v>EPA
^tes Office of Acid Deposition, Environmental
.nental Protection Monitoring and Quality Assurance
/ Washington DC 20460
EPA/600/4-86/026
December 1986
Research and Development
National
Surface Water
Survey:
National Stream Survey
Phase I—Pilot Survey
-------
EPA/600/4-86/026
December 1986
National Surface Water Survey:
National Stream Survey
Phase l-Pilot Survey
A contribution to the
National Acid Precipitation Assessment Program
By:
J. J. Messer, C. W. Ariss, R. Baker, S. K. Drouse, K. N. Eshleman,
P. R. Kaufmann, R. A. Linthurst, J. M. Omernik, W. S. Overton, M. J. Sale,
R. D. Schonbrod, S. M. Stambaugh, and J. R. Tuschall, Jr.
U.S. Environmental Protection Agency
Office of Research and Development, Washington, DC 20460
Environmental Research Laboratory, Corvallis, OR 97333
Environmental Monitoring Systems Laboratory, Las Vegas, NX' 89114
U.S. Environmental Protection Agency
Region 5, Library (5PL-16)
230 £. Dearborn Street, Room 1670
Chicago, JL 60604
-------
Notice
The research described in this document has been funded wholly or in part
by the U.S. Environmental Protection Agency under Contract No. 68-03-3050
to Lockheed Engineering and Management Services Company, Inc., No. 68-
02-3889 to Radian Corporation, and No. 68-03-3246 to Northrop Services,
Inc., and under Interagency Agreement No. 40-1557-85 with the U.S.
Department of Energy (Contract No. DE-AC05-840R21400 with Martin
Marietta Energy Systems, Inc.). It has been subject to the Agency's peer and
administrative review, and it has been approved for publication as an EPA
document. This report is also listed as contribution #2841 for Oak Ridge
National Laboratory (Environmental Sciences Division).
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:
Messer, J.J.1, C.W. Ariss2, J.R. Baker3, S.K. Drouse3, K.N. Eshleman4, P.R.
Kaufmann1, R.A. Linthurst5, J.M. Omernik6, W.S. Overton7, M.J. Sale8, R.D.
Schonbrod9, S.M. Stambaugh4, and J.R. Tuschall, Jr.10, 1986. National
Surface Water Survey: National Stream Survey, Phase l-Pilot Survey. EPA/
600/4-86/026, U.S. Environmental Protection Agency, Washington, DC.
Inquiries regarding the availability of the NSS Phase l-Pilot Survey data base
should be directed, in writing, to:
Chief, Air Branch
U.S. Environmental Protection Agency
Environmental Research Laboratory
200 SW 35th Street
Corvallis, Oregon 97333
'Utah State University, Utah Water Research Laboratory, Logan, Utah 843,22. Present address: U.S.
Environmental Protection Agency, Environmental Research Laboratory, 200 SW 35th Street, Corvallis, Oregon
97333.
2Utah State University, Utah Water Research Laboratory, Logan, Utah 84322.
3Lockheed Engineering and Management Services Company, Inc., Las Vegas, Nevada 89119.
"Northrop Services, Inc., U.S. Environmental Protection Agency, 200 SW 35th Street, Corvallis, Oregon 97333.
5U.S. Environmental Protection Agency, Office of Research and Development, 401 M Street, SW, Washington,
DC 20460. Present address: U.S. EPA, Environmental Monitoring Systems Laboratory, Mail Drop 39, Research
Triangle Park, North Carolina 27711.
6U.S. Environmental Protection Agency, Environmental Research Laboratory, 200 SW 35th Street, Corvallis,
Oregon 97333.
'Oregon State University, Department of Statistics, Kidder Hall No. 8, Corvallis, Oregon 97331.
'Environmental Sciences Division, Oak Ridge National Laboratory, Post Office Box X, Oak Ridge, Tennessee
37831. Operated by Martin Marietta Engergy Systems, Inc., under Contract No. DE-AC05-840R21400 for
the U.S. Department of Energy.
9U.S. Environmental Protection Agency, Environmental Monitoring Systems Laboratory, 944 E. Harmon Avenue,
Las Vegas, Nevada 89114.
'"Northrop Services, Inc., P.O. Box 12313, Research Triangle Park, North Carolina 27709.
-------
Abstract
A pilot survey of streams in the Southern Blue Ridge Province was conducted
by the U.S. Environmental Protection Agency during the spring and summer
of 1985 as a means of testing a proposed methodology for (1) determining
the present extent and location of acidic and low acid neutralizing capacity
(ANC) streams in the United States and (2) classifying sampled streams that
are representative of important classes of streams and, therefore, should be
selected for intensive study or long-term monitoring; Data from the National
Stream Survey Phase l-Pilot Survey are presented in the context of evaluating
a statistical sampling design, logistics plan, quality assurance plan, and data
management program. Results indicate that the design is capable of producing
robust population estimates for important chemical variables using a single
synoptic sampling of streams, and that it has the potential of producing a
relatively simple geochemical classification of streams. The study showed that,
with 95% confidence, less than 3,2% of the pombined length of streams in
the target population exhibited average spring non-episodic pH values below
6.4 (the lowest value for which a confidence level could be used). The best
estimate of the percentage of stream length with ANC less than or equal
to 200 /ueq L"1 was 74.4%.
-------
Executive Summary
Introduction and Objectives
The National Stream Survey (NSS) Phase l-Pilot Survey was conducted in
the spring and summer of 1985 as part of the U.S. Environmental Protection
Agency's National Surface Water Survey (NSWS). The NSWS is an important
contribution to the National Acid Precipitation Assessment Program, which
is charged by the U.S. Congress with providing sound technical information
regarding the effects of acid deposition on the environment. The three primary
objectives of Phase I of the NSWS are:
• To determine the percentage, extent, and location of low pH lakes and
streams in potentially susceptible regions of the United States.
• To determine the percentage, extent, and location of lakes and streams
in such regions that have low acid neutralizing capacity (ANC).
• To determine which lakes and streams are representative of important
classes of water bodies in each region, and thus should be selected for
additional study or long-term monitoring.
The NSS Phase l-Pilot Survey was designed to provide an otherwise unavailable
data base with which to answer certain questions relating to the proper design
and implementation of a full Phase I effort in 1986. The Phase l-Pilot Survey
objectives were:
• To test the ability of a proposed sampling design to meet the Phase I
objectives.
• To evaluate the proposed Phase I logistics plan, together with alternative
sample collection, preparation, and analytical techniques.
• To develop and test a data analysis plan for Phase I results.
The results of the study, conducted in the mountains of the Southern Blue
Ridge Province, were deemed to be adequate for meeting both sets of objectives
for the region.
Sampling and Logistical Design
To accomplish the survey objectives, a probability sample of 54 stream reaches
was drawn from a target population represented by the blue line streams
on 1:250,000-scale topographic maps, draining catchments of less than 60
square miles and satisfying certain site inclusion criteria. The resulting
statistical sample can be used to make quantitative population estimates with
known confidence limits for any characteristic associated with the reaches.
The characteristics measured during the survey include a suite of geographic,
physical, and chemical variables appropriate to the NSWS objectives. All
variables were measured using extensively reviewed techniques and protocols,
and were subjected to a high degree of quality control and assurance, from
sample collection to the final disposition in the data base.
IV
-------
In addition to the 54-stream probability sample, seven "special interest"
reaches also were included in the field sampling. The data from these streams
were not used to generate population estimates, but they allow the estimates
to be compared to historical stream data collected in the region.
Prior to field sampling, site reconnaissance activities were carried out for each
stream in conjunction with local district soil conservationists to identify and
resolve any physical or legal access problems. Water samples were collected
on three occasions at approximately biweekly intervals during the spring (17
March-30 April) and on one occasion in the summer (30 May-17 July), at
the downstream node of each reach. Samples also were collected at the
upstream node at 17 sites during the last spring sampling interval, and at
all 54 probability sample reaches on the summer sampling data. Site access
was by foot, four-wheel drive vehicle, boat, or.horseback, with samples returned
to a mobile laboratory for processing within 12 hours of collection. In all,
724 field and audit samples were analyzed during the survey.
Survey Results
Techniques and Protocols
A number of field evaluations of instruments and protocols were carried out
to test the logistics plan and field sampling protocols for the 1986 field activities.
This experience proved extremely helpful in selecting and/or modifying the
field measurement techniques, all of which were found to be acceptable for
use in the 1986 field work. Among the most important findings in this regard
was that the important chemical constituents in a wide variety of field samples
were found to be stable when held at 4°C for at least 24 hours following
collection, and that the plastic syringes used to hold dissolved inorganic carbon
and pH samples during transport to the field laboratories were impervious
to carbon dioxide when maintained at 4°C. These findings were deemed
sufficient to recommend locating the "mobile" processing laboratories at a
central location for Phase I field work, thus allowing many more sites across
a wider geographic range to be sampled. A simple field pH measurement
technique also was found to produce results equivalent to those of more
complex techniques involving closed-headspace measurements and research-
grade apparatus.
Population Estimates
Univariate population distributions are described in terms of an index value,
which is the mean value of the chemical variable for the three spring
measurements (excluding samples collected during rainfall episodes) made
at the downstream node of each reach. Distribution estimates for pH and
ANC were found to be similar, whether expressed on the basis of numbers,
length, or surface area of the stream target population. Two additional
measurement variables involving discharge and mass export coefficients
appear possible, but are presently incomplete. The inclusion of samples
collected during episodes tended to depress ANC and pH values below their
relatively stable index values by 24% and 0.19 units, respectively.
When episodes were excluded, population estimates based on any of the three
spring sampling intervals were essentially identical. The summer sample
clearly produced higher population estimates for ANC, however. Samples
collected at the upstream nodes exhibited markedly lower concentrations for
pH, ANC, sulfate, and nitrate tl . i did the corresponding samples at the
downstream nodes on both spring and summer sampling dates.
A "worst case" estimate based on spring index chemistry and expressed in
terms of length of reaches indicates that, at the 95% confidence level, fewer
than 3.2% of the target population exhibited pH values below 6.4. Indeed,
-------
no pH measurement made during the survey using the most consistent and
reliable technique exhibited a value below 6.0, including measurements made
during episodes and at upstream nodes. This does not mean that low pH
conditions do not occur in the study area in very small headwater reaches,
during rain-driven episodes, or during other seasons or other years. However,
it suggests that chronic acidification of medium-sized streams in the study
region during a season commonly associated with the shortest hydrologic
residence times in the watersheds is not common. Despite the fact that the
pH values observed during the survey are well above the levels usually
associated with fish mortality, some estimate of transient chemical changes
that may occur during hydrologic episodes is needed before a critical evaluation
of chemical habitat quality can be complete.
Despite the generally circumneutral pH values, the population estimates for
ANC indicate that a majority of target streams were characterized as possessing
relatively low acid neutralizing capacity. Again, based on index chemistry and
expressed on a length basis, 6.3% of the combined reach length was estimated
to exhibit ANC values of 50 /ueq L~1 or less, while 74.4% was estimated to
be less than 200/jeq L~1. Although these values have been cited in the literature
as "extremely" and "moderately" sensitive waters, respectively, the
susceptibility of streams in the region cannot be fully evaluated without
additional consideration of soil chemistry, which may act to delay the surface
water response to acid deposition, according to some theories.
Classification
With respect to the potential for classification, analysis of the survey data
provided several lines of subjective and objective evidence indicating that a
reasonable geochemical classification is possible. Geographic analysis
indicates that reaches within broad ANC classes tend to cluster spatially. The
highest ANC sites were located along the western border of the study region,
while intermediate ANC sites were located in the Broad and French Broad
River valleys that contain the main population centers of the region. The lowest
ANC sites occurred in the north and central highlands, including Great Smoky
Mountains National Park. ANC appears to be highly correlated with weathering
of one of the dominant minerals in the area (K-feldspar), which suggests an
underlying geochemical control of ANC in the region. Finally, agglomerative
cluster analysis, an objective multivariate statistical technique, when applied
to a full chemical data set, produced classes very similar to those based on
ANC alone. This analysis also indicated that the special interest sites included
in the survey were typical of the low end of the ANC spectrum in the area,
but none was found to be an outlier.
Conclusions and Recommendations
The Phase l-Pilot Survey demonstrated that a regional scale synoptic survey
of streams will produce population estimates, with known confidence bounds,
for important chemical variables such as pH and ANC. The population estimates
appear to be robust, and are not particularly sensitive to small changes in
chemistry that occur over weekly time scales during the spring. Intra-site
temporal variability does not preclude chemical classification of target streams
in the Southern Blue Ridge, if effects of episodes are removed.
The Phase l-Pilot Survey was also useful in increasing the probability of success
and decreasing the cost of a full Phase I survey. It was determined that the
proposed design could be modified slightly to meet the needs and increase
the efficiency of the 1986 Phase I effort. Major recommendations included:
• Make minor alterations in the inclusion criteria and the statistical sampling
method to better address the assessment objectives of the survey and to
increase the sampling efficiency.
vi
-------
• Reduce sampling to two visits in the spring prior to leafout, to satisfy the
classification objective, or to one visit to satisfy the objective of population
estimation.
• Sample the reaches at both their upstream and downstream nodes on each
visit to characterize intra-reach spatial variability.
• Increase sample holding time protocols to 24 hours to allow central
placement of the mobile analytical laboratories, and thereby greatly increase
the logistical efficiency of the survey.
• Adopt certain field measurement techniques that provide accurate and
reliable data.
• Alter certain quality control/quality assurance and data management
techniques to increase efficiency and decrease lags in data availability, to
the extent that data quality can be maintained.
• Further develop new data analysis techniques that aid in data interpretation
in an assessment context.
These recommended changes were incorporated into the draft planning
documents for the NSS Mid-Atlantic Phase I and Southeast Screening Surveys,
which were peer reviewed in January, 1986. We believe that the NSS Phase
I design can provide important incremental information in the assessment
process, and will serve as an important stepping stone to the regionalization
of site-specific results gathered during both past and future studies.
VII
-------
Project Contributors
The NSS Phase l-Pilot Survey and this report are the result of the cooperative
efforts of many individuals and organizations. The project was administered
by Rick Linthurst (U.S. EPA) with Jay J. Messer {Utah State University) as
Technical Director. W. Scott Overton (Oregon State University) designed the
statistical sampling plan and provided guidance on the data analysis. James
Omernik, research geographer at U.S. EPA-CERL, co-authored the alkalinity
maps which served as a basis for delineating the study area. Jack Tuschall
(Northrop Services, Inc.) helped to draft the project plan and served as liaison
with the local cooperators before the field sampling. Charles W. Ariss (Utah
State University) served as technical liaison during the field effort and later
provided data analysis support, as did Barry Gall (Western Washington
University). Finally, the participants at the peer review workshop held in
Washington, DC, in December 1984 gave valuable suggestions for improving
the study design.
Under the direction of James Omernik, Andrew Kinney (Northrop Services,
Inc.) oversaw the detailed map work, assisted by geographers Anastasia B.
Allen, Douglas B. Brown, and Suzanne Pierson (all of Northrop).
Field operations commenced under the direction of Project Officer Robert E.
Crowe (U.S. EPA-EMSL-LV) and Project Manager, Steve L Pierett (Lockheed
EMSCO, Inc.). John R. Baker (Lockheed) supervised all field operations and
logistics; as base coordinator/science supervisor, he was responsible for the
field implementation of the survey. Ken Asbury (Lockheed) served as technical
supervisor.
On-site field coordinators Frank A. Morris, Randy G. Cameron, Al W. Groeger,
C. Mel Knapp, Ky B. Ostergaard, Cindy L. Mayer, Cindy A. Hagley, and Barry
Baldigo, all of Lockheed EMSCO, Inc., supervised the dedicated efforts of the
large crews who accomplished the field sampling.
David V. Peck served as training coordinator and Gerald J. Filbin as laboratory
supervisor for the Las Vegas laboratories (both of Lockheed). Lab analysts
were Linda A. Drewes, C. Hunter Holen, and J. M. Henshaw (all of Lockheed).
Kevin J. Cabbie (Lockheed) provided technical support to the laboratory
operations.
Robert A. Schonbrod, U.S. EPA-EMSL-LV, served as project officer for quality
assurance/quality control and analytical methods development. Sevda Drous<§
(Lockheed) managed QA operations and reporting. Bryant C. Hess and Carol
MacLeod, QA scientists, and Martin Stapanian, statistician, along with
programmers David T. Hoff, In Seung Lau, Rick K. Maul, and Joseph Scanlan,
all of Lockheed, supported the QA operations.
Dan C. J. Hillman (Lockheed) served as technical supervisor for analytical
methods. Technical writer for the QA/QC and Methods Group was Jan Engles
(Lockheed).
Michael J. Sale (Martin Marietta Energy Systems, Oak Ridge National
Laboratory) coordinated the NSS Phase l-Pilot Survey data base management
and oversaw the many exacting data transfers among the reporting
viii
-------
laboratories. Jan M. Coe (Martin Marietta, ORNL) managed the data base;
Henriette I. Jager (Science Applications International Corp.) and Mary Alice
Faulkner (Martin Marietta, ORNL) developed statistical program applications.
The synthesis of all of the above efforts at U.S. EPA-CERL to produce this
report was directed by Jay J. Messer, with the assistance of Keith N. Eshleman
(Northrop Services, Inc.). Chapters 1 (Introduction) and 6 (Conclusions) were
co-written by Messer and Eshleman. Philip R. Kaufmann (Utah State University)
contributed to Chapter 2 (Study Design) and Chapter 5 (Population Estimates
and Classification). Sharmon M. Stambaugh (Northrop) contributed to Chapter
3 (Field Operations) and provided technical editing/production assistance.
Chapter 2 (Study Design) contributors included W. Scott Overton, Jay J. Messer,
James Omernik, and Andrew Kinney. The contributors to Chapter 3 (Field
Operations) were John Baker and David M. Peck.
The quality assurance and data base management report (Chapter 4) was
written by Sveda Drous6, Michael J. Sale, and Charles W. Ariss.
Chapter 5 addresses the two major objectives of the NSS Phase l-Pilot Survey.
Messer, Eshleman, and Kaufmann co-wrote this chapter with statistical
graphics provided by ORNL Barry Gall (Western Washington University)
performed the cluster analyses on the ANC data for the classification section.
Additional project administration at U.S. EPA headquarters was provided by
William Fallen (Battelle NW Laboratories). Barbara Emmel (Radian Corporation)
served as technical writer throughout the project. Nancy Lanpheare (Northrop)
ably typed the draft of this report.
The NSS Phase l-Pilot Survey also acknowledges the illustrative experiences
and advice from the Aquatic Effects Research Team, in particular, the National
Lake Survey project team.
The NSS thanks the principal reviewers of this report, distinguished by their
expertise in appropriate disciplines and knowledge of streams in the study
region. They were: Donald Porcella (EPRI), Gary Larson (Oregon State
University), Jerry Elwood (Oak Ridge National Laboratories-Environmental
Sciences Division), and Ken Reckhow (Duke University). In addition, this report
was reviewed by state air and water quality staffs in North Carolina, Tennessee,
South Carolina, and Georgia and by U.S. EPA Region IV staff.
The authors gratefully acknowledge all who contributed to the NSS Phase
l-Pilot Survey but who may not have been named in this section. The success
of the project reflects these participants' contributions of ideas, efficiency,
enthusiasm, and hard work.
IX
-------
Contents
Page
Notice ii
Abstract iii
Executive Summary iv
Project Contributors viii
List of Figures xiv
List of Tables xvi
Ancillary Reports xviii
Chapter 1. Introduction .1
1.1 Overview 1
1.2 The National Surface Water Survey 2
1.3 National Stream Survey—Phase I 3
1.3.1 Phase I Planning 3
1.3.2 Phase l-Pilot Survey ....... 4
1.4 Phase l-Pilot Survey Report 4
1.5 Project Organization 5
Chapter 2. Study Design 6
2.1 Overview 6
2.2 RIS and the "Index" Concept 6
2.2.1 Regionalized Integrated Studies 6
2.2.2 The "Index" Concept 6
2.2.3 Data Quality Objectives 7
2.3 Identifying the Target Population 7
2.3.1 Selection of the Study Area 8
2.3.2 Stream Population of Interest 8
2.4 Target Population Estimates 11
2.4.1 Methods for Identifying the Target Population 11
2.4.2 First Stage of Sampling 11
2.4.3 Site Inclusion Criteria ("Site Rules") 12
2.4.4 First Stage Data 14
2.4.5 Population Estimates 15
2.4.6 Second Stage of Sampling 16
2.4.7 Target Population Geographic Estimates 16
2.4.8 Special Interest Reaches 18
2.5 Third State of Sampling 18
2.5.1 Variables Measured 19
2.5.2 Sampling Season 21
2.5.3 Sampling Locations on Each Reach 22
2.6 The Watershed Alternative to the Reach Frame 22
Chapter 3. Field Operations 24
3.1 Introduction .24
3.2 Preparation for Field Operations 24
xi
-------
Contents (Cont'd)
Page
3.2.1 Protocol Development 24
3.2.2 Training Programs 24
3.2.3 Field Station Site Selection and Site
Reconnaissance 25
3.3 Field Operations 25
3.3.1 Field Station Operations 25
3.3.2 Field Sampling Operations 25
3.3.3 Field Laboratory Operations 29
3.4 Evaluation of Equipment and Methods 32
3.4.1 Equipment Evaluation 32
3.4.2 Methods Evaluation 32
3.4.2.1 Filtration Methods 32
3.4.2.2 Streamside pH Measurements 33
3.4.2.3 Aluminum Methods 33
3.4.3 Holding Time Studies 35
3.4.3.1 Syringe Experiment 35
3.4.3.2 Cubitainer Experiment 35
3.5 Summary of Field Operations 37
Chapter 4. Quality Assurance and Data Management 38
4.1 Introduction 38
4.2 Quality Assurance/Quality Control Operations 38
4.2.1 Selection of Contract Analytical Laboratories 38
4.2.2 Training 38
4.2.3 Daily Quality Assurance Contact 38
4.2.4 Field and Contract Laboratory Audits 39
4.2.5 Field Sampling Quality Control Procedures 39
4.2.6 Field Laboratory Quality Control Procedures 39
4.2.7 Quality Assurance/Quality Control Samples 39
4.2.7.1 Quality Control Samples 40
4.2.7.2 Quality Assurance Samples 40
4.2.8 Data Review 41
4.3 Data Base Management 42
4.3.1 Data Structure and Flow 42
4.3.2 Primary IMSS Data Sets 42
4.3.3 Enhanced Data Files 44
4.3.4 Data Change and Qualifiers 44
4.4 Data Verification 45
4.4.1 Review of Field Data Forms 45
4.4.2 Initial Review of Sample Data Package 45
4.4.3 Review of Quality Assurance/Quality Control Data 45
4.4.4 Follow-Up with Contract Laboratories 46
4.4.5 Preparation and Delivery of Verification Tapes 46
4.5 Data Validation 46
4.5.1 Frequency Analyses 47
4.5.2 Univariate Analyses 47
4.5.3 Multivariate Scoping 47
4.5.4 Bivariate/Multivariate Linear Regression
Analyses 48
4.5.5 Multivariate Analyses 49
x/7
-------
Contents (Cont'd)
Page
4.5.6 Episodes Screening 49
4.5.7 Reverification/Validation and Data
Correction/Flagging 49
4.6 Data Management and Quality Assurance Results 50
4.6.1 Data Base Management Performance 50
4.6.2 Verification/Validation Performance 51
4.6.3 Data Quality 52
4.6.3.1 Detection Limits 52
4.6.3.2 Precision 52
4.6.4 Summary 54
Chapter 5. Population Estimates and Stream Classification 55
5.1 Introduction 55
5.2 Population Estimates 55
5.2.1 Graphical Displays 55
5.2.2 Alternative Measurement Variables 62
5.2.3 Reference Values 63
5.2.4 Sample Timing and Frequency 64
5.2.5 Spatial Aspects of Reach Chemistry 65
5.2.6 Interpretation of Regional Assessments 69
5.3 Stream Classification 76
5.3.1 Univariate Models 76
5.3.2 Geographic Distributions 76
5.3.3 Cluster Analysis 83
5.3.4 Utility of Classification for Regional Assessment 85
5.4 Future Analyses 87
Chapter 6. Conclusions and Recommendations 88
6.1 Conclusions 88
6.2 Recommendations for Phase I 89
6.3 Related Documents 90
Chapter 7. References 91
Appendix A 95
Appendix B 119
Appendix C 121
Appendix D 125
Kill
-------
List of Figures
Number Page
1 -1 Organization of the National Surface Water Survey,
showing two major components (lake and
stream surveys), each consisting of three phases 2
2-1 Location of the Southern Blue Ridge Phase l-PMot Survey
study area 9
2-2 Geography of the NSS Phase l-Pilot Survey study area 10
2-3 Represention of the point frame sampling procedure for NSS
study reaches 13
2-4 NSS Phase l-Pilot Survey study area showing second
stage (II) probability sites and special interest reaches 17
3-1 NSS Data Form 7: Watershed Characteristics 26
3-2 Daily field station activities in the Phase l-Pilot Survey 27
3-3 Daily activities of the field sampling teams during the
Phase l-Pilot Survey 28
3-4 NSS Data Form 4: Stream Data 30
3-5 Daily activities at the field laboratory during the
Phase l-Pilot Survey 31
3-6 Comparisons of three pH methods used in the Phase l-Pilot
Survey 34
4-1 NSS data structure and flows 43
5-1 Population distribution estimates for average spring
downstream pH conducted at the mobile laboratory on
samples held in syringes closed to the atmosphere in the
NSS Phase l-Pilot Survey 56
5-2 Population distribution estimates for average spring
downstream acid neutralizing capacity (ANC) in streams
in the NSS Phase l-Pilot Survey 57
5-3 Population distribution estimates for average spring
downstream sulfate concentrations in streams in the NSS
Phase l-Pilot Survey 58
5-4 Population distribution estimates for average spring
downstream nitrate concentrations in streams in the NSS
Phase l-Pilot Survey 59
5-5 Population distribution estimates for average spring
downstream chloride concentrations in streams in the NSS
Phase l-Pilot Survey 60
5-6 Population distribution estimates for average spring
downstream extractable aluminum concentrations in
streams in the NSS Phase l-Pilot Survey 61
xiv
-------
List of Figures (Cont'd)
Number Page
5-7 Comparison of population length distribution estimates for
pH and ANC from the three spring and one summer
sampling intervals 66
5-8 Comparison of population length distribution estimates for
sulfate and nitrate from the three spring and one summer
sampling intervals 67
5-9 Comparison of population length distribution estimates for
chloride and extractable aluminum based on the three spring
and one summer sampling intervals 68
5-10 Comparisons of frequency distribution estimates for pH and
ANC in Phase l-Pilot Survey streams based on upstream
versus downstream sampling locations during the summer
sampling interval 70
5-11 Comparisons of frequency distribution estimates for sulfate
and nitrate concentrations in Phase l-Pilot Survey
streams based on upstream versus downstream sampling
locations during the summer sampling interval 71
5-12 Comparisons of frequency distribution estimates for
chloride and aluminum concentrations in Phase l-Pilot
Survey streams based on upstream versus downstream
sampling locations during the summer sampling interval 72
5-13 ANC distribution in the Southern Blue Ridge based on
downstream spring average chemistry with effects from
storm events removed 75
5-14 Geographic distribution of average springtime downstream
pH in the NSS Phase l-Pilot Survey streams 77
5-15 Geographic distribution of average springtime downstream
ANC in the NSS Phase l-Pilot Survey streams 78
5-16 Geographic distribution of average springtime downstream
sulfate concentrations in the NSS Phase l-Pilot Survey
streams 79
5-17 Geographic distribution of average springtime downstream
nitrate concentrations in the NSS Phase l-Pilot
Survey streams 80
5-18 Geographic distribution of average springtime downstream
chloride concentrations in the NSS Phase l-Pilot Survey
streams 81
5-19 Geographic distribution of average springtime downstream
extractable aluminum concentrations in the NSS
Phase l-Pilot Survey streams 82
5-20 Hierarchial cluster diagram of all NSS Phase l-Pilot Survey
sites based on downstream spring average values for 39
chemical variables 84
5-21 Potassium-feldspar mineral stability diagram for streams in
the NSS Phase l-Pilot Survey 86
xv
-------
List of Tables
Number Page
1-1 Objectives of tne National Surface Water Survey 3
2-1 NSS Phase l-Pilot Survey Site Inclusion Criteria 14
2-2 Geographic Attribute Estimates and Standard Errors for the
NSS Phase l-Pilot Survey Target Populations Based on
Stage I and Stage II Samples 18
2-3 Variables Measured During the NSS Phase l-Pi!ot Survey 19
2-4 Chemical Variables and Corresponding Measurement
Methods for the National Stream Survey 21
3-1 Summary of Routine Samples Collected During the NSS
Phase I- Pilot Survey 24
3-2 List of NSS Aliquots, Containers and Preservatives 32
3-3 Dissolved Inorganic Carbon Concentrations (mg L~1 ± 1 s.d.) in
Samples Initially Sub- or Supersaturated with C02 and Held
for 7-8 Days .35
3-4 Changes in Constituent Concentrations in Duplicate Field
Samples and Big Moose Lake QA Audits Held at 4°C for 12,
24, 48, and 84 Hours Prior to Stabilization 36
4-1 Types, Sources and Applications of Quality Control Samples
Used in the Phase l-Pilot Survey (Drous6, 1987) 40
4-2 Types, Sources and Applications of Quality Assurance
Samples Used in the Phase l-Pilot Survey (Drous6, 1987) .41
4-3 Composition of Big Moose Lake (FN4) and Bagley Lake (FN5)
Natural Audit Samples 42
4-4 Data Set Members for the Raw, Verified, and Validated
Versions of the NSS Phase l-Pilot Survey Data Base 44
4-5 Exception Generating Programs Within the AQUARIUS Data
Review and Verification System (Fountain and Hoff, 1985) 46
4-6 Variable Suites Obtained from Multivariate Scoping 48
4-7 NSS Validation Flags 50
4-8 Results of Verification/Validation: Numbers of Observations
Flagged and Numeric Changes Made (and percent of
total observations) in the NSS PIPS Data Base (excluding
episode flags) 51
4-9 System Decision Limits and Precision Estimates Based on
Interbatch Analysis of Field Audits and Intrabatch Analyses of
Field, Trailer, and Laboratory Duplicates (Drouse, 1987) 53
5-1 Phase l-Pilot Survey Length Distribution Estimates
Associated with Reference Values Based on Natural
Univariate Groupings of Streams (Except Where Noted for
ANC) 64
xvi
-------
List of Tables (Cont'd)
Number Page
5-2 Effects of Rainfall Events on ANC and pH at Seven
Downstream Phase l-Pilot Survey Sampling Sites 65
5-3 Statistically Significant (p = 0.05) Differences Between Mean
Concentrations of Primary Variables Between Spring (SP1,
SP2, SP3,) and Between Summer (SU) and Average Spring (SP)
Sampling Intervals (downstream nodes) for Streams with
< 250 yeq L"1 ANC 69
5-4 Comparison of Upstream/Downstream Chemistry During the
Third Spring (SP3) and Summer (SU) Sampling Intervals,
Based on a Paired t-Test with Differences Weighted to
Reflect Inclusion Probabilities (wj 73
XVII
-------
Ancillary Reports
In addition to this data report, supplemental information on the National Stream
Survey Phase l-Pilot Survey can be found in the series of ancillary manuals
and reports. Many of the technical manuals used in working draft form at
the time the Phase l-Pilot Survey was conducted. These publications include:
Field Operations Report, National Surface Water Survey, National Stream
Survey, Pilot Survey. 1986. Knapp, C. H., C. L. Mayer, D. V. Peck, J. R.
Baker, and G. J. Filbin. Lockheed Engineering and Management Services
Company, Inc., Las Vegas, Nevada 89109 (draft).
Quality Assurance Plan for the National Surface Water Survey. Stream Survey
(Middle Atlantic Phase I, Southeast Screening and Middle Atlantic Episodes
Pilot). 1986. Drouse, S. K., D. C. Hillman, L. W. Creelman, and S. J. Simon.
Lockheed Engineering and Management Services Company, Inc., Las Vegas,
Nevada 89114 (draft).
Evaluation of Quality Assurance and Quality Control Sample Data for the
National Stream Survey (Phase l-Pilot Survey). 1986. Drouse, S. K. Lockheed
Engineering and Management Services Company, Inc., Las Vegas, Nevada
89109 (draft).
Analytical Methods Manual for the National Surface Water Survey. Stream
Survey (Middle Atlantic Phase I, Southeast Screening, and Middle Atlantic
Episodes Pilot). 1986. Hillman, D. C., S. H. Pia, and S. J. Simon. Lockheed
Engineering Management Services Company, Inc., Las Vegas, Nevada 89114
(draft).
Data Management and Analysis Procedures for the National Stream Survey.
1987. Sale, M. J. (editor). ORNL/TM. Oak Ridge National Laboratory, Oak
Ridge, Tennessee 37831 (draft).
Draft Research Plan, National Surface Water Survey: National Stream Survey,
Mid-Atlantic Phase I and Southeast Screening. 1985. U.S. Environmental
Protection Agency, Office of Research and Development, Washington, DC
20460.
Draft Sampling Plan for Streams in the National Surface Water Survey. 1985.
Technical Report 114 (July 1986). Overton, W. S. Department of Statistics,
Oregon State University, Corvallis, Oregon 97331.
A Sampling and Analysis Plan for Streams in the National Surface Water
Survey. 1987. Technical Report 117. Overton, W. S. Department of Statistics,
Oregon State University, Corvallis, Oregon 97331.
xvin
-------
1. Introduction
1.1 Overview
The relationship between acid deposition and the
acidification of surface waters has become one of
the most critical environmental issues of the 1980s.
Studies on a variety of individual water bodies and
regional populations of lakes and streams have
produced data that suggest that surface waters in
some areas of Europe and North America have
experienced declines in pH and/or acid neutralizing
capacity over the past half century (Beamish and
Harvey, 1972; Beamish et al., 1975; Oden, 1976;
Wright and Gjessing, 1976; Watt etal., 1979; Pfeiffer
and Festa, 1980; Haines and Akielaszek, 1983; Smith
and Alexander, 1983). Acidic atmospheric deposition
arising from the combustion of fossil fuels has been
the most commonly attributed cause for such
declines (Drablos and Tollan, 1980; National
Research Council, 1981, 1983, 1984; Office of
Science and Technology Policy, 1984; Office of
Technology Assessment, 1984; U.S. EPA, 1984a;
Jeffries et ai., 1985). Alternative hypotheses and
discrepancies in the atmospheric acidification
scenario also have been discussed and debated in
the recent literature (e.g., Havas et al., 1984;
Howells, 1984; Cogbill et al., 1984; Lefohn and
Brocksen, 1984; Krug et al., 1985; Pierson and
Chang, 1986).
Th© latter arguments notwithstanding, previous
studies have left two critical gaps in our ability to
assess the quantitative risk associated with the
effects of acid deposition on surface water resources
in the United States:
1. It is impossible to combine the results of
previously conducted independent regional
surveys and historical data from monitoring
networks or site-specific research projects in
order to produce a quantitative estimate with
known confidence bounds of the present extent
of low pH waters, or of waters whose chemistry
is indicative of potential susceptibility to acid
deposition inputs. The problems stem primarily
from an inadequate statistical sampling plan,
inconsistencies in field or laboratory methods,
insufficient chemical measurements to ade-
quately characterize water quality, or inade-
quate quality assurance data by which to
evaluate potential bias between or among data
collected during the different studies.
2. It is virtually impossible to quantitatively
extrapolate the results from intensive, process-
oriented (cause and effect) research in a few
watersheds to the larger lake or stream
population comprising the resource at risk in
a given geographic region. This inability stems
from the lack of statistically defensible popu-
lation estimates noted above, together with the
absence of a companion lake or stream
classification strategy based on the regional
distribution of water body characteristics. It is
seldom quantitatively known whether research
sites are broadly typical of the majority of other
systems in the region, representative of a
relatively small (but perhaps potentially impor-
tant) subpopulation, or relatively unique. Given
the common research requirements that a
study site be relatively pristine, the possibility
that the site is pristine because it is otherwise
relatively unique is not unlikely.
The National Surface Water Survey was designed
to overcome these obstacles to assessment by
sampling water quality in lakes and streams on a
regional basis using a statistically rigorous survey
design, appropriate field and analytical techniques,
a sufficient set of measurement variables, and an
adequate quality assurance and control program to
maximize the confidence of the resulting data. The
initial survey component (Phase I) would provide a
snapshot of the present condition of surface water
in the regions most likely to exhibit effects from acid
deposition. The Phase I data would also serve as
a basis for classification of the lakes and streams,
so that results from past and subsequent intensive
studies on subpopulations of interest or at individual
study sites would be extrapolated with known
confidence to the regional populations.
The purpose of this report is to describe the results
of the Phase l-Pilot Survey, a component of the
National Stream Survey conducted in the Southern
Blue Ridge Province of the southeastern U.S.
conducted in 1985. The objectives of the Phase !-
Pilot Survey were to test the logistics plan and
statistical sampling design proposed for a full Phase
I effort in 1986. We will demonstrate the adequacy
of a modification to the original design by examining
the types of project outputs that could be expected,
based on the Phase l-Pilot Survey results. At this
-------
level of analysis, no attempt has been made to
interpret the data with respect to the likelihood of
past or future changes from acid deposition in the
region. Such analyses are presently the target of
considerable research effort, however, and will be
the subject of future project outputs.
1.2 The National Surface Water Survey
In response to the need for knowledge regarding the
present extent of the acidic or potentially susceptible
aquatic resource and its associated biota, the U.S.
Environmental Protection Agency and cooperating
scientists were asked in 1983 to design a program
to satisfy five major goals:
1. Characterize the chemistry of surface waters
(both lakes and streams) in regions of the
United States presently believed to be poten-
tially susceptible to change as a result of acid
deposition.
2. Examine associations among chemical constit-
uents and define important factors that may
affect surface water chemistry.
3. Determine the biological resources within
these systems.
4. Evaluate correlations among surface water
chemistry and the corresponding biological
resources.
5. Quantify any regional trends in surface water
chemistry and/or biological resources.
The resulting program designed to meet these goals
was designated the National Surface Water Survey
(NSWS). The NSWS became an integral part of the
National Acid Precipitation Assessment Program
(NAPAP), an interagency research, monitoring, and
assessment effort mandated by Congress in 1980.
NAPAP provides policy makers with technical
information concerning the extent and severity of
the effects of acid deposition ("acid rain") on human,
terrestrial, aquatic, and material resources.
In order to satisfy its five major research goals, the
NSWS was designed in two parallel components,
the National Lake Survey (NLS) and the National
Stream Survey (NSS) (Figure 1-1). Both components
consist of phases, each of which depends on the
preceding phases to satisfy its objectives (Table 1 -1).
This design grew out of the recognition that while
it is clearly not feasible to perform intensive, process-
oriented studies or monitoring programs on all
surface waters within the U.S., it is equally
inappropriate to study a few systems that later may
be found to have atypical biological and chemical
characteristics. Therefore, each component of the
Figure 1-1.
Organization of the National Surface Water
Survey, showing two major components (lake
and stream surveys), each consisting of three
phases.
National Surface Water Survey (NSWS)
National Lake Survey (NLS) National Stream Survey (NSS)
Synoptic Chemistry
Eastern Survey (1984)
Western Survey (1985)
Synoptic Chemistry
Pilot Survey (198S)
Synoptic Survey (1986-87)
Temporal Variability (1986-87)
Biologial Resources (1986)
Episodic Effects (1988)
Biological Resources (1988)
Long-Term Monitoring (1988)
NSWS begins with Phase I, a synoptic survey phase
designed to characterize and quantify the chemistry
of lakes and streams throughout the U.S., focusing
on the areas expected to contain the majority of low-
alkalinity waters.
Phase I data cannot be used to prove that a causal
link exists between observed aquatic effects and acid
deposition. Although the major concern over the
aquatic effects of acid deposition is its impact on
biological resources, it is more efficient to first
characterize surface waters in terms of the physico-
chemical factors that are expected to impact biota,
rather than to begin the process with a biotic survey
of all surface waters in a region, regardless of water
quality. The present study design, based on the Phase
I chemical classification, can be used not only to
quantify the present status of the aquatic resource,
but also to allow correlative relationships to be
examined among relatively homogeneous lake and
stream types. It also allows the selection of
geochemically representative sites for more studies
of intensive biological characteristics, temporal
variability in water chemistry, and long-term
changes.
The second phase of the NSWS will quantify the
biota and short-term (seasonal, weekly, or episodic)
variability in water chemistry within and among
-------
Table 1-1. Objectives of the National Surface Water Survey
Phase I
Synoptic Chemical Survey
Biological
Resources/Temporal Variability
Long-Term Monitoring
1. Quantitatively estimate the
percentages (number/length/
area) and location of acidic
streams in regions of the U.S.
potentially susceptible to acid
deposition.
2. Quantitatively estimate the
percentages (number/length/
area) and locations of lakes/
streams with low acid
neutralizing capacity in regions of
the U.S. potentially sensitive to
acid deposition.
3. Determine which lakes/streams
are representative of important
aquatic resources in the region
and should be selected for further
study in later phases.
1. Determine how many
representative Jakes and streams
are fishless.
2. Assess the temporal variability in
chemistry in representative lakes
and streams.
3. Determine the lake and stream
chemical characteristics
associated with fish presence/
absence.
4. Determine which chemically and
biologically representative
systems should be selected for
long-term monitoring.
1. Determine what chemical and
biological changes are occurring
over time in representative lakes
and streams.
2. Measure the rate at which
changes are occurring.
representative lakes and streams in each geographic
region. The definition of representativeness will be
based on Phase I water chemistry, hydrology, biotic
composition, regional acid deposition inputs, land
use, physiographic features, and other characteris-
tics. Some regionally representative sites will later
become the foundation for a long-term monitoring
program to detect and quantify any future changes
in the chemistry and biology of potentially suscept-
ible aquatic ecosystems in the region. Many lakes
and streams that have been the focus of intensive
and/or long-term studies in the past are included
in the Survey as "special interest" sites. Such sites
that are found to be representative of large numbers
of other aquatic systems in their respective regions
could serve as the nucleus of a long-term monitoring
effort.
Phase I of the Eastern Lake Survey has been
completed. A total of 1798 lakes in the eastern U.S.
were sampled in the fall of 1984, and 752 lakes
were sampled in selected areas of the western U.S.
in the fall of 1985. Phase II field work was begun
to determine seasonal chemical variability in
northeastern lakes in the spring of 1986. The status
of the National Stream Survey is discussed below.
1.3 National Stream Survey
1.3.1 Phase I Planning
Planning for Phase I of the National Stream Survey
(NSS) began in mid-1984 and resulted in a Draft
Research Plan (U.S. EPA, 1984b). Phase I of the NSS
was designed to chemically and physically charac-
terize a target population of streams existing within
any relatively homogeneous physiographic region,
based on a probability sample of those streams. It
has the joint major goals of description and
classification of the streams in the target population.
More specifically, the primary objectives of Phase
I of the NSS are to determine:
1. The percentage, extent (e.g., number, length,
and drainage area), and location of streams in
the United States that are presently acidic.
2. The percentage, extent, and location of streams
that have low acid-neutralizing capacity, and
thus might become acidic in the future.
3. Which streams are representative of important
classes of streams in each region and should
be selected for more intensive studies or long-
term monitoring.
The NSS was specifically designed to achieve these
objectives within known confidence limits. It was
also designed to allow the objectives to be met for
any chemical variable measured. For example, the
percentage of the population of stream reaches
within a given region that have sulfate, nitrate,
aluminum, and/or calcium concentrations above or
below any criterion value of interest could also be
determined. Should sensitivity to acidification be
acceptably defined in the future, based on one or
several of the variables being measured, the Survey
design will also permit post-stratification to deter-
mine the number and areal extent of streams that
fall into such sensitivity categories.
The sampling design also lends itself to many
comparative evaluations. For example, other ques-
tions that could be answered by the design include:
-------
1. Are acidic streams found primarily at high
elevation?
2. Are acidic streams found in small watersheds?
3. Are acidic streams found within areas with the
highest acid deposition rates?
4. Are sulfate and base cation concentrations in
different regions of the U.S. correlated with
regional deposition chemistry or with the
nature of watershed soils or geology?
5. Can existing alkalinity maps be refined?
6. What associations exist among water chem-
istry, land use, vegetation type, and geographic
data?
The principal restriction on these secondary
objectives was that they must not result in a design
that compromises the Phase I primary objectives.
In many cases, such secondary objectives might best
be met in later phases of the project.
1.3.2 Phase I—Pilot Survey
The initial research plan for Phase I underwent peer
review at a workshop in Washington, D.C., in
December, 1984. The workshop participants recom-
mended that a full Phase I survey should be preceded
by a pilot study whose findings might increase the
efficiency and quality of future field efforts. Planning
was begun immediately for such a pilot study with
the following objectives:
1. test the ability of the proposed sampling design
to meet the Phase I objectives, based on
analysis of data collected during the Pilot
Survey;
2. evaluate the Phase I logistics plan (including
safety issues and questions of legal and
physical site access) and alternative sample
collection, preparation, and analytical tech-
niques; and
3. develop and test a data analysis plan for Phase
I using actual data collected in the Pilot Survey.
Field work for the Phase l-Pilot Survey began in the
Southern Blue Ridge Province (Figure 2-1) in March,
1985, and was completed in June of the same year.
1.4 Phase I—Pilot Survey Report
This report summarizes the design and results of
the Phase l-Pilot Survey. A description of the Phase
l-Pilot Survey design is presented in Chapter 2. The
Survey employed the random placement of a
systematic sampling grid over 1:250,000-scale
topographic maps of the Southern Blue Ridge
Province to obtain a sample of stream reaches within
a pre-selected approximate size range and which met
certain other site inclusion criteria. By this method,
a sample of 115 reaches was selected for the
estimation of stream length, drainage area, and other
geographical characteristics. A random systematic
subsample of 54 reaches was selected from the
initial 115, to be visited by field crews to make on-
site physical and chemical measurements and to
collect water samples for laboratory chemical
analysis. The Pilot Survey utilized an "index" sample
to describe the chemical characteristics of each of
the 54 reaches. The average, non-event, spring
stream chemistry is analogous to the index samples
taken from the deepest part of lakes during fall
overturn in the Eastern Lake Survey (Linthurst et
al., 1986).
Chapter 3 describes the field and laboratory methods
used to collect data, as well as results of field and
laboratory experiments and evaluations. Such
information served as the basis for changes in
protocols of sample handling and analysis used in
subsequent Phase I Survey activities.
Chapter 4 presents, in detail, the quality assurance
and data base management programs employed in
the Survey. A variety of quality assurance and quality
control samples were employed to evaluate the
performance of the field sampling and analytical
activities, and to ensure that field and laboratory
activities were being conducted according to
established guidelines. Chapter 4 also summarizes
the QA results of the Pilot Survey. The data base
management tasks described in the section include
protocols for data flows and the statistical techniques
used to ensure data quality.
Chapter 5 evaluates the ability of the Phase S-Pilot
Survey design to meet the NSS Phase I objectives
and includes:
1. population distribution estimates for water
quality index variables, along with their
associated upper confidence bounds, including
an evaluation of the number and timing of field
data collections which led to the construction
of such estimates;
2. examples of potential classification approaches
for Phase I streams that could be used in future
phases of the study, and an evaluation of the
impact of sample timing and frequency on such
classifications; and
3. promising directions for further analysis of
synoptic data.
-------
Chapter 6 summarizes the conclusions from the
study and presents recommendations for the design
and implementation of the Phase I Survey.
It is important to recognize that the results from the
Phase l-Pilot Survey are strictly applicable only to
a defined target population of small to medium-sized
streams in the Southern Blue Ridge Province during
the spring and summer of 1985. The chemical
characteristics of this target population were
described using chemical index samples, excluding
those collected during major rainfall events. No data
were collected on very small intermittent or
headwater streams, and no attempt was made to
determine the lowest pH during storm-related
episodes which may be a critical factor affecting
survival of sensitive life stages or particular species
of fish. Further interpretation from an assessment
standpoint is being addressed in both present and
planned projects within the NAPAP Aquatic Effects
Research Program, and will be the subject of future
reports. Many of these projects will require additional
data collection (e.g., to determine low pH conditions
during storm-related episodes).
With respect to future design decisions, it was
recognized that differences in weather patterns,
hydrology, and watershed biogeochemistry may alter
many of the relationships observed in streams in
the Southern Blue Ridge. Therefore, the Southern
Blue Ridge results cannot be extrapolated quantit-
atively to streams in other parts of the country.
Consequently, conclusions regarding design recom-
mendations are generally not based on rigorous
statistical tests, but on finding consistent, reasonable
patterns in the data that allow important differences
of potential assessment significance to be discerned.
Ultimately, completion of the Phase I field work will
allow full appraisal of the success of the final design.
1.5 Project Organization
The National Stream Survey is administered by the
U.S. Environmental Protection Agency, Office of Acid
Deposition, Environmental Monitoring, and Quality
Assurance in Washington, D.C. The Environmental
Research Laboratory—Corvallis (ERL-C) is respon-
sible for coordinating the activities of the Survey and
for project design, data validation, and data
interpretation. The Environmental Monitoring
Systems Laboratory—Las Vegas (EMSL-LV) is
responsible for quality assurance/control, logistics,
and analytical support. Oak Ridge National Labor-
atory (ORNL) is responsible for developing and
maintaining the data base management system for
the Survey. ORNL also provided statistical program-
ming to implement the target population character-
ization, as well as mapping and other geographic
analyses for the survey
-------
2. Study Design
2.1 Overview
The design of the National Stream Survey is guided
by the general goals and approach of the National
Surface Water Survey, as described in Chapter 1,
and by a set of data quality objectives (DQOs) that
are intended to assure that resources expended in
sampling and analysis yield sufficiently precise and
accurate data to enable a useful interpretation of
that information with quantifiable confidence. This
chapter reiterates the overall goals of Phase I of the
NSS in the context of the NAPAP Aquatic Effects
Research Team "index" concept and the Regional-
ized Integrative Studies (RIS) approach and presents
a concise statement of the DQOs. Subsequent
sections describe the process of defining the target
population, drawing statistical samples from that
target population, and characterizing some of the
physical and chemical attributes of the streams
therein.
2.2 RIS and the ' 'Index" Concept
2.2.1 Regionalized Integrated Studies
The NSWS seeks to achieve the regional charac-
terization of surface waters by linking chemical,
biological, soils, and watershed studies through the
selection of common study sites. Such linkages are
critical to the Regionalized Integrative Studies (RIS)
approach. This approach begins with a large scale
classification study, such as the NSWS, to identify
regionally characteristic systems. Subsequently, a
smaller number of such characteristic systems will
be the focus of detailed research designed to
elucidate the mechanisms responsible for acidifica-
tion, to determine relationships between chemistry
and biological resources, or to detect long-term
changes, should they occur. Processes, relation-
ships, or changes observed in lakes and streams that
are typical of the various types of systems comprising
the regional population then can be extrapolated with
quantifiable confidence to a regional scale. Much
of the NAPAP aquatic effects research will hinge
on this "regional classification" approach, the
cornerstone of the RIS concept.
2.2.2 The "Index " Concept
Extrapolating from intensive studies to regional
population estimates relies on prior estimates of the
total population resource in a region, and of the
fraction of that population represented by the
intensively studied systems. The classification, or
determination of "representativeness," of the test
system must be based on the major factors thought
to control acidification in the population. Such
underlying factors could include regional hydrology,
geochemistry, and major vegetation types. Because
such factors are quite complex, however, it will
probably be necessary to rely on "indices" to
represent many of them. Data upon which to base
such indices must be available (or derivable from
maps and remote imagery) for a sufficiently large
and representative sample to estimate their fre-
quency of occurrence in the population. Examples
of simple physical and biological indices include
stream drainage density, mean watershed slope,
elevation, and percent coniferous vegetation. Phase
I of the NSWS relies on grab samples taken from
a number of water bodies during an appropriate
season to provide an "index" of the chemical
characteristics of the population of water bodies in
the region. Index samples for the lakes in the NSWS
were collected following fall overturn, when
intralake spatial variability is minimized. Ideally, if
the integrative capacity of a lake basin is sufficient,
a single sample collected at this time may provide
an "index" of chemical conditions at other times of
the year. The predictive success of a fall overturn
water chemistry index is influenced by many
processes, but is generally proportional to the
hydraulic residence time of the lake. Long residence
times tend to integrate the inputs of water and
dissolved materials from the lake watershed,
reducing that portion of temporal variability due to
changes in input rates. In streams, which have little
or no temporal integrative capacity in their channels,
it is necessary to draw the index sample during a
period of the year that is expected to exhibit
characteristics most closely linked to acid deposition
or to its most deleterious effects. Spring appears to
be the most appropriate period because stream water
acid neutralizing capacity (ANC) is typically low and
life stages of aquatic biota that are sensitive to low
pH are likely to be present at this time.
