RESEARCH  TRIANGLE   INSTITUTE
                                                                  Contract No. 68-C9-0013
                                                                          March 1990
                                                     Background Paper:
                              Use  Support Assessment Methods
                                                                          Prepared by

                                                                          Randy Dodd
                                                                        Mike McCarthy
                                                                           Keith Little
                                                                       Pat Cunningham
                                                                           Julie Duffin
                                                                 Research Triangle Institute
                                                                        PO. Box 12194
                                                           Research Triangle Park, NC 27709


                                                                       Wayne Praskins
                                                          U.S. Environmental Protection Agency
                                                    Assessment and Watershed Protection Division
                                                                          401 M Street
                                                                  Washington, DC 20460


                                                                       Ann Armstrong
                                                     School of Forestry and Environmental Studies
                                                                       Duke University
                                                                    Durham, NC 27706
POST OFFICE BOX 121 94  RESEARCH TRIANGLE PARK,  NORTH CAROLINA 27709-2194

-------
Contract No 68-C9-0013 March 1990
Background Paper:
Use Support Assessment Methods
Prepared by
Randy Dodd
Mike McCarthy
Keith Little
Pat Cunningham
Julie Duffin
Research Triangle Institute
P0 Box 12194
Research Triangle Park, NC 27709
Wayne Praskins
US Environmental Protection Agency
Assessment and Watershed Protection Division
401 M Street
Washington, DC 20460
Ann Armstrong
School of Forestry and Environmental Studies
Duke University
Durham, NC 27706

-------
CONTENTS
CONTENTS
Section Page
Figures . iv
Table . iv
Introduction 1-1
2 Assessment Methods Used by States in 1988 §305(b) Reports 2-1
2.1 Rivers and Streams 2-1
2.2 Lakes 2-2
2.3 Estuaries/Coastal Waters 2-2
3. Overview of Assessment Methods 3-1
3.1 “Monitored” Methods 3-1
3.1 .1 Interpreting Pollutant-Specific Data 3-1
3 1.2 Physical/Chemical Indices 3-3
3 1.3 Interpreting Biological Data 3-5
3 1.4 Interpreting Toxicological Data 3-5
3.1.5 Combining Different Types of Monitoring Data 3-5
32 “Evaluated” Methods 3-6
3.2.1 Use Restrictions 3-6
3.2.2 Fishery Status 3-7
3 2 3 Presence of Sources . 3-8
32.4 Public Input 3-8
3 3 Combining Evaluated and Monitored Information 3-9
4 References 4-1
APPENDIXES
A 1988 Use Support Approaches . . A-i
B Use of Reference Distributions to Assess Water Quality Criteria
Violations . . B-i
C Trend Assessments and Threatened Uses C-i
D Water Quality Indices . . .. . . . .. D-1
E Biological Methods E-i
F Toxicological Methods . F-i
G Remote Sensing . G-1
H Use of Flow Diagrams . . H-i
III

-------
CONTENTS
FIGURES
Number Page
1-1 Criteria for designated use support classification 1-2
3-1 Minnesota’s lake observer survey 3-11
TABLE
Number
1-1 Designated Use Support in Rivers and Streams 1-4
Iv

-------
1. INTRODUCTION
1. INTRODUCTION
The Current Process
Section 305(b) of the Clean Water Act requires each State, Territory, and
Interstate Commission (hereafter collectively referred to as States) to submit a
report to the EPA every 2 years that describes the quality of its surface and
ground waters. In 1988, States reported on 29 percent of total stream miles in
the United States, 41 percent of total lake acres, and approximately 76
percent of total estuarine miles. These §305(b) reports are summarized by
EPA and transmitted to Congress. This reporting process is designed to
provide each State with a comprehensive, systematic water quality
assessment and EPA with information needed to assess water quality
nationwide
In preparing their biennial §305(b) assessments, States are encouraged to
compile a wide variety of chemical, toxicological, ecological, and other data
describing the condition of their waters (and drainage areas) States are
increasingly using data collected in biological surveys to complement more
traditional physical/chemical measurements. States interpret the data to
report the degree to which the designated uses specified in the State water
quality standards are met. Föürb tegoriès of use support are reported: fully
suppoiling uses, partially supporting uses, not supporting uses, and fully
supporting uses but threatened. Threatened waters are identified by
downward trends in water quality or other evaluations indicating possible
future degradation (U S. EPA, 1989)
The Problem
EPA has provided only limited, and fairly simplistic, guidance on how to make
use support determinations:
• States are asked to judge the quality of the data they use by classifying
assessed waters as ‘monitored waters’ or “evaluated waters”
• Section 305(b) guidelines recommend that chemical data be used to
classify waters as fully, partially, or not supporting their uses based on the
percentage of measurements exceeding criteria (These recommenda-
tions, included in this document as Figure 1-1, are referred to as “Figure
1” in the 1990 §305(b) guidance They were developed in the “States’
Evaluation of Progress” project in 1981
1—1

-------
Assessment
BasIs
Support of
Designated Us.
Assessment Description
Fully Supporting
Partially Supporting
Not SupportIng
Evaluated
No site-specific ambient
data. Assessment is based
on land use, location of
sources, citizen complaints,
etc. Predictive models use
estimated inputs; are not
calibratedNerif led.
No sources (point or non-
point) are present that could
Interfere with the use, or
sources present but informs-
tion Indicates uses fully
attained. Criteria attainment
predicted.
Sources are present and
Information indicates uses
are partially supported or
uncertainty about use
support. Complaints on
record.
Sources are present and
Information dearly Indicates
use not supported. Criteria
exceedences predicted.
Monitored
Fixed station sampling or
For all pollutants, criteria
For any one pollutant, criteria
For any one pollutant, criteria
(Chemistry)
survey sampling. Chemical
analysis of waler, sediment,
or biota.
exceeded In 10% of meas-
urements and mean of
measurements Is lass than
criteria. Pollutants not found
at levels of concern, where
criteria not available,
exceeded 11-25% and mean
of measurements Is less
than criteria; or criteria
exceeded 1 0% and mean Is
greater than criteria. Pollut-
ants not found at levels of
concern, where criteria not
available.
exceeded >25% or criteria
exceeded 11-25% and mean
of measurements Is greater
than criteria. Pollutants found
at levels of concern, where
crIteria not available.
Monitored
Site visit by qualified biologl-
Use fully supported; no
Some uncertaInty about use
Use dearly not supported;
(Biology)
cal personnel. Rapid bio-
assessment protocols may
be used.
evidence of modification of
comnxinhty (within natural
range of controtlecoreglon).
support; some modification
of community noted.
definite modification of
comnunity.
Classification Guldetines for Multiple Use Waterbodies
Fully supporting — All uses are fully supported
Partially supportIng — One or more uses partially supported and remaining uses are fully supported
Not supporting — One or more uses not supported
FIgure 1-1. CrIteria for designated use support classification.
-I
0
0
z

-------
1. INTRODUCTION
• c ariyôñè po utant implies nonattainment of water quality standards, the
waterbody is reported as not supporting its use
• FF e lack of national guidance on how to make use support determinations
has contributed to inconsistencies among States in how use support
determinations are made. Therefore, interstate comparisons of
assessment results are suspect, and results in national summary statistics
are of unknown validity.
• Some States rely exclusively on monitoring data to make use support
determinations (e g., Massachusetts, Minnesota, Ohio, South Carolina,
Texas, Virginia), whereas other States rely largely on “evaluative”
information (e.g., New York, Mississippi, Alabama, Montana).
• Some States report waters where evaluative information indicates a
problem as threatened; other States ignore evaluative information.
• Some States use a wide variety of chemical and biological monitoring
data; others rely solely on fixed-station water column chemistry sampling.
• Many States use the “Figure 1” guidance to analyze their chemical data;
others rely on professional judgment.
• Several States assign partial support status to waters experiencing severe
fish kills; other States assign nonsupport status.
• Five States report 90 percent or more of their waters fully supporting uses;
whereas one State reports only 1 percent fully supporting uses, reflecting
differences in water quality and use support methodology. (Table 1-1
presents summary statistics from 1988 State § 305(b) reports.)
The lack of guidance is a problem not only for EPA, but for those States that
seek technical assistance to improve the quality of their §305(b) reports.
States have requested assistance on how to deal with apparent conflicts
between different types of data, how to account for frequency and duration
specifications in water quality criteria, how to interpret unfamiliar types of data,
and other topics.
Project Objectives
This background paper was prepared for the EPA/State “305(b) Consistency
Workgroup” The workgroup was created to assist EPA in achieving two
goals:
• Improve national reporting by increasing consistency among States in
data collection, data analysis methods, total waters reported, and/or the
fraction of waters assessed
• Improve the accuracy and coverage of State §305(b) assessments by
providing additional guidance on data collection and/or analysis methods.
1-3

-------
1. INTRODUCTION
Table 1-1. Designated Use Support In Rlve, and Stieams
btal
River
Mu Assessed
es
Percent
Miles
Fully
Percent
Miles
Part Ially
Percent
Miles
Not
Peicent
Percent
State
Miles
Total
Evaluated
Monitored
Supporting
Supporting
Supporting
Alabama
40,600
11,174
85
15
91
6
4
Anzona
8,671
2,279
—
—
69
9
21
Arkansas
11,508
4,107
48
54
42
1
58
California
26,970
9,885
—
—
67
22
11
Colorado
14,655
10,000
54
46
86
7
7
Connecticut
8,400
880
33
68
68
27
7
Delaware
500
467
0
100
60
33
7
Delaware River Basin
206
206
—
—
94
0
6
District of Columbia
36
26
0
100
0
0
100
Flonda
12,659
7,943
27
73
67
25
8
Georgia
20,000
20,000
66
34
97
2
1
Hawaii
349
349
28
72
76
23
1
Illinois
14,080
12,970
23
77
45
54
1
Indiana
90,000
5,181
28
72
68
19
13
Iowa
18,300
8,235
75
25
1
79
20
Kansas
19,791
6,888
57
43
58
11
31
Kentucky
18,465
8,653
63
37
71
10
18
Louisiana
14,180
8,483
—
—
68
25
7
Maine
31672
31672
—
—
99
0
1
Maryland
9,300
9,300
84
16
93
5
2
Massachusetts
10,704
1,646
0
100
43
36
20
Michigan
36,350
36,350
—
—
98
0
2
Minnesota
91,944
4,443
0
100
35
13
52
Mississippi
15,623
15,623
87
13
89
9
3
Missoun
19,630
19,630
77
23
52
48
0
Montana
20532
19505
85
15
63
34
3
Nebraska
10,212
5,690
—
—
57
21
22
New Hampshire
14,544
1,331
77
23
71
16
13
NewMexico
3,500
1,152
—
—
50
48
2
New York
70,000
69,988
95
5
76
12
12
North Carolina
37,378
33,275
45
55
67
28
5
North Dakota
11,284
9,850
44
56
69
31
0
Ohio
43,917
7,045
0
100
32
21
47
Ohio River Velley
981
981
17
83
0
100
0
Oklahoma
19,791
9,248
36
64
36
38
26
Oregon
90,000
27,738
—
—
45
31
24
Pennsylvania
50,000
13.242
39
61
73
13
14
Puerto Rico
5,373
5,373
67
33
46
21
33
Rhode Island
724
581
43
57
84
2
13
South Carolina
9,900
3,795
0
100
74
10
15
South Dakota
9,937
3,750
18
82
37
34
29
Tennessee
19,124
9,428
—
—
63
28
10
Texas
80,000
13,998
0
100
87
0
13
Vermont
5,162
5,162
83
17
88
7
5
Virginia
27,240
3,532
0
100
34
40
26
Washington
40.492
4,621
22
78
50
35
16
West Virginia
28,361
14,301
46
54
20
71
9
Wyoming
19,437
19,437
67
33
83
17
0
Totals
1,150,482
519,413
70
20
10
— Noi reported
Source 1
988 State Section 305(
b) reports
1-4

-------
1. INTRODUCTION
The workgroup will be asked to recommend practical, technically sound
improvements to the §305(b) process on the type or quality of data that are
used and on methods for interpreting data
What Is in This Document?
This document focuses on:
• A review of 1988 use support approaches
• An overview of both monitored and evaluated assessment methods
• Detailed discussions and examples of several specific topics in
appendixes: use of reference distributions for providing a more site-
specific assessment of water quality criteria; water quality indices; use of
flowcharts; trend analysis; and the utility of pollutant-specific, biological,
toxicological, and remote sensing methods for determining use support.
1-5

-------
2. ASSESSMENT METHODS
2. ASSESSMENT METHODS USED BY STATES
IN 1988 §305(b) REPORTS
As a result of limited guidance, and diversity in States’ waters and programs,
§305(b) assessment results vary widely. Table 1-1 presents data on the
percent of assessed rivers and streams reported as fully supporting or
impaired (partially or not supporting) by each State in 1988. Although
interstate variability resulting from differences in types of waterbodies, criteria,
available resources, pollution problems, percent of total waters assessed, and
other factors is to be expected, it is desirable for national assessments (i e.,
§305 [ b]) to contain minimal variability due to different methods.
A review of 1988 use support approaches used in rivers, lakes and estuaries
is summarized below A more complete analysis is provided for each Region
and State in Appendix A.
2.1 Rivers and Streams
Of the 53 States, 18 reported using the EPA Figure 1 approach, 12 States
reported using a modified EPA Figure 1 approach, 10 States reported using
some type of water quality index (WOl) based on chemical data (including 3
States that used a WQI in conjunction with a modified Figure 1 approach),
and 15 States provided insufficient information to determine the method used
in their use support determinations. The U.S. Virgin Islands has no
permanent streams for which to determine use support
Of the 18 States that reported using the EPA Figure 1 approach, many
mentioned Figure 1 only once and gave no further indications that the
approach was consistently used or that the results of this analysis would
override best professional judgment
Twelve States used a modified EPA Figure 1 approach That is, they adopted
the concept of making use support determinations based on evaluated vs
monitored information, percent of samples exceeding water quality criteria, or
other features of the EPA Figure 1 approach, but they took liberties with the
exact selection criteria used Again, how rigorously some of these States
applied their Figure 1-style decision criteria is unknown Professional
judgment rather than systematic analyses may feature more prominently in
use support determinations than the §305(b) reports indicate
2-1

-------
2. ASSESSMENT METHODS
2.2 Lakes
Review of the 1988 §305(b) reports does not provide a clear picture of how
States make use support determination for lakes. Of the 53 States, 14 used
the EPA Figure 1 approach or a modified EPA Figure 1 approach, 10 States
used Carison’s Trophic State Index (TSI), 15 States used other trophic indices
or other methods, and 12 States did not provide sufficient information to
determine the method used in their use support determinations. It is not
clearly stated, however, if States using trophic indices actually correlated the
TSI values to use support determinations. Hawaii and the U.S. Virgin Islands
have no lakes for which to determine use support.
A few States mentioned using toxics data in their assessments. States using
the EPA Figure 1 approach probably screen for toxics exceedances during
their STORET runs States that used only trophic status information may not
consider toxics in their routine lake assessments, but may incorporate toxics
concerns through the use of best professional judgment.
Of the 14 States that reported using the EPA Figure 1 approach, many
mentioned Figure 1 only once and provided no further information as to how
this approach was actually applied or whether the results of this analysis
overrode best professional judgment.
2.3 Estuaries/Coastal Waters
Little specific information on methods used in estuarine use support
determination is available in the §305(b) reports Of the 26 States having
coastal waters, 13 States gave insufficient information to classify their
assessment methods Twelve States reported using Figure 1 or a
modification, and one (Florida) used a TSI.
2-2

