United States
Environmental Protection
Agency
Region III
Chesapeake Bay
Program Office
Region III
Water Protection
Division
EPA 903-R-07-003
CBP/TRS 285/07
July 2007
In coordination with the Office of Water/Office of Science and Technology, Washington, D.C., and the states
of Delaware, Maryland, New York, Pennsylvania, Virginia and West Virginia and the District of Columbia
Ambient Water Quality
Criteria for Dissolved
Oxygen, Water Clarity and
Chlorophyll a for the
Chesapeake Bay and Its
Tidal Tributaries
2007 Addendum
July 2007
-------
Ambient Water Quality Criteria
for Dissolved Oxygen, Water Clarity
and Chlorophyll a for the Chesapeake Bay
and Its Tidal Tributaries
2007 Addendum
July 2007
U.S. Environmental Protection Agency
Region III
Chesapeake Bay Program Office
Annapolis, Maryland
and
Region III
Water Protection Division
Philadelphia, Pennsylvania
in coordination with
Office of Water
Office of Science and Technology
Washington, D.C.
and
the states of
Delaware, Maryland, New York,
Pennsylvania, Virginia and
West Virginia and the District of Columbia
-------
Contents
Acknowledgments v
I. Introduction 1
Literature Cited 3
II. Refinements to the Chesapeake Bay Water Quality Criteria
Assessment Methodology 4
Background 4
Overview of the CFD Assessment Methodology 5
Description and Evaluation of the CFD-Based Assessment Methodology .. 9
Example CFD-based criteria assessment 9
CFD reference curves 12
Comparing assessment and reference curves 15
Development of a statistical decision-making framework 16
Results of the scientific evaluation 19
Application of the CFD-Based Assessment Methodology 21
Recommendations for application of the CFD-based methodology . 21
Recommendations for future refinement of the
CFD-based assessment methodology 23
Literature Cited 24
III. Application of Chesapeake Bay Water Quality Criteria
Assessment Procedures 25
Background 25
Assessment Units, Segmentation, and Sub-Segmentation 26
Data to be Used in Chesapeake Bay Criteria Assessments 28
Updating the Criteria Attainment Assessment Framework 31
Literature Cited 32
IV. Refinements to the Chesapeake Bay Dissolved Oxygen Criteria
Assessment Procedures 33
Background 33
Temporal Periods for Assessment of Dissolved Oxygen Criteria 36
Dissolved Oxygen Criteria Assessments in Shallow Waters
Versus Open Waters 37
Assessment of Short Duration Dissolved Oxygen Criteria 38
Dissolved Oxygen Criteria Reference Curves 39
Contents
-------
IV
Summer open-water and deep-water 39
dissolved oxygen criteria reference curves 42
Non-summer open-water dissolved oxygen criteria
reference curve 42
Assessment of deep-channel instantaneous minimum
dissolved oxygen criteria 42
Use of Percent Saturation as Dissolved Oxygen Criteria 43
Literature Cited 45
V. to the
Use 47
Background 47
Shallow-Water Designated Use Attainment Assessment 48
Assessment based on the single best year ofSAV 50
Assessment based on water clarity acres 53
Assessment based on CFD-based water clarity criteria attainment . 56
Shallow-Water Designated Uses and SAV No-Grow Zones 56
Water Clarity Criteria Reference Curves 57
Literature Cited 59
VI. a 61
State Water Quality Standards 61
Chlorophyll a Criteria Assessment Procedures 62
Literature Cited 62
VII.
63
Design and Approach for Chesapeake Bay Shallow-Water Monitoring . 63
Shallow-water monitoring design 65
Continuous monitoring component 66
Water quality mapping component 67
Schedule for Assessment of Shallow-Water Designated Use Habitats .. 68
Extending the timeframe 69
Additional resources 69
Assessment based on reduced monitoring 70
Segment prioritization schedule 74
Dissolved Oxygen Criteria Assessments Using
Shallow-water Monitoring Data 75
Temporal standardization 75
Scaling and interpolation issues 77
Water Clarity Criteria Assessments Using
Shallow-water Monitoring Data 78
Analysis issues 79
Statistical modeling 79
Interpolation 80
Chlorophyll a Criteria Assessments Using
Shallow-water Monitoring Data 82
Statistical modeling 83
Contents
-------
Modeling Approach 84
Analysis issues 84
Literature Cited 88
III. for
89
Background 89
Listing Category Decisions 89
Criteria attainment assessments 90
Dissolved oxygen criteria attainment assessments 90
Water clarity criteria attainment assessments 91
Chlorophyll a criteria attainment assessments 91
Benthic index of biotic integrity assessments 92
Assessment reporting framework 92
Listing Decision Framework 95
Segments previously listed as impaired 96
Segments not previously listed as impaired 96
Shallow-water designated use listing decisions 96
Literature Cited 97
A, The Cumulative Frequency Diagram Method for Determining Water
Quality Attainment: Report of the Chesapeake Bay Program
STAC Panel to Review Chesapeake Bay Program Analytical Tools . A-l
B. Detailed Chesapeake Bay Water Quality Criteria Assessment
Methodology B-l
C, Evaluation of Options for Spatial Interpolation C-l
D, User Guide and Documentation for the
Chesapeake Bay Interpolator D-l
E, Potential Methods for Assessing Shorter Duration
Dissolved Oxygen Criteria E-1
F. Data Used in Deriving the Open-Water, Deep-Water, and
Deep-Channel Dissolved Oxygen Criteria Summer Biological
Reference Curves F-l
G, Equations for the Open-Water, Deep-Water, and Deep-Channel
Dissolved Oxygen Criteria Summer Biological Reference Curves . . G-l
H, Equations for the Water Clarity Criteria Biological Reference
Curves H-l
I, Evaluation of Maryland and Virginia Chesapeake Bay Segment SAV
Acreages from 2003 to 2005 for Prioritizing Shallow-Water Monitoring
by Segment 1-1
Contents
-------
VI
J, Chesapeake Bay Estuarine Benthic Communities Assessment Protocol
for Maryland and Virginia 305b/303d Integrated Reports J-l
K. 2006 303(d) Assessment Methods for Chesapeake Bay Benthos . . . . K-l
L, Addendum to the Report: Development of Diagnostic Approaches to
Determine Sources of Anthropogenic Stress Affecting Benthic
Community Condition in the Chesapeake Bay L-l
-------
VII
Acknowledgments
This second addendum to the April 2003 Ambient Water Quality Criteria for
Dissolved Oxygen, Water Clarity, and Chlorophyll a for Chesapeake Bay and Its
Tidal Tributaries was developed and documented through the collaborative efforts of
the members of the Chesapeake Bay Program's Criteria Assessment Procedures
Workgroup and Water Quality Steering Committee.
PRINCIPAL AND CONTRIBUTING AUTHORS
This document resulted from the collaborative expertise and talents of the
Chesapeake Bay Program's state agency, federal agency, and academic institutional
partners. The 25 principal authors (listed first) and contributing authors (listed in
alphabetical order) follow by chapter. Unless noted, author affiliations are listed
under the specific workgroup or committee acknowledgments. Chapter 1: Richard
Batiuk; Chapter 2: Steve Preston; Chapter 3: Steve Preston; Chapter 4: Richard
Batiuk, David Jasinski, Marcia Olson, and Gary Shenk; Chapter 5: Richard Batiuk;
Chapter 6: Elgin Perry, Richard Batiuk, and Larry Harding (University of Maryland
Center for Environmental Science); Chapter 7: Bruce Michael, Rick Hoffman, Mary
Ellen Ley, Ken Moore, Elgin Perry, and Mark Trice; Chapter 8: Larry Merrill, Mark
Barath, Richard Batiuk, and Richard Eskin; Appendix A: David Secor; Mary
Christman; Frank Curriero; David Jasinski; Steve Preston; Ken Reckhow; and Mark
Trice; Appendix B: Gary Shenk; Appendix C: Steve Preston; Appendix D: Lowell
Banner (NOAA Chesapeake Bay Office), David Jasinski, and Gary Shenk; Appendix
E: Gary Shenk, Marcia Olson, and Elgin Perry; Appendices F, G, and H: Gary
Shenk; Appendix I: Bruce Michael; Appendix J: Mark Barath; Appendix K: Roberto
Llanso (Versar), Dan Dauer (Old Dominion University), Mike Lane (Old Dominion
University), and Jon Volstead (Versar); and Appendix L: Dan Dauer, Mike Lane, and
Roberto Llanso.
CRITERIA ASSESSMENT PROTOCOLS WORKGROUP
Steve Preston, chair (U.S. Geological Survey/Chesapeake Bay Program Office),
Harry Augustine (Virginia Department of Environmental Quality); Mark Barath
Acknowledgments
-------
VIM
(U.S. Environmental Protection Agency Region III); Thomas Barron (Pennsylvania
Department of Environment); Joe Beaman (Maryland Department of the
Environment); Jerusalem Bekele (District of Columbia Department of the
Environment); Stephen Cioccia (Virginia Department of Environmental Quality);
Elleanore Daub (Virginia Department of Environmental Quality); Sherm Garrison
(Maryland Department of Natural Resources); Darryl Glover (Virginia Department
of Environmental Quality); Peter Gold (U.S. Environmental Protection Agency
Region III); John Hill (Maryland Department of the Environment); Rick Hoffman
(Virginia Department of Environmental Quality); Dave Jasinski (University of
Maryland Center for Environmental Science/Chesapeake Bay Program Office); Jim
Keating (U.S. Environmental Protection Agency Office of Water); Rodney Kime
(Pennsylvania State Department of the Environment); Larry Merrill (U.S.
Environmental Protection Agency Region III); Bruce Michael (Maryland
Department of Natural Resources); Ken Moore (Virginia Institute of Marine
Science); Shah Nawaz (District of Columbia Department of Health); Roland Owens
(Virginia Department of Environmental Quality); Jennifer Palmore (Virginia
Department of Environmental Quality); Elgin Perry (Statistics Consultant); Charley
Poukish (Maryland Department of the Environment); Matt Rowe (Maryland
Department of the Environment); John Schneider (Delaware Department of Natural
Resources and Environmental Control); Gary Shenk (U.S. Environmental Protection
Agency Chesapeake Bay Program Office); Nicoline Shulterbrandt (District of
Columbia Department of Health); Donald Smith (Virginia Department of
Environmental Quality); Matt Stover (Maryland Department of the Environment);
Robert Swanson (Virginia Department of Environmental Quality); Bryant Thomas
(Virginia Department of Environmental Quality); Mark Trice (Maryland Department
of Natural Resources); Michael Williams (University of Maryland Center for
Environmental Sciences/Chesapeake Bay Program Office); and Dave Wolanski
(Delaware Department of Natural Resources and Environmental Control).
Diana Esher, Chair (U.S. Environmental Protection Agency Chesapeake Bay
Program Office); Richard Batiuk (U.S. Environmental Protection Agency
Chesapeake Bay Program Office); Sheila Besse (District of Columbia Department of
the Environment); Bill Brannon (West Virginia Department of Environmental
Protection); Patricia Buckley (Pennsylvania Department of Environmental
Protection); Katherine Bunting-Howarth (Delaware Department of Natural
Resources and Environmental Control); Jennifer Capagnini (Delaware Department
of Natural Resources and Environmental Control); Moira Croghan (Virginia
Department of Conservation and Recreation); Frank Dawson (Maryland Department
of Natural Resources); Rusty Diamond (Department of Environmental Protection);
Ron Entringer (New York Department of Environmental Conservation); Richard
Eskin (Maryland Department of the Environment); Stuart Gansell (Pennsylvania
Department of Environmental Protection); Dave Goshorn (Maryland Department of
Natural Resources); Carlton Hay wood (Interstate Commission on the Potomac River
Foreword
-------
IX
Basin; Teresa Koon (West Virginia Soil Conservation Association); Bruce Michael
(Maryland Department of Natural Resources); Matt Monroe (West Virginia
Department of Agriculture); Kenn Pattison (Pennsylvania Department of
Environmental Protection); Alan Pollock (Virginia Department of Environmental
Quality); John Schneider (Delaware Department of Natural Resources and
Environmental Control); Rick Shertzer (Pennsylvania State Department of
Environmental Protection); Tom Simpson (University of Maryland); Randolph Sovic
(West Virginia Department of Environmental Protection); Pat Stuntz (Chesapeake
Bay Commission); Ann Swanson (Chesapeake Bay Commission); and Robert
Yowell (Pennsylvania Department of Environmental Protection).
Y. - rcc
The support and expert advice of all the members of the Chesapeake Bay Program's
Scientific and Technical Advisory Committee, under the leadership of Dr. Carl
Hershner (Virginia Institute of Marine Science) and the Executive Secretarial sup-
port of Dr. Kevin Sellner (Chesapeake Research Consortium) are hereby acknowl-
edged. The Scientific and Technical Advisory Committee convened a panel of inde-
pendent scientific experts to provide expert advice and direction on a set of criteria
assessment issues and procedures. The members of the Panel for Review of
Chesapeake Bay Program Analytical Tools were: Dr. David Secor, Chair (University
of Maryland Center for Environmental Science); Dr. Frank Curriero (Johns Hopkins
University); Dr. Mary Christman (University of Florida); and Dr. Ken Rechow (Duke
University). Elgin Perry, independent statistical consultant, provided the statistical
analysis support to the panel. Chapter 2 summarizes the panel's findings and recom-
mendations; Appendix A describes the findings and recommendations in full.
The contributions of the independent scientific peer reviewers—selected and con-
vened by the Chesapeake Bay Program's Scientific and Technical Advisory
Committee based on their recognized national expertise and drawn from institutions
and agencies across the country—are hereby acknowledged.
Without the efforts of the hundreds of colleagues involved in all aspects of field col-
lection, laboratory analysis, management, and interpretation of Chesapeake Bay
Monitoring Program data over the past two decades, these criteria could not have
been derived and the criteria assessment procedures could not have been developed.
Technical editing by Nina Fisher, independent technical editor, and document prepa-
ration by Jamie Bosiljevac, Chesapeake Research Consortium/Chesapeake Bay
Program Office, are also acknowledged.
-------
chapter |
In April 2003, the U.S. Environmental Protection Agency (EPA) published the Am-
bient Water Quality Criteria for Dissolved Oxygen, Water Clarity and Chlorophyll a
for the Chesapeake Bay and Its Tidal Tributaries (Regional Criteria Guidance) in
cooperation with and on behalf of the six watershed states—New York, Pennsyl-
vania, Maryland, Delaware, Virginia, and West Virginia—and the District of
Columbia. The culmination of three years of work, the criteria document resulted
directly from the collective contributions of hundreds of regional scientists, technical
staff, and agency managers as well as the independent review by recognized scien-
tific experts across the country (U.S. EPA 2003).
In October 2004, EPA published the first addendum to the 2003 Regional Criteria
Guidance (U.S. EPA 2004). The addendum provided additional guidance on:
• Applying the temperature-based open-water dissolved oxygen criteria required
to protect the endangered shortnose sturgeon;
• Assessing attainment of the instantaneous minimum and 7-day mean dissolved
oxygen criteria using monthly mean water quality monitoring data;
• Deriving site-specific dissolved oxygen criteria and assessing criteria attain-
ment of those tidal systems where the extensive adjacent tidal wetlands cause
naturally low dissolved oxygen levels;
• Delineating the upper and lower boundaries of the pycnocline that defines the
vertical boundaries distinguishing open-water, deep-water, and deep-channel
designated uses;
• Applying, in combination, the numerical water clarity criteria to shallow water
habitats and submerged aquatic vegetation restoration goal acreages for
defining attainment of the shallow-water bay grass designated use; and
• Determining where numerical chlorophyll a criteria should apply to local
Chesapeake Bay and tidal tributary waters.
From 2004 through early 2006, Delaware, Maryland, Virginia, and the District of
Columbia adopted: the EPA-published Chesapeake Bay water quality criteria for
chapter i » Introduction
-------
dissolved oxygen, water clarity, and chlorophyll a; the EPA-recommended tidal
water designated uses; and the EPA-established criteria assessment procedures into
their respective state water quality standards regulations. All four jurisdictions1
promulgated narrative chlorophyll a criteria in their standards regulations. Virginia
promulgated numerical segment- and season-specific chlorophyll a criteria for the
tidal James River. The District of Columbia promulgated numerical chlorophyll a
criteria for its reach of the tidal Potomac River and its remaining tidal waters, having
previously adopted numerical chlorophyll a criteria for protection of the tidal
Anacostia River.
The April 2003 Regional Criteria Guidance and the October 2004 addendum docu-
ments published the criteria attainment assessment methods (U.S. EPA 2003, 2004).
These methods characterize the spatial and temporal variability of the appropriate
water quality parameters and provide a clear basis for deciding whether a criterion
or set of criteria protecting a designated use in a specific segment of the mainstem
Chesapeake Bay or one of the tidal tributaries or embayments were in attainment.
The methods were quite detailed; however, specific technical and procedural issues
remained in applying the methods as specified in the original publication by EPA
from April 2003. These issues required resolution to allow Delaware, Maryland,
Virginia, and the District of Columbia to assess attainment of their new Chesapeake
Bay water quality standards regulations fully.
This second addendum documents the revised, refined, and new criteria assessment
methods for the published Chesapeake Bay dissolved oxygen, water clarity, and
chlorophyll a criteria.
• Chapter 2 documents refinements to and recommendations for further devel-
opment of the spatial interpolation and statistical aspects of the overall
Chesapeake Bay water quality criteria attainment assessment methodology.
• Chapter 3 documents the resolution of and recommended procedures for
addressing a series of overarching Chesapeake Bay water quality criteria
assessment issues.
• Chapter 4 documents refinements and additions to the procedures for
assessing the previously published Chesapeake Bay dissolved oxygen criteria.
• Chapter 5 documents refinements and additions to the procedures for
assessing the previously published Chesapeake Bay water clarity criteria and
determining attainment of the shallow-water bay grass designated use.
• Chapter 6 documents refinements and additions to the procedures for
assessing attainment of state-adopted numerical concentration-based chloro-
phyll a criteria.
'References throughout the text to "states" or "jurisdictions" means a collective reference to the states
of Delaware and Maryland, the Commonwealth of Virginia, and the District of Columbia. All four have
Chesapeake Bay tidal waters within their jurisdictional boundaries.
chapter
-------
• Chapter 7 documents new recommended methodologies and procedures for
using shallow-water monitoring data in assessing attainment of Chesapeake
Bay water quality criteria and tidal water designated uses.
• Chapter 8 documents a recommended 303(d) list decision-making framework
for assessment of Chesapeake Bay and its tidal tributaries and embayments.
This document represents the second formal addendum to the 2003 Chesapeake Bay
water quality criteria document; as such, readers should regard the sections in this
document as new or replacement chapters and appendices to the original published
report. The criteria attainment assessment procedures published in this addendum
replace and otherwise supercede similar criteria assessment procedures originally
published in the 2003 Regional Criteria Guidance and 2004 addendum (U.S. EPA
2003, 2004). Publication of future addendums by EPA on behalf of the Chesapeake
Bay Program watershed jurisdictional partners is likely as continued scientific
research and management applications reveal new insights and knowledge that
should be incorporated into revisions of state water quality standards regulations in
upcoming triennial reviews.
U.S. Environmental Protection Agency. 2003. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a for Chesapeake Bay and Its Tidal Tributaries. EPA
903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2004. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a for Chesapeake Bay and Its Tidal Tributaries -
2004 Addendum. EPA 903-R-04-005. Region III Chesapeake Bay Program Office,
Annapolis, MD.
chapter i » Introduction
-------
chapter||
Refinements to the Chesapeake
Bay Water Quality Criteria
Assessment Methodology
BACKGROUND
The Chesapeake Bay water quality criteria were designed to protect the ecological
integrity of the Bay's tidal waters. To ensure that the criteria are being attained and
the Chesapeake Bay ecosystem is, in fact, protected, adequate means to measure and
evaluate water quality relative to the criteria must exist. The Bay is a highly diverse
and variable system; these characteristics make precise assessment of water quality
criteria attainment difficult. Thus, it is critical to design both a data collection system
and a data analysis methodology carefully to make the best use of existing resources
and provide the best possible assessment of water quality criteria attainment. Such a
design can inform stakeholders about the status of impairments and whether the
impairments have been removed once management actions have resulted in the
achievement of the desired restoration goals.
To address the need for enhanced water quality criteria assessments brought on by
the states' adoption of new Chesapeake Bay water quality standards, the Chesapeake
Bay Program1 redesigned its tidal monitoring network to provide a framework for
interpreting the data. To the extent possible (within funding constraints), existing
monitoring programs were either enhanced to support criteria assessment or new
monitoring programs were established to address monitoring gaps. Given the diver-
sity of tidal habitats throughout the Bay, establishing a comprehensive tidal
monitoring network required different types of monitoring.
lrThe Chesapeake Bay Program, formed in 1983 by the first Chesapeake Bay agreement, is a unique
regional partnership guiding the restoration of the Chesapeake Bay and its tidal tributaries. On water
quality issues, the Chesapeake Bay Program partners include Delaware, Maryland, New York,
Pennsylvania, Virginia, West Virginia, the District of Columbia, the Chesapeake Bay Commission, the
U.S. Environmental Protection Agency, over 20 other federal agencies, academic institutions, local gov-
ernments, and citizen groups.
chapter ii • Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
Developing a methodology for assessing criteria attainment using these data was
also critical. Ideally the criteria assessment methodology would prove useful in
several ways: 1) it could be applied consistently for many water quality criteria
components; 2) it would provide a common framework for assessing data collected
over multiple scales; 3) it would provide a basis for using as much of the informa-
tion contained in the collected data as possible; 4) it would provide a clear basis for
making decisions on criteria attainment; and 5) it would provide diagnostic infor-
mation regarding the spatial and temporal patterns of criteria violations. The
cumulative frequency diagram (CFD) approach, described in the original 2003
Chesapeake Bay water quality criteria document, was designed with many of these
objectives in mind (U.S. EPA 2003a).
OF THE
The original 2003 Chesapeake Bay water-quality criteria document fully describes
the CFD methodology (Chapter 6, pages 154-178), but is summarized briefly here
(U.S. EPA 2003a). Criteria assessment using the CFD methodology is based on
interpolation within a spatially defined grid. Described later in this chapter, this grid-
based interpolation provides the spatial framework for use of all of the data. It
weights each data location according to the amount of area (or volume) it represents.
Water quality parameter levels in all interpolator grid cells are estimated based on
interpolation algorithms, providing a complete "map" of water quality throughout
the assessed area (Figure II-1). Water quality parameter levels in each grid cell are
compared to the applicable criteria levels to establish an estimate of the spatial extent
of criteria exceedance (non-attainment). Aggregating the total amount of space (area
or volume) in which the criteria are exceeded provides a basis for estimating the
percentage of the spatial assessment unit (designated use within a segment) in which
the criteria were exceeded for that monitoring cruise. These measures of criteria
exceedance are then compiled over the entire assessment period to develop a cumu-
lative frequency diagram, or CFD. The CFD is a well-known and well-established
statistical procedure commonly used to describe hydrologic and environmental data
(Helsel and Hirsch 1992).
The CFD assessment methodology evolved from the need to allow for variability in
water quality parameters due to unusual events. For the water quality parameter to
be assessed, a criterion threshold is established; when the threshold is exceeded, the
system is considered impaired. All water quality parameters, however, are inherently
variable in space and time. Because of this variability, it is unlikely that even a
healthy Chesapeake Bay ecosystem will attain the threshold absolutely in all places
and at all times.
Spatially, small regions may persistently exceed the criteria's threshold due to poor
flushing or other natural conditions. Such areas should not automatically lead to the
assumption that the entire assessment unit is impaired. Similar logic applies in the
chapter ii « Refinements to the Chesapeake Bay 'Water Quality Criteria Assessment Methodology
-------
6
Figure 11-1. Example of interpolation of Chesapeake Bay water quality data.
temporal dimension. Water quality in a large area of a segment may exceed the
criteria's threshold for a short time. If this degradation proves infrequent and short-
lived with the segment quickly returning to a healthy state, this situation does not
represent an impairment of the ecologically defined designated use of the segment.
Recognition that ephemeral exceedances of the criterion's threshold in time or space
do not represent persistent impairment of the segment's designated use ultimately
led to the development of a criteria assessment methodology that deems such
exceedances as acceptable. Persistent, widespread criteria exceedance, however, is
considered an impairment of the segment's designated use (U.S. EPA 2003a).
The criteria assessment methodology determines how much of the spatial assess-
ment unit is not in compliance with the criteria (percent of space) for each moment
in time. In the second step of the methodology, a determination is made of how often
(percent of time) a segment is out of attainment by more than a fixed percent of
chapter ii • Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
space. The results of these queries can be presented in graphical form with percent
of time plotted against percent of space.
Figure 11-2 illustrates a typical CFD based on 12 measures of spatial extent of criteria
exceedance over time. In general, if a segment is in attainment with the criterion,
then one expects a high frequency of dates for which the percent out of attainment
is low. In this case, the CFD should descend rapidly from the upper left corner, pass
not far from the lower left corner, and then proceed to the lower right corner. The line
in Figure II-2 shows the typical hyperbolic shape commonly observed using the CFD
to assess water quality criteria in the Chesapeake Bay. The closer the CFD curve
comes to the origin (lower left corner), the better the attainment of the assessed
segment. A curve that is far from the origin indicates that a larger percent of space
in the segment is out of attainment and the probability of use impairment increases.
The CFD methodology offers many advantages over other criteria assessment
approaches. Through interpolation, it provides a method for using data collected in
areas surrounding the area of interest (the spatial assessment unit). This factor is
important since the sample size of observations within a spatial assessment unit may
not be sufficient to determine the area (or volume) of exceedance within the unit
accurately. The method also weights the data collected from a given location
according to the amount of area (or volume) that the location represents. This capa-
bility is important because data may be collected from locations that do not represent
1.0
0.9
0.8
0.7
~ 0.6
"o
o 0.5
o
§• 0.4
Q.
0.3
0.2
0.1
O.Ol
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Proportion of Space
0.8
0.9
1.0
II-2. A water quality criteria attainment assessment cumulative frequency diagram
(CFD) based on 12 measures of the spatial extent of criteria exceedance over time.
-------
all areas of the spatial assessment unit; providing equal weight to such data could
bias the assessments.
A second advantage is that the CFD incorporates the spatial-temporal pattern of
criteria exceedance into the assessment. The shape of the curve offers information on
patterns of exceedance in space and time. Such information may prove helpful in
understanding the causes of impairments (see page 162 in U.S. EPA 2003a).
A third advantage is that it bases the assessment on biologically determined patterns
of allowable criteria exceedance. Reference curves are ideally developed in the same
way as assessment curves and should reflect the degree of criteria exceedance that
can be withstood by the ecological communities without impairing the designated
use. Thus, comparison of the assessment curve to the reference curve ensures that
any allowable criteria exceedances do not occur in a spatial or temporal pattern that
could, in reality, represent impairment at the scale of the entire assessment unit (see
pages 162-178 in U.S. EPA 2003a). Local persistent effects could still have high
impairment.
Finally, the combined elements of the CFD criteria assessment methodology fully
and effectively address all five factors used to determine attainment of designated
uses: magnitude, duration, frequency, space, and time. After conducting a national
review of TMDL programs, the National Research Council (2001) concluded that
establishing these conditions is crucial for successful application of state water
quality standards.
The CFD methodology is a new and innovative method of water quality criteria
assessment, representing an improvement over methods used in other parts of the
country (STAC 2006). The standard practice for assessing compliance with water
quality criteria throughout the United States is by sampling monthly at a fixed set of
stations and gauging compliance strictly from a count of exceedances of those
samples. Sampling stations are typically located for convenience (e.g., accessibility).
Consequently, reluctance to re-evaluate and change location (so as to maintain a time
series at a fixed point) is common; no consideration is given to the representative-
ness of the sample for the space/time not sampled.
Most assessments are based simply on EPA assessment guidance in which all
samples in a given area were compiled; attainment was assumed if no more than 10
percent of the samples exceeded the standard (U.S. EPA 1997). In this approach, all
samples are assumed to be fully representative of the specified space and time and
are simply combined as if they were random samples from a uniform population.
This approach was necessary in the past because the technology did not exist for a
more rigorous method of data analysis; however, it neglected spatial and temporal
patterns in the criteria measures. The CFD approach was designed to characterize
these spatial and temporal patterns and weight samples more accurately based on the
amount of space or time that they actually represent.
chapter
-------
The CFD methodology was first applied in the Chesapeake Bay for the most recent
303(d) listing cycle, completed in the spring of 2006 and based on data from 2002
through 2004. The CFDs were developed and used primarily for the dissolved
oxygen open-water and deep-water 30-day mean criteria because insufficient data
and data analysis techniques existed to assess the higher-frequency dissolved oxygen
criteria components. Similarly, the water clarity criteria were not assessed based on
the CFD because few tidal systems had sufficient shallow-water monitoring data for
an assessment.
In fall 2005, the Chesapeake Bay Program's Scientific and Technical Advisory
Committee (STAC) established a scientific panel to review and refine the CFD
assessment methodology. Nationally recognized academic experts in spatial and
environmental statistics made up the panel. The STAC-convened panel concluded
that the CFD approach is both feasible and innovative, qualifies as the best available
science, and represents an improvement over criteria assessment methods used in the
past (STAC 2006).
The panel also recognized, however, that the approach remains in the early stages of
management application. It stated that the CFD approach deserves further directed
study and analysis to evaluate the bias and imprecision that can occur due to limita-
tions in available data and in current interpolation and CFD algorithms (STAC
2006). This chapter provides guidance for criteria assessment application, summa-
rizes findings from the CFD evaluations, and offers recommendations for further
refinement of the CFD assessment methodology. Appendix A provides a complete
copy of the scientific panel's final report.
DESCRIPTION AND EVALUATION OF THE CFD-BASED
ASSESSMENT METHODOLOGY
The methodology for estimating the CFD is most easily described as a series of eight
steps as shown in Table II-1. These steps, described below, provide a framework for
considering the process and are elucidated by a simple example. More detailed
discussions of each step follow later in this chapter.
EXAMPLE CFD-BASED CRITERIA ASSESSMENT
To illustrate the CFD criteria assessment methodology, a simple theoretical example
based on a small data set can prove useful. Assume a segment for which the inter-
polation grid is 4 cells by 4 cells. In reality, the number of grid cells is much larger
(hundreds to thousands), but this small grid is illustrative. Also assume that data
were collected on five distinct dates, and that each date is representative of the appro-
priate time scale (in an actual application, data would be collected over many more
dates). The criterion threshold for this fictitious water quality parameter is 3.
chapter ii • Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
10
-1 Steps for constructing and assessing criteria attainment using cumulative
frequency diagrams (CFDs).
1. Collect data from a spatial network of locations on several dates during the
assessment period.
2. For each date, interpolate the data spatially over the entire system to obtain esti-
mates of water quality using a two- or three-dimensional grid of interpolation
cells.
3. Aggregate interpolations to the appropriate temporal scale (e.g., if evaluating the
30-day mean, take the average of all interpolations for each date in the month).
4. For each interpolator cell, assess whether the applicable criterion is exceeded.
5. For each assessment unit, compute the percentage of interpolator cells that exceed
the criterion as an estimate of the percent of area (or volume) within the spatial
assessment unit that exceeds the criterion.
6. Rank the percent of area estimates for the set of all sample days in the assessment
period from largest to smallest and sequentially assign to these ranked percents a
value that estimates percent of time. Add the end points of (100%, 0%) and (0%,
100%).
7. Plot the paired percent of area (or volume) and percent of time data on a graph
with the percent of area on the x-axis and percent of time on the y-axis. The
resultant plot is the assessment cumulative frequency diagram or CFD.
8. Compare the assessment CFD (from step 7) to the appropriate reference CFD. If at
any point the assessment CFD exceeds the reference CFD (i.e., a given level of
spatial noncompliance occurs more often than allowed for a given amount of
time), then the criterion is in non-attainment. Consequently, the segment fails to
meet that designated use.
An illustration of the eight steps for computing the CFD for these simplified
constraints is shown on the facing page. The three columns show the first three steps.
Column 1 provides fictional data for five dates for five fixed locations in a two-
dimensional grid. Column 2 shows a fictional interpolation of these data to cover the
entire grid. Column 3 gives the compliance status of each cell in the grid with 1 indi-
cating non-attainment and 0 signifying attainment.
In this hypothetical example, the assessment curve is clearly greater than the refer-
ence curve and in non-attainment of the criterion, therefore, the designated use is not
met. EPA recommends that any exceedance of the attainment CFD above the refer-
ence CFD should be considered non-attainment of the criterion and, consequently,
the designated use.
chapter ii » Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
11
Step 1. Collect data at
known locations.
Step 2. Interpolate the
data to grid cells.
Steps 3-4. Determine
attainment status of
each cell.
Date 1
Date 1
Date 1
3
2
5
3
1
Date 2
1
1
3
1
1
Date 3
4
1
2
2
1
Date 4
1
4
2
4
1
DateS
1
1
2
3
1
3
4
3
2
4
4
3
3
5
5
4
3
3
2
1
1
Date 2
Date 3
Date 4
DateS
1
2
1
1
2
2
3
1
3
3
2
1
1
2
1
1
4
3
2
1
3
2
2
1
2
2
1
1
2
1
1
1
1
2
3
4
2
2
3
3
3
2
2
1
4
3
1
1
1
2
1
1
2
2
1
1
3
2
1
1
3
2
1
1
Date 3
Date 4
DateS
1
1
1
0
1
1
1
1
1
1
1
1
1
0
0
0
Date 2
0
0
0
0
0
0
1
0
1
1
0
0
0
0
0
0
1
1
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
0
1
1
1
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
Step 5: Determine
percent attainment by date.
Sample date
Date 1
Date 2
DateS
Date 4
DateS
Percent
space
75.00%
18.75%
18.75%
43.75%
12.50%
Step 6. Rank the percent of
space values and assign percent
of time as (100*R/(N+1)), where
R is rank and N is sample size.
Sample date
Date 1
Date 4
Date 2
Date 3
DateS
Ranked
percent
space
100%
75.00%
43.75%
18.75%
18.75%
12.50%
0%
Percent time
0%
16.67%
33.33%
50.00%
66.67%
83.33%
100%
Steps 7 and 8. Figure II-3 illustrates the plot of this
theoretical assessment CFD and the comparison to a
hypothetical reference curve. In this hypothetical ex-
ample, the assessment area shows non-attainment. For a
percent area of 18.75, the allowable frequency on the
reference curve is about 17 percent. That is, 18.75 per-
cent of the segment area should not be out of attainment
more that 17 percent of the time. For Date 3, the esti-
mated frequency of 18.75 percent of segment area in
non-attainment is 66.67 percent. Thus the frequency of
18.75 percent of space out of attainment exceeds the 17
percent allowed. The reference curve is exceeded for
dates 4 and 1 as well.2
2In this cumulative distribution framework, the actual date is
not relevant. One should not infer that non-attainment occurred
on that date if the data point associated with a date falls above
the reference. The date is used here as a label for each coordi-
nate pair.
chapter ii • Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
12
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Area Exceeding the Criterion
II-3. Graphical representation of the CFD from the above theoretical example
assessment curve (blue) with a hypothetical reference curve (black).
CFD
Two approaches are feasible in defining the reference curves proposed for use in the
CFD assessment methodology. One is biologically based and identifies appropriate
regions of the Bay, its tidal tributaries, and its embayments that have healthy biolog-
ical indicators and are in attainment of their designated use (U.S. EPA 2003a). The
CFDs are developed for these areas in the same way that assessment CFDs would be
developed elsewhere. Curves generated for biologically healthy tidal areas are
considered "reference" curves.
For example, healthy benthic indices of biotic integrity (IBI) scores might be used as
indicators of adequate bottom dissolved oxygen (Weisberg et al. 1997; U.S. EPA
2003a). Thus, after stratifying by salinity zone and perhaps other factors, a series of
dissolved oxygen reference CFD curves could be developed from the existing moni-
toring database. One advantage of this approach is that each biological reference
curve could be tailored to each designated-use-based criteria component. This tech-
nique tailors the pattern of criteria exceedance that the part of the Bay ecosystem
could tolerate and remain healthy to the protected species and biological communi-
ties and the specific criterion component. Thus, each reference curve may have a
somewhat different shape (see pages 168-177 in U.S. EPA 2003a).
chapter
-------
13
In some cases, development of a biologically-based reference curve is not possible
due to lack of data describing the health of the relevant species or biological commu-
nities. Such cases require a different approach. The EPA recommends use of a
default reference curve in situations for which a biologically based reference curve
remains unavailable. This default reference curve is defined as a hyperbolic curve
that encompasses no more than 10 percent of the area of the CFD graph (percent of
space multiplied by percent of time) (see page 174 in U.S. EPA 2003a) (Figure II-4).
The default reference curve has the following important properties: 1) the plot is
symmetric about the 1:1 line; 2) the plot is hyperbolic; 3) the total area under the
100%
"o
o
5
o .0
S0* ~ 7n°/
c 0
8 °!
i- £ Rn%
Q_ W
_ "C
^ Q)
~ g OU/o
'o X
giu
Q- 0> 4QO/ _
w E 40/0
(0 3
o> o
jZ ^S
Iz ®
o < 20%
c
0>
o
Q.
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Area/Volume Exceeding the Criterion
Default reference curve for application in the attainment assessment of
Chesapeake Bay water quality criteria for which biologically based reference curves
cannot be derived.
curve equals 10 percent; and 4) the ends of the curve pass through x- and y-axis
intercepts (100, 0) and (0, 100), respectively.
Figure II-4 is defined by the equation:
(x + b)(y + b) = a
where: b = 0.0429945 and a = b2 + b.
Equation 1
-------
14
No specific theoretical basis underlies this definition of the default reference curve,
but the definition does provide equal weight to exceedances occurring in either space
or time. This approach is appropriate since no information exists to indicate that
either time or space should take precedence. Selection of the 10 percent value is
based on its consistency with past national EPA guidance (U.S. EPA 1997). The
default reference curve is hyperbolic, making it similar in shape to biologically
based reference curves. In fact, the shape of the default reference curve is quite
similar to some of the established biologically based reference curves, such as the
30-day mean open-water dissolved oxygen reference curve (Figure II-5).
"o
o
o
8,1
IB 70/0
80
0)0) 60 ^
Q. £
~° «
o> TJ ^n% -
_ -a au/0
o a>
S S
a 1
o
Q.
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Volume Exceeding the Criterion
Biological reference curve for 30-day mean open-water dissolved oxygen
criterion applied for assessment during the summer months (June-September) only.
A default reference curve, defined as a hyperbolic curve encompassing no more than
10 percent frequency exceedances, was also considered. Such a curve is based on a
simple model:
y = u + a; + bj
Equation 2
where a is temporal term with variance 2a and b is spatial term with 2b. The vari-
ance of Xy is $2a + $2b = $2. The standard deviation of Xy is sqrt(2) = $. Ten
percent of the xy- should fall above u + 1.2815 * $ where 1.2815 is the 90th
percentile of the standard normal distribution. Thus, assuming normality, a popula-
tion with equal spatial and temporal variance and a mean that is 1.2815 * below
the threshold criterion should have an exceedance rate of 10 percent over space and
time. Figure II-6 shows the CFD for the 10 percent frequency exceedance default
reference curve in black.
chapter
-------
15
Also plotted on this same axis in blue in Figure II-6 is a default reference curve
based on 10 percent of the area of the percent space x percent time (the default refer-
ence curve described previously and illustrated in Figure II-4). This evaluation was
undertaken given an approach to deriving and assessing attainment of numerical
chlorophyll a criteria is based largely on thresholds that should rarely be exceeded
in healthy populations (e.g., the 90th percentile). These two curves are very close in
shape, further supporting the use of the default reference based on a 10 percent area
under the curve. The EPA recommends use of the default reference curve, illustrated
in Figure III-4 and defined by Equation 1, when an applicable biologically-based
reference curve is not available.
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
II-6. Comparison of hyperbolic curves based on 10 percent of area under the curve
(blue) and 10 percent frequency exceedance (black).
< ',
Reference curves are more or less continuously defined while assessment curves
have relatively few discrete measures. Biological reference curves can contain
hundreds of points; the default reference curve has an infinite number of points. By
contrast, curves for three-year assessments of summer (June-September) monthly
means will have 12 data points with the curve defined by linear interpolation
between neighboring points. For this reason, it is possible for portions of the assess-
ment curve to be above the reference curve even without any measured point
exceeding the reference curve. This situation becomes more comprehensible by
chapter in « Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
16
understanding that reference curves typically have positive curvature and that this
curvature can dip below the line between consecutive points on the assessment
curve, causing a spurious, non-allowable exceedance.
To address this problem, the EPA recommends that reference curves be evaluated
only at the temporal axis points in the assessment curve as illustrated in Figure II-7.
For non-continuous biological reference curves, the points should be interpolated
from neighboring points. Appendix B provides a detailed description of the complete
Chesapeake Bay water quality criteria attainment assessment methodology.
Dissolved Oxygen DW Monthly new curve discrete points CB7PH 2002—2004
1.0
0.9
0.8
0.7(
E 0.6'
F
0 <
c 0.5
lo,
0.3
0.2
0.1
3 ja j a
D
\
|V.
\
r \
i \
V X.
- """"""•SN
Ox s%.!tv
^-V
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Fraction of Space
-O- assessment curve
""""' reference curve
-
-
_
-
_
_
-
0.9
0.8
0.7(
E 0.61
F
0 <
c 0.5
~
I 0.4
0.3
0.2
0.1
0.0
-\
\
- \
$ \
\ V,
- to- "\
- """V
- s X '***'»,
e- - - a s**'*^.^^^
0.8 0.9 1.0 Fraction of Space
-O- assessment curve
,„„,,,, reference curve
-
_
_
_
_
-
_
-
_
— r~— ._
0.8 0.9 1.0
II-7. The graph on the left (A) shows spurious non-attainment as the reference curve passes below the
assessment curve between points. The graph on the right (B) shows attainment as the reference and assessment
curves are evaluated at the same temporal axis points.
OF A
A statistical framework for making decisions on water quality criteria attainment
based on the CFD methodology would yield additional information on the certainty
of the attainment decisions. It would also help direct appropriate monitoring strate-
gies to reduce uncertainties. However, many theoretical obstacles remain in
developing such a framework. The CFD methodology is a new and innovative
approach to water quality criteria assessment. The relatively recent application of
this methodology to criteria assessment suggests that conducting further evaluations
and making improvements should prove constructive. The following section
discusses the steps in applying the CFD methodology.
1 —
One of the advantages of the CFD approach is that it can accommodate a variety of
input data and still arrive at the same assessment endpoint. Data collection methods
currently in place include: fixed-station data, cruise track data, continuous moni-
toring data, aircraft flight path data, and satellite imagery data. Because of the
chapter ii » Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
17
interpolation step, all of these data can be used with varying degrees of success to
estimate the total spatial distribution (to the limit of interpolator pixel size) of a water
quality parameter.
Interpolation can place data collected at various spatial densities on a common footing.
On the one hand, this capability is advantageous because data collected at different
spatial densities are available for the criteria assessment process. On the other hand, it
can be misleading to accept interpolated surfaces from different data sources as equiv-
alent without qualifying each interpolation with a measure of the estimation error
associated with each data type. Clearly, an interpolation based on hundreds of points per
segment (such as cruise track data) more accurately reflects the true non-attainment
percentage when compared to an interpolation based on two or three points per segment
(such as a fixed-station data). Of the various types of interpolation algorithms available
and reviewed, kriging is best positioned to address this issue (STAC 2006). Kriging
offers advantages over inverse distance weighting in that it provides the best assessment
of the estimation error associated with interpolation, but has not been implemented to
date. Other methods, such as interpolating polynomials, splines, and locally weighted
regression methods, should also be explored.
3— of
Depending on the interpolation method and the statistics available, it may be
possible to calculate the probability of exceedance of the temporal mean at each
point given the likely variance and the value(s) observed during the period. This step
is necessary to calculate probabilities in the following step.
4-
Determining the percent attainment of each grid cell from each interpolation seems
simple. If the estimated value for a grid cell is above (chlorophyll a) or below
(dissolved oxygen, water clarity) the criterion, then that cell is not in attainment.
While interpolation allows for standardization of many types of data, pointwise
attainment determination allows for standardization of many criteria. Because attain-
ment is determined at moments in time and points in space, it is possible to vary the
criterion in time and space. If different levels of a water quality constituent are
acceptable in different seasons, then the criterion can vary seasonally. It is possible
to implement different criteria over space for a segment that bridges, for example,
oligohaline and mesohaline salinity regimes. It might even be possible to let the
criterion be a continuous function of some ancillary variable such as temperature or
salinity, although this situation requires that such data exist for every interpolator
cell. The only requirement is that the final attainment determination be "yes" or "no"
for each interpolator cell.
chapter in » Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
18
Currently, limited pointwise attainment determination compliance has been im-
plemented. For example, the open-water 30-day mean dissolved oxygen criterion is
5 mgxliter1, except when the ambient salinity drops below 0.5 psu and the criterion
becomes 5.5 mgxliter1 (U.S. EPA 2003a). During the summer months, the open-
water designated use boundaries are selected based on local density conditions
reflecting stratification of the water column.
Even the simplicity of this concept diminishes when examining interpolation error.
Consider the assessment of two interpolator cells from an interpolation based on
cruise track data. While both interpolations could have the same value, each could
have a different level of error. Such different levels of error could mean that these
were different probabilities that the criteria were actually exceeded. For the simple
assessment of non-attainment, however, they count the same. Thus, one advantage of
a statistical framework is that it accounts for different levels of error throughout the
interpolation grid and these error levels could be incorporated into a single overall
assessment of attainment.
Stt; ; ~\ in
Computing a percentage should also be simple. The estimate is simply 100 times the
number of cells not in attainment divided by the total number of cells. As a rule, the
uncertainty of a binary process can be modeled using a binomial distribution. The
issue of uncertainty described in step 3 propagates into computing the percent of
attainment for a segment. In addition, estimated values for interpolator cells have a
complex dependence structure, ruling out a simple binomial model. The rules
governing the uncertainty of this step are also complex. The mathematics for
modeling this propagation of error are feasible, but have not yet been developed.
• "ne
While the CFD's percent-of-space coordinate provides a simple interpretation of the
percent of the spatial assessment unit that is out of attainment on a given date, the
percent-of-time coordinate is not simply the percent of time out of attainment at a
given point. Instead this coordinate is interpreted similarly to that of a cumulative
distribution function; it represents the percent of time that the associated spatial
percent of non-attainment is exceeded. For example, if the (percent space, percent
time) coordinates for a point on the CFD are (90, 10), the spatial percent of non-
attainment is greater than or equal to 90 percent about 10 percent of the time.
This step is very similar to computing an empirical distribution function, which is an
estimator of a cumulative distribution function. This similarity brings to mind
statistical inference tools associated with empirical distribution functions—the
Kolmogorov-Smirnov, Shapiro-Wilk, Anderson-Darling, or Cramer-von Mises — as
candidates for inference about the CFD (STAC 2006). These procedures model
uncertainty as a function of sample size only (in this case, the number of sample
dates). Since they do not account for uncertainty associated with the number of
chapter ii » Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
19
samples collected in space (i.e., number of sampling stations), this indicates that they
are to provide a statistical framework that truly accounts for error in Chesapeake Bay
water quality criteria assessments.
7 8—
When comparing the assessment curve to the reference curve, the issue of uncer-
tainty becomes most important. The preceding discussion clearly indicates that
uncertainty in the assessment curve represents an accumulation of uncertainty gener-
ated in and propagated through the preceding steps. If the reference curve is
biologically based, it is derived under the same system of error propagation. Devel-
oping the statistical algorithms to quantify this uncertainty poses a challenge.
Even if the uncertainty can be properly quantified, the issue of who gets the benefit
of doubt due to this uncertainty can prove difficult to resolve.
This problem of uncertainty in the regulatory process is widespread and not limited
to the CFD approach. Nonetheless, it must be dealt with. One option is to require that
the assessment curve be significantly above the reference curve to establish non-
attainment. This option protects the regulated party from being deemed out of
attainment due to random effects. If assessment CFD curves are not accurately deter-
mined, however, it could lead to poor protection of environmental health and
designated uses. A second option is to require that the assessment curve be signifi-
cantly below the reference curve to establish attainment. This option results in strong
protection of the environmental resource, but could lead to the regulated party imple-
menting unnecessary and expensive management actions.
Some compromise between these extremes is needed. The simplest compromise is
to ignore variability and compare the assessment curve to the reference curve. As
long as unbiased estimation is implemented for both the assessment curve and the
reference curve, this third option will result in roughly equal numbers of false posi-
tive (declaring non-attainment when, in fact, compliance exists) and false negative
(declaring attainment when, in fact, non-attainment exists) results. This last
approach is balanced and the one currently recommended by EPA. Under this
approach, however, no mechanism exists to motivate error reduction by improving
the data sets on which the criteria assessments are based.
OF TIIF
Beginning in fall 2005, the Chesapeake Bay Program's Scientific and Technical
Advisory Committee (STAC) appointed a panel of scientists to evaluate and refine
the CFD water quality criteria assessment methodology. Evaluations included tests
on the effects of: 1) sample densities in time and space; 2) varying levels of attain-
ment; and 3) varying degrees of spatial and temporal covariance. Appendix A
provides a complete copy of the panel's final report while Appendix C offers a narra-
tive evaluation of the options for spatial interpolation.
chapter in » Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
20
In general, the STAC panel analysis and review indicated that the CFD approach can
combine spatial and temporal data to support inferences on attainment or exceedance
of the water quality criteria (STAC 2006). The panel viewed the CFD approach as
innovative — one that has general application in water quality criteria assessments.
In comparison to other jurisdictional authorities, the Chesapeake Bay Program has
taken a lead in monitoring and assessment based upon scientific design (designated
uses) and emphasis on statistical evidence. Advancement in the CFD approach
should provide an important precedent for states outside the Chesapeake Bay region.
Because the CFD is both feasible and innovative, the panel felt that it qualifies as the
best available approach. On the other hand, the panel recognized that the approach
remains nascent and deserves further directed study and analyses to evaluate the bias
and imprecision that can occur due to small sample densities, non-independence in
temporal trends, and inadequate spatial interpolations.
The panel found that the CFD approach in its current form is feasible, but requires
additional research to further refine and strengthen it as a statistical tool. The CFD
builds on important statistical theory related to cumulative distribution functions; as
such, its statistical properties can be simulated and deduced. In its analyses, the
STAC panel showed that constructing confidence ellipses that support inferences
related to threshold curves or other tests of spatial and temporal compliance are
feasible. Understanding fundamental properties of how the CFD represents likely
covariances of attainment in time and space and how temporal and spatial correla-
tions interact with sample size effects require additional research. Further,
researchers must also analyze biases across regions and designated-use segments.
The panel expects that two to three years of directed research and development are
necessary to identify and measure potential sources of bias and imprecision for
criteria attainment determinations.
In the near future, the panel foresees that the CFD approach will prove particularly
powerful when linked to continuous spatial data streams through the cruise-track
monitoring program, and when able to utilize continuous temporal data generated
through further deployment of remote sensing platforms in the Chesapeake Bay
(e.g., Chesapeake Bay Observing System). These data sets will allow greater preci-
sion and accuracy in both threshold and attainment determinations made using the
CFD approach.
The STAC panel concluded that success of the CFD-based assessment rests upon
decision rules related to the biological reference curves. These curves represent
desired segment-designated use water quality outcomes and reflect sources of
acceptable natural variability (STAC 2006). The reference and attainment curves
should follow the same general approach in derivation: water quality data collection,
spatial interpolation, comparison to biologically based water quality criteria, and
combination of space-time attainment data through a CFD. Therefore, the biological
reference curves allow implementation of a tolerance threshold presuming the data
used to derive the reference curve were sampled similarly to the assessment curve.
chapter
-------
21
That is, the reference curve defines the degree to which criteria violations can be
tolerated without resulting in impairment of the designated use.
Bias and uncertainty are driven in CFD curves by sample densities in time and space.
Therefore, the STAC panel advised that similar sample densities be used in the deri-
vation of assessment and reference curves. As such densities are not always feasible,
additional analytical methods are needed to weight sampling densities equally
between attainment and reference curves.
APPLICATION OF THE CFD-BASED
ASSESSMENT METHODOLOGY
RECOMMENDATIONS FOR APPLICATION OF THE CFD-BASED
METHODOLOGY
As stated above, the CFD-based water quality criteria assessment methodology
offers the potential for significant benefit in accurately assessing Chesapeake Bay
water quality criteria attainment. As the STAC CFD Review Panel has indicated,
however, that the methodology is new and additional evaluations and refinements
should be performed (STAC 2006) (Appendix A). The EPA agrees with the panel's
conclusions, strongly supports the findings that further research is needed, and will
support those efforts in whatever way possible in the coming years. In the meantime,
the EPA recommends the following approach in undertaking Chesapeake Bay water
quality criteria assessments.
As described above, the Chesapeake Bay Program collects data at two different
scales for water quality criteria attainment assessment. In each case, the design of
data collection program focuses on assessments at a specific scale. The fixed-station
data are designed for segment and baywide assessments and the shallow-water moni-
toring data are designed to assess the small tidal tributaries and the Bay's
shallow-water habitats. Given the different scales, separate interpolations are likely
necessary using the most appropriate interpolation algorithm. The STAC CFD
Review Panel evaluated two possible options for spatial interpolation, recom-
mending kriging as the better of the two alternatives (STAC 2006). Kriging,
however, has not been fully developed for application in Chesapeake Bay water
quality criteria attainment assessment.
Until kriging is fully developed as an option for whole-Bay assessment based on the
fixed-station data, the EPA recommends that spatial interpolations continue using the
current Chesapeake Bay Program's inverse distance weighting (IDW) algorithm-
based interpolator (Appendix D). Spatial interpolation of the fixed-station data for
assessment of criteria attainment in the mainstem Bay and major tidal tributaries
requires several specific capabilities including: 1) the data must be interpolated in
three-dimensions (i.e., with depth); 2) the data must be interpolated into the tidal
tributaries and around bends in these tidal rivers; and 3) the interpolation needs to be
chapter ii • Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
22
automated to complete large number of criteria assessments efficiently and routinely.
These capabilities are not currently available using a kriging algorithm, but the
Chesapeake Bay Program IDW interpolator is designed with these capabilities in
mind. Thus, the EPA recommends that large-scale interpolations (segment, baywide)
continue to be based on the fixed-stations data be performed using the Chesapeake
Bay Program IDW interpolator. As kriging is developed further for use, this option
may be recommended in the future.
For the criteria assessment of small tidal tributaries and the Bay's shallow-water
habitats based on data from the shallow-water monitoring program, the EPA recom-
mends implementation of a kriging algorithm, where possible. The shallow-water
monitoring program yields data to assess criteria attainment in relatively few
systems at any one time. Thus, it is possible to provide the more focused evaluations
of individual interpolations that kriging requires. Furthermore, the intensive data
collection provided by the shallow-water monitoring program is particularly
conducive to detailed statistical analysis. To utilize the data's information fully, a
more thorough statistical interpolation procedure, such as kriging, should be imple-
mented. The shallow-water systems are highly dynamic and thus better characterized
by more intensive data collection combined with a more rigorous statistical interpo-
lation algorithm. For these reasons, the EPA recommends that kriging be
implemented, where possible, for criteria assessment based on shallow-water moni-
toring data.
Given the recommendation above, the EPA further advises that the states develop the
expertise to perform spatial interpolation based on statistical methods. Assessment
of the shallow waters will largely fall to the states, with some support from the
Chesapeake Bay Program Office. Guidelines are being developed for the use of
kriging in shallow-water criteria assessment. The procedure is detailed, however, and
requires expertise in geographic information systems, spatial statistics, and computer
programming. Questions remain about how best to implement kriging as an option
for spatial interpolation. The EPA plans to provide support through the Chesapeake
Bay Program Office to ensure that spatial interpolations based on kriging are
performed consistently for all shallow waters of the Bay when practical.
In general, most of the tidal waters of the Chesapeake Bay mainstem and major trib-
utaries remain impaired. This judgment was confirmed by the assessments
performed during the 2006 303(d) listing cycle and by listing decisions made prior
to that time. The 2006 assessments indicated that many of the assessment units were
far out of attainment with little need to confirm the conclusions through statistical
analysis. As restoration efforts proceed and more Bay tidal waters approach attain-
ment of their designated uses, then statistical procedures may become important to
ensure that waters are properly removed from the 303(d) list as soon as possible.
Given that it may require several years for the Bay to respond to management
actions, there is ample time to conduct the studies necessary to develop the required
statistical decision-making framework based on the CFD. The EPA recommends that
assessment of criteria attainment continue as in 2006 when the decision rule was that
chapter
-------
23
any criterion exceedance greater than that of the appropriate reference curve indi-
cates non-attainment of that criterion and, therefore, the designated use.
FOR OF
THE
As part of its conclusions, the STAC CFD Review Panel identified several critical
remaining issues requiring resolution in the near future (STAC 2006). The EPA
agrees with the recommendations for future development and advises that the Chesa-
peake Bay Program partners ensure that the work is completed in a timely,
appropriate manner.
The following list identifies some of the critical aspects requiring further research as
recommended by STAC (2006). See Appendix A for additional details.
1. Effects of Sampling Density on CFD Results. The CFD is a special case of
an unbiased estimator for a cumulative distribution function of a population.
Like the cumulative distribution function, the CFD is a function of the mean
and the variance of the population under assessment. The better the mean and
variance are characterized with sample data, the more accurate the shape of the
CFD. As the sampling density increases, the estimated CFD begins to approach
the true CFD. If the sampling density is low, however, then sampling error
could become important with the potential to affect the shape of the CFD and
ultimately the accuracy of the compliance assessment. Furthermore, the poten-
tial for sample size to affect the shape could create an attainment assessment
bias if the reference curve and assessment curve are based on different
sampling densities. Conditional simulation methods developed by the STAC
panel show promise in resolving these issues and mitigating potential biases
caused by sample size differences.
2. Choice of Interpolation Method, The STAC panel's research considered
several interpolation methods and outlined the features of each (Table C-l in
Appendix C). These features illustrated tradeoffs between ease of implementa-
tion and maximizing information garnered from the data. Further work is
needed to compare the features to the requirements of wide-scale implementa-
tion of Chesapeake Bay criteria assessment procedures and to formulate a plan
for tractable implementation that results in credible assessments. One strategy
is to implement easily performed analysis (e.g., IDW) as a screening tool to
identify cases for which attainment/non-attainment is clear, and then imple-
ment more labor-intensive methods (e.g., kriging) for cases in which
attainment is more difficult to resolve.
3. Three-Dimensional Interpolation. Assessments of the dissolved oxygen
criteria attainment requires three-dimensional interpolation. The field of three-
dimensional interpolation, however, is not as highly developed as that of
two-dimensional interpolation. Efforts are needed to evaluate research in three-
dimensional interpolation further and to seek additional outside scientific input
chapter ii * Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
24
and review to implement the best available technology for this aspect of criteria
assessment.
4. High-Density Temporal Data. As currently formulated, criteria assessment
for most of the Bay's open waters are based on "snapshots" in time of the
spatial extent of criteria exceedance estimated through interpolation. Data
collected for use in interpolation span several days given the large area being
sampled. New technologies should soon be capable of producing high-density
data in both space and time. Interpolation should accommodate data collected
densely in space. It is unclear, however, how the CFD process will accommo-
date data that are densely clustered in time. Further work is needed to evaluate
methods to fully utilize the temporally intensive data currently being collected.
5. Implementation and Review. As a rule of thumb, the best test of any new
procedure is putting it to work with stakeholder involvement. Through its
Criteria Assessment Protocols Workgroup, the Chesapeake Bay Program has
already established a forum for resolving the details of CFD implementation.
At appropriate intervals in this process, however, the Chesapeake Bay Program
should seek independent scientific and technical review of the implementation
status of the assessment methodology.
Helsel, D.R. and R.M. Hirsch.1992. Statistical Methods in Water Resources. Studies in Envi-
ronmental Science #49. Elsevier Science Publishers, Amsterdam, Netherlands. 552 pp.
Scientific and Technical Advisory Committee (STAC). 2006. The Cumulative Frequency
Diagram Method for Determining Water Quality Attainment: Report of the Chesapeake Bay
Program STAC Panel to Review of Chesapeake Bay Analytical Tools. STAC Publication 06-
003. 9 October 2006. Chesapeake Bay Program Scientific and Technical Advisory
Committee. Chesapeake Research Consortium, Edgewater, MD.
U.S. Environmental Protection Agency (U.S. EPA). 1997. Guidelines for Preparation of the
Comprehensive State Water Quality Assessments (305(b) reports) and Electronic Updates.
Assessment and Watershed Protection Division, Office of Wetlands, Oceans and Watersheds,
Office of Water, U.S. Environmental Protection Agency, Washington, D.C.
U.S. Environmental Protection Agency. 2003a. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and Its Tidal Tributaries.
EPA 903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2003b. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability. EPA 903-R-03-004. Region III Chesapeake Bay
Program Office Annapolis, MD.
Weisberg, S.B., J.A. Ranasinghe, D.M. Dauer, L.C. Schaffner, RJ. Diaz, and J.B. Frithsen.
1997. An estuarine benthic index of biotic integrity (B-IBI) for Chesapeake Bay. Estuaries
20:149-158.
chapter ii » Refinements to the Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
25
chapter|||
Application of Chesapeake Bay
Water Quality Criteria
Assessment Procedures
BACKGROUND
Beginning in the late 1990s and continuing through 2003, the Chesapeake Bay
Program partners developed new Chesapeake Bay water quality criteria designed
specifically to protect the ecological health of the Bay (U.S. EPA 2003a). Delaware,
Maryland, and Virginia, along with the District of Columbia, then adopted these
criteria and new tidal water designated uses into their water quality standards regu-
lations. The states' new Chesapeake Bay water quality standards were applied for the
first time in each state's 2006 Clean Water Act 303(d) listing cycle.
The four jurisdictions also adopted criteria assessment methods — published by EPA
in 2003 and in a 2004 addendum — into state water quality standards regulations (U.S.
EPA 2003 a, 2004a). The methods characterize the spatial and temporal variability of
the appropriate water quality parameters, while providing a clear basis for determining
whether a portion of the Bay's tidal waters reached attainment of the applicable desig-
nated use. Despite the methods' detail, technical limitations remained for their
complete application. This chapter and those that follow address many of the prior
technical limitations. Continued efforts to develop further refinements to the criteria
assessment methodology in specific areas, however, will likely remain.
In addition to the technical limitations, obstacles related to the states' transition from
an old set of water quality standards to the newer, more detailed Chesapeake Bay
water quality standards also existed. Differences occurred in the spatial extent of past
listing/delisting decisions. New water quality criteria components also exist that
have never been previously assessed. Furthermore, the mechanisms and processes
used to report listings in the past required updating to allow reporting based on the
states' new Chesapeake Bay water quality standards regulations. As with the tech-
nical limitations referenced above, an ongoing effort to refine and update the
methodology for making future listing decisions based on the new Chesapeake Bay
water quality standards will also be required (see Chapter 8 for further details).
chapter iii • Application of Chesapeake Bay Water Quality Criteria Assessment Procedures
-------
26
To assess attainment of the Chesapeake Bay water quality criteria, the spatial and
temporal extent over which they apply must be defined. The temporal extent is
defined implicitly for each component of the states' Chesapeake Bay water quality
standards. Described on page 150 in the 2003 EPA Chesapeake Bay water quality
criteria document (U.S. EPA 2003a) and adopted into the jurisdictions' water quality
standards regulations, the spatial extent is defined by the intersection of a Chesa-
peake Bay Program segment (U.S. EPA 2004b, 2005a) and each tidal water
designated use (U.S. EPA 2003b, 2004c). The spatial units defined by this intersec-
tion are referred to as "spatial assessment units." The intent is for each unit to be
assessed and listed independently on each jurisdiction's 303(d) list (part 1 through
part 5) (see Chapter 8 for further details).
The scale of the Chesapeake Bay spatial assessment units is large, with selection
based specifically on conditions in the Bay and on the factors affecting these condi-
tions. The Chesapeake Bay Program segments themselves were based on salinity
regimes, circulation patterns, and other natural physical features, but are generally
reflective of variations in water quality conditions and living resource communities
(U.S. EPA 2004b, 2005a). Thus, these segments serve as appropriate spatial units for
measuring the scope of water quality impairments in the Chesapeake Bay, its tidal
tributaries, and its embayments. They also work at a logical scale for developing
necessary management plans (TMDLs). Many of the water quality impairments
currently extend over large areas of the Bay and its tidal tributaries, so performing
assessments and reporting on these impairments at the segment scale are both appro-
priate. Developing management plans at this scale is also appropriate since multiple
jurisdictions often contribute to impairments.
Even though the scale of the spatial assessment units is suitable, in many cases it
varied from the scale of past tidal water quality criteria attainment assessments. The
change in scale introduced several challenges to the states as they implemented the
new Chesapeake Bay water quality criteria and tidal water designated uses. Bound-
aries of some previously established state assessment units were moved or shrunk to
address the spatial variability in some state water quality standards assessment meas-
ures. Furthermore, management decisions (e.g., listing certain waters as impaired,
developing TMDLs) had already been made based on the previously established
assessment units and were being implemented at the time the new Chesapeake Bay
water quality standards were adopted into state regulation. Thus, it was necessary to
establish procedures for transitioning to new spatial assessment units and relating
prior management decisions to new assessments that were sometimes defined at a
different spatial scale.
In general, the states could address the differences in boundary locations by making
small adjustments to state-defined spatial units. Primarily, adjustments consisted of
small changes in the boundaries of the previously state-defined assessment units to
chapter iii * Application of Chesapeake Bay Water Quality Criteria Assessment Procedures
-------
27
make them coincident with the boundaries of the
larger Chesapeake Bay Program segments. This
way, the smaller assessment units nest within the
larger ones and the larger-scale assessment results
can be attributed to each of the smaller units within.
The approach allows states to remain consistent
with previous listing decisions while accounting for
the broader designated-use-segment-assessment
results on their 303(d) lists.
In some cases, adoption of the new Chesapeake Bay
spatial assessment units represented a less detailed
and possibly less precise assessment of water-quality
criteria attainment. For example, Figure III-l illus-
trates Chesapeake Bay Program segment CB7PH,
which covers the southeastern portion of the main-
stem Chesapeake within Virginia. As is typical in
most of the Bay, the shoreline is extremely complex
with many small tidal rivers, creeks, and embay-
ments. These smaller tidal habitats may have different
water quality than the mainstem Bay section of the
segment due to different circulation patterns or land
uses or pollution sources that dominate local water
quality conditions. These smaller tidal habitats may
even have monitoring information that demonstrates
the differences in water quality conditions. In such a
case, it may make sense to separate the smaller tidal
river, creek, or embayment from the main assess-
ment unit by subdividing it to create a new smaller
spatial unit for separate assessment. Thus, the states have the option to "sub-segment"
larger units to characterize conditions in specific parts of the Bay, its tidal tributaries,
and embayments more precisely.
Allowing jurisdictions to subdivide the larger segments is consistent with national
EPA guidance and with EPA-published Chesapeake Bay water quality criteria assess-
ment guidance, which both provide specific considerations for sub-segmenting water
bodies for criteria assessment and listing decisions (U.S. EPA 2003a, 2005b).
Published EPA guidance states that waters can be partitioned "to represent homo-
geneity in physical, biological or chemical conditions." The EPA recommends that
jurisdictions use similar principles in deciding to subdivide the larger Chesapeake
Bay assessment units. A state's decision to sub-segment an existing segment should
be based on: 1) clear physical, biological, or chemical differences that can be docu-
mented; 2) homogeneity of water quality in the water body under consideration; and
3) confirmed future availability of monitoring data in the new sub-segment to provide
the capability to assess conditions and allow a determination regarding its 303(d) list
status. In all cases, there should be a priori knowledge of the conditions that support
a decision to subdivide, and preferably specific data that demonstrate how conditions
Figure 111-1. Segment CB7PH covering the southeastern
portion of the Chesapeake Bay in Virginia.
Source: U.S. EPA 2004b.
chapter iii • Application of Chesapeake Bay Water Quality Criteria Assessment Procedures
-------
28
differ in the area under scrutiny. Documentation of this information should be made
available for review as part of the 303 (d) listing cycle for which a new subdivided
segment is initially assessed. Jurisdictions need to ensure that any sub-segmentation
is fully consistent with their state's water quality standard regulations.
The EPA discourages states from subdividing segments simply to remove smaller
areas from an impaired waters list. Given the tidal exchange that occurs among all
segments, conditions in one segment can potentially affect adjacent segments. A sub-
segment that is prematurely removed from the impaired waters list might require
placement back on the impaired waters list in the next listing cycle due to adverse
conditions in the original segment.
Maryland and Virginia have already adopted specific sub-segments into their state's
water quality standards regulations in several tidal tributaries and embayments. The
2004 addendum to the 2003 Chesapeake Bay use attainability and designated-use
document contains detailed documentation supporting these state-defined, adopted
sub-segmentations (U.S. EPA 2004c).
USE IN BAY
To assess Chesapeake Bay water quality criteria attainment, the data used must prove
adequate. Consistent with the 2003 EPA Chesapeake Bay criteria assessment guid-
ance, the data should be of known quality and adequate quantity, as well as
representative of the tidal water designated use habitat under assessment (U.S. EPA
2003a). Documented QA/QC programs should ensure data quality; such documen-
tation should be publicly available for evaluation. A sufficient amount of data should
exist to provide a defensible degree of accuracy and precision given the expected
level of variability in the assessed tidal water body. The data should also be repre-
sentative of the spatial assessment unit as a whole so the resulting assessment is not
biased toward any one portion. While the EPA provides no minimum requirements
for each of these data characteristics, they should be maximized to the extent
possible to ensure that criteria assessments are scientifically defensible.
Opinions range broadly on the quantity of data required for criteria assessment. On
one extreme, some believe that sufficient data should be collected to capture all the
temporal and spatial variability to ensure that the criteria and designated uses are
attained in space and time. On the other extreme, some suggest that the state agency
manager should determine if a designated use is being attained based on available
information—even if it is anecdotal.
For the Chesapeake Bay and its tidal tributaries, the EPA recommends basing all
water quality criteria assessments on monitoring data. These data should be collected
over a three-year period immediately prior to the year of the listing cycle, unless
non-attainment is definitively established in less time (as described in Chapter 7).
Furthermore, the monitoring program for data collection should optimize quality,
quantity, and representativeness as described above.
chapter iii « Application of Chesapeake Bay Water Quality Criteria Assessment Procedures
-------
29
The Chesapeake Bay Program partners continue to fund and conduct an extensive
baywide, coordinated water quality monitoring program, much of which supports
water quality criteria assessment. Water quality monitoring takes place at more than
150 sites throughout the mainstem Chesapeake Bay and its tidal tributary waters
(Figure III-2). Samples are collected at each of the fixed stations on a monthly or semi-
monthly basis with data gathered since the mid-1980s (Chesapeake Bay Program
1989). The fixed-station network provides consistent data over the entire mainstem
Bay, major tidal tributaries, and larger embayments. The data are useful in assessing
the published Bay water quality criteria in the open-water, deep-water, deep-channel,
and migratory and spawning designated uses.
Use of the fixed-station network is limited for criteria assessments in the shallow-
water designated use habitats because the data scale is not appropriate. This network
also proves limited in many smaller tidal tribu-
taries and embayments, which have no or very few
stations. To address these limitations, the Chesa-
peake Bay Program partners developed a
Shallow-water Monitoring Program to provide
data collected intensively in space and time in the
Bay's shallow-water habitats. Chapter 7 describes
this program and the details of data application for
criteria assessment.
The 2003 EPA Chesapeake Bay water quality
criteria document describes the extent of data
collection needed to assess the state's Chesapeake
Bay water quality standards (U.S. EPA 2003a).
Three levels of effort are described for each crite-
rion: marginal, adequate, and recommended (see
pages 178-196 in U.S. EPA 2003a). The "mar-
ginal" level of monitoring is the minimum data
collection needed to support criteria assessment.
At this level, data may not be of the right type or
in sufficient quantity to assess all of the applicable
criteria components. In general, this level of
monitoring assumes that only the fixed-station
data are available for criteria assessment. The
"adequate" level of monitoring assumes that the
fixed-station monitoring program will be
combined with limited intensive data collection
(e.g., temporally continuous monitoring for
dissolved oxygen) to ensure that data are collected
to support the assessment of all the applicable
criteria components (e.g., 30-day, 7-day, and 1-
day means, instantaneous minimum) in some
spatial assessment units. The "recommended"
level of monitoring assumes that the fixed-station
Figure 111-2. Locations of the sites that make up the
fixed station network of the Chesapeake Bay Water
Quality Monitoring Program.
Source: Chesapeake Bay Program 1989.
chapter
Application of Chesapeake Bay Water Quality Criteria Assessment Procedures
-------
30
monitoring program will be combined with intensive data collection in all spatial
assessment units. Funding is not currently available to support monitoring at the
"recommended" level. The fixed-station monitoring is expected to continue into the
future, so data should be available at the "marginal" level for all spatial assessment
units. With the implementation of the Chesapeake Bay shallow-water monitoring
program in 2001, combined with a growing network of high-frequency observing
system deployed in the Bay tidal waters, monitoring will reach the "adequate" level
across all spatial assessment units with time.
To enhance the monitoring information from the coordinated Chesapeake Bay water
quality and shallow-water monitoring programs, jurisdictions are encouraged to
include data from other sources as appropriate. Consistent with the 2003 EPA-
published Chesapeake Bay water quality criteria assessment guidance, the states and
the District are encouraged to compile data from sources such as state and federal
monitoring agencies, local governments, universities, environmental organizations,
and citizen monitoring groups (U.S. EPA 2003a). Such data could prove significant
in enhancing the spatial coverage of the existing Chesapeake Bay water quality
monitoring program. The jurisdictions must ensure, however, that the data are appro-
priate for use in the Chesapeake Bay criteria attainment assessment methodology.
Data need to be of documented quality and adequate quantity as indicated above.
The jurisdictions also must ensure that the data are collected at an appropriate scale
and are representative of a given area or volume of a specific spatial assessment unit.
The Chesapeake Bay Program spatial interpolator uses data collected at all locations
and defines how much of that area or volume can be characterized by data from a
particular location (see Chapter 2 and Appendix D for details). Thus, a small tidal
embayment may be characterized by data from a single site. If that site is not located
properly (e.g., in a small creek, off a pier in shallow water, off a beach), the assess-
ment of the entire embayment may rest on potentially biased information. Similarly,
if data are collected intermittently at some sites, the spatial assessment unit may be
characterized inconsistently at times.
To use data collected through non-Chesapeake Bay Program monitoring programs in
Chesapeake Bay water quality criteria assessments, they must be merged with the Chesa-
peake Bay Program monitoring program data appropriately. The assumption is that these
water quality data were collected on different time (more infrequent) and space (well
away from the mid-channel river, mainstem) scales than the Chesapeake Bay Water
Quality Monitoring Program data. Therefore, these other data will be assigned a cruise
designation based on the monthly collection time so that they can be interpolated along
with the Chesapeake Bay Water Quality Monitoring Program data to generate the cumu-
lative frequency distribution (see Chapter 2 and Appendix B for details).
The states are encouraged to seek data from sources beyond the Chesapeake Bay
Water Quality Monitoring Program, but should use such data with care to avoid
biasing the assessment results for any particular portion of the tidal waters. Ideally,
the states would work with the collecting agencies and institutions in advance to
chapter iii » Application of Chesapeake Bay Water Quality Criteria Assessment Procedures
-------
31
ensure that the data are collected appropriately for use in interpolation and the
overall CFD-based criteria assessment methodology.