Although pH and ANC depressions also can occur
during other seasons, short hydraulic residence
-------
times in soil zones in the spring minimize acid
neutralization. Also, acid-sensitive swim-up fry life
stages of key fish species are typically present in
the streams during the spring in many parts of the
country. However, the short hydraulic residence
times that contribute to low pH and low ANC
conditions in the spring also may result in order-
of-magnitude changes in some chemical variables
over the course of hours or days during hydrologic
events. In order to reduce error in the population
estimates caused by such "atypical" samples,
replicate measurements on each reach were planned
for the Phase l-Pilot Survey, so that atypical values
could either be averaged, or identified and excluded.
The philosophy of indexing, if not the exact
methodology, is identical in the stream and lake
components of the NSWS. The multiple samples of
reach water chemistry in the NSS should be thought
of as replicates, with averages'replacing the single
NLS index sample in making regional population
estimates. Examples of how this index sample might
be used in stream classification are provided in
Chapter 5.
2.2.3 Data Quality Objectives
The data quality objectives (DQOs) of the National
Stream Survey were designed to overcome some of
the historical problems in past data sets noted in
Chapter 1. Few of the DQOs in a descriptive and
classificatory project such as the NSS can be
specified in terms of narrowly defined precision
limits, outside of which the data would be rendered
useless. Instead, the DQOs represent ideal targets.
However, even the qualitative specification of DQOs
in the planning process has resulted in significant
improvements to project designs and protocols.
Specifically, the DQOs are as follows:
1. The target population should accurately
represent the population of streams that
constitute the most important resource at risk
from acid deposition.
2. The NSS data should describe a probability
sample of streams from the target population.
3. The set of variables measured should be
sufficiently complete to provide information on
the suitability of the stream for key fish species
and on the geochemical parameters that can
be used to classify the streams and hypothesize
mechanisms relating to past and future
acidification (Phase I data may be supple-
mented during later phases to meet certain of
these objectives).
4. The data must be of high quality, with low and
quantifiable analytical error, and with known
precision, representative of the state-of-the-art
attainable in high-volume contract analytical
laboratories.
5. Sample variances must be sufficiently small to
provide useful population estimates and robust
stream classifications, to the extent that natural
classes exist in the target populations.
In most cases the data quality objectives can only
be met subjectively. In the case of DQO #4, however,
the results of the Phase l-Pilot Survey can be used
to provide a benchmark against which future
analytical data quality can be compared.
r,
2.3 Identifying the Target Population
A sampling design depends upon the identification
of a "target population," i.e., a collection of entities
about which we want to make estimates (and
ultimately management decisions). Only when such
a target population is explicitly defined can samples
be drawn from it in order to make statistical
inferences regarding the properties of that popula-
tion. In the case of the National Stream Survey, DQO
#1 indicates that the target population should best
represent the resource at risk. In order to design
the NSS sampling plan, we have construed this to
mean that the target population should be located
in an area of historically low ANC surface water that
receives acid deposition, and in which streams (as
opposed to lakes) are the predominant surface water
resource. We further presume that the primary
resource of interest is sport fisheries, and that the
size of streams of interest should reflect the portion
of the stream continuum that provides the majority
of fish habitat for critical life stages.
These general criteria define a conceptual population
of interest which is not easily defined in explicit
terms. Such a definition must be quantified before
it lends itself to statistical sampling. Ideally, there
may be a single size range of streams that satisfies
these abstract criteria. In practice, however, even
with well-defined habitat characteristics, it has been
very difficult to arrive at an explicit definition of this
population so that it might be "targeted" for
sampling.
There are two aspects to the problem of explicitly
and quantitatively defining the conceptual popula-
tion of interest. The first deals with actual stream
locations and characteristics; the second with the
correspondence between these "on-the-ground"
characteristics and abstract representations of them.
A simple solution to the first aspect of the problem
is confounded by the regional, temporal, species, and
life stage differences in fish habitat requirements
(not to mention different ways of defining stream
"size"), which make a precise definition of the size
-------
of streams in such a population impossible. Once
satisfactorily defined, we must then contend with
bias and imprecision associated with our abstract
representation of the assemblage of streams on
maps, lists and remote imagery.
Given the difficulty in defining the conceptual
population of interest, the most expedient approach
for the National Stream Survey was to explicitly
define a target population in terms of blue-line
representation of streams on 1:250,000-scale
topographic maps, modified by certain site inclusion
criteria, and to proceed with an evaluation of whether
that target population is a reasonable representation
of the population of interest. The target population
is, therefore, our best attempt to make explicit this
conceptual population of interest. Its precise
definition was influenced by the expertise of local
fisheries biologists in a number of regions and was
tempered with our understanding of watershed
response to acid deposition. The decision was also
constrained by logistical considerations that influ-
enced the number of sites which could be sampled.
2.3.1 Selection of the Study Area
The Southern Blue Ridge was chosen for the location
of the Phase l-Pilot Survey for two reasons. The first
was the need for a geographically compact, physi-
ographically homogeneous area expected to contain
predominantly low ANC streams, and to provide a
range of logistical difficulties that would serve as
a reasonable test of the field sampling design. The
second was that the Southern Blue Ridge would
provide "delayed response" types of watersheds that
could be compared with "direct response"
watersheds associated with northeastern U.S. lakes
studied in the NLS (Galloway et al., 1983; U.S. EPA,
1985c; U.S. EPA, 1985d). Delayed response
watersheds usually contain thick soil mantles and
have geochemical properties that tend to neutralize
hydrogen ions added by acid deposition. Because of
this buffering effect, streams draining such
watersheds may not exhibit significant pH changes
for 10-100 years under present acid deposition rates.
Direct response watersheds generally have shallow
soils that exhibit little or no acid neutralizing capacity.
They are in virtual equilibrium with acid deposition
inputs. Streams draining such watersheds are
expected to exhibit pH and ANC depressions
relatively quickly following changes in acid deposi-
tion loading.
Figure 2-1 shows the location and boundaries of the
Southern Blue Ridge study area as determined from
physiographic maps of Fenneman (1946) and the
EPA alkalinity map for the region (Omernik and
Powers, 1983). A northern spur of the Southern Blue
Ridge Province (not shown in Figure 2-1) was
excluded from the survey in order to avoid sites with
driving times greater than two hours from the base
operations site. It should be noted that the NSS
Phase l-Pilot Survey area falls within the larger
subregion 3A of the ELS-Phase I, but does not share
the same boundaries. The geography of the area
(Figure 2-2) is composed of uniformly dissected
mountains with elevations ranging from 500-2000
m above mean sea level. The 27,000 km2 area is
drained by clearwater streams with dendritic
drainage patterns. Three major river valleys, the
Hiwassee, Little Tennessee, and French Broad
provide the major relief in the area as they drain
northward into the Tennessee River. All of the major
rivers in the area have been dammed, and reservoirs
of all sizes are common. Based on historical data,
streams draining watersheds < 70 km2, especially
at elevations > 750 m above mean sea level, exhibit
low acid neutralizing capacity, particularly in the
spring of the year (Silsbee and Larson, 1982; Talbot
and Elzerman, 1985).
Vegetation in the area is mostly Appalachian oak
forest with pockets of northern hardwoods. Land use
is mostly forest and ungrazed woodlands, with mixed
cropland and pasture in the valley bottoms. The
French Broad River Valley contains the cities of
Asheville and Hendersonville, and is moderately
urbanized or farmed in most places. The highlands
are sparsely settled, and Great Smoky Mountains
National Park occupies much of the northwestern
portion of the study area.
2.3.2 Stream Population of Interest
Identification of the target population of streams
(DQO #1) required consideration of the characteris-
tics of large versus small streams with respect to
the aquatic resource potentially at risk from acid
deposition. Provided that differences among fish
species are ignored, larger streams provide consid-
erably more fish habitat per unit stream length than
do very small streams, and thus are relatively more
important from a fishery resource standpoint.
However, in most regions of the U.S., very large
streams or rivers generally do not experience low
pH conditions, because natural and anthropogenic
buffering sources (e.g., agricultural liming or
discharge of treated wastewater) tend to buffer any
atmospherically-derived acidity once a river has
descended into populated valley bottoms.
At the other end of the size spectrum, low order,
high elevation streams within a given basin are
expected to exhibit lower pH and ANC than their
downstream counterparts and are therefore more
likely to serve as "early warning" indicators of acid
deposition impacts. These smaller streams, however.
8
-------
Figure Z-1. Location of the Southern Blue Ridge Phase I- Pilot Survey study area.
42° N
38° N
34° N
30° N
26° N
97° W
73° W
93° W
89° W
85° W
81°W
77° W
offer quantitatively less fish habitat, and for that
reason may not best represent the biological
resource at risk. While the impacts of acidification
on spawning and detritus processing in very small
headwater and intermittent streams should not be
discounted, it would be very difficult, given the
present state-of-the-science, to relate headwater
acidification to fish productivity further downstream.
Given this uncertainty, together with the frequently
observed pattern of maximum productivity and
species diversity of fish and invertebrates in mid-
order reaches (Platts, 1979; Vannote et a I.', 1980;
Minshall et al., 1983), the small to medium-size
stream category appeared to be the best target for
Phase I sampling from a biological resource
standpoint.
Rivers and streams at opposite ends of the size
spectrum also present special logistic and sampling
design difficulties. Larger rivers require substantially
different physical sampling (measurement) tech-
niques and equipment than are used on smaller
streams. The geographical point sampling frame that
was used for the statistical sampling design (Section
2.4) also works less effectively on watersheds of
drastically different sizes (Overton, 1987). On the
other hand, populations of very small streams are
poorly represented on maps, are often very difficult
to access physically, and their flow may dry up
entirely in some years.
A decision ultimately was made to target the NSS
on the population of medium-sized streams draining
-------
Figure 2-2. Geography of the NSS Phase l-Pilot Survey study area.
84
82°
36° -
35°
-36°
- 35°
watersheds of approximately 1 to 200 km2. Such
streams in the Southern Blue Ridge typically are less
than 1 m in depth and less than 10 m in width during
spring "baseflow" conditions. They typically repres-
ent streams of Strahler order 2 to 4, as determined
from 1:24,000-scale USGS topographic maps.
There was also some question of how to deal with
anthropogenic impacts that may mitigate or exac-
erbate the effects of acid deposition on streamwater
chemistry. Inclusion of streams affected by non-
atmospheric, non-point source pollutants would
decrease our ability to apply geochemical models that
depend on relatively "pristine" geochemistry to infer
acid deposition impacts. Conversely, restricting the
target population to pristine streams would preclude
making robust and meaningful population estimates
for ail streams of interest in a region. This dilemma
reflects the inability to optimize on both primary and
secondary objectives, as noted in Chapter 1. Based
on the primary Phase I objectives of identifying the
regional extent of all low pH and low ANC streams,
it was reasoned that only grossly polluted streams
(e.g., urban drainage ditches) should be excluded.
The effects of nonpoint source pollution on streams
otherwise affected by acid deposition are "part and
parcel" of the existing environmental conditions and
these streams were, therefore, included in popula-
tion estimates. Such streams may make poor
candidates for further study, however, and will likely
be excluded from field study in subsequent phases
of the survey.
Another problem arose in delineating the geographic
boundaries expected to contain the majority of low
ANC streams. A decision originally was made to
strictly adhere to the 400 peq L"1 ANC boundaries
10
-------
shown on the most recent (working) versions of the
U.S. EPA regional alkalinity maps. This decision
resulted in exclusion of two small "islands" of higher
(> 400 //eq L"1) ANC surface waters in the center
of the study area (Omernik and Powers, 1983).
However, local water quality experts indicated on
the basis of recent data that the maps were in error
in this regard, and these areas were subsequently
included in the Survey. Owing to similar uncertain-
ties in the accuracy of historical data bases,
subregional boundaries have been drawn with a
"broad brush" in the areas covered by the NSS in
the 1986 Phase I design (U.S. EPA 1985a; U.S. EPA
1985b).
2.4 Target Population Estimates
2.4.1 Methods for Identifying the Target
Population
In Section 2.3.2, we identified the conceptual
population of interest as all reaches that are not
grossly polluted, that drain watersheds of interme-
diate size, and that occur within certain relatively
homogeneous physiographic areas expected to
contain surface waters with acid neutralizing
capacity (ANC) predominantly less than 400/ueq L"1.
A probability sampling technique was used to choose
a set of such streams upon which to make field
measurements. The sampling plan began by iden-
tifying a sampling "frame." Sampling frames are
often "list frames" which literally list the units of
the target universe that are available for sampling.
The NLS employed a list frame: the names (or site
descriptors) of each lake of surface area greater than
4 hectares (ha) in each region of interest, as shown
on 1:250,000-scale topographic maps. The first step
in creating a sampling frame for the NSS was
determining how to specify sampling units: whether
to specify individual stream reaches for the frame
or to identify collections of reaches specified within
networks or watersheds. Individual reaches were
chosen over networks or watersheds, for reasons
discussed in Section 2.6.
Next, the reach units were identified. Alternatives
included blue-line representations on different
scales of topographic maps, remote imagery
collected by satellite or aircraft, and various
computerized data files originally constructed for
other purposes. The use of existing computerized
lists of streams was rejected, because they tend to
describe large streams and rivers. For example, the
U.S. EPA REACH data file (Olsen et al., 1981) is
comprised of only those reaches large enough to
appear on 1:500,000-scale topographic maps. At this
scale, a large number of smaller streams that could
be potentially sensitive to acid deposition but are
still large enough to offer abundant fish habitat are
not included (Sports Fishing Institute Bulletin, 1984).
Such streams, which appear on larger scale (e.g.,
1:250,000 and 1:24,000) maps, are generally too
small to be of interest to water supply managers
and therefore have not historically been represented
in computerized water resources data bases.
For the foregoing reasons, reaches represented on
larger-scale maps or by remote imagery were
deemed the best alternatives. Remote imagery was
rejected as being too costly and time-consuming for
constructing a frame of thousands of reaches. Of
the two applicable map scales (1:250,000 and
1:24,000), the former was chosen because it best
represented the conceptual population of interest.
An earlier survey (TIE, 1981) indicated that historical
fishery and aquatic resource values are more closely
associated with blue-line streams on the smaller
scale (1:2F.»0,000) maps. Although the relationship
varies from map to map, 1:250,000-scale maps
generally exclude the first and second order streams
that appear on the corresponding 1:24,000-scale
maps in the eastern U.S. While it would be possible
to identify streams to be included in the sampling
frame on 1:24,000-scale maps by excluding head-
water reaches according to some specified protocol,
the somewhat arbitrary way in which headwater
reaches are interpreted on such maps (Chorley and
Dale, 1972) makes any such representation equally
arbitrary. While there is no ideal way to identify the
true universe of streams of management interest,
identification of streams on 1:250,000-scale maps
appeared to be the most reasonable delineation of
the target population for the NSS sampling frame.
There is theoretically no reason why a list frame
could not have been created to identify the target
population of streams in the Phase l-Pilot Survey
area. However, we now estimate that the combined
mid-Atlantic and Southeast study areas planned for
1986 field work may contain well over 90,000
individual reaches. It thus would have required
approximately 4-5 workyears to create a quality-
assured, computerized list of such reaches for the
full Phase I Survey. The alternative was to design
a sampling procedure that estimates the number,
length, and other geographical characteristics of the
target population from a sample of reaches drawn
from that population. A subsample of streams also
must be selected for making physical and chemical
measurements.
2.4.2 First Stage of Sampling
The previous sections described in some detail the
decisions involved in developing an explicit definition
of the target population. The target population was
explicitly defined using blue-line representation of
streams on USGS topographic maps of 1:250,000-
11
-------
scale in combination with site inclusion criteria
described in Section 2.4.3. Throughout the
remainder of this report, this target population was
assumed to adequately represent the subset of the
total population of streams which is in a size range
of interest from the standpoint of the resource at
risk (sport fisheries). The lack of exact correspon-
dence is acknowledged and will be clarified in future
work. Additional data collection and analysis is
presently underway to identify the relationships
among these populations and to aid in interpreting
the target population estimates in the context of the
various regional conceptual populations of interest.
In the sampling design chosen here, the procedure
for sampling stream reaches from which to estimate
the structure of the target population was termed
the "first stage" of sampling. This activity utilized
geographic data only, and was not concerned with
describing the chemical conditions in the population.
The method of selecting stream reaches for making
physical and chemical measurements was called the
"second stage" of sampling, and is described in
Section 2.4.6.
In order to avoid the delay associated with construct-
ing a list frame for all stream reaches in an area,
it was decided instead to construct a "point frame."
A point frame employs the random placement of a
systematic sampling grid over a region to choose
study reaches. The probabilistic sample obtained is
more efficient at representing the spatial variability
of reaches in the region than a totally random sample
because this systematic sample is more evenly
distributed over the land area. The point frame used
in the Phase l-Pilot Survey was a grid of dots on
an acetate transparency placed at random on
1:250,000-scale topographic maps. To select
reaches corresponding to grid dots, a line was drawn
perpendicular to the elevation contours, proceeding
downslope from each grid point toward a stream
reach (Figure 2-3). A stream "reach" was included
in the first stage sample if it was the first reach
intersected by the line, and was defined as the
stream segment bounded by an upstream and a
downstream node. The downstream node was
determined by the first confluence with another
1 -.250,000 blue-line stream. The upstream node was
determined by either a similar upstream confluence,
or by the origin, as indicated on the 1:250,000-scale
topographic map. The example in Figure 2-3 shows
sampling frame points corresponding to a uniform
rectangular geographic grid with 8 mi between each
point. Point 98 in the figure results in the selection
of a blue-line non-headwater reach, while point 99
results in selection of a blue line headwater reach.
The direct drainage area of downstream reach
sampling nodes are identified as "ai." The area
draining into the upstream node of the non-
headwater reach is represented by "a2." Total
drainage to the upstream node of the headwater
reach is represented by "a3."
It is convenient, although not critical, that the grid
points in the point sampling frame be spaced with
a sufficiently low density that no two grid points could
correspond to the same reach. Such sampling
overlaps can be accommodated by the statistical
models used in data analysis and have no effect on
the validity of the population estimates (Overton,
1985, 1987). An 8-mile (approximately 13 km)
distance between points has thus far yielded an
appropriate grid density.
2.4.3 Site Inclusion Criteria ("Site Rules")
The reaches identified by the grid points are
categorized into various "target" or "non-target"
categories according to criteria discussed below. The
target population thus defined is identical to that
which might have been defined by a list frame. Unlike
the exhaustive population defined by a list frame,
however, the point frame (with inclusion criteria)
defines a probability sample of that population. The
inclusion criteria that were used for drawing the first
stage sample of Phase l-Pilot Survey sites are shown
in Table 2-1. Specific decision protocols provided by
the site inclusion criteria were used by the project
geographers to identify the resource at risk, as
addressed in general by DQO #1 (Section 2.2.3).
Each grid dot may lead to a non-target reach, a target
reach, or no reach at all. A grid dot identifies no
reach if the topographic fall line identifies a reach
wholly outside the study area boundary, or if the
dot identifies something other than a reach (e.g.,
a lake, reservoir, swamp, or closed basin). Non-target
reaches are excluded because some characteristic
puts them into a non-interest category. Boundary
reaches may penetrate sufficiently into suspected
high ANC regions external to the study area that
such reaches are unlikely to have low alkalinity over
much of their length. A reach was excluded if any
part of the blue line was outside the study area or
if > 25% of the drainage area defined by the
downstream reach node was outside the boundary.
Large rivers were excluded by the site rules for
reasons cited above (2.3.2). Sixty square miles was
subjectively chosen as an upper limit for total
drainage area. The use of watershed area to express
stream size was chosen because of the objective,
relatively precise way in which watershed areas are
determined from topographic maps, as compared to
stream order (Hughes and Omernik, 1981; 1983).
Reaches draining into or out of reservoirs also were
excluded from the Phase l-Pilot Survey. It was
reasoned that reservoir taifwaters could be domi-
12
-------
Figure 2-3. Representation of the point frame sampling procedure for NSS study reaches. The sampling frame points
correspond to a uniform rectangular geographic grid with 8 miles between each point. The lower left point
results in inclusion of the reach shown.
Non-Headwater Reach
Headwater Reach
13
-------
Table 2-1. NSS Phase l-Pilot Survey Site Inclusion Criteria
Non-Reach Grid Dots
Reach Out:
No Reach:
Non-Target Reaches
Boundary Reach:
Watershed Out:
Large River:
Reservoir Reach:
Target Reaches
Target Reach:
Topographic fall line yields a reach lying entirely outside the study area.
Dot identifies a lake, reservoir, or wetland.
Any part of the blue line crosses the study area boundary.
> 25% of drainage area outside study area.
Total drainage area above downstream node is > 60 mi2 (ca. 155 km2
Reach drains into or out of a reservoir.
Reach lying entirely inside study area boundary, not draining into or out of a reservoir, with a
watershed of < 60 mi2, at least 75% of which lies within the study area.
nated by unusual water quality characteristics
because of hypolimnetic processes in the reservoir.
Downstream nodes of streams draining into reser-
voirs were difficult to identify due to inaccurate map
representations and fluctuating reservoir operating
schedules.
2.4.4 First Stage Data
The first stage data base includes a listing for each
grid point, including:
1. Site identification code: a seven-digit code (e.g.,
2A08901) containing three fields indicating the
NSS Phase I subregion code (2A), the
1:250,000-scale map ID (089), and the grid dot
sequence (01). The last field has been increased
to 3 digits in Phase I.
2. Stream name: recorded from 1:250,000-scale,
or 1:24,000-scale map, where indicated.
3. Site inclusion criteria applicable to the grid
point (Section 2.4.3). For target reaches, certain
additional information was collected that
locates the reach geographically for sampling
and identification purposes, including:
4. County(ies) and state(s) in which the reach is
located.
5. State(s) in which the associated watershed is
located.
6. Administrative jurisdiction, if sites lie within
national or state parks or on military
reservations.
7. Miscellaneous comments.
The latter information was critical in the reconnais-
sance procedures described in Chapter 3.
In addition, data were collected for certain quantit-
ative geographic variables including:
8. The area of direct drainage, ai (Figure 2-3): This
is the portion of the watershed that drains
directly into the chosen reach, and also the area
within which a grid point will select this same
reach. This variable is very important, as it is
a measure of the probability of selecting the
reach, and is used in making all population
estimates. It was measured as accurately as
possible on the 1:24,000-scale maps.
9. Reach order, R: The number of reach origins
(headwaters) in the watershed above and
including the selected reach, as identified on
the 1:250,000-scale maps. This reach ordering
system is basically that of Shreve (1966), and
has certain topological advantages over the
more familiar Strahler or Horton ordering
systems.
10. For reaches of order higher than 1, the area,
a2, of the upstream watershed. The variable
aa is the area of the entire watershed that
produces the streamflow that enters the
selected reach at the upstream node, deter-
mined from surface topography on 1:24,000-
scale maps. (This value is zero for first order
reaches.)
11. Reach length, L, is the length of the selected
reach. Locations of the reach ends were
determined on 1:250,000-scale maps, but
measurement of L is made on 1:24,000-scale
maps as noted above.
12. Headwater drainage area, a3, being the area
draining into the upper node of each stream
(identical to &z for reaches with R > 1).
14
-------
13. In the field, it was not always possible or
desirable to collect stream samples exactly at
the coordinates indicated on the field maps.
In each case, actual sampling coordinates were
marked by field personnel on the field maps,
and new variables (a4, as, and La) analogous
to a-i, a2, and L, were created, based on actual
field sampling locations.
In all cases, measurements were made based on
triplicate measurements with a Model 1250 Numon-
ics planimeter with a resolution of ± 0.010 seconds
and an accuracy of ± 0.020 seconds.
2.4.5 Population Estimates
The sampling design outlined above produced a
probability sample of 115 reaches with expected
probability of inclusion proportional to the direct
drainage area, a-i. of each reach. This design has
two advantages over a list frame, from which reaches
are chosen randomly:
1. The reaches are approximately uniformly
distributed in space, so that any intraregional
geographic chemical patterns should become
apparent.
2. It is less time-consuming than constructing a
list frame because reach attributes need only
be measured on the selected sample reaches.
Many attributes of the target universe of stream
reaches can be estimated from the first stage sample
(or second stage sample). The basic estimator is of
the form:
Z indicates summation over the number of sample
s reaches, s, in the stratum of interest; and
w is a weighting factor (d/ai) which is inversely
proportional to the inclusion probabiiity for each
particular stream reach.
The estimation of some attribute of interest in the
target universe is accomplished by employing
Equation 2.1. When an estimation of that attribute
is desired for a subset (stratum or population) of the
target universe, summation is restricted to the
pertinent subset. This is accomplished by setting of
the value of the indicator variable z to 1 if the sample
reach is in the population or to 0 if it is not. For
example, to estimate total stream miles by category
of watershed size, one can sum over only those
reaches belonging to that watershed size category.
Similarly, to estimate the total stream length which
lies in some criterion region (e.g., those stream miles
with values of ANC within some specified range),
we only sum over sample reaches having ANC values
in that range.
The number of populations or classes of reaches that
can be identified by physical or chemical charac-
teristics is essentially unlimited. The following
examples are chosen only to illustrate the possibil-
ities of the estimation procedure. The total number
of reaches (N), within any subset, z, of the target
universe (such as those reaches with ANC < 100
/ueq L"1) can be determined as the sum of interest
reaches, each multiplied by its individual probability
weighting factor:
Ty = d Z
s
= Z zyd/ai = Z zyw
s s
[2.1]
where:
y is any reach attribute of interest (e.g., length);
Ty is an estimate of the sum of that attribute over
the target universe (e.g., combined stream length);
z is an indicator variable (0, 1) for a particular class
of interest (e.g., all reaches with 81 + aa < 10 mi2);
81 is the area of direct drainage-of the sample reach,
as defined above;
d is the areal density of points in the geographic
grid used to construct the point frame (the NSS
Pilot uses an 8 mile grid resulting, respectively,
in 64 mi2 and 128 mi2 per point for Stage I and
Stage II sampling);
Nz = d I zy/ai = Z
s s
= I zw
s
[2.2]
A
Similarly, Az, the total area of watershed directly
drained by the reaches in subset z of the target
universe can be estimated by:
Az = d I zy/ai = I
s s
= Z zd
s
[2.3]
An estimate of the total length (Lz) of reaches in
subset z can be calculated as:
Lz =
d Z
s
= Z
s
= Z zLw
s
[2.4]
Examples of other types of classifications (stratifi-
cations) that were found to be useful in the Phase
l-Pilot Survey analysis included:
1. Categories of watershed size:
15
-------
a. Class A: reaches with 0 < (a1+aa) ^ 5 mi2
(13km2).
b. Class B: reaches with 5 < (ai+aa) < 15 mi2
(39 km2).
c. Class C: reaches with 15 < (ai+a2) < 60
mi2 (155 km2).
2. Shreve reach order (1:250,000 blue lines):
a. R = 1 (headwaters)
b. R = 2
c. R>2
The basic estimator can be generalized to yield the
following formula for a reach attribute:
Ty = Z yw,
[2.5]
where summation is over the sample data that are
in the subpopulation of interest. Variances are
calculated for each parameter of interest according
to the equation:
n
V(Ty) = I y2[w(w - 1)] + I Z
s i
where:
-Wj.i)
[2.6]
[2.7]
resource constraints of the project. Previous
experience in Phase I of the Eastern Lake Survey
had shown a second stage sample size of approx-
imately 50 per stratum to be satisfactory. It was also
desired that the second stage sites be well dispersed
geographically within the region so that any
correlations of aquatic chemistry with geologic type
or acid deposition loading could be detected.
A systematic random sample was chosen as the best
means to draw such a sample. Every other grid dot
was drawn in the second stage sample, beginning
with a random start, and without regard to the site
rules associated with each dot (i.e., non-reach and
non-target dots were included). The site rules then
were applied, and the resulting second stage sample
was found to contain 54 target reaches. The locations
of these sites are shown in Figure 2-4, along with
the last four digits of the site identification code.
Geographic site data for these reaches are provided
in Appendix B (Table B.1).
Population estimates and their associated variances
were calculated as in the first stage sample, although
n was smaller in the second stage sample. A
systematic random subsample from the first stage
sample retains the characteristic of non-uniform
inclusion probabilities of the reaches in the first stage
sample. Although the non-uniform inclusion proba-
bilities can be accommodated by the sampling
statistics, it is critical that future users not treat the
sample as if the inclusion probabilities are equal.
That is, population statistics associated with the
sample should not be calculated as unweighted
medians, means, and standard deviations.
if i and j are from the same stratum, and
j.i = Wj
[2.8]
if i and j are from different strata (Overton, 1985).
The effective sample size, n, is the number of grid
points that fall in the study area, and includes non-
target reaches and points that do not lead to a reach.
For any subpopulation, the formulae are identical
and summation is made over the sample data from
that exact subpopulation. A detailed discussion of
the variance estimation procedure is presented by
Overton (1987).
2.4.6 Second Stage of Sampling
The second stage of sampling was designed to
subsample the 115 target reaches in the first stage
sample to obtain a reduced number of reaches for
making chemical measurements within the time and
2.4.7 Target Population Geographic Estimates
Examples of the types of estimates that can be made
for some geographical attributes of the target
population are presented in Table 2-2. Estimates of
the numbers, length (L), direct drainage areas (ai),
and discharge indices of target reaches in the
Southern Blue Ridge are shown, along with the
associated standard errors of the estimates. The
percentages of stream reaches, length, and water
surface areas represented by the subpopulations of
drainage area and order categories, as described in
Section 2.4.5, are also shown. Interpretation of the
total discharge index and the significance of the
various geographic estimates will be discussed in
Chapter 5 in the context of reach chemistry.
Based on the first stage sample, the target population
is estimated to contain 2,156 reaches with a
combined length of 9,508 km (5,908 mi). The streams
are estimated to directly drain 19,062 km2 (7,360
mi2) or 7,360/10,501 = 70% of the study area. The
remainder of the area drains directly into reservoirs.
16
-------
Figure 2-4. NSS Phase l-Pilot Survey study area showing second stage (II) probability sites and special interest reaches.
Numbers represent the last four digits of the corresponding site code.
36° N
35° N •
34° N
Stage II Sites
Special Interest Sites
! ^805 8806
\ ^804 • 9 .8903
8808
8905
85° W
84° W
83° W
82° W
large rivers, or into boundary reaches. Sixty-six
percent of the target reaches are portrayed as
headwaters (R = 1) on 1:250,000-scale maps, and
50.6% are estimated to drain watersheds < 5 mi2
(13 km2). The percentages for headwater reaches
are somewhat higher (ca. 74%) if the estimates are
based on length or drainage area, thus indicating
that the headwater reaches typically are longer and
have larger direct drainages than the stream
segments lower in the watersheds (R > 1).
The same types of estimates can be made from the
smaller (n = 54) second stage sample. Differences
in the estimates (e.g., 8,963 versus 9,508 kilometers)
between the two samples are generally small for
the entire population, and for headwater reaches
(e.g., 6,714 versus 7,047 kilometers). Standard
errors based on the second stage sample are higher,
due to the smaller sample size. The estimates diverge
the most for relatively small subsets of the sample
(e.g., second order reaches). The differences in the
17
-------
Table 2-2. Geographic Attribute Estimates and Standard Errors for the NSS Phase I-Pilot Survey Target Populations Based
on Stage I and Stage II Samples
Attribute Description
Stage I"
Est.
S.E.
Stage II"
%c
Est.
S.E.
%c
Number of Target Reaches 2,155.6 265.0 2,020.9 326.7
Headwater Reaches 1,422.4 180.5 66.0 1,432.7 296.1 70.9
Second Order Reaches 372.5 199.1 17.3 79.1 53.7 3.9
Remaining Reaches 360.7 106.0 16.7 509.0 191.9 25.2
Class A Reaches 1,090.9 185.9 50.6 1,120.9 302.8 55.5
Class B Reaches 698.5 204.9 32.4 449.9 119.7 22.3
Class C Reaches 366.1 97.4 17.0 450.1 170.2 22.3
Direct Watershed Area (Sq. km) 19,062.4 954.7 17,902.1 1,417.7
Headwater Reaches 14,089.6 1,019.5 73.9 13,260.8 1,477.6 74.1
Second Order Reaches 1,989.1 525.6 10.4 994.6 550.1 5.6
Remaining Reaches 2,983.7 628.1 15.7 3,646.7 997.8 20.4
Reach Length, Total (km) 9,508.0 645.0 8.963.2 952.7
Headwater Reaches 7,046.5 636.7 74.1 6,714.1 938.3 74.9
Second Order Reaches 1,054.3 315.4 11.1 482.5 296.4 5.4
Remaining Reaches 1,407.2 339.0 14.8 1,766.6 553.7 19.7
Total Discharge Index (Sq. km) 47,286.6 7,368.8 51,123.7 12,487.4
Headwater Reaches
Second Order Reaches
Remaining Reaches
Class A Reaches
Class B Reaches
Class C Reaches
14,089.6
9,471.6
23,725.4
6,464.6
14,981.2
25,840.8
1,019.5
3,608.8
6,913.3
890.7
3,715.0
6,845.0
29.8
20.0
50.2
13.7
31.7
54.6
13,260.8
2,362.6
35,500.3
5,967.4
10,038.7
35,117.7
1,477.6
1,317.5
12,864.3
1,240.5
3,116.8
12,613.9
25.9
4.6
69.4
11.7
19.6
68.7
"n=115
bn = 54
°Percentages refer to the estimated number (length, areas) of streams in the correpsonding interest categories (see Section 2.4.5).
two estimates are discussed further in the context
of the chemical distributions in Chapter 5.
2.4.8 Special Interest Reaches
In addition to the reaches in the probability sample,
seven "special interest" sites also were visited
during the survey. Four of these sites were being
monitored for episodic pH depressions in conjunction
with another NAPAP Task Group E project (Olem,
1984), and two sites represented long-term mon-
itoring sites on control watersheds at the Coweeta
Hydrologic Laboratory (gauges 8 and 36). Although
the data gathered at these sites were not analyzed
as part of the probability sample, they ultimately will
serve as part of the data needed to establish the
representativeness of these sites with respect to the
Southern Blue Ridge target population. Locations of
the special interest sites, along with their corres-
ponding site codes, also are shown in Figure 2-4.
2.5 Third Stage of Sampling
The sampling design establishes the physical and
chemical measurements that are made on each.
second stage reach, and when and where to take
them in order to best characterize the reach. These
decisions were based on the expected temporal and
spatial variability in the chemical concentrations
within each reach and on the potential utility of the
field information relative to the project objectives to:
1. Estimate the current population distribution of
streams at risk (e.g., having low pH or high
concentrations of toxic aluminum species at
times when sensitive life stages of fish are
present).
2. Estimate the population distribution of streams
potentially at risk in the future (e.g., having low
ANC during periods when sensitive biota are
present).
18
-------
3. Classify reaches into representative "types"for
future intensive studies.
The first two objectives require that measurements
be taken during geochemically and ecologically
relevant periods of time, i.e., during seasons of low
pH, highest relative proportion of "recent" hydrologic
inputs to the system, and presence of sensitive biota.
From a population estimation standpoint, it also
would be desirable to minimize within-stream
sampling variance in order to reduce the error
bounds on the population estimates.
variables are to be measured, where on a reach,
and how frequently. The first issue relates to the
ecological and geochemical objectives of interest,
including ecological effects, mechanisms of acid
deposition processing within watersheds, and
geochemical classification. The spatial issue relates
to the spatial variability known or expected to occur
within the sampling unit (the reach). The final issue
is a function of the expected temporal variance of
the chemical variables at a site, and the desired
precision or robustness of the population descrip-
tions and classification. These issues are addressed
in the following sections.
The last objective requires that a balance be struck
between measuring chemistry at a time when
within-stream variation is minimized and when
between-stream variability is maximized, in order to
provide classes that are both distinct and robust. Any
such classification should separate streams into
categories that ultimately represent the most
important ecological and geochemical types with
respect to the first two objectives. Ideally, a sampling.
design would simultaneously meet all three of these
objectives; in reality, it probably cannot.
In addition to specifying the sampling season or
seasons, it also is necessary to specify which
2.5.1 Variables Measured
Data quality objective #3 (Section 2.2.3) specifies
that sufficient variables be measured so that one
can determine: (1) the chemical and physical quality
of the streams with respect to fish habitat; and (2)
the geochemical nature of the waters with respect
to past and future susceptibility to acid deposition.
It was not cost effective to measure all possible
variables on a large number of streams, but it was
necessary to measure the critical ones. Table 2-3
lists the measurements made on second stage
sample reaches, except for the geographic variables
noted above.
Table 2-3. Variables Measured During the NSS Phase I-Pilot Survey
Site Data
In situ Measurements
Laboratory Measurements
gage height (stage)
stream width
stream depth
land use
bank vegetation
stream substrate
cloud cover
weather conditions
pH (open head space)"
pH (closed head space)"
temperature
specific conductance
dissolved oxygen
pH (closed head space)"*"
pH (equilibrated, 300 ppm CO2)
DIG"
DOC
true color8
turbidity8
conductivity
ANC
BNC
Aluminum (total)
Aluminum (MIBK extractable)"
Aluminum (non-exchangeable)"
suspended solids
calcium
magnesium
potassium
sodium
nitrate
sulfate
chloride
fluoride
silica
ammonium ion
iron ^
manganese
total phosphorus
"In open headspace pH determinations, samples were exposed to the atmosphere during collection and measurement; in closed
headspace determinations they were not. See Section 3.3.2 for more detailed information.
"Samples prepared at field lab, then measured at analytical lab.
19
-------
The site information recorded at each sampling
location was primarily meant to assist in the initial
interpretation of physical/chemical data from each
site, and to aid in locating the site in future studies.
This site information ("Site Data" in Table 2-3) was
not quality-assured and, although it is recorded in
an NSS data file, it should not be used to draw
quantitative inferences about the other chemical or
physical data. For example, it would be inappropriate
to regress turbidity or temperature against substrate
type or stream bank vegetation, even though
reasonable relationships may exist. It would be
useful, however, if a site with high turbidity could
be determined to be bordered by unvegetated stream
banks, to have a silty bottom, or to have experienced
a rainy period prior to sampling.
The physical and chemical variables may have many
interpretations, depending on the way in which they
are used. Some variables are of primary interest with
respect to the immediate NSS objectives (e.g., pH
and ANC). Other variables are important in inter-
preting the primary variable data (e.g., DOC, color,
and fluoride are useful in understanding the
speciation of aluminum). Other variables such as
nitrate, sulfate, and DOC are needed to describe the
ionic composition of waters, and some may be useful
indicators of non-atmospheric pollution (e.g.,
chloride, total phosphorus, and ammonium-
nitrogen). Finally, some variables may provide clues
to the geochemical processes controlling water
chemistry in a region, and also may be useful in
classification of stream reaches for further study
(e.g., silica, sodium, potassium, or calcium). Com-
plete chemical analysis for all major ions is needed
for conducting verification checks on the accuracy
of chemical analyses on the basis of cation/anion
balances and conductivity checks (see Chapter 4).
Brief descriptions of the chemical variables mea-
sured during the Phase l-Pilot Survey are presented
below:
1. pH: The pH of a stream is a direct indication
of free hydrogen ion activity. The pH is an
important geochemical constituent and affects
toxicity through its effects on fish physiology
and the speciation of toxic metals such as
aluminum.
2. Base Neutralizing Capacity (BNC): The BNC is
a measure of acids in water including both
terrestrial and atmospheric mineral acids,
carbon dioxide, and organic acids associated
with decomposition of plants and detritus. This
term is used interchangeably with acidity
throughout this report.
3. Acid Neutralizing Capacity (ANC): ANC is a
measure of all bases and is an indication of
buffering capacity. Alkalinity is a more approp-
riate term if the ANC is primarily controlled by
the inorganic carbonate system. Alkalinity is
used synonymously with ANC throughout this
report.
4. Specific Conductance: The specific conduc-
tance of stream water is a measure of the
resistance of the water to electrical current.
Because resistance to electron flow is inversely
proportional to the concentration of ions in
solution, specific conductance can be used to
check the overall accuracy of ion analyses.
5. True Color: True color is a potential indicator
of naturally occurring organic protolytes and
DOC. Substances that impart color may also
be important natural chelators of aluminum
and other metals.
6. Dissolved Inorganic Carbon (DIC): In carbonate
systems, a measure of DIC (and either pH or
ANC) can be used to describe the equilibrium
distribution of carbonate solutes, and deter-
mine whether the solution is saturated with
respect to atmospheric COa.
7. Dissolved Organic Carbon (DOC): DOC is an
important source of energy for stream metab-
olism, but also provides an indication of the
presence of natural organic acids which can
influence pH. DOC is also a natural chelator
of aluminum and other trace metals.
8. Dissolved ions (Na+, K+, Ca2+, Mg2+, Fe3+, Mn2+,
NH4+, F", Cl~, S042~, and NOs ): These constit-
uents are measured in order to chemically
characterize streams and calculate ion
balances.
9. Total Extractable Aluminum: Total extractable
aluminum is an estimate of dissolved alum-
inum and includes most mononuclear alum-
inum species. Aluminum is considered to be
highly toxic, especially to fish. It was further
fractionated into operationally-defined inor-
ganic and less-toxic organic monomeric forms
based on affinity for a cation exchange resin.
10. Total Aluminum: Total aluminum is associated
with the weathering rate of soils in a
watershed, and is often associated with high
flow in streams.
11. Dissolved Silica (Si02): Dissolved silica is a
potentially important factor in identifying
mineral weathering reactions and source
materials in poorly buffered streams.
20
-------
12. Total Phosphorus: This is an indicator of
potentially available nutrients for periphyton
productivity and may provide a check on
unobserved pollutant sources.
13. Turbidity: Turbidity is a measure of suspended
material present in water at low concentrations
and is often a useful indicator of increased
discharge.
14. Total Non-Filterable Residue (Suspended
Solids): This parameter is a useful estimate of
the amount of paniculate material entering the
stream and potentially capable of interacting
with chemical species in the water. It also is
often a useful indicator of episodes resulting
from storms or snowmelt.
Those variables in Table 2-3 that are least stable
in a sample following collection due to temperature
changes or gas exchange, and for which portable
instrumentation was available, were measured in
situ or at streamside. A second set of variables that
are less labile and/or for which portable field
instrumentation was inadequate (e.g., aluminum
fractionation) or for which apparatus was unavail-
able (e.g., dissolved inorganic carbon) are indicated
in Table 2-3. These variables were measured within
12 hours of sample collection at a specially equipped
mobile analytical laboratory that was centrally-
located in the study area (Sylva, North Carolina).
Suspended solids were measured at a local contract
laboratory. The remaining variables were stabilized
(if necessary) and samples were sent to a central
contract analytical laboratory (New York State
Department of Health, Albany) for analysis. The
specific analytical procedures for these variables are
summarized in Table 2-4. Sample collection and
handling protocols are described in Chapter 3.
2.5.2 Sampling Season
In order to determine the optimum seasonal
sampling window, a literature search was followed
by meetings with hydrologists, biogeochemists, and
fishery experts in the study area. Although many
of the data discussed at these meetings were still
being prepared for publication, the following
Table 2-4. Chemical Variables and Corresponding Measurement Methods for the National Stream Survey
Parameter Method"
1. Acidity (BNC)
2. Alkalinity (ANC)
3. Aluminum, total
4. Aluminum, extractable
5. Aluminum, non-exchangeable
6. Ammonium, dissolved
7. Calcium, dissolved
8. Chloride, dissolved
9. Fluoride, dissolved-tola I
10. Inorganic carbon, dissolved (DIG)
11. Iron, dissolved
12. Magnesium, dissolved
13. Manganese, dissolved
14. Nitrate, dissolved
15. Organic carbon, dissolved (DOC)
16. pH
17. Potassium, dissolved
18. Silica, dissolved
19. Sodium, dissolved
20. Sulfate, dissolved
21. Specific conductance
22. Phosphorus, total
Titration with Gran plot
Titration with Gran plot
EPAJVIethod 202.2—AAS (furnace)
Extraction with 8-hydroxyquinolme into
MIBK followed by AAS (furnace)
Cation exchange, followed by extraction with
8-hydroxyquinoline into MIBK followed by
AAS (furnace)
EPA Method 350.1
EPA Method 215.1—AAS (flame)
Ion chromatography
Ion selective electrode
Instrumental (Similar to DOC)
EPA Method 236.1—AAS (furnace)
EPA Method 242.1—AAS (flame)
EPA Method 243 1—AAS (flame)
Ion chromatography
EPA Method 415.2
pH electrode and meter
EPA Method 258.1—AAS (flame)
USGS Method I-2700-78
EPA Method 273-1—AAS (flame)
Ion chromatography
EPA Method 120.1
USGS Method I-4600-78 or Modified
USGS Method
"AAS methods are taken from U.S. EPA (1983). Laboratories that have ICP instrumentation may use EPA Method 200.7, reproduced
in Appendix A of Hillman et al. (1986a) for determining Ca, Fe, Mg, and Mn, providing they can demonstrate the specified detection
limits. If the ICP instrumentation is not able to meet the required detection limits, it may still be used to analyze samples which
contain the analytes at concentrations greater than 10 times the ICP detection limit. Other samples must be analyzed by furnace
or flame AAS.
21
-------
generalizations were based on quantitative data
when available, and on expert opinion where data
were lacking:
1. Alkalinity and pH are typically low during March
15-May 15 of normal years in the region.
2. Low pH episodes may also occur in the fall or
winter in streams which also experience such
episodes in the spring. However, the fall
episodes do not appear to be "worse" than
those occurring in the spring, and some
streams with low spring pH may not exhibit
such conditions in the fall.
3. Chemistry may be highly variable (e.g., from
hours to days) during the spring in streams in
the area. Temporal variability during other
times of the year is usually (although not
always) lower.
4. Studies have indicated that all life stages of
fish are not equally susceptible to acidity and
other chemical parameters that accompany low
pH episodes in surface waters. Some of these
studies involved observations of acidic lakes
and streams in which viable eggs were found
together with older age classes of fish that
appeared to be spawning successfully, but in
which young age classes were absent (e.g,
Beamish et al., 1975; Muniz and Leivestad,
1980; Kelso and Gunn, 1982; Gunn and Keller,
1984; Sharpe et al., 1984). Such a population
structure suggests more pronounced effects of
acidity on larval fish than on egg hatching or
adult survival. These field observations are in
agreement with laboratory bioassays that also
indicate increased sensitivity of fry to low pH
conditions (Schofield, 1976; Haines, 1981). Fry
of the most important sport fisheries are
present in the study area during the March 15-
May 15 period. Fry of some trout (Salmo)
populations may also be present at other times
of the year.
5. Physical access to most sites during spring was
not expected to present significant problems.
Winter access is difficult in places because of
seasonal closure of unimproved roads and icy
conditions on intermediate roads and
highways.
Based on the above considerations, two sampling
periods were chosen for the Phase l-Pilot Survey:
March 17-May 30 (during which 3 samples were
collected from each reach at three-week intervals)
and July 1-July 17 (during which each reach was
sampled once). No attempt was made to either target
or avoid sampling during rainstorm events. The first
sampling period coincided with periods of highest
biogeochemical interest (i.e., low seasonal pH and
the presence of sensitive fry), but in which temporal
variability was potentially high enough to make
robust population descriptions and reach classifica-
tion impossible. The second sampling period was
investigated to determine whether the presumably
more stable summer "baseflow" chemistry could be
used to provide an "index" of the spring conditions.
2.5.3 Sampling Locations on Each Reach
Alternative locations for sampling on any particular
reach included the upstream node, centroid, and
downstream nodes. The downstream node was
selected in order to provide not only an index of the
chemistry over the entire reach, but also an
integrated index of water draining the watersheds
above the reach (a-i + aa). While the latter concept
is fundamental to most watershed studies, it was
anticipated that representing the chemistry of the
entire reach by the chemistry at the downstream
node would overestimate pH and ANC. It was
subsequently decided that the upstream node of each
reach could be added to the sampling itinerary
without jeopardizing the primary design, and 21
upstream sites were added to the itinerary during
the final spring sample. The chemistry of all
probability sample reaches was measured at both
upstream and downstream nodes during the summer
sampling period. Special interest sites were sampled
at the locations specified by the permanent
investigators.
2.6 The Watershed Alternative to the
Reach Frame
Almost everyone associated with the NSS was at
one time attracted to the idea of employing
watersheds, rather than reaches, as the statistical
sampling unit. This proclivity no doubt arose in
response to the proven utility of watersheds as units
with convenient external boundaries for constructing
mass balances useful in biogeochemical process
research. The ordering of stream networks within
watersheds also is useful in studying "continuum
processes" with a strong hierarchical gradient. As
sampling units of a large target population, however,
watersheds offer some critical drawbacks, given the
NSS primary objectives.
The principal problems with the watershed approach
have to do with maintaining a large number of
sampling units for making population estimates, and
the amount of data needed to describe the extent
of chemical conditions within a sampling unit. The
chemistry of a reach can be approximated by
measurements taken at two nodes. The chemistry
of the simplest topological network (three reaches)
22
-------
requires measurements at four nodes, resulting in
a three-row matrix of chemical variable scalars to
specify the water chemistry. Under an equal effort
sampling constraint, this reduces the sample size
from 50 to 25 units, while greatly increasing the
complexity of each unit for classification purposes.
Increasing the network to Shreve order 3 (5 reaches)
requires 6 measurements thus reducing n to 16.
This problem is further exacerbated by the fact that
areas do not partition uniquely into small
watersheds. If all watersheds smaller than 60 mi2
are included, a region is partitioned into many
different orders of watersheds, some large mainstem
sideslopes, and sideslopes of adjacent areas. Small
watersheds draining into large mainstems will have
different chemical patterns than their equal ordered
counterparts draining into intermediate-ordered
streams at higher elevations. There is no "typical"
order 3 (or any other) watershed in an area. These
characteristics have rendered artificial, if not
completely unworkable or inappropriate, any
watershed-based sampling designs thus far
considered.
In fact, the reach-based sampling plan can be used
to construct "artificial" watersheds, based on the
reach orders sampled in each subregion. Population
estimates can be made for "headwater" reaches,
second order reaches, (e.g., Table 2-3) and so on,
provided enough reaches have been sampled in that
category to provide meaningful estimates. Of course,
these estimates are valid for all reaches of order
R, regardless of where they occur, but they cannot
be used to estimate the chemical pattern in any
particular watershed. Nonetheless, the reach-based
sampling design appeared to be the best way of
meeting all of the NSS primary Phase I objectives.
23
-------
3. Field Operations
3.1 Introduction
Several activities were performed before and during
NSS Phase i-Pilot Survey field work to assure that
samples were collected, processed, and analyzed in
a consistent, safe, and timely manner. Protocols for
collecting samples and making in situ or streamside
measurements were developed prior to field
sampling and were documented in a field operations
manual (unpublished).Equipment and supplies for
field sampling teams and the field laboratory were
procured and evaluated, and the necessary person-
nel were hired and trained. Potential field station
sites were visited and evaluated, and a reconnais-
sance program was established for acquiring
pertinent access information on each stream reach
selected for sampling. Safety protocols for sampling
sites in rugged, unfamiliar terrain were developed.
Protocols were occasionally modified during the
course of field operations, and alternative method-
ologies and equipment were evaluated. In all cases,
appropriate QA/QC protocols were developed for
each procedure.
