRALIZING CAPACITY (ANC) less than or
equal to 50 jueq L~1.
LOW CALCIUM LAKE—for this report, a lake with
a CALCIUM concentration less than or equal to
50 peq L'\
LOW CONDUCTANCE—for this report, a lake with
a CONDUCTANCE less than 10 /uS cm"1 (synonym-
ous with DILUTE LAKE).
LOW DOC LAKE—for this report, a lake with a
DISSOLVED ORGANIC CARBON (DOC) concentra-
tion less than or equal to 2 mg L"1.
LOW pH LAKE—for this report, a lake with a pH
less than or equal to 5.0.
MAFIC—pertaining to or composed dominantly of the
magnesium rock-forming silicates; said of some
IGNEOUS rocks and their constituent minerals.
MAJOR ANION—for this report, an ANION with a
concentration > 5 /ueq L~1.
MAJOR CATION—for this report, a CATION with a
concentration > 5 jueq L"1.
MAP POPULATION—within each STRATUM, the
164
ordered list of lakes depicted on 1:100,000-scale
U.S. Geological Survey TOPOGRAPHIC MAPS.
MATRIX—the physical and chemical makeup of a
sample being analyzed.
MATRIX SPIKE—a QUALITY CONTROL sample
analyzed at an ANALYTICAL LABORATORY that
was prepared by adding a known concentration
of ANALYTE to a sample. Matrix spike samples
were used to determine possible chemical inter-
ferences within a sample that might affect the
analytical result.
MEDIAN (M)—the value of x such that F(x) or G(x)
= 0.5; the 50th percentile.
MESOZOIC—relating to an era of geologic time
including the interval between the PERMIAN and
the TERTIARY; relating to the system of rocks
formed in this era.
METAMORPHIC—rock derived from pre-existing
rocks as they are altered by heat, pressure, and
other processes.
METASEDIMENTARY—SEDIMENTARY rock altered
by heat, pressure, and other processes but still
retaining some sedimentary characteristics.
METAVOLCANIC—volcanic rock altered by heat,
pressure, and other processes but still retaining
some volcanic characteristics.
MINERAL WEATHERING—dissolution of rocks and
minerals by erosive forces.
MIXING—see TURNOVER.
MORPHOMETRY—lake shape, size, depth; physical
description of lake basin.
NBS-TRACEABLE—describing a REAGENT or stand-
ard, the accuracy of which has been checked
against a certified National Bureau of Standards
standard.
NONSTRATIFIED—see ISOTHERMAL
NON-TARGET LAKE—a lake that either was not the
focus of the WLS-I objectives or could not be
sampled within the constraints of the survey.
NOT VISITED LAKE—a PROBABILITY SAMPLE lake
scheduled for sampling that was not visited or
sampled during field activities; thus, its target or
non-target status is unknown. See TARGET
POPULATION and NON-TARGET LAKE.
NOMOGRAPH—a graph reducing a mathematical
formula to curves so that the resultant value can
-------
be read on the chart coordinates for any value
assigned to the variable involved.
ON-SITE INSPECTION—inspection of the operation
of FIELD and ANALYTICAL LABORATORIES.
OPEN SYSTEM—a measurement of pH or DIS-
SOLVED INORGANIC CARBON obtained from a
sample that was exposed to the atmosphere during
collection and/or measurement.
ORGANIC ACIDS—acids possessing a carboxyl
(-COOH) group; includes fulvic and humic acids.
ORGANIC ANION—a weakly dissociated organic
molecule with a negative net ionic charge.
OROGRAPHIC—associated with or induced by the
presence of mountains.
OUTLIERS—observations not typical of the popula-
tion from which the sample is drawn.
OVERALL AMONG-BATCH PRECISION—estimate of
PRECISION calculated from FIELD AUDIT SAM-
PLES that include the effects of different ANA-
LYTICAL LABORATORIES and day-to-day differen-
ces within a single laboratory.
OVERALL WITHIN-BATCH PRECISION—an estimate
of PRECISION calculated from FIELD DUPLICATE
SAMPLES, LABORATORY DUPLICATE SAMPLES,
or TRAILER DUPLICATES and reported as ROOT
MEAN SQUARE. See WITHIN BATCH PRECISION).
PALEOZOIC—one of the eras of geologic time
comprising the Cambrian, Ordovician, Silurian,
Devonian, Carboniferous (Mississippian and
PENNSYLVANIAN), and PERMIAN systems; relat-
ing to the system of rocks formed in this era.
PENNSYLVANIAN-AGE—in the United States, the
sixth period of the PALEOZOIC, approximately
equivalent to the Upper Carboniferous period.
PERCENT ION BALANCE DIFFERENCE—a QUALITY
ASSURANCE procedure used to check that the
sum of the anion EQUIVALENTS equals the sum
of the cation EQUIVALENTS. See ANION-CATION
BALANCE.