In addition to data collected by government and non-profit agencies, the states are
also encouraged to work with agencies, organizations, or other entities subject to
regulation, but with an interest in contributing data for use in the criteria attainment
assessment process. Such agencies may be able to provide additional monitoring
resources and significant amounts of supplementary data. Provided that an adequate
QA/QC program is in place to ensure that the data are accurate, representative, and
of known quality, these regulated agencies or entities may significantly benefit the
criteria assessment process.
The Hampton Roads Sanitation District in Virginia is one such example. The District
has worked with the Virginia Department of Environmental Quality and the Virginia
Institute of Marine Science to establish its own shallow-water monitoring program.
The Virginia Department of Environmental Quality can use the data generated by the
program to assess the state's dissolved oxygen, water clarity, and chlorophyll a
criteria in the lower tidal James River. Other similar organizations of regulated stake-
holders may also wish to provide similar data.
mr-
The criteria assessment methodology developed for the Chesapeake Bay water
quality criteria standards will require continued refinement into the future. The tech-
nical details of the methodology continue to be refined through research and
experience with application. This document describes many new refinements that
will assist the jurisdictions with their criteria assessment process and listing deci-
sions. More refinements are expected over the coming years. Furthermore, better
understanding is developing with time as more data are collected. New monitoring
programs (e.g., shallow-water monitoring) are offering new insight into the
processes that affect water quality conditions in the Chesapeake Bay. This enhanced
understanding will help fine-tune the requirements necessary for protection of the
Bay ecosystem. Given that continued refinements of the criteria assessment method-
ology are expected, it is recommended that the jurisdictions plan continued updates
to their Chesapeake Bay water quality standards regulations through their existing
triennial review process. The EPA commits to providing the information needed for
updating the states' water quality standards through publication of recommended
refinements to the criteria assessment procedures (such as in this addendum). The
publication of any future addendums to the 2003 Chesapeake Bay criteria document
will come in advance of the jurisdictions' triennial reviews for use in justifying
needed changes to the state's water quality standards regulations.
One example of the expected refinements to the criteria assessment methodology is
the development of a statistical basis for decision-making using the CFD (see
Chapter 2 and Appendices A and C for further details). Since the Chesapeake Bay
criteria assessment methodology was first published in 2003, interest has grown in
chapter iii « Application of Chesapeake Bay Water Quality Criteria Assessment Procedures
-------
32
developing an accounting of error in the assessment process. Research has been
underway over the past years to develop such a methodology. The technical details
are challenging, however, and research has not yet led to a solution. Progress has
occurred over the last year; a statistical framework could possibly be developed for
adoption into the state's water quality standards in upcoming 303(d) listing cycles.
Other refinements may be developed for monitoring programs and the interpolation
procedures. The EPA encourages states to prepare for adopting such refinements to
their criteria assessment procedures into future regulations.
Reference curves provide a second example of expected refinements. As more data
are collected, the capability for better defining the amount and pattern of criteria
exceedance that the system can withstand will continually improve. While major
changes to the reference curves are not expected, updating the reference curves with
additional data will improve the states' ability to assess Chesapeake tidal waters
accurately. With the prior agreement of the watershed jurisdictions, the EPA will
update the reference curves with new data and publish the revised curves in future
criteria document addenda. The jurisdictions will then need to adopt the new refer-
ence curves into their water quality standards regulations through their regular
triennial review processes.
Chesapeake Bay Program. 1989. Chesapeake Bay Monitoring Program Atlas-Volume 1:
Water Quality and Other Physio chemical Monitoring Programs. CBP/TRS 34/89. U.S. Envi-
ronmental Protection Agency Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency (U.S. EPA). 2003a. Ambient Water Quality Criteria for
Dissolved Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and Its Tidal Trib-
utaries. EPA 903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2003b. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability. EPA 903-R-03-004. Region III Chesapeake Bay
Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2004a. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a Chesapeake Bay and its Tidal Tributaries-2004
Addendum. EPA 903-R-04-005. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2004b. Chesapeake Bay Program Analytical
Segmentation Scheme: Revisions, Decisions and Rationales 1983-2003. EPA 903-R-04-008.
CBP/TRS 268/04. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2004c. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability'-2004 Addendum. EPA 903-R-04-006. Region III
Chesapeake Bay Program Office Annapolis, Maryland.
U.S. Environmental Protection Agency. 2005a. Chesapeake Bay Program Analytical
Segmentation Scheme: Revisions, Decisions and Rationales 1983-2003. 2005 Addendum.
EPA 903-R-05-004. CBP/TRS 278-06. Region III Chesapeake Bay Program Office,
Annapolis, MD.
U.S. Environmental Protection Agency. 2005b. Guidance for 2006 Assessment, Listing and
Reporting Requirements Pursuant to Sections 303(d) and 314 of the Clean Water Act.
Watershed Branch, Assessment and Watershed Protection Division. Office of Wetlands,
Oceans and Watersheds, Office of Water, U.S. EPA, Washington, D.C.
chapter iii * Application of Chesapeake Bay Water Quality Criteria Assessment Procedures
-------
33
chapter
Refinements to the Chesapeake
Bay Dissolved Oxygen Criteria
Assessment Procedures
BACKGROUND
In 2003, the EPA published detailed criteria for dissolved oxygen tailored to
different habitats within the Chesapeake Bay and its tidal tributaries (U.S. EPA
2003a) (Table IV-1). Oxygen is critical to most forms of life in the Bay; it must be
available in adequate concentrations to support overall ecosystem health. Minimum
concentrations of oxygen must be present to support the wide range of species
requiring protection as well as their various life stages.
Dissolved oxygen criteria were established for Chesapeake Bay that varied in space
and time to provide levels of protection for different key species and communities.
The criteria were also designed around several lengths of time to reflect the varying
oxygen tolerances for different life stages (e.g., larval, juvenile, adult) and effects
(e.g., mortality, growth, behavior). Thus, the dissolved oxygen criteria include
multiple components. Each component includes a target of dissolved oxygen
concentration, the duration of time over which the concentration is averaged, the
space (designated-use area) where the criterion applies, and a time (season, month)
when the criterion applies.
The dissolved oxygen criteria include 30-day, 7-day, and 1-day means along with an
instantaneous minimum. Each of these criteria components applies to a specific
season, such as the migratory spawning nursery period or the summer months (June
through September) or all-year round. Each also relates to one of four tidal-water
designated uses, according to the species and biological communities to be protected
(U.S. EPA 2003a, 2003c). The EPA published, and the states adopted into their water
quality standards regulations, dissolved oxygen criteria protective of migratory and
spawning, open-water, deep-water, and deep-channel designated-use habitats (U.S.
EPA 2003a) (Table IV-1).
chapter iv • Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
34
fa:'; e IV
Chesapeake Bay dissolved oxygen criteria.
Designated
Use
Migratory
fish
spawning
and
nursery use
Shallow-
water Bay
grass use
Open-water
fish and
shellfish use
Deep-water
seasonal fish
and shellfish
use
Deep-
channel
seasonal
refuge use
Criteria
Concentration/Duration
7 -day mean > 6 mg- liter"1
(tidal habitats with 0-0.5 ppt
salinity)
Instantaneous minimum > 5
mg-liter"1
Protection Provided
Survival/growth of larval/juvenile tidal-
fresh resident fish; protective of
threatened/endangered species
Survival and growth of larval/juvenile
migratory fish; protective of
threatened/endangered species
Open-water fish and shellfish designated-use criteria apply
Open-water fish and shellfish designated-use criteria apply
30-day mean > 5.5 mg-liter"
(tidal habitats with 0-0.5 ppt
salinity)
30 -day mean > 5 mg-liter"
(tidal habitats with > 0.5 ppt
salinity)
7 -day mean > 4 mg-liter"
Instantaneous minimum > 3.2
mg-liter"
30 -day mean > 3 mg-liter"
1-day mean > 2.3 mg-liter"1
Instantaneous minimum > 1 .7
mg-liter"1
Growth of tidal-fresh juvenile and adult
fish; protective of threatened/
endangered species
Growth of larval, juvenile and adult fish
and shellfish; protective of threatened/
endangered species
Survival of open-water fish larvae
Survival of threatened/endangered
sturgeon species1
Survival and recruitment of bay
anchovy eggs and larvae
Survival of open-water juvenile and
adult fish
Survival of bay anchovy eggs and larvae
Open-water fish and shellfish designated-use criteria apply
Instantaneous minimum > 1
mg-liter"1
Survival of bottom-dwelling worms and
clams
Open-water fish and shellfish designated-use criteria apply
Temporal
Application
February 1 -
May 31
June 1 -
January 3 1
Year-round
Year-round
June 1 -
September 30
October 1 -
May 31
June 1 -
September 30
October 1 -
May 31
1 At temperatures considered stressful to shortnose sturgeon (> 29EC), dissolved oxygen concentrations above an
instantaneous minimum of 4.3 mg-liter1 will protect survival of this listed sturgeon species.
chapter iv
-------
35
Assessing dissolved oxygen criteria attainment is challenging because of the
complexity of both the criteria and the Bay itself. To fully assess all the criteria
components, data need to be collected at a spatial intensity that adequately represents
the four designated-use habitats of Chesapeake Bay tidal waters at different times of
the year (U.S. EPA 2003c, 2004b). Similarly, data must be collected during all the
applicable seasons and at frequencies sufficient to address the various criteria dura-
tion components. The different dissolved oxygen criteria apply to different
designated-use areas and multiple criteria apply to the same designated-use area. The
dissolved oxygen criteria components also apply over different time periods to
protect species during critical life stages or during particularly stressful times of the
year. To fully assess each dissolved oxygen component in each designated-use
habitat over the appropriate time periods will require an extensive monitoring
program and a detailed assessment methodology. The Chesapeake Bay Program
currently conducts extensive water quality monitoring throughout the Bay tidal
waters and the EPA published a detailed dissolved oxygen criteria assessment
methodology with the new water quality criteria (Chesapeake Bay Program 1989;
U.S. EPA 2003a, 2004a). The existing Bay water quality monitoring was not suffi-
cient to cover all the criteria components, however, and some details in the
assessment methodology remain unresolved.
For the 2006 303(d) listing cycle, the states'
listing decisions were based primarily on
previous listings. Tidal waters that had been
listed as impaired in 2004 were not removed
from part 5 of their listing unless all the appli-
cable criteria components were shown in
attainment (see Chapter 8 for further details).
The Chesapeake Bay Program partners had the
capacity (data, assessment methodology) to
assess only the 30-day mean dissolved oxygen
criteria and, in some cases, the instantaneous
minimum dissolved oxygen criteria. The
remaining dissolved oxygen criteria were not
assessed because the existing water quality
monitoring programs and the published assess-
ment methodologies were inadequate for full
assessment. In most spatial assessment units,
the 30-day mean criterion was not attained and
those assessment units would have been listed
whether or not the other applicable dissolved
oxygen criteria were also assessed (Figure IV-
1). In many smaller tidal tributaries, however,
the 30-day mean criterion was attained and
those spatial assessment units were listed either
as "impaired" (part 5) due to previous listing or
DO Impairment
30 Day Mean
All Designated Uses
• Afiairwwitof Criteria
- msufftteni Data !o Assess
~ N/A
Figure IV-1. Listing status of the Chesapeake Bay open-water
designated use based on dissolved oxygen standards.
chapter iv • Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
36
as having "insufficient data to assess" (part 3). As nutrient loads are reduced and Bay
water quality improves, assessing the complete array of applicable dissolved oxygen
criteria to remove spatial assessment units from the "impaired" list will become
more critical.
Since Chesapeake Bay dissolved oxygen criteria were published in 2003, the capa-
bility of fully assessing all the dissolved oxygen criteria for all four designated uses
over all applicable time periods has progressed, but some limitations remain. The
refined and expanded dissolved oxygen criteria assessment methodologies docu-
mented in this chapter replace the methodologies previously published by U.S. EPA
(2003a, 2004a). Work should continue in refining these methodologies to reduce
uncertainty further and to increase confidence in the resulting assessments. Devel-
oping, validating, and publishing EPA-recommended methodologies for assessing
the full array of Chesapeake Bay dissolved oxygen criteria duration components will
also prove critical.
FOR
OF
To assess dissolved oxygen criteria attainment, the time span over which the criteria
apply must be clearly defined. In some cases, the temporal period is defined implic-
itly as part of the criteria. For example, the dissolved oxygen criteria protective of
the migratory fish spawning and nursery habitat designated use apply only to that
time of year when spawning fish (and the resultant eggs and early juveniles) require
higher dissolved oxygen levels compared to the rest of the year. In this example,
dissolved oxygen criteria attainment should be assessed over the entire spawning
season (February 1 through May 31) (U.S. EPA 2003a). Similarly, dissolved oxygen
criteria in the deep-water and deep-channel designated uses apply only during the
summer months — June 1 through September 30 — when the Bay stratifies and
naturally reinforces the potential for lower dissolved oxygen concentrations in
deeper waters. Therefore, assessment of dissolved oxygen criteria attainment in the
deep-water and deep-channel designated uses should also be performed over the
entire 4-month summer season (U.S. EPA 2003a). In all these cases, data are
collected over the entire criteria season in each of the three years of the assessment
period and these data are used to develop the cumulative frequency diagram (CFD)
for assessing dissolved oxygen criteria attainment (see Chapter 2 and Appendix B for
additional details).
Open water is the only tidal water designated use in which the dissolved oxygen
criteria apply year-round (U.S. EPA 2003a). In general, the Bay is most vulnerable to
low dissolved oxygen during the summer when temperatures are high, oxygen solu-
bility is low, and biological consumption of oxygen rises to its greatest level. Periods
of low dissolved oxygen can also occur during the rest of the year, sometimes caused
by high loading with subsequent slow consumption of organic material. The open-
chapter iv * Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
37
water dissolved oxygen criteria are designed to provide protection of open-water
habitat fish and shellfish communities at all times of the year. In spite of the year-
round application of these criteria, natural processes complicate the use of a single,
year-round assessment. Cooler temperatures affect the solubility of oxygen and allow
higher concentrations compared to similar organic loading conditions in warmer
months. Consequently, dissolved oxygen concentrations have a large natural vari-
ability range. Detecting human effects in the presence of that greater variability often
proves difficult. For this reason, as part of the dissolved oxygen criteria development
process, the EPA originally intended that the year-round open-water dissolved oxygen
criteria (see Table 111-10, page 66 in U.S. EPA 2003a) be assessed in each season (see
pages 150-151 in U.S. EPA 2003a). During the 2006 303(d) listing cycle, confusion
arose as to the appropriate time period for open-water dissolved oxygen assessment.
The criteria were clearly defined over the full annual cycle, but the stated intent was
to assess them on a seasonal basis. Furthermore, the 2003 EPA Chesapeake Bay
criteria document itself did not provide consistent guidance; it referred to assessment
on both an annual basis and a seasonal one (U.S. EPA 2003a).
Based on a re-evaluation of the underlying scientific basis for Chesapeake Bay
dissolved oxygen criteria, the EPA recommends that jurisdictions assess attainment
of the open-water dissolved oxygen criteria separately over two time periods:
summer (June 1 through September 30) and non-summer (January 1 through May 31
and October 1 through December 31). The open-water dissolved oxygen criteria
were largely derived to protect open-water species during the summer when elevated
temperatures, higher salinities, and naturally low dissolved oxygen levels occur
(U.S. EPA 2003a). Given that summer is a critical period for many species, it should
be assessed separately. The potential for dissolved oxygen impairments are lower in
the non-summer period due to greater natural dissolved oxygen solubility and lower
biological oxygen consumption—both due to lower water column temperatures.
Nevertheless, low dissolved oxygen levels sometimes occur during other times of the
year making a separate dissolved oxygen criteria assessment necessary for the non-
summer period. The separate criteria assessments for summer and non-summer
seasons will support year-round protective dissolved oxygen concentrations in the
open-water designated-use habitats.
The open-water designated-use boundary is explicitly defined as including "tidally
influenced waters extending horizontally from the shoreline to the adjacent shore-
line" (see page 71 in U.S. EPA 2003c). Further, on page 68, the U.S. EPA (2003c)
states that:
The shallow-water bay grass designated use is intended specifically to delin-
eate the habitats where the water clarity criteria would apply. The
chapter iv * Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
38
open-water fish and shellfish designated use and the accompanying dis-
solved oxygen criteria will fully protect the biological communities
inhabiting shallow-water habitats. The open-water designated use extends
into the intertidal zone and protects shallow-water organisms beyond under-
water bay grasses.
Unless a state has specifically delineated a sub-segment within a segment, attainment
of the open-water designated use will be based on dissolved oxygen criteria attain-
ment for the entire volume of the open-water designated use within the segment.
Neither the need nor the requirement exists for a separate assessment of dissolved
oxygen criteria attainment strictly within shallow waters (0-2 meters in depth). The
importance of acquiring better temporal and spatial coverage of dissolved oxygen
conditions in these shallow-depth habitats is not diminished however, since condi-
tions in these areas vary greatly from the open water of the mid channels where the
fixed stations are located. Shallow-water monitoring will provide the data needed to
characterize dissolved oxygen conditions in shallow-water habitats more fully (see
Chapter 7 for further details).
OF
Historically, the Chesapeake Bay Water Quality Monitoring Program consisted
primarily of fixed-station monitoring conducted on a monthly or twice-monthly
basis (Chesapeake Bay Program 1989). This sampling design was primarily intended
to assess long-term trends in water quality and the status of living resources,
capturing variability over decadal, annual, and seasonal time scales. The fixed-
station monitoring was adapted to assess the 30-day mean dissolved oxygen criteria
to measure dissolved oxygen throughout the Bay and its tidal tributaries and embay-
ments. This system ensures at least one set of measurements for each month.
The individual monthly estimates are considered accurate, although imprecise, since
the sample sizes are small (n = 1 or 2). This imprecision is likely to be mitigated by
the many estimates of monthly means (e.g., multiple months over the 3-year assess-
ment period), which are combined into each single assessment of criteria attainment
(see Chapter 2 and Appendix B for additional details). The monthly and twice-
monthly fixed-station data are not adequate to assess attainment of the 7-day and
1-day mean dissolved oxygen criteria directly because the sampling frequency rests
outside the defined time intervals and is unable to capture the short-term variability.
For the 2006 303(d) listing cycle, only three of the dissolved oxygen criteria compo-
nents were assessed. The 30-day mean open-water criterion was determined in all of
the assessment units of Chesapeake Bay using the fixed-station data and the CFD
assessment methodology. In spatial assessment units where deep-water and/or deep-
channel designated uses exist, the 30-day mean deep-water criterion and the
chapter iv « Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
39
instantaneous minimum deep-channel criterion were also determined using the
fixed-station data.
The rationale behind the assessment of the instantaneous minimum deep-channel
criterion was based on the long-term fixed station data record in the deep-channel
locations which shows that dissolved oxygen does not vary strongly through time in
the deep channel during the summer months because of the physical isolation from
the atmosphere and the photic zone. Dissolved oxygen concentrations remain rela-
tively constant; therefore, a 30-day mean should be similar to any instantaneous
measure (see section below).
No assessments were made of the 7-day and 1-day mean dissolved oxygen criteria
because the data were considered inadequate (as described above). In most cases,
this situation did not affect listing decisions because many spatial assessment units
did not attain the 30-day mean criterion (see Figure IV-1) and all criteria components
need to be attained to justify removal from the impaired list (part 5). The 30-day
mean criterion was attained in some cases. These spatial assessment units, if not
previously listed on part 5, were placed in part 3 of the states' lists for waters with
insufficient data (see Chapter 8 for further details). As water quality conditions
improve in Chesapeake Bay, a method to assess higher frequency dissolved oxygen
criteria will be needed so that spatial assessment units in attainment with all appli-
cable dissolved oxygen criteria components can be removed from the state's
impaired waters list (see Appendix E).
Until the EPA publishes methodologies for assessing the 7-day mean, 1-day mean
and instantaneous minimum open-water and deep-water dissolved oxygen criteria
components, the agency recommends that the states rely strictly on the assessment
of the 30-day mean open-water and deep-water dissolved oxygen criteria for listing
decisions. For those open-water and deep-water designated-use segments in which
the 30-day mean criteria are not in attainment, the jurisdictions should list the desig-
nated-use-segment on part 5 as impaired in the absence of data and/or methodologies
for assessing the remaining criteria components. For those designated-use segments
in which the 30-day mean criteria are in attainment, the jurisdictions should generate
additional data and apply the criteria assessment procedures to assess attainment of
the 7-day mean, 1-day mean, and instantaneous minimum criteria components.
DISSOLVED OXYGEN REFERENCE CURVES
SUMMER OPEN-WATER AND DEEP-WATER DISSOLVED OXYGEN
CRITERIA REFERENCE CURVES
Reference curves for both the 30-day mean open-water (June 1-September 30 only)
and 30-day mean deep-water dissolved oxygen criteria were based on criteria levels
that would not impair biological communities (U.S. EPA 2003a). Reference areas for
chapter iv • Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
40
derivation of the 2003 published deep-water reference curves were identified using
a measure of benthic community health—the Chesapeake Bay benthic index of
biological integrity or benthic-IBI (Weisberg et al. 1997). Sessile benthic communi-
ties are good indicators of the water quality of the overlying waters. Although
relatively tolerant of lower oxygen concentrations, a dissolved oxygen concentration
of 2 mg-liter1 is the threshold below which benthic infaunal communities become
severely stressed (numerous references cited in Chapter 3 of U.S. EPA 2003a). A
healthy benthic community, therefore, could indicate allowable time and space
exceedances of the dissolved oxygen criteria that will not impair the biological
community.
Benthic infaunal community samples are collected as part of the long-term Chesa-
peake Bay Benthic Monitoring Program at fixed and random locations during the
summer, usually in August to September. If the benthic-IBI of that sample is "good,"
(in this case 3 or more on a scale of 1 to 5), dissolved oxygen conditions were likely
adequate for the previous one to two months (Dauer et al. 2005).
In order to ensure greater consistency in deriving the open-water and deep-water
reference curves, factor in the state-adopted designated-use boundaries and take
advantage of a full two decades on monitoring data, both reference curves were
updated. To develop updated open-water and deep-water reference curves, the
monthly fixed and random station locations for the benthic-IBI data from 1985 to
2005 were matched with the monthly open-water and deep-water designated-use
boundaries for the same time period. This updated approach differs from the original
method published by the EPA (2003a), which used a single designated-use boundary
coverage for the entire data record. An additional difference is that previously this
method was used to define only the deep-water reference curve. The open-water
reference curve was based on an analysis in which "good" water quality conditions
were defined for reference segments by year (see Appendix H in U.S. EPA 2003a).
Reference locations were identified by sorting the resulting data set by year,
segment, and designated use. If a designated use in a given segment in a given year
had only "good" benthic-IBI scores (>3), then the dissolved oxygen data for that
segment, designated use, and summer period (June-September) can be used to
compute a reference curve. Appendix F lists these use-segment-year combinations.
Separate CFDs were generated for open-water and deep-water designated-use habi-
tats from the entire data set of summer dissolved oxygen data from all reference
locations over the 1985-2005 data record. Figures IV-2 and IV-3 respectively illus-
trate the resultant June-September open-water and deep-water dissolved oxygen
criteria reference curves. Appendix G documents the equations for the reference
curves.
chapter iv » Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
41
Percent of Time ^
0%
Open Water Monthly Dissolved Oxygen Biological Reference Curve
'••
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Volume
IV-2, Chesapeake Bay open-water 30-day mean dissolved oxygen criterion
biological reference curve applicable only during the June 1 through September 30
assessment period.
100% -
90%
80%
70%
d>
Efin% -
P
"O cno/
C
0>
o
Q.
ono/i.
m%
no/, _
Deep Water Monthly Dissolved Oxygen Biological Reference Curve
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Volume
IV-3, Chesapeake Bay deep-water 30-day mean dissolved oxygen criterion
biological reference curve.
-------
42
The default reference curve, illustrated in Figure II-4 in Chapter 2, should be used in
the assessment of the 30-day mean, open-water dissolved oxygen criteria during the
non-summer months (January 1 through May 31 and October 1 through December
31). The necessary biological indices and data were not available to support deriva-
tion of a biologically based reference curve for open-water habitats during the
non-summer months.
0)
F
H
o
c
01
o
Q_
100%'
90%-
ono/ 1
50%.
40%',
in%-
20%-
10%-
0
Deep Channel Dissolved Oxygen Biological Reference Curve
' 1
i««Hafi
v^
i**ia^f«''-**MiiS.s,MB,
v^"iffl'™'™«.,.«,,jav,w&_mi ^
% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Volume
IV-4, Chesapeake Bay deep-channel dissolved oxygen criterion biological reference
curve.
ur DJ
The April 2003 Chesapeake Bay water quality criteria document provides conflicting
guidance in the use of reference curves for assessing attainment of the four instanta-
neous minimum dissolved oxygen criteria. Pages 170 to 173 in U.S. EPA 2003a
display and discuss reference curves for migratory spawning and nursery, open-
water, deep-water, and deep-channel criteria attainment assessment. All four sets of
designated-use specific criteria include a use-specific instantaneous minimum crite-
rion. With the exception of the deep-channel criteria (page 173 in U.S. EPA 2003a),
none of these sections specifically describe whether a reference curve should be
chapter iv « Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
43
applied in assessing attainment of the respective instantaneous minimum criteria.
The reader is left with the sense that the published reference curves should be
applied to all the dissolved oxygen criteria, regardless of the stated duration.
All four instantaneous minimum criteria for protection of the four designated uses—
migratory spawning and nursery, open-water, deep-water, and deep-channel—protect
against mortality from very short-term exposure to low dissolved oxygen concentra-
tions (U.S. EPA 2003a). The other dissolved oxygen criteria with specific averaging
periods (30-day, 7-day, and 1-day means) protect against impairments—including
growth, respiration, and behavioral/avoidance—for which the impairments will not
impact the designated use. The 2003 EPA criteria guidance stated that there were no
"biologically acceptable exceedances of the applicable criteria" for the instantaneous
minimum criteria, given that the impairment is death (page 151 in U.S. EPA 2003a).
While updating the methodology for deriving the open-water and deep-water desig-
nated-use dissolved oxygen criteria reference curves for the 30-day mean criteria
(described above), there were times and locations in the Chesapeake Bay for which
healthy benthic infaunal communities still existed despite exceedance of the 1
mg-liter1 instantaneous minimum criterion. The EPA recommends, therefore, that
attainment assessment of the instantaneous minimum deep-channel dissolved
oxygen criteria be conducted with the CFD methodology using the deep-channel
biological reference curve (Figure IV-4; Appendices F and G).
USE OF AS
Several Chesapeake Bay scientists have called for future publication of dissolved
oxygen criteria based on percent saturation (not concentration) and for state adop-
tion of such percent-saturation-based criteria into the states' water quality standards
regulations. They cite fisheries physiology research showing that the pressure
gradient between the surrounding water and the blood running through the fishes'
gills that truly determines whether sufficient oxygen exists in the water to support
aquatic life. For example, Dutil and Chabot (2001) reported:
Fishes have developed several mechanisms to secure more oxygen from their
environment in critical situations such as low oxygen availability (Hoar and
Randall 1984). When the partial pressure of oxygen in the environment
drops below some critical limit, however, the pressure gradient between
blood and water may not allow the fish to deliver as much oxygen to its
tissues as needed to meet metabolic requirements associated with ingestion,
digestion, growth and activity. Thus critical thresholds may vary through
time in demersal fish species and are best described in terms of partial pres-
sure of oxygen or percent saturation.
chapter iv * Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
44
These scientists also note that the amount of dissolved oxygen dissolved declines as
temperature and salinity increase. For example, fully saturated freshwater at 20°C
holds 9.28 mg-liter1 of oxygen, but fully saturated seawater at the same temperature
only contains 7.58 mg-liter1 of oxygen. Seawater at 1°C can hold 11.38 mg-liter1 of
oxygen; at 30°C it can hold only 6.37 mg-liter"1 oxygen. As for the aquatic organ-
isms, research indicates that percent saturation drives the oxygen diffusion supplying
their respiratory demands.
Concentration-based, not percent-saturation-based, criteria were published given the
lack of reporting dissolved oxygen concentrations in terms of percent saturation in the
extensive effects database used to derive the Chesapeake Bay dissolved oxygen criteria
(U.S. EPA 2000). In addition, the lack of salinity and temperature values for each data
point in the laboratory-based low dissolved oxygen effects database prevented calcula-
tion of the concentration-based effects data into percent saturation numbers.
Following publication of the Ambient Aquatic Life Water Quality Criteria for Dissolved
Oxygen (Saltwater): Cape Cod to Cape Hatteras, EPA scientists evaluated the implica-
tions of recommending dissolved oxygen criteria as percent saturation versus
concentration (U.S. EPA 2000). In an addendum to the 2000 Virginian Province salt-
water dissolved oxygen criteria document, the U.S. EPA (2003b) reported:
A standard based on percent saturation has a wide range of differences in
partial pressure (2.14-4.01 Torr), which decreases with increasing temper-
ature. The opposite is more desirable, however, since respiratory demand
increases with temperature. Thus standards based on percent saturation are
likely to overprotect during winter and potentially underprotect in summer,
when organisms need the most oxygen. A standard based on concentration
provides a more uniform difference in partial pressure over the temperatures
used (2.45-2.72 Torr). Even though the range of difference is smaller, it still
increases with temperature. Thus a standard based on absolute concentra-
tion is more likely to create stable physiological conditions for animals
throughout the year.
Scientists from the EPA have generated a version of the EPA Virginian Province salt-
water dissolved oxygen criteria as percent saturation for the State of Maine (G.
Thursby, personal communication). At this time, however, the EPA does not have the
scientific basis to recommend a set of Chesapeake Bay dissolved oxygen criteria in
terms of percent saturation.
chapter iv » Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
45
Chesapeake Bay Program. 1989. Chesapeake Bay Monitoring Program Atlas - Volumel:
Water Quality and Other Physio chemical Monitoring Programs. CBP/TRS 34/89. U.S. Envi-
ronmental Protection Agency, Chesapeake Bay Program Office, Annapolis, MD.
Dauer, D.M., M.F. Lane, and RJ. Llanso. 2005. Addendum to the Report: Development of
Diagnostic Approaches to Determine Sources of Anthropogenic Stress Affecting Benthic
Community Condition in the Chesapeake Bay. Prepared for U.S. Environmental Protection
Agency, Chesapeake Bay Program Office, by Department of Biological Sciences, Old
Dominion University, Norfolk, VA, and Versar, Inc., Columbia, MD.
Dutil, J. D. and D. Chabot. 2001. Impact of hypoxia on Atlantic Cod in the Northern Gulf of
St. Lawrence, p. 51-60 In R.V. Thurston (Ed.), Fish Physiology, Toxicology, and Water
Quality. Proceedings of the Sixth International Symposium, La Paz, Mexico, January 22-26,
2001. U.S. Environmental Protection Agency Office of Research and Development, Ecosys-
tems Research Division, Athens, GA. EPA/600/R-02/097. 372 pp.
Hoar, W.S. and DJ. Randall. 1984. Fish Physiology. Volume X. Gills. Part A. Anatomy, Gas
Transfer, and Acid-Base Regulation. Academic Press, 456 pp.
U.S. Environmental Protection Agency (U.S. EPA). 2000. Ambient Aquatic Life Water
Quality Criteria for Dissolved Oxygen (Saltwater): Cape Cod to Cape Hatteras. EPA 822-
R-00-012. U.S. Environmental Protection Agency, Office of Water, Washington, D.C.
U.S. Environmental Protection Agency. 2003a. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and its Tidal Tributaries.
EPA 903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2003b. National Saltwater Criteria for Dissolved
Oxygen: Potential Addendum to Virginian Province Saltwater Criteria for Warmer and
Colder Waters. AED-03-113. U.S. Environmental Protection Agency, Office of Research and
Development, National Health and Environmental Effects Laboratory, Atlantic Ecology Divi-
sion, Narragansett, RI.
U.S. Environmental Protection Agency. 2003c. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability. EPA 903-R-03-004. Region III Chesapeake Bay
Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2004a. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a Chesapeake Bay and its Tidal Tributaries - 2004
Addendum. EPA 903-R-04-005. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2004b. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability - 2004 Addendum. EPA 903-R-04-006. Region III
Chesapeake Bay Program Office, Annapolis, MD.
chapter iv • Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
46
Weisberg, S.B., J.A. Ranasinghe, D.M. Dauer, L.C. Schaffner, RJ. Diaz, and J.B. Frithsen.
1997. An estuarine benthic index of biotic integrity (B-IBI) for Chesapeake Bay. Estuaries
20: 149-158.
chapter iv • Refinements to the Chesapeake Bay Dissolved Oxygen Criteria Assessment Procedures
-------
47
chapter
Refinements to the
Shallow-Water Designated-Use
Assessment Procedures
BACKGROUND
Submerged aquatic vegetation (SAV) is a critically important component of the
Chesapeake Bay ecosystem. These underwater plants provide habitat used by many
fish and shellfish species and provide food for migratory waterfowl, while also
improving water quality by generating oxygen, stabilizing sediment, and taking up
nutrients. Historically, the Chesapeake Bay was once known for its extensive SAV
beds. During the 1960s, however, much of the SAV disappeared. Poor water clarity,
caused by excessive algal growth and high levels of suspended sediment (Dennison
et al. 1993), was the primary factor in the decline of these beds. Both of these water
quality impairments result from human activities in the Chesapeake watershed that
cause excessive nutrient and sediment loadings to the Bay.
In 2003, after consultation with the watershed jurisdictions, the EPA published water
clarity criteria, SAV restoration goals, and shallow-water Bay grass designated-use
delineations for the Chesapeake Bay as well as its tidal tributaries and embayments
(U.S. EPA 2003a, 2003b). When applied as state water quality standards regulations,
these standards define the water clarity needed in delineated shallow-water habitats
to support SAV restoration to agreed-upon acreages.
The water clarity criteria and SAV restoration goals were designed to define attain-
ment of the shallow-water Bay grass designated use in three ways (U.S. EPA 2003a,
2004a). First, once the targeted acreage of SAV in a given segment is reached, that
segment is considered in attainment of the shallow-water Bay grass designated use.
Measurement of SAV goal restoration attainment is based on annual aerial surveys
in which the beds are photographed and mapped, acreages quantified, and the single
best year of acreage determined. Second, if sufficient shallow-water area with the
water clarity necessary to achieve restoration of the targeted SAV acres exists, then
the segment is considered in attainment. These "water clarity acres" are measured by
routinely mapping water clarity using data from the Chesapeake Bay Shallow-water
Monitoring Program (see Chapter 7 for details). Third, if the water-clarity criteria
were attained throughout the shallow-water designated use reaching to a specific
chapter v • Refinements to the Shallow-Water Designated-Use Assessment Procedures
-------
48
depth contour (segment-specific water clarity criteria application depth) based on the
cumulative frequency diagram (CFD) assessment methodology, then the segment is
also considered in attainment of this designated use. Like the water clarity acres
approach, the CFD-based assessment would be performed using data from the
shallow-water monitoring program (see Chapter 7 for details).
For the 2006 Impaired Water 303(d) listing cycle, insufficient data existed to assess
water clarity criteria attainment in nearly all of the Chesapeake Bay segments'
shallow-water bay grass designated-use habitats. The SAV acreages have been quan-
tified for many years (annually since 1984), however, and this data collection is
expected to continue. Thus, the 2006 assessments used SAV acreages over the three-
year assessment period from 2001 through 2004. If the single best year of SAV
coverage from that period exceeded the established, state-adopted SAV restoration
goal, then the segment's shallow-water designated use was deemed in attainment. If
the SAV restoration goal was not attained, then the segment's shallow-water desig-
nated use was listed either as impaired (category 5) or as insufficient data (category
3) since shallow-water monitoring data were unavailable for the segment.
The procedures for assessing attainment of the Chesapeake Bay shallow-water
designated use using the water clarity criteria and SAV restoration acreages, first
published by EPA in 2003, were broadly defined and had not been extensively
applied in the Chesapeake Bay prior to the 2006 303(d) listing cycle (U.S. EPA
2003a, 2003b). The jurisdictions and the EPA identified and resolved many issues
during the first baywide application. This chapter provides detailed and refined guid-
ance on the assessment of the water clarity criteria and the SAV restoration goals.
Ultimately, the chapter evaluates attainment of the shallow-water bay grass desig-
nated use. This guidance replaces the applicable criteria assessment methodologies
previously published by the U.S. EPA (2003a, 2003b, 2004a, 2004c).
The shallow-water bay grass designated use is considered in attainment if sufficient
acres of SAV are observed within the segment or enough acres of shallow-water
habitat meet the applicable water clarity criteria to support restoration of the desired
SAV acreage for that segment (U.S. EPA 2003a, 2003b). Assessment of either
measure, or a combination of both, serves as the basis for determining attainment or
impairment of the shallow-water bay grass designated use.
Given SAV is the ultimate biological measure of attainment of the designated use, in
the absence of sufficient shallow-water monitoring data necessary to determine the
available water clarity acres or assess water clarity criteria attainment using the
CFD-based criteria assessment procedure, the EPA recommends the States assess
shallow-water bay grass designated use attainment/impairment based on the acres of
aerial mapped SAV.
chapter v * Refinements to the Shallow-Water Designated-Use Assessment Procedures
-------
49
If a shallow-water bay grass designated use segment meets its SAV restoration
acreage, that designated use-segment is considered in attainment of the designated
use and should be listed on part 2.
If a shallow-water bay grass designated use segment does not meet its SAV restora-
tion acreage and sufficient shallow-water monitoring data is available, the
jurisdiction can then assess attainment of the designated use using water clarity acres
or water clarity criteria as described below. If the water clarity acres/water clarity
criteria are met/attained based on assessment of spatially intensive shallow-water
monitoring data, then that designated use-segment is considered in attainment of the
shallow-water bay grasses designated use and should be listed on part 2.
If the water clarity acres/water clarity criteria are not met/attained based on assess-
ment of shallow-water monitoring data, or if there is insufficient data to make a
determination using water clarity acres, then that designated use-segment is consid-
ered not in attainment of the shallow-water bay grasses designated use and should be
listed on part 5.
For those segments that contain the shallow-water bay grass designated use, attain-
ment of this use should be assessed with the following procedure:
If the segment's single best year SAV acreage (described below) drawn
from the most recent three-year period of available data is equal to or greater
then the state adopted SAV restoration acreage for that segment, then that
segment is considered to be in attainment of its shallow-water bay grass
designated use. If the segment's single best year SAV acreage is less than
the state adopted SAV restoration acreage for that segment, the state should
then proceed to assess water clarity acres if sufficient shallow-water data is
available, otherwise, the segment is not in attainment.
If the segment's water clarity acres (defined below) calculated from the
most recent three-year period of available shallow-water monitoring water
clarity data is equal to or greater than state adopted water clarity restoration
acreage for that segment, then that segment is considered to be in attainment
of its shallow-water bay grass designated use. If the segment's water clarity
acres are less than the state adopted water clarity restoration acreage for that
segment, then that segment is considered not to be in attainment of its
shallow-water designated use unless SAV acreage data indicate attainment.
A jurisdiction may also choose to apply the CFD-based assessment of water
clarity criteria, described in more detail below, in place of water clarity acres, to
assess attainment of the segment's shallow-water bay grass designated use.
Given that SAV is the ultimate biological measure of attainment of the designated
use, the EPA recommends a specific sequence of criteria assessment: assessment of
SAV acres meeting the segment-specific restoration acres first, followed by assess-
ment of water clarity acres or water clarity criteria attainment. In the absence of
sufficient shallow-water monitoring data to determine the available water clarity
acres or assess water clarity criteria attainment using the CFD-based procedure, the
chapter v • Refinements to the Shallow-Water Designated-Use Assessment Procedures
-------
50
EPA recommends that the states assess shallow-water bay grass designated-use
attainment based on the acres of mapped SAV (see Chapter 8).
A5SI-;SSM!>J. " -.'..- >; .. >' M THE SIM :• - - K "' '., -r'.K •' '•! >;W
Bay wide and segment-specific SAV restoration goals were defined for the Chesa-
peake Bay by evaluating the historical (1930s-1970s) and more recent (1980s-2000)
SAV distributions (U.S. EPA 2003b). Historical aerial photographs, available for
selected years in the 1930s, 1950s, and 1960s, were converted to digital maps. Then
acreages of SAV for all photographed shallow-water areas in Chesapeake Bay, its
tidal tributaries and embayments were quantified. To set restoration goals for the
Bay, the single best year of SAV coverage was defined as the restoration goal for
each segment. The combined individual restoration acreages yielded a baywide goal
of 185,000 acres. (See pages 105-122 in U.S. EPA 2003b for detailed documenta-
tion on the entire goal-setting process.)
This baywide restoration goal was established "to reflect the historical abundance,
measured as acreage and density from the 1930's to present" as committed to in the
Chesapeake 2000 agreement (Chesapeake Executive Council 2000) and essentially
represents the "existing use" as defined by the Clean Water Act. The single-best-year
approach was necessary because a common basis was needed to define the area of
SAV that should be present. The historical photography was not consistent through
time for all areas of the Bay and SAV acreages varied through time. Since at least
some coverage was available for most of the Bay, the single best year offered the best
option for setting goals (in selected cases with little or incomplete historical data, a
composite of multiple years of historical data was used to define the "single best
year") (U.S. EPA 2003b).
Because the segment-based SAV restoration goals were established based on the prin-
ciple of a "single best year," the assessment of attaining that goal within an individual
Chesapeake Bay Program segment's shallow-water bay grass designated-use habitat is
defined in a similar manner. Attainment of the SAV restoration goal is reached when
the single best year of SAV acreage during the applicable preceding three-year assess-
ment period equals or exceeds the established goal (defined as "SAV restoration acres"
in the states' water quality standards regulations) for that segment.
In nine segments, SAV restoration goals were not published in 2003 because no SAV
was mapped in the available historical photography or through the baywide aerial
survey (U.S. EPA 2003b). At the same time, existing information does not support
delineation of these entire segments as SAV no-grow zones following the detailed
decision rules documented by the U.S. EPA (2003b). The EPA recommends the
jurisdictions maintain the shallow-water designated use in the nine segments that
didn't have an SAV restoration goal published in 2003 but were previously deter-
mined not to be an SAV no-grow zone (Table V-l).
chapter v • Refinements to the Shallow-Water Designated-Use Assessment Procedures
-------
51
V-1. Recommended tidal-water designated uses by Chesapeake Bay Program segment and state-adopted sub-
segment. Updated version of Table IV-3 originally published on pages 62-63 of the 2003 Technical Support Document for
Identification of Chesapeake Bay Designated Uses and Attainability (U.S. EPA 2003b). The asterisks (*) indicate that no
numerical SAV restoration acreage goal was published in 2003 for the shallow-water designated use of that segment.
See Table V-2 for the nine new segment numerical SAV restoration averages. The absence of an "X" in the shallow-water
designated-use column indicates that segment has been entirely delineated as an SAV no-grow zone and the shallow-
water bay grass designated use should not apply to that segment.
Migratory
Spawning Shallow-
and Open- Deep- Deep- Water
Nursery Water Water Channel (SAV
Chesapeake Bay Program CBP Juris- (Feb. 1- (Year- (June 1- (June 1- growing
Segment Name Segment diction May 31) Round) Sept. 30) Sept. 30) season)
Northern Chesapeake Bay
Northern Chesapeake Bay
Upper Chesapeake Bay
Upper Central Chesapeake Bay
Middle Central Chesapeake Bay
Lower Central Chesapeake Bay
Lower Central Chesapeake Bay
Western Lower Chesapeake Bay
Eastern Lower Chesapeake Bay
Mouth of the Chesapeake Bay
Bush River
Gunpowder River
Gunpowder River
Middle River
Back River
Patapsco River
Magothy River
Severn River
South River
Rhode River
West River
Upper Patuxent River
Western Branch (Paruxent R.)
Middle Paruxent River
Lower Paruxent River
Lower Paruxent River
Lower Paruxent River
Lower Paruxent River
Lower Paruxent River
Lower Paruxent River
Upper Potomac River
Upper Potomac River
Upper Potomac River
CB1TF1
CB1TF2
CB2OH
CB3MH
CB4MH
CB5MH
CB5MH
CB6PH
CB7PH
CB8PH
BSHOH
GUNOH1
GUNOH2
MIDOH
BACOH
PATMH
MAGMH
SEVMH
SOUMH
RHDMH
WSTMH
PAXTF
WBRTF
PAXOH
PAXMH1
PAXMH2
PAXMH3
PAXMH4
PAXMH5
PAXMH6
POTTF
POTTF
POTTF
MD
MD
MD
MD
MD
MD
VA
VA
VA
VA
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
DC
MD
VA
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X*
X
X
X
X
X
X
X
X*
X
X
X
X
X
X
X
X
X
X
continued
chapter v * Refinements to the Shallow-Water Designated-Use Assessment Procedures
-------
52
Tab
(continued)
Anacostia River
Anacostia River
Piscataway Creek
Mattawoman Creek
Middle Potomac River
Middle Potomac River
Middle Potomac River
Middle Potomac River
Lower Potomac River
Lower Potomac River
Upper Rappahannock River
Middle Rappahannock River
Lower Rappahannock River
Corrotoman River
Piankatank River
Upper Mattaponi River
Lower Mattaponi River
Upper Pamunkey River
Lower Pamunkey River
Middle York River
Lower York River
Mobjack Bay
Upper James River
Upper James River
Appomattox River
Middle James River
Chickahominy River
Lower James River
Mouth of the James River
Western Branch Elizabeth River
Southern Branch Elizabeth River
Eastern Branch Elizabeth River
Lafayette River
Mouth of the Elizabeth River
Lynnhaven River
Northeast River
C&D Canal
C&D Canal
Bohemia River
Elk River
Elk River
Sassafras River
ANATF
ANATF
PISTF
MATTF
POTOH1
POTOH2
POTOH3
POTOH
POTMH
POTMH
RPPTF
RPPOH
RPPMH
CRRMH
PIAMH
MPNTF
MPNOH
PMKTF
PMKOH
YRKMH
YRKPH
MOBPH
JMSTF1
JMSTF2
APPTF
JMSOH
CHKOH
JMSMH
JMSPH
WBEMH
SBEMH
EBEMH
LAFMH
ELIPH
LYNPH
NORTF
C&DOH
C&DOH
BOHOH
ELKOH1
ELKOH2
SASOH1
DC
MD
MD
MD
MD
MD
MD
VA
MD
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
DE
MD
MD
MD
MD
MD
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X*
X
X
X
X
X*
X
X*
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
chapter v • Refinements to the Shallow-Water Designated-Use Assessment Procedures
-------
53
Table V-1. (continued)
Sassafras River
Upper Chester River
Middle Chester River
Lower Chester River
Eastern Bay
Upper Choptank River
Middle Choptank River
Lower Choptank River
Mouth of the Choptank River
Little Choptank River
Honga River
Fishing Bay
Upper Nanticoke River
Upper Nanticoke River
Middle Nanticoke River
Lower Nanticoke River
Wicomico River
Manokin River
Manokin River
Big Annemessex River
Big Annemessex River
Upper Pocomoke River
Middle Pocomoke River
Middle Pocomoke River
Lower Pocomoke River
Lower Pocomoke River
Tangier Sound
Tangier Sound
Tangier Sound
SASOH2
CHSTF
CHSOH
CHSMH
EASMH
CHOTF
CHOOH
CHOMH2
CHOMH1
LCHMH
HNGMH
FSBMH
NANTF
NANTF
NANOH
NANMH
WICMH
MANMH1
MANMH2
BIGMH1
BIGMH2
POCTF
POCOH
POCOH
POCMH
POCMH
TANMH1
TANMH2
TANMH
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
MD
DE
MD
MD
MD
MD
MD
MD
MD
MD
MD
VA
MD
VA
MD
MD
VA
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X*
X
X
X
X
X
X
X
X
X
X*
X
X
X
X
X
X
X
X*
X*
X
X
X
X
X
Source: U.S. EPA 2003b, 2004b, 2004c, 2005
To determine attainment of the shallow-water bay grass designated use, SAV restora-
tion goals for these nine segments were established based on the total surface acre
between the shoreline and the 0.5-meter depth contour divided by the 2.5 water
clarity acres multiplier (Table V-2). Any SAV no-grow zones within the individual
segments were removed before conducting the above calculations.
ON
The EPA has determined that the shallow-water designated use is protected when
there is restoration of SAV to the targeted restoration acreages or when a sufficient
area of shallow-water habitat contains required levels of water clarity, accounting for
chapter v » Refinements to the Shallow-Water Pesignated-Use Assessment Procedures
-------
54
¥-2, SAV restoration goals for segments that had no published acreage goals
in 2003.
Chesapeake
Bay Program
Segment
ANATF (MD)
BACHOH
C&DOH (DE)
C&DOH (MD)
CHSTF
MPNOH
NANTF (DE)
PAXMH3
PAXMH6
PMKOH
POCOH (MD)
POCOH (VA)
RPPOH
WBRTF
Segment Name
Anacostia River-Maryland
Back River
C&D Canal-Delaware
C&D Canal-Maryland
Upper Chester River
Lower Mattaponi River
Upper Nanticoke River-Delaware
Lower Patuxent River Sub-Segment 3
Lower Patuxent River Sub-Segment 6
Lower Pamunkey River
Middle Pocomoke River-Maryland
Middle Pocomoke River- Virginia
Middle Rappahannock River
Western Branch Patuxent River
Shallow-Water
Habitat Area1
(Acres)
3
850
15
83
574
323
370
3
3
423
56
167
1,226
3
SAV
Restoration
Goal2
(Acres)
3
340
6
33
230
129
148
3
3
169
22
67
490
3
Determined as total surface area of the segment from adjacent shoreline out to the 0.5-meter depth
contour at mean low water minus any delineated SAV no-grow zone within the segment.
Calculated as the shallow-water habitat area divided by 2.5 (the water clarity acres multiplier) (see
U.S. EPA 2003a).
3No (or very limited) bathymetry data were available, therefore, no shallow-water habitat area or
SAV restoration goal acreage could be calculated.
vegetated bottom habitat. Based on the decades long record of published documen-
tation on SAV light requirements (Batiuk et al. 1992, 2000; Dennison et al 1993;
Kemp et al. 2001; U.S. EPA 2003a, 2004a), the EPA recommends that an attainment
determination based on water clarity acres be based on 2.5 times each acre needed
to meet the restoration goal acreage.
A water clarity acre is defined as an acre of shallow-water bay grass designated-use
bottom habitat, located anywhere between the 2-meter depth contour and the adja-
cent shoreline inclusively, which has been observed to achieve the applicable
salinity-regime-specific water clarity criteria. A water clarity acre cannot be defined
within a delineated SAV no-grow zone (see pages 41-55 in U.S. EPA 2004c for
narrative descriptions and maps of those zones). For segments in which the resultant
water clarity acreage exceeds the total acres of shallow-water habitat from the
shoreline out to the 2-meter depth contour, the water clarity restoration acreage will
be set at the total acreage out to the 2-meter depth contour.
Assessment of attaining a segment's water clarity restoration acreage should be
based on calculation of the arithmetic mean of the year-by-year arithmetic means of
a month-by-month accounting of water clarity acres over the three-year SAV
growing season assessment period. Calculation of water clarity acres should be
chapter v * Refinements to the Shallow-Water Designated-Use Assessment Procedures
-------
55
based on spatially intensive shallow-water monitoring turbidity data converted to Kd
(light attenuation coefficient), interpolated as described in Chapter 2 and then
compared to the corresponding Kd threshold assigned to each interpolator grid cell.
The total acreage of an interpolator grid cell is added to the running total water
clarity acres for a segment when the interpolated Kd for that cell is less than or equal
to the Kd threshold assigned to that cell.
The Kd value based on achieving the applicable water clarity criteria at the 2-meter
depth will apply to all interpolator grid cells with centroids within the 2-meter to 1-
meter depth contours. All interpolator grid cells with centroids that lie within the
area bounded by the shoreline and the 1-meter contour will be assigned the Kd value
for the 1-meter depth.
If the segment's single best year of water clarity acres, as calculated above, is equal
to or greater than the segment's water clarity restoration acreage, then that segment
has attained the shallow-water bay grass designated use. If the segment's single best
year of water clarity acres is less than the segment's water clarity restoration acreage,
then the segment is in non-attainment of this designated use.
The EPA recommends the states adopt one of two approaches to calculating water
clarity acres. Both methodologies directly account for progress towards meeting the
SAV restoration goal acreage and measurement of suitable shallow water habitat
acreage necessary to support restoration of the remaining SAV beds needed to reach
the goal acreage.
The first methodology was originally published in the 2004 Chesapeake Bay water
quality criteria addendum (U.S. EPA 2004a). This methodology assesses attainment
of the shallow-water bay grass designated use in a segment through a combination
of mapped SAV acreage and meeting the applicable water clarity criteria in an addi-
tional, unvegetated shallow water surface area equal to 2.5 times the remaining SAV
acreage necessary to meet the segment's restoration goal (SAV restoration goal
acreage minus the mapped SAV acreage). In other words, a segment's shallow-water
bay grass designated use would be considered in attainment if there is sufficient
acres of shallow-water habitat meeting the applicable water clarity criteria to support
restoration of the remaining acres of SAV, beyond the SAV beds already mapped,
necessary to reach that segment's SAV restoration goal acreage. These measure-
ments of SAV acreages and water clarity levels would be drawn from three years of
data as previously described in the Regional Criteria Guidance (U.S. EPA 2003a).
Here's a hypothetical example of this first methodology for determining attainment
of the shallow-water bay grass designated use using both mapped SAV acreage and
shallow-water habitat acreage meeting the water clarity criteria. Segment X has an
SAV restoration goal acreage of 1,400 acres. Over the past three years, SAV beds
totaling 1,100 acres have been mapped within the segment. Therefore, the remaining
SAV acreage necessary to meet the segment's restoration goal is 1,400 acres
(segment SAV restoration goal) minus 1,100 acres (SAV acres currently mapped) or
300 acres. Beyond the currently vegetated shallow-water habitat, an additional
chapter v • Refinements to the Shallow-Water Designated-Use Assessment Procedures
-------
56
750 acres of shallow-water habitat (2.5 multiplier times 300 acres) is needed to attain
the water clarity criteria to determine this segment is attaining its shallow-water bay
grass designated use.
The second methodology directly accounts for mapped acres of SAV within the
calculation of water clarity acres. As part of the month-by-month accounting of
water clarity acres, over the three-year SAV growing season assessment period,
interpolator cells containing any mapped SAV beds are counted towards the total
water clarity acres.
Here's a hypothetical example of this second methodology for determining attain-
ment of the shallow-water bay grass use using both mapped SAV acreage and
shallow-water habitat acreage meeting the water clarity criteria. Segment Y has an
SAV restoration goal acreage of 1,400 acres. Applying the 2.5 multiplier, this
segment also has a water clarity restoration acreage of 3,500 acres. Over the past
three years, SAV beds totaling 1,100 acres have been mapped within the segment
each year. During each growing season's accounting of water clarity acres, these
1,100 acres of mapped SAV beds are directly counted towards the growing season
arithmetic mean water clarity acreage. Therefore, accounting directly for 1,100 acres
of mapped SAV beds as water clarity acres, an additional 2,400 acres (3,500 water
clarity restoration acres minus 1,100 acres of mapped SAV) of shallow-water habitat
is needed to attain the water clarity criteria to determine this segment is attaining its
shallow-water bay grass designated use.
ASSESSMENT BASED ON CFD-BASED WATER CLARITY
CRITERIA ATTAINMENT
A jurisdiction may choose to apply the CFD-based assessment of water clarity
criteria to evaluate attainment of the segment's shallow-water bay grass designated
use (U.S. EPA 2003a, 2004a). To attain the designated use, the segment must meet
the applicable water clarity criteria throughout the applicable shallow-water habitat
(from the shoreline out to the segment-specific water clarity criteria application
depth contour) (see Table IV-13 on pages 115-117 in U.S. EPA 2003b) over three
SAV growing seasons, factoring in allowable exceedances using the appropriate
salinity-regime-based biological reference curve (see Figures V-l, V-2). Chapter 2
and Appendix B document the application of the CFD-based criteria attainment
assessment in detail. Chapter 7 deals with the specific elements of the shallow-water
criteria attainment assessment procedures using a CFD-based evaluation.
SHALLOW-WATER DESIGNATED USES
AND SAV NO-GROW ZONES
Shoreline habitats of 2 meters or less (where SAV is never expected to grow due to
extreme wave energy, permanent physical alterations, natural discoloration of the
water, and no functional shallow-water habitat from river channeling) were
chapter v • Refinements to the Shallow-Water Designated-Use Assessment Procedures
-------
57
designated as SAV no-grow zones (see pages 108-110 in U.S. EPA 2003b). In the 39
segments with SAV no-grow zones, 31 of the segments have such zones extending over
a portion of the segment (see Table V-l on page 42 in U.S. EPA 2004c). In these
segments, an area delineated as an SAV no-grow zone should simply be left out of any
assessment of shallow-water designated-use attainment based on water-clarity acres or
on a CFD-based assessment of water clarity criteria attainment.
In the case of the eight segments where the entire shallow-water area was delineated
as an SAV no-grow zone (see pages 108-110 in U.S. EPA 2003b), the best available
information indicates the shallow-water bay grass designated use is not appropriate.
The EPA recommends that this designated use not apply to (or that it be removed
from) any segment in which the area encompassing the entire 2 meters or less
shallow-water habitat be delineated as an SAV no-grow zone (Table V-l).
Table V-l is an updated version of Table IV-3 originally published on pages 62-63
in the 2003 Technical Support Document for Identification of Chesapeake Bay
Designated Uses and Attainability (U.S. EPA 2003b). This revised table documents
the above-described segments that are entirely SAV no-grow zones (where the
shallow-water bay grass designated use does not apply) or had no previously estab-
lished SAV restoration goal. This table includes a list of all the Chesapeake Bay
Program segments in the Chesapeake Bay, its tidal tributaries, and its embayments
(U.S. EPA 2004b, 2005) as well as the sub-segments delineated by Maryland and
Virginia (U.S. EPA 2004c).
R
The original 2003 Chesapeake Bay water quality criteria document included biolog-
ical reference curves to assess attainment of the water clarity criteria using the CFD
methodology (see pages 173-176 and Appendix H in U.S. EPA 2003a). Those refer-
ence curves were developed using data collected as part of the Chesapeake Bay
Water Quality Monitoring Program in which the monitoring stations are located in
open, mid-channel areas of Chesapeake Bay, its tidal tributaries, and its embay-
ments. Use of the fixed-station, mid-channel water quality data was necessary even
though these data are not necessarily representative of the Bay's shallow-water habi-
tats; sufficient data more representative of the shallow-water habitats were not
available (see Chapter 9 in Batiuk et al. 2000).
Efforts are underway through the Chesapeake Bay Shallow-water Monitoring
Program to collect water clarity data for use in generating more appropriate biolog-
ical reference curves. These data are being collected (see Chapter 7 for additional
detail) in the same way that shallow-water designated use areas will be assessed. The
resulting biological reference curves will, therefore, be directly comparable to the
CFD assessment curves (see Chapter 2 for further details). Further refinement of the
existing published water clarity criteria biological reference curves (e.g., updating
with more recent mid-channel data, developing four salinity-regime-based curves) is
chapter v » Refinements to the Shallow-Water Pesignated-Use Assessment Procedures
-------
58
not warranted at this time given ongoing collection of more appropriate shallow-
water data. In the interim, the EPA recommends that states assess their water clarity
criteria using the CFD methodology which uses existing published biological refer-
ence curves to define the amount and pattern of allowable criteria exceedances.
Figure V-l illustrates the biological reference curve that states should apply in the
CFD-based water clarity criteria assessment of tidal fresh and oligohaline segments
with shallow-water bay grass designated uses. Figure V-2 illustrates the biological
reference curve that should be applied in the assessment of mesohaline and polyha-
line segments with shallow-water bay grass designated uses. Appendix H in this
document provides the equations for the Chesapeake Bay water clarity criteria
biological reference curves. Preliminary results from evaluation of limited shallow-
water monitoring data indicate that biological reference curves generated from
mid-channel data (Figures V-l and V-2) and those generated from shallow-water
monitoring data (see Figure VII-11 in Chapter 7) are quite similar in overall shape
and levels of allowable exceedances.
Oligohaline and Tidal Fresh Monthly Clarity Biological Reference Curve
HAAO/
QAO/
Qflp/
7AO/
<1>
F
§
o
Ju dfi0/
Q_
OAO/
HAP/
AO/
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Volume
¥-1. Chesapeake Bay water clarity criterion biological reference curve for application
to tidal fresh and oligohaline shallow-water designated-use habitats.
chapter v
-------
59
1 nn°/ -,
QAO/
7O°/
(U
H
"5 jrOo/
c
(D
O
»r ArtO/
Q_
OAO/
m°/
0<
Polyhaline and Mesohaline Monthly Clarity Biological Reference Curve
S
L'X
''"*'\,,
*'"""""\,,,,^
"""*\-,m,f<,,,rnf
'~U**~~*%mmff
"5!»my,5H!%tes^^^^^^^^^
& 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Volume
V-2. Chesapeake Bay water clarity criterion biological reference curve for application
to mesohaline and polyhaline shallow-water designated-use habitats.
LITERATURE CITED
Batiuk, R.A., P. Bergstrom, M. Kemp, E. Koch, L. Murray, J.C. Stevenson, R. Bartleson, V.
Carter, N.B. Rybicki, J.M. Landwehr, C. Gallegos, L. Karrh, M. Naylor, D. Wilcox, K.A.
Moore, S. Ailstock, and M. Teichberg. 2000. Chesapeake Bay Submerged Aquatic Vegetation
Water Quality and Habitat-Based Requirements and Restoration Targets: A Second Technical
Synthesis. CBP/TRS 245/00 EPA 903-R-00-014. U.S. EPA Chesapeake Bay Program,
Annapolis, MD.
Batiuk, R.A., R. Orth, K. Moore, J.C. Stevenson, W. Dennison, L. Staver, V. Carter, N.B.
Rybicki, R. Hickman, S. Kollar, and S. Bieber. 1992. Chesapeake Bay Submerged Aquatic
Vegetation Habitat Requirements and Restoration Targets: A Technical Synthesis. CBP/TRS
83/92. U.S. EPA Chesapeake Bay Program, Annapolis, MD.
Chesapeake Executive Council. 2000. Chesapeake 2000. Chesapeake Bay Program,
Annapolis, MD.
Dennison, W.C., RJ. Orth, K.A. Moore, J.C. Stevenson, V. Carter, S. Kollar, P. Bergstrom,
and R.A. Batiuk. 1993. Assessing water quality with submerged aquatic plants. Bioscience
43:86-94.
Kemp, W.M., R.A. Batiuk, R. Bartleson, P. Bergstrom, V. Carter, C.L. Gallegos, W. Hunley,
L. Karrh, E. Koch, J.M. Landwehr, K.A. Moore, L. Murray, M. Naylor, N.B. Rybicki, J.C.
chapter v » Refinements to the Shallow-Water Designated-Use Assessment: Procedures
-------
60
Stevenson, and DJ. Wilcox. 2004. Habitat requirements for submerged aquatic vegetation in
Chesapeake Bay: Water quality, light regime and physical-chemical factors. Estuaries 27:
363-377.
U.S. Environmental Protection Agency (U.S. EPA). 2003a. Ambient Water Quality Criteria
for Dissolved Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and Its Tidal
Tributaries. EPA 903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis,
MD.
U.S. Environmental Protection Agency. 2003b. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability. EPA 903-R-03-004. Region III Chesapeake Bay
Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2004a. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a Chesapeake Bay and Its Tidal Tributaries - 2004
Addendum. EPA 903-R-04-005. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2004b. Chesapeake Bay Program Analytical
Segmentation Scheme: Revisions, Decisions and Rationales 1983-2003. EPA 903-R-04-008.
CBP/TRS 268/04. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2004c. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability - 2004 Addendum. EPA 903-R-04-006. Region III
Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2005. Chesapeake Bay Program Analytical Segmen-
tation Scheme: Revisions, Decisions and Rationales (1983-2003) - 2005 Addendum. EPA
903-R-05-004. CBP/TRS 278-06. Region III Chesapeake Bay Program Office, Annapolis,
MD.
chapter v • Refinements to the Shallow-Water Designated-Use Assessment Procedures
-------
61
chapter
Chlorophyll a Criteria
Assessment Procedures
STATE WATER QUALITY STANDARDS
With publication of the April 2003 Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and Its Tidal Trib-
utaries, the EPA provided the states with a recommended narrative (non-numerical)
chlorophyll a criterion applicable to all of the Chesapeake Bay and its tidal tributary
waters (Table VI-1) (U.S. EPA 2003). From 2004 through early 2006, Virginia and
the District of Columbia adopted numerical chlorophyll a criteria for application in
the tidal James River (Virginia) and across all the District's jurisdictional tidal
waters. Both jurisdictions determined that algae-related designated use impairments
would likely persist in these tidal waters even after attainment of applicable
dissolved oxygen and water clarity criteria. The technical information supporting
adoption of numerical chlorophyll a criteria by Virginia and the District was
published in the 2003 Chesapeake Bay water quality criteria document (U.S. EPA
2003). Maryland and Delaware adopted narrative chlorophyll a criteria into their
water quality standards regulations (Table VI-1).
Table VI-1. Chesapeake Bay narrative chlorophyll a criteria.
Concentrations of chlorophyll a in free-floating microscopic aquatic plants (algae) shall not
exceed levels that result in ecologically undesirable consequences—such as reduced water
clarity, low dissolved oxygen, food supply imbalances, proliferation of species deemed
potentially harmful to aquatic life or humans or aesthetically objectionable conditions—
or otherwise render tidal waters unsuitable for designated uses.
Source: U.S. EPA 2003.
chapter vi • Chlorophyll a Criteria Assessment Procedures
-------
62
CHLOROPHYLL A CRITERIA ASSESSMENT
PROCEDURES
CHLOROPHYLLS CRITERIA REFERENCE CURVE
To assess attainment of the State adopted numerical chlorophyll a concentration-
based criteria, it was necessary to establish a reference curve for use in the CFD
criteria attainment assessment process (U.S. EPA 2003). In the case of chlorophyll
a criteria where a biologically-based reference curve is not available, EPA recom-
mends the states use of the default reference curve described in Chapter 2 (see Figure
II-4 and Equation 1).
CHLOROPHYLLS CRITERIA ASSESSMENT
A criterion threshold is a concentration that should rarely be exceeded by a "popu-
lation" of concentration data exhibiting healthy levels. The state-adopted
concentration-based chlorophyll a criteria values are threshold concentrations that
should only be exceeded infrequently since a low number of naturally occurring
exceedances occur even in a healthy phytoplankton population. The assessment of
chlorophyll a criteria attainment, therefore, should use the CFD-based assessment
method described in Chapter 2 that applies the default reference curve. These
Chesapeake Bay chlorophyll a criteria apply only to those seasons and salinity-based
habitats for which they were defined to protect against applicable human health
and aquatic life impairments. Each season—spring (March 1-May 31) and summer
(July 1-September 30)—should be assessed separately to evaluate chlorophyll a
criteria attainment.
Assessments of seasonal mean chlorophyll a criteria should be based on seasonal
averages of interpolated data sets. To calculate the seasonal averages, each interpo-
lated cruise within a season should be averaged on a point-by-point basis in
matching interpolator grid cells. Spatial violation rates should be calculated for each
seasonally aggregated interpolation in an assessment period. For example, for a
summer open-water seasonal chlorophyll a criteria assessment of a three-year
assessment period, three seasonal average interpolations representing each season
(Year 1 Summer, Year 2 summer, Year 3 summer) should be used.
LITERATURE CITED
U.S. Environmental Protection Agency. 2003. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and Its Tidal Tributaries.
EPA 903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis, MD.
chapter vi • Chlorophyll a Criteria Assessment Procedures
-------
63
chapter \/||
Shallow-water Monitoring and
Application for Criteria
Assessment
DESIGN AND APPROACH FOR CHESAPEAKE BAY
SHALLOW-WATER MONITORING
In July 2001, the Chesapeake Bay Program Monitoring and Analysis Subcom-
mittee's Tidal Monitoring and Analysis Workgroup formed a Tidal Monitoring
Design Team that undertook the redesigning of the Chesapeake Bay Tidal Moni-
toring Network. Over the next two years, the Design Team set goals and objectives,
reviewed the existing Chesapeake Bay monitoring design, evaluated potential new
monitoring strategies, and made recommendations for implementing a network to
provide the requisite data and support to address the Chesapeake Bay Program's
programmatic goals and objectives.
The new Tidal Monitoring Network focused on meeting the water quality protection
and restoration goals and objectives of the Chesapeake 2000 agreement (Chesapeake
Executive Council 2000). The network's primary objective is to supply the water
quality monitoring information needed to assess the new water quality criteria for
dissolved oxygen, water clarity, and chlorophyll a — ultimately with the goal of
removing the Chesapeake Bay and its tidal rivers from the list of impaired waters.
Secondary network objectives are to provide information for defining the nutrient
and sediment conditions necessary for protecting living resources and vital habitats.
Water quality data would also support refinement, calibration, and validation of the
Chesapeake Bay Water Quality/Sediment Transport Model.
The design of the new Tidal Monitoring Network emphasized monitoring of the
shallow-water designated use areas. In a 1999 study, the Maryland Department of
Natural Resources investigated the validity of using mid-channel data to assess
nearshore areas. The 13-tributary study examined water quality at 127 nearshore
stations and compared the data to 54 adjacent mid-channel stations (Karrh 1999;
Batiuk et al. 2000). The study found wide variations between nearshore and mid-
channel data, both within and between tributaries. Based on this finding, the
researchers concluded that decisions to use mid-channel data to characterize
nearshore conditions should be made on a site-by-site basis. Figure VII-1 illustrates
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
64
this variability, showing situations in which a single, mid-channel data point would
not adequately represent suspended solids and chlorophyll a in shallow areas. The
Design Team concluded that monitoring of shallow, nearshore waters must have
greater spatial coverage to obtain an accurate representation of these parameters.
Turbidity
EH 0.0 - 5.0 C10.0 -15.0 • 20.0 - 25.0 • 30.0 - 40.0 • 50.0 - 60.0
O 5.0 -10.0 IZ 15.0-20.0 • 25.0 -30.0 •40.0-50.0 • >60.0
B
Chlorophyll
Q 0.0 - 5.0
Q 5.0 -10.0
10.0-15.0
15.0-20.0
I 20.0 - 25.0
I 25.0 - 30.0
| 30.0 - 40.0
I 40.0 - 50.0
50.0 - 60.0
>60.0
Figure VII-1. Spatial distribution of turbidity (A) and chlorophyll a (B) in the tidal James River.
Source: Virginia Institute of Marine Science—www2.vims.edu/vecos/
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
65
To capture the temporal variability of dissolved oxygen, the new Tidal Monitoring
Network incorporated high-frequency monitoring stations in surface and nearshore
locations. Since then, the dissolved oxygen criteria assessment procedure has been
modified to project the results of open-water dissolved oxygen assessments onto
adjacent shallow-water, designated-use areas, instead of conducting a separate
shallow-water assessment (see Chapter 4 for details). The design for collecting high-
frequency dissolved oxygen data will likely be modified to represent dissolved
oxygen concentrations in open-water, designated-use habitats more accurately.
••
The intensive shallow-water monitoring program design is based on two innovative
technologies that were extensively tested in Maryland's Magothy and Severn rivers
as well as Tangier Sound from 1999 to 2002. The Dataflow water quality mapping
component collects high-resolution surface data from both open tidal-tributary and
shallow waters. The shallow-water buoy system collects high-frequency (near-
continuous) temporal data at specific locations, resulting in a data set that better
represents dissolved oxygen, chlorophyll a, and water clarity in time and space in
smaller tidal tributaries, small embayments, and shallow-water habitats. In 2003, the
Maryland Department of Natural Resources, the University of Maryland's Chesa-
peake Biological Laboratory, the Virginia Department of Environmental Quality, and
the Virginia Institute of Marine Sciences initiated the new Chesapeake Bay Shallow-
water Monitoring Program. The two states and their partners closely coordinate
development of the monitoring schedules, equipment, methodologies, and quality
assurance procedures to ensure baywide compatibility and comparability.
The Shallow-water Monitoring Program is based on two components that collect
spatially and temporally intensive data. Known as "Dataflow," the spatially intensive
component includes a sensor array and a GPS system that provide data continuously
along a boat track in both shallow- and open-water designated-use areas. These data
can be used to develop detailed maps of water quality conditions. The temporally
intensive component is known as "continuous monitoring" and includes a sensor array
at fixed locations that provides data continuously through time. These data reflect
episodic changes in water quality or signify extremes in water quality conditions.
The existing shallow-water monitoring design is based upon a three-year assessment
period. Data are collected from all segments within a tidal tributary or embayment
during the same three years. Both Dataflow sampling and continuous buoys are
deployed for the same time period. The three-year assessment provides adequate
time to account for variation in both weather and hydrologic conditions (see page
151 in U.S.EPA 2003a). Assessments using fewer than three years of shallow-water
monitoring data are discussed in the section Schedule for Assessment of Shallow-
water Designated Use Habitats below.
To adequately assess water quality criteria in shallow-water habitats and tidal tribu-
tary open-water designated-use habitats, the EPA recommends that the states
chapter vii * Shallow-water Monitoring and Application for Criteria Assessment
-------
66
conduct Dataflow monitoring from April through October in tidal fresh, oligohaline,
and mesohaline segments and from March through November in polyhaline
segments. These assessment periods for the water clarity criteria were based on the
growing seasons for the salinity-based SAV plant communities (U.S. EPA 2003a).
Continuous monitoring data are collected to assess the variability of water quality
parameters throughout the day. Temporally intensive data help explain the relation-
ships and timing among algal blooms, low dissolved oxygen, and nutrient additions.
Although previous convention suggested that shallow-water habitats did not experi-
ence significant low dissolved oxygen levels, continuous monitoring data are
proving otherwise. The lowest dissolved oxygen levels often occur between 4:00 and
6:00 a.m. when, historically, little information has been collected.
The continuous monitoring program component employs automated YSI 6600 EDS
water quality data sondes. Maryland and Virginia have agreed to use similar instru-
ments, when possible, to ensure consistent methodology and comparability across
Chesapeake Bay segments. The YSI 6600 sonde directly measures dissolved oxygen,
fluorescence (an indication of chlorophyll a), turbidity (an indication of water
clarity), temperature, salinity, and pH. The Maryland Department of Natural
Resources Chesapeake Bay Shallow-water Monitoring Program Quality Assurance
Project Plan (see page 32 in Maryland Department of Natural Resources 2006) docu-
ments the YSI instrument parameters, range, resolution, units, and accuracy.
Fluorescence is correlated to chlorophyll a, the measurement used for assessing
attainment of the chlorophyll a criteria. Turbidity is correlated to Kj (light attenuation
coefficient), the measurement used to assess attainment of the water clarity criteria.
The initial design recommended two shallow-water buoy deployments in each
segment, but often, resources limit the number of buoys to one per site. The buoys
are programmed to take measurements every 15 minutes for the six parameters listed
above. They are deployed off piers or pylons, either 1-meter below the surface or at
a fixed depth of 0.3 meters above the bottom, generally in waters of 2-meters or less
in depth (Figure VII-2).