Field sampling and iaboratory operations were
conducted between March 1 and July 16,1985. Over
the course of the study, 339 routine samples were
collected from 61 stream reaches (Table 3-1) and
a total of 724 samples were analyzed during the
project. A detailed description of the field opera-
tions,including planning and preparation, reconnais-
sance, field sampling, and field laboratory activities
can be found in Knapp et at. (1987). These activities
are briefly summarized in the following sections.
3.2 Preparation for Fieid Operations
3.2.1 Protocol Development
While most analytical protocols for the field and
contract analytical laboratories were adopted from
the Eastern Lake Survey (ELS) component of the
NSWS, much of the collection and measurement
apparatus used in the ELS was unsuitable for use
in streams, primarily because of large size and
limited portability. Consequently, new field equip-
ment was procured and tested, and protocols were
written for the new procedures. In addition, a
protocol was developed for fractionation of the total
extractable aluminum (monomeric) aliquot in the
field laboratory. This protocol was based on the
methodology of Driscoll (1984), and involved passing
filtered samples through a cation exchange column
prior to complexing the nonexchangeable fraction
with 8-hydroxyquinoline and extracting into methyl
isobutyl ketone (MIBK). A protocol for measuring
conductivity in the field laboratory also was
developed late in the Survey.
3.2.2 Training Programs
Training programs for field sampling and field
laboratory personnel were conducted in Las Vegas
over a five-day period before field work was begun.
The field training program was designed to famil-
iarize personnel with the objectives and research
design of the NSS, sampling and analytical protocols,
site reconnaissance, equipment troubleshooting,
and field safety. Additional training, conducted in
the field, included basic stream hydrology and site
Tabie 3-1. Summary of Routine Samples Collected During the NSS Phase (-Pilot Survey
Sampling
Interval
SPO"
SP1
SP2
SP3
SU
Dates
3/17
3/20
4/03
4/17
6/30
-3/19
-4/02
-4/16
-4/30
-7/16
Upstream
0
0
0
23
54
Downstream
18
54
54
54
54
Special
Interest
Reaches
0
7
7
7
7
"These samples represent a three-day training run; they are included in the data base but were not used for population estimates.
24
-------
coordination responsibilities. Field laboratory
personnel underwent a five-day training course that
covered all aspects of field laboratory operations,
safety, and quality assurance.
3.2.3 Field Station Site Selection and
Reconnaissance
Potential sites from which to conduct Phase l-Pilot
Survey field operations were visited in early 1985
and evaluated on their ability to support field
sampling and field laboratory operations. The field
laboratory was eventually located at Southwest
Technical College in Sylva, North Carolina (Figure
2-2). This location also served as a base for field
sampling activities. Field station personnel and a
local communications center were housed m
Cullowhee, North Carolina, approximately five miles
from the field laboratory.
After stream reaches had been identified on USGS
1:24,000-scale topographic maps, it was necessary
to assemble access information for each sampling-
site. This was accomplished through telephone
contact with "local cooperators" who were familiar
with areas where reaches were located, and included
personnel from federal, state and local agencies.
Information on ease of access, driving or hiking
times, and names of landowners to be contacted for
access permission were obtained from the cooper-
ators, and dossiers were compiled for each sampling
site. Each dossier contained maps, telephone
numbers of local cooperators, landowners, emer-
gency contacts, and information on travel routes and
site access. The dossiers were updated as new
information was gathered in the field.
Each site was visited by field sampling personnel
before sampling commenced. This field reconnais-
sance visit served to verify access'information and
to obtain access permission if necessary. Descriptive
information on site characteristics in the immediate
area was recorded on a standardized field form (Form
7, Figure 3-1), and the area was photographed to
aid in describing the site and locating it on
subsequent visits. A hydrologic staff gauge was
installed and an initial reading was taken at each
site; the cross-sectional area at the gauge was also
measured.
3.3 Field Operations
3.3.1 Field Station Operations
The Phase l-Pilot Survey was conducted in 1985
during two separate periods: a spring sampling
period (March 17-April 30), and a summer sampling
period (June 30-July 16). The field station in Sylva,
North Carolina, was staffed by 11 people: a site
coordinator, six field samplers, a laboratory super-
visor, and three analysts. The site coordinator was
responsible for the overall operation of the field
station (Figure 3-2). Duties included devising
sampling itineraries, organizing each day's samples
into a batch for processing, shipping preserved
samples to the contract analytical laboratory, and
shipping data forms to the data management center
at Oak Ridge National Laboratory (ORNL) and the
QA support group at the EPA Environmental
Monitoring Systems Laboratory at Las Vegas (EMSL-
LV). The site coordinator also filed a daily operations
report with the NSS communications center in Las
3.3.2 Field Sampling Operations
Samples were collected and field measurements
were made by two-person teams who accessed
stream sites by four-wheel drive vehicle, hiking,
boating, or horseback. Each team was responsible
for sampling 20-21 stream reaches during each two-
week sampling interval, and visited one to three sites
on each working day. The activities conducted at
each site are summarized in Figure 3-3.
Water samples (termed "routine" samples) were
collected from each stream by pumping water
through 1/4-ineh surgical grade Tygon tubing held
in the center of the stream cross-section using a
6-foot sampling boom. Samples were pumped using
portable peristaltic pumps driven by gel/cell
batteries, each sample representing an integrated
10-minute sample of the streamflow. Samples were
pumped into containers that had been aeidrwashed,
rinsed, and quality-assured by the supplier. A bulk
sample was collected into a four-liter Cubitainer.
Three 60 ml polypropylene syringes equipped with
gastight valves were filled in such a way that the
samples were not exposed to the atmosphere. An
aliquot for total suspended solids analysis was
collected into a 500-ml amber high-density polyeth-
ylene wide-mouth bottle. Each container was rinsed
three times with sample water prior to filling with
the routine sample, and new Tygon tubing was used
at each site.
Water samples were transported from streamside
in portable soft coolers containing chemical refrig-
erant packs. Cooler temperatures were checked
periodically and found to be approximately 4°C.
When field crews arrived at a vehicle, samples were
transferred to rigid insulated containers containing
chemical refrigerant packs. Samples were held in
these containers at approximately 4°C until they
arrived at the field laboratory, usually within 10
hours of collection.
Two types of quality assurance samples were
collected each day. Each team collected a "field
25
-------
Figure 3-1. NSS Data Form 7: Watershed Characteristics.
NATIONAL SURFACE WATER SURVEY
WATERSHED CHARACTERISTICS
FORM?
D D M M M Y
DATE
COUNTY
STATE
STREAM ID
STREAM NAME
LATITUDE' i i i 1°. i >_. <_i <_J' LONGITUDE.,
ELEVATION:
PHOTOGRAPHS
FRAME ID AZIMUTH
i_, , i (^} LAP CARD
i_J v__i Q_)
— — o
1 — 1 1— I 1 — 1°
1250.000 MAP NAME
1:24.000 MAP NAME
VISUAL ESTIMATES:
STRFAM WIDTH' ,
QTpc&M DFPTH
GAGE HEIGHT. u_» i_
units ^-
-j.»_i ft O
WATERSHED ACTIVITIES/DISTURBANCES
(Check all that apply)
units
D Roadways D Paved
O Bridged
D Unpaved
O Culvert
O Dwellings- O Single unit(s) O Multiple unit(s)
O Agriculture. O Cropland
O Fenced
O Industry. Specify Type
O Logging-
O Mining. Specify Type:
O Quarries
O Beaver dams: O Above Site
O Livestock-. D Cattle
O Horses
Cl Other-
O Pasture
O Unfenced
O Below Site
O Sheep
P r>h»r
BANK VEGETATION WITHIN 100 METERS OF
STREAM BED (Check all that apply)
Type Absent Sparse Moderate Heavy
Deciduous Trees: O
Coniferous Trees1 O
Shrubs O
Wetland Areas. D
Grasses- O
Rocky/Bare: O
COMMENTS D. SEE REVtHSt
ODD
0 O O
ODD
ODD
ODD
ODD
SIDE
O Grade Distance from Stream:
Distance from Stream-
DistanrB from Stream-
Distance from Stream:
Distance from Stream-
Distance from Stream -
Distance frQro Strearp- _.
Distance from Stream:
Instance f'om St'eanrv ., ...
Distance from Stream*
STREAM SUBSTRATE
(Check all that apply)
Type Absent Sparse N
Boulders: O O •
Cobble: O O
Gravel: O D
Sand: O O
Silt: D O
Autwuchs-. O O
DATA QUALIFIERS
f)?) = Other*
o
c\
w
n
w
o
w
n
\**>
o
o
\_x
o
oderate Heavy
0 O
D D
O D
O 0
O D
D 0
FIELD CREW DATA
r.HFW in
SAMPL FH 1
RAMPI fay
r~.nfr-.iffn HV
26
-------
Figure 3-2. Daily field station activities in the Phase I-Pilot Survey.
Instrument Calibrations
and Quality Control Checks
Pack Equipment and Supplies
Sampling Teams
Depart
Sample Collection
Field Measurements
Coordinator Files Daily Report
with Las Vegas Communications Center
Requests Audit Samples for Next Day
Sampling Teams Return
to Field Station
Audit Samples Shipped
Field Laboratory
Preparatory Activities
Samples and Data Forms
Transferred to
Field Laboratory
Field Crews
Next Day
1
Receive Audit Samples
at Field Laboratory
Samples Organized
into Batch for Processing
Daily
Debriefing
Schedule
Next Day's
Activities
1
Samples Analyzed
and Aliquoted
Next Day
Aliquots Shipped
to Contract Laboratory
blank" sample at the first site visited. Reagent grade
deionized water (meeting ASTM Type I specifica-
tions), prepared daily with a Millipore Milli-RO/
Super Q System in the mobile laboratory, was
pumped from the four-liter Cubitainers carried to the
site into clean sample containers. One team chosen
at random also filled a second set of sample
containers (Cubitainer, syringes, and 500 ml bottle)
with stream water from the pump immediately after
the routine sample had been collected. This second
sample was labeled as a "field duplicate" sample.
One of the most critical in situ measurements was
pH. At the beginning of the study, there was concern
that C02 present under highly supersaturated
conditions in streamwater would de-gas rapidly from
water samples held in open containers during the
measurement, thus preventing a stable pH reading.
In order to determine whether de-gassing might
present a problem, two measurements of pH were
conducted at streamside on different aliquots of
sample. One measurement was made on an aliquot
collected in a beaker. This aliquot was exposed to
27
-------
Figure 3-3. Daily activities of the field sampling teams during the Phase I-Pilot Survey.
Assemble Equipment
and Supplies
Calibrate and
Check Instruments
Load Vehicles
Travel to Site
Conduct pH
Measurements
Open-System
Closed-System
No
Read Staff Gauge
Record Site Observations
Conduct In Situ
Measurements of
Dissolved Oxygen
and Conductivity
Set Up Pump and
Sample Collection
Equipment
Collect
Field Blank
(if required)
Duplicate
Measurements
Required?
Yes
Final Instrument
Checks at Site
I
Pack Samples and Equipment
Return to Vehicle
Transfer Samples
and Field Data Forms
to Field Laboratory
No
Yes
Return
to Field Station
28
-------
the atmosphere during collection and measurement,
and was operationally defined as an open-system
measurement. The second measurement was made
on a sample collected in a 60 ml syringe and a flow-
through sample chamber (described in Hillman et
al., 1985). This arrangement allowed pH to be
determined on aliquots that were not exposed to the
atmosphere during collection and measurement.
This was operationally defined as a closed-system
measurement.
In both cases, pH was measured using a portable
Beckman pHI-21 meter (automatic temperature
compensated) equipped with a Ross combination
electrode (Orion models 81-52 or 81-04). Field
instruments were calibrated each day before use at
the field station using commercially available high
ionic strength buffer solutions (pH 7.00 and 4.00).
A pH 4.00 (10~4 N H2S04) Quality Control Check
Sample (QCCS) was used to check the calibration
of the meter at each stream site. The meter was
recalibrated if it failed to read between pH 3.90 and
4.10. At any site where a field duplicate sample was
collected, duplicate open and closed pH measure-
ments were made on a second sample collected from
the stream (beaker or syringe). Protocols for both
pH measurements are described in Knapp et al.
(1987).
Specific conductance was measured using a YSI
Model 33 meter. A 10~3 N KCI solution (specific
conductance = 147 //S cm'1 at 25°C) was used to
check the factory calibration of the meters prior to
each day's sampling. The calibration was checked
at each stream site using a 10~4 N KCI solution
(specific conductance = 74 //S cm"1 at 25°C) as a
QCCS. Failure to meet acceptable values for these
checks (64-84 /uS cm"1) required recalibration of the
meter. Following the QCCS, the probe was attached
to the sampling boom and immersed into the stream
in an area of flowing water. Measurements were
taken at a depth of approximately 10 cm (mid-depth
if the site depth was < 10 cm). Conductance and
water temperature were recorded when the conduc-
tance reading changed less than 5 jjS cm"1 over a
1 -minute period. A check was made using the QCCS
after each in situ measurement.
Dissolved oxygen (DO) was measured using YSI
Model 54 or 57 meters. The calibration was checked
at the field station at the beginning and end of each
day with a QCCS of water, saturated with bubbled
air. Acceptable values of these checks were within
± 0.5 mg 02/l of the calibration value. The meters
were calibrated at each site using a chamber
fabricated from a metal tube and rubber stoppers,
containing water-saturated air. The DO.probe was
inserted into the chamber and the chamber was
submerged in the stream for 15-20 minutes to
provide a water-saturated atmosphere within the
chamber. The meter was calibrated at each site
based on the theoretical partial pressure of oxygen
at ambient temperature and elevation. After
calibration, the probe was removed from the
chamber, attached to the sampling boom, and
immersed into an area of flowing water to a depth
of approximately 10 cm. Dissolved oxygen concen-
tration and water temperature were recorded when
the oxygen reading changed less than 0.5 mg L"1
over a 1-minute period.
Additional measurements and observations recorded
for each site included water stage, weather
conditions, and any problems associated with sample
collection or measurements. All data and observa-
tions were recorded in logbooks at streamside.
Observations and final measurement values were
later transcribed to a standardized field data form
(Form 4, Figure 3-4). Failure to meet QC checks for
any measurement was noted on the field form. Any
changes in local watershed characteristics since the
previous visit were recorded, and the site dossier
was updated. A complete list of equipment and
supplies used by the field teams is presented in
Knapp etal. (1987).
3.3.3 Field Laboratory Operations
A field laboratory was used for the Phase l-Pilot
Survey in order to meet the 12-hour holding time
requirement for preliminary analyses and the
preservation/aliquoting steps. This field laboratory
was housed in a trailer originally designed for the
ELS and provided a "clean" environment for
analyses and preparation of aliquots for analysis of
chemical variables critical to the NSWS. Field
laboratory analyses included pH, DIG, specific
conductance, true color, turbidity, and aluminum
fractionation. The specifications for this laboratory
are described in Morris et al. (1986). The daily
activities at the field laboratory are summarized in
Figure 3-5.
The site coordinator received audit samples from an
independent laboratory (Drouse et al., 1986) and
streamwater samples from the field crews daily.
Each sample was assigned an ID number, and all
samples received and processed at the field
laboratory on a given day constituted a field batch,
which also was given a unique ID number. All
containers associated with a given stream or audit
sample were labeled with the appropriate batch and
sample ID numbers. Once a batch was organized
and labeled, the bulk water and syringe samples
were transferred to the laboratory supervisor for
processing and analysis. Aliquots for total suspended
solids were transferred to the environmental science
laboratory at Southwestern Technical College for
29
-------
Figure 3-4. NSS Data Form 4: Stream Data.
NATIONAL SURFACE WATER SURVEY
STREAM DATA
FORM 4
DATE
TIME
STATE
LATITUDE
STREAM ID
STREAM NAME
D D M M M Y Y
pH METER ID i i i i
T/COND ID i 11 i
DISSOLVED O ID L_J u_
. u_i i—J <—> LONGITUDE i__. i_. i—i°
SAMPLE REPLICATE
NUMBER
1 250000 MAP NAME
1 24,000 MAP NAME
CLOUD COVER k_i i — i i — i %
RAIN Dmfv D NO D UGHT D HEAVY
GAGE HEIGHT , ,., , , •'<("")
D RISING D FALLING ^"^
DATA QUALIFIERS
(A) INSTRUMENT UNSTABLE
(§) REDONE FIRST READING NOT
ACCEPTABLE
(£) SLOW STABILIZATION
(N) DOES NOT MEET QCC
f7\ DTHPO fo.plam)
pH
Y N
METER CALIBRATION D D
QCCS = pH 4.00
QCCS INITIAL 1 — ii_J.
ROUTINE OPEN. ._,._,.
SAMPLE TEMP. . — . i — i
ROUTINE CLOSED . — 11 — ,.
DUPLICATE OPEN. i_j i_i.
SAMPLE TEMP i— < i— '
DUPLICATE CLOSED , ,i_j.
QCCS FINAL v_i i__i
i— Ji— >O
^^o
- cQ
^^o
^^o
-°cO
^^o
CONDUCTIVITY y«S
Cond QCCS = 74 @ 25 • C
QCCS INITIAL. • — i , — , , — .
IN SITU' i ii it i
STREAM TEMP.: !__,,__•.. ,' C
QCCS FINAL i — •• — >•—*
DISSOLVED OXYGEN mg /
o
^^*
o
o
o
t
QCC = Theoretical — Measured
^INITIAL iHn J.i i
IN SITU i ii i.i i
AFINAL- LLii i.i i
o
o
COMMENTS
D NOT SAMPLED. SEE BELOW
REASON
INACCESSIBLE D NO ACCESS PERMIT D TOO SHALLOW D OTHER
FIELD LAB USE ONLY
TRAILER ID
BATCH ID _
SAMPLE ID .
FIELD CREW DATA
CREW ID
SAMPLER 1 _
SAMPLER 2 _
CHECKED BY .
FORM DISTRIBUTION
WHITE COPY — ORNL
PINK COPY - EMSL-LV
YELLOW COPY — FIELD
30
-------
Figure 3-5. Daily activities at the field laboratory during the Phase I-Pilot Survey.
Field Sites — — • - •
Quality Assurance Samples
Audit Sample Preparation Laboratory -*
Routine
Samples
Syringes
4-L Container
500-mL Bottle
Field Blank
(deionized water)
4-L Container
500-mL Bottle
Transp
. Fiolri 1 a
at
orted to
boratory
4°C
Field Duplicate
Syringes
4-L Container
500-ml
. Bottle
Audit Samples
Field Laboratory
2-L Container
Eight
Preserved
Aliquots
- Field Laboratory - •
Shipped to
Field Laboratory
at4°C
Transl
Southwes
College f<
erred to ^_^_2i*i
t Technical
>r Analysis
..,, t , '
Total
Suspended
Solids
~jL
Data Transcribed
to Data Form
-mL Bottles
Syringes
, ^ s«
dissolved Closed-Systen
norganic PH Measureme
Carbon •—— .
I
Next Day
1
1
* t
imples Organized
into Batch
t
i Turbidity
m Measureme
Data Transcribed
to Data Form
4-L Containers
True Color
Measurement
Aliquot
Preparation
Filtration
Preservation
Aluminum Extractions
Hold at 4°C
Send Copies to
Data Management Center and
Quality Assurance Personnel
Next Day
Prepare Shipping Form(s)
Ship Aliquots for
Contract Laboratory
analysis. Details of the field laboratory analytical and
sample processing protocols are presented in
Hillman et al. (1985).
One syringe sample from each stream or audit
sample was allowed to reach ambient temperature
for pH determination. All other containers were
stored at approximately 4°C until analysis or
processing. The laboratory supervisor conducted the
DIG and pH determinations. One analyst prepared
fractions for subsequent analysis of total extractable
and non-exchangeable aluminum, and the other
analyst prepared the other six aliquots from each
bulk water or audit sample indicated in Table 3-2.
All aliquoting was conducted in a laminar flow hood
to avoid contamination. The third analyst conducted
turbidity and true color determinations, and pre-
served the aliquots as they were prepared by adding
concentrated acid and/or refrigerating them.
One sample in each batch was designated as a
"trailer duplicate" for purposes of analyzing
duplicate precision for mobile laboratory analyses.
Two aliquots of this sample from each syringe were
analyzed for DIG and pH, respectively. Two subsam-
ples were prepared from the bulk sample and
analyzed for turbidity and true color. All quality
assurance protocols for these analyses and process-
ing steps are described in Drous£ et al. (1986).
After all aliquots had been prepared and preserved,
they were sealed, bagged individually, and the data
transcribed to a standardized form. Aliquots were
held at approximately 4°C overnight. The following
morning, aliquots were packed with standardized
shipping forms into insulated containers with
enough chemical refrigerant packs to maintain
samples at approximately 4°C during transport.
Aliquots were shipped to the contract analytical
31
-------
Table 3-2. List of NSS Aliquots, Containers, and Preservatives*
Aliquot (Container)
Preservative
and
Description
Parameters
1
(250 ml)
Filtered,
pH<2
with
HNO3
Ca(180)
Mg(180)
K(180)
Na (180)
2a
(10ml)
Filtered
MIBK-HQ
Extract
Total
Extractable
A! (7)
3a
(250 ml)
Filtered, no
Preservative
CI" (28)
F- (28)
SOf (28)
NO; (7)
4a
<125ml)
Filtered,
pH<2
with
H2S04
DOC (14)
NH< (28)
5a
(500 ml)
Unfiltered
no
Preservative
pH(14)
BNC(14)
ANC(14)
Specific
Conductance
(14)
6°
(125ml)
Filtered,
pH<2
with
H2S04
Total P
(28)
7
(125ml)
Filtered
pH<2
with
HNOa
Total Al
(180)
8a
(10ml)
Filtered
MIBK-HQ
Extract
Non-
exchangeable
extra ctable
Al(7)
Mn(180)
Fe(180)
Si02(28)
DIC (14)
*Maximum permitted holding times from date of sampling are shown in days in parentheses with each variable.
"Stored at 4°C in the dark.
laboratory via overnight courier service. Copies of
the field and field laboratory data forms were sent
to the data management center at ORNL and to
quality assurance personnel at EMSL-LV. Copies of
the shipping forms were sent to the NSS sample
management office (Viar and Company, Alexandria,
Virginia).
3.4 Evaiuation of Equipment and
Methods
Selection of equipment and protocols initially
proposed for use in the Phase l-Pilot Survey was
based on consultation with experienced researchers,
previous experience in NSWS projects, and procure-
ment constraints. The three tasks identified for
investigation included: (1) evaluation of meters for
suitability in the field; (2) evaluations of different
techniques; and (3) study of the possible effects of
extending sample holding times beyond 12 hours.
Details of the equipment and methods evaluation
are presented in Knapp et al. (1987).
3.4.1 Equipment Evaluation
Field meters were evaluated on the basis of field
tests, laboratory tests, ease of use, portability, and
overall durability. The Beckman pHI-21 pH meter
equipped with an Orion Ross 81-04 pH electrode
etched with 50% NaOH prior to use was used for
all streamside pH measurements. Enclosing the
meter in a plastic bag and devising a special carry
case similar to that used for small cameras greatly
increased the suitability of the instrument for field
use. The YSI Model 33 S-C-T meter, although not
temperature compensating, was used for making
field conductivity measurements. YSI Model 54 and
57 meters with membrane-type probes were used
for measuring dissolved oxygen. All .meters were
found to be satisfactory, and were recommended for
use in Phase I.
3.4.2 Methods Evaluations
Field methods were developed based upon recom-
mendations of instrument manufacturers and
researchers and on similarity to methods described
in the ELS methods manual (Hillman et al., 1986b).
Modifications of these methods were evaluated
during field sampling. Some modifications were
developed and adopted immediately (e.g., pH) and
some were evaluated and rejected. The following
sections summarize the evaluations of several field
methods. Details of the procedures as adopted for
Phase I can be found in Hagley et al. (1986).
3.4.2.1 Filtration Methods
Streamside filtration of samples was attempted in
an effort to avoid potential deterioration of samples
before delivery to the mobile laboratory. In a field
evaluation conducted over several days, a filtration
apparatus (Nalgene cartridge filtrator) which used
disposable filters (Gelman 47 mm diameter, 0.45 /am
pore size Metricel) was fitted into the Tygon pump
line. The filtration apparatus was used on both the
suction and the discharge side of the pump in an
evaluation conducted over a period of three days by
teams collecting samples from a total of 16 streams.
Drawbacks included a high potential for sample
contamination during filter replacement or filter
rupture, unacceptably long filtration times, and a
32
-------
requirement for additional rinse water and supplies.
It was concluded that in-line filtration at streamside
was not practical.
3.4.2.2 Streamside pH Measurements
Two methods of streamside pH measurement were
performed at each stream throughout the Phase I-
Pilot Survey. A closed-system method using a
syringe was designed to measure the pH of a sample
without atmospheric contact. An open-system
method, developed by the U.S. Geological Survey,
was also evaluated. Early in the study, an experiment
was conducted to evaluate the equivalence of these
two methods and to compare each pH measurement
technique on samples collected using a pump versus
grab samples collected directly from a stream. Three
replicate samples of each treatment combination
(method x collection device) were measured at each
of three streams. A two-way analysis of variance
detected no significant differences (p = 0.05). Several
experimental devices designed to make in situ mea-
surements without developing a streaming potential
also were tested and showed no statistically
significant differences from either open or closed-
headspace streamside measurements (Knapp et al.,
1986).
Following the completion of field data collection, the
open and closed field pM measurements were
compared with the mobile lab pH measurement
(variable PHSTVL) to determine the degree of
equivalence among the three techniques. All
analyses were based on samples collected during
the summer period, during which all field protocols
had been finalized. Paired t-tests, unweighted for
each inclusion probability, indicated no significant
difference between the two streamside measure-
ments, and a statistically significant but unimportant
difference of 0.03 units between the streamside and
mobile laboratory closed pH measurements (p =
0.05). (Water equilibrated with 300 ppm v/v COa
at the contract analytical laboratory showed a
significantly higher pH value than the closed
headspace measurements, owing most likely to the
COa supersaturation common in small streams).
Linear regressions were performed to compare the
ability to predict the field laboratory closed pH value
on the basis of either of the field methods (Figure
3-6). The slopes of the regression lines were virtually
identical and not significantly different from unity
(0.995). Although the closed field pH was a slightly
better predictor of field lab pH based on a smaller
mean standard error of the estimate, the open field
pH measurement was chosen because of its logistic
simplicity, it is important to note that the open and
closed field pH techniques gave very similar results;
a bias adjustment of 0.03 units yielded virtually
identical population distributions. However, it is not
known which pH measurement technique (open or
closed) is more accurate. The choice of the field lab
closed measurement to express most of the Phase
l-Pilot Survey population estimates was based
primarily on consistency with the NLS data (Linthurst
etal, 1986).
3.4.2.3 Aluminum Methods
The Phase l-Pilot Survey employed a previously
untested (in the NSWS) protocol for fractionation of
MIBK extractable (monomeric) aluminum into non-
exchangeable (organic) and exchangeable (inor-
ganic) forms. The exchangeable fraction was
calculated as the difference between total extrac-
table aluminum and the non-exchangeable fraction
which are measured directly. The determination of
non-exchangeable aluminum involved passing
aliquot #8 (Table 3-2) through a cation exchange
column prior to complexation with 8-
hydroxyquinoline and extraction into methyl isobutyf
ketone (MIBK). Total extractable aluminum was
determined similarly, except the aliquot was not
passed through the exchange column. Details of the
procedure are described by Hillman et al., (1986a).
The ion exchange resin (Amberlite 125) had to be
conditioned before use so that the pH of the resin
column was within 0.5 pH unit of the expected
sample pH. Columns were conditioned by adjusting
a solution of 10~5 N NaCI to the desired pH with
HCI or NaOH. This solution was passed through the
resin column, collected, and the pH measured. This
process was repeated until the desired column pH
was achieved.
Following preparation of the column, a 125 ml aliquot
of sample was filtered into a 250 ml Pyrex beaker
that had been washed with 5% HN03 and rinsed
with deionized water. Portions of the filtered aliquot
were used to rinse a 50 ml polycarbonate centrifuge
tube. The remainder was pumped through the ion-
exchange column (30 ml/min). The first 30 ml of
sample from the column were discarded. The next
20 ml were collected and analyzed for pH, and the
following 25.0 ml volume was collected in the
centrifuge tube. The column was then flushed with
the buffer solution, and an aliquot of the buffer was
collected from the column. The pH of this aliquot
was measured to ensure that the column was
conditioned properly for the next sample. The aliquot
in the centrifuge tube was complexed and extracted
into MIBK.
Adjusting the pH of the NaCI solution was often very
time-consuming, and the solution was not stable
over time. Allowing the solution to equilibrate with
the atmosphere overnight before adjusting the pH
sometimes improved stability, however. During
33
-------
Figure 3-6. Comparison of three pH methods used in the Phase I-Pilot Survey. Confidence bounds (90 and 95%) about
the regressions are shown.
Q.
o
9.0 i—T
8.5
8.0
7.5
7.0
6.5
6.0
i l I I i i I T
y = 0.088 + 0.995x
r2 = 0.981
6.0
6.5
7.0 7.5 - 8.0
Open Field pH
8.5
9.0
9.0
8.5 -
8.0
7.5
I
Q.
-0
CD
S 7.0
o
6.5
6.0
I I I I I I I I I [ I I I I I I I I I I I I I I I I .1
y = 0.051 +0.995x
r2 = 0.990
I I I ' ' ' ' I ' ' « t I I I I I
6.1
6.6
7.1 7.6
Closed Field pH
8.1
8.6
9.1
34
-------
laboratory operation,'3 to 4 different columns had
to be prepared daily to cover the range of sample
pH values, and there was no standard solution of
non-exchangeable extractable aluminum that could
be used as an audit sample to check on the accuracy
of the procedure. A new, automated monomeric
aluminum speciation and measurement technique
using pyrocatechol violet has been instituted in
Phase I to overcome some of these difficulties
(Dougan and Wilson, 1974; RogebergandHenriksen,
1985).
3.4.3 Holding Time Studies
The 12 hour holding time protocols established for
the National Lake Survey (in which helicopters were
used to collect and transport samples) set significant
limits on the area that could be served by a mobile
laboratory because of driving time constraints. A set
of pilot experiments were conducted on five streams
in the Southern Blue Ridge to evaluate the stability
of syringe and Cubitainer samples over periods of
5, 12, 24, and 48 hours. Although the experiment
yielded little indication of changes in any of the
parameters, the number of audit samples included
was insufficient to establish the degree of within-
treatment analytical variability or among-treatment
analytical bias.
Two experiments were designed to overcome these
obstacles: (1) a laboratory study aimed at establishing
the COa-impermeability of the syringes under
markedly sub- or super-saturated conditions; and (2)
a field test of Cubitainer samples collected on a
number of lakes and streams in the Eastern U.S.
These experiments are documented thoroughly by
Burke and Hillman (1986) and Stepanian et al.
(1986), respectively, and the results are briefly
summarized here.
3.4.3.1 Syringe Experiment
In the syringe experiment, syringes filled with 1
mg L"1 Na2CO3 solution at pCO2 levels of Ox, 1x,
10x, and 100x atmospheric levels (atm) were held
for 1 to 8 days at 4°C and 25°C. Companion
experiments were conducted to test the ability of
the experiment to detect C02 equilibration in open
systems that were similarly sub- or super-saturated
and blanks, 10 ppm C02, and pH 4 H2SO4 QCCS
samples. Open containers containing O and 100 x
atm pCOa equilibrated within 24 hours. Conversely,
none of the samples showed significant changes in
DIC or pH over 7 to 8 days when held in syringes
at 4°C and 11°C (Table 3-3). Syringes did gain or
lose DIC when held at room temperature, however,
apparently due to increased permeability of the
polypropylene syringe walls. Experiments with
actual lake samples at 0.1-0.2 x atm pCOa produced
similar results (Burke and Hillman, 1986).
Table 3-3.
Dissolved Inorganic Carbon Concentrations
(mg L~1 ± 1 s.d.) in Samples Initially Sub- or
Supersaturated with CO2 and held for 7-8 Days
Time (Days)
PC02
Temp
7-8
0 x atm
0 x atm
1 x atm
1 x atm
10 x atm
10 x atm
100 x atm
100 x atm
10
23—26
11
23—27
8
22—26
4
22—26
1.548 + 0.023
1.548 + 0.023
2.134 + 0.003
2.134 + 0.003
2.840 + 0.038
2.840 + 0.038
12.41 ±0.12
12.41 +0.12
1.514 ±0.036
1.757 + 0.023
2.078 ± 0.037
2.286 ± 0.036
2.857 + 0.068
2.927 ± 0.088
12.02 + 0.15
9.44 + 0.22°
'Variance of 24-30 repeated measures on test samples in a
treatment < analytical variance of quality control check (QCCS)
samples analyzed along with the treatments (a = 0.05).
Based on these experiments, it was determined that
the holding protocol for DIC and pH held in syringes
at 4°C could safely be increased to 24 hours.
Although no experiments were performed on
aluminum, it has been assumed that syringe aliquots
can also be held for at least 24 hours prior to
aluminum extraction. It is assumed that pH changes
driven by COa degassing are the most significant
cause of alteration in aluminum speciation in
samples held for at least 5 to 6 hours prior to
extraction.
3.4.3.2 Cubitainer Experiment
In the Cubitainer holding time experiment, two 19-
liter samples were collected in June, 1985, from
three lakes in New York, three streams in Pennsyl-
vania, two streams in Maryland, and one stream each
in South Carolina and Tennessee. These water
bodies represent a wide range of water chemistry
types. Samples were transported at 4°C by air within
12 hours of collection to the field laboratory, where
they were each split into eight aliquots. Two aliquots
were processed immediately (12 hours), while the
remainder were held at 4°C, and duplicates
processed after 24, 48, and 84 hours, respectively.
Duplicate QA audit samples from Big Moose Lake,
New York, were analyzed with each batch. All
analyses were performed according to regular NSS
protocols.
The results of the Cubitainer holding time experiment
are presented in Table 3-4. Two criteria were utilized
in assessing the significance of observed changes.
First, all data that are below the limit of detection
for that variable were excluded (see Table 4-9). For
the remaining sample pairs, the percentage increase
or decrease between each analyte concentration at
12 hours holding time and the 24, 48, or 84 hours
holding times was calculated. Each percentage then
was compared with the maximum root mean square
35
-------
Table 3-4.
Changes in Constituent Concentrations in Duplicate Field Samples and Big Moose Lake OA Audits Held at 4°C
for 12, 24, 48, and 84 Hours Prior to Stabilization*
Mean Percent Change
in Field Samples
Chemical
Variable
ANC
Sulfate
Nitrate
Chloride
Silica
Fluoride
Calcium
Magnesium
Sodium
Potassium
DIC (equil)
pH (equil)
DOC
Ammonium
Total P
Total Al
Extractable Al
Iron
Manganese
Pairs
(n)
9
10
10
10
10
10
10
10
10
10
9
10
9
4
3
6
6
9
9
12-24
hr
1.1
-2.3
-4.9
-2.0
1.2
0.0
3.1
1.2
-0.2
2.4
2.5
0.0
0.4
0.1
-3.7
39"
-2.5
20
4.4
12-48
hr
1.7
-2.0
-5.2
-1.4
0.2
-1.2
! M
0.5
-0.3
3.0
8.5
-0.02
-5.5
14"
14°
~on
12
25
-21"
12-84
hr
0.5
-2.8
-4.5b
-3.4"
0.6
-2.2"
0.8
1.4
-1.4b
0.7
7.9
-0.02
4.8
13"
4.4
12
3.8
-3§
^16
System
Precision
RMS RSD (%)"
5.0
3.3
5.9
2.2
8.0
2.1
2.3
1.1
1.1
3.8
9.8
0.1
6.2
10
5.1
20
12
25
8.9
Mean Percent Change
in Audit Samples
12-24
hr
C
-2.6
-12.7
-0.5
0.2
0.7
2.6
1.9
0.8
0.9
C
0.0
-0.7
2.8
C
28
4.8
3.0
5.6
12-48
hr
C
-1.1
-17.0
2.7
-0.2
-4.6
5.3
2.1
-0.1
1.6
C
0.0
2.3
1.4
C
8.3
25
5.9
3.9
12-84
hr
C
2.8
-1.9
1.1
0.0
-0.7
0.5
1.8
-0.7
-0.8
C
0.0
5.0
-1.4
C
12
-8.2
-20
1.1
*Underlined values exceed the RMS %RSD for routine/duplicate precision.
aRoot mean square percent residual standard deviation.
"Exceeds interbatch bias as determined by changes in the audits.
GNo bias estimate could be calculated.
percent residual standard deviation (RMS % RSD)
for routine duplicate sample pairs analyzed during
the Phase l-Pilot Survey (Table 4-9). If the percentage
difference between the analyses at two different
holding times was less than the routine/duplicate
differences typically expected to occur within the
same batch, then differences resulting from > 12 h
holding times were indistinguishable from routine
sampling and analytical error. Several variables
show discernible differences, but only four variables
exhibited potentially distinguishable changes
between 12 and 24 hours (nitrate, calcium, mag-
nesium, and total aluminum).
The foregoing analysis does not include the effects
of interbatch analytical bias, however. A measure
of interbatch analytical bias was obtained by
analyzing duplicate QA audit samples from Big
Moose Lake, New York, with each batch of samples.
This QA audit has been shown to be chemically
stable for the variables of interest over holding
periods of several months (Table 4-3; Drousd et al.,
1987). Percentage changes in each variable were
calculated for the QA duplicates for the four holding
time intervals and are also shown in Table 3-4. If
the chemical analytes are truly stable in the audit
samples, then these changes represent the degree
of interbatch analytical bias in the holding time
experiment (again, concentrations below the limit
of detection were excluded).
Of the discernible analyte changes based on the
estimated within-batch precision, most were
accompanied by at least the same degree of
interbatch bias. The differences that exceeded the
apparent interbatch bias are asterisked in Table 3-4.
Only two variables, calcium and total aluminum,
appear to show possible changes between 12 and
24 hours. The 3% change in calcium is barely
detectable above the analytical precision and is of
no practical interpretive significance. The large
percentage changes in total aluminum, as with iron,
manganese, and total phosphorus after 48 hours,
probably result in part from sampling errors
associated with the colloidal nature of these
constituents in streams. A great deal of precision
is neither expected nor needed for these variables
in streams. Ammonium is the only other constituent
seen to change in 48 hours. Although the increase
may be a result of organic nitrogen mineralization,
the increases also are barely greater than the
analytical variance.
It should be noted that the choices of 1 RMS RSD
and various other decision criteria in this analysis
are rather arbitrary, and no rigorous statistical
36
-------
testing is implied. Instead, this interpretation should
be thought of as a screening procedure by which
to focus attention on the variables most likely to have
experienced changes after 12 hours. Also, the
experiment does not assess the probability of
chemical changes during the 12 hours between
sample collection and analysis. Any regional survey
activity must be predicated on the stability of samples
for this period, or on the assumption that any
changes that do occur are minimal, quantifiable by
calibration, or of no interest (e.g., speciation changes
for some variables may not affect the types of
interpretations to be expected from synoptic data
sets). Secondly, it is very difficult to perform a holding
time experiment such as the one described above
that incorporates a reasonable range of geographic
variability with a high degree of statistical discrim-
inatory power. Thus far, the data have been analyzed
using at least three different approaches, including
various group means and treatment of the audit data
(Overton, personal comm.; Stapanian et al., 1986),
including a multivariate statistical Hoteling-Lawley
trace approach.
Training for field sampling personnel in the future
will provide more details on the selection criteria
for a suitable sampling location at a given stream
site. Additional training in field hydrology will ensure
that staff gauges are placed correctly at all locations.
Experience in the field demonstrated that, in most
cases, equipment initially selected for use in the
Survey was adequate. A sensitive, portable conduc-
tivity meter will not be needed because conductivity
will also be measured at the field laboratory on a
research-grade instrument. Comparisons of the two
streamside pH protocols indicate that the more
difficult "closed" measurement is not needed,
because COa de-gassing is apparently sufficiently
slow in unstirred natural waters. Results from the
holding time experiments indicate that the holding
time protocols can probably be safely increased to
at least 24 hours, provided aluminum samples are
held in the COa-tight syringes.
All analysts have concluded that there is no
important effect of increasing the sample holding
times to as much as 48 hours. This is not to say
that no sample will change in this time, but that
the frequency and magnitude of such changes are
probably acceptable in terms of the data quality
objective of the project.
3.5 Summary of Field Operations
The Phase l-Pilot Survey was completed as sche-
duled on July 17, 1985. In completing the sampling
activities, the three field crews traveled approxi-
mately 45,000 miles by vehicle, averaging 270 miles
per day to access 1 to 3 stream sites. Most stream
sites were accessible by vehicle alone. A few sites
required additional transport by horseback or boat.
The longest hike required to a site was 16 miles
round trip. A total of 339 reach sites were sampled
during the survey, 724 samples were processed by
the field laboratory, and 668 samples were shipped
to the contract analytical laboratory. Only one
shipment of samples was delayed during the entire
operation.
The field operations plan implemented during the
Survey worked very well and was not modified during
the study. The preliminary reconnaissance activities
and contacts with local cooperators were integral
to the success of the field operations plan, and served
to minimize unexpected problems associated with
site access and daily sampling itineraries. The few
problems that were encountered were caused by
outdated maps and these were quickly rectified.
37
-------
4. Quality Assurance and Data Management
4.1 Introduction
The water quality data gathered in the National
Stream Survey constitute a large and important
research data base that requires a high degree of
quality assurance (QA) and quality control (QC).
Maintaining a high degree of QA/QC involves two
separate but highly integrated tasks. The first task
is to establish a QA/QC program to ensure that all
samples are collected and analyzed in a consistent
manner and to establish the accuracy and precision
of the reported values with a known degree of
confidence. Second, a data management program
must be designed to store and track the data, identify
and eliminate entry errors, and keep a record of such
changes. Ultimately, the product of these tasks will
be well-documented data files that are readily
accessible to project scientists and extramural users.
4.2 Quality Assurance/Quality Control
Operations
This section provides an overview of the QA/QC
activities in field sampling, field laboratory, and
contract analytical laboratory operations. Elements
of the QA/QC program include contract laboratory
performance evaluations, on-site auditing of field
and contract laboratories, specification of all sample
handling protocols, and utilization of a variety of QA/
QC samples. Detailed discussion of these elements
can be found in the Project QA Plan (Drous6 et al.,
1986).
solicit contractor support; and (3) evaluation of the
lowest bidders to ensure that qualified laboratories
were selected.
An SOW was prepared to document the analytical
methods and the QA/QC requirements that are
defined in the Analytical Methods Manual (Hillman
et al., 1986b) and Project QA Plan (Drouse et al.,
1986), and to specify these requirements in a
contractual format. An IFB was advertised in the
Commerce Business Daily in December 1984.
Approximately 180 laboratories responded and were
sent copies of the SOW. The lowest bidders were
sent pre-award performance evaluation (PE) sam-
ples. These laboratories were required to analyze
high- and low-concentration PE samples and to
report results within 15 days of sample receipt. The
data reports were evaluated for quality and
completeness.
Two laboratories scoring 88 percent subsequently
were visited by an EPA team to verify their
qualifications and capabilities to meet the contrac-
tual requirements. Both laboratories passed the PE
sample analysis and on-site evaluations and were
awarded contracts to provide analytical support to
the Phase l-Pilot Survey. Only the New York State
Department of Health (NYSDOH) laboratory received
samples during the survey, however, because they
were the lowest bidder and possessed ample
capacity to analyze all of the samples.
4.2.1 Selection of Contract Analytical
Laboratories
The objective of the analytical laboratory selection
process was to award contracts to the smallest
number of laboratories possible in order to minimize
potential interlaboratory bias, while ensuring that
each laboratory chosen could analyze the required
number of samples within the specified holding time
and quality performance criteria. The Contract
Laboratory Program (CLP) established to support the
EPA's hazardous waste monitoring activities was
used in laboratory procurement. The contract
process required: (1) preparation of a statement of
work (SOW) that defined the analytical and QA/QC
requirements in a contractual format; (2) preparation
and advertisement of an Invitation for Bids (IFB) to
4.2.2 Training
Data quality depends on the ability of the project
personnel to properly collect, process, and analyze
samples, and training is essential in ensuring
consistent application of all operational and quality
assurance procedures. Field laboratory personnel
underwent a five-day training period in Las Vegas,
Nevada, in all technical aspects of laboratory
operations.
4.2.3 Daily Quality Assurance Contact
During sampling and analysis, the QA staff com-
municated daily with the field station and the
contract laboratory to monitor logistics, methods, and
QA/QC activities. These communications were
38
-------
crucial and effective in identifying and resolving
issues affecting data quality at an early stage (see
Section 4.6.2). Each communication was logged
either on a field communication form or in a bound
laboratory notebook.
4.2.4 Field and Contract Laboratory Audits
On-site evaluations of the contract laboratory and
the field station were conducted during the survey
to assure that sampling and analysis activities were
being performed as planned. The contract laboratory
was visited once before sampling started and once
during field activities. The purpose of the first on-
site visit was to assure that the analytical laboratory
had the capability to perform the required analyses.
During the second on-site evaluation, QA/QC data
were reviewed and several issues were identified
and resolved. For example, it was discovered that
NYSDOH was analyzing pH and DIC at different
times, and was experiencing problems with the air-
equilibrated pH measurement. All observations were
summarized in an on-site laboratory evaluation
report.
Auditors also conducted an in-depth review of field
laboratory operations and interviewed the sampling
teams. During the on-site evaluation, the auditors
observed that the trailer was near a road where a
large amount of dust was present, resulting in
elevated total aluminum values in some samples.
The auditors recommended moving the field labor-
atory to a nearby dust-free location. Samples
processed after relocation of the trailer indicated no
further contamination. Also as a result of the review,
calibration activities were relocated to a heated
building to avoid slow meter response times on cold
mornings.
4.2.5 Field Sampling Quality Control
Procedures
The QC procedures consisted of calibrating all
instruments before and after each sampling trip and
monitoring any changes between calibrations. The
procedures are described in detail in Knapp et al.
(1987) and are discussed in Chapter 3 of this report.
The calibration check for temperature was to
compare the field meter reading to that determined
using an NBS-traceable thermometer. The reading
had to be within 2°C to meet QC criteria. A QCCS
having a theoretical pH value of 4.00 was analyzed
prior to and following all streamside pH determina-
tions. If any QCCS reading deviated from the
theoretical pH by more than ±0.1 pH unit, the
instrument was recalibrated and the pH of the QCCS
was remeasured. If the reading still did not meet
the specifications, then a data qualifier was recorded
on the Stream Data Form 4 (Figure 3-4). The in situ
specific conductance measurement was verified by
checking the factory calibration of the conductivity
meter by measuring QCCSs of 147 /uS cm"1 and 74
pS crrf1. The allowed error for the QCCSs were
± 15 /uS cnrf1 and ± 10 pS cm"1, respectively. The
QC check for dissolved oxygen consisted of calibrat-
ing the meter with water-saturated air, and then
measuring the dissolved oxygen in a sample taken
from a carboy of water saturated by bubbling with
compressed air. The readings had to be within 0.5
mg L"1. There were no QC checks for staff gauge
and other stream site data (Table 2-3).
All streamside and-in situ measurements and QC
data were recorded on Stream Data Form 4 (Figure
3-4). This multipart form was checked for complete-
ness and internal consistency at the field station.
One part of each form was sent to Oak Ridge National
Laboratory (ORNL) for entry into the NSS raw data
base. A second part of the form was sent to the
QA group in Las Vegas where it was checked to
identify and correct transcription errors and to
ensure that QCCS criteria were met. All forms were
sent by overnight mail.
4.2.6 Field Laboratory Quality Control
Procedures
The primary functions of the field laboratory were
to chemically stabilize aliquots of field samples and
to perform limited analyses for those variables that
are relatively unstable. The objectives of preservation
were to inhibit biological and chemical activity and
prevent changes due to volatility, precipitation, and
adsorption. Preservatives for each aliquot are
described in Table 3-2. Filtration through a 0.45-
/ym membrane filter removed suspended material,
including large colloids, and provided subsamples
that contained only dissolved analytes and smaller
colloidal material. Aliquots 1, 4, 6, and 7 were
preserved with strong acid to prevent loss of
dissolved analytes through precipitation or chemi-
cal/biological reactions. Storage at 4°C was required
to reduce biological activity in all aliquots except 1
and 7 and MIBK or volatilization in aliquots 2 and
8.
After sample preparation and preservation steps,
holding times were monitored to assure that the
samples were analyzed before any significant
degradation had occurred. The maximum permitted
holding times are shown in parentheses after each
variable in Table 3-2.
4.2.7 Quality Assurance/Quality Control
Samples
The QA program utilized a variety of QA/QC samples
to assure that the sampling and analytical activities
were performed according to the QA Plan and the
39
-------
data quality objectives. Every effort was made to keep
the number and costs of QA/QC samples within
logistic and budgetary constraints while providing
adequate information to the QA staff. Because little
information was available on the chemical stability
of low ionic strength waters, EPA protocols for
analysis of water and wastewater samples were
used (U.S. EPA, 1983).
4.2.7.1 Quality Control Samples
All QC activities related to field laboratory measure-
ments of DIG, pH, true color, and turbidity are
described in the QA Plan (Drouse" et al., 1986). QC
sample type, source applications, and frequency of
use are shown in Table 4-1 and described further
below.
Calibration Blank—Analysis of a calibration blank
was required for each batch of samples. This blank
(ASTM Type I deionized water) was analyzed after
the initial calibration to check for drift in the
measured signal and for contamination. The
observed concentration was required to be less than
or equal to twice the detection limit required by the
SOW contract.
Reagent Blank—A reagent blank was required for
dissolved Si02 and total aluminum analyses because
additional reagents were added in the digestion step
prior to analysis. The reagent blank was essentially
a calibration blank that had undergone the digestion
steps prior to analysis.
Matrix Spike—A matrix spike was required for each
batch of samples. A matrix spike is a routine sample
to which a known quantity of analyte at a concen-
tration of approximately twice the indigenous level
or ten times the detection limit (whichever was
greater) was added. The purpose of the matrix spike
was to verify the accuracy of the analysis in a matrix
typical of the samples being analyzed. The contract
laboratory met the limits for spike recovery for every
batch and no matrix interferences were observed.
Laboratory Duplicate—A contract laboratory dupli-
cate was required for each batch of samples. The
duplicate analyses provide estimates of within-batch
analytical precision, which must be met for the
samples in each batch to meet the QA limits
established for these variables.
Quality Control Check Sample—Each QCCS was a
commercially or laboratory-prepared sample that
was obtained from a source different from that used
for the calibration standards for the analyte. It was
analyzed to verify calibration at the beginning, after
every ten samples, and at the end of each batch.
The observed concentrations were required to be
within specified control limits. A low concentration
QCCS also was analyzed for some variables to
determine and verify the detection limits for these
analytes.
4.2.7.2 Quality Assurance Samples
External QA samples were used to judge the overall
performance of the sampling and analytical activities
and to establish the quality of the data with known
confidence limits. Table 4-2 lists types, sources, and
applications of QA samples used in the Phase l-Pilot
Survey. These samples were processed through the
field station and were "double blinds" to the contract
laboratories (i.e., the laboratory did not know that
they were QA samples and did not know their
composition).
Field Blank—A field blank was a deionized water
sample (meeting specification for ASTM Type I
reagent-grade water) that was carried to the stream
and processed through the sampling pump as though
Table 4-1. Types, Sources, and Applications of Quality Control Samples Used in the Phase l-Piiot Survey (Drouse, 1987)
Sample Type Description/Source Application Frequency
Quality Control
Check Sample
(QCCS)
Contract Laboratory
Blank3
Trailer Duplicate3
Contract Laboratory
Duplicate3
Matrix Spike
Standard; source other than
calibration standard
Reagent-grade water (zero
analyte concentration)
Stream sample; split
Stream sample; split
Sample plus known quan-
tity of analyte
Indicates accuracy and
consistency of calibra-
tion
Indicates signal drift
and sample contami-
nation
Indicates within-batch
precision
Indicates within-batch
precision
Indicates sample ma-
trix effect on analysis
Before, after every
ten, and after final
sample analysis
One per batch
One per batch
One per batch
One per batch
"Samples serve both as QA and QC samples.