-------
3. OVERVIEW OF ASSESSMENT METHODS
3. OVERVIEW OF ASSESSMENT METHODS
Section 305(b) guidelines distinguish between monitored and evaluated
approaches. M it ied assessments are those based on recent ambient,
physical/chemical/bacteriological data, ambient biological data, or
toxicological data. Biological data include quantitative information such as
the relative abundance of different invertebrate or fish species Toxicological
data consist of effluent bioassay results or, less frequently, ambient toxicity
testing results.
[ ,Evaluated assessments are those based on information other than recent
monitoring data Such information includes fish kill records; remote sensing;
water quality and land use data; questionnaire and survey results; public
input; and professional judgment The distinction between monitored and
evaluated assessments can be expected to conflict in certain circumstances.
This might occur where monitoring results fail to identify a perceived loss of
uses (i e , where monitoring results and professional judgment are in conflict),
or where monitoring data exist only for pollutants and media without water
quality criteria, such as for metals in sediments
3.1 “MONITORED” METHODS
3.1.1 ‘interpeting Pollutant-Specific Data
Virtually all States compare pollutant-specific ambient data to water quality
standards to determine use support. Standards are generally not intended to
be fixed limits but to specify levels not to be exceeded for a specified duration
at a specified frequency. Newer EPA criteria for toxic pollutants specify two
values, each with its own duration and frequency specifications:
• The acute value should protect aquatic life if the 1-hour average concen-
tration does not exceed the criterion value more than once every 3 years
on the average
• The chronic value should protect aquatic life if the 4-day average concen-
tration does not exceed the criterion value more than once every 3 years
on the average
The duration and frequency aspects reflect the random or probabilistic nature
of water quality variables
3-1

-------
3. OVERVIEW OF ASSESSMENT METHODS
One fairly simplistic approach to accounting for the random behavior of water
quality variables is the “EPA Figure 1” approach (see Figure 1-1), which
recommends that the determination of use support be based on the number
of measurements exceeding a criterion A waterbody is “fully supporting” its
uses if fewer than 10 percent of the measurements exceed the criterion;
“partially supporting” if between 10 and 25 percent of the measurements
exceed the criterion; and “not supporting” if greater than 25 percent of the
measurements exceed the criterion. The figure does not make use of the new
two-number criteria. One State that has modified the Figure 1 approach to
consider both acute and chronic criteria is Rhode Island. They report partial
use support, for example, if between 10 and 33 percent of samples exceed
the chronic criterion but no samples exceed the acute criterion.
Other techniques for interpreting pollutant-specific data are discussed in
Appendixes B and C. The approach described in Appendix B improves on
“EPA Figure 1” by replacing the arbitrary categories (0 to 10 percent, 10 to 25
percent, over 25 percent) with site-specific or region-specific categories The
categories are calculated from reference probability distributions determined
from data describing clean water sites. Appendix C discusses the use of
trend analyses to determine threatened waters.
Another issue raised by Figure 1 is how to determine use support when
measurements of several different parameters are available at a single site.
Figure 1 implies a “worst case approach” in which an exceedance of a criteri-
on for any one pollutant implies less than full use support An alternative
approach is to use a water quality index that weights each parameters (see
Section 3.1.2)
The Center for Exposure Assessment has recently developed support for a
model (DYNTOX) that predicts frequency and duration of criteria exceedance
for a given stream and waste source. The model addresses the problem of
extrapolating monitoring data and enables the user to choose from three
approaches: a continuous simulation approach (requiring continuous flow
and concentration data), a Monte Carlo approach, and a log normal
approach, which requires daily mean and variance flow and concentration
data. Although this model should be useful for interpreting instream data
when supported by intensive studies in high-priority waterbodies, it probably
does not provide a practical broad-scale solution to determining the
frequency and duration of criteria exceedances given current monitoring
resources
The advantages of using pollutant-specific data to assess use support
include:
• Numerical criteria are often available
• Standardized sampling and laboratory methods are available
• Ambient conditions can be related to sources
3-2

-------
3. OVERVIEW OF ASSESSMENT METHODS
• Historical data are available with reasonable spatial and temporal cover-
age
• Standardized data base management systems are available (e g.,
STORET, WATSTORE)
Physical/chemical measurements can also have less-than-desirable attributes
for use support determinations:
• Pollutants of interest must be known
• Criteria for the pollutants of interest often do not exist or are below analyt-
ical detection levels
• Ambient data are highly variable, resulting in an inability to test hypothe-
ses at high confidence levels
• Single-pollutant grab sampling may be difficult to relate to aquatic criteria
(e g., 3-day-average chronic criteria) or public health effects because of
issues of exposure, synergism/antagonism, and biomagnification
o Monitoring results can be difficult to communicate to laypersons
• A substantial number of chemical and physical parameters may be
needed to adequately represent water quality
Another major difficulty for national use is the difference in water quality
standards among States. This issue manifests itself as bias (where two
States have different standards for the same parameters) and as unequal
coverage (where one State has a standard and another does not)
In addition to water column data, physical/chemical parameters for monitored
assessments include toxics in sediment and fish tissue These measurents
offer the advantage of integrating the effects of pollutants over time, thus
reducing the impact of high water column variability. Their major
disadvantages to assessments is the lack of a clear relationship between
chemical levels and use support (i e., the lack of criteria)
3.1.2 Physical/Chemical Indices
Physical/chemical water quality indices (WQIs) provide a means of reducing
large amounts of information into a single value that can ease communication
between technical water quality experts and laypersons Indices can be
constructed in a “worst case” approach (i e , they can signal nonsupport
when any one parameter falls below an acceptable limit), but more typically
they individually weight and aggregate several parameters Appendix 0
discusses frequently used indices and their development in some detail
3-3

-------
3. OVERVIEW OF ASSESSMENT METHODS
A simple arithmetic WQl has the formula:
n
E q 1 w 1
1=1
where
n = number of parameters (pollutants)
= the given parameter
q. = the rating value for the ith parameter (from a special curve for the ith
parameter relating instream concentrations to quality ratings for a
particular use)
w. = weighting factor for the ith parameter (i.e., how important the 1 th
parameter is to the designated use).
Indices vary considerably in the number and type of parameters incorporated
and in weighting and aggregation schemes.
The advantages of physical/chemical WQIs include the following:
• They can reflect the opinions of a large group of experts as to necessary
weighting factors or rating curves, as well as deciding what constitutes
support of designated uses.
• Once an index is developed, results are reproducible and minimal
professional judgment is required.
• Programs are available to carry out WQI calculations automatically from
STORET data.
Potential disadvantages of WOls include:
• A loss of information in the aggregation process
• A lengthy development process (although several good WQls already
exist)
• A tendency to obscure individual high pollutant measurements
• Problems due to missing data or discontinued stations in some WOls
• Difficulty in making a connection to an ecological or human health
endpoint.
3-4

-------
3. OVERVIEW OF ASSESSMENT METHODS
3.1.3 Interpreting Biological Data
Many States now supplement their physical/chemical monitoring with
biological surveys of fish, macroinvertebrates, periphyton, plankton, or other
biological communities Biological survey data are usually expressed as the
number or condition of organisms of each species present. The data may be
compared to lists of “key” or “indicator’ species or may be reduced using a
biological index A general discussion of biological methods for aquatic
communities is given in Appendix E
Recent guidance on rapid bioassessment protocols (Plafkin et aL, 1989)
argues that biological (biosurvey) measures are the best methods for
determining aquatic life impairment, whereas physical/chemical and toxicity
testing (bioassay) data are more useful for detecting and assessing sources
The principal advantages of biological methods are their ability to integrate a
variety of processes and effects into a single direct measure of aquatic life use
support and their cost-effectiveness.
3.1.4 Interpreting Toxicological Data
Indications from State 3O5(b) reports are that toxicological methods are not
as widely used in making use support determinations as physical/chemical or
biological data. As pointed out in the “draft Ecopolicy” (see Section 3 1.5),
these methods can be an important tool in use support assessment by
predicting the toxicity of receiving waters. For example, a chronic toxicity test
might measure the minimum concentration of effluent causing reproductive
effects in laboratory test organisms; this endpoint concentration then can be
used to calculate whether indigenous organisms would experience chronic
effects at low stream flows or other flows of interest.
The limitations of toxicological methods include the fact that toxicity test
results, like chemical-specific ambient data, are indirect estimators of biolog-
ical integrity That is, they assess the suitability of the aquatic environment to
support a healthy community, but they do not assess the community itself
Like biosurveys, toxicity tests address only the aquatic-life uses of a
waterbody, while other methods may be more appropriate for other uses A
general discussion of toxicological methods is given in Appendix F
3.1.5 Combining Different Types of Monitoring Data
EPA has approved a new policy (“Ecopolicy”) on the use of ecological as-
sessment information, which states that
chemical-specific analyses, toxicity testing, and biosurveys can
each provide a valid and independent assessment of designated
aquatic life use impairment When any one of the three types of
assessments demonstrates that the standard is not attained, it is
EPA’s policy that appropriate action should be taken to achieve
attainment
3-5

-------
3. OVERVIEW OF ASSESSMENT METHODS
“Appropriate action” can be interpreted to include classifying a waterbody as
not supporting or partially supporting uses. In contrast, North Carolina decid-
ed to use biological ratings preferentially over chemical ratings, which in turn
were preferred over qualitative evaluations based on a series of public work-
shops and professional staff judgment for the 1988 §305(b) report. (Biological
ratings disagreed with physical/chemical ratings [ using an index] at about 30
percent of all sites examined) Ohio used a similar decision approach. For
their 1990 §305(b) reports, both Ohio and North Carolina are further
integrating biological and physical/chemical approaches by attempting to
validate physical/chemical indices with biological ratings (see Appendix D)
This approach is essentially an alternative to the EPA Figure 1 approach
Another approach for integrating physical/chemical and biological data is to
develop an integrated index. An example of this concept is the Ohio Lake
Condition Index (LCI), which is also discussed in Appendix D. The index is a
weighted sum of 13 quantitative and qualitative measures of water quality,
including the Index of Biotic Integrity and general physical/chemical
parameters.
3.2 “EVALUATED” METHODS
Evaluated assessments are made in the absence of current ambient chemical
or biological data. Sources of “evaluated” information can include fish kill
reports, fishery status reports, water supply closure reports, old assessments,
citizen complaint logs, remote sensing data, and information on potential
nonpoint source inputs. Professional judgment is used to assess the reliabili-
ty of the information, determine the severity of the problems indicated, and
rate use support. States often rely upon evaluated information to assess use
support because economic factors limit the spatial, temporal, and/or paramet-
ric coverage of monitoring networks Information on nonpoint source impacts
is often lacking For example, Oregon monitors approximately 4 percent of its
stream miles, which receive 90 percent of the States point source discharges
(Oregon Department of Environmental Quality, 1988).
States can assess a greater percentage of their waters by using evaluated
information For example, Virginia expects to increase its percentage of
assessed waters from 20 percent in the 1988 § 305(b) report to almost 100
percent for the 1990 report by including evaluated assessments (Virginia
Water Control Board, 1989) The following discussion will describe
advantages and drawbacks of determining use support with various types of
evaluated information
3.2.1 Use Restrictions
The main limitation of information on use restrictions (e g., drinking water
restrictions, fishing advisories or bans, shellfish harvesting closures, and
swimming prohibitions), is a lack of standardized criteria for imposing
restrictions For example, criteria for closing shellfish waters are based on
3-6

-------
3. OVERVIEW OF ASSESSMENT METHODS
measurements of fecal coliforms (Oregon, North Carolina), total coliforms
(Massachusetts), enterococcus bacteria (Delaware), and E. coil bacteria
(Maine) The variety of criteria can result in inconsistent application of use
restrictions and subsequent use support determinations.
States also take different approaches to assigning a use support category to
specific restrictions Maryland assesses chronic shellfish closures as not
supporting designated uses, while conditionally approved waters are partially
supporting their uses. In contrast, North Carolina assigns a partial support
assessment to all shellfish closures because other uses are still supported
(e.g , aquatic life and contact recreational uses).
Within-State variation is perhaps a more serious drawback associated with
swimming advisories and beach closures, which are often inconsistently
monitored and imposed by local health agencies Some States are reluctant
to base assessments on swimming restrictions because regions that seldom
sample for bacterial violations (and therefore infrequently impose swimming
advisories) will appear to have better water quality than regions that strictly
enforce closure criteria Despite inconsistent application of swimming
restrictions, this approach does provide the most direct measure of attain-
ment of Clean Water Act (CWA) swimmable goals
State or local health officials are responsible for implementing drinking water
closures resulting from violations of Federal maximum contaminant levels
(MCLs) and State standards States are required to report public supply
closures in the §305(b) report, and the EPA Office of Water tracks these
closures in the Federal Reporting Data System. Therefore, the information is
accessible
3.2.2 Fishery Status
Fisheries information can indicate nonattainment of fishable goals or
impairment of aquatic habitat and fishery use designations The extent and
severity of fish kills, diseases, and abnormalities can be acquired from fish kill
reports submitted to the EPA and by surveying State fishery biologists,
commercial fishermen, and recreational fishermen
Fish kill characteristics, such as size and frequency of kill events, may indicate
partial and nonsupport of aquatic life survival and propagation uses In
Maryland, isolated fish kills caused by spills are interpreted as partial support
of designated uses; areas affected by recurring fish kills are assessed as not
supporting designated uses Establishing strict numerical criteria to rate the
severity or frequency of fish kills may result in misleading assessments The
species affected and cause of the fish kill (some are natural) must also be
considered
3-7

-------
3. OVERVIEW OF ASSESSMENT METHODS
Accurate numerical fish kill data may not be available because of the varia-
bility in the States’ abilities to quickly investigate reported kills. Rapid
response to fish kill discovery is important to establish size, severity, and
causes, but may not be possible Discovery will also be biased toward
reporting kills near developed areas where people are available to identify and
report incidents. Natural conditions, abandoned mines, and forestry activities
might also cause fish kills that could go undetected in isolated areas.
Occurrences of fish diseases and abnormalities (i.e., lesions, sores, tumors,
or eroded fins) are also difficult to associate with use support status. Cause
and effect relationships between contaminants and abnormalities are often
unknown, susceptibility to abnormalities and diseases varies among species,
and background frequencies of abnormalities are unknown. In some
instances, the majority of a catch exhibits stress, and it is apparent that
aquatic life use is impaired. Hbwever, local conditions may not be responsi-
ble for the abnormalities. Fish exhibiting stress symptoms may be exposed
to contaminants elsewhere during their migratory life cycle.
States use different approaches to associate qualitative fishery status
measures with use support assessments. For example, Colorado considers
severe and frequent fish kills an indicator of partial support, while Delaware
considers such waters to be not supporting designated use Inconsistency is
aggravated by the lack of criteria for defining “frequent and severe” fish kills.
Improved reporting requirements might provide an initial step toward
improving the quality and reproducibility of evaluated assessments.
3.2.3 Presence of Sources
The presence of land-disturbing activities, or large wastewater inputs relative
to stream flow, can indicate waters with a high probability of not fully
supporting designated uses. However, States hesitate to assess designated
uses based on this information Nonetheless, a number of States used land
use and land cover information to prepare their 319 nonpoint source
assessments. Many States relied upon State and Federal soil conservation
agents to identify areas with highly erodable conditions and agricultural
activities likely to degrade nearby waterbodies. Soil conservation agents
acquire specific knowledge of localized problems through frequent field visits
Local planning officials and foresters can also map land uses that might
degrade waters
3.2.4 Public Input
Public perceptions of waterbody value and degradation may not correspond
to water quality data but can provide valuable input to decisionmakers User
surveys, citizen complaints, and public input at meetings indicate directly to
natural resource managers user perceptions of use support, which can be
associated with water quality measurements Public input can also notify
State personnel of localized water quality problems
3-8