Instruments are exchanged every one to two weeks, depending on biofouling and
following strict calibration protocols (Virginia Institute of Marine Science 2005).
Field crews collect samples to calibrate fluorescence and turbidity instrument read-
ings, respectively, with chlorophyll a and light attenuation. The monitors are
positioned at representative sites both up- and down-river.
Both Maryland and Virginia have rigorous shallow-water monitoring quality assur-
ance/quality control (QA/QC) programs. The QA/QC protocols remain consistent
between states; the Chesapeake Bay Program Quality Assurance Officer and the
Chesapeake Bay Program's Analytical Methods and Quality Assurance Workgroup
have reviewed these protocols.
chapter vii » Shallow-water Monitoring and Application for Criteria Assessment
-------
67
I)
Figure VII-2. Example of a continuous monitoring site and the generated 2004 dissolved oxygen data record
at Fenwick Point in the Potomac River, Maryland.
Source: Maryland Department of Natural Resources — www.eyesonthebay.net
Overlap periods occur at each continuous monitoring site by using multiple sondes
during routine biweekly maintenance runs to determine instrument drift. Instruments
are pre- and post-calibrated and must meet rigorous QA/QC protocols. Two instru-
ments are dedicated to each site. When one instrument is removed from the site for
maintenance, it is measured against the newly calibrated instrument. At the same time,
a field crew member takes a full suite of calibration samples for laboratory analysis.
Satellite and cellular telemetry are implemented at a subset of continuous monitoring
sites where resources permit. Data from these sites are assessed on a daily basis.
Maryland shallow-water continuous monitoring data are available in near- or real-
time on the Department of Natural Resources "Eyes on the Bay" website
(www.eyesonthebay.net) (Figure VII-3). Virginia shallow-water continuous moni-
toring data are available on the Virginia Institute of Marine Sciences website
(www2.vims.edu/vecos/). The Chesapeake Bay Program website's data hub
(www.chesapeakebay.net/data) offers access to the complete quality assured
Shallow-water Monitoring Program datasets for Maryland and Virginia.
WATER QUALITY MAPPING COMPONENT
The main purpose for collecting high-resolution water quality data is to provide reli-
able water quality criteria assessments. However, Dataflow monitoring also provides
insight into spatial complexities and localized phenomena and information for water
quality modeling in shallow waters (STAC 2005). The data are useful in producing
maps of the extent and patchiness of algal blooms, seasonal and inter-annual
progressions, and localized water quality impairments.
The Dataflow system is a small, fast-moving vessel that pumps surface water contin-
uously from 0.5 meters below the water surface through a chamber surrounding the
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
68
Click Stations for Data
Click Legend Symbols to
Toggle Stations On OH
Fixed Monthly
Stations - Current &
Historical Data
Real-Time
Continuous Monitors
O Near-Time
Continuous Monitors
Figure VII-3. The 2005 fixed dataflow and continuous monitoring station map from the
Maryland Department of Natural Resources' "Eyes on the Bay" website.
Source: Maryland Department of Natural Resources — www.eyesonthebay.net.
probes of aYSI 6600 sonde (http://mddnr.chesapeakebay.net/sinVdataflow_instrumen-
tation.cfm). The system uses the same YSI 6600 sonde as the continuous monitoring
buoys and measures the same suite of six parameters—dissolved oxygen, fluorescence,
turbidity, temperature, salinity, and pH. A Global Positioning System (GPS) unit is inte-
grated into the computer system to measure the spatial position of each recorded
measurement. Data are collected every four seconds as the boat follows a cruise track
that traverses between shallow and open waters. These data are then interpolated to
provide a high-resolution map of surface water quality conditions (see Chapter 2 for
further details). Each segment is mapped monthly from April through October or March
through November. The vessel stops at different locations throughout a segment for
discrete measurements of photosynthetically active radiation (PAR), Secchi depth, and
dissolved oxygen along with collection of water samples for laboratory analysis of
chlorophyll a (for use as calibration data). These "calibration" sites often overlap with
existing open-water fixed-station sites and continuous monitoring sites; they represent
the dynamic range of water quality in that segment.
SCHEDULE FOR ASSESSMENT OF SHALLOW-WATER
DESIGNATED-USE HABITATS
The current level of shallow-water monitoring is insufficient to conduct detailed
water quality criteria assessments in all Chesapeake Bay shallow-water habitats by
the Chesapeake 2000 agreement deadline of 2010 (Chesapeake Executive Council
2000). Three possible actions might remedy this problem. The first is extending the
deadline beyond 2010 for assessment of all Bay shallow-water habitats. The second
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
69
is identifying additional resources to expand the monitoring needed to meet the 2010
deadline. The third option is assessing segments for fewer than three years if
noncompliance of the segment is established. All three options are addressed below.
Accurately assessing how many segments can be assessed by each action remains
impossible however, since determining the availability of additional resources or
establishing how many segments might need fewer than three years of monitoring if
noncompliance is established cannot be predicted.
-V VI
The Chesapeake Bay Program partners have not approved extending the shallow-
water clarity criteria assessment timeframe beyond 2010. The current deadline will
not be met due to a lack of adequate resources to implement the shallow-water moni-
toring program design agreed upon by the Chesapeake Bay Program and the
participating states and thoroughly reviewed by the Chesapeake Bay Program Scien-
tific Technical Advisory Committee (STAC 2005). Significant progress has been
made to accelerate the assessment schedule. Although intensive shallow-water moni-
toring water clarity monitoring data will not be available in all segments, attainment
of the shallow-water bay grass designated use for those segments that contain an
SAV restoration acreage would be assessed by comparing each segment's single-best
SAV acreage from the most recent three-year period with the jurisdiction's adopted
segment-specific SAV restoration acreage (see Chapters 5 and 8 for further details).
In this way, each shallow-water designated-use segment could have some assessment
completed each year.
•U
Maryland, Virginia, and the EPA are actively seeking additional resources to expand
shallow-water monitoring in order to accelerate the schedule for completing baywide
assessments. In 2003, when Maryland and Virginia implemented shallow-water
monitoring in 11 Maryland segments and seven Virginia segments, it was estimated
that it would take until 2018 to assess all 78 Chesapeake Bay Program segments over
a three-year period on a rotating basis. Since the Shallow-water Monitoring
Program's initial implementation, both Maryland and Virginia have developed part-
nerships with county governments (e.g., Anne Arundel and Harford counties in
Maryland), municipal agencies (e.g., Hampton Roads Sanitation District in
Virginia), and federal agencies (e.g., NOAA's National Estuarine Research System)
and secured additional state funding to accelerate monitoring of all segments. Based
on these new partnerships, current expanded resources, and segment assessment over
a three-year period, it is estimated that Maryland will complete all its shallow-water
assessments by the year 2014 and Virginia will complete all its shallow-water assess-
ments by 2015. Figures VII-4 and VII-5 depict the current tentative schedule for
shallow-water monitoring and assessment of Maryland and Virginia segments,
respectively. New sources of funding continue to materialize and the schedules indi-
cated by Figures VII-4 and VII-5 will change in response to funding adjustments.
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
70
T
Year Assessment Complete
2006 | | 2010
2007 | | 2011
2008 | | 2012
2009 f~ ~l 2014
Segments
] BACOH
| | BIGMH
| | BOHOH
| | BSHOH
|| C&DOH
|| CB1TF
| | CB2OH
| | CB3MH
| | CB4MH
| | CB5MH
| | CHOMH1
|| | CHOMH2
|l | CHOOH
HH CHOTF
| | CHSMH
| | CHSOH
| | CHSTF
| | EASMH
| | ELKOH
| | FSBMH
| | GUNOH
| | HNGMH
| | LCHMH
| | MAGMH
~1 MANMH
J MATTF
| | MIDOH
| | NANMH
| | NANOH
| | NANTF
| | NOFTTF
| | PATMH
| | PAXMH
| | PAXOH
| | PAXTF
| | PISTF
| | POCMH
| | POCOH
| | POCTF
| | POTMH
| | POTTF
| |RHDMH
| | SASOH
| | SEVMH
| |SOUMH
| | TANMH
| | WBRTF
| | WICMH
~~] WSTMH
Figure VII-4. Schedule for shallow-water monitoring of Maryland's Chesapeake Bay segments.
Source: Maryland Department of Natural Resources
ASSESSMENTS BASED ON REDUCED MONITORING
The three-year assessment period was established to account for inter-annual variations
in weather and hydrologic conditions (U.S. EPA 2003a). If conditions are seriously
degraded, a state having fewer than three years of data can establish noncompliance by
applying the CFD-based criteria assessment methodology as follows.
First, at the start of a segment's shallow-water monitoring, assume 100 percent
compliance in all three years of the coming assessment period. Second, after the first
year of monitoring, a state should develop a CFD based on the collected data,
assuming all other planned sampling dates for the next two years had 100 percent
compliance with the applicable criterion. Finally, if the resultant assessment CFD
indicates that the segment will be in violation (compared to the applicable reference
CFD) no matter what happens in the following two years, then conclude that the
segment is out of compliance for the full assessment period and move the Shallow-
water Monitoring Program to another segment.
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
71
YEAR ASSESSMENT
COMPLETE
SCHEDULE
HH 2006
^•1 2008
I I 2009
II 2010
2012
2013
2015
Figure VII-5. Schedule for shallow-water monitoring of Virginia's Chesapeake Bay segments.
Source: Maryland Department of Natural Resources
To illustrate this approach, two hypothetical scenarios are illustrated below. In the
first example (Figure VII-6), it is assumed that monitoring was conducted for one
year and that full attainment was achieved during all scheduled sampling dates over
the next two years. The shallow-water monitoring over the first year indicated that
on all of the dates, the applicable criterion was violated in 15 percent or more of the
segment. The CFD indicates that the segment would be in noncompliance even if all
future sampling dates had 100 percent compliance. In this case, the state could have
decided to move the monitoring effort to a new shallow-water segment even after a
single year of study.
In the second example (Figure VII-7), the same assumptions are made and moni-
toring is conducted over one year. In this case, however, criteria exceedance is much
less extensive spatially and the CFD indicates that full compliance could be possible
if the current level of attainment is found in future monitoring. Since neither compli-
ance nor noncompliance could be established during the first year, shallow-water
monitoring would need to continue. The same analysis could take place after the
second year of monitoring and the decision could be revisited. It may turn out that a
full three years of monitoring are necessary to determine if the segment remained in
full compliance.
Although determining noncompliance in fewer than three years works in theory, the
yearly segment data must be analyzed in time to adequately design and implement a
sampling scheme for a new segment. The states must have the flexibility to deploy
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
72
Cruise Cruise
Number Year
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
2
2
2
10 2
11 2
12 2
13 3
14 3
15 3
16 3
17 3
18 3
0.9
8 0.8
1
o 07
£ '
I 0.6
^
& 05
™
1 0.4
1 0.3
1
0.1
0 -
I \
t
1 \
\ X
\
\
\
\ '••!
\
^T^^^
x<-\
0 0.1 0.2
Cruise
Month
May
June
July
August
September
October
May
June
July
August
September
October
May
June
July
August
September
October
^_
0.3 0.4
%0f
Space
1
0.7
0.45
0.3
0.25
0.2
0.15
0
0
0
0
0
0
0
0
0
0
0
0
% of Ref
Time Curve
0.00 0.00
0.05 0.02
0.11 0.05
0.16 0.09
0.21 0.11
0.26 0.14
0.32 0.19
0.37 1.00
0.37 1.00
0.37 1.00
0.37 1.00
0.37 1.00
0.37 1.00
0.37 1.00
0.37 1.00
0.37 1.00
0.37 1.00
0.37 1.00
0.37 1.00
Detailed Reference Curve
• "Scaled" Reference Curve
^— ^— Assessment Curve
— - — — — _
0.5 0.6
0.7 0.8 0.9 1
Percent Spatial Standard Exceedence
Scenario 1: noncompliance established after one year of shallow-water
monitoring.
resources to different systems. Often, implementation of a monitoring program for a
segment requires the coordination of various stakeholders and potential partners, the
leveraging of resources, and the allocation of field crews.
The Chesapeake Bay Program's Scientific and Technical Advisory Committee has
recommended that the tributary systems be assessed in their entirety for the full
three-year period rather than evaluating individual segments of a tributary in
different years (STAC 2005). This recommendation is particularly important for the
larger tidal tributaries such as the Patuxent, Potomac, Rappahannock, York, and
James rivers. These systems have tidal fresh, oligohaline, mesohaline, and poly-
haline segments, all of which influence each other. To understand the vast ecosystem
ihapter vii
-------
73
Cruise Cruise
Number Year
1 1
2 1
3 1
4 1
5 1
6 1
7 2
8 2
9 2
10 2
11 2
12 2
13 3
14 3
15 3
16 3
17 3
18 3
0.9 -\
0.8 I
!°M
i °-6 *.
•o
| 0.5- '•(
? 0.4
! \ \
| 0.3 N. vv
§ 0.2 ^\ *S
* V^ K
n- ^^^ -• ^ . .
0.1 ^^-ii:
0 -I 1 1 1
0 0.1 0.2
Cruise
Month
May
June
July
August
September
October
May
June
July
August
September
October
May
June
July
August
September
October
:iiife*ai^
0.3 0.4
%0f
Space
1
0.4
0.2
0.12
0.09
0.05
0.03
0
0
0
0
0
0
0
0
0
0
0
0
%of
Time
0.00
0.05
0.11
0.16
0.21
0.26
0.32
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
Detailed Reference Curve
"Scaled" Reference Curve
Assessment Curve
>_
0.5 0.6
0.7 0.8
Ref
Curve
0.00
0.06
0.14
0.23
0.29
0.44
0.57
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.9 1
Percent Spatial Standard Exceedence
VII-7. Scenario 2: noncompliance not established after one year of shallow-water
monitoring.
complexities and interactions between adjacent segments of a single tributary, it is
imperative to assess these tidal tributaries and segments as whole systems and not
discontinue monitoring in one segment if noncompliance occurs after only a year or
two of assessment.
The states should make the decision whether to continue shallow-water monitoring
for the full three years or to move the monitoring to another segment after a year or
two of sampling. In making such decisions, the state should consider the need to
gather shallow-water data for the assessment of multiple criteria (dissolved oxygen,
water clarity, and chlorophyll a) as well as other uses of the data (e.g., shallow-water
water quality model development and calibration). The states will also need to
ihapter vii
-------
74
consider if it makes sense (in terms of leveraging resources, coordinating, and under-
standing the relationship between segments and restoration activities) to discontinue
a segment's monitoring after one or two years if noncompliance of the segment is
shown. Finally, in the case of segments crossing two or more jurisdictional bound-
aries, all affected states will be involved in any decision to discontinue monitoring
prior to the end of the full three-year assessment period.
Importantly, the scenario described above and illustrated in Figure VII-6 does not
form an assessment that is lower in quality than one based on three years data. Non-
compliance is clearly established; that status would not change no matter what takes
place in ensuing years. The same approach may not be viable using alternative
assessment strategies such as the water clarity-acres approach for the clarity criteria.
Since the water clarity-acres assessment method relies on the mean of three years of
data, non-compliance could not be established in fewer than three years. The reverse,
however, may be true. However, if the segment's SAV restoration acreage goal was
attained during any single year, then compliance would be established and the deci-
sion could be made to discontinue monitoring.
The states' prioritization schedule for assessing shallow water monitoring segments,
(Figures VII-4 and VII-5) is based on several criteria—SAV coverage, maximization
of resources, partnerships, and management needs such as dissolved oxygen criteria.
Segment prioritization through SAV coverage is based on assessing segments that
are close to meeting the state-adopted SAV restoration acreage goal for the indi-
vidual segment. All states have agreed to assess attainment by each segment's
single-best SAV acreage for the most recent three-year period with the jurisdiction's
adopted segment-specific SAV restoration acreage (see Chapter 5 for further
details). Many Chesapeake Bay segments range between 50 and 100 percent of
meeting their restoration goals.
Appendix G lists all the Delaware, Maryland, Virginia, and the District of Columbia
segments and their relative success (by percent) in reaching their respective state-
adopted SAV restoration acreages. Those segments that have already met their SAV
restoration acreages constitute a lower priority for shallow-water assessment.
Segments that have not achieved any acres in meeting their SAV restoration acreage
form a lower priority as well. The higher the percentage attainment in meeting a
segment's SAV restoration acreage, the greater the priority was given for assessing
the shallow waters of that segment.
On the states' 2006 303(d) lists, eight Maryland segments and six Virginia segments
have met their adopted SAV restoration acreages. The segments that have already
attained their shallow-water designated use are low priority for shallow-water assess-
ment. Fourteen Maryland segments and five Virginia segments range between 50 and
100 percent of meeting their SAV restoration acreages (Appendix I). These segments
chapter vii » Shallow-water Monitoring and Application for Criteria Assessment
-------
75
were granted the highest priority for shallow-water monitoring (see Figures VII-4
and VII-5).
DISSOLVED OXYGEN CRITERIA ASSESSMENTS
USING SHALLOW-WATER MONITORING DATA
The Chesapeake Bay Shallow Water Monitoring Program has provided unprece-
dented volumes of spatially and temporally intensive Chesapeake Bay, tidal
tributary, and embayment data to assess water quality criteria attainment. This
wealth of data, however, provides new and unique analytical challenges within the
regulatory framework. In the case of dissolved oxygen criteria, these challenges
include: temporal variation of water quality parameters, spatial interpolations, and
scaling and interpolation issues. Specific procedures for evaluation of the 7-day, 1-
day, and instantaneous minimum open-water and deep-water dissolved oxygen
criteria have not been fully developed at this time.
The assessment of the 30-day mean dissolved oxygen criteria for open-water desig-
nated-use habitats will rely on mid-channel fixed station data combined with
Dataflow and Dataflow calibration profile data. As noted previously, the Dataflow
vessel stops at five to eight locations throughout a segment to collect calibration
measurements. Dissolved oxygen is measured from the surface to the bottom at these
sites using the same procedure as the mid-channel data collection. The dissolved
oxygen calibration data will provide an additional day of dissolved data each month,
at five locations instead of one or two. The dissolved oxygen Dataflow and the corre-
sponding Dataflow dissolved oxygen calibration data will be interpolated and
analyzed, along with fixed-station dissolved oxygen data, using the Chesapeake Bay
Program's interpolator and the CFD approach described in Chapter 2.
TEMPORAL VARIATION
Dataflow cruises collect between 3,000 and 10,000 points over several hours in a
segment. Data are normally collected between 7:00 a.m. and 5:00 p.m. with the boat
traversing open and shallow waters on one side of a tidal tributary or embayment and
repeating the process on the opposite side. The measurements can be interpolated to
produce a continuous surface of data that can be evaluated for the percentage area of
a segment that fails the applicable criterion.
The diel patterns of surface dissolved oxygen are well documented in both the liter-
ature and continuous monitoring data (www.eyesonthebay.net). In summer,
dissolved oxygen normally declines to its lowest level during the early morning
hours (3:00 a.m.) when algal and plant communities have been respiring throughout
the night; it reaches its peak in mid-afternoon (3:00 p.m.) following photosynthetic
activity. In some cases, this diel fluctuation can reach more than 15 mg-liter1
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
76
dissolved oxygen. When interpolating water quality mapping data collected
throughout the day, this variability presents a potential problem that is best illus-
trated by a map. Figure VII-8 shows that data collected early in the morning on one
side of the Severn River in Maryland is substantially lower than data collected later
in the day on the other side. If these measures were interpolated, it would appear that
one side of the river is faring more poorly than the other when, in fact, the dichotomy
merely represents a temporal artifact.
To produce a more representative spatial interpolation of surface dissolved oxygen
data, estimating the diel dissolved oxygen trend from continuous monitoring instru-
ments and using that trend estimate to adjust the Dataflow dissolved oxygen may
prove more feasible. The University of Maryland investigated this procedure by
comparing data from a nearshore continuous meter with those from a mid-channel
continuous buoy. They found that the dissolved oxygen in the two locations
responded differently to the local habitats and that nearshore dissolved oxygen
dropped at night and the mid-channel dissolved oxygen was highly variable, often
exceeding dissolved oxygen saturation during the day. Although the adjustment
procedure improved the data set, the prediction error was high. Further research is
needed to integrate the spatial and temporal monitoring data.
Rising DO Values
During the Day
Severn River
6/28/2001
<: rA
s-x
Figure VII-8. Illustration of rising dissolved oxygen concentrations during the day
(June 28, 2001) in the Severn River, Maryland.
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
77
'." .' .)N
The frequency and spatial coverage of water quality sampling will always remain
lower in relation to the temporal and spatial scales at which estuarine phenomena
occur. To overcome this reality, researchers must use innovative sample designs and
statistical methods. Throughout the Chesapeake Bay Program's tidal data analysis
and monitoring network design meetings, many of these issues regarding the inter-
pretation of shallow water monitoring data have been raised, but all were not solved.
The major issues relating to dissolved oxygen are highlighted below.
Water quality mapping of dissolved oxygen uses measures from a half-meter below
the surface. Some consider this type of measurement a weakness given that most
hypoxic events occur in deep-water or deep-channel habitats. The last five years of
water quality mapping, however, have revealed that hypoxic events can affect surface
and shallow waters more than initially recognized. Each mapping cruise collects
calibration samples and water quality depth profiles at five to eight stations per
segment. In much the same fashion that fixed station profiles are interpolated in
three-dimensions using the Chesapeake Bay interpolator (see Chapter 2 and
Appendix D), the surface mapping data could be interpolated along with calibration
station and mid-channel, fixed-station depth profiles to enhance volumetric estimates
of dissolved oxygen. Advancements in monitoring attainment technology that enable
deployment of automated vertical profilers and surface and bottom buoy monitors
could also support this effort. Overall, the integration of data types such as contin-
uous monitoring, mapping, remote sensing, and fixed-station profiles poses one of
the greatest challenges in criteria assessment.
Water quality mapping cruises cannot cover every shallow-water cove and creek in
a segment, thus presenting a problem for spatial extrapolation of the data. Criteria
assessment using the CFD method requires the use of an interpolated/extrapolated
surface from the entire segment and does not allow for exclusion of unsampled areas.
Almost certainly, many of the areas outside of the sampling boundary have far
different conditions than those measured in the shallow waters of the main segment.
These areas represent only a small percentage of each segment, but the question
remains whether they contain more valuable habitat than the space they occupy on a
percentage basis.
Annually, many of the larger fish kills in Chesapeake Bay occur in these small tidal
creeks and embayments due to anthropogenic influences or natural conditions. Two
months after torrential rains in June 2006, a Maryland Department of Natural
Resources aerial photography survey of the state's Eastern Shore tributaries revealed
that most small embayments were still clouded by silt and algal blooms to a far
greater extent than adjacent open waters. To assess conditions adequately in these
shallow-water tidal creeks and embayments, a probabilistic approach may be needed
in conjunction with current shallow-water sampling design in which representative
small tidal creeks and embayments are sampled by the surface mapping and the
chapter vii * Shallow-water Monitoring and Application for Criteria Assessment
-------
78
results become a surrogate for the percentage area that these creeks represent in a
segment.
A STAC-convened expert panel (described in detail in Chapter 2) has reviewed the
interpolation of spatial data. Several standardization decisions for interpolation
methodology will need to be made to address the panel's recommendations for
addressing shallow-water monitoring data (STAC 2006).
The water clarity assessment uses data from the shallow-water water quality
mapping to obtain high-resolution data in nearshore shallow waters. This section
describes the data analysis protocols for application of high-resolution turbidity
measurements to assess attainment of state-adopted water clarity criteria in shallow-
water monitored tidal tributaries and embayments of the Chesapeake Bay.
During each day of water quality mapping with the Dataflow, the operator stops at
five to eight locations (calibration stations) to measure photosynthetic active radia-
tion (PAR) so that the light attenuation coefficient (Kj) can be calculated and
correlated with the in situ turbidity values recorded simultaneously. The protocol
followed to derive this correlation is described below.
The Chesapeake Bay water clarity criteria were published as the percent of light
through water (see Table IV-1 on page 96 in U.S. EPA 2003a). Through the applica-
tion of the equation:
PLW= 100 exp(-KdZ) Equation 3
the appropriate percent light-through-water value and the selected water clarity
criteria application depth (Z) are inserted and the equation is solved for Kj. The
methodology developed by the Chesapeake Bay Program for assessing criteria
attainment involves a sequence of steps that leads to a cumulative frequency diagram
(CFD) as described in eight steps in Table II-l in Chapter 2. As part of step 3,
equating the in situ collected values of turbidity to estimated IQ values becomes
necessary to determine exceedance of the water clarity criterion. It is critical to
convert in-situ turbidity to estimates of IQ prior to any data interpolation in order to
reduce the error potential.
The relationship between turbidity and Kj, therefore, needs to be quantified to deter-
mine the turbidity threshold of the applicable water clarity criteria. This
determination narrows the scope considerably from the traditional calibration curve
in which the estimation of Kj is based on measurements for a wide range of turbidity
concentrations. In the current application, it is only necessary to accurately estimate
Kj from in situ measurements of turbidity in the neighborhood of the exceedance of
the water clarity criteria.
chapter vii * Shallow-water Monitoring and Application for Criteria Assessment
-------
79
In conducting the analysis to formulate the decision rules and calibration curves that
relate in situ turbidity measurements with calibration station K
-------
80
5J
3
73
O
Q. 21
04
01 234567
1 .5 root Turb
system - Chickahominy — James
-- Mattaponi
— York
-• Piankatank
9 10
— Lynnhaven
— Pumunkey
11
12
Figure VII-9. Simple linear regression of predicted Kd versus the 1.5 root of measured
turbidity using shallow water monitoring data from seven Virginia tidal tributaries
(2003-2005).
Source: Virginia Institute of Marine Science—www2.vims.edu/vecos.
coefficients and intercepts occur to form groupings of tributary data for calibration
purposes. The groupings developed to date reflect a strong geographic pattern, which
strengthens their validity.
INTERPOLATION
The very dense in situ measurements of turbidity from each sampling cruise track
(Figure VII-10) are first converted to Kj. The natural log of the converted IQ values
are then interpolated using a standardized ordinary kriging procedure with ARC/GIS
into a 25-meter square grid over the segment's entire surface area. Once interpo-
lated, the resultant interpolated Kd values are transformed back. Each interpolator
cell within a segment's shallow-water area is then assessed against a specific Kd
value for each applicable water clarity criterion application depth. An interpolator
cell value equal to or below this IQ value is considered in attainment of the appli-
cable water clarity criterion. A number above this value has failed to meet the
applicable water clarity criterion.
The entire area within the shallow-water designated-use zone for each sampling
cruise is then aggregated on an interpolator cell-by-cell basis to determine the total
area either in attainment or failing to meet the applicable water clarity criterion.
Water clarity attainment acres are determined for the total area within the shallow-
water area of each segment from the shoreline out to the 2-meter depth contour
excluding the delineated SAV no-grow zones (see Chapter 5 for details).
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
81
Mobjack
Bay
York
River
YRKPH
'15!-
76°30'
Figure VII-10. Example lower York River polyhaline segment YRKPH Dataflow sam-
pling cruise track on August 25, 2003.
Source: Virginia Institute of Marine Science—www2.vims.edu/vecos..
Water clarity criteria attainment can also be assessed in each segment's shallow-
water designated-use habitat through application of the CFD-based methodology
described in Chapter 2 for each three-year assessment period. Exceedance is the
cumulative frequency distribution of the portion of this zone that failed the Kd-
equivalent of the application depth specific water clarity criterion determined for that
segment compared to a reference CFD curve.
Naturally, environmental conditions will result in periodic exceedances of bay grass
water clarity requirements; such exceedances are allowable for bay grass survival
(U.S. EPA 2003a). Since allowable exceedances can be specific to salinity-based bay
grass communities, biologically based reference curves are applied using measured
water clarity exceedances established from existing bay grass beds for each salinity
region using mid-channel water quality data (see Figures VI-1 and VI-2 in Chapter
5). Figure VII-11 shows a preliminary example of a biological reference curve of
water clarity exceedances based on shallow-water monitoring data for established
bay grass beds in the polyhaline lower York River segment (YRKPH) during the
2003 and 2004 growing seasons. This curve is plotted along with the previously
published water clarity reference curve for mesohaline/polyhaline shallow-water bay
grass designated-use habitats (U.S. EPA 2003a).
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
82
100 •
90 ;
80 :
I eo.
0 20 40 60 80 100
Percent of Area Exceeding Criteria
VII-11. Comparison of the published mesohaline/polyhaline water clarity criteria
biological reference curve based on mid-channel water clarity measurements and a
preliminary example of a shallow-water monitoring-based water clarity criteria biological
reference curve.
Source: : U.S. EPA 2003a; Virginia Institute of Marine Science—www2.vims.edu/vecos.
A
Attainment of the chlorophyll a criteria in the shallow-water designated use areas
will be based upon the adjacent open-water designated use assessments. As with
dissolved oxygen assessments, open-water chlorophyll a assessments will rely on
the mid-channel fixed station data combined with Dataflow and Dataflow calibration
profile data. These data will be interpolated and analyzed, along with the fixed-
station chlorophyll a data, using the Chesapeake Bay Program's interpolator and
CFD approach described in Chapter 2. The following sections describe the rationale
for and development of protocols for using the in-situ fluorescence measurements
from the Dataflow system to assess chlorophyll a criteria attainment in shallow and
open-water tidal tributaries and embayments of Chesapeake Bay.
The Dataflow system generates a data set that better represents the spatial variability
of chlorophyll. The Dataflow cruise track transverses both the open and shallow
water designated use areas (see Figure VII-10), recording hundreds of fluorescent
measurements, very quickly and less expensively than the collection and laboratory
analysis of individual samples. However, the conversion of the fluorescence data to
chlorophyll a must be done carefully to ensure that they are comparable to the
chlorophyll a data upon which the chlorophyll a criteria were based.
chapter vii
-------
83
The in-situ fluorescence method is more susceptible to bias and interferences than
the laboratory method. Instrument manufacturers recognize that low temperatures
and high turbidities can affect the fluorescence response and note that different
phytoplankton species can fluoresce differently in-situ even if the actual chlorophyll
content is the same (YSI, Inc. 1999). To overcome these effects, it is a common prac-
tice to "calibrate" the in-situ data to the laboratory results by collecting and
analyzing a set of chlorophyll a samples in the laboratory concurrent with in-situ
measurements, and establishing a quantitative relationship, or "calibration" between
the methods via simple linear regression. The calibration may be done for each day
of sampling but better estimates may result if greater numbers of observations are
incorporated into a statistical model.
7 ' •'
The usual approach for calibrating in situ fluorescence to in vitro chlorophyll is to
develop a model of the form:
Chlorophyll = f(fluorescence, other variables). Equation 4
Usually the function f is a linear regression model and the estimates of the coeffi-
cients for this model are obtained using least squares. With this model, a measured
value of fluorescence may be used as an argument to obtain a predicted chlorophyll
value. By evaluating other water quality variables measured by the monitoring
program, it was determined that fluorescence, temperature, turbidity, pH, and
seasonal variables be used as independent variables as described above.
One problem with this standard approach is that least squares estimation requires
that data used as independent variables be measured without error. Clearly this
assumption is not satisfied for fluorescence. An alternative approach that treats both
in vitro and in situ chlorophyll as variables with measurement error estimates the
logarithm of their ratio with a linear regression model:
Log (R) = LogCChlj / Chl2) = f (other variables) Equation 5
where:
Chlj = in vitro chlorophyll
Chl2 = in situ chlorophyll (note: fluorometers used to collect data for this
study convert the fluorescence signal to chlorophyll with a standard
algorithm and this is the number recorded); and
R = the ratio of these two chlorophyll measures.
An estimate of in vitro chlorophyll is obtained from the in situ measurement by first
estimating the logarithm of R given the independent variables, back-transforming to
obtain an estimate of the ratio, and multiplying the in situ chlorophyll by the ratio to
estimate the in vitro chlorophyll.
chapter vii * Shallow-water Monitoring arid Application for Criteria Assessment
-------
84
.\
Continuous monitoring data for Maryland and Virginia were analyzed to determine
a method of post-calibrating fluorescence/chlorophyll to match extractive chloro-
phyll more precisely. Because the instruments are identical, it was assumed that the
relationships between the Dataflow fluorescence and chlorophyll a would show
similar patterns. Maryland data were available for 2003 through 2005 for approxi-
mately 21 tidal tributaries (not all tributaries were sampled in all three years).
Virginia data came from the York River. Initial tests indicated that no more variation
occurred between Maryland and Virginia data than among the tidal tributaries in
Maryland. This finding simplified the post-calibration model geographically by
allowing combination of data from both states.
A second test of the data evaluated potential differences among years. This test also
proved negative, which signified that all three years of data could be combined when
developing the post-calibration model. Tests of season and tributary differences
suggested that the final model would need to account for temporal and spatial differ-
ences. Further analyses indicated the need for two tributary groups and two season
groups, meaning that four calibration curves will be required. Significant variables
in the model also included water temperature, turbidity, and pH. Significance is
defined here as a p-value of less than 0.05.
Initial results indicate that four calibration curves would be needed, two for season
and two for tributary. All four models contain fluorescence, water temperature,
turbidity, and pH.
Several issues were addressed in conducting the analysis to formulate the decision
rules and calibration curves. Similar to the turbidity/IQj relationship, many of the
issues related directly to the decision to lump or divide the data when computing
calibration curves and decision rules. The argument in favor of lumping (to perform
the analysis on a data aggregate) reasons that better estimates result when large
numbers of observations are averaged. On the other hand, the in situ to in vitro rela-
tionship may not be consistent across all subsets of the data (i.e., between different
tidal tributaries and embayments). If so, dividing the data and developing algorithms
for each set may lead to better overall precision.
Because species composition can affect the relationship of in situ to in vitro chloro-
phyll measurements, this relationship may change with the seasons. Thus, one
aggregate-or-divide issue requiring resolution is the effect of seasons.
The in situ/in vitro difference generally follows a seasonal pattern consistent with
known species composition patterns for Chesapeake Bay and its tidal tributaries. In
situ chlorophyll measurements have a negative bias when phytoplankton populations
chapter vii * Shallow-water Monitoring and Application for Criteria Assessment
-------
85
shift toward a large component of blue-green algae. Blue-green algae increase in
abundance during mid to late summer, particularly in tidal-fresh to low-salinity habi-
tats. The calibration data from both the continuous monitors and the Dataflow water
quality mapping show that the negative bias of the in situ measure becomes greater
in summer. It was determined that two season groups would be needed.
It must be recognized that forming two season groups implements a model that
captures the average condition, but may not capture the condition that exists in a
particular tributary on a given date. The seasonal appearance of blue-green algae is
not the same across tributaries and not even the same within a tributary from year to
year. Even if the model predictions agree well with the observed data for the past
three years, it is quite possible that a blue-green bloom could form at some unusual
time of year in the future and lead to biased prediction. Truly reliable calibration of
in situ chlorophyll to in vitro chlorophyll requires that some information on the
concentration of blue-green cells be included in the calibration model.
Geography is another general factor that may influence the in situ to in vitro chloro-
phyll a relationship. Again, this influence is likely to be a phytoplankton species
composition effect. Other factors (e.g., turbidity), however, may play a role. It is
recommended that the analysis model the geography by treating locations (fixed-
stations for continuous monitors or river systems for Dataflow) as discrete
categorical predictors. If these predictors are statistically significant, the geography
portion of the model should be simplified using surrogate variables, such as salinity
and turbidity.
Spatial patterns emerge with data set analysis. These patterns, when viewed
geographically, appear to follow arrangements expected based on phytoplankton
species composition. In the Virginia Dataflow data, the trend is longitudinal within
the estuaries. In the tidal-fresh region, the in situ and in vitro measurements appear
similar, with a negative bias of in situ relative to in vitro emerging in downstream
stations (Figure VII-12). In the upper tidal Mattaponi River, one region occurs in
which in situ has a positive bias relative to in vitro. This situation may occur due to
high background fluorescence from tannins (dissolved organic carbon) in the water.
In Maryland, the negative bias (yellow squares) appears in regions where blue-green
populations have been identified; however, the data do not show a longitudinal
gradient similar to the Virginia data (Figure VII-13).
Diel
In continuous monitoring data, many locations exhibit distinct diel patterns in the in
situ chlorophyll. This diel pattern often shows that chlorophyll is higher at night and
lower during the day. Other research has shown that fluorometric chlorophyll read-
ings made in direct sunlight will be biased low because sunlight inhibits
phytoplankton fluorescence. This finding, coupled with the observed pattern of lower
in situ chlorophyll during the day, raised the concern that continuous monitoring of
chapter vii * Shallow-water Monitoring arid Application for Criteria Assessment
-------
86
Figure VII-12. Locations of the Virginia Chesapeake Bay Shallow-
water Monitoring Program calibration stations. In each location, a
circle indicates that no significant difference occurs between the in
situ chlorophyll measures and the in vitro chlorophyll measures. A
square indicates that the in situ measures are less than the in vitro
measures. An X indicates that in-situ measures are greater than the
in-vitro measures.
Source: Virginia Institute of Marine Science—www2.vims.edu/vecos.
in situ chlorophyll might be biased low during the day because of this measurement
problem. A special study was conducted at the Jug Bay station on the tidal Patuxent
River collecting hourly calibration samples for 24 hours. One set of samples was
collected monthly from March to December in 2005. Analysis of the in situ/in vitro
difference shows a very slight diel pattern in these data, but this variability became
trivial when compared to other sources of variance.
Collection Agency
The two principal agencies collecting these data—the Maryland Department of
Natural Resources and the Virginia Institute of Marine Sciences—have devoted
considerable effort to maintaining comparable shallow-water monitoring program
field collection methodologies, instrumentation, and QA/QC procedures. Even so,
because the two agencies work in geographically distinct regions, comparing results
between agencies to determine if these data can be combined to estimate calibration
curves should prove useful. Initial data evaluations indicate that no more variation
exists between Maryland and Virginia data than among the tidal tributaries in Mary-
land. These evaluations suggest that any differences between Maryland and Virginia
data may actually result from variations among the tidal tributaries and not from
dissimilarities between the data-collecting agencies.
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
87
.^r. r-t-r*
•• KMJJ.VN >* *Jr~-x VK
l/A *, .•""HP' V
'S*'' >» "•
,.\^,^ ^
Figure VII-13. Locations of the Maryland Chesapeake Bay Shallow-water
Monitoring Program continuous monitors. In each location, a circle indicates
that no significant difference exists between the in situ chlorophyll measures
and those for in situ chlorophyll. A square indicates that the in situ measures
are less than the in vitro measures.
Source: Department of Natural Resources--www2.eyesonthebay.net.
Background Fluorescence
In some Bay areas, the background fluorescence constitutes a significant component
of the total fluorescence signal due to freshwater input from blackwater streams.
Background fluorescence is the fluorescence measured on filtered water. This study
will identify those areas where background fluorescence requires measurement and
develop an algorithm to adjust for background fluorescence. Analysis indicates that
background fluorescence is not significant in the systems assessed to date.
Ancillary Data
While conventional wisdom holds that in vitro methods produce more accurate
measures of chlorophyll than in situ methods, both are still subject to error. Using
data collected independently of either type, the relative accuracy of the two method-
ologies will be assessed. For example, measurements taken as part of the nutrient
suite (e.g., particulate nitrogen, total nitrogen, etc.) have some predictive power for
chlorophyll. In cases where the in situ and in vitro measurements differ by more than
expected due to sampling error, these ancillary data may resolve which is more reliable.
Often a time series of both in situ and in vitro chlorophyll will show that the two
measurements compare quite well for much of the data record, with occasional large
discrepancies. Because these large discrepancies are most problematic from a
chapter vii • Shallow-water Monitoring and Application for Criteria Assessment
-------
decision-rule point of view, they warrant special consideration. If one of the methods
is more likely to be in error when these discrepancies occur, this finding will affect
use of that method in the regulatory process.
To address this issue, separate models of in situ chlorophyll and in vitro chlorophyll
need to be developed for which the independent variables are taken from the suite of
nutrient measurements (e.g., total nitrogen, paniculate nitrogen, etc.). A pilot project has
shown that these models are fairly predictive. In a case where a large discrepancy
between the in situ and the in vitro measurements exists, if one is in agreement with its
predictive model and the other is not, then the one out of agreement is likely in error.
Batiuk, R.A., P. Bergstrom, M. Kemp, E. Koch, L. Murray, J.C. Stevenson, R. Bartleson, V.
Carter, N.B. Rybicki, J.M. Landwehr, C. Gallegos, L. Karrh, M. Naylor, D. Wilcox, K.A.
Moore, S. Ailstock, and M. Teichberg. 2000. Chesapeake Bay Submerged Aquatic Vegetation
Water Quality and Habitat-Based Requirements and Restoration Targets: A Second Technical
Synthesis. CBP/TR 245/00 EPA 903-R-00-014. U.S. EPA Chesapeake Bay Program,
Annapolis, MD.
Chesapeake Executive Council. 2000. Chesapeake 2000. Chesapeake Bay Program,
Annapolis, MD.
Karrh, L. 1999. Comparison ofNearshore and Midchannel Water Quality Conditions. 200
pp. Chesapeake Bay Program, Annapolis, MD.
Maryland Department of Natural Resources. 2006. Quality Assurance Project Plan for the
Maryland Department of Natural Resources Chesapeake Bay Shallow Water Quality Moni-
toring Program for the period of July 1, 2006 - June 30, 2007. Maryland Department of
Natural Resources, Annapolis, MD.
Scientific and Technical Advisory Committee (STAC). 2005. Final Report of the Chesapeake
Bay Scientific and Technical Advisory Committee Workshop: Evaluating the Design and
Implementation of the Chesapeake Bay Shallow Water Monitoring Program Chesapeake Bay
Program Scientific and Technical Advisory Committee Publication 05-003.
Scientific and Technical Advisory Committee (STAC). 2006. The Cumulative Frequency
Diagram Method for Determining Water Quality Attainment: Report of the Chesapeake Bay
Program STAC Panel to Review Chesapeake Bay Analytical Tools. STAC Publication 06-003.
9 October 2006. Chesapeake Bay Program Scientific and Technical Advisory Committee.
Chesapeake Research Consortium, Edgewater, MD.
U.S. Environmental Protection Agency. 2003a. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and its Tidal Tributaries.
EPA 903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis, MD.
Virginia Institute of Marine Science. 2005. Quality Assurance Project Plan for Shallow
Water Monitoring. Virginia Institute of Marine Science, College of William and Mary,
Gloucester Point, VA.
YSI, Inc. 1999. Environmental Monitoring Systems, 6-Series Operations Manual.
chapter vii * Shallow-water Monitoring and Application for Criteria Assessment
-------
89
chapter\/l||
Framework for Chesapeake Bay
Tidal Waters 303(d) List
Decision-Making
BACKGROUND
Section 303(d) of the Clean Water Act and EPA Regulation 40CFR 130.7 requires
biennial identification of water segments that are not attaining water quality stan-
dards. These segments must have a total maximum daily load (TMDL) analysis
completed and allocations established that result in water quality standards attain-
ment. The states comply with this requirement through a process known as the
Integrated Reporting Requirements which covers the assessment and listing require-
ments through Clean Water Act sections 305(d), 305(b), and 314 (U.S. EPA 2005b).
Given that the 2006 integrated reporting documents would be the first prepared
under the states' newly adopted Chesapeake Bay water quality standards regulations,
a collaborative effort (among the EPA and watershed states) began in spring 2005 to
develop a decision-making framework for that portion of the 2006 submittals
addressing the Chesapeake Bay system. The Chesapeake Bay Program partners
reached agreement on several key assessment and listing issues. This chapter docu-
ments these agreements and presents the resultant flowchart for Chesapeake Bay
tidal-water listing decisions to guide Delaware, Maryland, Virginia, and the District
of Columbia in future 303(d) listing cycles.
LISTING CATEGORY DECISIONS
Each state-adopted, tidal-water designated use by Chesapeake Bay Program segment
(or formally adopted state sub-segment) is considered an individual spatial assess-
ment unit for the purposes of each state's 303(d) list (U.S. EPA 2003a, 2003b, 2004a,
2004b, 2005a).
If a segment has been previously listed in category 5—recognizing the recent adop-
tion of new Chesapeake Bay water quality standards—the original listing decision
chapter viii • Framework for Chesapeake Bay Tidal Waters 303(d) List Decision-Making
-------
90
should stand until sufficient data are available to fully assess attainment for all appli-
cable criteria components in each designated-use segment assessment unit. With
sufficient data, states can justify moving an individual designated-use segment, or
the segment as a whole, to another listing category. The lack of sufficient data for
full assessment of the applicable criteria is not justification for moving a category 5
(impaired) segment to category 3 (insufficient data).
If a segment's designated use was not previously listed in category 5, it can be listed
under category 3 if insufficient data exist to assess attainment of all applicable
criteria components. Because an individual segment may have up to five tidal-water
designated uses (see Table V-l in Chapter 5), the states can place individual
segments in multiple listing categories based on the criteria assessment results for
each designated use in the segment.
The preceding chapters document the different Chesapeake Bay water quality
criteria assessments. Across all Bay criteria, non-attainment is defined as any
percentage of non-attainment (even less than 1 percent) given that the CFD-based
criteria attainment assessment method already factors in the small percentage of
circumstances (in time and space) in which the criteria may be exceeded and still
fully protect the tidal-water designated use (U.S. EPA 2003a).
Given that multiple criteria often protect an individual designated use (e.g., separate
30-day mean, 7-day mean, and instantaneous minimum criteria required for protec-
tion of the open-water fish and shellfish designated use), full attainment of the
dissolved oxygen criteria must involve assessment of each applicable criterion indi-
vidually (U.S. EPA 2003a). In designated-use-segment assessment units for which
data are available to assess all applicable dissolved oxygen criteria, the states can
proceed with a full assessment of attainment of that segment's designated use. For
those units with insufficient data for one or more of these criteria, states should not
make any decisions on removing that designated-use segment from part 5 during that
listing cycle.
Until the EPA publishes methodologies for assessing the 7-day and 1-day mean,
along with the instantaneous minimum open-water and deep-water dissolved oxygen
criteria components, the EPA recommends the states rely strictly on the assessment
of the 30-day mean open-water and deep-water dissolved oxygen criteria for listing
decisions. For those open- and deep-water designated-use segments for which the
30-day mean criteria are in non-attainment, the jurisdictions should list the segment
on part 5 as impaired in the absence of data or methodologies for assessing the
remaining criteria components. For those designated-use segments in which the 30-
day mean open- or deep-water criteria are in attainment, the jurisdictions should
generate additional data and apply criteria assessment procedures to determine
chapter viii « Framework for Chesapeake Bay Tidal Waters 3Q3(cl) List Decision-Making
-------
91
attainment of the 7- and 1-day means as well as the instantaneous minimum criteria
components. If a segment was first listed in 2006 based on the 30-day mean open-
water and/or deep-water criteria and subsequent 30-day mean open-water and/or
deep-water criteria data now shows the segment to be in attainment, then the segment
may be delisted for these criteria.
The shallow-water bay grass designated use is in attainment if a sufficient number of
acres of SAV occur within the segment or if enough acres of shallow-water habitat
exist that meet the applicable water clarity criteria to support restoration of the
desired acreage of SAV for that segment (U.S. EPA 2003a, 2003b). Assessment of
either measure, or a combination of both, can serve as the basis for determining
attainment or impairment of the shallow-water bay grass designated use.
Since SAV is the ultimate biological measure of attainment of the designated use, in
the absence of sufficient shallow-water monitoring data necessary to determine the
available water clarity acres or assess water clarity criteria attainment using the
CFD-based criteria assessment procedure, EPA recommends the States assess
shallow-water bay grass designated use attainment/impairment based on the acres of
mapped SAV.
If a shallow-water bay grass designated-use segment meets its SAV restoration
acreage, that designated use-segment is in attainment of the designated use and
should be listed on part 2.
If such a segment does not meet its restoration acreage, the jurisdiction can then
assess attainment using water clarity acres or water clarity criteria as described in
Chapter 5. If the water clarity acres or water clarity criteria are attained based on
shallow-water monitoring data, then that segment is in attainment of the shallow-
water bay grasses designated use and should be listed on part 2.
Finally, if the water clarity restoration acres or water clarity criteria are not attained
using the same data, or if there are insufficient data to make a determination using
water clarity acres or water clarity criteria, then that segment is not in attainment of
the shallow-water bay grasses designated use and should be listed on part 5.
Any attainment/non-attainment determination of water clarity criteria based on mid-
channel-based monitoring is strictly diagnostic. These mid-channel data should not
directly form the basis for any listing decision based on attainment/non-attainment
of a segment's shallow-water bay grass designated use.
As described in Chapter 6, numerical chlorophyll a criteria attainment is assessed by
applying the appropriate numerical criteria over the applicable season for three years
using the CFD-based criteria assessment methodology.
chapter viii * Framework for Chesapeake Bay Tidal Waters 303(d) List Decision-Making
-------
92
OF
The benthic community health assessment is conducted in three phases to support
the states' tidal waters listing decisions (Llanso et al. 2005) (Appendices J and K).
Phase I evaluates the sample size from the segment during the five-year assessment
window. An impairment assessment based on benthic community health is not
possible if the sample size requirement is not met. The data, however, may still prove
useful as an adjunct to other aquatic life use data. If the sample size satisfies the
requirements of the statistical method (n > 10), a formal assessment of status (i.e.,
impaired vs. supports aquatic life use) is determined using the "percent degraded
area" statistical methodology (Phase II).
Phase II assesses aquatic life use impairment based on a comparison of the Chesa-
peake Bay benthic index of biotic integrity or benthic-IBI scores (Weisberg et al.
1997). This assessment is possible only when the number of benthic-IBI scores
within a segment is sufficient to meet the sample size requirement of the approved
statistical method (n > 10). Phase II can result in one of two possible outcomes: 1)
the segment is not impaired for aquatic life use due to benthic community status
(note that the segment may still be impaired due to failure of the other aquatic life
use subcategories or criteria); or 2) the segment fails to support aquatic life use due
to benthic community status and is assessed as impaired (part 5).
Phase III identifies the probable causes of assessed benthic impairment of the
segment using a diagnostic tool that can pinpoint potential sources of stress affecting
benthic community conditions in the Chesapeake Bay (Dauer et al. 2005). This
methodology can also identify causes of stress and quantify the magnitude of degra-
dation. In addition, it distinguishes stress due to contaminants from stress due to
other factors (Appendix L).
A Chesapeake Bay tidal-water designated-use criteria attainment assessment spread-
sheet has been developed to assist the states in reporting listing decisions for each
designated-use segment (Table VIII-1). The assessment reporting framework effi-
ciently documents relevant information as each segment goes through the listing
decision flowchart described below.
Table VIII-2 shows the example results of the Chesapeake Bay benthic analysis for
the 2006 303(d) reporting cycle. The benthic-IBI assessments are separate from the
Chesapeake Bay water quality criteria attainment assessment determinations and
reported for the segments as stand-alone or supplemental information for the states
to use in their 303(d) listing cycle decisions.
chapter viii « Framework for Chesapeake Bay Tidal Waters 3Q3(cl) List Decision-Making
-------
93
E
o
M—
Ol
C
o
Q_
0)
C
01
c
0)
E
c
ro
TO
u
0)
T3
0)
+-•
TO
C
gi
't/i
0)
T3
TO
TO
GO
TO
0)
Q_
TO I
U
(£>
O
- ^
1
CN CN
rn ro
f%J fSJ
IT) CO
1
: Q_ Q_ -^ -^l
: Q_ Q_ .^ J2
-^1 D_ Q_ -^2 D_
Q Q Q
JL' Ji! Q_ D_
1
:§•
•
> > O O Q
000°
-------
94
c
01
_Q
ID
O
O
rxl
t/i
0)
+J
TO
^H
O ^
!J1 '2
o> a
use
i. a o
a I -s
= S cs
to o T=
T3 U 2
£ o wt
0 .pj D
2-1°
s ^
rTi PQ
^ ^ -*^
^ ^ S3
S ^ * o
s-S |u
S "3 "^ -^
>_. S o ffs
5 g a S
Frt ' ' & ®
wo So
Q *
JS 11
§•£ su
| •§ .2 o
to *> ?? ^
"^ ^ o "«
Q> Q> g ^j
"» ja ca o
2 S T= H
Sf ^ §=S
Q -5s?
<1 ^
-a * TS
"!? 3 ^ ^ • ^ w
^ J .2 3 ^ o
gj g g^ rt ®
•3, « 1 •§ ii 's
|*.H
S to
B HH
^PQ
TS "S j^
•" "2 «8
'S 2 2
a wo 3
SO* ^H
hH O
"S
I
wt
t/}
"3
Kj
1
O
U
"3
1
•3
OJ
GO
O
O
o
•3
OJ
GO
O
O
o
CS'
r^
| JMSMHb
d
o
oo
oo
'
oo
oo
*/")
^2.
oo
CN
r^
| ELIPHa
O
Q
o
ON
CS
OO
oo
*/")
1^
1^
/">
^
r^i
^
O
ON
r^
| YRKPHa
O
Q
o
CS
CS
>/">
•
^
o
CS'
r^
| CHSMH
d
o
1
o
o
o
^
^
| APPTFa
•2
o
Q.
O
^H
P-l
O
o
o
*~~*
o
(N
^
^
\ CHSTF
d
o
1
o
o
o
^
^
| CHOTF
d
o
1
o
o
^.
CS
CS
^^
\ JMSTFa
d
o
1
o
CN
1^
^°.
^O
So
ON
-------
95
The Chesapeake Bay Program partners reached agreement on how to apply the
results of the refined criteria assessment procedures most effectively in making
Chesapeake Bay tidal-water listing decisions. The resultant decision-making frame-
work, presented here in the form of a flowchart, can guide jurisdictional decisions in
preparing future integrated reporting cycle submissions for the Chesapeake Bay
system (Figure VIII-1).
All of the designated-use segment combinations for the five possible tidal-water
designated uses—migratory fish spawning and nursery, shallow-water bay grass,
open-water fish and shellfish, deep-water seasonal fish, and shellfish and deep-
Are the designated-use segment
currently listed as impaired
(Part 5) for the water
qualityVbiological2 parameter?
Yes
I
No
Are the data available
to reassess criteria
attainment?
I
Are the data available
to assess criteria
attainment?
Are the available data
sufficient to make an
attainment decision?
No
I
Are the available data
sufficient to make an
attainment decision?
Yes
Criteria Attained?
List Part 2
List Part 5
Dissolved oxygen, water clarity and chlorophyll a.
2SAV acreage and benthic index of biotic integrity.
VIII-1. The 303(d) listing decision flow chart for assessing tidal waters designated
uses in Chesapeake Bay and tidal tributaries.
chapter viii « Framework for Chesapeake Bay Tidal Waters 3(B(d) List DecisiorvMaking
-------
96
channel seasonal refuge—along with the relevant dissolved oxygen, water
clarity/SAV restoration acreage, and chlorophyll a criteria are applied through this
listing decision flowchart. Benthic index of biotic integrity data are also evaluated
for listing decisions.
The listing decision flowchart starts with each designated use-segment-applicable
criterion combination, asking whether that segment was previously listed in category
5 as impaired based on the specific water quality (dissolved oxygen, water clarity,
chlorophyll a) or biological (SAV acreage) criterion parameter. If yes, its initial
listing status remains in category 5 pending new criteria attainment assessments. If
no, then the flowchart questions whether data now exist to assess criteria attainment.
SEGMENTS PREVIOUSLY LISTED AS IMPAIRED
At the second level, the flowchart queries whether the available data are sufficient to
reassess criteria or index attainment. If yes, the third level asks if the applicable
criteria or index is attained. If all applicable criteria components and indices have
been attained, the designated-use segment is then listed in part 2. If no, the desig-
nated-use segment remains in part 5. If insufficient data exist at the second level to
assess criteria attainment/index attainment, the designated-use segment previously
listed as impaired remains in part 5.
SEGMENTS NOT PREVIOUSLY LISTED AS IMPAIRED
At the second level, the flowchart queries whether the available data are sufficient
to reassess criteria or index attainment. If yes, the third level determines whether
the applicable criteria or index is attained. If all applicable criteria components and
indices have been attained, the designated-use segment is then listed in part 2. If
no, the designated-use segment is listed as impaired in part 5. If insufficient data
exist at the second level to assess criteria attainment/index attainment, the desig-
nated-use segment remains in part 3.
SHALLOW-WATER DESIGNATED-USE LISTING DECISIONS
If a shallow-water designated-use segment does not meet its SAV restoration
acreage, the EPA recommends that the state list this designated-use segment in cate-
gory 5 presuming the shallow-water monitoring data needed to assess water clarity
acres/criteria attainment do not exist.
LITERATURE CITED
Dauer, D.M., M.F. Lane, and RJ. Llanso. 2005. Addendum to the Report: Development of
Diagnostic Approaches to Determine Sources of Anthropogenic Stress Affecting Benthic
Community Condition in the Chesapeake Bay. Prepared for U.S. Environmental Protection
chapter viii • Framework for Chesapeake Bay Tidal Waters 303(d) List Decision-Making
-------
97
Agency, Chesapeake Bay Program Office, by Department of Biological Sciences, Old
Dominion University, Norfolk, VA, and Versar, Inc., Columbia, MD.
Llanso, R.J., J.H. V01stad, D.M. Dauer, and M.F. Lane. 2005. 2006 303(D) Assessment
Methods For Chesapeake Bay Benthos. Final report submitted to Virginia Department of
Environmental Quality, September 2005. Versar Inc., Columbia, MD, and Old Dominion
University, Norfolk, VA.
U.S. Environmental Protection Agency (U.S. EPA). 2003a. Ambient Water Quality Criteria
for Dissolved Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and its Tidal
Tributaries. EPA 903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis,
MD.
U.S. Environmental Protection Agency. 2003b. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability. EPA 903-R-03-004. Region III Chesapeake Bay
Program Office Annapolis, MD.
U.S. Environmental Protection Agency. 2004a. Chesapeake Bay Program Analytical
Segmentation Scheme: Revisions, Decisions and Rationales 1983-2003. EPA 903-R-04-008.
CBP/TRS 268/04. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2004b. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability: 2004 Addendum. EPA 903-R-04-006. Region III
Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2005a. Chesapeake Bay Program Analytical
Segmentation Scheme: Revisions, Decisions and Rationales 1983-2003: 2005 Addendum.
EPA 903-R-05-004. CBP/TRS 278-06. Region III Chesapeake Bay Program Office,
Annapolis, MD.
U.S. Environmental Protection Agency. 2005b. Guidance for 2006 Assessment, Listing and
Reporting Requirements Pursuant to Sections 303(d), 305(b) and 314 of the Clean Water Act.
July 29, 2005. Office of Water, Office of Wetlands, Oceans and Watersheds, Assessment and
Watershed Protection Division. Washington, D.C.
Weisberg, S.B., J.A. Ranasinghe, D.M. Dauer, L.C. Schaffner, RJ. Diaz, and J.B. Frithsen.
1997. An estuarine benthic index of biotic integrity (B-IBI) for Chesapeake Bay. Estuaries
20: 149-158.
-------
98
Id-2
°c
CART
CBP
CDOM
CFD
cells/ml
Chla
DIN
DO
gCm-'d'1
GLM
HAB
IDW
kg m3m
km
LOAEL
m
mV
mg
inverse of the distance
squared
degrees Celsius
classification and
regression tree
Chesapeake Bay Program
colored dissolved organic
matter
cumulative frequency diagram
cells per milliliter
chlorophyll a
dissolved inorganic nitrogen
dissolved oxygen
grams of carbon per meter
squared per day
general linear model
harmful algal bloom
inverse-distance weighting
kilograms per cubic meter
per meter
kilometers
lowest observable acute
effects level
meter
cubic meters per second
milligram
mg chla m
mg liter'1
NASS
NH4
NO2
NO3
O2
PAR
PO4
ppt
PSU
QA/QC
SAV
STAC
TMDL
TSS
U.S. EPA
milligrams of chlorophyll a
per meter squared
milligrams per liter
non-algal suspended solids
ammonium
nitrite
nitrate
oxygen
photosynthetically active
radiation
dissolved inorganic
phosphorous/
orthophosphorous
parts per thousand
practical salinity unit
quality assurance/quality
control
submerged aquatic
vegetation
Science and Technical
Advisory Committee
total maximum daily load
total suspended solids
United States
Environmental
Protection Agency
//g/kg micrograms per kilogram
jug liter"1
% saturation
micrograms per liter
percent oxygen saturation
-------
A-1
appendix
The Cumulative Frequency
Diagram Method for
Determining Water Quality
Attainment
Report of the Chesapeake Bay Program STAC Panel to
Review of Chesapeake Bay Program Analytical Tools
STAC Publication 06-003
9 October 2006
Panel Members:
David Secor, Chair (Chesapeake Biological Laboratory, University of Maryland
Center for Environmental Science)
Mary Christman (Dept. of Statistics, University of Florida)
Frank Curriero (Departments of Environmental Health Sciences and Biostatistics,
Johns Hopkins Bloomberg School of Public Health)
David Jasinski (University of Maryland Center for Environmental Science)
Elgin Perry (statistics consultant)
Steven Preston (US Geological Survey, Annapolis)
Ken Reckhow (Dept. Environmental Sciences & Policy Nicholas School of the
Environment and Earth Sciences, Duke University)
Mark Trice (Maryland Department of Natural Resources)
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-2
In accordance with the Chesapeake 2000 Agreement, the Chesapeake Bay Program
has recently implemented important modifications to (1) ambient water quality
criteria for living resources and, (2) the procedures to determine attainment of those
criteria. A novel statistical tool for attainment, termed the Cumulative Frequency
Diagram (CFD) approach, was developed as a substantial revision of previous attain-
ment procedures, which relied upon a simple statistical summary of observed
samples. The approach was viewed as advantageous in its capacity to represent
degrees of attainment in both time and space. In particular, it was recognized that the
CFD could represent spatial data in a synoptic way: data that is extensively collected
across diverse platforms by the Chesapeake Bay Program Water Quality Monitoring
Program. Because the CFD approach is new to Bay Program applications, under-
lying statistical properties need to be fully established. Such properties are critical if
the CFD approach is to be used to rigorously define regional attainments in the
Chesapeake Bay.
In Fall 2005, the Chesapeake Bay Program Scientific, Technical and Advisory
Committee charged our working group to provide review and recommendations on
the CFD attainment approach. As terms of reference we used guidelines of Best
Available Science recently published by the American Fisheries Society and the
Estuarine Research Federation. Statistical issues that we reviewed included,
1. What are the specific analytical/statistical steps entailed in constructing CFD
attainment curves and how are CFDs currently implemented? (Section 2)
2. How rigorous is the spatial interpolation process that feeds into the CFD
approach? Would alternative spatial modeling procedures (e.g., kriging)
substantially improve estimation of water quality attainment? (Section 3)
3. What are the specific analytical/statistical steps entailed in constructing CFD
reference curves? (Section 4)
4. What are the statistical properties of CFD curves? How does sampling density,
levels of attainment, and spatial covariance affect the shape of CFD curves?
What procedures are reliable for estimating error bounds for CFD curves?
(Section 5)
5. From a statistical viewpoint, does the CFD approach qualify as best available
science? (Section 6)
6. What are the most important remaining issues and what course of directed
research will lead to a more statistically rigorous CFD approach over the next
three years? (Section 7)
The central element of our work was a series of exercises on simulated datasets
undertaken by Dr. Perry to better evaluate 1) sample densities in time and space, 2)
varying levels of attainment, and 3) varying degrees of spatial and temporal
appendix a * The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-3
covariance. Further, trials of spatial modeling on fixed station Chesapeake Bay water
quality data by Dr.s Christman and Curriero were conducted to begin to evaluate
spatial modeling procedures. These exercises, literature review and discussions
leading to consensus opinion are the basis of our findings. In August 2006, the
working group supplied preliminary findings and related text for use in the 2006
CBP Addendum to Ambient Water Quality Criteria that is now under review.
1. The CFD approach is feasible and efficient in representing water quality
attainment.
The CFD approach can effectively represent the spatial and temporal dimensions
of water quality data to support inferences on whether regions within the Chesa-
peake Bay attain or exceed water quality standards. The CFD approach is
innovative but could support general application in water quality attainment
assessments in the Chesapeake Bay and elsewhere. The CFD approach meshes
well within the Chesapeake Bay Program's monitoring and assessment
approaches, which have important conceptual underpinnings (e.g., segments
defined by designated uses).
In accepting the CFD as the best available approach for using time-space data, the
panel contrasted it with the previous method and those sustained by other juris-
dictions. The previous method used by the Chesapeake Bay Program, similar to
the approaches used in other states, was simply based on EPA assessment guid-
ance in which all samples in a given spatial area were compiled and attainment
was assumed as long as > 10% of the samples did not exceed the standard. In this
past approach all samples were assumed to be fully representative of the specified
space and time and were simply combined as if they were random samples from
a uniform population. This approach was necessary at the time because the tech-
nology was not available for a more rigorous approach. But it neglected spatial
and temporal patterns that are known to exist in the standards measures. The CFD
approach was designed to better characterize those spatial and temporal patterns
and weight samples according to the amount of space or time that they actually
represent.
2. CFD curves are influenced by sampling density and spatial and temporal
covariance. These effects merit additional research. Conditional simulation
offers a productive means to further discover underlying statistical proper-
ties and to construct confidence bounds on CFD curves, but further directed
analyses are needed to test the feasibility of this modeling approach.
The panel finds that the CFD approach in its current form is feasible, but that
additional research is needed to further refine and strengthen it as a statistical tool.
The CFD builds on important statistical theory related to the cumulative distribu-
tion function and as such, its statistical properties can be simulated and deduced.
Through conditional simulation exercises, we have also shown that it is feasible
to construct confidence ellipses that support inferences related to threshold curves
appendix a * The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-4
or other tests of spatial and temporal compliance. Work remains to be done in
understanding fundamental properties of how the CFD represents likely covari-
ances of attainment in time and space and how temporal and spatial correlations
interact with sample size effects. Further, more work is needed in analyzing biases
across different types of designated use segments. The panel expects that a two-
three year time frame of directed research and development will be required to
identify and measure these sources of bias and imprecision in support of attain-
ment determinations.
3. The success of the CFD-based assessment will be dependent upon decision
rules related to CFD reference curves. For valid comparisons, both reference
and attainment CFDs should be underlain by similar sampling densities and
spatial covariance structures.
CFD reference curves represent desired segment-designated use water quality
outcomes and reflect sources of acceptable natural variability. The reference and
attainment curves follow the same general approach in derivation: water quality
data collection, spatial interpolation, comparison to biologically-based water
quality criteria, and combination of space-time attainment data through a CFD.
Therefore, the biological reference curve allows for implementation of threshold
uncertainty as long as the reference curve is sampled similarly to the attainment
curve. Therefore, we advise that similar sample densities are used in the deriva-
tion of attainment and reference curves. As this is not always feasible, analytical
methods are needed in the future to equally weight sampling densities between
attainment and reference curves.
4. In comparison with the current IDW spatial interpolation method, kriging
represents a more robust method and was needed in our investigations on
how spatial covariance affects CFD statistical inferences. Still, the IDW
approach may sufficiently represent water quality data in many instances
and lead to accurate estimation of attainment. A suggested strategy is to use
a mix of IDW and kriging dependent upon situations where attainment was
grossly exceeded or clearly met (IDW) versus more-or-less "borderline"
cases (kriging).
The current modeling approach for obtaining predicted attainment values in space
is Inverse Distance Weighting (IDW), a non-statistical spatial interpolator that uses
the observed data to calculate a weighted average as a predicted value for each loca-
tion on the prediction grid. IDW has several advantages. It is a spatial interpolator
and in general such methods have been shown to provide good prediction maps. In
addition, it is easy to implement and automate because it does not require any deci-
sion points during an interpolation session. IDW also has a major disadvantage - it
is not a statistical method that can account for sampling error.
Kriging is also a weighted average but first uses the data to estimate the weights
to provide statistically optimal spatial predictions. As a recognized class of statis-
tical methods with many years of dedicated research into model selection and
-------
A-5
estimation, kriging is designed to permit inferences from sampled data in the pres-
ence of uncertainty. Thus the quantity and distribution of the sample data are
reflected in those inferences. Indeed, the panel's initial trials on the role of spatial
sources of error in the CFD have depended upon the ability to propagate kriging
interpolation uncertainty through the CFD process in generating confidence inter-
vals of attainment.
In comparison to IDW, kriging is more sophisticated but requires greater expertise
in implementation. Kriging is available in commercial statistical software and
also in the free open source R Statistical Computing Environment, and requires
geostatistical expertise and programming skills for those software packages.
Segment by segment variogram estimation and subsequent procedures would
require substantial expert supervision and decision-making. Thus, this approach is
not conducive to automation. On the other hand, there may be CBP applications
where the decision on attainment is clearly not influenced to any substantial
degree by the method of spatial interpolation. One suggested strategy is to use a
mix of IDW and kriging - dependent upon situations where attainment was
grossly exceeded or clearly met (IDW) versus more-or-less "borderline" cases
(kriging).
5. More intensive spatial and temporal monitoring of water quality will
improve the CFD approach but will require further investigations on the
influence of spatial and temporal covariance structures on the shape of the
CFD curve. This issue is relevant in bringing 3-dimensional interpolations
and continuous monitoring streams into the CFD approach.
In the near future, the panel sees that the CFD approach is particularly powerful
when linked to continuous spatial data streams made available through the cruise-
track monitoring program, and the promise of continuous temporal data through
further deployment of remote sensing platforms in the Chesapeake Bay (Chesa-
peake Bay Observing System: http://www.cbos.org/). These data sets will support
greater precision and accuracy in both threshold and attainment determinations
made through the CFD approach but will require directed investigations into how
data covary over different intervals of time and space. Further, there may be
important space-time interactions that confound the CFD attainment procedure.
Some of the assessments for the Bay such as that for dissolved oxygen require
three dimensional interpolation, but the field of three dimensional interpolation is
not as highly developed as that of two dimensional interpolation. Kriging can be
advantageously applied in that it can use information from the data to develop
direction dependent weighted interpolations (anisotropy). Kriging can include
covariates like depth. Options for implementing 3-D interpolation include:
custom IDW software, custom kriging software using GMS routines, or custom
kriging software using the R-package.
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-6
The panel identified critical research tasks that need resolution in the near future.
The following is a list of critical aspects of that needed research. These research
tasks appear roughly in order of priority. However, it must be recognized that it is
difficult to formulate as set of tasks that can proceed with complete independence.
For example, research on task 1 may show that the ability to conditionally simulate
the water quality surface is critical to resolving the sample size bias issue. This
discovery might eliminate IDW as a choice of interpolation under task 3. The Panel
has made significant progress on several of these research tasks and CBP is encour-
aged to implement continued study in a way that maintains the momentum
established by our panel.
1. Effects of Sampling Design on CFD Results
(a) Continue simulation work to evaluate CFD bias reduction via conditional
simulation.
(b) Investigate conditional simulation for interpolation methods other than
kriging—this may lead to more simulation work.
(c) Implement and apply interpolation with condition simulation on CBP data.
2. Statistical inference framework for the CFD
(a) Conduct confidence interval coverage experiments.
(b) Investigate confidence interval methods for non-kriging interpolation
methods.
(c) Implement and evaluate confidence interval procedures.
3. Choice of Interpolation Method
(a) Implement a file system and software utilizing kriging interpolation for CBP
data.
(b) Compare interpolations and CFDs based on kriging and inverse distance
weighting (IDW).
(c) Investigate nonparametric interpolation methods such as LOESS and spline
approaches.
4. Three-Dimensional Interpolation
(a) Implement 2-D kriging in layers to compare to current approach of 2-D IDW
in layers.
(b) Conduct studies of 3-D anisotrophy in CBP data.
(c) Investigate software for full 3-D interpolation.
5. High Density Temporal Data
(a) Develop methods to use these data to improve temporal aspect of CFD
implementation.
(b) Investigate feasibility of 4-Dimensional interpolation.
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-7
In June 2000, Chesapeake Bay Program (CBP) partners adopted the Chesapeake
2000 agreement (http://www.chesapeakebay.net/agreement.htm), a strategic plan
that calls for defining the water quality conditions necessary to protect aquatic living
resources. These water quality conditions are being defined through the development
of Chesapeake Bay specific water quality criteria for dissolved oxygen, water clarity,
and chlorophyll_a to be implemented as state water quality standards by 2005. One
element of the newly defined standards is an assessment tool that addresses the
spatial and temporal variability of these water quality measures in establishing
compliance. This tool has become known as the Cumulative Frequency Diagram
(CFD).
The (CFD) was first proposed as an assessment tool by Paul Jacobson, of Langhei
Ecology (www.LangheiEcology.com). At that time Dr. Jacobson was consulting
with the Chesapeake Bay Program as a member of the Tidal Monitoring Network
Redesign Team. Within this group, the CFD concept gained immediate recognition
and support as a novel approach that permitted independent modeling of the time and
space dimensions of the continuous domain that underlies Chesapeake Bay water
quality parameters. In addition, because preparation of the CFD uses spatial inter-
polation, the approach can allow integration of data collected on different spatial
scales such as fixed station data and cruise track data.
While the benefits of the CFD approach has been recognized (U.S. EPA 2003) and
the the CBP has begun implementation of the approach for certain water quality
parameters and segments of the Chesapeake Bay, investigations of the statistical
properties revealed that the underlying shape parameters of the CFD were sensitive
not only to rates of compliance but also to sampling design elements such as sample
density. The novelty of the approach coupled with concerns about its statistical
validity motivated the Chesapeake Bay Program to request that its Scientific and
Technical Advisory Committee (http://www.chesapeake.org/stac/) empanel a group
with expertise in criteria assessment, spatial data interpolation, and statistics to
assess the scientific defensibility of the CFD. Here we report the findings of this
panel.
The primary goal of this panel is to provide an initial scientific review of the CFD
compliance approach. This review addresses a wide range of issues including: bias
and statistical rigor, uncertainty, practical implementation issues, and formulation of
reference curves. Because of the novelty of the CFD approach, the panel has endeav-
ored to research and explain the properties of the CFD and spatial modeling upon
which the CFD approach depends to provide a basis for this evaluation. These activ-
ities are beyond the scope of the typical review. However, because so little is known
about the CFD, it was necessary to expand the knowledge base.
The report is organized into 7 sections. In Section 2 of this report we present the
CFD approach as a series of steps, each of which needs to be considered carefully in
evaluating its statistical properties. Spatial interpolation is a critical but the most
statistically nuanced step in the CFD approach. Spatial interpolation of water quality
data in the CBP has to date received little statistical review. In Section 3 we evaluate
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-8
alternative geostatistical methods as they pertain to the CFD approach. The CFD
approach is an attainment procedure, which depends upon statistical comparison
between attainment and reference curves. In Section 4, we present alternative types
of references curves and discuss statistical properties of each. In Section 5 the statis-
tical properties of CFD curves (applicable to both attainment and reference curves)
is elucidated through a series of conditional simulation trials.
In addition to this primary charge, the panel is sensitive to the fact that the CFD will
be employed in the enforcement of water quality standards. Use as a regulatory tool
imposes a standard of credibility, which we review in Section 6. We use here "best
available science" and "best science" criteria to evaluate the overall validity and
feasibility of the CFD approach, following guidelines established by the American
Fisheries Society and Estuarine Research Federation (Sullivan et al. 2006). These
follow other similar criteria (e.g., The Daubert Criteria (Daubert v. Merrell Dow
Pharmaceuticals, Inc., 1993) and include:
1. A clear statement of objective
2. A conceptual model, which is a framework for characterizing systems, sating
assumptions, making predictions, and testing hypotheses.