PERCENT RELATIVE STANDARD DEVIATION (%
RSD)—the STANDARD DEVIATION divided by the
mean, multiplied by 100.
PERMIAN—the last period of the PALEOZOIC Era.
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.
PHYSIOGRAPHY—the elevation, SLOPE, and shape
of the landform, including its spatial relationships
with other physical features.
PLAGIOCLASE—common rock-forming minerals in
the FELDSPAR group, containing CALCIUM and
sodium.
PLATINUM COBALT UNIT—measure of the color of
a water sample defined by a potassium hexach-
loroplatinate and cobalt chloride standard color
series.
PLEISTOCENE—the earlier of the two epochs in the
QUATERNARY period, relating to the series of
sediments deposited during this epoch.
PLUTON—a mass of INTRUSIVE IGNEOUS rock,
most commonly granite.
PLUTONIC—referring to rocks, usually IGNEOUS,
formed at a great depth.
POPULATION ESTIMATE—a statistical estimate that
applies to a specific population of TARGET LAKES,
not only to the sample.
POTENTIOMETER—an instrument for measuring
electromotive forces. A source of adjustable
voltage.
PRECAMBRIAN—relating to the earliest era of
geologic history or the corresponding system of
rocks, relating to the system of rocks formed before
the Cambrian Era.
PRECISION—a measure of the capacity of a method
to provide reproducible measurements of a
particular ANALYTE.
PRE-JURASSIC—that period before the JURASSIC
period which was the second period of the
MESOZOIC Era, approximately 135-190 million
years before present.
PRIMARY VARIABLES—variables of foremost con-
cern in the Survey (pH, ACID NEUTRALIZING
CAPACITY, EXTRACTABLE ALUMINUM, sulfate,
CALCIUM, DISSOLVED ORGANIC CARBON).
PROBABILITY SAMPLE—a sample in which each
unit (lake, in this case) has a known probability
of being selected.
PROTOLYTE—that portion of a molecule that reacts
with either H+ or OH~ in solution.
165
-------
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 SAMPLE—a sample of
known concentration used to verify continued
calibration of an instrument.
QUANTITATION LIMIT—for each chemical variable,
a value calculated from BLANK samples that
represents the lowest concentration that can be
measured with reasonable PRECISION.
QUARTILE—any of the three values that divide the
population of a frequency distribution into four
equal classes, each representing 25 percent; used
to measure range or variation.
QUATERNARY—the second period of the Cenozoic
Era representing the period from two to three
million years before present to the present.
QUATERNARY PERIOD—relating to the geologic
period from the end of the TERTIARY to the present
time or the corresponding system of rocks.
QUARTZITE—sandstone consisting chiefly of quartz
grains welded so firmly that, when broken, the
rock breaks through, rather than around, the
grains.
QUINTILE—any of the four values (Qi, Q2, Qs, Q4)
that divide the population of a frequency distri-
bution into five equal classes, each representing
20 percent; used to provide additional values to
compare among populations of lakes (see CUMUL-
ATIVE FREQUENCY DISTRIBUTION).
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.
REAGENT—a substance added to a water sample
(because of its chemical reactivity) to determine
the concentration of a specific ANALYTE.
REAGENT BLANK—a LABORATORY BLANK SAM-
PLE that contained all the reagents required to
prepare a sample for analysis of silica and total
aluminum.
REFERENCE VALUE (Xc)—a concentration of interest
for a given chemical variable.
REGION—for this report, a major area of the
contermmous United States where a substantial
number of lakes with ALKALINITY less than 400
//eq L"1 can be found.
RELATIVE BIAS—the expected difference between
a measured value and the true value, expressed
as a percentage of the true value.
REMOTE BASE SITE—location serving as a base of
operations for sampling crews working more than
150 miles from the FIELD LABORATORY, requiring
that samples be flown to the FIELD LABORATORY
daily.
REPRESENTATIVENESS—a measure of data quality;
the degree to which sample data accurately and
precisely reflect the characteristics of a population.
REQUIRED DETECTION LIMIT—for each chemical
variable, the highest INSTRUMENT DETECTION
LIMIT allowable in the ANALYTICAL LABORA-
TORY contract.
RHYOLITE—light gray volcanic rock with large quartz
and FELDSPAR crystals in a finer groundmass; the
fine-grained extrusive equivalent of granite. See
EXTRUSIVE ROCK.
ROOT MEAN SQUARE—a summary statistic of the
relative or absolute STANDARD DEVIATION; a
pooled STANDARD DEVIATION of the % RELATIVE
STANDARD DEVIATION (%RSD), calculated by the
formula: [x2%RSD + SD2%RSDln-1/n']1/2
ROUTINE SAMPLE—the first lake sample collected
at a site in accordance with standardized protocols.