-------
3. OVERVIEW OF ASSESSMENT METHODS
The public can identify impaired waters with complaints of foul odors, floating
scum, water discoloration, and fish kills and diseases. A drawback to relying
upon citizen complaints for assessing use support is that the information may
be inaccurate and often the State staff are informed after the suspected
problem has dissipated. A followup investigation may not be warranted if the
complaint was improperly reported and the State’s ability to perform investi-
gations is limited by staff restrictions Kentucky has improved the reliability of
public complaints by developing a massive public education program called
Water Watch Water Watch projects train citizen volunteers to investigate
suspected spills and illegal discharges and to monitor local streams. The
program has sucessfully identified impaired waters and unpermitted
discharges (Cooke, personal communication). Citizen monitoring has also
been used effectively by other States such as Rhode Island, Illinois, and New
York Guidance highlighting these programs is currently under preparation by
the EPA Assessment and Watershed Protection Division
Public workshops played a prominent role in the development of a number of
State NPS Assessments North Carolina held 14 NPS Assessment work-
shops across the State and invited Federal, State, municipal, county, and
general public representatives to attend. The attendees identified many
impaired waterbody segments that were not monitored Oregon s
Department of Environmental Quality also held public meetings and
distributed questionnaires to citizens while developing a Clean Water
Strategy
3.3 COMBINING EVALUATED AND MONITORED INFORMATION
From discussions with State officials and examinations of §305(b) reports, it
appears that professional judgment based on evaluated data is sometimes
used to overrule monitored assessments For example, this might occur
where water column monitoring has not detected criteria exceedances, but
observations by fisheries biologists indicate severe biological impacts
guidance does not address this sue
It would be difficult.or impossible to develop a standardized ranking for
sources of evaluated information because of the variation in their reliability
Some States have special units to quickly respond to citizen complaints and
fish kill reports, elevating the reliability of such information Other States may
have access to refined land use information on a Geographic Information
System Each State may have to prioritize its own information sources, or
even integrate all available information for each waterbody individually. Some
States rely on the judgment of a single §305(b) Report Coordinator to make
the final use support judgments Other States (e g , Virginia and Tennessee)
request regional staff to develop consensus use support judgments for
waterbodies in their jurisdiction
3-9

-------
3. OVERVIEW OF ASSESSMENT METHODS
Special committees are another option developed by several States while
preparing NPS Assessments, but it is not clear whether committees can reach
a consensus about use support determinations. Idaho formed the Technical
Advisory Committee with Federal, State, and local agency personnel, as well
as representatives of interested citizen groups. Committee members were
solicited for information about NPS impacts, the sources of the information,
and the reliability of the information. All monitored and evaluated data were
entered into a data base that generated lists of sometimes contradictory
information The Technical Advisory Committee avoided ranking the
information by presenting a range of NPS impacts for each waterbody rather
than resolving conflicts in the information and consolidating the varied data
into a single NPS assessment.
Another approach is to use surveys to test public perceptions of resource
value against monitored information For example: Minnesota circulated a
questionnaire asking the respondent to rate physical condition and recre-
ational suitability of a lake, based primarily upon visual observations of algal
abundance (Figure 3-1). Perceptions of recreational value were plotted
against monitored trophic state indicators (total phosphorus concentration,
Secchi disk depth, and chlorophyll a levels) The results presented the per-
centage of respondents who perceived each level of use support at lakes with
varying trophic indicator measurements. The results could suggest ranges of
trophic indicator values that correspond to use support categories, or con-
versely associate future survey results with trophic state and use support
status. For example, fully supporting assessments could be assigned to the
range of Secchi disk depths that 80 percent of the survey respondents per-
ceive as fully supporting swimming use.
3-10

-------
3. OVERVIEW OF ASSESSMENT METHODS
A. Please circle the on. number that best describes the physicel
B. Pleas, circle lbs one number that beet describes yot opinion
conddon of the lak. water today:
on how suiteble the lake water la for recreation aid aesthetic
1. Crystal clew waler.
e*yment dsy:
2. Plot quite aystel dear, a little algae presentMslble.
1. Besutiful, could not be any nicer.
3. Definite algal green, yellow 1 or brown color apparent.
2 Very minor aesthetic problems; excellent for swimming,
4. High algal levels with limited deity and/or mlid odor
boating, and enjoyment
apparent.
3. Swimming aid aesthetic enjoyment slightly impaired
5. Severely high algae levels with one or more of the following:
becam. of algae levels.
massive floating sa ms on lake or washed upon shorn,
4. Desire to swim and level of enjoyment of the lake substan-
svong foul odor or fish kIN.
‘Jelly redeced beoatae of alga. levels (would not swim, but
bo. ig s okay).
5. Swimming and aesthetic enjoyment of the lake nearly im-
possible beca se of algae levels.
I..
S.
S
I.
4.
g
2S
i;.—-- ___ L
SourcHiskaiy, S. A., and W. W. Walker (1988)
4• SI II III III 1.
•$SYNOAI. *POiAaasci
cavsiu. c i*a
S • SONO 14.04$
3 • OUSITE 14.04$
4. N 102$ *4.041
— S • SEVENS *4.041
C* STIFF SUNVEY
2 3 4 $
sIn. siw (NOuNS)
Physical appearance ratings s. lake watsI’ quality mSEJrsmsn .
II
4.
2•
S
0 1 3 3 4 $
IECC)U SEPT14 (MUllEN)
Figure 3-1. Minnesota’s lake observer survey.
3-11
I..
S.
.5
45
25
S
S
ISTUV*4. MUOlAIlS
I
-
j/\
25
T0T*4. P44OEPIIOSIIS ( )
‘EL ’
1/
•
I 21 40 55 •0 S
e*.oscewvu.-a (P 0)
/
/
;/ /
S.
S
us
45
2I
I*cNEATION POTENTIM.
S • MU 50I *ESThEYIC P*054.(M$
3- SWI *E
= 4- ISO
— S • NO
MPC* STAFF SUSV*V
ISTENVAL MUOSAISS
ill —
SI
- ‘\\ 1
II 41 II II III I ii I
toil ). PWOSPWOSU5 (P01)
H /
C*.ONOOWYU.-A (PPU
S
25
41
S.
Recrastion potsntlal ratings v . ak. wits, quality vn..sur.m.ne.

-------
4. REFERENCES
4. REFERENCES
Cooke, Ken, Kentucky Division of Water Personal communication.
Heiskary, S.A and W W. Walker, Jr., 1988. Developing Phosphorus Criteria
for Minnesota Lakes. Lake and Reservoir Management, 4(1):1-10
Oregon Department of Environmental Quality, 1988. Water Quality
Assessment and Program Plan (Draft) Salem, Oregon.
Plaf kin, J.L, M.T. Barbour, K.D. Porter, S K Gross, and R M Hughes 1989
Rapid Bioassessment Protocols for Use in Streams and Rivers: Benthic
Macroinvertebrates and Fish Assessment and Watershed Protection
Division, Office of Water, U S. Environmental Protection Agency,
Washington, DC 20460. Publication No. EPA 444/4-89-001, May
U S EPA, 1989 Guidelines for the Preparation of the 1990 State Water
Quality Assessment (305(b) Report). Office of Water, Washington, DC,
February.
Virginia Water Control Board. 1989. 305(b) Procedures Manual for the 1990
Edition. Pages 4(1) - 4(2). Richmond, VA.
Walker, W W., 1985 Statistical Bases for Mean Chlorophyll a Criteria. In
Lake and Reseivoir Management, North American Lake Management
Society
4-1

-------
Appendix A

-------
APPENDIX A
1988 USE SUPPORT APPROACHES

-------
APPENDIX A. 1988 USE SUPPORT APPROACHES
APPENDIX A. 1988 USE SUPPORT APPROACHES
This appendix describes assessment methods used by selected States in
their 1988 305(b) reports States that reported using a straight EPA Figure 1
approach are not mentioned here
A.1 Rivers and Streams
Variations on the Figure 1 Approach
Region 1
Rhode Island is one of the few States that distinguished between chronic and
acute criteria for priority pollutants For example, for partial support, priority
pollutants could exceed chronic water quality criteria (WOO) in 10 to 33
percent of samples, but could not be present at acute concentrations
Region 6
New Mexico used a modified Figure 1 approach that treated toxics differently
from other parameters For full support, New Mexico allowed a maximum
exceedance for any parameter of <150 percent of WOC, no parameter could
exceed WOC in >10 percent of samples; and no 307(a) toxics could be
present at levels of .concern We assume this means no exceedances were
allowed for any 307(a) toxic
Region 7
Iowa referred to a methods document by the Iowa Department of Natura
Resources for information on their assessment approach To be used, a
station had to have > 75 percent of the number of samples expected from a
2-year quarterly sampling period for conventional pollutants and ammonia, for
metals, all data points from 1962 to 1967 had to be present
Missouri s modified Figure 1 approach added a physical criteria description
that had to be met regarding the type of substrate
- A-3

-------
APPENDIX A. 1988 USE SUPPORT APPROACHES
Nebraska used both a modified EPA Figure 1 approach and a water quality
index (WOl). The State did not describe how the WOl was incorporated into
use support determinations The Figure 1 type matrix used separate
frequency-of-violation criteria for toxics vs. other pollutants. For example, for
toxics, full support allowed a maximum of a single parameter value in
exceedance of WQC and no fish tissue values in excess of U.S Food and
Drug Administration (FDA) or National Academy of Sciences (NAS)/NAE
action levels.
Region 8
Utah used a modified EPA Figure 1 approach that calculated a “severity
index” for each parameter and an “overall index” for each waterbody. This
approach may be further explained in the Region 8 WQI documentation.
Water Quality Index Approach
Ten States reported using some type of WQI based on chemical data. Four
States (Idaho, Illinois, New Jersey, Wisconsin) used the Region 10 WQI, and
one (Tennessee) used the Region 8 WQI At least four States—Ohio, Maine,
North Carolina, and Kansas—relied heavily on biological standards and/or
biological indexes Several other States mentioned using biological data to
designate use support, but gave no specific details on their implementation
Several WQIs and their application in State use support assessments are
mentioned below WQIs are discussed in more detail in Appendix 0
Region 4
Florida used a WQI that incorporates six water quality categories (water
clarity, dissolved oxygen, oxygen demanding substances, bacteria, nutrients,
and I ioIogical diversity) The 6 categories comprise 13 parameters that are
arithmetically averaged to create the WOI. Biological diversity is assessed for
macroinvertebrates (collected on both natural and artificial substrates) using
the Shannon-Wiener Index and Beck’s Biotic Index
Tennessee reported using the Region 8 WOl, which allows an “unlimited
number of parameters and multiple types of criteria.”
Region 5
Illinois used the Region 10 WQI and Biological Stream Characterizations
(BSCs) in a Figure 1 type matrix The BSC is used to determine use attain-
ment in streams The stream classification system is predicated largely on
attributes of lotic fish communities. In the absence of fisheries data,
macroinvertebrate data or physical habitat descriptors may be used The
BSC method is driven primarily by an assessment of fish community structure
as represented by the Index of Biotic Integrity (IBI) for fish. The IBI method
incorporates 12 measures (metrics) of fish community structure, including
A-4

-------
APPENDIX A. 1988 USE SUPPORT APPROACHES
species composition and species richness, trophic state composition, and
abundance and càndition A Macroinvertebrate Index used in Illinois is a
modification of the method of Hilsenhoff (1982)
Ohio gave excellent documentation of the methods used to determine use
support A habitat index was used to determine whether habitat was causing
partial or nonsupport. For biological data, results for three indices (IBI and
Index of Well Being [ lwb] for fish, Invertebrate Community Index [ lCl] for
macroinvertebrates) were compared to State biological water quality
standards For example, partial attainment was reported when one or two
indices did not meet the ecoregion criteria, but did not suggest severe toxic
impact. For chemical data, a modified Figure 1 approach was used that
distinguished between acute and chronic criteria. Biological data overrode
chemical data when both were available
Region 6
Louisiana referred to a use impairment index but did not clearly explain its
application The State used a very modified Figure 1 approach that employed
only a primary determinant parameter and a secondary parameter for each
use classification. A percent exceedance was then calculated for these
parameters
A.2 Lakes
Region 4
Florida set up its own lake index by averaging the Carlson values for
chlorophyll a, Secchi depth, and the limiting nutrient (nitrogen or
phosphorus) No specifics were given on the correlation between trophic
state index (TSI) and use support.
Region 5
Minnesota used STORET to analyze 11 years’ worth of data and calculate
Carlson TSIs for (1) epilimnetic concentrations of total phosphorus and chlo-
rophyll a and (2) summer Secchi disk transparency. These TSIs were then
compared to ranges of TSI values corresponding to levels of use support
For example, TSIs 50 indicated full use support for swimming and
aesthetics; ISIs of 51 to 59 indicated threatened use support TSIs of 60 to
65 indicated partial support and TSIs of >65 indicated nonsupport The TSI
ranges were selected based on a “user perception survey” (see Chapter 3).
A-5

-------
APPENDIX A. 1988 USE SUPPORT APPROACHES
Region 7
Kansas used the Carison TSI, which was calculated based on chlorophyll a
data The degree of aquatic life use support was estimated from the TSI.
TSIs <50 were considered fully supporting TSIs from 50 to 59 were consid-
ered partially supporting, and ISIs >59 were considered nonsupporting.
Other Trophic Indexes or Other Methods
Fifteen States used other trophic indexes or other methods in use support
determination for lakes Examples of those using other indexes or other
methods are discussed below
Region 1
Massachusetts used a severity index based on measurements of six critical
parameters. The six parameters included: hypolimnetic dissolved oxygen,
Secchi disk reading, phytoplankton count (or chlorophyll a), total ammonia,
nitrate-nitrogen, total phosphorus, and aquatic macrophyton
Region 2
New York used a PC program to analyze all lakes data for selected
parameters and determine trophic status The method for determining use
support was not presented
Puerto Rico suggested using special trophic status criteria from the Pan
American Health Organization for determining trophic status, and
presumably, use support
Region 3
Delaware used three Carison Indexes based on chlorophyll a, transparency,
and total phosphorus to characterize the trophic state of lakes surveyed
Because the State’s lakes contain extensive macrophytic and filamentous
algae, total nitrogen and oxygen deficit were added to the trophic state evalu-
ation How the TSI was used to determine use support was not reported.
Region 4
Kentucky clearly stated its use support criteria for lakes The criteria were
based not on a TSI, but rather on the qualitative degree of fish kills,
hypolimnetic and epilirnnetic oxygen depletion, nuisance algal blooms or
macrophytes, taste and odor problems, treatability problems, and suspended
sediment problems
A-6

-------
APPENDIX A. 1988 USE SUPPORT APPROACHES
South Carolina appears to have used three approaches: a modified EPA
Figure 1 approach, Carlson TSI, and the National Eutrophication Survey
Index How these three approaches were used together was not reported,
nor was their relation to use support determinations
Region 5
Illinois used a Lake Impairment Index that evaluates TSI values and the sever-
ity of impairment from sediment, algae, and macrophytes. Illinois also used
biological data, field observations and best professional judgment in lake
assessments.
Ohio proposed the use of a Lake Condition Index (Ohio LCI) composed of 13
parameters selected to provide a holistic evaluation of lake conditions and to
meet 305(b) guidance These parameters included both monitored and
evaluated biological, chemical, physical, and aesthetic information The
biological parameters included the IBI for fish, nuisance growths of
macrophytes, fecal coliform bacteria contamination, primary productivity
based on chlorophyll a, and fish tissue contamination For the biological
parameters, monitoring data were available primarily for the nuisance growths
of macrophytes, fecal coliform, and primary productivity based on chlorophyll
a values The chemical parameters included nonpriority pollutants, priority
organics, priority metals, nutrients based on spring total phosphorus, sedi-
ment contamination, and acid mine drainage For the chemical parameters,
monitoring data were available primarily for nonpriority pollutants, priority
metals, total phosphorus, and acid mine drainage The physical parameter,
volume loss due to sedimentation, had been monitored for some lakes, and
the public perception of lake condition (aesthetics) when monitored was a
measure of eutrophication based on chlorophyll a The 13 parameters used
to evaluate use support incorporated best professional judgment and results
of a lake questionnaire Some of the parameters/matrices were subjective in
nature, but this represented an innovative attempt to apply many types of
information to use support assessments, while reducing the level of purely
subjective judgement
Wisconsin has attempted to use LANDSAT data to measure trophic status A
description of how this related to use support was not provided
Region 6
Oklahoma mentioned using both statistical and TSI techniques to classify
lakes as to trophic state The link between trophic state and use support was
not explained
Region 10
Idaho developed a TSI to classify a subpopulatiori of its lakes through a one-
time sampling during peak productivity using a linear-weighed sum of 11
water quality variables
A-7

-------
APPENDIX A. 1988 USE SUPPORT APPROACHES
Reference
Hilsenhoff, W.L., 1982 Using a Biotic Index to Evaluate Water Quality in
Streams. Technical Bulletin No. 132. Department of Natural Resources,
Madison, Wisconsin.
A-8