40
-------
Tabie 4-2. Types,
Sample Type
Sources, and Applications of Quality Assurance Samples Used in the Phase I-Pilot Survey (Drouse. 1987)
Description/Source Application Frequency
Field Blank
Contract Laboratory
Blank3
Field Duplicate3
Trailer Duplicate3
Contract Laboratory
Duplicate9
Field Audit
Contract Laboratory
Audit
Reagent-grade water
treated as a stream sample
Reagent-grade water (zero
analyte concentration)
Duplicate stream sample
Stream sample; split
Stream sample; split
Synthetic samples and nat-
ural lake samples
Synthetic samples and nat-
ural lake samples
Estimate system deci-
sion limit and quanti-
tation limit
Estimate nonparamet-
ric detection limit
Estimate overall
within-batch precision
Estimate analytical
within-batch precision
Estimate analytical
within-batch precision
Estimate overall
among-batch preci-
sion; estimate labora-
tory bias
Estimate analytical
among-batch preci-
sion; estimate labora-
tory bias
One per day
One per batch
One per day
One per batch
One per batch
As scheduled
As scheduled
"Samples serve both as QA and QC samples.
it were a routine sample. One field blank was
collected by each sampling team on each operating
day.
These samples were intended to identify any
contamination problems that may have occurred in
the overall sampling and analytical processes. Field
blank data were used to establish estimated decision
limits, quantitation limits, and background values
expected for each variable.
Field Duplicate—A field duplicate was a second
sample collected at the stream site by the same
sampling team immediately after the routine sample
was collected. Field duplicate data were used to
estimate overall within-batch precision for the
sampling and analytical processes. One field
duplicate was collected on each sampling day.
Trailer Duplicate—/^ trailer duplicate was a spike of
a routine sample processed in the mobile laboratory.
One trailer duplicate was processed for each batch.
The trailer duplicate was used to establish the
analytical precision of the analyses performed in the
field laboratory.
Audit Samples—Two types of audit samples were
used as QA checks on field and contract laboratory
operations. Field audit samples were used to
establish overall field and contract laboratory
performance. Laboratory audit samples were used
to establish the performance of the contract
laboratory. The use of both types of samples enabled
field laboratory problems to be distinguished from
analytical laboratory problems.
Field audit samples were received in 2-liter aliquots
from Radian Corporation (Austin, Texas, laboratory)
and were processed as routine stream samples by
the field laboratory. Laboratory audit samples were
received at the field station already prepared as
aliquots 1 through 8 from Radian, which were then
sent to the contract laboratory for analysis. Labor-
atory audit samples thus were not subject to any
analytical errors arising at the field laboratory.
Two natural samples and two low-concentration
synthetic samples were also used during the Survey
(Table 4-3). The natural samples (from Big Moose
Lake in the Adirondack Mountains of New York and
from Bagley Lake in the State of Washington)
represented two types of low ANC and low ionic
strength surface waters expected to be encountered
in the Survey. Following collection, these samples
were filtered through a 0.45-jum filter and stored
at 4°C until use. Synthetic audit samples were
prepared just prior to being sent to the field
laboratory. A high-concentration synthetic sample
was not utilized because concentrations of analytes
in streams in the Phase l-Pilot Survey area were
anticipated to be quite low.
4.2.8 Data Review
The results of the various chemical analyses were
reported on appropriate field and laboratory reporting
41
-------
Table 4-3. Composition of Big Moose Lake (FN4) and Bag ley Lake (FN5) Natural Audit Samples
Field Audit Sample Concentration3
Variable
Big Moose Lakeb
Bagley Lake0
Al, organic ext.
Al, total ext.
Al, total
ANC, (yueq L~1)
BNC (Aieq L"1)
Ca
ci-
Conductance (fiS/cm)
DIG, airequilib.
DIG, initial
DOC
F~, total dissolved
Fe
K
Mg
Mn
Na
NH4+
N03-
P, total
pH, acidity
pH, alkalinity
pH, air equilib.
Si02
SO4
0.123
0.284
0.418
-25
133
1.96
0.469
33
0.167
0.320
7.53
0.076
0.134
0.659
0.367
0.092
0.628
0.038
2.35
0.006
4.63
4.63
4.72
4.45
6.46
0.002
0.005
0.037
156
41
1.96
0.22
18
1.62
1.74
0.63
0.029
0.003
0.37
0.24
0.001
1.06
0.026
0.085
0.005
7.04
7.02
7.41
10.8
0.937
aAII variables are measured in mg/l unless otherwise indicated.
bMean concentration of 37 analyses of Big Moose Lake sample processed at the field laboratory.
c,Mean concentration of 9 analyses of Bagley Lake sample processed at the field laboratory.
forms, each of which was checked for accuracy
before entry into the data base. Prior to describing
these procedures in detail, however, it is helpful to
understand the NSS data flows and data base
structure. The NSS data-base management system
is described in the following section.
4.3 Data Base Management
NSS data-base management activities are patterned
after procedures developed for the National Lake
Survey (Kanciruk et al., 1986). All NSS data sets
are maintained at Oak Ridge National Laboratory
(ORNL) on IBM 3033 mainframe computers using
the SAS software package (SAS Institute, Inc., 1983,
1985). Data sets are also periodically transferred to
the National Computer Center (NCC) at Research
Triangle Park, North Carolina, via magnetic tape,
where they can be accessed by NSS scientists at
the Las Vegas and Coeval Ms laboratories.
4.3.1 Data Structure and Flow
The basic structure and data flow employed during
the Phase l-Pilot Survey are schematized in Figure
4-1. Three data bases, "raw," "verified," and
"validated," represent increasing levels of data
scrutiny. Data initially were entered into a raw data
set from the various field and laboratory reporting
forms. When enough data became available, a data
tape was sent to NCC, where it could be accessed
by the QAteam. Changes tothe raw data set included
insertion of data qualifiers (tags and flags) and
substitutions for incorrect values discovered by the
QA team at EMSL-LV. Changes were sent to ORNL
via a "change" tape, which was used to update the
existing raw data. When more raw data became
available, the process was repeated.
The verified data base was in turn used in the process
of validation, wherein additional data qualifiers and
substitutions were made based on examination of
distributions of variable values among samples by
the technical staff at ERL-C. This process also was
iterative, and involved the generation of additional
change tapes. The use of such tapes allowed any
changes to be tracked for any raw datum in the data
base.
4.3.2 Primary NSS Data Sets
The raw data set contains data that received a
preliminary review by ORNL and EMSL-LV staff to
ensure that they conformed to proper formats, were
complete and legible, and were within plausible
ranges (Rosen and Kanciruk, 1985). The raw data
set was used internally by the management team
to screen data for problems, to perform trial data
42
-------
Figure 4-1. NSS data structure and flows.
Verification by
SC/CERL and QA/EMSL
Data Editing
^ 1 I site Heports
I VERIFIED DATA
M Maps
Validation by
SC/CERL, QA/EMSL
and ORNL
Data Editing
and Flagging
f VALIDATED DATA
\
Maps, Reports,
Statistical Analysis
Data Access
and
Distribution
^ *s
analyses, to test and debug computer codes and
programs, and to make design adjustments when
needed. The raw data was continually updated as
new data were received from the field and as errors
were corrected.
At the verified level, data were reviewed and any
errors in transcription, keying, or processing were
corrected. Error checking as part of the verification
process included intra-sample analyses such as
cation-anion balances and chemical equilibrium
checks as described below. Verified data are
assumed to represent the correct values that were
measured and recorded in the field or contract
laboratory. As in the raw data set, the verified data
set was revised several times during the verification
process. Verification changes were initiated by the
QA group at EMSL-LV and the required corrections
43
-------
were made by ORNL Following entry, the data set
was reverified to search for entry errors that occurred
during the data set editing process.
The validated set contains data which were subjected
to the highest level of review. In contrast to
verification, the emphasis of validation was on inter-
sample comparisons. Validation routines were
performed as described below, and data were flagged
or deleted from the validated data file. Data in this
validated file will be archived in STORE! (1985) as
well as in the official NSWS data base.
Each raw, verified, and validated data set contains
11 data files that correspond to the individual field
and laboratory forms on which the data were
reported (Table 4-4). This data structure provides a
logical basis for data entry and tracking that is
necessary for large data bases such as the NSS data
base.
4.3.3 Enhanced Data Files
An "enhanced" or "interpreted" data set (printed
out in Appendix D) was subsequently created from
the validated data set for specific purposes. In this
enhanced data set, data for routine/duplicate sample
pairs were averaged, any missing values from the
validated data set were replaced by averaging or
calibration of other chemical variables, and data
associated with episodes were identified. This
enhanced data set is a clean and compact set for
performing population distribution estimates and
certain mapping and statistical analyses. The
enhanced data set will also be released to external
scientists upon request.
4.3.4 Data Change and Qualifiers
Three types of data qualifiers are used in the data
base: tags, flags, and missing value codes (see Sale
et al., 1986 for values assigned to data qualifiers).
Tags are assigned based on field observations made
during sample collection (e.g., an erratic field meter).
Flags are assigned during data verification and
validation to indicate questionable values or values
that did not meet QA/QC standards. Missing value
codes are entered directly into the data base to
indicate the reason for a missing datum (e.g., sample
lost). Numeric changes and data qualifiers assigned
in the verification or validation processes are sent
to ORNL in the form of change records (i.e., "tuples;"
see Section 4.4.5). These change records are then
applied to a copy of the raw data set to generate
Table 4-4. Data Set Members for the Raw, Verified, and Validated Versions of the NSS Phase f-Pilot Survey Data Base
Member
Name
F04
FOB
F07
F11
F13
F18
F19
F20
F21
F22
F71
Description
Field measured variables from
Form 4
Trailer measured variables from
Form 5
Site and watershed characteristics
from Form 7
Analytical chemistry from contract
lab from Form 1 1
Titration data from Form 13 or
diskette
Detection limits (QA/QC) from
contract labs and Form 18
Holding times (QA/QC) from con-
tract labs and Form 19
Blanks (QA/QC) from contract labs
and Form 20
Spikes (QA/QC) from contract labs
and Form 21
Duplicates (QA/QC) from contract
labs and Form 22
Site location variables from vari-
Data
49
25
56
35
22
49
79
418
232
332
31
vii i ihrvsi wi vaiiafcsi
Tags
21
9
13
29
12
39
50
411
227
325
0
CO
Flags
19
9
0
23
12
26
50
411
227
325
0
Number of
Observations
339
724
117
668
45K
51
668
51
51
51
117
ous sources
44
-------
a series of partially verified data sets. A permanent
file of all change records is maintained for post-
processing audits. This three-tiered system of
independent checks (Figure 4-1) is essential to
achieving the NSS data quality objectives and
producing the high-quality data base required for
NSS analyses.
Data anomalies were reported to the field laboratory
coordinator for corrective action. Data reporting
errors were reported to ORNL to be corrected before
entry into the raw data set. Telephone communi-
cations were documented in a bound notebook, and
data changes were annotated on the appropriate
form.
4.4 Data Verification
Data verification involves the identification and
correction, flagging, or elimination of data of
unacceptable quality on the basis of intra-batch QA
criteria. Verification involved: (1) reviewing the
available QA/QC data from the field and contract
laboratories; (2) reviewing any comments or
questions associated with the batch or sample under
evaluation; (3) performing QA checks for data
consistency and chemical reasonableness; (4)
reviewing QA sample data; (5) obtaining confirma-
tion, correction, or reanalysis data from the
laboratories; and (6) providing the verified data for
entry into the ORNL data base. Computer programs
were developed to automate this procedure as much
as possible. A team of auditors evaluated each data
package on a sample-by-sample basis using the
procedures outlined below.
4.4.1 Review of Field Data Forms
Verification began with the receipt of the data forms
from the field. The auditor reviewed each form to
check the following items:
1. Stream ID—The Stream Data Form (Form 4)
was compared with the Batch QC Field Data
Form (Form 5) to identify transcription errors.
2. Trailer Duplicate—Form 5 had to have a
duplicate Stream ID that matched a routine
stream sample ID, and the field precision
criteria had to be met.
3. Calibration Data—pH and conductivity calibra-
tion data on Form 4 were compared to the data
from the field calibration forms to ensure that
initial calibration criteria were met, or that the
appropriate data qualifiers were recorded.
4. Streamside pH—The Form 4 pH values (open
and closed) were compared to the field
laboratory pH value on Form 5.
5. Field Laboratory pH and DIC—Form 5 values
for field audit samples were compared to
acceptance criteria. Routine/duplicate pairs
and trailer duplicates were evaluated for
within-batch precision.
6. pH and DIC QCCS Data—Form 5 QCCS data
were reviewed to ensure that criteria were met.
4.4.2 Initial Review of Sample Data Package
As they were received, the sample data packages
were reviewed for completeness, internal QC
compliance, and proper use of data qualifiers. A
checklist was used by the EMSL-LV auditor to assure
consistency in the review of all data packages. Any
problems were reported to the appropriate contract
laboratory manager for corrective action. Comments
provided by the laboratory with the data package also
were reviewed to determine any impact on data
quality or need for follow-up action by the laboratory.
4.4.3 Review of Quality Assurance/Quality
Control Data
Following entry of the data into the raw data set
at ORNL, a magnetic tape containing the data was
sent to NCC. The QA personnel then were able to
access the data by telecommunication. The verifi-
cation process utilized a series of computer programs
that comprise the AQUARIUS QA/QC system
(Fountain and Hoff, 1985). The programs listed in
Table 4-5 identify or flag results that were classed
as "exceptions" (i.e., that did not meet the expected
QA/QC limits). The AQUARIUS system automated
much of the routine QA review process, which
enabled the auditor to concentrate more effort on
the substantive tasks of correcting or flagging
questionable data. The auditor used the output from
these programs (along with original data and field
notebooks) to complete the NSWS Verification
Report called for in the QA Plan. The form of the
Verification Report was a work sheet designed to
systematically guide the auditor through the
verification process by explaining how to: (1) flag
data; (2) track data resubmissions and requests for
reanalysis and confirmation; (3) list the steps that
led to identification of QA exceptions; and (4)
summarize modifications to the raw data set (change
records).
Each sample was verified individually. Stream
sample analytical results had to meet checks for
anion-cation percent ion balance difference (% IBD)
and for percent conductivity difference (% CD) in
order not to generate an "exception," unless the
discrepancy could be explained by either the
presence of organic species (as indicated by the
Protolyte Analysis Program) or by an obvious and
correctable reporting error. The Protolyte Analysis
45
-------
Table 4-5.
Exception Generating Programs Within the AQUARIUS Data Review and Verification System (Fountain and
Hoff, 1985)
Program
Data Type0
Audit Sample Summary
Lab/Field Blank Summary
Field Duplicate Precision Summary
Instrumental Detection Limit Summary
Holding Time Summary
Conductance Check Calculations
Anion/Cation Balance Calculations
Batch QA/QC Summary
Comparison of Form 4 and Form 5
Comparison of Form 5 and Form 11
Protolyte Analysis
Audit Sample Window Generation
Raw Data Listing
QA/QC Flag Summary
Reagent/Calibration Blanks and QCCS
Calculation of Laboratory Penalties
Matrix Spike Summary
Modified Gran Analysis
(FL, LL, FN)
(B, LB)
(R, D, Pairs)
(All Species)
(All Species)
(All Species)
(All Species)
(All Exceptions)
(pH and DIG)
(pH and DIC)
(DIC, DOC, pH, ANC, and BNC
Data Evaluation)
"FL = Field Low Audit.
LL = Laboratory Low Audit.
FN = Field Natural Audit.
B = Blank.
D = Duplicate.
LB = Laboratory Blank.
R = Routine.
Program flagged field and contract laboratory
measurements of pH, DIG, ANC, BNC, and DOC when
carbonate equilibria, corrected for organic protolytes,
were not in internal agreement. Additional data
qualifiers were added to a given variable when the
QA samples within the same analytical batch (field
blanks, field duplicates, or audit samples) did not
meet the acceptance criteria. Additional data
qualifiers were added if internal QC checks such as
matrix spike recovery, calibration and reagent blank
analyses, internal duplicate precision, required
instrument detection limit, QCCS percent recovery,
and required holding times were not met. In all cases,
each flag generated by AQUARIUS was evaluated
by the auditor for reasonableness and consistency
before it was entered into the data set.
4.4.4 Follow-Up with Contract Laboratories
Completion of the verification steps in sections 4.4.2
and 4.4.3 required follow-up with the contract
laboratory to confirm or correct reported data and
to reanalyze samples, if required. This follow-up was
the most difficult and time-consuming step in the
verification process, particularly when requests to
the laboratory were not specified in the original
statement of work. Typically, responses to requests
for confirmation or correction of reported data were
completed within two to four weeks. Re-analyses
were completed only if specified holding times had
not been exceeded.
4.4.5 Preparation and Delivery of Verification
Tapes
After the previous steps were completed, the data
were used to construct the verified data set. This
process required a consistent and trackable method
for transferring the change records to ORNL. The
process chosen used data base entries called
"tuples." A tuple consists of an ordered set of seven
variables (batch ID, sample ID, variable, old flag, new
flag, old value, new value) which identifies a change
to the data set. Tuples can be generated automat-
ically by AQUARIUS or manually by the auditor (e.g.,
changes and deletions). Tuples are stored in separate
data files until the tuple listing is ready to be sent
to ORNL. At that time, a computer program combines
all of the tuple areas and appends the combined tuple
list to the data set (flag, tag, or value changes) only
if the batch ID, sample ID, variable name, and old
value match. The combined tuple list was written
to a magnetic tape and mailed to ORNL from NCC.
ORNL then processed the tuple list and returned it
to NCC via a magnetic tape. Any illegal tuples ("no-
go's") which could not be applied to the data set
had to be reexamined by the QA staff. This procedure
was repeated approximately five times before the
final verified data set was generated.
4.5 Data Validation
The process of data validation was intended to assure
that data generated during the Phase l-Pilot Survey
46
-------
accurately described the physical and chemical
characteristics of the study area. Validation, an
iterative process performed in conjunction with data
verification, highlights "unusual" values, which
subsequently are investigated for entry, transcrip-
tion, or analytical errors. Suspect values are checked
against all data forms and the verified data set, and
then flagged or changed, as appropriate.
Validation of the Phase l-Pilot Survey data consisted
of:
1. Frequency analyses
2. Univariate analyses
3. Multivariate scoping
4. Bivariate/multiple linear regression analyses
5. Multivariate analyses
6. Episodes screening
7. Reverification/validation checks and data
correction/flagging
4.5.1 Frequency Analyses
In order to develop an appropriate strategy for
validation, it was necessary to determine first the
basic structure of the Phase l-Pilot Survey data set.
The SASPROCFREQuency procedure (SAS Institute,
1985) was used to produce one, two, three, four,
and five-way frequency/cross tabulation analyses
of the verified data. The analyses were ordered on
various combinations of stream ID, batch ID, sample
ID, sample code, and individual chemical variables,
each of which provided information on the s, 'cture
and completeness of the data base. As in verification,
this procedure can uncover errors such as duplicate
sample entries within a batch of samples, missing
stream IDs, invalid or incorrect stream IDs or sample
codes, and transcription errors. Once the data base
structure was determined and preliminary correc-
tions were made, more advanced statistical proce-
dures were applied.
4.5.2 Univariate Analyses
The first approach to outlier detection was to
consider each chemical variable individually,
searching for values that were extreme with respect
to all other observations in the data set. Univariate
analysis of the data consisted of: basic summary
statistics with plots, and computation of univariate
fences. Univariate summary statistics, together with
histograms or stem and leaf diagrams, probability
plots, box plots, and the five extreme high and low
values were computed for all observed routine and
duplicate values of each variable. In addition to
identifying extreme values, these techniques
provided useful information on the underlying data
distributions and variability. For example, many of
the major anion and cation concentrations demon-
strated log-normal distributions, which required data
transformations prior to conducting multivariate
tests. Seven data combinations were evaluated using
univariate statistics: ail data, all spring data, all
summer data, spring downstream data, spring
upstream data, summer downstream data, and
summer upstream data.
A unique feature of the Phase l-Pilot Survey
compared to previous NSWS designs is the multiple
observations at each stream reach through time.
These multiple observations permitted computation
of univariate statistics for all samples (regular and
duplicate) collected from each reach. Univariate
fences (Velleman and Hoaglin, 1981) were computed
for each stream using custom SAS programming and
the SAS PROC UNIVARIATE (SAS Institute, 1985)
procedure. The fence procedure compares univariate
quartiles for each chemical variable computed under
SAS PROC UNIVARIATE definition one: weighted
average of Xnp. An inner quartile range (i.e., the
difference between the first and third quartiles) was
used to establish various "hinges:"
Inner lower hinge = Q1 - (1.5 x QDIFF)
Outer lower hinge = Q1 - (3.0 x QDIFF)
Inner upper hinge = Q3 + (1.5 x QDIFF)
Outer upper hinge = Q3 + (3.0 x QDIFF)
where Q1 = 25th percentile, Q3 = 75th percentile,
and QDIFF = Q3 - Q1. Any data value falling inside
the inner hinges, between the inner and outer
hinges, or outside the outer hinges was so noted
and identified for further checking.
4.5.3 Multivariate Scoping
To examine relationships among two or more sets
of variables in the validation process, it was first
necessary to specify which sets of variables should
be explored. Many such relationships could be based
on previous experience or upon formal geochemical
models. This process is most suitable for bivariate
analysis, but the 4,600 potential bivariate pairs in
the data set make this approach to validation
inefficient. A more efficient procedure was to
perform multivariate regressions using several
related parameters, rotating the dependent variable
and comparing predicted to observed values in order
to detect outliers.
Although multivariate suites could also be based on
geochemical models, we chose to take an empirical
approach. Correlation coefficients were computed
for all chemical variables measured during the
Survey. Highly correlated variables then were placed
into 14 suites of variables (Table 4-6). Twelve suites
47
-------
Table 4-6. Variable Suites Obtained from Multivariate Scoping
1. ANC (alkalinity)
2. Aluminum (total)
3. Calcium
4. Chloride
5. Specific Conductance
6. Aluminum (organic extract)
7. Potassium
8. Ammonium
9. Silica
10. Turbidity
11. pH (field lab)
12. DIC (field lab)
13. Calcium
vs.
vs.
vs.
vs.
vs.
vs.
vs.
vs.
vs.
vs.
vs.
vs.
vs.
14. BNC (Acidity)
vs.
Calcium
Specific Conductance
Magnesium
Silica
pH (field lab)
Ammonium
Turbidity
True Color
Specific Conductance
Total Dissolved fluoride
Sulfate
Silica
Specific Conductance
Sodium
Total Dissolved Fluoride
Potassium
Magnesium
Sodium
Silica
Sulfate
Potassium
Magnesium
Silica
Total Extractable Aluminum
Magnesium
Turbidity
True Color
BNC
pH (field lab)
Magnesium
True Color
pH (initial and air-equilibrated)
DIC (initial and air-equilibrated)
Magnesium
Potassium
ANC
Specific Conductance
Silica
Sulfate
pH (field lab)
Aluminum (total)
Ammonium
Turbidity
True Color
Dissolved Organic Carbon
were used in regression analyses (bivariate or
multiple linear), and two suites were used in the
SAS Principal Components Analysis and PROC
FASTCLUSter analysis.
4.5.4 Bivariate/Multivariate Linear Regression
Analyses
Although the concentrations of neither of two
variables in a single sample may be outliers within
their respective univariate distributions, the ratio of
the pair may reveal one of the values to be an outlier.
Scatter plots were used to examine relationships
between pairs of observed and predicted values for
a given variable using simple and/or multiple linear
regression analysis. For suites 1 through 4 and 6
through 12 (Table 4-6), simple linear or multiple
linear regression analyses were performed, in which
each variable was modeled as the dependent variate
on all other variables in the suite. Only specific
conductance was modeled as the dependent variate
for suite 5. Outliers were identified by a combination
of visual inspection of regression plots of observed
versus predicted dependent variates, and by use of
a studentized residual threshold. Observations were
identified as outliers if the absolute value of the
studentized residual [(actual - predicted) / (residual
standard deviation)] was greater than 4. Each
regression was repeated three times with outliers
identification and removal of outliers after each
iteration.
48
-------
4.5.5 Multivariate Analyses
In contrast to multiple linear regression, in which
a single dependent variate is modeled on two or more
theoretically (or practically) related independent
variables, multivariate analysis enables examination
of several variables simultaneously. Suites 13 and
14 (Table 4-6) were examined using cluster analysis
and principal components analysis. Cluster analysis
is a classification technique for identifying similar-
ities or dissimilarities among observations. Each
observation is compared to others in the set and
is assigned to a group or cluster using a measure
of similarity. The PROC FASTCLUS procedure in SAS
(SAS Institute, 1985), a non-hierarchical divisive
method that is sensitive to outliers, was used in the
validation process. Principal components analysis
forms factors from linear combinations of the original
variables, such that the first factor reflects most of
the dispersion in the data. Each successive factor
explains less variance. If the original data are
approximately normally distributed, the resulting
factors are also approximately normal, and a plot
of any two components results in an elliptical cluster
with outliers displaced from the ellipse.
4.5.6 Episodes Screening
The purpose of episodes screening was to identify
chemical values attributable to rain storms that
occurred immediately before or during field sam-
pling, in order to exclude these data from population
estimate computations. The purposes of this
exclusion are explained in Section 2.2.2. Preliminary
data screening precipitation data from the three
NOAA meteorological stations in the study area, field
(Data Form 4; Figure 3-4) records of precipitation
and cloud cover, date and time of sampling, staff
gauge height and direction of change (if any), and
turbidity data. Four screening criteria were devised:
1. Precipitation > 0.1 inch on the same date at
a meteorological station.
2. Indication of rain on field Data Form 4 (light,
heavy, or previous).
, 3. Gauge height > 0.25 ft over other spring
measurements.
4. Turbidity increase (4x if baseline value is > 10
NTU, 2x otherwise).
For a particular sample to be flagged as an episode,
three of the four criteria had to be met. Eight spring
episode samples were identified as a result of the
screening process and substituted by calibration in
the enhanced data set.
Summer episodes and upstream episodes Were
difficult to detect using the screening technique
described above, due to the lack of readily compar-
able staff gauge orturbidity data. Several alternatives
involving comparisons using spring downstream
data were explored, most of which failed to provide
clear decision criteria. During the validation process
described in the following section, samples with
multiple chemical outliers were flagged as potential
episodes, and cross-checked against criterion 1 or
2 above. Satisfying either criterion caused an episode
flag to be generated. All of these flagged values were
carried into the enhanced data set, because no
substitute numbers were available, but the values
have been treated as missing in some of the
statistical comparisons, as noted in Chapter 5.
4.5.7 Reverification/ Validation and Data
Correction/Flagging
The end product of the six validation steps was a
master matrix of samples containing outliers. This
matrix was ordered by stream ID, sampling point
location (upstream and downstream), and time of
sampling (any of four spring and one summer
sampling intervals); outliers were identified by
chemical variable, and coded by a symbol denoting
the particular test (or tests) that identified that
observation as an outlier. Each code also specified
whether the routine and/or duplicate value (if
available) was flagged.
The validation matrix was sent to the QA/QC group
for reverification. All questionable values were re-
examined for data entry or other errors. The
reverified data then were subjected to a sample-by-
sample examination by the NSS technical manage-
ment team that resulted in a series of validation flags
(Table 4-7).
Substitution (U) flags were set for variables under
three conditions:
1. Downstream episodes: datum replaced with
average of remaining two spring downstream
samples (U1).
2. Datum flagged for which a duplicate analysis
was not flagged: regular sample datum
replaced by duplicate or duplicate datum
dropped (U2-U4).
3. Datum was impossible (e.g., extractable
aluminum was higher than total aluminum) but
no duplicate was available: datum was replaced
by calibration using a bivariate or multivariate
model developed as part of the validation
process (Section 4.5.4) (U2-U4).
Very few data were substituted under the last rule.
Validation (K) flags were used when data were
identified as outliers during validation, but not during
49
-------
Table 4-7. NSS Validation Flags
Substitution Flags
U1
U2
U3
U4
Validation Flags
K1
K2
K3
K4
Verification/Validation Flags
W2
W3
W4
Downstream Spring Episode
Univariate Outlier
Multivariate Outlier
Univariate and Multivariate Outlier
Episode—(No substitute value available)
Univariate Outlier
Multivariate Outlier
Univariate and Multivariate Outlier
Univariate Outlier—(no substitute value available)
Multivariate Outlier—(no substitute value available)
Univariate and Multivariate Outlier—(no substitute available)
the verification process. Generally, it is assumed that
these data represent "unusual" but not necessarily
incorrect numbers. Summer or upstream episode
samples, for which no substitute values were
available, represent a special case (K1) of such
situations. Other examples may indicate transient
pollution or contamination of sample containers.
Verification/validation (W) flags were generated
when data were identified in both validation and
verification procedures. These data may well be
incorrect, but no clearly superior substitute values
were available. Virtually all such cases involved
small discrepancies in the validation models, or
involved chemical concentrations close to the
detection limit for the variable. In general, it was
assumed that the averaging process employed in
constructing the enhanced data base would decrease
the impact of most small analytical errors on
population distribution estimates.
Once validation was completed, a final list of
validation change tuples was produced and sent to
ORNL, accompanied by instructions for building an
enhanced data set. This data set is the final product
of verification and validation, although intermediate
raw and verified (but final) data sets also are
produced. In the complete version of the enhanced
data base, episode values have been given a sample
code of "EA" if substitute values are available, or
"E" if substitute values are not available (e.g., spring
upstream, summer upstream, or summer down-
stream observations). Routine/duplicate observation
pairs are averaged and given a sample code, "DA."
Appropriate verification and validation changes have
been made in the enhanced data set, but all tags,
flags, and comments were dropped. These QA/QC
data remain available in the final validated data set.
4.6 Data Management and Quality
Assurance Results
The success of the Phase l-Pilot Survey data
management and QA program can be judged on
several counts, including the efficient performance
of the system in recording and tracking data, the
efficiency of the verification and validation processes
in identifying and treating suspect values, and the
degree of accuracy and precision attained in the
analytical data themselves. These issues are
addressed in the following section.
4.6.1 Data Base Management Performance
One measure of data base management perform-
ance for the Phase l-Pilot Survey is the length of
time required to complete the various data bases
described in Section 4.3.2. The corresponding dates
are:
17 July 1985
30 August 1985
30 October 1985
30 January 1986
30 March 1986
Field sampling
complete
Raw data base
complete
Verified data
base complete
Validated data
base complete
Enhanced data
set complete
These dates reflect completion of the "first draft"
of each data set. Reverification changes (none of
which involve numeric data changes) were finalized
on 30 May 1986.
The eight-month period required to produce a near-
final enhanced data set was not unexpectedly long
considering that more than 22 x 103 numerical
chemical data alone are represented in the data base,
in addition to flags, tags, site, and geographic data.
Raw data were generally available in machine-
readable form within seven weeks of collection for
'preliminary analyses. Several data transfer protocols
are being initiated in Phase I that are expected to
shorten some communications delays, and many of
the validation procedures that had to be developed
50
-------
specifically for the Phase l-Pilot Survey can be
transferred with minor modifications to Phase I.
4.6.2 Verification/ Validation Performance
Because a strict QA/QC program was adhered to
throughout the period of operations, any problems
that were encountered were detected and resolved
quickly through the daily QA contact. Examples of
issues that were addressed as a result of such calls
included:
1. incorrect calculations in reporting inorganic
and organic extractable aluminum and chloride
data;
2. use of contaminated matrix modifier (lantha-
num chloride) for calcium analysis by flame AA
that resulted in high calcium values;
3. indications of negative bias in manganese
analysis by evaluation of audit sample data;
4. use of an analytical method for nitrate analysis
that was not specified by the IFB contract;
5. illegible data reported by the contract
laboratory;
6. a brief aluminum contamination episode at the
field laboratory due to presence of large amount
of dust;
7. temporary contamination of aliquot #3 at the
field laboratory; and
8. inconsistent temperature correction and
reporting of in situ conductivity and QCCS data.
the preliminary QA/QC sample data, obtained
during daily communications, provided guidance for
QA staff to identify and solve most of the issues
that arose, resulting in minimal impacts on the final
data set. Several protocol changes were imple-
mented during the Survey, and others were made
after the Survey as a result of data evaluation. All
changes were incorporated into the final QA Plan
for Phase I (Drous6 et al., 1986).
Table 4-8 presents the final results of the verification
and validation processes. The verification data
include all routine, duplicate, trailer duplicate, audit,
and blank samples; validation data address all but
audits and blanks. There were a total of 20,613
Table 4-8. Results of Verification/Validation: Numbers of Observations Flagged and Numeric Changes Made (and percent
of total observations) in the NSS PIPS Data Base (excluding episode flags)
Chemical Variable
Acidity
Al (extractable)
ANC
Al (organic)
Al (total)
Ca
Cl
Color
Conductivity (lab)
Conductivity (in situ)
DIC (equilibrated)
DIC (initial)
DIC (field lab)
DOC
Fe
F (total)
K
Mg
Mn
Na
NH4
N03
pH (field-closed)
pH (field-open)
pH (acidity)
pH (alkalinity)
pH (equilibrated)
pH (field lab)
P
Si02
SO4
Turbidity
Total
Number of
Observations
668
668
668
668
668
668
668
724
668
339
668
668
724
668
668
668
668
668
668
668
668
668
339
339
668
668
668
724
668
668
668
724
Number (Percent)
of Observations
Flagged in
Verification
206 (30.8)
97 (14.5)
130 (19.5)
86 (12.9)
374 (56.0)
176 (26.3)
201 (30.1)
0 (0)
45 (6.7)
10 (2.9)
150 (22.5)
214 (32.0)
70 (9.7)
113 (16.9)
112 (16.8)
89 (13.3)
48 (7.2)
27 (4.0)
306 (45.8)
31 (4.6)
38 (5.7)
275 (41.2)
0 (0)
0 (0)
311 (46.6)
0 (0)
0 (0)
42 (5.8)
166 (24.9)
120 (18.0)
165 (24.7)
0 (0)
Number (Percent)
of Numeric
Changes from
Verification
0 (0)
2 (0.3)
1 (0.1)
13 (1.9)
7 (1.0)
0 (0)
3 (0.4)
0 (0)
1 (0.1)
7 (2.1)
4 (0.6)
11 (1.6)
0 (0)
15 (2.2)
1 (0.1)
1 (0.1)
2 (0.3)
0 (0)
287 (43.0)
1 (0.1)
0 (0)
72 (10.8)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
1 (0.1)
8 (1.2)
2 (0.3)
1 (0.1)
Number (Percent)
of Observations
Flagged in
Validation
6 (0.9)
15 (2.2)
2 (0.3)
7 (1.0)
9 (0)
5 (0.7)
0 (1.3)
6 (0.9)
1 (0.1)
2 (2.9)
0 (0)
0 (0)
0 (0)
5 (0.7)
10 (1.5)
2 (0.3)
2 (0.3)
3 (0.4)
1 (0.1)
5 (0.7)
10 (1.5)
3 (0.4)
3 (0.9)
2 (0.6)
5 (0.7)
3 (0.4)
45 (6.7)
1 (0.1)
5 (0.7)
2 (0.3)
5 (0.7)
8 (1.1)
Number (Percent)
of Numeric
Changes from
Validation
0 (0)
2 (0.3)
1 (0.1)
9 (1.3)
0 (0)
0 (0)
2 (0.3)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
1 (0.1)
0 (0)
0 (0)
0 (0)
0 (0)
1 (0.1)
0 (0)
1 (0.1)
6 (0.9)
0 (0)
0 (0)
1 (0.1)
0 (0)
45 (6.7)
0 (0)
1 (0.1)
0 (0)
2 (0.3)
0 (0)
51
-------
individual observations subjected to verification. In
situ dissolved oxygen and stream temperature data,
although not included in these statistics, were
examined manually and were found to be realistic.
Of all the data, 17.5% (3,602 observations) were
flagged, but only 2% (440 observations) were
changed numerically. The majority of the numeric
changes resulted from chemical reanalyses, as
discussed in Section 4.4.4. Most of these numeric
changes were less than 1% of the original values
reported.
Since audits and blanks accounted for 5,027 of the
20,613 observations subject to verification (ca. 25%),
validation addressed the remaining 15,586 values.
Of this number, only 1% (173 observations) were
flagged during validation, and less than 0.5% (72)
were actually changed. Forty-five of the 72 numeric
changes involved the first seven batches of
equilibrated-pH values, for which calibrated values
were substituted based on mobile and laboratory pH
observations. The verification/validation results
indicate that very few values were found to be in
error and subsequently changed.
4.6.3 Data Quality
The success of the Phase l-Pilot Survey ultimately
will be judged on the ability of the data to produce
robust population distribution estimates for the
primary NSS variables of interest. Error in these
estimates can arise from two primary sources:
sampling error and analytical error. The first source
is a function of the variability of the natural
environment and the sampling design. The second
is largely a function of the degree to which sampling
and analytical protocols are capable of providing
accurate data with acceptable levels of precision.
The statistical data in this section can be used to
answer some of these questions. Some of the QA
results that have a bearing on the interpretation of
the data in Chapter 5 are summarized in this section.
Drous6 (1987) provides a more detailed treatment,
including the degree to which contractual analytical
targets were met.
4.6.3.1 Detection Limits
During the Survey, 71 field blanks were processed,
and the data provide an overall estimate of the
normal background contamination that occurred
during sampling and analysis. Table 4-9 shows the
nonparametric decision limit for each variable based
on a statistical evaluation of the verified blank data.
This value represents the concentration limit, above
which the analyte can be detected with a known
degree (p = 0.5) of confidence.
For most of the variables, the prespecified targets
of the NSS QA Plan were met (Drouse", 1987).
However, data for the following variables indicated
background sources of contamination that caused
the decision limit to be appreciably higher than the
required detection limit:
1. Ammonium—although the detection limit
exceeded the prespecified target for ELS lakes,
concentrations in streams below 21 /ug L~1 (1.1
yueq L~1) are unlikely to be of interpretive
significance.
2. Total aluminum—contamination may result
from digestion reagents or dust in the field or
contract laboratories. Again, 0.062 fjg L"1 is
probably an acceptable detection limit in
streams, where some colloidal aluminum may
pass through the 0.45-/um filters, but has little
interpretive significance.
3. DIG—the background level in a blank exposed
to air was approximately 0.2 to 0.3 mg L"1,
which affects the results for samples that have
low DIG as a result of CO2 undersaturation or
low ANC.
4. DOC—blank background levels were approxi-
mately 0.1 to 0.3 mg L"1, apparently from COa
contamination. This value is above the concen-
tration of DOC of most Southern Blue Ridge
streams sampled during the study.
5. Total P—0.008 mg L"1 should be adequate for
interpreting most stream data with respect to
acid deposition, although many unpolluted
streams will have P concentrations below this
value.
6. Nitrate—Contamination is suspected by HNO3
vapors in the hood where aliquots are prepared.
This detection limit is not unacceptably high,
but reducing it is desirable, given the potential
importance of the anion in terms of acid
deposition.
In summary, most detection limit goals were
achieved in the laboratory. However, in interpreting
the data, the data user must take the results from
the field blanks into consideration. If the background
value from sample collection and handling is higher
than the laboratory (system) detection limit, obtain-
ing extremely low detection limits in the laboratory
is meaningless. The decision limit and system
detection limits must be considered as the real limits
for data interpretation.
4.6.3.2 Precision
Sampling and analytical variance, apart from
temporal (> 1 hour) variations in stream chemistry,
can arise in the survey from three major sources:
52
-------
Table 4-9. System Decision Limits and Precision Estimates9 Based on Interbatch Analysis of Field Audits and Intrabatch
Analyses of Field, Trailer, and Laboratory Duplicates (Drouse, 1987)
Variables
Al, organic ext.
Al, total ext.
Al, total
ANC {fjeq I"1
. BNC (/aeq I/1)
Ca
cr
Conductance (juS cm"1)
DIG,, air equilib.
DIG, initial
DOC
F~, total dissolved
Fe
K
Mg
Mn
Na
NH/
N03"
P, total
pH, acidity
pH, alkalinity
pH, air equilib.
Si02
S0<
pH (field lab)
DIG (field lab)
True Color (PCU)
Turbidity (NTU)
Non-parametric
System
Decision
Limit(P95)b'c
0.002
0.002
0.062
-
-
0.04
0.06
0.92
0.36
0.20
0.54
0.005
0.004
0.009
0.004
0.002
0.011
0.021
0.028
0.008
-
-
-
0.062
0.040
-
-
-
-
Field
Audits
(FN4)d
32
23
11
h
14
3.0
h
0.99
h
h
6.6
2.7
10
2.1
1.5
5.4
1.5
h
7.0
h
0.05s
0.05"
0.03e
h
2.8
-
-
-
-
Field
Duplicates
39
12
20
5
9.5
2.3
2.2
0.8
7.1
9.8
6.2
2.1
25
3.8
1.1
8.9
1.1
10
5.9
5.1
0.086*
0.083*
0.11*
8
3.3
0.05
4.00
21.5
14.9
Lab
Duplicates
5.1
5.1
12.89
-
-
1.2
1.6
0.5
2.3
2.5
2.3
1.2
18.2
1.1
0.8
6.0
0.8
7.7
3.6
5.1
0.03*
0.02*
0.03*
1.6"
1.9
0.03
3.36
11.7
10.4
"Root-mean-square of % relative standard deviation based on pairs with x > 10 times standard deviation of field blank.
"All variables are measured in mg/l unless otherwise indicated.
cPas= the 95th percentile of 71 field blank measurements.
dBig Moose Lake (FN4).
"Absolute standard deviation.
'Root-mean-square of standard deviation.
9x > 10 times standard deviation of the reagent blank.(
hx< 10 times standard deviation of the field blank.
(1) a field component associated with short term
temporal variability in stream chemistry, (2) an
analytical component associated with subsampling
an aliquot of water and random 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 can be assessed by comparative
statistical evaluations of analyses of field audits, field
duplicates, and laboratory and trailer duplicates. The
relative degree of precision in these analyses also
is shown in Table 4-9.
Precision of the various sets of analyses, with the
exception of pH, are expressed as root mean squares
of percent relative standard deviations of all samples
or sample pairs above the system quantitation limit.
The system quantitation limit, represented by ten
times the standard deviation of the corresponding
blank concentrations, assures that individual
samples considered in the analysis have sufficiently
high analyte concentrations that their expected
precision is constant. This practice insures that
samples with analyte concentrations near the
detection limit do not provide a false picture of the
interbatch or duplicate precision.
For most variables, interbatch variance, as estimated
from repeated measurements of the Big Moose Lake
field audits, exhibits the lowest degree of precision.
However, except for metal species, the relative
standard deviation is typically less than 10%. Within-
batch duplicate precision was better for most
variables, with pairs exhibiting the highest precision
associated with species that travel in colloidal form
in streams, and thus may be expected to exhibit some
degree of sampling variability when compared with
53
-------
even single sample aliquots of lake water. Laboratory
duplicate precision was better still, as expected, and
represents the highest degree of precision that could
likely be achieved in a project such as the NSS.
4.6.4 Summary
The QA/QC and data management program func-
tioned well in the Phase l-Pilot Survey and produced
a data set of known and acceptable quality in time
to meet project objectives. Much was learned,
however, about how to avoid or minimize future QC
problems and delays in data transfers and verifica-
tion and validation procedures. The new protocols
were implemented in the Phase I Research Plan (U.S.
EPA, 1985b; Drouse et al., 1986).
54
-------
5. Population Estimates and Stream Classification
5.1 Introduction
The primary objectives of the National Stream Survey
are (1) to provide population estimates of streams
that are currently acidic (low pH) or potentially at
risk from acid deposition (low ANC), and (2) to classify
streams for further intensive studies. Future studies
will aim at determining temporal (e.g., episodic)
variability, biotic conditions, and long-term trends,
and will require that the results can be extrapolated
to some larger target population of streams with
known confidence. The approach to these objectives
taken in the Phase l-Pilot Survey was to "overdesign"
a synoptic survey of streams focused on a relatively
small geographic area. That is to say, more samples
were taken during the Pilot Survey than were
expected to be necessary, in order to establish the
minimum acceptable sampling design needed to
meet the NSS objectives on a regional basis in Phase
I. This chapter illustrates, on the basis of Phase I-
Pilot Survey data, the types of results that could be
expected from a full scale synoptic survey of streams
and establishes the minimum number of samples
required to meet the Phase I project objectives.
The results presented here thus fall into two
categories: population distribution estimates and
stream classifications. We first consider alternative
methods for calculating and displaying population
estimates. Results from the three spring sampling
replicates and the summer sampling are compared
in order to determine the effect of sampling date
and replication on population estimates and stream
classifications. Chemical data from upstream and
downstream nodes are compared in order to
establish the desirability of sampling two points on
each reach during Phase I field work. Following these
discussions relating to survey design and data
analyses, we consider the ways in which the Phase
I data will likely be interpreted in order to provide
incremental information inputs to the assessment
process. This discussion necessarily focuses on the
Southern Blue Ridge as an example. Some caveats
and pitfalls also are noted and discussed.
The chapter then turns to the issue of classification,
the second major goal of the NSS. Examples of
classification based both on subjective (geographical
and geochemical) and objective (cluster analysis)
interpretations are presented. Some examples of
how the resulting classification might be used to
interpret historical data collected at the special
interest sites in the Phase l-Pilot Survey also are
discussed. It is important to note that any future
classification schemes must depend on the specific
nature of the intended research.
5.2 Population Estimates
Just as population distributions for geographic
characteristics were estimated from the first and
second stage samples (Table 2-2), the distributions
of chemical variables also can be estimated based
on the chemical data collected from the reaches in
the second stage sample. The generation of
cumulative population distribution curves for each
chemical variable satisfies the first two primary
objectives of the NSS (Section 1.3). Graphical and
tabular outputs showing the distribution estimates
for six primary NSS variables (pH, ANC, sulfate,
nitrate, chloride, and extractable aluminum) are
shown in Figures 5-1 through 5-6. The six variables
presented here are of particular interest because
they indicate present levels of acidity (pH) or potential
susceptibility (ANC); they are critical determinants
of toxicity commonly associated with atmospheric
acidification (extractable aluminum); or they involve
anions (sulfate and nitrate) that are commonly,
though not uniquely, associated with atmospheric
acids. Chloride was included as a possible indicator
of nonpoint source pollution (from agricultural
runoff, road de-icing, or wastewater effluent
disposal) in these streams. The region is too far from
the ocean to exhibit significant chloride deposition
from marine aerosols. While few interpretations can
be based on single variables alone, the distribution
of these variables within and among streams in the
probability sample is useful in evaluating the utility
and modifying the design of future NSS Phase I
activities. Similar distribution estimates for the
remaining NSS chemical variables can be found in
Appendix A of this report (Figures A.1 -A.23).
5.2.1 Graphical Displays
The distribution estimates in Figures 5-1 - 5-6 are
based on the mean value for each constituent at
the downstream sampling node of each reach over
the three spring visits. Water samples collected
-------
J
Ol
O)
1
1
en
•1*
CSl
rv
00
O)
CO
00
00
CO
>j
10
o
'^
10
1
c
01
c
3
_Q'
c
CD
i
s
$
2
3
2
0)
ffl
Q)
CO
D
CO
CD
3
•a
CD
CO
s'
en
CO
0)
13
CD
§
2.
CO
1.
CO
X
C
3
•£•
oo a> ? *. to
oof oo
F F §' F F
m m _ m m
TT *S" ^ S" S*
I X x X x
cc £pc c
2 3. 2. D 3
Z S 3. cJ »
"sj "^ "sj O) O>
M O O CO 00
-» cn co -j o>
M b b co bo
to o> co -j ^
ro b b co bo
vj-a-JO)05
10 O O CD 00
10 at w co co
H
O
S
CO
0
s
en
CO
CO
00
CO
o>
CO
01
10
en
3) z
CD C
£U ^
0 3
CD CD
o> -<
0
— *1
_ W
ffl CD S
~* s^ CD
CD > -i
<" 7
— • CD
ai
ifl
<
a>
— CD
fit
CD
Q)
TO
O
•o
c
o'
3
m
CO
3
Q)
•-»•
(D
cn
Cumulative Proportion
p p p p
N) ifc. O>
Cumulative Proportion
p p p p
k> A b> bo
c
Cumulative Proportion
p p p o o
b M '^ b» bo
ca
5
en
Cumulative Proportion
p p o
M '.* bo
.-r
T3
X
c
•032.
-------
Figure 5-2. Population distribution estimates for average spring downstream acid neutralizing capacity ANC in streams in
the NSS Phase l-Pilot Survey.
1.0
0.8
fo.6
I
|0.4J
3
E
00.2
0.0
. Data Subset = Downstream Spring Averages
Number of Reaches
^Proportion S X
Upper 95% Cl
1.0
0.8
c
o
o 0.6
a.
I
lo.4
o
0.2
0.0
100
200
300
400
500
600
Acid Neut. Capacity (pieq L"1)
Variable = Acid Neut. Capacity
Water Surface Area
Proportion ^ X
Upper 95% Cl
100 200 300 400 500
Acid Neut. Capacity (//eq L"1)
600
1.0
0.8
to.6
I
§
10.4
3
E
5 0.2
0.0
Length'of Reaches
™ Proportion ^ X
Upper 95% Cl
1.0
§0.8
l"
ra
0.4
O
100
200
300
400
500
0.2
0.0
600 0
Total Drainage Area
——Proportion ^X
Upper 95% Cl
Acid Neut. Capacity (//eq L"1)
100 200 300 400 500 600
Acid Neut. Capacity (//eq L~1)
Population Estimates
Totals
20 %ILE Gueq L"1)
40 %ILE (peq L"1)
Median Oueq L"1)
60 %ILE fc/eq L"1)
80%ILE(peqL"1)
Sample Sizes
Actual Unique
Number of
Reaches
2021
86.64
102.59
119.61
134.26
197.69
Effective
Water
Surface Area
(Hectares)
4633
65.03
87.73
98.58
110.03
216.11
Min
Reach
Length
(km)
8963
72.67
102.48
115.53
138.41
217.53
Sample Weighted Statistics (//eq
Max Mean
Total
Watershed Area
(sq km)
51215
72.84
89.41
98.03
108.09
181.45
L-1)
SD
54
54
84
16.18
1710.5
252.02
399.14
57
-------
01
00
Ol
C7I
2
_,
(0
bo
-j
8
^
Co
u
00
CO
00
CO
IO
'••J
en
Actual
c
&
c
CD
1
I
i'
^
5
0)
X
CD
D>
CO
o
CO
D>
3
^
5"
CO
CO
CO
•o
CD
CD
CQ'
IT
CD
Q.
S
S
O
0>
^°
2
— '
§O> 5 •&• 10
0 CD 0 0
F F | n F"
rn m J mm
fffff
-^ — ^ — ^ ^-^
on io io io -»
oo vi M row
OD en oo co 01
at io >j vj oo
*IO 10 10 -»
CO 00 CO 10 00
coco ^ oo o
ro -> ~j ^j en
01 io to io -»
CJI CO CJI 10 00
00 -« 10 O 10
co en io o> oo
A 10 10 10 -•
-•^ CO 10 00
CJI -J CO CJI -*
vj ^1 -> IO CO
1
n"
s
en
CO
CO
00
CD
o
CO
en
NJ
01
z
S i
sT?
8 o
_ CO
ill
I >?
£* o
_ r~ ^g
^ 3 2
""" 5 3"
D>
i"!^
^" ® w
3 Q- *~~
!
TJ
3
»"
5'
1
3
u
0
w
Cumulative Proportion
pop
*> b>
Cumulative Proportion
o o o P
io » b> »
en
S
CD
CD
r- _,
Cumulative Proportion
P p p p P
o io *>. g> <°
p
b
Cumulative Proportion _
P o P
c_
i?
s
CD
10
CJI
s-
oi
8-
S
s
8
; I
III
^ -i CQ
o
2 3
CD S
" 3
cf S
S o
8 2
-------
Figure 5-4. Population distribution estimates for average spring downstream nitrate concentrations in streams in the NSS
Phase I-Pi lot Survey.