SANDSTONE—a cemented or otherwise compacted
detrital sediment composed predominantly of
quartz grains, the grades of the latter being those
of sand.
SAS—Statistical Analysis System, Inc. (Gary, NC).
A statistical DATA FILE manipulation package that
has data management, statistical, and graphical
analysis abilities. The WLS-I DATA BASE was
developed and analyzed primarily using SAS
software, and is distributed in SAS format.
SCHIST—a crystalline rock that can be easily split
into layers; examples are mica and hornblende,
which may contain high potassium and f lourine.
SECCHI DISK—a 20 cm-diameter, black-and-white
disk used to measure water TRANSPARENCY.
SECCHI DISK TRANSPARENCY—a measure of lake
water clarity determined by lowering a SECCHI
DISK into the lake, recording the depth at which
it disappears from view, then raising it until it
166
-------
comes into view, and recording the depth; the
mean value of these depths is then calculated.
SECONDARY VARIABLES—chemical variables
measured during the survey considered to be
important in providing additional data in quanti-
fying the chemical status of lakes, e.g., sodium,
magnesium, potassium, nitrate, chloride and total
aluminum.
SEDIMENTARY—referring to rocks composed of
particles of other rock transported and deposited
by water, wind, or ice.
SEEPAGE LAKE—a lake with no permanent surface
water inlets or outlets.
SHALE—fine-grained mudstone or claystone with
laminated or layered structure, with a division
plane that separates each successive layer.
SHALLOW LAKE—a lake from which a clean sample
could not be obtained at 1.5 meters below the
surface but could be obtained at 0.75 meters; these
lakes were generally less than 3 meters deep.
SLOPE—vertical distance divided by horizontal
distance.
SOFTWATER—water with low concentrations of
CALCIUM and magnesium salts.
SPARGING—a sample preparation process that
involved bubbling a gas into an ALIQUOT.
SPECIAL INTEREST LAKE—in this survey, a lake
selected non-randomly which is not part of the
PROBABILITY SAMPLE; selection was based on
quality and amount of data available or involve-
ment 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.
STANDARD DEVIATION—the square root of the
variance of a given statistic, calculated by the
equation [I x2 - (I x)2]1/2/(n - 1 )1/2.
STANDARD ERROR—STANDARD DEVIATION
divided by the square root of the sample size.
STRATIFICATION FACTORS—factors used to define
STRATA prior to lake selection; the factors used
in the WLS-I were REGION, SUBREGION, and
ALKALINITY MAP CLASS. See STRATUM.
STRATIFIED DESIGN—a statistical design in which
the population is divided into strata, and a sample
selected from each STRATUM.
STRATIFIED LAKE—in this report, a lake with a
temperature difference greater than 4°C between
the water layers at 1.5 meters below the surface
and 1.5 meters above the lake bottom. If the
temperature difference is also greater than 4°C
between the water layers at 1.5 meters below the
surface and 60 percent of site depth, then the lake
is strongly stratified, if not, it is weakly stratified.
STRATUM—in this Survey, a SUBPOPULATION of
lakes within an ALKALINITY MAP CLASS within
a SUBREGION and within a REGION, as defined
by the STRATIFIED DESIGN.
SUBPOPULATION—any defined subset of the
TARGET POPULATION.
SUBREGIONS—areas within each REGION that are
similar in water quality, PHYSIOGRAPHY, vege-
tation, climate, and soil; used as a STRATIFICA-
TION FACTOR in the WLS-I design.
SURFACE WATER—streams and lakes.
SURFICAL GEOLOGY—characteristics of the earth's
surface; especially consisting of unconsolidated
residual, ALLUVIAL or glacial deposits lying on the
BEDROCK.
SYNOPTIC—relating to or displaying conditions as
they exist simultaneously over a broad area.
SYSTEM DECISION LIMIT—for each chemical
variable, except pH, a value calculated from FIELD
BLANK data that indicates a concentration (with
95% confidence) that can be distinguished reliably
from background levels.
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. A first sampling unit is randomly drawn
from the first k units, and every kth unit afterward
is included in the sample (k being equal to N divided
by the sample size).
TAG—code on a data point that is added at the time
of collection or analysis to qualify the data.
TARGET POPULATION—in this survey, the lake
population of interest that was sampled. This
population was defined by the sampling protocol.
TERNARY DIAGRAMS—a diagram showing three
components; for this report, these diagrams were
used to examine the ratios of major CATIONS and
ANIONS (synonymous with trilinear plots or
trilinear diagrams).
167
-------
TERTIARY PERIOD—relating to the first period of the
Cenozoic Era or the corresponding system of rocks
marked by the formation of high mountains.
TITRATION DATA—individual data points from the
modified GRAN ANALYSIS of ACID NEUTRALIZ-
ING CAPACITY.
TOPOGRAPHIC MAP—a map showing contours of
surface elevation.