3. A good experimental design and a standardized method for collecting data.
4. Statistical rigor and sound logic for analysis and interpretation.
5. Clear documentation of methods, results, and conclusions
6. Peer review.
The panel has made progress in better understanding statistical properties of the
CFD approach and overall, we recommend it as a feasible approach and one that
qualifies under most criteria for best available science. Still, we believe that our
efforts should only represent the beginning of a longer term effort to (1) Use simu-
lations and other means to support statistical comparisons of CFD curves; and (2)
Support the CBP's efforts to model water quality data with sufficient rigor in both
spatial and temporal dimensions. Research and implementation recommendations
follow in Section 7.
The water quality criteria assessment methodology currently proposed by the E.P.A.
Chesapeake Bay Program (CBP) involves the use of a Cumulative Frequency
Diagram (CFD) curve. This curve is represented in a two dimensional plane of
percent time and percent space. This document briefly discusses the reasoning that
lead to the development of this assessment tool. The proposed algorithm for esti-
mating the CFD is given and illustrated with small data sets. Some properties and
unresolved issues regarding the use of the CFD are briefly discussed. In Section 5,
simulation studies explore in greater specificity the multiple issues related to error
and bias in the CFD approach.
appendix a * The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-9
CFD
The CFD assessment methodology evolved from a need to allow for variability in
water quality parameters due to unusual events. For the water quality parameter to
be assessed, a threshold criterion is established for which it is determined that water
quality that exceeds this threshold is in a degraded state (For simplicity, we will
speak of exceeding the threshold as representing degradation, even though for some
water quality constituents such as dissolved oxygen, it is falling below a threshold
that constitutes degradation). Because all water quality parameters are inherently
variable in space and time, it is unlikely that a healthy bay will remain below the
threshold in all places at all times. In the spatial dimension, there will be small
regions that persistently exceed the threshold due to poor flushing or other natural
conditions. It is recognized by CBP that these small regions of degraded condition
should not lead to a degraded assessment for the segment surrounding this small
region. Similar logic applies in the temporal dimension. For a short period of time,
water quality in a large proportion of a segment may exceed the threshold, but if this
condition is short lived and the segment quickly returns to a healthy state, this does
not represent an impairment of the designated use of the segment. Recognition that
ephemeral exceedances of the threshold in both time and space do not represent
persistent impairment of the segment leads to an assessment methodology that will
allow these conditions to be classed as acceptable while conditions of persistent and
wide spread impaired condition will be flagged as unacceptable. The assessment
methodology should first ask how much of the segment (for simplicity, a spatial
assessment unit is called a segment, but more detail is given on spatial assessment
units in Section 2) is not in compliance with the criteria (percent of space) for every
point in time. In a second step the process should ask how often (percent of time) is
a segment out of compliance by more than a fixed percent of space. The results from
these queries can be presented in graphical form where percent of time is plotted
against percent of space (Figure 2.1). It is arbitrary to treat space first and time
second. A similar diagram could be obtained by first computing percent noncompli-
ance in time and then considering the cumulative distribution of percent time over
space.
If a segment is generally in compliance with the criterion, then one expects a high
frequency of dates where the percent out of compliance is low. In this case, the CFD
should descend rapidly from the upper left corner and pass not too far from the lower
left corner and then proceed to the lower right corner. The trace in Figure 2.1 shows
the typical hyperbolic shape of the CFD. The closer the CFD passes to the origin
(lower left corner), the better the compliance of the segment being assessed. As the
CFD moves away from the origin, a higher frequency of large percents of space out
of compliance is indicated.
an of CFD
The algorithm developed by CBP for estimating the CFD is most easily described as
a series of steps. These steps are given in bullet form to provide a frame work for the
overall approach. The quickly defined framework is followed by a simple example.
This in turn is followed by more detailed discussion of each step.
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-10
LO
0.9
0.8
0.7
"S
_ 0.5
0.3
0.2
0.1
0.0
0.0 0.1 0.2 03 0.4 0.5 0.6
Fraction of Space
0.7
0.8
0.9 LO
2.1. Illustration of CFD for 12 dates.
The steps:
1. Collect data from a spatial network of locations on a series of dates in a three
year assessment period.
2. For each date, interpolate the data for the entire system (e.g. mainstem bay) to
obtain estimates of water quality in a grid of interpolation cells.
3. For each interpolation cell assess whether or not the criterion is exceeded.
4. For each assessment unit (e.g. segment), compute the percentage of interpolator
cells that exceed the criterion as an estimate of the percent of area that exceeds
the criterion.
5. Rank the percent of area estimates for the set of all sample days in the assess-
ment period from largest to smallest and sequentially assign to these ranked
percents a value that estimates percent of time.
6. Plot the paired percent of time and percent of area data on a graph with percent
of area on the abscissa and percent of time on the ordinate. The resulting curve
is the Cumulative Frequency Diagram.
1. Compare the CFD from a segment being assessed to a reference CFD. If at any
point the assessment CFD exceeds the reference CFD, that is, a given level of
spatial noncompliance occurs more often than is allowed, then the segment is
listed as failing to meet it's designated use.
appendix a « The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-11
For this example, assume a segment for which the interpolation grid is 4 cells by 4
cells. In reality, the number of grid cells is much larger. Also let data be collected on
5 dates. Typically data would be monthly for a total of 36 dates. Let the criterion
threshold for this fictitious water quality parameter be 3. In what follows, you will
find an illustration of the steps of computing the CFD for these simplified
constraints. The three columns of the next page show the first three steps. Column
1 shows fictional data for five dates for five fixed locations in a 2 dimensional grid.
Column 2 shows a fictional interpolation of these data to cover the entire grid.
Column 3 shows the compliance status of each cell in the grid where 1 indicates
noncompliance and 0 indicates compliance.
Step 1. Collect data at
known locations.
date 1
3
2
5
3
1
date2
1
1
3
1
1
date3
4
1
2
2
1
date4
1
4
2
4
1
dateS
1
1
2
3
1
Step 2. Interpolate the
data to grid cells.
date 1
3
4
3
2
4
4
3
3
5
5
4
3
3
2
1
1
date2
date3
date4
dateS
1
2
1
1
2
2
3
1
3
3
2
1
1
2
1
1
4
3
2
1
3
2
2
1
2
2
1
1
2
1
1
1
1
2
3
4
2
2
3
3
3
2
2
1
4
3
1
1
1
2
1
1
2
2
1
1
3
2
1
1
3
2
1
1
Step 3. Determine
compliance status of each
cell.
date 1
1
1
1
0
1
1
1
1
1
1
1
1
1
0
0
0
date2
date3
date4
dateS
0
0
0
0
0
0
1
0
1
1
0
0
0
0
0
0
1
1
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
0
1
1
1
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-12
Step 4: Percent compliance by date.
sample date
date 1
date 2
date3
date 4
dateS
percent
space
75.00%
18.75%
18.75%
43.75%
12.50%
Step 5. Rank the percent of space values
and assign percent of time = (100*R/(M+1.0)),
where R is rank and M is total number of dates.
sample date
date 1
date 4
date 2
date3
date5
ranked
percent
space
75.00%
43.75%
18.75%
18.75%
12.50%
cumulative
percent time
16.67
33.33
50.00
66.67
83.33
Steps 6 and 7: The plot of the CFD
and the comparison to the reference
curve are shown in Figure 2.2. For
this hypothetical case the assessment
area would be judged in noncompli-
ance. For a percent area of 18.75, the
allowable frequency on the reference
curve is about 53%. That is, 18.75%
of the segment area should not be out
of compliance more that 53% of the
time. For date 3, the estimated
frequency of 18.75% noncompliance
is 66.67%. Thus the frequency of
18.75% of space out of compliance is
in excess of the 53% allowed. The
reference curve is exceeded for dates
4 and 1 as well. Note: in this cumula-
tive distribution framework, the actual
date is not relevant. One should not
infer that noncompliance occurred on
that date if the data point associated
with a date falls above the reference.
Date is being used here as a label for
each coordinate pair.
£
I
•s
1.0
0.9
(LS
0.7
,S 0.6
0.5-
0,1
OJ
0,2
0.1
0.0-1,
0,0 0,1 02 0,3 0,4 0,5 0,6
Fraction of Space
0,7
0.8
0,9
to
2,2, Graphical representation of CFD from the above example (' + ') with hypothetical reference
curve ( smooth).
-------
A-13
CFD
As defined above, the CFD is a data driven formulation. But the data used to formu-
late the CFD are a sample of points taken from a population. Defining the CFD
becomes complex when one considers the many different levels for which it might
be defined. At one level, the CFD might be defined based on the true state of a
segment. Imagine that the state of a segment could be frozen for sufficient time to
permit deployment of an analog sampler (that is one that measures water quality
continuously rather than in discrete samples) to assess the percent of area out of
compliance at that instant. Now stretch that imagination one step further to relax the
condition that the segment be frozen and allow that these analog measurements of
percent of area out of compliance be determined continuously in time. With this
information, a determination of the CFD for the true state of the segment is possible.
While the information needed to construct the ideal CFD is not obtainable, it is
important to ask how well the CFD based on obtainable data represents this ideal
(see also Section 5). Is a data driven CFD consistent for the ideal CFD in the statis-
tical sense? Loosely speaking, consistency implies that the data driven CFD should
get closer to the ideal CFD as more data are used. Is the data driven CFD unbiased
for the ideal CFD? Unbiasedness implies that even with small amounts of data, the
data driven CFD on average covers the ideal CFD.
One might argue that if both the assessment CFD and the reference CFD are data
driven, then it is not important for the CFD to approximate the ideal. Even so, it is
important to understand the behavior of the CFD as a function of samples size and
the relative temporal and spatial contributions to the variance in the water quality
parameter. If the curve changes shape as a more data are used, this could result in
unfair comparisons between assessment and reference regions. In Section 4, statis-
tical properties for both types of reference curves are evaluated further.
Two approaches to defining the reference curve are being considered. One is a
biologically based definition. The idea is to identify appropriate reference regions
with healthy biological indicators and compute the reference CFD for these regions.
For example, healthy benthic IBI scores might be used as indicators of adequate
bottom dissolved oxygen. Thus after stratifying by salinity zone and perhaps other
factors, a series of dissolved oxygen reference CDF curves could be computed from
the existing 20+ year monitoring data base. When it is not possible to establish a
reference condition some more arbitrary device must be employed. Alternatives are
discussed in Section 4.0.
of
Step 1 - Data Collection. One of the advantages of the CFD approach is that it will
accommodate a variety of input data and still arrive at the same assessment endpoint.
Data collection methods currently in place include: fix station data, cruise track data,
continuous monitor data, aircraft flight path data, and satellite imagery data. Because
of the interpolation step, all of these data can be used (and potentially combined)
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-14
with varying degrees of success to estimate the total spatial (to the limit of interpo-
lator pixel size) distribution of a water quality constituent. As noted above, one could
construct this process by reversing the roles of time and space. That is, first interpo-
late over time and then build a cumulative distribution in space. In theory it is an
abitrary choice to first standardize the data over space by interpolation and then
construct the cumulative distribution in time. However, in practice, there is a greater
diversity of sampling designs over space and therefore it is the sampling in the
spatial dimension more than the temporal that creates many types of data that must
be forced to a common currency.
Step 2 - Interpolation. Interpolation is the step that puts data collected at various
spatial intensities on a common footing. On the one hand, this is advantageous
because data collected at many spatial intensities are available for the assessment
process. On the other hand, it can be misleading to accept interpolated surfaces from
different data sources as equivalent without qualifying each interpolation with a
measure of the estimation error that is associated with each type of data. Clearly an
interpolation based on hundreds of points per segment (such as cruise track data) will
more accurately reflect the true noncompliance percent when compared to an inter-
polation based on two or three points per segment (such a fixed station data). Of the
various types of interpolation algorithms available, the method proposed for this
assessment is kriging. Kriging offers the best available approach for the estimation
error associated with interpolation.
Step 3 - Pointwise Compliance. Determining the percent of compliance of each cell
from each interpolation would seem to be a simple step. If the estimated value for a
cell exceeds the criterion then that cell is out of compliance.
While interpolation allows for a standardization of many types of data, pointwise
compliance allows for standardization of many criteria. Because compliance is
determined at points in time and space, it is possible to vary the compliance criteria
in time and space. If different levels of a water quality constituent are acceptable in
different seasons, then the criterion can vary by season. It is possible to implement
different criteria over space for a segment that bridges oligohaline and mesohaline
salinity regimes. It would even be possible to let the criterion be a continuous func-
tion of some ancillary variable such as temperature or salinity. All that is required is
that the final determination be yes or no for each interpolator cell.
Even the simplicity of this concept becomes diminished when issues of interpolation
error are considered. Consider the assessment of two interpolator cells from an inter-
polation based on cruise track data. One cell near the cruise track has an estimated
value is 4 and a standard error of 0.1. A second cell far from the cruise track has an
estimated value of 4 and a standard error of 1.0. If the criterion were 3.0, it is fairly
certain that the first cell represents exceedance. It is much less certain that the second
cell represents exceedance. In the simple assessment of non-compliance, they count
the same.
Step 4 - Percent Non-compliance in Space. Computing a percentage should also
be a simple step. The estimate is simply 100 times the number of cells out of compli-
ance divided by the total number of cells. As a rule, the uncertainty of a binary
process can be modeled using a binomial distribution. However, the issue of uncer-
-------
A-15
tainty described for step 3 propagates into computing the percent of compliance for
a segment. Add to that the fact that estimated values for interpolator cells have a
complex dependence structure which rules out a simple binomial model and the rules
governing the uncertainty of this step are also complex. The number of interpolator
cells, N, is relatively constant and under an independent binomial model the variance
of the proportion of cells not in compliance, p, would be (p)(l-p)/N. Intuitively, one
expects the variance of p to decrease as the number of data points that feeds the inter-
polation increases. This expectation has been confirmed by simulation, but the
mathematical tools for modeling this propagation of error are yet to be developed.
Step 5 - Percent of Time. While the percent of space coordinate of the CFD has
simple interpretation of the percent of the segment out of compliance on a given
date, the percent of time coordinate is not simply the percent of time out of com-
pliance at a given point. Instead the percent of time coordinate has an interpretation
similar to that of a cumulative distribution function. The percent of time coordinate
is the percent of time that the associated spatial percent of noncompliance is
exceeded. For example, if the (percent space, percent time) coordinates for a point
on the CFD are (90,10), one would say that the spatial percent of noncompliance is
greater than or equal to 90% about 10% of the time.
This step is very similar to computing an empirical distribution function which is an
estimator of a cumulative distribution function. Because of this similarity, one imme-
diately thinks of statistical inference tools associated with empirical distribution
functions, such as the Kolmogorov-Smirnov, Shapiro-Wilk, Anderson-Darling, or
Cramer-von Mises, as candidates for inference about the CFD. These procedures
model uncertainty as a function of sample size only; in this case the number of
sample dates. The fact that it does not incorporate the uncertainty discussed the
previous steps seems unsatisfactory.
A quick review of probability plotting will reveal several methods on estimating the
percent of time coordinate in step 5. Formulae found in the literature include: (R/N),
(R - 0.5) / (N - 1). and (R - 0.375) / (N + 0.5), where R is rank and N is sample size.
These generally fall in to a family of given by (R - A)/(N - 2A + 1) for various values
of A. They are approximately equal, but the choice should be fixed for a rule.
Step 6 - Plotting the CFD. Even the plotting of the points is subject to variation,
although these variations are somewhat minor compared to the larger issue of
assessing the uncertainty of the assessment curve. The simple approach used in the
figures above is to connect the points by line segments. In the statistical literature, it
is more common to use a step function. If the graph represents an empirical distri-
bution function, each horizontal line segment is closed on the left and open on the
right. Because the CFD is an inversion of an EDF it would be appropriate for these
line segments to be closed on the right and open on the left.
Step 7 - Comparing the Curves. It is at the point of comparing the assessment
curve to the reference curve that the issue of uncertainty becomes most important.
From the preceding discussion it is clear that uncertainty in the assessment curve is
an accumulation of uncertainty generated in and propagated through the preceding 6
steps. If the reference curve is biologically based, it is derived under the same system
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-16
of error propagation. Developing the statistical algorithms to quantify this uncer-
tainty is challenging.
Even if the uncertainty can be properly quantified, the issue of who gets the benefit
of doubt due to this uncertainty is a difficult question to resolve. This is a broad
sweeping issue regarding uncertainty in the regulatory process, not a problem
specific to the CFD approach. None-the-less, it must be dealt with here as well as
elsewhere. One option is to require that the assessment curve be significantly above
the reference curve to establish noncompliance. This option protects the regulated
party from being deemed out of compliance due to random effects, but if assessment
CFD curves are not accurately determined, it could lead to poor protection of envi-
ronmental health and designated uses. A second option is to require that the
assessment curve be significantly below the reference curve to establish compliance.
This results in strong protection of the environmental resource, but could lead to the
regulated party implementing expensive management actions that are not necessary.
Some compromise between these extremes is needed. The simplest compromise is
to ignore variability and just compare the assessment curve to the reference curve.
As long as unbiased estimation is implemented for both the assessment curve and the
reference curve, this third option will result in roughly equal numbers of false posi-
tive (declaring noncompliance when in fact compliance exists) and false negative
(declaring compliance when in fact noncompliance exists) results. This offers a
balanced approach, but there is no mechanism to motivate a reduction of these false
positive and false negative errors.
,/.
r
The Chesapeake Bay monitoring program routinely monitors 19 directly measured
water quality paramenters at 49 stations in the mainstem Bay and 96 stations in the
tidal tributaries. The Water Quality Monitoring Program began in June 1984 with
stations sampled once each month during the colder late fall and winter months and
twice each month in the warmer months. A refinement in 1995 reduced the number
of mainstem monitoring cruises to 14 per year. "Special" cruises may be added to
record unique weather events. The collecting organizations coordinate the sampling
times of their respective stations, so that data for each sampling event, or "cruise",
represents a synoptic picture of the Bay at that point in time. At each station, a hydro-
graphic profile is made (including water temperature, salinity, and dissolved oxygen)
at approximately 1 to 2 meter intervals. Water samples for chemical analysis (e.g.,
nutrients and chlorophyll) are collected at the surface and bottom, and at two addi-
tional depths depending on the existence and location of a pycnocline (region(s) of
density discontinuity in the water column). Correlative data on sea state and climate
are also collected.
In addition, Chesapeake Bay Program partner organizations Maryland Department
of Natural Resources and the Virginia Institute of Marine Science have recently
begun monitoring using a technology known as data flow. DATAFLOW is a system
of shipboard water quality probes that measure spatial position, water depth, water
appendix a « The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-17
Baltimore wru
Patapsco R. «*-'•
in
Rappahannock R.
TFX3
Figure 2.3. Map of the tidal water quality monitoring stations.
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-18
temperature, salinity, dissolved oxygen, turbidity (clarity of the water), and chloro-
phyll (indicator of plankton concentrations) from a flow-through stream of water
collected near the water body's surface. This system allows data to be collected
rapidly (approximately every 4 seconds) and while the boat is traveling at speeds up
to 20 knots.
In 2005, the MDDNR Water Quality Mapping Program covered 16 Chesapeake Bay,
Coastal Bay and Tributary systems. The St. Mary's, Patuxent, West, Rhode, South,
Middle, Bush, Gunpowder, Chester, Eastern Bay, Miles/Wye, Little Choptank,
Chicamacomico and Transquaking Rivers will be mapped, as well as Fishing Bay
and the Maryland Coastal Bays. In Virginia, dataflow data are available for the
Piankatank, York, Pamunkey and Mataponi Rivers.
Beginning in 1990, Chlorophyll-a concentrations were measured over the mainstem
Chesapeake using aircraft remote sensing. From 1990-1995, the instrument used for
this study was the Ocean Data Acquisition System (ODAS) which had three
radiometers measuring water leaving radiance at 460, 490 and 520 nm. In 1996, an
additional instrument was added, the SeaWiFS Aircraft Simulator (SAS II). SAS II
has sensors at seen wavebands which improves detection of Chlorophyll in highly
turbid areas. Since 1990, 25-30 flights per year have been made during the most
productive times of year.
The data described above and additional information can be obtained from:
www.chesapekebay.netmddnr.chesapeakebay.net/eyesonthebay/index.cfm
www2.vims.edu/vecos/
of
The current Chesapeake Bay Interpolator is a cell-based interpolator. Water quality
predictions for each cell location are computed by averaging the nearest "n" neigh-
boring water quality measurements, where "n" is normally 4, but this number is
adjustable. Each neighbor included in the average is weighted by the inverse of the
square of Euclidean distance to the prediction cell (IDW). Cell size in the Chesa-
peake Bay was chosen to be 1km (east- west) x 1km (north-south) x 1m (vertical),
with columns of cells extending from surface to the bottom of the water column, thus
representing the 3-dimensional volume as a group of equal sized cells extending
throughout the volume. The tributaries are represented by various sized cells
depending on the geometry of the tributary, since the narrow upstream portions of
the rivers require smaller cells to accurately model the river's dimensions. This
configuration results in a total of 51,839 cells by depth for the mainstem Chesapeake
Bay (segments CB1TF-CB8PH), and a total of 238,669 cells by depth for all 77
segments which comprise the mainstem Bay and tidal tributaries.
The Chesapeake Bay Interpolator is unique in the way it computes values in 3
dimensions. The interpolator code is optimized to compute concentration values,
which closely reflect the physics of stratified water bodies, such as Chesapeake Bay.
The Bay is very shallow compared to its width or length; hence water quality varies
much more vertically than horizontally. The Chesapeake Bay Interpolator uses a
vertical filter to select the vertical range of data that are used in each calculation. For
appendix a * The Cumulative hrequency Diagram Method for Determining Water Quality Attainment
-------
A-19
instance, to compute a model cell value at 5m deep, monitoring data at 5m deep are
preferred. If fewer than n (typically 4) monitoring data values are found at the
preferred depth, the depth window is widened to search up to d (normally +/-2m)
meters above and below the preferred depth, with the window being widened in 0.5m
increments until n monitoring values have been found for the computation. The
smallest acceptable n value is selectable by the user. If fewer than n values are
located, a missing value (normally a -9) is calculated for that cell. A second search
radius filter is implemented to limit the horizontal distance of monitoring data from
the cell being computed. Data points outside the radius selected by the user
(normally 25,000m) are excluded from calculation. This filter is included so that
only data that are near the location being interpolated are used.
In this version of the Interpolator, Segment and Region filters have been added.
Segments are geographic limits for the interpolator model. For instance, the Main
Bay is composed of 8 segments (CB1TF, CB2OH, ...,CB8PH). The tributaries are
composed of 77 additional segments, using the CBP 2003 segmentation. These
segments divide the Bay into geographic areas that have somewhat homogeneous
environmental conditions. This segmentation also provides a means for reporting
results on a segment basis, which can show more localized changes compared to the
whole Bay ecosystem.
Segment and bathymetry information use by the interpolator is stored in auxiliary
files. Segment information allows the interpolator to report results on a segment
basis which can show more localized changes compared to the whole Bay
ecosystem. These segment and bathymetry files have been created for the main bay
and all of the larger tributaries. The CBP segmentation scheme was replicated in
these files by partitioning and coding the interpolator cells that fall within each
segment.
The interpolator also identifies the geographic boundary that limits which moni-
toring station data are included in interpolation for a given segment through a region
file. Use of data regions ensures that the interpolator does not "reach across land" to
obtain data from an adjacent river which would give erroneous results. By using data
regions, each segment of cells can be computed from their individual subset of moni-
toring data. Each adjacent data region should overlap by some amount so that there
is a continuous gradient, and not a seam, across segment boundaries.
of
The Chesapeake Bay Program has initiated implementation of the CFD as an assess-
ment tool. The Criteria Assessment Protocols (CAP) workgroup was formed in the
fall of 2005 to develop detailed procedures for implementing criteria assessment.
This workgroup has developed and implemented procedures that use the CFD
process and conducted a CFD evaluation of dissolved oxygen for many designated
assessment units.
The CFD methodology was first applied in the Chesapeake Bay for the most recent
listing cycle which was completed in the Spring of 2006 and was based on data
collected over the period 2002 through 2004. CFDs were developed and utilized
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-20
primarily for the dissolved oxygen (DO) open- and deep-water monthly mean
criteria because there were insufficient data collected to assess the higher-frequency
DO criteria components. The clarity criteria were not assessed based on the CFD
because there were few systems in which there was sufficient data for an assessment.
Chlorophyll criteria were not available from the Chlorophyll criteria team in time to
implement a chlorophyll assessment.
In general, the CFD analysis indicated that most of the Bay waters failed one or more
of the open-water or deep-water DO criteria components. However, there were also
many tributaries in which all of the DO criteria assessed indicated attainment. Thus
in this initial application, the CFD method did appear to distinguish between
impaired and unimpaired systems in a manner that is consistent with the expectations
of the many stakeholders in the CAP workgroup.
In the 2006 application of the assessment methodology, there were many details that
required resolution in order to fully implement the methodology. Procedures gener-
ally followed the theoretical description as described in Section 2.1, but some details
were modified to address unforeseen complications. The following describes some
of those details.
In general, data were obtained from the CBP CIMS data base and parameters
included date, location, depth, salinity, temperature and the water quality parameter
being assessed. Some State data were also incorporated and those data were obtained
directly from the relevant State. Once all the data were compiled, they were assigned
to a time period based on the sample date. Fixed-station data are normally collected
during a monitoring cruise that covers the entire tidal Chesapeake Bay over several
days. However, in order to provide a "snapshot" in water quality, the data collected
within a cruise are assumed to be contemporaneous in order to perform a single
spatial interpolation. For any data not associated with a cruise, a cruise number is
assigned representing the closest cruise in time to the collection of each datum. Co-
located data points in the same cruise were averaged.
The assessment procedure requires assessment over large areas rather than at points
in space. Spatial interpolation using the CBP IDW interpolator was performed for
each water-quality criteria parameter for each cruise. Clarity and surface chlorophyll
were interpolated in the two horizontal dimensions using inverse distance squared
weighting. Dissolved oxygen was first linearly interpolated in the vertical dimension
within each column of data beginning at 0.5 meters and continuing at one meter
intervals, not to exceed the deepest observation in that column. Each depth was then
interpolated horizontally using inverse distance squared weighting. Data regions
were specified for each segment in order to prevent the interpolation algorithm from
using data points in neighboring tributaries.
Designated uses in the Chesapeake Bay are defined vertically in order separate stable
water layers that have differing criteria levels for dissolved oxygen. The surface layer
(open water) is that layer defined to be above the pycnocline and thus exposed to the
atmosphere. The middle layer (deep water) is defined to be the layer between the
upper and lower pycnocline. And the lower layer (deep channel) is defined to be the
layer below the pycnocline. Given that the pycnocline is dynamic and moves up and
-------
A-21
down with each monitoring cruise, the designated use of each grid cell must also be
defined based on the available data for each cruise.
The pycnocline is defined by the water density gradient over depth. Temperature and
salinity are used to calculate density, which in turn is used to calculate pycnocline
boundaries. Density is calculated using the method described in: Algorithms for
Computation of Fundamental Properties of Seawater (Endorsed by
UNESCO/SCOR/ ICES/IAPSO Joint Panel on Oceanographic Tables and Standards
and SCOR Working Group 51. Fofonoff, N P; Millard, R C Jr. UNESCO technical
papers in marine science. Paris , no. 44, pp. 53. 1983). For each column of temper-
ature and salinity data, the existence of the upper and lower pycnocline boundary is
determined by looking for the shallowest robust vertical change in density of 0.1
kg/m3/m for the upper boundary and deepest change of 0.2 kg/m3/m for the lower
boundary. To be considered robust, the density gradient must not reverse direction at
the next measurement and must be accompanied by a change in salinity, not just
temperature.
The depths to the upper pycnocline boundary, where detected, and the fraction of the
water column below the lower boundary are interpolated in two dimensions. If no
lower boundary was detected the fraction was considered to be zero. The depth to the
upper pycnocline boundary tends to be stable across horizontal space and so spatial
definition of that boundary using interpolation generally worked well. However,
interpolation of the lower boundary is more complicated because the results can
conflict with the upper boundary definition or with the actual bathymetry of the Bay.
As a result, interpolation of the lower boundary was performed based on "fraction of
water column depth". In that way, the constraints of the upper pycnocline boundary
definition and the actual depth were imposed and errors related to boundary conflicts
were eliminated.
Assessments were performed based on criteria specific averaging periods. The
instantaneous assessment for deep channel dissolved oxygen was evaluated using the
individual cruise interpolations. All monthly assessments were based on monthly
averages of interpolated data sets. To calculate the monthly averages, each interpo-
lated cruise within a month was averaged on a point-by-point basis. Generally, there
were 2 cruises per month in the warmer months and 1 cruise per month in the cooler
months. Spatial violation rates are calculated for each temporally aggregated inter-
polation in an assessment period. For example, for a three-year summer open-water
dissolved oxygen assessment, the twelve monthly average interpolations repre-
senting the four summer months over three years were used.
The CFD approach uses the proportion of space in attainment in any given month
estimated using an approach based on a statistical model. The current method uses
data collected in a specific month at a set of sampling locations within the segment
of interest to estimate the parameters of the model. The estimated model is then used
to interpolate likely values at unsampled locations, specifically at a set of prediction
locations arranged in a grid over the segment. The predictions thus obtained are used
appendix a « The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-22
to calculate the proportion of space in compliance that month. The current estima-
tion procedure for obtaining predicted values is Inverse Distance Weighting (IDW),
a non-statistical spatial interpolator that uses the observed data to calculate a
weighted average as a predicted value for each location on the prediction grid. The
method calculates the weight associated with a given observation as the inverse of
the square of the distance between the prediction location and the observation.
The panel considered several interpolation methods in addition to IDW. Of these,
kriging methods emerged as a principal alternative approach for populating the grid
of prediction locations. Non-parametric methods were also considered. These
include Loess regression or cubic spline methods. These approaches could be advan-
tageous in that they are statistical methods that provide levels of error, but panel
analyses and deliberations have been insufficient to provide definitive statements on
this class of methods. Table 3.2 which appears in Section 3.3 summarizes our deter-
minations.
'
Kriging is a spatial interpolation technique that arose out of the field of geostatistics,
a subfield of statistics that deals with the analysis of spatial data. Kriging and the
field of geostatistics has been employed in a wide variety of environmental applica-
tions and is generally accepted as a method for performing statistically optimal
spatial interpolations (Cressie 1991, Schabenberger and Gotway 2004, Diggle and
Ribeiro 2006). Applications of kriging in water related research can be found in
(Kitanidis 1997, Wang and Liu 2005,Ouyang et al. 2006). References on kriging
methodology, geostatistics, and their related statistical development can be found in
(Cressie 1991, Diggle et al. 1998, Schabenberger and Gotway 2004, Diggle and
Ribeiro 2006).
Kriging can equivalently be formulated in terms of a general linear regression model
Y (s) = &, + fa Xj(s) • • • + /3p Xp(s) + e(s) (1)
with s representing a generic spatial location vector (usually 2-D) assumed to vary
continuously over some domain of interest, Y(s) the outcome of interest measured at
s, Xj(s), . . . ,Xp(s) potential covariates indexed by location s, and their associated
regression effects /?1; . . . , /?p. Note that covariates must be known at every predic-
tion location. The elements of the spatial vector s can be used as covariates for
modeling spatial trends. On the other hand water quality measures such as salinity
which may have a strong association with the outcome of interest, is of limited value
as a covariate because it is not known at all prediction locations. The uncertainty in
this regression relationship is modeled with the random error term e(s) assumed to
have zero mean and constant variance. Spatial data like the type sampled in the
Chesapeake Bay water-quality criteria assessments often exhibit a property known
as (positive) spatial dependence, observations closer together are more similar than
those further away. This property is accounted for in model (1) by allowing e(s) to
have a spatial correlation structure.
Some further specifics on e(s) are warranted. Common distributional assumptions on
include normality or log-normality, although kriging can be performed based on
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-23
other statistical distributions and data transformations (Christenson et al. 2001). The
spatial correlation in e(s) is represented by positive definite functions. These func-
tions can be assumed isotropic where correlation decay depends just on distance, or
anisotropic where correlation decay depends on distance and direction. Variograms
are another special type of mathematical function closely related to spatial correla-
tion functions that can and are more often used to represent spatial correlation. For
purposes here and in many kriging applications, variograms and spatial correlation
functions provide equivalent representations of spatial structure. For consistency in
what follows only the term variogram will be used in discussions of spatial structure.
While there is considerable flexibility in implementing the error structure of a
kriging model, it is possible to generalize somewhat with respect to the error struc-
ture of Chesapeake Bay water quality data. Of the three water quality parameters
being assessed, chlorophyll and clarity measures tend to follow the log-normal
distribution and dissolved oxygen is reasonably approximated by the normal distri-
bution. The horizontal decay rate of spatial correlation does not tend to be
directionally dependent. Thus if the bay is viewed as a composite of horizontal
layers, isotropic variograms are appropriate for kriging each layer. In a vertical direc-
tion, water quality can change rapidly and thus spatial correlation can decay over a
short distance. A 3-D interpolation procedure would benefit from use of an
anisotropic variogram in order to differentiate the vertical correlation decay from the
horizontal correlation decay.
Note, in the literature model (1) is referred to as a universal kriging model. When
covariates (the X's) are not considered to influence interpolation of Y the right hand
side of model (1) contains just the constant term /?0 and e(s). The resulting model is
referred to as the ordinary kriging model. When the spatial structure (variogram) for
model (1) is known, statistically optimal predictions for the variable Y at unsampled
locations (outside of estimation of possible regression effects) can be derived using
standard statistical principles. The optimality criteria results in spatial predictions
that are linear in the data, statistically unbiased, and minimize mean squared predic-
tion error, hence referred to as best linear unbiased predictions (BLUPs). The
minimized mean squared prediction error is also taken as a measure of prediction
uncertainty. In practice, however, spatial structure of the data is unknown, the esti-
mation of which via the variogram function is cornerstone to kriging applications.
To demonstrate let {Xsi)> • • • > Xsn)l represent a set of spatial data, for example a
water-quality parameter such as dissolved oxygen sampled at a set of n spatial loca-
tions s1; . . . , sn. Assume this data to be a realization of the ordinary kriging version
of model (1). The first step in kriging is variogram estimation. There are several
methods available, method of moments and statistical likelihood based being two of
the more common, all of which though are based on the sample data {Xsi)> • • • .
Xsn)l- Without going into detail, this process ends with a chosen variogram function
and its parameter estimation, describing the shape and strength (rate of decay) of
spatial correlation. There is also a determination, again based on the sampled data,
of whether the spatial structure is isotropic or anisotropic. The estimated variogram
is then assumed known and kriged interpolations and their interpolated uncertainty
are computationally straight forward to generate at numerous locations where data
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-24
were not observed. Accounting for uncertainty in variogram parameter estimation
has commonly been explored using Bayesian methods (Diggle and Ribeiro 2006).
'; :•
The inverse distance weighting method that is currently used in the CFD approach
has already been described. Hence, this section provides a short review of IDW's
technical details and a comparison of IDW to alternative interpolation methods.
The IDW method is essentially a deterministic, non-statistical approach to interpo-
lating a two or three dimensional space. As a result it lacks statistical rigor so that
estimates of the prediction errors are not calculable without additional assumptions.
Similar to kriging, IDW predicts a value () at an unobserved site, say at location %
using a weighted average of the N nearest observed neighbors (N specified by the
modeler):
where the weights, wt, are inversely related to the distance between locations s0 and
d(s0,Si) is the Euclidean distance between locations s0 and Sj, and the denominator of
the weight is to ensure that the weights sum to 1 . The IDW is an exact interpolator
in that the predicted values for observed locations are the observed values and the
maximum and minimum values of the interpolated surface can occur only at
observed sites.
Recent research has compared IDW to other interpolation techniques, most notably
variations in kriging (Table 3.1). The authors found that in some cases kriging was
at least as good an interpolator as IDW and in some instances better. The non-para-
metric techniques (splines and similar methods) were not as precise as kriging and
IDW. The method used for comparison in virtually all of the research was some
variant of cross-validation, a method where some data are kept aside and not used in
the model estimation phase and then using the resulting model to predict values for
the data kept aside. The predicted and observed values are then compared and a
statistic is calculated that summarizes the differences between the two sets of values
(observed and predicted).
None of these studies used datasets with highly irregular edges such as are found in
the Chesapeake Bay nor did they use any distance metric other than Euclidean
distance. Whether one method is preferable to another in these more difficult situa-
tions remains unexplored.
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-25
3.1. A short list of recent articles comparing the precision of IDW to a subset of
other possible interpolation methods.
Authors
Kravchenko (2003)
Dille, et al. (2002)
Valley, et al. (2005)
Lloyd (2005)
Reinstorf, et al.
(2005)
Zimmerman, et al.
(1999)
Methods Compared
Inverse Distance
Weighting (IDW),
Ordinary Kriging
(OK)
IDW, OK, Minimum
Surface Curvature
(MC), Multiquadric
Radial Basis Function
(MUL)
IDW, OK, Non-
parametric Detrend +
Splines
moving window
Regression (MWR),
IDW, OK, simple
kriging with locally
varying mean (SKlm),
kriging with external
drift (KED)
IDW, OK, KED +
deterministic
chemical transport
models
2 types of IDW, UK,
OK
Variables
Manipulated
spatial structure and
sample grid spacing
neighborhood size,
spatial structure,
power coefficient in
IDW, sample grid
spacing, quadrat size
spatial structure,
sample size, quadrat
size
spatial structure,
sample size
single dataset was
analyzed
spatial structure,
sampling pattern,
population variance
Conclusions
IDW better than OK
unless sample sizes
were fairly large
No interpolator
appears to be more
precise than another.
Sample grid spacing
and quadrat size were
deemed more
important.
OK tended to be more
precise but IDW was
very similar
KED and OK best
OK best
UK and OK better
than IDW
One final and important issue with IDW is that, as currently used, IDW is a deter-
ministic method which makes no assumptions as to the probability distribution of the
data being interpolated. Hence, it does not allow for estimating prediction errors, i.e.
it does not allow for the possibility of random variation at interpolation sites. A
simple question is whether IDW can be recast in the kriging framework given the
similarity in prediction method (weighted average) and hence can a method be found
to estimate prediction errors? The short answer is no - the distance function used by
IDW, which is an implicit assumption about the autocorrelation function in the
spatial field, does not meet the assumptions required for development of a valid vari-
ance-covariance matrix describing the spatial covariance. As a result, IDW cannot be
modified to take advantage of the statistical knowledge that has been developed for
geostatistical analyses such as kriging. This does not imply that other approaches to
estimating prediction error are also not possible.
A non-parametric approach for estimating variance was proposed (Tomczak, 1998)
in which jack-knifing was used to provide error estimates. 95% confidence intervals
for the mean were calculated and then compared to the actual observed values. Not
surprisingly, only 65% of the data were captured within their associated confidence
interval. The method appears to have been misapplied—the jackknifing method as
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-26
used estimates the standard error of the mean assuming independent observations.
As a result, the confidence interval is not capturing the effect of the spatial depend-
encies nor is it based on the fact that we are predicting a value for the unobserved
site rather than estimating a mean. The development described by Tomczak (1998)
should be explored further and other alternatives such as block bootstrapping for
variance estimation as well.
33 ..in,.-ic
There are many variations on spatial interpolation in addition to kriging and IDW.
See Cressie (1989) for a review. The committee did not have sufficient time to
compare all models, but CBP in encouraged to continue this research. One promising
category of models are for interpolation based on non-parametric methods that do
not rely on measuring and accounting for spatial autocorrelation. All of the non-para-
metric approaches would be based on the assumption that the autocorrelation
observed in the data is due to unobserved explanatory variables and hence alterna-
tive modeling approaches are not unreasonable. The particular set we mention are
the regression type analyses with the locational indices (northings, eastings) used as
explanatory variables. Examples include generalized additive models (Hastie and
Tibshirani, 1990), high-order polynomials (Kutner, Nachtsheim, Neter, and Li,
2004), splines (Wahba, 1990), and locally weighted regression ("loess" or "lowess",
Cleveland and Devlin, 1988). In some kriging and IDW methods, large-scale trend
is modeled relatively smoothly using locational indices and local smaller-scale vari-
ation is modeled using the estimated autocorrelation in conjunction with the values
of the variable at nearby observed sites. The nonparametric methods replace estima-
tion of the local variation based on correlation functions with models of the
large-scale trend that are less smooth and more responsive to the spatial variation in
the observed data. A visual demonstration is given in Figure 3.1 which shows a one-
dimensional dataset with Y as the variable to be predicted and X as the location along
the one dimensional axis. For example, X could be distance from the mouth of a river
and Y could be chlorophyll a concentration.
One advantage of these approaches is that each of the methods has extensive statis-
tical research into estimation of model parameters as well as standard errors for those
parameters and for predictions at interpolation sites. Another is that the main
modeling decisions are related to bandwidth selection or degree order of polynomial
to fit. These decisions can be automated by developing rules for roughness of fit
based on reduction in MSE as compared to modeling a straight line (in X). Disad-
vantages are the same as for kriging, all model estimation is data dependent which
means that the spatial configuration and number of sampling sites has a direct influ-
ence on the predictions and their error estimates. In addition, a study done by Laslett
(1994) comparing kriging and splines indicated that the two methods are similar in
predictive power but for certain sampling regimes kriging performs better. We
recommend more study since the non-parametric approaches would be easier to
implement than kriging.
appendix a * The Cumulative frequency Diagram Method for Determining Water Quality Attainment
-------
A-27
figure 3.1, Bivariate fit of Y By X. Straight line is a linear large-scale trend fit (R2 = 0.19);
the moderately wavy line around the straight line is a 6th-order polynomial fit (X enters
the model as X, X2, X3, ..., and Xs; R2 = 0.25); and the jagged line is a spline fit with a
very small bandwidth (neighborhood used in local estimation at each X; R2 = 0.90).
3,3 OF
The following describes some of the benefits and potential limitations of kriging in
regards to CBP application with some comparisons to the IDW approach towards
spatial interpolation outlined in the previous section. Nonparametric methods are not
sufficiently developed to include in this comparison. A primary benefit of the kriging
methodology compared to IDW is that it is a statistical technique. As such the field
of statistics (including kriging) is designed to make inference from sampled data in
the presence of uncertainty and the quantity and quality of the sample data are
reflected in those inferences. However, kriging is a less than routine type of statis-
tical analysis and requires a certain level of statistical expertise to carry out the
process. The short description on variogram estimation provided above merely intro-
duces this involved and often complicated step. This requirement for informed
decision making limits the degree to which kriging can be automated and still main-
tain its flexibility and optimal properties.
Further issues regarding kriging and CBP applications are listed below.
• Kriging is flexible in that it is based on an estimate of the strength of spatial
dependence in the data (variogram). Kriging can consider direction dependent
weighted interpolations (anisotropy) and can include covariates (universal
kriging) to potentially influence interpolations, either simple trends in easting
and northing coordinates or water related measures such as sea surface temper-
ature measured by satellite.
• A key feature of a statistical technique like kriging is that a measure of uncer-
tainty (called the kriged prediction variance) is generated along with kriged
interpolations. Research has been initiated (i.e., conditional simulation) to
appendix a
The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-28
propagate this interpolation uncertainty through the CFD process for gener-
ating confidence intervals for estimates of attainment.
• Kriging can be applied in situations where the data are sparse, as in CBP fixed
station data, or densely sampled, as in CBP shallow water monitoring. Kriged
and IDW spatial interpolations may very well produce near identical results for
these two extreme scenarios. However it is the kriging approach that provides
a statistical model, the uncertainty of which is influenced by the quantity and
quality of data. Knowledge of interpolation uncertainty is crucial for discrimi-
nating the improved water quality assessment obtained from densely sampled
networks relative to sparsely sampled networks.
As alluded to earlier kriging is an advanced statistical technique and like all such
techniques should be carried out by well trained statistician(s) with experience in
spatial or geostatistical methodology and experience analyzing water quality data.
Assessing model fits (of the variogram and regression model) and kriging accuracy
via cross validation and/or likelihood based criteria should be employed routinely.
To further exemplify this point consider kriging the densely sampled shallow water
monitoring data which is generated by the DATAFLOW sampling. In addition to the
other technical complexities mentioned within, this spatial sampling design may
raise other issues not immediately recognized by untrained users (Deutsch 1984).
For kriging in CBP applications one potential methodological drawback is the issue
of non-Euclidean distance (Curriero 2006). Current kriging methodology only
allows the use of the straight line Euclidean distance as the measure of proximity.
However, the irregular waterways in the Chesapeake Bay system may very well
suggest other non- standard measures of distance. For example, the spatial design of
the fixed station data including those in the Bay mainstem and tidal tributaries. The
straight line Euclidean distance may very well intersect land particularly in regions
containing convoluted shorelines. There has been research initiated on this topic
(Curriero 2006, Jensen et al. 2006, Ver Hoef et al. 2007), however, results are not yet
ready for universal use.
Three dimensional interpolations (including depth as the third dimension) are poten-
tially required for CBP applications. The IDW and kriging methodologies,
mathematically speaking, certainly extend to three dimensions. However the rapid
change of water quality over depth would lead to significant anisotropies in the
application three dimensional kriging that would complicate this approach far more
than the application of IDW. On the other hand, a simplistic implementation of IDW
that does not recognize the rapid decay of covariance over depth would inappropri-
ately reach across the pycnocline when choosing nearest neighbors. Clearly the
special properties of water quality in a highly stratified bay require innovation for 3-
dimensional interpolations. Another approach would be to apply universal kriging
where a third dimension (depth) is used as a covariate. The use of depth as an inde-
pendent variable is motivated by the observation that often water quality exhibits a
predictable trend over depth as for example the trend of DO decreasing with
increasing depth. To include depth as a covariate, model (1) would be written as
-------
A-29
A third approach to interpolation in three dimensions is to implement 2-D interpola-
tion in layers. Note that the IDW interpolator currently implemented by CBP
(Section 2.2) employs a layered strategy by severely restricting (+/- 2m) the vertical
distance that may be searched for nearest neighbors. A similar strategy could be
implemented using 2-D kriging to interpolate the layers. Which of these approaches
is best suited to 3-D interpolation for the bay will depend on the data available and
assumptions related to vertical structure. Full 3-D kriging interpolation treats the 3rd
dimension as a spatial dimension in the error term y (s). The covariate approach
requires that the change over depth be a predictable trend. Interpolation in layers
assumes that covariance decays so rapidly over depth that it is adequate to treat the
layers as independent entities. Data sufficiency requirements increase for all
approaches when considering three dimensional interpolations. When data are
sparse, again a statistical based approach like kriging allows this to be reflected in
prediction uncertainty.
In many applications, attainment or lack of attainment will be so extreme that the
assessment end point is clear even without optimizing the error estimation of the
CFD. In these extreme cases, IDW or kriging simplified for automation could be
sufficient to support the attainment ruling without precise quantification of estima-
tion uncertainty. For these cases, the customized IDW algorithm that is currently
implemented by CBP provides a tool with which to begin testing the CFD assess-
ment procedure, but kriging simplified for automation may offer some advantages.
Kriging can be simplified for automation by fixing the variogram model to one math-
ematical form, say exponential, for all applications. With the variogram model fixed,
kriging becomes like IDW in assuming the same mathematical form for the spatial
dependence for all cases, but it is more flexible than IDW in that the rate of spatial
correlation decay could be allowed to vary among applications. In addition, the
simplified kriging opens the door for conditional simulation, with potential benefits
that are discussed in Section 5. While a simplified kriging algorithm offers some
advantages, there are also some potential drawbacks. Because variogram estimation
typically entails use of an iterative procedure such as maximum likelihood or non-
linear least squares, there is the potential that lack of convergence of these
algorithms would be problematic for an automated implementation of kriging.
In terms of computing, IDW is available in commercial GIS software, requiring GIS
skills for application. Kriging is available in commercial statistical software and also
in the free open source R Statistical Computing Environment (R Development Core
Team 2005, Ribeiro and Diggle 2001) and requires programming skills for those
software packages.
In summary, kriging is more sophisticated than IDW, but requires greater expertise
during implementation to fully exploit its full benefit. Table 3.2 provides a com-
parison of the capabilities of assessments based simply on: 1) percent of samples,
2) spatial interpolation based on IDW and 3) spatial interpolation based on kriging.
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-30
Table 3.2. Comparison of the capabilities of methods available for interpreting data
collected for Chesapeake Bay water-quality criteria assessment.
Attributes
Provides Spatial
Prediction
Provides Prediction
Uncertainty
Uncertainty for CFD
Deal with Anisotropy
Can Include Cruise
Track/Fly over
Feasibility of 3
dimensional
interpolations
Feasibility of mainstem-
tributary interpolations
Inclusion of covariates to
improve prediction
Predictions of non-linear
functions of predicted
attainment surfaces
P(y>c)
Level of Sophistication
Automation
Sample-based
Yes
No
No
No
No
No
No
No
No
Lowest
Yes
IDW
Yes
not routine
No
Possible, but
not routine
Yes
Yes
Yes
No
No
Low
Yes
Kriging
Yes
Yes
Yes
Yes
Yes
Possible, but not
routine
Possible
Yes
Yes
Very High
Possible, but not
routine
4.0
There are several approaches to defining reference curves that are proposed for use
in the CFD assessment methodology. One is a biologically based definition and other
approaches are based on an arbitrary allowable frequency (see Section 2). Here we
review these options in greater detail.
The idea behind biological reference curves is to identify regions of the Bay that
have healthy biological indicators and are thus considered to be in attainment of their
designated use. CFDs would be developed for these areas in the same way that CFDs
would be developed elsewhere, but those curves developed for healthy areas would
be considered "reference" curves. For example, healthy benthic IBI scores might be
used as indicators of adequate bottom dissolved oxygen.
The success of the CFD-based assessment will be dependent upon decision rules
related to the biological reference curves. These curves represent desired segment-
designated use water quality outcomes and reflect sources of acceptable natural
appendix a « The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-31
variability. The reference and attainment curves follow the same general approach in
derivation—water quality data collection, spatial interpolation, comparison to
biologically-based water quality criteria, and combination of space-time attainment
data through a CFD. Therefore, the biological reference curve allows for implemen-
tation of threshold uncertainty as long as the reference curve is sampled similarly to
the attainment curve. Bias and uncertainty are driven in CFD curves by sample
densities in time and space. Therefore, we advise that similar sample densities are
used in the derivation of attainment and reference curves. As this is not always
feasible, analytical methods are needed in the future to equally weight sampling
densities between attainment and reference curves.
4,2, CBP
In some cases, the development of biologically-based reference curve is not possible
due to lack of data describing the health of the relevant species. In such cases, a more
arbitrary approach is required since better information is not available. EPA recom-
mends the use of a default curve in cases where a biologically-based one is not
available. That default curve is defined by these properties:
1. symmetric about the 1:1 line,
2. hyperbolic,
3. total area = 0.1, and
4. pass through (1,0) and (0,1)
(see EPA, 2003; page 174). The equation that describes this figure is defined by the
equation:
(x+b) * (y+b) = a
Where: b = 0.0429945
a = b2 + b
This reference curve is illustrated in Figure 4.1 by curve 1.
An alternative default reference curve might be formulated by extending the arbi-
trary allowance of 10% exceedance into the two dimensional framework of the CFD.
The criterion threshold is a value that should be rarely exceeded by a population at
healthy levels. When the population is unidimensional, say concentration in a point
source effluent, then one can obtain this upper threshold based on the simple distri-
bution of values in a healthy population (Figure 4.2). The ninetieth percentile of this
distribution might be chosen as the criterion threshold. Thus in this example, 10%
noncompliance is allowed because this level of noncompliance is expected in a
healthy population. A standard technique for estimating distribution percentiles is to
assume a mathematical form for the distribution, e.g., the normal distribution, and to
estimate the percentile as some number of standard deviations above the mean. The
90th percentile of the normal distribution is 1.2815 standard deviations above the
mean.
appendix a * The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-32
o.u
0.2 0.3
0.4 0.5 0.8
Fraction of Space
0,7
0.8
0.9
1.0
4.1. Illustrations of three reference curves: 1) the standard CBP reference curve
derived to cover 10% of the percent space by percent time plane (curve 1); 2) a reference
curve based on 10% exceedance frequency and a temporal-spatial variance ratio of 1.0
(curve 2); and 3) a reference curve based on 10% exceedance frequency and a temporal-
spatial variance derived from chlorophyll data (curve 3).
When regulating populations that are distributed in both space and time, this simple
concept for regulating noncompliance must be extended to account for the variability
in each dimension. While there is some added complexity in the mathematics, the
fundamental concept remains the same: That is, to set the criterion threshold at a
certain distance above the mean so that exceedance of that threshold will be rare in
a healthy population. In this case, the distance by which the threshold must exceed
the mean is a function of both the spatial and temporal variance components as
described below.
To establish these criteria thresholds for populations with two components of vari-
ance, assume the simple model:
where:
IJL is the desired mean level of chlorophyll (in log space)
«; is a random term for variation over time with variance o2,
/?i(Sj) is a random term for variation over space with variance O2p
YJ(SJ) is a water quality constituent measured at time i and location Sj.
The variance of xy is O2a + o2^ = o2. The standard dev of xis is sqrt(cr2) = o. It is
common to allow an overall 10% exceedance rate without declaring an assessment
unit out of compliance. We would expect 10% of the xis to fall above /u + 1.2815*cr
appendix a
-------
A-33
0.6
0.5:
0.4-
I
o
a.
&
0.2:
0.1
10
15
Chlorophyll
20
25
4.2. Hypothetical lognormal distribution that might be typical of Chlorophyll.
The figure illustrates the relation of the geometric mean and the criterion threshold
set at the 90th percentile.
where 1.2815 is the 90th percentile of the standard normal distribution. Thus (assum-
ing normality) a population with spatial and temporal variance characterized by O2a
and O2a that has a mean that is 1.2815*<7 below the threshold criterion should have
an exceedance rate of 10% over space and time. Note that the reference curve is
determined by the ratio O2alo2^ and the distance in standard deviations of the mean
from the threshold. The actual values of the variance components, the mean, and the
threshold, are not important as long as the relationships hold. Thus as long as the
variance ratio is consistent, and mean to threshold distance is a fixed number of stan-
dard deviations, the same reference curve will serve for all seasons and regions.
Letting chlorophyll observed in the decade of the 1960s serve as a reference popu-
lation, the parameters in Table 4.1 can be used to construct this reference curve based
on the variance ratio and the mean to threshold distance given in the table. The ratio
O2alo2p is computed as the ratio of the temporal variance term and the spatial vari-
ance term. The mean to threshold distance is computed to be 1.2815cr for all regions
and seasons. Based on there parameters, a reference curve for chlorophyll can be
derived (curve 3 , Figure 4.1). For comparison a reference curve based on a variance
ratio of 1.0 (curve 2, Figure 4.1) and the standard CBP reference curve (curve 1,
Figure 4.1) are also shown.
appendix a
-------
A-34
Table 4.1. Chlorophyll criteria derived by computing and upper threshold based on predicted
means for mid-flow1960s chlorphyll data.
Season
Spring
Summer
Spring
Summer
Spring
Summer
Salinity
Zone
OH
OH
MH
MH
PH
PH
Mean
Log(chl)
0.7684
1.1693
0.4137
0.8626
0.1386
0.218
GMmean
(chl)
5.87
14.77
2.59
7.29
1.38
1.65
Temporal
Variance
0.0233
0.0233
0.0233
0.0233
0.0233
0.0233
Spatial
Variance
0.0658
0.0658
0.0658
0.0658
0.0658
0.0658
Std
Dev
log(chl)
0.2985
0.2985
0.2985
0.2985
0.2985
0.2985
Threshold
Criterion
log(chl)
1 .2594
1 .6603
0.9047
1.3536
0.6296
0.709
Threshold
Criterion
(chl)
18.17
45.74
8.03
22.58
4.26
5.12
Relative to the standard reference curves, the curve based on the observed variance
ratio for chlorophyll is more restrictive of events where large portions of the popu-
lation are out of compliance. For example, the CBP standard reference (curve 1)
would allow 40% of area to exceed the criterion threshold up to about 6% of the
time. The proposed chlorophyll reference curve (curve 3) would restrict occurrences
of 40% of area out of compliance to about 2% of the time. Conversely, the proposed
curve (curve 3) allows a higher frequency of events where a small percentage of
space in out of compliance. For example, 10% of space is allowed out of compliance
36% of the time under the proposed curve and 27% of the time under the standard
curve.
While there is mathematical and statistical logic underpinning this proposed chloro-
phyll reference curve, it is important to remember that it is based on parametric
models and simplifying assumptions. It is recommended that validation exercises be
performed to insure that the general shape of CFD curves generated from data
collected in near reference conditions is approximated by the proposed curve.
4.3 ACCOMMODATING SEASONALITY IN REFERENCE CURVES
The degree of acceptable exceedance can vary with season. For example, benthos are
less tolerant of hypoxia in warmer water temperatures. In addition, the threshold
criterion may never be exceeded in some seasons and frequently be exceeded in
others. By combining seasons, the acuteness of a specific seasonal exceedence is
diluted by data from the acceptable season(s). To some extent, seasonal differences
can be accommodated by changing the threshold criterion among seasons. However,
there may still be a need to develop separate reference curves by season.
5.0 REVIEW CFD STATISTICAL PROPERTIES
INCLUDING BIAS, PRECISION, AND INFERENCE
The CFD as an assessment tool is a relatively new and unstudied concept. Its close
relationship to the empirical distribution function does give some insight on the
mathematical behavior of the CFD. In this section we review some of the properties
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-35
of the CFD and discuss the complications that arise from these properties when the
CFD is used as an assessment tool. After defining the population which determines
the CFD, we go on to discuss the currently proposed sampling and estimation
scheme, sources of error in the estimation scheme, and problems that result from
these. The goal is to succinctly define these problems and elucidate possible solu-
tions. This section will cover: the behavior of the CFD as a function of temporal and
spatial variance, methods for construction CFD reference curves, the influence of
sampling and estimation variance on the CFD shape, and feasible methods for devel-
oping statistical inference tools.
5 1 OF CFD
SM? m 1 l%i™Wli™SB \af 1 %«wi 1 imT I 1 %\a# 1 I™ 1 % 1 1 I™ W»
With any statistical application, it is important to distinguish between the true
descriptive model underlying the population being sampled and the estimate of this
model derived from the data collected in a sample. As described above, the CFD has
a data driven definition where the CFD is constructed based on a sample from a
population for some water quality parameter. This population is a continuous
random process over space and time.
In order to quantify the statistical properties of the CFD, the CFD is defined in terms
of a population of experimental units. This approach is a discrete approximation of
the continuous random process in both time and space. However, the estimation
scheme involves interpolation to discrete units in a spatial dimension and discrete
days in the temporal dimension. To facilitate an understanding of the relation of the
estimator to the true population, it seems reasonable to use a discrete approximation
as the model for the true population.
The population will be defined as having different sizes of experimental units in
much the way we think of a population that gives rise to a nested design or repeated
measures design. The Chesapeake Bay will be partitioned into segments. Assessment
will be done for each segment based on a three year record of the segment. Thus a
three year period for the segment defines the entire population that will be parti-
tioned into experimental units. The continuous time dimension is partitioned into
days to form the primary units which are the state of a segment for a day. Call this a
Segment-Day. Let there be M segment-days in the assessment period (typically 3 x
365). The continuous spatial dimension is partitioned into N 3-dimensional cells
(may range from hundreds to thousands). The state of each cell for a day will be a
unit nested within the segment-day. The attribute of interest will be a measure of
water quality for each cell for a day. Examples might be the mean level of Chloro-
phyll-a in the cell for one day or the minimum of dissolved oxygen in the cell during
the day. Let Y be a random variable for the attribute of interest and consider the
following model
YJ(SJ) = fi + a{ + /?i(Sj) Eqn 5.1.1.1
appendix a * The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-36
the vector a will be assumed to have expectation 0 and variance 2c
each vector ^ will be assumed to have expectation 0 and variance 2/8i.
i is the ordinal index for days and
s is a vector valued ordinal for spatial location.
Under this model, 2a defines the correlation over time at the segment-day level and
2/?i defines correlation over space that occurs cell to cell within a day.
Let C ;(Sj) be a collection of threshold limits that define the acceptable criterion for
the measured attribute. If Y ;(Sj) exceeds C ;(Sj) in a cell, that cell is called degraded.
The criterion is allowed to vary in both time and space so that in theory each Y ;(Sj)
might be compared to a unique C j(Sj).. It may vary over time because different
levels of Y may be acceptable in different seasons. It may vary over space because
different levels of Y may be acceptable in different salinity regimes so that even
within a segment, C may be a function of salinity. As a rule, it is anticipated that
C ;(Sj) will be constant for regions of space and time such as salinity zones and
seasons.
Now convert the measured attribute Y ;(Sj) to a Boolean response as follows
TY ;(Sj) = I(Y ;(Sj) > C i(Sj)) = 1 if Y ;(Sj) > C ;(Sj) Eqn 5.1.1.2
= 0 otherwise
Thus TY takes the value 1 when Y exceeds the threshold defined by C. Using TY,
we summarize the state of a segment on one day as the fraction of that segment that
is out of compliance
Eqn 5.1.1.3
The CFD that we wish to estimate is one minus the cumulative distribution function
of the PJ'S. If P(i) represents the ordered values of the P;'s for any assessment period,
then let
(l)p Eqn 5.1.1.4
G defines the CFD that if it were known would be used for an exact assessment. The
cumulative distribution function is determined by the mean and variance of the ideal
population. This population is defined with a spatial variance component and a
temporal variance component. The final CFD shows the cumulative percent of time
that a certain percent of space is below the criterion threshold. If the CFD shows that
water quality in a segment is beyond the threshold for too much space and too much
time, then the segment is classified as impaired.
For one assessment period, G can be considered exact as defined above, but recog-
nize that even this is only one observation of the many possible observations of G
that could result from sampling different assessment periods.
Assume for simplicity that Y is normal. If 2a were 0 so that Y had constant expec-
tation over time and if 2/3 were of the form a2! then each cell on each day would
have constant probability of exceeding a constant value of C given by 1 - <&(C)
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-37
where is the normal cumulative density function. In this greatly simplified
scenario, P; would be the outcome of N independent Bernoulli trials. The ideal CFD
would be the cumulative distribution function of M outcomes of a binomial random
variable with N trials. If we allow 21/j to have positive off diagonal elements, then the
Bernoulli trials become dependent (i.e. adjacent cells are more likely to either both
exceed or both meet the standard than distant cells). This should make the distribu-
tion of the Pj more variable than under the independent binomial model, but the
expectation of P; would be constant over time. If we relax the assumption that ^a
is 0, then the expectation of the P; would vary over time which would increase the
variability of the P; even more.
Under the simplifying assumptions of independence, constant mean, and constant
variance, it is possible to obtain an analytical formulation for the CFD based on the
parameters of Eqn 5.1.1.1. However, when the more realistic time dependent, space
dependent model with seasonal nonstationarity is considered, an analytical formula-
tion is not tractable. The lack of an analytical formulation for this estimator under
realistic dependence assumptions, e.g. non-trivial ^a and 2^, points toward com-
puter intensive simulation techniques to develop statistical inference procedures for
this problem. None-the-less, it is interesting to consider the behavior of the CFD
under the simplified model.
5.3 CFD A
In what follows, the behavior of the CFD under various parameter formulations for
Equation 5.1.1.1 are presented in graphical form. There are four parameters involved:
|u the population mean, ot the temporal variance, os the spatial variance, and C the
criterion threshold. In the examples that follow, three of these parameters are held
constant and the fourth is varied to illustrate the effect of the varied parameter.
In this exercise, the parameters of Equation 5.1.1.1 are simplified as follows: ^a = ot
I and ^ = os I, where I is the identity matrix. Thus in both the temporal and spatial
dimensions, independence and constant variance is assumed.
Example 1. Example 1 considers the effect of changing the population mean on the
shape of the CFD.
5.1. Parameter values and key for the family of curves shown in Figure 5.1.
V
5
4
3
2
1
CTt
1
1
1
1
1
CJs
1
1
1
1
1
C
5
5
5
5
5
color
Red
Orange
Brown
Green
Blue
curve
number
1
2
3
4
5
appendix a * The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-38
1.0;
0.9;
0.7;
0.6:
§• 0.4
0.3;
0.2;
0.1-
0.3 O.I OJ> 0.6
Proportion of Space
0.7
0.8
0.9
5.1. A family of curves illustrating the behavior of the CFD as the population mean
decreases from the criterion threshold. The parameter values for each curve and the
corresponding curve number are given in Table 5.1.
Note that when the population mean is equal to the criterion threshold, the CFD is a
diagonal line from upper left to lower right (Figure 5.1, curve 1). This is largely an
artifact of using symmetric distributions, the normal, for both the time and space
variance components. That is, when the population median is equal to the criterion
threshold, we expect an average of 50% noncompliance over time and we expect the
exceed 50% noncompliance 50% of the time.
As the overall population mean decreases from the criterion threshold, the family of
curves tends to move from the diagonal line toward the lower left corner. Thus a
reference population, which should have a small probability of exceeding the crite-
rion threshold might have a shape similar to the green curve. This illustrates the
importance of the shape of the CFD in measuring compliance. A CFD from a highly
compliant population will tend to hug to lower left corner similar to the blue and
green curves. As the population mean approaches the criterion threshold, the CFD
approaches curve 1. If the population mean were to exceed the criterion threshold,
the CFD would tend toward the upper right corner.
appendix a
-------
A-39
Example 2. Example 2 considers the effect of changing the temporal variance on the
shape of the CFD. Note that the population mean is held constant at 3 which corre-
sponds to curve 2 of the preceding example.
5,2, Parameter values and key for the family of curves shown in Figure 5.2
0
3
3
3
3
3
Ot
1
2
3
4
5
os
1
1
1
1
1
c
5
5
5
5
5
color
Red
Orange
Brown
Green
Blue
curve
number
1
2
3
4
5
1.0
0.9
O.S
0.7
a
P 0.6
"s
I 0.5J
|- 0.4'
£
0.3
0.2-
0.1
0.0i
0.0 0.1 0.2 0.3 0.4 OJi 0.6
Proportion of Spa«»
0.8 0.9
1.0
5.2, A family of curves illustrating the behavior of the CFD as the temporal popula-
tion variance increases. The parameter values for each curve and the corresponding curve
number are given in Table 5.2. Note that the curve 1 here has the same parameters as
curve 2 in Figure 5.1.
As temporal variance increases, the frequency of large proportions of space going
out of compliance increases (Figure 5.2, lower right). Conversely, the frequency of
small proportions of space out of compliance (i.e. large proportions of space being
in compliance) decreases (Figure 5.2, upper left). That is, shifting the daily mean
either down or up tends to shift the entire segment toward or away from compliance.
appendix
-------
A-40
In preparing water clarity CFDs for reference areas defined by having successful
SAV beds, it is not unusual to find a curve shape similar to Figure 5.2 orange or
yellow curves. This pattern suggests that SAV is tolerant of ephemeral events of
spatially broad degraded water clarity. If water clarity is persistently degraded over
portions of the area, SAV may be impaired.
Example 3. Example 3 considers the effect of changing the spatial variance on the
shape of the CFD. Again the population mean is held constant at 3 which corre-
sponds to the curve 2 of the first example.
Table 5.3, Parameter values and key for the family of curves shown in Figure 5.3
n
3
3
3
3
3
CTt
1
2
3
4
5
ortion of Space
0.8 0.9
1.0
5.3, A family of curves illustrating the behavior of the CFD as the spatial popula-
tion variance increases. The parameter values for each curve and the corresponding curve
number are given in Table 5.3.
appendix a * The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-41
Increasing the spatial variance results in a family of curves that is complementary to
those that follow an increase in temporal variance. Increasing spatial variance results
in a higher frequency of small proportions being out of compliance. It is not so much
an all-or-nothing phenomenon.
Example 4. Example 4 considers the effect of changing both temporal and spatial
variance on the shape of the CFD.
5,4, Parameter values and key for the family of curves shown in Figure 5.4.
u
3
3
3
3
3
CTt
1
2
o
3
4
5
CTs
1
2
o
3
4
5
c
5
5
5
5
5
color
Red
Orange
Brown
Green
Blue
curve
number
1
2
o
3
4
5
1.0
0.9
0.8
0.7
0.6
0-5
OA
0.3;
0.2:
0.1
0.0-1
0.0
0.1
0.3 0.4 0.5 0.6
Proportion of Spaa*
0.7
0.8
0.9
1.0
5,4, A family of curves illustrating the behavior of the CFD as both temporal and
spatial variance increases. The parameter values for each curve and the corresponding
curve number are given in Table 5.4.
appendix a
-------
A-42
Increasing the spatial and temporal variance together has the opposite effect of
decreasing the population mean. The CFD tends to move in a direction of noncom-
pliance. Thus compliance as measured by the CFD depends on the relative values of
the population mean, the temporal and spatial variance, and the criterion threshold.
Increasing the population mean has the same effect as decreasing the criterion
threshold. Increasing population variance has the same effect as increasing the mean
or decreasing the criterion threshold. In a sense, the CFD is measuring the distance
between the population mean and the criterion threshold in units of variance analo-
gous to a simple t-test. A nuance introduced here that has no analogy in the t-test is
that the ratio of spatial to temporal variance controls the symmetry of the curve.
I / '.»M 'M'.*«
In Section 5.1., it was shown that the shape of the CFD is a critical element to deter-
mining compliance. Thus it is important that this shape be primarily determined by
the state of compliance of a segment and not be influenced by factors not relating to
the status of compliance. Because the CFD is constructed based on data that are a
sample from the whole, it is clear that some uncertainty in the CFD will result. In
addition, the CFD is a function of the empirical distribution function (EDF) of frac-
tion of space in compliance. The shape of this EDF is determined by the mean and
variance of the sample. Thus any factor, such as sample size, that affects the preci-
sion of the fraction of space estimate, will affect the shape of the CFD. In this section
we review the effect of noncompliance factors on the shape of the CFD.
As noted, because the CFD is a function of the EDF of estimates of "fraction of
space", any factor affecting the precision of the estimate of fraction of space in
exceedance will affect the shape of the CFD. In particular, the number of samples
used for each p-hat (% exceedence) will affect precision. For a given segment, this
fraction will be estimated more accurately if twelve samples are used to form the
interpolated surface rather than six. Because of unknown spatial dependence in the
data, it is difficult to analytically quantify the magnitude of this sample size effect.
Therefore simulation analysis was employed to address this issue.
Numerous simulation tests were performed. These begin with a simulation of struc-
turally simple data that have no temporal or seasonal trend and progress to simulated
data that mimic the temporal and spatial structure of observed data. Because the
results from this latter simulation are most relevant, these are the results that are
presented and discussed.
Simulated data were created to mimic the properties of surface chlorophyll in the
Patuxent estuary. Data were created to fill a 5 by 60 cell grid which approximates
the long and thin nature of an estuary. These data have mean zero and a spatial
variance-covariance structure chosen to approximate the spatial variance-covariance
structure of cruise-track chlorophyll observed in the Patuxent estuary. Thirty-six
appendix a * The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-43
grids of data were simulated to represent 36 months in a three year assessment
period. The temporal and spatial trends were added to the simulated data by adding
in means computed for each month and river kilometer during the period Jan 1, 1991
to Dec 31,1993. Simulated data were created using the "grf' function of the Geosta-
tistical Package "geoR" of the R-package.
After the full population of data was simulated for 3 year assessment period, a
sampling experiment was conducted to assess the effect of sample size on the shape
of the CFD. First, as a benchmark, a CFD was computed using all of the simulated
data. To simulate the effect of sampling, a sample of fixed size was randomly
selected from each the 36 5x60 grids of data. Using these samples, kriging
(krige.conv function of geoR) was used to populate each monthly grid with esti-
mates. These estimated chlorophyll surfaces were used to compute an estimate of the
CFD which was graphically compared to the benchmark (Figure 5.5). For a fixed
sample size, the process was repeated until it was clear whether the differences
between the benchmark CFD and the estimated CFDs were due to variance or bias.
To assess the effect of sample size, the process was repeated for several sample sizes.
The effect of sample size on the shape of the CFD is consistent with expectations
based on the relation of the CFD to the empirical distribution function (Figure 5.5).
As sample size decreases, the variance of the estimated values of fraction of space
increases. This increase in variance results in the estimated CFD being to the left of
the true curve for low values of fraction of space and to the right of the true curve
for high values of fraction of space. This assessment has been repeated many times,
varying the threshold criterion, systematic vs. random sampling, the level of vari-
ability in the simulated data, and so on. This sample size effect persists for every case
where realistic estimation is employed.
As shown above (Figures 5.2-5.4) the shape of the CFD is a function of the ratio of
temporal and spatial variance. To the extent that the ratio of these variance compo-
nents in the data represent the true state of nature, this is acceptable. However, under
a model with strong spatial and temporal dependence, the ratio of these variance
components might be influenced by the scale of sampling in the spatial and temporal
dimensions. For example, samples collected far apart in time might reflect higher
variance than samples collected close in time. If the ratio of temporal and spatial
variance is influenced by the density of sampling in each dimension, then experi-
mental design will have an effect on the asymmetry of the CFD estimate.
An investigation into the use of conditional simulation to obtain confidence bounds
for the CFD showed that not only is this a promising technique for statistical infer-
ence, but also has potential in correcting bias associated with sample size effects that
has been identified as a central problem in implementing the CFD approach.
Correcting the bias of the CFD due to the sample size effect is important in obtaining
confidence bounds on the CFD that cover the true CFD for a segment. Because bias
correction is an important first step, this aspect of the conditional simulation exper-
appendix a - The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-44
CFD simulations
0.4 0.6
fraction of space
CFD simulations
0.4 0.6
fraction of space
CFD simulations
CFD simulations
0.0
0,2
0.4 0.6
fraction of space
0.8
1.0 00
0,2
0,4 0.6
fraction of space
true
n=80
0,8
1.0
Figure 5.5. Illustration of the effect of sample size (n) on the shape of the CFD for sample sizes 10, 20, 40, and 80.
appendix a « The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-45
iments will be discussed first. Conditional simulation will then be evaluated in its
efficacy in obtaining confidence intervals.
This section first outlines the basic concept of conditional simulation and provides
an algorithm that employs conditional simulation to estimate confidence bounds for
the CFD. The results of this experiment support the potential of conditional simula-
tion for correcting the sample size bias. A heuristic discussion of the mechanism
underlying this adjustment for sample size effect is presented with the hope of moti-
vating additional analytical investigation of this effect.
Conditional simulation (Journel, 1974; Gotway, 1994) is a geostastical term for
simulating a population conditional on information observed in a sample. In the case
of kriging, a sample from a spatial population is used to estimate the variogram and
mean for the population. The conditional simulation procedure generates a field of
simulated values conditioned on the estimated mean and variogram from the sample.
To the extent that the estimated mean and variogram approximate the true mean and
variogram and the assumed distribution is a reasonable model for the true distribu-
tion, repeated simulations of this virtual population will represent the variability
typical of the true population. It follows that statistics computed from the condition-
ally simulated fields will represent the expected variability of statistics from the true
distribution. The CFD is a graphical representation of ordered statistics of percent
compliance over time and it is a reasonable to assume that repeated conditional
simulations will lead to effective confidence bounds for the CFD.