-------
Appendix B

-------
APPENDIX B
USE OF REFERENCE DISTRIBUTIONS TO ASSESS
WATER QUALITY CRITERIA VIOLATIONS

-------
APPENDIX B. USE OF REFERENCE DISTRIBUTIONS TO ASSESS
WATER QUALITY CRITERIA VIOLATIONS
APPENDIX B. USE OF REFERENCE DISTRIBUTIONS TO ASSESS
WATER QUALITY CRITERIA VIOLATIONS
B.1 Introduction
For purposes of determining whether waterbodies support their designated
beneficial uses, the U.S. Environmental Protection Agency (EPA) has
recognized this shortcoming of traditional WQC by providing Section 305(b)
guidance based on percentile statistics. In “Guidelines for the Preparation of
the 1990 State Water Quality Assessment (305(b) Report” (U.S EPA, 1989), a
designated use is considered “fully supported” if at least 90 percent of the
chemical measurements comply with the relevant WQC Designated uses are
“partially supported” if at least 75 percent of the data comply and “not
supported” if less than 75 percent of the data comply (See Figure 1 of the
305(b) guidance) A shortcoming of these recommended acceptable risks is
that they are neither site- nor chemical-specific and therefore may not
represent true natural risks of criteria exceedances.
The purpose of this discussion is to present a statistically based approach to
assessing WQC “violations” and to demonstrate highlights of the method by
means of a case study.
B.2 Proposed Approach
The proposed approach is quite simple, yet powerful, and is based on
concepts previously described in the literature (Loftis and Ward, 1981; Loftis,
Ward, and Smillie, 1983) The approach uses site-specific historical water
quality data to construct a “reference distribution” for the constituent under
consideration as shown for an idealized case (Figure B-i; all figures appear at
the end of this appendix) In the figure, water quality data for an arbitrary
constituent X have been used to estimate the probability distribution function
(PDF) for X The horizontal axis represents concentration while the vertical
axis gives the probability or relative frequency of concentration. Data used to
construct the reference distribution would be selected so as to exclude man-
induced pollution events to the extent possible For example, the data may
come from clean waters upstream of the station in question, from a clean
stream in the same ecoregion, or from the station itself with outliers thought
to be related to unnatural causes removed
B-3

-------
APPENDIX B. USE OF REFERENCE DISTRIBUTIONS TO ASSESS
WATER QUALITY CRITERIA VIOLATIONS
The WOC has been imposed on the PDF in Figure 8-1 to identify the natural
risk of violation, represented by the shaded area under the PDF to the right of
the WOC. For example, If X is total dissolved solids (TDS) with a WOC of 250
mgIL, the natural risk might be, say, 1 percent. “Violations” then detected in
new monitoring data would be compared to this natural risk to determine if,
indeed, they exceed this natural risk level. If so, then the WQC has in fact
been violated; if not, the new data can be considered as not significantly
different from the reference distribution data. A single TDS sample of, for
example, 270 mg/L could be considered as having no more than a 1 percent
chance of occurring due only to natural variation; thus, one could conclude a
pollution “event” with at least 99 percent confidence.
For the more common situation where a set of samples, say N samples, are
to be compared to the reference distribution, it could be expected that no
more than 1 percent of them would exceed the WOC due to natural variation
alone If, for example, 3 percent of them are in violation, it could be
concluded that 3 minus 1 percent, or 2 percent, represent real pollution
events and the WQC has been violated. (Note that it would not be possible,
without additional information such as knowledge of spills, upsets, etc , to
differentiate between those 2 percent that were the actual pollution events and
the 1 percent that were due to natural violations.)
B.3 Ohio Ecoregion Case Study
B.3.1 Introduction
The State of Ohio has implemented the ecoregion concept in defining and
assessing water quality conditions. Ohio comprises portions of five eco-
regions The State has established a set of reference monitoring stations in
each one, selected on the basis of being the least impacted locations in the
ecoregions. These reference stations are used primarily for biomonitoring
from which biological criteria are developed. However, chemical data are also
monitored. Ohio’s reference stations should closely approximate natural,
background conditions for the ecoregions. They also apply ideally for the
development of reference distributions with which to assess WOC attainment
for nonreference stations
B.3.2 Results
B .3.2.1 Nonconditional Reference Distributions
Figure B-2 ((a) and (b)) contains frequency histograms for dissolved oxygen
(DO) and lead concentrations in the Huron-Erie Lake Plain (HELP) ecoregion
in northwestern Ohio. These histograms were developed using only
reference station data and thus can be considered as reference distributions
from which background risks of WOC violations can be estimated Within the
HELP ecoregion there are three reference stations: 500080, 500290, and
B-4

-------
APPENDIX B. USE OF REFERENCE DISTRIBUTIONS TO ASSESS
WATER QUALITY CRITERIA VIOLATIONS
500820. Data from these reference stations were downloaded from STORET
and comprised monthly samples from 1973 through most of 1989, except for
station 500820, which also includes some 1968 data. A few months are
missing in each station’s record. In all, 683 monthly DO samples were used
to generate the DO reference distribution in Figure B-2 while 607 monthly lead
samples were used to chart the lead reference distribution.
Of the 683 samples used in the DO reference distribution of Figure B-2, 14
(2.1 percent) were less than the Ohio WQC for DO of 4 5 mgIL. Thus, the
natural risk level for DO criterion violations is estimated at 2.1 percent. Of the
607 samples used in the lead reference distribution of Figure 8-2, only 4 (0.66
percent) exceeded the U S EPA maximum contaminant level (MCL) for lead
of 50 jig/L. The natural risk of lead WQC exceedance is then estimated at
0 66 percent.
Figure B-2 ((C) and (d)) also contains frequency histograms for DO and lead
at station 500510, a nonrelerence station arbitrarily selected out of three
nonreference stations in ecoregion HELP Samples are again monthly
samples from the period 1973-1989, with some missing months A total of
173 DO samples were available of which 5 (2.8 percent) were below the 4 5
mg/L WQC. Based on the natural risk of 2.1 percent from Figure B-i, one
would expect only 3 6 violations (0 021 x 173) out of 173 observations due to
background variation alone Although confidence limits have not been
constructed here (this should be done in practice), it is very unlikely that a
single additional violation beyond the 3.6, say 4, expected violations is statisti-
cally significant and one could reasonably conclude from this analysis that
DO WQC exceedances at station 500510 are within the natural range of
variation and that no DO problems are in evidence
The lead data show similar results. Graph (d) in Figure B-2 is the frequency
histogram for lead at the nonreference station 500510. Of 150 samples, it
would be expected using the natural risk of 0.66 percent that approximately 1
sample (0.0066 x 150) was in excess of 50 ig/L, under the null hypothesis of
no difference in distributions Indeed a single sample is all that was observed
to exceed 50 ig/L and the conclusion from this analysis is that lead at station
500510 does not “violate” the WQC
B.3.2.2 Conditional Reference Distributions
While the foregoing analyses are correct, they can be criticized for ignoring
seasonal differences in water quality because the data from which the
distributions were generated were selected without regard to sampling time-
of-year Seasonal differences in many water quality variables, especially DO,
can be substantial due to seasonally related factors such as flow and temper-
ature To improve the seasonal resolution of the analyses, it was decided to
split the data sets into two seasons--summer, from May through October, and
winter, from November through April These selections were arbitrary and it
B-5

-------
APPENDIX B. USE OF REFERENCE DISTRIBUTIONS TO ASSESS
WATER QUALITY CRITERIA VIOLATIONS
may be that a different scheme, perhaps with more seasons, is more appro-
priate. These split data sets were then used to generate “conditional” fre-
quency histograms (conditional on the season) for DO and lead both at the
reference stations and at station 500510. These conditional distributions are
shown in Figure B-3.
As shown in diagram (a) in Figure B-3, the natural risk for summer DO
violations is 1 9 percent (7 of 362) Interestingly, this is a marginally lesser
risk than the 2 1 percent resulting from the nonconditional analysis (Figure
B-i) but the difference is not likely to be significant. At station 500510,
diagram (b) in Figure B-3 shows that only 1.1 percent (1 of 90) of the summer
samples were below 4.5 mg/L, suggesting that summer DO conditions at this
station are even better than those at the reference station. (Based on these
data, it would be tempting to reduce the natural risk to around 1 percent.)
For winter DO violations, the conditional analysis yields a natural risk of 2.2
percent (7 of 321), as shown in graph (C) in Figure B-3, approximately equal
to the 2.1 percent risk obtained from the nonconditional analysis. From this
risk level, it would be expected that, of the 83 winter samples at station
500510, less than 2 would be below 4 5 mg/L if only natural variation were
present. Instead, as shown in graph (d) of Figure B-3, four samples were
below the WOC, suggesting that some problem may exist at this station in the
winter. Notice that the enhanced resolution provided by the winter-
conditioned distribution has detected a possible DO problem that was not
apparent from the nonconditional analysis. One explanation might be that
relatively greater ice cover may occur, thereby inhibiting atmospheric
reaeration. (Again, it should be noted that this analysis has not considered
placing confidence limits on the natural risk levels. Such limits might have
revealed here that these four violations are not significantly different from the
natural risk.)
The conditional analyses for lead (Figure B-4) do not alter the conclusions
from the prior, nonconditional analyses that lead is not a problem at station
500510 Natural summer risks of exceeding the MCL are 1.21 percent (4 of
332) with only 1 of 79 (1 27 percent) summer lead samples at station 500510
in violation. This number, slightly higher than expected, is not likely to be
statistically significant Natural winter risks are estimated from these data at 0
percent; i e., none of the 275 samples was in exceedance. Likewise, of the 71
winter samples at station 500510, none was in exceedance, indicating that
winter lead levels are also not a problem at this station.
B.4 Discussion
B4.i Bootstrapped Reference Distributions
Ohio’s use of ecoregions with relatively unimpacted reference monitoring
stations provides an ideal application for the natural risk approach to
assessing WQC attainment. However, Ohio is one of only a few States
B-6

-------
APPENDIX B. USE OF REFERENCE DISTRIBUTIONS TO ASSESS
WATER QUALITY CRITERIA VIOLATIONS
known to use such an approach for defining attainable water quality (others
include Arkansas, Oregon, and Minnesota (Hughes and Larsen, 1988)) The
remaining States must estimate natural risks from data that may be less well
suited for this purpose. Indeed, in many cases, data from the particular
monitoring station itself for which WQC attainment is being assessed will be
the only data available from which natural risks can be identified. The
reference distribution must then be “bootstrapped” from these data.
Figure B-5 illustrates such a bootstrapped reference distribution for winter DO
at station 500510 Data used to construct the distribution were winter data
from station 500510 prior to 1986. Data collected during and after 1986 could
be used to test against the reference distribution to assess WOC violations, in
a manner analogous to what might have happened in preparation of the 1988
305(b) use attainment analysis The natural risk estimate from the
bootstrapped reference distribution is 6.5 percent (4 of 62). The 1986 and
later data include no WQC violations and would be judged as fully supporting
the DO WOC regardless of the natural risk estimate.
This particular set of data was selected because winter DO data at station
500510 were found to exhibit DO WOC violations in excess of the number
expected as defined by the true natural risk (from the reference stations) The
point of the example is to illustrate how bootstrapped distributions may over-
estimate natural risk, thereby underestimating actual problems. The
bootstrapped estimate of natural risk, 6.5 percent, exceeds the “true” natural
risk of 2.2 percent as given by the actual winter reference distribution (Figure
B-3, diagram (C)) The problems, of course, are that there is no easy way to
screen the data to eliminate “unnatural” violations from “natural” violations
and that including these unnatural violations tends to distort the reference
distribution In some cases, State staff may have personal knowledge of plant
upsets, spills, or other anthropogenic causes of poor water quality and can
screen the data accordingly, but it is expected that this would be the
exception and not the rule.
One method that might mitigate the overestimate of natural risk from
bootstrapped reference distributions would be to perform a trend analysis on
the reference distribution’s candidate data If the candidate data were to
show a trend of declining water quality, then it would be appropriate to use
only that portion of the record (if any) that did not include the trend. At a
minimum, the use of trend-tree data would ensure that the estimated risk,
albeit perhaps larger than natural risk, would serve as a status quo, or
baseline, that could not be exceeded in future years. That is, the baseline risk
would preclude further degradation even it it did not define the waterbodys
potential water quality
B-7

-------
APPENDIX B. USE OF REFERENCE DISTRIBUTIONS TO ASSESS
WATER QUALITY CRITERIA VIOLATIONS
B.4.2 Confidence Limits on the Reference Distribution
Each reference distribution presented here was estimated from a particular
data set. Because that data set is merely a sample of the entire population of
constituent concentrations that completely define the reference distribution,
the derived distribution is only an estimate of the true, but unknown, reference
distribution. It is therefore desirable to explicitly acknowledge this source of
error by placing confidence limits on the estimated distributions. These limits
would be presented as upper and lower confidence limits and would be used
to define when the number of observed WQC exceedances is significantly
different from the expected number of exceedances so as to constitute a true
violation Both parametric and nonparametric methods for determining
confidence limits are discussed by Loftis and Ward (1981).
B.4.3 How to Avoid Zero Risk
Occasionally, an empirically derived reference distribution, such as those
presented here, will indicate that none of the data exceeded the WQC and,
consequently, that the empirical distribution shows no natural risk Such was
the case for the winter lead distribution depicted in diagram (c) of Figure B-4
It is very unlikely that there is zero risk of winter lead exceeding 50 j g/L due to
natural variation; it is more likely that the data set simply was not large
enough to contain such exceedances.
One method to avoid this problem is to fit a suitable, theoretical probability
model (e g., normal, lognormal) to the data. The theoretical model would
then be used to estimate the natural risk by extending the distribution’s tail
out beyond the WOC. It should be noted that this method has the
disadvantage of uncertainty with regard to selection of the appropriate theo-
retical model, especially for small data sets In addition, State expertise and
resources may discount the practicality of this approach.
Bayesian statistics offers another approach to the zero-risk problem by
incorporating into the risk estimator the subjective notion that just because
something did not happen in the past does not mean that it will not happen in
the future Box and Tiao (1973) discuss Bayesian estimators.
B.4.4 Conditioning for Other Factors
The use of reference distributions conditioned on time of year has been
presented. As discussed by Loftis, Ward, and Smillie (1983), this same tech-
nique may be used to account for other factors that might affect water quality
such as flow For example, it might be that low summer DO levels are more
strongly correlated to low flows than to time of year per Se. Accordingly, it
would be more appropriate to condition the DO reference distribution on flow
than on time of year
B-8

-------
APPENDIX B. USE OF REFERENCE DISTRIBUTIONS TO ASSESS
WATER QUALITY CRITERIA VIOLATIONS
B.5 References
Box, G.E.P and G.C Tiao, 1973 Bayesian Inference in Statistical Analysis,
Addison-Wesley Publishing Co, Reading, MA
Hughes, R M and D.P Larsen, 1988. Ecoregions: An Approach to Surface
Water Protection. Journal WPCF, 60(4), April
Loftis, J C and R C Ward, 1981. Evaluating Stream Standard Violations
Using a Water Quality Data Base. Water Resources Bulletin, 17(6),
December
Loftis, J.C., R.C. Ward, and G.M. Smillie, 1983 Statistical Models for Water
Quality Regulation Journal WPCF, 55(8), August
U S. EPA, 1989 Guidelines for the Preparation of the 1990 State Water
Quality Assessment (305(b) Report). Office of Water, Washington, DC,
February
B-9

-------
APPENDIX B. USE OF REFERENCE DISTRIBUTiONS TO ASSESS
WATER QUAUTY CRITERIA VIOLATIONS
Reference Distribution
Figure B.1. Reference piobability distribution function (PDF) for pollutant X,
with the Indicated water quality criteria (WQC).
I L ’
WQC
X (Concentration)
B-i 1