TRANSPARENCY—the clarity of unfiltered water.
TRAILER BLANK—an ASTM TYPE I REAGENT
GRADE WATER sample prepared and processed
at the FIELD LABORATORY, but analyzed at an
ANALYTICAL LABORATORY.
TRAILER DUPLICATE—a SPLIT SAMPLE prepared in
the FIELD LABORATORY.
TRIPLICATE LAKE SAMPLE—the third sample of lake
water collected by the same helicopter crew at
the same lake immediately after the ROUTINE and
DUPLICATE SAMPLES in accordance with stan-
darized protocols; this third sample was used only
as part of the CALIBRATION STUDY.
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 blue to blackish-brown.
TUNDRA—a level or undulating treeless plain
characteristic of arctic and subarctic regions,
characterized by a topsoil layer high in organic
content above a permanently frozen soil layer.
TURBIDITY—a measure of light scattering by
suspended particles in an unfiltered water sample.
TURNOVER—a period of water circulation in lakes,
occurring when little or no differences in temper-
ature are observed with depth.
UPPER CONFIDENCE LIMIT (95%)—a value that, in
association with a statistic, has a 95 percent
chance of being above the true value of the
population parameter of interest.
ULTRAMAFIC—INTRUSIVE rocks consisting of dark
coarse grained minerals with relatively low silica
content.
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 ASSUR-
ANCE plan.
WATERSHED—the geographic area from which
surface water drains into a particular lake.
WATERSHED DISTURBANCE—for this report, a
disturbance of the natural environment in a
watershed within 100 meters of the shore as noted
by field samplers. Disturbances included roads,
houses, logging, mining and livestock.
WEAK ACID SYSTEM—a lake in which more than
10 percent of the ANION charge results from
ORGANIC ACIDS or other weak acid ANIONS.
WEATHERING—see MINERAL WEATHERING.
WEIGHT—the inverse of a sample lake's INCLUSION
PROBABILITY; each sample lake represents this
number of lakes in the population.
WITHIN-BATCH PRECISION—the estimate of PRE-
CISION expected in the analysis of samples in a
BATCH by the same laboratory on any single day
(in this report, overall within-batch precision
includes the effects of sample collection, process-
ing and analysis; analytical within-batch precision
includes the effects of sample analysis within
ANALYTICAL LABORATORIES).
765
-------
Appendix
Contributed by W.S. Overton, Department of
Statistics, Oregon State University, Corvallis,
Oregon.
Calibration to Correct for the Effect of
Different Protocols
Use of different protocols involves the risk of bias,
with the result that data generated under one
protocol may not be comparable to data generated
under another. For this reason, the field protocols
of the National Lake Survey were rigidly standard-
ized, with holding times and handling procedures
narrowly prescribed. The need to vary this protocol
in the wilderness areas of the Western Lake Survey -
Phase I (WLS-I), by ground crew collection, rather
than the helicopter collection that was the Survey
standard, caused some concern about the effect of
the associated changes in protocol, and particularly
in handling and holding time. Differences in the
protocols for ground crew and helicopter are
discussed in Section 2.5.1.
A strategy was devised between Survey personnel
and the involved Forest Service personnel, to assess
the differences in the two protocols and to minimize
the effect of the change in protocol. This strategy
required a random subset of the wilderness sample
lakes to be "sampled" by both ground crew and by
helicopter, allowing study of the effect of the change
in mode of collection. The strategy followed was to
identify (a) those variables for which the values
obtained by the alternate method (ground sample)
could be used interchangeably with values obtained
by the standard helicopter method, (b) those variables
for which the standard value could be predicted
accurately from the alternate value, and (c) those
variables for which it was judged that the alternate
method gave values that were too poor to use in
the Survey.
Of the 50 lakes selected for this calibration study,
45 were actually sampled by both ground crew and
helicopter. Duplicate samples were taken for these
lakes in both the standard and the alternate visits.
Visits were scheduled for the same day, but in order
to ensure comparability of the data in the calibration
study, ground crews were not apprised of which
lakes were also to be sampled by helicopter. Security
was inadvertently breached in one subregion. The
sample for this subregion was redrawn, and, in order
not to indicate the newly selected lakes, duplicates
were not reassigned for the ground crews. For this
reason, a few sample lakes do not have duplicate
ground sample values. Additionally, some data
values were removed from the analysis as a result
of data validation.
The duplicate samples provided valuable evidence
of the repeatability of the measurements, expressed
as measurement error. One sample of each duplicate
pair was sent to each of the two analytical
laboratories contracted for the WLS-I. Although this
plan caused some difficulty in the interpretation of
measurement error, it created a welcome opportu-
nity to assess laboratory bias, as reflected in patterns
of differences in the results from the two labora-
tories. Additionally, analysis of audit samples
(Section 4.3) and field laboratory splits provide
information regarding laboratory bias.