In the computation of the CFD, conditional simulation is implemented at the inter-
polation step for each month. Interpolation produces an estimate of the spatial
surface of the target parameter. From that estimate of the surface is obtained an esti-
mate of the percent of noncompliance. Using conditional simulation, the surface can
be reconstructed 1000 times. From the 1000 simulated surfaces are computed 1000
estimates of the proportion of noncompliance. When this is repeated for each month
for say 36 months, the result is an array of 1000 sets of 36 values of the proportion
of noncompliance. Each of the 1000 sets of 36 can then be ranked from largest to
smallest to compute a CFD in the usual way which results in 1000 CFD estimates.
The variability among these 1000 CFDs can be used to estimate confidence intervals.
To evaluate this concept, the following simulation experiment was conducted
1. The first step is to simulate a population that will be considered the "true"
population for this exercise. A grid of dimensions 5x60 is populated using an
exponential spatial variance model with variogram parameters set to
(0.00625026, 2.67393446). These variogram parameters were estimated from
Patuxent cruise track chlorophyll data. This grid is populated 36 times to repre-
sent 36 months. The mean and variogram are held constant for the 36
simulations to create a simplistic case with no seasonal or spatial trend. Using
this set of data, the CFD is computed in the usual way and this is considered
the "true" CFD.
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-46
2. A sample of size 40 is selected from each of the 36 simulations at random loca-
tions on the grid. Ordinary kriging is used to estimate the spatial surface for
each simulation and from these 36 estimates of the monthly spatial surfaces, a
CFD is computed. This is called the 'estimated' CFD.
3. For each of the kriged monthly surfaces, 1000 conditional surfaces are simu-
lated based upon the mean and variogram estimated from the sample data. The
Cholesky decomposition is used to reconstruct the covariance structure indi-
cated by the estimated variogram. The conditionally simulated surfaces were
processed to yield 1000 estimates of the proportion of noncompliance. The
1000x36 noncompliance values are used to compute 1000 CFDs, which are
called the population of "conditionally simulated" CFDs.
4. Each "rank position" of the monthly ordered proportions of noncompliance
has 1000 values in this simulated population. To assess variability in the simu-
lated population, graphs of the miniumum, the 2.5th percentile, the 50th
percentile, the 97.5th percentile, and the maximum at each rank position are
plotted to illustrate a 95% confidence envelop for the CFD (Figure 5.6).
To test this procedure under various conditions, this basic simulation exercise was
repeated varying the sample size and adding temporal and spatial trend to the simu-
lation of the "true" population to reflect conditions more similar to real populations.
The results of this simulation exercise are presented graphically. In Figure 5.6 the
line 1 represents the CFD computed for the true population computed from the orig-
inal data. The line 2 is the estimated CFD computed from kriging estimates based on
samples from the true population. The line 3 lines represent the min and max of the
1000 conditionally simulated CFDs. The two line 4s represent the 2.5 and 97.5
percentiles of the 1000 conditionally simulated CFDs, which is the proposed 95
percent confidence interval. The line 5 is the median of the 1000 CFD curves.
The results in Figure 5.6 are unusual in several respects. First note that the line 2
shows the typical sample size bias for the CFD as described above (n=40). Relative
to the true CFD (line 1) the estimated CFD is below line 1 for half the curve and
above line 1 for the remainder. The first unusual feature is that the distribution of the
conditionally simulated CFD curves is not centered on estimated CFD. In fact the
estimated CFD is not completely within the bounds (min, max) of the conditionally
simulated population. A surprising feature is that the median of the simulated popu-
lation tracks fairly well with the true CFD (line 1). It is clear that the simulated CFD
population is estimating something other than what is estimated by the estimated
CFD (line 2). At the same time, it appears that the median of the simulated popula-
tion is a good estimator of the true CFD and the proposed confidence bands (line 3)
is reasonable confidence envelop about the true CFD.
What follows is a heuristic explanation for why CFD computed from conditional
simulations might be a better estimator of the true CFD than a CFD computed from
appendix a » The Cumulative Frequency Diagram for Determining Water Quality Attainment
-------
A-47
CFD simulations
CD
E
"o
o
op
o
(O
o
CNI
o
o
O
0.0 0.2 0.4 0.6
fraction of space
0.8
1.0
5,6, Confidence bounds computed based on quantiles of fraction of space
computed on conditionally simulated surface estimates using variogram estimates
from data. The base simulation has spatial correlation and no spatial or temporal trend.
Sample size is 40.
the kriging estimator. Additional analyses test whether this property might hold in
general or is an artifact of the simple conditions (no spatial or temporal trend) under
which this experiment was performed.
In prior discussions we have noted that the CFD is the inverse of the CDF of the
population of p's where p is fraction of space out of compliance with the criterion
threshold. It is the variance of the p's that determines the steepness of the CFD: the
smaller the variance, the steeper the CFD. In real applications, estimates of the p's
have two important variance components. One variance component comes from true
variance over time in the parameter being assessed. Another variance component
comes from imperfect estimates due to sampling variability. In the base simulation
with no spatial or temporal trend in the data, it is this second source of variance that
controls the shape of the CFD.
appendix a
-------
A-48
Because the variance of the p's is critical to the shape of the CFD, consider the vari-
ance of p's computed from three sources in the experiment outlined above: 1) the true
data, 2) a krig estimate based on a sample from the true data, and 3) conditionally
simulated data based on a krig estimate of 2). To enhance our understanding of this
comparison, the variance of the p's are discussed for two cases for each source. The
first case assumes complete independence in the base simulation and does not use
interpolation to estimate proportion of area out of compliance. This simplification
allows us to easily infer the behavior of the CFD using analytical methods. The
second case introduces an unknown spatial dependence in the base simulation and
uses interpolated data to estimate the proportion of area out of compliance. These
additional complexities make it difficult to implement analytical inference but
conclusions may still be inferred by analogy to the simple independent case.
Consider the sequence of sources where the base simulations are generated under the
simple constraints of constant mean, constant variance and the errors for each cell of
the grid that are independent. For this case the exceedance probability is:
where : C is the criterion threshold,
xs is the data at location s,
|i is the mean used in the simulation,
crns the variance used in the simulation, and
is the standard normal Cumulative Distribution Function.
The distribution of the true p's computed from all 300 cells of the 5x60 simulation
grid would behave like that of a independent binomial with N=300 with a variance
of (p(l-p)/300). From these independent data draw a sample of size 40. Using only
the proportion of the sample that is out of compliance to estimate the p's, the distri-
bution of the p's would be that of a independent binomial with N = 40 and variance
(p(l-p)/40). Clearly the p's estimated from the sample of 40 have much larger vari-
ance than p's from the base simulation with 300 cells. Thus the true CFD computed
using data from 300 cells will be steeper than the sample CFD computed from 40
data points. This pattern is illustrated by comparing the true CFD (line 1) and the
estimated CFD (curve 2) in Figure 5.6. This increase in the variance of the p's due to
small sample size is the kernel of the sample size problem with the CFD. Now
consider the behavior of p's computed from conditional simulations based on the
sample. Compute x and s as estimates of® and (D from the sample of 40 in the usual
way. The conditional simulation is done by populating the 5x60 grid with data from
a normal distribution with mean jc; and variance s2j. The exceedance probability for
these simulated data for the ith month is
where : xss is simulated data at location s
jq is the estimated mean used in the conditional simulation, and
Sj is the estimated standard deviation used in the conditional simulation.
If the p' were constant over months, the variance of the p's estimated by conditional
simulation would be (p'(l-p')/300). The sample size component of this variance has
been standardized to 300 which is the same as the sample size component of the true
p's, but the variability of conditionally simulated p's will be greater than that of true
-------
A-49
p's because estimates of jc; and s2j will vary over months. The parameter p and it's
estimate p' will be close if x and s are close to CD and (D. In the simple case with
constant mean and independent errors, the CFD estimated by conditional simulation
will better approximate the true CFD because both are based on binomial distribu-
tions with the same N and approximately the same p.
Now consider the same sequence of distributions where the assumption of inde-
pendence is relaxed and interpolation of the data is used to estimate the proportion
of noncompliance. The introduction of spatial covariance in the base simulation
changes distribution of the true p's to a dependent binomial. The dependent binomial
will have variance similar to an independent binomial with N < 300. Sample size that
approximates the variance of the dependent binomial is termed Nb. The variance of
the p's estimated from spatially dependent data is approximated by (p(l-p)/Nb)
where Nb < 300 and thus the CFD from the independent case will be steeper than
from the dependent case. The degree to which Nb is less than N will depend on the
strength of the spatial correlation.
Next consider the effect of dependent data and interpolation on the distribution of the
p's. When we interpolate the sample of 40 onto the grid of 300, the interpolated
surface is smooth relative to the original data (compare curves 1 and 4 in Figure 5.2).
Because of this increased dependence in the krig estimates, the estimates of p
computed from the interpolated data behave more like binomial data with N=Ns (the
sample size) than like binomial data with N=Nb (the number of grid cells). Because
Ns is smaller than Nb, the variance of the population of p's computed from interpo-
lated data will be greater. The greater variance explains why curve 1 in Figure 5.6is
much flatter than line 1.
Finally consider the effect of conditional simulation on the distribution of the p's.
When data are conditionally simulated and the mean and variogram estimated from
the sampled data are accurate, then the character of the simulated data will be similar
to that of the true data (compare the line 1 with line 3 in Figure 5.7). Like the simple
independent case, the population of p's computed from the conditionally simulated
data will have a binomial variance that is similar to a binomial with sample size Nb.
The simulation experiment shows that the CFD computed from these conditionally
simulated p's will have a shape similar to the true CFD. This effect is illustrated in
Figure 5.6 where the median of the conditionally simulated CFDs (blue line) is more
similar to the true CFD line 1 than is the CFD estimate based on kriging (red line).
Additional analytical work is needed to formalize the heuristic concepts presented
here, but this finding indicates a productive direction in developing statistical infer-
ence procedures in the CFD approach.
The most successful technique for computing confidence bounds for the CFD were
obtained using conditional simulation based on kriging interpolation of the sample
data. The 95% confidence bands (lines 2, Figure 5.6) are well centered over the true
CFD (line 1) for the simplistic case where the true data have spatial dependence but
no spatial or temporal trends. When these simplistic assumptions are relaxed (Figure
5.8) and the true data are simulated to have spatial dependence and temporal and
appendix a * The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-50
Center Transect
O)
o
o
o
(1)
—— krig estimate (2)
— Conditional Simulation (3)
I
10
I
20
I
30
northing
I
40
I
50
I
60
Figure 5,7. Simulated chlorophyll data, kriging estimates based on a sample of the
simulated data, and conditionally simulated data where the simulation is conditioned
on the data used obtain the kriging estimates.
spatial trends similar to chlorophyll data from the Patuxent estuary, the confidence
bands cover the true CFD in this case as well. Experiments that varied the sample
size also produced confidence bands with good coverage.
Additional evaluation of the confidence band procedure should include a series of
confidence band coverage experiments to assess the true coverage rate in comparison
to the nominal coverage rate (e.g., 95% in this example). This series of experiments
should be conducted with simulated data where the simulations are designed to
produce data with properties similar to the three primary assessment water quality
parameters.
appendix a « The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-51
CFD simulations
c
.
CO
o
CD
O
2
p
O
0.0
0.2
I I
0.4 0.6
fraction of space
0.8
1.0
5,8, Confidence bounds based on quantiles of fraction of space computed on
conditionally simulated surface estimates using variogram estimates from data.
The base simulation has spatial and temporal trend estimated from Patuxent data.
Sample size is 40.
6.0 OF
6.1. CFD AS
This report represents an initial expert review of the CFD compliance approach. In
addition the panel undertook simulation tests on the effects of 1) sample densities in
time and space, 2) varying levels of attainment, and 3) varying degrees of spatial and
temporal covariance. Further, trials of spatial modeling on fixed station Chesapeake
Bay water quality data were conducted to begin to evaluate spatial modeling proce-
dures. Based upon review of underlying theory, initial statistical assessments, and
implementation feasibility, the panel finds that the CFD approach currently repre-
sents best available science in its application to water quality attainment
appendix
-------
A-52
determinations in the Chesapeake Bay. Using criteria for Best Science and Best
Available Science developed by the American Fisheries Society and the Estuarine
Research Federation (Sullivan et al. 2006), we list relevant attributes of the CFD
approach (Table 6.1).
The CFD builds on important statistical theory related to the cumulative distribution
function and as such, its statistical properties can be simulated and deduced. We have
also shown that it is feasible to construct confidence ellipses that support inferences
related to threshold curves or other tests of spatial and temporal compliance. Work
remains to be done in understanding fundamental properties of how the CFD repre-
sents likely covariances of attainment in time and space and how temporal and
spatial correlations interact with sample size effects. Further, more work is needed
in analyzing biases across regions and designated use segments. The panel expects
that a two-three year time frame of directed research and development will be
required to identify and measure these sources of bias and imprecision in support of
attainment determinations.
Through simulations of the CFD approach, it is feasible to analyze bias and error for
both temporal and spatial sources of attainment variability. In particular, conditional
simulations merit additional investigation as a relatively unbiased approach for
supporting statistical comparisons among CFD curves. Much work remains to be
done in understanding fundamental properties of how the CFD represents likely
covariances of attainment in time and space. Still, the panel finds the approach
feasible: one which merits additional development, testing, and application. Indeed,
the CFD approach is beginning to attract scientific and management attention
outside the Chesapeake Bay community.
As shown by analyses in previous sections, the approach can efficiently combine
spatial and temporal data to support inferences on whether regions within the Chesa-
peake Bay attain or exceed water quality standards. On the other hand, we recognize
substantial bias and imprecision can occur due to small sample size, non-independ-
ence in temporal trends, and inadequate spatial interpolations. More work is needed
in analyzing these biases across regions and designated use segments. Further, the
old saw of needing more samples cannot be ignored. In particular, the panel is opti-
mistic in the application of continuous spatial data streams made available through
the cruise-track monitoring program, and the promise of continuous temporal data
through further deployment of remote sensing platforms in the Chesapeake Bay
(CBOS web site, etc). These data sets will support greater precision and accuracy in
both threshold and attainment determinations made through the CFD approach.
In classifying the CFD approach as best available science, we seek to make several
important distinctions (Table 6.1). First, the CFD approach is a scientifically based
approach based upon its clear purpose, conceptual and design framework, empirical
procedures, documentation, and intent to develop rigorous statistical and review
procedures (Sullivan et al. 2006, Daubert v. Merrell Dow Pharmaceuticals, Inc.,
1993). That the approach permits evaluation of uncertainty also supports its classi-
fication as best available science (Christman 2006). On the other hand, we do not
believe that the CFD approach yet constitutes best science. Here, further analyses of
underlying statistical properties of the approach (including sampling design and
-------
A-53
6-1. Evaluation of CFD approach as Best Science or Best Available Science accord-
ing to AFS/ERF "Defining and Implementing Best Available Science for Fisheries
and Environmental Science, Policy, and Management" (Sullivan et al. 2006).
Attribute
Clear Objective
Conceptual
Model
Experimental
Design
Statistical Rigor
Clear
Documentation
Peer Review
Best
Science
YES
YES
NO
NO
YES
NO
Best
Available
Science
YES
YES
YES
YES
YES
YES
Current State of Development of CFD
Approach
Using biological response standards, combine
available water quality in time and space to determine
levels of attainment of Bay segments.
1 . Bay divided into functional classifications -
"Designated Uses."
2. Reference curves establish biologically
relevant threshold levels for attainment.
3. CFD combines and weights equally temporal
and spatial sources of water quality
variability.
1 . Bay segments are quasi-stratified for water
quality data collection.
2. Stratification of water quality data by
designated units does not yet occur.
3. Seasonal assessment of water quality
attainment through spatial interpolation and
the CFD approach is feasible but incompletely
developed.
1 . Procedures for quantifying uncertainty
associated with sampling design, spatial
interpolation and CFD approach are feasible
but incompletely developed.
2. Procedures for interpolating water quality data
are feasible but incompletely developed,
particularly for 3-D interpolations of
dissolved oxygen.
3. Procedures for testing inferences related to the
CFD curve are feasible but incompletely
developed.
CFD approach, water quality sampling design, and
current interpolation procedures well documented in
Chesapeake Bay Program Reports and on website.
1 . CFD approach and sampling design upon
which it is based has not been peer-reviewed
in the scientific literature.
2. This report comprises the first external review
by scientists with statistical expertise.
3. Grey literature reports produced by CBP
received expert and stakeholder input.
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-54
interpolation elements) and vetting by outside experts is needed. Indeed, although
the CFD approach is beginning to get featured in scientific venues, it has not yet
been reviewed as part of the scientific literature. The panel sees this as an overdue
next step for necessary for its acceptance, further development, evaluation, and
application.
The panel contrasted the CFD approach with existing state and jurisdictional water
quality criteria and attainment procedures that are based strictly upon the observed
sample, where site selection is not based upon probability sampling, inferences are
not based upon error structure, and monitoring does not involve a scientifically
rational design. Indeed, standard practice for assessing compliance with water
quality criteria throughout the US is to sample monthly at a fixed set of stations and
make judgments about compliance strictly from those samples. Sampling stations
are typically located for convenience (e.g., bridge overpasses), there is reluctance to
re-evaluate and change location (so as to maintain a time series at a fixed point), and
no consideration is given to representativeness of the sample for the space/time not
sampled. Thus the previous method used by the Chesapeake Bay Program, similar to
the approaches used in other states, was simply based on EPA assessment guidance
in which all samples in a given spatial area were compiled and attainment was
assumed as long as > 10% of the samples did not exceed the standard. In this past
approach all samples were assumed to be fully representative of the specified space
and time and were simply combined as if they were random samples from a uniform
population. This approach was necessary at the time because the technology was not
available for a more rigorous approach. But it neglected spatial and temporal patterns
that are known to exist in the standards measures. The CFD approach was designed
to better characterize those spatial and temporal patterns and weight samples
according to the amount of space or time that they actually represent.
6,2 TilK CB03 ! I M,.i P< ER
The panel views the CFD approach as innovative, one that has general application in
water quality attainment assessments, but scientific acceptance of the approach will
require that it is subjected to more extensive and targeted peer-review in the scien-
tific literature. Because the CFD is a regulatory tool, it is particularly important that
the approach is effectively communicated to the scientific community at large, for
general acceptance but more critically for the sustained research and development
that the CFD, as a nascent approach, requires. As highlighted elsewhere, bias and
imprecision that can occur due to small sample densities, non-independence in
temporal trends, and inadequate spatial interpolations. Such work is novel and
should elicit interest among biostatisticians as it addresses questions of both funda-
mental and applied consequence.
Although, continued working groups, involvement through STAC of expert biosta-
tisticians, and related reports such as this one will remain important in scientific
acceptance of the CFD approach, the panel recommends immediate attention in
subjecting the CFD to traditional peer review. One or several review papers should
be submitted by CFD principals that lay out the theory, general approach and lists
emergent scientific issues to stimulate other scientists to begin to address such
issues. Several such papers might be appropriate given potential interest by
appendix a * The Cumulative hrequency Diagram Method for Determining Water Quality Attainment
-------
A-55
biostatisticians and environmental and regulatory scientists. Scientific interest will
also be garnered by public and stakeholder interest. The CFD approach here presents
a challenge as it is complex in explanation. Still with careful diagrams and examples,
a brochure on the CFD approach should be extremely useful in getting uninitiated
scientists and stakeholders on the same page.
6.3. BIOLOGICAL REFERENCE CURVES
The success of the CFD-based assessment will be dependent upon decision rules
related to the biological reference curves. These curves represent desired segment-
designated use water quality outcomes and reflect sources of acceptable natural
variability. The reference and attainment curves follow the same general approach in
derivation—water quality data collection, spatial interpolation, comparison to
biologically-based water quality criteria, and combination of space-time attainment
data through a CFD. Therefore, the biological reference curve allows for implemen-
tation of threshold uncertainty as long as the reference curve is sampled similarly to
the attainment curve. Bias and uncertainty are driven in CFD curves by sample
densities in time and space. Therefore, we advise that similar sample densities are
used in the derivation of attainment and reference curves. As this is not always
feasible, analytical methods are needed in the future to equally weight sampling
densities between attainment and reference curves.
Conceptually, the CFD approach builds on the underlying view that water quality
criteria are surrogates for Designated Uses (regions that define ecosystem function).
Implicit is a bottom up model based upon eutrophication, which is expected to
diminish the designated use. Reference curves represent thresholds related to the
functioning of designated use regions. Therefore, choice of reference regions or
periods and sampling design in developing reference curve is critical to the imple-
mentation of a scientifically-rigorous CFD approach. Choice of such regions is
beyond the scope of this review, but we emphasize several relevant statistical issues
in developing reference curves in Section 4.
7.0 RECOMMENDATIONS FOR FUTURE EVALUATION
AND REFINEMENT OF THE
CFD ASSESSMENT METHODOLOGY
As part of its conclusions, the STAC CFD Review Panel identified critical remaining
issues that need resolution in the near future. The following is a list of critical aspects
of that needed research. These research tasks appear roughly in order of priority.
However, it must be recognized that it is difficult to formulate as set of tasks that can
proceed with complete independence. For example, research on task 1 may show
that the ability to conditionally simulate the water quality surface is critical to
resolving the sample size bias issue. This discovery might eliminate IDW as a choice
of interpolation under task 3. The Panel has made significant progress on several of
these research tasks and CBP is encouraged to implement continued study in a way
that maintains the momentum established by this research group (Table 7.1.).
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-56
7.1. Research Tasks, examples of specific subtasks, and suggested time frame for
continued CFD research.
Task
Schedule
1. Effects of Sampling Design on CFD Results
(a) Continue simulation work to evaluate CFD bias reduction
via conditional simulation.
(b) Investigate conditional simulation for interpolation
methods other than kriging - this may lead to more simulation work.
(c) Implement and apply interpolation with condition
simulation on CBP data.
2006-2008
2. Statistical inference framework for the CFD
(a) Implement and evaluate confidence interval procedures.
(b) Conduct confidence interval coverage experiments.
(c) Investigate confidence interval methods for non-kriging
interpolation methods.
(c) Implement and evaluate confidence interval procedures.
2006-2008
3. Choice of Interpolation Method
(a) continue to investigate other more nonparametric
interpolation methods (e.g. loess and splines).
(b) implement a file system and software utilizing the "best"
interpolation for CBP data.
(b) compare interpolations and CFD's based on IDW and
"best" method.
2006-2008
4. Three-Dimensional Interpolation
(a) Implement 2-D kriging in layers to compare to current
approach of 2-D IDW in layers.
(b) Conduct studies of 3-D anisotrophy in CBP data.
(c) Investigate software for full 3-D interpolation. Examples
of options include: custom IDW software, custom kriging software
using GMS routines, custom kriging software using the R-package, or
some other off the shelf product.
2007-2009
5. High Density Temporal Data
(a) Develop methods to use these data to improve temporal
aspect of CFD in current implementation.
(b) Investigate feasibility of 4-dimensional interpolation.
2008-2010
appendix a
-------
A-57
1. Effects of Sampling Design on CFD Results. The CFD is a special case of an
unbiased estimator for a cumulative distribution function of a population. Like
the cumulative distribution function, the CFD is a function of the mean and the
variance of the population being assessed. And the better the mean and vari-
ance are characterized with sample data, the more accurate the shape of the
CFD will be. As the sampling density increases, the estimated CFD begins to
approach the true CFD. However, if the sampling density is low, then sampling
error could become important and there is potential that it could affect the
shape of the CFD and ultimately the accuracy of the compliance assessment.
Furthermore the potential for the sample size to affect the shape could create a
compliance assessment bias if the reference curve and assessment curve are
based on different sampling densities. Conditional simulation methods devel-
oped by STAC panel members showed promise toward resolving these issues
and mitigating potential biases caused by differences in sample size.
2. Statistical inference framework for the CFD. It is important in a regulatory
process to differentiate an exceedance that is small and might have resulted
from chance variability from those that are large and indicative of an inherent
problem. This differentiation will require mathematical tools to quantify the
variability in the CFD that occurs as a result of sampling. The STAC panel
made progress on this issue by demonstrating a confidence interval procedure
based on conditional simulation associated with kriging. It remains to be
assessed whether or not confidence intervals produced by this algorithm
perform at the nominal level of coverage, fore example, does a nominally 95%
CFD confidence interval cover the true CFD 95% of the time.
3. Choice of Interpolation Method. The STAC panel considered several inter-
polation methods and outlined the features of each. Those features illustrate
tradeoffs between ease of implementation and maximizing the information
garnered from the data. Further work is needed to compare the features to the
requirements of wide-scale implementation of assessment procedures and
formulate a plan for tractable implementation that results in credible assess-
ments. One strategy is to implement easily performed analysis (e.g. IDW) as a
screening tool to identify cases where compliance / non-compliance is clear,
and then implement more labor intensive methods (e.g. kriging) for cases
where compliance is more difficult to resolve. One difficulty with imple-
menting a full comparison of methods is that implementation of each method
requires considerable work in terms of setting up file systems, interfacing soft-
ware and data, and coupling the considerable bathymetry data of the bay. Thus
it would be prudent to narrow the choices based on theoretical considerations
where possible.
4. Three-Dimensional Interpolation. Assessments of the dissolved oxygen
criteria require three-dimensional interpolation. However, the field of three-
dimensional interpolation is not as highly developed as that of two-dimensional
interpolation. While the mathematics of each method extend easily to three
dimensions, there are relatively few examples of 3-D interpolation available in
the literature and issues such as data density requirements for reliable results
are not well studied. Efforts are needed to further evaluate research in three-
dimensional interpolation and seek additional outside scientific input and
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-58
review with the goal of implementing the best available technology for this
aspect of criteria assessment. One of the first efforts under this task is a study
of the 3-D variance stucture of the data to be interpolated. A short term option
is to implement the optimal 2-D interpolator in layers as is done with the
current IDW interpolator.
5. High Density Temporal Data. As currently formulated, assessment for most
of the open-waters of the Bay are based on "snapshots" in time of the spatial
extent of criteria exceedence estimated via interpolation. Data collected for use
in interpolation are actually spaced over multiple days due to the large expanse
over which sampling must be conducted. It is clear that technology is becoming
available that will produce high density data in both space and time. Interpola-
tion should accommodate data that are collected densely in space. However, it
is unclear how the CFD process will accommodate data that are high density in
time. Further work is needed to evaluate methods to fully utilize the temporally
intensive data that is currently being collected.
The panel discussed several mechanisms for the CBP to make progress on chal-
lenging tasks ahead (Table 7.1). We recommend that a review panel oversee the tasks
over the next 3-5 year time frame. This panel would periodically review trials and
other products conducted by individual external scientists (academic scientists or
consultants) and existing teams of CBP scientists (e.g., the Criteria Assessment
Protocols (CAP) workgroup). Tasks 1 and 2 are most immediate and critical and we
recommend that these tasks by contracted out to external scientists, exploiting state-
of-the-art approaches and knowledge. Task 3 could be conducted through CAP or
other group of CBP scientists. Task 4 and 5 are less immediate but again will require
substantial expertise and innovation and may be most efficiently accomplished by
scientific expertise outside the immediate CBP community.
Christensen OF, Diggle PJ, Ribeiro PJ. 2001. Analysing positive-valued spatial data: the
transformed Gaussian model. In: Monestiez P, Allard D, and Froidevaux, editors. GeoENV
III - Geostatistics for environmental applications. Quantitative Geology and Geostatistics.
Dordrecht (Netherlands): Kluwer Academic Publishers. 11:287-298.
Christman MC. 2006. The characterization and incorporation of uncertainty in fisheries
management, In Fisheries Ecosystem Planning for Chesapeake Bay. Bethesda (MD): Amer-
ican Fisheries Society. (In press).
Cressie N. 1989. The Many Faces of Spatial Prediction. Mathematical Geology 1:163- 176.
Cressie N. 1991. Statistics for Spatial Data. New York: Wiley. 928 p.
Curriero FC. 2006. On the Use of Non-Euclidean Distance Measures in Geostatistics. Math-
ematical Geology (in press).
Daubert v. Merrell Dow Pharmaceuticals. Inc. 1993. 509 U. S. Supreme Court. 579.
Deutsch CV. 1984. Kriging with Strings of Data. Mathematical Geology. 26:623-638.
Diggle PJ, Tawn JA, Moyeed RA. 1998. Model Based Geostatistics (with Discussion).
Applied Statistics 47:299-350.
Diggle PJ, Ribeiro PJ. 2006. Model-based Geostatistics. New York: Springer. 230 p.
appendix a * The Cumulative hrequency Diagram Method for Determining Water Quality Attainment
-------
A-59
Dille JA. 2003. How good is your weed map? A comparison of spatial interpolators. Weed
Science 51:44-55.
Gotway, CA. 1994. The Use of Conditional Simulation in Nuclear Waste-Site Performance
Assessment. Technometrics 36:2:129-141.
Hastie TJ, Tibshirani RJ. 1990. Generalized Additive Models. New York: Chapman and Hall
p335.
Jensen OP, Christman MC, Miller TJ. 2006. Landscape-based geostatistics: A case study of
the distribution of blue crab in Chesapeake Bay. Environmetrics 17:605-621.
Journel A. 1974. Geostatistics for conditional simulation of ore bodies. Economic Geology
69:673-687.
Kitanidis PK. 1997. Introduction to Geostatistics: Applications in Hydrogeology. New York:
Cambridge University Press. 271 p.
Kravchenko AN. 2003. Influence of Spatial Structure on Accuracy of Interpolation Methods.
Soil Sci. Soc. Am. J. 67:1564-1571.
Kutner MH, Nachtsheim CJ, Neter J, Li W. 2004. Applied Linear Statistical Models, 5th
edition. Boston: McGraw-Hill.
Laslett GM. 1994. Kriging and splines: an empirical comparison of their predictive perform-
ance in some applications. JASA 89:391-400.
Lloyd CD. 2005. Assessing the effect of integrating elevation data into the estimation of
monthly precipitation in Great Britain. Journal of Hydrology 308:128-150.
Ouyang Y, Zhang JE, Ou LT. 2006. Temporal and spatial distributions of sediment total
organic carbon in an estuary river. J. Environmental Quality. 35:93-100
Reinstorf F,Binder M, Schirmer M, Grimm-Strele J, Walther W. 2005. Comparative assess-
ment of regionalization methods of monitored atmospheric deposition loads. Atmospheric
Environment 39:3661-3674.
Ribeiro PJ, Diggle PJ. 2001. geoR: A package for geostatistical analysis. R News 1:2:15-18.
Schabenberger O, Gotway CA. 2005. Statistical Methods for Spatial Data Analysis.
Boca Raton, FL: Chapman and Hall/CRC Press. 512 p.
Spokas K, Graff C, Morcet M, Aran C. 2003. Implications of the spatial variability of land-
fill emission rates on geospatial analyses. Waste Management 23:599-607.
Sullivan PJ, Acheson J, Angermeier PL, Faast T, Flemma J, Jones CM, Knudsen EE, Minello
TJ, Secor DH, Wunderlich R, Zanatell BA. 2006. Defining and implementing best available
science for fisheries and environmental science, policy and management. Bethesda, Md:
American Fisheries Society and Port Republic, Md: Estuarine Research Federation. Port
Republic, Maryland, (available: www.fisheries.org/AFSniontana/AFS.ERF.BestScience.pdf)
Tomczak, M. 1998. Spatial interpolation and its uncertainty using automated anisotropic
inverse distance weighting (IDW) - cross-validation/jackknife approach. J. Geog. Infor. and
Decision Analysis 2:18-30.
[EPA] Environmental Protection Agency (US). 2003. Technical support documentation for
identification of Chesapeake Bay designated uses and attainability. Annapolis (MD): EPA.
177 p. EPA 903-R-03-002. (Available: http://www.chesapeakebay.net/search/pubs.htm)
Valley RD,Drake MT, Anderson CS. 2005. Evaluation of alternative interpolation techniques
for the mapping of remotely-sensed submersed vegetation abundance. Aquatic Botany 81:13-
25.
appendix a • The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
A-60
Ver Hoef JM, Peterson, E, and Theobald, D, 2007, Spatial Statistical Models that Use Flow
and Stream Distance, Environmental and Ecological Statistics (in press).
Wahba, G. 1990. Spline Models for Observational Data. Philadelphia (PA): Society for
Industrial and Applied Mathematics. 169 p.
Wang XJ, Liu RM. 2005. Spatial analysis and eutrophication assessment for chlorophyll a in
Taihu Lake. Environmental Monitoring and Assessment 101:167-174.
Zimmerman D., Pavlik C, Ruggles A, Armstrong MR 1999. An experimental comparison of
ordinary and universal kriging and inverse distance weighting. Mathematical Geol. 31:375-
390.
appendix a » The Cumulative Frequency Diagram Method for Determining Water Quality Attainment
-------
B-1
appendix
Detailed Chesapeake Bay Water
Quality Criteria Assessment
Methodology
The methods in this appendix apply specifically to the evaluation of dissolved
oxygen criteria. For water clarity criteria or chlorophyll a criteria evaluations, the
individual methods are very similar to those described here. See chapters 5 and 6,
respectively, for additional details. Chapter 7 also contains important information in
using shallow-water data for criteria attainment assessment of all three parameters.
Data come from the Chesapeake Bay Program's Chesapeake Information Manage-
ment System (CIMS) database or through the CIMS partners' networked databases.
The parameters extracted include date, location, depth, salinity, temperature, and the
water quality parameter under assessment. Data identified by the states, but collected
from other than the Chesapeake Bay Water Quality Monitoring Program and Chesa-
peake Bay Shallow-water Monitoring Program, are also obtained. These data must
be of known and documented quality as described in Chapter 3.
Once the data are compiled, they are assigned to a time period based on the sample
date. Fixed-station data are normally collected during a monitoring cruise that covers
the entire tidal Chesapeake Bay over several days. To provide a "snapshot" of water
quality, however, the data collected within one cruise are considered contempora-
neous to enable a single spatial interpolation. For information not associated with a
cruise, such as state-supplied data, a cruise number is assigned representing the
closest cruise in time to the collection of each data point. Co-located data points in
the same cruise are averaged.
The criteria assessment procedure requires evaluation over large areas rather than at
distinct points. Spatial interpolation is carried out for each water quality criteria
parameter for each cruise (see Appendix D for details on the Chesapeake Bay inter-
polator and the interpolation process) with water clarity and chlorophyll a data
interpolated in the two horizontal dimensions using inverse distance squared
weighting and natural logarithm transformation. Dissolved oxygen data are first
appendix b • Detailed Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
B-2
linearly interpolated vertically within each column of observed data beginning at
0.5 meters below the water surface and continuing at one-meter intervals, without
exceeding the deepest observation in that water column. Data at each depth is then
interpolated horizontally using inverse distance squared weighting. Data regions
were specified for each segment to prevent the interpolation algorithm from using
data points in neighboring tidal tributaries (described in the section below and in
detail in Appendix D).
Some designated uses for dissolved oxygen during the summer in the Chesapeake
Bay and its tidal tributaries and embayments are defined vertically to distinguish
stable water layers with different criteria levels (U.S. EPA 2003a, 2003b). In areas
and seasons for which vertical stratified criteria apply, the surface mixed layer (open
water) is that layer above the pycnocline and, thus, exposed to the atmosphere. The
transitional middle layer (deep water) is the layer between the upper and lower pycn-
ocline boundaries. The lower layer (deep channel) is the water below the lower
pycnocline boundary. Given that the pycnocline is dynamic and moves up and down
with each monitoring cruise, the designated use of each interpolator grid cell must
also be defined based on the data for each cruise.
Temperature and salinity are used to calculate density; density, in turn, is used to
calculate pycnocline boundaries. Density is calculated using the method described in
Algorithms for Computation of Fundamental Properties of Seawater1. For each
column of temperature and salinity data, the upper and lower pycnocline boundaries
are determined by looking for the shallowest robust vertical change in density of
0.1 kg/m3/m for the upper boundary and the deepest change of 0.2 kg/m3/m for the
lower boundary. To be considered robust, the density gradient must not reverse direc-
tion at the subsequent measurement and must also demonstrate a change in salinity
of at least 0.1 psu per meter (not merely a change in temperature). Chapter 7 in U.S.
EPA 2004, pages 85-87, documents the detailed method for determination of both
the vertical density profile and the pycnocline.
The depths to the upper pycnocline boundary (where detected) and the fraction of
the water column below the lower boundary are interpolated in two dimensions. If
no lower boundary was detected, then the fraction is set at zero. The depth to the
upper pycnocline boundary tends to remain stable in the horizontal dimension,
meaning that spatial definition of that boundary using interpolation generally works
well. Interpolation of the lower boundary is more complicated because the results
may conflict with the upper boundary definition or with the actual bathymetry of the
Chesapeake Bay. Consequently, interpolation of the lower boundary is based on the
fraction of water column depth. In this way, the constraints of the upper pycnocline
boundary definition and the actual Bay bottom depth are imposed, eliminating errors
related to boundary conflicts.
Endorsed by UNESCO/SCOR/ ICES/IAPSO Joint Panel on Oceanographic Tables and Standards and
SCOR Working Group 51. N.P. Fofonoff, and R.C. Millard, Jr., 1983. UNESCO Technical Papers in
Marine Science. Paris, France. No. 44, p. 53.
appendix b » Detailed Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
B-3
Criteria assessments are based on each component criterion's specific averaging
period. Assessments of attainment of the instantaneous minimum criteria are directly
evaluated using the individual cruise interpolations. All 30-day mean criteria assess-
ments rely on monthly averages of interpolated data sets. To calculate these
averages, each interpolated cruise within a month is averaged on a point-by-point
basis in matching interpolator grid cells. Generally, two cruises per month run
through the warm season with one cruise per month during the cooler period. Spatial
violation rates are calculated for each temporally aggregated interpolation in an
assessment period. For example, the 12 monthly average interpolations representing
the four summer months (June, July, August, September) over three years were used
for a three-year summer open-water dissolved oxygen assessment.
Cumulative frequency diagrams (CFD) are generated from the spatial violation rates
for each assessed designated use, water quality parameter, criterion, and averaging
period using the Weibull plotting position (rank/(n+l)).
The assessment CFD is compared to a reference CFD to determine if unallowable
exceedances of the criterion occur. The diagrams of both CFDs show three areas:
non-exceedance (above the assessment curve), allowable exceedance (below both
curves), and unallowable exceedance (below the assessment curve and above the
reference curve). If the assessment CFD surpasses the reference CFD at any point,
an unallowable exceedance exists.
Reference CFDs are continuous or generally have many more points than assessment
CFDs. This situation can lead to spurious unallowable exceedances even without
individual points in the assessment CFD topping reference CFD levels. To address
this problem, reference curves are evaluated only at the temporal axis points in the
assessment curve (see Figure II-7 in Chapter 2). For non-continuous biological refer-
ence curves, these points are interpolated from neighboring points.
The trapezoidal rule is used to calculate the areas. This rule is a method of approxi-
mate integration, which calculates the areas of discrete trapezoids that make up the
area below a curve when summed. Since both the assessment and reference curves
are piecewise linear, repeated application of the trapezoidal rule results in an exact,
rather than approximate, value.
For dissolved oxygen criteria assessed without reference curves, the assessment
space is divided in two—non-exceedance and unallowable exceedance.
U.S. Environmental Protection Agency (U.S. EPA). 2003a. Ambient Water Quality Criteria for
Dissolved Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and its Tidal Trib-
utaries. EPA 903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2003b. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability. EPA 903-R-03-004. Region III Chesapeake Bay
Program Office Annapolis, MD.
appendix b * Detailed Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
B-4
U.S. Environmental Protection Agency. 2004. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a Chesapeake Bay and its Tidal Tributaries: 2004
Addendum. EPA 903-R-04-005. Region III Chesapeake Bay Program Office, Annapolis, MD.
appendix b « Detailed Chesapeake Bay Water Quality Criteria Assessment Methodology
-------
C-1
appendix
Evaluation of Options for
Spatial Interpolation
Interpolation constitutes a critical element of CFD-based assessment methodology.
It provides the spatial framework for data integration while allotting the appropriate
weight to all data. The spatial framework consists of a grid made up of a network of
cells that vary in size to cover the entire spatial domain. The size of the cells deter-
mines the scale of the assessment; smaller and more numerous cells in a given area
provide a more spatially detailed assessment. Estimates for all cells come from a
spatial interpolation algorithm.
To date, two spatial interpolation algorithms have been considered: inverse distance
weighting (IDW) and kriging. In IDW, estimates of water quality levels are based on
a weighted average derived from the closest measured data values. Weights depend
upon the distance between the measurement point and the cell being estimated. Thus,
measurements from the closest points are weighted most heavily and have the most
influence. The second method is kriging—a well-known statistical form of spatial
interpolation. The statistical details of kriging rest on ample research. This method,
however, has not been used for water quality criteria. Both spatial algorithm methods
can prove valuable for Chesapeake Bay water quality criteria assessment; one or
both will likely be used in the future. Other methods (non-parametric regression
methods such as Loess regression or cubic splines) are also available and could also
be considered for future use. Further details on the IDW and kriging methods are
provided below.
SPATIAL INTERPOLATION NEEDS SPECIFIC TO CHESAPEAKE
BAY WATER QUALITY CRITERIA ASSESSMENT
The Chesapeake Bay water quality criteria were established using the spatial defini-
tion of designated-use areas for the tidal waters of Chesapeake Bay (U.S. EPA
2003a, 2003b). These spatial definitions, along with the characteristics of the Bay
itself, present several challenges for spatial interpolation. For example, the Chesa-
peake Bay shoreline is extremely complex with many small tidal tributaries,
embayments, and inlets that occur at various scales throughout the water body. The
small inlets present a challenge for spatial interpolation because they require
appendix c • Evaluation of Options for Spatial Interpolation
-------
C-2
extrapolation from measured areas into unmeasured areas, often around numerous
bends and twists in a tidal river. Furthermore, they create the potential for interpo-
lating from one tidal tributary to another, which may be inappropriate since tidal
tributaries are often hydrodynamically independent. Most spatial interpolation algo-
rithms operate in two dimensions in a relatively simple spatial domain. Thus,
specific refinements need to be made for the algorithms used in Chesapeake Bay
criteria assessment.
The Chesapeake Bay dissolved oxygen criteria depend on designated-use areas—
specific volumetric areas with both vertical and horizontal dimensions (U.S. EPA
2003a, 2003b). Dissolved oxygen levels are naturally lower in bottom waters. There-
fore, the designated-use areas were defined as vertically stratified layers to allow
establishment of criteria levels that support the ecological communities residing in
the lower depths of the Bay. Any spatial interpolation supporting dissolved oxygen
criteria assessment must allow interpolation throughout the designated-use volumes
in three dimensions. The IDW algorithm developed and used by the Chesapeake Bay
Program was designed in this way and has been used consistently to provide
baywide maps of dissolved oxygen concentrations (see Appendix D). Kriging,
however, has not been used for three-dimensional interpolation in the Chesapeake
Bay to date; in fact, only limited research has taken place to develop the capability
of three-dimensional kriging for any purpose (STAC 2006). Thus, more research
may be required for the use of kriging in the assessment of dissolved oxygen criteria.
The complexity of the Chesapeake Bay shoreline presents several obstacles for
spatial interpolation in Bay tidal waters, mostly related to interpolating across land
area. Most spatial interpolation algorithms assume a relatively simple spatial domain
(e.g., rectangular) and interpolation takes place without regard to direction. In
contrast, the Chesapeake Bay (for example, see Figure III-l in Chapter 3) displays
tidal flow patterns that make some locations independent or virtually independent.
For Bay water quality criteria assessment, therefore, the influence between some
locations must be limited when interpolating spatially. The current Chesapeake Bay
Program interpolator provides limits by using data regions in which the data used to
estimate values in given locations are limited to certain areas (see Appendix D for
additional details). Similar or alternative methods may be required to apply kriging
broadly.
As described above, the Chesapeake Bay Program collects two types of data for
criteria assessment; these two data types supply information at different spatial
scales. The fixed-station Chesapeake Bay Water Quality Monitoring Program
collects data consistently for the entire Bay as well as its tidal tributaries and embay-
ments. The Chesapeake Bay Shallow-water Monitoring Program offers much more
detailed information within Bay tidal tributaries and across all shallow-water habi-
tats. Given the different spatial scales of these two monitoring programs, it is
unlikely that they can be used in the same interpolations. Thus, two separate inter-
polation approaches—each designed for specific types of criteria attainment
assessments—may prove necessary.
Since the Chesapeake Bay water quality criteria and the CFD-based criteria assess-
ment methodology were developed and published, interest has developed in creating
appendix c • E<
-------
C-3
a statistical basis for decision-making using the CFD (see pages 164-165 in U.S.
EPA 2003a). Such a basis would allow the incorporation of error analysis into the
criteria attainment assessment methodology. It would also allow the differentiation
of an assessment based on a well-characterized system from one that was poorly
characterized. Estimates of interpolation error are important to develop such a statis-
tical framework. Such estimates allow decision-making to be based on the number
(density) of sampling locations and promote greater statistical certainty (i.e., greater
sampling density) in the assessment. The current Chesapeake Bay interpolation algo-
rithm does not yield spatial error estimates (Appendix D); however, kriging is a
possible alternative algorithm that can provide spatial interpolation error (STAC
2006).
Chesapeake Bay spatial interpolation requires the potential for automation. For
many reasons, the Chesapeake Bay Program must compute many interpolations
quickly. In developing the attainment figures for the 2006 listing cycle, for example,
the program performed a total of 2328 interpolations for the final criteria assessment
analysis of the 95 water quality segments). During development of the methodology,
these interpolations were carried out repeatedly. Also, water quality models are often
used to evaluate the potential benefits of management actions with the generation of
multiple scenarios. Management action success is often defined in terms of water
quality criteria, with results evaluated similarly to the actual measurements. Given
the large number of data sets, automating the criteria assessment methodology and
spatial interpolations would likely prove necessary. The current Chesapeake Bay
interpolator allows automation and has been used in this way (Appendix D). Kriging,
however, is a more detailed analysis that requires multiple decisions along the way,
is not conducive to automation, and may not necessarily remain consistent within
and between jurisdictions.
IO (
TV
As stated, the Chesapeake Bay Program redesigned the tidal monitoring program
specifically to support water quality criteria assessment. That redesign resulted in
multiple monitoring program components, all of which address one or more of the
objectives of the Chesapeake Bay Water Quality Monitoring Program. Two of the
components that serve most of the current needs of criteria assessment include the
Bay wide Fixed-station Water Quality Monitoring Program and the Shallow-water
Monitoring Program. These two long-term efforts will provide data useful at
different scales.
The fixed-station monitoring program began in the mid 1980s and was designed to
provide data for assessing long-term trends at key sites throughout the Chesapeake
Bay and its tidal tributaries (Chesapeake Bay Program 1989). The program collects
water quality samples at more than 150 sites (Figure C-l), including 49 stations in
the mainstem Chesapeake Bay and 96 stations in the tidal tributaries. The samples
go to a network of laboratories for analysis, compiling data on 19 water quality
parameters. Fixed-station monitoring cruises run on a monthly basis throughout
appendix c « Evaluation of Options for Spatial Interpolation
-------
C-4
"
'*»»/*# \
**
*
''
't
>
f
most of the year, but occur two times a month
during the summer. At each station, samples are
collected at multiple depths depending on the
location of the pycnocline. In addition, techni-
cians collect water quality sensor data—
including water temperature, salinity, and
dissolved oxygen—along vertical profiles at
regular intervals.
The fixed-station network provides data to assess
water quality in the mid-channel, open waters of
the Bay mainstem as well as in the major tidal
tributaries and embayments. The network does
not assess conditions in the shallows since many
of the stations were purposely located in the
main channels and open tidal waters.
The Chesapeake Bay Program recently began
monitoring shallow-water habitats using a tech-
nology known as DataFlow (see Chapter 7 for
details). This new technology uses a system of
shipboard water quality probes that measure
spatial position, water depth, water temperature,
salinity, dissolved oxygen, turbidity, and fluores-
cence from a flow-through stream of water
collected near the water surface. This system
allows rapid data collection (approximately
every 4 seconds) while the boat is traveling at
speeds up to 20 knots. Due to the speed of data
collection, each cruise provides extremely
detailed data sets useful for assessing highly
variable water quality conditions, such as those
expected in the Bay's shallow waters and small
tidal tributaries. Thus, this monitoring program
specifically assesses shallow waters (STAC
2005). The spatial density of data collected by
the DataFlow system allows spatial interpolation.
The current Chesapeake Bay Program interpolation software is not designed for data
of this density, however, so new methods of interpolation need to be developed.
Due to the cost of the Shallow-water Monitoring Program, it cannot be implemented
baywide concurrently. Rather, the program is being put into practice on a rotating
basis, with the monitoring system deployed to selected assessment units long enough
to evaluate attainment and then moved to another set of units (see Chapter 7 for
further details). This set-up means that all shallow-water areas will not be assessed
simultaneously, although a full assessment will take place over time. For example,
the Maryland Department of Natural Resources' Water Quality Mapping Program
covered 14 Chesapeake Bay and tributary systems in 2005.These systems include the
St. Mary's, Patuxent, West, Rhode, South, Middle, Bush, Gunpowder, Chester,
*<•
-j i -' . The sites that make up the fixed-station
network of the Chesapeake Bay Water Quality
Monitoring Program.
Source: Chesapeake Bay Program 1989.
appendix c • Evaluation of Options for Spatial Interpolation
-------
C-5
Eastern Bay, Miles/Wye, Little Choptank, Chicamacomico, and Transquaking rivers.
In Virginia, Dataflow data are available for the Piankatank, York, Pamunkey, and
Mattaponi rivers. Chapter 7 discusses additional details on plans for monitoring
shallow-water systems.
Other alternative monitoring programs have been considered, but not fully imple-
mented for criteria assessment. Beginning in 1990, chlorophyll a concentrations
have been measured over the mainstem Chesapeake using aircraft remote sensing
(Harding et al. 1992). Twenty-five to 30 flights per year took place during the most
productive time periods. In addition, satellite remote sensing data have been consid-
ered for evaluating chlorophyll a concentrations in the Bay (Harding et al. 2004)
although no detailed evaluation of the feasibility has been completed. Water quality
sensors and data loggers mounted on buoys have also been evaluated as the best
means to assess high-frequency dissolved oxygen criteria. This option is expensive,
however, and only a limited (but growing) number of buoy systems have been
deployed to date (http://www.cbos.org).
INTERPOLATION METHODS CURRENTLY
USED FOR CHESAPEAKE BAY WATER QUALITY
CRITERIA ASSESSMENT
The current Chesapeake Bay Interpolator is a grid-based algorithm in which criteria
measurement data are used to estimate values for all grid cells (see Appendix D for
a detailed description). Estimates for cell locations are computed by interpolating the
nearest "n" neighboring water quality measurements for which "n" is normally 4 but
is adjustable. The interpolation uses an inverse distance weighted (IDW) algorithm
in which the estimated value of each grid cell is based on the four nearest measure-
ments. Each of the neighboring points is weighted by the inverse of the distance
squared (i.e., 1 d~2), however, so close neighbors have more influence than those
farther away.
The cell size in the Chesapeake Bay interpolation grid is 1 km (east-west) x 1 km
(north-south) x 1 m (vertical), with columns of cells extending from the water
surface to the Bay bottom representing the three-dimensional volume as a group of
equal-sized cells. Each tidal tributary is represented by variously sized cells
depending on the river's geometry since the narrow upstream portions require
smaller cells to model the dimensions accurately. Interpolator grid cells, however,
remain the same size within individual segments. This designation results in a total
of 51,839 cells by depth for the mainstem Chesapeake Bay (segments
CB1TF-CB8PH), and a total of 238,669 cells by depth for all 78 segments making
up the mainstem Chesapeake Bay and its tidal tributaries and embayments.
The Chesapeake Bay interpolator is optimized to compute concentration values that
closely reflect the physics of stratified water bodies such as the Bay. Water quality
varies much more markedly vertically as opposed to horizontally. To accommodate
this attribute, each column of data is interpolated vertically to the same depths as the
centroids of the interpolator cells, (i.e. 0.5,1.5, 2.5 meters, etc). The interpolator then
interpolates only in the horizontal dimension.
appendix c • Evaluation of Options for Spatial Interpolation
-------
C-6
Up to four points are used for interpolation. If fewer than four points exist, interpo-
lation is still carried out given at least one measured point. Without any measured
data, a missing value (normally a -9) is calculated for that cell. A search radius filter
limits the horizontal distance of monitoring data from the cell being computed. Data
points outside the user-selected radius (normally 25,000 m or 25 km) are excluded
from calculation. This filter ensures that only data near the location being interpo-
lated are used.
Segment and region filters have also been added. Segments are aggregations of the
interpolator cells. For instance, eight segments make up the mainstem Chesapeake
Bay (CB1TF, CB2OH,.. .CB8PH). The tidal tributaries have 70 additional segments,
created by the Chesapeake Bay Program's 2003 segmentation scheme (U.S. EPA
2004, 2005). These segments divide the Bay into geographic areas with somewhat
homogeneous environmental conditions. This segmentation also allows the reporting
of results on a segment basis, revealing more localized changes compared to the
whole Bay ecosystem.
The region file identifies the geographic boundary that limits which monitoring
station data are included in interpolation for a given segment (see Appendix D). The
purpose of the data region is to select a subset of the monitoring data from the input
data file and to use that subset for computing the values for each grid cell in a
segment. Use of data regions ensures that the interpolator does not "reach across
land" to obtain data from an adjacent tidal tributary—a process that would give erro-
neous results. By using data regions, each segment of grid cells can be computed
from its individual monitoring data subset. Each adjacent data region overlaps so that
a continuous gradient—not a seam—exists across segment boundaries. Data regions
for criteria assessment vary somewhat from the data regions in the standard interpo-
lator. These new regions were developed to exclude tributary measurements from
mainstem interpolations and to include additional observed data from Virginia.
EVALUATION OF THE INVERSE DISTANCE
WEIGHTING SPATIAL INTERPOLATION
ALGORITHM FOR ASSESSING CHESAPEAKE BAY
WATER QUALITY CRITERIA
The current Chesapeake Bay interpolator is based on an IDW algorithm—a non-
statistical spatial interpolator that uses observed data to calculate a weighted average
(as a predicted value) for each location on the prediction grid (Appendix D). The
method calculates the weight associated with a given observation as the inverse of
the square of the distance between the prediction location and the observation. The
IDW is a spatial interpolator; in general, such methods have provided good predic-
tion maps (STAC 2006). Additionally, implementation is relatively simple since
software exists to map IDW automatically. Further, the method does not require any
decisions during an interpolation session. Commercial Geographic Information
Systems (GIS) software contains IDW, requiring only GIS skills for application.
The IDW algorithm has several advantages for use in Chesapeake Bay water quality
criteria attainment assessment (STAC 2006). First, since it is non-statistical, the
appendix c • Evaluation of Options for Spatial Interpolation
-------
C-7
algorithm is not constrained by prior theoretical assumptions concerning error struc-
ture. It is, therefore, simpler mathematically and can be adapted to interpolation in
three dimensions (i.e., with depth). Second, due to its simplicity, IDW does not
require operator decisions at interim steps. Thus, it is conducive to automation—
running large numbers of interpolation without having to make decisions as part of
the interpolation process. The algorithm is susceptible to problems with interpo-
lating across land; however, methods exist to prevent such problems for Chesapeake
Bay application (as described in previous sections and in detail in Appendix D). It
can be applied at any scale, but is most appropriate for large scales where three-
dimensional interpolation becomes a necessity and data collection sites may remain
too dispersed to provide good estimates of error structure no matter which algorithm
is used.
In addition to its advantages, IDW also has a major disadvantage: it is not a statis-
tical method. The method is a deterministic approach without any sampling or model
error assumed or accounted for (STAC 2006). In addition, IDW does not account for
potential spatial autocorrelation among the observations and, therefore, does not
fully utilize the information contained within the data. No method exists to estimate
either source of error associated with a set of predicted values when using IDW and
it cannot be used as a basis for statistical decision-making using the CFD. Dedicated
research could determine whether IDW could be made more statistically defensible.
EVALUATION OF KRIGING AS A SPATIAL
INTERPOLATION ALGORITHM FOR
ASSESSING CHESAPEAKE BAY WATER QUALITY
Kriging has been considered by the Chesapeake Bay Program as a principal alterna-
tive algorithm for spatial interpolation in CFD water quality criteria assessment
methodology. Kriging is a spatial interpolation technique that arose from geostatis-
tics, a subfield of statistics that analyzes spatial data. Kriging and the field of
geostatistics have been used in a wide variety of environmental applications and are
generally accepted methods for statistically optimal spatial interpolations (Cressie
1991, Schabenberger and Gotway 2004, Diggle and Ribeiro 2006). Kitanidis (1997),
Wang and Liu (2005), and Ouyang et al. (2006) elaborate on the application of
kriging in water-related research. References on kriging methodology, geostatistics,
and their related statistical development can be found in Cressie (1991), Diggle et al.
(1998), Schabenberger and Gotway (2004), and Diggle and Ribeiro (2006).
Kriging can be formulated equivalently in terms of a general linear regression
model:
Y (s) = y30 + fii Xj(s) •••+/?£, Xp(s) + e(s) Equation C-l
with s representing a generic spatial location assumed to vary continuously over
some domain of interest and Y (s) capturing the outcome of interest measured at s,
Xj(s), . . . ,Xp(s) potential covariates indexed by location s and their associated
regression effects /?1; . . . , /?p. The uncertainty in this regression relationship is
modeled with the random error term e(s) assumed to have zero mean and constant
appendix c • Evaluation of Options for Spatial Interpolation
-------
C-8
variance. Spatial data, similar to the type sampled in Chesapeake Bay water quality
criteria assessments, often exhibit a property known as (positive) spatial dependence;
observations closer together are more similar than those further away. This property
is accounted for in the model by allowing g(s) to contain a spatial correlation
structure.
Common distributional assumptions on g(s) include normality and log normality,
although kriging can be based on other statistical distributions and data transforma-
tions. Functions of a specific mathematical type (positive definite) represent the
spatial correlation in g(s) and are assumed isotropic (correlation depends only on
distance) or anisotropic (correlation depends on both distance and direction). Vari-
ograms constitute another special type of mathematical function—closely related to
spatial correlation functions—that are more often used to represent spatial correla-
tion. In this case, and in many kriging applications, variograms and spatial
correlation functions provide equivalent representations of spatial structure. For con-
sistency, only the term "variogram" is used here in discussions of spatial structure.
In the literature, Equation C-l is referred to as a universal kriging model. When
covariates (the X's) don't influence interpolation of Y, the right hand side of model
(Equation C-l) contains only the constant term/?0. The resulting model is called the
ordinary kriging model. When the spatial structure (variogram) for the model (Equa-
tion C-l) is known, statistically optimal predictions for the variable Y at unsampled
locations (outside of estimation of possible regression effects) can be derived using
standard statistical principles. The optimality criteria result in spatial predictions that
are linear in the data, statistically unbiased, and minimize mean squared prediction
error—known as best linear unbiased predictions. The minimized mean squared
prediction error is also a measure of prediction uncertainty. In practice, however, the
spatial structure of the data remains unknown. The estimation of the spatial structure
using the variogram function, therefore, is critical to kriging applications.
To demonstrate let {Xsi)> • • • Xsn)l represent a sample set of spatial data such as
dissolved oxygen collected at a set of n spatial locations s1;. . .sn. Assume this data
set to be a realization of the ordinary kriging version of model. The primary step in
kriging is variogram estimation with several methods available; the method of
moments and statistical likelihood based are two of the more common. All of these
methods are based on the sample data {Xsi)> • • • Xsn)l- This process ends with a
chosen variogram function and its parameter estimation, describing the shape and
strength (rate of decay) of spatial correlation. A determination, also based on the
sampled data, is made of whether the spatial structure is isotropic or anisotropic. The
estimated variogram is then assumed known. Kriged interpolations and their inter-
polated uncertainty at numerous locations are computationally straightforward to
generate.
The following describes some of the benefits and potential limitations of kriging for
the Chesapeake Bay Program to use in criteria attainment assessment application
(with some comparisons to the IDW approach of spatial interpolation outlined in the
previous section). A primary benefit of kriging compared to IDW is that it is a statis-
tical technique. Statistics (including kriging) can make inferences from sampled data
even in the presence of uncertainty; the quantity and quality of the sample data are
appendix c • E<
-------
C-9
reflected in these inferences. Kriging, however, is a less-than-routine type of analysis
and requires statistical expertise to execute. The short description on variogram esti-
mation above merely introduces this involved and often complicated step.
Further issues regarding kriging and Chesapeake Bay Program applications are
listed below.
• Kriging is flexible; it is based on an estimate of the strength of spatial dependence
in the data (variogram). Kriging can consider direction-dependent weighted inter-
polations (anisotropy) and can include covariates (universal kriging) to influence
interpolations—either simple trends in easting and northing coordinates or water-
related measures such as salinity.
• A key feature of kriging is that a measure of uncertainty (called the kriged
prediction variance) is generated along with kriged interpolations. Research
has started to propagate this interpolation uncertainty through the CFD.
• Kriging can be applied in situations for which the data remain sparse (such as
the Chesapeake Bay Water Quality Monitoring Program's fixed station data) or
dense (such as the Chesapeake Bay Shallow-water Monitoring Program).
Kriged and IDW spatial interpolations may very well produce near identical
results for these two extreme scenarios. The kriging approach, however,
provides a statistical model, the uncertainty of which is influenced by the quan-
tity and quality of data. Interpolation uncertainty information is crucial for both
sparsely and densely sampled networks.
In comparison to IDW, kriging is more sophisticated, but requires greater expertise
in implementation. Kriging is available in commercial statistical software and also
in free open-source applications, such at the R Statistical Computing Environment.
Use of the technique requires geostatistical expertise programming skills for these
two software packages. Segment-by-segment variogram estimation and subsequent
procedures would require substantial expert supervision and decision-making.
Chesapeake Bay Program managers may very well view this as a limitation in using
kriging for certain Chesapeake Bay Program activities, such as criteria assessments,
applications that need automated spatial interpolations. Furthermore, for some
Chesapeake Bay Program applications, the decision on criteria attainment is clearly
not influenced to any substantial degree by the method of spatial interpolation
because the water quality conditions remain far out of attainment. One possible
strategy is using a mix of IDW and kriging in situations for which attainment was
grossly exceeded or clearly met (IDW) versus borderline cases (kriging). Table C-l
provides a comparison of the capabilities of assessments based on lumping data,
spatial interpolation based on IDW, and spatial interpolation based on kriging.
appendix c
-------
C-10
Table C-1. Comparison of the capabilities of methods for interpreting data for Chesapeake
Bay water quality criteria assessment.
Attributes
Provides Spatial Prediction
Provides Prediction
Uncertainty
Uncertainty for CFD
Deal with Anisotropy
Can include cruise track/
fly-over data
Feasibility of 3-dimensional
interpolations
Feasibility of mainstem-
tributary interpolations
Inclusion of covariates to
improve prediction
Predictions of non-linear
functions of predicted
attainment surfaces P(y>c)
Level of sophistication
Automation
Sample-based
Yes
No
No
No
No
No
No
No
No
Lowest
Yes
IDW
Yes
No
No
Possible, but
not routine
No
Yes
Yes
No
No
Low
Yes
Kriging
Yes
Yes
Yes
Yes
Yes
Possible, but
not routine
Possible
Yes
Yes
Very High
No
Source: STAC 2006.
Chesapeake Bay Program. 1989. Chesapeake Bay Monitoring Program Atlas - Volume 1:
Water Quality and Other Physiochemical Monitoring Programs. CBP/TRS 34/89 U.S. Envi-
ronmental Protection Agency Chesapeake Bay Program Office, Annapolis, MD.
Cressie, N. 1991. Statistics for Spatial Data. Wiley, New York, NY, 928 pp.
Diggle, P.J., J.A. Tawn, and R.A. Moyeed. 1998. Model Based Geostatistics (with Discus-
sion). Applied Statistics 47:299-350.
Diggle, PJ. and P.J. Ribeiro. 2006. Model-based Geostatistics. Springer, New York, NY, 230
pp.
Harding, L.W., Jr., E.G. Itsweire, and W.E. Esaias. 1992. Determination of phytoplankton
chlorophyll concentrations in the Chesapeake Bay with aircraft remote sensing. Remote
Sensing of the Environment 40: 79-100.
Harding, L.W., Jr., J.G. Kramer, and J. Phinney. 2004. Estuarine and Watershed Monitoring
Using Remote Sensing Technology Present Status and Future Trends: A Workshop Report,
7-8 January 2002, Annapolis, Maryland. Scientific and Technical Advisory Committee and
Maryland Sea Grant College. Maryland Sea Grant Publication UM-6-SG-TS-2004-03.
College Park, MD.
Kitanidis, PK. 1997. Introduction to Geostatistics: Applications in Hydrogeology. Cambridge
University Press, New York, NY, 271 pp.
Ouyang Y, J.E. Zhang, and L.T. Ou. 2006. Temporal and spatial distributions of sediment
total organic carbon in an estuary river. Journal of Environmental Quality 35:93-100.
Schabenberger, O and C.A. Gotway. 2005. Statistical Methods for Spatial Data Analysis.
Chapmann and Hall/CRC Press, FL, 512 pp.
appendix c
Evaluation of Options for Spatial Interpolation
-------
C-11
Scientific and Technical Advisory Committee (STAC). 2006. The Cumulative Frequency
Diagram Method for Determining Water Quality Attainment: Report of the Chesapeake Bay
Program STAC Panel to Review o Chesapeake Bay Analytical Tools STAC Publication 06-
003. 9 October 2006. Chesapeake Bay Program Scientific and Technical Advisory
Committee. Chesapeake Research Consortium, Edgewater, MD.
U.S. Environmental Protection Agency. 2003a. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and its Tidal Tributaries.
EPA 903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2003b. Technical Support Document for Chesapeake
Bay Designated Uses and Attainability. EPA 903-R-03-004. Region III Chesapeake Bay
Program Office Annapolis, MD.
U.S. Environmental Protection Agency. 2004. Chesapeake Bay Program Analytical Segmen-
tation Scheme: Revisions, Decisions and Rationales 1983-2003. EPA 903-R-04-008.
CBP/TRS 268/04. Region III Chesapeake Bay Program Office, Annapolis, MD.
U.S. Environmental Protection Agency. 2005. Chesapeake Bay Program Analytical Segmen-
tation Scheme: Revisions, Decisions and Rationales 1983-2003: 2005 Addendum. EPA
903-R-05-004. CBP/TRS 278-06. Region III Chesapeake Bay Program Office, Annapolis,
MD.
Wang, XJ. and R.M. Liu. 2005. Spatial analysis and eutrophication assessment for chloro-
phyll a in Taihu Lake. Environmental Monitoring and Assessment 101:167-174.
appendix c
-------
-------
D-1
appendix
User Guide and
Documentation for the
Chesapeake Bay Interpolator
INTRODUCTION
The Chesapeake Bay and Tidal Tributary Interpolator computes water quality
concentrations throughout the Chesapeake Bay and/or tributary rivers from water
quality measured at point locations. The purpose of the Interpolator is in compute
water quality concentrations at all locations in the 2-dimensional plane (top or
bottom depth) or throughout the 3-dimensional water volume. Results of the inter-
polation can then be compared over time to compute trends or individual
interpolations can be overlaid with other data to visualize possible cause and effect
relationships. One example is to compare water quality with living resource (fish,
shellfish, aquatic vegetation) distributions. Results of the Chesapeake Bay Interpo-
lator have been used since 1988 to determine trends in water quality for the
Chesapeake Bay Program (http://www.chesapeakebay.net/).
Version 4.2 of the VOL3D software includes new code to: 1) import data from
Microsoft ACCESS data tables; 2) draw improved graphics of tributary segments; 3)
draw colors using categories, as before, or to draw using a color ramp of 255 colors;
4) draw longitudinal sections which represent the centerline of the Bay or Tributary
River segments; 5) draw images of all Tributary Rivers in addition to the Bay; and,
6) compute composite images that represent the minimums or maximums over a
time series.
Another tool, DART, which must be run on the CIMS network at the Chesapeake
Bay Program Office, creates data sets for the Interpolator for any parameter in the
historical water quality data base. DART is a very powerful tool which can create
many data sets in a very short time. Anyone who needs to interpolate data held by
the Bay Program, should investigate the use of DART.
appendix d • User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-2
M.:&I
The Chesapeake Bay Interpolator is a cell-based interpolator. Fixed cell locations are
computed by interpolating the nearest n neighboring water quality measurements,
where n is normally 4, but this number is adjustable. Cell size in Chesapeake Bay
was chosen to be 1km (east-west) x 1km (north-south) x 1m (vertical), with columns
of cells extending from surface to the bottom of the water column, thus representing
the 3-dimensional volume as a group of equal sized cells extending throughout the
volume. The tributaries are represented by various sized cells depending on the
geometry of the tributary, since the narrow upstream portions of the rivers require
smaller cells to accurately model the river's dimensions. This configuration results
in a total of 51,839 cells by depth for the Main Bay (Segments CB1TF-CB8PH), and
a total of 238,669 cells by depth for all 77 segments which comprise the Main Bay
and tributaries. Computation time on a Pentium 2 ghz PC running Windows XP is
approximately 15 seconds for the Bay and tributary interpolator model.
The Chesapeake Bay Interpolator is unique in the way it computes values in 3-
dimensions. The interpolator code is optimized to compute concentration values that
closely reflect the physics of stratified water bodies, such as Chesapeake Bay. The
Bay is very shallow compared to its width or length, hence water quality varies much
more vertically than horizontally. The Chesapeake Bay Interpolator uses a vertical
filter to select the vertical range of data that are used in each calculation. For
instance, to compute a model cell value at 5m deep, monitoring data at 5m deep are
preferred. If fewer than n (4) monitoring data values are found at the preferred depth,
the depth window is widened to search up to d (normally +/-2m) meters above and
below the preferred depth, with the window being widened in 0.5m increments until
n monitoring values have been found for the computation. The smallest acceptable n
value is selectable by the user. If fewer than n values are located, a missing value
(normally a -9) is calculated for that cell.
A second search radius filter is implemented to limit the horizontal distance of moni-
toring data from the cell being computed. Data points outside the radius selected by
the user (normally 25,000m) are excluded from calculation. This filter is included so
that only data that are near the location being interpolated are used.
In this version of the Interpolator, Segment and Region filters have been added.
Segments are geographic limits for the interpolator model. For instance, the Main
Bay is composed of 8 segments (CB1TF, CB2OH, ...,CB8PH). The tributaries are
composed of 69 additional segments, using the CBP 1998 segmentation scheme
(Figure D-l). These segments divide the Bay into geographic areas that have some-
what homogeneous environmental conditions. This segmentation also provides a
means for reporting results on a segment basis that can show more localized changes
compared to the whole Bay ecosystem. To replicate the segmentation scheme, the
segment boundaries were used to cookie-cutter out the Interpolator cells that fall
within each segment. Each set of these cells are then identified inside the corre-
sponding *.bth file that contains the bathymetry definitions. To compute the
interpolated values for the Main Bay, the corresponding bathymetry file is named
"cbayS.bth". This file contains the cell locations for the cells in the Main Bay Inter-
polator. A similar file, "bay_trib.bth" contains the cell definitions for the Main Bay
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-3
Up RTF
CBITF-.-
9SHOH
D-1. Chesapeake Bay Program 1998 segmentation design.
and tributary interpolator. Other .bth files have also been created for individual river
systems. Users that need specialized processing, such as finer resolution or addi-
tional segments in a particular area of interest, must create a new bathymetry file that
defines the bathymetry of the area of interest at the desire cell-size.
Regions filters (cbayS.drg, bay_trib.drg, etc) are files which contain a closed
polygon of x-y points that define an area larger than the corresponding *.bth file. The
region file identifies the geographic boundary that limits which monitoring station
data are included in interpolation for a given segment. The purpose of the data region
is to select a subset of the monitoring data from the input data file, and to then use
appendix d « User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-4
CHESAPEAKE BAY PROGRAM INTERPOLATOR
Data File Creation Tool
Figure D-2. Screen 1 includes seven navigation buttons and the Data
File Creation Tool for importing data from an ACCESS data base and
creating VOL3D data files.Once the user has selected the desired data
fields in the data table (Figure D-3), the Data File Creation Tool opens a
new screen that provides a range of options to the user for selecting
and subsetting data from the ACCESS data table (Figure D-4). The Data
Engine allows the user to select data by parameter, by date range, to
set interpolation control parameters, to choose the desired bathymetry,
to select data by depth ranges or layers, and finally to choose how the
group the resulting data in one or more output files. The "Create Files"
button, when pushed, will generate data files in the VOL3D ".d3d" file
format. These files are then ready for interpolation.
that subset for computing the values for each cell in a segment. Use of data regions
ensures that the interpolator does not "reach across land" to obtain data from an adja-
cent river which would give erroneous results. By using data regions, each segment
of cells can be computed from their individual subset of monitoring data. Each adja-
cent data region should overlap by some amount so that there is a continuous
gradient, and not a seam, across segment boundaries.
In the future, a pycnocline filter may be added to the Interpolator, so that water
above, within, and below the pycnocline are not interpolated together. Since the
water quality in various parts of the pycnocline can be so dramatically different, the
Interpolator file structure will be modified to handle this requirement.
INSTALLATION
The VolSD Interpolator code and auxiliary files have been bundled together into a
SETUP application and then PKZipped to reduce the overall file size. The Vol3D.zip
file must first be unzipped into a directory on any standard PC running the Windows
appendix d
-------
D-5
95/98/XP operating system. Once unzipped, double click the SETUP.EXE file to
start the installation process. It is suggested that the application be installed in the
C:\VOL3D directory. The original zipped file can be deleted to regain disk space. A
fast Pentium machine with 256 mb ram and 1 gb disk drive will prove useful.
THE BAY
FOR THE
Begin using the VOL3D software by double clicking the VOL3D.EXE icon on the PC.
The first screen provides 7 buttons (Geography, Parameter, Data Import, Interpolate,
Math, Graphics, and Reports) that step the user through the interpolation, graphics,
and reporting process. Also on the first screen, is a Data File Creation Tool, that can
be used to create VOL3D compatible data files from an external ACCESS data base.
The ACCESS data base needs to contain data necessary for interpolation, as identi-
fied on Figure D-2. Essential data fields include STATION, LONGITUDE,
LATITUDE, DEPTH, and VALUE. Other fields, including SAMPLE_DATE,
PARAMETER name, LAYER, Qualifier (<, >), CRUISE, and CRUISEJD, provide
data that can be used to select or subset the monitoring data by cruise, layer, or date.
CHESAPEAKE BAY PROGRAM INTERPOLATOR
|] Data Import | Inleipolale | Malh | Giaph.r,»
Data File Creation Tool
Figure D-3. Example of data fields selected from the "cbpwq99" data
table in the "data1999.mdb" ACCESS data base. The Data File
Creation Tool allows a user to extract data from an ACCESS table into
desired data files for VOL3D
appendix d « User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-6
Database Name: |C:Wo(3D403\data1999mdb
Table Name: [Cbpwq93
inter Descriptive Title
Chesapeake Bay Wale* Quality Analysis
Choose Parameter (cHLA.Chlorophyll
DO,DissoJved Oxygen
SALINITY.Salrtty
WTEMP.Water Temperature
NH4F Ammonia
NQ2F,Nitrte
N03F,Nitrate v
Choose Start Date J 1 /s/1 999
1 /6/1 999
1/7/1999
1/11/1999
1/12/1999 v
Transform
ff Linear r Log (Base 2)
Choose Geography
-------
D-7
CHESAPEAKE BAY PROGRAM INTERPOLATOR
Paiamelet Data Input Interpolate
•Choose Geographic Area for Analysis-
•0 Bap and Tiibs
lUillLOB'IAMMH.MUSI Pill
• Chmtui flwm
* Chop!** B««
* N«Micok« HIVE.