1.0
§ 0.8
c
o
I 0.6
>
+••
J5
0.4
o
0.2
0.0
Data Subset = Downstream Spring Averages
Number of Reaches "'
Proportion & X
Upper 95% Cl
10
20 30
Nitrate GueqL~1)
40
50
1.0
Variable = Nitrate
cO.8
o
C
o
Q.
1 0.6
I
§0.4
O
0.2
0.0
Water Surface Area
Proportion § X
Upper 95% Cl
10
20
30
Nitrate (y^eq L )
40
50
1.0
0.8
c
o
I 0.6
£
.1
o 0.4
3
E
o
0.2
0.0
Length of Reaches
;, Proportion g X
Upper 95% Cl
10
20 30
Nitrate (/ueq L"1)
40
1.0
50
0.8
c
o
C
o
§•0.6]
Q_
a 0.4
3
E
O
0.2
0.0
Total Drainage Area
Proportion S X
Upper 95% Cl
10
20 30
Nitrate (/jeq L"1)
40
50
Number of
Reaches
Population Estimates
Water
Surface Area
(Hectares)
Reach
Length
(km)
Total
Watershed Area
(sq km)
Totals
2021
Actual
54
Sample Sizes
Unique
54
Effective
84
4633
8963
51215
20 %ILE (Aieq L"')
40%ILE(//eqL~1)
Median (/*eq L'1)
60%ILE&ueqL'1)
80%ILE(/ueqL~1)
2.99
4.89
7.55
10.85
23.38
2.65
7.51
9.91
10.84
16.63
2.86
4.98
6.98
10.09
20.36
2.01
9.57
9.89
10.82
13.77
Min
0.66
Sample Weighted Statistics (yeq L 1)
Max
41.83
Mean
12.08
SD
10.55
59
-------
Figure S-5.
1.0
0.0
Population distribution estimates for average spring downstream chloride concentrations in streams in the NSS
Phase 1-Pilot Survey.
Data Subset = Downstream Spring Averages
1.0
Number of Reaches
— Proportion fe X
— — Upper 95% Cl
20
40 60
Chloride (pieq L"1)
80
0.8
Variable = Chloride
Water Surface Area
Proportion § x
Upper 95% Cl
20
40 60
Chloride 0"eq L"1)
80
100
o.o 4-
0
Length of Reaches
Proportion S X
1.0
0.8
20
40 60
Chloride (#eq L"1)
Total Drainage Area
Proportion fe X
Upper 95% Cl
20
40 60
Chloride (/aeq L"1)
80
100
Actual
54
Sample Sizes
Population Estimates
Totals
20 %ILE U/eq L"1)
40%ILE(A
-------
Figure 5-6. Population distribution estimates for average spring downstream extractable aluminum concentrations in streams
in the NSS Phase I-Pilot Survey.
Data Subset = Downstream Spring Averages Variable = Extractable Aluminum
1.0
0.0
Water Surface Area
Proportion ^ X
Upper 95% Cl
Number of Reaches
Proportion 2 X
Upper 95% Cl
5 10 15 20
Extractable Aluminum (fig L
5 10 15 20
Extractable Aluminum {/jg L"1)
Length of Reaches
Proportion ^ X
Upper 95% Cl
Total Drainage Area
Proportion ^ X
Upper 95% Cl
Extractable Aluminum (fjg L 1)
5 10 15 20
Extractable Aluminum (/jg L~
Population Estimates
Number of
Reaches
Totals
20 %ILE (iJtg L'1)
40 %ILE (jug L"1)
Median (jug L"1)
60 %ILE (jug L"1)
80 %ILE (jug L"1)
Sample Sizes
Actual Unique
54 37
2021
2.33
3.00
3.24
3.51
4.93
Effective
84
Water
Surface Area
(Hectares)
4633
2.33
3.20
3.50
3.67
6.18
Min
1.20
Reach Total
Length Watershed Area
(km) (sq km)
8963
2.33
3.00
3.24
3.67
5.51
Sample Weighted Statistics (jug L~1)
Max Mean
23.33 3.93
51215
2.58
3.37
3.47
3.67
5.44
SD
2.70
61
-------
during seven rainfall events that occurred during the
third sampling period have been excluded.
The curves represent the target reach population
distribution estimates for the various water chem-
istry variables in terms of number of reaches (upper
left), stream length (lower left), stream surface area
(upper right), and a preliminary discharge index
based solely on drainage area (lower right). The four
types of distributions in Figures 5-1 - 5-6 are
interpreted similarly. Values on the vertical axes in
the graphs represent the proportion of the total target
reach attribute (reach number, length, surface area,
or discharge index) within the survey area estimated
to have a value for any particular chemical variable
greater than or equal to the corresponding value of
that variable along the x axis (less than or equal
to for pH and ANC). The pH and ANC plots were
ordered differently because it is the lower values
that are of greater environmental concern, rather
than the higher values.
The dashed lines above the cumulative distribution
curves represent the 95% upper confidence bound
for the estimate. The NSWS estimates thus are
viewed from a "worst case" standpoint, i.e., the
maximum percentage of lakes or streams in the
respective target populations that could reasonably
be expected to be below some particular value for
pH or ANC. An alternative viewpoint might be based
on the minimum 95% confidence bound, i.e., the
minimum percentage that could be expected to
exhibit a particular pH or ANC concentration. A lower
one-sided 95% bound would appear symmetrical to
the upper bound about the cumulative distribution
curve.
The tabular data at the bottom of each figure include
values for the four quintiles and the median of each
distribution. These figures permit rapid quantitative
comparisons and provide an estimate of the total
resource in the target population based on the
second stage sample. For example. Figure 5-1
indicates that 20% of the 2021 reaches in the
Southern Blue Ridge target population are estimated
to have "index" pH values below 6.86 (the first
quintile), while half were below 7.03 (the median).
If the proportions are based on kilometers of reach,
20% of the 8963 km of streams in the target
population were characterized as having a pH less
than 6.84, and so on. Sample means and standard
deviations, weighted to account for the non-uniform
inclusion probabilities of each reach, are also shown.
Actua!, unique, and effective sample sizes (Overton,
1985) also are included in each figure. The actual
sample size is the number of reaches in the second
stage sample; the unique sample size is the number
of uniquely occurring values for each variable, and
the effective sample size is the number of grid points
(including non-target reaches) associated with the
second stage sample.
5.2.2 Alternative Measurement Variables
The representation of alternative distributions based
on four different reach attributes is provided to
stimulate discussions on the relative merits of these
(or other) forms of expressing the distributions of
the chemical variables for future Phase I results.
Frequency distribution curves indicate the proportion
of the total number of reaches which were above
or below some reference value and are relatively
easy to understand. Frequency distributions,
however, treat reaches of different length and
discharge equally. For this reason they may present,
for example, a misleading picture of low ANC waters
if ANC is correlated with reach length, drainage area,
discharge, or position in the discharge network.
Expressing the estimates as length distributions
(lower left frame) gives a better picture of the total
resource, but still treats large and small streams of
the same length equally. The length distribution
estimate, as presented in the figures, assumes that
the value of the chemical variable is uniform
throughout the length of the reach, and equal to
the value at the lower node. The extent to which
this approximation may be reasonable is evaluated
below. An alternative is to collect data at more than
one point on the reach and to interpolate the results
to a number of segments within each target reach.
A crude aquatic habitat area index was calculated
by multiplying reach length by mean stream width
measured at the two sampling nodes. The resulting
index ("water surface area") was used to construct
the areal distribution curves in the upper right frames
of Figures 5-1 - 5-6. These areal distribution curves
indicate the proportion of combined reach water
surface area above or below some reference value
of a chemical variable. Fisheries managers fre-
quently refer to "weighted usable area" as a
measure of the amount of aquatic habitat available
to any particular species. Weighted usable area is
often calculated on the basis of velocity, depth,
substrate, and other physical variables (e.g., Bovee
and Cochnauer, 1977). While it can be seen that
such a measure would quantify the "usable" portion
of the total aquatic surface area indexed in the Pilot
Survey, the calculation of that habitat portion for
the sample of 54 streams was beyond the scope
of work in the Survey.
The lower right frames in the distribution estimate
figures display preliminary discharge index distribu-
tions, which were calculated solely on the basis of
total watershed area (ai + ag) of the target reaches
(and are thus labeled). When multiplied by an
62
-------
appropriate net precipitation index value, discharge
index values estimate the discharge at the down-
stream nodes of target reaches. Total watershed area
(discharge index) distribution curves indicate, on the
basis of watershed area, the proportion of target
reaches above or below a given reference value of
a chemical variable.
The total population of target reaches within the Pilot
Survey area includes drainages ranging from 1 to
155 km2, with smaller drainages nested within larger
ones. The estimated total watershed area of 51,215
km2 for the target reaches in Pilot Survey area,
therefore, includes drainage areas counted more
than once and should not be construed to represent
the total land area drained by the network of target
reaches. Once adjusted for runoff, the total discharge
index would provide an estimate of the sum of the
discharges at the downstream nodes of ali target
reaches in the population, summing the discharge
of all reach segments within a hierarchical network
of target reaches.
While a stream discharge index is not a particularly
good measure of available fish habitat, the prelimi-
nary discharge index distributions, once refined,
would provide a useful picture of the chemical
composition of water moving through the target
stream population. The best interpretation of the
curves as they are presently shown is that they
estimate the instantaneous, discharge-weighted
distribution of the chemical variables over the
downstream nodes of all target reaches in the
population, assuming that discharge is proportional
to drainage area only. The discharge index distri-
bution estimates will be revised upon completion of
a predictive model for net precipitation that takes
into account spatial differences in precipitation,
evapotranspiration, and runoff. The accuracy of such
a revised discharge index (or the uncorrected index
as presently expressed) would be reduced in
drainage networks where groundwater is delivered
across topographic drainage divides (Toth, 1963).
Another useful target reach attribute would be the
concentration of some chemical variable (C) in runoff
contributed to a reach by direct drainage between
an upper and lower sampling location. For headwater
reaches (R = 1), this concentration would be equal
to that in the water at the downstream node. For
downstream reaches (R > 1), an appropriate
concentration variable (C8l) would be calculated
using a mass balance between the upstream and
downstream nodes, as estimated by the measured
chemical concentrations and measured or estimated
(indexed) discharges:
C.. =
- (Qo-Cp) - (Qu-Cu)
(Qo - Qu)
[5-1]
where Q and C refer to discharge and concentration,
respectively, and the subscripts U and D refer to the
upstream and downstream nodes, respectively. Such
estimates could only be calculated for the Phase I-
Pilot Survey summer sampling period and thus are
not included in the distribution figures.
Comparisons of the curves and quintile values for
each of the variables in Figures 5-1 - 5-6 show very
similar distributions. Apparently, there is very little
effect of the choice of a particular distribution index
on the interpretation of each of the six NSS chemical
distributions. Distribution estimates based on the
discharge index show the greatest differences, but
this may be caused by the incomplete nature of the
index. The similarity of the curves suggests that, at
least on a region-wide basis, there was little (if any)
correlation between concentrations of the NSS
primary variables and stream length or drainage
area, which was subsequently confirmed using
univariate and multivariate regression analyses.
Elevation is thus far the only geographic variable
tested that has shown any significant relationship
to pH or ANC concentrations, and even this
relationship proved too weak to be of any predictive
or descriptive value in partitioning these distributions
into a priori categories of interest.
5.2,3 Reference Values
A potentially useful way of expressing population
estimates is with respect to the proportion or number
(length, area, etc.) of reaches which are above or
below some particular chemical reference value. A
reference value could represent a criterion value
established on the basis of toxicological studies (e.g.,
a TLCsofor inorganic monomeric aluminum) or a legal
standard. No widely accepted criteria are presently
available for evaluating the quality of waters with
respect to acidification by atmospheric deposition,
but reference values can also be based on common
usage. For example, waters with negative ANC
values are acidic by definition, and those below 50
fjeq L~1 are often cited as being highly susceptible
to acidification (Pfeiffer and Festa, 1980; Linthurst
et al., 1986, Table 5-2). Alternatively, reference
values may simply partition a population into useful
categories. Such categories may be artificial, such
as (logarithmically) evenly spaced pH increments
(e.g., 4.0,4.5,5.0, 5.5), or they may represent natural
groupings (clusters) of geochemically similar waters,
as explained below. Although any partitioning
scheme provides an important starting point for most
detailed analyses, care must be exercised in
interpreting population distributions based on
criteria associated with any single chemical variable.
Examples of such partitioning of distribution
estimates for the Phase l-Pilot Survey streams are
63
-------
Table 5-1 .
Variable
Phase I- Pilot Survey Length Distribution Estimates Associated with Reference Values Based on Natural Univariate
Groupings of Streams (except where noted for ANC)*
Reference
Value
Proportion
Population Estimate*
Total Length
(km)
Upper 95% C.I.
(km)
pH
ANC
Gueq L~1)
Sulfate
Nitrate
0"eq I/1)
<6.7
<7.6
<25
<50a
<200a
<250
>40
>80
>120
>20
>35
7.4
87.0
92.0
1.2
6.3
48.0
74.4
84.5
30.8
10.7
3.7
40.7
22.9
2.9
662
7800
8247
108
561
4304
6666
7573
2761
957
331
3648
2054
259
1244
9373
9816
283
980
5637
8353
9155
3820
1625
710
4899
3088
557
Chloride
(//eq L'1)
>50
>100
>200
9.5
2.4
0.9
852
215
77
1356
455
190
*AII estimates based on spring average reach chemistry at the downstream sampling nodes.
"Values provided to allow comparison with commonly cited sensitivity criteria (Table 5-2 in Linthurst et al., 1986).
shown in Table 5-1. The reference values used in
Table 5-1 were derived by ordering sample sites
according to measured values of each chemical
variable (ordination) and searching for natural gaps
in the data (the interpretive value of this procedure
will become clear below). Because of the absence
of low pH values, there appears to be little value
at this time in partitioning these distributions into
a priori categories of interest. Extractable aluminum
concentrations were too low to make partitioning
meaningful and are not represented in Table 5-1.
Commonly cited ANC reference values (50 and 200
//eq L~1) are included to allow comparisons with other
data bases. Because the measurement variables do
not have a strong effect on the distribution estimates,
only those based on reach length are reported here.
New reference values that would aid in comparing
the Southern Blue Ridge with other NSWS target
populations will be computed for future data reports.
5.2.4 Sample Timing and Frequency
During the design phases of the NSS, concern was
expressed that temporal chemical variability may be
so high during the spring that more than one
sampling visit would yield widely diverging popu-
lation estimates. Temporal variability could include
both hourly/daily components due to hydrologic
events, and weekly/monthly components due to
vegetational (e.g., leafout) and climatic (e.g., soil
warming) effects. In order to determine the effect
of these variance components on the outcome of
a synoptic survey, three biweekly spring samples and
one summer sample were collected during the Phase
l-Pilot Survey without regard to present or antece-
dent meteorologic conditions.
Rainfall events were observed to cause temporal
variability in NSS target population streams. Table
5-2 demonstrates the effect of seven hydrologic
events on two primary NSS variables, pH and ANC.
In each case, identifying the occurrence of the event
was predicated on the occurrence of previous
precipitation, an increase in stream stage, and an
increase in turbidity and total afuminum indicative
of increased runoff in the watershed (Section 4.5.6).
ANC decreased by an average of 23 percent (range
-17% to -35%) and pH decreased by almost 0.2 units
(range -0.01 to -0.37 units) during the events.
Summer events were also typically characterized by
reduced pH and ANC concentrations, but the effects
were difficult to quantify precisely without a summer
benchmark against which to compare stage heights
during suspected events. The average spring ANC
depression due to hydrologic events would be
sufficient to move a given stream approximately 20
percentile units, relative to its position on the curves
in Figure 5-2, and thus could have a substantial
impact on stream classification, as explained below.
Once the episodic effects were removed, the
remaining temporal variance had little effect on the
64
-------
Table 5-2. Effects of Rainfall Events* on ANC and pH at
Seven Downstream Phase t-Pilot Survey
Sampling Sites
Stream ID
7702
7819
7831
8809
8902
8904
8906
Mean(±1 SO)
ANCa
{/ueq LM)
967
-21%
201
-19%
274
-22%
75
-17%
41
-35%
132
-29%
52
-28%
-23 ± 6%
pHa
(Units)
8.22
-0.29
7.02
-0.14
7.00
-0.12
6.78
-0.01
6.67
-0.17
6.76
-0.37
6.25
-0.24
-0.1 9 ±0.1 2
Units
A Stage
(Ft)
+0.25
+0.51
+0.55
+0.28
+0.51
+0.63
+0.30
• *Events are predicated on at least three of the following 7.5 cm
change in stage, evidence of precipitation on Data Form 4,
precipitation within 1 day at the nearest NOAA meteorological
station, or an increase in turbidity and total aluminum.
"ANC and pH represent values measured during episodes, and
percentages represent depressions below corresponding mean
spring values.
shape of the distribution estimates over the three
spring sampling periods. Figures 5-7 - 5-9 display
cumulative frequency curves based on the four
different sampling intervals, with the seven spring
episodes removed from the data base as explained
in Section 4.5.6. For all variables, the length
distribution estimates based on the three spring
samples are virtually superimposed. The total
extractabie aluminum concentrations are sufficiently
dose to the analytical decision limit (Table 4-8) that
differences probably include a substantial compo-
nent of analytical error (variance). Week-to-week
variation in stream chemistry during the spring
apparently has minimal effect on the distribution of
these species in streams of the target population,
if the large rainfall events can be separated from
typical spring flow conditions.
Seasonal variation, however, was sufficient to alter
the distribution estimates for some of the NSS
primary variables. Summer distributions were
virtually identical to spring distributions for pH,
sulfate, chloride, and total extractabie aluminum.
ANC increased substantially from spring to summer,
especially in streams with ANC < 250 jueqL'1. Nitrate
distributions were similar during both sampling
intervals, except for streams near the low (< 10
fjeq L~1) end of the distribution.
We also approached the question of differences
among the sampling periods by calculating paired-t
statistics for the six primary variables. Table 5-3
presents results based on reaches with ANC less
than 250 /jeq L~1, unweighted for reach inclusion
probability. High ANC reaches were excluded to avoid
possible differences in geochemical patterns
correlated with land use or geology. No calculations
were performed for extractabie aluminum, because
a majority of the values were below the quantitation
limit; pH calculations were performed on the log-
transformed value rather than on hydrogen ion
activity.
Chloride and pH were the only variables to increase
significantly between the three spring sampling
intervals, but the quantitative changes are small and
probably unimportant from a water quality assess-
ment perspective. The 12% increase in chloride,
followed by a subsequent 11% decline, probably
reflects the unusually dry conditions during the
second (SP2) sampling interval. ANC exhibited a
substantial 36% increase between the average
spring and summer sample, and pH increased by
0.04 units. The limitation of the t-test, which is a
test only for difference in the means, is illustrated
by nitrate, which exhibits a differential shift in
different parts of the distribution (Figure 5-8).
5.2.5 Spatial A spects of Reach Chemistry
The relatively rapid downstream flow which is
characteristic of streams yields a set of sampling
problems different from those encountered in the
study of lakes. Sampling at a point in the center
of an unstratified lake is widely accepted as providing
a useful index value for the central water column
chemistry. A single point sample at the downstream
node of a stream reach, however, may not provide
a particularly good representation of the chemistry
of the entire reach lying above it. Chemical
composition may change along the reach due to
instream processes (e.g., primary production),
confluence with streams not represented on
1:250,000-scale maps, and lateral inflows from
springs and seeps feeding the reach. In describing
a population of stream reaches, it is often the length
(or some transformation such as habitat area) of
reaches characterized by some particular chemical
value that is of interest. Any variation along the reach
should be accounted for at a level of resolution
appropriate to the population estimate.
During the Phase l-Pilot Survey, 23 reaches were
sampled at their upstream and downstream nodes
65
-------
Cumulative Proportion
Cumulative Proportion
p
o
p
to
p
b>
p
bo
s -
10
8
o
m
•a
o>
2. a>
I 8
8-
B
«
(D
0)
3
VI
a
•
?
I
3
O.
O
IP
0)
t
B
(O
»
g.
3
a
-------
Cumulative Proportion
p
bo
p
b
p
CD
O
CO
CO
o
O)
SI
*
3"
w w S.
to
(71
g -
0}
c
0)
ID
I 8
10
CJI
§ -
8
I 3
! t
c/> 6) S.
g ^ 3D
1 5- 8
m ^ °
»
in
a
0)
CO
§
o
1
I
ffi •
3
V)
m
•
8.
II
3-5
. »
88
3 =
•0.0
="5
51
i.5'
oT 3
-------
Figure 5-9. Comparison of population length distribution estimates for chloride and extractable aluminum based on the
three spring and one summer sampling intervals.
Data Subset = Four Downstream Sample Periods
1.0
0.8 -
I 0.6 -
o
BL
| ««J
i
o
0.2 .
0.0
Length of Reaches
. Spring
— Summer
20
40 60
Chloride (jueq L~1)
i
80
100
1.0
0.8.
c
o
I
I °-6'
Q.
0.4.
0.2,.
0.0
Length of Reaches
• Spring
——— Summer
10 15
Extractable Aluminum (/ug L~1)
7
20
1
25
30
68
-------
Table 5-3. Statistically Significant (p = 0.05) Differences
Between Mean Concentrations of Primary
Variables Between Spring (SP1, SP2, SP3) and
Between Summer (SU) and Average Spring (SP)
Sampling Intervals (downstream nodes) for
Streams with < 250 Aieq L"1 ANC
Mean Concentration Difference8
Chemical
Variable
ANC
PH
Sulfate
Nitrate
Chloride
SP2 - SP1
n = 38
NS
NS
NS
NS
+ 12%
SP3 - SP2
n = 39
NS
+0.05 units
NS
NS
-11%
SU-SP
n = 34
+36%
+0.04 units
NS
NS
NS
"Comparisons are not weighted according to reach inclusion
probability; episodes have been excluded.
NS = not significant at p = 0.05.
during the third spring sampling interval, and all sites
were sampled at both nodes during the summer.
The purpose was to determine:
1. Whether chemistry at the opposite nodes of a
single reach was substantially different.
2. Whether downstream chemistry could be
adjusted to reflect the entire reach chemistry
on the basis of a limited number of upstream
samples.
The answer to objective (1) was approached much
as the question of temporal variability was in the
previous section.
Figures 5-10 through 5-12 show frequency distri-
bution estimates for the six primary variables based
on the chemistry at the upstream versus the
downstream sampling nodes during the summer
sampling interval. With the exception of sulfate and
extractable aluminum, substantially lower values of
all variables were observed at the upstream sites.
The curves indicate only that the frequency distri-
bution estimates are affected by the sampling
position along the reach during the summer
sampling interval. An insufficient number of streams
was sampled at both nodes to construct meaningful
comparative spring distribution estimates.
It should be noted that the upstream node chemistry
could be used to indicate the chemical properties
of an unspecified target population of streams
draining smaller watersheds than those represented
by the actual NSS target population. The distribution
estimates based on upstream chemical data in ,
Figures 5-10 - 5-12 represent this unspecified >
population of streams. Median values (expressed as
proportions) for pH, ANC, nitrate and chloride would
be lower for this unspecified population. The NSS
geographic site data describe the drainage areas (a3)
for this population of smaller streams, but no
information on the length of streams in this
population is presently encoded. We are presently
investigating alternatives for making distribution
estimates for these lower order streams.
The effect of sampling position can also be seen in
Table 5-4 using paired-t comparisons for the sites
not exhibiting episodes on the last spring and
summer sampling dates. Again, sites with spring
ANC averages > 250 //eq L~1 were excluded and
aluminum has been excluded due to the low
concentrations observed. Virtually all primary
variables showed statistically significant within-
stream spatial differences, many of which were
numerically large enough to affect interpretation of
the data. The tabulated data also show that the
spatial sampling effects demonstrated in Figures
5-10 through 5-12 were not restricted to the summer
sampling period.
Given that the chemistry at the upstream and
downstream nodes of the reaches in the target
population were clearly different, it remained to be
determined whether one value could be inferred by
calibration on the basis of the other. Simple linear
regression equations were calculated and bivariate
plots were investigated visually for both spring and
summer data, with and without episodes removed,
both including and excluding high (> 250 /yeq L'1)
ANC reaches. Whereas most of the summer
relationships were highly significant, the 95%
confidence intervals about the predicted values were
on the order of ± 50% or more and this restricts
the utility of prediction of upstream values using
downstream chemical data.
5.2.6 Interpretation for Regional Assessments
One of the primary objectives of Phase I of the NSS
is to provide population distribution estimates of the
number of acidic (low pH) and potentially susceptible
(low ANC) streams in each NSS subregion. Ways
in which the population estimates can be constructed
and used to estimate the characteristics of the target
populations were presented and discussed in Section
5.2.2. Although not a specific objective of the NSS,
a relevant issue is how the resulting estimates could
be interpreted to provide incremental information
useful in a regional assessment. In this section we
will demonstrate, by example, some potential uses
of the data, as well as some caveats and potential
pitfalls in the interpretation process.
It is apparent from Figure 5-1 - 5-6 and Table 5-1
that the Phase l-Pilot Survey characterized a very
high proportion of target stream reaches and target
stream length as possessing low acid neutralizing
capacity (ANC). Half of the reaches and half of the
69
-------
Figure 5-10. Comparisons of frequency distribution estimates for pH and ANC in Phase I—Pilot Survey streams based on
upstream versus downstream sampling locations during the summer sampling interval.
Data Subset = Summer Averages
1.0
0.8.-
c
o
o
Q.
I
0.6 -
S 0.4-
3
O
0.2.
0.0
Number of Reaches
. Lower Site
——— Upper Site
8
pH (pH Units)
1.0
0.8-
c
o
C
o
Q.
O
£ 0.6
9
I °-4
0.2.
0.0
Number of Reaches
. Lower Site
———' Upper Site
i
100
200
300
400
500
600
Acid Neut. Capacity (/-eg L )
70
-------
Figure 5-11. Comparisons of frequency distribution estimates for sulfate and nitrate concentrations in Phase I-Pilot Survey
streams based on upstream versus downstream sampling locations during the summer sampling inverval.
0.0
25
Data Subset = Summer Averages
Number of Reaches
. Lower Site
Upper Site
50
75 100
Sulfate (A/eq L'1)
125
150
200
1.0
0.8
c
o
I
I 0.6
o.
O
0.4
0.2.
0.0
Number of Reaches
— . Lower Site
Upper Site
10
20
Nitrate iueq L"1)
I
30
40
50
71
-------
Figure 5-12. Comparisons of frequency distribution estimates for chloride and aiuminum concentrations in Phase l-Pilot
Survey streams based on upstream versus downstream sampling locations during the summer sampling interval.
Data Subset = Summer Averages
1.0.
0.8-
•| 0.6.
o
I °*1
3
O
0.2 .
0.0
I
20
Number of Reaches
. Lower Site
Upper Site
i
40
I
60
r
80
100
Chloride (fjeq L )
1.0
c
o
't
o
Q.
O
o
0.8.
0.6.
0.4.
0.2 .
0.0
Number of Reaches
• Lower Site
Upper Site
10
15
20
25
30
Extractable Aluminum (/jg L' )
72
-------
Table 5-4. Comparison of Upstream/Downstream Chem-
istry During the Third Spring (SP3) and Summer
(SU) Sampling Intervals, Based on a Paired t-
Test with Differences Weighted to Reflect
Inclusion Probabilities (wi)*
Sampling Interval
Chemical
Variable
ANC
pH
Sulfate
Nitrate
Chloride
SP3
n = 14
+ 14%
+0.06 units
+ 15%
+33%
IMS"
SU
n = 31
+26%
+0.13 units
+7%
+37%
+23%
"IMS = not significant at p = 0.05.
*0nly streams with mean spring ANC < 250 jueq/l are considered,
and samples collected during hydrologic events have been
excluded.
total reach length were estimated to contain water
of less than 120 //eq L"1 of ANC. Nearly 75% of the
estimated length distribution was below the
reference value of 200 yueq L"1, a value often cited
as separating potentially sensitive from relatively
insensitive systems (Linthurst et al., 1986). However,
only 6.3% of the target reach miles are expected
to have ANC concentrations less than 50 //eq L"1,
a value that has been used to identify particularly
acid-sensitive waters (Pfeiffer and Festa, 1980).
Despite the preponderance of low ANC in the target
population, fewer than 8% of the reaches (upper 95%
Cl = 14%) exhibited non-episodic, average spring pH
values below 6.7. Even when the episodes and the
upstream node measurements were included in the
data, no samples were collected during the Phase
l-Pilot Survey that exhibited a pH value below 6.0.
A "worst-case" estimate can be made for the spring
index pH value in the target population streams by
choosing a reference value at the low end of the
range observed during the survey, and calculating
the upper 95% confidence bound on the estimate.
The exercise leads us to conclude that, with 95%
confidence, less than 3.2% of the combined length
of streams in the target population (based on the
mean downstream spring average pH with episodes
excluded) exhibited a pH below 6.4 during 1985.
While it would be helpful to calculate the length of
stream reach below some more meaningful value
(e.g., 5.0), the method used to estimate confidence
intervals cannot be applied below the minimum index
value occurring in the sample (6.38). These obser-
vations and estimates are based on the closed-
headspace pH measurements made at the mobile
field laboratory, which were of consistently high
quality throughout the project. All pH values were
well above the range of 5.3 to 5.7 frequently cited
as representing geochemical neutrality. Although
this analysis does not address the question of
whether pH values in these streams would be
different in the absence of acid deposition nor what
the lowest pH values were experienced during the
spring, the "index values" are certainly not in a pH
range that has been associated with deterioration
of coldwater sport fisheries in the past (Howells,
1984; Magnuson et al., 1984). However, some
estimate of transient chemical changes that may
occur during hydrologic episodes is needed before
a critical evaluation of chemical habitat quality can
be complete.
Consistent with distributions for pH dominated by
neutral conditions, the median extractable aluminum
concentration (approximately 3 fjg L~1) was barely
above the analytical detection limit, and the
maximum concentration was only 23 yug L"1.
Inorganic monomeric aluminum concentrations,
estimated as the difference between total extractable
and non-exchangeable aluminum fractions, were
below the decision limit in virtually all samples and
are, therefore, not reported. Total extractable
aluminum concentrations in the range 2-20 //g L"1
are one to two orders of magnitude lower than the
lowest concentrations at which short-term exposure
of selected fish species have been observed to
produce significant mortality (Schofield and Trojnar,
1980; Baker, 1981; Henriksen et al., 1984; Johnson
eta)., 1985).
The foregoing statements exemplify the kinds of
univariate interpretations that represent one level
of incremental information that can be used to satisfy
the NSS primary objectives related to description.
These statements characterize the Southern Blue
Ridge as an area dominated by stream waters of
moderately low acid-neutralizing capacity, but in
which chronic acidic conditions are relatively rare.
Although we believe this description to be funda-
mentally accurate, several caveats should be borne
in mind.
First, the target population, represented by the
sampled population, focuses on second to fourth
order (Strahler order based on blue line represen-
tations on 1:24,000-scale topographic maps)
reaches and thus does not include the first order,
headwater reaches which might be expected to
be"early warning"indicators of acidification. Furth-
ermore, spring reach chemistry is characterized on
the basis of the chemistry at the downstream node.
Based on the limited spring data in Table 5-4, the
upper nodes of the target reaches might be expected
to be 14% lower in ANC than the corresponding lower
nodes, and 0.06 units lower in pH. These data
suggest that the target population estimates for ANC
and pH based on reach length are somewhat high,
although not markedly so. Almost half (44%) of the
upper nodes of the target population reaches drain
73
-------
first order reaches (1:24,000 blue line), and thus the
water draining from these headwater catchments
is, as expected, lower in ANC and pH than that
draining the larger catchments represented by the
downstream nodes of the target population. Again,
the differences are not large. This project did not
measure the chemistry at the upper limits of flowing
water, but such streams are likely to be extremely
small, and many are likely to be ephemeral.
An alternative sampling design could have focused
on these extremely small target reaches, with the
objective of detecting early signs of acidification.
However, the difficulty of access, together that the
possibility that many of the second stage sample
reaches may have been dry at their upper node or
misrepresented on maps would have greatly
increased the cost per site of field sampling. This
translates to a smaller sample, or less areal coverage
per research dollar, especially given the fixed cost
component of a project. Add to this factor the highest
degree of chemical variance in the smaller catch-
ments resulting from increased heterogeneity and
shorter hydraulic residence times, and the efficiency
of the survey decreases accordingly. By including
(or focusing on) first order streams, we might
discover more acidic streams, but they would
represent a very small percentage of the resource
at risk, and the confidence bounds on the estimate
might well overlap those obtained with the present
design. If the chemistry of the smallest streams in
a critical component of the assessment activity, an
alternative, more efficient approach is to focus future
headwater sampling on areas found to contain
significant percentages of low ANC systems in Phase
I.
In a temporal setting, the population estimates here
strictly refer to spring and summer of 1985.
Precipitation was approximately half of normal
during the spring, but 5% above normal during July.
If the storm tracks were also unusual, 1985 may
represent an unusual year. Some idea of the
representativeness of any single year synoptic
sample can be gained by comparing historical
records for the special interest sites sampled in each
NSS subregion. These data were not available for
the Southern Blue Ridge sites as this report was
being prepared.
Spring may not represent typical annual low pH or
low ANC conditions in the study area. Discussions
with local investigators indicated that (with the
possible exception of late winter) spring did appear
to represent typically low pH and ANC conditions
in the region, however, and that the sensitive swim-
up fry life stages of salmonids in the region were
present at that time. Therefore, lower pH conditions
in mid-winter (if they occur) may be of less ecological
interest. The population estimates do not include the
effects of episodes. Although episodes are likely to
be critical determinants offish survival, the duration
of such events may be extremely short in the
Southern Blue Ridge, and thus extremely difficult
to measure in a synoptic survey. Even though several
hydrologic events were sampled in the Survey (none
of which produced pH depressions of greater than
0.4 units or pH values below the minimum reported
above. Table 5-2), the study design does not provide
an estimate of the minimum pH experienced by these
streams during rainstorms, nor of the temporal or
spatial extents of pH depression associated with
rainstorms. Rather than attempting to quantify these
transient effects in a synoptic survey, it is planned
to target episodes monitoring at typical low ANC sites
in future studies, and subsequently to expand the
results via the Phase I population estimate to the
target population in each subregion. This plan
depends on classification of Phase I sites, and is
discussed further in Section 5.3.
Interpretation of the population estimates also
involves a philosophical viewpoint. For example, one
person may view a hypothetical population estimate
of 1% of combined stream length in acidic condition
as acceptable, while another may view the same
estimate expressed as 200 km as quite the opposite,
especially if the 200km coincides with the only blue-
ribbon trout streams in an area. While the first
consideration is beyond the scope of statistical
estimation, additional data analyses employing maps
and overlays may be useful. For example, the
geographic distribution of ANC for the target
population streams is depicted in Figure 5-13 with
respect to the 50 and 200 jueq L~1 reference values
noted earlier. The map shows that the lowest ANC
reaches are focused in the highlands in the north
central part of the region (see Figure 2-2). The
highest ANC reaches are located along the border
with the calcareous Valley and Ridge Province to
the west and in the agricultural valleys of the Broad
and French Broad Rivers. This map is not directly
comparable with the alkalinity map of Omernik and
Powers (1983) for the region, as the latter also
includes data from larger rivers and reservoirs, but
both maps convey a similar image of the proportions
of the region represented by the three ANC classes,
if not their specifications. The utility of using"non-
standard"reference values to delineate map iso-
pleths will be discussed in Section 5.3 in the context
of stream classification.
While the previous discussion has focused on
statistical population estimates based on single
variables, it should also be borne in mind that single
74
-------
Figure 5-13. ANC distribution in the Southern Blue Ridge based on downstream spring average chemistry with effects
from storm events removed.
36° N
35° N -
34° N
• Less than 50 /jeq L~1
El 50-200 fjeq L~1
* Greater than 200 peq L
TN
GA
85° W
84° W
83° W
82° W
75
-------
variables seldom provide adequate answers to
complex questions. For example, all other things
being equal, low ANC waters are by definition more
susceptible to acidification than are high ANC
waters. Indeed, high ANC waters are unlikely to be
susceptible to acid deposition in any near-term
scenario (e.g., < 100 years). However, low ANC
systems in catchments that never were exposed to
glaciation may have a variable degree of buffering
capacity in their soils to delay marked declines in
pH over 10-100 years of exposure to acid deposition
(Galloway et al., 1983; Rochelle et al., 1986).
Therefore, the term "potentially susceptible," based
on ANC alone, should best be thought of as being
opposed to "unlikely to be susceptible."
Also, univariate population estimates address the
question of whether the pH and ANC of streams in
an area are different than they would have been
in .the absence of acid deposition. One of the
strengths of the NSS approach, however, is that the
distribution of any derived datum from manipulations
involving ion ratios, models, or other transformations
can be estimated for the target population using the
Phase I sampling design. Also, estimates involving
parts of the population (e.g., streams above 1000
m elevation or draining watersheds < 10 km2) also
can be made. Relationships among chemical,
hydrologic, and land use variables can be explored.
While such inferences cannot prove cause and
effect, it can be extremely helpful in generating
testable hypotheses. This level of analysis is already
underway for the Phase l-Pilot Survey data, and will
be the subject of future reports.
5.3 Stream Classification
In addition to providing population estimates that are
useful for environmental assessment, classification
of streams for future intensive studies is the other
primary objective of the NSS. Future studies will
focus on determining temporal variability, biotic
conditions, and long-term trends on a relatively small
number of streams, and thus will require extrap-
olation of the results to a larger target population.
While classification could be based on arbitrary
criteria (such as streams with index ANC greater
than or less than 200 /ueq L~1), a preferable scheme
would be based on evidence for two or more distinct
natural chemical classes, with different expected
responses to acid deposition. If such natural classes
do exist, it should be possible to accurately classify
streams on the basis of a minimum number of
samples, with a low probability of misclassification.
Finally, any classification system should not only be
qualitatively consistent with current scientific
understanding, but should also be quantitatively
objective and repeatable.
In the following sections, we present results from
two subjective classification schemes based on
geochemistry and geography, and from cluster
analysis, a method of objective multivariate analysis.
Examples of ways in which such classification
schemes could be used in future phases of the NSS
are also provided.
5.3.1 Univariate Models
A potentially useful subjective classification scheme
appeared early in analysis of the Phase l-Pilot Survey
data. Preliminary examination of the data indicated
that many of the variables were highly correlated
with ANC, and that streams could be divided into
at least three ANC groups or classes separated by
large ranges of ANC over which no reaches were
observed: < 250 /ueq L"1, 250-600 //eq L~\ and >
600 fjeq L"1. A smaller break was observed at 25
/jeq L"1 along with a noticeable thinning of data in
the 100-125 /ueq L"1 range, prompting tentative
classification breaks at 25 and 115 /ieq L"1. Various
ion ratios [(Ca + Mg)/(Na + K), (Ca + Mg)/ANC,
(Ca+Mg)/(S04=), (Na)/(CI), (SO4=)/(NO3~ ] were
calculated for each group, which revealed similar
values in all of the 25-600 /aeq L"1 ANC groups.
However, the high ANC group demonstrated
(Ca+Mg)/ANC and (Ca+Mg)/(Na+K) ratios typical of
calcareous systems. The single low ANC (< 25
//eq L~1) site showed an unusually high ratio of
(Ca+Mg)/ANC and a low (SO4=)/(N03~) ratio. These
sites thus appeared to be atypical of most streams
in the target population. When plotted on a map,
the major ANC groups exhibited considerable spatia!
continuity, as discussed below (Section 5.3.2). The
initial univariate ANC classification scheme served
as a "straw man" for many of the subsequent data
analyses.
5.3.2 Geographic Distributions
The geographic distributions of spring downstream
average concentrations of the six primary NSS
variables across the Phase l-Pilot Survey study area
are shown in Figures 5-14 through 5-19. The
classifications are based on natural univariate
groupings of the data for each variable, as noted
above, and do not represent any particular water'
quality criterion with respect to acid deposition.
Special interest sites are not shown on the maps.
The ANC map (Figure 5-15) demonstrates the
approximately contiguous geographic distributions
of the major ANC classes noted above. Three high
ANC (> 600 /jeq L"1) sites are located along the
northern and western edges of the study area, where
limestone from the adjacent Ridge and Valley
Province frequently is mixed with the felsicsaprolites
of the Southern Blue Ridge (e.g.. Hunt, 1974). A
second, intermediate ANC group (250-600 //eq L"1)
is located in the predominantly agricultural valleys.
76
-------
Figure 5-14. Geographic distribution of average springtime downstream pH in the NSS Phase l-Pilot Survey streams.
36° N
35° N
34° N
Less than 6.7
6.7-7.6
v 7.6-8.1
85° W
84° W
83° W
82° W
77
-------
Figure 5-15. Geographic distribution of average springtime downstream ANC in the NSS Phase I Pilot Survey streams.
36° N -
35° N
34° N
8 Less than 25/ueq L"1 1
Kl25-115A/eqL"1
x 115-250 //eq L'1
A 250-600 jieq L1
O Greater than 600 /ueq L
TN
85° W
84° W
83° W
82° W
78
-------
Figure 5-16. Geographic distribution of average springtime downstream sulfate concentrations in the NSS Phase l-Pilot
Survey streams.
36° N
35° N
34° N
Greater than 120 Aieq L
120-80 Aieq L'1
X 80-40 (teq l_~1
A Less than 40/ueqL 1
85° W
84° W
83° W
82° W
79
-------
Figure 5-17. Geographic distribution of average springtime downstream nitrate concentrations in the NSS Phase l-Pilot
Survey streams.
36° N
35° N
34° N
Greater than 35 /ueq L"1
35-20 //eql/1
X 20-10AieqL'1
A Less than 10 ueq L"1
85° W
84° W
83° W
82° W
80
-------
Figure 5-18. Geographic distribution of average springtime downstream chloride concentrations in the NSS Phase l-Pilot
Survey streams.
36° N
35° N
34° N
• Greater than 200/ueq L"1
200-
100-50//eqL~1
Less than 50//eq L"1
85° W
84° W
83° W
82° W
81
-------
Figure 5-19. Geographic distribution of average springtime downstream extractable aluminum concentrations in the NSS
Phase l-Pilot Survey streams.
36° N
35° N -
34° N
• Greater than 15 fjg L~1
15-10/ugl/1
X 10-5/ug L~1
A Less than 5/yg L"1
85° W
84° W
83° W
8Z°W
82
-------
These reaches also exhibit elevated chloride
concentrations (Figure 5-18). It is not known at this
time whether the high chloride concentrations are
indicative of anthropogenic sources, or are simply
correlated with the characteristics of valley soils
suitable for farming. The consistently low ANC sites
(including the only site with ANC < 25 ^ueq L~1) form
an inverted L-shaped area that includes the
highlands just west of Asheville, NC and the high
elevation ridge that defines the North Carolina-
Tennessee border including Great Smoky Mountains
National Park. The remaining areas appear to form
a patchwork containing 25-115 fjeq L~1 and 115-
250 /jeq L~1 sites.
Low pH sites (Figure 5-14) show general agreement
with the low ANC patterns, although two relatively
low pH sites stand out in the Georgia section of the
Survey. A different anomalously high sulfate site
(Figure 5-V6) also stands out in Georgia, with the
remainder of the high sulfate sites co-occurring with
high ANC. The nitrate map (Figure 5-17) exhibits
a cluster of relatively high nitrate (20-35 jjeq L~1)
sites in the northeast part of the study area, and
a very high nitrate value corresponding to the lowest
ANC site located in Great Smoky Mountain National
Park. It has been suggested that this site might be
typical of old-growth forests in this part of the study
area, which are at steady state with respect to nitrate
inputs (/?. Turner and P. Mulholland, Oak Ridge
National Laboratory, personal communication). No
pattern is evident in the extractable aluminum
concentrations, although concentrations greater
than 5 jug L"1 are apparently rare in the western
part of the study region. Possible geochemical or
anthropogenic causes for the remaining "unusual"
sites are being investigated as geology and land use
data are acquired for the Phase l-Pilot Survey
watersheds.
ANC maps (such as Figure 5-15) were plotted and
compared for each spring and summer sampling
interval. Although a few sites changed categories
on each map, the overall geographic distribution of
ANC within the study area remains identical
throughout the spring. During summer, the area of
higher ANC categories expanded, apparently due to
changes in the relative volumes of source water
detained in the soil mantle for short and long periods
of time before entering the channel. While all spring
samples during 1985 provided similar geographic
ANC maps of the region, the summer sample
provided a map with considerably wider areas" of
high-ANC stream water.
5.3.3 Cluster Analysis
Previous sections have dealt with subjective
interpretation and classification of multivariate data.
Cluster analysis (Romesburg, 1984) is a multivariate
statistical technique that can be used to make such
classifications more objective, but the results are
sufficiently dependent on the particular algorithm
used that such classifications should be used with
caution. Phase l-Pilot Survey data were subjected
to hierarchical cluster analysis using a number of
different clustering methods and sets of chemical.
variables, both with and without episodes removed.
A polythetic agglomerative technique (in contrast to
a devisive technique) based upon Euclidean distance
and the maximization of the average linkages within
sample site clusters appeared to produce the most
robust and useful classifications and was employed
using a "full"set of 30 chemical variables to produce
the dendrograms discussed below.The agglomera-
tive clustering technique used is more efficient at
identifying outlying groups than at minimizing
within-group variance.
The dendrogram resulting from a clustering run
based on the spring downstream averages with
episodes removed is shown in Figure 5-20. This
dendrogram is typical of all runs on individual spring
sampling intervals, in that the three high-ANC sites
cluster far from the remaining variables, the
intermediate ANC (250-600 //eq L"1) sites form a
second cluster, and the lower ANC « 250 ^ueq L"1)
sites form a third cluster with similar ANC groups
appearing near each other. Two episodes occurring
during the third spring sampling interval changed
the average spring chemistry sufficiently to cause
the sites to appear as outliers on a similar dendro-
gram. Major episodes thus cause sufficient changes
to confound reach classification based on agglomer-
ative cluster analysis of stream chemistry. On the
other hand, cluster analysis may be useful for
identifying putative episodes in cases where stage
changes are not available. The special interest sites
are indicated by arrows in the diagram, and most
can be seen to be typical of the 25-115 /ueq L~1 ANC
class reaches.
During the course of analysis, it became evident that
the presence of two small classes of high ANC sites
with highly distinct chemistry dominated site
classification in the Pilot Survey. Furthermore, these
high ANC sites are not particularly interesting from
an acidification standpoint. Cluster analysis of the
same variables, after removing the three high ANC
sites, caused one of the intermediate ANC sites
consistently to appear as an outlier. Removal of this
site resulted in a strong cluster containing the
intermediate ANC sites, another cluster breaking at
approximately 190 jueq L"1, and a weaker cluster
breaking at 115 fjeq L"1. These clusters were most
pronounced during the first spring sample, and each
< 250 //eq L~1 cluster expanded or contracted by
83
-------
Figure 5-20. Hierarchical cluster diagram of all NSS Phase I—Pilot Survey sites based on downstream spring average values
for 39 chemical variables. Episodes have been removed. Arrows indicate special interest sites.
ANC Classes
foieq/L)
0-25
25-115
115-250
b 250-600
I >600
*
*
*
•K
*
•K
*
•*
*
*
*
*
*
*
*J
A
A
A-J
10
5 10 15 20 25
Rescaled Distance Clusters Combine
30
84
-------
fewer than five sites during the second and third
spring sampling interval (occasionally the interme-
diate cluster split into two). Interestingly, the lowest
ANC site did not appear as an outlier, despite its
unusual ion ratios. The geochemical significance of
the remaining clusters, and the effect of removing
all > 250 fjeq L"1 sites from the clustering process,
is presently under investigation. Use of devisive
clustering techniques (which are less sensitive to
sample outliers than are agglomerative techniques)
may enable robust classification of sample streams
without necessitating the removal of sites of unusual
chemical composition.
5.3.4 Utility of Classification for Regional
Assessment
One of the two primary data quality objectives (DQOs)
of the National Stream Survey is to classify streams
for future intensive studies. The foregoing examples
illustrated how univariate and multivariate analyses
of water chemistry data could provide a useful
framework for understanding the spatial distribution
of chemical variables and the relationships among
sites arranged according to single and multiple
variables. All of the classification results were
consistent with a relatively simple geochemical
interpretation of the chemistry of streams in the
Southern Blue Ridge.
Figure 5-21 shows the Phase l-Survey data plotted
on a mineral stability diagram for potassium feldspar,
an important component of the parent geologic
material in the area. The position and slope of data
plotted on such a diagram enable one to hypothesize
the geochemical weathering processes controlling
water chemistry in natural waters in contact with
soil and rock. The data points are coded according
to ANC classes described in the preceding section.
As shown in the lower (eft region of the diagram,
a paucity of potassium and silica is accompanied
by low values for ANC, and is probably indicative
of intense weathering due to high precipitation
loadings (Velbel 1985a, 19855). This relationship is
not evident in the highest ANC stream group, whose
numbers are characterized by considerably lower
ratios of potassium (and higher ratios of calcium)
to ANC than those in lower ANC groupings. The
streams with ANC > 600 fjeq L~n drain watersheds
which contain underlying limestone inclusions (see
below). One of the 250-600 fjeq L~1 ANC streams
also appears as an outlier in this analysis. In the
lower ANC range of watersheds in the Southern Blue
Ridge, ANC thus appears to be correlated with
feldspar weathering.
The geochemistry of streams in the Southern Blue
Ridge has been hypothesized to be controlled
primarily by weathering kinetics (Velbel, 1985a,b).
To draw a simple analogy, water moves through the
silacious saprolites (subsoils) in these watersheds,
dissolving minerals that comprise the pseudomor-
phous parent materials, much as water acquires a
pleasant flavor when in contact with a tea bag. As
in the case of tea, the strength of the brew is
controlled by the residence time of the water in the
bag, and the number of times the bag was previously
used, in the field, the former part of the analogy
involves the hydrology of the watershed and the
second part depends upon the age and degree of
weathering of the predominant geologic formation
in the area (Coweeta Group or Tallulah Falls
Formation). It would be convenient for classification
purposes if acid neutralizing capacity in the region
could be predicted largely on the basis of the degree
of weathering in a watershed, which could in turri
be related to hydrology.
The chemical classification of stream reaches
described above is a first step toward understanding
the regional relationships between watershed
characteristics, acid deposition, and water chemistry
variables. In the future, such chemical classifications
will be refined as additional data on Phase I streams
and watersheds are obtained. Water chemistry data
will be related to such watershed characteristics as
topography, drainage area, bedrock geology, soil
residence time, vegetation, and land use data.
Subsequent reclassification on the basis of these
variables will provide a framework which can be
utilized for the following purposes:
1. Identification of scientifically-based groupings
of sites according to water chemistry and
presumed vulnerability to acidification.
2. Generation or refinement of hypotheses
regarding the relationships among watershed,
atmospheric, and water quality variables.
3. Identification and delineation of distinct classes
of sites which can be considered separately
with regard to a variety of NSWS objectives.
The identification of distinct classes of sites is a
particularly important goal of classification, because
it allows individual stream reaches and their
watersheds to be identified for intensive research
or long-term monitoring. The index chemistry of the
special interest sites studied in the Phase (-Pilot
Survey exemplifies the utility of classification. The
estimated population distributions for the various
chemical variables in the Southern Blue Ridge target
population are particularly interesting when com-
pared to the chemistry of the special interest sites.
85
-------
Figure 5-21. Potassium-feldspar mineral stability diagram for streams in the NSS Phase l-Pilot Survey. Values represent
spring downstream average with episodes removed. pH is based on 300 ppm CO: equilibrated values.
4.00
3.60
3.20
2.80
2.40
2.00
1.60
4.4
4.2
4.0
3.8
3.6
3.4
pSi02
86
-------
These sites have been chosen by previous inves-
tigators as being "typical" of the region, in addition
to providing a reasonable degree of accessibility for
field work. Six of the seven special interest sites
had mean spring ANC values below 50//eq L"1 during
the study, and thus represent less than 7% of the
target population streams in the area. On the other
hand, they did not appear as outliers in the
multivariate chemical classification of streams in the
region, and pH values were higher than at several
of the probability sites. Results gained from
geochemical research and monitoring programs at
these sites can thus probably best be considered as
typical of a small but significant group of target
population streams in the area.