The objective of the calibration study was to identify
those variables for which the two protocols generate
interchangeable results, those for which the
helicopter values are different, but predictable from
the ground crew values, and those for which such
prediction involves unacceptable error, so that the
ground crew data must be considered unusable in
the population descriptions.
Regression analysis is the basic method for
assessment of a linear relationship between the two
values of a variable, representing the same reality,
but determined by different protocols. There are
many alternate ways these analyses can be made;
discussion of the rationale for the choices made in
this study are provided by Overton (1986).
As the ground value is to be used to predict a
helicopter value, the latter is analysed as the
dependent variable (y) and the ground value as the
independent variable (x). Each lake in the calibration
sample is represented for each variable by two y's
and most lakes by two x's. Individual y values were
169
-------
regressed on the mean of the x values of that lake.
All regressions were forced through the origin, with
the structural model given by:
Y = BX + e
resulting in a simple calibration equation:
y = bx
When replacement is implemented, the value of b
is set to 1, and for calibration by prediction, b is
the least squares estimate of B.
Three versions of this model were used, one
following each of the following specifications of
variance of e:
Model 1. variance proportional to x's
Model 2. variance inversely proportional to
x's,
Model 3. variance homogeneous.
Analysis began with plots of the data, y against x,
in the form of box-duplicate plots (Figures A-1 and
A-2). Lakes with duplicates in each variable are
represented by a box, one side determined by the
two duplicate values of y and the other by the two
duplicate values of x. Missing values cause missing
sides of some boxes. The identity transfer line
indicates the locus on which the two protocols give
identical values.
Inspection of these plots gives a strong visual sense
of the patterns of the relations. If the boxes are
arrayed along the transfer line, the relation is close.
One can readily see if the array tends not to pass
through the origin. If the boxes tend to change size
as a function of x, then one should consider either
Model 1 or 2; if the boxes are homogeneous (in size)
with x, then Model 3 is a candidate. The plots also
provide evidence for erroneous data, and were used
in flagging values for validation. The data ultimately
used in analysis excluded those values that were
confirmed to be erroneous.
From inspection of the graphs, each variable was
assigned to one or more of the above models, and
analyzed for each by least squares with transfor-
mation appropriate to the model. The model giving
the best fit was then used for that variable
throughout the assessment, with Model 3 prevailing
unless one of the others was clearly better. Each
variable was also analyzed with a y-intercept, and
the same error structure; this allowed assessment
of the consequences of the choice of the model
through the origin. Most variables were adequately
analyzed by Model 3, with most of the rest
conforming to Model 1. One pH variable (closed
system pH) required Model 2, and the complement
of this variable (pOH) was analysed by Model 1 to
illustrate that the assessment was identical. In
general, assessment was robust to model identifi-
cation, and variables that fit one model equally well
as another got essentially the same assessment from
both models.
In establishing criteria for calibration, it is necessary
to consider the consequences of failure of the
alternate protocol value to be identical to the value
from the standard protocol. First, the alternate
method may be biased, so that adjustment of the
x values by prediction is required to remove the bias.
This adjustment contributes an additional variance
component to the error of the calibrated value, this
having the effect of additional measurement error.
Even if there is no bias between protocols, it is
possible for the deviation about regression to be
substantially greater than the inherent measure-
ment error, and this also must be considered
measurement error of the alternate protocol as
representing the standard. Bias of measurement is
easily detected and removed by the methods used
here, but the residual increase in error remains after
correction for bias and is still cause for concern.
This concern is because of the bias in the estimated
distribution, F(x), due to measurement error and to
other sources of variation in the values. The
magnitude of these biases was investigated by
Overton and Blick (1986). They showed that
maximum bias is closely related to the relative
variance:
d2 = 1 + a zm/al.
where a m and a* are measurement variance and
population variance, respectively.
Table A-1 shows the bias associated with F(x) as
a function of d2, as determined by Overton and Blick
(1986). They showed that these values closely
approximate maximal bias in cases like Model 1, as
well as Model 3, and for some distributions other
than normal.
There is no general standard by which one can select
from this table a criterion by which to assess the
calibration; choice is necessarily subjective, and
related to the intended uses of the analyses. A
Table A-1. Approximate Bias of F (x) as a Function of
Relative Variance, d2. This is Calculated for the
Simple Case of a Normal Population with Normal
Measurement Error, and Assessed at ± 1 Stand-
ard Deviation. (From Overton and Blick 1986)
d2 1.001 1.01 1.05 1.10 1.20 1.50 2.00
Bias 0.0001 0.0012 0.0059 0.0115 0.0220 0.0485 0.0811
170
-------
600
Ca
300 .