-------
APPENDIX B.
USE OF REFERENCE DISTRIBUTIONS TO ASSESS
WATER QUALITY CRITERIA VIOLATIONS
(b)
DO Reference Distribution
N .613
No. WQC a 14
ii, 11 1
160
170
160
150
1
130
120
110
90
60
70
60
50
30
20
I0
0
17.5 20
DO C auVM (mgIL)
(c) Cd)
DO Distribution, Station 500510
0 2.5 5 7.5 10 12.5 15 17.5 20
DO Cu .....iuu . (mgIL)
IS
S
Figure B-2. Nonconditional distributIons for Huron.ErIe Lake Plain ecoregion in Ohio.
(a)
Lead Reference Distribution
50
40
30’
20’
10
0 ’
N .607
No. WQC .4
F’
ri
0
r IF IF IF
25 5 7.5 tO 12.5 15
L.sdC .n1i *(ug/L)
85 90 95
N • 173
No. c WQC —5
Lead Distribution, Sta 500510
20 ’
19
1$
17
16
I,
14
13
12
ii
10
S
7
6
5
4.
3.
2’
0
45
40
35
30
25
20
nr
I
I
10
0
05 1015202530354045505560657075808 59095
L..d C .uIrWo.(uWL
B- 12

-------
APPENDIX B. USE OF REFERENCE DISTRIBUTiONS TO ASSESS
WATER QUAUTY CRITERIA VIOLATIONS
(a) (b)
DO Reference Distribution DO Distribution, Station 500510
35’ r i 1$
17
fl c i N.362 16’
30 II No.WQC.7 Is’ No.
-------
APPENDIX B. USE OF REFERENCE DISTRIBUTiONS TO ASSESS
WATER OUAUTY CRITERIA VIOLATIONS
(a) (b)
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
0
(c) (d)
____ Lead Distribution, Station 500510
24
n
20
Is
16
14
12
10
8
6
4
2
0
Figure B-4. Seasonal lead dIstributIons for the Huron-Erie Lake Plain ecoreglon In Ohio.
Lead Reference Distribution
19
Lead Distribution, Station 500510
100
90
60
70
60
50
40
30
20
10
0
30 35 40 45 50 55 60 65 70 75 80 85 90 95

LaadC . u.. eiiL o (vejL)
Lead Reference Distribution
W ir
70
50
40
30
20
10
0
05 1015202530354045505560657075 50*59 095
L4adCoocc ios (NIL)
La Cc cn*r.floa (VIIL)
B- 14

-------
Appendix C

-------
APPENDIX C
TREND ASSESSMENTS AND THREATENED USES

-------
APPENDIX C. TREND ASSESSMENTS AND THREATENED USES
APPENDIX C. TREND ASSESSMENTS AND THREATENED USES
The EPA’s 1990 305(b) guidance document defines a ‘threatened”
waterbody as one that is currently fully supportive of designated uses but
poses the threat of not fully supporting those uses in the future due to some
adverse temporal trend.
Detection of time trends in water quality data can be accomplished either by
statistical methods such as the Seasonal Kendall Tau Test, or by essentially
subjective methods such as visual inspection of a time series plot When
adverse trends are clear and unambiguous, such trends can be detected
satisfactorily by either method and the waterbody can be judged threatened
with little possibility of error However, when adverse trends are not so clear,
the possibility of decision errors becomes proportionately greater For these
situations, the subjective methods suffer the considerable disadvantage of
being unable to quantify these errors
With either method, two types of decision errors are possible A “Type I”
error occurs when a waterbody is judged threatened that in fact is not (no
trend is present) Conversely, a “Type II ” error occurs when a waterbody that
truly does have an adverse trend is judged nonthreatened The implication of
either type of error is a misallocation of monitoring or control resources In
the former case, a falsely identified threatened waterbody may unknowingly
shift resources away from higher priority waterbodies In the latter case, a
resource shift should be made that will not occur because of the failure to
identify the threatened use
It is an unfortunate fact that the possibility of these two errors can never be
completely eliminated, regardless of how many data are available or how
carefully they were selected. Moreover, these two errors are not independent,
indeed, there is a tradeoff between them in the sense that, for a given amount
of data, a monitoring program design cannot be chosen that will reduce the
probability of both errors simultaneously A reduction in Type I error occurs at
the expense of an increase in Type II error and vice versa Only by collecting
more data can both be reduced at the same time
Despite the fact that water quality managers must learn to live with the
consequences of these two decision errors, statistical theory does at least
otter a means by which they can be quantified Type I error is quantified by
the “p” value, or “significance ’ level, used in conventional hypothesis tests
C-3

-------
APPENDIX C. TREND ASSESSMENTS AND THREATENED USES
Type I error is a single number (probability) corresponding to the fact that
there is a single null hypothesis (no trend) that is either accepted or not. Type
II error is not a single number because, unlike the null hypothesis, an infinite
number of alternative hypotheses exist and a Type II error is associated with
each of them Consequently, Type II error is quantified by constructing
curves that give Type II error probabilities as a function of specific alternative
hypotheses These curves are commonly called “power” or “operating
characteristics” curves
For data from an established monitoring program design, the probability of
each error type can be determined, but it cannot be changed Significantly,
for new monitoring programs, the relative magnitude of each error can be
predetermined by its design For example, a policy decision might be that the
detection of truly threatened waters with high probability outweighs the cost of
judging some waters to be threatened when they are not. In this case, a
design would be selected that would reduce Type II at the expense of Type I
error.
In summary, statistical methods of trend analysis to identify threatened uses
offer significant advantages over subjective methods due to their ability to
quantify decision errors, not because they can somehow avoid them This
advantage is particularly important when trends are ambiguous due to noisy
and/or sparse data An important issue for analysis of existing data, and for
the design of new monitoring programs, is the relative importance of the two
types of decision errors
C-4

-------
Appendix D

-------
APPENDIX D
WATER QUALITY INDICES

-------
APPENDIX D. WATER QUALITY INDICES
I ’) L o .&L1 4o aA )aki r
4 ( MAcu
d iiip -
APPENDIX D. WATER QUALITY INDICES
Selection and Development
Indices can be categorized as general or specialized. General indices use-the
same equation for different uses but have differing ranges for use support
classification (example in Figure 0-1) Specific indices are tailor-made to best
represent the water use that is being indexed Both the parameters and the
weights for the indices are specialized for the use. This is beneficial for
comparing waterbodies with the same use but does not allow comparison
between two different uses
Parameter Choice
Parameter choice is one of the most important steps in developing an index
because the index will only reflect the water quality and use support status of
the waterbody if the parameters are actually important to the intended use.
For example, North Carolina has a use designation for Trout Waters. If
parameters were chosen which placed a heavy emphasis on nutrients and
pesticides and little or no emphasis on temperature and dissolved oxygen
content, then the results of the index could, potentially, incorrectly indicate
that the stream was supporting its use as a trout fishery For the general
water quality indices (WOls), conventional parameters such as turbidity,
temperature, pH, dissolved oxygen, chemical oxygen demand (COD), and
nutrients are included to try to give a general idea as to the quality of water in
all of the uses. Specific use indices often contain these parameters and, in
addition, also contain specialized parameters which best indicate the status of
the water for the given use
Another important consideration for parameter choice is data availability
Often, because of budget considerations or sampling techniques, data
availability is highly variable. A third guideline for parameter choice is the
existence of water quality standards or criteria for the parameter Indices are
designed to be objective representations of water quality, and as a result, the
frame of reference provided by standards and criteria is quite important
One technique which has been employed in the past to increase objectivity is
the use of a DELPHI questionnaire process The true DELPHI method:
D-3

-------
.. ,WATERUSES
Level
of
Poll”-
lion
(100
Best)
I Public
I Water
I Supply
RecreatIon
Fish
Shellfish
Agricultural
Industrial
Purification
Not
Necessary
Minor
Purification
Required
Necessary
Treatment
Becoming
More
Extensive
Doubtful
N
0
I
A
C
C
E
P
I
A
B
L
E
Acceptable
for
All
Water
Spoils
Becoming
Polluted
Still
Acc ptabIe
Baôteria
Count
Doubtful
for
Water
Cothact
Only
Boating,
No Water
Contact
Obvious
Pollution
Appearing
Obvious
Pollution
Not
Acceptable
Acceptable
for
All
Fish
Marginal
for
Sensitive
Fish
Doubtful
• for
Sensitive
Fish
Hardy
Fish
Only
Coarse
Fish
Only
N
0
T
A
C
C
E
P
T
A
B
L
E
Acceptable
for
AU
Shellfish
Marginal
for
Sensitive
Shellfish
Doubtful
for
Sensitive
Shellfish
Hardy
Shellfish
Only
Coarse
Shellfish
Only
N
0
I
A
C
C
E
P
T
A
B
L
E
Purification
Not
Necessary
Minor
Purification
for
Crops
Requinng
High
Quality
Water
No
Treatment
Necessary
for Most
Crops
Extensive
Treatment
for Most
Crops
Use Only
for Very
Hardy Crops
Not
Acceptable
Purification
Not
Necessary
Minor
Purification
for
Industries
Requiring
High
Quality
Water
No
Treatment
Necessary
for Normal
Industry
Extensive
Treatment
for Most
Industry
Rough
Industrial
Use Only
Not
Acceptable
Source: Dinius, 1987
FigUre 0-1. General rating scale for water quality.

-------
APPENDIX D. WATER QUALITY INDICES
is an effective method of integrating the opinions of experts without
the disadvantageous effects of the committee process .A series of
sequential questionnaires, separated by additional information and
response feedback from earlier questionnaires, replaces
committee debate. Important elements of the technique are ano-
nymity of response, statistical analysis of responses, and increas-
ingly refined feedback The technique is designed to control
negative committee characteristics such as pressure from
dominant individuals and irrelevant and often lengthy
discussions.. Each round [ of questionnairesj should bring the
panel member’s views on the questions closer together so that by
the final round they are in approximate accord (Dinius, 1987,
p 834)
Many of the studies involved in creating indices use a “modified DELPHI”;
these are shortened versions of the principle In most modified DELPHI
questionnaires there are only one or two rounds and the reaching of a
consensus is not as crucial. This method is often shortened in the interest of
time For instance, the Tennessee Valley Authority (TVA) DELPHI procedure
took 7 months to run three rounds Experience has shown that if there are
too many rounds the experts will begin to drop out, leaving the index
developer with no statistically significant results
If the parameters are to be developed into a large-scale, widely usable index,
they must be widely monitored on a regular basis Problems occur, on a
State level, when budget cuts require the discontinuation of monitoring at
some stations and parameters from the regularly monitored schedule.
Indices can be extremely sensitive to missing data and as a result, in order to
get defensible results, typically no more than one or two parameters may be
missing at a given time Index developers have approached this problem
from a variety of standpoints. The National Sanitation Foundation (NSF)
index requires a perfect data set (no missing data); the index developed by
House allows for up to two of the minor parameters but none of the major
parameters to be missing; and Region 10 allows for wholesale substitution of
parameters to ensure a complete data set
Complications also evolve from seasonally monitored data Parameters such
as fecal coliform and chlorophyll a are important indicators of microbiological
quality and trophic state Both of these are commonly only monitored during
the summer From a monitoring standpoint this schedule is fine, but from an
indexing standpoint, it impedes their use in annual or longer-term
measurements One solution to this is the development of indices which run
only for the summer and, therefore, include these parameters
There are several methods of dealing with missing data for indices (Landwehr,
1989, personal communication). The first is to skip over that sampling date
for that station, as NSF has The second is to try to calculate a value based
on historical values This is impractical for large-scale use from two
D-5

-------
APPENDIX D. WATER QUALITY INDICES
standpoints: (1) for parameters that are monitored seasonally, there are no
good historical data for the rest of the year; and (2) the effort involved in
reentering each missing parameter for each date for each station for an entire
State may be prohibitive A third approach, use of regression techniques, has
the same drawbacks as the historical extrapolation. Once again, the time
required to complete the data set may be greater than that of perusing it and
there is the danger of actually creating, through regression, a large
percentage of the entire data set. This could occur, for example, if many of
the parameters are monitored only on a quarterly basis.
Parameter Transformation
Parameters have different values and ranges based on the units that they are
measured in and the natural levels found in water. These differences require
indices to transform the raw data before aggregating it into a single index
value While each index uses a slightly different method of standardizing its
results, most use some type of rating curve transformation The methods of
developing rating curves differ based on the objective of the index. For
example, “standard based” indices incorporating toxicants use the standard
or final chronic value (FCV) and assign a rating between 50 and 100 or 5 and
10, based on the rating scale (Armstrong, 1989; House, 1989, personal
communication). The final acute value (FAV) is also assigned a rating value
ranging from 25 (or 2.5) to 0 to demonstrate the low end of the rating curve.
A line is formed using nonparametric functions which connects the points and
results in a simple transformation from a data value to a standardized rating.
In most cases, rating curves do not differ between general and specific use
indices developed by the same person or agency
There may be a great deal of professional judgment in the placement of
coordinates (e g., the FCV and FAV) on the rating curve The effort is made to
define a range of water quality from “clearly impacted” (FAV) to ‘clearly
unimpacted,” e.g., the detection limit An effort is also made not to lose the
detail of the water quality which became evident through sampling. In most
cases the transformation is the most complex mathematical step to the index,
but once established, it is also the most stable aspect of the index An
example of a rating curve for a new North Carolina index is shown later in this
appendix (Figure D-3).
Weightings
All indices have some form of weighting factors on the parameters Those
indices which claim to be unweighted actually assign an equal weight of one
to all of the parameters. This “unweighted” system is not used widely in the
indices currently in use because it is unreasonable to say that, for example,
color is as important as temperature to the quality of water for cold water
fishery uses. A second and widely used method is unequal weights of all the
parameters based on some expert’s or group of experts’ opinion of the
D-6

-------
APPENDIX D. WATER QUALITY INDICES
importance of the parameter for the given water use. In most instances
weights are determined using either a DELPHI survey or one-time survey In
either case, a panel of water quality experts is chosen and questioned about
the importance of each parameter for indicating water quality. The results are
then compiled using a method as simple as an arithmetic mean The
weighting prbcess, too, is highly subjective but the survey approach is
designed to reduce the subjectivity as much as possible.
Aggregation Methods
General and specialized indices have been developed using both arithmetic
and multiplicative aggregation. The arithmetic mean formula is:
n
E q.w.
1=1
where:
n = number of parameters,
= the given parameter,
q 1 = the rating value for the ith parameter, and
w 1 = weighting factor for the ith parameter.
This formula is widely used both because it is easy to understand and
because all weighted values are equally important.
The multiplicative mean takes the form of:
n
i q.w.
1=1 ‘
This formula has the characteristic of being more sensitive to extreme values
for a given parameter.
Examples of Indices
The majority of indices have been developed, or at least verified, for streams
and rivers The scope of indices and the number available make the task of
reviewing them quite a challenge, so we have reviewed only indices for which
we have a clear, technical explanation Ground water indices are not widely
published at this point The general applicability of indices to estuaries and
wetlands is currently unstudied. One inventory, six river quality indices, and
0-7

-------
APPENDIX 0. WATER QUALITY INDICES
two lake indices are evaluated below. A review of trophic-state indices is not
included.
The Maryland Inventory
The State of Maryland uses a modified Figure 1 inventory system to classify
its streams and lakes for the Section 305(b) reports Their method has a high
degree of accuracy and fits the Clean Water Act’s requirements of status
toward the fishable-swimmable goals. The State uses a combination of
monitored water quality data and evaluations where chemical data are not
available. The inventory relies quite heavily on the State biologists for incident
reports of fish kills and blooms as well as for judgment regarding general
quality of all of the streams and rivers. Data are compiled onto worksheets
(Figure 0-2) and the severity of water quality impact is assessed based on the
percentage of exceedance (for chemical data) and professional judgment
(Maryland Water Quality Inventory 1985-1987, p. B5). Use support decisions
are made based on a set of “rules of assessment.”
This method is thorough and clear in its treatment of conventional pollutants,
sources, and (most notably) evaluated information It can also be used, with
the same accuracy, for different waterbody types Potential disadvantages
include labor requirements and reproducibility (a second expert examining
the same data may come up with a different view of use support) This
approach may be a useful alternative to an index for those who are very
opposed to indices.
River Water Quality Indices
Dinius’ Water Quality Index
Dinius has refined the Social Accounting System originally published in 1972
(Ott, 1978a, p 219), resulting in a scientifically based water quality index The
objective of this index is to provide a quantified method of determining the
extent of increasing water pollution in order to estimate the cost of pollution
prevention and control The index uses a geometric mean (multiplicative
mean) to aggregate data and uses a full four-round DELPHI questionnaire to
determine both the standards and the weights for the index It was developed
for 6 water uses: public water supply, recreation, fish, shellfish, agriculture,
and industry; and for 12 parameters. dissolved oxygen (DO), BOD-5, total
colilorm, fecal coliform, alkalinity, hardness, chloride, specific conductance,
pH, nitrate, temperature, and color (Dinius, 1987, p 840) Dinius formulated
rating curves and made data transformations based on the DELPHI panel’s
responses
This index provides a very good template for developing an index The
process is well documented, with the exception of the rating curves A
D-8