The identification of individual reaches or
watersheds for intensive study of episodic acidifi-
cation due to hydrologic events will require that these
systems be suitably representative of appropriate
classes described above. Current understanding of
episodes suggests that deposition loading,
watershed hydrology, baseflow chemistry, and land
use may be the important components of a clas-
sification scheme for studying episodes. Developing
indices for each of these components is a critical
task prior to undertaking detailed, intensive studies
of episodes in regions receiving acid deposition.
8. Developing robust classification schemes for
selecting sites for intensive process-oriented
research, ecological effects studies, and long-
term monitoring.
The results of these studies will be published as
open-file reports or in the professional scientific
literature as they become available.
5.4 Future Analyses
While the findings of the Phase l-Pilot Survey were
sufficient to allow Phase I to proceed with approp-
riate modifications, several additional analyses of the
Phase l-Pilot Survey are presently underway. These
include:
1. Applying empirical models to search for
evidence (if any) of acidification by atmospheric
deposition.
2. Investigating effects of runoff and subsurface
geology on ANC and chemical variables
associated with weathering.
3. Linking chemistry to geography and land use,
including nonpoint sources of pollution.
4. Creating or revising ANC maps.
5. Calibrating target population data obtained
from 1:250,000-scale maps with smaller scale
maps and remote imagery.
6. Devising population estimates based on area!
export coefficients (using Equation [5.1]).
7. Comparing the Phase l-Pilot Survey data with
other intensive and synoptic stream and lake
data available for the Southern Blue Ridge.
87
-------
6. Conclusions and Recommendations
6.1 Conclusions
The primary objectives of the Phase l-Pilot Survey
were to (1) test the ability of the proposed sampling
design to meet the Phase I objectives and (2) evaluate
the logistics plan and data analysis plan for Phase
I. The Phase l-Piiot Survey demonstrated that a
regional-scale synoptic survey of streams of the size
targeted in the data quality objectives can be
accomplished logistically, will produce robust
population estimates for important chemical varia-
bles based on a single spring sampling, and has the
potential of producing a relatively simple geochem-
ical grouping of streams with reasonable support in
the scientific literature. The results of the study were
deemed to be adequate to meet the Phase I
objectives, so no further Phase I field work is planned
for the Southern Blue Ridge Province. It is important
to avoid sampling during major hydrologic events
or post-stratify the data, so that the distributions are
not affected by episodes, during which pH and ANC
are temporarily depressed. For population estimation
alone, there is apparently little value in replication,
if sampling is confined to the spring of the year.
in the Southern Biue Ridge, pH showed virtually
identical population distributions in spring and
summer. However, ANC increased substantially
between the two periods. Summer sampling did not
affect the interpretation of sulfate and nitrate
distribution estimates generated from spring data.
The Phase l-Pilot Survey used a sample size of 54
to generate distribution estimates for chemical
variables, but subsequent analyses could be
performed to determine the effects of sample size
on the distributions. Such experiments would also
be useful in evaluating the absolute minimum
sample size that is likely to be useful for assessment
purposes.
There were significant differences in the concen-
trations of most important NSWS variables between
the upstream and downstream nodes of the reaches
in the target population. The changes varied
sufficiently in both magnitude and direction that
calibration of one value on another does not appear
to bepossible. Univariate and multivariate regres-
sions indicated that of all of the geographic variables
tested, only elevation showed any significant
relationship to pH or ANC concentration on a region-
wide basis. Even this relationship proved to be too
weak for predictive or descriptive purposes.
Classification of streams for further study also
appears to be possible on the basis of a single
synoptic sampling of the Southern Blue Ridge
streams. However, it is desirable to have two samples
to make chemical changes associated with hydro-
logic or pollution events easier to identify, and to
provide an estimate of the degree of robustness of
the classification. The latter factor is particularly
important if the Phase I streams themselves are to
be the primary candidates for study in future phases
of field work. However, it would be valuable even
if they serve only to identify the desired character-
istics of other sites chosen for their greater
accessibility.
The Phase l-Pilot Survey was useful in greatly
increasing the probability of success and decreasing
the cost of a full Phase I Survey. Evaluation of the
logistics plan indicated that a ground-based synoptic
survey of randomly chosen streams over a large
geographic region could be carried out safely and
successfully. Field experiments and evaluations
identified the most promising techniques and
protocols, as well as potential problems ranging from
instrument malfunctions to disbursing pay envelopes
to field crews. The data management and QA/QC
programs proved largely successful, and experience
gained in these programs will undoubtedly reduce
unnecessary sample processing costs and eliminate
many troublesome communication bottlenecks. The
successful completion of the pilot study and the
timely analysis of the data were critical in producing
a scientifically acceptable and cost-effective Phase
I research plan.
Finally, the Phase l-Pilot Survey data set can and
will certainly be used in an assessment context, as
regulators, resource managers, and others charged
with environmental assessment seek to quantify the
extent of acidic and low ANC waters of the United
States that are potentially susceptible to acidification
by atmospheric deposition. While the target popu-
lation of reaches was characterized by a high
proportion of low ANC reaches, less than 3.2% of
the stream length in the target population was
estimated to have an index pH below 6.4. Thus, the
88
-------
Phase l-Pilot Survey data base provides a snapshot
in time of the resource at risk as represented by
the target population. This information will become
increasingly meaningful as our overall understand-
ing of the relationships between acid deposition,
water quality, and aquatic biota improves.
6.2 Recommendations for Phase I
Analysis of the Phase l-Pilot Survey results were
used in formulating recommendations for changes
in the proposed Phase I sampling design, logistics
plan, QA and methods plan, and in the data
management system. It is not known to what extent
the recommendations made are based on conclu-
sions drawn about a unique geographic region, the
Southern Blue Ridge Province. Therefore, scientific
judgment exercised by EPA, contractor, and external
scientists was in many cases used to extrapolate
the Pilot Survey results to other regions of study.
In particular, the input of scientists who work
extensively or exclusively on ecosystems in these
new regions of study were crucial in altering the
proposed Phase I design in a responsible and timely
way. The modified Phase I Research Plan was
deemed sufficient to proceed with a full scale Phase
I Survey in the Mid-Atlantic region during the spring
of 1986 (U.S. EPA, 1985b). The following list
summarizes the modifications in the proposed Phase
I design which were recommended and enacted.
1. Population estimates based on data collected
during the spring sampling season provide a
good "index" for characterizing the chemical
status of mid-sized streams. Because ANC and
pH are not lower during the summer and the
most sensitive life stages of fish are typically
present during the spring, summer sampling
is unnecessary for population description
purposes.
2. Whereas one spring sampling appears ade-
quate for population description, replication is
desirable for classifying streams. This is
particularly true if the goal of classification is
to identify Phase I streams for further intensive
studies. Therefore, two samplings are recom-
mended for meeting both primary Phase I
objectives.
3. Because the goal of Phase I is to describe the
population of target reaches using the "index"
concept, it is important to avoid sampling under
transient hydrologic conditions such as major
rain storms or snowmelt events, during which
relatively large changes in many chemical
variables may occur. Regional studies of
episodic acidification, like long term monitoring
and studies of biological resources, will almost
certainly be performed on a limited number of
aquatic systems that can be considered as
"regionally representative"based on Phase I
classification, acid deposition inputs, land use,
physiography, and other characteristics. There-
fore, it is recommended that Phase I sampling
be conducted so that field sampling does not
occur during or immediately after major rainfall
or snowmelt events.
4. Observed differences irj chemical concentra-
tions measured at upstream and downstream
nodes were statistically significant for five of
the six major variables. In Phase I, it will be
desirable to estimate the chemistry of the entire
reach, so it is recommended that both nodes
of each reach be sampled. The chemistry of
the intervening water will then be estimated
by interpolation. Although not totally satisfac-
tory, this practice should improve the popula-
tion estimates to some extent. As an added
benefit, chemistry from the upstream node can
be used to make estimates for the smaller
watersheds draining into the NSS target
population reaches, as well as to estimate an
area! contribution index for the watersheds
contained in the target population itself.
5. In subsequent Phase I surveys, it is recom-
mended that revised site inclusion criteria be
used to identify the target population. The
following criteria are suggested, and these are
summarized in the 1986 Draft Research Plan
(U.S. EPA, 1985b).
a. The boundary reach criteria should be
consolidated into one rule: reaches should
be considered non-reaches to a given
subregion if greater than 50% of the blue
line length lies outside the region boundary.
b. It is now assumed that reservoirs in
watersheds of < 60 mi2 are unlikely to
significantly affect reach chemistry, and
that downstream nodes of influent reaches
can be identified adequately during field
reconnaissance. Reservoir tailwater
reaches are included as a special class of
interest reaches (N2R) in Phase I.
c. New categories were subsequently added
to the Phase I site rules as problems with
the original rules were encountered in new
regions. Urban reaches (based on areas
indicated in yellow on USGS 1:24,000-
scale topographic maps) are defined as non-
interest reaches in Phase I. This decision
was made because of the operational
difficulty of determining drainage boundar-
ies when no contours are shown within the
mapped urban areas, and also because of
89
-------
the perception that these reaches are not
at risk. Inappropriate physical habitat or
gross point-source pollution reduce the
importance of possible impacts due to
acidification. Special categories of interest
reaches have been added to the Phase I
site rules to include wetland reaches with
indistinct topographic drainage boundaries
and large "headwater" reaches with
drainage areas > 60 mi2.
An addition to these changes in sampling design,
the following changes in the logistics and QA plans
were recommended for Phase I:
1. As a result of the comparability of the two field
pH methods, it is recommended that the closed-
system measurement be dropped in Phase I.
C02 degassing is apparently slow enough in
quiescently stirred solutions, so that electrode
instability is not a problem in the open-system
measurement. The open-system measurement
is much easier to perform in the field, and
requires less equipment.
2. An automated aluminum speciation and
measurement technique using pyrocatechol
violet (Dougan and Wilson, 1974; Rogeberg and
Henriksen, 1985) will be instituted in Phase
I to avoid problems associated with the manual
technique used in the Phase l-Pilot Survey.
3. Sample holding times for syringe and Cubitain-
ers will be increased from 12 to 24 hours based
on results from several holding time experi-
ments that we performed during the Phase I-
Pilot Survey. Given this decision, it is recom-
mended that the mobile field laboratory be
deployed in Las Vegas during Phase I, rather
than at base sites in the field. Samples will
be transported by overnight air courier in
coolers kept at 4°C to Las Vegas, and pres-
ervation at the laboratory will occur within 24
hours of sample collection.
4. Because the contract laboratory met the limits
for matrix spike recovery for every batch of
samples analyzed in the Phase l-Pilot Survey
and no matrix interferences were observed,
matrix spike QC samples will not be used during
the Phase I Survey.
6.3 Related Documents
In addition to this report, supplemental information
on the National Stream Survey Phase l-Pilot Survey
can be found in the series of ancillary manuals and
reports. Many of the technical manuals used were
in draft form at the time the Phase l-Pilot Survey
was conducted. If substantive changes were not
anticipated to technical manuals to be used for the'
full-scale Phase I Survey, then separate Pilot Survey
manuals will not be published. Major changes in
Pilot Survey methods and procedures planned for
Phase I were summarized in this report. The related
documents include:
1. Field Operations Report, National Surface
Water Survey, National Stream Survey, Pilot
Survey. 1986. Knapp, C. H., C. L. Mayer, D.
V. Peck, J. R. Baker, and G. J. Filbin. Lockheed
Engineering and Management Services Com-
pany, Inc., Las Vegas, Nevada 89109 (draft).
2. Quality Assurance Plan for the National
Surface Water Survey. Stream Survey (Middle
Atlantic Phase I, Southeast Screening and
Middle Atlantic Episodes Pilot). 1986. Drouse,
S. K., D. C. Hillman, L W. Creelman, and S. J.
Simon. Lockheed Engineering and Manage-
ment Services Company, Inc., Las Vegas,
Nevada 89109 (draft).
3. Evaluation of Quality Assurance and Quality
Control Sample Data for the National Stream
Survey (Phase l-Pilot Survey). 1986. Drouse,
S. K. Lockheed Engineering and Management
Services Company, Inc., Las Vegas, Nevada
89109 (draft).
4. Analytical Methods Manual: National Surface
Water Survey, Stream Survey (Middle Atlantic
Phase I, Southeast Screening, and Middle
Atlantic Episodes Pilot). 1986a. Hillman, D. C.,
S. H. Pia, and S. J. Simon. Lockheed Engineer-
ing and Management Services Company, Inc.,
Las Vegas, Nevada 89109 (draft).
5. Data Management and Analysis Procedures for
the National Stream Survey. 1987. Sale, M. J.
(editor). ORNL/TM. Oak Ridge National Labor-
atory, Oak Ridge, Tennessee 37831 (draft).
6. Draft Research Plan, National Surface Water
Survey: National Stream Survey, Mid-Atlantic
Phase I and Southeast Screening. 1985. U.S.
Environmental Protection Agency, Office of
Research and Development, Washington, D.C.
20460.
7. Draft Sampling Plan for Streams in the National
Surface Water Survey. Technical Report 114
(July 1986) Overton, W. S. Department of
Statistics, Oregon State University, Corvallis,
Oregon 97331.
90
-------
7. References
Baker, J. 1981. Aluminum Toxicity to Fish as Related
to Acid Precipitation and Adirondack Surface
Water Quality. Ph.D. Thesis, Cornell University,
Ithaca, NY.
Beamish, R. J., and H. H. Harvey. 1972. Acidification
of the LaCloche Mountain Lakes, Ontario, and
Resulting Fish Mortalities. J. Fish. Res. Board Can.
29:1131-1143.
Beamish, R. J., W. L. Lockhart, J. C. Van Loon, and
H. H. Harvey. 1975. Long-term acidification of a
lake and resulting effects on fishes. Ambio 4:98-
102.
Bovee, K. D., and D. Cochnauer. 1977. Development
and evaluation of weighted criteria probability-of-
use curves for instream flow assessments.
Fisheries Instream Flow Information Paper No. 3.
FWS/OBS-77/63. Cooperative Instream Flow
Service Group, Ft. Collins, Colorado 38 pp.
Burke, E. M., and D. C. Hillman. 1987. Syringe
Holding Time Study, pp. 59-75. In Knapp et al.
NSWS: NSS-Pilot Survey. Field Operations Report.
EPA/600/8-87/019.
Chorley, R. J., and P. F. Dale. 1972. Cartographic
problems in stream channel delineation. Carto-
graphy 7:150-162.
Cochran, W. G. 1977. Sampling Techniques, 3rd Ed.
Wiley: New York.
Cogbill, C. V., G. E. Likens, and T. A. Butler. 1984.
Uncertainties in historial aspects of acid precip-
itation: getting it straight (including comments by
Hansen and Hidy). Atmos. Environ. 18:2261 -2270.
Cosby, B. J., G. M. Hornberger, and J. N. Galloway.
1985. Time scales of catchment acidification.
Environ. Sci. and Technol. 19:1144-1149.
Dougan, W. K., and A. L. Wilson. 1974. The
absorptiometric determination of aluminum in
water: A comparison of some chromogenic
reagents and the development of an improved
method. Analyst 99:413-430.
Drablos, D., and A. Tollan, editors. 1980. Ecological
impact of acid precipitation. Proceedings of an
international conference, Sandefjord, Norway.
March 11-14, 1980.
Driscoll, C. T. 1984. A procedure for the fractionation
of aqueous aluminum in dilute acidic waters.
Internal. J. Environ. Anal. Chem. 16:267-283.
Drous§, S. K., D. C. Hillman, J. L. Engels, L W.
Creelman, and S. J. Simon. 1986. National Surface
Water Survey, National Stream Survey (Phase I-
Pilot, Mid-Atlantic Phase I, Southeast Screening,
and Episodes Pilot). Quality Assurance Plan. EPA/
600/4-86/044. U.S. EPA, Las Vegas, Nevada.
Drouse, S. K. 1987. National Surface Water Survey,
National Stream Survey (Phase l-Pilot Survey)
Summary of Quality Assurance Results. U.S. EPA,
Las Vegas, Nevada. In Press.
Fenneman, N. M. 1946. Physical divisions of the
United States. U.S. Department of Interior,
Geological Survey. Reston, Virginia.
Fountain, J., and D. T. Hoff. 1985. AQUARIUS User's
Guide. Lockheed Engineering and Management
Services Company, Inc., Las Vegas, Nevada
(unpublished MS).
Galloway, J. N., S. A. Norton, and M. R. Church.
1983. Freshwater acidification from atmospheric
deposition of sulfuric acid: a conceptual model.
Environ. Sci. and Technology 17(11):541-545.
Gunn, J. M., and W. Keller. 1984. Spawning site
water chemistry and lake trout (Salvelinus
namaycush) sac fry survival during spring snow-
melt. Can. J. Fish. Aquat. Sci. 41:319-329.
Hagley, C. A., C. M. Knapp, C. L. Mayer, and F. A.
Morris. 1986. National Surface Water Survey,
Stream Survey (Middle Atlantic Phase I, Southeast
Screening, and Middle Atlantic Episode Pilot). Field
Training and Operations Manual. Lockheed
Engineering and Management Services Company,
Inc., Las Vegas, Nevada.
Haines, T. A. 1981. Acidic precipitation and its
consequences for aquatic ecosystems: a review.
Trans. Am. Fish. Soc. 110:669-707.
Haines, T. A. 1982. Interpretation of aquatic pH
trends. Trans. Am. Fish. Soc. 111 -.779-786.
Haines, T. A., and J. J. Akielaszek. 1983. A regional
survey of the chemistry of headwater lakes and
streams in New England: Vulnerability to acidi-
fication. In Air Pollution and Acid Rain. Report No.
15, U.S. Dept. Interior, Fish and Wildlife Service,
FWS/OBS-80/40/15. 141 p.
Havas, M., T. C. Hutchinson, and G. E. Likens. 1984.
Red herrings in acid rain research. Environ. Sci.
Technol. 18:176A-186A.
Henriksen,.A., O. Skogheim, and B. Rosseland. 1984.
. Episodic changes in pH and alurninum-speciation
kill fish in a Norwegian salmon river. Vatten
40:255-260.
Hillman, D. C., D. V. Peck, J. R. Baker, F. A. Morris,
K. J. Cabbie, and S. L. Pierett. 1985. National
91
-------
Stream Survey Field Training and Operations
Manual. Lockheed Engineering and Management
Services Company, Inc., Las Vegas, Nevada.
Hillman, D. C., S. H. Pia, and S. J. Simon. 1986a.
National Surface Water Survey, National Stream
Survey (Phase l-Pilot, Mid-Atlantic Phase I,
Southeast Screening, and Episodes Pilot). Analyt-
ical Methods Manual. EPA/600/8-87/005. U.S.
EPA, Las Vegas, Nevada.
Hillman, D. C., J. F. Potter, and S. J. Simon. 1986b.
National Surface Water Survey, Eastern Lake
Survey (Phase l-Synoptic Chemistry), Analytical
Methods Manual. EPA-600/4-86-009, U.S. Envir-
onmental Protection Agency, Las Vegas, Nevada.
Howells, G. D. 1984. Fishery decline: mechanisms
and predictions. Phil. Trans. R. Soc. Lond. B305,
529-547. Lockheed-EMSCO, Las Vegas, Nevada.
Hughes, R. M., and J. M. Omernik. 1981. Use and
misuse of the terms watershed and stream order.
Am. Fish. Soc. Warmwater Streams Symposium,
pp.320-326.
Hughes, R. M., and J. M. Omernik. 1983. An
alternative for characterizing stream size. pp. 87-
101. In: T. D. Fontaine III and S. M. Bartell (eds.).
Dynamics of Lotic Ecosystems, Ann Arbor Science,
Ann Arbor, Michigan.
Hunt, C. B. 1974. Natural Regions of the United
States and Canada. W. H. Freeman and Company,
San Francisco.
Jeffries, D. S., B. LaZerte, and R. A. Linthurst. 1985.
Effects of acid deposition on the chemistry of
aquatic ecosystems in eastern North America.
International Symposium on Acidic Precipitation,
Muskoka Conference, September 1985 (Abstract).
Johnson, D., J. Colquhoun, F. Flack, and H. Simonin.
1985. In situ Bioassays of Fish in Acid Waters.
Final Report to EPA/NS4CSU Acid Deposition
Program, Raleigh, NC.
Kanciruk, P., R. J. Olson, and R. A. McCord. 1986.
Quality Control in Research Data Bases: The U.S.
Environmental Protection Agency National Sur-
face Water Survey Experience. In: W. K. Michener
(ed.). Research Data Management in the Ecological
Sciences, University of South Carolina Press,
Columbia, South Carolina, pp. 193-207.
Kelso, J. R. M., and J. M. Gunn. 1982. Responses
of fish communities to acid waters in Ontario. In
G. R. Hendrey (ed.). Chronic stress of acidification
on aquatic biota: pH 5 to 6. Am. Chem. Soc. Symp.
Ser., Ann Arbor Science Publishers, Ann Arbor,
Michigan.
Knapp C. M., C. L. Mayer, D. V. Peck, J. R. Baker,
and G. J. Filbin. 1987. NSWS: NSS-Pilot Survey;
Field Operations Report. EPA/600/8-87/019-
U.S. EPA, Las Vegas, Nevada. 94 pp.
Krug, E. C., P. J. Isaacson, and C. R. Frink. 198"5.
Appraisal of some current hypotheses describing
acidification of watersheds. JAPCA 35:109-114.
Lefohn, A. S., and R. W. Brocksen. 1984. Acid rain
effects research—a status report. JAPCA 34:1005-
1013.
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. U.S. Envir-
onmental Protection Agency (EPA-600/4-86-
007A).
Magnuson, J. J., J. P. Baker, and E. J. Rahel. 1984.
A critical assessment of acidification on fisheries
in North America. Phil. Trans. R. Soc. Lond. B305,
501-516.
Minshall, G. W., R. C. Petersen, K. W. Cummins,
T. L. Bott, J. R Sedell, C. E. Gushing, and R. L.
Vannote. 1983. Interbiome comparison of stream
ecosystem dynamics. Ecol. Monogr. 53:1-25.
Morris, F. A., D. V. Peck, M. B. Bonoff, K. J. Cabbie,
and S. L. Pierett. 1986. National Surface Water
Survey, Eastern Lake Survey-Phase l-Synoptic
Chemistry, Field Operations Report. EPA-600/4-
86-010. U.S. Environmental Protection Agency,
Las Vegas, Nevada.
Muniz, I. P., and H. Leivestad. 1980. Toxic effects
of aluminum on the brown trout, Safmo trutta L.
In: D. Drablos and A. Tollan (eds.). Ecological
Impacts of Acid Precipitation, Proc. Int. Cont,
Sandefjord, Norway, SNSF Project 1432 AS-NLW,
pp. 320-321.
National Research Council. 1981. Atmosphere-
Biosphere Interactions: Toward a Better Under-
standing of the Ecological Consequences of Fossil
Fuel Combustion. National Academy Press.
National Research Council. 1983. Acid Deposition:
Atmospheric Processes in Eastern North Amer-
ica—a Review of Current Scientific Understanding.
National Academy Press: Washington, D.C. 375
pp.
National Research Council. 1984. Acid Deposition:
Processes of Lake Acidification, National Academy ;
Press.
Oden, S. 1976. The acidity problem -- an outline
of concepts, pp. 1 -36. In Proc. of the First Intern.
Symp. on Acid Precip. and the Forest Ecosystem.
Dochinger, L. S. and T. A. Seliga, eds. U.S. Dept.
Agric. For. Serv. Gen. Tech. Rep. NE-23.
Office of Science and Technology Policy. 1984.
Report of the Acid Rain Peer Review Panel; July
1984.
Office of Technology Assessment. 1984. Acid Rain
and Transported Air Pollutants: Implications for
Public Policy, Report OTS-0-204; U.S. Congress,
Washington, D.C. June 1984.
Olem, H. 1984. Stream acidification during storm
events in southern Appalachian watersheds.
Aquatic Effects Task Group (E) and Terrestrial
Effects Task Group (F) Peer Review Research
Summaries. NAPAP. ADP. North Carolina State
Univ. pp. 61-66.
92
-------
Olsen, J. E., E. A. Whippo, and G. C. Horak. 1981.
REACH file Phase II report: A standardized method
for classifying status and type of fisheries. U.S.
Fish Wildl. Serv. FWS/OBS-81/31. 56 pp.
Omernik, J. M., and C. F. Powers. 1983. Total
alkalinity of surface waters—a national map. Ann.
Assoc. Amer. Geographers 73(1): 133-136.
Overton, W. S. 1985. (Draft) Sampling Plan for
Streams in the National Surface Water Survey.
Technical Report 114 (July 1986). Department of
Statistics, Oregon State University, Corvallis,
Oregon.
Overton, W. S. 1987. A Sampling and Analysis Plan
for Streams in the National Surface Water Survey.
Technical Report 117. Department of Statistics,
Oregon State University, Corvallis, Oregon.
Pfeiffer, M. H., and P. J. Festa. 1980. Acidity status
of lakes in the Adirondack region of New York in
relation to fish resources. New York State
Department of Environmental Conservation, FW-
PI68( 10/80).
Pierson, W. R., and T. Y. Chang. 1986. Acid rain
in western Europe and northeastern United
States—a technical appraisal. CRC Critical
Reviews in Environmental Control 16:167-192.
Platts, W. S. 1979. Relationships among stream
order, fish populations, and aquatic geomorphol-
ogy in an Idaho river drainage. Fisheries 4:5-9.
Rochelle, B. P., M. R. Church, and M. B. David. 1986.
Sulfur retention at intensively studied watersheds
in the U.S. and Canada. Water, Air, and Soil
Pollution (in press).
Rogeberg, E. J. S., and A. Henriksen. 1985. An
automatic method for fractionation and determi-
nation of aluminum species in fresh-waters.
Vatten 41:48-53.
Romesburg, F. 1984. Cluster Analysis for
Researchers. Lifetime Learning, Belmont,
Massachusetts.
Rosen, A. E., and P. Kanciruk. 1985. A Generic Data
Entry Quality Assurance Tool. In: SAS User's
Group International, Proceedings of the Tenth
Annual Conference, SAS Institute, Gary, North
Carolina. 1985. pp. 434-436.
Sale, M. J. (editor). 1987. Data Management and
Analysis Procedures for the National Stream
Survey. ORNL/TM. Oak Ridge National Laboratory,
Oak Ridge, Tennessee 37821 (draft).
Sale, M. J. 1986. Data Dictionary for the National
Stream Survey. Oak Ridge National Laboratory,
Oak Ridge, Tennessee 37831 (draft).
SAS Institute, Inc. 1983. SAS User's Guide: Basics,
Version 5 Edition. SAS Institute, Inc., Gary, North
Carolina. 1290 p.
SAS Institute, Inc. 1985. SAS User's Guide:
Statistics, Version 5 Edition. SAS Institute, Inc.,
Gary, North Carolina. 956 p.
Schofield, C. L. 1976. Acid precipitation: effects on
fish. Ambio 5:338-230.
Schofield, C. L., and J. R. Trojnar. 1980. Aluminum
toxicity to brook trout (salvelinus fontinalis) in
acidified waters, pp. 347-366. In Polluted Rain.
Toribarn, T. Y., M. W. Miller and P. E. Morrow,
Eds. Plenum Press, New York, NY.
Sharpe, W. E., D. R. DeWalle, R. T. Leibfried, R. S.
Dinocola, W. G. Kimmel, and L. S. Sherwin. 1984.
Causes of acidification of four streams on Laurel
Hill, southwestern Pennsylvania. J. Environ. Qua).
13(4):619-631.
Shreve, R. L. 1966. Statistical law of stream
numbers. Journal of Geology 74:17-37.
Silsbee, D. G. and G. L. Larson, 1982. Water quality
of streams in the Great Smoky Mountains National
Park. Hydrobiologia 89, 97-115.
Smith, R. A., and R. B. Alexander. 1983. Evidence
for acid-precipitation—induced trends in stream
chemistry at hydrologic benchmark stations. U.S.
Geological Survey Circular 910. US DOI. 12 p.
Sports Fishing Institute Bulletin. 1984. Nov./Dec.
No. 360.
Stapanian, M. A., A. K. Pollack, and B. C. Hess. 1986.
Holding Time for Lake and Stream Samples: Effects
on Twenty-Five (25) Analytes. Lockheed Engineer-
ing and Management Services, Inc., Las Vegas,
Nevada (unpublished MS). 12 pp.
Storet User's Handbook. 1985. Computer Sciences
Corporation, 6521 Arlington Boulevard, Falls
Church, Virginia 22024.
Talbot, R. W. and A. W. Elzerman. 1985. Acidification
of Southern Appalachian lakes. Environ. Sci.
Technol. 19:552-557.
TIE Staff. 1981. Regional assessment of aquatic
resources at risk from acid deposition (Inst. of
Ecology, Indianapolis, Ind.).
Toth, J. 1963. A theoretical analysis of groundwater
flow in small drainage basins. J. Geophys. Res.
67:4795-4812.
U.S. Environmental Protection Agency. 1983
(revised). Methods for Chemical Analysis of Water
and Wastes. EPA-600/4-79-020. U.S. EPA,
Cincinnati, Ohio.
U.S. Environmental Protection Agency. 1984a. The
Acidic Deposition Phenomenon and Its Effects:
Critical Assessment Review Papers. Volume 1.
Atmospheric Sciences; Volume 2. Effects Scien-
ces. ORD: Washington, D.C. EPA-600/8-83-
016AFand-016BF.
U.S. Environmental Protection Agency. 1984b.
National Surface Water Survey: National Stream
Survey Draft 1985 Research Plan (EPA-ERL/C).
U.S. Environmental Protection Agency. 1985a.
National Surface Water Survey: National Stream
Survey; Draft 1985 Research Plan (March, 1985).
ORD. Washington, D.C.
U.S. Environmental Protection Agency. 1985b.
National Surface Water Survey; National Stream
Survey; Mid-Atlantic Phase I and Southern
93
-------
Screening Draft Research Plan. ORD, Washington,
D.C.
U.S. Environmental Protection Agency. 1985c.
National Surface Water Survey; Eastern Lake
Survey Phase II Research Plan. EPA-ERL/C.
U.S. Environmental Protection Agency. 1985d.
Direct/Delayed Response Project; Vol. I-V, August
30, 1985 (EPA-ERL/C).
Vannote, R. L, G. W. Minshall, K. W. Cummins, J.
R. Sedell, and C. E. Gushing. 1980. A river
continuum concept. Can. J. Fish. Aquat. Sci.
37:130-137.
Velbel, M. A. 1985a. Hydrogeochemical constraints
on mass balances in forested watersheds of the
southern Appalachians. The Chemistry of Weath-
ering, J. I. Drever (ed.). D. Reidel Publishing
Company, p. 231-247.
Velbel, M. A. 1985b. Geochemical mass balances
and weathering rates in forested watersheds of
the Southern Blue Ridge. Am. Jour, of Science,
285:904-930.
Velleman, P. F., and D. C. Hoaglin. 1981. Applica-
tions, Basics, and Computing of Exploratory Data
Analysis. Duxbury Press, Boston, Massachusetts.
pp. 354.
Watt, W. D., D. Scott, and S. Ray. 1979. Acidification
and other chemical changes in Halifax County
lakes after 21 years. Limnol. Oceanogr. 24:1154-
1161.
Wright, R. F., and E. T. Gjessing. 1976. Acid
precipitation: changes in the chemical composition
of lakes. Ambio 5:219-223.
94
-------
Appendix A
Cumulative Distributions for Chemical Variables
95
-------
Figure A. 1. Population distribution estimate for CO2 acidity, based on spring downstream averages.
1.0
c
.o
0.8
0.6
n 0.4
D
E
3
o
0.2 \
0.0
0.0
Data Subset = Downstream Spring Averages Variable = CO2 Acidity
1.0
Number of Reaches
Proportion § X
Upper 95% Cl
20 40 60
CC>2 Acidity (/ueq L"1)
80
100
Water Surface Area
Proportion ^ X
Upper 95% Cl
60
CO2 Acidity (/ueq L"
80 100
Length of Reaches
Proportion S X
Upper 95% Cl
Total Drainage Area
—— Proportion ^ X
Upper 95% Cl
20
40 60
COz Acidity (/ueq L
20 40 60
CO2 Acidity (/ueq L"
Number of
Reaches
Population Estimates
Water
Surface Area
(Hectares)
Reach
Length
(km)
Total
Watershed Area
(sq km)
Totals
2021
Actual
54
Sample Sizes
Unique
52
Effective
84
4633
8963
51215
20%ILE(/ueqL'1)
40%ILEU/eqL"1)
Median (/ueq L"1)
60%ILE(jueqLM)
80%ILE(jueqL"1)
31.38
35.62
37.33
39.18
46.98
28.22
33.90
35.74
39.00
46.49
28.96
35.50
36.64
39.22
47.21
28.20
35.32
38.76
42.81
46.39
Min
0.03
Sample Weighted Statistics (/ueq L 1)
Max
116.42
Mean
41.16
SD
19.34
96
-------
Figure A.2. Population distribution estimate for organic aluminum, based on spring downstream averages.
Data Subset = Downstream Spring Averages Variable = Organic Aluminum
I.U
0.8
c
o
'€
§0.6
it
§
S 0.4
3
E
O
.0.2
o n
0 5 10 15
Number of Reaches
., Proportion ^ X
_. Upper 95% Cl
I.U
0.8
|
§• 0.6
Q-
.1
.2 0.4 J
3
3
0
n o •
is /
\l
r
j
,1
i/
Jf
n
tj
;l Water Surface Area
/ :
20 25 30 05 10 15
Organic Aluminum (/ug L~1) Organic Aluminum
0.8
c
o
e
o
§0.6
a.
.1
.20.4.
E
3
O
00
0 5 10 15
Length of Reaches
— « — ; Proportion S X
Upper 95% Cl
0.8
c
o
C
o0.6
I
30.4
E
3
O
ft f\
nn .
Proportion ^ X
Upper 95% Cl
20 25 30
(/ug L-')
Total Drainage Area
_
-
20 25 30 05 10 15
— . Proportion g X
Upper 95% Cl
20 25 30
Organic Aluminum (ug L"1) Organic Aluminum (/ug L"1)
Population Estimates
Water Reach
Number of Surface Area Length
Reaches (Hectares) (km)
Totals 2021 4633 8963
20 %ILE Oug L'1) 1-27 1.26 1.32
40 %ILE (,ug I'1) 1.68 1.97 1.71
Median (jugL'1) 2.08 2.03 2.00
60 %ILE (ug L-1) 2.65 2.64 2.31
80 %ILE (ug L"1) 3-03 4.49 3.51
Sample Sizes
Sample Weighted Statistics (/ug
Actual Unique Effective Min Max Mean
54 36
84 0.47 15.22 2.65
Total
Watershed Area
(sq km)
51215
1.14
2.02
2.12
2.64
3.46
L-1)
SD
2.04
97
-------
Figure A.3. Population distribution estimate for total aluminum, based on spring downstream averages.
1.0
0.8
§• °-6
I
jo 0.4
1
o
0.2
0.0
1.0
0.8
c
o
a 0.6 •
ct
1
50.4
O
0.2
0.0 . —
0
Data Subset = Downstream Spring Averages Variable = Total Aluminum
1.0
r
Number of Reaches
.. I. . Proportion S X
Upper 95% Cl
0.8
c
o
0.6
ct
§
S 0.4 -I
3
E
3
O
0.2
50 100 150
Total Aluminum (fig L"1)
200
Length of Reaches
. Proportion & X
Upper 95% Cl
0.0
1.0
0.8
c
o
o
.1
JS 0.4-
O
0.2
0.0
50 100 150
Total Aluminum (fig L"1)
200
Water Surface Area
Proportion § X
Upper 95% Cl
50 100 150
Total Aluminum (/jg L"1)
200
Total Drainage Area
_ Proportion § X
Upper 95% Cl
50 100 150
Total Aluminum (/ig L~1)
200
Population Estimates
Totals
20 %ILE (fig L'1)
40 %ILE (fig L"1)
Median (fig L )
60 %ILE (fig L'1)
80 %ILE (fig L'1)
Actual
Number of
Reaches
2021
58.30
76.98
81.08
90.12
142.59
Sample Sizes
Unique Effective
Water
Surface Area
(Hectares)
4633
56.66
75.04
78.34
82.23
131.22
Min
Reach
Length
(km)
8963
55.70
77.84
82.35
93.62
142.63
Sample Weighted Statistics (fig
Max Mean
Total
Watershed Area
(sq km)
51215
55.87
74.87
75.38
76.24
102.61
L-1)
SD
54
54
84
38.67
355.33
101.09
54.06
98
-------
Figure A.4. Population distribution estimate for ammonium, based on spring downstream averages.
Data Subset = Downstream Spring Averages Variable = Ammonium
0.0
Number of Reaches
—— Proportion S X
Upper 95% Cl
1.0
c
o
0.8'
0.6
0.4
i
D
O
0.2
0.0
Water Surface Area
__ Proportion % X
Upper 95% Cl
Ammonium (/ueq L"1)
2 3
Ammonium (j/eq L"1)
1.0
0.8
c
o
c
I 0.6
Q-
I
| 0.4
B
3
O
0.2
0.0
Length of Reaches
___ Proportion ^ X
Upper 95% Cl
1.0
2 3
Ammonium (/ueq L"1)
0.8
c
o
'£
10.6
CL
I
| 0.4
E
o
0.2
0.0
Total Drainage Area
— Proportion S X
Upper 95% Cl
2 3
Ammonium (//eq L"1)
Population Estimates
Totals
20%ILEtueqLT1)
40%ILE(jueqL"1)
Media n(jueq L"1)
60 %ILE fr/eq L"1)
80%ILEOueqL"1)
Sample Sizes
Number of
Reaches
2021
0.54
0.61
0.86
0.90
1.36
Water
Surface Area
(Hectares)
4633
0.58
0.72
0.79
0.86
1.01
Reach
Length
(km)
8963
0.54
0.64
0.79
0.89
1.07
Sample Weighted Statistics Cueq
Total
Watershed Area
(sq km)
51215
0.59
0.78
0.86
0.90
1.01
L-1)
Actual
54
Unique
52
Effective
84
Min
0.25
Max
3.40
Mean
0.98
SD
0.62
-------
Figure A.6. Population distribution estimate for calcium, based on spring downstream averages.
Data Subset = Downstream Spring Averages _ Variable = Calcium
c
o
1.0
0.8
0.6
§ 0.4
1
I °'2
0.0
o
I
S
1.0
0.8
0.6
a o.4
"3 '
3 .0.2
0.0
Number of Reaches
Proportion £ X
.... Upper 95% Cl
1.0
0.8.
jo.6,
i'
•S0.4
£
3
u'o.2
100 200 300
Calcium foueq L"1)
Length of Reaches
Proportion SX
.... Upper 95% Cl
100 200 300
Calcium (/ueq L"1)
0.0
400 0
1.0
0.8,
.5
£0.6^
Q.
§0.4
I
3
5 0.2
0.0
400
Water Surface Area
Proportion s x
Upper 95% Cl
100 200 300
Calcium (f/eqL~1)
400
Total Drainage Area
__ Proportion S x
Upper 95% Cl
100 200 300
Calcium (jueq L"1)
400'
Population Estimates
Totals
20%ILE(peqL'1)
40%ILE(A/eqL"1)
Media n(peqL 1)
60%ILE(A(eqL"1)
80%ILEOueqL"1)
Actual
54
Number of
Reaches
2021
36.79
59.74
64.69
71.89
131.81
Sample Sizes
Unique Effective
54 84
Water
Surface Area
(Hectares)
4633
33.55
55.10
59.71
69.55
132.82
Min
25.45
Reach
Length
(km)
8963
35.82
59.80
65.86
84.67
136.73
Sample Weighted
Max
1588.5
Total
Watershed Area
(sq km)
51215
36.45
53.14
59.24
59.75
119.50
Statistics (//eq L'1)
Mean SD
190.79 362.59
700
-------
s
cn
u
00
ro
o>
CO
to
cn
S
10
o>
vl
CO
CO
Vl
CD
CO
|
C
s
c
CO
m
3:
CD
S
§
3,
5'
^fe
8
CD
2?
CO
o
CO
01
3
•o
CD
CO
8
CO
CO
m
13
CD
^
CD
CO
3-
CD
a
CO
m
S
o
co
n
ca a> 7 •> to
oo ifoo
f= F S F F
m m m m
33333
~~~~~
cn ro CD co cn
vl O CO vl vl
10 co cn on co
co vi o> cn ^>
CD CO 00 VI O>
vi on ^ vi to
O A ^ 00 CO
M io A cn -*
CD ro CD oo m
CD *« O) VI ^
*. M cn o oo
4* O> ^ CO -»
ai coco vi cn
10 ** -• CO O>
(O CO 00 O CO
IO CO A M CO
S
0
_f
a>
CO
CO
00
CO
O)
CO
cn
10
cn
.g Z
O CT
3- o
_ w
S 0 ffl
S > -i
Ul *"i
42. CD
01
3?^i?
3.S o
3-3-
0)
__ _ (p
1 12.
CD
U
|
CD
6'
3
m
S
3
Q)
Cumulative Proportion
Cumulative Proportion
O
Cumulative Proportion
p p p
cn
p
bo
-------
o
(71
2
00
0
8
8
en
10
CO
b
CO
^
en
CO
^
£2
C
C
3_
CD
!
1'
2
5'
2
Q)
X
2
i
CO
0
CO
CO
3
•a
CD
?P
8
01
CO
0)
3
•o
CD
1
CD
co"
3-
CD
a.
CO
£
CO*
o'
3"
CO
f",
— ^
O O 1" O O
c^ &** & S*^ 3""
FF™ FF
m m •* m m
Illll
-«^ --- "^^
fO _•_•_•_•
M 00 A CO O
-» o co -• a>
10 -.-«-. o
2co ro -> oo
C71 CD -> ^1
(0 -.-.-» o
a> oo ^ co CD
M -»-.-> O
-• CO 10 -» CO
4^ co oo 4^ en
I
01
10
§
*>
O)
CO
CO
CO
CO
o>
u
01
ho
01
^
^0 w
2 1
o or
3- CD
CD -"
v, o^
_ CO
cf » 5
™ * ffi
3 > -i
m
•S^c?
Is!
^
0)
-CD
^ 3 -»
?*" sT
->~
CD
0)
|
s
o
3
rn
J!J,
3
01
CO
3
Ei'
g
CO
CO
5"
o
ca
O
?
CQ
rj
5"
sL
g
CA
5"
CO
o
?
ca
j—
Cumulative Proportion
Cumulative Proportion
p
b
Cumulative Proportion
p p p p
'n *» b> oo
Cumulative Proportion
p p p P
o> °°
o •
10
O
O
I
3.
5'
a
a
3-
§
a: VT
2 |
sr o.
|
5'
CD
o.
s
-------
Figure A.8. Population distribution estimate for dissolved inorganic carbon, based on spring downstream averages.
1.0
0.8
c
I 0.6
a
3
c
o
0.4:
0-2
0.0
1.0
0.8
0.6
o
ol
§
1 0-4
3
E
5
0.2
0.0
Data Subset = Downstream Spring Averages
Number of Reaches
Prnpnrtinn S X
...-.Upper 95% Cl
5 10 15
Diss. Inorg. Carbon (mg L"1)
20
Length of Reaches
_^, Proportion § X
Upper 95% Cl
5 10 15
Diss. Inorg. Carbon (mg L~1)
20
0.0
1.0
0.8
c
o
C0.6
§0.4
1
3
E
50.2
0.0
Variable = Diss. Inorg. Carbon
Water Surface Area
Prnpnrtinn g X
. Upper 95% Cl
5 10 15
Diss. Inorg. Carbon (mg L"1)
Total Drainage Area
.—Proportion & X
Upper 95% Cl
5 10 15
Diss. Inorg. Carbon (mg L"1)
20
Population Estimates
Totals
20%ILE(mgL'1)
40%ILE(mgL'1)
Median (mg L"1)
60%ILE(mgL'1)
80%ILE(mgL"1)
Actual
Number of
Reaches
2021
1.12
1.36
1.61
1.74
2.54
Sample Sizes
Unique Effective
Water
Surface Area
(Hectares)
4633
0.92
1.26
1.41
1.56
2.66
Min
Reach
Length
(km)
8963
1.12
1.39
1.63
1.76
2.70
Sample Weighted Statistics (mg
Max Mean
Total
Watershed Area
(sq km)
51215
1.12
1.20
1.33
1.55
2.47
L-1)
SO
54
54
84
0.41
19.52
3.17
4.62
103
-------
en
^
en
CO
2
P
.
to
b
o
bo
to
o
CO
^
0
c
01
c
3
C
CD
3
G
1'
^
5'
2
01
X
CJ1
io en -» &• en
p p p O p
A en -» io en
-» o o o o
O ~J ^1 OJ en
-» en -» en tn
o o o o o
8^ en CD tn
to -• o en
S
en
to
IO
Ji
co
co
00
CD
0)
CO
01
IO
en
„ "Z.
CD §
01 3
O CT
3" CD
CD ""
Ul o
_ W
cfjg
r* O 0)
0) CD ?*
3 > ?
co 3
— ' 0)
Q}
i-a
I||
^
S
» 5 -i
fJLi
CD
01
TJ
O
•o
c
01
o'
3
m
S
3
01
CD
tn
g
y
fO
01
3_
o'
O
01
CT
O
3
CO
g
en
en
O
CO
01
3
0
0
01
5-
0
3
•g-
CO
^
Cumulative Proportion
p p p p
to ^ 05 bo
Cumulative Proportion
p p p p
to *» b) bo
01
3
'
O
01
3-
o
3
•o
o
•
Cumulative Proportion
o o o o
Cumulative Proportion
p p p p
Ko ^ en bo
D-
O
3
of
A
O
a
01
A
a>
V*
is
^«
o
S3.
a
o'
er
o
3
(35
a.
o
3
SO
•a
5'
to
a.
I
o
I
3
(Q
O
-------
Ol
.
^
00
Ol
8
§
8
2
Ol
CO
1
£
c
a
C
3
S'
c
a>
I
1
_
5"
s
s
CD
0)
3
CO
o
CO
I
"O
CD
CO
8
(0
CO
Q)
^
CD"
1
CO
3"
CD
Q.
CO
S
S
S™
TJ
0
c
oo o> ? ^ to
oof oo
gg ^3 Q. ^ ^g
F F f F F
m m ^ m m
O f? ^ o O
C c£ C C
00 ^ 10 -• O
— * Ol 00 G3 O>
00 O CD O CO
vl WM -«O
-• O 01 CO CO
to to o co O
vj CO 10 -> O
oo ro oo m -fc.
OI CO -» ^J CO
o> -• -» o o
CO '~J CO CO '-•
00 CO CO xj 4^
H
0)
CO
S
to
o>
CO
CO
00
CO
O)
co
Ol
to
Ol
.jj Z
S i
3-S"
co -•
CO o
_co
2o 1
§ * 3
Si- 3
ffi
*$s
3
p
co
ST
t
m
<°-S
01 o 2.
Cumulative Proportion
P P P P
io '*. b> bo
f
i
Cumulative Proportion
pop
k>
03
CD
-------
o
01
en
*»
2
2
^
o
io
O
00
i
u
•*>.
CO
CD
•^
O
b>
en
o
c
2L
C
3^
C
CD
i
o
1
2
j.
01
X
CD
01
O
CO
01
3
•a
o>
CO
i'
CO
01
1
CD"
<
CO
3-
CD
Q.
S
S
5'
en
"C
CO
o
^_i
§§|§S
^ cr^ — cr~ cf^
i- I- 2 r- I—
m m m m
r5i r$* col;* fjj
o o 2 o o
3, 3, 3,3,3,
NJ 10 10 -• -«
CO A O 00 CO
CO 4^ O) CO CO
O to to O CD
CO -.-.-»-»
O co >) o> to
•vi co bo io on
10 o> 10 10 co
CO 10 NJ -» -•
•c> co p >i co
en o 45* o> *^
A ^ co io on
to -.-»-»-»
~j CD OD en co
*> -» en ro to
S
CO
to
0
IO
S
CO
CO
oo
to
0
CO
CJ1
~*
en
» ^
tD J
3-S"
CD ""
(a o
CO
ill
01 CD S
-i >^ CD
CD > -i
42- S
01
s?§i?
3 CD S
— 5: 3-
fll
—^ CD
CO -i _j
"^ "T O
||E
CD
01
T)
0
•a
c
Q)
5'
^
m
01
g'
2i
CD
01
Cumulative Proportion
p p p p p
b io '*. b> bo
p
b
10
O
ST
O-
o
o
O)
O
00
o
8
?!?
Sfl
S? 3 CD
ollA01
~ X
Cumulative Proportion
p p p p
to ** b> bo
Cumulative Proportion
p p p p
Ko '*. b> bo
Cumulative Proportion
p p p p p
b io * b> bo
8-1
o
o
•fc- ro.
C/) O
o
3
§-
8
o
1
ili
I I CD
•a
2 o 01
CO 5. 8
to
c
3
?
•o
£
A
O
3
CT
A
O
3
I.
01
o-
o
I
1
•a
I
i
CD
01
3
I
CO
-------
s
g
00
VI
co
CO
_
CO
O)
8
to
to
VI
to
""•4
CO
^^
2
c
Q)
C
3
C
(D
i
s
$'
_
5*
3'
a
X
S
CD
0)
3
CO
o
CO
D>
3
•D
Weighted
CO
5?
S
0
VI
•fc
CO
o
3,
^_^
8§|S8
o^ o^ — • ^ °^
r- i- 2 i- i-
m m -1 m m
CO CO c" CO CO
o o Q o o
3l3Ai3i3i
10 -» _ _
10 VI U1 to CO
co 4^ 10 on o
-» vl O O) ^
to -.-»-»
-» en co -» co
CO -» vl CO fo
O) -^ Co en ^
to _._._>
CO O> *^ IO CO
en t bo !f> to
vi to 01 01 10
(O _> _. _i
-> ^. ^ to cc
-• -^ O -« O
co o> en to A
o1
Q)
5T
IO
O
ro
O)
co
co
00
CO
o>
co
01
en
31?
0} 3
3-S"
CD ->
cn o
CO
III
5? a> S
~* >*^ CD
-i
— (D
0)
If |
<
m
r*
•° S- o*
fjE
CD
ffl
^
•o
C
s
o
3
m
«-*•
i
S
CO
Cumulative Proportion
p p o o
Cumulative Proportion
p p p
' ' 01
IO
i
3
CO
9
3
D.
I
CO
-------
Figure A.13. Population distribution estimate for total fluoride, based on spring downstream averages.
Data Subset = Downstream Spring Averages Variable = Total Fluoride
0.0
1.0
0.8
c
o
'£;
i 0.6
a.
§
I
3
0.2
0.0
Number of Reaches
—__, Proportion S X
....Upper 95% Cl
246
Total Fluoride (//eq U"1)
Length of Reaches
__, Proportion S X
.... Upper 95% Cl
2 4
Total Fluoride (/Keq L"
0.8
c
o
S:
'§ 0.4
E
O
0.2
0.0
Water Surface Area
„„_ .Proportion S X
—.. Upper 95% Cl
Total Fluoride (/ueq L"1)
Total Drainage Area
Proportion
Upper 95% Cl
Total Fluoride (/ueq L )
Population Estimates
Number of
Reaches
Totals
20%ILE&ueqL~1)
40%ILE(//eq L"1)
Median (//eq L"1)
60%iLE(//eqL'1)
80%ILE(//eqL"1)
Sample Sizes
Actual Unique
54 51
2021
.03
.31
.42
.55
.64
Effective
84
Water
Surface Area
(Hectares)
4633
0.97
1.06
1.13
1.29
1.69
Min
0.82
Reach Total
Length Watershed Area
(km) (sq km)
8963
1.02
1.15
1.36
1.45
1.92
Sample Weighted Statistics (//eq L"1)
Max Mean
5.24 1.51
51215
0.97
1.02
1.11
1.41
1.62
SD
0.61
108
-------
en
2
00
o
CO
vl
M
a>
s
CO
en
CO
Ol
CO
CO
CO
CD
^
0
c
CD
C
3_
C
CD
rn
I'
2
5°
f
X
2
S
3
w
D
to
o
3
•o
CD
CO
§
CO
CO
0)
3
H.
g
CD
co'
3"
I
CO
S>
S
5'
CO
CO
r~
"•^
00 O> ? £i 10
OOJf 00
£ £ 5- S S
m m 3 m m
CO CO ^ CO CQ
*. CO hO ->
CO A vl CD O)
*. *. A b 00
Co eo o> to en
*> ho ho -•
NJ O) CO vl O>
en Idk A -» bo
co oo ho oo en
en co ho ->
CO 4s* C3) ^ O>
M -p. 01 CO (D
CO -» vl O 01
CO NJ N> -» -«
oo en en vi -»
CO vl -^ CO -*
co vi vi o en
3
CO
s
KJ
3)
CO
CO
00
o
J>
to
en
en
3>Z
||
_ CO
f ->> <
3 o m
ED CD r+
1-3 "
01
-,£"3!