300
8.5
pOH
7.0 _
5.5
50
5.5
7.0
Extractable Al
20 -
-10
650
600
8.5
325 -
8.5
7.0
5.5 ,
5.5
ANC
pH
T
325
650
I
7.0
8.5
-10
20
50
Figure A-1. Duplicate box plots of calibration study lakes showing helicopter samples (Y axis) compared to ground samples
(X axis). Shown are calcium, acid neutralizing capacity, negative logarithm of the hydroxyl ion. pH, and extractable
aluminum.
171
-------
740 ,
70-
SO,
I
70
6.50
140
3.25 -
O.OO
DOC
0.00
3.25
6.50
DIC
85.0
42.5 .
0.0
Conductance
0.0
42.5
850
14.O
70
-0.5
-05
NO
7.0
14.0
6.0
30 _
-0.5
NOl
-0.5
3.0
6.0
Figure A-2. Duplicate box plots of calibration study lakes showing helicopter samples (Y axis) compared to ground samples
(X axis). Shown are sulfate, dissolved organic carbon, dissolved inorganic carbon (closed system), conductance,
and nitrate (regular and expanded scale).
172
-------
criterion of d2 = 1.1 is seen associated with maximal
bias of 0.0115 (about 10%), and d2 = 1.01 with bias
of 0.0012 (about 1 % of the value of F(x) at the point
of maximal bias). Relative to the intended uses of
these data, these levels of bias seem reasonable to
serve as maximum acceptable bias, and design
standard bias, respectively.
In the calibration context, prediction error and error
of deviation from regression simply contribute to
"measurement error," so that assessment of these
errors is appropriately applied to the calibrated
ground values, represented by y. That is, d2 can be
considered a general form of relative error:
d2 = total variance/population variance
where total variance is the sum of population
variance, measurement error variance, and predic-
tion variance. It is possible, then, to utilize several
different values of d2 in the process of choosing the
best treatment of the ground crew data. Specifically,
d2 was estimated for the standard protocol, for the
replacement of y by x, and for the prediction of y
by the regression of y on x.
From the regression analyses, the estimated
variances of prediction were calculated, along with
mean square deviation from regression. Measure-
ment error variances of y and of x were calculated
from the duplicate values of those quantities, in a
manner appropriate to the model prescribed for each
variable. To complete the computations, the appro-
priate variance of x as a replacement for y was
calculated. For Model 3, all of these statistics are
constants over x, but for Models 1 and 2, they vary
with x. The estimated values of d2 are given in Table
A-2; for variables analyzed by Models 1 or 2, these
values are evaluated at the median of x in the
calibration sample.
In Table A-2, most variables can be seen to pose
no problem due to collection by ground crew. The
first 12 meet the suggested design standard; for four
of those, prediction by regression is optimal, but
comparison of the optimal and replacement d2
indicates no appreciable disadvantage from replace-
ment for these four, also.
The second group of nine variables has somewhat
greater d2, in the range greater than considered
desirable, but still acceptable. It is of note that for
six of these, d2 is nearly as great for the standard
survey protocol as it is for the optimal calibration.
That is, although the precision of these variables
is nearing the level of concern, this is not due to
the change of protocol, but to inherent imprecision
in the standard protocol. The exceptions to this are
closed system pH and turbidity; these demonstrate
substantial additional variance due to the calibration
process. Again, for three of the five for which
regression is optimal, there is little penalty for using
replacement; closed system pH is the only variable
of the 29 that possibly justifies calibration by
regression.
This leaves the last group of eight "questionable"
variables, for each of which at least one value of
d2 exceeds the acceptable value, some by a
considerable margin. The first four of these are also
seen not to owe their undesirable properties to the
alternate protocol; in fact, three of them appear
"improved" by the regression, two substantially. This
is evidence that the estimates of measurement error
for these variables are somewhat inflated. The other
four questionable variables were either worsened
by calibration or have no basis for assessment of
measurement error.
From this analysis, the change in collection protocol
is seen to have created no substantial problem of
comparability of the data. Several variables (turbidity,
closed system pH, and extractable Al) show an
appreciable increase in d2 under calibration, and
several (Ca+z, total Al and SiO2) show an inexplicable
reduction in d2. These results can be interpreted in
terms of more familiar variance components, as
shown in Table A-3.
In Table A-3, it is seen that several variables
(turbidity, closed system pH, extractable Al) have very
much greater deviation about regression of y on x
than can be accounted for by measurement error,
and another group (Ca+2, total Al, Si02) have very
much less. These two groups were identified above
from Table A-2; the principal determinant of
differences between d? and dl is simply the ratio
between measurement error variance and the mean
square deviation from regression.
Similarly, the determinant of the general magnitude
of d2 is the ratio of error variance to population
variance. This follows the definition of d2, but its
significance will become more apparent in the
following discussions. Further, this assessment
relative to the population variation is appropriate
when concern is relative to the impact of errors and
bias on population assessment. There may be
circumstances in which another criterion is more
appropriate.