•it I'Dtomnke Hivei
• Magolhv Rivei
'H Patapsco Rival
Figure D-5. Geography screen. Select desired bathymetry.
CHESAPEAKE BAY PROGRAM INTERPOLATOR
Paiamelei I Data Inpoft I IntefpoMe
rChoose Geographic Area tor Analysts
Chcsapcjkc M.nnslcm
ICB1-C88I
Chesapeake Mains!em
(CD1 CU8. T ANMH^MOBPH I
_ Cheslei Rhmi
0 Choptank Rivei
& Bappahannecfc Rl
• N.jfilicoke liivci
• PalUKWll Rivei
• Hotomoke Hivt'i
• Pal.ipsco Rivei
• Back Rivei
Giaphio I Report!
Figure D-6. Parameter screen. Choose desired parameter for analysis.
appendix d « User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-8
Click the Data Import Button to select the Data Import screen. On this screen, click
the "Get File Name" button to select the data file that contains the data that are to be
interpolated (Figure D-7). The default file extension for data files is .d3d. d3d files
include the X and Y coordinates (UTM Zone 18, NAD83 is recommended for inter-
polation. These are also required for the graphics tools.)
Selected .Bata.File;
Look
My Recent
Documents
; _J VoSJD
i.^JSasBBl
J6DO97G716.D3O
J800970801.D3D
'JB00970816.030
'JBOO970901.030
Desktop
f% Doeumetits
My Computer
My ftetwortc
Places
filename:
Fifes of type:
Open
Data File f.d 3d)
Cancel
D-7. Data Import screen, "Get File Name" window. Select a data
file (.d3d) for analysis.
Once the file has been selected, the other fields on this screen will populate with
information about the data file, including, start and end dates of the data, the number
of observations, the date the file was created, the parameter name and code, and title
(Figure D-8). Normally data do not need to be transformed, however, some data such
as chlorophyll or TSS should be transformed with the log-transform to normalize the
data. The data are transformed as they are read into the interpolator and the results
are back-transformed to the original units in the output file. If the parameter is to be
transformed by the natural log transform, any data values that are negative or zero
will be set to a value of 0.0001. If the parameter is to be transformed by the square
root transform, any data values that are negative will be set to 0.0.
Two buttons at the bottom of the screen can be used to convert latitude and longitude
coordinates to UTM coordinates, which are recommended for interpolation (Figures
D-9, D-10). The first converts the longitude and latitude coordinates in d3d formatted
files to UTM coordinates, and vice versa. This is handy for checking data locations on
maps. The second converts individual longitudes and latitudes to and from UTM coor-
dinates. NAD27 to NAD83 conversion is not supported in this code. Improper use of
NAD27 or NAD83 can result in coordinate errors in the 100 to 300 meter range.
appendix d
-------
D-9
CHESAPEAKE BAY PROGRAM INTERPOLATOR
CHESAPEAKE SAY BY CRUISE BstolveiJ Oxygen • ln» IrterpoMed Data - 07JUU93M5IUU337
H
Figure D-8. Data Import
screen with fields popu-
lated with data from the
selected .d3d input file.
Longitude/Latitude and UTM
Converter for Interpolator Data Files
Choose an input file in ,
-------
D-10
lajnniON
Click the Interpolate Button to select the Interpolate screen (Figure D-ll). Select
the interpolator settings that match your requirements. The 3D Inverse-Distance
Squared model is the 3-dimensional interpolator model. The 2D Inverse-Distance
Squared model uses the same code as the 3D interpolator model except that only one
layer of cells are computed—cells for each depth below the surface cell are set to
missing (normally -9). The 2D Octant Search model computes values for cells in
only one layer, however, the data used for computing each cell value are selected
from data in each surrounding octant. For instance, for a given cell, the data used for
calculation would include 4 data points from each surrounding octant, or a total of
32 data points. The model will use fewer than the total data in each octant if insuffi-
cient data exist. The model uses as many data as are available for each octant, up to
the maximum requested number of data points. The octant search model is used to
reduce the bias from sampling schemes that collect continuous strings of data, such
as aircraft monitoring that collect many data points in well defined flight tracks. The
run-time for the octant search model is significantly longer due to the extensive
sorting required to select data from each data octant.
The "Trace Level" selects the amount of detail written to the ".LOG" file. A "Trace
Level" of "2" provides general interpolator statistics. A "Trace Level" of "3"
provides information about the data values used in the computations for each region.
A "Trace Level" of "4" provides information about individual cell computations. A
"Trace Level" of "5", "6", or "7" provides increasing information about data values,
distances, and octants. Increasing the "Trace Level" value is useful for investigating
the performance of the interpolator.
The "Convert .EST to .TXT" button will create a .txt file that can be imported into
Arc/Info or Arc View. The .txt files are a full matrix of values, 57 columns wide, with
all missing or non-existent cell values designated as missing values (normally -9),
comma delimited, and column headings and text strings are enclosed in quotes. Each
row in the .txt file represents numbers from 1 column of water from top to bottom,
1 cell wide by n cells deep. Additional columns are appended to the .txt file for
bottom, minimum, maximum, mean, and sum values.
The "Convert .EST to .T3D" button will create a .txt file that can be imported other
applications. The .t3d files are 4 columns wide, comma delimited, contain the x
value, y value, negative z value, and the estimated value. There are no column head-
ings. Missing values are included and are coded based on what was selected during
the interpolation.
The interpolator mode can be set to "Interactive" or "Batch". In interactive mode, the
chosen file is interpolated as defined in Figure D-ll. In "Batch" mode, a job file is
selected which provides the information needed to interpolate a series of files under
machine control (Figure D-12). The ".job" file can be built interactively by pressing
the "Save to Batch Job" button after selecting the run parameters for each desired file
(Figure D-13). The "Batch Job" can be executed by pushing the "Run Batch Job"
button.
appendix, d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-11
CHESAPEAKE BAY PROGRAM INTERPOLATOR
I Data Impoil I Mnuti/f I Math I Giaphict 1
Figure D-11. 3D Inverse-Distance Squared Interpolate screen populated
with entries after having made choices on previous screens. The 2D
Inverse-Distance Squared Interpolate screen uses the same format as
the 3D Inverse-Distance Squared model; however, only the surface
depth value has computed values. Cell values at depths below surface
are set to missing (generally -9). The 2D Octant Search Interpolator
does not rely only on the closest data in all directions, but rather uses
data from data from surrounding octants. For example, if 4 nearest
neighbors are requested in each of 8 octants surrounding the cell being
computed—up to 32 nearest neighbor values will be used to compute
the value. If no nearest neighbor values are available, a missing value
will be computed. Other buttons are available for creating data using
specific formats for various GIS (.txt files) and graphics applications
(.t3d files).
Choose Batch File Name
Look in: Q Vol3D
JLl
My Recent
Dcxxarcerts
Desktop
My Computer
1% Netwoik Fte name:
Places
Res of type:
|bdo-9?07-9709jofa
Batch Job He T job)
Figure D-12. Batch Job File Name selection window that displays after
choosing "Batch" radio button on Interpolation screen.
-------
D-12
CHESAPEAKE BAY PROGRAM INTERPOLATOR
I Data Impoil 1 Mw&rt- I Math I Graphic* 1
mmmmmmmmmmmmmm
Figure D-13. Saving or running a batch job through the Interpolate
screen. "Save to Batch Job" saves the values that have been entered in
the fields on this screen into the "Batch File ('filename'.job)". If the 'file-
name'.job file already exists, the new entry is appended to the existing
file. If it does not exist, a new file is created. The "Output File" file name
entry is also written to a file ('filename'.fls) for use in creating volume
and mass estimates by running batch jobs. The .fls file is simply a list of
interpolated (.est) file names that can be processed sequentially. The
"Test Batch Job" button executes a batch job but does not run the inter-
polator. This button can be used to test whether the needed files exist
and the batch job is sound prior to execution of the interpolator.
Click the Math Button if you need to conduct special operations on one or more
files. Four functions are provided: 1) Math operations which include adding,
subtracting, multiplying, or dividing one interpolated file by another, or by a
constant; 2) Receding values to new values; 3) Conducting a change analysis over
time; and, 4) Calculating the minimum or maximum values from a set of files.
Math functionality is provided so that special parameters can be calculated. Math is
conducted on a cell by cell basis. For instance, to add two interpolated files, Cell 1
of input file A is added to Cell 1 of input file B and the sum is stored in Cell 1 of
output file C, and so forth. Subtracting one file from another can be used to show
change from one time to another (Figure D-14). Missing values are handled as in
regular math—a non-missing value becomes missing if a math operation attempts to
compare a real value with a missing value. Division by zero or other illegal math
operation will cause the operation to stop.
The "Derive New Parameter" math operations can be performed sequentially to
provide additional capability. For instance, five interpolated files (.est files) could be
sequentially added together, then the resulting file could be divided by 5 to compute
appendix d * User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-13
r _^
[";. r';; •<- -x'
CHESAPEAKE BAY PROGRAM INTERPOLATOR
D-14. Math screen with files chosen to "Derive New Parameter"
of Dissolved Oxygen by subtracting File 2 from File 1 to create the
output file.
the mean for the five files. Another example would be to subtract interpolated
dissolved oxygen data from an interpolated saturated dissolved oxygen file to
compute the oxygen deficit.
The code checks whether the input files have the same number of segments. If the
input files ("Input .est File 1" and "Input .est File 2") do not have the same number
of segments, they were generated from different bathymetry files, and the cell values
in the two files can not be properly combined. An error message will be displayed if
this condition occurs.
The "Recode Parameter Value" radio button provides the means to convert calculated
values to new values (Figure D-15). The input file is not changed, but a new output
file is created with new values in each cell which classify the data into new values
or categories. For example, to compute the interaction of dissolved oxygen and water
temperature:
1) Recode the dissolved oxygen .est file so that oxygen below 3 mg/1 is set to "1"
and oxygen above 3 is set to "0" (also set missing to -9).
2) Recode the water temperature .est file so that temperature below 25C is set to "0"
and temperature above 25C is set to "10" (also set missing to -9).
3) Derive a new parameter "WD" by adding the receded dissolved oxygen and water
temperature .est files. The result is a wd.est file where: "0"=acceptable oxygen and
temperature; "l"=unacceptable oxygen; "10"=unacceptable temperature; and "11"=
unacceptable oxygen and unacceptable temperature (missing cells will = -9). This
file can be graphed to show the distribution of these categories. The water column
appendix d * User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-14
CHESAPEAKE BAY PROGRAM INTERPOLATOR
I Data Import I Intrapolale I tfM I Gidphici I
Figure D-15. Math screen with files chosen to "Recede Parameter
Value" of Dissolved Oxygen by receding values from 0-3 to the new
value of "1", receding values of above 3 to the new value of "0" and
retaining missing values as -9, to create the new output file. Choosing
a Range File is provided to load an existing set of ranges, which can
then be modified on this screen for specialized analyses.
volume of these categories can also be computed to show critical ranges for habitat
analysis.
Missing values are handled in a special way. Since missing values have no real value,
they are not used in math operations. If a cell in either "Input File 1" or "Input File
2" are flagged as missing (normally -9), then no math is done, and the "Output File"
value for that cell is set to missing (The "Missing Value" is set on the Interpolate
screen).
Interpolated values can be analyzed for trends. The "Change Over Time" button
allows the user to create a 3-dimensional (.est) file with linear percentage changes
over time for each cell in the bathymetry (Figures D-16 and D-17).
As a simple example, a station may be sampled several times over a period of time.
The measured values can be plotted with time on the x-axis and value on the y-axis.
The resulting linear regression line can be plotted through these points and the slope
and intercept can be used to compute the percentage increase or decrease between
the beginning and end of the time series.
This same technique can be used with the interpolator. Each cell value from a series
of .est files can be used to compute a linear regression, so that each cell has its own
regression and resulting percentage change, either up or down, over time. By coding
appendix d « User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-15
CHESAPEAKE BAY PROGRAM INTERPOLATOR
1 Pai.imeta I Data Import m Intwpolale I ,t-'.t'/> I Graphic! 1 Reports
Figure D-16. Math screen with files
chosen to "Change Over Time". The
bdo-9707-9709.fls file contains file
names of dissolved oxygen .EST files.
The corresponding Julian dates for
these .EST files are read from the do.jul
file. In this example, a new .EST file -
Change.est - is created which contains
the linear trend for each cell over the
time interval of the bdo-9707-9709.fls
file. The Change.est cell values are per-
centage change over time, categorized
by the selection criteria identified in
the pc.rng file. In this example, "Ignore
missing values" has not been selected.
CHESAPEAKE BAY
"NJ* Water Quality Analysis
Total Phosphorus Percent Change Jul 2,1984 Dec 18,1997
PLANVIEW
-»* i •«
. , - . f/tvtfi -* ^ ~ ' -
-Wiai.t^ "* ^
TOPTH PROFILE
D-17. Plot of total phosphorus as a "Change Over Time". In this
example, a new .EST file - tp8497.est - was created which contains the linear
trend for each cell over the time interval (July 1984 through December 1997)
of the tp8497.fls file. The tp8497.est cell values are percentage change over
time, categorized by the selection criteria identified in the pc.rng file. The
percentage change categories for total phosphorus mass (kg) are: >10%
increase (red); 5 to 10% increase (pink); 0 to 5% increase (yellow); 0 to 5%
decrease (light blue); 5 to 10% decrease (dark blue); and greater than 10%
decrease (green). In this example, "Ignore missing values" was selected so
that a trend on any available data was calculated.
appendix d • User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-16
the result as the percent change, a .est file can be created that has a percentage
change value for each cell. This .est file represents a 3D file of "Change Over Time".
The plot of this file provides a graphical representation of the change. The categories
used to display the changes graphically are defined in the pc.rng file. The default
pc.rng file provides categories of: >+10%, +5 to +10%, 0 to +5%, -5 to 0%,-10 to
-5 %, and >-10 % change. These categories should be modified to reflect the needs
of the analysis.
Missing values in the analysis can be treated in two ways: 1) included, meaning they
are propagated through the analysis; or, 2) ignored. The default is to include missing
values. The result of including missing values is that if one value for a specific cell
is missing anytime in the times series, then that cell is set to missing. The single
missing value forces the whole series of values at that cell to be missing and no
percentage change is calculated. The percentage change value is set to missing (-9
by default).
If missing values are set to be ignored, then each missing value in a time series for
a given cell is ignored and the rest of the time series observations are used to
compute the percentage change over time. The potential problem with this approach
is that the trend may be skewed by the lack of having all of the desired data.
At least two points are required to compute a time series change. If the number of
observations for any cell is less than 2, the resulting value for the percentage change
is set to missing.
The "Min-Max" button can be selected to locate the minimum or maximum values
in a series of interpolated values. For instance, this function could be used to read ten
interpolated files, and find for Cell 1 the minimum value and write that minimum
value to the output file Cell 1. This process would be repeated for each cell, so the
resulting output file would contain the minimum value for each cell in the series. The
Maximum function could be chosen if desired to find the maximum cell values in a
series of files. These functions are useful for determining, for example, the lowest
salinity over a 10-year period, or the highest temperature over a year period, for each
cell in the interpolated files. (Figures D-18 and D-19).
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-17
CHESAPEAKE BAY PROGRAM INTERPOLATOR
Geography • Parameter I Data Import I Intetpolo'e 1 «*» II Graphki 1 Reporti
Input Ffcs | (is) »C \VonO\Mo-9707-97Mll!
Temp File »C \Vol3ti\IeroFJeesl
Output We (est) lc VM»\DO M«nume:>
Figure D-18. Min-Max screen to
capture the minimum oxygen
values in each cell over the July-
September timeframe in 1997. The
bdo-9707-9709.fls file contains file
names of interpolated dissolved
oxygen .EST files. The Temp file is
an intermediate working file that
can be deleted after the job is com-
pleted. In this example, a new .EST
file - do-minimum.est - is created
which contains the minimum vale
for each cell over the time interval
of the bdo-9707-9709.fls file. In
this example, "Ignore missing
values" has been selected.
CHESAPEAKE BAY PROGRAM INTERPOLATOR
Paiametef i Data Inport I Intaipolate
Selcct Mmttmim
Numl.ei nl LdlwJs
:»§•;
'
Figure D-19. Math screen with files
chosen to "Change Over Time". The
do97.fls file contains file names of
dissolved oxygen .EST files. The
corresponding Julian dates for
these .EST files are read from the
do.97.jul file. In this example, a
new .EST file - dope.est - is created
which contains the linear trend for
each cell over the time interval of
the do97.fls file. The dope.est cell
values are percentage change over
time, categorized by the selection
criteria identified
appendix d « User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-18
Click the Graphics Button to select the Graphics screen. In this version, the graph-
ical representation of the data is limited to a Plan view (looking down on the Bay and
tidal tributary rivers) and a Side view (looking at the vertical dimension of the Bay
and tidal tributaries from the West).
The Graphics screen provides a means to choose all of the variables need to create
the Bay/Trib graphic (Figure D-20). Most of the choices are driven by the files being
graphed, to help minimize typing in all of the required information. The graphic can
be printed or saved to a .BMP file (Figure D-21).
Figure D-20. Graphics screen
with default values for graphing
the selected ".est" file.
« INTERPOLATOR GRAPHICS
CHESAPEAKE BAY PROGRAM INTERPOLATOR
Figure D-21, Graphics screen with
titles imported from the selected
".est" file. Click the "Interpolated
File" button to load these titles to
this screen. The titles can be edited
directly on this screen.
CHESAPEAKE BAY PROGRAM INTERPOLATOR
appendix d • User Guide and Documentation for the Chesapeake Bay Interpolator
-------
The Batch checkbox can be selected to process a group of files (.est file names are
read from a *.fls file) using the choices selected as shown in Figure D-22. If Batch
is chosen, the program prompts the user to choose a *.fls file. Title 3 and Graphics
File will automatically change for each plot based on the information contained in
the .est file. Other titles and legends will display based on what is displayed when
the DRAW button is pushed. Each graphic will automatically be saved to the default
graphics file name (D-23).
D-19
« INTERPOLATOR GRAPHICS
CHESAPEAKE BAY PROGRAM INTERPOLATOR
Figure D-22. Graphics screen
with legend imported from the
selected ".rng" file. Click the
"Categories File" button to load
these range categories to this
screen. These category values and
colors can be edited directly on
the screen.
PLANVIEW
Svsqitet)3K®a
Legend:
• 0.0-0.2
•i 0.2-1.0
cu 1.0-3.0
a 3.0-5.0
Bi >50MG/l.
• Not Calculated
OCPTM PROFILE
CHESAPEAKE BAY
Water Quality Analysis
Dissolved Ojygen - Jul7, ISWT-Jul IS, 1997
Pt«*BMPtofiW«|
.,—-j
Figure D-23. Example graphic
of interpolated Chesapeake Bay
mainstem dissolved oxygen. The
data are displayed so that the
worst case data (low dissolved
oxygen, in this case), regardless
of depth, are visible in both the
Plan- and Side views.
appendix d
-------
D-20
Click the Reports Button to generate files for volumetric and mass analysis (Figure
D-24).
The "Layer Thickness" is set to 0-50 meters deep to include all cells in the interpo-
lated file. This thickness could be set, for example, to 3-6 meters to calculate the
volume and mass for the water 3 to 6 meters deep (Figure D-25).
Figure D-24. The Reports screen is
used to compute volume and mass.
In Interactive mode, an interpolated
file (.est) is processed to create a
file of water volumes by parameter
range category by segment. A
file of parameter mass is also
computed by segment. If the
"Compute Mass by Concentration
Range" is checked on, then the
mass calculations are also
separated by the same category
ranges as the volume calculations.
CHESAPEAKE BAY PROGRAM INTERPOLATOR
Figure D-25. The Reports screen
with "Bottom Layer Thickness"
selected and set to the bottom
3 meters of water column depth.
Only the cells in the selected
bottom layer will be processed for
volume and mass calculations.
CHESAPEAKE BAY PROGRAM INTERPOLATOR
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-21
The resulting volume (.vol) and mass (.mas) files can be used for creating numerical
or graphical reports, such as trends plots (Figure D-26).
Each successive set of computed numbers are appended to the same specified
"Output File" (.vol and .mas).
Each successive set of computed numbers are appended to the same specified
"Output File" (.vol and .mas) (Figure D-21).
CHESAPEAKE BAY PROGRAM INTERPOLATOR
Figure D-26, The Reports screen
with "Batch" mode selected. In
Batch mode, a list of interpolated
file names (.est) are processed
sequentially to create a file of
water volumes by parameter
range category by segment. A file
of parameter mass is also com-
puted by segment. If the
"Compute Mass by Concentration
Range" is checked on, then the
mass calculations are also sepa-
rated by the same category
ranges as the volume calculations.
This example calculates volume
and mass for the top
CHESAPEAKE BAY PROGRAM INTERPOLATOR
Figure D-27. The Reports screen
is used to compute volume and
mass. In Interactive mode, an
interpolated file (.est) is processed
to create a file of water volumes
by parameter range category by
segment. A file of parameter mass
is also computed by segment. If
the "Limit Report to Selected
Segments" is checked on, then
the volume and mass calculations
are computed only for the seg-
ments identified in the "file-
name".1st file.
appendix d « User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-22
Figure D-28 illustrates a time series plot of the mass of total phosphorus computed
by the procedure described in. The mass of total phosphorus was computed for
Chesapeake Bay and tidal tributary rivers using monthly mean data for each station
at each depth for the period of record. The "Date" and "Total" (sum of all columns
in mass file) columns were used from the mass file (.mas) to make this plot. The
linear trend line is superimposed to show the general rate of decline. This plot was
created by opening the tp8497.mas file in Excel, selecting the line chart button,
selecting the "Start_Date" and "Total" columns, and adjusting the legends and titles
as necessary to create the time series plot. The linear trend was added by selecting
the time series followed by "Chart:Add Linear Trendline".
Mass of Total Phosphorus in Chesapeake Bay and
Tidal Tributaries Computed by Interpolation
6000000
o
in 1000000
U)
re
Date
D-28, TTime series plot of the mass of total phosphorus.
I-.' U U:
Monitoring data are required for the Interpolator to compute values. The file should
contain one value per depth per station for which data exist. If replicate values were
measured at some or all stations, they should be averaged at each station depth so
that only one value exists per depth per station. The overall data can represent one
cruise, a season of cruises, or a decade of data—there are no limitations on what the
data represent—that is up to the user to determine. It is best, statistically, to provide
as many data as possible. One method is to linearly interpolate values from surface
to bottom before creating the data file for the Interpolator. This will provide more
data for the Interpolator if it is valid to do so for the desired data. For the 2D inter-
appendix d
-------
D-23
polation models, only one value per station should be used, since the depth value is
ignored. The file naming convention for the input file is 'filename'.d3d. The input
file has the following structure:
Line 1> contains a title that is meaningful to the user that identifies the contents of this dataset.
Line 2> contains a 2-digit parameter code, comma, and the spelled-out parameter name
Line 3> contains the start date, comma, and end date of the data
Line 4> contains the date and time the data were compiled
Line 5> contains the number of observations that follow
Lines 6+> contain the easting in UTM Zone 18 meters NAD83, comma, the northing in UTM Zone 18
meters NAD83, comma, the sample depth in meters, comma, the measured value of the parameter,
comma, and the station ID
CHESAPEAKE BAY AND TRIES - Dissolved Oxygen - Measured Data -06JUL199315JUL1993
DO.Dissolved Oxygen
07/06/1993,07/15/1993
08/11/1997:15:11
1128
407056,4377577, 0.5, 7.7000,CB1.1
407056,4377577, 1.0, 6.8000,CB1.1
407056,4377577, 2.0, 6.2000,CB1.1
407056,4377577, 3.0, 5.7000,CB1.1
407056,4377577,4.0, 5.5000,CB1.1
407056,4377577, 5.0, 5.2000,CB1.1
411793,4365898,0.5, 5.9000,CB2.1
411793,4365898, 1.0, 5.7000,CB2.1
411793,4365898,2.0, 5.7000,CB2.1
411793,4365898,3.0, 5.7000,CB2.1
366939,4301041, 0.5, 7.5000,WT8.3
366939,4301041, 1.0, 7.3000, WT8.3
A metadata (documentation) file is created during the job. The default filename is
'filename'.met.
Check this file (using the Notepad editor) to see what calculations were performed
during the job.
Statistics Report for C:\Vol3D\BDO930701.est
Title: CHESAPEAKE BAY AND TRIES Dissolved Oxygen Measured Data
06JUL199315JUL1993
Parameter: Dissolved Oxygen Parameter Code: DO
Data Period: 07/06/199307/15/1993
Data File Date: 08/11/1997:15:11
Observations: 1128
Maximum Number of Nearest Neighbors: 4
Minimum Number of Nearest Neighbors: 1
appendix d » User Guide arid Documentation for the Chesapeake Bay Interpolator
-------
D-24
Maximum Vertical Search Window: 4
Minimum Vertical Search Window: 0
Vertical Search Window Step Size: .5
Maximum Horizontal Search Radius: 25000
Missing Value: 9
Interpolator Model: DepthRadiusInterpolator
Interpolation Date: 10/6/97 10:55:55 AM
Bathymetry File: bay_trib.bth
Number of Bathymetry Regions: 68
Data Region File: bay_trib.reg
Number of Data Regions: 68
- Bathymetry Region ID: 1001 Region Name: CB1TF Data Points: 28 in data region 1001
- Cell Size EW: 1000 NS: 1000 Vertical: 1
— 360 cells were interpolated in region CB1TF Subtotal: 360 total cells
— Region was calculated in 4 seconds.
- Bathymetry Region ID: 1002 Region Name: CB2OH Data Points: 48 in data region 1002
- Cell Size EW: 1000 NS: 1000 Vertical: 1
— 1237 cells were interpolated in region CB2OH Subtotal: 1597 total cells
Region was calculated in 1 seconds.
Total Number of Cells Interpolated: 173805
Total Number of NonMissing Value Cells Interpolated: 160558
Total Number of Missing Value Cells Interpolated: 13247
Nearest Neighbors: 4 # of Cells: 101978
Nearest Neighbors: 3 # of Cells: 12358
Nearest Neighbors: 2 # of Cells: 33517
Nearest Neighbors: 1 # of Cells: 12705
173805 cells were calculated in 244 seconds.
E^.fV:*r£S FILE
An interpolated estimates file is created during the job. The default filename is 'file-
name'.est. For the 3D interpolator model, this file contains the values for each cell
interpolated during the job, from surface to bottom for each cell location. For the 2D
interpolator models, this file contains the values for the top cell at each cell location.
While the interpolated value is written to the surface cell location in this file, its
value might represent the bottom value—i.e., the value might represent bottom layer
dissolved oxygen. All cell values below the top value will be set to missing (usually
-9). The file contents include:
Line l>Input data file name
Line 2>Data file description
Line 3> 2 digit parameter code and parameter name
Line 4>Start and end dates of data
Line 5>Date and time data file was compiled
Line 6>Number of data points, nearest neighbors, minimum neighbors, maximum vertical window,
minimum vertical window, vertical window step increase size, maximum search radius, missing value
Line 7>Name of interpolator used
Line 8>Date and time of job
Line 9>Bathymetry file used
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-25
Line 10>Data region file used
Line ll>Number of segments to interpolate
Line 12>Cell description for this segmentnumber of surface cells in segment, segment id, segment
name, cell e-w dimension in meters, cell n-s dimension in meters, cell vertical depth in meters
Line 13+>cell easting, cell northing, cells deep, interpolated values from surface to bottom.
C:\VOL3D\BDO970601.D3D
CHESAPEAKE BAY BY CRUISE - Dissolved Oxygen - Linear Interpolated Data -
3JUN199712JUN1997
DO,Dissolved Oxygen
06/03/1997,06/12/1997
06/10/1998:8:55
1254,4,1,4,0,.5,25000,9
Interpolator Model: DepthRadiusInterpolator
6/17/98 10:24:26 AM
cbayS.bth
cbayS.reg
8
132,1001,CB1TF,1000,1000,1
403000,4384000,2,9.1,8.9
404000,4384000,5,9.1,8.9,8.8,8.8,8.8
404000,4383000,3,9.1,8.9,8.8
405000,4383000,8,10.3,9.9,9.5,8.8,8.8,8.8,8.8,8.8
405000,4382000,3,10.3,9.9,9.5
406000,4382000,1,9.9
405000,4082000,1,10.5
410000,4082000,4,10.5,10.5,10.4,10.3
404000,4081000,1,10.6
410000,4081000,3,10.5,10.5,10.4
405000,4079000,1,10.6
";(f -._E I1-: • I •.'•'..-
Interpolated estimates files ('filename'.est) can be reformatted as 'filename'.txt files
which can be readily imported into other applications, including Arc/Info and
Arc View. The .TXT file contains the values for each cell in the original Estimates
file, from surface to bottom for each cell location. In addition, each line in the file is
padded with -9 values. So the file is a rectangular matrix of data with all values
having a value. The file is comma delimited, and all extraneous blanks have been
removed. The precision of the reported parameter values are assigned by the values
set in the 'parameter.sys' file. The .TXT file contents include:
Line l>Column Headings
Line 2+>cell easting, cell northing, segment name, cell e-w dimension in meters, cell n-s dimension in
meters, cell vertical depth in meters, bathymetry depth in meters, interpolated values from surface to
bottom, additional depths padded with -9 down to layer_45, then bottom, minimum, maximum, mean,
and sum values for non-missing cells in this water column.
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-26
UTM_X,UTM_Y,Segment,EW_dim,NS_dim,Vert_dim,n,layer_l,layer_2,layer_3,layer_4,layer_5,laye
r_6,layer_7,layer_8,layer_9,layer_10,layer_ll,layer_12,layer_13,layer_14,layer_15,layer_16,layer_17
,layer_18,layer_19,layer_20,layer_21,layer_22,layer_23,layer_24,layer_25,layer_26,layer_27,layer_2
8,layer_29,layer_30,layer_31,layer_32,layer_33,layer_34,layer_35,layer_36,layer_37,layer_38,layer_
39,layer_40,layer_41,layer_42,layer_43,layer_44,layer_45,Bottom,Minimum,Maximum,Mean,Sum
403000,4384000,CBlTF,1000,1000,l,2,-9.0,-9.0,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-
9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9.0,-9.0,-9.0,-9.0,-9.0
404000,4384000,CBlTF,1000,1000,l,5,-9.0,-9.0,-9.0,-9.0,-9.0,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-
9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9.0,-9.0,-9.0,-9.0,-9.0
383000,4338000,CB3MH,1000,1000,l,6,8.7,8.2,8.0,7.4,6.2,4.7,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-
9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,4.7,4.7,8.7,7.2,43.2
384000,4338000,CB3MH,1000,1000,l,5,8.7,8.2,8.0,7.4,6.2,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-
9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,6.2,6.2,8.7,7.7,38.5
385000,4338000,CB3MH,1000,1000,l,6,8.7,8.2,7.9,7.3,6.2,4.8,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-
9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,4.8,4.8,8.7,7.2,43.1
386000,4338000,CB3MH,1000,1000,l,6,8.7,8.1,7.9,7.3,6.2,4.9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-
9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,-9,4.9,4.9,8.7,7.2,43.1
' * .» l ' • FILE
Interpolated estimates files ('filename' .est) can be reformatted as 'filename' .t3d files
which can be readily imported into other applications, such as, NoeSys and T3D.
The .T3D file contains the values for each cell in the original Estimates file, with one
cell value per line in the output file. The file is comma delimited, and all extraneous
blanks have been removed. The precision of the reported parameter values are
assigned by the values set in the 'parameter.sys' file. The .T3D file contents include:
Line l+>cell centroid easting, cell centroid northing, negative cell centroid depth in meters, interpo-
lated value for parameter
403000,4384000,-0.5,-9.0
403000,4384000,-!.5,-9.0
404000,4384000,-Q.5,-9.0
404000,4384000,-!.5,-9.0
391500,4304000,-9.5,0.8
391500,4304000,40.5,0.2
391500,4304000,41.5,0.1
391500,4304000,42.5,0.1
i
Each interpolator job requires a bathymetry file which defines the cell structure of
the desired body of water that is being interpolated. The following shows the
contents of the cbayS.bth file:
Line l>Number of segments to interpolate
Line 2>Number of surface cells in segment 1, segment id, segment name, e-w cell size in meters, n-s
cell size in meters, cell depth in meters
appendix d * User Guide and Documentation for the Chesapeake Bay interpolator
-------
D-27
Line 3>Cell centroid easting in meters, cell centroid northing in meters, number of cells (>0) from
surface to bottom, cell centroid depths from surface to bottom. The Interpolator computes a value for
each cell centroid identified in the .bth file which is output to the .est file.
(Repeat 2 & 3 for each segment.)
132,1001,CB1TF,1000,1000,1
403000,4384000,2,0.5,1.5
404000,4384000,5,0.5,1.5,2.5,3.5,4.5
404000,4383000,3,0.5,1.5,2.5
405000,4383000,8,0.5,1.5,2.5,3.5,4.5,5.5,6.5,7.5
405000,4382000,3,0.5,1.5,2.5
410000,4363000,6,0.5,1.5,2.5,3.5,4.5,5.5
411000,4363000,10,0.5,1.5,2.5,3.5,4.5,5.5,6.5,7.5,8.5,9.5
412000,4363000,4,0.5,1.5,2.5,3.5
270,1002,CB 2OH, 1000,1000,1
403000,4363000,3,0.5,1.5,2.5
404000,4363000,4,0.5,1.5,2.5,3.5
405000,4363000,3,0.5,1.5,2.5
406000,4363000,3,0.5,1.5,2.5
400000,4362000,1,0.5
385000,4107000,1,0.5
381000,4106000,2,0.5,1.5
381000,4105000,2,0.5,1.5
381000,4104000,1,0.5
• • ;••
Each interpolator job requires a data regions file which defines the geographic
boundary of the data for the body of water that is being interpolated. 77 data regions
have been created, one for each CBP segment. The data region is used to clip off data
that fall outside the desired geographic area that is being interpolated. A data regions
file includes one or more data region definitions that must match the bathymetry
being interpolated. These data region file names are stored in a file, regions, sys,
which is required by the Interpolator. This file can contain 25 defined regions files.
The order of the entries in this file define the order presented to the user in the
GEOGRAPHY screen during the job. The structure of the regions.sys file is:
Line l+>Item identifier (sequential number of 1 to 25), comma, data region name, comma,
corresponding bathymetry file name, comma, corresponding data region file name.
Repeat for each defined data region.
l,Bay and Tribs,bay_trib.bth,bay_trib.reg
2,Chesapeake Mainstem (CBlCB8),cbay8.bth,cbay8.reg
Each .reg file defined in the regions.sys file must have the following structure. The following shows the
contents of the cbayS.reg file:
Line l>Bathymetry file name
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-28
Line 2>Data region file name
Line 3>Number of segments to interpolate
Line 4>First segment ID and name
Line 5> Data region ID
Line 6>Number of x-y points in this data region
Line 7+>Data region x-y points. First and last in each polygon must be the same to close the polygon.
Repeat for each data region.
cbayS.bth
cbayS.reg
8
1001,CB1TF
1001
8
398699,4385013
421073,4384159
413328,4355973
397839,4344869
383210,4352557
401281,4367077
398699,4385013
398699,4385013
1008,CB8PH
1008
13
410677,4131611
418631,4108780
422019,4095375
415538,4079614
408467,4086243
396978,4083150
384016,4087569
372968,4087569
372968,4095817
385194,4104508
374588,4117618
386372,4131464
410677,4131611
FILE
Each parameter is identified by a 2-digit parameter code and spelled out parameter
name. These codes and names are stored in the params.sys file. This file can accom-
modate 25 parameters. The order of the codes and names in this file determines the
order of the parameters in the PARAMETERS screen (Figure D-3). This file can be
edited as necessary by the user. The file structure is:
Line l>Item number (up to 25 lines), comma, spelled out parameter name, comma, 2-digit parameter
code, comma, number of digits precision to the right of the decimal in output file.
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-29
1,Dissolved Oxygen,DO,l
2,Chlorophyll,CH,l
3,Salinity,SA,l
4,Water Temperature, WT,1
5,Total Nitrogen,TN,2
6, Ammonia, NH,3
7,Nitrite,N2,3
8,Nitrate,N3,3
9,Total Phosphorus,TP,2
::M RAX : :'••-,)
Several files are required to compute report files of volume and mass and to graphi-
cally portray the interpolated results. These include:
1) the "filename".est file of estimated values;
2) the cbpotiny.bmp which is a small CBP logo file;
3) the aro.bmp which is a small north arrow;
4) shore_18.bnd which is a shoreline boundary file; and,
5) the "parameter_code".rng file. The range file is used by the volume and mass
report procedures to subset the computed results into categories for reporting. For
graphics, the .rng file defines how the graphics program assigns colors to each cell
value in the "filename".est file. For drawing purposes, the first range in the .rng table
has drawing priority over the second range, which has priority over the third range,
etc, so the first range color will paint over ranges lower in the table. This order deter-
mines which colors have priority in the final graphic. The do.rng file serves as an
example:
Line l>For dissolved oxygen values of 0.0 to but less than 0.2, color 12, pattern 0, title 0.0-0.2
Line 2>For dissolved oxygen values of 0.2 to but less than 1.0, color 13, pattern 2, title 0.2-1.0
Line 3>For dissolved oxygen values of 1.0 to but less than 3.0, color 14, pattern 12, title 1.0-3.0
Line 4>For dissolved oxygen values of 3.0 to but less than 5.0, color 11, pattern 12, title 3.0-5.0
Line 5>For dissolved oxygen values of 5.0 to but less than 25.0, color 19, pattern 21, title >5.0 MG/L
Line 6>For dissolved oxygen values of -10.0 to but less than -8.0 (-9=missing value), color 8, pattern
8, title Not Calculated
Pattern is currently ignored in this version.
To categorize integer value ranges, it is best to bracket the range, for instance, to
assign color 12 to the range of 2 (lower bound) to 2 (upper bound), set the lower
bound to 1.9 and the upper bound to 2.1. Set the title to "2" to convey the intent that
"2" is the range being presented. This bracketing is required to allow the code to
select values equal to or greater than the lower bound and less than the upper bound.
appendix d * User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-30
Acceptable color codes are:
0 Black
1 Blue
2 Green
3 Cyan
4 Red
5 Magenta
6 YeUow
7 White
8 Gray
9 Light Blue
10 Light Green
11 Light Cyan
12 Light Red
13 Light Magenta
14 Light Yellow
15 Bright White
0.0,0.2,12,0,0.00.2
0.2,1.0,13,2,0.21.0
1.0,3.0,14,12,1.03.0
3.0,5.0,11,12,3.05.0
5.0,25.0,9,21,>5.0 MG/L
10.0,8.0,8,8,Not Calculated
Individual interpolator runs can be computed sequentially by saving the necessary
information for the run in a "Batch File" ('filename'.job). This "Batch File" is then
used to calculate each of the files identified in the .job file. This file can be edited
as necessary by the user. The file structure is:
Line l>bathymetry file
Line 2>regions file
Line 3>input data file
Line 4> output interpolated (.est) file
Line 5> output metadata (.met) file
Line 6>parameter transformation
Line 7>minimum number of neighbors
Line 8>maximum number of neighbors
Line 9>horizontal range (m)
Line 10>vertical range minimum
Line ll>vertical range maximum
Line 12>Vertical step size
Line 13>missing value
Line 14>interactive/batch flag
Repeat lines 1-14 for each file to be interpolated
appendix d • User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-31
cbayS.bth
cbayS.reg
C:\Vol3D\BDO970601.D3D
C:\Vol3D\BDO970601.est
C:\Vol3D\BDO97060Lmet
None
1
4
25000
0
4
.5
-9
1
cbayS.bth
cbayS.reg
C:\Vol3D\BDO970701.D3D
C:\Vol3D\BDO970701.est
C:\Vol3D\BDO970701 .met
None
1
4
25000
0
4
.5
-9
1
Calculations on individual interpolator .est files can be computed sequentially by
reading the .est file names from a "batch file list" ('filename'.fls). This file is created
when the Batch Job File ('filename'.job) is created. This file can be edited as neces-
sary by the user. The file structure is:
Line 1 interpolator (.est) file name
Repeat for each file to be processed.
C:\Vol3D\BDO970601.est
C:\Vol3D\BDO970701.est
.'4, i ^ .• }• , i!)
Calculations of "Change Over Time" require a Julian date file which contains the
dates which relate to the .est files identified in the .fls file. For this analysis, the Julian
appendix d * User Guide and Documentation for the Chesapeake Bay interpolator
-------
D-32
dates are the X variable of the time series and the .est files are the Y variables of the
time series. The Julian dates represent dates and times that are based on the decimal
numbering system, rather than years, months, and days (and time). The Julian (or any
linearly numbered scheme) date file must be created by the user. The following
example was created by opening the appropriate .mas file in Excel, converting the
"Start_Date" from mm/dd/yy format to decimal format, and cutting and pasting the
reformatted date column into a flat file. The file structure is:
Line l>julian date
Repeat for each file to be processed.
30865.00
30895.00
30929.00
30956.00
LIST FILE {.1st}
Reports on individual segments can be computed sequentially by reading the .est file
names from a "segments list file" ('filename'.1st). This file is created manually by
the user. The file structure is:
Line l>Number of segments to process
Line 2+>Segment name (spelling must match those in Appendix A).
Repeat for each segment to be processed.
3
POCTF
POCOH
POCMH
FILE
The volume of water that contains a specified range of concentrations of a parameter
can be computed and saved to a volume report file ('filename'.vol). Volume esti-
appendix d * User Guide and Documentation for the Chesapeake Bay interpolator
-------
D-33
mates are reported in liters. The entire Bay and tributary volume based on the "Bay
and Tributary" 77 segment bathymetry is 75,199,817,500 mA3, or 75.2xlOA12 liters.
The volume of Main Bay segments CB1TF-CB8PH totals 51.839xlOA12 liters. Data
from one job to another may be appended to the same output file so that a time series
file is created that can be opened in a spreadsheet or database program for further
graphing or analysis. The first line of the file is composed of 'column headings'
contained within quotes. This file can be edited as necessary by the user. The file
structure is:
Line l>Column headings defined by the 'Report' job that is run, including data start date, data end
date, depth of top layer analyzed, depth of bottom layer analyzed, volume for segment, volume by
concentration range for that segment,...,..., repeat for each segment,... .grand total volume
Line 2>data accumulated from the input interpolated file (.est) for each column in line 1
Repeat for each interpolated file processed.
Note: Since data that are calculated may be appended to an existing file, there is a
risk that the user may append data from different bathymetry jobs. The user must be
careful not to mix 8 segment mainstem data with 77 segment mainstem and tributary
data in this report, or else the column headings will not represent the data.
"Start Date","End Date","Layer Top","Layer
Bottom" ,"CB1 TF","CB1 TF_0.0_0.2","CB1 TF_0.2_1.0","CB1 TF_1.0_3.0","CB1 TF_3.0_5.0","CB
1 TF_5.0_25.0","CB1 T F _ - 1 0 . 0 _ -
8.0","CB2OH","CB2OH_0.0_0.2","CB2OH_0.2_1.0","CB2OH_1.0_3.0","CB2OH_3.0_5.0","CB
2OH_5.0_25.0","CB2OH_-1 0 . 0 _ -
8.0","CB3MH","CB3MH_0.0_0.2","CB3MH_0.2_1.0","CB3MH_1.0_3.0","CB3MH_3.0_5.0","CB
3MH_5.0_25.0","CB3MH_-10.0_-
8.0","CB4MH","CB4MH_0.0_0.2","CB4MH_0.2_1.0","CB4MH_1.0_3.0","CB4MH_3.0_5.0","CB
4MH_5.0_25.0","CB4MH_-10.0_-
8.0","CB5MH","CB5MH_0.0_0.2","CB5MH_0.2_1.0","CB5MH_1.0_3.0","CB5MH_3.0_5.0","CB
5MH_5.0_25.0","CB5MH_-1 0 . 0 _ -
8.0","CB6PH","CB6PH_0.0_0.2","CB6PH_0.2_1.0","CB6PH_1.0_3.0","CB6PH_3.0_5.0","CB6
PH_5.0_25.0","CB6PH_-1 0 . 0 _ -
8.0","CB7PH","CB7PH_0.0_0.2","CB7PH_0.2_1.0","CB7PH_1.0_3.0","CB7PH_3.0_5.0","CB7
PH_5.0_25.0","CB7PH_-10.0_-
8.0","CB8PH","CB8PH_0.0_0.2","CB8PH_0.2_1.0","CB8PH_1.0_3.0","CB8PH_3.0_5.0","CB8
PH_5.0_25.0","CB8PH_-10.0_-8.0","Total"
"06/03/1997" ,"06/12/1997',0., 50., 359000000000. ,0,0,0,0, 359000000000. ,0., 12370000000
00., 0,0,0, 56000000000. ,1181000000000. ,0., 2391 000000000. ,0., 2000000000. ,366000000
000.,361000000000.,1662000000000.,0.,9237000000000.,0.,19000000000.,197400000000
0,1104000000000., 6140000000000., 0,15377000000000. ,0,0,0,851000000000.,1452600
0000000., 0., 6503000000000., 0., O.,0.,0., 6503000000000. ,0.,13488000000000. ,0.,0.,0.,0., 13
488000000000.,0.,3150000000000.,0.,0.,0.,0.,3150000000000.,0.,51742000000000.,0.,210
00000000.,2340000000000.,2372000000000.,47009000000000.,0.
"07/07/1997" ,"07/15/1997", 0., 50. ,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 2391000000000. ,13400
0000000.,51000000000.,150000000000.,168000000000.,1888000000000.,0.,92370000000
00.,1937000000000.,1253000000000.,933000000000.,620000000000.,4494000000000.,0.,
15395000000000.,1391000000000.,2236000000000.,1909000000000.,1069000000000.,87
90000000000.,0.,6503000000000.,0.,0.,524000000000.,1063000000000.,4916000000000.,
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-34
0., 13482000000000. ,0,0, 661000000000. ,2978000000000., 9843000000000. ,0., 247400000
0000., 0,0,0. ,114000000000., 2360000000000. ,0., 49482000000000. ,3462000000000. ,3540
000000000.,4177000000000.,6012000000000.,32291000000000.,0.
"07/1 6/1997" ,"07/31/1997",0, 50., 360000000000. ,0,0,0,0, 360000000000. ,0., 12370000000
00.,0.,0.,7000000000.,24000000000.,1206000000000.,0.,2391 000000000.,0.,13300000000
0.,159000000000.,282000000000.,1817000000000.,0.,9237000000000.,0.,1620000000000.
, 1644000000000. ,942000000000. ,5031000000000. ,0.,15388000000000. ,0., 1453000000000
.,2710000000000.,2005000000000.,9220000000000.,0.,6503000000000.,0.,0.,5500000000
0., 1231000000000. ,5217000000000. ,0., 13491000000000. ,0,0,9000000000. ,17930000000
00.,11689000000000., 0., 3160000000000. ,0,0,0,0, 3160000000000. ,0., 51767000000000.,
0.,3206000000000.,4584000000000.,6277000000000.,37700000000000.,0.
: ; " .-j
The mass file report contains the mass of a parameter computed for each cell in the
interpolated (.est) file then summed in one of two ways. The default method (below
example) is to sum the mass by segment and total for all segments in the bathymetry.
The second method follows the format of the volume report and computes the mass
by concentration range for each segment.
The mass that is computed and summed is saved to a mass report file ('file-
name'.mas). It is assumed the input data are measured in [units]/[liter], such as mg/1
or ug/1 or counts/liter. In the mass report, the resulting mass estimates are computed
by multiplying the [estimated concentration in the cell (often in mg/1)] * [the volume
of the cell in mA3 (for instance, 1000m east-west x 1000m north-south x 1m deep)]
* [1000 l/mA3 to convert from mA3 to liters]. Hence, if the input data were in mg/1
and then the concentration is estimated to be 6mg/l in a cell, the resulting mass will
be 6*10A9 mg for a 1km x 1km x 1 m cell. As a second example, if the input data
were in mg/mA3, which is equivalent to ug/1, then the reported mass values would be
in micrograms to account for the volume being reported in mA3 rather than liters. If
the input data are counts (such as organism counts) per liter, then the mass report
units would be total counts. If the input data are counts (such as organism counts)
per cubic meter, then the total counts in the mass report must be divided by 1000 to
account for the conversion from cubic meters to liters between the input data and the
interpolated counts. The mass (or counts) for each cell is then summed for a total
mass (or count) in the segment and also a grand sum of mass (or count) for the total
for all segments under analysis. For instance, if the input data for CHLA were meas-
ured as ug/1 and the resulting mass in Segment CB2OH was reported after
interpolation as 13,000,000,000,000, that represents 1.3A13 ug CHLA for Segment
CB2OH—i.e 1.3A13ug / 1.237M2 liters in CB2OH=10.5 ug/1 average. As a second
example, if the input data were for mg biomass of organisms per cubic meter and the
resulting mass in Segment CB2OH was reported as 132,627,709,873,200, that repre-
sents 1.326A14 / 1000 mg for Segment CB2OH, since an adjustment for the input
data must be made for the per cubic meter to per liter basis. A quick check can be
made by multiplying the average input data value by the volume of a segment to
determine if the results are within reason. For instance, if there were approximately
appendix d • User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-35
150 mg biomass per cubic meter in the monitoring data for CB2OH, that would be
[150 mg/mA3] * [1,237,000,000 cubic meters in Segment CB2OH] = 1.86A11 mg
biomass in the Segment CB2OH, which is close to the interpolated value of
1.326A11 mg, above.
Data from one job to another may be appended to the same output file so that a time
series file is created that can be opened in a spreadsheet or database program for
further graphing or analysis. The first line of the file is composed of 'column head-
ings' contained within quotes. This file can be edited as necessary by the user. The
file structure is:
Line l>Column headings defined by the 'Report' job that is run, including data start date, data end
date, depth of top layer analyzed, depth of bottom layer analyzed, mass of parameter by
segment,...,..., repeat for each segment,...,grand total mass
Line 2>data accumulated from the input interpolated file (.est) for each column in line 1
Repeat for each interpolated file processed.
Note: Since data that are calculated may be appended to an existing file, there is a
risk that the user may append data from different bathymetry jobs. The user must be
careful not to mix 8 segment mainstem data with 77 segment mainstem and tributary
data in this report, or else the column headings will not represent the data.
"Start Date","End Date","Layer Top","Layer
Bottom","CBlTF","CB2OH","CB3MH","CB4MH","CB5MH","CB6PH","CB7PH","CB8PH","Total"
"06/03/1997","06/12/1997",0.,50.,33956000.,99418000.,165733000.,602087001.,1229455000.,56352
3000.,! 121378003.,301820000.,4117370004.
"07/07/1997","07/15/1997",0.,50.,0.,0.,149745000.,378545000.,726926999.,429706000.,871456001.,
173612000.,2729991001.
"07/16/1997","07/31/1997",0.,50.,21746000.,82626000.,150816000.,438919000.,795246002.,428770
O00.,917109001.,243880000.,3079112003.
"08/04/1997","08/14/1997",0.,50.,22096000.,73961000.,128601000.,465053000.,824646000.,456154
001.,940162001.,236345001.,3147018002.
"08/18/1997","08/28/1997",0.,50.,19932000.,82306000.,158793000.,460818001.,885485001.,470413
001.,1010405001.,231651001.,3319803005.
"09/02/1997","09/15/1997",0.,50.,19748000.,76169000.,128181000.,517683000.,1009546002.,45963
2001.,949621002.,227962000.,3388542005.
"10/06/1997","10/15/1997",0.,50.,0.,0.,173127001.,576579000.,1004260004.,482547000.,963827000.
,200433996.,3400774001.
, ' , ' • , B
1) Double click the Vol3D.exe icon to run the Interpolator program.
2) Click the GEOGRAPHY button to display the GEOGRAPHY screen.
3) Choose "Chesapeake Mainstem (CB1-CB8)" to interpolate the Main Bay.
4) Click the PARAMETER button to display the PARAMETER screen.
appendix d • User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-36
5) Choose "DO- Dissolved Oxygen" as the parameter.
6) Click the DATA IMPORT button to display the DATA IMPORT screen.
7) Click Get File Name button and select the "C:\Vol3D\BDO970701.D3D" data
file. The file name, start and end dates, number of observations, file date, parameter,
code, and title should appear in the Data Import screen. If not, you selected an incor-
rect file.
8) Click the INTERPOLATE button to display the INTERPOLATE screen. The
input file should read "C:\Vol3D\BDO970701.D3D", the output file should read
"C:\Vol3D\BDO970701.est", Bathy file should read "cbayS.bth", and metadata file
should read "C:\Vol3D\BDO970701.met".
9) Click the "Run Interpolation" button to create the standard *.est file or click the
"Also Create TXT File" to create a "filename".txt file that can be imported into
Arc/Info or Arc View as a table. The text Arc View file will be approximately 2.3 mb
in size.
10) If you created an interpolated .est file, you can view the results by clicking the
GRAPHICS button to display the GRAPHICS screen. If you created a .txt file to
load into an Arc View table, you can quit the Interpolator program and continue
working with the output file in Arc View.
11) At the GRAPHICS screen, the Interpolated file should read
"C:\Vol3D\BDO970701.est". Click the Interpolated File button to load titles and
dates for the graphic.
12) The Bathymetry file should read "cbayS.bth", the Logo File 1 should read
".\cbpotiny.bmp", the Logo File 3 should read ".\aro.bmp", the Categories File
should read ".\DO.rng". Click the Categories File button to load the categories for
the graphic. The Boundary File should read ".\shore_18.bnd", the output Graphics
File name should read "C:\Vol3D\BDO970701.bmp". The titles and legends were
loaded by pushing the Interpolated File and Categories File buttons. The background
color of the boundary file is set by clicking the small grey box to the right of the
Boundary File name. Click "Plot Points" ON if you want to display the location of
the monitoring stations. Click "Plot Data Regions" ON if you wish to see the Data
Region polygons. Choose "Minimum" if you wish to display the minimum color
value (where minimum is the worst case, such as dissolved oxygen), or choose
"Maximum" if you wish to display the maximum color value (where maximum is
the worst case, such as temperature), or choose the "Top/West" edge or
"Bottom/East" edge to display the desired side.
Titles, categories, colors, and legends can be modified on this screen and will be
reflected in the resulting drawing. Clicking the "DRAW" button will draw the image
in a graphics window. The graphics window can be saved to a file or printed. This
version of the Interpolator does not allow graphical editing. The saved "file-
name" .bmp file can be edited in a commercial graphics editing package, such as
Lview Pro or Corel Draw. The .bmp file can be converted to gif or jpeg format for
publication on the web.
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-37
APPENDIX
Segment Name, EW-Dimension, NS-Dimension, Depth Dimension, Number Cells in Segment,
Segment Volume
CB1TF, 1000,1000,1,3 60,3 60000000
CB2OH, 1000,1000,1,1237,1237000000
CB3MH, 1000,1000,1,2391,2391000000
CB4MH, 1000,1000,1,9237,9237000000
CB5MH,1000,1000,1,15416,15416000000
CB6PH,1000,1000,1,6503,6503000000
CB7PH,1000,1000,1,13523,13523000000
CB8PH,1000,1000,1,3172,3172000000
NORTF,500,500,1,106,26500000
C&DOH,100,100,1,2413,24130000
ELKOH,500,500,1,405,101250000
BOHOH,250,250,1,272,17000000
SASOH,250,250,1,1347,84187500
CHSTF,50,50,1,1345,3362500
CHSOH,250,250,1,462,28875000
CHSMH,500,500,1,1821,455250000
EASMH,500,500,1,3987,996750000
CHOTF,50,50,1,6129,15322500
CHOOH,250,250,1,722,45125000
CHOMH2,500,500,1,1067,266750000
CHOMH1,1000,1000,1,945,945000000
LCHMH,500,500,1,833,208250000
HNGMH, 100,100,1,18568,185680000
FSBMH,1000,1000,1,143,143000000
NANTF,50,50,1,2646,6615000
NANOH,50,50,1,18000,45000000
NANMH,500,500,1,389,97250000
WICMH,100,100,1,5642,56420000
MANMH,500,500,1,358,89500000
BIGMH,250,250,1,698,43625000
POCTF,50,50,1,1788,4470000
POCOH,50,50,1,7200,18000000
POCMH,500,500,1,1418,354500000
T ANMH, 1000,1000,1,4019,4019000000
BSHOH,500,500,1,197,49250000
GUNOH,500,500,1,257,64250000
MIDOH,250,250,1,400,25000000
BACOH,250,250,1,358,22375000
PATMH,500,500,1,1806,451500000
MAGMH,250,250,1,1224,76500000
SEVMH,250,250,1,1815,113437500
SOUMH,250,250,1,1072,67000000
RHDMH,250,250,1,325,20312500
WSTMH,250,250,1,326,20375000
P AXTF,50,50,1,4410,11025000
PAXOH,100,100,1,2718,27180000
PAXMH,500,500,1,2244,561000000
PISTF,100,100,1,285,2850000
MATTF,250,250,1,152,9500000
POTTF,500,500,1,1939,484750000
POTOH,500,500,1,3409,852250000
POTMH,1000,1000,1,5792,5792000000
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolator
-------
D-38
RPPTF.250,250,1,1719,107437500
RPPOH,100,100,1,5358,53580000
RPPMH,500,500,1,5929,1482250000
CRRMH,250,250,1,1051,65687500
PIAMH,250,250,1,3223,201437500
MPNTF,50,50,1,6135,15337500
MPNOH, 100,100,1,3539,35390000
PMKTF,50,50,1,11452,28630000
PMKOH,100,100,1,6668,66680000
YRKMH,500,500,1,1102,275500000
YRKPH,500,500,1,1603,400750000
MOBPH,500,500,1,5370,1342500000
APPTF,100,100,1,151,1510000
CHKOH,250,250,1,777,48562500
JMSTF,250,250,1,4579,286187500
JMSOH,500,500,1,1726,431500000
JMSMH, 1000,1000,1,977,977000000
JMSPH, 1000,1000,1,434,434000000
WBEMH,100,100,1,631,6310000
SBEMH, 100,100,1,2773,27730000
EBEMH,50,50,1,2584,6460000
ELIMH,100,100,1,5339,53390000
LAFMH,100,100,1,339,3390000
ELIPH,500,500,1,246,61500000
LYNPH, 100,100,1,1673,16730000
Total Volume (mA3) = 75199817500
appendix d » User Guide and Documentation for the Chesapeake Bay Interpolate
-------
E-1
appendix
Potential Methods for Assessing
Shorter Duration Dissolved
Oxygen Criteria
POTENTIAL METHODS
The 2003 Chesapeake Bay water-quality criteria document described three alternatives
for assessing attainment of the short duration dissolved oxygen criteria (U.S. EPA
2003). Those include: 1) logistic regression; 2) a time series statistical method and 3)
continuous dissolved oxygen data collection using meters that are deployed for an
extended period of time. Each of these approaches has strengths and drawbacks.
Appropriate implementation of logistic regression or time series statistical methods
may require continuous dissolved oxygen data. To develop the full capacity to assess
the shorter duration dissolved oxygen criteria—7-day mean, 1-day mean and instanta-
neous minimum, EPA recommends a phased approach in which the methods that are
easiest to implement are employed initially while continuing to work on development
and implementation of the more detailed and/or expensive methods.
LOGISTIC REGRESSION
The instantaneous minimum criteria imply the requirement that waters within the
respective designated use be at or above the defined concentration everywhere all the
time. Stated in this way, the logistic regression approach clearly has application to
the challenge of assessing attainment of instantaneous minimum criteria. In the
context of criteria attainment, logistic equations are developed from the long term
dissolved oxygen data record, which predict the probability that the defined criteria
concentrations were met, based on observed monthly mean concentration.
The logistic regression approach utilizes a well-established statistical procedure
(U.S. EPA 2004) and has been employed in the past in Chesapeake Bay to estimate
instantaneous minima (Jordan et al. 1992). It is relatively simple to use and only
requires regular updating to keep the predictive models relevant to current condi-
tions. The limitation of this approach is that it is based on an extrapolation of the
fixed-station data and is likely to have higher error than the other methods.
appendix e • Potential Methods for Assessing Shorter Duration Dissolved Oxygen Criteria
-------
E-2
The logistic regression approach could be also be adapted to assess attainment of the
7-day and 1-day mean criteria components as well as other duration-specific criteria,
where and when a body of observational data is available at frequencies relevant to
the time frame. High frequency 'buoy' data sited at sentinel locations, where contin-
uous records extend over days, weeks and months, would offer opportunities to
develop logistic models of the relationship between exceedance/attainment and the
temporal means. EPA recommends that this method be actively developed for
possible employment for attainment assessments of the instantaneous minimum
dissolved oxygen criteria (see next section for details) while additional high
frequency data are collected and more complex, detailed methods described below
are being developed.
The time series approach utilizes a statistical procedure known as spectral analysis
to synthesize a complete record of dissolved oxygen concentrations at short interval
time steps over time. The synthetic record is developed using continuous measure-
ment data from nearby locations to develop a model that predicts the short-interval
variations in concentration. That model is combined with the long-term pattern of
variability derived from data collected routinely, monthly to twice monthly, at the
fixed-stations located in the assessment unit. The synthetic dissolved oxygen record
can then be used in the same way that data collected using a continuous meter would
be used. This time series approach has only been applied in a limited way to date and
further development is needed in order for it to fully meet the needs of a publishable
Chesapeake Bay dissolved oxygen criteria assessment methodology (see pages 183-
185 in U.S. EPA 2003). EPA recommends that this development work proceed
simultaneously with the development of the logistic regression and that the spectral
analysis method replace the logistic regression in the future should it prove a more
robust method.
The most rigorous approach for assessing attainment of the high frequency dissolved
oxygen criteria would be to collect continuous measures of dissolved oxygen
concentration at representative locations and depths throughout each spatial assess-
ment unit. The temporal and spatial density of such data would need to be sufficient
to enable all of the dissolved oxygen criteria to be assessed simply by calculating
means at the appropriate time scales (e.g. 30-day, 7-day, 1-day) or by observing
violations of the instantaneous minimum criteria values. However, continuous
collection of high frequency dissolved oxygen concentration in the Bay is expensive
both in purchasing the equipment and maintaining it. It is also difficult or impossible
to find sufficiently representative locations where the equipment can be affixed to
buoys or fixed pilings. Finally, it is expensive and labor-intensive to maintain the
equipment and sensor calibration once it is deployed due to the effects of weather,
turbulence, biological fouling and human interferences (e.g. accidents, thefts).
Nevertheless, the collection of at least some continuous dissolved oxygen data will
be critical for use in the other two statistical analysis-based assessment methods
appendix e • Potential Methods for Assessing Shorter Duration Dissolved Oxygen Criteria
-------
E-3
described above. Therefore, EPA recommends that the States continue to seek funds
to support this type of data collection in order to directly generate the data supporting
attainment assessment of the full array of applicable dissolved oxygen criteria.
OF TO
In the prior sections, it was noted that the data collection frequency of the long term,
fixed-station water quality monitoring program is inadequate to assess attainment of
short-duration criteria components. However, the greater than 20-year record of
dissolved oxygen measurements collected relatively synoptically throughout the
mainstem Bay, tidal tributaries and embayments, and collected regularly throughout
the annual cycle provides a very substantial data base from which to derive infer-
ences and define quantitative relationships between seasonal and monthly mean
dissolved oxygen concentrations and the frequency of observations above and below
specified criterion concentrations. Where relationships are strong, the logistic regres-
sion procedure produces models in the form of simple equations that
estimate/predict the likelihood that the criterion threshold concentration was attained
or violated during the period.
This method was explored originally to measure attainment of the 1992 Chesapeake
Bay dissolved oxygen restoration goal (Jordan et al. 1992) and was adapted for
assessing attainment of the 2003 Chesapeake Bay dissolved oxygen instantaneous
minimum (see Chapter 5, pages 27-62, in U.S. EPA 2004). The 2003 method modi-
fications included spatial and temporal refinements to the predictive models, with
consequent improvements to their goodness of fit. The early (1992) models esti-
mated exceedance based on segment-specific seasonal means and whether the means
were from depths above or below pycnocline. The 2003 method update was enriched
with an additional decade of monitoring data (1990-2000) for the regression analysis
and provided segment-specific models for individual months and depths. Recent
progress on this work again includes several additional years of new fixed-station
and continuous monitoring buoy data (2001-2005) and modifications to implemen-
tation procedures that could provide results for attainment assessment through the
CFD methodology in a format consistent with other dissolved oxygen criteria.
In this latest iteration, logistic regression models for the individual instantaneous
minima are developed for each station. The independent variables are, as before,
mean dissolved oxygen, month and water depth. The addition of a depth-squared
variable for deep stations is being tested, but not yet implemented. The dependent
variable is an indicator that the minimum threshold (e.g., the instantaneous criterion
concentration) is violated. (Since the CFD methodology is based on percent failure,
the dependent variable is based on exceedance rather than attainment.) This model-
building step currently uses the entire 1985-2005 water quality data record at each
station. Over time, however, if trends in ambient dissolved oxygen indicate signifi-
cant, sustained change in a segment, then the extent of the historical record to be
included in this step should be re-examined.
The collection of station models is used to estimate a predicted probability of
exceedance for each station, for each month in the 3-year, multi-month seasonal
assessment period, at each meter of depth. Then, for each month, the predicted prob-
appendix e « Potential Methods for Assessing Shorter Duration Dissolved Oxygen Criteria
-------
E-4
abilities are spatially interpolated to estimate probabilities for all interpolator cells
that represent the bathymetry of the Bay, its tidal tributaries and embayments. The
interpolator cells that are contained within the designated use where the criterion
applies are parsed out by segment and the probabilities calculated for each cell are
evaluated cell-by-cell against a threshold of probability which indicates an unac-
ceptably high risk that the dissolved oxygen criterion was exceeded (Jordan et al
1992). The volume of water represented by the interpolator cells exceeding the
threshold as a percentage of the total volume in the designated use is tallied for each
segment, for each month in the assessment period.
There are several elements of the logistic regression approach which should be eval-
uated as part of the attainment assessment procedure. Each of the station-specific
logistic models has its own goodness-of-fit measure. Each station will have a result
from the predictive model, i.e., the probability of exceeding the instantaneous
minimum over the assessment unit. Each segment will have an estimate of the
percent volume exceeding the criteria, based on spatial interpolation of the station
probabilities. As with other components of the dissolved oxygen criteria, these
results can also be assessed and visualized using the CFD methodology, although
this is not mandatory.
The limitations of this methodology have been noted earlier, particularly the
temporal frequency on which the models are based. In addition, the lack of good
spatial representation in the tidal tributaries and embayments is a concern. Most of
the fixed-stations are situated more or less longitudinally in mid-channel and there
is insufficient lateral coverage of the flanks, where different oxygen conditions and
different model relationships may exist. Data now being collected through the
Chesapeake Bay Shallow Monitoring Program will help answer where and to what
extent this is true.
Jordan, J., C. Stenger, M. Olson, R. Batiuk and K. Mountford. 1992. Chesapeake Bay
Dissolved Oxygen Goal for Restoration of Living Resource Habitats. CBP/TRS 88/93.
Chesapeake Bay Program, Annapolis, Maryland.
U.S. Environmental Protection Agency. 2003. Ambient Water Quality Criteria for Dissolved
Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and Its Tidal Tributaries.
EPA 903-R-03-002. Region III Chesapeake Bay Program Office, Annapolis, Maryland.
appendix e » Potential Methods for Assessing Shorter Duration Dissolved Oxygen Criteria
-------
F-1
appendix
Data Used in Deriving the
Open-Water, Deep-Water and
Deep-Channel Dissolved Oxygen
Criteria Summer Biological
Table F-1. Designated use, segment, year combinations found to be "good" using
the Benthic-IBI summer reference curve area locator method described in
Chapter 4.
CBP Segment
CB6PH
CB7PH
CB8PH
JMSPH
LCHMH
NANOH
RPPOH
CB6PH
CB8PH
CHOMH1
CHSMH
CHSMH
CHSMH
JMSOH
JMSPH
RPPOH
YRKMH
CB3MH
CB6PH
CB8PH
CHOMH1
CHSMH
JMSOH
JMSPH
NANMH
PMKTF
RPPMH
RPPMH
RPPOH
CB2OH
Year
1985
1985
1985
1985
1985
1985
1985
1986
1986
1986
1986
1986
1986
1986
1986
1986
1986
1987
1987
1987
1987
1987
1987
1987
1987
1987
1987
1987
1987
1988
Designated Use
DW
OW
OW
OW
OW
OW
OW
DW
OW
OW
OW
DW
DC
OW
OW
OW
OW
DW
DW
OW
OW
DW
OW
OW
OW
OW
OW
DW
OW
OW
CBP Segment
CB7PH
JMSMH
JMSPH
NANMH
PAXMH
PMKTF
RPPMH
YRKMH
CB2OH
CB3MH
CB8PH
JMSPH
POTMH
CB1TF
CB7PH
CB8PH
CHOOH
CHSMH
CHSOH
JMSPH
JMSTF
PAXOH
RPPMH
CB6PH
CB7PH
CB8PH
CHOMH2
JMSMH
JMSPH
JMSTF
Year
1988
1988
1988
1988
1988
1988
1988
1988
1989
1989
1989
1989
1989
1990
1990
1990
1990
1990
1990
1990
1990
1990
1990
1991
1991
1991
1991
1991
1991
1991
Designated Use
OW
OW
OW
OW
OW
OW
OW
OW
OW
DW
OW
OW
OW
OW
OW
OW
OW
DW
OW
OW
OW
OW
OW
DW
OW
OW
OW
OW
OW
OW
appendix f • Data Used in Deriving Summer Biological Reference Curves
-------
F-2
CBP Segment
PMKTF
POTMH
RPPMH
RPPMH
CB1TF
CB2OH
CB5MH
CB6PH
CB6PH
CB8PH
CHOTF
CHSMH
CHSMH
CHSOH
ELKOH
JMSPH
JMSTF
PMKTF
POTMH
POTTF
RPPMH
SASOH
CB3MH
CB6PH
CB6PH
CB7PH
CB8PH
CHOMH2
CHSMH
CHSMH
JMSPH
JMSTF
PMKTF
CB2OH
CB5MH
CB7PH
CB8PH
CHOMH2
CHSMH
HNGMH
JMSMH
JMSPH
LCHMH
PMKTF
BSHOH
CB1TF
CB3MH
CB6PH
CB6PH
Year
1991
1991
1991
1991
1992
1992
1992
1992
1992
1992
1992
1992
1992
1992
1992
1992
1992
1992
1992
1992
1992
1992
1993
1993
1993
1993
1993
1993
1993
1993
1993
1993
1993
1994
1994
1994
1994
1994
1994
1994
1994
1994
1994
1994
1995
1995
1995
1995
1995
Designated Use
OW
DW
OW
DW
OW
OW
OW
OW
DW
OW
OW
OW
DC
OW
OW
OW
OW
OW
DW
OW
OW
OW
DW
OW
DW
OW
OW
OW
DW
DC
OW
OW
OW
OW
DW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
DW
CBP Segment
CB8PH
JMSPH
MIDOH
NANMH
PAXTF
PMKTF
RPPMH
SASOH
SEVMH
SOUMH
TANMH
YRKPH
CB7PH
CB8PH
CHOOH
CHSMH
FSBMH
JMSPH
LCHMH
MIDOH
MPNOH
NANMH
PMKOH
RPPTF
SASOH
SEVMH
WICMH
WSTMH
BIGMH
CB3MH
CB6PH
CB8PH
CHOMH2
CHSOH
FSBMH
JMSTF
MANMH
MIDOH
MPNTF
NANMH
RHDMH
RPPTF
SOUMH
BIGMH
CB3MH
CB3MH
CB4MH
CB6PH
CB6PH
Year
1995
1995
1995
1995
1995
1995
1995
1995
1995
1995
1995
1995
1996
1996
1996
1996
1996
1996
1996
1996
1996
1996
1996
1996
1996
1996
1996
1996
1997
1997
1997
1997
1997
1997
1997
1997
1997
1997
1997
1997
1997
1997
1997
1998
1998
1998
1998
1998
1998
Designated Use
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
DW
DW
OW
OW
DC
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
DW
DW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
DW
DW
OW
DW
appendix f • Data Used in Deriving Summer Biological Reference Curves
-------
F-3
CBP Segment
CB8PH
CHOMH2
CHOOH
CHSMH
CHSMH
GUNOH
JMSPH
MPNOH
MPNTF
PAXTF
POCOH
POTTF
RPPTF
WICMH
CB3MH
CB4MH
CB6PH
CB7PH
CB7PH
CB8PH
CHSMH
CHSMH
JMSPH
JMSTF
LYNPH
POCMH
RHDMH
WICMH
WSTMH
BSHOH
CB2OH
CB7PH
CB8PH
CHKOH
CHSOH
EASMH
ELKOH
HNGMH
JMSPH
JMSTF
LAFMH
MIDOH
MPNTF
NANOH
PMKOH
PMKTF
POTOH
RPPTF
SEVMH
Year
1998
1998
1998
1998
1998
1998
1998
1998
1998
1998
1998
1998
1998
1998
1999
1999
1999
1999
1999
1999
1999
1999
1999
1999
1999
1999
1999
1999
1999
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
Designated Use
OW
ow
OW
ow
DW
OW
OW
ow
OW
OW
OW
OW
OW
OW
DW
DW
OW
OW
DW
OW
OW
DC
OW
OW
OW
OW
OW
OW
OW
OW
OW
DW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
CBP Segment
YRKPH
CB2OH
CB3MH
CB6PH
CHSMH
ELKOH
FSBMH
HNGMH
MANMH
MOBPH
PMKTF
RPPTF
SASOH
WICMH
CB2OH
CB5MH
CB7PH
CHKOH
CHOMH1
CRRMH
NANOH
PAXTF
PMKTF
RPPOH
RPPTF
YRKPH
BIGMH
CB2OH
CB6PH
CB8PH
CHSOH
JMSPH
MIDOH
MPNOH
POCMH
APPTF
BOHOH
CB1TF
CB2OH
CB6PH
CB8PH
CHKOH
CHOMH1
CHOTF
CHSMH
CHSOH
CRRMH
GUNOH
MANMH
Year
2000
2001
2001
2001
2001
2001
2001
2001
2001
2001
2001
2001
2001
2001
2002
2002
2002
2002
2002
2002
2002
2002
2002
2002
2002
2002
2003
2003
2003
2003
2003
2003
2003
2003
2003
2004
2004
2004
2004
2004
2004
2004
2004
2004
2004
2004
2004
2004
2004
Designated Use
OW
OW
DC
DW
DW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
DW
DW
OW
OW
OW
OW
OW
OW
OW
OW
DW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
OW
appendix f
-------
F-4
CBP Segment Year Designated Use
MPNTF 2004 OW
NORTF 2004 OW
RPPOH 2004 OW
CB1TF 2005 OW
CB7PH 2005 DW
CHOMH2 2005 OW
FSBMH 2005 OW
PMKOH 2005 OW
SASOH 2005 OW
TANMH 2005 OW
appendix f « Data Used in Deriving Summer Biological Reference Curves
-------
G-1
appendix
Equations for the Open-Water,
Deep-Water and Deep-Channel
Dissolved Oxygen Criteria
Summer Biological Reference
Curves
A biological reference curve of acceptable violation rates is generated using a cumu-
lative frequency distribution (CFD) of violation rates for "healthy" designated uses.