-ii
s
D)
^^« CD
||l
CD
01
?
Z
S"
5'
3
m
2
3
n
CO
g
55'
CO
o^
0.
~
o
D
' -t
CO
1—
g
2.
S
Q.
S
CO
I
^?
Cumulative Proportion
Cumulative Proportion
o
D
CO
O
b
Cumulative Proportion
p p o
to i» b>
Cumulative Proportion
p
to
p
'*.
p
en
p
bo
I
i
|
a
1'
o.
I
5-'
A
S
I
a
S
3"
^
a.
o.
-------
01
2
S
on
on
M
01
00
§
01
00
00
o
•fr
bo
CO
^
S
c
g>
C
D
a'
c
CD
i
3
S
s
5'
S
S
3
00
o
00
Q)
1
CD"
00
S~"
CO
00
0)
3
•o
CD
1
CO
s
CD
O.
en
S
— •
o'
CO
•c
CD
_^
— '
00 O> ? 4* N>
0 0 §• 00
S^ S*^ & 5** 5^
p f= m p p
m m ^ m m
Ifffl
•— • '-—' -~?-?
O) * 4* CO M
00 CD CO ^J -J
cn M GO co o>
CO 4^ 00 NJ 4*
O> A CO U 10
00 O -J N3 ~J
4» to K> o *.
-» 4^. M Oi CJ1
^J Ol 4* CO IO
ro o co o> ^i
w bi co Vj u
O) CO 00 O> CJ1
O> 4^ CO OO (O
vj M 00 -" -vl
bs o o to b>
CO C3> O> IO -J
s
en
s
CO
CO
00
CO
o>
CO
(Jl
J~*
en
Number of
Reaches
00
Water
urface An
(Hectares
g>
*f*
3 to f5
** * ^ 3-
<
S5
^-«.
Cumulative Proportion
c. -a v>
III
to 3.
010
-------
Figure A.16. Population distribution estimate for dissolved manganese, based on spring downstream averages.
Data Subset = Downstream Spring Averages Variable = Dissolved Manganese
1.0
0.8
c
o
r 0.6
o
Q.
o
0.4
o.o
Number of Reaches
_ Proportion g X
Upper 95% Cl
1.0 tr
II
Water Surface Area
Proportion &X
.... Upper 95% Cl
10
20
30
40
50
Dissolved Manganese (fig L }
0.0
1.0
0.8
10 20 30 40
Dissolved Manganese (//g L"1)
50
Length of Reaches
___ Proportion S X
.... Upper 95% Cl
0.0
0.0
Total Drainage Area
__,Proportion §X
.... Upper 95% Cl
10 20 30 40
Dissolved Manganese (//g L"1)
10 20 30 40
Dissolved Manganese (/jg L"1)
50
Population Estimates
Number of
Reaches
Totals
20 %ILE (fjg L'1)
40%ILE(^gL~1)
Median (fjg L"1)
60%ILE(AigL"1}
80 %ILE (jug L'1)
Sample Sizes
Actual Unique
2021
0.77
2.71
5.75
7.79
16.86
Effective
Water
Surface Area
(Hectares)
4633
O.67
1.42
2.85
7.69
12.56
Min
Reach Total
Length Watershed Area
(km) (sq km)
8963
0.73
1.83
5.96
9.19
17.64
Sample Weighted Statistics (fig L"1)
Max Mean
51215
0.90
1.89
3.50
9.18
9.83
SD
54
47
84
0.00
49.33
10.37
12.99
-------
en
CO
CO
2
00
Ol
i
CO
bo
o>
o
3
^
o
c
SL
C
2.
^
s
rn
I
s
5'
0)
X
2
2
D
(O
O
CO
01
3
^
CD
CO
S~"
ca
CO
0)
3
e Weighte
a.
CO
at
2
— •
CO
ca
rj
. _T*
O O ^ O O
0 Q S g ^Q
" — Q) — ; ~
1 1 T 1 •
m rn ^ m m
3"? 3 "g"?
3 .3 ,Q -J J
O O CO CO 00
CO — * CO 00 00
Ol 00 *• >J -4
o o o co co
^ co "-' co b>
-> o cnoo *>.
O O CO CO CO
co b co bo *>
Ol «J ^ v) *.
O O O O CO
4> CO CO -» -J
2
Dl
CO
o
10
O)
CO
CO
at
CO
en
to
Ol
•9
? =
S 3
fS w
o o^
3" CD
CD ->
W o
co
Water
urface A
(Hectare
2-3
a
HI
$
s
„_ m
4o-
la"
~>
s
o
c
to
o'
3
m
£»
3
%
CO
Cumulative Proportion
CO
r*
o
'
I
to
CD
3
Cumulative Proportion
P P P
CD
3
ca
•n
ca'
I
Cumulative Proportion
Cumulative Proportion
O
b
vl .
5"
CO
§' ""I
3 1
« •_.,
™ t.
0 co.
S
CD
3
? ^ _
r-(
_/
«ik
1
!
•
o o P P
io '*>. o> on c
1 i
I ',
\
!'
Drainage Ai
. Proportion
• Upper 95%
Q
IA o>
u "*
X
L-N ^
^"llfcj" k
V
O O O O O ^"
_ b M '* a> cp P
XI •
ET
w oo -
c
o
_
O co .
X
2
CD
3
i" -
ca o
r;
— * .
%
Vi
^Xw
' 1 <
: 1
1 1 s
• "^
£5?
2 o c
^•03.
? 0 m
co5. S
en o r_
g?3 >
ol^S
~ X
•^^
^'^"^ir"*-^
^^\^_***%--
^^"WW^^T***
"%
%^
^»»
<
a>
2_
B
cr
CD
ii
3"
CO
c
g
S
?
S
CD
3
O
O
P
s
8
a
o
3
CB
T»
2
3'
CO
a.
<
S
§
3
i
5
S
CB
-------
co
2
§
2
05
to
00
en
00
vj
CO
CO
o
§
^
c
SL
C
3.
C
a>
i
S
i'
s
5
g-
Q)
X
g
co
=
CO
CO
D>
1
a
(A
S"
CO
CO
0)
•SL
CO
€
CD
3-
<»
Q.
J?
g
|.
CO
5"
C
3_
CO
§O> 7 .tk 10
of oo
r~ r~ °* P f~
m m_^ rn rn
xx^xx
cc?Jcc
3 3 2. 3 3
— •— '^
V) vl Vl Vl VI
01 co u io —
_. _. O vl M
VI Vl VI VIVI
bi ro io -i b
IO 00 >J CO CO
Vl VI Vl Vl Vl
en u ro 10 -»
00 ^ CO *^ — *
Vl VI VI Vl Vl
§M 10 -» -•
oo vi oo ro
i
(ft
§
£
C7)
W
CO
00
0)
U
01
M
C7I
Mumber of
Reaches
CO
•= c
fM
SSS
Icf5
o>
— f J
III
31
$
a
SSrl
111
5
CD
|
C
CD
5'
3
S1
3
0)
S"
Cumulative Proportion
Cumulative Proportion
m
a
c_
51
S
c
or
3
c
3
CO
Cumulative Proportion
p p o P P
b to v^ en CD
f
ur
i
•a
C
CO
i
• I SL
O
ox!
c_
cr
T3
c
-------
2
en
to
CO
en
at
8
io
CO
*
0
o
CO
b
1
c
ja"
c
X
CO
o
CO
m
3
•Q
ffl
CO
g™
en
CO
0)
mple Weighted
CO
&
w'
5
CO
-?
°|||°
F F= 2 F F
m m m m
fflfl
n n rj,ri n
CO -»-»-*
— en -« o tf)
en CD en en to
to co en *» en
OS CO -» O 00
on '-* '-* io *>.
en en A en co
10 -»-'-'
po on -> p po
^ en ^ en ^
CO O 10 ^ A
IO -•-•-»
co eo -» o co
co eo bo en ^
**l CD Cil Ol &
o"
S
en
CO
CO
CD
CO
en
CO
en
-*
en
z
ll
7
CO
Water
urface A
(Hectare
w 2
— 0)
V
III
J
T- O
3S.5L
1
T>
C
S
o
3
m
(0
timates
Cumulative Proportion
Cumulative Proportion
(0
a.
§
Q.
O"
I
I
8
f
3
•o
8
-------
2
2
2
00
-»
en
10
bo
00
6
00
00
c
01
c
£
c
CD
5
0>
O
1
_
A
5
2
01
X
CD
ffi
3
CO
o
CO
01
3
•a
CD
CO
i'
01
C/)
Q)
3
CD
co'
3
i
CO
S
S'
o'
CO
CD
3
§a> ? ^ 10
of oo
r= f= 2 r= r=
m m -1 m m
SE^.SS
»_ ^^ >_- • •
co en A eo ;-•
NJ ^j "si en NJ
o co *» ^ o
to -»-»-.-»
-> A co to o
a> en oo -^ ^i
-» en eo eo ^
|O _»-»_»-»
-• tn A co -«
to o> *k en eo
00 CO CO CO 10
en oo co en 10
to tn eo 10 o
51
at
01
O
10
a>
CO
CO
00
CO
en
CO
en
en
•^
lumber of
Reaches
_ OT
CD" -^
n £j
01 CD
S >
a s
01
r
9T ^
3 CQ
3-
1
JS 3
is.
CD
0)
Cumulative Proportion
Cumulative Proportion
§
-------
Ol
*
2
00
s
c
m
C
Si
c
CD
i
s
i'
c/>
CO
3
•Q
CD"
CO
CD
CO
CO
c?
00
5!
10
-J
f
a
2
i
M
O
CO
0)
I
a*
i Weighted
CO
s
5'
o'
CO
I"
CO
[—
§<3> » *» M
of oo
S? 3S 9: 6S 35
m m m m
?| 3|?
O CO 00 ^J O
A CO N> CO O3
CO ^ O5 CO Ol
CD CD -J ~J Ol
— NJ CD A K>
CO ^ ^ O> O
O 00 00 >4 O>
to CD en •-J M
00 00 00 ^1 O>
CD to O CO CD
to 10 00 -> CD
?
5T
to
o
10
^
m
CO
CO
00
CO
O)
CO
en
to
on
Number of
Reaches
CO
i c
® « S
S O Q)
3^ CD
> -i
0!
11 |
Q)
r*
fii
3
CO
3>
1
cF
o
1
§
1
CO
CO
o'
0)
CO
en
r^
8
•g
CO
r,
Cumulative Proportion
Cumulative Proportion
m
?
CD
i
>
•o
c
Q.
I
Cumulative Proportion
p p p o P
p ro ja. g> oo
Cumulative Proportion
pop
k> ^ b>
CO
-------
(71
(71
00
10
bo
OS
10
OS
A
CO
£
OS
o
J,
c
c
3
.5'
c
CD
!
i'
5
5"
I
X
g
CD
0)
CO
o
CO
m
3
CD"
co
§'
CO
CO
01
3
13
CD
2.
jj"
CD
Q,
CO
g
2.
o'
CO
-^
§o> ^ *. io
OCD °°
5^ cS 9r ^ 5^
m m m m
££%££
r; r; "^r; r;
— •• — """ • — •• — •
00 O> (71 (71 A
-• VI OB 10 A
(7i b> io bo io
-» 10 CO OS 00
vl (71 (71 A A
ps co ro vi ->
vl vl CO (71 O
00 OS (71 (71 4^
-• 10 Vl |0 -•
(31 vl (0 CO OS
CO (71 00 CO VI
vl (71 (71 A A
7* 00 10 vl vl
£71 CO
CO
"* ** C
III
» CD "
S >3
2-S
0)
^ r~ 3
T 2 (D
3 (D n
^™* ?• ^*
~^
S
— * to
!|!
CD
Q)
O
C
m
5'
i?
m
§.
3
i
Cumulative Proportion
Cumulative Proportion
I
3
>
io
•o
C
Cumulative Proportion
Cumulative Proportion
c/i
o
o.
E'
3
•5-
CD
(71
O
8-
§
•-* M
10
(71
O
(O
8
p
io
p
b>
p
oo
•D g t»
Iff
ollVg
~ x
p
b
p
io
s-
8
CO
I
" 1
10
or
O
8
p
b>
p
bo
;i i
:! t
^ 7 w
o>
-
CT
CD
II
cn
a
-------
8
2
8
o
io
o
o>
00
10
u
io
~*
to
8
a
•
c
.5'
c
CD
i
o
1
^
„
3
D)
X
2
CD
01
cn
0
en
0)
3
"O
-• O
fO *si O ^i O>
*» 10 10 -• O
O co co en ~j
10 CO O Ol **
*. ro io -> -«
01 CD CO vj O
CO -N| *» to O>
GO IO IO -» O
eji co io 01 ij
*» O -J -» 10
o
co"
0
l-^
A
u
CO
00
CO
o>
u
CJ1
10
01
3 ?
Si
o cr
J CO
CD ->
CO o
en
III
0) (D «-*
"^ -w ®
(D ^ "^
CO "^
~2
D)
WCff
-|l
^
— ^
c?
CO
—1
c
3;
5:
^
Z
H
c
3;
t
Z
H
Cumulative Proportion
Cumulative Proportion
Cumulative Proportion
c
3;
CL
•<
Z
•o. -
*-..
.. I
C"-D CO
III
co 3. 8
S°*
o'£S
n
•D
2.
3
to
a
o
II
IS
: I
I*
ty n
a! '
^'
-------
Appendix B
Geographic Data for Stage II Sites
Table B.1. Geographic Data for Phase I—Pilot Survey Second Stage sampling sites.
Reach
ID#
Stream
Name
County of
Location
Grid
Latitude
Site
Longitude
7-V2 Minute
Map Stream
Location
Number
of
Reaches
Drainage
Area
(Mi.2)
Crew
ID
Code
Chattanooga
2A00701
2A07702
2A07703
Knoxville
2A07801
2A07802
2A07803
2A07805
2A07806
2A07807
2A07808
2A07810
2A07811
2A07812
2A07813
2A07814
2A07815
2A07816
2A07817
2A07818
2A07819
2A07820
2A07821
2A07881
2A07882
2A07822
2A07823
2A07824
2A07825
2A07826
2A07827
2A07828
2A07829
Sugar Cove Branch
Childers Creek
Hall Creek
Big Fork
Puncheon Fork
Unnamed Trib. of
Ellejoy Crk.
Cosby Creek
Roaring Fork
North Fork
Armstrong Creek
Little River Gorge
False Gap Prong
Correll Branch
Little Sandy Mush
Creek
Reems Creek
Curtis Creek
Eagle Creek
(Gunna Creek)
Forney Creek
Bunches Creek
Crooked Creek
Pigeon River
Grassy Creek
Walnut Creek
Little Branch Cr.
Sweetwater Creek
Brush Creek
Middle Prong
S. Fork Mills R.
Henderson Creek
Welch Mill Creek
Whiteoak Creek
Catheys Creek
Monroe/FC
Polk
Polk
Cocke
Madison
Blount
Cocke
Madison
Buncombe
McDowell
Sevier/P
Sevier/P
Haywood/P
Madison
Buncombe
McDowell
Swain/P
Swain/P
Swain/I
McDowell
Haywood
Henderson
Madison
Haywood
Graham
Swain
Haywood/FP
Tran./FP
Henderson
Cherokee
Macon/FN
Tran.
35°19'56"
35°12'30"
35°05'56"
35°54'43"
35°55'35"
35°46'40"
35°47'37"
35°48'11"
35°48'57"
35°49'20"
35°40'04"
35°40'35"
35°41'0r
35°41'24"
35°41'50"
35°42'18"
35°32'57"
35°33'28"
35°33'52"
35°35'39"
35°27'48"
35°28'36"
35°48'4r
35°27'16"
35°19'15"
35°19'46"
35°20'41"
35=21 '07"
35°21'33"
35°12'00"
35°12'35"
35°13'57"
84°03'46"
84°28'30"
84°19'47"
83W03"
82°32'19"
83°47'47"
83°14'09"
82°57'24"
82°23'39"
82°06r57"
83°38'52"
83°22'11"
83°05'28"
82°48'42"
82°31'54"
82°15'09"
83°46'59"
83°13")1"
83°30'31"
82°06'33"
82°48'06"
82°14'44"
82°40'28"
83W47"
83°46'07"
83°29'31"
82°56'05"
82°39'21"
82°22'36"
83°54'09"
83°22'36"
82°47'23"
Big Junction (TN-NC)
McFarland (TN)
Isabella (TN-NC)
Neddy Mtn. (TN)
Sams Gap (NC-TN)
Wildwood
-------
Table B.1.
Q-j_ _L_
Heacn
ID#
(continued)
Stream
Name
County of
Location
Grid Site
Latitude
Longitude
7-1/2 Minute Number
Map Stream ~x
Ul
Location Reaches
Drainage
Area
(Mi.2)
Crew
ir\
IU
Code
Knoxville (continued)
2A07830
2A07831
2A07832
2A07833
2A07834
2A07835
2A07881
2A07882
2A07891*
2A07892*
2A07893*
2A07894*
2A07895*
2A07896*
Rome
2A08801
2A08802
2A08803
2A08804
2A08805
2A08806
2A08808
2A08809
2A08810
2A08811
2A08891*
Greenville
2A08901
2A08902
2A08903
2A08904
2A08905
2A08906
Mud Creek
N. Pacolet River
Tusquitee Cr.
Allison Creek
Brush Creek
Middle River
Walnut Creek
Little Branch Cr.
Cosby Creek
Twentymile Creek
Jarrett Creek
Slope Fork
Moses Creek
Pinnacle Branch
Unnamed Trib.
to Perry Creek
Dunn Mill Creek
Owenby Creek
Bear Creek
Weaver Creek
Unnamed Trib. to
Kiutuestia Creek
Whitpath Creek
Tickanetley Creek
Bryant Creek
Hinton Creek
Chester Creek
Persimmon Cr.
West Fork
Nottely River
She Creek
Chattahoochee
River
Deep Creek
Henderson
Polk
Clay
Macon
Macon
Greenville
Madison
Haywood
Cocke/P
Swain/P
Macon/FN
Macon
Jackson
Macon
Murray
Fannin
Fannin
Gilmer/Fch
Fannin
Union
Gilmer
Gilmer
Lumpkin/Fch
Pickens
Fannin/Fch
Rabun
Rabun
Union
Rabun
White
Hebersham
35°14'32"
35°14'54"
35°05'37"
35°06'08"
35°06'35"
35°07'30"
35°48'27"
35°03'39"
35°44'50"
35°28'00"
35°08'50"
35°03'48"
35°19'30"
35°03'27"
34°57'15"
34°57'48"
34°58'07"
35°50'45"
34°51'07"
34°51'33"
34°44'04"
34°37'27"
34°37'48"
34°30'26"
34°39'30"
34°56'56"
34°57'25"
34°49'07"
34°50'19"
34°42'35"
34°43'16"
82°30'48"
82°14'00"
82°45'23"
83°28'42"
83°12'13"
82°39'00"
83°43'36"
83°27'04"
83°12'00"
83°52'25"
83°37'00"
83°26'11"
83W15"
83°28'03"
84°43'33"
84°26'57"
84°10'21"
84°35'04"
84°18'33"
84"OV55"
84°26'22"
84°17'45"
84°01'15"
84°25'41"
84°10'40"
83°29'08"
83°12'46"
83°53'46"
83°20'37"
83"44'59"
83°28'27"
Hendersonville (NC)
Standingstone Mtn.
(NC)
Horse Shoe (NC)
Inman (SC-NC)
(15' map)
Tigersville (SC-NC)
(15' map)
Hayesville (NC)
Shooting Creek (NC)
Prentiss (NC)
Rainbow Springs
(NC)
Scaly Mountain (NC)
Highlands (NC-GA)
Cleveland (SO
Table Rock (SC-NC)
Marshall (NC)
Hazelwood (NC)
Hartford (TN-NC)
Luftee Knob (NC-TN)
Fontana Dam (NC)
Cades Cove (TN-NC)
Wayah Bald (NC)
Topton (NC)
Prentiss (NC)
Tuckasegee (NC)
Sam Knob (NC)
Prentiss (NC)
Tennga (GA-TN)
Epworth (GA-TN)
Culberson (GA-NC)
Dyer Gap (GA)
Blue Ridge (GA)
Mulkey Gap (GA)
Ellijay (GA)
Tickanetley (GA)
Suches, Dahlonega
Campbell Mtn. (GA)
Jasper (GA)
Dyke (GA)
Noontootla (GA)
Hightower Bald
(GA-NC)
Dillard (GA-NC)
Satulah (GA-SC-NC)
Coosa Bald (GA)
Rainy Mountain
(GA-SC)
Cowrock (GA)
Helen (GA)
Tallulah Falls (GA)
1
1
1
11
1
1
5
1
1
1
1
-
1
?
1
1
3
1
1
1
1
13
1
1
1
1
12
27
1
5
1
6
25
40
6
4
11
37
4
3-1/2
7
6
-
11
?
1-3/4
2
4
1
3
3/4
3
18
2
1/4
2
6-1/2
45
58
6
32
4-1/2
T1B
T1B
W2R
W2R
T1R
S2B
TH2G
T2R
TH2R
M2B
S1R
S2R
T2B
S2R
M1G
T2G
TH1G
M2R
TH1G
TH1R
TH1G
M1B
W2B
M1B
W2B
S2R
T1R
TH1R
M1R
T2R
M1R
•Special interest points.
720
-------
Appendix C
Geographic Data
121
-------
to
ox
30W
om
m<
rv) M «»*»e» M <^OVM o 04 -*MW -*oo>o -* •* f\> 030
m
r
Itl
SI
t« I* It It I* It |t |t |t |t |t |i |f |t |« |t
W W -A -k I\J -k
t it it it {fifi?*!! i» jl^i*!* it*i*iS°it iTit*it"ii°il"il^it0it^11°i^it0liT*
QOOOOOOOOOOOOOOOOOOOOOQOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOQO
-------
OJC
l-«w
9
M
*i *l *l
Or-
t| i| t|
u I •< it *( M i| •!
i! c| I I I I •!
UJ
_J
UJ
UJ
UJO
ceac
xo
to
• t»««flllllflllttl*ll|l 1)111 ••«•!( *f«*l»>tlt«<»l|
OOOOOOOCOOOOOOOOOOOOO OOOOOOv-vOOw-OOOOOOOO " OOOOO
0£t-t OOOQOOQOOOOOOaOOOOOOOOOOOOOQOOOOOOOOOQOOOOOOOQOOOOOOOOO
723
-------
O.C ••••it*
OOOOOOO
X
»-**
OS!
o in r*
« •» rsi
19X | • I (I • I
in rj oo
)_ fM N 00
X nj oo •»
W I tl »l *l
M*» r- O O
OX
(A -J
Ul
•CM
tu
KOI •*»- OON.V
IUK I r-l irti ^l
>ui
wo
KOC
xo
OOOOC30C3
OOOOOOO
wo Bowooeoooooeo
«M OOOOOOO
724
-------
Appendix D
Chemical Data
Glossary for Chemical Data
The variable SAMPLE identifies the sampling time intervals. Codes 0, 1, 2,
3, and 4 refer, respectively, to sampling intervals SPO, SP1, SP2, SP3, and
SP4 in Table 3-1 of this report. Streams Corresponding to the stream
identification numbers (STRM-ID) are listed in Appendix B. Other variables
are identified below. For calculating spring index values, observations marked
with "E" after SAMPLE number should be replaced by the average of the
remaining two observations in SAMPLES 1 through 3. SAMPLES numbered
0 are not included in the calculation of index values.
The variable "WGT" is equal to "[/a^. Whenever making explicit population
estimates and their variances, WGT must be multiplied by the Stage II grid
point density (128 miVdata point) to obtain values of "w" used in Equations
2.1 and 2.6.
Variable
Name
NA16
K16
CA16
MG16
FE11
MN11
H16
ALKA1 1
S0416
IM0316
CL16
FTL16
SI0211
COND11
ALTL1 1
ALEX1 1
ALOR11
DOC11
COLVAL
TURVAL
ORGION
PTL11
NH416
CONIS
TM PSTR
PH-CLO
PH-OPN
PHSTVL
DICVAL
ALKA1 1
PHEQ11
PHAC11
PHAL1 1
DICE1 1
Chemical Definition
Sodium
Potassium
Calcium
Magnesium
Iron
Manganese
Hydrogen ion activity
Alkalinity
Sulfate ion
Nitrate ion
Chloride ion
Fluoride ion, Total
Silica
Conductivity
Aluminum, total
Aluminum, extractable
Aluminum, organic
Dissolved organic carbon
Color value
Turbidity value
Organic ion
Total phosphorous
Ammonium ion
Conductivity, in situ
Stream temperature
pH, routine closed
pH, routine open
pH, station value
Dissolved inorganic carbon
Alkalinity
pH, air equilibrated
pH, initial acidity
pH, initial alkalinity
Dissolved inorganic carbon (equil.)
125
Units
//eq L~1
//eq L"1
//eq L"1
//eq L"1
;ueq L
//eq L"
A
-------
iX; f w IV* ' *J *v N* *\i fVJ I"1*1 $ ?v f\j f v r*j ^i
O O O O O O O O O O O O OX*
*sj -~^i -*^ "«i4 "Sj JNij *^ -^J ^ -«^ a**ij "^ -Nj "^£
** v -M 'M#* fXt '-^ '"^0 ow "*^ "^ ""*^ "^ "**J "^J J
o b ob ,00 o o o o o o o M
re re re re re re f
o o o o o o o o o o a o o
It «• •* I* 8 » *• •
-sO W^ i>/i 0^ 0"* *j^ 'w1^ *^ W^ cJ* v^- *O C»i
fv ui ui r\j r^ ro (xt O O *• ".»» flJ fJ
m >
Jt
TO
r
oooo
3B
'jo
m
• « * • •* • • • « » * t • * t • t * l • t* « • B I i Itiii* • g * « • * •jb
OOO»O OO OOO«^t O OOOO OO OOOO OO OOOO O GOQQt/l O-J OO^O -»
~* IVt-*-*OO OO
ooow oo oooo o oooui oo ooow oa aooo o ooou«o oo ooout •*
OOOO OO OOOO O OOOO OO OOOO OO OOOO O OOOOO OO OOOa
• * t t tt » • • t I til* »l • 1 • I • I l*»f * «•«•* »» »*il X
oo oooo o a-»-*o oo oooo rv»-* -*-*-»o o ooooo ^^t -*-»-*Q -»
-------
** QOOOQ
X * ttfft f • t » » • f tft« I till » * I * I t « » t » * «•!••
O OOOOO O OOOOO O OOOO O OOOO O OOOO O OOOO O C3OOOO
«~ O N.OOOO O «*mOOO G K>OOO O OOOO O OOQW1 O OOOO O N-OOlTiQ
T- I • * I 9 I I lilt* • (lit ( I • • I I • • • • • • • I • f I f I t t
%. T- OOT-«-(\l f\i fV»~O«*fn T" «r-t»O-*t T- OOO-* *•« * *(•• I I » I * •
uj ao N^otnor^ «o f»»T»o<»~h- *- oeor*^ tn r-orao T- oooo O
U. sr>T-C>J»- T-«M *- T-
•o t> tr-*so«s>«h »n ^.rsjoGr-sr o «p«r-*AO o
T- • I • I I • ( I • I • • • lit* t
o
IA
^111 (II III (Jill
si- io^fni>^«& o opaoooec N. sfo-sroo »•«? r- Ofs.>*h- on
N. O>O>Cf*-N. lA fNJOOOO Is- OO^Ol- ^rt MSlA^N- ift intA-Sf^ •* lOWM/^h. IT*
,_,_,_,__. r» r-T— T- (VjMMfSlfVl
o «TJOO«?O ^
* • • • » »
W) ;*: CC.1 (M twin's* rvf QOr-f^N- ^ •*•»*<•« *- ^-r-Kif^ O OO-OIS. N- fMf>j>*"<> o
»*"T"N-K>fvi r%f f^Of^1^ ^ wf^-c? «O oo
T*~ • » • • • « • 1(1*1 * I I » • • I • • I I III* I • • • • •
«£ in »^N.>OO*^ O MfOtrnA*? fM fOpvfVO- fw MK"irn~ fW --Opk-O1'* O Ws-OMO 60
SK >/* M3sON.fs.N- tn OOi^OO *O *OX3(A-O Sf fOMK1!^ pi r>j(NfMfO (M
UJ
-J
0.
f\i N,
O ID
O ts".
fs| IV
N- 6
o *v
I *
o o
or.- N.
^o o
N- rj
Of r»
o **
or-
IS, N,
o o
o
N-
o
It
CD
o
oo
f»
o
«r
«o
o
00
N.
o
rg
-J
o
N.
o
O w-
o
«r
N
o
tf
o -Jl
T- ra
T- »~
00 OO
N- N.
o o
«* <
-------
ON- r*SK3*CO N.
«~o fT-fiT- fv T-rv»**T» •*• *fo *~ r-oji*1)*- KI»» oorvi*- o
X •• I i I i i •• «iii * iiii i 1111 . . • . . , • 111* «
OO OOOOO OO OCOO O OOOO O OOOO «~ OOOO OO OOOO O
*• OO K»O«AOO OO «*OOO O OOOO O
f *• t t t « • II Illl I 1*11 *
3K O «M«-*-OC M K>OO«~ O v-
ar
O OOOO OO IAOOO O
I Itll II Illl I
OfV
t- OO OOWOQ OKI OOOO O OOOO Q OOOO O OOOO OO OVAOO O
f" •• tllll II Illl I Illl I Illl I 1*11 II Illl I
UJ
oo
XT
T-T-T.^_ r «~r\|(vrs( CM
IT I/MOW*!
C5 -O
V \C
O OOOO CO
"O
ir
•4-O-tAfvt
r-in
o
-*•
o
rv
o
I
tu
o.
K1-* T-'
O N.
r- O
O *•
OO
o
I
r~
O
O
O
o-
o
o
o
o
*
o
o
o
o
ff
I
c
N-
fM
O
I
O
f*
*
c;
o
3^ ^*
-------
O
T*r*'*«'r" O r"T""»»*~ i" »~«">»»»O 1"»O r"»"«"v rM
X I « I i » li tilt t III* t f*fi t* «*it i * i t i • tt »•••
OOOOO OO OOOO O OOOO O OOOO OO OOOO O O'OOOO OO OOOO
T- mQOOQ OO MOOO O OOOO O mOOO OO inOOO O K-iOOOO OO
X -OKiT-^nrv Gf\» T-Qwrv N- inrvvtn O
SET »•» oo^ *)v0o O
O
* i 1*11 ii • • • i •
VI ooo«o **
>o »-Or-^ »*
r«.ir> X>(S,voiri *M«» ^-N-norrt ao -^rvMr^-f^ ^N- aco-o»
** ***• * **t* * ii** «• •*•• i **•«» •• iii
tow oOf~f\» N> aoooc^^s oo
rv
O
UJ
-J
eu
at
v *c* ^ o o f*» tw rw rv rv f\~ c^ o Q*
•38 in in N. PW f> O rvi <%.i(v, fw fV <>O
«~ ^ K-, K*. T- in tnKi KHO& oo *^ »* »*
O Of^ »^ «- «~ T- r-O OO o «vi
I II II I* II II II
f" O O O O O O O C O O O O
-j r>j s> -i o»4 r>j r>-i 3-1
O O*~ r~fV r>jm fnxf -^fsn lf\ -O
^ CO OO fs flD 'X? <1O OO *XJ OO **" iSO 30
fv- p*». r^- fft^. fv f*»» N» ^v fv« p*-*, PS* ^» PV,
O OO OO OO OO OO OO
IM rv *v rv' rj
-------
Q
IX)
a
(Sj
o
fu
o
JO
I\J fV»
o o
c*j
O
ru
j*
O
•Ni
C
o
-J
JO
rNJ
(NJ
O
O
•>
fV
o
o
ce
O O
o -»
t» yt»
, ** ot
o
o
ru
O
CJ
o
o
V/l
(w
V/l
.3
i
N»
O SC
-o m
4/S
3P-
at
r-
m
'^J!S V^.JP^^^^* CJ ^O *-HO^ v^ijt3Os W4 v^^CjtXj^N^C^* ^p**^* ^NjJ^^N^* fsjfx^ v^'-H*j^i WJ LrJ ^^ J*s*-^*^«^tN "OC3 y**
C*JO^* ^VOt>O^t% ^O-^» 4?^\>1'^J'*Sji,^l C5 •,"& "*W^J -O -*1 ^fX> '^4j~nJ-.«*.fv -^ ~^i
«AMA TnA -^ naA i.i>
"*O rv^^^^o coco
OO ^3 <4> 0^ J9 *** *43-&°- ''OfNrfOO v/iC^ tK>-O-OO"** Ot?
h«A4M*>*lA«^ .^
l»V>4«Mk —»!** -SJ MjkO.^0,)^ !MA«Jh fV^Ao*^ 4MA«Jk '^4 J^J »«k (NJ JWI* f\J C£> O^ O1 CK O **^ JC
v/l ^i^u4^A>O OCS w4^}U£>43 OOi^l ^«^kiOM4^) ^ •*^*!>^*CK jt'CO CT>
t*^«^k O^i^^^^^ 4>^1^4 'I^'MCM^O^^^ Oti *•* ^O ty~ "^St O^ ^^W^ (VJ1*1^ *•»•"* wA ^4«OO^ ^^^4^^**"^ ^*K~^> ^tU&CDCi)*** •O f^JJ^^^Oi ^"•O1. f^l
OO OOO-O OO OOOOO O OUiQQQ OO OOOO OO OOOWOO O OOOO OO -*
ok
OO^OO <5 "^4 !\>fO~*O-O UxO "*OOO ««*** fSJfUOO-* —* Of^JOOOO "^•"g 2t
oo oooo oo oooorj a oaoo-~4 oo ooa-g oo oooooo o oooo. oo -*
OO OOOO OO OOOOO O OOOOO OO OOOO OO OOOOO O OOOO OO
0*0
(V
01 (Jt
X
fsj OOOO OK-* -*
-------
Id.
iv M IVJP ru r\j r\t f\< iv rv f\j ru 'v w
O OO OO OO OO OO 0 X»
(v *•* fv> iv *** *•* wt SMA ^ ^ 's/j iw ^?
r re re re re re r
o o a o o o o o o o
i ii i « it • » i
*» f+ a ao OPJ i\»«"* <-*
O O*^ J>W UJf*J f\J "s| ««J
W1 W1 O^ C^* f^ '^ ^V fx* '^ OO £T5
O O'j»i o4-i —* fsj f\>v/< wiH
VI
Co
r-
m
.
O VJ1WO-* i\>-* ^i^i-OO O
I till *
O fvlCK-O-* -»*sJ OOO»-*O O
-*-*
*« ooChwiO U!H> hj-*aw *»
>o
3t
se •sjoo O
vn vn-*oou* o<.oe O'Wico-^ rv •ONWv> *• a»"^ouio v>»w »>»uiwiv o.
-n
oo o-i^iNj *o. ^.-ooyiu* oo
OO-slO O OO-MO OO OOU1U4 O OOOU1 O OOOWIO OO OOOsO ~*
tv»-li ~*-*
'^CJCJCJ *** *^C3C*)"*^ 00*^ O^Oii'*i40^ **^ Ksi-i^rof^ 4""* fSjt'»M'4-iii*C3C3 C3C5
oooo oooo o ooo^fl oo oowo o *nao** o oboo« oo oooo
OOOO OOOO O OOOO OO OOOO O OOOO O OOOOO OO OOOO
• ill •«*• I • I • I II till I It** * i • i • • *i « • • •
Jv -*rw-*-» OO OOOO -* -*-*-*O *» IVJNi^^lv ~»~k OOOO
cm
-------
Z£l
r\j njix» rutv N> ru ro rursi fv ru rw
o oo oo oo o oo O o O;o
'jc. ao Ot t» .30 Cx> OS OO -si *^ -4 -s^ . ->J jfc
u> goo» -jooa o> -ju O» cw Cu o» (X> O» I
O OO OO OO O ^D-O *O "O '"Oi^
4S 4^ tsi O^fsJ fV<*^ x*^ O-- U1 W1 4> O4O
c re re re r re r r r
-» -»o oo oo o
I II It tl I II
O OW O4** 4f-U1 (J«
—k -j> O o > o-* -*
O OO QW t\>«* «*
3t
t)
I-
m
o oooo o uJw*^*^^ ->J*»NI*- o
• • « « • » • • • • * * » » » t iitit
** .CnOfsi-* --O "^^oui vO -*•« ("•>*•> "Ni U1-**OO O
o t\j-*oo
I lit* I III*
O
a*
i *ttt t tltt t tttl t ttttt tttt t tit* lilt I l I l I *k
o
o -si -si 0.0 rvt -^ -^oJ<>--vJc» r\jrxj-»iv o-i ut4>4>4>> t«4t\Jrorgi 4»o4oJUlu4 JE
o o*
I till I I
o o«-sj-*o 4> rv*Crt-»o*s< -*o-»w tsj ooruwi
-------
*v isjrw (NJ rv i\> j\» pa i\i Mfw ru rv t\t c/>
O OO OO OO OO OO OO O »
(A c» 96 oo os OB o> oo oo oa <» oou> jt) :*
to «• oe ou oo c» oo GO tx» eo c* cc oo oo|
O o-* -*-k «* O OO OO OO OM
—* <»»«jk ««*CJ O -O 'OOO OOO- O^VI U1O
c: re re re re re re r
O OO OO OO O~» «*O O
I • • It tl tt It t
**4 *>4Vl UlVt UIO4 O>4 O O«O4 O4
"Sj >Sti>i Uti«»k «A «A «*Qs (^ O O
x/i <-n -* -* O o -^ -^ O» O>rv) (\*
•O -O -* —* O OfJ rUQK O> -* -* (7j
*- *-"V» VI4N *»O O.1^ "^Ul WlH
V>
at
r
m
ce» -»N»O.O>O o
* »«•• • *•*•• i • • • • » • • i i i •
O r\JOf\)Ul CK 01O-NJWO -O oJ-*OOO -J U-O-A^O v» XJChUWn >| O^O--N-* O
Ov -WO>C»vfl*- Wl WCOCS-* O -OO->J-O O
* till * • I • I > « • > I • * tilt* t • • • • t fttt I
04 ^JS^SO- O4 U(M'J1*^xO O W*^C»-* 4S OOOCttOOO O OOONJO 06 *»<*-*UI O
o
*- O oo co oo o t*ijoo>o •* wno^^u^ ** v^>u»-*ui o
* till I • • • • • I (III I I I I t I I • I I I I I I I I I
04
ru nj-*-ooo -* wrvjsruo *> >oc»ooc» «si -goo-N* -J
I I I I I I I • I I I * I I I • I I I • I I I III! • till •
O co-»u<-* o ooou4-*-k o- o-*-»a- CK oooooa> o
O- *»*» *~ -»W-OiJ«l\» U1
<-* o ut-sjwoa -M MO.WINJ-* M OIU«M O WlOOvt O OOOO O OOOO -»
o *»ooo w *>u4**M*» o
• «tt* • *t*tt » I » » I I I I I I t * III! • lit* • •••(
O OOOO O OOOOO O VAOOO O OOOOO O OOOUl O OOO-J O OOO-O
O OOOO O OOOOO O OOOO O OOOOO O OOOO O OOOO O OOOO
* till • tttii t i • i • i • i i i i • «iit « i < * I • ••••
M rs>4NN»N> -» -*-*-*-*C> O -*-*-*O IV M«*04«*l\) -* OOO-* -* M-A-*fNi Ut
oo oeo-4-* oa -*-^4>ooo so -»M
-------
•O
*~
X
r- ojvfnjv wo sc*O*0-»<» >o«Q .^•("v.vi IAN.
«M Kir-*-*- r»ON-
as*- K*T-O»»-* Or- fMOO-st fMK>
»- OOOO OO OOOOO O OQIAO OO T-OOOO OO OOOO OO OOOOQ O
**" •••• VI VVVVII V IVRt !• VflVftV ft 9V9V Ri iitii 9
ut N.N»»*O "4TV ow>or-rM N. *tv>eo«j ON.
U. t-rnf^sf r- (Vf^Kt^rKi O
N-N-oo
oora
o
O
NO-
-OO
N-o
to
o
oo
*
O
-O r>K^«-,o o^o
*N. «vj»»
r-9c K*
sfN.riN.v- r-o
oiifi ii
ON-
inao
i* 11111 *
KN- Mr\»uM^»
O
O
r>
o
Ch
oc.
O
O
ru
O
O
or
o
134
-------
IV
o
"Sj
O
U(
r
rw
3*
O
•si
o»
o
v*
c
rv
*»
o
oo
O
se/
rv
>
C
-si
OO
O
*SJ
G
*-J
Cfc
O
IXJ
Q
-si
c;
O
•M
03
O
(XI
>
o
•M
"M
O
O
"-J
*>1
O
o
•M
-N»
O
fj
C
ru PJ
o o
-si "si
•sj -si
O O
O »
-si r
O f
o o
N*
O
;Jt
O
o o
I I
O i/J
o *-
O
a
•o
o
*
tv
O
i
K)
o o
O
i\t
o
i
rxl
O
I
ru
O
*
O
m
r*
m
Wrt
oo
OOU1O OO OOOWl O OOOuft OO OOOO O*0 OOU»O O OOOW»O OO
O-*
OO
o
*•
(A
rw
o
"Si
oo-o o«>-*-o
-sl-st yOuJO.0 O-
O
t
oo
CMJK
O-*
O
rvo
ooo
-*o
V)
M
O
OO OUirorv O OOO^t OO OQUifJ OO w»v/iOs« *«« *sifjfs»V«o O»M
r»
o
>
OOVftO OO OOOO O OOOO OO OOOO OO OOOO O OOOO** OO OOOO -*
-------
OOOO O OOOO O OOQQ O
r- O OOOOO Q »»moOQ O >*aaO
1AOOO O OOOOO
r- VA f-OCHAp w> OIAOOO O OO*OO r" QeCOO
*~ fwoGKtQOQ iA
MlMfvlr-T- O OONMKl
NO O WMTr
fv «.'Or—r- f~
O OOOO O
T- T*T-O Of'
_J
(V
o o
00 vC
fs, ^O
r-- O
« i
O O
KV fvl
•"O
o
t
o
o
•
o
(M
irv
o o
* •
o o
ft
*e no
m o
O rv
O O
CO «-
o •*
ry K>
i *
O O
fs. tn
O N-
•* O
h" »-
* *
O O
TK
OK
O
iO
fv
O
(M
-e
o
oc
N-
O
O O
o o
T36
O O
on f)
O
«r
O
eo O
O T-
ftTS -OJ?
o a
«t «*
rs j
O r-
o
f
AJ
TO
r-
o
l^"» ^f
O O
f\i r\i
rv K>
r- t«-
SO •*">
r> N.
o o
-------
ru
*"
o
•"•J
"O
c
f\»
js»
O
~vj
Ot.
«4>
•4..
r
O
-si
ze/
o o
CX/ ">J
r c
o o
•«I(J •ffcij
Ou OO
-Si O'
r c
3*
O
-J
U»
—»
a-
(V
'*»
o
-J
Cft
o
^
to
O
O
IV
*»
o
"Nj
H
:*
i
HI
O
O
a
O
fv
a
•
o
o
*
o
o
•o
o o
• *
o o
<) "O
o a>
o o.
NO -*
'
O
CO
-*W1 OO OOOOC OO -*
o
I
rvwtvfv r^fv fatvruru
wo
WIN!
-si «*
*
o
;\j-*
eo
O
OO
rv
O
o
o
CO
O OOOO
OOOOO OO
O
O OOOO OO OOOO O OU»Out O OOOO O OOOU VrtO OOOOtU OO -*
-------
*- eomoOO OO
r-
o
O OOOO O w*OOO OO OIAOQ' O NQmOQ OO «OOO
«- o
or-ooo
oinoin o
o oooo oo oooin in oo«o»no oo oooo
v-F"-. to *)Ooo- c^N-rgo, in
o
O
r" OO
oo
u.
^.0*- OO r-**T-i- »- T-T-T-T- w v
OOOO O Or-OOO OO Mfs!
r-^r •*
o
• •••
O (T"O«-
O
a;
*-tA O
Q-&
eo
o
PS.SOOCK
«~»-«M»-* r-
»* O»—
1O
o
o
in
T~
•O
O
O
*
o
o i*-
fw !>•
K) r-
* «
O O
O 00
T- O
* *
O O
(A W»
(V O>
fv ».
O »0
Sf
o o
t I
o o
00
o
o
in
o
*
o
o o
sr
ex:
o
«c
fM
o
rs»
a?
N.
O
00
IV
O
=> J
CM fv
O O
30 00
IV IS.
O O
(M
-------
*• OO «AOOO O 0*O
OOOO O*O OOOOO O -*OQOO - OO OOOO CO
mo tftf*»fwn
ofe M
T- O
* t
O O
CO
in
M
o
*
o
o
*
o
rvj
•O
NO
in
O
t
o
• O
jr.
rv
o
W
r>-
o
•c
r?
N.
eo
>o oo
fv iv
O O
*c *c
*M fM
O
<
M
(V
or>
(V
O
«t
(V
oo
N-
O
«r
fM
O
K*
00
O
fM
o
N N-
O O
«c <
O
•c
fM
IV
O
N.
O
«c
-------
tv
>
o
•si
Co
fv
3*
O
«sj
CK
•O
O
-M
03
|Sj
c
O
"M
Co
Ge
rv
>
o
o,
O9
> 3*
O O
-J «vi
OO O»
(JO W
«* Wl
r c
hj rv
> 3»
o o
**J "Sj
O9 Ot
N)
>
o
-J
oo
IV
>
o
•>j
00
o
-^
oa
O
o
o
*>
U1
o
t
a
•sj
o o
» I
o o
SI <>
*• *><
o o
I *
O K)
•O f\J
Ui fvi
N fv
IV
O
IV
(V
ru
tv
o
•^ c
jo Ch O400OO O^
aooo
2
O
o «*«*
OOOOO OO -*-*-*-*
ruo
oo
fvoo
oo
VI
M
o
oo
o c^^ro-x) -*o ixo^o -o woooo o
O OOOO OO OOOO O
ao- oui*-as -*
OU1
004
o
o
O OCfcOOJ •£>
OOOO OOOO O OOOO OO OOVAO O OOOO O OOOO-* OO OOOO «*
-------
rv
>
o
(X.
GO
a
ru rsj
o o
us oo
O O
** CM
r c
o o
CO 00
U3 0»
o o
M hJ
f" C
O o
00 CO
GO 00
O O
INJ -*
r c
N*
JB»
O
go
ru i\j
> >
o o
•si -M
CW OS
C <»
Ch wi
r- c
o
•si
00
•o
(V
>
o
x
cu
to w
> -H
O :»
•^ at
CO |
I
o
_*
O
-*
o
-* o
* I
O ijJ
-k >o
O •
•Si
O
O O
* «
*> Ul
o
t
ui
-*
-*J
>*
o>
I I
rs>
•
ui
O
CO -i
• •
O OOOi-J O W1WOO O OOOOO -k
O OOOO QOU4U1
OK >
Ul
I
o
-si -*!*» OOOOO
o
1-4
«*
O«
00
I
00
-^-* OOOO "* -*—»-*«* OO-*O OOOOO
O
o
oo
wi
O
o
O> OO-s)OOO
O
I
O UltsJOUl O U>O«OO O UtUIOoO "^ OOOOO OOOW1 UI
to
IS)
v^UtOO OoOf\tOO
I till I till I IIII »
-si ^co-slut ON OJMOOO. o a»*siae*sj >v
O UtOlOO O OOOWl O OOOO O OQQOUi OOOO O OWtOO OOOO OOOOU1
-------
*- COQQQ p "*QQCJ O NQQlA O T-aOOO O OOQO O K>OOOlA -O »A»AiAQ O
• t t • » • i • * i i i . • • « i t i « t t a a i » * « *
O ^Of^OCP1 fM fAsfr*»0iANC3 -^ OT"*AJCO v* st'^r^^OT*' %-t' 0osoo>3O O
O f\lf\IA4fV *** f\tfMfvJfVil IV *«J*st'IA1*O AJ ^p—^if-n*-^ ^t*» PMfVAJfVI A4 fsJAJ AJfVf-O C\l V™ v-
»A OOOO O IAlA*~r» O IAQQ o »AOOf«~ ', O P--SAGOO O OO«At^ IA
«V »llt t I « « • I i t i • « >«(t« v ««•* * *•**• t *»** o
T~ l""i^">4-*4" *~ »AIA|A>O *<> X5sO>OO Kl OOO(\IOA,< ^" t~OT"'T" O "^!AlA1C>6f IA
—I i»«* * i • t « « (lit « « i * t i i t(f« i ifft* « *f»* •
OOT»»-*» T- «-f-r-T" T" <-T-T"*-T- 1" OOOO O
i e • • « t » i « « i • • « t i •
•*-O O
AJ«VC *f NtKi^OsI1 Kt flO-^^fMi* O «O«OOSO fw
<«t- ««il « «**« • (**• « * I • « • ( I(|l t Ittll « «««! «
~f !A?VfvJ9D fSl JAfAQT* r- f^-O-a^Ki KV O-eOON-l"" •« nf.if*-«7«7 sT rv)O<>iE>O1€! C ^-^D-J-iA r»
r» OlAOO *O OOOO 'O (A-OOO O O^ir-OO O OOOtA O »AOOOtA O T~»Of«3O ^(30 ^i' T"* T"* T1'' ?A ^"* f^T **^fA^t*^ i(O
_J r-T_t—^- ^^T—^™.^. r» ^.^.jnj,^, ^_
•C
a.
*•
•9T
31 «-
f\!
O
O
*" •«
f\) y;>
O K
O
o- o
«r> •«•-
o- "o
o
O O
n
CD
O
o
eo
O
«T
rv
rc
O
«f
-------
«» GHAOO OO "OOOOQ O «*OOO QO JAQinOO OO QOQQ OO **OOQO O
O
SE
Q ^.^
K50h-r>-r>~ N ooor- oo
j it«* ii »iiii i iiii ii
t- r-Q*-O OO OOOOO O r»T-»-f- O*- i-T-r"*-
»~ r-r- OOOr- OO OOr-r-O O
ON-
Or-**«OO-
«««
Q
OO«M<\*
sOP-IASO OO fMr-
«O fs«O»— «* OITl
K> OOOO- «JO
O"
Q
o
o
0"
o
«r
rvi
o
o-
oo
o
a.
o
•t
tM
o
o
00
o
«*
o-
no
o
O
o
00
o
•r
AJ
-------
O
(K
O
CO Ql»0*fV «5O
I III! • t * t I I
T-O oo ooor-rg *- f»h-sf>c fMM «ot- O*» r-OOO
O
»- OOOO OO OOOOO O O«COO OO WiQOtn OO inOOO O lf»OOO OO OOOO
««
«
O
-J
T- o»«oo .00 OIAOOO o aooo oo oooo oo oooo o oooo oo oooo
ir* 1*11 it iiiit i tiii •* iiii it iiii i iiii it itii
X
gj
r- fvtAOO OQ (A4AOOO O OOOO OO OOOO OO (MOOO O IAOOO OO OOOO
CM
*-^ Ift
-4
a.
u>
PN. fs-fs, IVIN (\)t>> O><> Or* *-N
«V fVX> OO O*M MM fMIA (T>«V
S3 ^>O- O^r- T-h» fv^S- STO OO
O O*O "OtA ITiKl K>O •
rM fMi>« K>sTt mrvj fviro MO Of>J
«- T- T-M T~ r»«V fMM K>>«
MO OO O'O OO OO OO OO
If- fs, N. N.f«- N,«J f»«o eofo ooeo
jcr>> NK K i»- •^w^. n.^, N-N rs-r.