It was earlier indicated that measurement error
estimated from the calibration sample contained a
component of laboratory bias. In the context of
population descriptions, it should contain such a
component, but not necessarily in the mix found in
the analysis of duplicates. This is clearly of no
importance for most variables, because the relative
173
-------
Table A-2. Calibration Equations and Decision Criteria for the Twenty-Nine Variables Analyzed in the Calibration Study. The
Estimated b is Replaced by 1.0 When Replacement is Used; Replacement is Optimal When |t| < 1.
General Equation: y = bx.
d2
Variable3
Within Design Standard
ANC
Ca+2
Mg + 2
DIG, initial
DIG, air-equil.
DIG, closed system
HCO3-
Conductance
F~, total dissolved
S04-2
Na +
K +
t
-0.125
0.575
0.674
0.257
-0.650
0.198
0.833
-0.075
1.277
2.664
-2.682
1.390
b
0.99961
1.00261
1.00293
1.00156
0.99649
1 .00098
1 .00450
0.99967
1.02172
1.01314
0.98849
1.01000
Standard
Protocol
1.00108
1.00156
1 .00063
1.01081
1.00511
1.00131
1.00578
1 .00283
1 .00329
1 .00290
1 .00465
1 .00806
Optimal
Calibration
1.00123
1.00111
1 .00070
1.00818
1.00574
1. 00523°
1 .00500
1 .00447
1.00375
1 .00390
1 .00628
1.01172
Replacement1"
1.00376
1.00395
1 .00646
1.01175
Within Acceptable Standard
Fe
pH, alkalinity
pH, acidity
Depth
Turbidity
pH, closed system
P, total
DOC
cr
Questionable Variables
Al, total
SiO2
pH (air-equil.)
N03-
Al, extractable
True color
Secchi disk
transparency
Temperature
-0.806
-0.304
0.545
0.539
-0.872
-3.674
-1.037
-2.600
1.536
0.437
0.679
-1.783
2.386
0.628
-0.722
-1.702
-2.649
0.98219
0.99960
1 .00090
1.01214
0.98115
0.99313
0.99354
0.97454
1.02240
1.01089
1.01069
0.99619
1.09554
1.05926
0.96388
0.93799
0.93530
1.01696
1.03850
1.06130
d
1.01871
1.00976
1.03894
1.02431
1.02577
1.20802
1.13895
1.15562
1.13662
1.09462
1.41182
d
d
1.02620
1.05526
1.06833
1.05513
1 .07403°
1.06918C
1.03617
1.02909
1 .04009
1.16929
1.08965
1.14249
1.14743
1. 46587°
1. 70104°
1.23717
1.41990
1.07856
1.03621
1.03010
1 .04024
1.14593
1.14779
1.24613
1.47127
aVariables are defined in Volume II, Tables 2-1 and 5-3; units are those used in data analysis (Volume I).
bGiven only when the optimal calibration is predicted by regression.
°These variables apparently have an additional source of variation associated with the change in protocol.
dNot possible to compute.
magnitude of calculated measurement error is so
small, relative to population variance, that substan-
tial adjustment in the relative contribution of
laboratory error would have no importance. In only
six variables (all in the last group of Table A-2) is
the magnitude of estimated measurement error
sufficiently large that change of the bias term might
be of influence. Of these, only N0a~ has estimated
laboratory bias of sufficient magnitude (10%, Section
4.5.2) to warrant attention, and even this magnitude
is too small to have an effect, for the levels of N0a~
found in the calibration study.
Three groups of variables in Table A-2 can be
identified as a focus of summary considerations.
Closed system pH has estimated measurement error
within design standard, but the change in protocol
creates substantially greater variance in this
variable. Bias is slight, but bias correction by
regression is marginally justified. It is indicated that
for this variable, maximum bias of F(x) would be
changed from approximately 0.007 to 0.006 by use
of calibration (0.99313x) instead of replacement (use
of x). Further, maximum bias would not occur at an
important value of pH; values in the wilderness lakes
are generally well above the level of concern with
respect to lake acidity.
All the rest of the first two groups (totaling 21
variables) are clearly usable without calibration, and
calibration for closed system pH appears unjustified,
in that little reduction in potential bias is achieved.
The last group of variables in Table A-2 requires
more attention. The change in protocol has no
detrimental effect on NOa", total Al, SiO2, or air-
equilibrated pH and for each of these, the ground
174
-------
Table A-3. Estimated Variance Components for the Twenty-Nine Variables in the Calibration Study. These Components Provide
the Basis for the Calibration Methods. All Variance Components are Calculated in Accordance with the Indicated
Model, and Assessed at the Median of x. The Population Variance is Based Only on the Calibration Sample.