The violation rates are sorted in ascending order, ranked in descending order, and
graphed on a quantile plot:
• Violation rates are plotted on the x-axis, with plotting position on the y-axis.
• Plotting position represents the probability, i/n, of being less than or equal to a
given violation rate, or x, and is plotted on the y-axis as a function of rank, or
"i", and sample size, or "n".
• The x-axis is labeled "Percentage of Volume" because the violation rate repre-
sents the fraction of volume that is in violation.
• The y-axis is labeled as "Percentage of Time" because "probability" represents
the probable amount of time that a given violation rate will be observed.
• The Chesapeake Bay Program currently uses the Wiebull plotting position to
plot the cumulative distribution function. The Wiebull equation for calculating
probability, y, for each violation rate with rank "i" is: y = i/(n+l); i = rank.
In order to generate a graph of the CFD:
• X1; x2, x3,...xn= violation rates provided herein, sorted in ascending order,
with rank (i) assigned in descending order.
After plotting the data's violation rates and probabilities, two additional points
should be added to the distribution in order to complete the CFD curve:
Insert (x0, y0) = (0,1) before the first data point; and
Insert (xn+1, yn+1) = (1,0) after the last data point.
appendix g • Equations for the Summer Biological Reference Curves
-------
G-2
rank
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.1229698
0.1377778
0.1869919
0.192
0.1938775
0.2833333
0.3069767
0.3857374
0.5
0.6338462
0.7984496
1
1
1
Fraction
Time
1
0.975
0.95
0.925
0.9
0.875
0.85
0.825
0.8
0.775
0.75
0.725
0.7
0.675
0.65
0.625
0.6
0.575
0.55
0.525
0.5
0.475
0.45
0.425
0.4
0.375
0.35
0.325
0.3
0.275
0.25
0.225
0.2
0.175
0.15
0.125
0.1
0.075
0.05
0.025
0
appendix g » Equations for the Summer Biological Reference Curves
-------
G-3
rank
155
154
153
152
151
150
149
148
147
146
145
144
143
142
141
140
139
138
137
136
135
134
133
132
131
130
129
128
127
126
125
124
123
122
121
120
119
118
117
116
115
114
113
112
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
1
0.993589744
0.987179487
0.980769231
0.974358974
0.967948718
0.961538462
0.955128205
0.948717949
0.942307692
0.935897436
0.929487179
0.923076923
0.916666667
0.91025641
0.903846154
0.897435897
0.891025641
0.884615385
0.878205128
0.871794872
0.865384615
0.858974359
0.852564103
0.846153846
0.83974359
0.833333333
0.826923077
0.820512821
0.814102564
0.807692308
0.801282051
0.794871795
0.788461538
0.782051282
0.775641026
0.769230769
0.762820513
0.756410256
0.75
0.743589744
0.737179487
0.730769231
0.724358974
0.717948718
rank
111
110
109
108
107
106
105
104
103
102
101
100
99
98
97
96
95
94
93
92
91
90
89
88
87
86
85
84
83
82
81
80
79
78
77
76
75
74
73
72
71
70
69
68
67
Fraction
Volume
0
0
0
0
0
0.0011772
0.0027367
0.0053908
0.0058608
0.0071155
0.0082474
0.0086758
0.0105042
0.0119522
0.014231
0.0143416
0.015544
0.0186097
0.0186104
0.0186916
0.0229885
0.0242872
0.0290657
0.0303867
0.0341702
0.0372195
0.0394495
0.0442319
0.0468541
0.0492611
0.053407
0.0596184
0.0646766
0.0669035
0.0749625
0.0772947
0.0773381
0.0819209
0.0830704
0.0842912
0.0843786
0.0914286
0.0922064
0.096124
0.0967341
Fraction Time
0.711538462
0.705128205
0.698717949
0.692307692
0.685897436
0.679487179
0.673076923
0.666666667
0.66025641
0.653846154
0.647435897
0.641025641
0.634615385
0.628205128
0.621794872
0.615384615
0.608974359
0.602564103
0.596153846
0.58974359
0.583333333
0.576923077
0.570512821
0.564102564
0.557692308
0.551282051
0.544871795
0.538461538
0.532051282
0.525641026
0.519230769
0.512820513
0.506410256
0.5
0.493589744
0.487179487
0.480769231
0.474358974
0.467948718
0.461538462
0.455128205
0.448717949
0.442307692
0.435897436
0.429487179
appendix g
-------
G-4
rank
66
65
64
63
62
61
60
59
58
57
56
55
54
53
52
51
50
49
48
47
46
45
44
43
42
41
40
39
38
37
36
35
34
Fraction
Volume
0.0986842
0.1003289
0.1030177
0.1073883
0.1123967
0.1133005
0.1142857
0.1153846
0.1340996
0.1351351
0.1405229
0.1536643
0.1561065
0.1613475
0.1666667
0.1690574
0.177641
0.1888889
0.193999
0.2019704
0.2030651
0.2064298
0.2138837
0.2144487
0.2149758
0.2301587
0.2398477
0.2399356
0.2473721
0.2550629
0.2568941
0.2744511
0.2754491
Fraction Time
0.423076923
0.416666667
0.41025641
0.403846154
0.397435897
0.391025641
0.384615385
0.378205128
0.371794872
0.365384615
0.358974359
0.352564103
0.346153846
0.33974359
0.333333333
0.326923077
0.320512821
0.314102564
0.307692308
0.301282051
0.294871795
0.288461538
0.282051282
0.275641026
0.269230769
0.262820513
0.256410256
0.25
0.243589744
0.237179487
0.230769231
0.224358974
0.217948718
rank
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
Fraction
Volume
0.2863962
0.2887439
0.2992831
0.304324
0.3064989
0.3065134
0.3125
0.313253
0.3192771
0.3256059
0.3313559
0.3367199
0.3522608
0.3867069
0.4039409
0.4058394
0.4066776
0.4071428
0.4091904
0.4172932
0.4230019
0.4251208
0.4340449
0.4419155
0.4548346
0.4548849
0.4679803
0.5176327
0.5266618
0.5465729
0.5878661
1
1
1
Fraction Time
0.211538462
0.205128205
0.198717949
0.192307692
0.185897436
0.179487179
0.173076923
0.166666667
0.16025641
0.153846154
0.147435897
0.141025641
0.134615385
0.128205128
0.121794872
0.115384615
0.108974359
0.102564103
0.096153846
0.08974359
0.083333333
0.076923077
0.070512821
0.064102564
0.057692308
0.051282051
0.044871795
0.038461538
0.032051282
0.025641026
0.019230769
0.012820513
0.006410256
0
appendix g
-------
G-5
rank
868
867
866
865
864
863
862
861
860
859
858
857
856
855
854
853
852
851
850
849
848
847
846
845
844
843
842
841
840
839
838
837
836
835
834
833
832
831
830
829
828
827
826
825
824
823
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
1
0.998849252
0.997698504
0.996547756
0.995397008
0.99424626
0.993095512
0.991944764
0.990794016
0.989643268
0.98849252
0.987341772
0.986191024
0.985040276
0.983889528
0.98273878
0.981588032
0.980437284
0.979286536
0.978135788
0.97698504
0.975834292
0.974683544
0.973532796
0.972382048
0.9712313
0.970080552
0.968929804
0.967779056
0.966628308
0.96547756
0.964326812
0.963176064
0.962025316
0.960874568
0.95972382
0.958573072
0.957422325
0.956271577
0.955120829
0.953970081
0.952819333
0.951668585
0.950517837
0.949367089
0.948216341
0.947065593
rank
822
821
820
819
818
817
816
815
814
813
812
811
810
809
808
807
806
805
804
803
802
801
800
799
798
797
796
795
794
793
792
791
790
789
788
787
786
785
784
783
782
781
780
779
778
111
776
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
0.945914845
0.944764097
0.943613349
0.942462601
0.941311853
0.940161105
0.939010357
0.937859609
0.936708861
0.935558113
0.934407365
0.933256617
0.932105869
0.930955121
0.929804373
0.928653625
0.927502877
0.926352129
0.925201381
0.924050633
0.922899885
0.921749137
0.920598389
0.919447641
0.918296893
0.917146145
0.915995397
0.914844649
0.913693901
0.912543153
0.911392405
0.910241657
0.909090909
0.907940161
0.906789413
0.905638665
0.904487917
0.903337169
0.902186421
0.901035673
0.899884925
0.898734177
0.897583429
0.896432681
0.895281933
0.894131185
0.892980437
appendix g
-------
G-6
rank
775
774
773
772
771
770
769
768
767
766
765
764
763
762
761
760
759
758
757
756
755
754
753
752
751
750
749
748
747
746
745
744
743
742
741
740
739
738
737
736
735
734
733
732
731
730
729
728
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
0.891829689
0.890678941
0.889528193
0.888377445
0.887226697
0.886075949
0.884925201
0.883774453
0.882623705
0.881472957
0.880322209
0.879171461
0.878020713
0.876869965
0.875719217
0.87456847
0.873417722
0.872266974
0.871116226
0.869965478
0.86881473
0.867663982
0.866513234
0.865362486
0.864211738
0.86306099
0.861910242
0.860759494
0.859608746
0.858457998
0.85730725
0.856156502
0.855005754
0.853855006
0.852704258
0.85155351
0.850402762
0.849252014
0.848101266
0.846950518
0.84579977
0.844649022
0.843498274
0.842347526
0.841196778
0.84004603
0.838895282
0.837744534
rank
727
726
725
724
723
722
721
720
719
718
717
716
715
714
713
712
711
710
709
708
707
706
705
704
703
702
701
700
699
698
697
696
695
694
693
692
691
690
689
688
687
686
685
684
683
682
681
680
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
0.836593786
0.835443038
0.83429229
0.833141542
0.831990794
0.830840046
0.829689298
0.82853855
0.827387802
0.826237054
0.825086306
0.823935558
0.82278481
0.821634062
0.820483314
0.819332566
0.818181818
0.81703107
0.815880322
0.814729574
0.813578826
0.812428078
0.81127733
0.810126582
0.808975834
0.807825086
0.806674338
0.80552359
0.804372842
0.803222094
0.802071346
0.800920598
0.79976985
0.798619102
0.797468354
0.796317606
0.795166858
0.79401611
0.792865362
0.791714614
0.790563867
0.789413119
0.788262371
0.787111623
0.785960875
0.784810127
0.783659379
0.782508631
appendix g • Equations for the Summer Biological Reference Curves
-------
G-7
rank
679
678
677
676
675
674
673
672
671
670
669
668
667
666
665
664
663
662
661
660
659
658
657
656
655
654
653
652
651
650
649
648
647
646
645
644
643
642
641
640
639
638
637
636
635
634
633
632
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
0.781357883
0.780207135
0.779056387
0.777905639
0.776754891
0.775604143
0.774453395
0.773302647
0.772151899
0.771001151
0.769850403
0.768699655
0.767548907
0.766398159
0.765247411
0.764096663
0.762945915
0.761795167
0.760644419
0.759493671
0.758342923
0.757192175
0.756041427
0.754890679
0.753739931
0.752589183
0.751438435
0.750287687
0.749136939
0.747986191
0.746835443
0.745684695
0.744533947
0.743383199
0.742232451
0.741081703
0.739930955
0.738780207
0.737629459
0.736478711
0.735327963
0.734177215
0.733026467
0.731875719
0.730724971
0.729574223
0.728423475
0.727272727
rank
631
630
629
628
627
626
625
624
623
622
621
620
619
618
617
616
615
614
613
612
611
610
609
608
607
606
605
604
603
602
601
600
599
598
597
596
595
594
593
592
591
590
589
588
587
586
585
584
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
0.726121979
0.724971231
0.723820483
0.722669735
0.721518987
0.720368239
0.719217491
0.718066743
0.716915995
0.715765247
0.714614499
0.713463751
0.712313003
0.711162255
0.710011507
0.708860759
0.707710012
0.706559264
0.705408516
0.704257768
0.70310702
0.701956272
0.700805524
0.699654776
0.698504028
0.69735328
0.696202532
0.695051784
0.693901036
0.692750288
0.69159954
0.690448792
0.689298044
0.688147296
0.686996548
0.6858458
0.684695052
0.683544304
0.682393556
0.681242808
0.68009206
0.678941312
0.677790564
0.676639816
0.675489068
0.67433832
0.673187572
0.672036824
appendix g * Equations for the Summer Biological Reference Curves
-------
G-8
rank
583
582
581
580
579
578
577
576
575
574
573
572
571
570
569
568
567
566
565
564
563
562
561
560
559
558
557
556
555
554
553
552
551
550
549
548
547
546
545
544
543
542
541
540
539
538
537
536
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
0.670886076
0.669735328
0.66858458
0.667433832
0.666283084
0.665132336
0.663981588
0.66283084
0.661680092
0.660529344
0.659378596
0.658227848
0.6570771
0.655926352
0.654775604
0.653624856
0.652474108
0.65132336
0.650172612
0.649021864
0.647871116
0.646720368
0.64556962
0.644418872
0.643268124
0.642117376
0.640966628
0.63981588
0.638665132
0.637514384
0.636363636
0.635212888
0.63406214
0.632911392
0.631760644
0.630609896
0.629459148
0.6283084
0.627157652
0.626006904
0.624856157
0.623705409
0.622554661
0.621403913
0.620253165
0.619102417
0.617951669
0.616800921
rank
535
534
533
532
531
530
529
528
527
526
525
524
523
522
521
520
519
518
517
516
515
514
513
512
511
510
509
508
507
506
505
504
503
502
501
500
499
498
497
496
495
494
493
492
491
490
489
488
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
0.615650173
0.614499425
0.613348677
0.612197929
0.611047181
0.609896433
0.608745685
0.607594937
0.606444189
0.605293441
0.604142693
0.602991945
0.601841197
0.600690449
0.599539701
0.598388953
0.597238205
0.596087457
0.594936709
0.593785961
0.592635213
0.591484465
0.590333717
0.589182969
0.588032221
0.586881473
0.585730725
0.584579977
0.583429229
0.582278481
0.581127733
0.579976985
0.578826237
0.577675489
0.576524741
0.575373993
0.574223245
0.573072497
0.571921749
0.570771001
0.569620253
0.568469505
0.567318757
0.566168009
0.565017261
0.563866513
0.562715765
0.561565017
appendix g
-------
G-9
rank
487
486
485
484
483
482
481
480
479
478
477
476
475
474
473
472
471
470
469
468
467
466
465
464
463
462
461
460
459
458
457
456
455
454
453
452
451
450
449
448
447
446
445
444
443
442
441
440
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
0.560414269
0.559263521
0.558112773
0.556962025
0.555811277
0.554660529
0.553509781
0.552359033
0.551208285
0.550057537
0.548906789
0.547756041
0.546605293
0.545454545
0.544303797
0.543153049
0.542002301
0.540851554
0.539700806
0.538550058
0.53739931
0.536248562
0.535097814
0.533947066
0.532796318
0.53164557
0.530494822
0.529344074
0.528193326
0.527042578
0.52589183
0.524741082
0.523590334
0.522439586
0.521288838
0.52013809
0.518987342
0.517836594
0.516685846
0.515535098
0.51438435
0.513233602
0.512082854
0.510932106
0.509781358
0.50863061
0.507479862
0.506329114
rank
439
438
437
436
435
434
433
432
431
430
429
428
427
426
425
424
423
422
421
420
419
418
417
416
415
414
413
412
411
410
409
408
407
406
405
404
403
402
401
400
399
398
397
396
395
394
393
392
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fraction Time
0.505178366
0.504027618
0.50287687
0.501726122
0.500575374
0.499424626
0.498273878
0.49712313
0.495972382
0.494821634
0.493670886
0.492520138
0.49136939
0.490218642
0.489067894
0.487917146
0.486766398
0.48561565
0.484464902
0.483314154
0.482163406
0.481012658
0.47986191
0.478711162
0.477560414
0.476409666
0.475258918
0.47410817
0.472957422
0.471806674
0.470655926
0.469505178
0.46835443
0.467203682
0.466052934
0.464902186
0.463751438
0.46260069
0.461449942
0.460299194
0.459148446
0.457997699
0.456846951
0.455696203
0.454545455
0.453394707
0.452243959
0.451093211
appendix g * Equations for the Summer Biological Reference Curves
-------
G-10
rank
391
390
389
388
387
386
385
384
383
382
381
380
379
378
377
376
375
374
373
372
371
370
369
368
367
366
365
364
363
362
361
360
359
358
357
356
355
354
353
352
351
350
349
348
347
346
345
344
Fraction
Volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.0005089
0.0005304
0.0005965
0.0006126
0.000801
0.0009697
0.0011871
0.0014519
0.0016221
0.0016335
0.0021645
0.0021645
0.0021684
0.0022422
0.0024594
0.0024882
0.0025445
0.0029858
0.0035562
0.0042328
0.0044582
0.0048011
0.0049478
0.0050706
0.0055944
0.005598
Fraction Time
0.449942463
0.448791715
0.447640967
0.446490219
0.445339471
0.444188723
0.443037975
0.441887227
0.440736479
0.439585731
0.438434983
0.437284235
0.436133487
0.434982739
0.433831991
0.432681243
0.431530495
0.430379747
0.429228999
0.428078251
0.426927503
0.425776755
0.424626007
0.423475259
0.422324511
0.421173763
0.420023015
0.418872267
0.417721519
0.416570771
0.415420023
0.414269275
0.413118527
0.411967779
0.410817031
0.409666283
0.408515535
0.407364787
0.406214039
0.405063291
0.403912543
0.402761795
0.401611047
0.400460299
0.399309551
0.398158803
0.397008055
0.395857307
rank
343
342
341
340
339
338
337
336
335
334
333
332
331
330
329
328
327
326
325
324
323
322
321
320
319
318
317
316
315
314
313
312
311
310
309
308
307
306
305
304
303
302
301
300
299
298
297
296
Fraction
Volume
0.005618
0.0056497
0.0056497
0.0056497
0.0057741
0.0060024
0.0064935
0.0064935
0.0064935
0.0064935
0.006993
0.0072816
0.0077864
0.0081425
0.0083485
0.009176
0.009772
0.0100344
0.0108696
0.011236
0.0112994
0.0113636
0.0122877
0.0129206
0.0133457
0.0135135
0.0135135
0.0138099
0.0141643
0.0145985
0.0152119
0.0153473
0.0163934
0.0169367
0.0173661
0.0178716
0.0181209
0.0190476
0.0195954
0.0201126
0.020657
0.021121
0.0213471
0.0220096
0.0222782
0.0227457
0.0228091
0.0235199
Fraction Time
0.394706559
0.393555811
0.392405063
0.391254315
0.390103567
0.388952819
0.387802071
0.386651323
0.385500575
0.384349827
0.383199079
0.382048331
0.380897583
0.379746835
0.378596087
0.377445339
0.376294591
0.375143843
0.373993096
0.372842348
0.3716916
0.370540852
0.369390104
0.368239356
0.367088608
0.36593786
0.364787112
0.363636364
0.362485616
0.361334868
0.36018412
0.359033372
0.357882624
0.356731876
0.355581128
0.35443038
0.353279632
0.352128884
0.350978136
0.349827388
0.34867664
0.347525892
0.346375144
0.345224396
0.344073648
0.3429229
0.341772152
0.340621404
appendix g » Equations for the Summer Biological Reference Curves
-------
G-11
rank
295
294
293
292
291
290
289
288
287
286
285
284
283
282
281
280
279
278
277
276
275
274
273
272
271
270
269
268
267
266
265
264
263
262
261
260
259
258
257
256
255
254
253
252
251
250
249
248
Fraction
Volume
0.0238095
0.0238095
0.0238095
0.0238095
0.0238095
0.0238095
0.0242139
0.0243704
0.0248963
0.0251828
0.025533
0.0257611
0.025804
0.02595
0.0277429
0.0282486
0.0285016
0.0300546
0.0310473
0.0312355
0.0321381
0.0321381
0.0328082
0.0341186
0.0349206
0.0363636
0.0364963
0.0366795
0.0374777
0.0380952
0.0381803
0.0391608
0.0399501
0.0408163
0.0422037
0.042328
0.0423729
0.0427173
0.043956
0.0444942
0.0452794
0.0464169
0.0470397
0.0486772
0.0486772
0.0492197
0.0511788
0.0519084
Fraction Time
0.339470656
0.338319908
0.33716916
0.336018412
0.334867664
0.333716916
0.332566168
0.33141542
0.330264672
0.329113924
0.327963176
0.326812428
0.32566168
0.324510932
0.323360184
0.322209436
0.321058688
0.31990794
0.318757192
0.317606444
0.316455696
0.315304948
0.3141542
0.313003452
0.311852704
0.310701956
0.309551208
0.30840046
0.307249712
0.306098964
0.304948216
0.303797468
0.30264672
0.301495972
0.300345224
0.299194476
0.298043728
0.29689298
0.295742232
0.294591484
0.293440736
0.292289988
0.291139241
0.289988493
0.288837745
0.287686997
0.286536249
0.285385501
rank
247
246
245
244
243
242
241
240
239
238
237
236
235
234
233
232
231
230
229
228
227
226
225
224
223
222
221
220
219
218
217
216
215
214
213
212
211
210
209
208
207
206
205
204
203
202
201
200
Fraction
Volume
0.0521231
0.0529101
0.0529101
0.0530612
0.0536869
0.0536869
0.0536869
0.0541126
0.0541126
0.0541126
0.0541126
0.0541126
0.0579235
0.0586084
0.0587911
0.0596465
0.0604396
0.0614422
0.0618847
0.0620364
0.0625508
0.0653824
0.0658002
0.0684039
0.0695876
0.0708995
0.0713287
0.0714866
0.0717703
0.0739236
0.0756155
0.0757415
0.0760522
0.0779468
0.0781759
0.0784933
0.0793651
0.0813397
0.0836718
0.0852341
0.0852341
0.0852341
0.0858202
0.0859692
0.0862069
0.0867133
0.0889352
0.0924318
Fraction Time
0.284234753
0.283084005
0.281933257
0.280782509
0.279631761
0.278481013
0.277330265
0.276179517
0.275028769
0.273878021
0.272727273
0.271576525
0.270425777
0.269275029
0.268124281
0.266973533
0.265822785
0.264672037
0.263521289
0.262370541
0.261219793
0.260069045
0.258918297
0.257767549
0.256616801
0.255466053
0.254315305
0.253164557
0.252013809
0.250863061
0.249712313
0.248561565
0.247410817
0.246260069
0.245109321
0.243958573
0.242807825
0.241657077
0.240506329
0.239355581
0.238204833
0.237054085
0.235903337
0.234752589
0.233601841
0.232451093
0.231300345
0.230149597
appendix g
-------
G-12
Fraction
Volume
0.0926076
0.094402
0.0951807
0.0953661
0.0980392
0.0986222
0.0995439
0.1013514
0.1056534
0.1097062
0.1108631
0.1108631
0.1110075
0.1119293
0.1123596
0.1135513
0.113798
0.1141304
0.1160355
0.1222826
0.1235521
0.1259259
0.1260344
0.1269841
0.1270358
0.1300254
0.1310766
0.1316527
0.1342461
0.1372868
0.1389115
0.14
0.14
0.140647
0.1415645
0.1419069
0.1435523
0.1449735
0.1455978
0.1514339
0.1538896
0.1542142
0.1566434
0.1587452
0.1595922
0.1611479
0.1614429
0.162963
Fraction Time
0.228998849
0.227848101
0.226697353
0.225546605
0.224395857
0.223245109
0.222094361
0.220943613
0.219792865
0.218642117
0.217491369
0.216340621
0.215189873
0.214039125
0.212888377
0.211737629
0.210586881
0.209436133
0.208285386
0.207134638
0.20598389
0.204833142
0.203682394
0.202531646
0.201380898
0.20023015
0.199079402
0.197928654
0.196777906
0.195627158
0.19447641
0.193325662
0.192174914
0.191024166
0.189873418
0.18872267
0.187571922
0.186421174
0.185270426
0.184119678
0.18296893
0.181818182
0.180667434
0.179516686
0.178365938
0.17721519
0.176064442
0.174913694
rank
151
150
149
148
147
146
145
144
143
142
141
140
139
138
137
136
135
134
133
132
131
130
129
128
127
126
125
124
123
122
121
120
119
118
117
116
115
114
113
112
111
110
109
108
107
106
105
104
Fraction
Volume
0.1631854
0.1643766
0.1650579
0.1661578
0.1686848
0.1688742
0.1706892
0.1722272
0.1731602
0.1735369
0.1756757
0.1779041
0.1805116
0.1805379
0.1830601
0.190725
0.1914525
0.1941337
0.199403
0.201087
0.2013652
0.2039852
0.2189781
0.227972
0.2337085
0.2359882
0.2374406
0.2409669
0.2419833
0.2432432
0.2444856
0.2445605
0.2457132
0.2472826
0.2478753
0.2583187
0.2593284
0.2593284
0.2593284
0.2611517
0.2736842
0.2866706
0.2889755
0.2900763
0.2905174
0.3018642
0.3043062
0.3130724
Fraction Time
0.173762946
0.172612198
0.17146145
0.170310702
0.169159954
0.168009206
0.166858458
0.16570771
0.164556962
0.163406214
0.162255466
0.161104718
0.15995397
0.158803222
0.157652474
0.156501726
0.155350978
0.15420023
0.153049482
0.151898734
0.150747986
0.149597238
0.14844649
0.147295742
0.146144994
0.144994246
0.143843498
0.14269275
0.141542002
0.140391254
0.139240506
0.138089758
0.13693901
0.135788262
0.134637514
0.133486766
0.132336018
0.13118527
0.130034522
0.128883774
0.127733026
0.126582278
0.12543153
0.124280783
0.123130035
0.121979287
0.120828539
0.119677791
appendix g
-------
G-13
rank
103
102
101
100
99
98
97
96
95
94
93
92
91
90
89
88
87
86
85
84
83
82
81
80
79
78
77
76
75
74
73
72
71
70
69
68
67
66
65
64
63
62
61
60
59
58
57
56
Fraction
Volume
0.319855
0.3289125
0.3301158
0.3326345
0.3344764
0.350166
0.350417
0.3507276
0.3753799
0.3762215
0.37668
0.3773917
0.3793436
0.3807623
0.3879781
0.3998882
0.4010152
0.4038889
0.4038889
0.4038889
0.4038889
0.4038889
0.4146816
0.4163347
0.4233177
0.4269663
0.429676
0.4300699
0.4304933
0.4374046
0.4423898
0.4565826
0.456869
0.4570895
0.4570895
0.4570895
0.464702
0.4692463
0.4841679
0.5085324
0.5108851
0.5263544
0.545568
0.5625
0.5701425
0.575
0.576076
0.6036122
Fraction Time
0.118527043
0.117376295
0.116225547
0.115074799
0.113924051
0.112773303
0.111622555
0.110471807
0.109321059
0.108170311
0.107019563
0.105868815
0.104718067
0.103567319
0.102416571
0.101265823
0.100115075
0.098964327
0.097813579
0.096662831
0.095512083
0.094361335
0.093210587
0.092059839
0.090909091
0.089758343
0.088607595
0.087456847
0.086306099
0.085155351
0.084004603
0.082853855
0.081703107
0.080552359
0.079401611
0.078250863
0.077100115
0.075949367
0.074798619
0.073647871
0.072497123
0.071346375
0.070195627
0.069044879
0.067894131
0.066743383
0.065592635
0.064441887
rank
55
54
53
52
51
50
49
48
47
46
45
44
43
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
Fraction
Volume
0.6200294
0.6237785
0.6320347
0.6437941
0.6469252
0.6567556
0.663745
0.7578948
0.7717455
0.8227364
0.8384528
0.97
0.9949544
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Fraction Time
0.063291139
0.062140391
0.060989643
0.059838895
0.058688147
0.057537399
0.056386651
0.055235903
0.054085155
0.052934407
0.051783659
0.050632911
0.049482163
0.048331415
0.047180667
0.046029919
0.044879171
0.043728423
0.042577675
0.041426928
0.04027618
0.039125432
0.037974684
0.036823936
0.035673188
0.03452244
0.033371692
0.032220944
0.031070196
0.029919448
0.0287687
0.027617952
0.026467204
0.025316456
0.024165708
0.02301496
0.021864212
0.020713464
0.019562716
0.018411968
0.01726122
0.016110472
0.014959724
0.013808976
0.012658228
0.01150748
0.010356732
0.009205984
appendix g * Equations for the Summer Biological Reference Curves
-------
G-14
rank
7
6
5
4
3
2
1
Fraction
Volume
1
1
1
1
1
1
1
1
Fraction Time
0.008055236
0.006904488
0.00575374
0.004602992
0.003452244
0.002301496
0.001150748
0
appendix g
-------
H-1
appendix
hi
Equations for the
Water Clarity Criteria
Biological Reference Curves
A biological reference curve of acceptable violation rates is generated using a cumu-
lative frequency distribution (CFD) of violation rates for "healthy" designated uses.
The violation rates are sorted in ascending order, ranked in descending order, and
graphed on a quantile plot:
• Violation rates are plotted on the x-axis, with plotting position on the y axis.
• Plotting position represents the probability, i/n, of being less than or equal to a
given violation rate, or x, and is plotted on the y-axis as a function of rank, or
"i", and sample size, or "n".
• The x-axis is labeled "space" because the violation rate represents the fraction
of volume that is in violation.
• The y-axis is labeled as "time" because "probability" represents the probable
amount of time that a given violation rate will be observed.
• The Chesapeake Bay Program currently uses the Wiebull plotting position to
plot the cumulative distribution function. The Wiebull equation for calculating
probability, y, for each violation rate with rank "i" is: y = i/(n+l); i = rank.
In order to generate a graph of the CFD:
• Xj , x2, x3,...xn= violation rates provided herein, sorted in ascending order,
with rank (i) assigned in descending order.
After plotting the data's violation rates and probabilities, two additional points
should be added to the distribution in order to complete the CFD curve:
Insert (x0, y0) = (0,1) before the first data point; and
Insert (xn+1, yn+1) = (1,0) after the last data point.
appendix h • Equations for the Water Clarity Criteria Biological Reference Curves
-------
H-2
rank
406
405
404
403
402
401
400
399
398
397
396
395
394
393
392
391
390
389
388
387
386
385
384
383
382
381
380
379
378
377
376
375
374
373
372
371
370
369
368
367
366
365
364
363
362
361
360
359
volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
time
1
0.997542998
0.995085995
0.992628993
0.99017199
0.987714988
0.985257985
0.982800983
0.98034398
0.977886978
0.975429975
0.972972973
0.970515971
0.968058968
0.965601966
0.963144963
0.960687961
0.958230958
0.955773956
0.953316953
0.950859951
0.948402948
0.945945946
0.943488943
0.941031941
0.938574939
0.936117936
0.933660934
0.931203931
0.928746929
0.926289926
0.923832924
0.921375921
0.918918919
0.916461916
0.914004914
0.911547912
0.909090909
0.906633907
0.904176904
0.901719902
0.899262899
0.896805897
0.894348894
0.891891892
0.889434889
0.886977887
0.884520885
0.882063882
rank
358
357
356
355
354
353
352
351
350
349
348
347
346
345
344
343
342
341
340
339
338
337
336
335
334
333
332
331
330
329
328
327
326
325
324
323
322
321
320
319
318
317
316
315
314
313
312
311
310
309
308
volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
time
0.87960688
0.877149877
0.874692875
0.872235872
0.86977887
0.867321867
0.864864865
0.862407862
0.85995086
0.857493857
0.855036855
0.852579853
0.85012285
0.847665848
0.845208845
0.842751843
0.84029484
0.837837838
0.835380835
0.832923833
0.83046683
0.828009828
0.825552826
0.823095823
0.820638821
0.818181818
0.815724816
0.813267813
0.810810811
0.808353808
0.805896806
0.803439803
0.800982801
0.798525799
0.796068796
0.793611794
0.791154791
0.788697789
0.786240786
0.783783784
0.781326781
0.778869779
0.776412776
0.773955774
0.771498771
0.769041769
0.766584767
0.764127764
0.761670762
0.759213759
0.756756757
appendix h » Equations for the Water Clarity Criteria Biological Reference Curves
-------
H-3
rank
307
306
305
304
303
302
301
300
299
298
297
296
295
294
293
292
291
290
289
288
287
286
285
284
283
282
281
280
279
278
277
276
275
274
273
272
271
270
269
268
267
266
265
264
263
262
261
260
259
258
257
volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
time
0.754299754
0.751842752
0.749385749
0.746928747
0.744471744
0.742014742
0.73955774
0.737100737
0.734643735
0.732186732
0.72972973
0.727272727
0.724815725
0.722358722
0.71990172
0.717444717
0.714987715
0.712530713
0.71007371
0.707616708
0.705159705
0.702702703
0.7002457
0.697788698
0.695331695
0.692874693
0.69041769
0.687960688
0.685503686
0.683046683
0.680589681
0.678132678
0.675675676
0.673218673
0.670761671
0.668304668
0.665847666
0.663390663
0.660933661
0.658476658
0.656019656
0.653562654
0.651105651
0.648648649
0.646191646
0.643734644
0.641277641
0.638820639
0.636363636
0.633906634
0.631449631
rank
256
255
254
253
252
251
250
249
248
247
246
245
244
243
242
241
240
239
238
237
236
235
234
233
232
231
230
229
228
227
226
225
224
223
222
221
220
219
218
217
216
215
214
213
212
211
210
209
208
207
206
volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
time
0.628992629
0.626535627
0.624078624
0.621621622
0.619164619
0.616707617
0.614250614
0.611793612
0.609336609
0.606879607
0.604422604
0.601965602
0.5995086
0.597051597
0.594594595
0.592137592
0.58968059
0.587223587
0.584766585
0.582309582
0.57985258
0.577395577
0.574938575
0.572481572
0.57002457
0.567567568
0.565110565
0.562653563
0.56019656
0.557739558
0.555282555
0.552825553
0.55036855
0.547911548
0.545454545
0.542997543
0.540540541
0.538083538
0.535626536
0.533169533
0.530712531
0.528255528
0.525798526
0.523341523
0.520884521
0.518427518
0.515970516
0.513513514
0.511056511
0.508599509
0.506142506
appendix h * Equations for the Water Clarity Criteria Biological Reference Curves
-------
H-4
rank
205
204
203
202
201
200
199
198
197
196
195
194
193
192
191
190
189
188
187
186
185
184
183
182
181
180
179
178
177
176
175
174
173
172
171
170
169
168
167
166
165
164
163
162
161
160
159
158
157
156
155
volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
time
0.503685504
0.501228501
0.498771499
0.496314496
0.493857494
0.491400491
0.488943489
0.486486486
0.484029484
0.481572482
0.479115479
0.476658477
0.474201474
0.471744472
0.469287469
0.466830467
0.464373464
0.461916462
0.459459459
0.457002457
0.454545455
0.452088452
0.44963145
0.447174447
0.444717445
0.442260442
0.43980344
0.437346437
0.434889435
0.432432432
0.42997543
0.427518428
0.425061425
0.422604423
0.42014742
0.417690418
0.415233415
0.412776413
0.41031941
0.407862408
0.405405405
0.402948403
0.4004914
0.398034398
0.395577396
0.393120393
0.390663391
0.388206388
0.385749386
0.383292383
0.380835381
rank
154
153
152
151
150
149
148
147
146
145
144
143
142
141
140
139
138
137
136
135
134
133
132
131
130
129
128
127
126
125
124
123
122
121
120
119
118
117
116
115
114
113
112
111
110
109
108
107
106
105
104
volume
0.0054
0.0108
0.0108
0.0108
0.0108
0.0108
0.0196
0.0215
0.0215
0.0215
0.0215
0.0215
0.0261
0.0269
0.0269
0.0278
0.0278
0.0278
0.0323
0.0323
0.0455
0.0556
0.0719
0.0784
0.1111
0.1176
0.1237
0.1307
0.1307
0.1389
0.1389
0.1389
0.1389
0.1438
0.1505
0.1667
0.1667
0.1944
0.1989
0.2151
0.2222
0.25
0.2742
0.2778
0.3116
0.3203
0.3659
0.3889
0.3889
0.4167
0.4167
time
0.378378378
0.375921376
0.373464373
0.371007371
0.368550369
0.366093366
0.363636364
0.361179361
0.358722359
0.356265356
0.353808354
0.351351351
0.348894349
0.346437346
0.343980344
0.341523342
0.339066339
0.336609337
0.334152334
0.331695332
0.329238329
0.326781327
0.324324324
0.321867322
0.319410319
0.316953317
0.314496314
0.312039312
0.30958231
0.307125307
0.304668305
0.302211302
0.2997543
0.297297297
0.294840295
0.292383292
0.28992629
0.287469287
0.285012285
0.282555283
0.28009828
0.277641278
0.275184275
0.272727273
0.27027027
0.267813268
0.265356265
0.262899263
0.26044226
0.257985258
0.255528256
appendix h » Equations for the Water Clarity Criteria Biological Reference Curves
-------
H-5
rank
103
102
101
100
99
98
97
96
95
94
93
92
91
90
89
88
87
86
85
84
83
82
81
80
79
78
77
76
75
74
73
72
71
70
69
68
67
66
65
64
63
62
61
60
59
58
57
56
55
54
53
volume
0.4444
0.4444
0.5538
0.5581
0.5752
0.5948
0.6136
0.6136
0.6202
0.6237
0.6434
0.6434
0.6434
0.6434
0.6899
0.7209
0.7209
0.7364
0.7519
0.7597
0.7597
0.7674
0.7727
0.7796
0.7974
0.8062
0.8188
0.8225
0.8225
0.8333
0.837
0.8611
0.8864
0.8864
0.8889
0.8992
0.8992
0.8992
0.9058
0.938
0.9457
0.9477
0.9545
0.9545
0.9612
0.9674
0.9722
0.9739
0.9767
0.9767
0.9773
time
0.253071253
0.250614251
0.248157248
0.245700246
0.243243243
0.240786241
0.238329238
0.235872236
0.233415233
0.230958231
0.228501229
0.226044226
0.223587224
0.221130221
0.218673219
0.216216216
0.213759214
0.211302211
0.208845209
0.206388206
0.203931204
0.201474201
0.199017199
0.196560197
0.194103194
0.191646192
0.189189189
0.186732187
0.184275184
0.181818182
0.179361179
0.176904177
0.174447174
0.171990172
0.16953317
0.167076167
0.164619165
0.162162162
0.15970516
0.157248157
0.154791155
0.152334152
0.14987715
0.147420147
0.144963145
0.142506143
0.14004914
0.137592138
0.135135135
0.132678133
0.13022113
rank
52
51
50
49
48
47
46
45
44
43
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
volume
0.9804
0.9804
0.9804
0.9935
0.9935
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
time
0.127764128
0.125307125
0.122850123
0.12039312
0.117936118
0.115479115
0.113022113
0.110565111
0.108108108
0.105651106
0.103194103
0.100737101
0.098280098
0.095823096
0.093366093
0.090909091
0.088452088
0.085995086
0.083538084
0.081081081
0.078624079
0.076167076
0.073710074
0.071253071
0.068796069
0.066339066
0.063882064
0.061425061
0.058968059
0.056511057
0.054054054
0.051597052
0.049140049
0.046683047
0.044226044
0.041769042
0.039312039
0.036855037
0.034398034
0.031941032
0.029484029
0.027027027
0.024570025
0.022113022
0.01965602
0.017199017
0.014742015
0.012285012
0.00982801
0.007371007
0.004914005
appendix h * Equations for the Water Clarity Criteria Biological Reference Curves
-------
H-6
rank
1
volume
1
1
time
0.002457002
0
rank
390
389
388
387
386
385
384
383
382
381
380
379
378
377
376
375
374
373
372
371
370
369
368
367
366
365
364
363
362
361
360
359
358
357
356
355
354
353
352
351
350
349
348
volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
time
1
0.997442455
0.99488491
0.992327366
0.989769821
0.987212276
0.984654731
0.982097187
0.979539642
0.976982097
0.974424552
0.971867008
0.969309463
0.966751918
0.964194373
0.961636829
0.959079284
0.956521739
0.953964194
0.95140665
0.948849105
0.94629156
0.943734015
0.941176471
0.938618926
0.936061381
0.933503836
0.930946292
0.928388747
0.925831202
0.923273657
0.920716113
0.918158568
0.915601023
0.913043478
0.910485934
0.907928389
0.905370844
0.902813299
0.900255754
0.89769821
0.895140665
0.89258312
0.890025575
rank
347
346
345
344
343
342
341
340
339
338
337
336
335
334
333
332
331
330
329
328
327
326
325
324
323
322
321
320
319
318
317
316
315
314
313
312
311
310
309
308
307
306
305
304
303
302
301
300
299
298
297
volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
time
0.887468031
0.884910486
0.882352941
0.879795396
0.877237852
0.874680307
0.872122762
0.869565217
0.867007673
0.864450128
0.861892583
0.859335038
0.856777494
0.854219949
0.851662404
0.849104859
0.846547315
0.84398977
0.841432225
0.83887468
0.836317136
0.833759591
0.831202046
0.828644501
0.826086957
0.823529412
0.820971867
0.818414322
0.815856777
0.813299233
0.810741688
0.808184143
0.805626598
0.803069054
0.800511509
0.797953964
0.795396419
0.792838875
0.79028133
0.787723785
0.78516624
0.782608696
0.780051151
0.777493606
0.774936061
0.772378517
0.769820972
0.767263427
0.764705882
0.762148338
0.759590793
appendix h
-------
H-7
rank
296
295
294
293
292
291
290
289
288
287
286
285
284
283
282
281
280
279
278
277
276
275
274
273
272
271
270
269
268
267
266
265
264
263
262
261
260
259
258
257
256
255
254
253
252
251
250
249
248
247
246
volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
time
0.757033248
0.754475703
0.751918159
0.749360614
0.746803069
0.744245524
0.74168798
0.739130435
0.73657289
0.734015345
0.731457801
0.728900256
0.72634271 1
0.723785166
0.721227621
0.718670077
0.716112532
0.713554987
0.710997442
0.708439898
0.705882353
0.703324808
0.700767263
0.698209719
0.695652174
0.693094629
0.690537084
0.68797954
0.685421995
0.68286445
0.680306905
0.677749361
0.675191816
0.672634271
0.670076726
0.667519182
0.664961637
0.662404092
0.659846547
0.657289003
0.654731458
0.652173913
0.649616368
0.647058824
0.644501279
0.641943734
0.639386189
0.636828645
0.6342711
0.631713555
0.62915601
rank
245
244
243
242
241
240
239
238
237
236
235
234
233
232
231
230
229
228
227
226
225
224
223
222
221
220
219
218
217
216
215
214
213
212
211
210
209
208
207
206
205
204
203
202
201
200
199
198
197
196
195
volume
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.0037
0.0037
0.0037
0.0054
0.0054
0.0073
0.0094
0.011
0.0147
0.0147
0.0154
0.0154
0.0162
0.0162
0.0162
0.0162
0.0162
0.0162
0.0162
0.0204
time
0.626598465
0.624040921
0.621483376
0.618925831
0.616368286
0.613810742
0.611253197
0.608695652
0.606138107
0.603580563
0.601023018
0.598465473
0.595907928
0.593350384
0.590792839
0.588235294
0.585677749
0.583120205
0.58056266
0.578005115
0.57544757
0.572890026
0.570332481
0.567774936
0.565217391
0.562659847
0.560102302
0.557544757
0.554987212
0.552429668
0.549872123
0.547314578
0.544757033
0.542199488
0.539641944
0.537084399
0.534526854
0.531969309
0.529411765
0.52685422
0.524296675
0.52173913
0.519181586
0.516624041
0.514066496
0.511508951
0.508951407
0.506393862
0.503836317
0.501278772
0.498721228
appendix h * Equations for the Water Clarity Criteria Biological Reference Curves
-------
H-8
rank
194
193
192
191
190
189
188
187
186
185
184
183
182
181
180
179
178
177
176
175
174
173
172
171
170
169
168
167
166
165
164
163
162
161
160
159
158
157
156
155
154
153
152
151
150
149
148
147
146
145
144
volume
0.0204
0.022
0.022
0.0235
0.0256
0.0256
0.027
0.027
0.027
0.0282
0.0378
0.0432
0.0476
0.0476
0.0513
0.0513
0.0513
0.0612
0.0615
0.0615
0.0615
0.0615
0.0615
0.0615
0.0623
0.0659
0.0659
0.0696
0.0703
0.0728
0.0769
0.0806
0.0811
0.0939
0.102
0.102
0.1077
0.1077
0.1081
0.1127
0.1136
0.1209
0.1221
0.1224
0.1224
0.1224
0.1224
0.1224
0.1224
0.1224
0.1224
time
0.496163683
0.493606138
0.491048593
0.488491049
0.485933504
0.483375959
0.480818414
0.47826087
0.475703325
0.47314578
0.470588235
0.468030691
0.465473146
0.462915601
0.460358056
0.457800512
0.455242967
0.452685422
0.450127877
0.447570332
0.445012788
0.442455243
0.439897698
0.437340153
0.434782609
0.432225064
0.429667519
0.427109974
0.42455243
0.421994885
0.41943734
0.416879795
0.414322251
0.411764706
0.409207161
0.406649616
0.404092072
0.401534527
0.398976982
0.396419437
0.393861893
0.391304348
0.388746803
0.386189258
0.383631714
0.381074169
0.378516624
0.375959079
0.373401535
0.37084399
0.368286445
rank
143
142
141
140
139
138
137
136
135
134
133
132
131
130
129
128
127
126
125
124
123
122
121
120
119
118
117
116
115
114
113
112
111
110
109
108
107
106
105
104
103
102
101
100
99
98
97
96
95
94
93
volume
0.1224
0.1224
0.1224
0.1224
0.1231
0.1231
0.1231
0.1231
0.1282
0.1297
0.1319
0.1351
0.1385
0.1429
0.1502
0.1538
0.1538
0.1575
0.1612
0.1612
0.1633
0.1633
0.169
0.1795
0.1837
0.1868
0.1995
0.2
0.2041
0.2088
0.2108
0.2113
0.2254
0.2308
0.2308
0.2308
0.2308
0.2378
0.2394
0.2432
0.2449
0.2449
0.2486
0.2541
0.2564
0.2582
0.2653
0.2653
0.2811
0.2857
0.2857
time
0.3657289
0.363171355
0.360613811
0.358056266
0.355498721
0.352941176
0.350383632
0.347826087
0.345268542
0.342710997
0.340153453
0.337595908
0.335038363
0.332480818
0.329923274
0.327365729
0.324808184
0.322250639
0.319693095
0.31713555
0.314578005
0.31202046
0.309462916
0.306905371
0.304347826
0.301790281
0.299232737
0.296675192
0.294117647
0.291560102
0.289002558
0.286445013
0.283887468
0.281329923
0.278772379
0.276214834
0.273657289
0.271099744
0.268542199
0.265984655
0.26342711
0.260869565
0.25831202
0.255754476
0.253196931
0.250639386
0.248081841
0.245524297
0.242966752
0.240409207
0.237851662
appendix h » Equations for the Water Clarity Criteria Biological Reference Curves
-------
H-9
rank
92
91
90
89
88
87
86
85
84
83
82
81
80
79
78
77
76
75
74
73
72
71
70
69
68
67
66
65
64
63
62
61
60
59
58
57
56
55
54
53
52
51
50
49
48
47
46
45
44
43
42
volume
0.2857
0.2973
0.3081
0.3187
0.3216
0.3239
0.3243
0.3265
0.3405
0.3451
0.3469
0.348
0.359
0.359
0.359
0.3592
0.3622
0.3692
0.3838
0.3846
0.3892
0.3919
0.392
0.4
0.4
0.4
0.4054
0.4202
0.4286
0.4324
0.439
0.4432
0.4615
0.4703
0.4769
0.4812
0.4872
0.4872
0.4872
0.5077
0.5077
0.5092
0.5094
0.5102
0.5128
0.5128
0.5231
0.5385
0.5495
0.5657
0.5692
time
0.235294118
0.232736573
0.230179028
0.227621483
0.225063939
0.222506394
0.219948849
0.217391304
0.21483376
0.212276215
0.20971867
0.207161125
0.204603581
0.202046036
0.199488491
0.196930946
0.194373402
0.191815857
0.189258312
0.186700767
0.184143223
0.181585678
0.179028133
0.176470588
0.173913043
0.171355499
0.168797954
0.166240409
0.163682864
0.16112532
0.158567775
0.15601023
0.153452685
0.150895141
0.148337596
0.145780051
0.143222506
0.140664962
0.138107417
0.135549872
0.132992327
0.130434783
0.127877238
0.125319693
0.122762148
0.120204604
0.117647059
0.115089514
0.112531969
0.109974425
0.10741688
rank
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
volume
0.5838
0.5897
0.6
0.6054
0.6154
0.6315
0.641
0.6462
0.6667
0.6714
0.6923
0.6923
0.6923
0.7077
0.7077
0.7297
0.7347
0.7538
0.7551
0.7568
0.7949
0.7959
0.8239
0.838
0.8498
0.8571
0.8571
0.8615
0.9121
0.9385
0.9388
1
1
1
1
1
1
1
1
1
1
1
time
0.104859335
0.10230179
0.099744246
0.097186701
0.094629156
0.092071611
0.089514066
0.086956522
0.084398977
0.081841432
0.079283887
0.076726343
0.074168798
0.071611253
0.069053708
0.066496164
0.063938619
0.061381074
0.058823529
0.056265985
0.05370844
0.051150895
0.04859335
0.046035806
0.043478261
0.040920716
0.038363171
0.035805627
0.033248082
0.030690537
0.028132992
0.025575448
0.023017903
0.020460358
0.017902813
0.015345269
0.012787724
0.010230179
0.007672634
0.00511509
0.002557545
0
appendix h * Equations for the Water Clarity Criteria Biological Reference Curves
-------
appendix |
Evaluation of Maryland and
Virginia Chesapeake Bay
Segment SAV Acreages from
2003 to 2005 for Prioritizing
Shallow-water Monitoring
by Segment
MARYLAND
Chesapeake Bay
Program
Segments/
Subsegments
CHSOH
BSHOH
BOHOH
CB20H
PAXTF
SASOH
C&DOH
PAXOH
GUNOH
MATTF
ELKOH
PISTF
POTTF(MD)
NORTF
SEVMH
CB1TF
MIDOH
PATMH
2003 Acres
0
390
288
212
217
371
0
106
489
612
346
212
885
46
III
7,574
391
7
2004 Acres
4
1,025
730
1,303
220
1,272
8
106
2,392
601
1,913
507
1,256
84
388
10,110
671
183
2005 Acres
228
726
918
1,071
324
1,476
9
125
1,733
770
1,964
757
2,029
78
426
9,193
454
279
2003-2005 Single
Best Year Acres
228
1,025
918
1,303
324
1,476
9
125
2,392
770
1,964
757
2,029
84
426
10,110
671
279
State-adopted
SAV Restoration
Acreage
77
350
354
705
205
1,168
7
115
2,432
792
2,034
789
2,142
89
455
12,903
879
389
Single Best Year
as % of SAV
Restoration
Acreage Status
3 Pass
3 Pass
3 Pass
2 Pass
2 Pass
Pass
Pass
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
appendix i • Evaluation of SAV Acreages from 2003 to 2005 for Prioritizing Shallow-water Monitoring
-------
MARYLAND (continued)
Chesapeake Bay
Program
Segments/
Subsegments
POTOH(MD)
CB3MH
HNGMH
MAGMH
CHOMH1
POTMH(MD)
BIGMH
LCHMH
EASMH
CHSMH
TANMH(MD)
CB5MH(MD)
WSTMH
SOUMH
MANMH
FSBMH
POCMH(MD)
PAXMH
CB4MH
CHOMH2
CHOOH
NANMH
NANOH
RHDMH
WICMH
BACOH
CHOTF
CHSTF
NANTF
POCOH
POCTF
WBRTF
2003 Acres
1,384
23
2,844
169
2,972
2,430
451
784
1,639
117
4,725
700
23
14
235
15
58
37
21
0
0
0
0
0
0
0
0
0
0
0
0
0
2004 Acres
1,408
909
3,433
300
3,774
3,063
550
1,221
1,040
731
4,554
398
0
46
291
17
69
42
10
1
0
0
0
0
0
30
0
0
0
0
0
0
2005 Acres
1,888
567
4,376
308
2,293
2,893
710
260
768
462
5,801
919
0
10
410
7
69
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
2003-2005 Single
Best Year Acres
1,888
909
4,376
308
3,774
3,063
710
1,221
1,639
731
5,801
919
23
46
410
17
69
42
21
0
0
0
0
0
0
30
0
1
0
0
0
0
State-adopted
SAV Restoration
Acreage
2,802
1,370
7,761
579
8,184
7,088
2,043
4,076
6,209
2,928
24,757
8,270
238
479
4,353
197
877
1,634
2,533
1,621
72
3
12
60
3
-
-
-
-
-
-
.
Single Best Year
as % of SAV
Restoration
Acreage
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
N/A
-
-
-
-
-
-
Status
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
No SAV
No SAV
No SAV
No SAV
No SAV
No SAV
NO SAV
appendix i • Evaluation of SAV Acreages from 2003 to 2005 for Prioritizing Shallow-water Monitoring
-------
"
VIRGINIA
Chesapeake Bay
Program
Segments/
Subsegments
MPNTF
PMKTF
POTOH(VA)
CHKOH
RPPTF
POTTF(VA)
CB8PH
CB7PH
JMSOH
CB6PH
MOBPH
JMSPH
POCMH(VA)
CRRMH
TANMH(VA)
CB5MH(VA)
YRKPH
LYNPH
PIAMH
RPPMH
POTMH(VA)
JMSTF
JMSMH
APPTF
YRKMH
EBEMH
ELIMH
ELIPH
LAFMH
MPNOH
PMKOH
RPPOH
SBEMH
WBEMH
2003 Acres
184
217
1,950
425
0
761
5
9,192
9
707
8,457
132
1,608
43
4,682
*
887
0
447
21
55
75
2
0
0
0
0
0
0
0
0
0
0
0
2004 Acres
179
334
2,326
432
24
1,197
6
7,157
0
488
7,549
74
1,094
224
3,990
1,833
597
9
443
33
339
12
2
0
0
0
0
0
0
0
0
0
0
0
2005 Acres
296
585
2,627
697
81
2,336
9
8,139
0
642
7,205
0
1,716
292
5,036
2,464
438
19
561
198
444
53
0
0
0
0
0
0
0
0
0
4
0
0
2003-2005 Single
Best Year Acres
296
585
2,627
697
81
2,336
9
9,192
9
707
8,457
132
1,716
292
5,036
2,464
887
19
561
198
444
75
2
0
0
0
0
0
0
0
0
4
0
0
State-adopted
SAV Restoration
Acreage
85
187
1,503
535
66
2,093
11
15,107
15
1,267
15,901
300
4,066
768
13,579
7,633
2,793
107
3,479
1,700
4,250
1,200
200
379
239
-
-
-
-
-
-
-
-
-
Single Best Year
as % of SAV
Restoration
Acreage
3
3
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
-
-
-
-
-
-
-
-
Status
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Fail
Fail
Fail
Fail
Full
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
No SAV
No SAV
No SAV
No SAV
No SAV
No SAV
L No SAV
No SAV
No SAV
appendix i • Evaluation of SAV Acreages from 2003 to 2005 for Prioritizing Shallow-water Monitoring
-------
1-4
Chesapeake Bay
Program
Segments/
Subsegments
2003 Acres
2004 Acres
2005 Acres
State-adopted
2003-2005 Single SAV Restoration
Best Year Acres Acreage
Single Best Year
as % of SAV
Restoration
Acreage
Status
NANTF(DE)
POTTF(DC)
ANATF (DC)
OF
PASS (> 100% of Goal)
FAIL (50%-< 100% of Goal)
FAIL (<50% of Goal)
No SAV Goal
^Partial data available that year
appendix
-------
J-1
appendix I
Chesapeake Bay Estuarine
Benthic Communities
Assessment Protocol for
Maryland and Virginia
305b/303d Integrated Reports
Maryland (Department of the Environment, Department of Natural Resources),
Virginia (Department of Environmental Quality) and U.S. EPA (Region 3 Water
Protection Division and Chesapeake Bay Program Office) reached agreement on the
protocol to assess Chesapeake Bay benthic community health. This appendix docu-
mented the assessment protocol supporting the States evaluation of Chesapeake Bay
benthic community data as part of their 305b/303d Integrated Reports. This assess-
ment protocol builds directly on the more detailed assessment methods
recommended by Llanso et al. 2005 (see Appendix K).
The overall decision protocol is shown in Figure J-1. Phase I consists of the evalu-
ation of the sample size (i.e., number of B-IBI scores) available from the waterbody
segment during the five-year assessment window. If the sample size satisfies the
requirements of the statistical method (N > 10), a formal assessment of status (i.e.
impaired vs. supports aquatic life use) is determined utilizing the "percent degraded
area" statistical methodology (Phase II). If the sample size requirement is not met
an impairment assessment based solely on these analyses is not possible. Results for
segments with insufficient sample size should still be examined for possible use in
conjunction with other assessment data of the 305b/303d reporting process.
Phase II consists of the impairment assessment of aquatic life use attainment based
on a comparison of Benthic Index of Biotic Integrity (B-IBI) scores and can only be
performed when the number of B-IBI scores within a specified waterbody segment
is sufficient to meet the sample size requirement of the approved statistical method
(N > 10). Phase II can result in one of two possible outcomes: (1) the segment is not
impaired for Aquatic Life use due to benthic community status (note that the
segment may still be impaired for aquatic life use due to failure of other aquatic life
use criteria), or (2) the segment fails to support aquatic life use due to benthic
community status and is assessed as impaired. Best professional judgment can be
appendix J • Bay Estuarine Benthic Communities Assessment Protocol for Maryland and Virginia
-------
J-2
Phase!
Sample She
Evaluation
N<10
i\o
N>li
Yes —
Yes -.
Phase 11
Impairment Assessment
Insufficient sample size
Apply Degraded Area
Statistical method
1
Segment declared
'not impaired* for bcnthic
aquatic life communities
in 305b/303d Integrated
Report
1 No
Segment declared
'impaired' for benthic
aquatic life communities in
305b/303d Integrated
Report
Yes —
Yes -.
Ph»« in
Segment Characterization
(Identify Probable Causes)
Optional use of
B-EBE scores and diagnostic analyses
in conjunction with other available
data for305b/303d Integrated Report
Optional use of
B-IBI scores and diagnostic analyses
in conjunction with other available
data Ibr305b/303d Integrated Report
Apply diagnostic analyses for
assignment of suspected cause(s) of
degradation in 305b/303d Integrated
Report
J-1, Overall Chesapeake Bay benthic index of biotic integrity assessment decision
protocol.
applied to override (reverse) the outcome of the formal statistical analysis results, but
such reversals must be justified and documented.
Phase III consists of the identification of probable causes of benthic impairment of the
waterbody segment based upon benthic stressor diagnostic analyses. It is a two-step
procedure that involves (1) Site Classification, and (2) Segment Characterization.
1. Site classification: The first step is to assign probable cause of benthic degra-
dation to each individual "degraded" benthic sample. For purposed of these
diagnostic analyses, a sample is considered degraded if the B-IBI score is less
than 2.7.
Site Classification—Step la: The application of a formal statistical linear
discriminant function calculates the 'inclusion probability' of each degraded
site belonging to a 'contaminant caused' group or an 'other causes' group,
based upon its B-IBI score and associated metrics. If a site is assigned to the
'Contaminant' Group with a probability > 0.9, this site is considered impacted
by contaminated sediment and no further classification is required.
Site Classification—Step Ib: If a site is classified as degraded due to 'other
causes' (i.e., not contaminant-related), an evaluation of the relative abundance
(and/or biomass) of the benthos is examined. Scores for both abundance and
biomass are considered to be bipolar for the Chesapeake Bay Benthic IBI. For
either metric; a high score of 5, indicating desirable conditions, falls in the mid-
range of the abundance/biomass distributions, while a low score of 1,
indicating undesirable conditions, can result either from insufficient abun-
ap pen a ix j
-------
J-3
dance/biomass or excessive abundance/biomass. The scoring thresholds for
these two metrics vary with habitat type (salinity regime and substrate type) as
summarized in Figure J-2. In this process, a site is classified as degraded by
"low dissolved oxygen" if the abundance (and/or biomass) metric scores a 1
due to insufficient abundance (and/or biomass). Alternatively, if the abundance
(and/or biomass) metric scores a 1 because of excessive abundance (and/or
biomass) the site is classified as degraded by "eutrophication".
2. Segment classification: The assignment of probable causes of benthic degra-
dation for the overall segment is accomplished using a simple 25% rule. If the
percent of total sites in a segment impacted by a single cause (i.e. sediment
contaminants, low dissolved oxygen, or eutrophication) exceeds 25%, then that
cause is assigned. If no causes exceed 25%, the cause is considered unknown.
The cause(s) should be identified as a suspected (vs. verified) cause of benthic
community degradation in the ADB database.
Table J-l shows the possible conclusions from applying the above protocol. The
States should carefully review the results from application of the protocol to ensure
all findings and conclusions are rational and reasonable. Best profession judgment,
common sense, and ancillary information about each segment should be utilized as
necessary and available.
Habitat
Metric
Lower Limit
(Metric
Score=1)
Upper Limit
(Metric
Score=1)
Tidal Freshwater
Abundance (# m-2)
^Biomass(gm-2)^
<800
25500
Oligohaline
Low Mesohaline
Abundance (# m-2)
Biomass (g m-2)
Abundance (# m-2)
Biomass (g m-2)
<180
<500
<1
24050
26000
230
High Mesohaline Sand
Abundance (# m-2)
Biomass (g m-2)
<1000
<1
>5000
250
High Mesohaline Mud
Abundance (# m-2)
Biomass (g m-2)
<1000
<0.5
>5000
250
Polyhaline Sand
Abundance (# m-2)
Biomass (g m-2)
<1500
<1
58000
250
Polyhaline Mud
Abundance (# m-2)
Biomass (g m-2)
<1000
<0.5
>8000
230
J-2, Metric scoring
Source: Llanso 2002, Table 9,
for eutrophication and low dissolved oxygen causes.
pages 24-26.
-------
J-4
,--", Possible conclusions from application of the assessment protocol.
n>=10 - sufficient sample size for assessment
Scenario
1
Impairment Analysis
CL-L
(P-Po)
(Table 3 of
VERSAR
Technical
Report)
<0
Impaired:
Degraded Area
method?
(Table 3 of
VERSAR
Technical
Report)
No
Stressor Diagnostic Analyses
Samples with
contaminant
Posterior Prob.
p>= 0.90; % of
Total (Table 5 of
VERSAR Technical
Report)
review as
supplemental info
Degraded Samples with
excessive Abundance/Biomass;
% of Total w/o Cont. (Table 5 of
VERSAR Technical Report)
review as supplemental info
Degraded Samples with
Insufficient
Abundance/Biomass; % of
Total w/o Cont. (Table 5 of
VERSAR Technical Report)
review as supplemental info
A small, non-significant fraction of IBI scores are within or below the lower range of the reference distribution so water quality conditions in this
segment support the benthic community (no impairment).
Where community samples are degraded, the stressor analyses may provide information that supports other assessment data.
2
>0
Yes
< 25% of Total
Samples
< 25% of Total Samples
< 25% of Total Samples
A large, significant fraction of IBI scores are within or below the lower range of the reference distribution, so water quality conditions in this
segment do not support the benthic community (impaired condition).
Stressor diagnostic analyses do not suggest dominant stressors affecting community composition. Cause of degradation is "unknown".
3
>0
Yes
> 25% of Total
Samples
< 25% of Total Samples
< 25% of Total Samples
A large, significant fraction of IBI scores are within or below the lower range of the reference distribution, so water quality conditions in this
segment do not support the benthic community (impaired condition).
Stressor diagnostic analyses suggest sediment contaminants as a likely pollutant affecting benthic community structure.
4
>0
Yes
> 25% of Total
Samples
> 25% of Total Samples
< 25% of Total Samples
A large, significant fraction of IBI scores are within or below the lower range of the reference distribution, so water quality conditions in this
segment do not support the benthic community (impaired condition).
Stressor diagnostic analyses suggest sediment contaminants as a likely pollutant affecting benthic community structure. Observation of high
biomass or abundance is indicative of eutrophic conditions as an additional stressor affecting the benthic community.
5
>0
Yes
> 25% of Total
Samples
< 25% of Total Samples
> 25% of Total Samples
A large, significant fraction of IBI scores are within or below the lower range of the reference distribution, so water quality conditions in this
segment do not support the benthic community (impaired condition).
Stressor diagnostic analyses suggest sediment contaminants as a likely pollutant affecting benthic community structure. Samples observed with
low biomass or abundance are indicative of low dissolved oxygen as an additional stressor affecting the benthic community.
6
>0
Yes
< 25% of Total
Samples
> 25% of Total Samples
< 25% of Total Samples
A large, significant fraction of IBI scores are within or below the lower range of the reference distribution, so water quality conditions in this
segment do not support the benthic community (impaired condition).
Stressor diagnostic analyses do not suggest sediment contaminants as a stressors affecting community composition. Samples observed with
high biomass or abundance are indicative of eutrophic conditions (excessive nutrients) as a stressor affecting the benthic community.
7
>0
Yes
< 25% of Total
Samples
> 25% of Total Samples
> 25% of Total Samples
A large, significant fraction of IBI scores are within or below the lower range of the reference distribution, so water quality conditions in this
segment do not support the benthic community (impaired condition).
Stressor diagnostic analyses do not suggest sediment contaminants as stressor affecting community composition. Samples observed with high
biomass or abundance are indicative of eutrophic conditions within the segment while other samples observed with low biomass or abundance
are indicative of low dissolved oxygen as another stressor within the segment.
8
>0
Yes
< 25% of Total
Samples
< 25% of Total Samples
> 25% of Total Samples
A large, significant fraction of IBI scores are within or below the lower range of the reference distribution, so water quality conditions in this
segment do not support the benthic community (impaired condition).
Stressor diagnostic analyses do not suggest sediment contaminants as a stressor affecting community composition. Samples observed with
low biomass or abundance are indicative of low dissolved oxygen as a stressor affecting the segment.
9
>0
Yes
> 25% of Total
Samples
> 25% of Total Samples
> 25% of Total Samples
A large, significant fraction of IBI scores are within or below the lower range of the reference distribution, so water quality conditions in this
segment do not support the benthic community (impaired condition).
Stressor diagnostic analyses suggest sediment contaminants as a likely pollutant affecting benthic community structure. Samples observed with
high biomass or abundance are indicative of eutrophic conditions within the segment while other samples observed with low biomass or
abundance are indicative of low dissolved oxygen as an additional stressor within the segment.
appendix j • Bay Estuarine Benthic Communities Assessment Protocol for Maryland and Virginia
-------
J-5
J-l, (continued)
n<10 - small sample size, insufficient for analysis
Scenario
1
Impairment Analysis
CL-L
(P-Po)
(Table 3 of
VERSAR
Technical
Report)
n/a
Impaired:
Degraded
Area? (Table 3
of VERSAR
Technical
Report)
Unknown, Not
Assessed
Stressor Diagnostic Analyses
Samples with
contaminant
Posterior Prob.
p>= 0.90; % of
Total (Table 5 of
VERSAR
Technical
Report)
review as
supplemental info
Degraded Samples with
excessive
Abundance/Biomass; % of
Total w/o Cont. (Table 5 of
VERSAR Technical Report)
review as supplemental info
Degraded Samples with Insufficient
Abundance/Biomass; % of Total w/o
Cont. (Table 5 of VERSAR Technical
Report)
review as supplemental info
There are too few samples to define the confidence interval of benthic sample IBIs, so in this segment -the biological community condition is
unknown.
Where community samples are identified as degraded, information from the stressor diagnostic analyses may provide supplemental information
that may support other assessment data.
Llanso, RJ. 2002. Methods for Calculating the Chesapeake Bay Benthic Index of biotic
Integrity. Versar Inc., Columbia, Maryland http://www.baybenthos.versar.com/docs/Ches-
BayBIBI.PDF
Llanso , RJ., J.H. V01stad, D.M. Dauer, and M.F. Lane. 2005. 2006 303(D) Assessment
Methods For Chesapeake Bay Benthos. Final Report Submitted to Virginia Department of
Environmental Quality, Richmond, Virginia. September 2005.
appendix i * Evaluation of SAV Acreages from 2003 to 2005 for Prioritizing Shallow-water Monitoring
-------
-------
K-1
appendix
2006 303(d) Assessment
Methods for Chesapeake Bay
Benthos
Final Report Submitted to:
Virginia Department of Environmental Quality
629 East Main Street
Richmond, Virginia 23230
Submitted by:
Roberto J. Llanso
Jon H. V01stad
Versar, Inc., Columbia, Maryland
Daniel M. Dauer
Michael F. Lane
Department of Biological Sciences
Old Dominion University
Norfolk, Virginia
September 2005
FOREWORD
This report, 2006 303(d) Assessment Methods for Chesapeake Bay Benthos, was
prepared by Versar at the request of the Virginia Department of Environmental
Quality, under Purchase Order # 11646 between Versar, Inc. and the Commonwealth
of Virginia. Old Dominion University contributed to the diagnostic (discriminant
tool) assessment and to project conceptualization and evaluation. The statistical
analyses for the 2006 impairment assessment were conducted by Dr. Ed Weber and
Ms. Jody Dew, of Versar. Dr. Weber also contributed to the development of the
Degraded Area method presented in this report.
appendix k • 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-2
To meet the requirements of the Clean Water Act, the States of Maryland and
Virginia are using benthic biological criteria for reporting overall condition and iden-
tification of impaired waters in Chesapeake Bay. The Chesapeake Bay benthic index
of biotic integrity (B-IBI) is the basis for these biological criteria. Previous work
conducted by Versar and Old Dominion University had two objectives: to develop a
methodology for the assessment of benthic community status for 303(d) impairment
decisions and to produce an assessment for each of the Chesapeake Bay segments
and sub-segments containing benthic community data. A statistical procedure was
developed that tests whether the distribution of B-IBI scores from probability-based
samples collected from a Bay segment is significantly different from the distribution
of scores from reference sites (Llanso et al. 2003). This procedure, a stratified
Wilcoxon rank sum test, was evaluated and applied to the 2004 assessment data. The
assessment resulted in 26 segments considered impaired based upon benthic
community condition. The Wilcoxon approach, however, was sensitive to small
shifts in B-IBI scores relative to the reference condition, even in some cases where
a majority of the B-IBI scores in a segment met the restoration goals. For stratified
data (i.e., the habitat types of the B-IBI, see below) it was not possible to estimate
the magnitude of the shift, for example by using a Hodges-Lehman confidence
interval. Thus, with the Wilcoxon approach we were unable to estimate the magni-
tude of degradation: the difference between the segment and the reference condition.
A small difference could be statistically significant but of little ecological relevance.
It was recommended that alternative methods be evaluated, especially those that take
into account magnitude of departure from reference conditions and whether this
magnitude is above specific thresholds of protection that the States may wish to
implement. For the 2006 303(d) report, we developed a new method that quantifies
magnitude of degradation. We call this method "Degraded Area." In the present
report, we describe the Degraded Area method, apply this method and the Wilcoxon
approach to the 2006 assessment data, and compare the results.
In addition, a benthic diagnostic tool has been developed that can be used to identify
potential sources of stress affecting benthic community condition in the Chesapeake
Bay (Dauer et al. 2002). The tool can distinguish stress due to contaminants versus
stress due to other factors (e.g., low dissolved oxygen, or unknown). This screening
tool was used to identify which impaired segments have a high probability of sedi-
ment contamination. These segments could then be targeted for additional sampling
or evaluation. The B-IBI metric scores for abundance and biomass were also used to
identify (1) insufficient abundance patterns consistent with a low dissolved oxygen
effect and (2) excessive abundance patterns consistent with eutrophication effects.
appendix k » 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-3
1. Develop a new method for the assessment of Chesapeake Bay benthic commu-
nity status for 303(d) impairment decisions.
2. Produce an assessment for the 2006 303(d) report using both the new method
and the Wilcoxon approach.
3. Apply the benthic diagnostic tool and the insufficient/excessive abundance
criteria to the 2006 assessment data.
Like the Wilcoxon (described in Llanso et al. 2003), the Degraded Area method
compares reference data sets to assessment data sets. The reference data set
consisted of the calibration and validation data used to develop the Chesapeake Bay
benthic index of biotic integrity (B-IBI). The Chesapeake Bay B-IBI is described in
Weisberg et al. (1997) and Alden et al. (2002). The B-IBI consists of benthic
community metrics and scoring thresholds (metric values) that were developed sepa-
rately for seven habitat types (Table 1). The numbers of reference samples in each
habitat used to develop the B-IBI, the Wilcoxon approach, and the method described
in this report are listed in Table 2. The reference samples were either "good" (=unde-
graded, collected at sites known to have good sediment and water quality) or
"degraded" (collected at sites with low dissolved oxygen, organic enrichment, or
high sediment contaminant concentrations and toxicity). To develop the B-IBI, Weis-
berg et al. (1997) used averages of three replicate samples per site for mesohaline
and polyhaline habitats, while Alden et al. (2002) used single replicate samples for
tidal fresh and oligohaline habitats. We used the same metrics values produced by
these two studies, but re-calculated B-IBI scores from these metrics to be consistent
with the latest B-IBI methodology. The methods for the calculation of the Chesa-
peake B-IBI are described in the World Wide Web at: http://www.baybenthos.
versar.com/ referenc.htm.
The assessment data for the 2006 303(d) report consisted of random samples
collected from 2000 to 2004 throughout the Chesapeake Bay. A total of 1,430
samples (single replicates) were used, including 750 samples collected by the Mary-
land Chesapeake Bay benthic monitoring program, 500 samples collected by the
Virginia Chesapeake Bay benthic monitoring program, 150 samples collected by the
Elizabeth River benthic biological monitoring program, and 10 samples collected for
a gear comparison study in each of Mobjack Bay, the tidal fresh Mattaponi River,
and the Nansemond River. All assessment samples were collected with a Young grab
(440 cm2 surface area, 0.5-mm screen). For sample collection methods, see the
benthic monitoring program comprehensive reports posted at the World Wide Web
address given above.
appendix k * 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-4
Assessments were produced for each of 85 Chesapeake Bay Program segments and
sub-segments containing benthic data. Segments (TMWA 1999) are Chesapeake Bay
regions having similar salinity and hydrographic characteristics. In Virginia,
segments were sub-divided into smaller units by the Virginia Department of Envi-
ronmental Quality. Sub-segments were produced for each of the mainstems of rivers
and bays (e.g., James River mesohaline) and for some of the smaller systems
opening into the mainstem (e.g., Pagan River). Assessment samples were assigned to
segments and sub-segments using GIS software. Hydrographic data collected synop-
tically with the benthic data were used to assign each sample to one of seven habitat
classes used in the calculation of the B-IBI. These are the same habitat classes used
in the reference data set.
3.2.
The new method developed for the 2006 assessment was based on the confidence
limit and bootstrap simulation concepts described in Alden et al. (2002). Specifi-
cally, bootstrap simulation (Efron and Tibshirani 1998) was applied to incorporate
uncertainty in reference conditions. Bootstrap simulation is used to assess the accu-
racy of an estimate by randomly sampling n times, with replacement, from an
original data set. In our case, we wished to estimate the score corresponding to the
5th percentile of the B-IBI reference distributions for the good sites (by habitat).