CK O OO OQ OO OO OO OO
'
-------
9PI
no iv>
O U
•M -^
•J» i*i
fv> (V
> 3»
O O
**^ •HO
Ot> i>J
r c
o o
•jo *.X
-* —k
— O
r c
o o
* <*
-* o
O to
ro »\>
o o
-»
o
(xt
rv
tn
JC
-o
m
OM..>—i it.i*
00
Ui
OOOOO O OOOO O
O OOOO O OOOOC
OOOOrj O OOOOuo O
fu oaO-»O >
OW1OOO O OOU1O O «/IVAOO O OOOO O OOOO
U>4 r\J-*l^J—k CJ5
OU^OOO O OOOO O
V/l
r-
o
rvi
o oooo o
O OOOOO O OOOOui O -*
o
•-*O O OOO-* O OOOO O OOOO O OOOO O OO-»-*O •* OOOOO O
II I III! I Illl I Illl I III I I Illll t Illll I
o
o
O
OOOU1
O
OOOOO O f*
O OOOO O OOOO O OOOO O -*U4OQ N> <>.w4r\J-*O -*
CO
o
I O
-------
z com ofMO'*r». sro WOMM/I
O • • ttiit ii tti*
M K»O
U) (V
OC
O
O-OQKH- T» »»Ofno9 CM*-
OO O
« *»** • *•*! I* 11*1 I
40
*- oo ooinoo oin oooo o oooo
X
MJ
-I
o oooo o oooo oo amoo o
...... «*** *• I*** *
O
O
!>
o
I
o
O
*
O O
• I
o o
(M OO
O «-
'O NO
s^f *4£J
CO O>
o
I
o
*
o o
no O-
T- a
-o o
•C O
c^ ^^
O O"
• *
O O
o
o
I
o
O
*
O
rw
O
*
a
ar:
OK
tfl
-J
•>*•
t-
or<
o
•s*
*~
OR
146
O
in
T"
00
K
o
=3 J
in vO
r- T"
90 TO
fs. N-
O O
of> OP
N. IS.
O O
«e •«
(M rsj
o o
«e «c
rx* rvj
OO
N.
o
o
O
SJ
o-
IV
O
-------
in
in-* fv^ooofn o K»^«*O oo »-«o^oo Do rMruisioo »o K>v>voo»- -o-o O«MOO
•• •*•« I **«* • • • t • • • ••*• * ••••• •• I • I »
OO *»«*O»O" IA v«~
O
O
»*•<». ectnf^^o N «n«n«n«o in ^'nfu'Vv win ooooo <*
W OOOOO OO OOOO O OOOO O OOOO OO O«-f-*- T- OOOOfW O»- f-
o
o
T- otnooo oo ooom o oooo o K>rgr- i- T-OT-OJ rvj m>O>»(M «*fvi O>soooo
O
_J
< v ooooo oo omom o oooo o oooo oo inmoo o ooooo oo oooo
V) T- Itill • • tftf I ilil i »!*• • I lift * i • » • t t« • I • •
x >»«vmM«» -*K rv«©srKi K»
t- OOr-OOO OO «*OOO O OOOO O inOOO OO OOOO O NOOOO OO O^OOO
wo
(Mt- T-PO-frN. O
0.
9C
<
tfl
m IAN. N-O^ O>(M fMN. fv^ OO
r- f-K"t Kir- f~'V fMO Ot*^ f*iirs
00 OO O O>-U*> U"IK) Kf OO OO vj- >* st
O Of^ K>^. r- *• T«O OO O«M
i II t* II I* •• *«
O OO OO OO OO OO OO
-J =3 _J => _J 3 _J =»^ =) ^ =>~l
OO O r» »-fV fNf'* K»»» >*IA IA»O
MfM CJ 9300 <»*j aoao
UN r^-h- NN f-r- r^fs. rwr*. f*. is,
«o oo oo oo oo oo oo
K< ««r < « fM rvirvl f\J»M (M«M |M«V (VfV (M«v
147
-------
STRM.ID
2A07826U
2A07827L
^2A07827U
S
092A07828L
2A07828U
2A07829L
2A07829U
2A07830L
2A07830U
2A07831L
2A07831U
2A07832L
2A07832U
WST
0.245098
0.252525
0.252525
0.135870
0.135870
0.118203
0.118203
0.184343
0.184843
0.035842
0.035842
0.056625
0.056625
SAMPLE
I
1
2
3
4
4
0
1
2
4
3
4
1
2
3
4
3
4
0
1
2
3
4
4
0
1
2E
3
4
3
4
1
2
4
3
4
ALTL11
74000*0
70.5
104.0
73.0
101.0
47.0
43.4
78.1
55.0
102.0
1820.0
84.0
209.5
49.0
92.0
98.0
185.0
76.0
144.0
236.0
162.0
201.0
166.5
145.0
300.0
73.7
124.0
372.0
143.0
1890.0
121.0
2000.0
214.0
50.0
163.0
54.0
158.0
240.0
SAS
ALEX11
4.0
5.0
7.0
6.0
3.0
4.0
2.0
5.0
6.0
8.0
6.0
7.0
7.0
13.5
1.0
6.0
4.0
4.0
5.0
7.0
4.0
3.5
6.0
2.0
0.0
3.0
2.0
2.0
3.0
2.0
1.0
4.0
0.5
1:8
?:8
ALOR11
3.0
4.0
2.0
3.0
1.0
1.0
2.0
4.0
6.0
6.0
5.0
5.0
5.0
14.0
1.0
6.0
4.0
4.0
5.0
6.0
0.0
4.0
8.0
2.0
6.0
3.0
0.0
2.0
1.0
1.0
1.0
1.0
4.0
2.0
0.3
2,0
4.0
I-R
4.0
OOC11
1.4
7.7
0.5
1.3
0.5
0.9
0.6
0.6
0.6
0.6
0.7
2.3
1.2
1.7
0.8
0.9
0.6
0.8
0.6
0.8
0.9
1.1
3.3
1.7
1.1
1.8
0.5
1.0
1.2
1.3
1.0
1.0
1.4
0.6
0.5
0.6
1.2
oil
COLVAL
r!8
15
i§
20
15
10
10
1 0
1 5
35
15
18
15
1 5
10
10
15
10
15
20
30
20
25
35
15
20
30
15
25
18
18
10
25
IE
TURVAL
7.4
1800.0
2.4
ki
2.3
0.6
0.3
0.4
1.6
1.6
42.0
0.9
3.0
0.4
1.1
1.6
3.2
2.1
3.0
4.7
5.2
7.9
8.4
2.6
8.8
2.0
2.7
42.0
5.4
38.0
5?:J
|:8
49*0
ki
ORGICHi
5.8
46.5
1.<
3^2
3.C
3.1
2.J
3. 1
3.'
10.3
6.5
9.9
3.6
4.5
3.2
3.7
2.7
3.5
3.2
4.4
18.1
5*2
8.7
1.7
3.!
5.^
4.3
4.6
6.1
|-J
3*1
5.7
3.4
2.5
-------
3! Ot>»
-j
o
U OOOO OO O«-«-»-*» O O»-OO O «-OO ^O in m
^» ^* <\i M ^» ^» OK c^» >^ *4*
f*- N-fV fv)K> f»-* «*O O
.J =>,J =5 _J => «l 9UI J
|00 OO «O OOOO flQOO 00 00 00 W 00
«o oo oo oo oo oo o
>-«* < « «r «c <* «c< « <
-------
091
r\j ix» rv iv r\» f\> ru i\> iu rv rv rv is* M
» >> >> ;» >> *• > *» > > a» H
o oo oo oo o oo c o o »
3> CttCO CA O* OOOa C» *>( ~st -J ~ *0 •& -O Ot-4
4s •*•• w t« rv to -» -j> 0.^1 v^i «s MO
c re re re r re r r r
-* -» o oo o o o
S • • •< •• I II
O O t* ut *- *- ui wi
_, -i -*5 >O U1 U1 >O -O
o OWM uj«>»O»-^-^
OOOO(N>N>-*-»
_. _i -i -»«*«*(> O
r
m
S/l
00 -^QWS Os Ov/l-*U4 O WlM*"*- Ni OOO.O^O«* >O***«U1 IS*
• •*•• * i • « • • «*•* • i > • • • • • i • t it*t •••• > • i • • >*
o o^oo o ooom o oooo o oooo*» oooo o oooo ouioo ooovi-j -»
m
x
* tail • aiii • • • • • * *»tti liti • »*t» till • • • i > ^
O QW1OO O OOOO O OOOO O OOOOO OOOO O vnOOO OOOO OOOOO -*
r
o
tvt w»v/»**w-* *»-«4*»o rv O«*«*(\» '^i-*ruo i\*ivwi*«-* ?e
• • « • * i l«ll • •*!• • • I • • » • I • • • «••• • • • I tltlt -*
o oooo o oooo o ooooo o ooooo oooo o oooo ovnoo ooooo -*
o
o
O OOOO O OOOO O OOOO -* ~»O*»-»Q OOOO O OOOO OOOO OOOOO O
• *tl* • ill! t lit* * • • • • . Ill* I Illl III) Illll -k
oo
o
r
iv -*
O OOOWt O OUIO^ Wt muiWiOO OwiOO v^i mwiOOO
e
hj "»
tv -»PUOO *• O(>^i\j of> furjui-* -* oworu-* wkooo -* -»-•-*o rv-»oo o-*ooo <
« 11*1 I 1*1* I 11*1 I III** *•** • • I I • •*•• fill* >
oa
rv)
-------
Z
O
O QO»ut
«o
•
o
OS
f»T-
o
*- «*
O OOOO O
O
O OOOin O
U OOOO O «"O*M- *- OOOi- »- OOfMO*- «- OOO*- O OOOOf- O OOOO O
o
o
r- OIAOO O OOOO O WtOOlft O OOOlAO O OsTOO O inOOOlA O OOOO O
w- *••* I **•• t » • • I * • • I • • I tifi f titit t * • I t f
^ rvr-*- *» «*«s«¥" K> (vrvj«i-M •«• r- «- *• «rt- *-
J
<
w
< r- Otf^OO O OOOO O OOOIA O OOOOO O O-OOO O OOOOtn O OtAOO O
*• xCSjAOO O r-OOO O IvoOO O TOOOOO O OOOO O O^OOOm O O«nOO O
O Oirvh-Ki Ki fslAWflO ST «*O»*«M«M » r"«*9Q» O
tu
«J
0.
SC
UJ
O
*»
O «"•
«« *•
>O T»
w. O
ivO
«T>«M
OO
r-U-s
K»f««
IAN-
e i
OO
«
O
O ift
MO
(KO
OO
«o<»
soeo
OO
vOOO
OO
f^oo
sooc
OO
OO
«?co
aoeo
OO
OO
Or"
«o«o
eooo
00
Or-
OO
coeo
00
eo
O
-------
Z91
IV (vru oj rvj N>IVI hj »\J M rs> r^ */»
» » *• » > J» :» > > >> > 1
O OO 00 OO OO OO O 'X>
CO <» 09 JO OO i» CO 00 09 CO 00 «> ae
>O O SO V/I \SI &• *" I* O. ->4 -s( \A •* «*o orv» ev-*" *• -» -*
O O •«>• *» O4 U4 >O ««<• *^ ~» -»
rn m
a
•
O OOOOO OO OOOO OO OOOO^ OO OOOOO O OOOOISJ OO OOJiO -k
rs»
I Illll I* Illl II Illll II Illl I Illll II lilt _» to
O OOOOO OO OOOO OO OOOOO OO OUIOO O OOOOWl OO OOOO -* >
HA
rv>
I Illll • I Illl It Illll II Illl I
O OOOOO OO OOOO OO OOOOO OO OWlOO O OOOOO OO OOOO **
oooo oo -»at*ao oo oooo -* ^-*-*
Illl II I I I I I || Illl I III** l|
o INJ-^-^-WW out
o ooooo uno ouivjiin aiut ooouiv^ 010 o»uto w/i aowiuiu out oov^ivn r
o ut
01 tvrjo-* is» ui
w» vj»rv>o-»w» -*o -*o-*o
I Illll II lilt
o oooo>*» *-*j> -•-jo'Ji oo ownoos'O WICK aoij40KO o MOOUIO^-O wiw ot\»*^w r
o
aalt (ij p^t ,,t
O
09
-------
3AS
SAMPLE
PTL11
MH416
TUSVAL
CQNIS
TWPSTt?
pi
CO
2A07701L
2A07701U
2A07702L
2A07702U
2A077G3L'
2A07703U
2A07801L
2A07301U
2A07802L
2A07802U
2A07833L
2A07803U
2A07805L
0.120627
0.120627
0.236967
0.236967
0.255102
0.255102
0.42372"
0.423729
0.136426
0.136426
0.505051
0.505051
0.729927
1
2
3
4
3
4
0
1
2€
3
4
4
1
2
3
4
3
4
1
2
3
4
3
4
1
2
3
4
4
1
2
3
4
3
4
1
2
3
4
0
4
14
11
13
10
8
12
26
20
21
0.0
0.4
0,6
0.2
1.9
0.1
0.4
1.2
0.3
0.8
0.9
S.2
.8
1.3
0.7
o.a
0.1
0.5
0.8
10.0
1.6
2.2
1 2
1 0
15
18
10
7
90
92
85
119
135
7.6
8.0
13.8
13.0
12.0
12.0
13.0
17.2
16.0
19.9
20.0
4
63
0,3
0.8
1.8
7.S
50
32
21.0
14
16
17
21
15
15
105
26
46
26
44
12
18
25
31
0.3
1.2
0.8
0.8
0.8
0.7
0.6
0.7
ill
0.9
1.6
0.9
0.6
o.a
0.7
3.0
7.9
4.9
6.2
2.0
3,4
6.9
1.4
2.4
14.5
3.3
10.5
1.4
3.1
2.0
7.4
12
15
20
23
15
16
34
35
39
87
40
75
20
24
28
32
11.5
11.0
17.0
18.5
15.0
17.0
9.1
10.3
18.0
22.0
19.8
20.5
8.3
7.0
13.0
22.0
13.0
9
15
21
73
37
50
22
1470
18
31
0.9
0.7
1.2
1.6
1.0
0.5
1.2
0.8
4.2
1.2
2.0
2.1
3,9
14.2
1.8
3.0
0.5
0.4
0.9
1.5
132
111
165
211
164
204
11
11
15
22
14.5
12.0
18.4
27.5
16.0
21.5
10.0
9.0
15.0
18.0
-------
O OCHriO-O O QWOOQO O OO r-OOtr-
M »— T-*»T-T»«- rs) O Or-MeafVI r- Qt-fMK*>O «M OOKl«- O OOOO O OOOO O OOQO O
or
fV)
r- | IHfll « «t*«* « » » » » I I • I I • III* I !••! I *!**•
•* r- r-OQOO O OOvOO O OOOO O T-OOO O OOOr- O OOr-O O OOO*-O
X
3R
OOr-O h- -^f-Or-iA »A <«OOON- ur*
a,
U.'
—'
a
5T
«*.
K- »- r-t^ K»«\( fMKl K>00 (TN- N.IA
fy N. f^w- »- *" r-^O ^>O ^O?> C^N»
csin snr^i f>j*r» Vih- N.(V pyrM
oeo « >o OOG «c>o >ooo »r- »~»*
Mh- |V->O •OK* K»K1 rOO O«» -^fO
N- r" v- O OO OO Q
-------
STRM_ia
2A07813U
2AQ7814L
2AQ7314U
2A07815L
2A07315U
2AQ7816L
2A07816U
2A07817L
2A07817U
2A07318L
2A07518U
2A07819L
WST
0.104275
0.072516
0.072516
0.059952
0.059952
0.034602
0.084602
0.096618
0.096618
0.990099
0.990099
0.027739
SAMPLE
3
4
0
1
3
4
3
4
1
2
4
4
1
2
4
4
1
2
3
4
4
2
4
3
4
1
2
3£
4
SAS
PTL11
69
361
11
n
27
173
\\
5
5
II
13
6
10
15
8
3
6
14
6
2
ll-
19
12
24
19
19
65
47
UH416
2.5
1.8
0.5
1.2
1.0
1.9
1.9
0.7
0.6
0.4
0.6
0.8
0.8
2.1
0.7
0.5
0.7
0.6
0.9
0.4
0.4
0.8
0.5
0.7
0.6
0.4
0.8
1.6
0.8
0,9
0.4
0.4
1.1
1.0
TUHVAL
24.0
38.0
1.9
fc!
4.1
76.0
3.9
7.6
0.3
0.7
3.5
1.3
1.3
0.2
I:!
0.5
0.1
0.3
8.6
.2
0.1
8:1
1.8
3.7
1.9
3.2
2.8
3.1
31.0
7.9
CONIS
29
25
35
35
50
44
52
H
13
15
17
16
15
9
10
'?
2
5
8
9
7
10
6
8
n
12
12
21
25
25
24
TMPSTR
15.3
15.8
9.0
10.0
10.4
16.3
19.0
«:I
12.0
11.2
15.9
20.0
19.0
6.0
7.0
till
14.0
6.0
7.1
12.0
12.0
5.0
6.8
11.5
18.0
12.0
17.0
10.0
10.2
16.9
23.0
2A07819U
0.027739
63
0.5
5.8
26
18.0
-------
Ol
o>
STRH.IO
2A07820L
2A07820U
2A07S21L
2AQ7S21U
2A07322L
2A07822U
2A07823L
2A07323U
2A07824L
2A07824U
2A07825L
2A07S25U
2A07S26L
MST
0.081566
0.081566
0,393701
0.393701
0.151976
0.151976
0.132275
0.132275
0.080775
0.080775
0.043995
0.043995
0.245098
Si
0
2
3
4
3
4
1
2
3
4
4
1
2
3
4
4
1
2
3
4
3
4
1
2
3
4
4
0
1
2
3
4
3
4
1
2
3
4
SAMPLE
SAS
PTL11
28
6
1 2
17
6
7
13
56
25
28
16
9
13
37
21
33
7
11
37
19
42
20
6
6
3
6
NH416
0.5
1.0
0.6
0.3
0.2
0.4
0.4
0.0
0.5
0^7
0.9
1.1
0.3
ila
1.6
0.4
0.6
1.2
0.7
1.2
0.5
0.0
0.4
0.7
0.5
TU«VAL
0.5
0.5
0.8
1.8
1.3
0.5
0.4
1.7
4.6
9.8
9.3
5.9
1.3
1.9
1.4
2.0
1.8
1.1
1.8
1.6
6.0
2.0
25.0
0,2
P,-?
0.2
0.8
CONIS
11
17
11
14
9
9
7
12
12
10
16
22
15
24
27
29
16
16
22
24
14
39
8
10
10
13
THPSTR
10.8
14.3
11.1
13.9
18.0
13.8
17.0
7.0
10.5
16.0
16.0
15.8
12.0
9.9
15.0
18.5
18.5
10.5
7.4
18.5
22.0
14.0
19.0
5.0
3.9
11.4
18.0
0.4
0.3
14.0
10
4
10
9
18
9
20
16
17
26
84
1.1
0.6
0.6
0.6
0.1
0.5
0.4
0,4
0.6
2.5
2.6
0.3
0.5
0.6
7,0
1.1
0.6
5.6
2,0
2,2
6.0
5.0
8
13
9
10
3
10
7
38
36
50
52
8.0
14.0
11.4
15.8
17.0
13.0
15.0
9.0
9.Q
14,8
17.0
-------
2A07S26U
2A07827L
01
MST
0.245098
0.252525
SA«PLE
4
2A07827U
2AQ7823L
2A07828U
2A07829L
2A07829U
2A07830L
2A07S30U
2A07831L
2A07331U
2A07332L
2A07S32U
0.252525
0.135370
0.135870
0.118203
0.118203
0.184843
0.184843
0.035342
0.035342'
0.056625
0.056625
4
0
1
2
3
4
3
4
1
2
3
4
3
4
0
I
3
4
4
0
1
2€
3
4
3
1
2
3
4
3
4
SAS
PTL11
§1
11
it
10
90
NH416
1
9
.9
.0
?:?
0.4
0.3
0.1
20
TURVAL
180
hi
3.5
2.3
0.6
3.8
COMIS
49
46
25
20
20
29
12
12
THPSTR
12.8
19.0
16.0
8
11
12
108
10
20
194
166
53
59
12
12
29
1 20
43
11
1.1
0.5
0.1
1.8
1.8
0.4
0.5
1.9
2.4
0.2-
0.0
0.8
1.2
1.6
6.6
2.1
1.1
0.3
0.4
1.6
1.6
42.0
0.9
3.0
0.4
1.6
3.2
2.1
3.0
4.7
5.2
7.9
8.4
2.6
7
8
3
9
16
6
8
9
9
10
12
9
9
23
25
34
22
39
7.3
10.5
11.5
13.5
14.0
13.0
15.2
8.8
8.0
14,5
16.0
12.2
14.0
12.0
12.5
13.0
15.3
20.0
16.0
26
51
48
189
61
35
155
5
1 0
21
68
18
13
0.5
0.9
1.3
1.2
1.0
0.5
1.6
0.3
1.1
1.7
0.9
1.2
0.4
2.0
2.7
42.0
5.4
38.0
6.1
57.0
1.0
2.0
1.7
49.0
1.2
2.8
42
60
42
112
42
II
11
10
13
19
12
13
8.0
11.0
13.5
14.0
18.0
1I-J
11.0
8.2
14.5
19.0
15.0
18.0
-------
tn
co
2A07333L
2A07533U
2AQ7834L
2A07334U
2A07335L
2A07332U
2A07891L
2A37392L
0,167785
0,167785
0.222222
0.222222
0.003721
0.450450
SAMPLE
1
3
4
3
4
0
1
3
4
1
SAS
PTL11
2A07S35U
2A07881L
2A07S91U
2A07332L
0.093721
0. 074963
0.074963
0.450450
4
1
2
3
4
3
4
1
2
3
4
1
3
4
1
2
3
4
10
10
HH416
0.2
0.6
TURVAL
COMIS
3.2
3.0
12
14
TMPSTR
6
10
12
13
14
13
4
6
15
13
10
1.0
0.0
0.9
0.8
0.4
0.3
0.3
0,7
1.0
0.4
0.2
1.0
2.4
1.7
6.6
• 0.4
1,4
0,7
1.1
2.9
1.5
1.8
17
1?
23
30
22
30
6
6
a
8
11
9.0
7.5
17.0
17.8
12.0
10.5
7.0
8.0
11.5
13.0
17.0
13.0
4
11
12
0.6
0.7
0.1
0.9
0.4
0.6
1.4
2.8
15
14
12
11.1
12.4
18.2
20.0
19.0
18
27
39
62
53
57
4
3
11
17
16
5
11
12
11
8
11
12
21
0.4
0.9
1.3
0.9
1.5
1.6
0.6
0.8
Q.5
0.3
0.2
0.7
3.1
0.9
0.1
0.6
0.6
0.9
0.4
3.5
9.6
8.0
14.0
4.3
3,9
0.3
1.4
0.4
0.6
1.6
0.1
0.3
0.1
0.1
0.5
Q.4
1.3
2.7
72
70
92
149
62
54
10
10
13
19
10
10
11
11
17
9
9
9
12
14.0
11.3
19.0
22.0
13.1
20.0
5,8
35. 5
12.2
16.0
13.3
9.5
8.7
12.5
16.0
8.0
8.0
15.4
18.0
-------
691
Oft
O
4S
c
r\J fv
*» >
0 U
Ote Ct>
CO 00
0 0
4«* v^i
f" Ci
>
o
w
o
t>J
r
2AQ38G
r*w
er
2A08SO
PO
r
o
co
o-
w*
c
ro
o
ec
o
amfc
r
2A07S?
o*»
r~
a*
o
•%!
oo
t/i
c
2A07B9
" wA
r
Nt
••0
^*»
r-
-«a as
QO t
O M
4J(f^ Jp^v^l v/t
«A 4 (>* O> ""^ *•<*
_L »* -x| «S| '>. O^ ""J *SJ
O OO OfV* N» •* •*
31
•v
r
rn
1J
•H
r
(SJ
O K»O(VJ>O O **OOW O
a:
x
O OO~*«* -* **OOO O OOO-* O OOOOO OOOO O OOOO OO-*O rv)O-*-*-* •*>
I • I I • t I • t I I lit* • itltt •»•* ( I • • I III! • t I • I -*
-*U4OO»
c
ro
I Hit* « **•§ t «»l* • • • 1 » •
fo •UlvJ'^O' O* U1-NJO** 4* •OUtMOK •*" O-*-*OO
o
o
•* «* Ul
o
O OOOV/l >J» OWOOO O OCKHU1O O-*OO C» O'OOO OO*-OO
-------
09 /
O OO O G O O O O OO O O O
CO OO "M CW OC OJ ;. Ju oo <.* "jj oo
>O x» -» -»-i -.O OO OO OO O
-* ~» —k -* O O •*> I •
*Si "^ V< >JH •** •** "-* *^ Ch O^ O O
\j\ Vi
x/> iji*» *>O> O"^ -sjwl vl —t
IV —* -J> —i—4 I\J-*»A _a ixj—*-^J i\>
rrs *.
r"
•* oooo o ooaoo o ro-*oo o -*oooo o oaoo o -*tMOw o oooo
I 1*11 * I**** | **** I ***** * ***< I *»*• I • > I 9 >
NO **OOQ»> w wi-oouJo ots >owi~>if\> ot> owivo o OGWV/IN* r*
o
o
2!
«* -oo-^ao w *»rvO-^i*i » «ow-» o wi-*~*ivu< o 040-*^ o*. -ouoo
-------
2AQ89Q1L
2AQ8901U
2A03902L
2A03902U
2A03903L
2A03903U
2A03904L
2A03904U
2AQS9Q5L
2AQ89Q5U
2A089Q6L
2A08906U
0.133511
0.133511
0.124344
0.124344
Qi44Q529
0.440529
0.199203
0,199203
0.507614
0.507614
0.196850
0.196S50
SANPLF
1
3
4
3
4
0
2€
3
4
1
3
4
3
4
0
2€
3
4
3
4
1
2
4
3
4
0
2£
3
4
SAS
PTi.11
NH416
TU8VAL
CONIS
TffPSTR
19
49
23
25
16
10
3
5
18
12
9
1.7
1.6
0.7
1.1
0.0
0.2
0.2
0.9
0,4
0.3
0.5
2.3
12.4
6.2
10.0
2,3
4.5
0.9
1.6
9.5
2.8
5.2
11
12
15
24
9
8
9
10
15
13.0
11,9
20.5
21.5
14.0
15.0
5.5
9.2
11.5
15.0
20.0
14
0.5
7.0
18,0
3 ,
16
17
18
16
20
3
9
11
14
27
12
16
4
9
10
8
10
16
23
4
44
40
0.9
1.2
1.0
1.2
3.2
0,1
0.2
9.5
0.9
1.4
0.5
0.1
1.0
0.9
0.8
0.2
0.4
0.2
1.1
2.2
16
3.3
0.9
1.0
1.6
4.3
3.8
4.6
4.5
5.9
3.8
1550.0
6.5
20.0
4.0
4.0
0.5
1.0
0.7
1.1
0.5
1.4
5,4
11,6
1000.0
22.0
25.0
15 9,5
12 11.2
17 16.0
20 1
20 1
8.0
6.0
20 17,0
18 1
18
22
22
22 1
§,"!
.0
4.1
7.9
9.0
15 15,9
19 17.0
10 1
I2.0
a.o
10 16.5
15 23.0
3 1
6.0
14 23.5
11
8.0
11 15.7
17 14.0
12 1
7 1
I5.2
I7.5
4.9
55.0
12
17,0
-------
Z91
w
o
O
-si
oo
o
OO
-g «sj
uaoo
oo
UJ (V
re
O
-si
cw
o
oe
o
O
-j
oo
o
o
0-<
c
OO
-j -«a
->i "«J
oo
UifNJ
re
o
ISj
r
tv
>
O
-j
--J
o
-i
e
O SO
-j at
o
O
o
•
m
O
oo oo
*• It
m -* «k«»
o w tji t\t
oo
••
oo
II
o
•
o
•
o
I
-*
is*
(v»o
oo
OOEWUBQ9
x
I
r
o
OO
OOCO 09--J'Nt» O^Ch O«OsOsO«. O- SOOOOOOOOO
I
o
10
o
woo
TJ
OO
OOOO O
o
M
oo •»-»-*-» o
oom ooo-»
fvt
COW1
00
>
r
UIO
O-* ChOMO -* OOt»-»O O*- OKO*-O N»->l -J»0
-------
£91
O O OO OO OO OO OO O O »
-»4 -sj -sJ-M «s* -»J ~J ~W -J -xl -J -N -SJ "Nl ;*
oo us 00 co OM oe co yj at go o> jo 00 c* i
-* -» _* *j> *i _i —k Q OO OO OOM
V^JNI INJ -± -* o o as eo-sj -go o> wt o
re re re re: re re re
OO OO OO OO OO OO O
• » • • • • > • • • ft *
-* UJ ut fv» N»O OO OO O-» -»
O*« *»Q O^ W^ O<5> »- -J -^
^ -J> -i CC CO O O 00 O»O« O f» CO
f\*(\> r^*j -ju* u»i\j MW< ojv« wt
••jo >oo>. 0.0- «K-» «*-» -*-«j -^
-k 4S
i i*tt • • • « • i > • • • i • • * • t i i • • • i «*«*« • r>
C» -*O-O'» (V U14SU4JS) O L»)sO->4i> -* O4-*00-k -O r>JL^-A-*l>J O. -OO <><>-• O r
w M-A^-O *• ooooo «• w-siO-» <^ wi>i'«ju« 1^4 wi(\jwi-»ru •<> woe*4«w<» wow
O-
I O
oo t)
t»
Ob O-»OO IV *»^»o*«>o-* *»
o
M
•* -*•*•*-* O OOOO O «*OOO M M-»N>-* «* fv>r\j-*rv>-* •* «*»jk.A.4.» -A rt
wn v/i*»*«i>i -o OJ-oooo u*t u«wio^> o» wirsjooce o uiO4N>r>jts) »> >
ut ->JN>t^>o ut o>utr^<^ OB ww*»ixto •* oovfluio o r
OB uiooo rs> w
ui wwoo w4 -J-4-^o4 *• t»Jooooo
ooooo *> oauj«4 i> ooui*. oui o oooo o ooooo o aN**oa o
-------
t- OQ ooooo oo «OT»
*
*
f\*f\J
•*
80
o
t
M
oooo O
iiii a
~ O-v-
_J «OflO fNJ*-
<
v O OOOO O
r-T- i«*- fSIKlfvJM >*
T* OO
Vt
a.
ft,
O
|
Q.
o
i
«O
OfV
r-o
ae
4,
rt.fs.rs.fs.
NN.
I
fv
WJ
-J
a.
UJ
i
O
fM
fs.
O
I
O
(A
(M
N.
O
I
o
O
*A
O
I
o
rvi
O
«* ^
«! O*
o o
I I
o o
OC- O-
v O
«o o
XJ O
o* o*
o o-
I I
o o
O^1 0**
CD1 ^**
o N.
a- ra
o-. o
i i
o o
to
r-
O
I
O
O K>
fc-l T-
I 00
5E fs.
ac O
>~ <
w «ss
N.
o
It
CO
r>-
o
•r
PVJ
in
T*
CO
»s.
o
o o
-------
*• Qv>ot«oor- N «-O»Of O>«o mcMOrs »n o«ov»00 00 ^infO**" K»00 fs.fvfw<* fs, COOOOOQOQ flOts. "OOf*"*
O O«0«»«> f-«N ooooO'M *- crfipji* CM «cor»K> O»<* ^in^o^ in
U till! II lilt t 1*1* I lit! f I till t.
I
CL
UJ
J
a
JE
«e
*«t r-
>o o *• «— 1.0 sJin min inin in oo
>O '•O O ON- Is. f*. |s. fs. |S-c^ Os>
tn m N- fs. ^ c^ M ^ji fw fs. ^ o* o
T» *-K» fn»- r» CM njo OM Mtn
C?5 OO i^ OK tj^ l/^ l1^ |*^ 60 O^1-1 vf *4?' "*^h
O OK> Kit- *-•— T-O OO O*M
• it II il it II li
o oo oo oo oo oo oo
oo Or* T"W OOOft WOO W!»3
XI N» fSt fW fS» f**i fS. N» f^w ^« P**. ^» ^** fs.
Of O OO OO OO OO OO OO
(WM fsjM rjfM ranf (vnj rvirvt
765
-------
991
o
"si
OB
1^1
IV
c;
KJ
O
-si
o
-si
jo
O
-si
CO
-g
00
V4
O
C
O
•>i
uo
W4
O
'*•
O
-si
CO
IV)
O
-si
00
o
•s|
oo
ro
00
c
O O
•si -si
•» oo
t\i ro
OO "si
r c
AJ
S»
O
•sj
jj
rw
-sf
O X
-si -Jt
uu I
c;
*
o
o
9
o
o
v^
w»
00
O O
* *
O -Jk
u-4 O»
Wi 4t»
09 00
O
*
O
OS
r\»
o
o
1
00
fSi
o
O
*
V/l
o o
• *
£»
-si
o
i
TV
Wt
IV
O
ru
O C
-*O *-
"0
r-
m
-si-si •>|->l-Nl*sl
ChO
O
•
-si
t>
r*
o
O-
-o
X
o
Q009"sl~s(
-sj-sl-sj-sj
oo
OO ~*<-*QQO -*
ii 11111 « 11*1
i\» ooo-sj>
O
<;
oo
ra
O-* Uq««l»i~si«» -si
** •***• I II** »* «Jk
uio o«*oo»o o oooo oo -*
oo
oo ooooo o ouiooo o-J
-------
OO O"0<\i<*Q O N-N-«*Q O OOOO OO «Ar>OO O (MO-O^sf
( f tit** * i » • • * *«** * * *t*i * **t*
O NKIN-N fl flOnNH^ O-«0 IT>xO«3Q K>
v KIT-SON- fMu"< QCM-VO w
fN» K»,fOinm *5 OeO-^O VON- T-N"*«A O»
«c
>
OOOOO O »•*•»•»• T* OvnvOO «*8Q T», o»-oo o- ^-rwrufv O>N> oooo <>
O
a. oO'Ofo «0Mt> sr»*'<»y!3 so in N-N.N.OQ oc ofMr-fM N.M? vooaof- oo
O **** ** ***** * ***« * • • • • i * *•*« *
X
a,
©
-J
_J
a
at
*n sn rs* f\j T" T~ f*n f^ O O
so oo ?\i fv f*J rv NO ^5 in tn
^" ^* ^1 ^-J ^*~ ^*» (^^ C^ v^ s^
fs^ ^»« f%| f\| ?>A ^> "^ ^T C2 C3
^3 NOfV MO O- fw N> kn tA
T- r-fSJ fVO OO O-* •^>
* »< II t* t* •
O OO OO OO OO O
j "~5 i *™4 i *"^ i **t i ^ i i
«J ^J^Ml ^»«J ^ ^ iwl ««l Z-U^JI »J
O ^ f"l ^ ^f 1A W^ T~ v* fM fV ip— fV
M Kl h^K^ K^r^ X10G OC;QO OTj<> O
I on $e> 0> ^r> P«- oc* op
-------
891
o
06
US
O
rv *v
o O
<* Co
o o
r c
•V IV
*• >
O O
CO <#
00 ,1C
O O
<*d IV
r c
> >•
o o
C"j CW
O
IN*
r
O
~*
c
O
t»
»
o
.-j f\i
*» 3»
Q 0
•N %i
M U»
-o *o
O v/»
r c
o
CO
•o
rv
a»
O
O ^
•SJ "?
go f
^) ^
1/4 £
r
O
-*
o
_a
O
-* o
I I
O w*<
-A ^>
O !•"«
_» -J
o o
o
•
O
o
t
o o
*- wi
Ol -O
o- «*
IV -*
o
I
I i
«* *>
35
t>
r
•si
•si
"0
X
r>
r*
o
(\J
-*-»hj'sj
00
* I • * i O
IK oas-wo- -o
O-»OO 00
ooa>cx»
o ~»ooo
I •***
oooo -*
ui
r
Ov
M
*
O.
oo
ro
O OOON> O
O OOOOO
O OOOO
-------
f~ QlAOO
v « • i »
«C
OOOO O W>OOO O O-ifir-O-O O OOOW» O IT>QQOV> O r-NOWO o T-
O
w
Ct
(V (MfSjrkifM fM
>,f>.fs.
•an.
0.
O
i
X
a,
o
oo
rs»
so
in
•ft
o
u
X
0.
oo
O
oo
*- oo
Ui
-J
0
to T-
WW* V
fM
o
*
o
T™ O
V" *0
rg *3
O ^0
e *
O -~
fv
T- O
-
* o
in NO
o o
no ec
•o «t"
O O
«r «c
Q O
oo
o
o
O
W
OO
o
O
00
o
fM fM
O r-
V T"
o- rr
rrt oo
o o
< «e
fW (N
«i ec
W no
O O
*t <€
fM fM
cc
oo
o
<
rvl
/69
-------
OLl
> >> >> a* >• >> >> > H
o oo p a .: oo oo oo o so
cc co co £» —> _* JP-. J>>> _i —k _i
« -O O O >o -O ** OO O*"1 4>W O<
oo fio ex 4^ f*o fw v/i v^i *-js ot wi wi
wn ui _» »i o oiv* ru .*- *-^ _» _i
O O ** 4suii U4^C -O *- *-. —i —»
HI
m
O*"* O^OO'^ *sit>^ ^0^0^***^ VK/l fw 'ONJ*-^ fsj CKf\J'usj'OaW *«Jk^s 00*^4NUFt O
t^ Ovooo>o-
* t » « « g *• • i * i o
Ov ooowiw* ^c^* ^s> rsj "si ^>"^j 0*0 |*^''OOO'«D c& ixuooot • • ** •!•«
•Nl OiNJOO^rv O-M *OO>-*-xJ wu« O-Nlw4a>rv> oao
OO OO\«OO Oce OOOO m -»M«UiO «*>O OOUIO -*
-------
oo omooo o otnoo o»o oooo oo mooo o mooo oo ow>oo
ti t i t i • • t i t t •• • • t • •• *••• i •••• •• i t * •
«• ooowv* NCJ ooar^ «»i« K>«OT-IA is. oooo oo
^*r»r~r»ra «o monnoo oi« «»^»fM o
IU lilt «• ttltl i *««* it tttt tt «tt« I
OO
ooo
O-^O
tt
"C-O «O«5h-KN I'. r-XJO^O XJiO NNN-W N-N KSON-N N flCNNN
X
0.
r-KOr- T"
V» tttt tt ttttt t tttt tt tttt tt tttt I
X
a,
VI
«e
a
UJ jy.rs.fwhK, (S.N- aSW«C'OOW N. fs-N.fs.fv f««r>- oafv(S.flO f>-CO fWN-Nf«. N 40 «0 OB 00
X
IX
-J
a
*:
< «J
•* r-
ursfM oj*o »oo Orst rvifvi cvir> i^\f\»
3!>O soo ar- r-N fvsr sf'O OO.
o ooin »nr^ K»O >o»r% m^
rv) r-j t*> ro in (A (M ra K> MO O fw
T-. T-.J 3.J 33.J 3.J => «J =>U
Or- *• «M PJf^ K> •• »-(M P4fO K»!A
MO OO OO OO OO OO OO
IIs- fs,|v N-h- fvco ecoa aoao OOOD
3KN- fs,fs, fs-h- N.fs. fw N. fwfw fwfw
WO 00 OO 00 00 00 OO
t- < << << ic
-------
ZLi
ruru ruru ro ru ru r%> ruru ru ru ru ru c/t
oo oo oo oo oo oo o o »
o^J -**J *\J *«ij «^ «\J *%! «^| w^j «»^ MH^ MI^ fe^j «^ ^|E
oo us oo oo oo ce i» co * oo a»oa to a> |
«^«i ^s«* •* jk -*o oo OO OOM
u»ru ru~* «* o ooo c»-st "sta- ovicy
rc re re re re re re
oo oo oo oo oo oo oo
*» t * * * t 8 •* It I*
-*uj uru ruo OO oo o«* -* -sj
o ^** 4^ o o uj mt uJ t/4 o 0* "**J "sj ru
** «A ~A OS) CO O OO& Ctt O OCO CC <3
ruru ru**4 *^ui mru rucrf MV/I wt^
1^1 -sj ~w4 ujiv ruu4 ui«>% —* "sj
r
m
IB
X
m
-**•-?» at ooaoo^ o
ot» uoww'O *•• -*joou< ra
ae
O^ •slOOO O OOOO O OOOO *si OOO1^ O OOO^sJO O OOOOO O 2^
• • • • • * *»«« II *>»• • •«•« • *««ll » «•««! » <•»
i*i '"siimi**si**j C)
CO i^4trf44^*i\l *^
"0
OO*OO» OK O-OO-O -si ChO^O-sl O OOO-sJO O> OOOOO Ch "»>
*ll* • »•*» 1 ««*« t «(•«* * ••••* I p
*» •sj*»o-si o ^ooow -o «-sj«M«*^t -sj jw 09 «*
o
M
o •»£«»*•»*.»* o oooo O ~*ooo ru ru-*-*—*> «* K»-i-*«A-* o «>*oO"*-«* •»» n
ti^ »ii» i*«ift • iAftfi • ttttittt a attti* a m
o& w<^«M*c,3w« ru wvu^o^ru *si ru&^^vo t^ CMOIOSO^ ^t ^"si^sj^oio *si fuod^o^j ru <*•&
09 UtOO^O IV W-*-*»* o -AOO-O O -*IVsJO -sj MOOUI-Mut 06 W^O1*-*-* •* K
ut wooo-'Q o«> ru UiNluru ru cs-o-G-sios M Mgiooso -* >
OOOOO *• OOUJ--J O, OOUt*" *O O
-------
fSj 3\J l\> f\> pyj ftj tV t\J f\J f\J f\) |\> M M
> > *• >> >> »> a>> > > H
O OO OO OO OO OO OO*
CK> JO CO t_
•O -OO OO OO OO o -*
"si -JO O O- O- *- 4*>^O O (Nj f\j *»
UI UI *3 OO f^vt fO fsd i\) CH Os ^^1 *>*4
3*
•o
X
m
* » » • » I* *•»* * till I I*** I lill Illllll *IC»
• **•• tl It** *
O •OCh~*> *M0 0
OOO
•o
X
• !**• ii i«ii i **!* i *i«* i
o lO-si**^ out ->(*«a«oo o *-wcrf-*g U4 -g*^v/ich wt
ooru oox^^o ^ a-'O-si-* *• ^ownut oo UI»>N»W NJW -J»>o*-O4vi 0-04 -*
-*o -*ooo o oooo o oooo o oo-*o
* ***» i* **** i **ii i **** « **i*
rv crfaoo-si ru-g ooocooo ivt *-w*>wi w*> •^GKO^UI ~O ooul
UI fVN»(V(vt •* •* -* Uiro 4>UIMUlUJ MM
"6 OO^*-' -*O -»-OOoai -* 4SMMUI UI ««jmO«4s 00
04 *«-»-»^. v/J*>i uirjt>—k ru o-M-sio -* MOO^OK *-
t *««• • « i**t » *i»* i t*»i i
O OOOO OM O*«i^fV O OO.UIO *J -*M-«*. 00 OuJ-»W OO OOOOO OO •*
-------
PLl
OO OO OO OO OO 00 O»
"«*J *ti «sj -mj -^J "^ «O «*J ""^ *^ ™*ij *^J **\J *j£
Oo . on vX> .J6 <3O 'JO Oi uJ 00 (jo tAOb u>i
rviv r«j)v i\jr\j rvi INJ i\iro ru i\> (u>^
QKUI wi*» 4i*m i,^ r j K» —* "*'O oo
r- c re r-c re re re r
OO OO OO OO OO OO O
II II II II II *l I
IMO op o-* -* -» -*w i^o o
Uf uJ uJ O O tJ "^'"J "W"^ "-4O OO1- CKff»
m
TJ
X
>
o
-si CS-NJ-^-Nl fvj-O -»>OCiOt» CO O^OSt* O
**-* -slO*~*»w< O
OK
***« i* **•» *
00
ui
oo oooo-* o oooo **«* ra-*-»-* N> iw«*-»«* -* -k-*«*^, oo -*ooo-» o
«• *ii** « i**i ii 11*1 * it** * iiii *i iiiii nn
o>
V/t
04 tvOfJwi oo-o -»<>sO-* ->i **
-------
r- OO OOOO O O*OO»-O CM/i OOOt^Q O OOOOO OC NftiNO OO
r- ti i«i« t * * t t • *t «ttt •• • • • • « > i • • i • ** iiii t«
r-O •omrjfv Ktc* ONr-KiiO N. rwO-»tir\)« fs-T- oo»-rM«o o*>
ooa-moo* m oir* oo«o*t~ »••«-
r-T-«M(V (M KiKifvi^K-. 4O rj K>M«v-*ni *"«\i OOr-«~
M
r- CM- (V*»OO *O m>©lf>N«' SffM N*»«AN N-fv
X
Q.
eoo-
M *• •••• t tfttt • • •••• it ••tin « ••iii it ««i»
< ->0»O N.|s.f>-P«- O vOXJOO-O OO -OO-OO ONO N-N-OO'Ci O N-N-OOO O«O O>0O
X
«O
r- -oc> iAstinN r- OOOf^w .o«o r-QO-*- r-»~ ^VONON.^ N- ir*N-fw^«r o-N.
O
UJ
X
0-
UJ
-I
a.
eom mo OKI K»»A »««\i rviio in
C*OJ fM^ r*O ON* •*xt %***• -*m IA>O o
sf m U-KI K> ^. ,^'eo eoK> KIIA IA
fvifui rvjr- ». ^ v.^. T- O OO O
** •• i« • « > • it *
OO OO OO O'O OO OO O
rot*** BOO CKO o i~ T-«M rg
OJ K>fO K-)K> K»
Wqo «5W or>nn , N.N- fs-N- f»»
oo oo oo oo oo o
"*< << << << << «t
MPJ (M(\i CVJfV)
-------
9 LI
p^ \\i r^ fw f\j tw NJ f^o fw r\J f">j fvi w^
3t* 3* jtfr !i* 5* 5* ^* X1* 3* 3** 3«* 5* ""^
o oo oo oo oo oo o »
"^1 **^ ™*^ ***^ ""''J **^ """^ *^J ™^J ""^ ™"^ **^ *2i*
oa jo uo oo ui ttt \M wS 'Jo -ja u» uw I
^ ^Oco OtOC OOW ^ \M ist CM 5-H »^
fvj -* ru ru -^ —* v/i <-f> 4-* 4> i,/ji v^o
r re re re re re: r
o oo oo oo oo o
I • • « • • • • • •
*. JSO OO OfU t\>-* -*
Ml •t "°*i *^4 ^5 "O fW FU 0^ 'i^
O O 4*** ^*. w^ CH (V IV *M ""^
UH W1O- O>rO ISJISJ *VQO *ff!
Ci di wJ ^^ *^ "** f^J 'V C^ ^ ~~^
-o
r
m
ti
X
OO^XsCOOO *%i ^"""^"Nl***! CN *s( ^^O^*1*^ *sl*sl -^JNI'*W"*J tVi
lilt *«t* * **•• ** •••• • •*•* t •*»*• i « lilt jo
C5C3Ci*** ""•^'"•DCS'*^ *O (j^fW'-^w^ •**'^J *%J{^)O?tO fy w4~—^(wa^tjJ ^*
1 »*«•<* £15
X
>
r»
*> a>
ou4*»oo w»i^ ivut-oui c& owoo* OB
"0
x
00
00
OOOO OOOO O -»-*-»-* t»v» O*J<^iO> •* «j>o«*"* O OOOOO
• «»• i * » • i » i i < • • tiii i iii* i tii«t i« «•** nt
•Nl
»
r
W OOOOOI OfS'O W~>JO40O W -Nl-^WNj O OMXJ'^'O** POtM
oo oooo o a**-j"«j o oofvo^o oo
-------
Ul
ru
>
a
oo
oo
o
ru PU
> a*
O O
oo oo
co a>
o o
*•> v*l
r c
ru ru
O O
CO OJ
CM «»
o o
oi ru
r c
O O
O3 OP
CO CO
o o
(V -k
r c
fa
>
O
Of
os
o
f\» fU
> >
O O
-J ^
C8 O3
O O
Q\ Ul
r c
O
•sj
tt>
(V
>
o
3» H
O X3
^sl Jf
S5 i-H
l-»4 O
r
a
O
_»
O
_k
o
-* o
> I
o w
-A «
o w
-A «4
o o
o o
» a
w *•
•O WI
UJ Os
•si O*
O (VJ
o o
O. -M
ru -A
O
*
m
x
m
o -*
to
•
VI
oo
"M
ui
oo
wn«*u«c»
O
"0
X
•
V/»
00
O
O
00
O O-»
o oooo ru
i iit» t
os
««* OOOO Oe OGfr"A*sl'>0 OOOO
O QOQ*»O
O
-*OOO -*OOO-* O
««*« lisa* m
ru
ru
t
o
» *•*«*
O OOOOO
UiruOOOO O OOOtvl O
O OOOO OOU4U<
-------
*- o»neo
oooo o tnooo o
tttf • iiii *
O-
o mooom o *-NO«OQ
NO
«n
90
JM
0? r-
in
lu
O
O
*• "r-Or-Or- O r-r»*"T- <~
«- »" OOOO O
O
o
>O
X
a,
T- Q!»VC»T"
X
a.
->O
N.
I
SO
W* «O(VOO«» CK
IA
K*
O
(A
«M
•*
I
K*OCf*"<*A O
T- N-OC^OO fM
X
0,
ai
a.
z:
sr W-
na
o
O
in K
^* so
O NO
Kt NO
I I
o *-
NO in
NO f"
-O f-
•« K>
I I
r- O
CM O
«n IM
»- o
«~
to oo
o o
3 -I
NO «0
O O
«•> «f>
CO W
o o
"* <
IM rvi
CO 0»
O O
TO «3
CO 00
O O
< <
M l\t
*> O
O «-
«o w
CO 00
o o
O T"
»— T~
so «e
DO eo
O O
•c <
«> co
03 00
O O
00
00
o
178
-------
«" Otnoo O-t- O(A«S-o» o* N-N.rs.ch o«o ON^NIT •*
TMp»T-r»»» «r-»-
oaoino rao
Ul Illl II Illll I Illl ll*llllt II lllili Illll I
in) ^pfr^f^m^m OO OOOOO O f^T^t1*^*" Of fVTl*v*T"rvt T"ifw OOO^*" OO OOOOO O
O
V* MJISfMC^ ^f^C)
Ii iiii ii 11111 i iiii ii iiiit ii iiii ii iiiii i
X
Ou
U 1111 II Illll • 1111 ii 11111 ii till || 11111 I
X
a
o
^** NNNN NN NNNNN NN NNNN NN ONOON
:r
u;
J
a
ItJ
v) *-
Kt
•
O
i/i
i
O
ori
«*
i
O
eoirt
ii
OO
O
(M
tn
O
i
O
KI
O
r-j
O
(M
O
N-
O
O
f«.
O
(n
i
O
0f>
O
O
,«
i
O
O
IA
90
O
O
Or-
MO
OiO
O
9-
rg
O
O
OO
OO
«C <
KI
O
O
«*
O
<*•
O
O~
00
O
O
<>
00
O
yi
O
o
O
O
O
O1-
CO
O
*U.S.GOVERNMENTPWNTINGOmCE:1987 -7U8-121/ 67063
/7S
------- |