Measurement Error
Variable3
ANC
Ca+2
Al, extractable
pOH
F~, total dissolved
Mg+2
Fe
Al, total
P, total
Secchi disk transparency
pH, closed system
so4-2
DOC
pH, acidity
pH, alkalinity
pH, air-equilibrated
DIG, closed system
DIC, initial
DIG, air-equilibrated
HCO3"
cr
N03~
Na +
K +
Si02
Conductance
Depth
True color
Turbidity
Temperature
y
20.363
21.531
6.4988
0.00222
0.04987
0.9889
17.469
63.150
0.3922
b
0.00222
2.0171
0.01797
0.01051
0.00657
0.02348
0.00334
0.02195
0.01052
82.341
0.4323
0.6007
0.9680
0.2127
0.3192
0.5954
b
9.7222
0.00389
b
X
26.563
24.146
8.2466
0.00728
0.01781
0.8111
22.142
47.144
0.3360
b
0.00726
2.1582
0.01216
0.01792
0.01117
0.01221
0.00228
0.03147
0.02330
81.093
0.2516
0.2259
0.3308
0.1265
0.2055
0.6071
b
5.9211
0.00908
b
Deviation
from
regression
23.331
15.339
31.895
0.01519
0.05650
1.1068
26.863
51.296
0.3604
2.3506
0.01548
2.7079
0.02139
0.01169
0.00933
0.02127
0.01337
0.01660
0.01 180
71.041
0.6760
0.6480
1.3018
0.3086
0.2055
0.9392
5.4354
16.477
0.01536
1.7046
Population
variance
18941.0
13794.6
68.682
0.22696
15.152
1581.35
1030.16
303.57
10.071
10.113
0.22696
695.64
0.73918
0.17151
0.17066
0.15089
2.5576
2.0307
2.0566
14236.2
16.907
4.3973
208.05
26.388
2.2971
210.11
98.790
23.608
0.2079
4.1438
Model
1
1
1
1
1
1
1
1
1
1
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3Vanables are defined in Volume II, Tables 2-1 and 5-3, units are those used in data analysis (Volume I); pOH was calculated
from the closed-system pH value.
bMeasurement error not possible to calculate.
crew data should clearly be used as though they
were helicopter data. However, this analysis has
disclosed significant relative imprecision in these
four variables, and each should be examined for
utility.
The levels of NO3~ are low in the calibration sample,
with a median of 0.45, and greater levels would have
increased the estimated population variation and
thereby decreased the value of dz. A moderate Assessment of SiC>2 is less clear. The median of the
80 percent were below 10.4//gL 1, andthethreshold
of biological concern is approximately 20 /JQ L~1; the
median is below the detection limit (Volume I, Table
4-1). Lastly, the data are accurate enough to indicate
that the level of extractable Al in the wilderness lakes
is very low, even if the precision of the measured
values is questionable. A similar argument applies
to total Al.
change would shift NO3 from the third group of
variables to the second group, with acceptable
status. Further, the levels in the sample are low
relative to the detection limit, having the median well
below the assessed value (Volume I, Table 4-1).
Precision of extractable Al is also inherently marginal
in this analysis, and is greatly worsened by the
change in protocol, so that one must question the
use of the ground data, even after calibration.
However, the observed values of extractable Al are
low, relative to levels that would be of concern, and
near the detection limit. In the calibration sample,
calibration is 2.00 mg L well above the detection
limit of 0.18 mg L"1 (Volume I, Table 4-1), but close
to the quantitation limit of 2.07 mg L~1 (Volume I,
Table 4-4). At the concentrations of Si02 present
in the wilderness lakes, the precision estimated from
the calibration study (Table A-3) is substantially less
than the precision anticipated from the estimates
from the field synthetic audit samples (Volume I,
Table 4-2).
The three physical variables in the set of suspicious
variables have somewhat different interpretation.
Secchi disk transparency and temperature values
775
-------
had no duplicates, so one can only guess that the
protocols led to substantial variation in these
variables. True color did have the benefit of
duplicates, and it is seen that protocol added
substantial variation. True color has coarse resolu-
tion, so one might not expect much more, particularly
with the restricted range of values found in this
study. The apparent coarse resolution of temperature
is surprising; the measurement of temperature is
routine, and its unsatisfactory properties may also
be due only to the narrow range found in these data.
These three physical variables have low information
content, and so use of the ground data is judged
acceptable.
Reference
Overton, W. S., D. J. Blick. 1986. Effects of
measurement and other extraneous errors on
estimated cumulative distributions in the National
Lake Survey. Presented at the Western Regional
Meeting of the Institute of Mathematical Statistics,
July, 1986. Seattle, Washington.
176
. S . GOVERNMENT PRINTING OFFICES 1987-748-121/67054
-------
O k-
o o
Q_ ^
o
o
,-jvirontnental Protection Agency
. .,', L" -irary ( M-L-16)
. I'^arborn St eet, Room 1670
Cm0^.30,. It 60604
------- |