-------
Monitored: ___________ Evaluated: ___________ Total: ___________
Monitored: ___________ Evaluated: ___________ Total: ___________
Monitored: ___________ Evaluated; ___________ Total: ___________
Monitored: ___________ Evaluated: ___________ Total: ___________
Monitored: ___________ Evaluated: ___________ Total: ___________
Monitored: ___________ Evaluated: ___________ Total: ___________
River (mu)
Estuary (ml)
Lake (ac)
Coast Cmi)
Wetland (ac)
Groundwater ____________ ____________ ____________
LAND USE•
Urban; Agricultural: Forest; _______ Wetland: ________ Mines: _______
MONITORED
CSO/Stormwater WB Type Compliance MGD Pretreat
1985-1987 Maryland Water Quality Inventory
Water Quality Assessment Summary
Basin: _______________________ Name:
Class: __________________________ Priority:
1985-1 987 Maryland Water Quality Inventory
Water Quality Assessment Summary
I. I
Station
I . : • .. :.:
State WO Standards
WB Type Temp DO pH Turb Bact
Subjective
N P Chia
SIP
WB Type
Compliance
MGD
Pretreat
industry
WB Type
Compliance
MGD
Pretreat
½. . ..2.4...a. ,...5 .,,Y
Basin: Name:
EVALUATEDI
1
Swimming Ban
Area WB Type Source Cause
Fish Kill
Area WB Type Source Cause
SHELLFISH CLOSURES/CONDITIONAL ACRES 1
Permanent
1
Area WB Type Source Cause
Conditional
Area WB Type Source Cause
FIsh AdvIsory
Area WB Type Source Cause
Algae Bloom
Area WB Type Source Cause
Agricultural Water Pollution Complaints
Area WB Type Source Cause
Mine Impact
Area WB Type Source Cause
Figure D-2. Example of watershed worksheet for water quality assessment.

-------
APPENDIX D. WATER QUALITY INDtCES
disadvantage in using the index for use support determinations is that the
rating curves are not based on the standards of any State or on EPA criteria.
Another potential problem is the choice of parameters (e.g., toxicants are not
included for drinking water). Unfortunately, parameters may not be readily
substituted into and out of this index due to the lengthy process needed to
weight the parameters
House’s Water Quality Index
House has developed several indices to be used by the United Kingdom
Public Water company as both an enforcement and an assessment tool for
the country’s river waters. House has proposed both multiplicative and
arithmetic indices but has found, through verification, that the arithmetic index
best represents the quality of the streams based on the national standards.
This index is based not only on Great Britain’s standards but also on the
European Economic Community’s (EEC’s) and EPA’s standards and criteria.
The House index will be implemented in Great Britain starting in early 1990,
and as a result, it has been streamlined a great deal. The Water company
has developed software that automatically computes the water quality index
value for a given sampling date (House, 1989, personal communication).
The “index” is actually a family of indices which House reports may be used
either separately or combined. The first index is the General Water Quality
Index, which has nine parameters. These are: DO, BOD-5, NH 3 -N, total
coliform, suspended solids, pH, nitrates, chlorides, and temperature. In the
general index each parameter is assigned the same weighting for all
classifications. The general index is an arithmetic mean index using rating
curves developed from the available standards and criteria and weights based
on a modified DELPHI questionnaire House deals with missing data by
allowing the program to run if all of the first five parameters (listed above) are
present and if no more than one of the others is missing (House, 1989,
personal communication). In addition to the General Water Quality Index, she
has also developed three other indices which are based on the specific
parameters important to human drinking water, aquatic toxicity, and drinking
water standards Because in each of these indices the exceedance of a
standard is unacceptable, a geometric mean aggregation was used to
emphasize any exceedance problems. Also, the rating curves were based on
a 0 to 10 range instead of the 10 to 100 range of the general index. Another
major difference is that the specific indices were not given any weighting
factors because House felt that if there was an exceedance in one parameter
it was equally toxic to humans and aquatic organisms
The indices developed by House suit many of the needs for a national index
because they are standard-based and designed to be applied to a relatively
large range of geographic areas and classifications. Another advantage to
the House index is that it is very well documented. House did not use
complex math or a full DELPHI questionnaire and as a result, modifications to
D- 10

-------
APPENDIX D. WATER QUALITY INDICES
this index are the easiest and most legitimate of the indices reviewed Other
advantages to this index include, the rating curves are both easy to construct
and adjust, and the index has been fully statistically tested and verified in
Great Britain Disadvantages include a bias towards drinking water
parameters because Great Britain does not have a designated class for cold
water fisheries Along the same line, many States do not have standards for
parameters such as chlorides in any classification other than drinking water
Also, the rating curves are based primarily on Great Britain’s standards and
as a result would need modification before being adopted in the United
States House does not use the “fully supported,” “partially supported,” and
‘not supported” categories to rate waters within a classification; however, it
would not be difficult to add these delineations into the index results and in
this way address the use support question.
Re9ion 8 Water Quality Index
The Region 8 Water Quality Index is a chemical index designed to express
water quality in terms of use impairment The first step in using the index is to
determine the frequency and amount of standard exceedance for individual
parameters A severity index (SI) is calculated from the exceedance data for
each applicable use Station SI values within a reach are averaged to
determine a reach SI A use impairment value (UIV) is then calculated for
each reach as the sum of the SI values for different uses. EPA calculates UIV
values The States then augment these values with subjective, evaluated
information for 305(b) reports
Region 10 Water Quality Index
Region 10 has developed a comprehensive index based on a complex
arithmetic aggregation This index was designed to be highly adaptable for
each of the States in Region 10 and as a result it has addressed many of the
problems faced by index developers The index is based on the following
categories of parameters: temperature, oxygen, pH, bacteria, trophic state,
aesthetic state, solids, organic toxicity, and metal toxicity. This index is
applied to all streams, regardless of classification, and ratings of blue, yellow,
and red are produced The ratings may be interpreted for uses as fully,
partially, and nonsupporting. The report on the index states that
the WOl compares measured water quality with recommended
“fishable/swimmable” Federal criteria by default These criteria are
a synthesis of national criteria, State standards, information in the
technical literature, and professional judgment... .An overall WOl
number is calculated for every selected water quality sampling
station with sufficient data The overall WQI number (calculated
monthly) is an aggregation of subindices for 10 pollution
categories which are weighted by the relative severity of criteria
exceedances for each group (Peterson and Bogue, 1989)
D-1 1

-------
APPENDIX D. WATER QUALITY INDICES
Each of the parameters is weighted by a value assigned to each use category
of water The subindex value, Ziki is calculated by (Ott, 1978b, p. 293):
Z. _q. *F. *W ‘ (1)
where:
0 j,k = quality rating for station j and category k
= frequency of violation of recommended limit
th
Wk = weight for the k category.
= l/2 E [ zJ+lk ZJ k] [ d + 1 _d ] (2)
where:
= water quality subindex for category k along the reach
z. = water quality value of the kttl category at the th station in the
reach
d. = river mile distance of the station (d < d. +
n = number of stations.
These values are then aggregated using a summation for each of the
categories for the index value (I) for the reach
k
1= E I
1=1
This index addresses the problem of missing values by calculating
replacement values for a group of parameters, not just a single parameter.
For example, if nitrite were the parameter required for the index, but it was not
sampled, then the STORET retrieval program would look for any of five other
nitrogen parameters; if one of these were present, it would be scaled and
substituted Rating (‘severity”) curves are employed to adjust all of the
parameters into the same scale; documentation of these curves has not been
received to date
This index has many attributes which are not seen in the other indices. The
first is that it is designed to use STORET data and as a result is also designed
D- 12

-------
APPENDIX 0. WATER QUALITY INDICES
to deal with factors such as changes in monitoring schedules. Second, it has
been used in the United States for at least 10 years and as a result many of
the problems of the newer, untested indices have been worked out A third
major advantage is the flexibility of the program to be adapted to different
States’ needs (Bogue, 1989, personal communication). Several questions
and potential disadvantages remain. The documentation that has been
reviewed includes very little technical information as to the aggregation
technique, the rating curves, and the origin of the weights. The use of linear
interpolation to fill gaps in the data that cannot be filled by direct substitution
may be misleading in some instances (e.g., seasonal data sets). Another
concern is that metals data cannot fall into the highest category, regardless of
lab results. In many cases the cutoff for the middle (yellow) category is also
the detection limit, which means that index values may indicate that a support
is threatened because of the inability to measure lower levels. A final concern
is the large-scale substitution which takes place in order to maintain a
complete data set. In the case of nitrogen, for example, ammonia is more
toxic (at the same concentration) than nitrate, yet the index has the ability to
calculate an index value based on either compound. Where parametric
coverage is not uniform, assurances need to be made that this degree of
substitution is not leading to the comparison of “apples and oranges” in the
index results.
National Sanitation Foundation Water Quality Index
The NSF index is one of the older indices (1973) and as a result has had
many modifications made to it The index is based on nine parameters: DO,
fecal coliform, pH, BOD-5, N03, P04, temperature (degrees centigrade from
the equilibrium), turbidity, and total solids. There are three use categories
which this index considers: public water supply, fish and wildlife propagation,
and contact recreation (Brown and McClelland, 1974, p 3) The parameter
selection and the weightings of the parameters were both accomplished
using a full DELPHI questionnaire and five rounds of questioning As part of
the DELPHI questionnaire the experts were asked to create rating curves with
levels of water quality from 0 to 100 These responses were then averaged
together to form the curves. The early papers published on the index used an
arithmetic aggregation; however, a later report on the index, “WQI Enhancing
Appreciation of Quality of Improvement” (McClelland et al., 1973) reports that
the aggregation process is a multiplicative one. The change occurred as a
result of a Ph 0 dissertation presented by Landwehr (1974) which tested both
the arithmetic and multiplicative aggregation processes on the NSF index and
found better correlation between the multiplicative and experts’ opinions of
the water quality at given sites One additional characteristic which this index
has is the ability to work within data availability constraints. This excerpt from
the 1976 report (McClelland, Brown, and Deininger) has been included
because it is true for all of the American indices which use STORET:
D- 13

-------
APPENDIX D. WATER QUALITY INDICES
Naturally it is disappointing to those who developed and maintain
STORET that so little of the water quality data it contains lends
itself to retrospective determination of index values by means of
WQI or any other proposed formulation. This is not critical of
STORET, but of the inconsistency and inadequacy of data (for the
index development purposes) entered by agencies supplying the
data. Data deficiencies can frequently be handled by applying
estimation techniques. They can be entirely overcome in the future
by adopting a few simple but important policy decisions relating to
parameter selection and use of STORET (McClelland, Brown, and
Deininger, 1976, p. 24).
The index could be adjusted to a use support index simply by altering their
current scale of “very bad’ to “excellent” into, for example, use supporting =
excellent and good; partially supporting = medium; and nonsupporting =
bad and very bad. This is advantageous from two standpoints: first, these
designations were determined as part of the questionnaires and as a result do
not represent arbitrary scorings; and second, the scaling process is helpful
when coupled with EPA’s support scorings to lend meaning to “use
supporting’ for the layperson. This index was developed using the opinions
of experts from all over the United States. As a result of the experts’
distribution, the parameters chosen also are useful indicators throughout the
United States. This is the only index which has published thorough accounts
of its validation process and as a result it has demonstrated itself as a sound
index with few reasoning errors and a strong user-support network. The
index is in a finished, verified state and as a result, would require little or no
alteration for future 305(b) use.
Proposed North Carolina Water Quality Index
The proposed North Carolina index is based on House’s general index, with
modifications to better suit the State’s needs The index being developed will
include 12 parameters for both conventional water quality indicators and
toxics This index is a monitor of the water column only and as a result does
not include parameters that rapidly adsorb to particulates (e g., pesticides)
Similarly, sediment and fish tissue data will not be incorporated For fresh
waters this index will use 12 parameters: DO, total solids, temperature,
turbidity, arsenic, cadmium, total chromium, copper, lead, mercury, nickel,
and zinc. These parameters will be applied to all waters except Nutrient
Sensitive Waters since nutrients and chlorophyll have not been included. Salt
waters will have the same parameters, with the exception of the toxic metals
because of analytical difficulties in measuring water column metals in saline
waters
The index, like many of the others, used a modified DELPHI questionnaire to
determine both the parameters which are included and the weights of each
parameter Rating curves are being developed using the lowest
concentration measured in North Carolina (or the detection limit), the State
0-14

-------
APPENDIX D. WATER QUALITY INDICES
standard for the parameter, the EPA FAV concentration and the concentration
which falls at 95 percent on a distribution curve for concentrations measured
in North Carolina This method of determining the high score is designed to
choose the true highest recorded score and not anomalies resulting from lab
error an example showing development of a rating curve for zinc is given in
Figure D-3
The greatest deviation from House’s index is in the weightings In the North
Carolina index the weights are based on the parameter and on the the classi-
fication This modification was made to allow the, importance of individual
parameters to vary based on the projected use of the water. On the other
hand, to allow the comparison of unlike use classifications, a constant set of
parameters was used in each classification
Once data have been retrieved from STORET and transformed through the
rating curves, aggregation takes place. Like House’s index, this index uses
the Solway aggregation method to total the parameters’ values for each date
This formula is:
n
WQI = 1/2 ( E q 1 w 1 ) 2
i =1
For a given station there would be anywhere from 3 to 13 of these values for a
2-year period, and in order to aggregate them to a single index value for that
station for that time period, a distribution curve is created. The final index
value for that station is the 5-percent value on the distribution curve This
aggregation method is chosen based on House’s work Many authors have
noted that the most obvious approach of taking the mean of the aggregates
overestimates the quality of water when using an arithmetic approach. Taking
the 5-percent value eliminates this overestimation.
North Carolina’s index is still in the early stages of development and has not
undergone the degree of verification of the other indices discussed The
State does plan to use the index, in its unfinished state, for the 1990 Section
305(b) report This index has several advantages over older indices: the
rating curves and standards are based on the most recent toxicological data,
the index is using the experience and criticisms of the older indices to avoid
many of the problems of index development, and it is being written in SAS
within the STORET environment so that there is no need to actually handle
the raw data at all This is a tremendous time saver and allows less
technically oriented staff members to work with the index The index has the
ability to have 7 more parameters added onto it, with the same statistical
significance as the first 12, when consistently monitored data become avail-
able This index is designed to address the use support classifications direct-
ly as a result of being developed for the 305(b) report and therefore is ideal for
the goals of the EPA To a large extent this is mathematically the simplest
0-15

-------
Figure D-3. Development of a rating curve for Zinc.
Step 1: STORET retrieval to find distribution of concer
Step 2: Isolate the 5% and 95% concentrations.
Step 3: Set up table:
Rating USE Water Supoly
100 25 (Detection Limit)
75 50 (Aquatic Life Standard)
25 130 (FAV)
0 240(95% Recorded in North Carolina)
Step 4: Graph and find equation for curve.
Step 5: Test on actual data to ensure that concentrations fall within the 100-0 range.
Step 6: Apply the rating to the index to find a subindex value.
Source: Proposed North Carolina WOl
Rating Curve for Zinc
100
90
80
70
60
C
40
30
20
10
0
Zn (j&g/L)