Because the reference distributions were based on small sample sizes, the percentiles
were not well defined and would likely vary if different sets of reference sites were
sampled. Thus the need to estimate this parameter more accurately with bootstrap
simulations. Bootstrap simulations make no assumptions, except that the reference
data are a representative sample from a "super population" of reference sites.
For each habitat, a threshold based on the 5th percentile B-IBI score of the reference
data set for the good sites (or the maximum B-IBI score observed for the degraded
sites, see below), was determined. This threshold was not intended to serve as a crite-
rion for classifying individual B-IBI scores, rather it was used to categorize the
segment as impaired or not based on the proportion of sites below the threshold (i.e.,
degraded area) and the variance associated with this estimate. The variance in the
estimates of proportions for each segment was estimated by the simulations.
The B-IBI scores for the reference good and degraded sites had degrees of overlap
that ranged from quite high in the tidal freshwater and oligohaline habitats to moder-
ately low in the mesohaline and polyhaline habitats. An assessment sample is more
likely to come from an impaired benthic community if the B-IBI score for this
sample is within the range of scores observed for sites known to be degraded. There-
fore, two criteria were established for determining the threshold: its score had to be
within the lower bound of the good reference distribution (i.e., 5th percentile), and
it had to be within the upper range of observed scores for known degraded sites (i.e.,
the reference degraded sites). If the 5th percentile score for a simulation run was not
within the range of scores for the reference degraded sites, then the maximum B-IBI
score for the reference degraded sites was selected as the threshold. Thus, in this
study, sites with low B-IBI scores below thresholds were likely to be impaired and
unlikely to come from good reference areas.
appendix k * 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-5
In each simulation run, a subset of the reference good sites for each habitat was
selected at random, and the B-IBI threshold for this subset was determined (i.e., the
IBI score at the 5th percentile, or the maximum score for the reference degraded
samples). The scores of the assessment data for each habitat were then compared to
the threshold to estimate the proportion of sites below the threshold. By repeating
this process over and over again (5,000 runs) we were able to estimate the variance
in the proportion of sites below the threshold from the bootstrap estimates. This vari-
ance reflects variability in the thresholds as well as sampling variability in the
assessment data.
In the final step of the method, segments were declared impaired if the proportion of
sites below the threshold (i.e., degraded area) was significantly higher than expected
under the null hypothesis. Under the null hypothesis, a small number of sites
(defined as 5 % of the sites) would be expected to have low IBI scores even if all sites
in a segment were in good condition (i.e., no low dissolved oxygen, contaminant, or
nutrient enrichment problems). This is because of natural variability in the benthic
communities, the effects of natural stressors, and sampling and methodological error.
For a segment to be declared as impaired, the lower bound of the 95% confidence
interval of the estimate had to be higher than 5% (the expected proportion under the
null hypothesis), with a minimum sample size of 10. A 5% level was used in agree-
ment with standard statistical practice.
The steps described above are summarized below and in Appendix A:
1. Thresholds are set for each of seven benthic habitats in Chesapeake Bay.
2. The threshold is set as the smaller of two values: 5th percentile IBI score for
the good reference sites or maximum observed IBI score for the degraded refer-
ence sites.
3. The 5th percentile score and its variance is estimated by bootstrap simulations.
4. For each iteration of the bootstrap simulation, a subset (of same sample size)
of the good reference sites for each habitat is selected at random (with replace-
ment), and the 5th percentile score determined.
5. At each iteration, the threshold is set according to #2.
6. At each iteration, the assessment data are compared to the reference data to
estimate the proportion of sites (P) with scores below the threshold. This is
done for each of one or more habitats within a segment.
7. P is averaged over all the iterations.
8. Under the null hypothesis, 5% of the sites (Po) would be expected to have low
IBI scores, even if all sites in a segment were in good condition.
9. Segments are declared impaired if P — Po > 0 (greater than expected under the
null hypothesis, with 95% confidence) (See Schenker and Gentleman 2001).
,
A stratified Wilcoxon rank sum test was applied as described in Llanso et al. (2003)
using Proc-StatXact 5 software (Cytel Software Corporation 2002). B-IBI scores
were grouped into three ordered condition categories (1.0-2.0, 2.1-2.9, 3.0-5.0) and
appendix k » 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-6
the distribution of scores in each category within a segment was compared for each
habitat to the distribution of scores for the good reference condition. Under the null
hypothesis (Ho) of no impairment, the two populations (segment and reference)
were considered to have the same underlying multinomial distributions of samples
among the ordered categories. The assessment of impairment was based on a one-
sided exact test of Ho against the alternative hypothesis that the segment had a
distribution shifted towards lower B-IBI scores than for the reference condition. The
ranking was done separately by habitat, and then combined across habitats.
Segments with a minimum of 10 samples for which the test was significant at the 1%
alpha level and 90% power, were considered impaired under this method.
3.4.
The benthic diagnostic tool allows environmental managers to identify potential
sources of anthropogenic stress to benthic communities within Chesapeake Bay. The
development and application of the tool was described in detail in Dauer et al. (2002,
2005). The benthic diagnostic tool is based on a linear discriminant function that
classifies sites in Chesapeake Bay identified as having degraded benthic communi-
ties into categories distinguished by the type of stress experienced by those
communities. Presently, the function is capable of discriminating contaminated sites
from sites affected by all other potential sources of stress in any of the seven benthic
habitat types of Chesapeake Bay. Sites are classified into two groups: 1) a contami-
nant group and 2) the other group representing all other potential sources of stress
(eutrophication, low dissolved oxygen, etc.). This function is a linear combination of
variables that includes over 60 measures of diversity, dominance, and function of
benthic communities. The score for the function is used to calculate the probabilities
that a sample is drawn from both groups and the sample is assigned to the group to
which it has the highest probability of belonging. These probabilities are typically
referred to as posterior probabilities of group membership.
For this assessment, sites with B-IBI scores < 2.7 were defined as "degraded" for
benthic diagnostic tool application purposes. A score of 2.7 is used in the Chesa-
peake Bay benthic monitoring programs to define benthic community degradation.
This cutoff value may differ from the threshold used by the Degraded Area method
to determine proportion of sites with degraded benthic communities, but it should be
very close to that threshold. Because cutoff values differ, diagnostic tool percentages
should only be used as a general guide for identifying potential causes of degrada-
tion. For each "degraded" site, benthic metric values were submitted to the function
and posterior probabilities of group membership calculated. Posterior probabilities
for impaired segments were then used to identify the most likely source of stress
affecting benthic communities in these segments. Sites with posterior probabilities
of membership in the contaminant group that were greater than 0.50 were classified
as putatively contaminated.
"'HI
Insufficient and excessive abundance or biomass was determined from the abun-
dance and biomass metric scores for all sites not classified as putatively
appendix k « 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-7
contaminated. In the B-IBI, a score of 1 is assigned to total species abundance and
total biomass if the value of these metrics for the site being evaluated is below the
5th percentile or above the 95th percentile of corresponding reference values. A
score of 1 is assigned for both insufficient and excessive abundance or biomass
because abundance and biomass of organisms respond bimodally to pollution. An
increase in abundance or biomass is expected at polluted sites when stress from
pollution is moderate, such as at sites where there is organic enrichment of the sedi-
ment. Excessive abundance and excessive biomass are phenomena usually
associated with eutrophic conditions. A decrease in abundance and biomass is
expected at sites with high degrees of stress from pollution; for example, sites
affected by low dissolved oxygen. The insufficient and excessive abundance or
biomass criteria can then be used to determine the likelihood of low dissolved
oxygen problems versus eutrophic conditions for each of the Chesapeake Bay
segments evaluated.
4.0
Based on the bootstrap-degraded area procedure, 22 segments with sample size of at
least 10 were considered impaired (Table 3). Impaired segments were sorted according
to the lower 95% bound of the confidence interval of the difference between the
proportion of sites in the segment below threshold (P) and the proportion of sites below
threshold under the null hypothesis (Po), from high to low. The estimated P for the
impaired segments ranged from 28 to 76%, and the average B-IBI score was below 3.0
for most segments (Table 3). The estimates for CB4MH and CB5MH exclude the deep
trough (>12 m) of the mainstem which is not monitored because this area is subjected
to summer anoxia and has consistently be found to be azoic.
Nineteen of the segments declared impaired in this assessment were also declared
impaired by the Wilcoxon test in the 2004 assessment. Three segments (JMSMHb,
PMKOHa, MOBPHa) were declared impaired in this assessment but not in the 2004
assessment, and seven segments (LAFMHa, POCMH, POTOH, GUNOH, TANMH,
NANMH, CB7PHa) were declared impaired in the 2004 assessment but not in the
current assessment. Of the new impaired segments, the Nansemond River
(JMSMHb) and Mobjack Bay (MOBPHa) were sampled with additional effort in
2004. Previously, these two segments and the Pamunkey River (PMKOHa) had
sample size <10. Of the segments that are no longer classified as impaired, only the
Pocomoke River mesohaline (POCMH) had sample size <10 in the current assess-
ment.
4.2.
The stratified Wilcoxon rank sum test identified 27 segments with sample size of at
least 10 as impaired (Table 3). Segments impaired by the Wilcoxon test but not
impaired by the Degraded Area method were the lower Bay meainstem (CB7PHa),
Tangier Sound (TANMH), the Lafayette River (LAFMHa), Severn River (SEVMH),
appendix k * 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-8
and Gunpowder River (GUNOH). Except for the Severn River, these segments were
also identified as impaired in the 2004 assessment.
4,3.
The diagnostic tool and the insufficient and excessive abundance/biomass criteria
can be used as ancillary information to determine most likely source of stress
affecting benthic communities in segments classified as impaired. The results of this
part of the assessment should be used only as a screening tool to identify probable
causes of degradation and to prioritize segments for further study.
There is always a risk of misclassifying sites as affected by toxic contamination, low
dissolved oxygen, or nutrient enrichment, so independent measurements of sediment
and water quality should be made whenever possible. Table 4 presents the results of
the diagnostic tool and the insufficient and excessive abundance/biomass character-
ization for sites with contaminant group posterior probabilities >=0.50, and Table 5
presents the results for sites with contaminant group posterior probabilities >=0.90.
A general decision tree for segment assessment and characterization is provided in
Figure 1. Results are summarized below.
The percentages of degraded samples with a contaminant effect ranged from 67% in
the upper James River (JMSTFa) to 78% in the middle James River (JMSOHa) for
P >=0.5, with average contaminant group posterior probabilities ranging from 0.64
to 0.79. At P >=0.9 contaminant percentages ranged from 33-50% (Table 4). At the
James River mouth (JMSPHa) no samples were classified as contaminated. In addi-
tion, an examination of all samples collected indicated that only one sample had
excessive abundance/biomass and only one had insufficient abundance/biomass. In
the Nansemond River (JMSMHb), 90% of the degraded samples were classified as
contaminated with an average contaminant group posterior probability of 0.87.
Eighty percent of degraded samples had contaminant group posterior probabilities of
at least 0.90. Only three samples were collected in the Chuckatuck River/Pagan
River segment (JMSMHc), and three in the Warwick River (JMSMHd). Although the
low number of samples makes reliable assessments difficult, degraded samples were
collected in both segments and each was classified as contaminated with high poste-
rior probabilities of contaminant group membership. Although only three samples
were collected in Willoughby Bay (JMSPHd), each sample was classified as contam-
inated. Contaminated samples in this segment had an average contaminant group
posterior probability of 0.84. Additional samples are required in these segments to
determine the extent of benthic degradation and potential sources of stress.
In summary, results indicate that contaminants may account for a large portion of the
degradation in the James River, except for the James River mouth. The primary
source of degradation in the Nansemond River appears to be anthropogenic contam-
ination. Sampling was not sufficient for a reliable assessment in the Chucktuck/
Pagan River and Warwick River segments.
appendix k « 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-9
Percentages of degraded samples with a contaminant effect ranged from 50% in the
lower Elizabeth River mainstem (ELIPHa) to nearly 91% in the Eastern Branch
(EBEMHa). At least 80% of degraded samples were classified as contaminated in
both the Southern Branch (SBEMHa) and the Lafayette River (LAFMHa) and 68%
were classified as contaminated in the upper Elizabeth River mainstem (ELIMHa).
Of the remaining degraded samples without a contaminant effect, excessive abun-
dance/biomass was found in 9.1%, 12.5%, and 5.3% in the Western Branch
(WBEMHa), Southern Branch (SBEMHa) and upper Elizabeth River mainstem
(ELIMHa), respectively, indicating the potential of stress due to eutrophication.
Only one sample had excessive abundance in the lower Elizabeth River mainstem
(ELIPHa). Insufficient abundance/biomass was found in 12.5%, 5.9%, and 15.8% of
the degraded samples without a contaminant effect in the Southern Branch
(SBEMHa), the Lafayette River (LAFMHa) and the upper Elizabeth River
(ELIMHa), respectively, indicating low dissolved oxygen as an additional source of
stress to benthic communities in these segments.
In summary, the predominant source of stress to benthic communities within the
Elizabeth River is anthropogenic contamination. Both eutrophication and low
dissolved oxygen appear to be additional sources of stress within the Southern
Branch (SBEMHa) and upper Elizabeth River mainstem (ELIMHa).
None of the upper Pamunkey River (PMKTF) samples had B-IBI scores <2.7, so
none were assessed by the diagnostic tool. Over 57% of the lower Pamunkey River
(PMKOH) degraded samples were classified as contaminated by the tool, but the
average contaminant group posterior probability was low at 0.62. One additional
sample in this last segment was not classified as contaminated and had insufficient
abundance/biomass. Few samples were degraded in the upper Mattaponi River
(MPNTFa), and 67% of these were classified as contaminated. However, the average
contaminant group posterior probability was low at 0.65 and no samples collected
had a probability of contaminant group membership >=0.90. No samples were clas-
sified as having excessive or insufficient abundance/biomass within this segment. In
the lower Mattaponi River (MPNOHa) 80% of the degraded samples were classified
as contaminated. The average contaminant group posterior probability in this
segment was high at 0.87 and group membership probabilities for all samples clas-
sified as contaminated were >=0.90. No uncontaminated degraded samples had
excessive or insufficient abundance/biomass. In the middle York River (YRKMHa)
64% of the degraded samples were classified as contaminated. An additional 9.1%
of degraded samples had excessive abundance/biomass and were not classified as
contaminated by the tool, while 12.1% of the uncontaminated degraded samples had
insufficient abundance/biomass. In the lower York River (YRKPHa) only 46% of the
degraded samples were classified as contaminated. An additional 9.1% and 27.3% of
uncontaminated degraded samples were found with excessive abundance/biomass
and insufficient abundance/biomass, respectively, in this segment. In Mobjack Bay
(MOBPHa), 50% of the degraded samples were classified as contaminated, all with
contaminant group posterior probabilities >=0.90. An additional 12.5% and 25% of
appendix k » 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-10
uncontaminated degraded samples were found with excessive abundance/biomass
and insufficient abundance/biomass, respectively. Insufficient sample size in Severn
Creek (MOBPHe), Ware River (MOBPHf), and East River (MOBPHh), precluded
reliable assessments of degradation within these segments.
In summary, contaminants are likely to be substantial contributors to benthic
community degradation in the York River, particularly in the lower Mattaponi River
(MPNOHa) and the middle York River (YRKMHa). Contamination sources of stress
are unlikely in both the lower York River (YRKPHa) and Mobjack Bay (MOBPHa),
but both eutrophication and low dissolved oxygen may affect benthic communities
in these segments, as well as in the lower York River (YRKMHa).
All of the degraded samples in the upper Rappahannock River (RPPTFa) were clas-
sified as contaminated. Only five samples were collected in the middle
Rappahannock River (RPPOH), making assessments of benthic community degra-
dation unreliable. In the lower Rappahannock River (RPPMHa), 67% of the
degraded samples were classified as contaminated, with an average contaminant
group posterior probability of 0.67. The remaining degraded samples that were not
classified into the contaminant group had insufficient abundance/biomass. Only
eight samples were collected in the Corrotoman River. One of these samples was
classified as contaminated and another as uncontaminated with insufficient abun-
dance/biomass.
In summary, degradation in the upper Rappahannock River (RPPTFa) appears to be
the result of anthropogenic contamination while degradation in the lower Rappa-
hannock River may be the result of a combination of contamination and low
dissolved oxygen effects. The small number of samples collected makes assessments
of overall benthic community condition in the middle Rappahannock River
(RPPOHa) and Corrotoman River (CRRMHa) difficult but, the degradation observed
appears to be from a variety of sources in both segments.
Fifty percent of the degraded samples in the upper Potomac River (POTTF) were
classified as contaminated by the diagnostic tool. None of the uncontaminated
degraded samples had excessive or insufficient abundance/biomass. In the middle
Potomac River (POTOH), 80% of the degraded samples were classified as contami-
nated. Of the uncontaminated degraded samples, 20% had excessive
abundance/biomass and none had insufficient abundance/biomass. In the lower
Potomac River (POTMH), 31% of the degraded samples were classified as contam-
inated. Of the remaining degraded samples classified as uncontaminated, 65% had
insufficient abundance/biomass while only 2.6% had excessive abundance/biomass.
In summary, benthic community degradation in much of the upper Potomac River
(POTTF) appears to be the result of anthroprogenic contamination. In the middle
Potomac River (POTOH), the primary source of stress appears to be contamination;
appendix k * 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-11
however, eutrophication is likely to also affect benthic communities in this segment,
as indicated by the samples with excessive abundance/biomass.
The predominant source of stress in the lower Potomac River (POTMH) appears to
be from low dissolved oxygen, as indicated by the high percentage of samples clas-
sified as uncontaminated and having insufficient abundance/ biomass.
An inadequate number of samples were collected in the upper Patuxent River
(PAXTF) and middle Patuxent River (PAXOH) for assessing benthic community
degradation using the benthic diagnostic tool. In the upper Patuxent River (PAXTF),
two samples were classified as contaminated and one had excessive
abundance/biomass without likelihood of contamination. In the middle Patuxent
River (PAXOH), three samples were classified as contaminated and none had exces-
sive or insufficient abundance/biomass. In the lower Patuxent River (PAXMH), 46%
of the degraded samples were classified as contaminated, with an average posterior
probability of contaminant group membership of 0.51. Of the remaining uncontam-
inated samples, 50% had insufficient abundance/biomass while only 1.5% had
excessive abundance/biomass.
In summary, accurate assessment of benthic community degradation in the upper
Patuxent River (PAXTF) and middle Patuxent River (PAXOH) requires additional
sampling; however, available data suggest contaminants may be a source of stress in
these segments. Degradation in the lower Patuxent River (PAXMH) is likely to be
the result of a combination of contamination and low dissolved oxygen stress.
Over 38% of the degraded samples in the lower Chester River (CHSMH) were clas-
sified as contaminated. Of the remaining uncontaminated samples, 11% had
excessive abundance/biomass and 33% had insufficient abundance/biomass. Benthic
community degradation in this segment would appear to be the result of contamina-
tion, eutrophication, and low dissolved oxygen effects. All other segments in the
Chester River had low sample size.
Accurate assessment of benthic degradation the upper Choptank River (CHOTF),
middle Choptank River (CHOOH) and Choptank River mouth (CHOMH1) will
require additional sampling. In the lower Choptank River (CHOMH2), 67% of the
degraded samples were classified as contaminated, with group membership proba-
bilities >0.90. Of the remaining uncontaminated degraded samples, 22% had
excessive abundance/biomass while 11% had insufficient abundance/biomass. Cont-
amination appears to account for most of the benthic community degradation in the
lower Choptank River (CHOMH2), but eutrophication and low dissolved oxygen are
also likely to play a role.
appendix k » 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-12
Pocomoke River segments had low sample size; however, most of the degraded
samples in the lower Pocomoke were classified as contaminated.
Again, Pocomoke Sound had low sample size; however, 75% of the degraded
samples were classified as contaminated by the benthic diagnostic tool. Twenty-five
percent of the uncontaminated samples had insufficient abundance/biomass. Results
suggest that benthic community degradation in this segment stems from a combina-
tion of contaminants and low dissolved oxygen.
Of the Maryland small Eastern Tributaries, only the Manokin River (MANMH) had
adequate sample size. Seventy-five percent of the degraded samples were classified
as contaminated. Of the remaining uncontaminated and degraded samples, 25% had
insufficient abundance/biomass.
In the Gunpowder River (GUNOH), only 17% of the samples were classified as
contaminated. Of the uncontaminated samples, 50% had insufficient
abundance/biomass and another 17% had excessive abundance/biomass. The
predominant source of stress to benthic communities in this segment appears to be
low dissolved oxygen. In the Magothy River (MAGMH), 38% of the degraded
samples were classified as contaminated. Excessive abundance/ biomass was
observed in 13% and insufficient abundance/biomass in 50% of the uncontaminated
degraded samples. Results suggest a mixed source of stress. In the Patapsco River
(PATMH), 58% of the degraded samples were classified as contaminated. The
remaining degraded samples had insufficient abundance/biomass, suggesting
contaminants and low dissolved oxygen as sources of stress. In the Severn River
(SEVMH), 60% of the degraded samples were classified as contaminated. An addi-
tional 20% and 40% of the uncontaminated degraded samples had excessive and
insufficient abundance/biomass, respectively. Results suggest a variety of sources of
stress for this segment.
Sixty-seven percent of the upper Chesapeake Bay (CB1TF) degraded samples had
possible contaminant effects, and 17% of the remaining degraded samples had
excessive abundance/biomass. Segment CB2OH, on the other hand, had no degraded
samples. In Segment CB3MH, 55% of the degraded samples were classified as
contaminated while 32% of the remaining degraded samples had insufficient abun-
dance/biomass. In Segment CB4MH, 35% of the degraded samples were classified
as contaminated, 25% of the uncontaminated degraded samples had excessive
appendix k » 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-13
abundance/biomass, and 35% had insufficient abundance/biomass. Few samples in
Tangier Sound were degraded. In Segment CB5MH, 18% of degraded samples were
classified as contaminated and 82% of the uncontaminated degraded had insufficient
abundance/biomass, indicating a low dissolved oxygen effect. In the lower main-
stem, Segment CB6PH had 67% of the degraded samples classified as contaminated
and 33% of the uncontaminated degraded samples classified with insufficient abun-
dance/biomass. Segment CB7PHa had 63% of the degraded samples classified as
contaminated, but none had contaminant group posterior probabilities above 0.90
and the average probability for the segment was 0.58. Of the degraded samples not
classified as contaminated in this last segment, 13% had excessive abundance/
biomass and 25% had insufficient abundance/biomass. Finally, none of the samples
near the Bay mouth in Segment CB8PHa were classified as contaminated.
In summary, contaminants were likely sources of stress to benthic communities in
CB1TF and CB3MH, while a variety of stresses were likely in CB4MH. Low
dissolved oxygen was the predominant source of stress in CB5MH, contaminants
and low dissolved oxygen in CB6PHa and CB7PHa, and low dissolved oxygen alone
in CB8PHa.
Alden, R.A. Ill, D.M. Dauer, J.A. Ranasinghe, L.C. Scott, and RJ. Llanso. 2002. Statistical
verification of the Chesapeake Bay benthic index of biotic integrity. Environmetrics 13:473-
498.
Cytel Software Corporation. 2002. Pro-StatXact for SAS users. Statistical Software for Exact
Non-Parametric Inference.
Dauer, D.M., M.F. Lane, and RJ. Llanso. 2002. Development of Diagnostic Approaches to
Determine Sources of Anthropogenic Stress Affecting Benthic Community Condition in the
Chesapeake Bay. Prepared for U.S. Environmental Protection Agency, Chesapeake Bay
Program Office, by Department of Biological Sciences, Old Dominion University, Norfolk,
VA.
Dauer, D.M., M.F. Lane, and RJ. Llanso. 2005. Addendum to the Report: Development of
Diagnostic Approaches to Determine Sources of Anthropogenic Stress Affecting Benthic
Community Condition in the Chesapeake Bay. Prepared for U.S. Environmental Protection
Agency, Chesapeake Bay Program Office, by Department of Biological Sciences, Old
Dominion University, Norfolk, VA., and Versar, Inc., Columbia, MD.
Efron, B. and R. Tibshirani. 1998. An Introduction to the Bootstrap. Chapman & Hall/CRC.
Llanso, RJ., J.H. V01stad, and D.M. Dauer. 2003. Decision Process for Identification of
Estuarine Benthic Impairments. Prepared for Maryland Department of Natural resources,
Tidewater Ecosystem Assessments, Annapolis, MD., by Versar, Inc., Columbia, MD., and
Department of Biological Sciences, Old Dominion University, Norfolk, VA.
Schenker, N. and J.F Gentleman. 2001 . On judging the significance of differences by exam-
ining the overlap between confidence intervals. The American Statistician 55:182-186.
TMWA (Tidal Monitoring and Analysis Workgroup). 1999. Chesapeake Bay Program,
Analytical Segmentation Scheme for the 1997 Re-evaluation and Beyond. Prepared for the
U.S. Environmental Protection Agency, Chesapeake Bay Program Office, by the Tidal Moni-
appendix k * 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-14
toring and Analysis Workgroup of the Chesapeake Bay Program Monitoring and Assessment
Subcommittee, Annapolis, MD.
Weisberg, S.B., J.A. Ranasinghe, D.M. Dauer, L.C., Schaffner, RJ. Diaz, and J.B. Frithsen.
1997. An estuarine benthic index of biotic integrity (B-IBI) for Chesapeake Bay. Estuaries
20:149-158.
appendix k * 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-15
1, Habitat classification for the Chesapeake Bay B-1 B1.
Habitat Class
1 . Tidal freshwater
2. Oligohaline
3. Low mesohaline
4-1 . High mesohaline sand
4-2. High mesohaline mud
5-1. Polyhaline sand
5-2. Polyhaline mud
Bottom Salinity (psu)
0-0.5
>0.5-5
>5-12
>12-18
>12-18
>18
>18
Silt-clay (<62 ji) content by
Weight (%)
N/A
N/A
N/A
0-40
>40
0-40
>40
2, Number of samples by habitat in the original index development data files used by Weisberg et al. (1997) and
Alden et al. (2002). Calibration (Cal) and validation (Val) samples combined. Habitat Class designations as in
Table 1.
Cal + Val
Reference Degraded
Reference Good
Habitat Class
1
136
75
2
92
32
3
49
20
4-1
5
14
4-2
81
39
5-1
7
39
5-2
136
24
appendix k
2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-16
_
-C O H=
-
O
O
T3
o
13
C .ti
-l°l
01 ^ O %
t S>- P
c
01
o ""» I,
OI '4^ i; ^ 0)
Dl 0)
O 01
^ 0)
' =3 ~ .
in
01 '
: _ O
• CQ u
'C +J u O
I/! — "" **~
•^ D CQ ,-t^
2 0) CQ .§
u cu -^ dj
— u
a c
•* 0)
i;-£^
59 <" *E
Dl u
-C Si o
4-. 10 Q.
3 .E oi-
£ So
£-^
c
ro
O E a> D- -^
:3 DiE <"•§
- Q) 0{ , 10 ^
i •£ -7-, 'aT i- c
> ^^ " 8 ?
; £•- oJ "> E-o
' ^^ TO i— — Ql (U
.-t g 2. .!= - CQ c g
1/1 > ro^;c^ — —
g
g
.
. c o
- 0) CL
N is^
2|^r
Q- OJ ^^ M-
Eu aj ^,
r- U T3
TO aJ c c
^ i_ (U D
^ ^ aj o
I'.Is
O •*-> ni C
^ ^ ^
i c- *<-
^ S °
c E •£
ell
DI "" jn
01 r- 2;
w .— Q-
u
rs
s_ J c a.
dj _Q ro c
5 ££ I
O 3 i/i
— c fe =
£ ^T m ro
. _ r^ TO ^
5 rxi Dl 10
•o
1
'53
p,
a
£
X
3
a
o
X
o
g
«
3
«
i
1
s
&•
Dl
O
o
£7
^® fS
^ A
O
A
S
z !
Is
IS
a
o
X
s
>
•a
S3 «
o
9
"3
o.
a
1
PH
£
&
-.
O
,—
PH
PH
°
O
PH
PH
&
PH
"5. «
a -3
x
Segment
C*1
"n
~
X
03
X
03
O
O
0
0
p
tN
OO
0
o
0
l>
0
•n
0
o
\0
l>
o
^
POTMH
. in
^ tN
m
•— i tN
tN tN
X X
03 03
X X
03 03
>H >H
O O
O O
0 0
0 0
p o
O tN
OO OO
0 0
0 0
Ti tN
0 0
•n m
0 0
o o
o r-
r-- \o
o o
O 00
•^t tN
SBEMHa
CB4MH
O Ti
^
m m
*f 00
tN tN
X X
03 03
X X
03 03
O O
O O
0 0
0 0
o p
OO OI
0 0
0 0
•*t •>*•
0 0
•n m
0 0
o o
Ch tN
*3- in
o o
,__, •=*•
PAXMH
PATMH
m
<3- *±
in o
•— ' 01
tN tN
X X
03 03
i* >"
X X
03 03
>H >H
0 0
o o
0 0
0 0
o p
00 Tt
0 0
0 0
oo a\
0 0
in in
0 0
o o
01 •>*•
in •=*•
o o
OI •— i
OI VO
CHSMH
CB3MH
0,
*f tN
Ti tN
tN tN
X X
03 03
X X
03 03
O O
O O
0 0
0 0
p p
"3- tN
\O 00
o o
0 0
OI tN
•st W^
0 0
in in
p p
o o
00 1>
•sj- Ti
0 0
OI •— i
ELIMHa
EBEMHa
O 00
•n Tt-
•*t 01
tN tN
X X
03 03
i* >"
X X
03 03
>H >H
0 0
o o
o o
0 0
o o
o o
in •— i
0 0
0 0
tN 00
OI OI
0 0
in in
0 0
0 0
f~- OI
01 ^t
o o
00 TJ-
RPPMHa
YRKMHa
Tt in
vo
•rf in
tN -—I
tN tN
X X
03 03
X X
03 03
O tN
O O
o o
0 0
o o
o o
-H 0
0 0
0 0
tN VO
0 0
in in
0 0
0 0
t~- •— '
01 -^
o o
VO tN
•rj- tN
JMSMHa
CHOMH2
00 00
"*• Tt-
•— i tN
VO ^O
O 00
tN -—I
l> O
tN 01
X X
03 03
X X
03 03
>H >H
0 0
o o
o o
0 0
o o
o o
OI VO
0 0
0 0
tN OI
0 0
in in
0 0
o o
tN 00
OI OI
O O
•^t tN
CB5MH
YRKPHa
oo
tN 01
X X
03 03
X X
03 03
>H >H
g 0
o o
o o
0 0
o o
o o
in \sO
0 0
o o
0 0
Tf OI
0 0
in in
0 0
0 0
\0 \0
•* 01
o o
•—i O
•— ' tN
PMKOHa
MOBPHa
appendix k
-------
K-17
1 2
S ffl
O -^ •—i CO
(N (N (N
H >H >H fc
O O O O
ooooooooooo
•—'OfiOOOOOl
O O O O O O O
O(No
ooof^coocoooo
,_;,_;,_;o'o',_;o',_;,_;,_;
o
oooooooooooooooooooo
QQ^pppppO'—i '—''—! i—i '—! i—^CNCNCNCN
ooo^^^^^^^^^^^^^^1^^
o o o o
o o o o o o o
pppppppppppppppppppppppppp
oooooooooooooooooooooooooo
oooooooooooooooooooooooooo
01
D
C
'+-•
C
o
u
cot^m-— i
K ^3
K ^
Se
ffl S
5 ffl
CB
tS K K K S K K
£ S S o I S I
53 u o K 3 ^ 2
§ £ 2 CQ 2^ M U
appendix k
-------
K-18
•o
2
'53
&
1
B!
a
o
X
S
a
Vl
13
W)
3
1
si
o
o
*I
*3
S
z 7
A
§3
12
a
o
X
S
^
p
•o
•o «
53 «
M^
0
u
1
B
u
&
£
'o
PH
-.
U
^
PH
fe
°
O
PH
O
PH
PH
±
g"|
x
Segment
o
S
-
«
^
(N
'
,
t>
O
O
o
o
o
~
1
•
00
o
o
o
f-1
(N
O
00
CRRMHa
•n^r-c^c^c^c^oooog'r)0o'nog0
P2S9SSSgSSg°SSPPSoo
— — — — — — — — — °
^^^-^^^^,^-^^^-^^^^^^,^0
ri(^oon(^t^^^>n^O(N(^C^
mO>C*^OOO>OOOOOOOOOOOf^t~^C*^OO>OO
000000^-^^^0^0000-^00
0^).00^VO(N(NVOVOrfrsu-i00^oomu-ivo
ddddddddddocpdddddod
ooooooooooooooooooo
ooooooooooooooooooo
^^mo^t-t-^^^mo^t^^ooo^
^^t^^tc^^^cN«>«>n>n>«^-^-^-^-^-^-m
J— < , , iTJHHrt HHHHHH^ ^ T3
§§l|||§llilll§ilill
rTi HI ^ HI 1 1 o HI PI o ^ nn o rTi nn ^ TT i . HI ^
m
t~
-
«
oo
(N
'
,
r-
oo
(N
p
o
,__,
o>
o
1
•
<3\
o
o
o
^
C^l
o
m
JMSMHd
1^
G
«
«
rf
r-r>
'
,
o
o
o
00
o
1
•
00
o
o
o
m
m
o
f-1
JMSMHc
o
o
o
o
m
m
^
m
'
,
o
o
(N
^
o
1
•
u-1
o
°
o
o
o
o
o
m
MIDOH
r-
G
(N
(N
(N
f-1
'
,
(N
o
0>
o>
o
1
•
u-1
°
o
o
o
o
o
m
SASOH
o
o
-
rf
fl
'
,
(N
o
«n
00
o
1
'
«n
o
o
o
o
o
(N
RPPMHm
T3
0)
C
O
u
appendix k
-------
K-19
oooooooooo^^^oo
(NOOOOOOO^O^^^^O
t^t^OOOOt^"^
i A
•a «
2 2
$
I
o ^ *r*
^t- i
^H T3 « O HH M
KHhHtijHHhH rT tti rn rn
S£HS£^H^O£
H^hwwSofcffi"
^^<^r^ffiffieL,Og
&>nS>^>^oo<;mS
appendix k
-------
K-20
01
> 01
-* c
s?
Q.J'
T l-\
~
23 II-
-
.2
°|^|-^||
o en CD — —
~ 01 c °--o
_ ' "D 'F E <"
±;'o to H- c CD HE
co o o J3
-^o f
^ 0 ^
Q S
o
^o "^
^ a
0 (D
Q
2
"o
<« - ^
° -2 o
0s H ^
^*
.
^1|
0 (D -<
Q S
T3
-^ Q
|1
0 (D
Q
2
"o
H
t<-i —
o J2
^0 0
o- H
"S
o "^
VO g)
Q
^
"o
H
§ o a
O PH DH
(D (D
r^ f "^
| °|
9
m
1
5b
(D
t/1
0
d
i
,
0
p
d
i
,
0
o
0
,— J
,
.
, 1
p
fi
APPTFa
0 0
o o
d d
1 o
q
d
1 0
o
0
0 0
o 8
d d
1 o
p
d
1 t>
vo
vo
O •— '
o t>
o "n
d co
fN
1 t>
VO
vo
vo
1 o
^
o
in Tf
t> tN
t*i t*i
CHKOHa
JMSTFa
in
vo
vo
vo
vo
p
d
Q
d
| !
1 1
1—1
^
fN
CO
,_;
^
CO
t>
t>
t>
fN
CO
t>
o
fN
fN
0>
fN
JMSOHa
^ 0
i-; p
tN d
o o
o o
•* d
0 0
o o
o o
^ Tf
0 •*
t> o
"1 p
fN d
o o
o o
•* d
0 0
o o
•^t o
fN t>
VO t>
t*i in
•— i fN
cK vd
0 0
o o
fN O
t^ o>
^
fN O
fN O>
•* VO
VO CO
o o
vo vo
t- Tf
fN fN
JMSMHa
JMSMHb
o o
o o
d d
o o
p p
d d
0 0
o o
o o
2
-H o
§ o
d d
o o
p p
d d
0 0
o o
0 0
^ ^
f-1 f-1
t*"l t*"l
iri c^i
c^i m
0 0
O O
^ ^
in t>
in ^t
co in
0 0
fi t*i
i— i CO
C^i tN
JMSMHc
JMSMHd
o o
o o
d d
1 o
p
d
1 0
o
0
0 0
§ o
d d
1 o
p
d
1 0
o
0
o o
0 0
o o
d d
0
1 0
o
^
• 00
CO
CO
0
O c*"i
"*• ^
^ , ;
JMSPHa
JMSPHd
0 0
o t>
0 r^
O CO
O CO
d "n
t^ in
fN VO
t> t>
tN •— i
^ ^
tN O
in d
o> o
o o
cK d
**• -H
vo t>
vo vo
t> ,— 1
^ ^
i— i in
tN "n
rf "n
t^ Tf
t> fN
tN CO
t> CO
^
t*"l fl
CO O>
fi t^
t> CO
o o
o> t^
•— i fN
^ ^
tN fN
WBEMHa
LAFMHa
^
vo
d
Q
p
CO
vo
0
o
Q
CO
tn
vo
CO
o
t>
p
fN
SBEMHa
0
o
d
o
o
d
o>
^
o
p
d
o
o
d
^
vo
fl
vo
t>
t>
vd
vo
| !
O>
O
^
^
o
CO
o
in
fN
fN
EBEMHa
i— i CO
co in
o> t^
in vd
i i i i
*± m
co m
VO fi
^ ^
t> fN
O CO
t> CO
fN m'
vo t>
fN VO
in vd
tN t>
fi VO
vo vo
fN •— i
in •— i
rf "n
in t>
m -H
fN O
^" O
CO O
vo "n
^
CO O>
in -et
t> CO
vo •*
o o
t> t>
in co
fN fN
ELIMHa
ELIPHa
appendix k
-------
K-21
T3
0)
C
O
u
11
JD g
m
Q £
§ I S
a
a
•—' O
l^,—I,—IOO-—'OOOO
en en en
en -—I en en
•—i o in •—i
O •—i •—i
•—i Tf •—i O>
encNen(N
-------
K-22
T3
0)
C
O
u
CJ
3
1
o
o
"3
O
o
.
I
I
•a
es
(D O
o o
Q fi
13
W
i
PH »
|£
Q S
~a °
£°°
Q S
o 2
^j2
111
O PH PH
1-sl
I I
m
5
o •—i •—i
O -—( ^H
fe K
I °
5 5
TF
§ 3
9 y
8||«
appendix k
-------
K-23
T3
0)
C
O
u
1
rn S
"5 O
'5 ffl
|> M-|
(/; fl}
fl} O
!"!
rt S
™ M
"o ^
"^ *-
£3 H
fl} O
1
$
.ts o
^ m
U
-a s>
-§ •*•
S §
txi-Z
IB ^
Q IB
W
«
|g
3 ^
O -
s
«> 'i
o> jjj
1- o
| PH
GO
"^
to "^ ^
^ H _o
^
,
fl} "S
^^8
*£!
T3
fl}
^M "^
^ S1
*
*rt
o
H
^J
fl
Q g ,Q
vO O W
^ H o
. fl} "S
^|3
Q "^
T3
Q— -rt
O ^
^ OD
Q
2
"o
H
VO £
ff- H
=« -§
o a
,~ hn
0
o
o
o
o
p
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
00
oo
•— '
o
p
o
o
en
oo
o
^_l
•— '
en
RPPTFa
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
^
o
o
o
o
o
o
,__,
^
o
m
en
RPPOHa
en
en
vo
V~i
vo
en
en
'xf
, I
•n
en
o
o
o
o
o
o
en
en
vo
•— '
00
^
vo
en
en
en
vo
en
o
CN
VO
o
00
o^
vo
RPPMHa
o o
o o
o o
o o
o o
o o
o o
o o
o o
o
•— ' O
o o
o o
o o
o o
o o
o o
o o
o o
o o
o
O •— '
o o
o o
o o
o «n
^
o o
o o
o o
o o
t> -H
^ s
o o
•— i
vo
vo
vo
rs
o
o
o
o
o
o
o
o
o
o
o
oi
en
en
en
en
en
o^
vo
o
00
(N
CRRMHa
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
^
o
o
o
o
o
o
o
o
,__,
0?
o
,__,
^
MATTF
o o
o o
o o
o o
o o
o o
o o
o o
o o
O (N
0 VO
p t>
o -^t
o o
o o
o o
(N
o o
o o
o o
V~i "3"
«n
•— ' •— '
O O
p p
o o
«n oo
\C -H
o oo
vo vo
o o
(N •— '
•— ' (N
en en
POTTF
POTOH
«n
O"i
•^j-
T-j-
O"i
s
,__,
en
00
00
oo
vo
o
CN
(N
O
(N
0>
'xf
VO
-
^
en
vo
(N
•— ;
en
(N
^
en
vo
en
o
,__,
ON
^
POTMH
o
o
o
o
p
o
o
o
o
o
^
vo
vo
o
o
CN
o
o
»n
(N
^
en
en
en
en
o
p
o
,__,
vo
oo
en
o
vo
vo
(N
PAXTF
o vo
o en
o o
en
o o
o o
o o
O 00
o oo
>/~i o
(N 00
^
o oo
o o
0 t>
p ^
o •-<
o
(N
^ ^
VO 00
oo vo
(N l>
^
O V~>
iS ^t
en
t> 0
t-^ 00
vo o
vo «n
0 0
t>
-------
K-24
T3
0)
C
O
u
11
CD $>
Q
•s
+3
I
•S
a
I
H
es
w
w
?
J I
"a o
8 "
E-i o
O
o -<
Q »
8
la
H o
S O
g>-°
Q £
^
^
o
m
5
m o ^ o ^
o ^ ^
^ o ^ m ^ ^
oo oo en en
a a a
2oo
2 § ^
ra S w
a a a a a
2 g 2 Q 2
Z 5 Z
MA
NA
C
appendix k
-------
K-25
T3
0)
C
O
u
CJ
1
•g
H
CJ
&
CC
1
-•s .2
j> M-|
X CD
CD O
"75 fl
II
1X1 ^
IB ^
&b -^
(D y
B
KH
s
-B 1
.tS 0
* m
f J
X §
"^ rC
-§ <
«d
o
^
^
(N
a
BACO
,_,
^
en
en
en
en
en
en
en
•—(
o
p
o
o
p
o
en
en
en
I~H
r-j
CJ
CJ
t>
vo
vq
CJ
K
O
o
o
o
CN
O
o
o
o
o
o
en
^
vq
vo
[-^
vq
vo
^
vq
vo
i~H
vq
vo
t>
vo
vo
^H
0>
vo
o
^
^
CJ
K
o
o
en
en
o
o
o
o
p
vo
CO
CO
o
en
en
vo
o
o
o
en
M
(
t~^
CO
en
t>
^
(N
vo
vo
vo
CJ
o
p
o
o
p
o
en
en
en
en
I~H
CO
t>
vo
vo
vo
vq
o
p
o
CJ
o
o
o
vo
en
CO
o
en
o
p
o
vo
en
fx|
2;
t>
CO
(N
O
p
o
o
p
o
^.j.
•— ;
«n
^
en
t>
o
CO
-------
K-26
T3
0)
C
O
u
I
So
CJ
C/3
'u
o
•a
m
I
8,
o
o
1
r* &
^ O
'S ffl
X 03
03 C3
t/3 S
"o <
S. §
(tl
a
KH
1
.tS 0
S m
|1
||
C/3 -3
"^ J^
-S <
a g
bj, ^
03 ^
Q aS
o
X
w
+-
§
n ^
I?
"S ^
o -
^ o
03 g
1|
5^
"^
^o *rt °
^ H ^
^
,
I"3 3
^
v§
t~
^d
>J3
OJ
o
p
o
o
p
o
01
01
01
01
'~H
o
p
•n
01
r*"i
^H
O
o
^j-
CJ
EASMH
o
o
•n
O
p
o
o
o
o
o
o
^H
o
p
o
o
p
o
o
o
o
o
o
p
o
o
o
o
o
2
o
^-j-
«
FSBMH
o 01
O O1
O O1
0 O
o \o
§ vS
0 O
o ^o
o vd
O '^O
i-H CS
0 t>
o ^
o ^J
O 01
p 01
O 01
o 01
O 01
o 01
O -— (
o o
p p
o o
o o
o o
o o
o o
vo l>
*3~
o
p
in
'
^
t> '
\£)
'^O
Tf O
•n o
o o
i — i o
t> '
^q
Tl- 0
i>
in
o
o> o
-H rt
•— i 00
CBITF
CB2OH
0>
O1
\sO
'^O
cs
S
0>
^
t>
O1
CN
Tl-
, — J
01
cs
01
oo
^o
o>
01
^
00
[-^!
cs
Tl-
co
Tf
m
o
^
00
Tl-
•n
o
,__<
\o
S
CB3MH
o
o
m
— < in
•n 01
0
-—<
m o
o o
01 in
t -H
S S
CB7PHa
CBSPHa
appendix k
-------
K-27
0)
> 01
oirSoTC. ?>•£
m £ U g +; Q- .&.
.
TO 01
_C CD o CD to n
•1- = E8>E =
> 0) O CD <" S —
_ ;- u 10 -o .2 >
S^^ U -So.
o
c
CD ^ CD Ol
.2
j_ -O O Q)
o a; >, Q.
*- c^-o
o 01 CD -^ '-
- -
.•t: o i/i H- c CD ^
c o o o ^5 10 '(/,
E"occno,
S ol fe
CD l/l l/l
•J3 (- O
' -°
QtCQ-DicDO.b
m
rt
1
W
•a
03
^H
U
>
rt
I
H m
S 8
"ft 9
a -a
a g
"o t^p'nco
ooocoinoooooinc^icjoc^in
1 'O^OOOO 'O^COOO(Nt~^
O^OOOO OOCOOOc^^)
d^coddd do^m'ind^d^d
"T 04 i— i CN i— i
• 'or-0000 'or-mo^^c^
O^OOOO OCJ^OOCOC'^
°g§§§° °stsCTNmm
^
000>J=0^--OOOmmO-l^(N
OOOOCMOOOOOVOOCTNO^OO
OOOOTiOOOOOC^t^ooO — oo
OOOO^OOOOOOl/"lc*^*3'Oooi/~i
1 '000000 'OCTNOOOOCTNt^
OOOOOO OOOOinOt^^O
OOCJOOO OCTN^t^O^^
' 'G-oooo '8S"oSSG
S-?i°oo °SSSSSS
00««vot^««00t^«vot^in«
^ cmpmmqmoo-.'TpCTN>c
OOr^^H^?ic^c^Oc^c^^w"iS^-^
1 •OmOOOO'mTt^OIN'TO
oc^oooo C^^O^T'/^OOOOO
§o c^i c^i ^d t^ — ^d o
^0 m>cl^>coomin
« tN
1 'oc^cso1/^^ 'ooc^c^^o^roo^
^OO^T^OOO"/^ C^C^t^^^f^OO
OOOOOO OOOOOOO
^'/"l^C'J^O^OC^C^OC'^^C^C^i/^C^C^
«(NT« « «(NT«m«
Ot^C^^t^^^oo^t^TTTTOCS'/^oo
aaacacs^o-aa-iscaacscsaa
feffi^aaaffiffiffiffiaffiaaaa
appendix k
-------
K-28
T3
0)
C
O
u
rt
•w
m
1
_ a
-£ o
'I? rn
!> ty
M O
*& B
11
"o ^
S S
W) -^
Q S
S
a>
J= I
.tS .0
3l
d, C
S -^
M §
•^ "5
•rt
rt O
ob -^
a> M
Q o
X
W
-g
-S ^
1 ?
o ..^
^ S
•^ PH
's .s
J2 -2
||
GO
"tf
^o 13 o
^ H ^
S
.
^ 1 0
Q ~g
*^
Q— -rt
O 5
^ oc
Q
2
^s
H
=4H —
O ^3
VO 0
ff- H
T3
. p
° s
^ o1
Q
^te
1
H
0 0 2
QJ PH PH
a> a>
-^ Cj, *Q.
1 ° 1
a
ffl
(D
1
o
o
o
o
o
o
o
o
o
(N
O
p
o
o
o
o
o
p
o
^-
«
\0
en
\^j
en
o
p
o
CO
•^t
^
CO
CO
o
"
-
MPNOHa
o
o
o
o
o
o
o
o
o
o
o
p
o
o
o
o
o
o
o
o
o
p
o
o
o
o
o
I_H
o
^o
o
en
en
MPNTFa
o>
o
o>
o>
CJ
T^
•— (
0>
CN
i-H
>— <
O
p
o
o
o
o
^o
CO
CJ
^
m
t^
C-l
[-^
CJ
\0
CO
CJ
en
o
o
^o
o
"
S
PMKOHa
o
o
o
1
,
o
o
p
o
1
1
o
o
p
o
,
o
,
S
PMKTFa
CO O
en o
o> o
CO O
•— i O
CO O
•— (
o o
en o
o o
en
o o
^
CO O
en p
o> o
o
CO O
•— i O
CO O
--H O
-H
o o
p p
T} O
CO O
•*t o
CO O
\^o o
en r-
en en
^ *—!
o o
^0
T} t>
YRKMHa
YRKMHb
-3- o
en o
o o
•— < O
^
l> o
o
o o
CTs O
o> o
p p
o> o
•— < O
-3- o
en p
o o
t~^ o
CN O
l> O
en o
\o ^o
CN -^t
o o
MOBPHa
MOBPHe
o
o
o
o
^
o
o
o
o
o
o
o
o
"~H
>— 1
o
p
o
o
o
o
o
o
o
o
o
p
o
o
o
o
o
^
o
o
en
MOBPHf
o o
o o
o o
o
^
O '
o
o
o
O '
o
o
o
"~H
•— < O
o o
o p
o o
O '
o
o
O '
o
o
o o
0 0
o p
o o
O '
p
o
o o
CO '
••/"I
o
t> r-;
MOBPHg
MOBPHh
appendix k
-------
K-29
0)
D
C
'+->
C
o
u
'g m
>•§
(D ^
Q IB
w
"S £
a::
o w
H o
"i S
"!1
1||
O PH PH
m
5
•—< O CN
CO O '—>
O -—( CO O -—( O -—(
•—I .—I O -—( O -—(
-H O t> -H
rt rt
PH a
appendix k
-------
K-30
T3
0)
C
O
U
I
o
"B
CJ
.fe
^
I
6
O
1
H
a
es
w
-
es
a -2
u "S
03 O
fl
11
3 §
£ in
o. a
11
-------
K-31
T3
0)
C
O
u
.
I
I
•g
+3
I
es
w
•a
X
X
(H 3
•S o
'>? CO
u "33
1-3
"2^
H
(D y
Q fe
a
KH
»
.s a
*S ,0
o "53
"^H fl
1|
5 §
"^ r^
03 -^
« 03
00 -S
^ X
Q 03
X
W
•g
a °
'a 3
a 0
"S
o ^
O o
-£H CL-
5 0
CO 'C
oi> jj
& o
3 ^
c/3
"o
^
"o
^
o
xO
'o
^
o
o
;_<
-2
|
•a
"o
H
Ti
1
00
03
Q
o
+-
0
•rt
B
1
00
03
Q
"o
•sO
t*H
o
0^
^
o
^x
°
"x
O
CM
t!_,
°
m
m
Segment
0
O
0
£
"S
0
O
o
Is
(D
1
Q
^te
•a
"o
H
oo
o
>n
0>
K
1
o
p
o
1
,
o
o
p
o
1
,
o
o
p
o
1
o
1
,__<
p
K
BOHO
o
p
o
o
o
o
o
o
o
o
o
p
o
o
o
o
"-^
o
o
o
o
CN
o
p
o
o
o
o
o
^
o
00
^o
o
p
>/~i
CJ
^H
1
o
en
rt
(
en
K
MANM
o
p
o
o
o
o
o
o
o
o
o
o
o
o
p
o
o
p
o
o
'~H
o
o
p
en
K
NANM
o
p
o
o
p
o
o
p
o
o
o
o
o
o
p
o
o
p
o
o
o
o
o
o
p
o
o
1
o
en
CN
en
K
O
t/i
o
o
o
o
p
o
o
p
,_H
o
o
o
o
p
o
o
p
o
CJ
en
en
en
en
o
p
>/~i
t^
en
en
oo
o
o>
oo
a
o
appendix k
-------
K-32
CJ
1
•a
H
CJ
&
C
o
u
&i
•a s
a .a
2 ,a
a i
Q
"83°
^t2^
=" T3 O
° 2 O
=* -a o
° 2 O
^ S>-°
Q £
•5 ts •§
6 (2 <§
na
S
m
O ^H
O ^H
O ^H
o ^H cn C^ oo
m m o ^H
^H o oo t^
m oo ^H
C^ t~^ ^H c~J
CN CN tN tN
a a
o o
Q a
<; x
CQ CQ
a a a a a a a
2 § § I
o
Q > 5
a w o
S «> x
appendix k
-------
K-33
0)
D
C
'+->
C
O
u
1
rn S
"5 O
'I? rn
j> y-j
03 O
C/3 3
"03 <3
•^ *-
s. §
i>-s
Q £
>9
X
a
.a a
.tS 0
£ ffl
-1 8
f 1
5/3 3
"^ Js
•8 <
§)•£
$1
o
X
W
+-
IS
§ II
o ^
j* in
'•S o
X 'C
Q, X
3 o
9
t/i
"JH
to "^ °
"Q O
^ H .0
^
,
03 "S
^ T3 O
° 2u
^ l-a
Q £
o
'o "«
*!
%
13
o
H
^_J
fl
Q g ,Q
vo 0 ^
Is
.
03 "S
^ T3 O
0 rt Q
^
(N
0
^
VO
FSBMH
0
o
o
CN
O
o
0
o
o
o
o
0
(
o
q
o
o
q
o
0
o
0
o
o
q
o
o
o
o
vo
vo
O
vd o
01 '
01
01
01
01 '
01
01
01
•— i O
o o
0 0
0 0
q
o
o o
,__< 1
vo
0
VO 01
in *f
(N 01
LCHMH
MIDOH
0
o
in
CN
O
q
o
m
o
o
o
in
(
O
q
•n
O
q
O
m
0
O
0
in
^
O
q
o
o
o
o
t>
vo
(N
m
0
^
rs
01
NORTF
(N O
^" O
0 0
o o
in O
(N 0
VO
§o
o
in O
t>
vo o
O O1
0 m
0 0
O O1
O O1
O O1
01
0 O
o vo
o vo
vo
o -3-
CO 01
0 m
(N 0
O O1
in 01
oi 01
'—> 01
^ -H
o> vo
(N in
0 0
CO O>
(N --H
01 01
TANMH
CBITF
o r-
O CO
O l>
CN
1 T-j-
co
in
, ^
i-H
^
o
O 01
(N
o -3-
o vo
o --5
1 01
CN
O1
i co
vo
o>
O 01
O in
q t>
O ^
1 01
q
CN
O O>
, t^
CO
m
0
0 -H
CO !>;
01 (N
CB20H
CB3MH
vo o>
CO O
(N -3-
^" 01
o -3-
O (N
O CO
VO CO
o o
q q
o o
(N l>
01 t>
1—5 oi
(N
O CO
O CO
O in
§co
CO
in in
01
t> -H
^ m
K 5
q t^
O --H
(N (N
01 m
0 (N
01 (N
0 0
CO Tf
(N ^
^ t-
(N (N
CB4MH
CB5MH
(
i-H
^H
i-H
t^
s
^
vo
vd
vo
(N
vo
•n
in
01
01
s
01
01
01
01
^
vo
•n
01
01
01
CO
t>
o
0
CO
01
CB6PHa
in 01
o> 01
01 01
i-H i-H
o o
o o
in O
l> O
o o
q q
•n O
l> O
VO
vo o>
in O
0 0
01 in
01 "^t
01 01
CB7PHa
CBSPHa
appendix k
-------
K-34
T3
C
TO
+J
C
0)
c
o
'l/l
'u
0)
T3
hH
hH
hH
ft
64
PH
a ^
O »
Characteriza
Probable Cai
8 '-B
S a
»* ^
^^
U
^
'S ®
al application
:ool for stress
racterization
a .. 54
•2-5 "S
o. g
0 a
54
•d
u
^
'S ®
al application
;ool for stress
racterization
|.a|
al "
0 a
64
•d
•s
'35*
V
c2 s
*• s!
iagnostic tool
of probable c
egradation
•° a -d
>, .2
11
•< SB
a
2
T T
J GC GC
hH
hH
ft
64
PH
a
x
v
35
%
•<
•*-
a
1
L.
=.
hH
1
V
=.
g
54
-f-
a
-
'3
1
a
hH
64
•^
•d
11
e s
"E.
^
^
•d
^ "^ .
54 W
•3 -a
— Q,
U ^_
g O
M P
O
•d
v
64 '
It
1 •§
I
T T
hH
ft
64
PH
1 §
« "54
a. 5
1 *
0
^H
V
z
o
Z
o
^H
Al
Z
appendix k
-------
K-35
APPENDIX
Benthic Index of Biotic Integrity
(B-IBI) for 2006 303(d) List
Alternative Assessment
Methodology
Roberto Llanso, Jon Velsted, Ed Weber
Versar, Inc.
Daniel Dauer
Old Dominion University
(co-Pis)
August 23, 2005
Summary
The impairment assessment for each segment is based on the
proportion of samples with "low" B-IBI scores (i.e., below a
threshold)
Two steps, estimate:
1. Proportion of sites in a segment with scores below a threshold (P)
2. Difference between P and the expected proportion under the null
hypothesis (P0), i.e., if the segment were in good condition (no low
DO, contaminant, or nutrient enrichment problems), we would still
expect a small proportion of sites to have low" scores (e.g., because
of natural variability); this proportion under the null hypothesis is
defined as 5%.
appendix k * 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-36
Summary (cont.)
Thresholds are set for each of seven benthic habitats in Chesapeake
Bay: tidal fresh, oligohaline, low mesohaline, high mesohaline sand, high
mesohaiine mud, polyhaline sand, polyhaline mud.
The threshold is set as the smaller of two values:
1, 5""1 percentile IBI score for the good reference distribution (i.e., sites
with low scores are unlikely to come from good reference conditions)
2, Maximum observed IBI score for the degraded reference distribution
(i.e, sites with low scores are likely to come from degraded
conditions)
See example next slide for two hypothetical habitats: 1) Habitat A, the
distributions of scores for the good and the degraded reference sites do
not overlap, 2) Habitat B, the distributions overlap.
Habitat A
IBI scores
2,0 2.7
Habitat B
IBI scores
2.2 3.0
-------
K-37
Summary (cont.)
Reference distributions are sometimes based on a small number of
samples; therefore the 5th percentile score is not well defined
The 5th percentile score and its variance was estimated by bootstrap
simulations
For each iteration of the bootstrap simulation, a subset of the good
reference data for each habitat was selected at random, and the 511
percentile score determined
Over all the iterations, the 5* percentile score varied, and at each iteration
the threshold was established according to the rule described earlier
See next slide for the two habitat examples
181 scores
H.ihtll I
-------
K-38
Summary (cont.)
For each iteration of the bootstrap simulation, the assessment data are
compared to the reference data to estimate proportion of sites with scores
below the threshold
This is done for each of one or more habitats within a segment (i.e.,sonne
segments have sites in more than one habitat)
See next slides for the two examples
Hnb tot
IBI scores
2.0
-2.7-
5%
3,3
\ \
Habitat A
Habitat B
-------
K-39
Hah tit B
Habital A
Habitat B
c-
8
r\
Summary (cont.)
Example of calculations for a hypothetical segment with two habitats:
Iteration
1
2
3
n
n
10
10
1D
10
Habitat A
threshold P<*reshold
20 0.4C
2.0 a AC
2.0 0.4C
20 0.4C
n
-0
as
iO
^0
Habitat B
threshold P
-------
K-40
Summary (cont.)
Under the null hypothesis, 5% of the sites (P0) would be expected to have
low IBI scores, even if all sites in a segment were in good condition {i.e. no
low DO, contaminant, nutrient enrichment problems)
_-—JT 5% of sites
Segments declared impaired if P greater than expected under the null
hypothesis
P- P. > 0 (with 95% confidence)
Summary (cont.)
Variance components in P added
1 Variance in P due to estimating thresholds - from bootstrap
« Sampling variation within segment - binomial
Confidence interval of P - P0 =
P - Pa ± 1.96(SEP + SEP ) = P - P., + 1,96*SQRT(Varp 4 VarP )
Var^ = Variance from bootstrap =
from segment = (pq/N-1)
5000 -1
plus variance
appendix k » 2006 303(d) Assessment Methods for Chesapeake Bay Benthos
-------
K-41
Advantages of new method over Wilcoxon's
Wilcoxon
• evaluates differences in distributions based on ranks, cannot quantify
magnitude of shift
• sensitive to small shifts in distribution of B-IBI scores
New method
• estimates proportion of area below thresholds and magnitude of
departure from reference conditions
• tests if this magnitude is above specific thresholds of protection
• incorporates uncertainty in reference conditions as well as sampling
variability in the assessment data
• does not require purchase of special statistical analysis package
(Wilcoxon does)
* Both methods are suitable for data segregated into multiple habitats for
which reference distributions are not homogeneous
appendix k « 2006 303(d) Assessment Methods tor Chesapeake Bay Benthos
-------
L-1
appendix |
Addendum to the Report
Development of Diagnostic Approaches to
Determine Sources of Anthropogenic Stress
Affecting Benthic Community Conditions
in the Chesapeake Bay
Prepared for:
U.S. EPA Chesapeake Bay Program Office
410 Severn Avenue, Suite 109
Annapolis, Maryland 21403
Prepared by:
Daniel M. Dauer1
Michael F. Lane1
Roberto J. Llanso2
Department of Biological Sciences
Old Dominion University
Norfolk, VA 23529-0456
2Versar, Inc.
9200 Rumsey Road
Columbia, Maryland 21045
June 2005
appendix I • Addendum to the Report
-------
L-2
1. INTRODUCTION
Dauer et al. (2002) submitted a report to the US EPA Chesapeake Bay Program
Office on the development of diagnostic approaches to determine sources of anthro-
pogenic stress affecting benthic community condition in the Chesapeake Bay. The
objective of the study was to develop analytical tools capable of classifying regions
in Chesapeake Bay identified as having degraded benthic communities into cate-
gories distinguished by the type of stress experienced by those communities. The
tool was successful at identifying regions with high probabilities of sediment
contamination. However, prior to implementation, it was recommended that the
operational effectiveness of the diagnostic tool be further tested using additional
validation data sets.
In this Addendum the results of two additional tasks are presented. First, the linear
discriminant function was independently derived to verify the accuracy of the devel-
opment of the function. Second, two additional putative validation data sets were
used to assess the validity of the linear discriminant function.
2. LINEAR DISCRIMINANT FUNCTION
In this task it was discovered that four samples from the original calibration data set
were not included in the derivation of the final linear discriminant function originally
reported in Dauer et al. 2002. The final validation of the linear discriminant func-
tion with these additional four samples was identical to that reported in Table 21 for
the Baywide scenario, i.e. using the All Province sediment contaminant classifica-
tion, namely, with an overall percent correct classification of 75.14%. The new
coefficients for this function are given in Table 1 of this Addendum (revised Table
24 of Dauer etal. 2002).
3. ADDITIONAL VALIDATION DATA SETS
Two putative data sets were used for further validation of the Contaminant Discrim-
inant Tool (CDT) as presented in Dauer et al. 2002.
ELIZABETH RIVER WATERSHED
The first putative data set consisted of 125 random samples collected in 1999 from
the Elizabeth River watershed (Dauer and Llanso 2003). An additional 100 random
samples collected 25 per year from 2000-2003 were also used (Dauer 2001, 2002,
2003, 2004). All samples were analyzed using the CDT function and placed into
categories based upon the posterior probability of inclusion into the Contaminant
Group. Due to the high levels of contaminants recorded historically in the Elizabeth
River watershed (Hall et al., 1992, 1997, 2002; Padma et al. 1998; Conrad et al.
2004), the a priori expectation was that a high percentage of samples declared
degraded by the Benthic Index of Biotic Integrity would be placed into the Contam-
inant Group. The results from the Elizabeth River watershed are compared to results
appendix I • Addendum to the Report
-------
L-3
from the Virginia Mainstem that is characterized as having low levels of contami-
nants and accordingly classified as of no environmental concern (USEPA 1999).
Our a priori expectation was that all branches of the Elizabeth River would show a
higher percent area placed into the Contaminant Group compared to the Virginia
Mainstem. For the Virginia Mainstem the number of sites placed into the Contami-
nant Group represented 11% of the entire stratum. Consistent with our a priori
expectation, all strata in the Elizabeth River had higher proportions placed into the
Contaminant Group, ranging from 40-92% (Table 2; Figure 1). These results indi-
cate strong support for the CDT.
FOR BAY
The second putative data set consisted of random samples collected as part of the
Maryland and Virginia Benthic Monitoring Program from 1996-2002. All samples
were analyzed using the CDT function and placed into categories based upon the
posterior probability of inclusion into the sediment Contaminant Group. The a
priori expectation was that more samples collected near highly urbanized or indus-
trialized watersheds would be placed into the Contaminant Group. Results are more
difficult to interpret but the pattern of location of samples placed into the Contami-
nant Group is non-random (Table 3; Figure 2), and can be interpreted to be consistent
with known patterns of sediment contaminant distributions for the entire Chesapeake
Bay (e.g. see USEPA 1999). GIS maps show patterns of location that agree well
with a priori expectations within highly contaminated regions of the Bay such as
Baltimore Harbor (Figure 3) and the Elizabeth River (Figure 4). The maps were
made with data placed on a 100 m grid and interpolated using a two-dimensional
surface fitting algorithm.
Conrad, C.F. and CJ. Chisholm-Brause. 2004. Spatial survey of trace metal contaminants
in the sediments of the Elizabeth River, Virginia. Marine Pollution Bulletin 49:319324.
Dauer, D.M. 2001. Benthic Biological Monitoring Program of the Elizabeth River Water-
shed (2000). Final Report to the Virginia Department of Environmental Quality, Chesapeake
Bay Program, 35 pp. plus Appendix.
Dauer, D.M. 2002. Benthic Biological Monitoring Program of the Elizabeth River Water-
shed (2001) with a study of Paradise Creek. Final Report to the Virginia Department of
Environmental Quality, Chesapeake Bay Program, 45 pp.
Dauer, D.M. 2003. Benthic Biological Monitoring Program of the Elizabeth River Water-
shed (2002). Final Report to the Virginia Department of Environmental Quality, Chesapeake
Bay Program, 56 pp.
Dauer, D.M. 2004. Benthic Biological Monitoring Program of the Elizabeth River Water-
shed (2003). Final Report to the Virginia Department of Environmental Quality, Chesapeake
Bay Program, 88 pp.
Dauer, D.M., M.F Fane and RJ. Flanso. 2002. Development of Diagnostic Approaches to
Determine Sources of Anthropogenic Stress Affecting Benthic Community Condition in the
appendix I » Addendum to the Report
-------
L-4
Chesapeake Bay. Final Report to the U.S. Environmental Protection Agency, Chesapeake
Bay Program Office, Annapolis, Maryland, 64 pp.
Dauer, D.M. and RJ. Llanso. 2003. Spatial scales and probability based sampling in deter-
mining levels of benthic community degradation in the Chesapeake Bay. Environmental
Monitoring and Assessment 81:175-186.
Hall, L.W. Jr. and R.W. Alden, III. 1997. A review of concurrent ambient water column and
sediment toxicity testing in the Chesapeake Bay watershed: 1990-1994. Environmental
Toxicology and Chemistry 16:16061617.
Hall, L.W. Jr., R.D. Anderson and R.W. Alden, III. 2002. A ten-year summary of concurrent
ambient water column and sediment toxicity tests in the Chesapeake Bay watershed:
19901999. Environmental Monitoring and Assessment 76:311352.
Hall, L.W. Jr., M.C. Ziegenfuss and S.A. Fischer. 1992. Ambient toxicity testing in the
Chesapeake Bay watershed using freshwater and estuarine water column tests. Environ-
mental Toxicology and Chemistry 11:14091425.
Padma, T.V., R.C. Hale, andM.H. Roberts. 1998. Toxicity of water-soluble fractions derived
from whole creosote and creosote-contaminated sediments. Environmental Toxicology and
Chemistry 17:16061610.
USEPA. 1999. Targeting Toxics: A Characterization Report. A Tool for Directing Manage-
ment and Monitoring Actions in the Chesapeake Bay's Tidal Rivers, 1999. U.S.
Environmental Protection Agency, Chesapeake Bay Program Office, Annapolis, Maryland,
49pp.
appendix I » Addendum to the Report
-------
L-5
1. Revised Table 24 of Dauer et al. (2002). Coefficients and cutoff values for the Baywide linear discriminant
function for classifying severely degraded and degraded sites into the Contaminant and Other stress
groups using "uncorrected" data.
Variable
Isopoda abundance
Isopoda diversity
Isopoda proportional abundance
Amphipoda abundance
Amphipoda richness
Amphipoda proportional abun.
Haustoriidae abundance
Haustoriidae diversity
Haustoriidae proportional abun.
Ampeliscidae abundance
Ampeliscidae richness
Ampeliscidae proportional abun.
Corophiidae abundance
Corophiidae richness
Corophiidae proportional abun.
Mollusca abundance
Mollusca richness
Mollusca proportional abundance
Bivalvia abundance
Bivalvia richness
Bivalvia proportional abundance
Gastropoda abundance
Gastropoda richness
Gastropoda proportional abun.
Polychaeta abundance
Polychaeta richness
Polychaeta proportional abun.
Spionidae abundance
Spionidae richness
Spionidae proportional abundance
Capitellidae abundance
Capitellidae richness
Capitellidae proportional abun.
Coefficient
2.01518
-3.07226
9.45420
0.38084
-0.32010
-4.25029
-3.85522
-1.39235
34.61687
-1.57316
-1.79716
25.88958
37.26499
-18.36548
-2329.15377
2.52241
0.74909
-1.43165
-4.43466
1.28499
-0.27727
-1.23734
-0.15477
-3.82240
0.05506
0.46294
-5.08183
-0.02286
-1.89087
4.02486
0.48588
2.55550
-1.67289
Variable
Nereidae abundance
Nereidae richness
Nereidae proportional abundance
Oligochaeta abundance
Oligochaeta richness
Oligochaeta proportional abundance
Tubificidae abundance
Tubificidae richness
Tubificidae proportional abundance
Deep deposit feeder abundance
Deep deposit feeder richness
Deep deposit feeder proportional abun.
Suspension feeder abundance
Suspension feeder richness
Suspension feeder proportional abun.
Interface feeder abundance
Interface feeder richness
Coefficient
-0.28511
-0.53535
12.23099
0.43911
1.37409
-5.05367
0.33669
0.96057
-2.27273
-1.07320
-2.43057
12.57963
1.05255
-1.25065
2.17966
0.84134
-0.47052
Interface feeder proportional abundance 4.50630
Carnivore-Omnivore abundance
Carnivore-Omnivore richness
-0.05179
-0.00602
Carnivore-Omnivore proportional abun. 3.13784
Total Abundance
Total biomass
Biomass to abundance ratio
Infaunal species richness
Infaunal Shannon Wiener diversity
Infaunal species evenness
Epifauna to Infaunal abundance ratio
Epifauna species richness
Epifaunal Shannon Wiener diversity
Epifaunal species evenness
0.18311
4.75310
-123.97124
-0.04107
1.22042
-2.50732
4.41998
-0.96505
-1.11725
5.85736
Cutoff Value = 2.56645
appendix
-------
L-6
2, Percent of the Elizabeth River 1999 strata placed into the sediment
contaminant effect group using the contaminant discriminant function of
Dauer et al. 2002 (posterior probability > 0.5). Scuffletown, Gilligan, Jones,
and Paradise creeks are subsystems of the Southern Branch. Paradise Creek
sampled in 2000. The Elizabeth River strata are compared to the Virginia
Mainstem Stratum.
Stratum
Percentage of Stratum
in Contaminant Group
Mainstem of the Elizabeth River
Lafayette River
Eastern Branch
Western Branch
Southern Branch
Scuffletown Creek
Gilligan/Jones Creek
Paradise Creek (2000)
Entire Elizabeth River watershed*
Virginia Mainstem
* Area weighted value
40
60
64
72
64
60
68
92
54
11
100
90
80
70
60
50
40
30
20
10
0
>0.5
D
w
1. Percentage of stratum with a B-IBI value < 2.7 and placed into the
Contaminant Group with a posterior probability > 0.5.
-------
L-7
3, Percent of the stratum placed into the sediment contaminant effect
group using the contaminant discriminant function of Dauer et al. 2002
(posterior probability > 0.5). Data from 1996-2002. Elizabeth River data
includes the intensive 1999 event and 25 random samples of the watershed
from 2000-2002.
Percentage of stratum
Stratum N in Contaminant Group
Lower (VA) Mainstem 175
Upper Bay Mainstem 175
MD Eastern Tributaries 175
Patuxent River 175
MD Middle Mainstem 175
MD Western Tributaries 175
Potomac River 175
James River 175
Rappahannock River 175
York River 175
Elizabeth River 275
10.9
17.7
16.6
20.0
17.1
24.6
31.4
30.9
37.1
38.3
52.4
% of stratum
50 |
40
30
20 | 1
:m Q D D
SI fe « s £ B.I 1
si Ii si £ il |s
§1 3i §^ | gl *f=
a E
— |
. .
— 1 1 — 1
ffi & | ,_ S S
1 ™ I 1
2, Percentage of stratum with a B-IBI value < 2.7 and placed into the
Contaminant Group with a posterior probability > 0.5.
-------
L-8
Figure 3. Diagnostic discriminant tool results and an interpolation fitting algorithm were
used to classify Baltimore Harbor benthic communities into categories distinguished by
the type of stress experienced by those communities. Red shading indicates degraded
benthic communities stressed by toxic contamination (posterior probability in
Contaminant Group > 0.5), with higher color intensity indicating higher probabilities of
contaminant effects (>0.5 to <0.7; >=0.7 to <0.9; > = 0.9). Salmon shading indicates
degraded benthic communities stressed by other sources, most likely low dissolved oxy-
gen (posterior probability in Contaminant Group <=0.5). Green indicates good benthic
community condition. Middle Branch (mb), Curtis Creek (cc). Stony Creek (sc), and Bear
Creek (be) show contamination as likely source of stress. The deep basin north of Curtis
Bay and the deep channel southwest of Sparrows Point (sp) shows other stress (low DO)
as probable cause of degradation.
appendix I • Addendum to the Report
-------
L-9
Figure 4. Diagnostic discriminant tool results and an interpolation fitting algorithm used
here to classify lower James River benthic communities into categories distinguished by
the type of stress experienced by those communities. Red shading indicates degraded
benthic communities stressed by toxic contamination (posterior probability in
Contaminant Group > 0.5), with higher color intensity indicating higher probabilities of
contaminant effects (>0.5 to <0.7; >=0.7 to <0.9; > = 0.9). Salmon shading indicates
degraded benthic communities stressed by other sources (posterior probability in
Contaminant Group <=0.5). Green indicates good benthic community condition. The
Elizabeth River (er), Craney Island (ci), Willoughby Bay (wb), Nansemond River (nr), and
Pagan River (pr) show contamination as likely source of stress.
appendix I • Addendum to the Report
------- |