-------
APPENDIX D. WATER QUALITY INDICES
index discussed and as a result is the easiest to manipulate to fit other States’
needs. A final important distinction is that this index, unlike the others, is
being calibrated and verified using biological data, not a panel of experts.
The chemical index values will tall in the range of 100 to 0 with use-category
cutoffs (calibration) being determined by the biological data and criteria
Verification will take place in a similar manner using different stations. The
most notable exception to this is the calibration for water supply waters.
Macrobenthos data may indicate pristine water conditions in water supplies
which are not supporting their use due to coliform contamination. As a result,
a panel of experts will be used to designate the cutoff ratings for water supply
Lake Water Quality Indices
The Ohio Lake Condition Index
This index was introduced in November 1988 and is titled as a ‘new
approach to lake classification.” One of the main objectives of this index is to
address overall water quality and not just trophic state. The index is
aggregated using an arithmetic mean and is reportedly easily modified to
accommodate revisions of water quality standards and changes in
management objectives. The objectives of this index are to meet the Water
Quality Act of 1987 and as a result the use support designation has been
included in this index from the beginning. The index is based on 13
parameters or matrices which include: index of biotic integrity (IBI), nuisance
growths of macrophytes, fecal coliforms, primary productivity based on
chorophyll a, fish tissue contamination, nonpriority pollutants, toxic organics,
toxic metals, sediment contamination, nutrients based on spring total
phosphorous, acid mine drainage, volume loss due to sediment, and aesthet-
ics Those parameters which had standards were based on the standards
while those that did not, like aesthetics, were based on best professional
judgment Each matrix was judged as either supporting, partially supporting,
or nonsupporting for each sampling date for each station The support
categories were each assigned a number of points with the best quality being
one and the worst being ten To avoid the problem of missing data, this index
totals up all of these subindex scores and divides them by the number of
matrices assessed. The resulting index value is a three-part number which
identifies the overall condition of the lake, any “support threatened” or
“nonsupporting” matrices, and finally the number of matrices which are
missing This index does not make use of rating curves per se and does
allow for more data to be missing than any of the other indices
Potentially this index shows more information than just a number in its results,
but the final value is complex enough that laypersons may need more
guidance to understand its meaning The index is based on standards and
does address the use category approach of water quality determination
D- 17

-------
APPENDIX D. WATER QUALITY INDICES
Potential drawbacks are that it is very labor intensive to apply and allows for
up to 50 percent of the data to be missing.
The Tennessee Valley Authority (WA) Water Quality Index
This index is essentially the same as Dinius’ index with some modifications
made to the rating curves to include water quality standards and additional
parameters. This index was developed to allow NA to simplify its data so
that it could be interpreted by laypersons without losing its technical validity.
It has been developed specifically for the Southeast. This index is designed
for shallow southeastern lakes and reservoirs with weak summer stratification
and short retention times. Butkus (1989) used a three-round DELPHI
questionnaire to determine which parameters to include in the index as well
as weightings and ratings. The parameters included were: suspended solids,
fecal coliform, true color, DO, chlorophyll a, temperature, and pH. The index
was designed to examine the quality of water for four water uses: fish and
aquatic life, recreation, domestic water supply, and industrial water supply
This index is still in its development stages. In fact, this review is based on
the draft sent out to the experts taking part in the questionnaire for the index.
It has been written up in a program to be loaded onto a PC Once again, this
may be a factor for STORET users because data must be manually
downloaded in order to use the program. In terms of application for use
support determinations, this index has the same drawbacks as that of Dinius:
it is not based on the standards of a State and the water uses are not
necessarily those of any State. The problem of missing data was not
addressed in this draft of the index and as a result further development and
documentation are needed to fully explain its applicability
References
Armstrong, Ann, 1989. The Development of an Arithmetic Mean Index for
Determining North Carolina’s Water Quality. Master’s Thesis Draft
Bogue, Bill, 1989 Personal communication, August.
Brown, A M. and N.I McCIeIIand, 1974 Up from Chaos: The Water Quality
Index as an Effective Instrument in Water Quality Management National
Sanitation Foundation, Ann Arbor, Ml
Brown, R M, N I McCleIIand, R A Deininger, and J.M Landwehr, 1973
Validating the WQI Presented at ASCE National Meeting on Water
Resources Engineering, Washington, DC
Butkus, Steve, 1989 Personal communication, May
Butkus, Steve and Donald W Anderson, 1989. Development of a Reservoir
Water Quality Index DRAFT from Tennessee Valley Authority,
Chattanooga, TN
D-18

-------
APPENDIX D. WATER QUALITY INDICES
Davic, Robert D and Jeffery E Deshon, 1988 The Ohio Lake Condition
Index: A New Multiparameter Approach to Lake Classification State of
Ohio Environmental Protection Agency, Division of Water Quality Monitor-
ing Assessment
Dinius, S H, 1987. Design of an Index of Water Quality Water Resources
Bulletin, 23(5) :833-43
Entzminger, Tom. Region 8 WQI EPA Region 8, Denver
EPA Region 8, 1981. EPA Water Quality Trends, Region 8
Florida DER, 1988. A Guide to the Interpretation of Metal Concentrations in
Estuarine Sediments Tallahassee, FL, April.
House, Margaret, 1989 Personal communication, July
House, M and J B Ellis, 1980 Water Quality Indices: An Additional Manage-
ment Tool” Prog Wat Tech, 13, pp 413-23
House, M A and D H. Newsome, 1989. Water Quality Indices for the Man-
agement of Surface Water Quality Wat Sci Tech, 13, pp. 1137-1148
House, M A, 1990. A Water Quality Index for Use in the Operational Manage-
ment of River Water Quality in Europe, to be published in IAWPRC Series:
Advances in Water Pollution Series, Watershed ‘89: The Future of Water
Quality in Europe, V2.
Landwehr, J, 1974 Water Quality Indices--Construction and Analysis Ph.D
Dissertation.
Maryland Water Quality Inventory, 1985-1987 Volume II of the Section 305(b)
Report.
McCIeIland, Nina I, Robert M Brown, RoIf A Deininger, and Jurate M
Landwehr, 1973 Wate 1 1 4 uaIity Index Application in the Kansas River
Basin Presented at 46 Annual Conference, Water Pollution Control
Federation, Cleveland, OH
McCIeIland, Nina I, Robert M Brown, and Roll A Deininger, 1976 WQI
Enhancement Appreciation of Quality Improvement Presented at
American Chemical Society, New York, NY
Ott, Wayne A, 1978a Environmental Indices--Theory and Practice Ann
Arbor Science Publishers, Inc
Ott, Wayne A, 1978b Water Quality Indices A Survey of Indices Used in the
United States U.S Environmental Protection Agency, Washington, D C
Publication No EPA-600/4-78-005
Peterson, Ray and Bill Bogue, 1989 Water Quality Index (Used in Environ-
mental Assessments) EPA Region 10, Seattle, Washington
D- 19

-------
APPENDIX D. WATER QUALITY INDICES
Research Triangle Institute, 1989 Summary of State Use Support
Methodologies. Prepared for EPA AWPD, April
Saylor, Charles and Edwin M Scott, Jr., 1987. Application of the Index of
Biotic Integrity to Existing WA Data Tennessee Valley Authority, Norris,
TN.
Schaeffer, D.J. and K G. Janardan, 1977. Communicating Environmental
Information to the Public: A New Water Quality Index Journal of Environ-
mental Education, 8(4), Summer.
Tyson, J.M. and M A. House, 1989. The Application of a Water Quality Index
to River Management. Wat. Sd. Tech., 21, pp. 1149-1159.
U S EPA, 1985. Technical Support Document for Water Quality-Based Toxics
Control Oftice of Water, Washington, DC, September.
U S EPA, 1989a. Water Quality Program Highlights: Ohio EPA’s Use of
Biological Survey Information. Oftice of Water Regulations and
Standards, May. Draft
U S. EPA, 1 989b. Resource Document for Workshop on Environmental
Indicators for Surface Water Program, Alexandria, VA, March.
U S EPA, 1 989c. Guidelines for the Preparation of the 1990 State Water
Quality Assessment (305(b) Report) Office of Water, Washington, DC,
February.
D-20

-------
Appendix E

-------
APPENDIX E
BIOLOGICAL METHODS

-------
APPENDIX E. BIOLOGICAL METHODS
APPENDIX E. BIOLOGICAL METHODS
Historically, most aquatic life criteria have been chemical-specific limits based
on laboratory toxicity tests These methods have been heavily relied on to
protect designated aquatic life uses, in part because proven measures of
natural communities had not been developed In a sense, the chemical-
specific results have been used as surrogates for measures of the actual
biological health of aquatic communities However, many direct measures of
community health have now been developed, tested and presented in
national guidance. These biological methods surpass chemical-specific
methods in certain areas, including assessing impacts of habitat modification,
of complex toxic inputs, and of nontoxic pollutants (0 g, nutrients and
sediment from nonpoint sources).
Rapid Bioassessment Protocols for Use in Streams and Rivers: Benthic
Macroinvertebrates and Fish (Plafkin et al, 1989) presents tested procedures
for conducting biosurveys and analyzing results These macrobenthos and
fish methods measure community structure and detect impairment at various
levels of certainty and technical effort. Other methods are available for
periphyton, plankton, and macrophyton communities The advantages of
using biological assessment methods in concert with physical/chemical
methods are described in Section 2.3 and in EPA’s draft Policy on the Use of
Biological Assessments and Criteria in the Water QuaIi Program (the draft
Ecopolicy”)
E.1 PHYTOPLANKTON
Because of their position at the base of food chains, phytoplankton
monitoring may be a very useful assessment tool, particularly in standing or
slow-flowing waters Phytoplankton populations, under favorable growth
conditions, may reach “bloom” conditions where they may cause taste and
odor problems and extreme diurnal dissolved oxygen fluctuations leading to
fish kills
E.2 PERIPHYTON AND MACROPHYTON
The periphyton, those organisms attached to underwater substrates, are an
assemblage of a wide variety of organisms The principal components are
algae, fungi, bacteria and protozoa In the shallow waters of lakes, ponds,
E-3

-------
APPENDIX E. BIOLOGICAL METHODS
and streams, the periphyton are often the dominant primary producers
Because they are fixed in place, they provide a time-integrated record of the
quality of the water passing by More subtle changes, such as species
composition, can indicate subtle changes in water quality
The diatoms, which are often the dominant algae of the periphyton, are partic-
ularly useful because there are many species and each can be identified The
environmental requirements and pollution tolerance of several hundred
species are documented Because of this, diatoms have been widely used
for high sensitivity water quality monitoring. The silicon cell walls of diatoms
preserve very well and make them useful for monitoring environmental
changes through longer time periods (decades or centuries). Recent
research on the effects of acid rain has used diatom data extensively. The
primary disadvantage is the high level of expertise and time required to make
the population counts and species identifications.
Macrophyton have been used much less than plankton and periphyton for
water quality monitoring. The macrophyton community has many of the
attributes of the periphyton community, but the diversity is much lower,
consisting of only a few dozen species of vascular plants. Macrophytes have
provided an estimate of habitat quality, particularly in the southern United
States
E.3 BENTHIC MACROIN VERTEBRATES
Macroinvertebrate community measures include indicators of abundance,
biomass, species composition, trophic (feeding) levels, pollution tolerances,
and life history requirements of benthos. They are used extensively to charac-
terize and evaluate the biological integrity of surface waters. The benthos are
relatively sedentary, inhabit the bottom sediment and other benthic substrates
for their life functions, and are sensitive to both long and short-term changes
in the sediment and water quality of various aquatic ecosystems
Use of macroinvertebrate community measures reflects overall ecological
integrity (chemical, physical, and biological), integrates the present stress
and, over time, provides an ecological measure of fluctuating environmental
conditions, and is relatively inexpensive. A limitation of macroinvertebrate
methods is the uneven distribution of the organisms in the sediment and on
substrates resulting in sampling data that may reflect more variability than is
generally accepted with chemical data
Rapid bioassessment protocols (RBPs) have been developed for conducting
biological assessments of lotic ecosystems and were designed as inex-
pensive screening and/or intensive tools for determining if a stream is or is
not supporting a designated aquatic life use (Plafkin et al 1989) Rapid
E-4

-------
APPENDIX E. BIOLOGICAL METHODS
bioassessment techniques are gaining in popularity because they serve as
cost-effective sbreening procedures for determining if a stream is or is not
supporting the designated use These macroinvertebrate community
measurement methods are described in methods manuals developed by both
government and private standards-setting organizations. State biomonitoring
programs specify the use of many of these techniques for regular surveillance
activities Rapid bioassessment techniques are most useful in wadeable
streams (streams sampled without a boat) Larger streams and rivers require
other types of sampling equipment and more intensive resources.
E.4 FISH
Fish are good indicators of long-term (several years) effects and broad habitat
conditions because they are relatively long-lived and mobile. Fish community
measures are appropriate direct measures of aquatic life resources
Several measures are employed to evaluate the biological integrity of rivers
and streams. Top carnivore population size and structure can be used to
relate directly to the “fishable” goal of the Clean Water Act. Relative
abundances of juveniles, adults, and large adults are used to measure a site
for its adequacy for rearing and reproduction and the potential impact of
bioaccumulation Where piscivorous (fish-eating) birds, reptiles, and
mammals are expected, the relative abundance of fish offers insights for
further food chain effects and provides an important assessment of species
commonly valued by the public. Fish are sensitive to over-harvest, acid
deposition, contaminants, and physical habitat/flow modification.
Another fish community measure commonly used is fish gross pathology
Stressors generally produce functional change in particular cells, tissues, and
organs If the duration or intensity of stressors is sufficient, fish structural
changes occur, followed by changes iri fish populations and assemblages
Typically, as the level of contaminants and pathogens rises, pathology
increases
Eutrophication has been found to affect gills and fin structure, acid deposition
has been found to affect sex organs,.and contaminants affect the liver and
gills One particular fish community measure used in one form or another by
more than 30 States in the United States is the Index of Biotic Integrity (IBI)
The IBI integrates several fish community measures such as top carnivore
size and structure and fish pathology with others such as age class, number
of tolerant species, number of native fish species, darter/benthic-intolerant
species, omnivorous individuals, insectivorous (insect-eating) individuals, and
a number of exotic/hybrid individuals
E-5

-------
APPENDIX E. BIOLOGICAL METhODS
Fish community measures such as the IBI are good indicators of whether
designated uses are being attained The IBI can provide a measure of health
and complexity of a fish assemblage relative to those of a series of minimally
impacted sites of similar size from the same ecological region Several
advantages of using fish community measures have already been mentioned
Others include: fish communities generally include a range of species that
represent a variety of trophic levels and therefore tend to integrate effects of
lower trophic levels and integrated environmental health; fish are at the top of
the food chain and are consumed by humans, making them important
subjects in assessing contamination; fish are easy to collect and identify;
environmental requirements of common fish are comparatively well known,
life history information is extensive for most species, and information on fish
distributions is commonly available; aquatic life uses (water quality standards)
are typically characterized in terms of fisheries (coldwater, warmwater, sport,
forage); and fish account for nearly half of the endangered vertebrate species
and subspecies in the United States.
The limitations of fish community measures are that they may be more diffi-
cult to obtain and use in large rivers, estuaries, and lakes/reservoirs. This is
due mainly to the logistics of capturing fish and the fact that fish communities
in large systems are the least understood. Coidwater fisheries such as trout
fisheries are more difficult to evaluate because of their low species richness
and lower relative abundance Variance is high in perturbed habitats, and
temporal and spatial variability affects larger systems more than it affects
smaller streams
Protocols for two fish community assessment techniques have been
published by the USEPA in the aforementioned RBPs The use of the IBI is
described in this document. The protocols also include examples and
discussions of how various States are using fish community assessments in
their monitoring programs
E.5 HABITAT
Evaluations of habitat quality is critical to any assessment of biological integri-
ty. The information needed is usually collected during biological surveys.
Examples of physical habitat measures are: general land use and physical
stream or lake/reservoir characteristics (width, depth, flow, and substrate for
streams; average depth, residence time, and average surface acreage for
lakes and reservoirs) Physical characterization starts with the riparian zone
(stream bank or lake shore leatures, and drainage area) and proceeds in-lake
or in-stream to sediment/substrate descriptions Such information can pro-
vide insight as to what organisms should be present or are expected to be
present.
E-6

-------
APPENDIX E. BIOLOGICAL METHODS
Predominant surrounding land use information is useful because of its
potential effect on water quality Local watershed erosion, existing or
potential, is characterized because detachment of soil from within the local
watershed and its movement into a stream or lake will alter the physical habi-
tat and reduce the expected species richness and abundance. Other
nonpoint source pollution that affects habitat alteration, such as agricultural
and urban runoff that contributes to factors other than siltation (feedlots,
wetlands, septic systems, dams and impoundments, and/or mine seepage
and tailings), is also identified and characterized.
Physical integrity or habitat is an important aspect of the abiotic component of
any ecosystem and can be evaluated qualitatively or quantitatively It is well
recognized by past and recent nonpoint source national assessments and
State 305(b) reports that habitat alteration is the major cause for reduced or
damaged ecological condition in streams, rivers, and lakes Guidance for
evaluating use attainment in EPA ’s National Water Quality Standards program
includes habitat assessment More recent policy development and guidance
for establishing biocriteria for lakes and streams include a habitat assessment
protocol (Platkin et al., 1989).
Habitat assessments of small and medium-sized streams and rivers are
expected to be straightforward with few problems Large rivers and lakes and
reservoirs will need research for the development of standardized assessment
and quantification methods. Physical habitat alteration due to sedimentation,
flow alteration, and channelization must be determined so that impacts to
biological communities can be separated into those caused by discharges of
contaminants and those related to physical habitat damage
Some modifications of existing habitat evaluation assessment methods may
be necessary depending upon the predominant fish species present
(coldwater vs warmwater) For example, assessments of trout fishery
habitats would emphasize different substrate requirements than would
assessments of warmwater fishery habitats for species such as bass or
sunfish Along with these fishery differences, benthic community
assemblages will also vary and, therefore, change the habitat characteristics
emphasized in a physical habitat assessment Many States have already
accounted for such considerations, and their techniques should be readily
usable by States having similar fisheries and benthic communities
Several States are known to use habitat assessments to supplement
macroinvertebrate and fish surveys The States of Arkansas, Maine, Missouri,
New York, North Carolina, Ohio, and Virginia use some aspect of habitat
evaluation to assist in their monitoring efforts Federal Agencies such as the
Bureau of Recreation, U S Forest Service, Bureau of Land Management, and
U S. Fish and Wildlife Service utilize habitat assessments to help sort out
differences between water column effects and physical habitat alterations
E-7

-------
APPENDIX E. BIOLOGICAL METHODS
References
Ohio EPA (Environmental Protection Agency). 1987. Biological Criteria for
the Protection of Aquatic Life: Volumes I, II, and Ill. The Role of Biological
Data in Water Quality Assessment (Final Draft). User’s Manual for
Biological Assessment of Ohio Surface Waters. Standardized Biological
Field Sampling and Laboratory Methods for Assessing Fish and
Macroinvertebrate Communities Ohio EPA, Columbus, Ohio 43212.
Plafkin, J. L., M. 1. Barbour, K. D. Porter, S. K. Gross, and R. M Hughes
1989. Rapid Bioassessment Protocols for Use in Streams and Rivers:
Benthic Macroinvertebrates and Fish. Assessment and Watershed
Protection Division, Office of Water, U.S. Environmental Protection
Agency, Washington, DC 20460. Publication No. EPA-444/4-89-001 May
These protocols include a habitat assessment method.
E-8

-------
Appendix F

-------
APPENDIX F
TOXICOLOGICAL METHODS

-------
APPENDIX F. TOXICOLOGICAL METHODS
APPENDIX F. TOXICOLOGICAL METHODS
F.1 GENERAL
Toxicity tests are used to estimate one or more of the following: (1) the
toxicity of an effluent collected at the end of the discharge pipe and tested
with a standard dilution water; (2) the toxicity of effluents collected at the end
of the discharge pipe and tested with dilution water consisting of nontoxic
receiving water collected upstream or beyond the influence of the outfall, or
with other uncontaminated surface water or standard dilution water having
approximately the same hardness or salinity as the receiving water, depend-
ing on the nature of the receiving water (fresh or saline) and test organisms;
(3) the toxicity of diluted effluent in the receiving water downstream or at
increasing distance from the outfall; and (4) the effects of multiple discharges
on the quality of the receiving water
F.1 ACUTE TOXICITY TESTS
Acute toxicity tests are used to predict potential acute and chronic toxicity in
the receiving water, based on the LC5O (lethal concentration to 50 percent of
exposed organisms) or EC5O (effective concentration for death, immobiliza-
tion, or other adverse effect for 50 percent of exposed organisms) and
appropriate dilution, application, and persistence factors
Two types of acute toxicity tests, static and flowthrough, are available
Static tests include (1) nonrenewal tests in which the test organisms are
exposed to the same test solution for the duration of the test, and (2) renewal
tests in which the test organisms are exposed to a fresh solution of the same
concentration of test solution every 24 hours or another prescribed interval,
either by transferring the test organisms from one test chamber to another or
by replacing all or a portion of the test solution in the test chambers The
renewal system is preferred because interfering factors such as toxicant
adsorption on the walls of the test chambers, volatilization, uptake by test
organisms, and metabojism may affect toxicity
F-3

-------
APPENDIX F. TOXICOLOGICAL METHODS
Two approaches to flow-through testing are available: (1) the test solution is
pumped continuously from the sampling point directly to the diluter system;
and (2) grab or composite samples are collected periodically, placed in a tank
adjacent to the test laboratory, and pumped continuously from the tank to the
diluter system
Standard acute toxicity tests involve the exposure of 20 test organisms to
each of five toxicant concentrations and a control water. The test duration
depends on the test species and ranges from a few hours for screening tests
to 96 hours for definitive tests with fish and some invertebrates. The results of
the test are reported as acute toxicity (LC5O or EC5O) usually expressed as a
percent, which is the concentration of test solution causing death (or
immobilization, or other adverse effect) in 50 percent of the test organisms.
These methods are applicable to all freshwater and near coastal marine
waters for assessing impacts from both point and nonpoint sources. In
general, the advantages of acute test methods are that they are short,
inexpensive, and easily reproducible In addition, they measure a direct
endpoint, such as lethality or immobilization, which is easily determined by
the investigator. A limitation of these tests is that they cannot be used directly
to measure long-term impact of toxic chemicals or effluents. The single
laboratory and multilaboratory precision of acute toxicity tests with several
common test species and reference toxicants ranges from 38 to 50 percent.
F.3 CHRONIC TOXICITY TESTS
The endpoints generally used in chronic tests are growth and reproduction.
The effects include the synergistic, antagonistic, and additive effects of all the
chemical, physical, and biological components that adversely affect the physi-
ological and biochemical functions of the test organisms. The results of the
short-term chronic toxicity tests are expressed as the NOEC, which is the
highest percent concentration at which no adverse effect on survival, growth,
or reproduction is observed.
Freshwater methods to measure the chronic toxicity of effluents and receiving
waters to three freshwater species include an invertebrate, a plant, and a fish:
the cladoceran, Ceriodaphnia dubia; the alga, Selenastrum capricornutum:
and the fathead minnow, Pimephales promelas The methods are 4- to 7-day
tests using the various test endpoints observed
Methods to measure the chronic toxicity of effluents and receiving waters to
five marine species include two fish species, two invertebrates and an alga:
the sheepshead minnow, Cyprinodon variegatus, and the inland silverside,
Menidia beryllina; the mysid, Mysidopsis bahia; the sea urchin, Arbacia
punctulata; and the red macroalga, Champia parvula. The methods are 1-
hour to 9-day tests using the various test endpoints as described by Weber
and Peltier (1988)
F-4

-------
APPENDIX F. TOXICOLOGICAL METHODS
These methods are applicable to all freshwater and near coastal marine
waters for assessing impacts from both point and nonpoint sources. In
general, the advantages of these chronic methods are that they are shorter
than most life-cycle tests while demonstrating good correlations of sensitivity
with the longer-cycle tests proposed earlier In addition, the tests are cost-
effective and are easily reproducible A limitation of these tests is that they
are short-term estimates for predicting chronic toxicity that need to be
checked periodically as to their sensitivity, compared to longer term, full life-
cycle tests. These methods can be used to determine receiving water
impacts from toxic effluents and evaluate single-chemical toxicity for
developing water quality criteria. The sensitivity of the tests depends in part
on the number of replicates, the probability level selected, and the type of
statistical analysis chosen. Precision depends on the dilution factor chosen
References
U S EPA, 1989 Guidelines Establishing Test Procedures for the Analysis of
Pollutants Under the Clean Water Act Proposed rule. Deliverable No
0749 [ A]/EMSL-Cincinnati No. 1141, “Report Promulgate Rapid Chronic
Toxicity Tests in Section 304(h) 40 CFR 136” Environmental Monitoring
System Laboratory, Cincinnati, OH
Weber, C I., W B Horning, II, D J. Klemm, T W Nieheisel, P A Lewis, E L
Robinson, J R. Menkedick, and F A Kessler (eds.). 1988. Short-term
Methods for Estimating the Chronic Toxicity of Effluents and Receiving
Waters to Marine and Estuarine Organisms Environmental Monitoring
and Support Laboratory, U.S Environmental Protection Agency,
Cincinnati, Ohio Publication No EPA/600/4-87/028
Weber, C.I , W H Peltier, I J. Norberg-King, W B Horning, II, F A. Kessler,
J.R. Menkedick, T.W Neiheisel, P A Lewis, D J Klemm, Q.H. Pickering,
E.L. Robinson, J M Lazorchak, L.J Wymer, and R W Freyberg. 1989
Short-term Methods for Estimating the Chronic Toxicity of Effluents and
Surface Waters to Freshwater Organisms, Second Edition Environmental
Monitoring Systems Laboratory, U S Environmental Protection Agency,
Cincinnati, Ohio Publication No. EPA/600/4-89/001
F-5

-------
Appendix C

-------
APPENDIX G
REMOTE SENSING

-------
APPENDIX G. REMOTE SENSING
APPENDIX G. REMOTE SENSING
Remote sensing technology consists of aerial photography as well as
airborne and satellite digital (multispectral scanners, or MSS) sensors.
Photography is typically used to map spills, features associated with landfills
and hazardous waste sites, and wetland delineations. MSS have been used
to assess quality: to detect changes over time; to map algal, macrophyte,
intertidal, and wetland species and extent; to detect thermal plumes; and to
record point and nonpoint runoff in lakes and rivers throughout the United
States Products for management applications include aerial photography
prints, digital image map products, and geographical information systems
(GIS) data bases
Aerial cameras or MSS are appropriate for (1) mapping large areas,
(2) obtaining emergency or simultaneous views of an area, (3) developing
complete aerial coverage rather than point data, (4) developing a digital data
sets, and (5) mapping areas that are inaccessible or remote.
Several measures of water quality and character may be acquired using
remote sensing devices Because light penetration in water is directly related
to the amount of suspended material in the water column, measures of
turbidity and clarity may be made using remote sensors If a relationship
between sediments and other parameters of concern can be established,
then sediment concentration may be used as a surrogate measure for such
parameters as conductivity, nutrients and chemicals In waters with low
turbidity, passive remote sensing can be used to characterize subsurface
features Under optimal conditions, dense stands of submerged macro-
phytes and shallow water (<10 m) bottom types can be delineated. Addi-
tionally, remote sensing devices may give a synoptic view of the watershed
Such a view is particularly useful because water quality problems seldom
originate at the shoreline or within a waterbody itself
The advantages of remote sensing techniques are they provide synoptic
views of a resource; they permit data for large areas to be acquired
simultaneously or over a very short period of time; and they cost little in com-
parison to intensive field investigations. Perhaps the major question with
using remote sensing for use support determinations is whether MSS data
can be correlated to a particular measurement of interest Several studies
have demonstrated that MSS data correlate directly with water clarity
However, icr other factors such as dissolved organic compounds, pH, and
conductivity, direct measurements cannot be made by MSS and indirect
correlations alone may or may not adequately represent the measurement of
interest Regardless of the features of being assessed by cameras or MSS,
some field data must be collected to define the accuracy of the final product.
G-3

-------
Appendix H

-------
APPENDIX H
USE OF FLOW DIAGRAMS

-------
APPENDIX H. USE OF FLOW DIAGRAMS
APPENDIX H. USE OF FLOW DIAGRAMS
One of the major deficiencies in national reporting for the 1988 §305(b)
reports was that it was very difficult to follow the use support decision process
through the data analysis and interpretation steps. One simple method for
improving reporting and assisting in structuring the decision process is to
incorporate flow charts or decision trees in the reporting process.
A schematic of an idealized use support decision process for protecting
aquatic life is presented in Figure H-i. For each waterbody, the process
could be framed as follows:
1 “Direct” measures of use support are evaluated Has a fish kill
occurred ’ Are sublethal impacts known to have occurred (e g , fish/
shellfish disease, noxious algal blooms, etc.)” If such events are
sufficiently documented, a preliminary determination (before review of
ambient monitoring information) that a waterbody is not fully supporting
could be reached Ambient biological monitoring data are reviewed, and
tentative use support determinations are made
2 All other recent “indirect” measures of use support are reviewed. If data
are available, this step could include:
• Reviewing water column data
• Calculating a water quality index value
• Determining if fish or sediments are contaminated
• Assessing habitat integrity
• Toxicity data analysis.
“Weights” are assigned to the different measures of water quality within
these categories. This assigning requires implicit or explicit judgments of
the importance of, for example, toxic vs. conventional pollutants, and
bioassay results relative to pollutant-specific results Indices might be
used for aggregating data into an overall measure of water quality.
Decisions on how to integrate direct and indirect use support measures
are made
3 Lacking recent or sufficient monitoring data, “evaluated” information is
used to make the use support decision
This flowchart (Figure H-i), shows that a single use support decision could
require a complex series of decisions based on analyses of diverse data sets
H-3

-------
APPENDIX H. USE OF FLOW DIAGRAMS
NonIeth
(e.g., dIe .
blooms)
None or + T end
Figure H-i. Hypothetical use support decision process for aquatic life uses.
Go to®
H-4

-------
APPENDIX H. USE OF FLOW DIAGRAMS
from Sections &
Figure H-i (continued).
from Section®
Tfend
H-5

-------
APPENDIX H. USE OF FLOW DIAGRAMS
Also at issue is whether or not different methods should be developed for
different uses. Using the flow chart as a starting point, a hypothetical State
could develop a list of indicators to use for use support determinations (see
Figure H-2).
Aquatic Life Survival and Propagation
A. Direct Measures
• Fish kills
• Disease and stress indicators
• Algal blooms
B Comprehensive measures
(biological, chemical, physical)
C. Community Measures
• Benthic invertebrates
• Fish
• Submerged aquatic vegetation
0. Habitat Measures
• Sedimentation
• Flow alterations
E. Toxics Measures
• Toxicity
• Tissue contamination
• Sediment contamination
• 304(l) lists
F Chemical and Physical Measures
• Indices
• Parameter-specific methods
— Toxicant assessments
— Nontoxicant assessments
(DO, nutrients, etc.)
G Evaluated Measures
• Public and professional opinion
(questionnaires, hearings,
correspondence)
• 319 lists
II. Public Water Supply
A. Drinking Water (Raw)
• Direct evaluations
— Water system closures
— Waterborne disease
— Algal blooms
• Pathogen analysis
• Toxicant criteria analysis
• Aesthetic measures
— Color
— Turbidity/suspended solids
— Taste and odor
III. Contact Recreation
A. Recreation Closures
B. Pathogen Measurements
C. Aesthetic Measures
• Turbidity/suspended solids
• Algal blooms
• Oil and grease
• Scums and floating debris
IV. Fish and Shellfish Consumption
A. Closures
B Tissue contamination
C Pathogens
V. Outstanding Resource Waters
A Water Quality Measures Based on
Other Uses
B Resource Value Measures for
Identified Resource
• Fishery production
• Water-based recreational use
Figure H-2. Examples of methods for determining designated use support.
H-6

-------